Browse the corpus

A recent review of seasonal influenza during 1998–2009 revealed a paucity of epidemiological data throughout most of sub-Saharan Africa, including Kenya [1]. In particular, there were few influenza-associated data on age-specific incidence, mortality, seasonal variation, and relationship with common co-occurring conditions, including human immunodeficiency virus (HIV) infection, malnutrition, bacterial pneumonia, and malaria. In this study, we report the results of surveillance for influenza among patients presenting to a rural district hospital and outpatient clinics in coastal Kenya during 2007–2010. We used molecular diagnostic methods to distinguish influenza A, B, and C viruses and to subtype influenza A viruses. At admission to Kilifi District Hospital (KDH), detailed clinical and laboratory data were systematically collected, and residency within the surrounding Kilifi Health and Demographic Surveillance System (KHDSS) was ascertained. The study spans a period in which a new strain of influenza A virus, 2009 pandemic influenza A virus subtype H1N1 (A[H1N1]pdm09), spread worldwide and entered the study population, the occurrence of which was monitored through influenza virus subtyping.

ealth and Demographic Surveillance System (KHDSS) was ascertained. The study spans a period in which a new strain of influenza A virus, 2009 pandemic influenza A virus subtype H1N1 (A[H1N1]pdm09), spread worldwide and entered the study population, the occurrence of which was monitored through influenza virus subtyping. MATERIALS AND METHODS Study Population and Samples We sampled children presenting to KDH and surrounding outpatient clinics located in Kilifi District on the coast of Kenya, approximately 60 km north of Mombasa. The district comprises a largely rural population of subsistence farmers and has an equatorial climate, with rain predominantly falling during April–July and October–December. KDH is the principal hospital facility for the population of the KHDSS. Further details of the study area and respiratory disease surveillance at KDH and local clinics can be found in previous reports (2–4, 13). Nasopharyngeal wash or aspirate specimens were collected from eligible children aged 1 day to 12 years from January 2007 through December 2010 and were either stored in viral transport medium at −80°C (2007–2009) prior to molecular screening or screened prior to freezing in raw form.

ound in previous reports (2–4, 13). Nasopharyngeal wash or aspirate specimens were collected from eligible children aged 1 day to 12 years from January 2007 through December 2010 and were either stored in viral transport medium at −80°C (2007–2009) prior to molecular screening or screened prior to freezing in raw form. Inpatients were eligible if they were admitted to the hospital with cough or difficulty breathing and either lower-chest-wall indrawing (defined as severe pneumonia) or 1 or more of the following: cyanosis, prostration, unconsciousness, or an oxygen saturation level <90% (a modification of the World Health Organization criteria for very severe pneumonia). Children who were not residing in the KHDSS at admission, were admitted in extremis, were admitted for elective surgery, or received a diagnosis of neonatal tetanus were excluded. The following clinical and laboratory features obtained on admission or that relate to discharge outcome were compared between influenza-positive and influenza-negative children: duration of hospitalization >14 days, very severe pneumonia, wheezing, hypoxia (oxygen saturation level <90%, by fingertip pulse oximetry), circulatory shock (capillary refill time ≥3 seconds), severe anemia (hemoglobin level <5 g/dL), prematurity, congenital heart disease, positivity for HIV antibody (by 2 rapid tests), severe underweight (weight for age Z score ≤3), slide positivity for Plasmodium species, bacteremia, concurrent viral infection diagnosis, and death before discharge [2]. Outpatient recruits were a convenience sample of children aged <13 years, enrolled for broad comparison with hospital-admitted patients, who presented with either no signs of acute respiratory infection (non-ARI) or signs of upper respiratory tract infection (URTI) [2, 5]. Individuals with URTI had 1 or more of the following: cough, difficulty breathing, nasal discharge, runny or blocked nose, or sore throat. Written informed consent was obtained from the parent or guardian of subjects. This study was approved by the Kenyan National Ethical Review Committee and the University of Warwick Biomedical Research Ethics Subcommittee.

he following: cough, difficulty breathing, nasal discharge, runny or blocked nose, or sore throat. Written informed consent was obtained from the parent or guardian of subjects. This study was approved by the Kenyan National Ethical Review Committee and the University of Warwick Biomedical Research Ethics Subcommittee. Diagnostic Real-Time Reverse-Transcription Polymerase Chain Reaction (RT-PCR) RNA was extracted from 140 µL of nasopharyngeal samples, using the Qiagen Viral RNA miniprep kit (Qiagen, United Kingdom), or from 200 µL of nasopharyngeal samples, using the total nucleic acid extraction kit (Roche Applied Science, Germany) with a MagNA Pure LC32 automated nucleic acid extractor, following the manufacturer's instructions. Diagnostic screening for viral targets was performed using real-time RT-PCR. For specimens from KDH inpatients in 2007 and from outpatients in 2007–2008, reactions were tested by multiplex real-time RT-PCR, using FRET hybridization probes as described by Lassaunière et al [6], with primers and probe targeting NS1 of influenza A virus and nucleoprotein of influenza B virus. All other samples were screened using the Taqman Qiagen Quantifast multiplex method on the ABI 7500 platform described by Hammitt et al, targeting the matrix protein for influenza A and C viruses and targeting NS for influenza B virus [3]. Concurrent targets in both RT-PCR assays included respiratory syncytial virus (RSV), adenovirus, rhinovirus, parainfluenza virus (PIV) 1–3, human metapneumovirus, coronavirus (CoV 229e, NL69, and OC43), and, differentially by assay, coronavirus Hong Kong [6] and PIV4 [3].

ses and targeting NS for influenza B virus [3]. Concurrent targets in both RT-PCR assays included respiratory syncytial virus (RSV), adenovirus, rhinovirus, parainfluenza virus (PIV) 1–3, human metapneumovirus, coronavirus (CoV 229e, NL69, and OC43), and, differentially by assay, coronavirus Hong Kong [6] and PIV4 [3]. Subtype Analysis An aliquot of each of the influenza A virus–positive specimens was shipped to the National Influenza Center (NIC) in Nairobi for subtyping. Samples were subjected to RNA extraction using the QIAamp viral RNA isolation kit (Qiagen). Detection was performed using Invitrogen SuperScript III Platinum One-Step quantitative kit with primers and probes targeting seasonal influenza A virus H1N1 (A[H1N1]), A(H1N1)pdm09, and influenza A virus subtype H3N2 (A[H3N2]).

Samples were subjected to RNA extraction using the QIAamp viral RNA isolation kit (Qiagen). Detection was performed using Invitrogen SuperScript III Platinum One-Step quantitative kit with primers and probes targeting seasonal influenza A virus H1N1 (A[H1N1]), A(H1N1)pdm09, and influenza A virus subtype H3N2 (A[H3N2]). Statistical Analysis Statistical analysis was undertaken using Stata, version 11.0 (StataCorp, College Station, TX) and Microsoft Office Excel 2003 (Microsoft, Redmond, WA). The incidence of influenza among inpatients for age group i, I(i), per 100 000 population per year was estimated on the basis of the equation I(i) = [C(i)/N(i)p(i)].100,100, where C(i) is the average number of children per year admitted who were resident in the KHDSS in age group i, N(i) the midsurvey KHDSS population in age group i, and p(i) is the proportion of eligible children tested for influenza (ie, we assumed that children who were not tested would have had the same prevalence of influenza as those who were tested and scaled the incidence accordingly). For pneumonia incidence estimates, p(i) is set to 1. The KDHSS population on 1 January 2009 was estimated to be 9451 individuals aged <1 year, 45 644 aged <5 years, and 108 708 aged <13 years. The population size and incidence estimation procedures have been described elsewhere [4]. The incidence estimation for the population proximal to the hospital was undertaken using cases involving children aged <5 years admitted from administrative sublocations within a 5-km radius of the hospital and the corresponding midpoint population estimate from the KHDSS (12 339 as of 1 January 2009). The Wilcoxon rank sum test was used to compare median ages; the χ2 or Fisher exact test was used to compare proportions, as appropriate; the Score test (procedure tabodds) was used to assess the trend in prevalence, by age; and the Poisson probability distribution was used to assess whether observed cases of influenza in specified quarters of the year exceeded the expected number of cases. Spearman rank correlation was used to test for a temporal association between monthly or quarterly numbers of influenza cases or influenza A virus infections and the number of cases of bacteremia or Streptococcus pneumoniae infection. The analysis was undertaken with cases of influenza and bacteremia temporally in phase or between 1 and 4 months time step out of phase.

mporal association between monthly or quarterly numbers of influenza cases or influenza A virus infections and the number of cases of bacteremia or Streptococcus pneumoniae infection. The analysis was undertaken with cases of influenza and bacteremia temporally in phase or between 1 and 4 months time step out of phase. The association between positivity for any influenza type and laboratory or clinical features on admission was assessed using logistic regression, adjusted for age group, to obtain odds ratios (ORs) and 95% confidence intervals (95% CIs).

mporal association between monthly or quarterly numbers of influenza cases or influenza A virus infections and the number of cases of bacteremia or Streptococcus pneumoniae infection. The analysis was undertaken with cases of influenza and bacteremia temporally in phase or between 1 and 4 months time step out of phase. The association between positivity for any influenza type and laboratory or clinical features on admission was assessed using logistic regression, adjusted for age group, to obtain odds ratios (ORs) and 95% confidence intervals (95% CIs). RESULTS Over the 4-year period, there were 2429 admissions to KDH involving individuals who were eligible for the study (57% were boys; median age, 9 months [interquartile range {IQR}, 3–22 months]). A total of 503 (21%) had very severe pneumonia (50% were boys; median age, 10 months [IQR, 2–34 months]; 55% were infants), and the in-hospital case-fatality rate was 6.5%. Of the eligible inpatients, 2002 (82%) were tested for influenza (57% were boys; median age, 9 months [IQR, 3–21 months]; 58% were infants), and this percentage did not differ among those aged <1 year, 1–4 years, and ≥5 years (P = .333); 387 (19%) had very severe pneumonia (52% were boys; median age, 10 months [IQR, 2–34 months]; 53% were infants). Stratified by severity, 84% of eligible inpatients with severe pneumonia were tested, compared with 77% of eligible inpatients with very severe pneumonia (P = .001). The case-fatality rate among inpatients who were untested was significantly higher than among those who were tested (69 of 427 [16%] vs 88 of 2002 [4.4%]; P < .0001). Reasons for not testing comprised consent refusal (63%), early discharge (26%), and death (11%) before sampling.

s with very severe pneumonia (P = .001). The case-fatality rate among inpatients who were untested was significantly higher than among those who were tested (69 of 427 [16%] vs 88 of 2002 [4.4%]; P < .0001). Reasons for not testing comprised consent refusal (63%), early discharge (26%), and death (11%) before sampling. Within the same period, 527 outpatients were recruited and tested for influenza from May 2007 through March 2008 (57 with non-ARI and 96 with URTI) and again from March 2010 through December 2010 (139 with non-ARI and 235 with URTI). Cumulative non-ARI cases numbered 15, 40, 76 and 65, for quarters 1–4, respectively, and cumulative URTI cases numbered 47, 95, 99, and 90, respectively. Overall, 196 outpatients had non-ARI (median age, 13 months [IQR, 5–26 months]; 47% were infants), and 331 had URTI (median age, 19 months [IQR, 8–39 months]; 37% were infants). Compared with the median age of inpatients, the median ages of outpatients with non-ARI (P = .004) or URTI (P ≤ .0001) were higher.

pectively. Overall, 196 outpatients had non-ARI (median age, 13 months [IQR, 5–26 months]; 47% were infants), and 331 had URTI (median age, 19 months [IQR, 8–39 months]; 37% were infants). Compared with the median age of inpatients, the median ages of outpatients with non-ARI (P = .004) or URTI (P ≤ .0001) were higher. Prevalence, Disease Association, and Incidence The prevalence of influenza virus of any type was 4.9% (99 of 2002 cases) among inpatients; 4.7% (76 of 1615) had severe pneumonia, and 5.9% (23 of 387) had very severe pneumonia (P = .299). Among outpatients, the prevalence of influenza virus of any type was 3.9% (13 of 331) among those with URTI and 0.5% (1 of 196) among those with non-ARI. Data stratified by virus type are presented in Table 1. Influenza A virus was the most prevalent type among outpatients with pneumonia (3.5%) and outpatients with URTI (3.3%). These proportions were unaltered by restricting the analysis to children <5 years of age. Among outpatients classified as having non-ARI, there were no cases of influenza B or C virus infection; influenza A virus was detected in 1 child. Table 1. Distribution of Influenza Viruses Diagnosed by Molecular Methods Among Children, by Presenting Condition, From Kilifi, Kenya, 2007–2010

o children <5 years of age. Among outpatients classified as having non-ARI, there were no cases of influenza B or C virus infection; influenza A virus was detected in 1 child. Table 1. Distribution of Influenza Viruses Diagnosed by Molecular Methods Among Children, by Presenting Condition, From Kilifi, Kenya, 2007–2010 Condition No. of Children Influenza Virus Detected, by Type, No. (%b) of Children Eligible Tested Anya A B Cc NARI 196 196 1 (0.5) 1 (0.5) 0 0 URTI 331 331 13 (3.9) 11 (3.3) 1 (0.3) 2 (0.9) Severe pneumonia 1926 1615 76 (4.7) 57 (3.5) 13 (0.8) 7 (0.6) Very severe pneumonia 503 387 23 (5.9) 14 (3.6) 6 (1.6) 4 (1.4) All pneumonia (severe and very severe) 2429 2002 99 (4.9) 71 (3.5) 19 (0.9) 11 (0.8) Total 2956 2529 113 (4.5) 83 (3.3) 20 (0.8) 13 (0.7) All children resided in the Kilifi Health and Demographic Surveillance System. Abbreviations: NARI, no acute respiratory infection; URTI, upper respiratory tract infection. a Includes influenza A, B, or C viruses. There were 13 influenza diagnoses among outpatients with URTI, but 14 influenza viruses were detected because 1 child was coinfected (with influenza A and C viruses), and there were 99 influenza diagnoses among all inpatients with pneumonia, but 101 influenza viruses were detected because 2 children were coinfected (with influenza A and B viruses in one and influenza A and C viruses in the other). b % is number detected / number tested

a Includes influenza A, B, or C viruses. There were 13 influenza diagnoses among outpatients with URTI, but 14 influenza viruses were detected because 1 child was coinfected (with influenza A and C viruses), and there were 99 influenza diagnoses among all inpatients with pneumonia, but 101 influenza viruses were detected because 2 children were coinfected (with influenza A and B viruses in one and influenza A and C viruses in the other). b % is number detected / number tested c Because samples collected during 2007 were not tested for influenza C virus, percentages are based on the following denominators: NARI, 139; URTI, 235; severe pneumonia, 1126; and very severe pneumonia, 288.

a Includes influenza A, B, or C viruses. There were 13 influenza diagnoses among outpatients with URTI, but 14 influenza viruses were detected because 1 child was coinfected (with influenza A and C viruses), and there were 99 influenza diagnoses among all inpatients with pneumonia, but 101 influenza viruses were detected because 2 children were coinfected (with influenza A and B viruses in one and influenza A and C viruses in the other). b % is number detected / number tested c Because samples collected during 2007 were not tested for influenza C virus, percentages are based on the following denominators: NARI, 139; URTI, 235; severe pneumonia, 1126; and very severe pneumonia, 288. The estimated incidence of severe or very severe pneumonia among children <1 year of age, <5 years of age, and <13 years of age was 3902 (95% CI, 3708–4106), 1321 (95% CI, 1269–1375), and 589 (95% CI, 567–612) cases per 100 000 population per year, respectively. Correspondingly, for all inpatients with influenza, the incidence of severe or very severe pneumonia for those aged <1 year, <5 years, and <13 years was 154 (95% CI, 116–204), 60 (95% CI, 49–74), and 28 (95% CI, 23–34) cases per 100 000 per year, respectively, after scaling for the proportion of patients who were tested for pneumonia (ie, 0.82). For inpatients with influenza A virus infection, the incidence of severe or very severe pneumonia among children aged <1 year, <5 years, and <13 years, was 106 (95% CI, 75–149), 43 (95% CI, 34–55), and 20 (95% CI, 16–25) cases per 100 000 per year, respectively. Considering only inpatients <5 years of age who were admitted from locations within a 5-km radius of KDH, the incidence estimates (using a scaling factor, p, of 0.78) of severe or very severe pneumonia, influenza, and influenza A virus infection were 1815 (95% CI, 1700–1938), 106 (95% CI, 78–144), and 70 (95% CI, 48–102) cases per 100 000 per year, respectively. Comparative estimates of incidence by age group for other virus groups are presented in Supplementary Table 2. Rhinovirus and RSV infection had higher incidences than that for influenza virus by factors of 4–6 and 5–8, respectively, according to age group.

d 70 (95% CI, 48–102) cases per 100 000 per year, respectively. Comparative estimates of incidence by age group for other virus groups are presented in Supplementary Table 2. Rhinovirus and RSV infection had higher incidences than that for influenza virus by factors of 4–6 and 5–8, respectively, according to age group. Distribution of Influenza Virus Infection, by Season and by Age The seasonal patterns of influenza A and B virus infection are shown in Figure 1. Over the 4-year period, influenza A and B virus infection showed a significantly higher occurrence among inpatients during quarters 3 and 4, relative to the average for all quarters (22.5 cases expected vs 37.5 cases observed; P = .003). There was very little influenza activity in 2010. The majority of influenza B virus infections occurred in the fourth quarter of 2009. Most influenza cases occurred after the main period of rainfall (April–July) and before peak temperatures (first quarter; Figure 1). Cases of bacteremia, and specifically S. pneumoniae infection, by quarter, are shown in Figure 1. No statistically significant correlation between influenza cases (or influenza A virus infections) and occurrence of bacteremia (or S. pneumoniae infection) was identified, either concurrently or delayed (P > .05). Figure 1. Temporal distribution, by quarter, of molecular diagnoses of influenza A virus infection (dark bars) and influenza B virus infection (light bars) in nasopharyngeal samples collected from children with severe or very severe pneumonia who were admitted to Kilifi District Hospital, 2007–2010. Also, on the same axis are shown the quarterly number of admissions with bacteremia (x markers) and S. pneumoniae (o markers). The numbers of nasopharyngeal swab specimens collected each quarter are shown on the secondary Y axis (triangle markers). B, Monthly weather patterns averaged during 2007–2010.

ital, 2007–2010. Also, on the same axis are shown the quarterly number of admissions with bacteremia (x markers) and S. pneumoniae (o markers). The numbers of nasopharyngeal swab specimens collected each quarter are shown on the secondary Y axis (triangle markers). B, Monthly weather patterns averaged during 2007–2010. Of the 70 of 83 influenza A virus–positive samples sent to the NIC for subtyping, 42 (60%) were A(H3N2) (median age, 13 months [range, 0–134 months]), 10 (14%) were A(H1N1) (median age, 21 months [range, 4–127 months]), and 4 (6%) were A(H1N1)pdm09 (median age, 31 months [range, 17–43 months]), with 14 failing to subtype (7 were unconfirmed as influenza A virus). Among outpatients aged <13 years who presented with URTI, 7 had A(H3N2), 3 had A(H1N1), and 0 had A(H1N1)pdm09. The distribution of subtypes over time is shown in Figure 2. A(H3N2) circulated in all years, and A(H1N1) was confined to 2007; A(H1N1)pdm09 occurred in late 2009 and again in late 2010. Figure 2. Temporal distribution, by quarter, of influenza A virus subtypes for outpatients and inpatients combined from Kilifi District, coastal Kenya, 2007–2010.

over time is shown in Figure 2. A(H3N2) circulated in all years, and A(H1N1) was confined to 2007; A(H1N1)pdm09 occurred in late 2009 and again in late 2010. Figure 2. Temporal distribution, by quarter, of influenza A virus subtypes for outpatients and inpatients combined from Kilifi District, coastal Kenya, 2007–2010. The age distributions of diagnoses of influenza A, B, or C virus infection over the 4 years of surveillance are shown in Supplementary Figure 1 and follow a similar pattern (Supplementary Figure 1A), with around 50% of cases in infants (46% had influenza A virus infection, 47% had influenza B virus infection, and 66% had influenza C virus infection; P = .632). The proportion of inpatients with pneumonia who were found to be influenza positive showed a trend for increase with increasing age (Supplementary Figure 1B). In the case of influenza A virus infection, the trend was significant: 1.7% of children aged 0–2 months had influenza A virus infection, compared with >5% of children aged ≥24 months (P = .005).

pneumonia who were found to be influenza positive showed a trend for increase with increasing age (Supplementary Figure 1B). In the case of influenza A virus infection, the trend was significant: 1.7% of children aged 0–2 months had influenza A virus infection, compared with >5% of children aged ≥24 months (P = .005). Clinical Characteristics and Disease Severity Analysis of the association between a diagnosis of infection with any influenza virus and components of a set of severity features yielded an increased odds of hypoxia among children with influenza, compared with those without influenza (age-adjusted OR, 1.78; 95% CI, 1.04–1.96). No other severity feature was significantly associated with influenza (Supplementary Table 1). There were 4 deaths (age range, 7–32 months) among the 99 inpatients with influenza; one was infected with influenza A virus, 2 were infected with influenza B virus, and 1 was infected with influenza C virus. Three were positive for HIV antibody, severely malnourished (weight-for-age Z score, ≤4), and had a discharge diagnosis including immunosuppression, and 1 inpatient (who was infected with influenza A virus) had chronic heart disease.

us, 2 were infected with influenza B virus, and 1 was infected with influenza C virus. Three were positive for HIV antibody, severely malnourished (weight-for-age Z score, ≤4), and had a discharge diagnosis including immunosuppression, and 1 inpatient (who was infected with influenza A virus) had chronic heart disease. DISCUSSION Our study identified all 3 influenza viruses in circulation in this rural coastal Kenya location among patients hospitalized with severe or very severe pneumonia and among outpatients with URTI. However, the prevalence of infection with any influenza virus was low among inpatients with severe or very severe pneumonia (4.9%) and among outpatients with URTI (3.9%). Influenza A virus predominated, with identification in 3.5% of inpatients and 3.3% of outpatients with URTI. Correspondingly, relative to the incidence of severe or very severe pneumonia among hospitalized children aged <5 years (1323 cases per 100 000 per year), the incidences of influenza (64 cases per 100 000 per year) and influenza A virus infection (46 cases per 100 000 per year) were low. There was near absence of influenza in the convenience sample of children without signs of respiratory illness. Our inpatient data are consistent with the results of a recent review of data on seasonal influenza from 15 published studies in sub-Saharan Africa [1], which reported a median prevalence of 6.0% among hospitalized pediatric patients, with a range of 0%–16%. In the same review, the median prevalence of influenza among outpatients with ARI was higher than we found, at 10% (range, 1%–25%; 11 studies), but comparisons should be cautioned because of a number of methodological differences. Exploration of a range of severity features and concurrent illnesses revealed hypoxia to be more commonly associated with influenza among hospitalized children.

s with ARI was higher than we found, at 10% (range, 1%–25%; 11 studies), but comparisons should be cautioned because of a number of methodological differences. Exploration of a range of severity features and concurrent illnesses revealed hypoxia to be more commonly associated with influenza among hospitalized children. Calculation of the incidence of influenza-associated severe disease on the basis of hospital admission data is likely to underestimate the true burden in the community, as a result of the relationship between healthcare access and distance from the hospital. This is supported in the analysis, where it was shown that the incidence of influenza-associated admissions among children with severe or very severe pneumonia was about 70% greater in the population proximal to the hospital. We have previously shown a similar distance decay for severe rotavirus diarrhea [7] and severe RSV-associated pneumonia [4] and pneumonia and meningitis [8]. Furthermore, a previous study of RSV among infants and young children in the HDSS revealed that roughly 4 in 5 children identified with severe pneumonia in the outpatient setting were not admitted to the local hospital [9].

rotavirus diarrhea [7] and severe RSV-associated pneumonia [4] and pneumonia and meningitis [8]. Furthermore, a previous study of RSV among infants and young children in the HDSS revealed that roughly 4 in 5 children identified with severe pneumonia in the outpatient setting were not admitted to the local hospital [9]. Notwithstanding this underestimation, it is clear that the incidence of influenza-associated hospital admissions is significantly lower than that associated with either rhinovirus or RSV infection. While the etiology of rhinovirus as the causative agent of lower respiratory tract disease may be in question [10], this is not the case for RSV, which is known to be a major cause of infant and childhood lower bronchiolitis and pneumonia in sub-Saharan Africa and globally [11]. While RSV is invariably among the most prevalent viruses in children admitted with lower respiratory tract illness, it is not always dominant over influenza [1]. In Kenya, contemporary data on the relative prevalence of respiratory viruses among pneumonia-related admissions to the hospital are largely absent. Further data are clearly needed in Kenya to gauge the relative burden of disease due to respiratory viruses and thereby help support future health policy planning.

za [1]. In Kenya, contemporary data on the relative prevalence of respiratory viruses among pneumonia-related admissions to the hospital are largely absent. Further data are clearly needed in Kenya to gauge the relative burden of disease due to respiratory viruses and thereby help support future health policy planning. In terms of seasonality, there was increased occurrence in the third and fourth quarters of each year, most notably for influenza A virus infection, except in 2010. These periods are characteristically times of lower rainfall levels (referred to locally as “second rains”), intermediate temperatures, and relative lower humidity. During the study period, A(H1N1)pdm09 entered Kenya [12], and cases of A(H1N1)pdm09 infection were identified in KDH from late 2010. It is possible that the introduction of A(H1N1)pdm09 disrupted the normal pattern of A(H3N2) activity in 2010; only A(H1N1)pdm09 was observed in the latter quarters of 2010. Continued surveillance will reveal whether A(H1N1)pdm09 has any long-term effect on the circulation patterns of other influenza subtypes. However, in general, the contribution of A(H1N1)pdm09 to the burden of hospitalization-associated pneumonia was minimal in this setting. The possibility exists that the burden of influenza was underestimated because of associated, but delayed, invasive bacterial disease. However, we identified no evidence for an increased number of admissions in which bacteria (or S. pneumoniae, in particular) were detected in blood cultures during the quarter following peak occurrences of influenza.

burden of influenza was underestimated because of associated, but delayed, invasive bacterial disease. However, we identified no evidence for an increased number of admissions in which bacteria (or S. pneumoniae, in particular) were detected in blood cultures during the quarter following peak occurrences of influenza. During the A(H1N1)pdm09 infection pandemic, antiviral therapy (oseltamivir) was prescribed to children admitted to KDH with severe acute respiratory illness on a presumptive basis (ie, prior to laboratory confirmation of influenza.) This would not have altered the pattern of observation of influenza described in this study, because nasopharyngeal specimens were collected prior to treatment with the antiviral. Within the surrounding community, A(H1N1)pdm09 vaccination was undertaken in 2010 but was limited to target groups, including healthcare workers, pregnant women, and patients with chronic disease, totaling 2203 subjects (Kenya Ministry of Health, personal communication). This number and the age group of subjects receiving vaccine would not have altered the pattern of A(H1N1) infection occurrence described in this study. Half of the influenza cases occurred in infants, and the proportion of cases rapidly declined with age into older age groups. However, we noted that within the age group that we studied, the prevalence of influenza increased with age, suggesting that relative to other causes of cases of severe or very severe pneumonia associated with admission, those caused by influenza decline less rapidly with age.

y declined with age into older age groups. However, we noted that within the age group that we studied, the prevalence of influenza increased with age, suggesting that relative to other causes of cases of severe or very severe pneumonia associated with admission, those caused by influenza decline less rapidly with age. A limitation of the present study is the failure to recruit and test samples from approximately 20% of eligible children. The reasons for this have been previously described [2, 4] and were primarily due to parental refusal and, to a lesser extent, to discharge or death before sampling. Failure to recruit and test was more common for critically ill children and for those who died while in the hospital. This almost certainly led to an underestimation of severity associated with influenza. A further limitation of the study is that results are based only on nasopharyngeal samples, and we now have definitive evidence that an oropharyngeal swab specimen provides added diagnostic value for detecting influenza in our setting (22% increased detection; 95% CI, 9–42), compared with nasopharyngeal specimens alone [3]. The low prevalence of influenza viruses in this study limits the power of the analysis to identify associations between virus presence and specific clinical features or coinfections. Our study involved sampling of 2000 children; a definitive investigation of clinical associations will require a considerably larger sample size or a location with a markedly higher incidence of influenza A virus infection. We report very low prevalence of influenza in outpatient children without signs of acute respiratory infection, suggesting that influenza virus is rarely the cause of asymptomatic infection, and the data also suggest influenza is the cause of only 4% of URTI cases. Further interpretation of these data with regard to the association between influenza and severe disease is unwarranted since outpatient sampling was not contemporaneous throughout the period of surveillance of hospitalized patients, and collection was not frequency matched by age and location within the KHDSS. We therefore await the complete results of a larger and better-designed case-control study, which is ongoing.

se is unwarranted since outpatient sampling was not contemporaneous throughout the period of surveillance of hospitalized patients, and collection was not frequency matched by age and location within the KHDSS. We therefore await the complete results of a larger and better-designed case-control study, which is ongoing. In conclusion, although the incidence of influenza was underestimated in this study, it is clear that influenza contributes only a small proportion of the total burden of hospitalization-associated severe and very severe pneumonia among children in this rural coastal Kenya setting. Influenza A virus is the dominant influenza virus causing pediatric severe and very severe pneumonia. A seasonal signature for influenza was evident, but no temporal association was identified with invasive bacterial disease. Although A(H1N1)pdm09 infection was observed, its contribution to disease was not substantial. Hypoxia was more frequently identified among patients with influenza, and immunosuppression, severe malnutrition, or chronic heart disease were identified in all of the 4 influenza-associated deaths. Given the low influenza prevalence, larger studies are required to investigate associations between influenza and disease severity or prevalent conditions, such as malaria, HIV infection, or malnutrition. Additional comparative studies on viral diagnoses in severe pneumonia hospital admissions are warranted elsewhere in Kenya. Such data may be informative to the Kenya Ministry of Health in their assessment of the role for influenza antivirals and vaccination in Kenya.

ns, such as malaria, HIV infection, or malnutrition. Additional comparative studies on viral diagnoses in severe pneumonia hospital admissions are warranted elsewhere in Kenya. Such data may be informative to the Kenya Ministry of Health in their assessment of the role for influenza antivirals and vaccination in Kenya. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank the clinical and laboratory staff, for their hard work in collection and processing of the specimens; Mwanajuma Ngama, for the coordination of patient recruitment and sample collection at the clinic; Dr Marietjie Venter of NCID South Africa, for screening the 2006–2007 samples described here; and Janet Majanja and Meshack Wadegu of the National Influenza Centre-Nairobi, Ann Bett, Caroline Gitahi, Martin Mutunga, Alexander Gichuki, Clement Lewa, Getrude Ndanu, and James Kipkoech, who screened some of the samples for respiratory viruses. We also acknowledge the support of the Kilifi HDSS team. This article is published with permission from the director of the Kenya Medical Research Institute.

tt, Caroline Gitahi, Martin Mutunga, Alexander Gichuki, Clement Lewa, Getrude Ndanu, and James Kipkoech, who screened some of the samples for respiratory viruses. We also acknowledge the support of the Kilifi HDSS team. This article is published with permission from the director of the Kenya Medical Research Institute. Financial support. This work was supported by the Wellcome Trust (grants 084633, 081835, and 081186 to D. J. N., J. A. G. S., and J. A. B., respectively), as well as by Bill and Melinda Gates Foundation, through PERCH (Pneumonia Etiology Research for Child Health). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

The immune reconstitution inflammatory syndrome (IRIS) is a commonly encountered immunological complication of the combined therapies for human immunodeficiency virus type 1 (HIV-1) infection and tuberculosis [1, 2]. Two forms of tuberculosis-associated IRIS (tuberculosis-IRIS) exist. The condition is best recognized as a paradoxical immune-mediated worsening of tuberculosis symptoms and other symptoms occurring in patients prescribed combined antitubercular and antiretroviral therapy [3, 4]. Common clinical manifestations of tuberculosis-IRIS include recurrent tuberculosis symptoms, fever, and enlarging and suppuration of lymph nodes [4]. Understanding of the immunopathogenesis of tuberculosis-IRIS remains incomplete, although it has become the subject of increased investigation in recent years.

py [3, 4]. Common clinical manifestations of tuberculosis-IRIS include recurrent tuberculosis symptoms, fever, and enlarging and suppuration of lymph nodes [4]. Understanding of the immunopathogenesis of tuberculosis-IRIS remains incomplete, although it has become the subject of increased investigation in recent years. We previously investigated the involvement of multiple cytokines and demonstrated that paradoxical tuberculosis-IRIS is associated with hypercytokinemia [5, 6]. The interleukin 10 (IL-10) family comprises cytokines structurally related to IL-10, such as interleukin 19 (IL-19), interleukin 20 (IL-20), interleukin 22 (IL-22), interleukin 24 (IL-24; MDA-7), interleukin 26 (IL-26; AK155), interleukin 28 (IL-28), and interleukin 29 (IL-29) in humans, in addition to some encoded in the genomes of herpes and pox viruses [7]. IL-10–related cytokines are highly pleiotropic but are linked together through genetic similarity and intron-exon gene structure [8], although other than the activities of IL-10, little is known about the biological activities of the IL-10 family [7]. It has been suggested that these cytokines may participate in T-cell–mediated diseases by regulation of T-cell cytokine profiles [9]. Human IL-10 is the most well-characterized molecule in this group, and it has multiple biological effects, including immunoregulatory and antiinflammatory effects on different cell types. IL-10 is produced by various immune cells, which include activated monocytes, T cells (CD4+ and CD8+), macrophages, and dendritic cells [10]. IL-10 functions mainly to regulate T-helper 1 cytokines, major histocompatibility class II and B7 molecules, and expression of costimulator molecules on macrophages that are necessary for optimal T-cell activation [7, 8, 11]. By inhibiting expression of these molecules on antigen-presenting cells, IL-10 directly suppresses the activation of T cells and the production of T-cell–derived cytokines, such as interferon γ (IFN-γ) and interleukin 2 [11].

imulator molecules on macrophages that are necessary for optimal T-cell activation [7, 8, 11]. By inhibiting expression of these molecules on antigen-presenting cells, IL-10 directly suppresses the activation of T cells and the production of T-cell–derived cytokines, such as interferon γ (IFN-γ) and interleukin 2 [11]. Similar to IL-10, the related members of the IL-10 family are secreted as α helical proteins whose amino acid sequences are up to 30% identical to that of IL-10, with the encoding genes located on 2 clusters in the human genome [12, 13]. One major similarity among these cytokines is the sharing of signaling receptors and the use of conserved signaling cascades [8]. However, knowledge of the biology of many of these IL-10 homologs remains incomplete. IL-22 is a proinflammatory cytokine that plays an important role in innate pathogen defense [12,14]. Expression of IL-22 has been shown to be highest in activated memory CD4+ T cells and lowest in activated natural killer cells [8], with little or no expression observed in other immune cells [13]. The preferential production of IL-22 by T cells suggests that increased expression of this cytokine may occur in T-cell–mediated diseases [8]. IL-22 acts on nonimmune cells and has been linked with severe inflammation in chronic T-cell–mediated inflammatory diseases, such as psoriasis, Crohn disease, and rheumatoid arthritis [12, 13]. IL-22 also contributes to the human antimycobacterial immune response [15, 16].

ine may occur in T-cell–mediated diseases [8]. IL-22 acts on nonimmune cells and has been linked with severe inflammation in chronic T-cell–mediated inflammatory diseases, such as psoriasis, Crohn disease, and rheumatoid arthritis [12, 13]. IL-22 also contributes to the human antimycobacterial immune response [15, 16]. The resolution of inflammation is dependent on the host's ability to mount an immunoregulatory response to regulate the magnitude of inflammatory responses. IRIS results from excessive pathological immune responses occurring during immune recovery in HIV-infected patients who are commencing antiretroviral therapy (ART). In this study, we therefore hypothesized that IL-10 and its homologs may be deficient and that this immune dysregulation may be involved in the immunopathology associated with tuberculosis-IRIS.

l immune responses occurring during immune recovery in HIV-infected patients who are commencing antiretroviral therapy (ART). In this study, we therefore hypothesized that IL-10 and its homologs may be deficient and that this immune dysregulation may be involved in the immunopathology associated with tuberculosis-IRIS. METHODS Ethical approval for these studies was provided by the University of Cape Town Research Ethics committee (REC references 337/2004 and 173/2005). Participants were recruited at G. F. Jooste Hospital and at the Ubuntu Clinic, Site B Khayelitsha, in Cape Town, South Africa, between March 2005 and December 2007 to 2 prospective studies whose designs have been previously reported [17, 18]. Patients had a confirmed diagnosis of paradoxical tuberculosis-IRIS that was based on the International Network for the Study of HIV-associated IRIS consensus case definition [3], which has been independently validated [19–21]. Patients were >18 years of age, not pregnant, and ART naive. Known rifampicin-resistant tuberculosis was an exclusion criterion. At entry into these studies, a clinical diagnostic work-up was performed to exclude an alternative diagnosis for clinical deterioration.

n [3], which has been independently validated [19–21]. Patients were >18 years of age, not pregnant, and ART naive. Known rifampicin-resistant tuberculosis was an exclusion criterion. At entry into these studies, a clinical diagnostic work-up was performed to exclude an alternative diagnosis for clinical deterioration. Cross-sectional Analysis of Paradoxical Tuberculosis-IRIS and Non-IRIS Patients For the case-control analysis, 20 patients (of 32 available) with paradoxical tuberculosis-IRIS were selected randomly from participants who were enrolled at the time of diagnosis with paradoxical tuberculosis-IRIS. Cases were matched with 20 non-IRIS controls who were enrolled using similar criteria (CD4+ T-cell count, age, sex, and duration of anti-tuberculosis treatment) but did not develop tuberculosis-IRIS 2 weeks after starting combination ART (cART). The study design, patient selection criteria, and treatment regimens have been previously described [5]. Blood Processing and Cell Cultures Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll from 30 mL of blood collected in Na-heparin BD vacutainers and were plated at 2.5 × 106 cells/well. PBMCs were restimulated with heat-killed H37Rv Mycobacterium tuberculosis for 6 or 24 hours at a multiplicity of infection (MOI) of 1:1. After incubation, PBMCs were harvested and lysed for RNA extraction. Cell lysates from both the 6- and 24-hour time points were stored for RNA analysis. Serum was also collected and cryopreserved at −80°C until further analysis.

v Mycobacterium tuberculosis for 6 or 24 hours at a multiplicity of infection (MOI) of 1:1. After incubation, PBMCs were harvested and lysed for RNA extraction. Cell lysates from both the 6- and 24-hour time points were stored for RNA analysis. Serum was also collected and cryopreserved at −80°C until further analysis. Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR) Messenger RNA (mRNA) was extracted from PBMC lysates by using the RNeasy Mini Kit Spin Protocol (Qiagen, Valencia, CA) in accordance with the manufacturer's recommendations and was cryopreserved at −80°C until further analysis. Primers and probes were purchased from Applied Biosystems as inventoried TaqMan Gene Expression Assays (Applied Biosystems, Qiagen, Valencia, CA). The TaqMan RNA-Ct 1 Step kit protocol was used in these assays. A lower difference in cycle threshold (ΔCt) indicates a higher transcript abundance. Fold-induction was used to obtain a relative measure of gene induction by M. tuberculosis and was determined using the ΔΔCt method. Analysis of Cytokine Levels in Serum Samples The level of soluble IL-10 protein in serum samples was measured using customized commercial Milliplex XMAP kits, whereas the level of IL-22 was measured using the human IL-22 Immunoassay Quantikine ELISA kit from R&D Systems (Minneapolis, MN).

Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR) Messenger RNA (mRNA) was extracted from PBMC lysates by using the RNeasy Mini Kit Spin Protocol (Qiagen, Valencia, CA) in accordance with the manufacturer's recommendations and was cryopreserved at −80°C until further analysis. Primers and probes were purchased from Applied Biosystems as inventoried TaqMan Gene Expression Assays (Applied Biosystems, Qiagen, Valencia, CA). The TaqMan RNA-Ct 1 Step kit protocol was used in these assays. A lower difference in cycle threshold (ΔCt) indicates a higher transcript abundance. Fold-induction was used to obtain a relative measure of gene induction by M. tuberculosis and was determined using the ΔΔCt method. Analysis of Cytokine Levels in Serum Samples The level of soluble IL-10 protein in serum samples was measured using customized commercial Milliplex XMAP kits, whereas the level of IL-22 was measured using the human IL-22 Immunoassay Quantikine ELISA kit from R&D Systems (Minneapolis, MN). Flow Cytometric Analysis Frozen PBMCs from 4 patients with tuberculosis-IRIS were thawed, resuspended in Roswell Park Memorial Institute medium (Sigma, UK) with 10% heat-inactivated fetal calf serum (FCS; Sigma, UK), and stimulated with heat-killed M. tuberculosis H37Rv (hkH37Rv; MOI 1:1) for 4 hours at 37°C in 5% CO2. Brefeldin A (10 μg/mL; Sigma, UK) was added to each tube, vortexed, and incubated for a further 20 hours. After 24 hours, cells were washed twice with phosphate-buffered saline (PBS; Sigma, UK) and stained for surface CD3-FITC and CD14-APC (both from BD Biosciences) for 30 minutes in the dark at 4°C. PBMCs were washed in fluorescence-activated cell sorter (FACS) wash buffer (PBS with 0.5% FCS); BD Cytofix/Cytoperm was then added, and PBMCs were incubated for 20 minutes in the dark at 4°C. PBMCs were washed with 1X BD Perm Wash and stained for intracellular IL-10-PE (BD Biosciences) and IL-22-PerCP-eFluor 710 (eBiosciences) for 1 hour at 4°C in the dark. Cells were washed 3 times with 1X BD Perm Wash buffer and acquired on a BD FACS Calibur. Analysis was performed using FlowJo, v9.4.3 (available at: http://www.treestar.com).

h 1X BD Perm Wash and stained for intracellular IL-10-PE (BD Biosciences) and IL-22-PerCP-eFluor 710 (eBiosciences) for 1 hour at 4°C in the dark. Cells were washed 3 times with 1X BD Perm Wash buffer and acquired on a BD FACS Calibur. Analysis was performed using FlowJo, v9.4.3 (available at: http://www.treestar.com). RESULTS Baseline Characteristics of Participants A total of 20 patients with paradoxical tuberculosis-IRIS and 20 non-IRIS controls were analyzed. Supplementary Table 1 shows a summary of the baseline demographic and clinical characteristics for the patients analyzed in this study. The 2 groups of patients were well matched with respect to sex, age, and baseline CD4+ T-cell count. There were no significant differences between the tuberculosis-IRIS and non-IRIS groups in terms of previous tuberculosis, tuberculosis form, and median days from cART initiation to IRIS onset (or to sample collection, in the case of non-IRIS controls). However, tuberculosis-IRIS patients were more likely to have a shorter duration between tuberculosis treatment and commencement of cART (P = .028).

groups in terms of previous tuberculosis, tuberculosis form, and median days from cART initiation to IRIS onset (or to sample collection, in the case of non-IRIS controls). However, tuberculosis-IRIS patients were more likely to have a shorter duration between tuberculosis treatment and commencement of cART (P = .028). Transcript Abundance of IL-10–Related Genes PBMCs from 20 tuberculosis-IRIS patients and 20 non-IRIS controls were cultured in the presence or absence of heat-killed H37Rv M. tuberculosis antigen for 6 and 24 hours. Evaluation at 6 hours revealed that M. tuberculosis stimulation had increased the transcript abundance of several of the cytokines in both the tuberculosis-IRIS and non-IRIS groups (Table 1). A lower ΔCt indicates a higher transcript abundance. Significantly higher transcript levels were observed for IL-22 in tuberculosis-IRIS patients after stimulation (P = .009), whereas levels of IL-24 transcripts were higher for non-IRIS patients (P = .020). IL-10 levels were significantly higher in unstimulated non-IRIS cultures (P = .04) and increased more in tuberculosis-IRIS cultures, compared with non-IRIS cultures, after stimulation. IL-26 transcript levels were significantly higher in both stimulated and unstimulated cultures of tuberculosis-IRIS PBMCs (P = .008 and P = .042, respectively). IL-28 transcript levels did not differ between unstimulated tuberculosis-IRIS and non-IRIS cultures but increased significantly in non-IRIS cultures after stimulation (P = .013). Table 1. Cycle Threshold Differences (ΔCt) for Interleukin 10 (IL-10)–Related Cytokine Genes After 6 Hours of In Vitro Stimulation With Heat-Killed Mycobacterium tuberculosis

iffer between unstimulated tuberculosis-IRIS and non-IRIS cultures but increased significantly in non-IRIS cultures after stimulation (P = .013). Table 1. Cycle Threshold Differences (ΔCt) for Interleukin 10 (IL-10)–Related Cytokine Genes After 6 Hours of In Vitro Stimulation With Heat-Killed Mycobacterium tuberculosis Cytokine P Unstimulated PBMCs, by Study Group, Median (IQR) Stimulated PBMCs, by Study Group, Median (IQR) Tuberculosis-IRIS vs Non-IRIS Unstimulated vs Stimulated Tuberculosis-IRIS Non-IRIS Tuberculosis-IRIS Non-IRIS Unstimulated Stimulated Tuberculosis-IRIS Non-IRIS IL-10 7.2 (6.4–8.2) 6.6 (5.8–7.6) 5.9 (3.4–7.1) 6.1 (5.0–7.5) .04 .25 .003 .27 IL-19 13.4 (5.5–17.8) 13.6 (10.8–19.2) 8.8 (6.5–11.6) 8.4 (5.1–12.1) .935 .804 <.0001 <.0001 IL-20 18.9 (11.2–21.3) 18.4 (9.6–21.2) 14.3 (10.6–19.8) 13.2 (8.5–18.8) .236 .103 <.0001 <.0001 IL-22 19.5 (15.6–21.6) 18.6 (13.8–21.3) 13.0 (8.3–20.5) 17.1 (10.1–21.4) .073 .009 <.0001 .086 IL-24 14.1 (7.6–20.8) 13.3 (5.8–18.3) 11.8 (7.7–15.7) 9.5 (5.9–13.7) .625 .020 .001 <.0001 IL-26 16.9 (14.4–20.0) 18.6 (17.1–21.0) 16.4 (11.3–19.4) 17.5 (13.0–20.7) .008 .042 .002 .010 IL-28 11.6 (8.6–16.3) 11.7 (9.7–13.0) 12.5 (10.1–14.3) 11.4 (8.8–12.8) .882 .013 .351 .130 Data are for patients with HIV infection and tuberculosis who had IRIS at clinical presentation (tuberculosis-IRIS) and similar patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS). At 6 hours, levels of IL-22 transcripts in stimulated PBMC cultures were significantly higher in tuberculosis-IRIS patients, whereas levels of IL-24 transcripts stimulated cultures were higher in non-IRIS patients. IL-26 had significantly higher transcript levels in both stimulated and unstimulated PBMC cultures for tuberculosis-IRIS patients. Levels of IL-28 transcripts were marginally higher in the stimulated PBMC cultures for non-IRIS patients (P = .013). IL-29 transcripts were barely detectable (results not shown). A lower ΔCt indicates a higher transcript abundance.

levels in both stimulated and unstimulated PBMC cultures for tuberculosis-IRIS patients. Levels of IL-28 transcripts were marginally higher in the stimulated PBMC cultures for non-IRIS patients (P = .013). IL-29 transcripts were barely detectable (results not shown). A lower ΔCt indicates a higher transcript abundance. Abbreviations: HIV, human immunodeficiency virus; IL-19, interleukin 19; IL-20, interleukin 20; IL-22, interleukin 22; IL-24, interleukin 24; IL-26, interleukin 26; IL-28, interleukin 28; IQR, interquartile range; IRIS, immune reconstitution inflammatory syndrome; PBMC, peripheral blood mononuclear cell. At 24 hours, several of the cytokines (IL-10, IL-20, and IL-26) had significantly higher transcript levels in the stimulated tuberculosis-IRIS cultures (P < .001, P = .022, and P = .004, respectively). IL-22 transcript levels were higher in both stimulated and unstimulated tuberculosis-IRIS cultures (P = .004 and P = .015, respectively). IL-28 transcript levels were significantly higher in unstimulated non-IRIS PBMCs (P = .012) but not in stimulated cultures. In PBMCs from both tuberculosis-IRIS and non-IRIS patients, M. tuberculosis stimulation resulted in increased transcript abundance for many of the cytokines, most prominently IL-19, IL-20, IL-24, and IL-26 (Table 2). Table 2. Cycle Threshold Differences (ΔCt) for Interleukin 10 (IL-10)–Related Cytokine Genes After 24 Hours of In Vitro Stimulation With Heat-Killed Mycobacterium tuberculosis

sis stimulation resulted in increased transcript abundance for many of the cytokines, most prominently IL-19, IL-20, IL-24, and IL-26 (Table 2). Table 2. Cycle Threshold Differences (ΔCt) for Interleukin 10 (IL-10)–Related Cytokine Genes After 24 Hours of In Vitro Stimulation With Heat-Killed Mycobacterium tuberculosis Cytokine P Unstimulated PBMCs, by Study Group, Median (IQR) Stimulated PBMCs, by Study Group, Median (IQR) Tuberculosis-IRIS vs Non-IRIS Unstimulated vs Stimulated Tuberculosis-IRIS Non-IRIS Tuberculosis-IRIS Non-IRIS Unstimulated Stimulated Tuberculosis-IRIS Non-IRIS IL-10 6.9 (5.1–8.1) 6.9 (6.0–7.1) 4.9 (3.1–6.4) 6.8 (5.9–8.5) .74 <.001 <.001 1.0 IL-19 11.7 (2.3–16.9) 10.9 (4.3–18.8) 6.9 (2.9–10.1) 5.9 (2.0–18.2) .472 .433 <.0001 .0002 IL-20 14.0 (7.9–19.1) 16.2 (4.3–20.8) 9.6 (5.9–15.9) 12.5 (6.4–19.1) .386 .028 .0003 <.0001 IL-22 18.3 (12.4–19.9) 15.5 (12.5–19.7) 12.0 (6.8–14.9) 14.7 (8.0–19.2) .004 .015 <.0001 .199 IL-24 13.2 (6.4–18.3) 13.1 (3.1–14.0) 10.4 (5.5–15.1) 9.9 (12.9–15.1) .944 .661 .007 .020 IL-26 13.1 (10.2–16.5) 13.3 (11.4–18.2) 8.4 (4.4–14.5) 11.1 (7.2–16.2) .273 .004 <.0001 .019 IL-28 10.7 (8.6–14.2) 9.7 (7.5–11.8) 10.2 (6.4–12.3) 10.2 (8.3–11.9) .012 .609 .034 .025 Data are for patients with HIV infection and tuberculosis who had IRIS at clinical presentation (tuberculosis-IRIS) and similar patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS). At 24 hours, IL-10, IL-20, and IL-26 transcript levels were significantly higher in the stimulated cultures for tuberculosis-IRIS patients but not in unstimulated cultures. IL-22 transcript levels were higher in unstimulated non-IRIS cultures but increased significantly after M. tuberculosis stimulation in tuberculosis-IRIS patients only (P = .015). IL-28 transcript levels were significantly higher in non-IRIS patients at 24 hours in unstimulated cultures but not in stimulated cultures. A lower ΔCt indicates a higher transcript abundance.

non-IRIS cultures but increased significantly after M. tuberculosis stimulation in tuberculosis-IRIS patients only (P = .015). IL-28 transcript levels were significantly higher in non-IRIS patients at 24 hours in unstimulated cultures but not in stimulated cultures. A lower ΔCt indicates a higher transcript abundance. Abbreviations: HIV, human immunodeficiency virus; IL-19, interleukin 19; IL-20, interleukin 20; IL-22, interleukin 22; IL-24, interleukin 24; IL-26, interleukin 26; IL-28, interleukin 28; IQR, interquartile range; IRIS, immune reconstitution inflammatory syndrome; PBMC, peripheral blood mononuclear cell.

non-IRIS cultures but increased significantly after M. tuberculosis stimulation in tuberculosis-IRIS patients only (P = .015). IL-28 transcript levels were significantly higher in non-IRIS patients at 24 hours in unstimulated cultures but not in stimulated cultures. A lower ΔCt indicates a higher transcript abundance. Abbreviations: HIV, human immunodeficiency virus; IL-19, interleukin 19; IL-20, interleukin 20; IL-22, interleukin 22; IL-24, interleukin 24; IL-26, interleukin 26; IL-28, interleukin 28; IQR, interquartile range; IRIS, immune reconstitution inflammatory syndrome; PBMC, peripheral blood mononuclear cell. Fold-Induction Analysis of IL-10–Related Genes IL-10 mRNA was highly and significantly induced in tuberculosis-IRIS patients at 6 hours and more so at 24 hours (P = .04 and P = .0002, respectively; Figure 1). Similarly, IL-22 was found to be significantly upregulated at both 6 and 24 hours in cultures from tuberculosis-IRIS patients, compared with non-IRIS control patients (P = .004 and P = .0015, respectively). Although both IL-19 and IL-20 were highly induced by M. tuberculosis stimulation, no significant differences were observed when comparing mRNA from tuberculosis-IRIS patients and mRNA from non-IRIS patients in cultures from the 6- and 24-hour time points. Although IL-24 tended to be more induced in cultures for the non-IRIS group at both time points, the difference between the tuberculosis-IRIS and non-IRIS groups was not significant. Similarly, there were no significant differences observed between tuberculosis-IRIS and non-IRIS patients at either time point for IL-26, although there tended to be more induction of IL-26 in the tuberculosis-IRIS patients at 24 hours. IL-28 and IL-29 were seldom expressed at high levels in these samples, and no significant differences were observed between induction/repression in the 2 patient groups. Although IL-29 was marginally detectable at 6 hours, it was not detectable at 24 hours (Figure 1). Figure 1. Fold-induction analysis of induction of interleukin 10 (IL-10)–related cytokines by heat-killed Mycobacterium tuberculosis. Peripheral blood mononuclear cells (PBMCs) from 20 patients with human immunodeficiency virus (HIV) infection and tuberculosis who had immune reconstitution inflammatory syndrome (IRIS) at clinical presentation (tuberculosis-IRIS) and 20 control patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS) were cultured in the presence or absence of M. tuberculosis for 6 and 24 hours and then lysed for messenger RNA transcript analysis by quantitative reverse transcription polymerase chain reaction. Fold-induction of the genes was determined using the ΔΔ cycle threshold method.

id not develop tuberculosis-IRIS (non-IRIS) were cultured in the presence or absence of M. tuberculosis for 6 and 24 hours and then lysed for messenger RNA transcript analysis by quantitative reverse transcription polymerase chain reaction. Fold-induction of the genes was determined using the ΔΔ cycle threshold method. This method compares the difference in the cycle threshold between the gene of interest to that of a normalization gene (β-actin). IL-10 and interleukin 22 (IL-22) were differentially induced in tuberculosis-IRIS patients at both 6 and 24 hours. For interleukin 19 (IL-19), interleukin 20 (IL-20), interleukin 24 (IL-24), interleukin 26 (IL-26), interleukin 28 (IL-28), and interleukin 29 (IL-29), no significant differences were noted between tuberculosis-IRIS and non-IRIS patients, irrespective of restimulation of PBMCs with M. tuberculosis. Tuberculosis-IRIS patients are represented by the black boxes, whereas the open white boxes represent non-IRIS patients.

interleukin 28 (IL-28), and interleukin 29 (IL-29), no significant differences were noted between tuberculosis-IRIS and non-IRIS patients, irrespective of restimulation of PBMCs with M. tuberculosis. Tuberculosis-IRIS patients are represented by the black boxes, whereas the open white boxes represent non-IRIS patients. Analysis of Serum Cytokine Levels To further investigate the RNA results, protein concentrations were determined in serum samples for genes that showed consistent and significant differences in the mRNA analysis (ie, IL-10 and IL-22). Significantly higher concentrations of IL-10 (P = .0004; Figure 2) were detected in the serum of tuberculosis-IRIS patients (median, 875.6 pg/mL [interquartile range {IQR}, 64.5–9982 pg/mL]), compared with non-IRIS patients (median, 137.2 pg/mL [IQR, 24.3–1346 pg/mL]). Similarly, significantly higher levels of IL-22 (P = .007) were detected in the serum samples of tuberculosis-IRIS patients (median, 53.2 pg/mL [IQR, 11.31–188.8 pg/mL]), compared with non-IRIS patients (median, 24.4 pg/mL [IQR, 15.7–43.8 pg/mL]). Figure 2. Analysis of interleukin 10 (IL-10) and interleukin 22 (IL-22) concentrations in serum. Consistent with the fold-induction analysis, IL-10 and IL-22 protein levels were differentially higher in the serum of patients with human immunodeficiency virus (HIV) infection and tuberculosis who had immune reconstitution inflammatory syndrome (IRIS) at clinical presentation (tuberculosis-IRIS). Tuberculosis-IRIS patients are shown by the white circles, whereas the open black circles represent the control patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS).

rculosis who had immune reconstitution inflammatory syndrome (IRIS) at clinical presentation (tuberculosis-IRIS). Tuberculosis-IRIS patients are shown by the white circles, whereas the open black circles represent the control patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS). Correlation of IL-10 and IL-22 Serum Concentrations To determine whether the elevated concentrations of IL-10 and IL-22 observed in the serum samples of tuberculosis-IRIS patients correlated, Spearman correlation was calculated. There was an inverse correlation between the serum concentrations of IL-10 and IL-22 (Spearman r = −0.69; P = .007) in the tuberculosis-IRIS group. However, no significant correlation was observed in the non-IRIS group (Spearman r = −0.07; P = .83).

culosis-IRIS patients correlated, Spearman correlation was calculated. There was an inverse correlation between the serum concentrations of IL-10 and IL-22 (Spearman r = −0.69; P = .007) in the tuberculosis-IRIS group. However, no significant correlation was observed in the non-IRIS group (Spearman r = −0.07; P = .83). Cellular Origin of IL-10 and IL-22 To investigate the cellular origin of IL-10 and IL-22 in tuberculosis-IRIS patients, frozen PBMCs from 4 patients who presented with IRIS were stimulated with heat-killed M. tuberculosis H37Rv (MOI 1:1) for 4 hours, followed by addition of brefeldin A for a further 20 hours. At this time, cells were stained for surface CD3 and CD14. Cells were then stained for intracellular IL-10 and IL-22. We found that CD14+ cells were more likely to be positive for IL-22 (median, 0.6% [IQR, 0.02%–2.1%]), compared with CD3+ lymphocytes (median, 0.02% [IQR, 0%–0.04%]). The difference was not statistically significant, owing to the small number of samples. Similarly, CD14+ cells were also more likely to be positive for IL-10 (median, 0.06% [IQR, 0%–0.15%), compared with CD3+ cells (median, 0% [IQR, 0%–0.01%]).

02%–2.1%]), compared with CD3+ lymphocytes (median, 0.02% [IQR, 0%–0.04%]). The difference was not statistically significant, owing to the small number of samples. Similarly, CD14+ cells were also more likely to be positive for IL-10 (median, 0.06% [IQR, 0%–0.15%), compared with CD3+ cells (median, 0% [IQR, 0%–0.01%]). DISCUSSION Our analysis of IL-10–related cytokines showed an increase in the transcript levels of IL-10 and IL-22 in tuberculosis-IRIS patients, compared with non-IRIS controls. The corresponding serum samples showed significantly higher concentrations of IL-10 and IL-22 protein in tuberculosis-IRIS patients. Thus, our hypothesis that IRIS is associated with depressed IL-10 responses was falsified, as the opposite appears to have been the case. Whereas analysis of fold-induction showed significant induction of IL-19, IL-20, IL-24, and IL-26 by M. tuberculosis, the differences detected between the tuberculosis-IRIS and non-IRIS groups were neither great nor statistically significant.

IL-10 responses was falsified, as the opposite appears to have been the case. Whereas analysis of fold-induction showed significant induction of IL-19, IL-20, IL-24, and IL-26 by M. tuberculosis, the differences detected between the tuberculosis-IRIS and non-IRIS groups were neither great nor statistically significant. Whereas the regulatory role of IL-10 in vitro and in animal models is well documented, the role of this cytokine in clinical situations, particularly in infectious diseases, remains the subject of investigation [9]. Production of IL-10 can also be stimulated by bacteria, viruses, and parasites and is regulated both at the transcriptional and translation levels. The absence of IL-10 in knockout mice results in reduced M. tuberculosis loads in the lung. However, this reduction was preceded by an accelerated and enhanced IFN-γ response in the lung, an increased influx of CD4+ T cells into the lung, and enhanced production of chemokines and cytokines, including CXCL10 and IL-17, in both the lung and the serum [22]. This suggests that, although the early production of IL-10 in response to tuberculosis may be detrimental, this cytokine is ultimately necessary to prevent immunopathology.

f CD4+ T cells into the lung, and enhanced production of chemokines and cytokines, including CXCL10 and IL-17, in both the lung and the serum [22]. This suggests that, although the early production of IL-10 in response to tuberculosis may be detrimental, this cytokine is ultimately necessary to prevent immunopathology. IL-10 has been reported to modulate the innate and adaptive immune responses, potentially creating a favorable environment for the persistence of microbes, intracellular pathogens, and chronic infections [9]. The increased ability of macrophages to produce IL-10 when stimulated with Toll-like receptor ligands is also associated with an increased tendency to develop primary progressive tuberculosis [22]. Production of IL-10 has also been reported to be higher in patients who had active tuberculosis, compared with tuberculin skin test responders [23]. In this study, we show that consistently higher levels of IL-10 mRNA and serum protein were observed in tuberculosis-IRIS patients. Tuberculosis-IRIS is a highly inflammatory condition. We hypothesize that the high levels of IL-10 observed in the peripheral blood of tuberculosis-IRIS patients in this study reflect an overspill of IL-10 from the sites of inflammation where IL-10 would be involved in regulating and resolving inflammation.

S patients. Tuberculosis-IRIS is a highly inflammatory condition. We hypothesize that the high levels of IL-10 observed in the peripheral blood of tuberculosis-IRIS patients in this study reflect an overspill of IL-10 from the sites of inflammation where IL-10 would be involved in regulating and resolving inflammation. Other human studies have shown that IL-10 may be associated with susceptibility to infections caused by rapidly growing mycobacteria [24]. Significantly higher levels of circulating IL-10 have been demonstrated in HIV-infected patients than in healthy controls [25]. IL-10 production has previously been shown to increase in Mycobacterium avium–stimulated monocytes from HIV-infected patients, with the highest expression observed in patients with advanced AIDS [24].

antly higher levels of circulating IL-10 have been demonstrated in HIV-infected patients than in healthy controls [25]. IL-10 production has previously been shown to increase in Mycobacterium avium–stimulated monocytes from HIV-infected patients, with the highest expression observed in patients with advanced AIDS [24]. It is interesting that levels of IL-22 gene expression and serum protein were, like those of IL-10, found to be significantly higher in tuberculosis-IRIS patients. IL-22 is implicated in T-lymphocyte disease, innate pathogen defense, and acute phase responses [9] and has been associated with increased innate immunity in tissues [13], with expression being higher in outer body barriers, such as skin, the respiratory system, and the digestive system. Previous work on IL-22 and IL-17 in tuberculosis immunity has demonstrated elevated levels of IL-22 in the bronchoalveolar lavage fluid of pulmonary tuberculosis patients [15]. We show that IL-22 may be implicated not only in tuberculosis pathology, but also in tuberculosis-IRIS immunopathology, as indicated by the higher concentrations of IL-22 we observed in tuberculosis-IRIS patients. The coexpression of IL-10 and IL-22 seen in our study and their inverse correlation in tuberculosis-IRIS patients supports the suggestion that there may be interaction between these cytokines in this condition (and not in controls), as has been postulated for other diseases [26]. We also performed preliminary analysis of the cellular source of IL-10 and IL-22 and found both more frequently in CD14+ cells (Table 3), consistent with increasing evidence that innate immune system activation is a component of the immunopathology of tuberculosis-IRIS [5, 27–30]. Table 3. Cellular Sources of Interleukin 10 (IL-10) and Interleukin 22 (IL-22)

f the cellular source of IL-10 and IL-22 and found both more frequently in CD14+ cells (Table 3), consistent with increasing evidence that innate immune system activation is a component of the immunopathology of tuberculosis-IRIS [5, 27–30]. Table 3. Cellular Sources of Interleukin 10 (IL-10) and Interleukin 22 (IL-22) Cells (n = 4) IL-10 IL-22 Unstimulated hkH37Rv (1:1) Unstimulated hkH37Rv (1:1) CD14+ 0 (0–0.6) 0.06 (0–0.15) 0 (0–0.01) 0.6 (0.02–2.1) CD3+ 0 (0–0.01) 0 (0–0.01) 0.01 (0–0.03) 0.02 (0–0.04) Data are median % (interquartile range) of positive cells. Abbreviation: hkH37Rv, heat-killed Mycobacterium tuberculosis H37Rv. IL-19, like IL-20, has been shown to be produced under inflammatory conditions and is thought to play an important role in the pathogenesis of some inflammatory diseases, such as psoriasis. It has been suggested that these cytokines may be important in autoimmune diseases and may enhance antibacterial and antiviral immunity [8]. However, from our study, we did not show a link between IL-19 or IL-20 with tuberculosis-IRIS immunopathology. A major strength of our study was close matching of patients with paradoxical tuberculosis-IRIS and controls who did not develop the condition. Apart from the longer duration between initiating tuberculosis treatment and ART in the non-IRIS group, the groups were well matched. We acknowledge some limitations to our study, an important one of which is the insufficiency of 24-hour tissue culture supernatants from these patients. This may have limited our ability to validate and confirm the mRNA transcript findings.

rculosis treatment and ART in the non-IRIS group, the groups were well matched. We acknowledge some limitations to our study, an important one of which is the insufficiency of 24-hour tissue culture supernatants from these patients. This may have limited our ability to validate and confirm the mRNA transcript findings. In summary, our findings show elevated levels of IL-10 and IL-22 in patients with tuberculosis-IRIS. The higher levels of IL-10 may represent compensatory antiinflammatory and immune-regulatory responses, whereas elevated IL-22 levels suggest an association with immunopathology in tuberculosis-IRIS. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. Priscilla Mouton and Musaed Abrahams are especially thanked for assistance with recruitment and clinical assessment of study participants. This study was undertaken in healthcare facilities of the Provincial Government of the Western Cape (PGWC), and we thank PGWC staff for their support and assistance.

cknowledgments. Priscilla Mouton and Musaed Abrahams are especially thanked for assistance with recruitment and clinical assessment of study participants. This study was undertaken in healthcare facilities of the Provincial Government of the Western Cape (PGWC), and we thank PGWC staff for their support and assistance. R. T., G. A. M., K. A. W., M. X. R., and R. J. W. conceived and designed the study; R. T., K. A. W., K. M., K. S., R. S., and R. J. W. performed experimental work; R. T., K. A. W., G. A. M., and R. J. W. performed analysis and interpretation; and R. T., G. M., K. A. W., and R. J. W. drafted the manuscript for important intellectual content. Financial support. This work was supported by the Wellcome Trust (references 084323, 081667, and 088316), with additional support from the Medical Research Councils of the UK (U1175.02.002.00014.01) and South Africa; by the European Union (Sante/2006/105-061); and by the European and Developing Countries Clinical Trials Partnership (EDCTP 060613). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Chronic infection with Plasmodium and helminths causes an enormous public health burden in the tropics [1–5]. Malaria in pregnancy has been associated with increased risk of maternal anemia, low birth weight, stillbirth, and maternal death [6, 7]. Recent evidence suggests that prenatal exposure to Plasmodium falciparum may increase malaria risk in early childhood [8, 9]; however, little is known about the long-term consequences for young children. In the 1990s, estimates suggested that 44 million of the world's pregnant women harbored hookworm [10], and it was suggested that helminth infections might be particularly detrimental to the mother during pregnancy [11, 12]. However, in the Entebbe Mother and Baby Study (EMaBS) we found unexpectedly little association between maternal helminths and maternal anemia and none of the expected benefits of anthelminthic treatment in pregnancy for birth outcomes [13, 14]. Still, there is a dearth of literature on the consequences of helminth infections in pregnancy for the child, and effects of prenatal exposure to malaria–helminth coinfections on childhood malaria have not been addressed.

none of the expected benefits of anthelminthic treatment in pregnancy for birth outcomes [13, 14]. Still, there is a dearth of literature on the consequences of helminth infections in pregnancy for the child, and effects of prenatal exposure to malaria–helminth coinfections on childhood malaria have not been addressed. Studies including the EMaBS have reported associations between malaria and helminth infections in pregnant women [15, 16], and others have reported associations in children [17, 18]. Hookworm infection has been related to increased susceptibility to malaria infection in pregnancy [2, 16, 19] and in children [2, 20]. However, results have been inconsistent [15, 21]. Immunoregulatory mechanisms have been proposed. Helminth infections induce T-cell hyporesponsiveness, downmodulating immunity to their own as well as other antigens [22–25], and Plasmodia may possibly modulate responses to helminth coinfections [26]. Prenatal exposure to pathogen antigens might enhance fetal tolerance or sensitization to the antigens, leading to a failure to mount a response to the infection [9] or the development of infection resistance [27], respectively. Whether such immunological effects have a measurable impact on the incidence of malaria is uncertain [28, 29].

to pathogen antigens might enhance fetal tolerance or sensitization to the antigens, leading to a failure to mount a response to the infection [9] or the development of infection resistance [27], respectively. Whether such immunological effects have a measurable impact on the incidence of malaria is uncertain [28, 29]. In the EMaBS, treatment with anthelminthics in pregnancy had no effect on childhood malaria incidence. However, we speculated that the single-dose treatment might be insufficient to change an effect of helminths established earlier during pregnancy. This study used EMaBS data (ISRCTN32849447) [30] to examine associations between helminth and malaria infections in pregnancy and malaria in the offspring. METHODS The study was approved by the Science and Ethics Committee of the Uganda Virus Research Institute, the Uganda National Council for Science and Technology and the ethics committee of the London School of Hygiene and Tropical Medicine. Project staff explained the study to the participants in the local language and provided participants with a study information letter to take home. Written informed consent was obtained from the mother during pregnancy and from the mother or caregiver when the child was aged 1 year. Study Population The study area is on the northern shores of Lake Victoria in Uganda, a high malaria transmission area. P. falciparum is the dominant Plasmodium species; Anopheles gambiae and Anopheles fenestus are the dominant vectors [31]. The area consists of urban, rural, and fishing communities.

METHODS The study was approved by the Science and Ethics Committee of the Uganda Virus Research Institute, the Uganda National Council for Science and Technology and the ethics committee of the London School of Hygiene and Tropical Medicine. Project staff explained the study to the participants in the local language and provided participants with a study information letter to take home. Written informed consent was obtained from the mother during pregnancy and from the mother or caregiver when the child was aged 1 year. Study Population The study area is on the northern shores of Lake Victoria in Uganda, a high malaria transmission area. P. falciparum is the dominant Plasmodium species; Anopheles gambiae and Anopheles fenestus are the dominant vectors [31]. The area consists of urban, rural, and fishing communities. Study Design The EMaBS enrolled 2507 pregnant women in their second or third trimester between April 2003 and November 2005; 2345 live births were recorded. Inclusion and exclusion criteria are described elsewhere [30]. We present an observational analysis of the trial cohort.

Study Population The study area is on the northern shores of Lake Victoria in Uganda, a high malaria transmission area. P. falciparum is the dominant Plasmodium species; Anopheles gambiae and Anopheles fenestus are the dominant vectors [31]. The area consists of urban, rural, and fishing communities. Study Design The EMaBS enrolled 2507 pregnant women in their second or third trimester between April 2003 and November 2005; 2345 live births were recorded. Inclusion and exclusion criteria are described elsewhere [30]. We present an observational analysis of the trial cohort. At enrolment, the median gestational age was 27 weeks (interquartile range, 22–31). Before receiving the trial intervention, women gave a blood and stool sample for assessment of parasite infections. Women were then randomized to single-dose albendazole (400 mg) or placebo, and praziquantel (40 mg/kg) or placebo; their offspring were randomized to receive quarterly single-dose albendazole (200 mg from age 15 to 21 months and 400 mg from age 24 to 60 months) or placebo. Ferrous sulphate was provided monthly, and intermittent presumptive treatment for malaria (sulphadoxine-pyrimethamine) twice during the pregnancy. All women received anthelminthic treatment 6 weeks after delivery. After delivery, mothers were invited to bring the children to the research clinic for routine immunizations and any illness and for quarterly study visits to age 5 years. Community workers visited the children fortnightly and referred sick children to the clinic. Clinical malaria episodes were recorded prospectively. At annual scheduled visits, the children were examined for P. falciparum and helminth infections and were treated according to clinical guidelines if infections were found.

Community workers visited the children fortnightly and referred sick children to the clinic. Clinical malaria episodes were recorded prospectively. At annual scheduled visits, the children were examined for P. falciparum and helminth infections and were treated according to clinical guidelines if infections were found. Participants' addresses at enrollment were geo-referenced using handheld GPS receivers, and geographical zones were established based on features such as coastline, forest, location of settlements, and altitude [16]. Diagnosis of Infections Women provided blood and stool samples at enrollment and after delivery; children provided samples at illness and routine annual visits. Blood samples were examined for Mansonella perstans using the modified Knott's method [32]. Thick blood films were stained with Leishman's stain, malaria parasites were counted against 200 leucocytes, and at least 100 high-power fields were examined before a film was declared negative. Duplicate Kato–Katz slides were prepared and examined within 30 minutes for hookworm and the following day for other helminth eggs [33, 34]. Human immunodeficiency virus (HIV) serology was performed for mothers and children aged ≥18 months using a rapid test algorithm [13]; for infants, RNA and DNA polymerase chain reaction methods were used. Vector Control Division, Ministry of Health, Uganda, provided quality control for Kato–Katz slides, and the Medical Research Council Laboratories at Uganda Virus Research Institute provide quality control for malaria films.

rapid test algorithm [13]; for infants, RNA and DNA polymerase chain reaction methods were used. Vector Control Division, Ministry of Health, Uganda, provided quality control for Kato–Katz slides, and the Medical Research Council Laboratories at Uganda Virus Research Institute provide quality control for malaria films. Statistical Methods The aim was to examine the association between maternal helminth and malaria infections in pregnancy and malaria in the offspring. The sample size for the study was determined for the original trial objectives. To test the hypothesis that maternal albendazole or praziquantel in pregnancy would influence the incidence of malaria in infancy (assumed to be 0.5 per person-year in the maternal placebo group), a study with 2500 participants would have 80% power to show an 18% reduction or a 19% increase in the incidence of malaria in infancy, with P < .05, assuming that each intervention had an independent effect. Either direction of effect could happen, depending on whether helminth coinfection increases susceptibility to malaria infection and disease or decreases inflammation and hence reduces malaria-induced morbidity.

he incidence of malaria in infancy, with P < .05, assuming that each intervention had an independent effect. Either direction of effect could happen, depending on whether helminth coinfection increases susceptibility to malaria infection and disease or decreases inflammation and hence reduces malaria-induced morbidity. Of 2345 live births, 53 twins and 3 triplets were excluded, leaving 2289 children for inclusion in the analysis. Primary outcome was incidence of childhood clinical malaria from birth to age 5 years, and secondary outcome was prevalence of asymptomatic P. falciparum parasitemia as determined at annual visits to age 5 years. Childhood clinical malaria was defined as a history of recent fever or axillary temperature of ≥37.5°C and any parasitemia. Asymptomatic P. falciparum parasitemia was defined as a positive malaria slide in the absence of fever on the sampling day. Key predictor variables were maternal P. falciparum, hookworm, M. perstans, and S. mansoni infections at enrollment. Malaria was defined as peripheral parasitemia.

of ≥37.5°C and any parasitemia. Asymptomatic P. falciparum parasitemia was defined as a positive malaria slide in the absence of fever on the sampling day. Key predictor variables were maternal P. falciparum, hookworm, M. perstans, and S. mansoni infections at enrollment. Malaria was defined as peripheral parasitemia. For the primary outcome, time at risk began at birth and was censored at loss to follow-up, death, or age 5 years. Clinical malaria episodes within 14 days of an initial presentation were regarded as a recrudescence and excluded from the analysis; time at risk was adjusted accordingly, excluding these 14-day periods from the total person-time denominator. Crude hazard ratios (HRs) for the effect of malaria and helminth infections in pregnancy on the incidence of childhood malaria were calculated using Cox regression with robust standard errors to allow for within-child clustering of malaria episodes. Independent risk factors for maternal infections and childhood malaria that were significant (P ≤ .10) at the univariable analyses were entered into multivariable models. Variables included in the models were maternal age, education, parity, HIV status, mosquito net ownership, socioeconomic status, and geographical residential zone. Maternal P. falciparum and HIV infections and child albendazole were assessed as potential effect modifiers of the association between each maternal infection and childhood malaria. The secondary outcome was analyzed by combining data from all annual visits and comparing repeated prevalence of parasitemia between maternal malaria and helminth infection groups using random effects logistic regression, adjusting for the same confounders. Adjusted P values were calculated using likelihood ratio tests. Albendazole and praziquantel treatment in pregnancy had no effect on the incidence of clinical malaria [35]; hence there was no need to allow for them in the analysis. Statistical analysis was performed using Stata version 11. Rather than formally adjusting for multiple testing, we interpreted consistent results for related outcomes as providing evidence of a true association. Significance was defined as P values ≤ .05.

[35]; hence there was no need to allow for them in the analysis. Statistical analysis was performed using Stata version 11. Rather than formally adjusting for multiple testing, we interpreted consistent results for related outcomes as providing evidence of a true association. Significance was defined as P values ≤ .05. RESULTS Of 2345 live births, 1622 children (69%) were still under follow-up at 5 years, and a total of 33 178 clinic visits for illnesses were recorded [35]. A trial flow chart has been reported previously [35] and is available as Supplementary Figure 1. The total number of clinic visits was similar across maternal helminth groups (data not shown). Table 1 shows characteristics of the participating women and children. A complete description of maternal infections in pregnancy, previously reported [36], is provided in Supplementary Appendix 1. The overall mean P. falciparum parasite count in pregnancy was 163 (standard deviation [SD], 357), and only 51 (22%) of 236 mothers with parasitemia had >1000 parasite/µL blood. Two hundred two (8%) women were infected with P. falciparum and at least one of hookworm or S. mansoni or M. perstans. At delivery, only 69 of 2133 (3%) of the women tested had malaria parasitemia. Maternal HIV prevalence was 11%. Table 1. Characteristics of Mothers and Children Who Participated in the Study

/µL blood. Two hundred two (8%) women were infected with P. falciparum and at least one of hookworm or S. mansoni or M. perstans. At delivery, only 69 of 2133 (3%) of the women tested had malaria parasitemia. Maternal HIV prevalence was 11%. Table 1. Characteristics of Mothers and Children Who Participated in the Study Group Characteristic Summary Mothers (n = 2289) Mean age (±SD) at enrollment Gravidity 24 (5.4) 1 611 (27%) 2–4 1307 (57%) ≥5 371 (16%) Trimester (4 mv) 2 1051 (46%) 3 1234 (54%) Highest educational level attained (4mv) None 81 (4%) Primary 1152 (50%) Secondary 860 (38%) Tertiary 192 (8%) Socioeconomic status (44 mv) Lower 1028 (45%) Higher 1217 (53%) Infections Any helminth (29 mv) 1545 (68%) Hookworm (9 mv) 1004 (45%) Schistosoma mansoni (9 mv) 415 (18%) Mansonella perstans (8 mv) 492 (21%) Plasmodium falciparum (43 mv) 236 (10%) HIV 261 (11%) Received IPTp 2211 (97%) Owns mosquito net 1131 (49%) Primary source of water (5 mv) Open source 1910 (83%) Piped source 374 (16%) Primary source of fuel (6 mv) Firewood 408 (18%) Charcoal 1626 (71%) Paraffin 49 (2%) Gas/electricity 200 (9%) Children (n = 2289) Male 1167 (51%) Mean birthweight (±SD) 3.19 (±0.49) Abbreviations: HIV, human immunodeficiency virus; IPTp, intermittent presumptive treatment for malaria; mv, missing values; SD, standard deviation.

urce of fuel (6 mv) Firewood 408 (18%) Charcoal 1626 (71%) Paraffin 49 (2%) Gas/electricity 200 (9%) Children (n = 2289) Male 1167 (51%) Mean birthweight (±SD) 3.19 (±0.49) Abbreviations: HIV, human immunodeficiency virus; IPTp, intermittent presumptive treatment for malaria; mv, missing values; SD, standard deviation. The prevalence of helminth infections was low among the children: 2.9% (95% confidence interval [CI], 2.0–3.8), 5.2% (95% CI, 4.0–6.3), 7.7% (95% CI, 6.3–9.1), 9.0% (95% CI, 7.4–10.5), and 9.5% (95% CI, 7.9–11.1) at annual visits 1, 2, 3, 4, and 5 years, respectively. Of 2289 children, 1161 (51%) had at least 1 malaria episode; 459 (20%) had 1 episode, and 702 (31%) had ≥2 episodes (18 children had >10 episodes). The overall malaria incidence rate was 34 per 100 child-years, higher in the first 2 years (41 and 53 per 100 child-years, respectively), than in years 3, 4, and 5 (31, 20, and 20 per 100 child-years, respectively). The annual prevalence of asymptomatic parasitemia among the children was 5.9% (95% CI, 4.8–7.2), 7.1% (95% CI, 5.9–8.5), 4.7% 95% CI, (3.7–6.0), 4.3% (95% CI, 3.3–5.6) and 4.8% (95% CI, 3.8–6.1) at 1, 2, 3, 4, and 5 years, respectively. The overall mean parasite density was 8292 parasites/µL (SD, 18 576).

espectively). The annual prevalence of asymptomatic parasitemia among the children was 5.9% (95% CI, 4.8–7.2), 7.1% (95% CI, 5.9–8.5), 4.7% 95% CI, (3.7–6.0), 4.3% (95% CI, 3.3–5.6) and 4.8% (95% CI, 3.8–6.1) at 1, 2, 3, 4, and 5 years, respectively. The overall mean parasite density was 8292 parasites/µL (SD, 18 576). Variables Associated With Helminth and Malaria Infections in Pregnancy Older age, higher education, higher socioeconomic status, and mosquito net ownership were associated with a lower prevalence of maternal helminth and malaria infections [37]. HIV infection was negatively associated with hookworm, whereas HIV, hookworm, and M. perstans infections were associated with increased odds of maternal malaria. The risk of maternal helminth and malaria infection in pregnancy varied significantly by geographical zone [16]. Variables Associated With Childhood Malaria Table 2 shows that children of younger, poorer mothers, who had a low education level and did not own a mosquito net had a higher risk of childhood clinical malaria. The risk also varied significantly by both parity and geographical location of residence. Table 2. Variables Associated With Clinical Malaria Episodes in Childhood (aged 0–5 years)

n of younger, poorer mothers, who had a low education level and did not own a mosquito net had a higher risk of childhood clinical malaria. The risk also varied significantly by both parity and geographical location of residence. Table 2. Variables Associated With Clinical Malaria Episodes in Childhood (aged 0–5 years) Maternal Risk Factor Incidence Rate per 100 Child-Years (95% CI) Crude HR (95% CI) Adjusted HRa (95% CI) P Value Age (years) <25 34.70 (33.21–36.26) 1 1 ≥25 32.57 (30.79–34.45) 0.94 (.83–1.08) 0.84 (.71–.99) .04 Parity Primipara 31.78 (29.60–34.11) 1 1 Multipara (2–4) 32.83 (31.35–34.39) 1.04 (.90–1.20) 1.13 (.97–1.30) Grand multipara ( ≥5) 40.55 (37.53–43.80) 1.29 (1.06–1.57) 1.40 (1.10–1.78) .03 HIV Negative 34.38 (33.15–35.64) 1 1 Positive 29.21 (25.96–32.86) 0.85 (.68–1.06) 0.82 (.67–1.01) .14 Socioeconomic Lower status 37.57 (35.76–39.48) 1 1 Higher status 30.30 (28.82–31.85) 0.81 (.71–.92) 0.85 (.75–.96) .01 Education None/Primary 39.01 (37.32–40.78) 1 1 Postprimary 28.12 (26.60–29.73) 0.73 (.64–.82) 0.83 (.73–.94) .003 Owns bed net Yes 27.00 (25.55–28.53) 1 1 .01 No 40.67 (38.90–42.52) 1.50 (1.32–1.71) 1.17 (1.04–1.33) Water source Piped 29.93 (28.75–31.17) 1 1 Open 53.10 (49.63–56.82) 1.78 (1.54–2.06) 1.39 (1.19–1.61) <.001 Fuel source Indoor (electricity/gas) 14.30 (11.92–17.16) 1 1 Outdoor (paraffin/charcoal/wood) 35.79 (34.55–37.08) 2.53 (2.01–3.18) 1.83 (1.44–2.32) <.001 Geographical zone 1 15.19 (13.04–17.69) 1 1 2 19.32 (16.87–22.13) 1.26 (.93–1.71) 1.57 (1.15–2.13) <.001 3 46.98 (44.00–50.18) 3.07 (2.39–3.95) 3.28 (2.54–4.25) 4 16.49 (13.98–19.45) 1.08 (.79–1.48) 1.28 (.93–1.76) 5 28.47 (25.78–31.43) 1.85 (1.42–2.40) 2.14 (1.63–2.80) 6 38.72 (27.86–30.78) 2.52 (1.91–3.33) 2.60 (1.96–3.44) 7 38.71 (33.91–44.19) 2.54 (1.83–3.52) 2.63 (1.90–3.65) 8 32.95 (29.02–37.42) 2.17 (1.60–2.95) 2.21 (1.63–3.01) 9 64.60 (58.58–71.24) 4.23 (3.22–5.57) 3.96 (3.00–5.22) 10 30.38 (23.15–39.87) 2.00 (1.15–3.47) 1.99 (1.14–3.44) 11 67.79 (54.52–84.28) 4.42 (2.72–7.18) 3.67 (2.21–6.10) 12 57.00 (38.21–85.05) 3.63 (1.23–10.73) 1.04 (.28–3.82) 13 77.28 (63.46–94.11) 5.12 (3.64–7.20) 3.57 (2.46–5.17) Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio.

10 30.38 (23.15–39.87) 2.00 (1.15–3.47) 1.99 (1.14–3.44) 11 67.79 (54.52–84.28) 4.42 (2.72–7.18) 3.67 (2.21–6.10) 12 57.00 (38.21–85.05) 3.63 (1.23–10.73) 1.04 (.28–3.82) 13 77.28 (63.46–94.11) 5.12 (3.64–7.20) 3.57 (2.46–5.17) Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio. a Variables adjusted for each other. Association Between Malaria in Pregnancy and Childhood Malaria After adjusting for maternal age, parity, education, mosquito net ownership, household socioeconomic status, maternal HIV status, and location of residence, maternal malaria was associated with a significantly higher incidence of childhood malaria (adjusted hazard ratio [aHR], 1.23; 95% CI, 1.01–1.51]; P = .04). However, the positive association observed between maternal malaria and childhood asymptomatic parasitemia did not reach statistical significance (adjusted odds ratio [aOR], 1.27; 95% CI, .83–1.97; P = .28). Infection intensity did not affect the association. The association was not modified by maternal HIV (Pinteraction = .20) or child albendazole (P = .60).

d between maternal malaria and childhood asymptomatic parasitemia did not reach statistical significance (adjusted odds ratio [aOR], 1.27; 95% CI, .83–1.97; P = .28). Infection intensity did not affect the association. The association was not modified by maternal HIV (Pinteraction = .20) or child albendazole (P = .60). Association Between Maternal Helminth Infections in Pregnancy and Childhood Malaria Children of mothers with hookworm or M. perstans in pregnancy had significantly higher rates of clinical malaria and increased odds of asymptomatic parasitemia compared with children of uninfected mothers (Tables 3 and 4). Simultaneously adjusting for each helminth did not change the association between maternal hookworm and childhood clinical malaria (aHR, 1.18; 95% CI, 1.04–1.34; P = .01) or childhood asymptomatic parasitemia (aOR, 1.43; 95% CI, 1.08–1.90; P = .01) but weakened the association between maternal M. perstans and childhood clinical malaria (aHR, 1.14; 95% CI, 1.00–1.30; P = .06) or childhood asymptomatic parasitemia (aOR, 1.26; 95% CI, .92–1.74; P = .15). Overall, there was no association between maternal S. mansoni infection and childhood clinical malaria or asymptomatic parasitaemia (Tables 3 and 4). Maternal HIV did not modify the associations between childhood malaria and maternal hookworm (P = .10), M. perstans (P = .80), or S. mansoni (P = .80). Similarly, child albendazole did not modify the associations between childhood malaria and maternal hookworm (P = .30), M. perstans (P = .10), or S. mansoni (P = .70). Table 3. Association Between Maternal Helminth Infections and Clinical Malaria Episodes in Childhood (aged 0–5 years)

P = .80), or S. mansoni (P = .80). Similarly, child albendazole did not modify the associations between childhood malaria and maternal hookworm (P = .30), M. perstans (P = .10), or S. mansoni (P = .70). Table 3. Association Between Maternal Helminth Infections and Clinical Malaria Episodes in Childhood (aged 0–5 years) Maternal Infection Status Incidence Rate per 100 Person-Years (95% CI) Cox HR (95% CI) Adjusted HRa (95% CI) P Value Adjusted HRb (95% CI) P Value No hookworm 29.28 (27.86–30.78) 1 1 1 Hookworm 39.93 (38.04–41.91) 1.36 (1.20–1.54) 1.26 (1.10–1.43) <.001 1.23 (1.09–1.40) .001 No Mansonella perstans 31.83 (30.58–33.15) 1 1 1 M. perstans 40.75 (38.06–43.64) 1.30 (1.12–1.51) 1.24 (1.07–1.42) .003 1.19 (1.03–1.37) .02 No Schistosoma mansoni 33.07 (31.81–34.38) 1 1 1 S. mansoni 37.74 (34.96–40.74) 1.14 (.97–1.35) 1.07 (.91–1.26) .43 1.06 (.90–1.25) .48 Abbreviations: CI, confidence interval; HR, hazard ratio. a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal malaria and human immunodeficiency virus infections. b In addition adjusted for geographical zone. Table 4. Association Between Maternal Helminth Infections in Pregnancy and the Prevalence of Childhood Asymptomatic Parasitemia (aged 0–5 years)

a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal malaria and human immunodeficiency virus infections. b In addition adjusted for geographical zone. Table 4. Association Between Maternal Helminth Infections in Pregnancy and the Prevalence of Childhood Asymptomatic Parasitemia (aged 0–5 years) Maternal Infection Status No. of Children Ever Parasitemic at Any Time Point (%) Crude OR (95% CI) Adjusted ORa (95% CI) P Value Adjusted ORb (95% CI) P Value No hookworm 160 (4.2) 1 1 1 Hookworm 179 (6.6) 1.76 (1.31–2.38) 1.63 (1.22–2.17) .001 1.57 (1.18–2.08) .002 No Mansonella perstans 282 (4.9) 1 1 1 M. perstans 113 (7.4) 1.71 (1.24–2.35) 1.49 (1.08–2.06) .02 1.36 (.99–1.88) .06 No Schistosoma mansoni 322 (5.4) 1 1 S. mansoni 76 (5.7) 1.07 (.75–1.53) 1.00 (.69–1.44) 1.00 0.99 (.69–1.41) .96 Abbreviations: CI, confidence interval; OR, odds ratio. a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal malaria and human immunodeficiency virus infection. b In addition adjusted for geographical zone. Association Between Malaria–Helminth Coinfections in Pregnancy and Childhood Malaria Associations between childhood malaria and maternal hookworm or M. perstans did not differ when stratified by maternal malaria status. An association between maternal S. mansoni and childhood parasitemia was observed only in the presence of maternal malaria (Tables 5 and 6). Table 5. Association Between Maternal Helminth Infections in Pregnancy and Childhood Clinical Malaria (aged 0–5 years), Stratified by Maternal Malaria

by maternal malaria status. An association between maternal S. mansoni and childhood parasitemia was observed only in the presence of maternal malaria (Tables 5 and 6). Table 5. Association Between Maternal Helminth Infections in Pregnancy and Childhood Clinical Malaria (aged 0–5 years), Stratified by Maternal Malaria Maternal Infection Status Incidence rate per 100 pyrs (95% CI) Cox HR (95% CI) Adjusted HRa (95% CI) P Value Adjusted HRb (95% CI) P Value No malaria No hookworm 28.81 (27.33,30.38) 1 1 1 Hookworm 37.93 (35.95,40.01) 1.31 (1.15–1.50) 1.20 (1.05–1.38) .01 1.17 (1.03–1.34) .02 Had malaria No hookworm 35.09 (29.96,41.09) 1 1 1 Hookworm 52.46 (46.48,59.21) 1.48 (1.03–2.14) 1.63 (1.15–2.32) .01 1.60 (1.13–2.26) .01 Pinteraction = .23c No malaria No Mansonella perstans 30.79 (29.49,32.15) 1 1 1 M. perstans 40.05 (37.13,43.21) 1.30 (1.12–1.51) 1.26 (1.08–1.46) .003 1.21 (1.03–1.40) .02 Had malaria No M. perstans 44.33 (39.35,49.93) 1 1 1 M. perstans 44.34 (37.68,52.18) 0.99 (.70–1.40) 1.25 (.87–1.79) .23 1.22 (.85–1.76) .28 Pinteraction = .32c No malaria No Schistosoma mansoni 31.85 (30.53,33.22) 1 1 1 S. mansoni 36.41 (33.52,39,54) 1.14 (.96–1.36) 1.10 (.92–1.31) .29 1.09 (.92–1.29) .33 Had malaria No S. mansoni 43.91 (39.49,48.83) 1 1 1 S. mansoni 46.36 (36.97,58.14) 1.06 (.61–1.86) 0.86 (.53–1.41) .56 0.88 (.53–1.44) .60 Pinteraction = .50c Abbreviations: CI, confidence interval; HR, hazard ratio. a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal human immunodeficiency virus infection.

Maternal Infection Status Incidence rate per 100 pyrs (95% CI) Cox HR (95% CI) Adjusted HRa (95% CI) P Value Adjusted HRb (95% CI) P Value No malaria No hookworm 28.81 (27.33,30.38) 1 1 1 Hookworm 37.93 (35.95,40.01) 1.31 (1.15–1.50) 1.20 (1.05–1.38) .01 1.17 (1.03–1.34) .02 Had malaria No hookworm 35.09 (29.96,41.09) 1 1 1 Hookworm 52.46 (46.48,59.21) 1.48 (1.03–2.14) 1.63 (1.15–2.32) .01 1.60 (1.13–2.26) .01 Pinteraction = .23c No malaria No Mansonella perstans 30.79 (29.49,32.15) 1 1 1 M. perstans 40.05 (37.13,43.21) 1.30 (1.12–1.51) 1.26 (1.08–1.46) .003 1.21 (1.03–1.40) .02 Had malaria No M. perstans 44.33 (39.35,49.93) 1 1 1 M. perstans 44.34 (37.68,52.18) 0.99 (.70–1.40) 1.25 (.87–1.79) .23 1.22 (.85–1.76) .28 Pinteraction = .32c No malaria No Schistosoma mansoni 31.85 (30.53,33.22) 1 1 1 S. mansoni 36.41 (33.52,39,54) 1.14 (.96–1.36) 1.10 (.92–1.31) .29 1.09 (.92–1.29) .33 Had malaria No S. mansoni 43.91 (39.49,48.83) 1 1 1 S. mansoni 46.36 (36.97,58.14) 1.06 (.61–1.86) 0.86 (.53–1.41) .56 0.88 (.53–1.44) .60 Pinteraction = .50c Abbreviations: CI, confidence interval; HR, hazard ratio. a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal human immunodeficiency virus infection. b In addition adjusted for geographical zone. c Interaction test corresponding to the model adjusted for geographical zone. Table 6. Association Between Maternal Helminth Infections in Pregnancy and the Prevalence of Childhood Asymptomatic Parasitemia (aged 0–5 years), Stratified by Maternal Malaria

a Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal human immunodeficiency virus infection. b In addition adjusted for geographical zone. c Interaction test corresponding to the model adjusted for geographical zone. Table 6. Association Between Maternal Helminth Infections in Pregnancy and the Prevalence of Childhood Asymptomatic Parasitemia (aged 0–5 years), Stratified by Maternal Malaria Maternal Infection Status No. of Children Ever Parasitemic at Any Time Point (%) Crude OR (95% CI) Adjusted ORb (95% CI) P Value Adjusted ORc (95% CI) P Value No malaria No hookworm 160 (4.2) 1 1 1 Hookworm 179 (6.6) 1.76 (1.31–2.38) 1.48 (1.09–2.00) .01 1.41 (1.05–1.90) .02 Had malaria No hookworm 15 (4.5) 1 1 1 Hookworm 38 (10.4) 2.70 (1.14–6.44) 3.38 (1.33–8.63) .01 3.29 (1.30–8.34) .01 Pinteraction = .11a No malaria No Mansonella perstans 249 (4.7) 1 1 1 M. perstans 90 (7.0) 1.63 (1.15–2.31) 1.44 (1.02–2.04) .04 1.31 (.93–1.84) .12 Had malaria No M. perstans 31 (6.7) 1 1 1 M. perstans 22 (9.3) 1.62 (.67–3.90) 2.17 (.80–5.88) .13 2.09 (.78–5.65) .15 Pinteraction = .77a No malaria No Schistosoma mansoni 279 (5.3) 1 1 1 S. mansoni 60 (5.0) 0.95 (.64–1.40) 0.85 (.57–1.26) .42 0.84 (.57–1.23) .36 Had malaria No S. mansoni 37 (6.3) 1 1 1 S. mansoni 16 (14.2) 2.98 (1.08–8.24) 3.12 (1.05–9.27) .04 3.15 (1.06–9.39) .04 Pinteraction = .02a Abbreviations: CI, confidence interval; OR, odds ratio. a Interaction test corresponding to the model adjusted for geographical zone.

Maternal Infection Status No. of Children Ever Parasitemic at Any Time Point (%) Crude OR (95% CI) Adjusted ORb (95% CI) P Value Adjusted ORc (95% CI) P Value No malaria No hookworm 160 (4.2) 1 1 1 Hookworm 179 (6.6) 1.76 (1.31–2.38) 1.48 (1.09–2.00) .01 1.41 (1.05–1.90) .02 Had malaria No hookworm 15 (4.5) 1 1 1 Hookworm 38 (10.4) 2.70 (1.14–6.44) 3.38 (1.33–8.63) .01 3.29 (1.30–8.34) .01 Pinteraction = .11a No malaria No Mansonella perstans 249 (4.7) 1 1 1 M. perstans 90 (7.0) 1.63 (1.15–2.31) 1.44 (1.02–2.04) .04 1.31 (.93–1.84) .12 Had malaria No M. perstans 31 (6.7) 1 1 1 M. perstans 22 (9.3) 1.62 (.67–3.90) 2.17 (.80–5.88) .13 2.09 (.78–5.65) .15 Pinteraction = .77a No malaria No Schistosoma mansoni 279 (5.3) 1 1 1 S. mansoni 60 (5.0) 0.95 (.64–1.40) 0.85 (.57–1.26) .42 0.84 (.57–1.23) .36 Had malaria No S. mansoni 37 (6.3) 1 1 1 S. mansoni 16 (14.2) 2.98 (1.08–8.24) 3.12 (1.05–9.27) .04 3.15 (1.06–9.39) .04 Pinteraction = .02a Abbreviations: CI, confidence interval; OR, odds ratio. a Interaction test corresponding to the model adjusted for geographical zone. b Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal human immunodeficiency virus infection. c In addition adjusted for geographical zone.

Maternal Infection Status No. of Children Ever Parasitemic at Any Time Point (%) Crude OR (95% CI) Adjusted ORb (95% CI) P Value Adjusted ORc (95% CI) P Value No malaria No hookworm 160 (4.2) 1 1 1 Hookworm 179 (6.6) 1.76 (1.31–2.38) 1.48 (1.09–2.00) .01 1.41 (1.05–1.90) .02 Had malaria No hookworm 15 (4.5) 1 1 1 Hookworm 38 (10.4) 2.70 (1.14–6.44) 3.38 (1.33–8.63) .01 3.29 (1.30–8.34) .01 Pinteraction = .11a No malaria No Mansonella perstans 249 (4.7) 1 1 1 M. perstans 90 (7.0) 1.63 (1.15–2.31) 1.44 (1.02–2.04) .04 1.31 (.93–1.84) .12 Had malaria No M. perstans 31 (6.7) 1 1 1 M. perstans 22 (9.3) 1.62 (.67–3.90) 2.17 (.80–5.88) .13 2.09 (.78–5.65) .15 Pinteraction = .77a No malaria No Schistosoma mansoni 279 (5.3) 1 1 1 S. mansoni 60 (5.0) 0.95 (.64–1.40) 0.85 (.57–1.26) .42 0.84 (.57–1.23) .36 Had malaria No S. mansoni 37 (6.3) 1 1 1 S. mansoni 16 (14.2) 2.98 (1.08–8.24) 3.12 (1.05–9.27) .04 3.15 (1.06–9.39) .04 Pinteraction = .02a Abbreviations: CI, confidence interval; OR, odds ratio. a Interaction test corresponding to the model adjusted for geographical zone. b Adjusted for maternal age, education, parity, net ownership, socioeconomic status, and maternal human immunodeficiency virus infection. c In addition adjusted for geographical zone. DISCUSSION To our knowledge, this is the first report of a birth cohort showing an association between helminth infections in pregnancy and childhood malaria. Earlier studies exploring the influence of helminth infections on the course of malaria and the effect of malaria–helminth coinfections [18, 38–41] used different study designs and showed both beneficial and detrimental association [36].

ing an association between helminth infections in pregnancy and childhood malaria. Earlier studies exploring the influence of helminth infections on the course of malaria and the effect of malaria–helminth coinfections [18, 38–41] used different study designs and showed both beneficial and detrimental association [36]. Our main finding was higher malaria morbidity (both in terms of clinical episodes and of asymptomatic parasitemia) among children of mothers with hookworm and M. perstans infections in pregnancy compared with children of uninfected mothers. Additionally we observed an increased rate of childhood clinical malaria in children of mothers with malaria compared with children of mothers without malaria. Whereas other studies have described associations between placental malaria infection and childhood malaria [8, 9, 42, 43], we report associations with maternal peripheral parasitemia.

an increased rate of childhood clinical malaria in children of mothers with malaria compared with children of mothers without malaria. Whereas other studies have described associations between placental malaria infection and childhood malaria [8, 9, 42, 43], we report associations with maternal peripheral parasitemia. At enrolment, pregnant women with hookworm and M. perstans infections were at increased risk of peripheral malaria parasitemia [16], suggesting that the association we observed between the maternal helminth infections and childhood malaria might be explained by the association with malaria in pregnancy. However, in the analyses stratified by maternal malaria status, maternal hookworm infection was associated with an increased rate of childhood clinical malaria and an increased prevalence of childhood parasitemia, irrespective of whether the mother had malaria or not. Also, the association between maternal hookworm and childhood malaria remained consistent after simultaneously adjusting for each helminth, whereas the adjustment weakened the association between maternal M. perstans and childhood malaria. This suggests that the association between maternal hookworm and childhood malaria may be independent of the association with maternal malaria and the other helminth infections.

imultaneously adjusting for each helminth, whereas the adjustment weakened the association between maternal M. perstans and childhood malaria. This suggests that the association between maternal hookworm and childhood malaria may be independent of the association with maternal malaria and the other helminth infections. A possible explanation for these findings is that mothers with helminths and malaria in pregnancy came from a high malaria transmission environment and that the child's increased malaria risk was partly due to this. The risk of maternal malaria varied significantly by geographical zone of residence [16], suggesting that children of women living in high malaria transmission areas would be more exposed than children of mothers living in areas with lower transmission. Nevertheless, the association between maternal malaria, hookworm, and M. perstans and childhood malaria persisted after adjusting for location of residence. However, we were unable to accurately assess the contribution of malaria transmission in the observed associations because we did not measure malaria exposure at the household or individual level. In the multivariable analyses, we adjusted for geographical location of residence assuming homogenous malaria transmission zones within an area of 4 km in diameter, which may not be sensitive to within-area transmission variations, and therefore we cannot exclude a role for malaria transmission in the observed associations.

e multivariable analyses, we adjusted for geographical location of residence assuming homogenous malaria transmission zones within an area of 4 km in diameter, which may not be sensitive to within-area transmission variations, and therefore we cannot exclude a role for malaria transmission in the observed associations. Previous studies have reported greater malaria morbidity associated with S. mansoni coinfection [44]. In our study, maternal schistosome infections were mostly light to moderate, and only 2% (37 of 2237) of the mothers had malaria–S. mansoni coinfection, but we observed that children of S. mansoni–infected mothers were at higher risk of malaria parasitemia only if the mother also had malaria. This result should be interpreted with caution because it is not consistent with the association between clinical malaria and maternal S. mansoni, and we cannot rule out that this finding was due to chance alone. However, the result suggests the hypothesis that exposure to both malaria and helminths is required to alter the way the fetus's initial response to malaria is primed. In fact this may also be the case for the malaria–hookworm interaction. Although the associations between maternal hookworm and childhood malaria were statistically similar between children of mothers with and without malaria, the point estimates were lower in the strata of mothers without malaria. There may have been some misclassification of maternal malaria and, if exposure of the fetus to both malaria and hookworm is required for the observed effects, the observed trend in hazard ratios could have occurred if about 20% of the “no malaria” mothers actually had malaria at some time during the pregnancy. This would explain the absence of interaction between maternal malaria and hookworm or M. perstans.

to both malaria and hookworm is required for the observed effects, the observed trend in hazard ratios could have occurred if about 20% of the “no malaria” mothers actually had malaria at some time during the pregnancy. This would explain the absence of interaction between maternal malaria and hookworm or M. perstans. A possible immunological explanation for the observed results is that fetal exposure to maternal helminth and malaria infections may induce T-cell hyporesponsiveness, downmodulating immunity to helminths [24] and malaria antigens and modifying fetal acquisition of immunity to malaria [9, 21, 45]. Tolerance in offspring exposed to parasite antigens in utero [46], resulting in increased susceptibility to infection, has been reported. Alternatively, coexposure to helminths and malaria might bias the profile of the antimalarial response toward a T-helper 2 profile or regulatory profile [24]. Another explanation is that individuals susceptible to helminths are more susceptible to malaria. Studies on helminth infections have shown that only a minority of individuals account for the majority of infection burden [47]. This might be due to variation in parasite exposure but could also be due to variation in individual genetic susceptibility [48]. Studies have suggested genetic susceptibility to polyparasitism [49]; hence pregnant women with a genetic susceptibility to hookworm or M. perstans infection might be more susceptible to malaria infection. However, adjusting for maternal malaria did not alter the associations between maternal hookworm or M. perstans infections and childhood malaria, suggesting that these helminth infections are not simply a marker for genetic susceptibility to malaria.

M. perstans infection might be more susceptible to malaria infection. However, adjusting for maternal malaria did not alter the associations between maternal hookworm or M. perstans infections and childhood malaria, suggesting that these helminth infections are not simply a marker for genetic susceptibility to malaria. We used data collected from the EMaBS, a trial that investigated whether anthelminthic treatment during pregnancy could alter the effects of prenatal helminth exposure. Anthelminthic treatment in pregnancy was effective [14], but albendazole and praziquantel had no effect on childhood malaria overall or in subgroup analyses by maternal hookworm and S. mansoni infections [35]. This could imply that the association between maternal hookworm and childhood malaria was established early in pregnancy and that single-dose albendazole or praziquantel in the second or third trimester was not sufficient to eliminate or reverse any effect of helminth infection in pregnancy on malaria susceptibility in the offspring. This is particularly likely to be true because this analysis was based on assessment of maternal helminth and malaria status at enrollment to the study, before the trial intervention and intermittent presumptive treatment for malaria were provided to the women (>90% of the mothers received intermittent presumptive treatment for malaria in the second and/or third trimesters, greatly reducing maternal malaria prevalence). It is likely that most of the fetal exposure to maternal malaria occurred before the anthelminthic trial intervention. In contrast, childhood quarterly albendazole was associated with a 15% reduction in the incidence of clinical malaria [35].

econd and/or third trimesters, greatly reducing maternal malaria prevalence). It is likely that most of the fetal exposure to maternal malaria occurred before the anthelminthic trial intervention. In contrast, childhood quarterly albendazole was associated with a 15% reduction in the incidence of clinical malaria [35]. Our major limitation was the use of single samples for ascertainment of maternal malaria and helminth infections. Although microscopy is the gold standard in malaria diagnosis, sensitivity is low in pregnancy due to low parasite densities and placental sequestration. In a review [50], the pooled prevalence estimate for peripheral malaria in East and Southern Africa was 32.0% (95% CI 25.9–38.0; n = 11 688), considerably higher than the prevalence we observed. Misclassification of malaria-infected mothers classified as malaria-negative would weaken the strength of observed associations between maternal and childhood malaria. Similarly, for helminths, some mothers may have been misclassified as uninfected. The underestimation of these key exposures may also account for the lack of interaction between maternal malaria and helminths as discussed above. Lastly, we cannot exclude the possibility that confounding by unmeasured covariables could explain some of the observations. Nonetheless, this study had the unique advantage of a prospective birth cohort design, with a large sample size, a long follow-up period, and comprehensive data on potential confounders minimizing residual confounding.

de the possibility that confounding by unmeasured covariables could explain some of the observations. Nonetheless, this study had the unique advantage of a prospective birth cohort design, with a large sample size, a long follow-up period, and comprehensive data on potential confounders minimizing residual confounding. This study provides the first report of an association between helminth infections in pregnancy and malaria in the offspring and suggests that helminth infections in pregnancy may increase the overall burden of childhood malaria in regions of coendemicity. The association between maternal hookworm and childhood malaria was consistent for clinical malaria episodes and asymptomatic parasitemia. However, the mechanism is unclear, and studies are needed to elucidate the significance of maternal helminth infections on fetal and early childhood antimalarial responses. Our findings support the strategy of integrated malaria–helminth control to accelerate the reduction of malaria morbidity and provide pertinent knowledge for the evaluation of malaria vaccine trials because results might be modified by concurrent helminth infections in pregnancy.

early childhood antimalarial responses. Our findings support the strategy of integrated malaria–helminth control to accelerate the reduction of malaria morbidity and provide pertinent knowledge for the evaluation of malaria vaccine trials because results might be modified by concurrent helminth infections in pregnancy. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgements. We thank all staff and participants of the EMaB Study, the fields team in Entebbe and Katabi, and the staff of Entebbe hospital and MRC/UVRI Uganda Research Unit on AIDS. We thank Barbara Willey for her statistical advice. Financial support. The Entebbe Mother and Baby Study is supported by the Wellcome Trust (grant numbers 064693, 079110, 95778) with additional funding from the European Community's Seventh Framework Programme (FP7/2007–2013) under EC-GA number 241642. J. N. was supported in part by a PhD fellowship from the Malaria Capacity Development Consortium which is funded by Wellcome Trust (grant number WT084289MA). Potential conflict of interest. All other authors report no potential conflicts.

Financial support. The Entebbe Mother and Baby Study is supported by the Wellcome Trust (grant numbers 064693, 079110, 95778) with additional funding from the European Community's Seventh Framework Programme (FP7/2007–2013) under EC-GA number 241642. J. N. was supported in part by a PhD fellowship from the Malaria Capacity Development Consortium which is funded by Wellcome Trust (grant number WT084289MA). Potential conflict of interest. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Staphylococcus aureus is an important opportunistic human pathogen and the cause of a large burden of morbidity and mortality. The ability of the pathogen to bind to and activate platelets, small (2–4 µm), blood cells responsible for maintaining normal hemostasis leads to the formation of platelet-bacteria thrombi on the surface of heart valves, which is required for the development of endocarditis since platelets attached to damaged valves serve as foci for attachment of bacteria circulating in the blood [1]. Several studies have shown that S. aureus binds platelets and induces their aggregation. The pathogen possesses a variety of surface proteins known as microbial surface components reacting with adhesive matrix molecules (MSCRAMMs), some of which are virulence factors in models of S. aureus endocarditis. MSCRAMMs attach the bacterium to platelets, either indirectly, by binding to fibrinogen, simultaneously bound to the platelet surface by integrin αIIbβ3, or by binding directly to αIIbβ3, thus inducing outside-in signaling and platelet activation [2, 3]. Such observations rely on washed S. aureus cells and thus ignore the contribution of bacterial molecules secreted into the extracellular milieu.

brinogen, simultaneously bound to the platelet surface by integrin αIIbβ3, or by binding directly to αIIbβ3, thus inducing outside-in signaling and platelet activation [2, 3]. Such observations rely on washed S. aureus cells and thus ignore the contribution of bacterial molecules secreted into the extracellular milieu. Although induction of thrombus formation by S. aureus has been characterized extensively, infection of wounds by this pathogen frequently results in impaired healing, the mechanisms of which are not fully understood [4]. S. aureus extracellular proteins Efb inhibits platelet aggregation by binding to fibrinogen [5]. Inhibition of platelet activity by Efb or pharmacological antagonists causes decreased killing of S. aureus in whole blood and increases the lethality of S. aureus infection in a mouse model [6]. S. aureus lipoteichoic acid (LTA) was previously shown to inhibit activation of platelets, although a role in hemostasis its relevance to the S. aureus-platelet interaction and the mechanism(s) by which inhibition is achieved are not understood [7]. In this study, LTA inhibited activation of human platelets by physiological agonists and S. aureus. Furthermore, LTA inhibits platelet function and thrombus formation in vivo by binding platelet activating factor receptor (PafR), a phospholipid receptor that binds LTA and is associated with an increased platelet intracellular cyclic adenosine monophosphate (cAMP) concentration.

y physiological agonists and S. aureus. Furthermore, LTA inhibits platelet function and thrombus formation in vivo by binding platelet activating factor receptor (PafR), a phospholipid receptor that binds LTA and is associated with an increased platelet intracellular cyclic adenosine monophosphate (cAMP) concentration. METHODS Reagents Collagen was obtained from Nycomed, thrombin, Ginkoglide B, Mouse immunoglobulin G (IgG) from Sigma, and cross-linked collagen-related peptide (CRP-XL) from R. Farndale (University of Cambridge, UK). Anti-LTA (pagibaximab) was provided by Biosynexus Inc. Anti-TLR2 (T2.5) was purchased from eBioscience, anti-CD14 (UCH-M1) from AbD serotec, anti-PafR from Cayman Chemical, anti-PhosphoVASP from Cell Signaling Technology. Bacterial Strains Used Wild-type S. aureus SA113 was used with S. aureus SA113 ΔdltABCD [8], S. aureus SA113 ΔtagO [9], and S. aureus SA113 Δlgt [10]. S. aureus SEJ1 and isogenic strains ΔgdpP, ΔgdpPΔltaS, and pCN34-ltaS were used for mutant studies [11]. Bacillus subtilis 128 and Streptococcus pneumoniae D34 were used. LTA Extraction S. aureus was grown in BHI 37°C and centrifuged at 20 463 g for 15 minutes. The pellet was resuspended in 50% butanol/water. LTA was resuspended in a 1 mM sodium acetate, 15% 1-propanol buffer followed by a 15%–60% 1-propanol elution gradient, dialyzed against dH20, and the concentration determined by phosphate assay [12].

ureus was grown in BHI 37°C and centrifuged at 20 463 g for 15 minutes. The pellet was resuspended in 50% butanol/water. LTA was resuspended in a 1 mM sodium acetate, 15% 1-propanol buffer followed by a 15%–60% 1-propanol elution gradient, dialyzed against dH20, and the concentration determined by phosphate assay [12]. LTA From S. aureus Culture Supernatants S. aureus cultures were centrifuged at c. 12 000 g for 10 minutes to remove cells. In total, 2.3M (NH4)2SO4 was added to the supernant overnight at 4°C. The supernatant was centrifuged at 20 000 g for 20 minutes at 4°C and the pellet resuspended in 2 mL of phosphate-buffered saline (PBS). To standardize supernatant preparations, including those lacking LTA, proteins carried over were quantified by Bradford assay. Where appropriate, LTA concentrations were determined as above. Preparation of Human Platelets Human blood was obtained from healthy volunteers who gave informed consent. Ethical approval was obtained from the University of Reading Research Ethics Committee. Platelets were prepared as described elsewhere [13]. In total, 4 × 108 platelets/mL were incubated for 15 minutes with LTA and stimulated by agonists. Aggregation was measured in an optical aggregometer (Chronolog). Percentage inhibition of aggregation by LTA was calculated by dividing maximal aggregation of LTA-treated samples by the aggregation achieved by the given agonist alone. Centrifuged S. aureus were washed 3 times in Tyrodes buffer and adjusted for a final experimental OD600 0.3. Aggregation was measured up to 15 minutes.

ercentage inhibition of aggregation by LTA was calculated by dividing maximal aggregation of LTA-treated samples by the aggregation achieved by the given agonist alone. Centrifuged S. aureus were washed 3 times in Tyrodes buffer and adjusted for a final experimental OD600 0.3. Aggregation was measured up to 15 minutes. Flow Cytometry In total, 5 µL of platelet-rich plasma (PRP; 4 × 108 cells/mL) was incubated with anti-PafR (50 µg/mL), IgG2a (50 µg/mL), or Tyrode buffer for 30 minutes, then incubated with various concentrations of FITC-LTA for 15 minutes. Fluorescence intensity of the sample was measured using a BD Accuri C6 flow cytometry; 10 000 events per sample were measured. Measurement of Intracellular [Ca2+]i Platelets were preloaded with the fluorescent dye Fluo-4NW as described elsewhere [13]. PRP was preincubated with LTA for 15 minutes before being stimulated with CRP-XL, and calcium release was measured using a Fluoroskan reader (Thermolab Systems) at 485/530 nm. In Vitro Thrombus Formation Whole citrated blood was perfused through a Vena8Biochip (Cellex, Dublin). Z-stack images were taken every 30 seconds using a Nikon eclipse (TE2000-U) microscope; data were analyzed using Slidebook5 software (Intelligent Imaging Innovations, Denver, USA) Tail-bleeding Assay Procedures were approved by the University of Reading Animal Ethics Committee and the Home Office. Sixteen age-matched C57BL/6 mice were killed using ketamine (80 mg/kg) and xylazine (5 mg/kg) administered intraperitoneally prior to a tail biopsy. Time to cessation of bleeding was measured up to 20 minutes.

eding Assay Procedures were approved by the University of Reading Animal Ethics Committee and the Home Office. Sixteen age-matched C57BL/6 mice were killed using ketamine (80 mg/kg) and xylazine (5 mg/kg) administered intraperitoneally prior to a tail biopsy. Time to cessation of bleeding was measured up to 20 minutes. RESULTS S. aureus LTA Inhibits Platelets Aggregation by Various Platelet Agonists LTA is found at high concentrations within S. aureus cells and the extracellular milieu [14] and could thus interact with platelets during infection. LTA present in the supernatant and S. aureus cell envelope have the same structure [15]. In this study we investigated LTA inhibition of platelet aggregation. First, we confirmed equal inhibitory activity for LTA purified from the supernatant and cells (Supplementary Figure 1A). For efficiency, remaining experiments used LTA extracted from the cell envelope. To determine which signaling pathways LTA inhibits, aggregation assays were performed with well-characterized platelet agonists. Purified LTA from S. aureus SA113 was preincubated with washed human platelets before activation with CRP-XL, a collagen receptor Glycoprotein VI selective agonist (Figure 1Ai and 1Aii), platelet activating factor (paf; Figure 1Bi and 1Bii) or thrombin (Figure 1Ci and 1Cii). LTA inhibited aggregation in a dose-dependent manner with all agonists. LTA was incubated for varying periods of time with washed platelets to observe any time-dependent effects on aggregation. Using 4 µg/mL LTA, inhibition of aggregation increased in a time-dependent manner (Figure 1Di and 1Dii). Platelet activation was inhibited over extended time periods (Supplementary Figure 1B). The highest concentration of LTA used with thrombin (3 µg/mL), CRP-XL (4 µg/mL), and paf (16 µg/mL) as platelet agonists produced a 40%, 85%, and 50% reduction in platelet aggregation respectively, thus showing LTA to be a potent inhibitor of platelet aggregation. Figure 1. LTA from Staphylococcus aureus inhibits platelet aggregation. Washed human platelets (4 × 108 cells/mL) or PRP were preincubated with LTA or tyrodes buffer and stimulated with various platelet agonists. Aggregation was measured as change in light transmission. Ai–Di, Representative aggregation traces of platelets incubated with LTA and stimulated with various platelet agonists. All aggregation traces commence upon addition of agonist.

eincubated with LTA or tyrodes buffer and stimulated with various platelet agonists. Aggregation was measured as change in light transmission. Ai–Di, Representative aggregation traces of platelets incubated with LTA and stimulated with various platelet agonists. All aggregation traces commence upon addition of agonist. Aii–Dii, Data are plotted as percentage inhibition of aggregation or percentage aggregation (vehicle treated representing 100% aggregation) and represent mean values ± SEM. A, Platelets were pre-incubated with LTA at a range of concentrations followed by stimulation with CRP-XL (0.5 µg/mL). Aggregation was measured for 90 seconds. B, Platelets were preincubated with LTA at a range of concentrations followed by stimulation with Paf (37.5 µg/mL). Aggregation was measured for 90 seconds. C, Platelets were pre-incubated with LTA at a range of concentrations followed by stimulation with thrombin (0.05 units/mL). Aggregation was measured for 90 seconds. D, Platelets were preincubated with LTA (4 µg/mL) for 8, 10, 12, 14, and 16 minutes followed by stimulation with CRP-XL (0.5 µg/mL). Aggregation was measured for 90 seconds. E, PRP was pre-incubated with various concentrations of LTA for 15 minutes followed by stimulation with whole S. aureus SA113 cells (2 × 108 cells/mL). Aggregation was measured for up to 20 minutes. Data are plotted as increase in lag time and represent mean values ± SEM. *P < .05 Abbreviations: CRP-XL, cross-linked collagen-related peptide; LTA, lipoteichoic acid; PRP, platelet-rich plasma; SEM, standard error of the mean.

le S. aureus SA113 cells (2 × 108 cells/mL). Aggregation was measured for up to 20 minutes. Data are plotted as increase in lag time and represent mean values ± SEM. *P < .05 Abbreviations: CRP-XL, cross-linked collagen-related peptide; LTA, lipoteichoic acid; PRP, platelet-rich plasma; SEM, standard error of the mean. Whole bacterial cells stimulate platelet activation via formation of fibrinogen or fibronectin bridges between integrin αIIbβ3 and S. aureus MSCRAMMs [3, 16, 17]. Having demonstrated LTA inhibition of platelet activation by physiological agonists, we examined the ability of exogenous LTA to inhibit S. aureus induced platelet aggregation in PRP (Figure 1E). Increasing LTA concentrations increased the lag time to platelet aggregation.

rin αIIbβ3 and S. aureus MSCRAMMs [3, 16, 17]. Having demonstrated LTA inhibition of platelet activation by physiological agonists, we examined the ability of exogenous LTA to inhibit S. aureus induced platelet aggregation in PRP (Figure 1E). Increasing LTA concentrations increased the lag time to platelet aggregation. To confirm that the observed inhibition was due to LTA rather than a copurifying contaminant, platelets were pretreated with monoclonal anti-LTA and LTA before stimulation with CRP-XL. This blocked the LTA inhibitory effect (Figures 2Ai and 2Aii). An isotype (IgG1) matched control had no effect (results not shown). Lipoproteins that can copurify with LTA are sometimes responsible for the immunological activities that have been assigned to LTA [18]. S. aureus wall teichoic acid (WTA) has a similar structure to LTA. LTA was extracted from SA113 ΔtagO and SA113 Δlgt, which lack WTA and lipoproteins, respectively, and tested in the same manner. LTA extracted from both of these strains inhibited platelets to the same levels as LTA from SA113 (Supplementary Figure 2), confirming LTA platelet inhibitory activity and excluding any effect from lipoproteins or WTA. Figure 2. The inhibition is due to the presence of Staphylococcus aureus LTA. A, Washed human platelets (4 × 108 cells/mL) were preincubated with anti-LTA antibody (1 µg/mL) before incubation for 15 minutes with LTA (2 µg/mL) followed by stimulation with CRP-XL (0.5 µg/mL). Aggregation was measured as change in light transmission for 90 seconds. NB: In Ai, lines representing platelets treated with 2 µg/mL + anti-LTA antibodies and untreated platelets, overlap extensively. B, Washed platelets were incubated for 15 minutes with LTA extracted from S. aureus SA113 (4 µg/mL), Bacillus subtilis, or Streptococcus pnuemoniae D34 (12.5 µg/mL) followed by stimulation with CRP-XL (0.5 µg/mL). C, Washed platelets were incubated for 15 minutes with LTA extracted from B. subtilis at a range of concentrations followed by stimulation with CRP-XL (0.5 µg/mL). D, Washed platelets were incubated for 15 minutes with supernatant from S. aureus SEJ1, SEJ1 ΔgdpP, SEJ1 ΔltaS ΔgdpP or SEJ1 ΔltaS ΔgdpP pCN34-ltaS (10 µg/mL), followed by stimulation with CRP-XL (0.5 µg/mL). (Ai, Bi, Di) Representative aggregation traces of washed platelets. Aggregation was measured for 90 seconds. (Aii, Bii, C, Dii) Data are plotted as percentage inhibition of aggregation represent mean values ± SEM. *P < .05, **P < .001.

ΔltaS ΔgdpP pCN34-ltaS (10 µg/mL), followed by stimulation with CRP-XL (0.5 µg/mL). (Ai, Bi, Di) Representative aggregation traces of washed platelets. Aggregation was measured for 90 seconds. (Aii, Bii, C, Dii) Data are plotted as percentage inhibition of aggregation represent mean values ± SEM. *P < .05, **P < .001. Abbreviations: CRP-XL, cross-linked collagen-related peptide; LTA, lipoteichoic acid; SEM, standard error of the mean. To determine whether this inhibition was restricted to S. aureus LTA, the molecule was extracted and purified from 2 other Gram-positive bacteria and tested for its ability to inhibit platelet aggregation. Bacillus subtilis is a nonpathogenic species that, like S. aureus, produces LTA with a 1,3-linked polyglycerolphosphate chain tethered to the membrane by a diglucosyl-diacylglycerol glycolipid. Glycerolphosphate subunits are esterified with D-alanine [19]. Streptococcus pneumoniae LTA consists of a repeating ribitol-galactose backbone with phosphocholine and D-alanine residues attached [20]. S. pneumoniae LTA showed no inhibition, whereas B. subtilis LTA caused a dose-dependent inhibition up to a maximum of 20% (Figure 2Bi, 2Bii, and 2C); we were unable to solubilize high enough concentrations of B. subtilis LTA to achieve saturation.

ctose backbone with phosphocholine and D-alanine residues attached [20]. S. pneumoniae LTA showed no inhibition, whereas B. subtilis LTA caused a dose-dependent inhibition up to a maximum of 20% (Figure 2Bi, 2Bii, and 2C); we were unable to solubilize high enough concentrations of B. subtilis LTA to achieve saturation. The Supernatant of an S. aureus Mutant Which Lacks LTA Is Unable to Inhibit Platelet Activation S. aureus mutant strains lacking LTA only grow under osmotically stabilizing conditions or by acquiring compensatory mutations [11]. S. aureus SEJ1 (RN4220 spa) was used to construct an LTA-deficient (ΔltaS) strain, but in order for the ΔltaS to be viable, gdpP must also be mutated [11]. Cells of the parental SEJ1 and isogenic strains ΔgdpP, ΔgdpPΔltaS, and ΔgdpPΔltaS containing pCN34-ltaS, a complementation plasmid expressing ltaS from its native promoter, were grown and OD600 of c. 0.5 and LTA in the supernatant was precipitated using (NH4)2SO4. To ensure that consistent amounts of material were used, the amount of protein precipitated along with the LTA was determined by Bradford assay to standardize the preparations. As expected, only supernatant from the ΔgdpPΔltaS lacked LTA upon Western blotting (results not shown). In each experiment, precipitate was used to a final concentration of 10 μg/mL of exoprotein. Supernatant from all strains, except ΔgdpPΔltaS, inhibited platelet activation (Figure 2Di and 2Dii).

ize the preparations. As expected, only supernatant from the ΔgdpPΔltaS lacked LTA upon Western blotting (results not shown). In each experiment, precipitate was used to a final concentration of 10 μg/mL of exoprotein. Supernatant from all strains, except ΔgdpPΔltaS, inhibited platelet activation (Figure 2Di and 2Dii). We were unable to assess the ability of these stains to induce platelet aggregation, as S. aureus RN4220 proved unable to induce aggregation despite incubations of up to 1 hour (results not shown).

ize the preparations. As expected, only supernatant from the ΔgdpPΔltaS lacked LTA upon Western blotting (results not shown). In each experiment, precipitate was used to a final concentration of 10 μg/mL of exoprotein. Supernatant from all strains, except ΔgdpPΔltaS, inhibited platelet activation (Figure 2Di and 2Dii). We were unable to assess the ability of these stains to induce platelet aggregation, as S. aureus RN4220 proved unable to induce aggregation despite incubations of up to 1 hour (results not shown). LTA Produced From S. aureus ΔdltABCD Has a Reduced Ability to Inhibit Platelet Aggregation D-alanine residues are important for various functions of LTA in different biological systems. In S. aureus the addition of D-alanine to the LTA chain is encoded by the dlt operon [8]. LTA was extracted and purified from S. aureus SA113 ΔdltABCD and preincubated with washed human platelets. LTA purified from S. aureus SA113 ΔdltABCD showed a significant reduction (c. 60%) in its ability to inhibit the platelet aggregation (Figure 3Ai), compared to LTA from the parental wild-type strain (Figure 3Aii). Activation of platelets by whole S. aureus ΔdltABCD was indistinguishable from the parent (results not shown); thus it appears that LTA needs to be released from S. aureus to exert its inhibitory effect. Figure 3. LTA from Staphylococcus aureus ΔdltABCD has a reduced ability to inhibit platelet aggregation. A, Washed human platelets (4 × 108 cells/mL) were incubated for 15 minutes with LTA extracted from S. aureus strains SA113 or SA113 ΔdltABCD (4 µg/mL) followed by stimulation with CRP-XL (0.5 µg/mL). B, Platelets in PRP were preloaded with Fluo-4NW dye. Platelets were then pre-incubated with LTA extracted from S. aureus strains SA113 or SA113 ΔdltABCD (4 µg/mL) or Tyrodes buffer for 15 minutes followed by stimulation by CRP-XL (0.5 µg/mL). Intracellular mobilization of calcium was measured by spectrofluorimetry for 120 seconds. Ai, Representative aggregation traces of washed platelets. Aggregation was measured for 90 seconds. Ai, Representative aggregation trace. Aii, Data are plotted as percentage maximum fluorescence (vehicle treated representing 100% aggregation) and represent mean values ± SEM. Bi, Representative calcium flux trace. Bii, Data are plotted as percentage endpoint fluorescence (vehicle treated representing 100% aggregation) and represent mean values ± SEM. *<.01, **P < .005. Abbreviations: CRP-XL, cross-linked collagen-related peptide; LTA, lipoteichoic acid; SEM, standard error of the mean.

SEM. Bi, Representative calcium flux trace. Bii, Data are plotted as percentage endpoint fluorescence (vehicle treated representing 100% aggregation) and represent mean values ± SEM. *<.01, **P < .005. Abbreviations: CRP-XL, cross-linked collagen-related peptide; LTA, lipoteichoic acid; SEM, standard error of the mean. LTA Inhibits [Ca2+] Mobilization During the initial stages of platelet activation, intracellular calcium stores are mobilized to modulate downstream signaling. In order to further characterize the effect of LTA on platelets during the early stages of aggregation, an assay to determine intracellular calcium mobilization and influx was performed. Platelets preincubated with LTA showed a significantly reduced ability to mobilize calcium when challenged with CRP-XL compared to a vehicle treated control (P < .001; Figure 3Bi). Over the course of the assay this equated to a 70% reduction in intracellular calcium levels (Figure 3Bii). Because of its reduced inhibitory potency, calcium release stimulated by LTA from SA113 ΔdltABCD was assayed. From the mean trace of fluorescence over the course of the experiment a repeatable difference was observed (Figure 3Bi). The peak end point fluorescence caused by LTA derived from wild-type and ΔdltABCD strains were significantly different (Figure 3Bii).

alcium release stimulated by LTA from SA113 ΔdltABCD was assayed. From the mean trace of fluorescence over the course of the experiment a repeatable difference was observed (Figure 3Bi). The peak end point fluorescence caused by LTA derived from wild-type and ΔdltABCD strains were significantly different (Figure 3Bii). LTA Inhibition of Platelet Activation Can Be Blocked Using Anti-Paf Receptor Antibodies and Ginkgolide B In different cell types, LTA has previously been reported to bind 4 receptors CD14 [21], CD36 [22], TLR2 [23], and platelet activating factor receptor (PafR) [24]. Monoclonal antibodies with blocking activity for TLR2 and CD14, along with anti-PafR and –CD36 monoclonal antibodies were each tested for their ability to block LTA inhibition (Figure 4Ai). The anti-CD14 and –PafR antibodies were isotype matched. No significant blocking of LTA inhibition occurred with the anti-CD14, -CD36, -TLR2, or mouse IgG (negative control). However anti-PafR abolished LTA-mediated inhibition (Figure 4Ai and 4Aii). Furthermore, ginkgolide B, a specific PafR antagonist [25] blocked LTA-mediated platelet inhibition (Figure 4Bi), reducing it to 0% inhibition (Figure 4Bii). These data demonstrate a role for PafR on platelets as an LTA receptor. Figure 4. LTA acts through PafR to inhibit platelets. A, Washed platelets were incubated with anti-PafR (4 µg/mL), anti-TLR2 (4 µg/mL), anti-CD14 (4 µg/mL), Mouse IgG (4 µg/mL) or Tyrodes buffer for 30 minutes before addition of LTA (4 µg/mL) for 15 minutes. B, Washed platelets were incubated with Ginkgolide B (2 mM) or Tyrodes buffer for 30 minutes before the addition of LTA (4 µg/mL) for 15 minutes. Platelets were then stimulated with CRP-XL (0.5 µg/mL). NB: In Bi, lines representing platelets treated with 4 µg/mL LTA + 2 µM Ginkgolide B and 0 µg mL−1 LTA + 2 µM Ginkgolide B, overlap extensively. A and B, Platelet aggregation was measured as change in light transmission and recorded for 90 seconds. Data are plotted as percentage inhibition of aggregation (normalized so that LTA treatment represents 100% inhibition) and represent mean values ± SEM. *P < .0001, **P < .01. C, PRP (4 × 108 cells/mL) was incubated for 30 minutes with anti-PafR (50 µg/mL), IgG2a (50 µg/mL) or Tyrode buffer before the addition of FITC-LTA at several concentrations for 15 minutes. Samples were run through a BD Accuri C6 flow cytometer and median fluorescence was recorded.

values ± SEM. *P < .0001, **P < .01. C, PRP (4 × 108 cells/mL) was incubated for 30 minutes with anti-PafR (50 µg/mL), IgG2a (50 µg/mL) or Tyrode buffer before the addition of FITC-LTA at several concentrations for 15 minutes. Samples were run through a BD Accuri C6 flow cytometer and median fluorescence was recorded. Data are plotted as percentage median increase in fluorescence when compared to a Tyrode buffer only control and represent mean values ± SEM. Abbreviations: CRP-XL, cross-linked collagen-related peptide; IgG, immunoglobulin G; LTA, lipoteichoic acid; PafR, platelet activating factor receptor; SEM, standard error of the mean. Using flow cytometry, anti-PafR antibody abolished binding of FITC labelled LTA to platelets, demonstrating that PafR is the only receptor for the molecule (Figure 4C). We tested whether anti-PafR antibody blocking was due to a nonspecific interaction with LTA, rather than blocking of PafR. We were unable to deplete samples of their inhibitory effect using anti-PafR antibodies, linked to protein A coated magnetic beads (results not shown). The blocking effect of anti-PafR was not due to cross-reactivity with LTA.

ntibody blocking was due to a nonspecific interaction with LTA, rather than blocking of PafR. We were unable to deplete samples of their inhibitory effect using anti-PafR antibodies, linked to protein A coated magnetic beads (results not shown). The blocking effect of anti-PafR was not due to cross-reactivity with LTA. LTA Causes Raised cAMP Levels to Cause Inhibition of Platelet Activation Our finding that LTA inhibited Ca2+ flux in platelets led us to hypothesize that an increase of cAMP would occur in platelets upon incubation with LTA, which could be blocked with ginkgolide B. Increased cAMP concentrations attenuate Ca2+ mobilization, which is necessary for platelet activation. In platelets, raised levels of cAMP result in increased levels of phosphorylated vasodilator-stimulated phosphoprotein (VASP) [26]. Qualitative assessment of VASP phosphorylation was carried out by Western blot. Samples were prepared that had been pretreated with ginkgolide B, incubated with LTA, and were compared with control samples (Figure 5A). Levels of phosphorylated VASP, shown by the upper band which represents phosphorylation at residue Ser 157 [26], were increased in both the positive control (prostacyclin treated) and LTA-treated samples indicating a possible role for increased platelet cAMP concentrations in LTA-mediated inhibition. To confirm that this inhibition occurred via PafR, 2 samples were treated with ginkgolide B and one of these with LTA also. In both samples no increase in VASP phosphorylation was observed, demonstrating that VASP phosphorylation resulting from LTA treatment can be blocked by the PafR antagonist. Furthermore, supernatant from S. aureus SEJ1 and SEJ1 ΔgdpP were able to cause VASP phosphorylation. Supernatant from S. aureus SEJ1 ΔgdpP ΔltaS was not able to induce phosphorylation, but the effect was restored in supernatant of the complemented strain (Figure 5A). These data provide genetic proof of the role of LTA in VASP phosphorylation. Figure 5. Incubation of platelets with LTA increases cAMP concentrations. A, Platelets (8 × 108 cells/mL) were pretreated with either Ginkgolide B (2 mM) or tyrodes buffer for 30 minutes. Platelets were then treated with LTA from Staphylococcus aureus SA113 (4 µg/mL), PGI2 (0.25 µg/mL), supernatant from S. aureus SEJ1, SEJ1 ΔgdpP, SEJ1 ΔltaS ΔgdpP or SEJ1 ΔltaS ΔgdpP pCN34-ltaS (10 µg/mL), or Tyrodes buffer. Lysates were immunoblotted with an anti-VASP antibody.

(2 mM) or tyrodes buffer for 30 minutes. Platelets were then treated with LTA from Staphylococcus aureus SA113 (4 µg/mL), PGI2 (0.25 µg/mL), supernatant from S. aureus SEJ1, SEJ1 ΔgdpP, SEJ1 ΔltaS ΔgdpP or SEJ1 ΔltaS ΔgdpP pCN34-ltaS (10 µg/mL), or Tyrodes buffer. Lysates were immunoblotted with an anti-VASP antibody. B, Platelets (8 × 108 cells/mL) were pretreated with either Ginkgolide B (2 mM) or tyrodes buffer for 30 minutes. Platelets were then treated with LTA (4 µg/mL), PGI2 (0.25 µg/mL), or tyrodes buffer. Samples were then assayed for cAMP concentration determined by ELISA. Data represent mean values ± SEM. *P < .05. Abbreviations: cAMP, cyclic adenosine monophosphate; ELISA, enzyme-linked immunosorbent assay; LTA, lipoteichoic acid; SEM, standard error of the mean; VASP, vasodilator-stimulated phosphoprotein. The cAMP concentration in platelet lysates was assayed (Figure 5B). A mean c. 350% increase of cAMP concentration occurred in platelets incubated with LTA. In platelets pretreated with ginkgolide B, no increase was observed, correlating with the results from the Western blot for VASP phosphorylation. No increase in VASP phosphorylation or cAMP levels was observed in samples treated solely with ginkgolide B.

e of cAMP concentration occurred in platelets incubated with LTA. In platelets pretreated with ginkgolide B, no increase was observed, correlating with the results from the Western blot for VASP phosphorylation. No increase in VASP phosphorylation or cAMP levels was observed in samples treated solely with ginkgolide B. LTA Reduces Platelet Thrombus Formation In Vitro and Causes Extended Bleeding Time In Vivo In order to investigate whether LTA could inhibit thrombus formation in whole blood and under arterial flow conditions, whole human blood was perfused through collagen coated biochips in the presence of tyrodes buffer and LTA from wild-type S. aureus or from S. aureus SA113 ΔdltABCD (Figure 6A). In blood pretreated with LTA, a significantly reduced thrombus size, compared to the vehicle treated control, was observed (P < .001; Figure 6A and 6B). Additionally peak fluorescence was reduced by approximately 85% (Figure 6C). Blood treated with LTA extracted from SA113 ΔdltABCD showed a slower rate of thrombus formation however at the end of the 10 minutes the thrombi present were no different in size than vehicle treated (P > .05; Figure 6B). The mean peak fluorescence of thrombi formed in the presence of LTA from S. aureus ΔdltABCD showed no difference to that of the control but was significantly different than wild-type LTA treated (P < .05; Figure 6C). Inhibition of thrombus formation in blood was consistent with the reduced platelet function observed in washed platelets (Figure 1Ai–Eii). Figure 6. LTA inhibits thrombus formation in vitro. A, Platelets within whole human blood were labelled with a lipophilic dye DIOC6. Whole blood was then treated with (Ai) tyrodes buffer, (Aii) LTA extracted from Staphylococcus aureus strains SA113 (10 µg/mL) or (Aiii) LTA extracted from S. aureus strain SA113 ΔdltABCD (10 µg/mL) for 15 minutes. Whole blood was then perfused through collagen coated (400 µg/mL) Vena8Biochip at a flow rate of 20 dynes cm−2. Formation of thrombi was recorded using a Z stack capture every 30 seconds for 10 minutes using a Nikon eclipse (TE2000-U) microscope. Thrombus fluorescence intensity was calculated using Slidebook 5 software. B, Data represents mean of thrombi volume over the experiment duration. C, Data represent mean ± SEM of peak fluorescence intensity. *P < .05. **P < .01. Abbreviations: LTA, lipoteichoic acid; SEM, standard error of the mean.

2000-U) microscope. Thrombus fluorescence intensity was calculated using Slidebook 5 software. B, Data represents mean of thrombi volume over the experiment duration. C, Data represent mean ± SEM of peak fluorescence intensity. *P < .05. **P < .01. Abbreviations: LTA, lipoteichoic acid; SEM, standard error of the mean. To substantiate these findings, an in vivo mouse model of bleeding was used. Previous reports have proposed that mouse platelets may lack PafR [27]; however, by Western blotting, we identified a c. 48 kDa band in mouse platelets that comigrates with PafR on human platelets (Supplementary Figure 3A). Similarly, Paf was a weak agonist, compared to collagen, for mouse and human platelet activation (Supplementary Figure 3B) and LTA inhibited platelet activation by collagen in mouse platelets (Supplementary Figure 3C). The effect of LTA on maintenance of hemostasis was measured by a tail bleed assay. Infusion of S. aureus LTA into rodents does not induce shock or affect blood pressure [28]. The mean bleeding time of vehicle-treated (PBS) mice was 340 seconds following tail biopsy (Figure 7). In LTA-treated mice mean time to cessation of bleeding increased significantly (P < .01), more than doubling to 690 seconds. Figure 7. LTA affects hemostasis in vivo. LTA (10 µg/mL) (n = 11) or PBS (n = 11) was administered intravenously to mice and time to cessation of bleeding following a tail biopsy was measured. Data represent individual mice and horizontal lines refer to mean values of seconds until cessation of bleeding. *P < .01. Abbreviations: LTA, lipoteichoic acid; PBS, phosphate-buffered saline.

or PBS (n = 11) was administered intravenously to mice and time to cessation of bleeding following a tail biopsy was measured. Data represent individual mice and horizontal lines refer to mean values of seconds until cessation of bleeding. *P < .01. Abbreviations: LTA, lipoteichoic acid; PBS, phosphate-buffered saline. LTA has multiple roles on many different host cell types, and we cannot rule out an additional role for the endothelium in the tail bleeding experiments. But by measuring thrombus formation in the absence of endothelial cells, we have confirmed a role for LTA alteration of platelet function in thrombus formation. Taken together, these data demonstrate that LTA has a role in the inhibition of thrombosis and hemostasis.

or the endothelium in the tail bleeding experiments. But by measuring thrombus formation in the absence of endothelial cells, we have confirmed a role for LTA alteration of platelet function in thrombus formation. Taken together, these data demonstrate that LTA has a role in the inhibition of thrombosis and hemostasis. DISCUSSION The interplay between S. aureus and its human host is complex. Numerous host and pathogen factors are involved in the interaction, and many have multiple activities. S. aureus LTA is a polymeric glycerol-phosphate molecule that can be fixed to the cell membrane by a lipid anchor and has a well-established role in several host-pathogen interactions. Although attached to the cell envelope, LTA is also released into the S. aureus supernatant, a process accelerated by some antibiotics [29]. In vivo during S. aureus infections, LTA has been detected, albeit by a different method to ours, at up to 10 µg/mL [30]. S. aureus modifies LTA with D-alanine, which confers multiple effects on its function. Interestingly, the presence of D-alanine on S. aureus teichoic acids confers resistance to platelet microbicidal protein, a product of activated platelets [31]. We have shown that D-alanylation increases the platelet-inhibitory potency of LTA. Thus the dltABCD operon plays a dual role in protecting S. aureus against the bactericidal effects of platelet activation.

aureus teichoic acids confers resistance to platelet microbicidal protein, a product of activated platelets [31]. We have shown that D-alanylation increases the platelet-inhibitory potency of LTA. Thus the dltABCD operon plays a dual role in protecting S. aureus against the bactericidal effects of platelet activation. Interestingly, LTA from wild-type S. aureus was more inhibitory than that from the isogenic dltABCD mutant. However, there was no difference in the ability of S. aureus cells to induce platelet activation in the absence of exogenous LTA. Both D-alanine residues and lipid anchor have been reported to be required for stimulation of cytokine production in human whole blood and mouse monocytes [32]. This may explain our observation, as the lipid, which is usually embedded within the bacterial membrane, would only be exposed to host cell receptors upon release of LTA. Alternatively, measurements of the S. aureus cell wall show it to be sufficiently thick that the commonly depicted schematic, in which LTA chains extend through the cell wall and are exposed on the surface, may be incorrect [33].

the bacterial membrane, would only be exposed to host cell receptors upon release of LTA. Alternatively, measurements of the S. aureus cell wall show it to be sufficiently thick that the commonly depicted schematic, in which LTA chains extend through the cell wall and are exposed on the surface, may be incorrect [33]. Many of the ascribed functions of LTA have been determined using commercial preparations [24, 34], which were subsequently shown to contain contaminants responsible for the activity [10, 18, 35]. Thus results obtained using such preparations may remain open to question. We confirmed that properly purified LTA does inhibit platelet activation. Moreover, LTA inhibits activation of platelets by multiple physiological agonists and whole S. aureus cells, each of which trigger activation in different ways, suggesting that LTA blocks a common downstream effect. Ca2+ mobilization, a critical stage in platelet activation, was inhibited and is accompanied by increased cellular cAMP concentrations. LTA interacts with PafR, a cell surface receptor that couples to different G proteins to activate cellular responses that differ between cell types [36–40]. In platelets, the G-proteins that interact with PafR remain to be determined. However, PafR does not signal through the pertussis toxin-sensitive Gi and Go proteins in platelets [41]. It is well documented that platelet GPCRs can influence cellular cAMP concentrations, thereby inhibiting Ca2+ flux [42]. In other cells, PafR interacts with multiple G proteins, resulting in activation of distinct signaling pathways. Indeed, differential PafR signaling in response to agonists and inverse agonists has been reported [43]. Leukocyte responses to Paf utilize pertussis toxin-insensitive and -sensitive G protein(s) [44, 45]. In CHO cells, Paf activation of p38 MAPK occurs through Gq protein, but Paf activation of extracellular signal-regulated kinases 1 and 2 occurs via signaling through Go protein [46]. PafR signaling in HUVECs, including cAMP production by stimulation of adenylate cyclase, occurs via the Gq protein [47]. The exact nature of the signaling events that lead to the increased platelet cAMP levels will be the topic of future pharmacological studies.

ses 1 and 2 occurs via signaling through Go protein [46]. PafR signaling in HUVECs, including cAMP production by stimulation of adenylate cyclase, occurs via the Gq protein [47]. The exact nature of the signaling events that lead to the increased platelet cAMP levels will be the topic of future pharmacological studies. The ability of S. aureus to induce platelet activation is well documented [3] and is presumed to be important in the development of bacterial endocarditis. S. aureus therefore possesses the ability both to positively and to negatively influence thrombus formation. It seems questionable that platelet activation is advantageous for any bacteria. In doing so, a pathogen becomes enmeshed in a thrombus that can lead to the death of the host, leaving the bacterium unable to continue the infectious cycle. Indeed, the ability to inhibit platelet activation presumably confers an advantage to pathogens during infection. Because they are rich sources of bioactive molecules, some of which are bactericidal, platelets have roles in modulating other cellular functions, including those of the innate and acquired immune systems [48] and as such serve to alert the host to the presence of an infection. Activated platelets can engulf S. aureus, although whether this occurs in vivo or has any role in infection remains to be determined [49]. Furthermore, direct interaction with activated platelets induces hyperactivation of neutrophils, enhancing their already potent antibacterial activity [50].

resence of an infection. Activated platelets can engulf S. aureus, although whether this occurs in vivo or has any role in infection remains to be determined [49]. Furthermore, direct interaction with activated platelets induces hyperactivation of neutrophils, enhancing their already potent antibacterial activity [50]. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank John Kokai-Kun, Biosynexus Inc, for providing the monoclonal anti-LTA (pagibaximab) antibodies and Angelika Gründling, Imperial College and Andreas Peschel, University of Tübingen, for strains. Financial support. This work was funded in part by a British Heart Foundation Project Grant (PG/11/65/28969) to S. R. C. A. K. W. was the recipient of a University of Reading PhD studentship. T. S. was funded by a British Heart Foundation Programme Grant (PG/09/011/28094) to J. M. G. Potential conflict of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Approximately 2.6 billion people are at risk of acquiring Plasmodium vivax infection worldwide, of whom half live in Southeast Asia [1]. In contrast with Plasmodium falciparum malaria, P. vivax can cause relapse infections emerging from dormant hypnozoite forms in the liver. Strains in tropical regions such as Sumatera are characterized by frequent (>30%) and early (around 1 month) relapses [2]. Radical cure can only be achieved by adding a hypnozoitocidal drug, and the 8-aminoquinolone primaquine (PQ) is the only widely available drug for this purpose [3]. However, the drug is used infrequently because of concerns about its oxidative side effects causing intravascular hemolysis and methemoglobinemia in populations in whom glucose-6-phosphate dehydrogenase (G6PD) deficiency is common and facilities for assessing G6PD status are not readily available (ie, most malaria-endemic areas). The G6PD gene is located on the X chromosome and there are >180 genetic polymorphisms, most of which confer reductions in G6PD-enzyme activity [4]. The common variants differ importantly in their effect on enzyme activity; hence, the associated risk of hemolysis after PQ treatment varies enormously. The prevalence of G6PD deficiency is approximately 5% in North Sumatra [5], but which variants are prevalent and the risks vs benefits of deploying PQ are not known.

. The common variants differ importantly in their effect on enzyme activity; hence, the associated risk of hemolysis after PQ treatment varies enormously. The prevalence of G6PD deficiency is approximately 5% in North Sumatra [5], but which variants are prevalent and the risks vs benefits of deploying PQ are not known. Plasmodium vivax resistance to chloroquine is prominent in many parts of Indonesia, ranging from 43% in Sumatera island to >80% in Papua [6–8], In 2008, artesunate-amodiaquine (AAQ) and, more recently, dihydroartemisinin-piperaquine (DHP) have replaced chloroquine as first-line treatments [9, 10]. However, it has not been established which of these artemisinin combination therapies (ACTs) is most effective in Sumatera. We compared the efficacy and safety of AAQ + PQ and DHP + PQ for the treatment of uncomplicated vivax malaria in the operationally realistic context without prior testing for G6PD deficiency to identify the optimal treatment of vivax malaria. MATERIALS AND METHODS We performed a prospective, open-label, randomized study comparing AAQ + PQ and DHP + PQ for the treatment of uncomplicated symptomatic P. vivax monoinfection in nonpregnant adults and children aged >1 year presenting at a rural clinic in Tanjung Leidong village, Labuhan Batu, North Sumatera, Indonesia. Routine G6PD testing is not available here. Clinical malaria incidence is 400–500 per year among a population of 32 837 (in 2010), equally divided between P. vivax and P. falciparum infections (written communication, July 2011, from Ministry of Health, Indonesia).

idong village, Labuhan Batu, North Sumatera, Indonesia. Routine G6PD testing is not available here. Clinical malaria incidence is 400–500 per year among a population of 32 837 (in 2010), equally divided between P. vivax and P. falciparum infections (written communication, July 2011, from Ministry of Health, Indonesia). Patients with fever (or recent fever <48 hours) and microscopically confirmed P. vivax monoinfection (≥250/µL) were eligible. Exclusion criteria included any feature of severe malaria [3], severe malnutrition, recurrent vomiting, concomitant infections, pregnancy or lactation, known allergies to the study medication, and inability to follow up. Written informed consent was obtained from patients or their attending relatives before enrollment. The study was approved by the Ethics Committee of the National Institute of Health Research and Development, Indonesian Ministry of Health, Jakarta, Indonesia; Faculty of Tropical Medicine, Mahidol University, Thailand; and the Oxford Tropical Research Ethics Committee, Oxford University, United Kingdom.

d with early NAI 3652 5583 1085 Treated with late NAI 6549 11 993 1226 Untreated with NAI 1738 4818 186 Studies adjusting for potential confounders, No. (%) 8 (18) 11 (21) 0 (0) NOS score, median (range) 6 (4–9) 6 (4–9) 5 (4–8) Abbreviations: NAI, neuraminidase inhibitor; NOS, Newcastle-Ottawa Quality Assessment Scale. a Some studies examined multiple outcomes. b Severe outcome was defined as critical care admission or death. c The breakdown by sex was unknown in a small number of studies (3 each in the mortality and severe outcome analyses). d Overall, 7 studies provided information on combined oseltamivir and peramivir use in 14 patients (see Supplementary Table 2). e Overall, 8 studies reported combined use of NAI and non-NAI (rimantadine, amantadine, or ribavarin) therapy (n = 77 patients; see Supplementary Table 2). f Some studies examined multiple exposure comparisons. Early NAI was defined as treatment beginning within ≤2 d after symptom onset; late NAI, treatment beginning >2 d after symptom onset. g Best estimates of numbers of patients (some publications provided insufficient data on patient numbers by treatment category).

Patients with fever (or recent fever <48 hours) and microscopically confirmed P. vivax monoinfection (≥250/µL) were eligible. Exclusion criteria included any feature of severe malaria [3], severe malnutrition, recurrent vomiting, concomitant infections, pregnancy or lactation, known allergies to the study medication, and inability to follow up. Written informed consent was obtained from patients or their attending relatives before enrollment. The study was approved by the Ethics Committee of the National Institute of Health Research and Development, Indonesian Ministry of Health, Jakarta, Indonesia; Faculty of Tropical Medicine, Mahidol University, Thailand; and the Oxford Tropical Research Ethics Committee, Oxford University, United Kingdom. Parasite density was assessed per 200 white blood cells on a Giemsa-stained thick film, and assumed to be absent if not detected in 200 high-power fields. Gametocytes were counted per 1000 white blood cells. Parasite species was confirmed in thin smear, and 10% of slides were cross-checked at the Faculty of Tropical Medicine, Mahidol University. Other investigations included hemoglobin measurement (Hemocue201+), hemoglobin-methemoglobinemia by pulse oximetry (Masimo-Set, Masimo), and G6PD genotyping from a filter paper blood spot (Whatman 3M). Genotyping by polymerase chain reaction–restriction fragment-length polymorphism (PCR-RFLP) enabled identification of 3 common mutations (Mediterranean, Mahidol, and Viangchan) [11]. In patients developing hemolysis or methemoglobinemia with no mutation by PCR-RFLP, and in patients identified as G6PD deficient by a fluorescent spot test at the end of the study (see below), sequencing of the whole G6PD gene was performed (Macrogen).

of 3 common mutations (Mediterranean, Mahidol, and Viangchan) [11]. In patients developing hemolysis or methemoglobinemia with no mutation by PCR-RFLP, and in patients identified as G6PD deficient by a fluorescent spot test at the end of the study (see below), sequencing of the whole G6PD gene was performed (Macrogen). Patients were not screened for G6PD status before the start of therapy and were managed as outpatients, both current practice in Sumatera. All patients were followed daily for 14 days and then weekly until 42 days, followed by monthly visits up to a year, or in between in case of symptoms. Hemoglobin levels were assessed on days 0, 2, and 7, and then weekly. During PQ therapy, methemoglobinemia was monitored daily. PQ therapy was discontinued in case of macroscopic hemoglobinuria, a drop in hemoglobin >2 g/dL, or when methemoglobin increased to >20% of total hemoglobin. At the end of the study, all patients were invited to test for G6PD status using a NADPH qualitative spot test (SQMMR720 kit, R&D Diagnostics).

s monitored daily. PQ therapy was discontinued in case of macroscopic hemoglobinuria, a drop in hemoglobin >2 g/dL, or when methemoglobin increased to >20% of total hemoglobin. At the end of the study, all patients were invited to test for G6PD status using a NADPH qualitative spot test (SQMMR720 kit, R&D Diagnostics). Patients randomized to AAQ (Arsuamoon, Guilin Pharmaceuticals) received artesunate 12 mg/kg and amodiaquine 30 mg/kg divided over 3 days. Patients randomized to DHP (Arterakine, Pharbaco Central Pharmaceuticals), received dihydroartemisinin 6.75 mg/kg and piperaquine 54 mg/kg in divided doses over 3 days. All patients also received PQ (Phapros Inc) in a dose of 0.25 mg base/kg (or 15 mg for >40 kg) for 14 days started on the first day. All treatment doses were given directly observed and together with some biscuits (ie, cookies). If the patient vomited within 30 minutes, the dose was repeated. Recurrent vivax malaria infections occurring in the first 42 days of follow-up were treated with quinine/doxycycline following Indonesian guidelines; episodes occurring after this point were treated with the same regimen as the initial treatment. All patients were provided with insecticide-treated bednets. Patients were randomized by an independent statistician in blocks of 10, with each treatment allocation concealed in an opaque, sealed envelope, opened only after enrollment.

Patients randomized to AAQ (Arsuamoon, Guilin Pharmaceuticals) received artesunate 12 mg/kg and amodiaquine 30 mg/kg divided over 3 days. Patients randomized to DHP (Arterakine, Pharbaco Central Pharmaceuticals), received dihydroartemisinin 6.75 mg/kg and piperaquine 54 mg/kg in divided doses over 3 days. All patients also received PQ (Phapros Inc) in a dose of 0.25 mg base/kg (or 15 mg for >40 kg) for 14 days started on the first day. All treatment doses were given directly observed and together with some biscuits (ie, cookies). If the patient vomited within 30 minutes, the dose was repeated. Recurrent vivax malaria infections occurring in the first 42 days of follow-up were treated with quinine/doxycycline following Indonesian guidelines; episodes occurring after this point were treated with the same regimen as the initial treatment. All patients were provided with insecticide-treated bednets. Patients were randomized by an independent statistician in blocks of 10, with each treatment allocation concealed in an opaque, sealed envelope, opened only after enrollment. Outcome Patient outcomes, including early treatment failure, late treatment failure, and adequate clinical and parasitological response, were classified according to World Health Organization guidelines [12]. The primary outcome was 42-day efficacy. Secondary outcomes included risk of recurrent P. vivax infection during 1-year follow-up, fever and parasitemia clearance times, gametocyte carriage rates and clearance times, hematological recovery, and safety and tolerability of treatments.

Health Organization guidelines [12]. The primary outcome was 42-day efficacy. Secondary outcomes included risk of recurrent P. vivax infection during 1-year follow-up, fever and parasitemia clearance times, gametocyte carriage rates and clearance times, hematological recovery, and safety and tolerability of treatments. Statistical Analysis Including a 10% anticipated loss, a sample size of 165 patients per study arm was calculated to detect a difference in 42-day cure rate of 90% with AAQ + PQ vs 98% with DHP + PQ with 95% confidence and 80% power. Data were anonymized and double entered into a secured database (OpenClinica). Analysis was done using Stata software (StataCorp). The primary intention-to-treat analysis included all randomized patients and per-protocol analysis of all patients who completed 42 days of follow-up. Comparisons between groups were made by Mann–Whitney U test, Student t test, χ2 test, and Fisher exact test where appropriate. Efficacy at 42 days and after 1 year of follow-up were assessed by Kaplan–Meier survival analysis with log-rank test for statistical significance.

l patients who completed 42 days of follow-up. Comparisons between groups were made by Mann–Whitney U test, Student t test, χ2 test, and Fisher exact test where appropriate. Efficacy at 42 days and after 1 year of follow-up were assessed by Kaplan–Meier survival analysis with log-rank test for statistical significance. RESULTS Between December 2010 and April 2012, 3168 patients were screened, of whom 331 were enrolled in the study. A total of 167 patients were treated with AAQ + PQ and 164 with DHP + PQ (Figure 1). Baseline characteristics were similar between treatment arms (Table 1). Follow-up until day 42 was achieved for 138 of 167 (83%) patients treated with AAQ + PQ and 151 of 164 (91%) with DHP + PQ. One-year follow-up was completed in 130 of 167 (78%) patients treated with AAQ + PQ and 143 of 164 (87%) with DHP + PQ. The median number of missed visits per patient completing 1 year of follow-up was 1 (range, 0–9) for both treatment arms. Table 1. Patient Characteristics at Baseline

d 151 of 164 (91%) with DHP + PQ. One-year follow-up was completed in 130 of 167 (78%) patients treated with AAQ + PQ and 143 of 164 (87%) with DHP + PQ. The median number of missed visits per patient completing 1 year of follow-up was 1 (range, 0–9) for both treatment arms. Table 1. Patient Characteristics at Baseline Characteristic AAQ + PQ (n = 167) DHP + PQ (n = 164) Geometric mean of asexual Plasmodium vivax/µL (95% CI) 1061 (876–1285) 981 (811–1187) Patients with gametocytes on admission 67 (40.1) 74 (45.1) Sex Female 66 (39.5) 79 (48.2) Male 101 (60.5) 85 (51.8) Weight, kg, median (range) 38 (9–99) 37 (10–80) Age, y, median (range) 13 (2–63) 14.5 (2–70) Age group <18 y 106 (64.2) 96 (59.3) ≥18 y 59 (35.8) 66 (40.8) Temperature, mean (SD) 37.7 (1.0) 37.7 (1.0) ≥37.5°C, No. (%) 92 (55.1) 96 (58.5) <37.5°C, No. (%) 75 (44.9) 68 (41.5) Hemoglobin concentration (mean, SD) 12 (1.5) 11.7 (1.4) ≥10 g/dL, No. (%) 151 (90.4) 148 (90.2) <10 g/dL, No. (%) 16 (9.6) 16 (9.8) Methemoglobin concentration, mean (SD) 1.63 (0.82) 1.59 (0.95) Repellent use 38 (29.7) 39 (32.7) Insecticide-treated net use 96 (60.4) 105 (67.7) History of antimalarial use 28 (20) 22 (15.9) Occupation Unemployed 20 (12.1) 19 (11.8) Fisherman 56 (33.9) 52 (32.3) Laborer 27 (16.4) 31 (19.2) Housewife 8 (4.9) 7 (4.3) Businessman 9 (5.4) 6 (3.7) Teacher 4 (2.4) 4 (2.5) Student 26 (15.8) 26 (16.1) Policeman 3 (1.8) 3 (1.8) Farmer 12 (7.3) 13 (8.1) Education Primary 1 (0.8) 3 (2.4) Junior high 70 (53.4) 61 (48.4) Senior high 27 (20.6) 29 (23.0) University 29 (22.1) 23 (18.2) No education 4 (3.1) 10 (7.9) Data are presented as No. (%) unless otherwise indicated.

Teacher 4 (2.4) 4 (2.5) Student 26 (15.8) 26 (16.1) Policeman 3 (1.8) 3 (1.8) Farmer 12 (7.3) 13 (8.1) Education Primary 1 (0.8) 3 (2.4) Junior high 70 (53.4) 61 (48.4) Senior high 27 (20.6) 29 (23.0) University 29 (22.1) 23 (18.2) No education 4 (3.1) 10 (7.9) Data are presented as No. (%) unless otherwise indicated. Abbreviations: AAQ, artesunate-amodiaquine; CI, confidence interval; DHP, dihydroartemisinin-piperaquine; PQ, primaquine. Figure 1. Study flowchart. Abbreviations: P.f., Plasmodium falciparum; SAE, severe adverse event.

Teacher 4 (2.4) 4 (2.5) Student 26 (15.8) 26 (16.1) Policeman 3 (1.8) 3 (1.8) Farmer 12 (7.3) 13 (8.1) Education Primary 1 (0.8) 3 (2.4) Junior high 70 (53.4) 61 (48.4) Senior high 27 (20.6) 29 (23.0) University 29 (22.1) 23 (18.2) No education 4 (3.1) 10 (7.9) Data are presented as No. (%) unless otherwise indicated. Abbreviations: AAQ, artesunate-amodiaquine; CI, confidence interval; DHP, dihydroartemisinin-piperaquine; PQ, primaquine. Figure 1. Study flowchart. Abbreviations: P.f., Plasmodium falciparum; SAE, severe adverse event. Therapeutic Response Intention-to-treat survival analysis showed an adequate parasitological cure rate at 42 days of 91% (95% confidence interval [CI], 86%–95%) with AAQ + PQ and 94% (95% CI, 91%–98%) with DHP + PQ (Figure 2, log-rank P = .51). Per-protocol analysis of patients with complete 42-day follow-up showed cure rates of 100% (95% CI, 98%–100%; 138 of 138 patients) with AAQ + PQ and 99.3% (95% CI, 97%–99.9%; 150 of 151 patients) with DHP + PQ (P = .31). Parasite clearance was within 48 hours in both treatment arms, except for 1 patient with early treatment failure after DHP + PQ (who received rescue treatment) and another 2 patients after DHP + PQ who cleared parasites after >72 hours; neither showed recurrent infection during follow-up. No late treatment failures until day 42 were found in either treatment group. During 1-year follow-up, recurrent infections were observed in 15 of 130 (11.5%) patients after AAQ + PQ (of whom 2 had a second recurrent P. vivax infection) and 13 of 143 (9.1%) after DHP + PQ (Figure 3, log-rank P = .48). The earliest recurrence after treatment with AAQ + PQ was at day 54 compared to 83 days after DHP + PQ. After 1 year, the mean day of recurrence was day 165 (SD, 70) for patients treated with AAQ + PQ and day 203 (SD, 91) for those treated with DHP + PQ (P = .23). Among 28 patients with recurrent infections, 24 had monoinfection with P. vivax, 2 had monoinfection with P. falciparum, and 2 had mixed infection (P. falciparum/P. vivax). Cumulative risk of recurrence for the total group during the 1-year follow-up period was 17.5 per 100 person-years. Figure 2. Kaplan–Meier survival efficacy analysis of all randomized patients. Abbreviations: AAQ + PQ, artesunate-amodiaquine plus primaquine; CI, confidence interval; DHP + PQ, dihydroartemisinin-piperaquine plus primaquine.

currence for the total group during the 1-year follow-up period was 17.5 per 100 person-years. Figure 2. Kaplan–Meier survival efficacy analysis of all randomized patients. Abbreviations: AAQ + PQ, artesunate-amodiaquine plus primaquine; CI, confidence interval; DHP + PQ, dihydroartemisinin-piperaquine plus primaquine. Figure 3. Kaplan–Meier analysis for recurrent infection during the 1-year follow-up period. Abbreviations: AAQ + PQ, artesunate-amodiaquine plus primaquine; DHP + PQ, dihydroartemisinin-piperaquine plus primaquine. On admission, 92 of 167 (55.1%) patients in the AAQ + PQ arm and 96 of 164 (58.5%) in the DHP + PQ arm had fever (≥37.5°C). All patients treated with DHP + PQ cleared their fever within 1 day, compared to 89 of 92 (97%) with AAQ + PQ (P = .07). In patients presenting with gametocytemia, 55 of 67 (82%) of patients treated with AAQ + PQ and 63 of 74 (85%) with DHP + PQ cleared gametocytemia within day 1 (P = .63), and all patients cleared gametocytemia by day 2. At day 42, the mean hemoglobin was 11.9 g/dL (95% CI, 11.8–12.1 g/dL) with DHP + PQ vs 11.9 g/dL (95% CI, 11.7–12.1 g/dL) with AAQ + PQ (P = .91). Hemoglobin levels did not differ between treatment arms at any time point.

PQ cleared gametocytemia within day 1 (P = .63), and all patients cleared gametocytemia by day 2. At day 42, the mean hemoglobin was 11.9 g/dL (95% CI, 11.8–12.1 g/dL) with DHP + PQ vs 11.9 g/dL (95% CI, 11.7–12.1 g/dL) with AAQ + PQ (P = .91). Hemoglobin levels did not differ between treatment arms at any time point. Adverse Events In patients treated with AAQ + PQ, 3 had a drop in hemoglobin level >2 g/dL (to 7.9 g/dL, 12.3 g/dL, and 10.9 g/dL, respectively), of whom 2 developed cola-colored urine temporarily without other complications. One patient had an increased methemoglobin level of 20.3%, after which PQ was discontinued. One patient developed a generalized urticarial rash half an hour after the first dose of AAQ + PQ. This patient recovered after treatment with an antihistamine and was subsequently treated with quinine/doxycycline. In patients treated with DHP + PQ, 1 male and 1 female patient had a drop in hemoglobin level >2 g/dL (to 8.8 g/dL and 7.8 g/dL, respectively) and 2 had increased methemoglobin levels to 20.2% and 21.6% respectively, after which PQ was discontinued. None of the patients with intravascular hemolysis needed blood transfusion, and hemoglobin levels returned to normal (>10 g/dL) after a median of 14 (range, 7–35) days. An increase of >10% in methemoglobin level occurred in 17 of 167 (10.2%) patients treated with AAQ + PQ compared to 24 of 164 (14.6%) treated with DHP + PQ (P = .22).

ients with intravascular hemolysis needed blood transfusion, and hemoglobin levels returned to normal (>10 g/dL) after a median of 14 (range, 7–35) days. An increase of >10% in methemoglobin level occurred in 17 of 167 (10.2%) patients treated with AAQ + PQ compared to 24 of 164 (14.6%) treated with DHP + PQ (P = .22). All 8 patients with PQ-related hemolysis or methemoglobinemia were genotyped. Three male patients with hemolysis were hemizygous for the Mahidol variant of the G6PD gene. One male and 1 female patient with hemolysis had normal results on both PCR-RFLP and complete gene sequencing. The 3 patients with methemoglobinemia also had normal results on PCR-RFLP. Another 52 patients without hemolysis or methemoglobinemia were genotyped. All had the normal reference genotype, except for 1 female patient who was heterozygous for the Mahidol variant. At the end of the study, 212 of 273 (78%) patients were screened for G6PD status by fluorescence spot test. Two males and 5 females (2.6%) were G6PD deficient according to the screening test. The median reduction in hemoglobin levels in these patients was 1.4 g/dL (range, 0.9–2 g/dL). Gene sequencing showed that 1 male patient was hemizygous for the Mahidol variant and another male carried the 1311C→T intron 11 nt93T→C mutation. One of the 5 females was heterozygous for the C 1311 T/C intron 11 nt 93 T/C and intron 2 nt 8 C/A mutations, whereas the other 4 had wild-type genotype (Table 2). Table 2. Summary of G6PD Status Analysis

ale patient was hemizygous for the Mahidol variant and another male carried the 1311C→T intron 11 nt93T→C mutation. One of the 5 females was heterozygous for the C 1311 T/C intron 11 nt 93 T/C and intron 2 nt 8 C/A mutations, whereas the other 4 had wild-type genotype (Table 2). Table 2. Summary of G6PD Status Analysis Patient No. Sex Symptom Hb Drop, g/dL FST Genotyping Sequencing 1 M Dark urine/Hb drop 10.9 to 7.9 − Mahidol − 2 M Dark urine/Hb drop 14.9 to 12.3 + Mahidol − 3 M Hb drop 13.7 to 10.9 − Normal Normal 4 M Hb drop 12.7 to 8.8 − Mahidol − 5 F Hb drop 10.5 to 7.8 − Normal Normal 6 F MetHb rise Normal Normal − 7 F MetHb rise Normal Normal − 8 M MetHb rise Normal Normal − 9 F − + Mahidol (heterozygous) − 10 M − + − Mahidol 11 M − + − 1311 C→T intron 11 nt 93 T→C 12 F − + − Normal 13 F − + − Normal 14 F − + − Normal 15 F − + − Normal 16 F − + − C 1311 T/C intron 11 nt 93 T/C and intron 2 nt 8 C/A (heterozygous) Abbreviations: FST, fluorescent spot test; Hb, hemoglobin; MetHb, methemoglobin. Minor adverse events were more commonly reported in patients receiving AAQ + PQ compared to those receiving DHP + PQ (Table 3). Table 3. Adverse Events Adverse Event AAQ + PQ (n = 167),No. (%) DHP + PQ (n = 164), No. (%) P Value Headache 92 (55.1) 50 (30.5) .001 Dizziness 24 (14.4) 7 (4.4) .002 Vomiting 86 (51.5) 8 (4.9) <.001 Diarrhea 27 (16.2) 8 (4.9) .08 Skin rash 4 (2.4) 1 (0.6) .37 Dyspnea 6 (3.6) 0 (0.0) .03 Abdominal pain 46 (27.5) 14 (8.5) .001 Hemolysis 3 (1.8) 2 (1.2) >.50 Abbreviations: AAQ, artesunate-amodiaquine; DHP, dihydroartemisinin-piperaquine; PQ, primaquine.

izziness 24 (14.4) 7 (4.4) .002 Vomiting 86 (51.5) 8 (4.9) <.001 Diarrhea 27 (16.2) 8 (4.9) .08 Skin rash 4 (2.4) 1 (0.6) .37 Dyspnea 6 (3.6) 0 (0.0) .03 Abdominal pain 46 (27.5) 14 (8.5) .001 Hemolysis 3 (1.8) 2 (1.2) >.50 Abbreviations: AAQ, artesunate-amodiaquine; DHP, dihydroartemisinin-piperaquine; PQ, primaquine. Three patients had a severe adverse event during the first year of follow-up, none of which seemed to be related to the study drugs or malaria infection. One patient developed pericarditis 10 days after treatment with DHP + PQ. The malaria slide was negative at the time of this event. Primaquine was discontinued, and the patient made a full recovery. Two patients treated with AAQ + PQ died during the 1-year follow-up period, unrelated to malaria or study drugs. A 50-year-old diabetic male patient died 9 months after treatment after an acute myocardial infarction. A 50-year-old man died 7 months after treatment; his cause of death was unknown but followed hemoptysis in the days prior to death.

, their demographic characteristics, and both their and their children's water contact behavior and history of schistosomiasis treatment. The current study draws on baseline data and sera collected in April 2009 from 426 children of 213 mothers living in the villages of Bugoigo and Piida, Bulissa District, Lake Albert. SmTAL1 (Sm22.6; XP_002575844) and SmTAL2 (Sm21.7; XP_002569898) were prepared as previously described [5]. Serum from blood samples obtained by finger prick was stored at −80°C until required. Levels of IgE and IgG4 to SmTAL1 and SmTAL2 were measured using biotinylated isotype-specific monoclonal antibodies, as described elsewhere [4]. Sample sera and plasma from noninfected European controls were assayed in duplicate at concentrations of 1:20 (IgE) and 1:200 (IgG4). A 3-fold serial dilution of purified human IgG4 (Sigma-Aldrich, United States) or IgE myeloma (Calbiochem, Germany) was added to each plate, forming a 14-point standard curve, starting at 30 µg/mL. Plates were read at dual wavelengths (490 and 630 nm) on a Powerwave HT microplate reader (BioTek Instruments). Results were interpolated from standard curves with a 5 parameter curve fit, using Gen5 analysis software (BioTek Instruments).

Q + PQ died during the 1-year follow-up period, unrelated to malaria or study drugs. A 50-year-old diabetic male patient died 9 months after treatment after an acute myocardial infarction. A 50-year-old man died 7 months after treatment; his cause of death was unknown but followed hemoptysis in the days prior to death. DISCUSSION The recent guideline of the Indonesian Ministry of Health for treatment of uncomplicated vivax malaria includes 2 first-line ACTs, AAQ and DHP [10]. We compared the efficacy and safety of these combinations in radical treatment regimens with PQ in the normal context of use (ie, without G6PD testing). In the setting of North Sumatera, both treatment regimens were safe and efficacious for cure of the blood-stage infection. Hemolysis after treatment with PQ (0.25 mg/kg for 14 days), not requiring transfusion, was a rare event. This was because the prevalence of G6PD deficiency was relatively low (<5%) by comparison with other areas of the tropics, and the prevalent genotypes were not associated with severe deficiency. A study from Thailand found a similar low risk for hemolysis after treatment with PQ in the same dosing scheme, without prior G6PD testing [13]. The Mahidol variant (487G→A) is also the most common G6PD variant in the western part of Thailand.

DISCUSSION The recent guideline of the Indonesian Ministry of Health for treatment of uncomplicated vivax malaria includes 2 first-line ACTs, AAQ and DHP [10]. We compared the efficacy and safety of these combinations in radical treatment regimens with PQ in the normal context of use (ie, without G6PD testing). In the setting of North Sumatera, both treatment regimens were safe and efficacious for cure of the blood-stage infection. Hemolysis after treatment with PQ (0.25 mg/kg for 14 days), not requiring transfusion, was a rare event. This was because the prevalence of G6PD deficiency was relatively low (<5%) by comparison with other areas of the tropics, and the prevalent genotypes were not associated with severe deficiency. A study from Thailand found a similar low risk for hemolysis after treatment with PQ in the same dosing scheme, without prior G6PD testing [13]. The Mahidol variant (487G→A) is also the most common G6PD variant in the western part of Thailand. We screened patients for G6PD deficiency at the end of follow-up with a fluorescent spot test. This identified another 7 patients who were G6PD deficient according to this test, of whom 1 male was hemizygous for the Mahidol variant and another male showed the relatively common 1311C→T intron 11 nt93T→C mutation, both associated with mild G6PD deficiency [14, 15]. In total, 3.3% of patients had a variant G6PD genotype, which compares to an earlier study in North Sumatera showing a 5% prevalence of G6PD deficiency [5]; the slightly lower prevalence in vivax patients in the current study might relate to the protective effect of G6PD deficiency against malaria [16–18]. A total of 4 of 9 (44%) patients with a positive fluorescent screening test denoting G6PD deficiency had a normal G6PD genotype, indicating suboptimal specificity of the test, which could be related to the presence of additional sources of oxidative stress (eg, deriving from food or drugs) not accounted for in the test. Only 5 of 331 (1.5%) patients developed significant intravascular hemolysis (>2 g/dL hemoglobin drop), none of whom required a blood transfusion. Another 3 of 331 (0.9%) had methemoglobin levels >20% related to PQ treatment, without any other clinical signs. Most (7 of 8 [87.5%]) adverse events occurred within the first 7 days of treatment and all quickly resolved. Our findings suggest that both regimens including low-dose PQ can be deployed safely in this setting of low prevalence and “mild-type” G6PD deficiency, provided that the risks are acknowledged and that adequate follow-up can be assured. It should be noted that PQ is contraindicated during pregnancy. Implementation of G6PD testing should be a priority in P. vivax endemic settings, but where this is currently not feasible, a suggested follow-up scheme is a daily visit during the first 7 days of treatment with hematocrit or hemoglobin levels measured at diagnosis and 3 and 7 days after start of treatment. If hemoglobinuria occurs, then PQ should be stopped. Simple color cards to aid detection of hemoglobinuria may be useful.

not feasible, a suggested follow-up scheme is a daily visit during the first 7 days of treatment with hematocrit or hemoglobin levels measured at diagnosis and 3 and 7 days after start of treatment. If hemoglobinuria occurs, then PQ should be stopped. Simple color cards to aid detection of hemoglobinuria may be useful. Both treatments resulted in a rapid clinical and parasitological cure, fast gametocyte clearance, and good therapeutic efficacy at 42 days. Only 1 patient treated with DHP + PQ had early treatment failure. In vivax malaria, genotyping cannot distinguish between relapse and reinfection, as more than half of the relapse infections in endemic areas are caused by reactivation of liver schizonts with a different genotype [19]. Because the natural history of relapse infections in North Sumatera is not known and this study did not include a control arm without PQ administration, we cannot assess with certainty the efficacy of this low-dose PQ regimen for preventing relapse infection. In our study, 28 of 289 (9.7%) patients had recurrent infections after 1 year of follow-up. In comparison, in patients returning from highly endemic Papua Indonesia to nonendemic Java, relapse rates were comparable, with 2 of 36 (6%) relapses after treatment with DHP + PQ combined with a higher dose (30 mg) of PQ [20]. However, hypnozoite sensitivity may vary geographically. In our study, the ratio between P. falciparum and P. vivax infections was 6.5:1 during screening and 2:1 during follow-up, suggesting that a proportion of the late recurrent infections were relapse infections. Efficacy trials of ACT regimens with and without PQ are now being planned and implemented throughout Asia to assess the dose-dependent relapse-preventing efficacy of PQ in the treatment of vivax malaria.

ing and 2:1 during follow-up, suggesting that a proportion of the late recurrent infections were relapse infections. Efficacy trials of ACT regimens with and without PQ are now being planned and implemented throughout Asia to assess the dose-dependent relapse-preventing efficacy of PQ in the treatment of vivax malaria. Both relapse and recurrent infections are suppressed by the posttreatment prophylactic effect of the long half-life partner drug in the ACT used for treatment. The terminal half-life of the active metabolite of amodiaquine, desethylamodiaquine, is approximately 21 days [21], compared to 28–35 days for piperaquine [22]. In our study the earliest recurrence with AAQ + PQ was indeed earlier (at 54 days) than with DHP + PQ (at 83 days), but with longer follow-up this advantage disappeared. After 1 year, the time to recurrent infection was no longer statistically different between treatment groups. Both regimens used in this study were well tolerated, although DHP + PQ was associated with significantly fewer (mild) adverse events than AAQ + PQ, as has also been reported in other studies [23, 24]. In addition to its longer posttreatment prophylactic effect, this makes DHP + PQ an attractive alternative to AAQ + PQ for the treatment of uncomplicated vivax malaria, and could be a further step to harmonization of the treatment of falciparum and vivax malaria in Indonesia.

so been reported in other studies [23, 24]. In addition to its longer posttreatment prophylactic effect, this makes DHP + PQ an attractive alternative to AAQ + PQ for the treatment of uncomplicated vivax malaria, and could be a further step to harmonization of the treatment of falciparum and vivax malaria in Indonesia. This study has several limitations: 12% of patients were lost for follow-up at day 42, related to poor accessibility of some areas in rural northern Sumatera, and 22% were not tested for G6PD status at the end of the study, so our prevalence estimate may be imprecise. Patients with hemolysis were not formally assessed for changes in renal function, but no patient reported anuria or developed symptoms of renal failure during follow-up. The number of G6PD-deficient patients in the current study was low, and because enzyme activity can vary considerably even within specific genotypes, assessment of the hemolysis risk after low-dose PQ within specific genotypes requires larger studies. Further prevalence studies on the genetic variants of G6PD and their corresponding phenotypes in various parts of Indonesia will be required to generalize our current findings to other parts of Indonesia. In conclusion, radical treatment with AAQ or DHP, both combined with low-dose PQ (0.25 mg/kg for 14 days), without prior testing for G6PD deficiency proved a safe and efficacious treatment for uncomplicated P. vivax in North Sumatera. DHP + PQ was better tolerated and had a longer posttherapeutic prophylactic effect.

The number of G6PD-deficient patients in the current study was low, and because enzyme activity can vary considerably even within specific genotypes, assessment of the hemolysis risk after low-dose PQ within specific genotypes requires larger studies. Further prevalence studies on the genetic variants of G6PD and their corresponding phenotypes in various parts of Indonesia will be required to generalize our current findings to other parts of Indonesia. In conclusion, radical treatment with AAQ or DHP, both combined with low-dose PQ (0.25 mg/kg for 14 days), without prior testing for G6PD deficiency proved a safe and efficacious treatment for uncomplicated P. vivax in North Sumatera. DHP + PQ was better tolerated and had a longer posttherapeutic prophylactic effect. Notes Acknowledgments. We thank all our staff members in the field, and the patients and their family members who participated in this study. Financial support. This work was supported by University of Sumatera Utara, the Indonesian Ministry of Health, and the Directorate General of Higher Education. Additional support was provided by the Lee Foundation, Singapore, the Wellcome Trust of Great Britain, and the Office of the Higher Education Commission and Mahidol University under the National Research Universities Initiative. Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

(See the editorial commentary by Aoki and Hayden, on pages 547–9.) The neuraminidase inhibitors (NAIs), oseltamivir and zanamivir are licensed for the treatment of influenza A and B. Before the 2009–2010 pandemic, evidence from randomized trials suggested modest reductions in time to alleviation of symptoms and symptom severity [1–4] and possibly a reduction in antibiotic use for secondary complications [5–7]. Further evidence from methodologically weaker observational studies, derived mainly from prepandemic data (seasonal influenza), suggests that oral oseltamivir reduces mortality by about 75%, hospitalization by 25% and symptom duration compared with no treatment, with broadly similar findings for zanamivir, based on fewer studies [8]. Despite limited usage since launch, except in Japan, both drugs, especially oseltamivir, were widely stockpiled for pandemic purposes and subsequently deployed during the influenza A(H1N1)pdm09 pandemic. A subsequent analysis of oseltamivir safety data published by F. Hoffman–La Roche estimated that 18.3 million individuals worldwide received the drug during the pandemic period between 1 May and 31 December 2009 [9], and data from the United States shows that 97.5% of prescriptions for NAIs during the pandemic period were for oseltamivir [10].

oseltamivir safety data published by F. Hoffman–La Roche estimated that 18.3 million individuals worldwide received the drug during the pandemic period between 1 May and 31 December 2009 [9], and data from the United States shows that 97.5% of prescriptions for NAIs during the pandemic period were for oseltamivir [10]. Published studies from the recent pandemic period suggest that early (≤48 hours after symptom onset) versus “late” (delayed >48 hours after symptom onset) treatment of healthy and at-risk adults reduced the likelihood of hospitalization or requirement for critical care [11–15]. Similarly, a small number of studies suggest that increased in-hospital mortality might be related to the late initiation of NAI therapy [16–19]. However, many studies are too small to produce conclusive individual findings; some adjust for possible confounders, but most do not. Considerable uncertainty remains among public health policy-makers and governments regarding the public health benefits of NAI usage during the 2009–2010 pandemic. We therefore present a systematic review and meta-analysis of studies specifically from that period, assessing the impact of NAI treatment in hospitalized patients on mortality, requirement for critical care, and influenza-related pneumonia.

garding the public health benefits of NAI usage during the 2009–2010 pandemic. We therefore present a systematic review and meta-analysis of studies specifically from that period, assessing the impact of NAI treatment in hospitalized patients on mortality, requirement for critical care, and influenza-related pneumonia. METHODS Eligibility Criteria and Assessment Types of Studies We included all comparative epidemiological studies (case series, case-control, and cohort studies) and randomized controlled trials conducted during the time period between 1 March 2009 (Mexico), or 1 April 2009 (rest of the world) until the WHO declaration of the end of the pandemic (10 August, 2010); assessing the association between NAI treatment and clinical outcomes. Studies with <10 participants were excluded. Types of Participants Subjects of all ages hospitalized with a clinical or laboratory diagnosis of A(H1N1)pdm09. Types of Interventions Treatment with an NAI (oseltamivir, zanamivir and peramivir [20]) administered via any route for A(H1N1)pdm09. Articles reporting combined results with other influenza virus types, subtypes, and strains were excluded. Types of Outcome Measures Mortality, admission to critical care, and influenza-related pneumonia.

Types of Interventions Treatment with an NAI (oseltamivir, zanamivir and peramivir [20]) administered via any route for A(H1N1)pdm09. Articles reporting combined results with other influenza virus types, subtypes, and strains were excluded. Types of Outcome Measures Mortality, admission to critical care, and influenza-related pneumonia. Search Strategy We searched Medline, EmBase, CINAHL, CAB Abstracts, ISI Web of Science, PubMed UK, PubMed central, Scopus, WHO regional indexes, LILAC, and J-STAGE (to 19 April 2012), imposing no language restrictions. Further studies were also identified from scanning reference lists of identified studies and through contact with subject area experts (via J. S. N. V. T.). We used Boolean logic and core search terms relating to pandemic influenza (including influenza A virus OR H1N1 subtype OR swine origin influenza AH1N1 virus) AND exposure of interest that is, antiviral drugs (including neuraminidase inhibitors OR oseltamivir OR zanamivir OR peramivir) AND clinical outcome measures (including pneumonia, or critical care/intensive care, or mortality). Our detailed search strategy is shown in Supplementary Table 1.

rigin influenza AH1N1 virus) AND exposure of interest that is, antiviral drugs (including neuraminidase inhibitors OR oseltamivir OR zanamivir OR peramivir) AND clinical outcome measures (including pneumonia, or critical care/intensive care, or mortality). Our detailed search strategy is shown in Supplementary Table 1. Screening, Data Extraction, and Quality Assessment Titles, abstracts, and full texts of identified studies were screened independently by 2 reviewers (S. G. M., S. V.) with differences being resolved through discussion with a third reviewer (P. R. M.). Data from included studies were independently extracted by 2 investigators (S. G. M. and S. V.) using a previously piloted data extraction form, and scored for methodological quality using the Newcastle-Ottawa Quality Assessment Scale (NOS) [21]. This scale awards a maximum score of 9 points to each included study based on representativeness of the cohort, adjustment for confounders and assessment of the outcome/exposure. Where relevant and possible, supplementary data were sought from corresponding authors of included studies. Differences in quality assessment were resolved by referral to a third investigator (P. R. M.).

study based on representativeness of the cohort, adjustment for confounders and assessment of the outcome/exposure. Where relevant and possible, supplementary data were sought from corresponding authors of included studies. Differences in quality assessment were resolved by referral to a third investigator (P. R. M.). Data Analysis Results from individual studies were extracted directly as odds ratios (ORs), with 95% confidence intervals (CIs; or standard errors), or as tabulated data, from which ORs were estimated based on adjustment for the greatest number of covariates possible in each analysis. The data were pooled using random effects meta-analysis. Separate analyses were performed for the following 3 treatment exposures: NAI treatment (irrespective of timing) versus none; early NAI treatment (≤48 hours after symptom onset) versus late treatment (delayed >48 hours after symptom onset); and early NAI treatment versus no treatment. Heterogeneity between studies was assessed using the I2 statistic [22]; when at least moderate (I2 > 50%), subgroup analyses were conducted to explore the effects of age; ascertainment of A(H1N1)pdm09 diagnosis; special health states (eg, pregnancy, intensive care unit admission, pneumonia); and study quality (Newcastle-Ottawa Quality Assessment Scale >6 vs ≤6). Publication bias was determined using funnel plots and Egger's tests [23]; all analyses were conducted using Stata v11.2 software (StataCorp).

A(H1N1)pdm09 diagnosis; special health states (eg, pregnancy, intensive care unit admission, pneumonia); and study quality (Newcastle-Ottawa Quality Assessment Scale >6 vs ≤6). Publication bias was determined using funnel plots and Egger's tests [23]; all analyses were conducted using Stata v11.2 software (StataCorp). Protocol and Registration We adhered to the recommendations for Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [24], and the protocol is published in the National Institute for Health Research international prospective register of systematic reviews (PROSPERO) [25]. RESULTS Study Selection and Characteristics Of 8783 records identified from electronic searches, 1495 titles were judged potentially relevant, and their abstracts screened for relevance, yielding 259 full-text records. After these were assessed, 107 articles were eligible (Figure 1). A full reference list of eligible articles is provided in Supplementary Table 2. Of the 107 articles, 53 assessed mortality, 59 assessed severe outcome (defined as critical care admission or death) and 14 assessed A(H1N1)pdm09-related pneumonia (Supplementary Table 2). Seventeen articles could not be included in the meta-analyses because they were partially or completely included as part of a national surveillance dataset or larger study earmarked for inclusion within the overall meta-analysis (Figure 1); reasons for exclusions are provided in Supplementary Table 3. Figure 1. Summary of article selection process. Abbreviation: NAI, neuraminidase inhibitor.

were partially or completely included as part of a national surveillance dataset or larger study earmarked for inclusion within the overall meta-analysis (Figure 1); reasons for exclusions are provided in Supplementary Table 3. Figure 1. Summary of article selection process. Abbreviation: NAI, neuraminidase inhibitor. Characteristics of the 90 studies eligible for meta-analyses are summarized in Table 1. Eighty (89%) reported exclusively laboratory-confirmed diagnoses, positive by A(H1N1)pdm09-specific polymerase chain reaction (PCR) or positive by PCR for influenza A but nontypeable for human subtypes H1 (seasonal); 8 (9%) studied hospitalized patients with confirmed, probable, or suspected A(H1N1)pdm09 infection. Two studies reported A(H1N1)pdm09 cases but did not specify methods of diagnosis. Table 1. Summary of 90 Studies Included in Meta-Analysis, by Outcome Measure

influenza A but nontypeable for human subtypes H1 (seasonal); 8 (9%) studied hospitalized patients with confirmed, probable, or suspected A(H1N1)pdm09 infection. Two studies reported A(H1N1)pdm09 cases but did not specify methods of diagnosis. Table 1. Summary of 90 Studies Included in Meta-Analysis, by Outcome Measure Outcome Measurea Mortality Severe Outcomeb Pneumonia Studies, No. 44 52 13 Total sample size, No. of patients 23 723 31 428 3271 Male patients, No.c 11 558 13 608 1602 Age range, y <1 to 93 <1 to 91 <1 to 93 Population groups, No. of studies Mixed age groups 20 21 5 Adults 8 8 3 Children 5 10 2 Pregnant women 4 4 3 Other 7 9 … Regions, No. of studies North America 9 18 2 Latin America 9 4 2 Europe and Australia/New Zealand 10 18 2 Asia-Pacific 14 12 7 Others 2 … … A(H1N1)pdm09 diagnosis, No. of studies Laboratory confirmed 37 49 11 Laboratory confirmed or clinically diagnosed cases 6 3 1 Confirmed cases but method of confirmation not stated 1 … 1 A(H1N1)pdm09 diagnosis, No. of patients Laboratory confirmed 15 998 29 574 3059 Laboratory confirmed or clinically diagnosed cases 7707 1854 31 Confirmed cases but method of confirmation not stated 18 … 181 Antiviral agents used, No. of studiesd Oseltamivir only 23 24 9 NAI only 8 10 4 NAI and non-NAI antivirale 5 5 … NAI drug name not specified 8 13 … Exposure comparison, No. of studiesf Any NAI vs none 23 29 6 Early NAI vs late NAI 27 30 11 Early NAI vs none 9 13 4 Early NAI vs late NAI or no treatment … 2 … Preadmission NAI vs no preadmission NAI 2 3 … Patients, No.g Treated with any NAI 14 920 25 246 2964 Treated with early NAI 3652 5583 1085 Treated with late NAI 6549 11 993 1226 Untreated with NAI 1738 4818 186 Studies adjusting for potential confounders, No. (%) 8 (18) 11 (21) 0 (0) NOS score, median (range) 6 (4–9) 6 (4–9) 5 (4–8) Abbreviations: NAI, neuraminidase inhibitor; NOS, Newcastle-Ottawa Quality Assessment Scale.

e Overall, 8 studies reported combined use of NAI and non-NAI (rimantadine, amantadine, or ribavarin) therapy (n = 77 patients; see Supplementary Table 2). f Some studies examined multiple exposure comparisons. Early NAI was defined as treatment beginning within ≤2 d after symptom onset; late NAI, treatment beginning >2 d after symptom onset. g Best estimates of numbers of patients (some publications provided insufficient data on patient numbers by treatment category). Forty-five studies (50%) in the meta-analyses reported treatment with oseltamivir only, 21 (23%) reported treatment with NAIs (oseltamivir, zanamivir, and/or peramivir), 8 (9%) reported monotherapy with NAI and non-NAI antiviral drugs (amantadine, rimantadine, ribavirin), and in 16 studies (18%) the name of NAI drug was not specified. Overall, 34 895 patients were treated with an NAI, of whom 14 (0.0004%), across 7 studies, were treated with peramivir either alone or as dual therapy with oseltamivir. Seventy-seven patients (0.002%) across 8 studies also received combined therapy using NAI and non-NAI antiviral drugs (typically NAI plus either adamantane or ribavirin). Because we did not have access to individual-level raw data, it was not possible to exclude such patients without sacrificing eligible whole studies.

Seventy-seven patients (0.002%) across 8 studies also received combined therapy using NAI and non-NAI antiviral drugs (typically NAI plus either adamantane or ribavirin). Because we did not have access to individual-level raw data, it was not possible to exclude such patients without sacrificing eligible whole studies. Meta-Analysis Findings Mortality Fifty-three studies presented data on the association between NAI treatment and mortality. Nine studies were unsuitable for meta-analyses and were excluded (Supplementary Tables 2 and 3). Analyses of the remaining 44 are summarized in Figure 2. The pooled analysis of 20 studies comparing NAI treatment (at any time) versus none revealed a nonsignificant reduction in risk of mortality (OR, 0.72 [95% CI, .51–1.01]), with moderate statistical heterogeneity (I2, 49%) and no evidence of publication bias (Egger's test, P = .894). Moreover, meta-analysis of 2 studies examining preadmission NAI treatment versus no preadmission NAI in subsequently hospitalized patients did not find a statistically significant reduction in mortality (OR, 0.59 [95% CI, .21–1.71]) (Table 2). Table 2. Summary of Results (Random Effects Model) Including Subgroup Analyses for Mortality, Severe Outcome, and A(H1N1)pdm09-Related Pneumonia Hospitalized Patients Studies Included in Analysis, No.

Meta-Analysis Findings Mortality Fifty-three studies presented data on the association between NAI treatment and mortality. Nine studies were unsuitable for meta-analyses and were excluded (Supplementary Tables 2 and 3). Analyses of the remaining 44 are summarized in Figure 2. The pooled analysis of 20 studies comparing NAI treatment (at any time) versus none revealed a nonsignificant reduction in risk of mortality (OR, 0.72 [95% CI, .51–1.01]), with moderate statistical heterogeneity (I2, 49%) and no evidence of publication bias (Egger's test, P = .894). Moreover, meta-analysis of 2 studies examining preadmission NAI treatment versus no preadmission NAI in subsequently hospitalized patients did not find a statistically significant reduction in mortality (OR, 0.59 [95% CI, .21–1.71]) (Table 2). Table 2. Summary of Results (Random Effects Model) Including Subgroup Analyses for Mortality, Severe Outcome, and A(H1N1)pdm09-Related Pneumonia Hospitalized Patients Studies Included in Analysis, No. Pooled OR (95% CI) I2, % Referencesa Mortality (Died vs Survived) NAI vs no NAI treatment (overall) 20 .72 (.51–1.01) 49 [3–5, 16, 17, 26, 38, 41, 43, 62, 63, 69, 78, 83, 85, 92, 94, 97, 99, 104] Unadjusted studies 18 .73 (.53–1.00) 44 [4, 5, 16, 17, 26, 38, 41, 43, 62, 63, 69, 78, 83, 85, 92, 97, 99, 104] Adjusted studies 2 1.22 (.01–172.42) 85 [3, 94] A(H1N1)pdm09 diagnosis Laboratory confirmed cases 16 .77 (.54–1.08) 48 [5, 17, 26, 38, 41, 43, 62, 63, 69, 78, 83, 85, 92, 97, 99, 104] Laboratory confirmed or   clinically diagnosed 4 .50 (.14– 1.78) 54 [3, 4, 16, 94] Mixed age groups 12 .75 (.50–1.13) 61 [3–5, 26, 38, 62, 69, 85, 88, 92, 94, 104] Adults 7 .43 (.20–.97) 63 [16, 37, 43, 63, 85, 96, 104] Children 6 .72 (.36–1.44) 12 [17, 41, 85, 97, 99, 104] Pregnant women 1 .34 (.14–.81) … [78] Patients with pneumonia 5 .74 (.13–4.28) 70 [43, 83, 88, 97, 104] ICU patients 8 .61 (.41–.90) 5 [3, 16, 17, 41, 63, 78, 92, 99] Others … … … [63] Preadmission NAI treatment vs no preadmission NAI treatment 2 .59 (.21–1.71) 0 [26, 93] Early treatment vs late treatment (overall) 25 .38 (.27–.53) 52 [2, 8, 13, 21, 26, 31, 32, 36, 37, 42, 49, 50, 52, 58, 65, 69, 72, 78, 84, 90, 92, 101, 102, 104, 107] Unadjusted studies 23 .35 (.24–.51) 53 [2, 8, 13, 21, 26, 31, 32, 36, 37, 42, 49, 50, 52, 58, 69, 78, 84, 90, 92, 101, 102, 104, 107] Adjusted studies 2 .61 (.31–1.19) 26 [65, 72] A(H1N1)pdm09 diagnosis Laboratory confirmed cases 23 .37 (.26–.52) 53 [2, 8, 13, 21, 26, 31, 32, 37, 42, 49, 50, 52, 58, 65, 69, 72, 78, 84, 90, 92, 102, 104, 107] Laboratory confirmed or   clinically diagnosed cases 2 .33 (.03–3.73) 61 [36, 101] Mixed age groups 14 .51 (.36–.72) 50 [13, 26, 32, 36, 37, 65, 69, 72, 84, 85, 90, 92, 102, 104] Adults 10 .41 (.28–.59) 0 [8, 21, 37, 42, 49, 58, 82, 85, 104, 107] Children 4 .37 (.20–.68) 0 [50, 52, 85, 104] Pregnant women 4 .09 (.04–.21) 0 [31, 78, 98, 101] ICU patients 9 .33 (.17–.64) 59 [8, 21, 36, 42, 52, 78, 92, 98, 102] Patients with pneumonia 1 .53 (.19–1.5) … [104] Others … … … [20, 32] Early treatment vs no treatment  (overall) 9 .35 (.18–.71) 77 [26, 31, 37, 65, 69, 78, 85, 92, 104] Mixed age groups 6 .43 (.23–.80) 69 [26, 65, 69, 85, 92, 104] Adult

78, 98, 101] ICU patients 9 .33 (.17–.64) 59 [8, 21, 36, 42, 52, 78, 92, 98, 102] Patients with pneumonia 1 .53 (.19–1.5) … [104] Others … … … [20, 32] Early treatment vs no treatment  (overall) 9 .35 (.18–.71) 77 [26, 31, 37, 65, 69, 78, 85, 92, 104] Mixed age groups 6 .43 (.23–.80) 69 [26, 65, 69, 85, 92, 104] Adult s 5 .22 (.07–.66) 73 [31, 37, 78, 85, 104] Children 2 .12 (.02–.76) 0 [85, 104] Pregnant women 2 .07 (.02–.20) 0 [31, 78] ICU patients 2 .28 (.02–3.88) 94 [78, 92] Severe Outcome (Required Critical Care or Died vs Hospitalized and Survived) NAI vs no NAI treatment (overall) 23 1.76 (1.22–2.54) 86 [2, 5, 6, 9, 10, 18, 22, 24, 26, 27, 30, 33, 35, 45, 55, 61, 62, 64, 70, 77, 86, 95, 100] Unadjusted studies 23 1.76 (1.22–2.54) 86 [2, 5, 6, 9, 10, 18, 22, 24, 26, 27, 30, 33, 35, 45, 55, 61, 62, 64, 70, 77, 86, 95, 100] Mixed age groups 13 1.68 (1.05–2.70) 89 [2, 5, 6, 10, 18, 19, 26, 30, 55, 62, 64, 70, 77] Adults 5 1.26 (.64–2.46) 60 [19, 45, 61, 67, 104] Children 12 2.97 (1.81–4.89) 38 [9, 19, 22, 24, 27, 33, 53, 71, 74, 86, 95, 104] Pregnant women 2 2.41 (1.71–3.39) 0 [14, 70] Other … … … [95, 100] Preadmission NAI treatment  (before hospital admission) 3 .51 (.29–.89) 0 [26, 28, 93] Early treatment vs late  treatment (overall) 24 .41 (.30–.56) 82 [5, 8, 12, 13, 15, 18, 22, 24, 26, 30–32, 40, 44, 49, 58, 61, 64, 65, 70, 72, 73, 101, 105] Unadjusted studies 19 .45 (.31–.66) 82 [5, 8, 13, 15, 18, 22, 24, 26, 32, 40, 44, 49, 58, 61, 64, 70, 73, 101, 105] Adjusted studies 5 .33 (.19–.55) 77 [12, 30, 64, 65, 72] Mixed age groups 11 .44 (.31–.62) 86 [5, 12, 13, 18, 26, 30, 44, 64, 65, 70, 72] Adults 8 .63 (.38–1.02) 69 [8, 40, 47, 49, 58, 61, 66, 104] Children 6 1.01 (.58–1.96) 57 [22, 24, 59, 71, 73, 104] Pregnant women 4 .16 (.04–.60) 90 [15, 31, 70, 101] Patients with pneumonia 3 .51 (.11–2.36) 91 [12, 66, 104] Other (ARDS, diabetics,  cancer, HIV) … … … [32, 100, 105, 106] Early treatment vs no treatment  (overall) 11 .94 (.50–1.76) 93 [5, 18, 24, 26, 30, 31, 61, 64, 65, 70, 73] Mixed age groups 5 1.58 (.70–3.58) 96 [18, 26, 30, 70, 104] Adults 2 1.01 (.18–5.74) 83 [61, 104] Children 3 5.77 (.83–40.29) 79 [24, 71, 104] Pregnant women 2 .59 (.07–5.19) 95 [31, 70] Patients with pneumonia 1 3.77 (1.78–7.96) … [104] Diabetics 1 .18 (.05–.61) … [100] Early treatment vs late treatment or no NAI 2 .27 (.04–2.00) 23 [38, 39] Pneumonia vs No Pneumonia NAI treatment vs none (overall) 6 2.29 (1.16–4.53) 26 [1, 33, 57, 60, 81, 108] Unadjusted studies 6 2.29 (1.16–4.53) 26 [1, 3

5 [31, 70] Patients with pneumonia 1 3.77 (1.78–7.96) … [104] Diabetics 1 .18 (.05–.61) … [100] Early treatment vs late treatment or no NAI 2 .27 (.04–2.00) 23 [38, 39] Pneumonia vs No Pneumonia NAI treatment vs none (overall) 6 2.29 (1.16–4.53) 26 [1, 33, 57, 60, 81, 108] Unadjusted studies 6 2.29 (1.16–4.53) 26 [1, 3 3, 57, 60, 81, 108] A(H1N1)pdm09 diagnosis Laboratory confirmation 5 2.83 (1.65–4.85) 0 [1, 33, 57, 81, 108] Not specified 1 .26 (.02–3.04) … [60] Pneumonia confirmation Chest radiographs 4 2.72 (1.54–4.82) 3 [1, 33, 81, 108] Not specified 2 1.17 (.07–20.09) 66 [57, 60] Mixed age groups 3 2.01 (.57–7.11) 21 [1, 57, 108] Adults 2 1.10 (.12–10.16) 66 [60, 81] Children 1 3.53 (1.63–7.66) … [33] Pregnant women 1 .26 (.26–3.04) … [60] Early treatment vs late treatment (overall) 11 .35 (.24–.50) 50 [1, 11, 49, 51, 57, 60, 76, 79, 81, 101, 103] Unadjusted studies 10 .37 (.23–.58) 55 [1, 11, 49, 51, 57, 60, 76, 79, 101, 103] Adjusted studies 1 .29 (.19–.45) … [81] A(H1N1)pdm09 diagnosis Laboratory confirmation 9 .37 (.25–.55) 57 [1, 11, 49, 51, 57, 76, 79, 81, 103] Laboratory and/or clinical   confirmation 1 .19 (.02–1.78) … [101] Not specified 1 .12 (.02–.66) … [60] Pneumonia confirmation Chest radiographs 8 .36 (.24–.53) 58 [1, 49, 51, 76, 79, 81, 101, 103] Not specified 3 .24 (.06–1.05) 38 [11, 57, 60] Mixed age groups 3 .35 (.11–1.08) 76 [1, 57, 76] Adults 7 .35 (.25–.47) 14 [11, 49, 51, 60, 81, 101, 103] Children 1 .81 (.25–2.63) … [79] Pregnant women 3 .31 (.04–.45) 0 [11, 60, 101] ICU patients 2 .05 (.01–.20) 0 [11, 76] Early treatment vs no treatment (overall) 4 .73 (.27–2.02) 48 [1, 57, 60, 81] Mixed age groups 2 1.00 (.11–9.05) 69 [1, 57] Adults 2 .62 (.11–3.89) 54 [60, 81] Children 0 … … … Pregnant women 1 .18 (.02–2.02) … [60] Early treatment vs late treatment or no NAI 1 6.67 (2.61–17.06) … [40] Abbreviations: ARDS, acute respiratory distress syndrome; CI, confidence interval; HIV, human immunodeficiency virus; ICU, intensive care unit; NAI, neuraminidase inhibitor; OR, odds ratio.

81] Children 0 … … … Pregnant women 1 .18 (.02–2.02) … [60] Early treatment vs late treatment or no NAI 1 6.67 (2.61–17.06) … [40] Abbreviations: ARDS, acute respiratory distress syndrome; CI, confidence interval; HIV, human immunodeficiency virus; ICU, intensive care unit; NAI, neuraminidase inhibitor; OR, odds ratio. aSee reference list in Supplementary Table 2. Figure 2. Summary of pooled analyses from studies examining mortality. Abbreviations: CI, confidence interval; NAI, neuraminidase inhibitor; OR, odds ratio. Separate meta-analyses showed that early NAI treatment versus late (25 studies) was associated with a significant reduction in mortality (OR, 0.37 [95% CI, .27–.53]; I2, 52%), although there was evidence of asymmetry in tests for publication bias (Egger's test, P = .004). Pooled analyses for early NAI therapy compared with no treatment (9 studies) also found a significant reduction in mortality (OR, 0.35 [95% CI, .18–.71]; I2, 77%; Egger's test, P = .142). The high level of heterogeneity in this meta-analysis was partly attributable to the heterogeneous populations. Our subgroup analysis based on subpopulations found no evidence of heterogeneity for studies in children or pregnant women but high heterogeneity in intensive care unit–based studies (Table 2).

st, P = .142). The high level of heterogeneity in this meta-analysis was partly attributable to the heterogeneous populations. Our subgroup analysis based on subpopulations found no evidence of heterogeneity for studies in children or pregnant women but high heterogeneity in intensive care unit–based studies (Table 2). Severe Outcome (Critical Care Admission or Death) Using a composite variable for “severe outcome” based on receiving critical care or death, 59 studies reported this outcome of which 52 were suitable for inclusion in meta-analyses; these are summarized in Figure 3 and Table 2. For NAI treatment (at any time) versus none (23 studies), a statistically significant increase in severe outcomes with NAI therapy was observed (OR, 1.76 [95% CI, 1.22–2.54]; I2, 86%; Egger's test, P = .036). We pooled 3 studies providing data on preadmission NAI use in hospitalized patients and found a statistically significant reduction in severe outcomes compared with no preadmission NAI (OR, 0.51 [95% CI, .29–.89]; I2, 0%; Egger's test, P = .46). Early NAI treatment compared with late (24 studies) also significantly reduced the likelihood of severe outcome (OR, 0.41 [95% CI, .30–.56]; I2, 82%; Egger's test, P = .016); however, early NAI treatment versus none (11 studies) revealed no statistically significant decrease in the likelihood of severe outcome (OR, 0.94 [95% CI, .50–1.76]; I2, 93%; Egger's test, P = .023). Two studies that assessed early NAI treatment versus late or none (combined) also revealed no significant reduction in severe outcomes (OR, 0.27 [95% CI, .04–2.00]; I2, 23%; Egger's test, not calculable; Table 2). Findings from all of these analyses were subject to high levels of heterogeneity (I2 > 75%) which were neither explained by subgroup analyses (Table 2) nor attributable to methodological quality (data not shown). Figure 3. Summary of pooled analyses from studies examining severe outcome. Abbreviations: CI, confidence interval; NAI, neuraminidase inhibitor; OR, odds ratio.

gh levels of heterogeneity (I2 > 75%) which were neither explained by subgroup analyses (Table 2) nor attributable to methodological quality (data not shown). Figure 3. Summary of pooled analyses from studies examining severe outcome. Abbreviations: CI, confidence interval; NAI, neuraminidase inhibitor; OR, odds ratio. Pneumonia Associated With A(H1N1)pdm09 Infection Fourteen studies reported data on hospitalized patients with A(H1N1)pdm09 infection and documented the presence or absence of pneumonia . Most reported radiographic pneumonia, whereas 3 did not provide information on ascertainment; the latter were still included in the meta-analysis but apportioned lower scores during quality assessment (Table 2). The meta-analysis based on 13 articles is summarized in Figure 4. The pooled analysis comparing NAI treatment (at any time) versus none (6 studies) revealed a significantly increased likelihood of pneumonia associated with NAI treatment (OR, 2.29 [95% CI, 1.16–4.53]; I2, 26%; Egger's test, P = .282). However, early versus late treatment (11 studies) significantly reduced the likelihood of pneumonia (OR, 0.35 [95% CI, .24–.50]; I2, 50%; Egger's test, P = .646). A comparison between early treatment and none (4 studies) revealed no statistically significant decrease in the likelihood of pneumonia (OR, 0.73 [95% CI, .27–2.02]; I2, 48%; Egger's test, P = .826). Figure 4. Summary of pooled analyses from studies examining A(H1N1)pdm09-associated pneumonia. Abbreviations: CI, confidence interval; NAI, neuraminidase inhibitor; OR, odds ratio.

vealed no statistically significant decrease in the likelihood of pneumonia (OR, 0.73 [95% CI, .27–2.02]; I2, 48%; Egger's test, P = .826). Figure 4. Summary of pooled analyses from studies examining A(H1N1)pdm09-associated pneumonia. Abbreviations: CI, confidence interval; NAI, neuraminidase inhibitor; OR, odds ratio. One pneumonia study (reference 40 in Supplementary Table 2) was unsuitable for inclusion in any of the pooled analyses because treatment exposure was measured as early versus late or none (combined). This study showed early oseltamivir treatment to be associated with a significantly increased likelihood of pneumonia (unadjusted OR, 6.67 [95% CI, 2.61–17.06]; P < .001).

Table 2) was unsuitable for inclusion in any of the pooled analyses because treatment exposure was measured as early versus late or none (combined). This study showed early oseltamivir treatment to be associated with a significantly increased likelihood of pneumonia (unadjusted OR, 6.67 [95% CI, 2.61–17.06]; P < .001). DISCUSSION Mortality Overall, our meta-analyses suggest that NAI treatment of A(H1N1)pdm09 in hospitalized cases reduced mortality. Although comparison of treatment (at any time) with none revealed a 28% nonsignificant reduction in mortality, when comparing early with late treatment we observed a significant 63% reduction in mortality, albeit with significant publication bias. Finally, we noted a significant 65% reduction in mortality when comparing early treatment with none, along with high levels of heterogeneity. This suggests that early initiation of treatment following symptom onset is key for reducing mortality. We did not detect a significant reduction in mortality associated with preadmission NAI treatment in subsequently hospitalized patients; very few studies were available to address this question, and the absence of data from cases that remained in the community does not allow us to draw conclusions about whether community NAI treatment prevented hospital admission.

ortality associated with preadmission NAI treatment in subsequently hospitalized patients; very few studies were available to address this question, and the absence of data from cases that remained in the community does not allow us to draw conclusions about whether community NAI treatment prevented hospital admission. Severe Outcome Alongside mortality, critical care admission due to influenza is an undesirable outcome of public health importance, worth preventing. Many studies described “severe outcome” using a common definition of critical care admission or mortality, reflecting the occurrence of severe but sometimes survivable A(H1N1)pdm09 infection. It should however be appreciated that some patients with severe disease might have failed to access critical care because of limited availability, which may have introduced bias. Notwithstanding, we observed that NAI treatment (at any time) was associated with a 76% significant increase in the likelihood of severe outcome compared with none. In contrast, a 59% significant reduction in the likelihood of severe outcome was seen for early versus late NAI treatment, but no significant reduction for early NAI treatment versus none. Our data also suggest that preadmission NAIs in patients subsequently hospitalized significantly reduced the likelihood of severe outcome by 49%, albeit based on only 3 studies.

e likelihood of severe outcome was seen for early versus late NAI treatment, but no significant reduction for early NAI treatment versus none. Our data also suggest that preadmission NAIs in patients subsequently hospitalized significantly reduced the likelihood of severe outcome by 49%, albeit based on only 3 studies. Pneumonia Our findings on pneumonia may have been influenced by differential ascertainment and classification of pneumonia. We therefore gave a lower quality score to studies in which information pneumonia ascertainment was not available and performed a subgroup analysis to take this into account (Table 2). We found the likelihood of pneumonia to be significantly increased by 129% for the comparison of any NAI treatment with none, whereas early versus late NAI treatment significantly reduced the likelihood of pneumonia by 65%; we did not find a statistically significant reduction when comparing early treatment versus none.

e found the likelihood of pneumonia to be significantly increased by 129% for the comparison of any NAI treatment with none, whereas early versus late NAI treatment significantly reduced the likelihood of pneumonia by 65%; we did not find a statistically significant reduction when comparing early treatment versus none. Interpretation Our findings are consistent with earlier data on seasonal influenza, showing that the magnitude of symptomatic benefit due to oseltamivir treatment is increased by early instigation of therapy [8, 26]. We believe the 3 different comparisons in our analyses (treatment at any time vs none, early vs late, and early vs none) help reveal confounding related to treatment propensity but at the same time offer important clinical coherence. We hypothesize that patients with mild illness, more likely to survive and less likely to develop pneumonia, were also less likely to be offered antiviral treatment in most settings during the 2009 pandemic, because of either physician preference or patient care-seeking behavior. Furthermore, we surmise that access to rapid diagnostic testing was variable across settings and that A(H1N1)pdm09 was either not suspected and/or not confirmed in many patients until late in their illness (or late in their admission), by which time either they were recovering or their condition had deteriorated. This may explain the apparent increase in severe outcomes associated with NAI use at any time. It is most likely that those with mild illness who were recovering were left untreated with NAIs and that those with initially mild but later severe illness were treated late as a final attempt at disease reversal. Indeed, unpublished data from the UK FLU-CIN study [27] reveal that among patients with a length of stay ≤4 days (as a proxy for mild to moderate disease) the proportions of patients receiving early, late, or no NAI treatment were 36%, 27%, and 37% respectively, compared with 22%, 41%, and 36% respectively in patients with length of stay >4 days (χ2 trend, P = .008; data available on request [P. R. M. and J. S. N. V. T., unpublished data].) Thus, comparisons of early treatment versus late may have overestimated treatment effectiveness, whereas comparisons of treatment versus none and early treatment versus none may have underestimated it.

ength of stay >4 days (χ2 trend, P = .008; data available on request [P. R. M. and J. S. N. V. T., unpublished data].) Thus, comparisons of early treatment versus late may have overestimated treatment effectiveness, whereas comparisons of treatment versus none and early treatment versus none may have underestimated it. In that context, our findings on mortality (early treatment vs none and any treatment vs none), suggest potentially important public health effects because untreated patients were likely to have had milder disease, and our finding of an association between NAI treatment and increased severe outcome seems explainable.

ength of stay >4 days (χ2 trend, P = .008; data available on request [P. R. M. and J. S. N. V. T., unpublished data].) Thus, comparisons of early treatment versus late may have overestimated treatment effectiveness, whereas comparisons of treatment versus none and early treatment versus none may have underestimated it. In that context, our findings on mortality (early treatment vs none and any treatment vs none), suggest potentially important public health effects because untreated patients were likely to have had milder disease, and our finding of an association between NAI treatment and increased severe outcome seems explainable. Limitations We observed a high degree of heterogeneity among studies examining severe outcome, and although we performed subgroup analyses and stratified by methodological quality, this finding remained largely unexplained. For some of the outcomes we found evidence of publication bias, which may have overestimated the observed pooled effect. All of the studies included in the systematic review were observational designs. This is, in itself, a limitation that cannot be overcome, but it can be argued that such observational data provide a more realistic estimate of the field effectiveness of NAIs in a pandemic situation. Most studies did not provide adjusted risk estimates, but even when these were available there were differences in the extent to which adjustment had been made for potential confounding. Another limitation is the inability to adjust for propensity to treatment. In the absence of random allocation to antivirals, one of the inherent biases in observational studies is the likelihood of receiving treatment. Some of the studies included in the meta-analysis are from low-resource countries and it is likely that treatment was given preferentially to more severely ill patients, thereby underestimating the effectiveness of antiviral therapy in reducing severe outcomes. Finally, a very small proportion of patients received intravenous peramivir (alone or as dual therapy) or dual therapy with oseltamivir and zanamivir. Such patients were well dispersed between studies, and excluding them would have sacrificed too much data. However, because they account for such a small proportion of cases overall, we do not believe they have introduced meaningful bias into the results.

as dual therapy) or dual therapy with oseltamivir and zanamivir. Such patients were well dispersed between studies, and excluding them would have sacrificed too much data. However, because they account for such a small proportion of cases overall, we do not believe they have introduced meaningful bias into the results. The question of whether NAI treatment has an impact on patient outcomes in a pandemic situation can only ever be answered by using observational data because of the ethical implications of randomization to treatment during a public health emergency. The logical next step is to conduct an individual patient level meta-analysis based on obtaining raw data from observational studies around the world and reanalyzing pooled data [28]. This approach will allow more complete adjustment for confounders, such as comorbid conditions, disease severity, concomitant therapies, propensity for NAI treatment, and the assessment of different NAI treatment regimens.

obtaining raw data from observational studies around the world and reanalyzing pooled data [28]. This approach will allow more complete adjustment for confounders, such as comorbid conditions, disease severity, concomitant therapies, propensity for NAI treatment, and the assessment of different NAI treatment regimens. In conclusion, this systematic review and meta-analysis is to our knowledge the first to examine the effectiveness of NAI treatment solely during the 2009–2010 pandemic, measured against clinical outcomes of likely importance to public health policy-makers. The findings suggest that mortality was reduced among hospitalized patients through early NAI treatment, although the magnitude of benefit offered by early versus late treatment may have been overestimated by treatment propensity. Nevertheless, our finding of a 65% mortality reduction in early treated versus untreated patients suggests a meaningful public health benefit, of relevance to pandemic policy-makers, because it is more likely that untreated cases were less severe rather than more severe and the true effect may therefore have been underestimated. If this is so, pandemic preparedness policies need to emphasize not only the issue of appropriate NAI stockpiling but also practical mechanisms for ensuring easy and early access to treatment during a pandemic.

ed cases were less severe rather than more severe and the true effect may therefore have been underestimated. If this is so, pandemic preparedness policies need to emphasize not only the issue of appropriate NAI stockpiling but also practical mechanisms for ensuring easy and early access to treatment during a pandemic. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank colleagues in the Division of Epidemiology and Public Health, especially Charles Beck, for advice and comments on the manuscript. Author contributions. Review and concept design, J. S. N. V. T., J. L. B., P. R. M., S. G. M.; critical appraisal and acquisition of data, S. G. M., S. V., P. R. M.; analysis and interpretation of data, S. G. M., S. V., P. R. M., J. L. B., J. S. N. V. T.; manuscript preparation and contribution of intellectual content, S. G. M., S. V., P. R. M., J. L. B., J. S. N. V. T.; final manuscript approval, S. G. M., S. V., P. R. M., J. S. N. V. T., J. L. B.

data, S. G. M., S. V., P. R. M.; analysis and interpretation of data, S. G. M., S. V., P. R. M., J. L. B., J. S. N. V. T.; manuscript preparation and contribution of intellectual content, S. G. M., S. V., P. R. M., J. L. B., J. S. N. V. T.; final manuscript approval, S. G. M., S. V., P. R. M., J. S. N. V. T., J. L. B. Disclaimers. The study was undertaken fully independently of the funder, who had no involvement in the design, data collection, analysis and interpretation of data or preparation of the manuscript. The funder has no rights to access the data. Financial support. This work was funded via an unrestricted grant from F. Hoffmann–La Roche. Details of the contract may be examined freely at http://www.nottingham.ac.uk/chs/research/projects/pride/index.aspx.

Disclaimers. The study was undertaken fully independently of the funder, who had no involvement in the design, data collection, analysis and interpretation of data or preparation of the manuscript. The funder has no rights to access the data. Financial support. This work was funded via an unrestricted grant from F. Hoffmann–La Roche. Details of the contract may be examined freely at http://www.nottingham.ac.uk/chs/research/projects/pride/index.aspx. Potential conflicts of interest. The University of Nottingham Health Protection and Influenza Research Group is currently in receipt of research funds from GlaxoSmithKline and an unrestricted grant from Astra Zeneca; this funding received from GlaxoSmithKline and Astra Zeneca did not support any aspect of this study. J. S. N. V. T. has received funding to attend influenza-related meetings and lecture and consultancy fees from several influenza antiviral drug and vaccine manufacturers, including GlaxoSmithKline and F. Hoffmann–La Roche; all forms of personal remuneration from such activities ceased in September 2010. He is a former employee of SmithKline Beecham (now GlaxoSmithKline), Roche Products, and Aventis-Pasteur (now Sanofi-Pasteur), all before 2005; he has no outstanding pecuniary interests in any of these companies through shareholdings, share options, or accrued pension rights. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Malaria caused an estimated 216 million cases and approximately 1 million deaths in 2010 [1], mainly in sub-Saharan Africa where most cases are caused by Plasmodium falciparum. Development of vaccines and new drugs and better understanding of immunological processes are essential to tackling this immense problem. Controlled human malaria infection (CHMI), in which healthy volunteers are exposed to bites of P. falciparum–infected mosquitoes, is a powerful tool to address questions regarding P. falciparum drug and vaccine efficacy, clinical signs and symptoms, parasite kinetics, and human immunology. Since the first CHMI by mosquitoes fed on cultures of P. falciparum, >1300 healthy volunteers have been exposed to CHMI with mainly the Nijmegen falciparum strain NF54 or its clone 3D7 [2]. Strain/parasite line NF54 stably produces sexual stages required for production of infectious mosquitoes. Parasites have been adapted to laboratory conditions by continuous in vitro culture for >3 decades. In the field, P. falciparum displays a wide genetic diversity, which is currently not represented by the available laboratory strains for CHMI. Other strains, including the South American 7G8 P. falciparum clone of the Brazilian strain IMTM22, have been sporadically used in limited number of volunteers [3–5]. We therefore aimed to identify, clone, and test an additional P. falciparum strain that can be used in CHMIs, and we developed several qualification criteria: The clone (a) must consistently produce gametocytes and sporozoites, (b) should be cloned to create a single genetically homogenous parasite population, (c) should be sensitive to commonly administered antimalarials, and (d) should be of non-African origin to be geographically and genetically distinct from the NF54 strain, an airport strain that probably originates from Africa [6]. Here we report the generation, characterization, and first CHMI for NF135.C10, a new Cambodian clone; findings include drug sensitivity, microsatellite profile, kinetics of parasitemia, and clinical and immunological properties in a direct comparison with NF54.

n airport strain that probably originates from Africa [6]. Here we report the generation, characterization, and first CHMI for NF135.C10, a new Cambodian clone; findings include drug sensitivity, microsatellite profile, kinetics of parasitemia, and clinical and immunological properties in a direct comparison with NF54. METHODS Blood collected from patients for diagnosis of malaria was cultured in Roswell Park Memorial Institute 1640 medium containing 10% human serum at 5% hematocrit in a semiautomated suspension culture system, cloned by limiting dilution, and fed to Anopheles stephensi mosquitoes, reared according to standard operating procedures, as described elsewhere [7]. Salivary glands of 10 mosquitoes were dissected for each strain to confirm the presence of sporozoites. The identities of NF135.C10 and NF54 were defined by polymerase chain reaction (PCR), microsatellite mapping of the P. falciparum rifin repetitive microsatellite (pfRRM), and drug sensitivity assay. (For descriptions of the respective techniques, please refer to the Supplementary Methods.) Volunteers, aged 18–35 years, were screened at the Leiden University Medical Centre for eligibility based on medical and family history, physical examination, and general hematological and biochemical tests. All volunteers gave written informed consent before inclusion.

METHODS Blood collected from patients for diagnosis of malaria was cultured in Roswell Park Memorial Institute 1640 medium containing 10% human serum at 5% hematocrit in a semiautomated suspension culture system, cloned by limiting dilution, and fed to Anopheles stephensi mosquitoes, reared according to standard operating procedures, as described elsewhere [7]. Salivary glands of 10 mosquitoes were dissected for each strain to confirm the presence of sporozoites. The identities of NF135.C10 and NF54 were defined by polymerase chain reaction (PCR), microsatellite mapping of the P. falciparum rifin repetitive microsatellite (pfRRM), and drug sensitivity assay. (For descriptions of the respective techniques, please refer to the Supplementary Methods.) Volunteers, aged 18–35 years, were screened at the Leiden University Medical Centre for eligibility based on medical and family history, physical examination, and general hematological and biochemical tests. All volunteers gave written informed consent before inclusion. Ten Dutch malaria-naive volunteers were randomized into 2 groups and exposed to bites of 5 A. stephensi mosquitoes infected with either NF54 or NF135.C10 for 10 minutes. Feeding sessions were repeated until each volunteer had been bitten by exactly 5 P. falciparum–infected mosquitoes.

Volunteers, aged 18–35 years, were screened at the Leiden University Medical Centre for eligibility based on medical and family history, physical examination, and general hematological and biochemical tests. All volunteers gave written informed consent before inclusion. Ten Dutch malaria-naive volunteers were randomized into 2 groups and exposed to bites of 5 A. stephensi mosquitoes infected with either NF54 or NF135.C10 for 10 minutes. Feeding sessions were repeated until each volunteer had been bitten by exactly 5 P. falciparum–infected mosquitoes. Starting on day 5 after infection, volunteers were subjected to intensive outpatient follow-up with up to thrice-daily visits. Signs and symptoms (solicited and unsolicited) were recorded and graded by the attending physician as follows: mild (easily tolerated), moderate (interferes with normal activity), or severe (prevents normal activity) or, for fever, grade 1 (>37.5°C to 38.0°C), 2 (>38.0°C to 39.0°C), or 3 (>39.0°C). Hematological and biochemical parameters were monitored daily. After identification of a positive blood smear, or if smears remained negative until day 21, volunteers were treated with a curative regimen of atovaquone and proguanil (1000 and 400 mg/d, respectively) for 3 days. The trial was performed in accordance with good clinical practice and approved by the Central Committee for Research Involving Human Subjects of The Netherlands (CCMO NL30350.058.09).

until day 21, volunteers were treated with a curative regimen of atovaquone and proguanil (1000 and 400 mg/d, respectively) for 3 days. The trial was performed in accordance with good clinical practice and approved by the Central Committee for Research Involving Human Subjects of The Netherlands (CCMO NL30350.058.09). Thick blood smears were examined by microscopy twice daily on days 5 and 6 after challenge, thrice daily on days 7–11, twice daily on days 12–15, and once daily on days 16–21. For each smear, 15 µL of ethylenediaminetetraacetic acid–anticoagulated blood was stained with Giemsa for 30 minutes and examined at ×1000 magnification, with assessment of approximately 0.5 µL of blood. A smear was considered positive if 2 unambiguously identifiable parasites were found. The prepatent period was defined as the period between exposure to infected mosquitoes and the first positive blood smear. Parasitemia was also measured retrospectively with real-time quantitative PCR (qPCR), using a technique described elsewhere [8], with minor changes (the MGB probe AAC AAT TGG AGG GCA AG was used instead of the turbo TaqMan probe sequence).

riod between exposure to infected mosquitoes and the first positive blood smear. Parasitemia was also measured retrospectively with real-time quantitative PCR (qPCR), using a technique described elsewhere [8], with minor changes (the MGB probe AAC AAT TGG AGG GCA AG was used instead of the turbo TaqMan probe sequence). In vitro immunological assays were performed on peripheral blood mononuclear cells isolated from venous whole blood on the day before challenge, on days 5, 35, and 140 after challenge, and on the first day of treatment. Cells were stored in liquid nitrogen and, after thawing, cultured in the presence of NF135.C10 or NF54 P. falciparum red blood cells (RBCs) at a 1:2 ratio (peripheral blood mononuclear cells to P. falciparum RBCs) for 24 hours. Flow cytometric staining was performed for CD4, CD45RO, CD3, CD62L, CD8a, γδ T-cell receptor, CD56, interferon γ (IFN-γ), tumor necrosis factor, and interleukin 2. A more detailed description can be found in the Supplementary Methods. Data were analyzed using GraphPad Prism5 software (GraphPad). Differences in parasite kinetics between subjects in the NF135.C10 and NF54 groups were analyzed using the nonparametric Mann–Whitney U test. Differences were considered statistically significant at P < .05 (2 sided).

In vitro immunological assays were performed on peripheral blood mononuclear cells isolated from venous whole blood on the day before challenge, on days 5, 35, and 140 after challenge, and on the first day of treatment. Cells were stored in liquid nitrogen and, after thawing, cultured in the presence of NF135.C10 or NF54 P. falciparum red blood cells (RBCs) at a 1:2 ratio (peripheral blood mononuclear cells to P. falciparum RBCs) for 24 hours. Flow cytometric staining was performed for CD4, CD45RO, CD3, CD62L, CD8a, γδ T-cell receptor, CD56, interferon γ (IFN-γ), tumor necrosis factor, and interleukin 2. A more detailed description can be found in the Supplementary Methods. Data were analyzed using GraphPad Prism5 software (GraphPad). Differences in parasite kinetics between subjects in the NF135.C10 and NF54 groups were analyzed using the nonparametric Mann–Whitney U test. Differences were considered statistically significant at P < .05 (2 sided). RESULTS Plasmodium falciparum strains obtained from 74 patients with malaria were adapted to culture; 21 strains produced gametocytes, and 16 were able to infect mosquitoes. Based on gametocyte production, exflagellation, and transmission to mosquitoes, 7 strains were cloned. Two of these clones produced at least 5 oocysts and 30 000 sporozoites in >70% of mosquitoes. The drug sensitivity profile of NF135.C10 is similar to that of NF54 for atovaquone, proguanil, dihydroartemisinin, and lumefantrine, but NF135.C10 is >8-fold less sensitive to chloroquine than NF54. The culture characteristics and drug sensitivity of NF135.C10 and NF54 are shown in Table 1. Comparison of NF135.C10 and NF54 genotypes using PCR and rifin microsatellite mapping showed distinct genetic profiles (Supplementary Figure 1). Table 1. NF135.C10 and NF54 Culture Characteristics

tive to chloroquine than NF54. The culture characteristics and drug sensitivity of NF135.C10 and NF54 are shown in Table 1. Comparison of NF135.C10 and NF54 genotypes using PCR and rifin microsatellite mapping showed distinct genetic profiles (Supplementary Figure 1). Table 1. NF135.C10 and NF54 Culture Characteristics NF135.C10 NF54 Restarted cultures until CHMI 7 306 Country of origin Cambodia West Africa (airport) Year of isolation 1993 1979 Period 2009–2010a Infection, % 74 (62–87) 86 (78–94) Oocysts 12 (7.3–16) 27 (22–33) Sporozoites/mosquito, ×103 39 (18–60) 99 (74–124) CHMI (April 2010) Infection, % 100 100 Oocysts 5.6 17 Sporozoites/mosquito, ×103 12.5 69 Gametocyte male-female ratio 1:5 1:3 Drug sensitivity, mean IC50 (SD)b Dihydroartemisinin, nmol/L 3.4 (1.8) 9.9 (6.0) Lumefantrine, nmol/L 89 (26) 78 (9.7) Proguanil, µmol/L 21 (3.3) 27 (4.0) Atovaquone, nmol/L 0.3 (0.1) 0.6 (0.3) Chloroquine, nmol/L 201 (45) 24 (1.1) Abbreviation: CHMI, controlled human malaria infection; IC50, half-inhibitory concentration; SD, standard deviation. Mosquito infection and drug sensitivity profiles of NF135.C10 and NF54 in the period 2009–2010 and for the specific batches used in this CHMI. a Data for the period 2009–2010 represent mean findings (95% confidence intervals) after 26 and 39 standard dissections, for NF135.C10 and NF54 respectively, from 10 mosquitoes per dissection. b The drug sensitivities of NF135.C10 and NF54 were tested by the malaria SYBR Green I–based fluorescence assay in triplicate experiments; values represent means from 3 independent experiments.

a Data for the period 2009–2010 represent mean findings (95% confidence intervals) after 26 and 39 standard dissections, for NF135.C10 and NF54 respectively, from 10 mosquitoes per dissection. b The drug sensitivities of NF135.C10 and NF54 were tested by the malaria SYBR Green I–based fluorescence assay in triplicate experiments; values represent means from 3 independent experiments. Three of 5 volunteers infected with NF135.C10 and 4 of 5 infected with NF54 parasites had a positive thick smear during follow-up. The remaining 3 smear-negative volunteers were qPCR negative for P. falciparum for 21 days. In P. falciparum–positive volunteers, kinetics of parasitemia for both strains were comparable to those in historical controls (n = 48; Figure 1A [9]). Patent parasitemia for NF135.C10 occurred slightly earlier than for NF54, as measured by both thick smear (median [range], 7.0 [7.0–9.0] vs 10.6 [10.6–11] days after infection; P = .05, Mann–Whitney U test) and qPCR (median [range], 7.0 [6.3–7.0] vs 7.3 [7.0–7.3] days after infection; P = .1, Mann–Whitney U test). In addition, the peak of the first cycle seemed higher in infected volunteers for NF135.C10 (geometric mean [GM], 1.2 [95% confidence interval {CI}, .61–2.4] parasites/µL) than for NF54 (GM [95% CI], 0.16 [055–.46] parasites/µL; P = .06, Mann–Whitney U test). In the same 2 groups of volunteers, the GMs (95% CIs) for peak parasitemia were 11 (1.8–73) and 30 (7.7–120) parasites/µL, respectively (P = .4). All parasites were cleared from the blood of all volunteers during follow-up, with slopes that were similar for both strains (Figure 1B). The PCR identities of both strains were confirmed by culture of smear-positive samples from several randomly selected infected volunteers. Figure 1. Parasite kinetics of Plasmodium falciparum strains NF135.C10, and NF54 assessed by quantitative real-time polymerase chain reaction. Volunteers were infected by bites of mosquitoes infected with either NF135.C10 or NF54. A, Parasitemia of volunteers until thick smear positivity; data are shown as geometric means and 95% confidence intervals for volunteers successfully infected with NF135.C10 (red) or NF54 (black) and historical controls infected with NF54 (gray area; n = 48). B, Parasitemia of volunteers at the time of smear positivity and subsequent start of treatment (T), followed up for 3 days and finally at 28 days after infection; data are shown as geometric means and 95% confidence intervals for volunteers successfully infected with NF135.C10 (red; n = 3) or NF54 (black; n = 4).

48). B, Parasitemia of volunteers at the time of smear positivity and subsequent start of treatment (T), followed up for 3 days and finally at 28 days after infection; data are shown as geometric means and 95% confidence intervals for volunteers successfully infected with NF135.C10 (red; n = 3) or NF54 (black; n = 4). All volunteers, including the smear-negative volunteers, reported solicited adverse events that were considered possibly or probably related to the trial procedures (Table 2), particularly headache, fatigue, myalgia, and nausea, without apparent differences between the 2 groups. One volunteer infected with NF54 reported severe malaise, headache, and vomiting. One infected volunteer in the NF135.C10 group had a decreased platelet count of 146 × 109/L at day 3 after treatment (cutoff, 150 × 109/L), which returned to normal values at routine examination on day 28. Levels of D-dimers did not increase in any of the volunteers before thick smear positivity. Highly sensitive troponin T values were always <0.05 µg/L. Table 2. Adverse Events in Volunteers

as added to each plate, forming a 14-point standard curve, starting at 30 µg/mL. Plates were read at dual wavelengths (490 and 630 nm) on a Powerwave HT microplate reader (BioTek Instruments). Results were interpolated from standard curves with a 5 parameter curve fit, using Gen5 analysis software (BioTek Instruments). For analysis, infection intensity was expressed as mean egg count per gram (epg); geometric means were calculated to allow for skewness of data. Detection thresholds for enzyme-linked immunosorbent assay readings for each antigen and isotype were calculated as the mean plus 3 SDs of noninfected European control plasma samples. Risk factors for infection were examined using forward-fitting 2-level logistic regression analysis, to allow for correlations between siblings. Sex-adjusted associations between seroprevalence, age, and village were similarly examined using 2-level logistic models; age-village interactions were tested to determine whether associations varied with age and village. Nonlinear associations were examined by testing quadratic terms and categorical variables. Multilevel models were fitted in MLwiN (Bristol University, United Kingdom); other analyses were conducted using Stata, version 10.1 (StataCorp, United States).

nt of 146 × 109/L at day 3 after treatment (cutoff, 150 × 109/L), which returned to normal values at routine examination on day 28. Levels of D-dimers did not increase in any of the volunteers before thick smear positivity. Highly sensitive troponin T values were always <0.05 µg/L. Table 2. Adverse Events in Volunteers NF135.C10 (n = 3) NF54 (n = 4) Smear and PCR Negative (n = 3) Adverse Event Events, No. Duration, Mean (SD), d Events, No. Duration, Mean (SD), d Events, No. Duration, Mean (SD), d Abdominal pain 2 0 (0.0) 0 … 0 … Arthralgia 0 … 0 … 0 … Chills 0 … 0 … 0 … Fatigue 3 2.4 (1.9) 1 3.0 (…) 2 13.5 (9.3) Fever 1 0.2 (…) 2 0.7 (0.8) 0 Headache 3 1.5 (2.4) 4 2.3 (2.0) 3 4.6 (3.8) Itching 0 … 4 3.1 (1.5) 2 5.2 (0.3) Malaise 0 … 4 2.5 (3.3) 0 … Myalgia 1 2.7 (…) 3 1.3 (1.4) 2 2.7 (2.4) Nausea 1 0.1 (…) 3 2 (2.0) 1 0.0 (…) Vomiting 0 … 1 0.4 (…) 0 … Any 3 1.3 (1.7) 4 2.2 (1.9) 3 5.7 (5.8) Grade 3 adverse event Headache 0 … 1 4.6 (…) 0 … Malaise 0 … 1 0.4 (…) 0 … Vomiting 0 … 1 0.4 (…) 0 … Any 0 … 1 1.8 (2.4) 0 … Reported solicited adverse events, collected throughout the postinoculation period, that were considered possibly, probably, or definitely related to the trial procedures. Abbreviations: SD, standard deviation; PCR, polymerase chain reaction.

NF135.C10 (n = 3) NF54 (n = 4) Smear and PCR Negative (n = 3) Adverse Event Events, No. Duration, Mean (SD), d Events, No. Duration, Mean (SD), d Events, No. Duration, Mean (SD), d Abdominal pain 2 0 (0.0) 0 … 0 … Arthralgia 0 … 0 … 0 … Chills 0 … 0 … 0 … Fatigue 3 2.4 (1.9) 1 3.0 (…) 2 13.5 (9.3) Fever 1 0.2 (…) 2 0.7 (0.8) 0 Headache 3 1.5 (2.4) 4 2.3 (2.0) 3 4.6 (3.8) Itching 0 … 4 3.1 (1.5) 2 5.2 (0.3) Malaise 0 … 4 2.5 (3.3) 0 … Myalgia 1 2.7 (…) 3 1.3 (1.4) 2 2.7 (2.4) Nausea 1 0.1 (…) 3 2 (2.0) 1 0.0 (…) Vomiting 0 … 1 0.4 (…) 0 … Any 3 1.3 (1.7) 4 2.2 (1.9) 3 5.7 (5.8) Grade 3 adverse event Headache 0 … 1 4.6 (…) 0 … Malaise 0 … 1 0.4 (…) 0 … Vomiting 0 … 1 0.4 (…) 0 … Any 0 … 1 1.8 (2.4) 0 … Reported solicited adverse events, collected throughout the postinoculation period, that were considered possibly, probably, or definitely related to the trial procedures. Abbreviations: SD, standard deviation; PCR, polymerase chain reaction. T lymphocytes of volunteers successfully infected with either NF135.C10 or NF54 showed similarly increased IFN-γ, tumor necrosis factor, and interleukin 2 recall responses 35 days after infection and the same kinetics for both homologous and heterologous stimulation (Supplementary Figure 2A–I). IFN-γ–producing cells were found in both the innate compartment (γδ-T, natural killer, natural killer–T) and the adaptive compartment (CD4 and CD8) with an effector memory phenotype which was generally consistent over time and in both groups (Supplementary Figure 2J and 2K and data not shown).

pplementary Figure 2A–I). IFN-γ–producing cells were found in both the innate compartment (γδ-T, natural killer, natural killer–T) and the adaptive compartment (CD4 and CD8) with an effector memory phenotype which was generally consistent over time and in both groups (Supplementary Figure 2J and 2K and data not shown). DISCUSSION We identified and characterized NF135.C10 as the first P. falciparum clone of Asian origin for successful infection of malaria-naive human volunteers by CHMI. Clone NF135.C10 consistently produced gametocytes in culture and was able to generate infections in laboratory-reared mosquitoes with high yields of sporozoites. NF135.C10 parasites were clearly distinct from NF54 parasites by genetic marker profiles and were sensitive to the most commonly used antimalarials. Clinical presentation after CHMI and characteristics of P. falciparum RBC–specific recall (T-)lymphocyte responses in vitro were similar to those in NF54.

of sporozoites. NF135.C10 parasites were clearly distinct from NF54 parasites by genetic marker profiles and were sensitive to the most commonly used antimalarials. Clinical presentation after CHMI and characteristics of P. falciparum RBC–specific recall (T-)lymphocyte responses in vitro were similar to those in NF54. For manufacturing purposes, cultures should ideally produce gametocytes that consistently infect ≥75% of the mosquitoes with ≥10 oocysts, resulting in 10 000–30 000 sporozoites per mosquito. Selection, identification, and cloning of P. falciparum field strains that meet those criteria pose technical difficulties. Only after extensive efforts on >70 strains were we able to identify a parasite clone, NF135.C10, that met these criteria and which is geographically and molecularly distinct from NF54. We consider NF135.C10 closely related to its original field strain because of the limited restarts of the culture. We showed that clinical signs and symptoms after infection with NF135.C10 or NF54 were similar despite a shorter prepatent period in NF135.C10-infected volunteers. The observed difference in the prepatent period may represent a true difference in infectivity or may be due to coincidental distribution within the previously observed variation related to the limited number of volunteers.

Supplementary Data Notes Acknowledgments. We thank the volunteers for their enthusiastic participation in this trial and Kitty Suijk for her nursing support. We thank Laura Pelser, Jolanda Klaassen, Astrid Pouwelsen, and Jacqueline Kuhnen for their work in the culturing and dissection of mosquitoes. We are indebted to all the slide readers in Leiden: Jan Kromhout, Jaco Verweij, Meriam Beljon, Jolanda van Schie, Jaqueline Schelfaut, Jeanette van der Slot, Heleen Gerritsma, Fons van der Sande, Eric Brienen, and Els van Oorschot. We thank Adriana Ahumada at Protein Potential, Jianbing Mu and Xin-zhuan Su at the Laboratory of Malaria and Vector Research, National Institute of Allergy and Infectious Diseases, National Institute of Health, for the microsatellite mapping studies; Chris Janse, Shahid Khan, and the malaria team for their hospitality in their laboratory in Leiden; and safety monitors Sandra Arend and Mark de Boer and independent physician Frank Kroon for their continuing support. Financial support. This work was supported by Top Institute Pharma (grant T4-102), the European Malaria Vaccine Development Association (A. C. T), a long-term EMBO fellowship (A. S.), and NWO Mozaiek (grant 017.005.011 to K. N.). Potential conflicts of interest. B. K. L. S. is an employee of Sanaria Inc. S. L. H. is a major shareholder of Sanaria Inc. All other authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

35.C10 or NF54 were similar despite a shorter prepatent period in NF135.C10-infected volunteers. The observed difference in the prepatent period may represent a true difference in infectivity or may be due to coincidental distribution within the previously observed variation related to the limited number of volunteers. Notably, not all volunteers exposed to NF54 infected mosquitoes became parasitemic, in contrast to findings in 22 previous CHMI trials infecting 128 naive volunteers with NF54 parasites [10]. Unsuccessful infection after bites from 5 mosquitoes, although rare, has been described elsewhere for 3D7 [11, 12]. Although the exact reason for this low infectivity is unclear, it might be due to a technical disturbance in our cultures leading to unusually low NF135.C10 and NF54 oocyst and sporozoite counts in this particular trial, although the relation of these parameters to infectivity has never been formally established [13]. Surprisingly, all 3 unsuccessfully infected volunteers reported adverse events that were considered possibly or probably related to the trial procedures, which might have been the result of overreporting in an intense follow-up schedule. More studies are required to determine whether 100% infection rates can be achieved and to fully establish NF135.C10 as a heterologous field clone to complement the current CHMI portfolio of P. falciparum parasites. We have established master and working cell banks required to produce aseptic, purified, cryopreserved P. falciparum sporozoites using NF135.C10 parasites (B. K. L. S. et al, unpublished data), enabling the potential future needle and syringe inoculation of a stable number of sporozoites, in analogy to NF54 [14].

ites. We have established master and working cell banks required to produce aseptic, purified, cryopreserved P. falciparum sporozoites using NF135.C10 parasites (B. K. L. S. et al, unpublished data), enabling the potential future needle and syringe inoculation of a stable number of sporozoites, in analogy to NF54 [14]. We found similar kinetics and composition of IFN-γ recall responses with homologous and heterologous Pf54 and Pf135.C10 restimulation, possibly suggesting a role for specific (conserved) antigens in the induction and maintenance of heterologous memory responses against P. falciparum [15, 16]. Whether these cross-strain T-lymphocyte responses also translate into or represent cross-strain protective immunity in vivo remains to be investigated. In conclusion, increasing the portfolio of new P. falciparum parasite strains, as achieved here for NF135.C10, will accelerate the evaluation of malaria vaccines candidates by facilitating the downstream selection process for further clinical vaccine development. Moreover, heterologous parasite clones may be a component of whole sporozoite combination vaccines in order to enhance cross-strain protection. Although more trials will be necessary to fine-tune the heterologous CHMI model with clone NF135.C10, the current results will boost the continued application of CHMIs as a crucial tool for malaria vaccine development.

be a component of whole sporozoite combination vaccines in order to enhance cross-strain protection. Although more trials will be necessary to fine-tune the heterologous CHMI model with clone NF135.C10, the current results will boost the continued application of CHMIs as a crucial tool for malaria vaccine development. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

rdjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank the volunteers for their enthusiastic participation in this trial and Kitty Suijk for her nursing support. We thank Laura Pelser, Jolanda Klaassen, Astrid Pouwelsen, and Jacqueline Kuhnen for their work in the culturing and dissection of mosquitoes. We are indebted to all the slide readers in Leiden: Jan Kromhout, Jaco Verweij, Meriam Beljon, Jolanda van Schie, Jaqueline Schelfaut, Jeanette van der Slot, Heleen Gerritsma, Fons van der Sande, Eric Brienen, and Els van Oorschot. We thank Adriana Ahumada at Protein Potential, Jianbing Mu and Xin-zhuan Su at the Laboratory of Malaria and Vector Research, National Institute of Allergy and Infectious Diseases, National Institute of Health, for the microsatellite mapping studies; Chris Janse, Shahid Khan, and the malaria team for their hospitality in their laboratory in Leiden; and safety monitors Sandra Arend and Mark de Boer and independent physician Frank Kroon for their continuing support. Financial support. This work was supported by Top Institute Pharma (grant T4-102), the European Malaria Vaccine Development Association (A. C. T), a long-term EMBO fellowship (A. S.), and NWO Mozaiek (grant 017.005.011 to K. N.).

ested to determine whether associations varied with age and village. Nonlinear associations were examined by testing quadratic terms and categorical variables. Multilevel models were fitted in MLwiN (Bristol University, United Kingdom); other analyses were conducted using Stata, version 10.1 (StataCorp, United States). RESULTS Overall, 42.1% of children had detectable S. mansoni, and the geometric mean infection intensity among those infected was 49.23 epg. The prevalence and intensity of infection varied significantly by village. In Bugoigo, the prevalence was 53.0%, compared with 27.5% in Piida (P < .001), and geometric mean intensity of infection among infected individuals was 61.38 epg in Bugoigo and 27.79 epg in Piida (P = .002).

Streptococcus pneumoniae (the “pneumococcus”) is an asymptomatic colonizer of the human nasopharynx and a major cause of otitis media, sinusitis, pneumonia, and meningitis, resulting in approximately 14.5 million annual serious disease episodes globally among young children [1]. The primary virulence determinant is its polysaccharide capsule, of which there are 94 known serotypes [2–6]. The capsule protects against phagocytosis during invasive pneumococcal disease and may also prevent clearance during nasopharyngeal colonization [7, 8]. Ninety-two capsule types are synthesized and exported through the “Wzy-dependent” pathway, whereby extracellular polymerization of component lipid-linked repeat units is preceded by “flippase-mediated” transfer across the cell membrane [2, 9]. The proteins involved are encoded by the capsule polysaccharide synthesis (cps) genes, located between dexB and aliA on the chromosome. The remaining 2 capsule types are synthesized through independent biochemical pathways. Sequence analyses of 88 reference cps loci revealed variable lengths (approximately 10–30 kb) and a range of genes specific to capsule production [2]. The serotype-nonspecific genes were present among all Wzy-dependent capsule types, whereas serotype-specific genes were only found among 1 or a subset of types. Unique serotypes, and further diversity within serogroups, evolved through a combination of mutation and interspecies/intraspecies recombination [2, 10–13].

]. The serotype-nonspecific genes were present among all Wzy-dependent capsule types, whereas serotype-specific genes were only found among 1 or a subset of types. Unique serotypes, and further diversity within serogroups, evolved through a combination of mutation and interspecies/intraspecies recombination [2, 10–13]. Serotype is a key determinant of invasive pneumococcal disease potential and prevalence; certain serotypes are more commonly associated with carriage and others more commonly with invasive pneumococcal disease [14, 15]. A 7-valent pneumococcal conjugate vaccine (PCV7) was introduced in the United States in 2000 [16] and, subsequently, into many other countries. Ten-valent and 13-valent vaccines (PHiD-CV and PCV13, respectively) contain polysaccharides targeted against the original 7 serotypes (4, 6B, 9V, 14, 18C, 19F, and 23F) plus serotypes 1, 5, and 7F (both vaccines) and 3, 6A, and 19A (PCV13 only) and have now replaced PCV7 [17, 18].

bsequently, into many other countries. Ten-valent and 13-valent vaccines (PHiD-CV and PCV13, respectively) contain polysaccharides targeted against the original 7 serotypes (4, 6B, 9V, 14, 18C, 19F, and 23F) plus serotypes 1, 5, and 7F (both vaccines) and 3, 6A, and 19A (PCV13 only) and have now replaced PCV7 [17, 18]. Surveillance following PCV7 introduction showed a decline of vaccine type (VT) and an increase of nonvaccine type (NVT) pneumococci in disease and nasopharyngeal carriage [19]. This can be attributed to the phenomena of “serotype replacement,” the expansion of preexisting NVT pneumococci, and/or “serotype switching,” a change of serotype of a single clone by alteration or exchange of its cps locus [20]. (Serotype switching was first described by Griffith in 1928 [21] and was the focus of the transformation studies by Avery and colleagues in 1944 [22].) These effects are not completely independent: capsular switch variants can subsequently expand within a population. Both phenomena can be studied by comparison of the serotypes and genotypes present in populations before and after introduction of pneumococcal vaccines. Pneumococcal genotypes, as defined by multilocus sequence typing (MLST) [23], show serotype-specific associations [14, 24]; any isolate exemplifying a different genotype/serotype combination may represent a capsular switch variant. Such variants usually arise by recombination at the cps locus, and studies of a limited number of such strains indicated that recombination fragment sizes varied from approximately 21.9 kb to approximately 56.5 kb [25–28]. In some cases, the fragments also included part or all of the pbp2x and pbp1a genes (2 of the 3 primary penicillin-resistance determining genes, located approximately 8 kb upstream and approximately 7 kb downstream of the cps locus).

ion fragment sizes varied from approximately 21.9 kb to approximately 56.5 kb [25–28]. In some cases, the fragments also included part or all of the pbp2x and pbp1a genes (2 of the 3 primary penicillin-resistance determining genes, located approximately 8 kb upstream and approximately 7 kb downstream of the cps locus). Vaccine-induced selective pressure is contributing to the postvaccination changes of serotype epidemiology, but natural fluctuations in serotype prevalence also play a role. Many studies have documented prevaccine temporal changes in relative serotype prevalence: all but 3 included pneumococci isolated no earlier than 1969; only 2 studies provided genotype data (Supplementary Materials). In this study, we used serotyping, MLST, and whole-genome data to study a large, genetically diverse collection of historical and modern pneumococci, with the aim to better understand the mechanisms and role of capsular switching in pneumococcal evolution.

nly 2 studies provided genotype data (Supplementary Materials). In this study, we used serotyping, MLST, and whole-genome data to study a large, genetically diverse collection of historical and modern pneumococci, with the aim to better understand the mechanisms and role of capsular switching in pneumococcal evolution. METHODS Strains and Genome Sequencing A global collection of 426 pneumococci recovered during 1937–2007 (Supplementary Materials) were previously serotyped by the Quellung reaction and genotyped by MLST [23] to determine the sequence type (ST). Closely related isolates were assigned to clonal complexes (CCs) by a modified goeBURST method [29] (Supplementary Materials). CCs were named after the predicted founder ST. When no single founder ST could be determined, CCs were named NoneX, where X was the ST of lowest numeric value within the CC. Isolates of the same CC but with different serotypes were presumed to be capsular switch variants. A total of 96 genetically diverse isolates from our collection were selected for whole-genome sequencing (Supplementary Materials and Table 1) on the Illumina platform as previously described [30]. Seven were excluded as technical failures. Raw sequence data were assembled using Velvet [31], and contigs were deposited in a BIGS database (BIGSdb) [32]. Sequence data were deposited in the European Nucleotide Archive (Supplementary Table 1).

terials and Table 1) on the Illumina platform as previously described [30]. Seven were excluded as technical failures. Raw sequence data were assembled using Velvet [31], and contigs were deposited in a BIGS database (BIGSdb) [32]. Sequence data were deposited in the European Nucleotide Archive (Supplementary Table 1). Nucleotide Sequence Analysis Eight CCs were represented by isolates subjected to whole-genome sequencing and were selected for further cps locus analysis; 7 CCs included capsular switch variants. Within each CC, the ancestral serotype was assumed to be the most common one or that of the oldest isolate. cps sequence alignments for study isolates and published loci were used to investigate serotype changes. Nucleotide sequences of the cps locus, and upstream/downstream flanking sequences where appropriate, were aligned using MUSCLE [33] and imported to MEGA5 [34] for visual inspection of variable sites. When recombination at the cps locus was predicted, potential donor representatives were sought from our whole-genome sequenced isolates, the cps locus reference sequences, and an additional 131 pneumococcal genomes retrieved from GenBank. Putative donors were identified as isolates for which the cps locus differed from that of the recombinant by a maximum of 3 nucleotides (excluding clusters of closely linked substitutions, which were considered to have arisen through recombination within the cps locus itself). MLST data were used to infer the CC most likely represented by the true donor isolate (the same as that of the donor representative). Recombination regions were identified as the minimum region over which the recombinant representative (hereafter, the “recombinant”) differed from the ancestral representative (hereafter, the “ancestor”) and was identical or highly similar (>99.7% sequence identity) to the donor representative (hereafter, the “donor”). Where no suitably matched donor was identified, the maximum recombination region was estimated as the maximum length over which the recombinant differed from the ancestor (Supplementary Materials).

tor”) and was identical or highly similar (>99.7% sequence identity) to the donor representative (hereafter, the “donor”). Where no suitably matched donor was identified, the maximum recombination region was estimated as the maximum length over which the recombinant differed from the ancestor (Supplementary Materials). RESULTS Indication of Serotype Switching Among the entire collection of 426 pneumococci, 21 of 163 unique CCs were represented by isolates of 2–6 serotypes. At least 36 independent changes of serotype within CCs were represented in our collection, 34 of which predated the introduction of pneumococcal conjugate vaccines (Figure 1). Figure 1. Temporal distribution of capsular switch variants among the historical isolate collection. Gray triangles indicate isolation dates for the oldest recombinant representatives of 36 independent capsular switch events identified among the historical pneumococcal collection. Abbreviation: PCV7, 7-valent pneumococcal conjugate vaccine.

tribution of capsular switch variants among the historical isolate collection. Gray triangles indicate isolation dates for the oldest recombinant representatives of 36 independent capsular switch events identified among the historical pneumococcal collection. Abbreviation: PCV7, 7-valent pneumococcal conjugate vaccine. Within-CC cps Diversity CC15 Seven of 10 CC1514 (CC15, serotype 14) cps locus sequences were unique (Table 1). Isolates ICE13 and ICE50 shared wciY locus nucleotide substitutions with 4 CC12414 representatives (isolates 14/9, USA6, Ala289, and ICE594), suggesting a possible recombination event between these CCs. The CC1514 representatives had a 432-bp deletion in the wciY gene, resulting in a predicted 250-amino-acid truncation of the protein. Isolate CGSP14 [35] also had an additional 52-bp deletion within the wciY gene; if this represented a true deletion event rather than a sequencing/assembly error, the resultant protein would be further reduced to 38 amino acids. Table 1. Genetic Evolution at the cps Locus Among Isolates Representing the Same Pneumococcal Serotypes and Clonal Complexes (CCs)

nal 52-bp deletion within the wciY gene; if this represented a true deletion event rather than a sequencing/assembly error, the resultant protein would be further reduced to 38 amino acids. Table 1. Genetic Evolution at the cps Locus Among Isolates Representing the Same Pneumococcal Serotypes and Clonal Complexes (CCs) CCSerotype Isolate Year Nucleotide Substitution(s) as Compared to Ancestor (gene) Recombination Regions 1514 14/5 1967 NA … PMEN10 1987 A8695G (wchM); missing bp 11 509–11 525 and bp 12 423–enda … SPnINV200 1995 C12439T (wciY) … ICE13 1998 C12466A and C12467A (wciY) … Ala243 1998 C12439T (wciY) … GA13338 1999 C12439T (wciY) … Ala317 2001 None … ICE50 2003 C12439T, C12466A, C12467A, and G12502A (wciY); 22-bp insertion at position 11 509 … CGSP14 2004–2005 C1010T (wzg) and C12439T (wciY); missing bp 11 488–11 539 cps bp 1–367 SP14-BS292 Unknown T1976A (wzh) and C12439T (wciY) … 669N 9N/6 1960 NA … USA8 2001 A4399G (wchA), A14000G, T14014G, C14037T, and T14042C (between ugd and IS1381 putative transposase) … 11318C 18C/2 1939 NA … 18C/1b 1940 A2203G (wzd) … 18C/3 1968 G3478A (wze) … Netherlands18C-36 1980 None … USA2 1999 C4622T (wchA) cps bp 15 861–17 005 ICE501 2002 None … 12414 14/2b 1952 NA … 14/4 1961 A3109G and G3478A (wze), A5616G and C5617T (wchK), and T6662A (wzy) cps bp 12 330–12 467 PMEN35 1980 None; missing bp 12 437–enda … 14/9 1982 T11491C, A11524G, C12466A, C12467A, G12495A, and A12505T (wciY) … 14/7 1992 A4141G (wchA) … Ala292 1998 C3816A and G4087A (wchA) … USA6 1999 C43A (wzg), A12403G, C12466A, C12467A, and G12502A (wciY) … Ala263 2002 None … Ala289 2002 C12466A, C12467A, and G12502A (wciY) … ICE46 2003 None … ICE594 2005 C12466A, C12467A, and G12502A (wciY) … CCR11974 G2455T (wzd) … 156/1629V 9V/5 1991 NA … PMEN3 1993 None … GA08780 1997 T5230C (wchO) … SP9-BS68 Unknown None … 1917F 7F/3 1962 NA; missing bp 20 430–enda … 7F/5b 1962 A16G (wzg) … Netherlands7F-39 1984 None; missing bp 20 434–enda … 7F/4 1986 None; missing bp 20 434–enda … ICE22 1993 C20435T, A20436G, T20441C, and A20450G (glf) … CDC1087–00 1999 T10697C (wchF) … USA16 2003 None … 21812F 12F/5 1988 NA … Denmark12F-34 1995 G16593T (between fnlA and fnlB) … 12F/6 1996 G16593T (between fnlA and fnlB) and C18121T (fnlC) … USA18 1999 G16593T (between fnlA and fnlB) … CDC0288-04 2003–2004 C7397T (wzy), G16593T (between fnlA and fnlB), and C18121T (fnlC) … 2187F 7F/2 1952 NA … ICE23 1993 C5369A (wchF), C6499T (wcwA), and A12622C (wcwH) … USA20 1999 C9059T (HG140), C11766G (wcwH); missing b

6593T (between fnlA and fnlB) and C18121T (fnlC) … USA18 1999 G16593T (between fnlA and fnlB) … CDC0288-04 2003–2004 C7397T (wzy), G16593T (between fnlA and fnlB), and C18121T (fnlC) … 2187F 7F/2 1952 NA … ICE23 1993 C5369A (wchF), C6499T (wcwA), and A12622C (wcwH) … USA20 1999 C9059T (HG140), C11766G (wcwH); missing b p 20 431–enda … Abbreviation: NA, not applicable. a Likely due to sequencing/assembly failure. b This isolate was used as the cps reference strain in reference [2]. CC66 Ten CC66 isolates, dated 1952–2005, represented 5 serotypes. The oldest isolate was serotype 7B; the next oldest was a serotype 9N from 1960. An approximately 53.4-kb region of sequence of the serotype 9N isolate, including pbp2x, the cps locus, and pbp1a, differed from the CC667B isolate but was highly similar to a CC37829N isolate dated 1952 (Table 2 and Figure 2). (Although the data strongly suggest that a CC37829N pneumococcus was the donor in this case, it is possible that another clone not represented in this sample was the true donor. Thus, we use the term “CCXX-like” when describing putative donors.) Within the cps locus, a second CC669N isolate differed from the oldest CC669N isolate by only 5 nucleotide substitutions (Table 1). Table 2. Genetic Changes Leading to Capsular Switching Events Among Pneumococci

ented in this sample was the true donor. Thus, we use the term “CCXX-like” when describing putative donors.) Within the cps locus, a second CC669N isolate differed from the oldest CC669N isolate by only 5 nucleotide substitutions (Table 1). Table 2. Genetic Changes Leading to Capsular Switching Events Among Pneumococci Recombination CC Serotype Change Yeara PM or R Point Mutation(s) Donor Breakpoints Import, kb cps Point Mutations as Compared to Donorc 66 7B→9N 1960, 2001 R … CCNone37829N ≥8901 bp 5’ pbp2x–7415 bp 3’ pbp1a ≥54.3 Noned 9N→19F 1972 R … Unk19F 2538 bp 5’ dexB–34 bp 5’ aliA ≤23.7 … 9N→19F 2005 R … Unk19F 115 bp 3’ dexB–bp 1173 aliA ≤20.6 … 9N→14 1995, 1999 R … CC1514 2158 bp 5’ dexB–between bp 88 aliA and 785 bp 3’ aliA 23.1–25.8 A8695G, G10976T, and C12439T 9N→14 1997 R … CC1514 Und Und A4486G, A8695G, and G10461Ae 9N→14 unka R … CC12414 ≥2520 bp 5’ dexB–4062 bp 3’ aliA ≥29.9 A1704G and T12439C 9N→23F 2001 R … Unk23F 1964 bp 5’ dexB–3013 bp 3’ aliA ≤30.4 … 113 18C→18B 1941b PM G12382T … … … … 18C→35C 1941, 1943b R … Unk35C 144 bp 3’ dexB–76 bp 5’ aliA ≤19.0 … 35C→17F 1939 R … CC57417F bp 549 wze–3278 bp 3’ pbp1a 30.9 A8794G 18C→9V 1968b R … Unk9V Und Und … 124 14→11C 1957b R … Unk11C Und Und … 14→9L 1952 R … Unk9L Und Und … 156/162 9V→9A 1962b PM G1543A, deletion bp 16 996 … … … … 9V→19A 2005, 2006 R … CC19919A ≥5701 bp 5’ dexB to ≥6022 bp 3’ aliA ≥ 50.4 None 191 7F→7A 1937b PM A2375G, A10521T, insertion bp 8606 … … … … 218 12F→7F 1952, 1993, 1999 R … CC1917F ≥9045 bp 5’ pbp2x–4782 bp 3’ pbp1a ≥ 58.2 C6854A, C8089T, and 2 putative recombinations marked by 39 nucleotide changes bp 1083–1279 and 7 changes bp 19 486–20 473 574 17Fb→2 1956b R … CCNone1282 Between bp 326 dexB and 639 bp 3’ dexB–5229 bp 3’ pbp1a 34.9–36.8 G3294T, T3339C, and G3956T Abbreviations: CC, clonal complex; PM, point mutation; R, recombination; Unk, unknown; Und, undetermined.

ecombinations marked by 39 nucleotide changes bp 1083–1279 and 7 changes bp 19 486–20 473 574 17Fb→2 1956b R … CCNone1282 Between bp 326 dexB and 639 bp 3’ dexB–5229 bp 3’ pbp1a 34.9–36.8 G3294T, T3339C, and G3956T Abbreviations: CC, clonal complex; PM, point mutation; R, recombination; Unk, unknown; Und, undetermined. a Year of recipient isolate(s) with the new serotype. b This isolate was used as the cps locus reference strain in reference [2]. c Calculated over the region spanning the synthesis-related genes only (Supplementary Materials). d Missing bp 14 006–15 542 (end) likely due to sequencing/assembly failure. e Missing bp 12 437–12 516 (end) likely due to sequencing/assembly failure. Figure 2. Genetic changes leading to capsular switching within clonal complexes (CCs) 66, 113, and 156/162. Bars represent the cps locus and flanking regions. Filled arrows indicate the relative positions of the dexB (left) and aliA (right) loci. Open arrows indicate the relative positions of the pbp2x (left) and pbp1a (right) loci. Vertical bisecting lines indicate single nucleotide substitutions. Recombination fragments are indicated by coloring. Identical sequences are indicated by identical coloring. Regions bounded by solid lines are depicted at their maximum length. Regions not bounded by solid lines are depicted at their minimum length (Table 2). The asterisk indicates a single-base-pair deletion. The schematic is drawn approximately to scale.

by coloring. Identical sequences are indicated by identical coloring. Regions bounded by solid lines are depicted at their maximum length. Regions not bounded by solid lines are depicted at their minimum length (Table 2). The asterisk indicates a single-base-pair deletion. The schematic is drawn approximately to scale. Sequences of the 2 CC6619F genomes suggested 2 independent capsular switches, but no potential 19F donors were identified. The putative imports were estimated as the regions over which these genomes clearly differed from those of CC669N; independent recombination breakpoints were indicated (Table 2 and Figure 2). There were 3 independent capsular switches among 4 CC6614 representatives. On the basis of sequence similarity, the donors for 2 events were CC1514-like, while the third was CC12414-like. The breakpoints for 1 event could not be determined because of extensive diversity in the cps flanking regions, but breakpoints for the others were estimated (Table 2 and Figure 2). Interestingly, the 5′ breakpoints for 2 serotype 14 switches and one of the aforementioned serotype 19F switches were all 2.2–2.5 kb upstream of dexB (in the ATCC700669 reference genome [36], this corresponds to the clpL locus, which has been shown to be involved in modulating the expression of virulence-related genes [37] and in the development of penicillin nonsusceptibility [38]).

ne of the aforementioned serotype 19F switches were all 2.2–2.5 kb upstream of dexB (in the ATCC700669 reference genome [36], this corresponds to the clpL locus, which has been shown to be involved in modulating the expression of virulence-related genes [37] and in the development of penicillin nonsusceptibility [38]). The single CC6623F representative in our collection differed from the CC669N isolates from approximately 2 kb upstream of dexB to approximately 3 kb downstream of aliA, suggesting a potential recombination import of approximately 30.4 kb, but no suitable donor was identified.

ne of the aforementioned serotype 19F switches were all 2.2–2.5 kb upstream of dexB (in the ATCC700669 reference genome [36], this corresponds to the clpL locus, which has been shown to be involved in modulating the expression of virulence-related genes [37] and in the development of penicillin nonsusceptibility [38]). The single CC6623F representative in our collection differed from the CC669N isolates from approximately 2 kb upstream of dexB to approximately 3 kb downstream of aliA, suggesting a potential recombination import of approximately 30.4 kb, but no suitable donor was identified. CC113 Eleven CC113 isolates represented serotypes 18C (n = 6), 35C (n = 2), and 9V, 17F, and 18B (n = 1 each). Serotype 18C was assumed to be the ancestral serotype and was also represented by 1 of the 2 oldest isolates, both dated 1939. The change from serotype 18C to 18B was due to a single nucleotide substitution within the wciX gene [2], while each of the other serotype changes was associated with a recombination event. Both CC11335C isolates likely arose from a single capsular switch event (Table 2 and Figure 2). The serotype 17F change appears to have involved DNA acquisition from a CC57417F-like pneumococcus by a CC11335C pneumococcus. The 5′ region of the CC11317F cps locus was distinct from that of 3 other serotype 17F isolates in our collection but differed by only 1 nucleotide substitution from the CC11335C representatives. The CC11317F genome was highly similar to that of a CC57417F isolate between position 549 of the wze locus and approximately 3 kb downstream of pbp1a, after which the CC11317F genome resembled that of the CC11318C isolates.

our collection but differed by only 1 nucleotide substitution from the CC11335C representatives. The CC11317F genome was highly similar to that of a CC57417F isolate between position 549 of the wze locus and approximately 3 kb downstream of pbp1a, after which the CC11317F genome resembled that of the CC11318C isolates. There was little to no variation at the cps locus among 5 CC11318C representatives (Table 1). A sixth isolate differed by 1 nucleotide in the wchA gene and by a putative import of approximately 1.1 kb (marked by 44 nucleotide substitutions) spanning the rmlA and rmlC genes (Figure 3A). Figure 3. Variable site maps for the cps loci of serogroups 18 and 7. Rows represent the nucleotide sequences of independent isolates and are labeled as isolate name (year). Nucleotide differences relative to row 1 are shown. Periods indicate an identical base, and hyphens indicate a missing base. Numbers above each column indicate the nucleotide position within the alignment. Nucleotide substitutions marking putative recombination regions are marked in bold and underlined. A, CC113 isolates include 1 of serotype 18B (18B/2) and 6 of serotype 18C. B, CC191 isolates include 1 of serotype 7A (7A/2) and 7 of serotype 7F. CC218 isolates all represent serotype 7F (B*). The arrow indicates the position of a single-base-pair insertion in the 7A/2 sequence, compared with the other serogroup 7 sequences.

3 isolates include 1 of serotype 18B (18B/2) and 6 of serotype 18C. B, CC191 isolates include 1 of serotype 7A (7A/2) and 7 of serotype 7F. CC218 isolates all represent serotype 7F (B*). The arrow indicates the position of a single-base-pair insertion in the 7A/2 sequence, compared with the other serogroup 7 sequences. CC124 Twelve of 14 CC124 representatives were serotype 14 and represented 8 unique but highly similar cps sequences (Table 1). Two isolates were serotypes 9L and 11C, but no suitable donors were identified, and the cps flanking regions of both isolates were highly divergent from those of the CC12414 representatives for ≥30 kb in either direction. Thus, the putative recombination breakpoints could not be estimated. CC156/162 Seven isolates (dated 1952–2006) representing serotypes 9V, 9A, and 19A were assigned to CC156/162. Four isolates were serotype 9V; the oldest was serotype 9A. The CC156/1629A cps locus differed from that of the oldest CC156/1629V locus by 2 nucleotides [2] (Table 2 and Figure 2). Three of the 4 CC156/1629V cps loci were identical, and the other differed by 1 nucleotide.

es 9V, 9A, and 19A were assigned to CC156/162. Four isolates were serotype 9V; the oldest was serotype 9A. The CC156/1629A cps locus differed from that of the oldest CC156/1629V locus by 2 nucleotides [2] (Table 2 and Figure 2). Three of the 4 CC156/1629V cps loci were identical, and the other differed by 1 nucleotide. Analysis of the cps loci and flanking regions of the 2 CC156/16219A representatives (GenBank accession numbers AGOR01000001–AGOR01000024 and AGQA01000001–AGQA01000007) suggested a single capsular switch event. Approximately 50.4-kb regions of the CC156/16219A genomes were highly similar to those of a CC19919A isolate (Table 2 and Figure 2) and included pbp2x, the cps locus, and pbp1a. The regions directly flanking the putative import did not resemble those of the CC156/1629V representatives, and thus the true recombination breakpoints could not be estimated. However, we deduced that the original import was ≥50.4 kb.

C19919A isolate (Table 2 and Figure 2) and included pbp2x, the cps locus, and pbp1a. The regions directly flanking the putative import did not resemble those of the CC156/1629V representatives, and thus the true recombination breakpoints could not be estimated. However, we deduced that the original import was ≥50.4 kb. CC191 Seven of the CC191 representatives were serotype 7F, and the eighth, and oldest, was serotype 7A. Serotypes 7A and 7F differed by 3 nucleotides (Table 2 and Figure 4). There was a repeated motif (5′-CTA AGA TGA ATA-3′) within the wcwC gene, and the number of repeats differed between isolates (n = 3, 4, or 6). Repeat motifs are difficult to sequence accurately, so further speculations about such changes cannot be made. Apart from the repeat region, the CC1917F cps loci each differed by a maximum of 4 nucleotides (Table 1 and Figure 3B). Figure 4. Genetic changes leading to capsular switching within clonal complexes (CCs) 191, 218, and 574. Bars represent the cps locus and flanking regions. Filled arrows indicate the relative positions of the dexB (left) and aliA (right) loci. Open arrows indicate the relative positions of the pbp2x (left) and pbp1a (right) loci. Vertical bisecting lines indicate single nucleotide substitutions. Recombination fragments are indicated by coloring. Regions bounded by solid lines are depicted at their maximum length. Regions not bounded by solid lines are represented at their minimum length (Table 2). The asterisk indicates a single-base-pair insertion. The schematic is drawn approximately to scale.

substitutions. Recombination fragments are indicated by coloring. Regions bounded by solid lines are depicted at their maximum length. Regions not bounded by solid lines are represented at their minimum length (Table 2). The asterisk indicates a single-base-pair insertion. The schematic is drawn approximately to scale. CC218 Eight CC218 isolates represented serotypes 12F (n = 5) and 7F (n = 3). Between approximately 9 kb upstream of pbp2x and approximately 4.8 kb downstream of pbp1a, CC2187F and CC1917F sequences were very similar, although differentiating nucleotides were identified (Table 2). The clustering of 44 nucleotide substitutions within the cps locus was indicative of recombination events (Figure 3B), but whether these events occurred before or after the cps switch is unknown. Note that approximately 585 bp were missing between pbp2x and dexB in both CC2187F and CC1917F representatives. Table 1 provides additional details about nucleotide substitutions among the cps loci of CC2187F and CC21812F isolates. CC574 Two CC574 isolates were examined: a serotype 17F isolate from 1952 and a serotype 2 isolate from 1956. The sequences differed from position 326 of dexB to approximately 5 kb downstream of pbp1a. A region of the serotype 2 representative (totaling approximately 35 kb, from approximately 0.6 kb downstream of dexB to approximately 5 kb downstream of pbp1a) was highly similar to that of a CCNone1282 representative dated 1916 (Table 2 and Figure 4). A short, 314-bp region of unknown sequence in the dexB locus was also present.

the serotype 2 representative (totaling approximately 35 kb, from approximately 0.6 kb downstream of dexB to approximately 5 kb downstream of pbp1a) was highly similar to that of a CCNone1282 representative dated 1916 (Table 2 and Figure 4). A short, 314-bp region of unknown sequence in the dexB locus was also present. DISCUSSION Within our collection of historical and modern pneumococci, we identified 36 independent capsular switching events. Approximately 94% of the variants were isolated prior to the introduction of PCV7 and were roughly evenly distributed through time (Figure 1). The collection was not designed for inferring a capsular switching rate; nevertheless, these data imply that this phenomenon has been a regular occurrence (ie, there is evidence of capsular switching within a diverse range of CCs) every decade throughout the past 7 decades). Analysis of the cps loci of 10 representatives of CC66 indicated multiple independent changes to the same serotype, supporting the notion that capsular switching may occur regularly among pneumococci. Within this CC, an initial capsular switch from serotype 7B to 9N no later than 1952 was followed by at least 2 independent changes to serotype 19F no later than 1972 and 2005 and by at least 3 independent changes to serotype 14, each no later than the mid-1990s. This is consistent with previous studies that demonstrated multiple changes of serotype within the same CC: CC156/1629V→14 [39–41], CC8123F→19F/A [30], and CC6954→19A [25, 28].

ent changes to serotype 19F no later than 1972 and 2005 and by at least 3 independent changes to serotype 14, each no later than the mid-1990s. This is consistent with previous studies that demonstrated multiple changes of serotype within the same CC: CC156/1629V→14 [39–41], CC8123F→19F/A [30], and CC6954→19A [25, 28]. We studied 18 capsular switching events in detail. Three of these were presumably the result of nucleotide substitution and/or deletion, a finding consistent with previous work [2]. The remaining 15 events appeared to be due to recombination, and breakpoints for 11 events could be estimated by comparison of the cps loci and flanking sequences of the putative ancestors, donors, and recombinants. The capsular switch recombination fragments identified here (ie, imports of various lengths, inserted at different points around the cps locus, with or without the adjacent pbp sequences) are consistent with fragments detected in previously published studies [25, 42, 43]. Figure 5 depicts the exchange of cps loci between CCs, as inferred by our data. Isolates identified as donors from our collection could possibly have represented different CCs than the true donors; however, even if this were true, our conclusions about the number of independent capsular switches and the range of recombination fragment sizes would remain unchanged. Figure 5. Inter-clonal complex (CC) cps locus transfers inferred in this study. Circles represent pneumococcal CCs as named. Arrows indicate transfer of cps loci between CCs, as inferred from nucleotide sequence analyses for 15 capsular switching events. Donor CCs could not be predicted for a total of 7 events because donor isolates were not present among this collection. Arrow labels indicate serotypes associated with the transferred cps loci and isolation year of the oldest recombinant pneumococcus in this collection.

quence analyses for 15 capsular switching events. Donor CCs could not be predicted for a total of 7 events because donor isolates were not present among this collection. Arrow labels indicate serotypes associated with the transferred cps loci and isolation year of the oldest recombinant pneumococcus in this collection. Our analyses did not indicate any recombination breakpoint hot spots around the cps locus, which had also not been indicated by any previous studies [25, 27, 28, 30]. It is impossible to know whether the putative imports were acquired through single or multiple recombination events, but the former is most parsimonious. The import lengths were estimated to range from approximately 19.0 to ≥58.2 kb (and apart from 1 example, always included the entire cps locus) and did not show any trends toward increasing/decreasing lengths through time. The CC21812F→7F and CC667B→9N events were characterized by large (>50 kb) recombination imports and must have taken place no later than 1952 and 1960, respectively. Large-scale recombination between pneumococci has therefore clearly been occurring for decades, although the technology capable of detecting and detailing such events has only recently become available [25, 28, 30, 42]. Additionally, both of these putative imports included the pbp2x + cps + pbp1a loci, as reported for other events [27, 28] and for another newly characterized event in this study (CC156/1629V→19A).

although the technology capable of detecting and detailing such events has only recently become available [25, 28, 30, 42]. Additionally, both of these putative imports included the pbp2x + cps + pbp1a loci, as reported for other events [27, 28] and for another newly characterized event in this study (CC156/1629V→19A). Recently reported in vivo pbp2x ± cps ± pbp1a recombination events were detected soon after widespread vaccination began in the United States. No such recombinants had been reported before in nature, and the imported pbp2x and pbp1a sequences conferred penicillin nonsusceptibility. Thus, it was theorized that vaccine-induced and/or antibiotic-induced selective pressures may play a role in driving these genetic changes [28, 44]. Penicillin was introduced in the 1940s; oral penicillins were not available until the mid-1950s, and their use was initially limited. Consequently, our analyses question the above theory because the CC21812F→7F and CC667B→9N events were both associated with penicillin-susceptible pneumococci and occurred before the widespread use of penicillins and before the introduction of PCV7. Our new data suggest that recombination of the cps locus and flanking regions might be “normal” biological processes, the evolution of which has undoubtedly been influenced by naturally occurring immunity and other selective pressures. Presumably, vaccine-induced immune pressures and/or the pressure of antibiotic use subsequently influence the spread and maintenance of advantageous genes (and/or alleles) by selecting recombinants that are best able to survive. However, the potential negative effects should not be underestimated: the CC6954→19A “vaccine escape” capsular switches in the United States were first recovered from patients with pediatric invasive pneumococcal disease only 3 years after PCV7 introduction and, 2 years later, were the third-most common serotype 19A CC causing invasive pneumococcal disease among all age groups [20]. These strains were penicillin nonsusceptible, owing to the simultaneous acquisition of altered pbp2x and pbp1a genes. Consequently, the increase in prevalence of these strains likely contributed to the increase in pneumococcal penicillin nonsusceptibility in the United States after PCV7 introduction [20, 45].

ups [20]. These strains were penicillin nonsusceptible, owing to the simultaneous acquisition of altered pbp2x and pbp1a genes. Consequently, the increase in prevalence of these strains likely contributed to the increase in pneumococcal penicillin nonsusceptibility in the United States after PCV7 introduction [20, 45]. Another interesting finding was that a CC19919A-like representative was the most probable donor of the 19A cps locus and flanking pbps to the CC156/16219A isolates described in this study. CC19919A representatives were also the donors of the cps locus ± pbps to the vaccine escape progeny [25, 28] and an ST32019→19A cps locus switch [27]. Future studies will attempt to uncover an explanation of why CC19919A representatives appear to be “good” cps locus donors.

C156/16219A isolates described in this study. CC19919A representatives were also the donors of the cps locus ± pbps to the vaccine escape progeny [25, 28] and an ST32019→19A cps locus switch [27]. Future studies will attempt to uncover an explanation of why CC19919A representatives appear to be “good” cps locus donors. Our analyses also revealed recombination at the cps locus that did not result in capsular switching, among isolates belonging to the same ancestral lineages (CC11318C and CC1917F) and different ancestral lineages (CC1917F vs CC2187F and CC1514 vs CC12414). Similar events have been noted within the CC8123F lineage [26] and among serogroup 6 and 19 isolates [10–12]. We speculate that this is more likely to be the result of recombination whereby some genes are coincidentally exchanged, rather than the result of exchange that occurred directly in response to selection pressure. It could also be the result of DNA repair mediated by recombination at or near the cps locus. A comparison of the cps loci of CC1514 and CC12414 representatives also indicated a deletion in the CC1514 wciY gene, resulting in a predicted truncation of the putative glycerol phosphotransferase encoded by this gene. Notwithstanding the deletion, the CC1514 isolates were successfully serotyped by the Quellung reaction, indicating successful capsule production. Indeed, an in silico analysis of the predicted cps protein coding regions failed to identify a specific reaction catalyzed by the serotype 14 version of this protein [46].

e. Notwithstanding the deletion, the CC1514 isolates were successfully serotyped by the Quellung reaction, indicating successful capsule production. Indeed, an in silico analysis of the predicted cps protein coding regions failed to identify a specific reaction catalyzed by the serotype 14 version of this protein [46]. Given the overall capacity for cps locus recombination, the associations between genotypes and serotypes are puzzling, as is our evidence of a high level of cps sequence conservation within some CCs (eg, CC1917F and CC21812F). Perhaps there is some synergism between serotype and the genetic background of a strain that conveys an advantage to certain combinations over others. This sort of synergistic effect has been invoked to explain changes in the pneumococcal population in South Korea, where there was a prevaccine reduction of multidrug-resistant CC271/32019F pneumococci and replacement by similarly multidrug-resistant CC271/32019A pneumococci over several years [47]. It is also possible that genes, such as those encoding sugar-biosynthesis enzymes, outside the cps locus contribute to capsular expression and facilitate a genotype/serotype association [2].

stant CC271/32019F pneumococci and replacement by similarly multidrug-resistant CC271/32019A pneumococci over several years [47]. It is also possible that genes, such as those encoding sugar-biosynthesis enzymes, outside the cps locus contribute to capsular expression and facilitate a genotype/serotype association [2]. Our collection of pneumococci provided a unique opportunity to study evolution at the cps locus over approximately 70 years. Capsular switching with/without simultaneous pbp transfer has occurred regularly and prior to both PCV7 introduction and widespread antibiotic use. It is highly likely that the proliferation of newly generated NVT capsular switch variants will continue to be favored by PHiD-CV and PCV13 vaccination programs, as was the case after PCV7 implementation. Penicillin-nonsusceptible variants will have an even greater advantage. Although the magnitude of these selective forces relative to those favoring established genotype/VT associations remains unclear, the implementation of vaccine programs across the globe will most likely favor the intercontinental spread of NVT and penicillin-nonsusceptible pneumococci. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

rdjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments.We thank Dalia Denapaite, Irma Ochigava, and Shwan Rachid, for DNA sequencing and MLST analysis; Andries van Tonder, for retrieval of genomic sequence data from GenBank; and the clinicians, microbiologists, and investigators of the Active Bacterial Core surveillance program of the Emerging Infections Program Network, United States. Financial support.This work was supported by the Wellcome Trust (grant 083511/Z/07/Z to A. B. B. and grant 098051 to J. P. and S. D. B.), Stiftung Rheinland Pfalz für Innovation (to R. H.), and Ciber de Enfermedades Respiratorias (Instituto de Salud Carlos III, Madrid Spain [to J. L.]). A. B. B. is a Wellcome Trust Career Development Fellow. Potential conflicts of interest.K. P. K. received research funding from and is a consultant for Pfizer Vaccines and is a consultant for GlaxoSmithKline Biologicals and Merck. A. B. B. received research funding from GlaxoSmithKline Biologicals. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Respiratory syncytial virus (RSV) is the most important cause of viral acute lower respiratory tract infection in young children worldwide [1]. Two distinct antigenic groups, A and B, have been identified on the basis of reaction with monoclonal antibodies and nucleotide sequence data [2, 3]. Epitopes on the fusion (F) and attachment (G) glycoproteins are the targets of neutralizing antibodies [4], which correlate well with resistance to infection [5] and disease [6]. There is some evidence of population-level interaction between RSV A and B [7], suggesting temporal variations in population-level group-specific immunity.

e fusion (F) and attachment (G) glycoproteins are the targets of neutralizing antibodies [4], which correlate well with resistance to infection [5] and disease [6]. There is some evidence of population-level interaction between RSV A and B [7], suggesting temporal variations in population-level group-specific immunity. In addition to group-specific differences, the RSV G gene is known to undergo molecular evolution characterized by progressive accumulation of amino acid changes at an estimated rate of 0.25% of amino acids per year over the length of the protein [8]. When considered together with evidence of the greater rate of nonsynonymous-to-synonymous nucleotide substitutions [8], as well as the existence of positively selected sites on the attachment proteins of both RSV A and B [9], it is reasonable to speculate that changes within the G protein are immune driven. Recently, the emergence of a novel strain of RSV B with a 60-nucleotide duplication in the variable region of the G gene has been described (the BA genotype) [10]. Since it was first reported about 10 years ago, the BA genotype has progressed from relative novelty to becoming the most dominant genotype of RSV B globally [11]. The factors that underpin its remarkable epidemiological success have so far not been described. We hypothesized that the BA genetic change conferred a neutralization-resistance phenotype that permits BA genotype strains to escape previous host immunity, allowing for increased transmissibility in susceptible populations.

e factors that underpin its remarkable epidemiological success have so far not been described. We hypothesized that the BA genetic change conferred a neutralization-resistance phenotype that permits BA genotype strains to escape previous host immunity, allowing for increased transmissibility in susceptible populations. In the present study, we investigated RSV group–specific responses to both contemporary and historical test viruses, as well as the role of the recent BA genetic change in abrogating neutralizing responses generated against wild-type group B strains that did not have the duplication. MATERIALS AND METHODS Patients, Samples, and Gene Sequencing Nasal washings were obtained from children admitted to Kilifi District Hospital with severe or very severe pneumonia, for whom RSV infection was diagnosed on the basis of immunofluorescent antibody test results (Millipore). Multiplex reverse transcription polymerase chain reaction was used to determine whether the infecting virus was from group A or B [12]. An acute-phase serum sample was collected from all children at admission, and a convalescent-phase serum sample was obtained from RSV-positive patients approximately 4 weeks later. Ethics approval for the study was granted by the Kenya Medical Research Institute Ethical Review Committee. Further details about the study population, sampling procedures, diagnostic methods, and clinical findings have been published elsewhere [13]. Details on the age and sex distribution of study participants are shown in Supplementary Table 1.

y was granted by the Kenya Medical Research Institute Ethical Review Committee. Further details about the study population, sampling procedures, diagnostic methods, and clinical findings have been published elsewhere [13]. Details on the age and sex distribution of study participants are shown in Supplementary Table 1. The study used the following test viruses: A2 (RSV A; isolated in Australia in 1961), Kil/A/2006 (RSV A; Kenya, 2006), 8/60 (RSV B; Sweden, 1960), and Kil/B/2008 (RSV B; Kenya, 2008). Of the 2 RSV B test viruses, 8/60 did not have the 60-nucleotide G gene duplication, while Kil/B/2008 did. The G genes of infecting viruses were sequenced between nucleotide 284 on the G gene and nucleotide 9 on the F gene (GenBank accession numbers JX453211–JX453270), while their F genes were sequenced between nucleotides 121 and 918 of the F gene (GenBank accession numbers JX453271–JX453330).

cation, while Kil/B/2008 did. The G genes of infecting viruses were sequenced between nucleotide 284 on the G gene and nucleotide 9 on the F gene (GenBank accession numbers JX453211–JX453270), while their F genes were sequenced between nucleotides 121 and 918 of the F gene (GenBank accession numbers JX453271–JX453330). Plaque Assay and Microplaque Reduction and Neutralization Assay Virus titers were determined by plaque assay. Ten-fold dilutions of test virus were made in minimum essential medium (MEM) and inoculated onto HEp-2 cultures for 48 hours in 96-well plates. Cells were fixed in methanol, washed, and incubated at room temperature with a primary mouse anti-RSV immunoglobulin G (IgG) monoclonal antibody (Leica Microsystems), followed by a secondary horseradish peroxidase–linked rabbit anti mouse IgG (Dako, Denmark). Plaques were developed using aminoethylcarbazole. An enzyme-linked immunosorbent spot reader was used to count the number of plaques in each well. The plaque-reduction neutralization assay was performed by preparing serial 2-fold dilutions of sera in MEM. Fifty plaque-forming units of test virus were added to each dilution, and after incubation for 1 hour at room temperature, the material undergoing the neutralization reaction was inoculated onto HEp-2 cells and incubated at 37°C for 48 hours. Plaque development and enumeration were done as described above. Neutralizing antibody titers were calculated as the 50% neutralizing dose, using the Spearman-Karber method [14], and were expressed as plaque-reduction neutralization titers. These titers were normalized using log10 transformation, for statistical analyses. Seroconversion was defined as a ≥4-fold rise in the neutralizing antibody titer between the acute and convalescent phases of infection.

using the Spearman-Karber method [14], and were expressed as plaque-reduction neutralization titers. These titers were normalized using log10 transformation, for statistical analyses. Seroconversion was defined as a ≥4-fold rise in the neutralizing antibody titer between the acute and convalescent phases of infection. Study Design and Statistical Analyses Two arms of the study were designed to measure group- and strain-specific neutralizing antibody responses. In the first arm, serum neutralizing antibody responses to contemporary and historical group A and B test viruses were measured in the sera of infants naturally infected with contemporary RSV A and B. In the second arm of the study, 2 separate groups of children who were naturally infected with wild-type BA or non-BA viruses were used to investigate the effect of the 60-nucleotide duplication on the neutralizing response.

viruses were measured in the sera of infants naturally infected with contemporary RSV A and B. In the second arm of the study, 2 separate groups of children who were naturally infected with wild-type BA or non-BA viruses were used to investigate the effect of the 60-nucleotide duplication on the neutralizing response. Data analyses were done using Stata (version 11.1; StataCorp). Group- and cross-specific neutralizing antibody responses in convalescent-phase sera were analyzed using a multilevel modeling approach. Random individual-level effects were estimated using a 1-level random-effects multiple linear regression model. In this model, convalescent-stage titers were the dependent variable, while acute-stage titers, type of response (homologous/heterologous), and age were independent variables. Comparison of homologous and heterologous responses in different age classes was done by use of a linear regression model in which the dependent variable was the log-transformed rise in titer, and the independent variables were the test and infecting viruses. Proportions seroconverting to homologous and heterologous virus were compared using the McNemar χ2 test.

sponses in different age classes was done by use of a linear regression model in which the dependent variable was the log-transformed rise in titer, and the independent variables were the test and infecting viruses. Proportions seroconverting to homologous and heterologous virus were compared using the McNemar χ2 test. RESULTS Group Specificity of the RSV Neutralizing Response Comparison of homologous and heterologous neutralizing antibody responses in different age classes was performed by testing the difference between homologous and heterologous fold-rises in titer. Results of regression analysis showed that the mean homologous response to RSV A by RSV A–infected individuals was significantly greater than their heterologous response to RSV B in the 0–5-month age class (1.8-fold vs 0.5-fold rise in titer; P < .0001), the 6–11-month age class (11.2-fold vs 3.8-fold rise in titer; P = .002), and the ≥12-month age class (7.1-fold vs 2.2-fold rise in titer; P = .001). Similarly, the homologous response to RSV B by RSV B–infected individuals was significantly greater than their heterologous response to RSV A in the 0–5-month age class (2.7-fold vs 0.7-fold rise in titer; P < .0001), the 6–11-month age class (5.9-fold vs 1.7-fold rise in titer; P < .0001), and the ≥12-month age class (4.3-fold vs 0.9-fold rise in titer; P < .0001). These data are shown in Figure 1. Group homologous and heterologous responses were further classified in terms of the ability to seroconvert. As shown in Table 1, the proportion of individuals with homologous seroconversion to both RSV A and B was significantly greater than the proportion with heterologous seroconversion. Analysis of genetic similarity between infecting viruses was done using partial F and G gene sequences. In concurrence with previous reports, there was a high level of sequence diversity on the G gene and a high level of sequence conservation on the F gene, as shown in Supplementary Table 2. Table 1. Proportion of Infants Infected With Different Strains of Respiratory Syncytial Virus (RSV) Who Seroconverted to Different Test Viruses

ce with previous reports, there was a high level of sequence diversity on the G gene and a high level of sequence conservation on the F gene, as shown in Supplementary Table 2. Table 1. Proportion of Infants Infected With Different Strains of Respiratory Syncytial Virus (RSV) Who Seroconverted to Different Test Viruses Test viruses % of Seroconverted Infants Infecting viruses Kil/A/2006 Kil/B/2008 Pa Group A (n = 32) 50 12.5 .0005 Group B (n = 25) 8 40 .008 A2 8/60 Group A (n = 18) 28 0 .06 Group B (n = 20) 10 65 .001 Kil/B/2008 8/60 BA genotype (n = 20) 50 65 .4 Non-BA genotype (n = 20) 35 50 .4 Both genotypes (n = 40) 43 58 .1 Kil/A/2006 A2 Group A (n = 33) 51.5 39.4 .13 Kil/B/2008 (+C’) Kil/B/2008 (−C’) BA genotype (n = 10) 50 30 .32 Non-BA genotype (n = 10) 40 30 .32 8/60 (+C’) 8/60(−C’) BA genotype (n = 10) 50 60 .16 Non-BA genotype (n = 10) 40 40 1 a By the McNemar χ2 test, comparing differences in proportions of infants who seroconverted.

l/A/2006 A2 Group A (n = 33) 51.5 39.4 .13 Kil/B/2008 (+C’) Kil/B/2008 (−C’) BA genotype (n = 10) 50 30 .32 Non-BA genotype (n = 10) 40 30 .32 8/60 (+C’) 8/60(−C’) BA genotype (n = 10) 50 60 .16 Non-BA genotype (n = 10) 40 40 1 a By the McNemar χ2 test, comparing differences in proportions of infants who seroconverted. Figure 1. Comparison of the magnitude of the homologous and heterologous neutralizing response to both RSV A and B. The first letter in each panel heading denotes the group designation of the infecting virus, while the second letter denotes the group designation of the test virus. The grey diamond markers indicate the distribution of the acute-phase response and their corresponding means and 95% confidence intervals, while the open markers denote the distribution of convalescent-phase responses. The number above each acute/convalescent-phase pair denotes the mean fold-rise in titer from acute to convalescent phases of infection. Comparison of the magnitude of response (in terms of fold-rise in titer) to homologous virus and heterologous virus is shown by the long bars traversing the panels. The P value denotes whether the difference between the homologous and heterologous response in a particular age class is statistically significant. The dashed line indicates the lower limit of detection of neutralizing antibodies in this assay (defined as a plaque-reduction neutralization titer [PRNT] of <20)

e panels. The P value denotes whether the difference between the homologous and heterologous response in a particular age class is statistically significant. The dashed line indicates the lower limit of detection of neutralizing antibodies in this assay (defined as a plaque-reduction neutralization titer [PRNT] of <20) Effect of the BA Genetic Change and Temporal Evolution on the RSV Neutralizing Response There was no significant difference between the magnitude of the neutralizing response mounted by infants infected with non-BA strains to the 8/60 strain (3.7-fold rise in titer) and the Kil/B/2008 strain (3.42-fold rise in titer; P = .78). There was also no significant difference between the magnitude of the neutralizing response mounted by infants infected with BA strains to the 8/60 strain (5.13-fold rise in titer) and the Kil/B/2008 strain (3.54-fold rise in titer; P = .1). As shown in Table 1, no differences were found in terms of the ability to seroconvert to these 2 test viruses by either group. The effect of complement on G-specific neutralizing antibodies was investigated in a subset of infants from either group. The data presented in Table 1 show that even with the addition of complement, the proportion of infants who seroconverted to either test virus was similar, irrespective of whether the infecting group B strain contained the 60-nucleotide duplication. The effect of cumulative genetic change over approximately 45 years of RSV A evolution was tested using the sera of 33 RSV A–infected individuals. As shown in Table 1, there was no difference in the proportion who seroconverted to the A2 (1961) and Kil/A/2006 (2006) test viruses.

ontained the 60-nucleotide duplication. The effect of cumulative genetic change over approximately 45 years of RSV A evolution was tested using the sera of 33 RSV A–infected individuals. As shown in Table 1, there was no difference in the proportion who seroconverted to the A2 (1961) and Kil/A/2006 (2006) test viruses. DISCUSSION The results presented in this study show that the infant serum neutralizing response to RSV is significantly group specific. Homologous seroconversion rates were significantly greater than heterologous seroconversion rates for both RSV A and B. This pattern of homologous versus heterologous reactivity was similar irrespective of whether the test viruses were contemporary or historical. Analysis of the magnitude of homologous and heterologous neutralizing responses to RSV A and B at different ages showed that homologous responses were of significantly greater magnitude than heterologous responses, irrespective of age. In this study, we were unable to definitively characterize the group specificity of the RSV neutralizing response following secondary exposure, since there were only 7 children who were >2 years old (6 with RSV A and 1 with RSV B) and who could therefore be presumed to have been undergoing secondary infection. As a result, we were unable to determine whether the pattern of responses reported here remain imprinted on secondary exposure. The data presented support the idea that sequential alternation in the transmission of RSV A and B could be the result of population-level group-specific immunity. They further provide the basis to assert that the benefit of vaccination may be enhanced if representative strains from both RSV A and B are included in future vaccines. The data also show that, despite evidence of progressive evolution over 40–50 years, the RSV neutralizing response was not altered, suggesting that future RSV vaccines may retain effectiveness over long periods, without the need for repeated antigenic updates.

ains from both RSV A and B are included in future vaccines. The data also show that, despite evidence of progressive evolution over 40–50 years, the RSV neutralizing response was not altered, suggesting that future RSV vaccines may retain effectiveness over long periods, without the need for repeated antigenic updates. Analysis of the neutralizing responses of infants who underwent natural infection with group B strains that did not contain the 60-nucleotide duplication showed that their neutralizing responses to the 8/60 strain were no different from their responses to the Kil/B/2008 strain. The proportion who seroconverted to the 8/60 strain was not statistically different from the proportion who seroconverted to the Kil/B/2008 strain, suggesting that the BA mutation does not confer the ability to escape the neutralizing responses to non-BA variants. It is possible that the neutralizing responses reported here may have been predominantly directed at the more conserved F protein or, alternatively, at the conserved region of the G protein, thus masking strain-specific responses directed at the variable parts of the G protein. Previous work has shown that G protein–specific neutralizing responses are enhanced in the presence of complement [15], suggesting that the inability to detect a difference in neutralization could potentially be attributed to this fact. To address this concern, the neutralization assays were repeated using sera from a set of infants with wild-type BA and non-BA infections. The incorporation of complement in the neutralization assays did not alter the pattern of reactivity toward the test viruses. Overall, the results suggest that the increased prevalence of the BA genotype is not accounted for by a lower susceptibility to neutralization as measured in serum antibody to non-BA variants, and the basis for the success of this new variant remains to be explained.

r the pattern of reactivity toward the test viruses. Overall, the results suggest that the increased prevalence of the BA genotype is not accounted for by a lower susceptibility to neutralization as measured in serum antibody to non-BA variants, and the basis for the success of this new variant remains to be explained. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. This article is published with permission of the director of the Kenya Medical Research Institute. Financial support. This work was supported by the Wellcome Trust (grant 084633) and a Wellcome Trust PhD studentship (grant 083085 to C. J. S.). Potential conflict of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Children <5 years old have the highest burden of malaria and malaria-associated mortality in sub-Saharan Africa [1–4]. In these moderate-to-high transmission areas, the diagnosis of severe malaria is challenging. Parasitemic children with severe febrile illness can suffer from severe malaria but the parasitemia can also be coincidental, with an alternative illness causing severe disease. This is because partial immunity develops early in life in regions of high malaria endemicity, and malaria parasites can be tolerated without development of symptoms [5, 6]. Community-based cross-sectional studies conducted in these settings typically show that >10% of children <5 years old are parasitemic by microscopy yet symptom free, with the prevalence varying by age, exposure to infection, transmission season, and other factors [7–10]. Commonly used case definitions of malaria rely on the presence of fever and detection of malaria parasites on peripheral blood films and thus lack specificity. In addition, symptoms of severe malaria are nonspecific and can have different etiologies [11–13].

Children <5 years old have the highest burden of malaria and malaria-associated mortality in sub-Saharan Africa [1–4]. In these moderate-to-high transmission areas, the diagnosis of severe malaria is challenging. Parasitemic children with severe febrile illness can suffer from severe malaria but the parasitemia can also be coincidental, with an alternative illness causing severe disease. This is because partial immunity develops early in life in regions of high malaria endemicity, and malaria parasites can be tolerated without development of symptoms [5, 6]. Community-based cross-sectional studies conducted in these settings typically show that >10% of children <5 years old are parasitemic by microscopy yet symptom free, with the prevalence varying by age, exposure to infection, transmission season, and other factors [7–10]. Commonly used case definitions of malaria rely on the presence of fever and detection of malaria parasites on peripheral blood films and thus lack specificity. In addition, symptoms of severe malaria are nonspecific and can have different etiologies [11–13]. More-accurate case definitions for clinical or severe malaria are required for clinical management and research purposes. The specificity of a malaria case definition can be improved by using a parasite density threshold based on peripheral blood parasitemia [7, 14, 15]. This approach is useful as an epidemiological tool, but it lacks accuracy for clinical management. Peripheral blood parasitemia does not represent the sequestered parasite burden, which is pivotal to the pathophysiology of severe falciparum malaria. Asexual parasites in the second half of the erythrocytic stage of the life cycle effectively adhere to the endothelial lining of microcirculation vessels, which prevents detection of these parasites in peripheral blood films [16].

ed parasite burden, which is pivotal to the pathophysiology of severe falciparum malaria. Asexual parasites in the second half of the erythrocytic stage of the life cycle effectively adhere to the endothelial lining of microcirculation vessels, which prevents detection of these parasites in peripheral blood films [16]. Plasmodium falciparum histidine-rich protein 2 (PfHRP2) is a parasite-derived water-soluble protein and is released in discrete amounts into the plasma, predominantly during schizont rupture [17]. Released PfHRP2 is distributed over the plasma volume and, therefore, the PfHRP2 concentration in plasma reflects the total body parasite burden, including the sequestered parasites. Studies involving Asian adults [18, 19] and African children [20, 21] show that, in contrast with the peripheral blood parasite density, the plasma PfHRP2 concentration correlates strongly with disease severity and outcome. We hypothesized that the plasma PfHRP2 concentration, as a measure of the total parasite burden determining disease severity, can be used to define malaria-attributable disease in malaria-endemic regions where coincidental peripheral blood parasitemia is common. In this study, we compared the distribution of peripheral blood parasitemia versus the plasma PfHRP2 concentration in healthy, rapid diagnostic test (RDT)-negative controls, in asymptomatic carriers, and in patients with uncomplicated or severe malaria and used this to estimate the malaria-attributable fraction of severe disease.

dy, we compared the distribution of peripheral blood parasitemia versus the plasma PfHRP2 concentration in healthy, rapid diagnostic test (RDT)-negative controls, in asymptomatic carriers, and in patients with uncomplicated or severe malaria and used this to estimate the malaria-attributable fraction of severe disease. METHODS The study was conducted in the rural lowlands of northeastern Tanzania. Peripheral blood slides and plasma PfHRP2 samples were collected in 1 community-based and 2 hospital-based studies in the neighboring districts of Handeni and Muheza, Tanga Region, that have similar intensities of malaria transmission [9, 22]. Four clinical severity groups were defined: patients with severe malaria, patients with uncomplicated malaria, asymptomatic carriers, and healthy control subjects with negative results of an RDT.

METHODS The study was conducted in the rural lowlands of northeastern Tanzania. Peripheral blood slides and plasma PfHRP2 samples were collected in 1 community-based and 2 hospital-based studies in the neighboring districts of Handeni and Muheza, Tanga Region, that have similar intensities of malaria transmission [9, 22]. Four clinical severity groups were defined: patients with severe malaria, patients with uncomplicated malaria, asymptomatic carriers, and healthy control subjects with negative results of an RDT. Cases of severe malaria were identified in patients from the hospital-based studies, using modified clinical World Health Organization criteria that were confirmed by positive results of a parasite lactate dehydrogenase (pLDH) RDT (OptiMAL-IT, DiaMed, Switzerland) and/or PfHRP2-based RDT (Paracheck, Orchid Biomedical, India). Severity criteria included decreased consciousness (coma or severe prostration), convulsions, respiratory distress or acidotic breathing, shock, severe symptomatic anemia (hemoglobin concentration <5 g/dL), and hypoglycemia (glucose concentration <2.5 mmol/L) [23]. Children with uncomplicated malaria, asymptomatic carriers, and healthy controls were identified in the community-based study on the basis of results of a pLDH-based RDT (CareStart, Access Bio, United States). Uncomplicated malaria was defined by fever (axillary temperature ≥37.5°C), absence of severity criteria, and a positive result of a pLDH-based RDT [24, 25]. Asymptomatic carriers were defined as afebrile children (on the basis of their clinical history and an axillary temperature of <37.5°C at presentation) with a positive result of a pLDH-based RDT. Controls were afebrile children with negative results of a pLDH-based RDT.

d a positive result of a pLDH-based RDT [24, 25]. Asymptomatic carriers were defined as afebrile children (on the basis of their clinical history and an axillary temperature of <37.5°C at presentation) with a positive result of a pLDH-based RDT. Controls were afebrile children with negative results of a pLDH-based RDT. In the community-based study, asymptomatic children were recruited between February and August 2008 in the context of the baseline screening for a randomized trial that assessed the effect of micronutrient supplementation on the incidence of uncomplicated malaria [25]. In 4 villages in Handeni District, all resident children aged 6–60 months were invited for the screening, and those with a height-for-age z score of ≤−1.5 SD, a weight-for-height z score of ≥−3 SDs, and a hemoglobin concentration of ≥7 g/dL were eligible to participate. Those who were unlikely to comply with interventions, whose parents/guardians refused to provide consent, or who had signs of severe or chronic disease on clinical examination were excluded.

core of ≤−1.5 SD, a weight-for-height z score of ≥−3 SDs, and a hemoglobin concentration of ≥7 g/dL were eligible to participate. Those who were unlikely to comply with interventions, whose parents/guardians refused to provide consent, or who had signs of severe or chronic disease on clinical examination were excluded. In total, 246 of 612 children had a plasma sample and a RDT positive for P. falciparum. Of these, 177 were afebrile on examination and reported absence of fever within the past 48 hours. Slide results were available for 172 asymptomatic individuals, who were included in the present study (termed “group 2”). All parasitemic children at baseline were treated with an effective antimalarial (artemether-lumefantrine). We selected the first 60 consecutively enrolled RDT-negative children as controls (termed “group 1”), of whom 11 were subsequently excluded because they had a history of or current fever. Uncomplicated malaria cases were detected during the follow-up period of the trial. Parents were requested to bring study children to the clinic if their child developed a fever or became unwell. Of these, 285 randomly selected febrile children with a positive result of a pLDH-based RDT (termed “group 3”) were included in the analysis (Figure 1). Figure 1. Selection of study subjects. Abbreviations: Pf, Plasmodium falciparum; PfHRP2, P. falciparum histidine-rich protein 2; pLDH, parasite lactate dehydrogenase; RDT, rapid diagnostic test.

cted febrile children with a positive result of a pLDH-based RDT (termed “group 3”) were included in the analysis (Figure 1). Figure 1. Selection of study subjects. Abbreviations: Pf, Plasmodium falciparum; PfHRP2, P. falciparum histidine-rich protein 2; pLDH, parasite lactate dehydrogenase; RDT, rapid diagnostic test. Severely ill parasitemic patients originated from 2 consecutive studies conducted at Teule Hospital (Muheza, Tanzania). The details of these studies have been published elsewhere [26, 27]. The first study assessed the causes of fever in 3639 febrile children admitted from June 2006 through May 2007 [26]. From this cohort, patients who had pathogens isolated by blood culture and a RDT positive for falciparum malaria, plus a random sample of patients with RDT-positive severe malaria but a negative blood culture result, were included in the analysis (termed “group 4”; n = 226). The second severe malaria group (termed “group 5”) was part of a severe malaria treatment trial (AQUAMAT) conducted from February 2007 through July 2010 (n = 703) [27]. These subjects were also part of a separate report describing the prognostic value of plasma PfHRP2 levels among 3826 children across all AQUAMAT study sites [21].

evere malaria group (termed “group 5”) was part of a severe malaria treatment trial (AQUAMAT) conducted from February 2007 through July 2010 (n = 703) [27]. These subjects were also part of a separate report describing the prognostic value of plasma PfHRP2 levels among 3826 children across all AQUAMAT study sites [21]. All 3 studies were approved by the Tanzania Medical Research Coordinating Committee. The community-based study was also approved by the Ethical Review Committee of Wageningen University. The hospital-based studies were also approved by the London School of Hygiene and Tropical Medicine and the Oxford Tropical Research Ethics Committee. In all studies, written informed consent was obtained from parents or guardians of each participating child. Experienced microscopists at the National Institute of Medical Research Tanga laboratory in Korogwe, Teule Hospital (Joint Malaria Programme), and the Mahidol-Oxford Tropical Medicine Research Unit in Bangkok read the malaria slides; the latter institution was also responsible for quality control. Parasitemia (parasites/µL) was calculated from the thick film per 200 white blood cells (WBCs) and the actual WBC count or, if missing, assuming 8000 WBC/µL (count/200 WBC × 40) [28]. In the AQUAMAT study, parasitemia was calculated from thin film per 1000 red blood cells (RBCs) (count/1000 RBCs × 125.6 × Hct) [29, 30].

arasitemia (parasites/µL) was calculated from the thick film per 200 white blood cells (WBCs) and the actual WBC count or, if missing, assuming 8000 WBC/µL (count/200 WBC × 40) [28]. In the AQUAMAT study, parasitemia was calculated from thin film per 1000 red blood cells (RBCs) (count/1000 RBCs × 125.6 × Hct) [29, 30]. Plasma PfHRP2 was assessed from freeze-thawed ethylenediaminetetraacetic acid plasma samples by a commercial sandwich enzyme-linked immunosorbent assay (ELISA) kit (Celisa, Cellabs, Sydney, Australia), according to the manufacturer's instructions, with minor modifications [18]. Reference plasma with a known PfHRP2 concentration was used to construct standard curves. Concentrations in diluted plasma dilutions were determined in duplicate according to the linear segment of the standard curve. Positive cases were defined as those in which duplicate derived concentrations were in agreement (ratio, 0.5–2) and the optical density relative to background was >3 SDs of the average background based on all plates. Statistical Analysis Data were analyzed with Stata, version 12 (StataCorp, United States). Parasite counts and PfHRP2 concentrations were normalized by log10 transformation. Normally distributed or log10-normalized variables were compared using a Student t test, and the remainder were compared by the Wilcoxon rank sum test. PfHRP2 concentrations between blood-culture-positive patients and blood-culture-negative patients were compared according to PfHRP2 quintiles for patients with severe malaria (groups 4 and 5).

log10-normalized variables were compared using a Student t test, and the remainder were compared by the Wilcoxon rank sum test. PfHRP2 concentrations between blood-culture-positive patients and blood-culture-negative patients were compared according to PfHRP2 quintiles for patients with severe malaria (groups 4 and 5). Modeling PfHRP2 Concentrations According to Diagnostic Group Analysis of the observed PfHRP2 concentrations suggested distinctive distributions according to severity of P. falciparum infection (Figure 2). In addition, the PfHRP2 concentrations observed in patients with clinically defined severe malaria suggested contributions of underlying plasma PfHRP2 distributions, as observed in RDT-negative controls, asymptomatic carriers, and patients with uncomplicated malaria (Figure 2), all representing severe illness with alternative causes. It was assumed that each diagnostic group (k) had a distinctive Weibull distribution of plasma PfHRP2 concentrations and that the observed plasma PfHRP2 distribution in the different clinical groups (j) was a composite of these Weibull distributions. The diagnostic groups (k) consisted of healthy controls (k = 1), asymptomatic carriers (k = 2), patients with uncomplicated malaria (k = 3), and patients with severe malaria (k = 4). The diagnostic groups of uncomplicated and severe malaria, in contrast with the clinically defined groups, exclude patients with coincidental parasitemia. A mechanistic model was constructed to infer the most likely Weibull distributions in each diagnostic group (k), described by the coefficients αk and βk. The probability (P) that an individual (i) has a particular plasma PfHRP2 concentration (P[hji]) is then determined by the probability (mjk) that this individual belongs to diagnostic group k. Figure 2. Frequency distributions of peripheral blood parasitemia, plasma Plasmodium falciparum histidine-rich protein 2 (PfHRP2) concentrations, and modeled fitted PfHRP2, according to malaria clinical group (1 = healthy rapid diagnostic test [RDT]–negative controls, 2 = asymptomatic carriers, 3 = uncomplicated malaria, 4 = severe malaria, 5 = severe malaria).

eral blood parasitemia, plasma Plasmodium falciparum histidine-rich protein 2 (PfHRP2) concentrations, and modeled fitted PfHRP2, according to malaria clinical group (1 = healthy rapid diagnostic test [RDT]–negative controls, 2 = asymptomatic carriers, 3 = uncomplicated malaria, 4 = severe malaria, 5 = severe malaria). The fitted PfHRP2 distributions (right column) show the modeled PfHRP2 distributions with the underlying contributing PfHRP2 distributions of different diagnostic groups (dotted lines), composed of RDT-negative controls (light green), asymptomatic carriers (green), and patients with uncomplicated malaria (blue turquoise) or severe malaria (bright blue and purple).

show the modeled PfHRP2 distributions with the underlying contributing PfHRP2 distributions of different diagnostic groups (dotted lines), composed of RDT-negative controls (light green), asymptomatic carriers (green), and patients with uncomplicated malaria (blue turquoise) or severe malaria (bright blue and purple). Two different groups with clinical severe malaria were included in the model, of which one was partly selected on the basis of the concomitant presence of bacteremia (group 4; see above). The model was used to define the plasma PfHRP2-based malaria-attributable fraction in the unselected group of parasitemic patients with a clinical diagnosis of severe malaria (group 5). It differentiates severe malaria from the population with asymptomatic parasitemia and the population with uncomplicated malaria, who have severe disease of a different origin. The proportion of malaria-attributable disease (y), according to PfHRP2 concentration (h), is given by that is, for each value of plasma PfHRP2 concentration (h), the malaria-attributable fraction of severe disease is m54W(α5,β5) divided by the total number of individuals with the same PfHRP2 concentration (h) predicted by the model as The parameters were estimated by implementing a mixture model within WinBUGS [30]. Three chains were run, for a burn-in of 5000 iterations, followed by a further 5000 iterations to obtain posterior distributions. The model parameters were estimated with 95% credible intervals (CIs). Sensitivity was calculated using the model-derived number of patients with severe malaria as a reference.

WinBUGS [30]. Three chains were run, for a burn-in of 5000 iterations, followed by a further 5000 iterations to obtain posterior distributions. The model parameters were estimated with 95% credible intervals (CIs). Sensitivity was calculated using the model-derived number of patients with severe malaria as a reference. RESULTS Subject Characteristics We analyzed data from 49 healthy RDT-negative controls (group 1), 172 children with asymptomatic parasitemia (group 2), 285 patients with uncomplicated malaria (group 3), and 226 patients (group 4) and 703 patients (group 5) with clinical severe malaria (Figure 1). Microscopy findings were negative in all RDT-negative controls, except for 1 child (145 parasites/µL). Baseline clinical and laboratory characteristics according to malaria clinical group are summarized in Table 1. Children with severe malaria were younger than children with uncomplicated malaria (P < .0001) or asymptomatic parasitemia (P < .0001) and also had lower hemoglobin concentrations (P < .0001). Admission characteristics and outcomes of patients with severe malaria (groups 4 and 5) are summarized in Table 2. Table 1. Baseline Characteristics of the Study Population, According to Malaria Clinical Group

(P < .0001) or asymptomatic parasitemia (P < .0001) and also had lower hemoglobin concentrations (P < .0001). Admission characteristics and outcomes of patients with severe malaria (groups 4 and 5) are summarized in Table 2. Table 1. Baseline Characteristics of the Study Population, According to Malaria Clinical Group Group 1: RDT-Negative Controls Group 2: Asymptomatic Carriers Group 3: Uncomplicated Malaria Group 4: Severe Malaria Group 5: Severe Malaria Characteristic (n = 49) (n = 172) (n = 285) (n = 226) (n = 703) Female sex 20 (41) 91 (53) 141 (49) 125 (55) 339 (48) Age, y 2.3 (1.5–3.6) 3.2 (2.3–4.1) 2.8 (1.9–4.0) 1.7 (1.1–2.6) 2.2 (1.2–3.1) Weight-for-age z scorea −1.6 ± 0.7 −1.6 ± 0.7 NA −1.5 ± 1.1 −1.1 ± 1.2 Temperature, °C 36.4 ± 0.4 36.5 ± 0.4 37.9 ± 1.3 38.0 ± 1.1 38.1 ± 1.0 Hemoglobin concentration,b g/dL 11.3 (10.4–11.9) 10.3 (9.3–11.2) 9.8 (8.9–10.8) 4.8 (3.7–6.4) 6.5 (4.4–8.2) Slide positive for P. falciparum 1 (2.0) 118 (68.6) 275 (96.5) 208 (92.0) 701 (99.7) Parasitemia, parasites/µL Geometric mean (95% CI) 145 1602 (1189–2157) 29 836 (24 390–36 498) 28 187 (22 312–35 607) 46 619 (39 476–55 054) Range … 19–35471 96–1 448 094 221–626 431 16–1 375 069 PfHRP2 concentration,c ng/mL Geometric mean (95% CI) 4 (1–11) 19 (15–23) 163 (137–194) 1510 (1180–1933) 1746 (1577–1934) Range 1–29 1–546 3–4343 4–87 199 5–56 818 Data are no. (%) of children, mean ± SD, or median (interquartile range), unless otherwise specified.

Range … 19–35471 96–1 448 094 221–626 431 16–1 375 069 PfHRP2 concentration,c ng/mL Geometric mean (95% CI) 4 (1–11) 19 (15–23) 163 (137–194) 1510 (1180–1933) 1746 (1577–1934) Range 1–29 1–546 3–4343 4–87 199 5–56 818 Data are no. (%) of children, mean ± SD, or median (interquartile range), unless otherwise specified. Abbreviations: CI, credible interval; NA, not available; P. falciparum, Plasmodium falciparum; PfHRP2, P. falciparum histidine-rich protein 2; RDT, rapid diagnostic test. a Data were missing for 7 children in group 4 and 1 child in group 5. b Data were missing for 1 child in group 1, 3 children in group 2, and 2 children in group 5. c Data are for individuals with detectable concentrations (8 in group 1, 156 in group 2, 269 in group 3, 222 in group 4, and 698 in group 5). Table 2. Admission Characteristics and Outcomes of Children With Severe Malaria Variable Group 4 (n = 226) Group 5 (n = 703) Coma (BCS ≤2 or GGS ≤10) 30 (13) 213 (30) Prostration (inability to sit) 106 (47) 403 (57) Convulsions (≥2 within 24 h) 40 (18) 268 (38) Severe anemia (hemoglobin concentration <5 g/dL) 128 (57) 221 (32) Hypoglycemia (glucose concentration <2.5 mmol/L) 27 (12) 145 (21) Acidosis (lactate concentration >5 mmol/L or base excess ≤−8 mmol/L)a 97 (43) 314 (49) Respiratory distressb 74 (33) 131 (19) Shockc 21 (9) 111 (16) Blood culture positivityd 47 (20.8) 36 (5.1) Mortality 31 (13.7) 99 (14.1) Data are no. (%) of children. Abbreviations: BCS, Blantyre coma scale; GCS, Glasgow coma scale. a Data were missing for 29 children in group 4 and 61 children in group 5.

Variable Group 4 (n = 226) Group 5 (n = 703) Coma (BCS ≤2 or GGS ≤10) 30 (13) 213 (30) Prostration (inability to sit) 106 (47) 403 (57) Convulsions (≥2 within 24 h) 40 (18) 268 (38) Severe anemia (hemoglobin concentration <5 g/dL) 128 (57) 221 (32) Hypoglycemia (glucose concentration <2.5 mmol/L) 27 (12) 145 (21) Acidosis (lactate concentration >5 mmol/L or base excess ≤−8 mmol/L)a 97 (43) 314 (49) Respiratory distressb 74 (33) 131 (19) Shockc 21 (9) 111 (16) Blood culture positivityd 47 (20.8) 36 (5.1) Mortality 31 (13.7) 99 (14.1) Data are no. (%) of children. Abbreviations: BCS, Blantyre coma scale; GCS, Glasgow coma scale. a Data were missing for 29 children in group 4 and 61 children in group 5. b Defined as nasal alar flaring, costal indrawing, use of accessory muscles, or severe tachypnea. c Compensated shock (capillary refill time of ≥3 seconds or presence of a temperature gradient with systolic blood pressure of ≥70 mm Hg) and decompensated shock (systolic blood pressure of <70 mmHg) combined. d Data were missing for 3 children in group 5.

b Defined as nasal alar flaring, costal indrawing, use of accessory muscles, or severe tachypnea. c Compensated shock (capillary refill time of ≥3 seconds or presence of a temperature gradient with systolic blood pressure of ≥70 mm Hg) and decompensated shock (systolic blood pressure of <70 mmHg) combined. d Data were missing for 3 children in group 5. PfHRP2 concentrations were detectable in 8 of 49 healthy pLDH negative controls (16%), 156 of 172 asymptomatic patients (91%), 269 of 285 patients with uncomplicated malaria (94%), and 222 of 226 patients (98%; group 4) and 698 of 703 patients (99%; group 5) with severe malaria (Table 1). The distributions of peripheral blood parasitemia and PfHRP2 concentrations according to clinical groups are displayed in Figure 2. Plasma PfHRP2 concentrations were associated with the severity of P. falciparum infection, whereas peripheral blood parasitemia was not.

ients (99%; group 5) with severe malaria (Table 1). The distributions of peripheral blood parasitemia and PfHRP2 concentrations according to clinical groups are displayed in Figure 2. Plasma PfHRP2 concentrations were associated with the severity of P. falciparum infection, whereas peripheral blood parasitemia was not. Plasma PfHRP2-Based Malaria-Attributable Disease in Parasitemic Severe Febrile Illness The observed PfHRP2 distributions in the clinical groups were modeled as a composite of the PfHRP2 distributions of the contributing diagnostic groups (Figure 2).The model-derived parameter estimates for mjk denoting the probability that an individual from clinical group j = 1–5 belongs to diagnostic groups k = 1–4 are given in the Supplementary Materials. From these parameter estimates, the predicted distributions were fitted to the observed distributions and used to derive malaria-attributable proportions according to the log10 plasma PfHRP2 concentration in the unselected clinical group of severely ill parasitemic children (group 5; Figure 3). This shows that PfHRP2 levels of >1000 ng/mL correspond to a malaria-attributable fraction of 99% (95% CI, 96%–100%), with a sensitivity of 74% (95% CI, 72%–77%). The proportion of malaria-attributable disease declined at lower PfHRP2 concentrations. Below 200 ng/mL, an alternative diagnosis than malaria was suggested in >10% (95% CI, 3%–27%) of patients, whereas this proportion increased to >50% (95% CI, 31%–67%) at concentrations of <50 ng/mL. Figure 3. Malaria-attributable proportion (left axis) and sensitivity (median, 95% credible interval; right axis) for severe disease, according to log10 plasma Plasmodium falciparum histidine-rich protein 2 (PfHRP2) concentration. The malaria-attributable proportion was derived from the predicted PfHRP2 distributions from the median (95% credible interval) values of the mij distributions of individuals in each malaria diagnostic group (see Supplementary Materials).

g10 plasma Plasmodium falciparum histidine-rich protein 2 (PfHRP2) concentration. The malaria-attributable proportion was derived from the predicted PfHRP2 distributions from the median (95% credible interval) values of the mij distributions of individuals in each malaria diagnostic group (see Supplementary Materials). Blood Cultures Blood culture results were positive among 83 patients with severe malaria (Table 2), and based on the sample selection criteria, this proportion was higher in group 4. Patients with a positive blood culture result were overrepresented in the lowest and highest plasma PfHRP2 quintiles (Figure 4). Of 90 patients with a PfHRP2 concentration below the threshold of 200 ng/mL, 16 (18%) had positive blood culture results. Figure 4. Blood culture positivity, according to plasma Plasmodium falciparum histidine-rich protein 2 (PfHRP2) quintile, in patients with severe malaria. Gram-positive bacteria included Streptococcus pneumonia, Staphylococcus aureus, β-hemolytic Streptococcus. Other gram-negative bacteria included Haemophilus influenza (type b), unspecified gram-negative rods, Salmonella typhi, Acinetobacter baumannii, Burkholderia cepacia, Kingella kingae, Neisseria species, Pseudomonas oryzihabitans, and Pasteurella species. Gram-negative bacteria included Salmonella species, Escherichia coli, Enterobacter cloacae, and Klebsiella species. Contaminants included Micrococcus species, Bacillus species, coagulase-negative Staphylococcus, yeast, Corynebacterium species (diphtheroids), unspecified gram-positive rods, mixed bacterial species, Ralstonia pickettii, α-hemolytic Streptococcus viridans, Sphingomonas paucimobilis, Pseudomonas stutzeri, Chryseomonas luteola, and Stenotrophomonas maltophilia.

, Bacillus species, coagulase-negative Staphylococcus, yeast, Corynebacterium species (diphtheroids), unspecified gram-positive rods, mixed bacterial species, Ralstonia pickettii, α-hemolytic Streptococcus viridans, Sphingomonas paucimobilis, Pseudomonas stutzeri, Chryseomonas luteola, and Stenotrophomonas maltophilia. DISCUSSION This study shows a clear stepwise increase in plasma PfHRP2 concentrations according to disease severity from asymptomatic parasitemia, to uncomplicated malaria, to severe malaria. There was substantially less overlap in the distributions of plasma PfHRP2 concentrations between groups as compared to the distributions of peripheral blood parasitemia. The distinct distributions between diagnostic groups enabled us to model the proportion of malaria-attributable disease on the basis of plasma PfHRP2 level at admission and to distinguish this from patients with coincidental peripheral blood parasitemia in whom severe disease is caused by an alternative disease. The PfHRP2-based model performed better than a previously described model that was based on peripheral blood parasitemia [15]. The proportion of malaria-attributable disease dropped to <50%, with a sensitivity of >99% at plasma PfHRP2 concentrations of <50 ng/mL, in which case additional diagnostic tests are indicated to identify alternative diseases. The current model also accurately identified patients with a very high probability of severe malaria, with acceptable sensitivity. A threshold of 1000 ng/mL defined a population of patients with severe malaria not diluted by patients with coincidental parasitemia (<1%), which is mainly useful for defining a study population in a research setting, but is also useful for the treating clinician. A low plasma PfHRP2 concentration in a parasitemic patient with severity signs should not result in withholding of treatment with antimalarials, but should prompt the treating physician to look for other possible diseases, depending on the clinical presentation and resources (eg, blood culture, lumbar puncture, chest radiograph, and computed tomography of the cerebrum). In African settings, were diagnostic facilities are scarce and treatment stock-outs occur, the PfHRP2 concentration can also help to prioritize these resources.

diseases, depending on the clinical presentation and resources (eg, blood culture, lumbar puncture, chest radiograph, and computed tomography of the cerebrum). In African settings, were diagnostic facilities are scarce and treatment stock-outs occur, the PfHRP2 concentration can also help to prioritize these resources. A previous study reported the strong prognostic significance of plasma PfHRP2 level for death in a large cohort of African children with severe malaria and modelled the malaria attributable fraction in fatal cases [21]. The current study enabled a more accurate definition of the probability of nonmalarial disease at low plasma PfHRP2 concentrations by incorporating children with asymptomatic parasitemia and uncomplicated malaria. It is reassuring that the identified plasma PfHRP2 thresholds denoting high or low probabilities of alternative disease were highly consistent between these studies, which used different modeling techniques.

plasma PfHRP2 concentrations by incorporating children with asymptomatic parasitemia and uncomplicated malaria. It is reassuring that the identified plasma PfHRP2 thresholds denoting high or low probabilities of alternative disease were highly consistent between these studies, which used different modeling techniques. Our findings are supported by 2 recent studies involving African children. A small study among Tanzanian children showed a mean PfHRP2 value of 1008 ng/mL in patients with cerebral malaria, compared with a PfHRP2 concentration of 443 ng/mL in patients with uncomplicated malaria [20]. The diagnostic potential of the plasma PfHRP2 concentration in pediatric cerebral malaria was also confirmed in a Malawian study, where the presence of malarial retinopathy was used as the reference test [32]. In contrast, 2 other studies in moderate-to-high transmission settings reported that the PfHRP2 concentration does not reflect severity in children. In Papuan children, the median PfHRP2 concentrations in uncomplicated and severe malaria were similar (584 vs 456 ng/mL) [33]. However, the case-fatality rate in the severe malaria group was <1%, suggesting moderately severe malaria in accordance with the low PfHRP2 concentrations reported. A small study involving 22 Kenyan children with severe malaria reported low median PfHRP2 concentrations of 63 ng/mL with absence of decay over 48 hours, which could be related to problems in the PfHRP2 assay [34].

was <1%, suggesting moderately severe malaria in accordance with the low PfHRP2 concentrations reported. A small study involving 22 Kenyan children with severe malaria reported low median PfHRP2 concentrations of 63 ng/mL with absence of decay over 48 hours, which could be related to problems in the PfHRP2 assay [34]. The prognostic usefulness of plasma PfHRP2 concentration is in line with previous reports involving adult populations. A study among Thai adults showed a similar stepwise increase in plasma PfHRP2 levels that was associated with disease severity [18]. In Indonesian adults, the mean PfHRP2 value among patients with severe malaria was 1863 ng/mL, compared with 314 ng/mL among patients with moderately severe malaria [19]. In both studies, the plasma PfHRP2 concentration was prognostic for a fatal outcome.

ma PfHRP2 levels that was associated with disease severity [18]. In Indonesian adults, the mean PfHRP2 value among patients with severe malaria was 1863 ng/mL, compared with 314 ng/mL among patients with moderately severe malaria [19]. In both studies, the plasma PfHRP2 concentration was prognostic for a fatal outcome. This is the first study to assess PfHRP2 concentrations in healthy asymptomatic children in a moderate-to-high transmission area. Parasite densities that can be tolerated without causing symptoms vary substantially between individuals of different age groups, transmission intensities, and seasons [10, 15, 35, 36]. In moderate-to-high transmission settings, children aged <5 years represent a heterogeneous group with regard to levels of immunity. This is reflected by the younger age of children with severe malaria and by the older age of asymptomatic children, of whom 13 of 172 (8%) had parasite densities of >10 000 parasites/µL. Similarly high parasite densities have been reported in cross-sectional surveys in settings with moderate-to-high malaria transmission [15, 35]. The accuracy of PfHRP2 concentration thresholds for defining malaria-attributable disease will vary with the level of acquired immunity in the population, because this factor determines the relative sizes of populations with asymptomatic parasitemia, compared with populations with uncomplicated or severe malaria, and thus determines the corresponding overlap of plasma PfHRP2 distributions. The model prediction as a function of transmission intensity will be explored in a separate study. In addition, the prevalence of bacteremia will also affect the size of the population of individuals who have asymptomatic parasite infections or uncomplicated malaria but present with severe illness. Indeed, in the current study, selection of patients with a positive blood culture result (group 4) resulted in a relatively higher proportion of parasitemic patients with severe illness due to diseases other than malaria.

ho have asymptomatic parasite infections or uncomplicated malaria but present with severe illness. Indeed, in the current study, selection of patients with a positive blood culture result (group 4) resulted in a relatively higher proportion of parasitemic patients with severe illness due to diseases other than malaria. Detection of malarial retinopathy by fundoscopy is an alternative diagnostic tool that has been evaluated for its ability to distinguish children with cerebral malaria from encephalopathic children with coincidental parasitemia [37–40]. In the African setting, this has only been evaluated in comatose patients and requires considerable expertise and training and appropriate equipment. In comparison, the plasma PfHRP2 concentration is positively associated with the entire clinical severity spectrum of P. falciparum infection. In this study, plasma PfHRP2 was assessed by a quantitative ELISA. Our findings call for the development of a low-cost semiquantitative rapid test for the detection of plasma PfHRP2 at suitable thresholds.

HRP2 concentration is positively associated with the entire clinical severity spectrum of P. falciparum infection. In this study, plasma PfHRP2 was assessed by a quantitative ELISA. Our findings call for the development of a low-cost semiquantitative rapid test for the detection of plasma PfHRP2 at suitable thresholds. Positive blood culture results, particularly those for gram-negative organisms, were overrepresented among patients within the lowest and highest PfHRP2 concentration quintiles. Blood cultures are known to have a limited sensitivity (around 40%) for detecting bacteremia [41]. The actual number of bacteremic patients could thus be 2.5-fold higher than detected, implying an actual proportion of bacteremic patients close to 50% among patients with a plasma PfHRP2 concentration of <200 ng/mL (2.5 times the observed proportion of 18%). This would be consistent with results from a Malawian autopsy series, in which invasive bacterial infection was reported as the cause of death in 4 of 7 parasitemic patients (64%) with an alternative diagnosis [42]. Positive blood culture results for patients with high PfHRP2 concentrations indicate concomitant bacteremia during severe malaria. There are several mechanisms that may explain this high rate of concomitant bacteremia, including a reduction in gut barrier function due to intense sequestration [43], which facilitates translocation of gut bacteria, or general immunosuppression due to macrophagocytic dysfunction induced by hemozoin and heme-oxygenase 1 [44–46]. Severe malarial anemia is particularly associated with invasive bacterial disease, mainly non-typhi Salmonella bacteremia [47]. P. falciparum infection predisposes to gram-negative bacteremia and can account for more than half of invasive bacterial disease in malaria-endemic areas [48]. Our data show that bacteremia contributes to severe illness but also occurs concomitantly in patients with severe malaria, warranting the use of broad-spectrum antibiotics in addition to prompt antimalarial treatment, preferably with parenteral artesunate.

lf of invasive bacterial disease in malaria-endemic areas [48]. Our data show that bacteremia contributes to severe illness but also occurs concomitantly in patients with severe malaria, warranting the use of broad-spectrum antibiotics in addition to prompt antimalarial treatment, preferably with parenteral artesunate. The current study has several limitations. This is a retrospective analysis of pooled data sets. Patients with severe malaria in group 5 were also included in a previous publication on the prognostic value of PfHRP2 concentration. Patients with severe malaria in group 4 were partly selected on the basis of blood culture positivity. However, patients were selected on the basis of clinical criteria and RDT results and not on the basis of PfHRP2 concentrations, and the PfHRP2 distributions in both severe malaria groups were similar. In patients with low parasitemia, the sensitivities of the peripheral blood slide and the RDT are relatively low, which could have affected the composition of the clinical groups.

eria and RDT results and not on the basis of PfHRP2 concentrations, and the PfHRP2 distributions in both severe malaria groups were similar. In patients with low parasitemia, the sensitivities of the peripheral blood slide and the RDT are relatively low, which could have affected the composition of the clinical groups. In conclusion, our study shows that the plasma PfHRP2 concentration can be used to estimate the proportion of malaria-attributable disease in African children in moderate-to-high transmission settings and can distinguish severe malaria from severe febrile illness with coincidental peripheral blood parasitemia. Bacteremia is prominent among patients with severe illness and low plasma PfHRP2 concentrations, suggesting that malaria may not be their primary diagnosis. Bacteremia is also more frequent among patients with high plasma PfHRP2 concentrations, denoting concomitant sepsis with severe malaria, which implies that administration of antibiotics is warranted for all patients with a clinical diagnosis of severe malaria. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

rdjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Notes Acknowledgments We thank Tedson Lukindo, from the Joint Malaria Programme–Tanzania, for assistance with the ELISA; Benjamas Intharabut, Ketsanee Srinamon, and Forradee Nuchsongsin, from the Mahidol-Oxford Tropical Medicine Research Unit (MORU), for malaria slide reading; Tharisara Sakulthaew, from the MORU, for coordinating the sample shipments; and Montri Rijaibun and Nuttapol Panachuenwongsakul, from the MORU, for data management. Permission to publish this work was given by the Director General, National Institute for Medical Research, Tanzania. Disclaimer. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Notes Acknowledgments We thank Tedson Lukindo, from the Joint Malaria Programme–Tanzania, for assistance with the ELISA; Benjamas Intharabut, Ketsanee Srinamon, and Forradee Nuchsongsin, from the Mahidol-Oxford Tropical Medicine Research Unit (MORU), for malaria slide reading; Tharisara Sakulthaew, from the MORU, for coordinating the sample shipments; and Montri Rijaibun and Nuttapol Panachuenwongsakul, from the MORU, for data management. Permission to publish this work was given by the Director General, National Institute for Medical Research, Tanzania. Disclaimer. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Financial support. This work was supported by the Wellcome Trust (grants 076908 and 082541) and was coordinated as part of the Wellcome Trust–Mahidol University Oxford Tropical Medicine Research Programme, funded by the Wellcome Trust of Great Britain. The community study was supported by the Netherlands Organization of Scientific Research, Foundation for the Advancement of Tropical Research (grants W 93–413, WAO 93–441), and the Cornelis Visser Foundation. H. V. is supported by the INSTAPA project, which receives funding from the European Union's Seventh Framework Programme (FP7/2007–2013; grant 211484). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Evidence is accumulating that preschool-aged children (PSAC) are at significant risk of schistosomiasis [1]. However, relatively little is known about the immunoepidemiology of Schistosoma species infection among these children and, hence, about the early development and regulation of the immune response to schistosomiasis in populations where Schistosoma species are endemic. Among older children and adults, chronic infection is associated with a skewed type 2 response, with elevated levels of specific immunoglobulin E (IgE) and eosinophilia [2]; these responses are also typical of allergy. In allergy, specific IgE induces a potentially lethal inflammatory response. A similar IgE response directed at antigen from relatively short-lived eggs that are trapped in host tissues everyday during schistosome infection [3] would be disastrous for both host and parasite. Instead, both have coevolved to produce/induce a tightly regulated immune response during infection, mediated by factors such as interleukin 10 and T-regulatory cells (Tregs), as well as immunoglobulin G4 (IgG4), which is capable of blocking IgE-allergen interaction [2].

would be disastrous for both host and parasite. Instead, both have coevolved to produce/induce a tightly regulated immune response during infection, mediated by factors such as interleukin 10 and T-regulatory cells (Tregs), as well as immunoglobulin G4 (IgG4), which is capable of blocking IgE-allergen interaction [2]. We have shown previously that IgE regulation depends on the extent and length of exposure to individual parasite allergen-like proteins (Jones et al, unpublished data). IgE responses to SmTAL2, a member of the tegumental allergen-like (TAL) family expressed throughout the parasite's life cycle, including the egg stage [4], were low among long-term residents of a Schistosoma mansoni–endemic area of Kenya but significantly higher among recent immigrants to the same area. In contrast, SmTAL2-IgG4 responses were higher among residents; removal of IgG from sera resulted in significantly higher SmTAL2-IgE levels among residents, to the extent that levels were higher than those detected in immigrants. This demonstrates IgG-dependent desensitization of SmTAL2-IgE responses among individuals with long-term exposure.

responses were higher among residents; removal of IgG from sera resulted in significantly higher SmTAL2-IgE levels among residents, to the extent that levels were higher than those detected in immigrants. This demonstrates IgG-dependent desensitization of SmTAL2-IgE responses among individuals with long-term exposure. SmTAL1 is another TAL protein but is principally expressed in adult worms; anti-SmTAL1 IgE is associated with immunity to infection [5, 6]. In the same study in Kenya, SmTAL1-IgE and SmTAL1-IgG4 levels were both high among residents and significantly lower among immigrants (Jones et al, unpublished data). In communities of endemicity, SmTAL1-IgE and SmTAL1-IgG4 responses increase with age and after chemotherapeutic drug treatment [4]. In the mouse, where schistosome worms outlive their host, SmTAL1-IgE responses only develop following repeated rounds of infection and praziquantel treatment, whereas SmTAL2-IgG and SmTAL2-IgE are seen relatively early (Jones et al, unpublished data). Taken together, this evidence suggests that SmTAL1 responses and associated immunity take much longer to develop after repeated exposure to dying worms.

develop following repeated rounds of infection and praziquantel treatment, whereas SmTAL2-IgG and SmTAL2-IgE are seen relatively early (Jones et al, unpublished data). Taken together, this evidence suggests that SmTAL1 responses and associated immunity take much longer to develop after repeated exposure to dying worms. In the current study, we investigate the development of IgE and IgG4 responses to SmTAL1 and SmTAL2 in PSAC. The study was conducted in 2 separate villages with different degrees of transmission. We compare age-related changes in IgE and IgG4 responses to SmTAL1 and SmTAL2, to determine how the extent of exposure determines the early development and regulation of these allergic-type responses. Previous findings would predict that few PSAC have anti-TAL1 responses, but that they might have greater, un-regulated, and potentially damaging, IgE-SmTAL2 levels. This combination of responses could result in increased susceptibility to infection and morbidity, highlighting the potential benefits of including PSAC in schistosomiasis control programs.

at few PSAC have anti-TAL1 responses, but that they might have greater, un-regulated, and potentially damaging, IgE-SmTAL2 levels. This combination of responses could result in increased susceptibility to infection and morbidity, highlighting the potential benefits of including PSAC in schistosomiasis control programs. METHODS This study forms part of a larger Schistosomiasis in Mothers and Infants (SIMI) project which was conducted in 6 S. mansoni–endemic communities in Uganda and described in detail elsewhere [7]. The London School of Hygiene and Topical Medicine and the Ugandan National Council of Science and Technology granted ethics approval. Briefly, mothers and up to 2 of their children (age, 0.5–5 years) were recruited, and written informed consent obtained on behalf of children. Stool samples were obtained from each child on 2 consecutive days, and two 41.7 mg Kato-Katz thick slides [8] were prepared from each specimen; 75-µL blood samples were obtained by finger prick. Mothers were interviewed in the local language about their knowledge of schistosomiasis, their demographic characteristics, and both their and their children's water contact behavior and history of schistosomiasis treatment. The current study draws on baseline data and sera collected in April 2009 from 426 children of 213 mothers living in the villages of Bugoigo and Piida, Bulissa District, Lake Albert.

y among those infected was 49.23 epg. The prevalence and intensity of infection varied significantly by village. In Bugoigo, the prevalence was 53.0%, compared with 27.5% in Piida (P < .001), and geometric mean intensity of infection among infected individuals was 61.38 epg in Bugoigo and 27.79 epg in Piida (P = .002). The prevalence of key demographic and behavioral risk factors, determined by the questionnaire, is presented in Table 1 by village; also displayed are associations between risk factors and infection. The likelihood of infection was increased among certain ethnic groups, with age, with the duration of water contact, and on learning to swim (P ≤ .03). Children from Bugoigo were more likely to be of “other” ethnic groups (which was associated with a greater odds of infection), to spend more time in the water, and to be brought to the water by their mother, compared with children from Piida (Table 1); these behavioral differences help explain the higher prevalence of infection among Bugoigo children, although environmental factors are also likely to be important. Table 1. Distribution of Risk Factors and Association Between Risk Factors and Schistosoma mansoni Infection Among Preschool-Aged Children (PSAC)

Table 1); these behavioral differences help explain the higher prevalence of infection among Bugoigo children, although environmental factors are also likely to be important. Table 1. Distribution of Risk Factors and Association Between Risk Factors and Schistosoma mansoni Infection Among Preschool-Aged Children (PSAC) Distributiona Association With Infection Risk Factor Bugoigo Piida P PSAC Infected, % Adjustedb OR (95% CI) Pc Village Bugoigo … … 53.02 Reference Piida … … 27.50 .20 (.09–.43) <.0001 Sex Female 111 (51.15) 77 (47.83) 42.25 Reference Male 106 (48.85) 84 (52.17) .52 41.71 1.15 (.60–2.20) .66 Age, y, mean 3.11 2.94 .26 1.37 (1.06–1.76) .02 Ethnic background Banyoro 18 (6.87) 14 (8.75) 41.38 .93 (.23–3.77) Bagungu 58 (22.14) 30 (18.75) 28.57 .31 (.12–.75) Alur 150 (57.25) 110 (68.75) 43.67 Reference Otherd 36 (13.74) 6 (3.75) .004 63.89 4.13 (1.03–16.62) .004 Water contact duration, h Never 97 (45.12) 68 (42.77) 30.00 Reference <0.5 42 (19.53) 53 (33.33) 39.36 .76 (.25–2.29) 0.5–1 27 (12.56) 21 (13.21) 55.32 3.06 (.90–10.45) >1 to 2 43 (20.00) 14 (8.81) 63.16 6.20 (1.71–22.50) >2 6 (2.79) 3 (1.89) .01 77.78 6.78 (.34–134.69) .01 Can swim No 185 (87.68) 111 (71.15) 39.45 Reference Yes 26 (12.32) 45 (28.85) <.001 53.52 3.26 (1.14–9.33) .03 Mother brings to water No 124 (65.26) 122 (80.26) 39.26 Reference Yes 66 (34.74) 30 (19.74) .002 50.00 .43 (.16–1.17) .10 Abbreviations: CI, confidence interval; OR, odds ratio. a Data are no. (%) of PSAC, unless otherwise indicated.

Distributiona Association With Infection Risk Factor Bugoigo Piida P PSAC Infected, % Adjustedb OR (95% CI) Pc Village Bugoigo … … 53.02 Reference Piida … … 27.50 .20 (.09–.43) <.0001 Sex Female 111 (51.15) 77 (47.83) 42.25 Reference Male 106 (48.85) 84 (52.17) .52 41.71 1.15 (.60–2.20) .66 Age, y, mean 3.11 2.94 .26 1.37 (1.06–1.76) .02 Ethnic background Banyoro 18 (6.87) 14 (8.75) 41.38 .93 (.23–3.77) Bagungu 58 (22.14) 30 (18.75) 28.57 .31 (.12–.75) Alur 150 (57.25) 110 (68.75) 43.67 Reference Otherd 36 (13.74) 6 (3.75) .004 63.89 4.13 (1.03–16.62) .004 Water contact duration, h Never 97 (45.12) 68 (42.77) 30.00 Reference <0.5 42 (19.53) 53 (33.33) 39.36 .76 (.25–2.29) 0.5–1 27 (12.56) 21 (13.21) 55.32 3.06 (.90–10.45) >1 to 2 43 (20.00) 14 (8.81) 63.16 6.20 (1.71–22.50) >2 6 (2.79) 3 (1.89) .01 77.78 6.78 (.34–134.69) .01 Can swim No 185 (87.68) 111 (71.15) 39.45 Reference Yes 26 (12.32) 45 (28.85) <.001 53.52 3.26 (1.14–9.33) .03 Mother brings to water No 124 (65.26) 122 (80.26) 39.26 Reference Yes 66 (34.74) 30 (19.74) .002 50.00 .43 (.16–1.17) .10 Abbreviations: CI, confidence interval; OR, odds ratio. a Data are no. (%) of PSAC, unless otherwise indicated. b Estimated using forward-fitting 2-level logistic regression. The following variables were included and retained if significant at the P < .1 level: village, age, water contact duration, child treated for schistosomiasis, child can swim, ethnic background, mother brings child to water, site where child is bathed (lake vs home), frequency of bathing, mother's occupation, and whether mother had heard of schistosomiasis (or Bilharzia). Infection was defined as ≥1 detectable S. mansoni eggs in Kato-Katz slides.

treated for schistosomiasis, child can swim, ethnic background, mother brings child to water, site where child is bathed (lake vs home), frequency of bathing, mother's occupation, and whether mother had heard of schistosomiasis (or Bilharzia). Infection was defined as ≥1 detectable S. mansoni eggs in Kato-Katz slides. c By likelihood ratio tests. d Congolese and other minority ethnic groups.

treated for schistosomiasis, child can swim, ethnic background, mother brings child to water, site where child is bathed (lake vs home), frequency of bathing, mother's occupation, and whether mother had heard of schistosomiasis (or Bilharzia). Infection was defined as ≥1 detectable S. mansoni eggs in Kato-Katz slides. c By likelihood ratio tests. d Congolese and other minority ethnic groups. To investigate how the degree of exposure influences the early development of immune responses to S. mansoni, we measured children's anti-SmTAL1 and anti-SmTAL2 IgE and IgG4 responses. Virtually none of the 301 children who donated serum produced SmTAL1-IgE or IgG4 responses: SmTAL1-IgE and SmTAL1-IgG4 were detected in 1 child (age, 4 years) and in 2 children (mean age, 4 years; both were treated previously), respectively, at very low levels. In contrast, 72 (23.9%) children had detectable SmTAL2-IgE, and 180 (59.8%) children had detectable SmTAL2-IgG4. Although there was no significant difference in the prevalence of SmTAL2-IgE responsiveness among infected versus noninfected children (prevalence, 25.9% among infected children and 22.4% among noninfected children; P = .65 after adjustment for age and sex), the prevalence of SmTAL2-IgG4 responsiveness was significantly greater among infected children (prevalence, 72.6% vs 48.4%; P = .01 after adjustment for age and sex). The prevalence of both responses varied by village and with age; for anti-SmTAL2-IgE, associations with age varied significantly by village (age-village interaction, P = .001). Overall, 13.9% of children from Bugoigo had detectable SmTAL2-IgE responses, compared with 38.8% of children from Piida (P < .001 after adjustment for age and sex). Figure 1A displays the predicted probability of SmTAL2-IgE responsiveness over age, by village. Among infants from Piida, the predicted anti-SmTAL2-IgE prevalence initially increased rapidly with age but peaked and then declined at around 4 years of age. Among infants from Bugoigo, in contrast, the predicted probability was overall lower and decreased with age. Figure 1. Predicted probability for TAL2 immunoglobulin E (A) and TAL2 immunoglobulin G4 (B) responsiveness among infants in Bugoigo (dashed line) and Piida (solid line). A total of 180 children in Bugoigo (68.7%) donated sera, 98 (56.0%) of whom had detectable infection. A total of 121 children in Piida (73.8%) donated sera, 37 (30.6%) of whom had detectable infection.

lin E (A) and TAL2 immunoglobulin G4 (B) responsiveness among infants in Bugoigo (dashed line) and Piida (solid line). A total of 180 children in Bugoigo (68.7%) donated sera, 98 (56.0%) of whom had detectable infection. A total of 121 children in Piida (73.8%) donated sera, 37 (30.6%) of whom had detectable infection. A, A statistically significant age-village interaction was observed (χ2 [3 df] = 16.01; P = .001); age was modeled as a categorical variable because of departure from linearity for Piida estimates (P = .01). Model-predicted odds ratios (ORs) were as follows: male sex, 0.68 (95% confidence interval [CI], .36–1.29); village (Piida), 0.77 (95% CI, .26–2.32); age 2.1–3 years, 0.82 (95% CI, .27–2.55); age 3.1–4 years, 0.52 (95% CI, .14–1.97); age 4.1–6 years, 0.41 (95% CI, .11–1.54); age 2.1–3 years*Piida interaction term, 7.29 (95% CI, 1.33–40.05); 3.1–4 years*Piida interaction term, 25.36 (95% CI, 4.35–147.68); and age 4.1–6 years*Piida interaction term, 14.18 (95% CI, 2.43–82.76). B, No significant age-village interaction was observed (χ2 [3 df] = 0.374; P = .541); age was modeled as a continuous variable because there was no departure from linearity (P = .274). Model-predicted ORs were as follows: male sex, 0.44 (95% CI, .20–.94); village (Piida), 0.01 (95% CI, .003–.02); and age, 2.03 (95% CI, 1.54–2.67).

icant age-village interaction was observed (χ2 [3 df] = 0.374; P = .541); age was modeled as a continuous variable because there was no departure from linearity (P = .274). Model-predicted ORs were as follows: male sex, 0.44 (95% CI, .20–.94); village (Piida), 0.01 (95% CI, .003–.02); and age, 2.03 (95% CI, 1.54–2.67). Figure 1B displays the predicted probability of an anti-SmTAL2-IgG4 response over age, by village. Unlike the predicted anti-SmTAL2-IgE prevalence, the predicted anti-SmTAL2-IgG4 prevalence increased linearly with age in both villages. Furthermore, the likelihood of a response was significantly greater among children from Bugoigo, compared with children from Piida (P < .0001 after adjustment for age and sex), with 89.4% of children from Bugoigo having a detectable SmTAL2-IgG4 response, compared with only 15.7% of children from Piida.

n both villages. Furthermore, the likelihood of a response was significantly greater among children from Bugoigo, compared with children from Piida (P < .0001 after adjustment for age and sex), with 89.4% of children from Bugoigo having a detectable SmTAL2-IgG4 response, compared with only 15.7% of children from Piida. DISCUSSION SmTAL1 is a member of the TAL family, a family of proteins differentially expressed throughout the schistosome life cycle that share structural homology with the EF-hand allergens, one of the most common group of clinical allergens [4]. It is principally expressed in the adult worm and thought to be sequestered from the immune system in live worms. In areas of endemicity, responses to SmTAL1 steadily increase with age, it is thought following gradual, accumulated exposure to antigen released from dying worms [4]. SmTAL2, another TAL, is expressed throughout the parasite's life cycle, including the egg stage; hence, exposure is continuous during infection because of the release of SmTAL2 from short-lived eggs trapped in tissue. In contrast to SmTAL1-IgE, SmTAL2-IgE responses are low among long-term exposed individuals but significantly higher among recently exposed individuals; there is strong evidence to suggest that this is due to IgG4-dependent SmTAL2-IgE desensitization (Jones et al, unpublished data).

short-lived eggs trapped in tissue. In contrast to SmTAL1-IgE, SmTAL2-IgE responses are low among long-term exposed individuals but significantly higher among recently exposed individuals; there is strong evidence to suggest that this is due to IgG4-dependent SmTAL2-IgE desensitization (Jones et al, unpublished data). In the current study, we examined SmTAL1- and SmTAL2-IgE and IgG4 responses among PSAC from an S. mansoni–endemic region of Uganda. On the basis of findings from previous studies, we hypothesized that children would have no or low anti-TAL1 responses but higher, unregulated TAL2-IgE responses. The children studied were from 2 villages with different levels of transmission: children from Bugoigo had significantly greater risk of infection than children from Piida. In Bugoigo, SmTAL2-IgE responses decreased with age and were overall lower than in Piida, where responses increased then decreased with age. In contrast, the prevalence of SmTAL2-IgG4 responsiveness was higher in Bugoigo, and the likelihood of a response increased with age in both villages. These findings are consistent with previous observations comparing SmTAL2 responses among resident and immigrant populations (Jones et al, unpublished data) and provide further support for our hypothesis that SmTAL2-IgE is an early human immune response to S. mansoni, which is downregulated during chronic infection, probably because of IgG4-dependent desensitization. The rapid SmTAL2-IgE desensitization observed in Bugoigo highlights the acute nature of this response. Since the average lifespan of S. mansoni adult worms is 7 years [9], the observed lack of SmTAL1 responsiveness among PSAC is entirely expected and confirms that this is a much later response that develops after repeated exposure to antigen following natural or induced worm death.

hts the acute nature of this response. Since the average lifespan of S. mansoni adult worms is 7 years [9], the observed lack of SmTAL1 responsiveness among PSAC is entirely expected and confirms that this is a much later response that develops after repeated exposure to antigen following natural or induced worm death. Chronic schistosomiasis morbidity is caused by T-helper 2 granulomatous responses to continuous deposition of eggs, which over years [10] can cause severe fibrotic disease [11]. Acute schistosomiasis is also thought to be a reaction provoked by eggs, as well as by migrating schistosomulae [12]. IgE-mediated inflammation, triggered by egg allergen-like antigens such as SmTAL2, could play a role in this and could also occur in very young children in schistosomiasis-endemic areas. If so, SmTAL2-IgE modulation would limit IgE-mediated tissue damage, similarly to allergen-specific immunotherapy (SIT), in which repeated allergen administration is used to induce IgE desensitization. Immunological changes associated with SIT include reductions in IgE, induction of Tregs, and increases in allergen-specific IgG, particularly IgG4 [13]. IgG is thought to directly compete for the same epitopes as IgE, downmodulating both IgE-dependent histamine release [14] and IgE-facilitated allergen presentation to T cells [15].

changes associated with SIT include reductions in IgE, induction of Tregs, and increases in allergen-specific IgG, particularly IgG4 [13]. IgG is thought to directly compete for the same epitopes as IgE, downmodulating both IgE-dependent histamine release [14] and IgE-facilitated allergen presentation to T cells [15]. In summary, the current study investigated the development of IgE and IgG4 responses to the allergen-like proteins SmTAL1 and SmTAL2 among PSAC from 2 separate villages with different degrees of S. mansoni transmission. We provided evidence for IgG4-dependent IgE desensitization to constitutively expressed SmTAL2; this desensitization occurred earlier in the higher transmission village. Almost no children had developed detectable responses to the worm antigen SmTAL1, most likely because of a lack of sufficient exposure to antigen. Our results confirm previous findings suggesting that the degree of IgE regulation is dependent on the extent and length of antigen exposure: we hypothesize that potentially pathogenic IgE responses to continuously-released SmTAL2 are tightly regulated among adults in regions of endemicity but that SmTAL1-IgE responses are less regulated, because of only periodic exposure following worm death. Findings will help our understanding of immune responses in schistosomiasis and in allergy, providing insights for the therapeutic treatment of both. A lack of immunity, combined with higher prevalence of pathogenic IgE responses, could increase the risk of severe morbidity among PSAC, highlighting the benefit for their inclusion in schistosomiasis control programs.

esponses in schistosomiasis and in allergy, providing insights for the therapeutic treatment of both. A lack of immunity, combined with higher prevalence of pathogenic IgE responses, could increase the risk of severe morbidity among PSAC, highlighting the benefit for their inclusion in schistosomiasis control programs. Notes Acknowledgments. We thank the families that took part in the study; the Ministry of Health of Uganda, Vector Control Division, and all their staff, for their support during the SIMI project; and the local Vector Control Division officers, Mr Juma Nabonge, Mr Ashuman Babyesiza, Mr Perez Isingoma, and Mr Chris Byalero, for making cohort studies possible in areas where populations are constantly migrating. Mr Byalero regrettably passed away shortly before submission of this manuscript, and we express our regret for the irreplaceable loss of a friend and colleague. J. C. S.-F. and J. R. S. thank Prof David Rollinson and the Natural History Museum, London, for continued support during the SIMI project. Disclaimer. The funders had no role in the study design, data collection, data analysis, decision to publish, or preparation of the manuscript. Financial support. This study was funded by a Wellcome Trust Programme (grant WT 083931/Z/07/Z to D. W. D.) and a Wellcome Trust Project (grant WT085440MA to J. R. S.). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Mycobacterium tuberculosis is the causative agent of human tuberculosis. The World Health Organization estimates that M. tuberculosis was been responsible for approximately 1.5 million deaths and approximately 8.8 million new infections in 2010 [1]. A key feature of tuberculosis is the presence of necrotic caseations in the lungs, which are mainly derived from the necrotic corpses of macrophages [2]. Induction of macrophage necrosis is a well-known virulence mechanism of M. tuberculosis and is dependent on the ESX-1/RD1 protein secretion system [3, 4]. The ESX-1 secretion system comprises 11 gene products. ESAT-6 and CFP-10 are 2 major ESX-1 substrates. They are required for the proper functioning of the secretion system [5, 6]. ESX-1's necrotic effect has been attributed to the membrane-perforating activity of ESAT-6 [4, 7]. This ESX-1 activity is responsible for the ESX-1–dependent phagosomal rupture [8–10] and the subsequent escape of M. tuberculosis [11]. ESAT-6 can interact indirectly with the cell-surface innate immune receptor Toll-like receptor 2 (TLR2), causing suppression of proinflammatory cytokine responses [12].

ESAT-6 [4, 7]. This ESX-1 activity is responsible for the ESX-1–dependent phagosomal rupture [8–10] and the subsequent escape of M. tuberculosis [11]. ESAT-6 can interact indirectly with the cell-surface innate immune receptor Toll-like receptor 2 (TLR2), causing suppression of proinflammatory cytokine responses [12]. Neutrophils undergo a novel death mechanism when subjected to infections or chemical stimulations [13]. The mechanism is characterized by the release of cellular DNA, which is decorated with citrullinated histones [14]. Initial work on PMA-stimulated neutrophils indicates that cell lysis, as measured by lactate dehydrogenase release, occurs after DNA release has been completed [15]. But a single-cell assay for lysis reveals that extracellular DNA is detected as soon as the plasma membrane is ruptured [13]. Rather than a passive leakage from cellular lysis, the DNA release is regulated by multiple enzymes, such as neutrophil elastase [16]. The released DNA is called an extracellular trap because of its ability to bind diverse pathogens ranging from yeast to bacteria [15]. Extracellular traps have reported antibacterial activities [17]. However, this model has been under scrutiny [18]. A new report indicates that bacteria released from extracellular traps can resume growth [19]. Thus, the antibacterial effect of extracellular traps should be determined in situ. Macrophages can produce extracellular traps under certain circumstance [20]. However, despite extensive studies on bacteria-induced macrophage deaths, it remains unclear whether death-inducing pathogens, such as M. tuberculosis, can induce human macrophages to produce extracellular traps.

should be determined in situ. Macrophages can produce extracellular traps under certain circumstance [20]. However, despite extensive studies on bacteria-induced macrophage deaths, it remains unclear whether death-inducing pathogens, such as M. tuberculosis, can induce human macrophages to produce extracellular traps. Formation of extracellular traps by neutrophils upon stimulation with complement factor 5a requires priming by interferon γ (IFN-γ) [21]. Murine macrophage studies clearly show that IFN-γ induces M. tuberculosis killing [22, 23]. But a similar effect of IFN-γ in human macrophages remains controversial [24–26]. IFN-γ can even enhance M. tuberculosis replication during human macrophage infections [26]. Vogt and Nathan have recently reported specific culture conditions for human macrophages, such as physiological O2 levels and the presence of the growth factor granulocyte macrophage colony-stimulating factor (GM-CSF), that are critical for the IFN-γ–mediated antimycobacterial activity [27]. Here we show that M. tuberculosis induced extracellular trap formation by infected human macrophages. This phenomenon was mediated by elastase activity. We also demonstrated that IFN-γ amplified extracellular trap formation in an ESX-1–dependent manner. As extracellular trap formation is linked to cell death, the effect of IFN-γ on ESX-1's necrotic effect was studied. Our present results suggest a novel role for IFN-γ in amplifying multiple effects of the mycobacterial ESX-1.

demonstrated that IFN-γ amplified extracellular trap formation in an ESX-1–dependent manner. As extracellular trap formation is linked to cell death, the effect of IFN-γ on ESX-1's necrotic effect was studied. Our present results suggest a novel role for IFN-γ in amplifying multiple effects of the mycobacterial ESX-1. METHODS Preparation of Primary Human Macrophages Peripheral blood monocytes were isolated from human peripheral blood (New York Blood Center) by density centrifugation on Ficoll-Paque PLUS (GE Healthcare), followed by positive selection, using CD14 magnetic beads (Miltenyi Biotec). CD14+ monocytes on the magnetic column were washed 3 times in phosphate-buffered saline (pH 7.2), 0.5% fetal bovine serum, and 2 mM ethylenediaminetetraacetic acid before elution by gravity. Selected CD14+ monocytes were resuspended in Roswell Park Memorial Institute (RPMI) 1640 (catalog no. 11875, Gibco-Invitrogen) containing 10% non–heat-inactivated human AB serum (Gemini Bio-Products) and 20 mM HEPES buffer with or without recombinant human GM-CSF (2 ng/mL) or macrophage colony-stimulating factor (M-CSF; 10 ng/mL; R&D Systems). Cell density was adjusted to 0.67 × 106 cells/mL. A total of 0.3 or 0.6 mL of cells was dispensed respectively into 48-well or 24-well tissue culture plate wells, the latter containing coverslips. Monocytes were differentiated into macrophages for 11–16 days at 37°C in a humidified chamber with 5% CO2 or in an O2-regulated humidified chamber flushed with N2 to achieve 10% O2 along with 5% CO2. Thirty percent of incubating medium was replaced every 3–4 days with fresh medium with or without the corresponding growth factor. Macrophages were activated by recombinant human IFN-γ at 0, 2.5, 25, or 100 U/mL (EMD-Calbiochem) overnight before infection.

ed chamber flushed with N2 to achieve 10% O2 along with 5% CO2. Thirty percent of incubating medium was replaced every 3–4 days with fresh medium with or without the corresponding growth factor. Macrophages were activated by recombinant human IFN-γ at 0, 2.5, 25, or 100 U/mL (EMD-Calbiochem) overnight before infection. M. tuberculosis Infection M. tuberculosis H37Rv, ΔESX-1 (or ΔRD1), and complemented ΔESX-1 have been described elsewhere [4] and were grown at 37°C in Middlebrook 7H9 (Difco) supplemented with 10% oleic acid-albumin-dextrose-catalase (BBL, Becton Dickinson), 0.5% glycerol, and 0.05% Tween-80. A total of 30 µg/mL kanamycin was added to 7H9 for the complemented strain. Bacteria were prepared for infection as described elsewhere [10]. Because the optical density of nonsonicated bacteria was typically twice that of sonicated bacteria, multiplicities of infection (MOIs) of 5 for nonsonicated bacteria and 10 for sonicated bacteria were used. Four hours after infection with M. tuberculosis, macrophages were washed twice with RPMI 1640 containing 20 mM HEPES buffer and then maintained in RPMI 1640 containing 10% non–heat-inactivated human AB serum and 20 mM HEPES buffer. Macrophages activated by IFN-γ were incubated in the same dose of IFN-γ after infection. When appropriate, 1,25-dihydroxyvitamin D (Enzo Life Sciences) was used at 20 nM along with IFN-γ as described elsewhere [28].

then maintained in RPMI 1640 containing 10% non–heat-inactivated human AB serum and 20 mM HEPES buffer. Macrophages activated by IFN-γ were incubated in the same dose of IFN-γ after infection. When appropriate, 1,25-dihydroxyvitamin D (Enzo Life Sciences) was used at 20 nM along with IFN-γ as described elsewhere [28]. Analyses of Infected Macrophages Necrotic macrophages were stained by 2.5 µg/mL propidium iodide (Sigma) in serum-free RPMI 1640 for 10 minutes at room temperature and fixed in 2% paraformaldehyde (Sigma) in RPMI 1640 for 1 hour. To stain M. tuberculosis, auramine-rhodamine staining was performed using the TB Fluorescent Stain Kit T (Becton Dickinson) according to manufacturer's protocol. To visualize extracellular traps, fixed cells were stained with Hoechst 33 258 (Sigma) or picogreen (Invitrogen-Life Technologies) for 5 minutes at room temperature. Processed cells were analyzed using Eclipse Ti (Nikon) or Axio Observer (Carl Zeiss) inverted fluorescence microscopes equipped with a Photometrics CoolSNAP HQ2 (Photometrics) or AxioCam Mrm (Carl Zeiss) charge-coupled device camera, respectively. Images were analyzed by ImageJ software. M. tuberculosis was defined as aggregate positive when rhodamine's intensity was >125 arbitrary units, based on the 0–255 scale of 8-bit images. Average sizes of the M. tuberculosis–positive aggregates were determined by ImageJ's Analyze Particle function. Cells were identified as necrotic when the intensity of propidium iodide was >45 arbitrary units on the 0–255 scale. Cells were identified as apoptotic when propidium iodide's intensity was <45 arbitrary units and Hoechst's intensity was >125 arbitrary units. The percentage of positive cells was defined as the number of cells with positive signals relative to the total number of cells. Data were displayed as the mean percentage obtained from 4 different fields; 400–900 cells were examined from each field.

<45 arbitrary units and Hoechst's intensity was >125 arbitrary units. The percentage of positive cells was defined as the number of cells with positive signals relative to the total number of cells. Data were displayed as the mean percentage obtained from 4 different fields; 400–900 cells were examined from each field. Mycobacterial growth in macrophages was enumerated by lysis in 0.5% Triton X100, followed by spotting serial dilutions on agar (7H10, 10% OADC) in triplicate. Colonies were counted after 2–3 weeks. Statistical Analysis Results were tested statistically by an unpaired 2-tailed Student t test in Microsoft Excel.

<45 arbitrary units and Hoechst's intensity was >125 arbitrary units. The percentage of positive cells was defined as the number of cells with positive signals relative to the total number of cells. Data were displayed as the mean percentage obtained from 4 different fields; 400–900 cells were examined from each field. Mycobacterial growth in macrophages was enumerated by lysis in 0.5% Triton X100, followed by spotting serial dilutions on agar (7H10, 10% OADC) in triplicate. Colonies were counted after 2–3 weeks. Statistical Analysis Results were tested statistically by an unpaired 2-tailed Student t test in Microsoft Excel. RESULTS M. tuberculosis Induces Release of Extracellular Traps From Human Macrophages Our preliminary observations indicated that M. tuberculosis infection induced PMA-differentiated THP-1 macrophages to produce Hoechst-positive fiber structures (data not shown). A similar phenomenon was also observed in human macrophages derived from peripheral blood mononuclear cells from healthy donors (Figure 1A). In this report, primary human macrophages from healthy donors were used throughout this study. The observed extracellular structures were morphologically identical to that of extracellular traps, which are produced by activated neutrophils and are composed mainly of DNA decorated with citrullinated histones. Staining with picogreen, a dye specific for double-stranded DNA, and DNase I treatment confirmed that the fiber structures contained DNA (Figure 1B). An antibody specific to histone h4 citrullinated at residue 3 also recognized the DNA fiber produced by M. tuberculosis–infected macrophages (Figure 1C). Figure 1. Extracellular trap formation by Mycobacterium tuberculosis–infected macrophages. Primary human macrophages were infected with M. tuberculosis H37Rv at a multiplicity of infection (MOI) of 5 without sonication of M. tuberculosis. A, Effect of elastase inhibitor AAPV. Macrophages differentiated in macrophage colony-stimulating factor (M-CSF) were pretreated 1 hour before and maintained after M. tuberculosis infection with either dimethyl sulfoxide (DMSO) or 100 nM AAPV. Twenty-seven hours after infection, cells were fixed and stained with Hoechst 258. B–F, Differentiation, activation, and infection of macrophages were performed in 10% O2. M. tuberculosis was not sonicated. Macrophages that were differentiated with granulocyte-macrophage colony-stimulating factor (GM-CSF) and activated with 100 U/mL of interferon γ were fixed 2 days after infection. Fixed cells were treated with DNase I (B) or processed for citrullinated histone 4 staining (C) or auramine-rhodamine staining to visualize M. tuberculosis (D and E), before staining with the double-stranded DNA–specific dye picogreen. The circles stained with picogreen represent intact nuclei.

2 days after infection. Fixed cells were treated with DNase I (B) or processed for citrullinated histone 4 staining (C) or auramine-rhodamine staining to visualize M. tuberculosis (D and E), before staining with the double-stranded DNA–specific dye picogreen. The circles stained with picogreen represent intact nuclei. F, Nucleic origin of extracellular traps. GM-CSF–differentiated macrophages were infected with sonicated M. tuberculosis at an MOI of 10 and stained with picogreen. Arrows indicate nuclei that are releasing chromosome DNA. Arrowheads indicate M. tuberculosis. Activated neutrophils undergo a novel cell death mechanism that leads to extracellular trap formation. Multiple enzymes, such as neutrophil elastase, regulate extracellular trap formation [16]. As macrophages also express various elastase activities [29, 30], we determined whether elastase activity affected extracellular trap formation in M. tuberculosis–infected macrophages. Indeed, the elastase inhibitor AAPV prevented primary human macrophages infected with M. tuberculosis from forming DNA fibers (Figure 1A). AAPV did not block macrophage necrosis triggered by M. tuberculosis (data not shown), indicating that DNA fiber formation was a regulated process, rather than a passive event due to cell lysis.

elastase inhibitor AAPV prevented primary human macrophages infected with M. tuberculosis from forming DNA fibers (Figure 1A). AAPV did not block macrophage necrosis triggered by M. tuberculosis (data not shown), indicating that DNA fiber formation was a regulated process, rather than a passive event due to cell lysis. Acid-fast staining revealed that extracellular M. tuberculosis colocalized with the DNA fibers after extensive washings, indicating that M. tuberculosis was bound to the DNA fibers (Figure 1D). Thus, we characterized the DNA fibers released from infected human macrophages as extracellular traps. Mouse macrophages derived from bone marrow or of cell-line origin did not produce extracellular traps as a result of M. tuberculosis infection, even after onset of necrosis (data not shown). In contrast to the extracellular mycobacteria that bind to extracellular traps, the majority of M. tuberculosis cells remained associated within macrophages, as expected (Figure 1D). As infections progressed, macrophages with a high burden of M. tuberculosis tended to cluster together with extracellular traps (Figure 1E). Higher-magnification imaging of these structures revealed that extracellular traps emerge from the nuclei of infected macrophages (Figure 1F). Some extracellular traps from multiple nuclei appeared to merge together in extracellular space.

M. tuberculosis tended to cluster together with extracellular traps (Figure 1E). Higher-magnification imaging of these structures revealed that extracellular traps emerge from the nuclei of infected macrophages (Figure 1F). Some extracellular traps from multiple nuclei appeared to merge together in extracellular space. Given that high mycobacterial burden was associated with extracellular trap formation in macrophages, we tested whether mycobacterial clumps were more efficient at promoting extracellular trap formation. Without sonication, mycobacterial clumps typically contained 5–20 mycobacteria. Given that nonsonicated bacteria induced the death of many more cells (data not shown), we used a lower MOI to normalize the cytotoxic effect. Generally, nonsonicated bacteria induced more extracellular trap formation than sonicated bacteria (data not shown).

mycobacterial clumps typically contained 5–20 mycobacteria. Given that nonsonicated bacteria induced the death of many more cells (data not shown), we used a lower MOI to normalize the cytotoxic effect. Generally, nonsonicated bacteria induced more extracellular trap formation than sonicated bacteria (data not shown). Human IFN-γ Promotes the Formation of Extracellular Traps by M. Tuberculosis–Infected Macrophages Neutrophils stimulated with lipopolysaccharide or complement factor 5a release extracellular traps only if the neutrophils have been primed with GM-CSF [31]. In contrast to neutrophils, GM-CSF–differentiated macrophages were less efficient at forming extracellular traps in response to challenge with nonsonicated M. tuberculosis (Figure 2A). GM-CSF represses IFN-γ signaling, and priming of IFN-γ is critical for mature neutrophils to form extracellular traps in response to stimulation [21]. We therefore tested IFN-γ's effect on M. tuberculosis–induced extracellular trap formation. IFN-γ was found to enhance M. tuberculosis–induced extracellular trap formation by macrophages differentiated with M-CSF but not with GM-CSF (Figure 2A). Figure 2. Interferon γ (IFN-γ) enhances macrophages to form extracellular traps. Hoechst staining reveals fine fiber structures that have emerged from nuclei that are morphologically indistinguishable from extracellular traps. A, The effect of granulocyte-macrophage colony-stimulating factor (GM-CSF) differentiation and IFN-γ activation is shown. Primary human macrophages differentiated with or without GM-CSF in the presence of 10% human serum were activated with 100 U/mL IFN-γ or left untreated before infection with sonicated M. tuberculosis H37Rv at a multiplicity of infection (MOI) of 10. B, The effect of O2 levels in shown. Primary human macrophages were differentiated with GM-CSF either at normoxia (approximately 20% O2) or at a physiological O2 level (10% O2) and pretreated with 100 U/mL IFN-γ before infection with nonsonicated M. tuberculosis at an MOI of 5. The corresponding phase contrast images are shown in Supplementary Figure 1. Scale bar, 100 µm.

rophages were differentiated with GM-CSF either at normoxia (approximately 20% O2) or at a physiological O2 level (10% O2) and pretreated with 100 U/mL IFN-γ before infection with nonsonicated M. tuberculosis at an MOI of 5. The corresponding phase contrast images are shown in Supplementary Figure 1. Scale bar, 100 µm. The preceding observation was made under atmospheric O2 levels (approximately 20%). However, macrophages are normally exposed to physiological O2 levels (5%–10%) [32]. Since the O2 level has a major impact on the metabolism and behavior of cultured human macrophages [33], we studied the effect of 10% O2 on extracellular trap formation in the presence or absence of IFN-γ. GM-CSF–differentiated macrophages cultured in 10% O2 produced no extracellular traps upon infection by M. tuberculosis, even when the bacteria were nonsonicated (Figure 2B). Remarkably, IFN-γ enabled the infected macrophages cultured in 10% O2 to release extracellular traps (Figure 2B). Under this condition, nearly all infected macrophages formed extracellular traps (Figures 1E and 2B). Thus, IFN-γ remained capable of inducing extracellular traps from macrophages regardless of whether they were cultured in approximately 20% or 10% O2.

ages cultured in 10% O2 to release extracellular traps (Figure 2B). Under this condition, nearly all infected macrophages formed extracellular traps (Figures 1E and 2B). Thus, IFN-γ remained capable of inducing extracellular traps from macrophages regardless of whether they were cultured in approximately 20% or 10% O2. The Effect of M. tuberculosis ESX-1 on Extracellular Trap Formation Is Further Enhanced by Human IFN-γ We next examined whether ESX-1 played a role in M. tuberculosis–induced extracellular trap formation, since ESX-1 is required for initiating caspase-1–independent cell death in human macrophages [10, 34] and because extracellular trap formation was unaffected by caspase-1 inhibition (data not shown). An M. tuberculosis ESX-1 deletion mutant did not induce M-CSF–differentiated macrophages to form extracellular traps (Figure 3A). Addition of human IFN-γ could not restore the ability of the ESX-1 mutant to induce extracellular trap formation by GM-CSF–differentiated macrophages (Figure 3B). These results indicated that human IFN-γ–potentiated extracellular trap production was dependent on ESX-1. Regardless of whether bacteria were sonicated, extracellular trap formation induced by M. tuberculosis remained strictly dependent on the ESX-1 and was highly inducible by IFN-γ (data now shown). Figure 3. ESX-1 mediates extracellular trap formation by Mycobacterium tuberculosis­–infected macrophages. A, Primary human macrophages were infected with nonsonicated M. tuberculosis H37Rv at a multiplicity of infection (MOI) of 5. Hoechst staining of macrophage colony-stimulating factor–differentiated macrophages infected with M. tuberculosis, ΔESX-1, or ΔESX-1::ESX-1 was performed 2 days after infection. B, Primary human macrophages differentiated with granulocyte-macrophage colony-stimulating factor in 10% O2 were activated with 0, 2.5, or 25 U/mL interferon γ (IFN-γ). Macrophages were analyzed 2 days after infection with sonicated H37Rv at an MOI of 10. Extracellular traps were stained by picogreen. Total numbers of extracellular traps per total cell counts were quantitated from 4 different fields.

ulating factor in 10% O2 were activated with 0, 2.5, or 25 U/mL interferon γ (IFN-γ). Macrophages were analyzed 2 days after infection with sonicated H37Rv at an MOI of 10. Extracellular traps were stained by picogreen. Total numbers of extracellular traps per total cell counts were quantitated from 4 different fields. Human IFN-γ Enhances ESX-1–Dependent M. tuberculosis Aggregation and Growth in Macrophages Our previous observation that extracellular traps were usually produced from heavily infected macrophages prompted us to examine whether IFN-γ would increase the number of heavily infected macrophages. Acid-fast staining of M. tuberculosis showed that IFN-γ induced M. tuberculosis aggregations in human macrophages (Figure 4A). Vogt and Nathan reported that IFN-γ levels of >25 U/mL are cytotoxic to mycobacteria-infected human macrophages, whereas a low IFN-γ concentration (2.5 U/mL) conferred some degree of protection to human macrophages [27]. In our case, a lower IFN-γ level (ie, 2.5 U/mL) still stimulated M. tuberculosis aggregation (Figure 4A). Figure 4. Interferon γ (IFN-γ)–enhanced Mycobacterium tuberculosis aggregation requires ESX-1. Primary human macrophages were differentiated with granulocyte-macrophage colony-stimulating factor (GM-CSF) at 10% O2 and were activated with the indicated dose of IFN-γ or left untreated. Macrophages were infected with sonicated H37Rv at a multiplicity of infection of 10. A, Auramine-rhodamine staining of fixed macrophages was performed 3 days after infection. The intensity of the rhodamine was displayed in a rainbow color scale. Scale bars, 100 µm. B and C, Results of quantification of the percentages of M. tuberculosis cells that were auramine-rhodamine positive (>125 on a rainbow color scale; B) and the average sizes of the M. tuberculosis aggregates (C) are shown. D, A role for ESX-1 on M. tuberculosis growth in the presence of different doses of IFN-γ is shown. Error bars indicate standard errors of the mean. Abbreviation: CFU, colony-forming units.

amine-rhodamine positive (>125 on a rainbow color scale; B) and the average sizes of the M. tuberculosis aggregates (C) are shown. D, A role for ESX-1 on M. tuberculosis growth in the presence of different doses of IFN-γ is shown. Error bars indicate standard errors of the mean. Abbreviation: CFU, colony-forming units. Because extracellular trap formation was linked to increased M. tuberculosis burden and because IFN-γ enhanced the ESX-1–dependent extracellular trap formation, we tested whether the IFN-γ–enhanced M. tuberculosis aggregation was dependent on ESX-1. Auramine-rhodamine staining revealed that increasing IFN-γ concentrations yielded a dose-dependent increase in the size and number of M. tuberculosis aggregates in a strictly ESX-1–dependent fashion (Figure 4A–C). These increases coincided with the ESX-1–dependent growth enhancement in the presence of a high dose (25 U/mL) of IFN-γ (Figure 4D).

g revealed that increasing IFN-γ concentrations yielded a dose-dependent increase in the size and number of M. tuberculosis aggregates in a strictly ESX-1–dependent fashion (Figure 4A–C). These increases coincided with the ESX-1–dependent growth enhancement in the presence of a high dose (25 U/mL) of IFN-γ (Figure 4D). Human IFN-γ Potentiates Macrophage Necrosis Induced by M. tuberculosis ESX-1 Finally, since extracellular trap formation was linked to M. tuberculosis­–induced necrosis, which in turn was promoted by a higher MOI, we suspected that IFN-γ would also promote the ESX-1–triggered necrosis in human macrophages. Indeed, IFN-γ pretreatment readily induced rapid necrotic death in macrophages infected with M. tuberculosis, whereas the same treatment had a minimal effect on macrophages infected with the ESX-1 mutant of M. tuberculosis (Figure 5). An increase in ESX-1–dependent extracellular trap formation due to increasing IFN-γ doses was also detected (Figure 5). Similar results were obtained from PMA-differentiated THP-1 human macrophages (data not shown). The absence of necrosis induction and extracellular trap formation by the ESX-1 deletion mutant was unlikely due to the mutant's survival defect within macrophages, because the ESX-1 deletion mutant survived as well as the wild-type when necrosis and extracellular trap formation were being quantified (Figure 4D). Thus, IFN-γ amplified ESX-1's multiple effects on inducing mycobacterial aggregation, extracellular trap formation, and macrophage necrosis. Figure 5. Interferon γ (IFN-γ) promotes ESX-1–triggered necrosis. Primary human macrophages differentiated with granulocyte-macrophage colony-stimulating factor (GM-CSF) in 10% O2 were activated with 0, 2.5, or 25 U/mL IFN-γ. Macrophages were analyzed 2 days after infection with sonicated H37Rv at a multiplicity of infection of 10. Necrotic cells were stained by propidium iodine (red) before fixation. Fixed cells were then stained with Hoechst stain for condensed nuclei as a marker for apoptotic cells (not shown for clarity). A, Necrotic (PI+) human macrophages are shown. Scale bar, 100 µm. B and C, Results of quantification of the frequency of necrotic (PI+; B) and apoptotic (Hoechst+ PI−; C) cells are shown. Error bars indicate standard errors of the mean. Images were quantified by ImageJ as described in Materials and Methods.

r clarity). A, Necrotic (PI+) human macrophages are shown. Scale bar, 100 µm. B and C, Results of quantification of the frequency of necrotic (PI+; B) and apoptotic (Hoechst+ PI−; C) cells are shown. Error bars indicate standard errors of the mean. Images were quantified by ImageJ as described in Materials and Methods. DISCUSSION Our present study revealed a novel aspect of macrophage death induced by M. tuberculosis: formation of DNA-containing extracellular traps from infected human macrophages. This phenomenon took place after M. tuberculosis resided within macrophages, which contrasts with neutrophil extracellular trap formation triggered by extracellular bacterial pathogens. Only a subset of M. tuberculosis was bound to the extracellular traps, while the majority remained within the macrophages. The extent to which extracellular traps were formed increased over time as M. tuberculosis continued to replicate intracellularly, indicating that extracellular traps did not affect intracellular mycobacterial growth Neutrophil extracellular traps do not affect the viability of M. tuberculosis [35]. Whether M. tuberculosis in direct contact with macrophage extracellular traps is similarly unaffected has not been addressed, but our preliminary results indicate that extracellular M. tuberculosis in contact with extracellular traps retained cell integrity, based on propidium iodide staining (data not shown).

is [35]. Whether M. tuberculosis in direct contact with macrophage extracellular traps is similarly unaffected has not been addressed, but our preliminary results indicate that extracellular M. tuberculosis in contact with extracellular traps retained cell integrity, based on propidium iodide staining (data not shown). ESX-1 is a major virulence factor for mycobacterial pathogenesis, encoding a specialized protein secretion system and triggering a caspase-1–independent cell death pathway. We report here that ESX-1 was critical for extracellular trap formation by human macrophages. Our result stood in contrast to a recent report based on mouse macrophages, which shows that M. tuberculosis at a high MOI induces a small degree of DNA-containing fiber structures independent of ESX-1 [36]. We failed to detect any extracellular trap formation by mouse bone marrow–derived macrophages infected by M. tuberculosis at an MOI of 10, suggesting that M. tuberculosis–induced extracellular trap formation was a phenotype specific to human macrophages.

ree of DNA-containing fiber structures independent of ESX-1 [36]. We failed to detect any extracellular trap formation by mouse bone marrow–derived macrophages infected by M. tuberculosis at an MOI of 10, suggesting that M. tuberculosis–induced extracellular trap formation was a phenotype specific to human macrophages. Observations from previous animal studies are suggestive of the existence of extracellular traps induced by M. tuberculosis infection. Extracellular traps can be induced in vivo by trehalose dimycolate, a cell wall component of M. tuberculosis in fibrinogen-deficient mice [37]. Extracellular traps are known to promote thrombosis in vitro and in vivo [38–40]. The presence of vascular thrombosis in necrotic lesions of M. tuberculosis–infected rabbits is consistent with the existence of extracellular traps induced by M. tuberculosis [41]. Conversely, the absence of vascular thrombosis is in line with our inability to detect M. tuberculosis­–infected murine macrophages to form extracellular traps.

lar thrombosis in necrotic lesions of M. tuberculosis–infected rabbits is consistent with the existence of extracellular traps induced by M. tuberculosis [41]. Conversely, the absence of vascular thrombosis is in line with our inability to detect M. tuberculosis­–infected murine macrophages to form extracellular traps. We unexpectedly observed that IFN-γ stimulated extracellular trap formation by macrophages cultured at 10% O2 with GM-CSF. IFN-γ is a well-known macrophage activation factor that stimulates antimycobacterial activities in murine macrophages [22, 42, 43]. Its protective role against M. tuberculosis has been established in animal infection studies [23, 44] and inhalation treatment for patients with active tuberculosis [45]. Consistent with our observed correlation between extracellular trap formation and the intracellular burden of macrophages, we demonstrated on the basis of acid-fast staining that IFN-γ increased M. tuberculosis aggregation. We also detected a significant ESX-1–dependent increase in mycobacterial growth stimulated by IFN-γ at a late time point (Figure 4D). It is likely that the ability of IFN-γ to induce further extracellular trap formation and necrosis via an ESX-1 mechanism was mediated, at least in part, by the ability of IFN-γ to stimulate ESX-1–dependent mycobacterial aggregation and growth.

mycobacterial growth stimulated by IFN-γ at a late time point (Figure 4D). It is likely that the ability of IFN-γ to induce further extracellular trap formation and necrosis via an ESX-1 mechanism was mediated, at least in part, by the ability of IFN-γ to stimulate ESX-1–dependent mycobacterial aggregation and growth. Our results collectively suggest that the mycobacterial ESX-1 secretion system subverts the IFN-γ responses of human macrophages, resulting in amplification of the pathological effects of ESX-1. To the best of our knowledge, the synergistic effect of IFN-γ and ESX-1 has not been reported. We demonstrated that ESX-1 was required for IFN-γ to promote extracellular trap formation by infected macrophages. Since necrosis induction is a primary virulence mechanism of ESX-1, the effect of IFN-γ on the ESX-1–induced necrosis was also examined. We found that IFN-γ synergized with ESX-1 to induce necrosis (Figure 3). IFN-γ is known to enhance cytotoxicity in M. tuberculosis–infected human macrophages [24, 36]. One report showed that IFN-γ enhanced necrosis in bacille Calmette-Guérin–infected macrophages, implying that ESX-1 was not involved [36]. Our work by contrast used the isogenic ESX-1 mutant of M. tuberculosis to address unequivocally the role for ESX-1.

otoxicity in M. tuberculosis–infected human macrophages [24, 36]. One report showed that IFN-γ enhanced necrosis in bacille Calmette-Guérin–infected macrophages, implying that ESX-1 was not involved [36]. Our work by contrast used the isogenic ESX-1 mutant of M. tuberculosis to address unequivocally the role for ESX-1. It remains unclear how IFN-γ interacts with ESX-1 and enhances ESX-1's cellular effects. Among the synergistic effects of IFN-γ and ESX-1, necrosis occurred earlier than M. tuberculosis aggregation and extracellular trap formation (data not shown). Macrophages undergoing necrosis might release, actively or passively, intercellular signaling molecules that promote macrophage-macrophage interactions. Such cellular aggregation might give rise to the observed aggregation of M. tuberculosis and could also contribute to extracellular trap formation because of a relative higher M. tuberculosis burden. The mechanism through which the ESX-1–dependent necrosis is amplified by IFN-γ is unknown. M. tuberculosis–triggered necrosis involves damage to phagolysosomes in which the bacteria are residing [10]. IFN-γ may promote such damage and facilitate macrophage necrosis. However, no IFN-γ–enhanced damage to the M. tuberculosis–containing phagolysosomes was observed (data not shown). We and others have shown that ESX-1 mediates phagosomal rupture [8–10]. ESAT-6 point mutants defective in damaging phagosomes are unable to induce necrosis [10] and are unlikely to cause extracellular trap formation. The elastase inhibitor AAPV, which blocked extracellular trap formation, did not affect damage of phagosomes or necrosis (data not shown; Figure 1A). Therefore, extracellular trap formation could be a downstream effect of ESX-1–dependent phagosomal damage and necrosis. Alternatively, phagosomal damage and extracellular trap formation could be independent processes simultaneously induced by ESX-1.

t affect damage of phagosomes or necrosis (data not shown; Figure 1A). Therefore, extracellular trap formation could be a downstream effect of ESX-1–dependent phagosomal damage and necrosis. Alternatively, phagosomal damage and extracellular trap formation could be independent processes simultaneously induced by ESX-1. Vogt and Nathan have recently reported that human macrophages cultured in the presence of GM-CSF and 40% serum at a physiological O2 level (10% O2) survive M. tuberculosis infection for several weeks [27]. Key differences between their study and ours included MOIs (0.1–0.2 vs 5–10 in this study), the strains used (Erdman vs H37Rv in this study), and serum percentage (40% vs 10% in this study). With 40% serum, a low IFN-γ level (2.5 U/mL) could still amplify ESX-1–induced necrosis (data not shown). Although we did not determine IFN-γ's effect by using a lower MOI of Erdman, which is less virulent than H37Rv [46], we speculate that the more virulent the infection is, such as one involving a higher MOI and a more virulent M. tuberculosis strain, the greater the likelihood that IFN-γ can enhance M. tuberculosis–triggered necrosis.

not determine IFN-γ's effect by using a lower MOI of Erdman, which is less virulent than H37Rv [46], we speculate that the more virulent the infection is, such as one involving a higher MOI and a more virulent M. tuberculosis strain, the greater the likelihood that IFN-γ can enhance M. tuberculosis–triggered necrosis. Necrosis induced by M. tuberculosis represents a nonresolving inflammation [47]. Necrotic cells release inflammatory cytosolic contents, yet these cells are unable to reduce the M. tuberculosis burden. As this necrosis is facilitated by IFN-γ as reported here, M. tuberculosis–infected human macrophages are more likely to undergo necrosis when IFN-γ is present, such as during the adaptive immune response against M. tuberculosis antigens. Because IFN-γ's potentiating effect is ESX-1 dependent, our work suggests that targeting ESX-1 activity should prevent IFN-γ from inducing human macrophages to undergo a nonprotective necrosis. Patients with active tuberculosis have elevated IFN-γ levels [48–50]. Patients receiving aerosolized IFN-γ show clinical improvement [45]. Thus, improving IFN-γ's efficacy against M. tuberculosis might be possible by targeting ESX-1's pronecrotic effect on M. tuberculosis–infected macrophages.

a nonprotective necrosis. Patients with active tuberculosis have elevated IFN-γ levels [48–50]. Patients receiving aerosolized IFN-γ show clinical improvement [45]. Thus, improving IFN-γ's efficacy against M. tuberculosis might be possible by targeting ESX-1's pronecrotic effect on M. tuberculosis–infected macrophages. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank Cody Colon-Berezin and Melissa McCoy, for careful review of the manuscript; and Yong Chen and John Chan, for obtaining PBMCs from New York Blood Center. Financial Support. This work was supported by the National Institutes of Health (grant RO1 AI26170 and PO1 AI063537) and the Albert Einstein College of Medicine Center for AIDS Research (grant AI0-51519). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

The evolution and spread of artemisinin-based combination therapy (ACT)–resistant Plasmodium falciparum may have drastic consequences on global malaria control and elimination efforts. Development of ACT resistance is likely to start with a decreased efficacy of the long–half-life partner drug, gradually transforming ACT into an unprotected artemisinin-derivative monotherapy. This notion underlies the World Health Organization policy that artemisinin-derivative agents should be exclusively used in combination with other drugs for the treatment of uncomplicated malaria and never together with a failing drug. In high-transmission areas, the development of resistance against long–half-life partner drugs is likely to occur through the posttreatment selection of less sensitive parasites, as reinfections are exposed to subtherapeutic blood levels of these slowly eliminated drugs. The posttreatment selection of drug resistance–associated single-nucleotide polymorphisms (SNPs) observed after both artemether-lumefantrine [1–3] and artesunate-amodiaquine [4, 5] treatment, involving polymorphisms of the P. falciparum multidrug resistance gene 1 (pfmdr1; National Center for Biotechnology Information Reference Sequence gene ID 813045), are examples of this.

single-nucleotide polymorphisms (SNPs) observed after both artemether-lumefantrine [1–3] and artesunate-amodiaquine [4, 5] treatment, involving polymorphisms of the P. falciparum multidrug resistance gene 1 (pfmdr1; National Center for Biotechnology Information Reference Sequence gene ID 813045), are examples of this. To identify useful surveillance tools such as genetic markers of resistance against the long-acting partners in ACT, a clear definition of the P. falciparum resistance phenotype is needed. Unfortunately, in vivo resistance has been difficult to define with precision. Treatment failure can be due to nonparasitological factors like poor patient drug bioavailability or adherence. Furthermore, the use of biodiversity markers such as P. falciparum merozoite surface protein 1 (pfmsp1), pfmsp2, and glutamate-rich protein (glurp) to distinguish between recrudescence (treatment failure) and reinfections may be less reliable than previously expected [6, 7]. As for field in vitro methods, these are only applicable to subsets of infections with appropriate parasitemia and tend to select for high-fitness parasites. We propose a complementary concept for defining molecular markers of in vivo P. falciparum susceptibility.

s may be less reliable than previously expected [6, 7]. As for field in vitro methods, these are only applicable to subsets of infections with appropriate parasitemia and tend to select for high-fitness parasites. We propose a complementary concept for defining molecular markers of in vivo P. falciparum susceptibility. In recent years, we have witnessed important technological developments in the determination of drug levels in blood, using samples collected on filter papers. The increased robustness of these techniques has allowed routine analysis of drug levels of the ACT partner drugs on the seventh day (D7) after treatment initiation as a surrogate marker for drug exposure (ie, area under the concentration-time curve [AUC]) [8]. Long-acting antimalarial drugs (eg, lumefantrine [LUM]) are in their terminal elimination phase at D7 and have a log-linear decrease of drug concentrations [8]. This enables later drug levels to be inferred using the measured D7 concentration and the terminal elimination half-life. The accuracy of the inferred drug concentrations depends on knowledge of the pharmacokinetic characteristics of the target group, especially the terminal elimination rate constant. Estimated drug concentrations can then be correlated to genotypes (SNPs, copy number variation) of recurrent parasites. As a proof of concept, we correlated polymorphisms in pfmdr1 with estimated LUM drug concentrations in patients treated with artemether-lumefantrine.

In recent years, we have witnessed important technological developments in the determination of drug levels in blood, using samples collected on filter papers. The increased robustness of these techniques has allowed routine analysis of drug levels of the ACT partner drugs on the seventh day (D7) after treatment initiation as a surrogate marker for drug exposure (ie, area under the concentration-time curve [AUC]) [8]. Long-acting antimalarial drugs (eg, lumefantrine [LUM]) are in their terminal elimination phase at D7 and have a log-linear decrease of drug concentrations [8]. This enables later drug levels to be inferred using the measured D7 concentration and the terminal elimination half-life. The accuracy of the inferred drug concentrations depends on knowledge of the pharmacokinetic characteristics of the target group, especially the terminal elimination rate constant. Estimated drug concentrations can then be correlated to genotypes (SNPs, copy number variation) of recurrent parasites. As a proof of concept, we correlated polymorphisms in pfmdr1 with estimated LUM drug concentrations in patients treated with artemether-lumefantrine. METHODS Clinical Trials The infections analyzed in this report came from 2 artemether-lumefantrine clinical efficacy/effectiveness trials. Full details of these studies have been reported previously and are summarized in brief in Table 1 [9, 10]. Table 1. Description of the Study Populations

As a proof of concept, we correlated polymorphisms in pfmdr1 with estimated LUM drug concentrations in patients treated with artemether-lumefantrine. METHODS Clinical Trials The infections analyzed in this report came from 2 artemether-lumefantrine clinical efficacy/effectiveness trials. Full details of these studies have been reported previously and are summarized in brief in Table 1 [9, 10]. Table 1. Description of the Study Populations Variable Study 1 (n = 359) Study 2 (n = 244) Overall (n = 603) Reinfection, no. of subjects 170 84 254 Recrudescence, no. of subjects 7 10 17 pfmdr1 N86 day 0, % of subjects (pure N86/total)a 43 (155/357) 49 (115/234) 46 (270/591) pfmdr1 N86 reinfection, % of subjects (pure N86/total) 61 (101/166) 79 (65/82) 67 (166/248) Time to reinfection, d, median (95% CI)b 35 (34–36) 28 (25–31) 32 (30–34) Abbreviations: CI, confidence interval; pfmdr1, Plasmodium falciparum multidrug resistance gene 1. a The present single-nucleotide polymorphism discrimination has been previously published [9, 10]. b The difference in time to reinfection is partly explained by the different follow-up durations in study 1 (56 days) and study 2 (42 days)

Variable Study 1 (n = 359) Study 2 (n = 244) Overall (n = 603) Reinfection, no. of subjects 170 84 254 Recrudescence, no. of subjects 7 10 17 pfmdr1 N86 day 0, % of subjects (pure N86/total)a 43 (155/357) 49 (115/234) 46 (270/591) pfmdr1 N86 reinfection, % of subjects (pure N86/total) 61 (101/166) 79 (65/82) 67 (166/248) Time to reinfection, d, median (95% CI)b 35 (34–36) 28 (25–31) 32 (30–34) Abbreviations: CI, confidence interval; pfmdr1, Plasmodium falciparum multidrug resistance gene 1. a The present single-nucleotide polymorphism discrimination has been previously published [9, 10]. b The difference in time to reinfection is partly explained by the different follow-up durations in study 1 (56 days) and study 2 (42 days) Study I was a randomized 2-arm artemether-lumefantrine (AL) clinical trial (efficacy vs effectiveness) among 359 febrile patients <5 years of age (age range, 3–59 months) in the Bagamoyo district of Tanzania [9] (ClinicalTrials.gov identifier ISRCTN69189899). All patients had confirmed P. falciparum parasitemia (2000–200 000 asexual parasites/μL) on admission and were followed weekly for 56 days. For the 161 patients with recurrent infection, a second AL treatment was given, and the patient was followed weekly for an additional 42 days.

rials.gov identifier ISRCTN69189899). All patients had confirmed P. falciparum parasitemia (2000–200 000 asexual parasites/μL) on admission and were followed weekly for 56 days. For the 161 patients with recurrent infection, a second AL treatment was given, and the patient was followed weekly for an additional 42 days. Study II was a single-arm effectiveness study of the standard 6 dose AL regimen involving 244 subjects from rural Kibaha, Ngeta, and Mwanabwito districts, Tanzania (ClinicalTrials.gov identifier NCT00454961) [10]. Inclusion criteria were identical to those described for study I, with the exception of including individuals with a parasite load of <2000 asexual parasites/μL. The patients were followed weekly for 42 days. For both studies, capillary blood samples were taken at enrollment (D0), at the weekly clinical assessment, and in the event of recurrent parasitemia (R0) and preserved on 3-mm filter paper.

ption of including individuals with a parasite load of <2000 asexual parasites/μL. The patients were followed weekly for 42 days. For both studies, capillary blood samples were taken at enrollment (D0), at the weekly clinical assessment, and in the event of recurrent parasitemia (R0) and preserved on 3-mm filter paper. Patients with polymerase chain reaction (PCR)–confirmed reinfections at D7 and onward, with LUM filter paper samples collected on D7 and concentrations successfully measured (see below), were included in the present work. Recrudescences were excluded because they, by definition, have survived treatment with both artemether and LUM. Recrudescent infections therefore represent a different population of parasites. Additional details regarding the patient population are specified in Table 1. Before enrollment, written informed consent was provided by parents or guardians. Both studies were approved by the National Institute for Medical Research, Tanzania, and the Regional Ethics Committee, Stockholm, Sweden. Quantification of LUM Concentrations at D7 Seven days after treatment initiation (D7), capillary blood samples were applied on filter papers pretreated with 0.75 M tartaric acid and were stored at −20°C. LUM whole-blood concentrations were measured by solid-phase extraction and liquid chromatography, as described elsewhere [11]. A total of 530 capillary samples were quantified and 34 excluded because measurements were below the limit of detection (100 nM). The within-study assay performance showed a precision (coefficient of variation [CV]) of 6.07%–11.5%.

asured by solid-phase extraction and liquid chromatography, as described elsewhere [11]. A total of 530 capillary samples were quantified and 34 excluded because measurements were below the limit of detection (100 nM). The within-study assay performance showed a precision (coefficient of variation [CV]) of 6.07%–11.5%. Estimation of LUM Concentrations After D7 LUM elimination is in the log-linear phase after D7, and individual drug concentrations can be extrapolated to the point of interest according to the individual pharmacokinetic characteristics. Population estimates were derived from a detailed pharmacokinetic study previously performed in the same setting [12]. The population mean of the terminal elimination half-life used in this work was 80 hours. This was in close agreement with previously published data [13]. The expected drug concentrations on the day of hepatocyte burst were calculated for all patients with reinfections (Equation 1). The estimated day of hepatocyte burst was assumed to occur 7 days before microscopy-based detection of recurrent parasitemia during follow up after AL treatment. This method permitted an in vivo estimate of the reinfecting parasite's ability to multiply under drug pressure. (1) where CEST is the estimated LUM blood concentration, CD7 is the individually measured D7 LUM blood concentration (in nanomoles), k is the terminal elimination rate constant set to 0.00865 hours−1 [6], and t is the time in hours from D7 to estimated hepatocyte burst.

rasite's ability to multiply under drug pressure. (1) where CEST is the estimated LUM blood concentration, CD7 is the individually measured D7 LUM blood concentration (in nanomoles), k is the terminal elimination rate constant set to 0.00865 hours−1 [6], and t is the time in hours from D7 to estimated hepatocyte burst. Data Transformation Price et al previously defined a venous plasma LUM cutoff concentration of 331 nM (175 ng/mL) to predict recrudescence (treatment failure) with 75% sensitivity and 84% specificity [14]. We converted this value to capillary blood samples, taking into account the hematocrit for our study population, and got an equivalent LUM cutoff concentration of 328 nM (Supplementary Material). Thus, LUM concentrations >328 nM at D7 are defined here as the threshold of exposure of the parasites to an adequate AUC.

]. We converted this value to capillary blood samples, taking into account the hematocrit for our study population, and got an equivalent LUM cutoff concentration of 328 nM (Supplementary Material). Thus, LUM concentrations >328 nM at D7 are defined here as the threshold of exposure of the parasites to an adequate AUC. Molecular Analysis The pfmdr1 N86Y status of every recurrent parasitemia (R0, initial day of microscopy-determined recurrent parasitemia), previously determined through PCR–restriction fragment length polymorphism analysis [9, 10], was reanalyzed and confirmed through reamplification and direct PCR amplicon sequencing. The pfmdr1 Y184F and D1246Y SNPs were analyzed by PCR amplicon sequencing. One patient, who experienced a second recurrent infection, with estimated LUM blood concentrations of >550 nM, was also added to the analysis. The PCR success rates for analysis of codon 86, 184, and 1246 were 97% (259 of 267), 92% (245 of 267), and 90% (241 of 267), respectively. The presence of pfmdr1 copy number variation, previously associated with in vivo artemether-lumefantrine response [14, 15], has been previously tested [9], with only 1 infection identified as carrying 2 copies (86Y). Reinfections and recrudescences were distinguished using previously published stepwise genotyping of pfmsp2, followed by pfmsp1 and glurp [9, 10]. For primer sequences, see Supplementary Table 1.

lumefantrine response [14, 15], has been previously tested [9], with only 1 infection identified as carrying 2 copies (86Y). Reinfections and recrudescences were distinguished using previously published stepwise genotyping of pfmsp2, followed by pfmsp1 and glurp [9, 10]. For primer sequences, see Supplementary Table 1. Statistical Analysis The statistical analyses were done using Stata v.12 and SigmaPlot 11. Statistical significance was defined as a P value of < .05. The 95% confidence intervals were computed in STATA v.12, for binomial variables, and by SigmaPlot 11, for survival data. Normally distributed continuous data were analyzed with the Student t test. A Mann–Whitney rank sum test was used to compare estimated LUM blood concentrations for different genotypes. Mixed infections with the presence of both alleles were excluded from analysis. Only pure infections were analyzed for pfmdr1 haplotypes, and rare haplotypes present in <3 infections were excluded.

udent t test. A Mann–Whitney rank sum test was used to compare estimated LUM blood concentrations for different genotypes. Mixed infections with the presence of both alleles were excluded from analysis. Only pure infections were analyzed for pfmdr1 haplotypes, and rare haplotypes present in <3 infections were excluded. RESULTS Reinfecting parasites carrying pfmdr1 N86, 184F, or D1246 pure alleles were able to survive at significantly higher median estimated LUM blood concentrations, compared with parasites harboring their alternative alleles (Figure 1 and Table 2). The largest difference was observed for the N86Y SNP, with concentrations of 25.4 nM versus 2.08 nM (a 12.2-fold difference) between the N- and the Y-carrying parasites. For pfmdr1 Y184F and D1246Y, the corresponding differences were 4.09 nM versus 34.5 nM (an 8.4-fold difference) and 15.9 nM versus 3.23 nM (a 4.9-fold difference), respectively. Table 2. Estimated Median Lumefantrine (LUM) Blood Concentrations for Different Plasmodium falciparum Multidrug Resistance Gene 1 (pfmdr1) Single-Nucleotide Polymorphisms (SNPs) and Haplotypes pfmdr1 No. LUM CEST, nM Interquartile Range P SNP N86 166 25.4 3.85–72.7 < .001 86Y 37 2.08 0.248–4.43 184F 80 34.5 10.5–87.5 < .001 Y184 127 4.09 0.879–25.4 D1246 195 15.9 2.26–46.4 .006 1246Y 23 3.23 0.293–10.9 Haplotype NFD 64 31.4 10.5–76.1 NYD 63 15.8 2.53–46 .045 YYY 15 2.16 0.293–3.77 ≤.001 YYD 15 0.678 0.108–3.87 ≤.001 Abbreviation: CEST, estimated LUM concentration.

25.4 3.85–72.7 < .001 86Y 37 2.08 0.248–4.43 184F 80 34.5 10.5–87.5 < .001 Y184 127 4.09 0.879–25.4 D1246 195 15.9 2.26–46.4 .006 1246Y 23 3.23 0.293–10.9 Haplotype NFD 64 31.4 10.5–76.1 NYD 63 15.8 2.53–46 .045 YYY 15 2.16 0.293–3.77 ≤.001 YYD 15 0.678 0.108–3.87 ≤.001 Abbreviation: CEST, estimated LUM concentration. Combinations of pfmdr1 polymorphisms at codon N86Y, Y184F, D1246Y. Mann–Whitney rank sum test was used to compare estimated lumefantrine concentrations between SNPs and the NFD haplotype against other haplotypes. Figure 1. Estimated lumefantrine (LUM) concentrations for reinfecting Plasmodium falciparum carrying different pfmdr1 single-nucleotide polymorphisms (SNPs) at codons 86, 184, and 1246. Each reinfection is represented 3 times, once for each SNP. Only pure infections (concerning the pfmdr1 SNPs) were included in the analysis. According to the Mann–Whitney rank sum test, there was a significant difference between N86 and 86Y (P < .001), 184F and Y184 (P < .001), and D1246 and 1246Y (P = .006; Table 2). Black lines, median values; grey lines, interquartile ranges.

nly pure infections (concerning the pfmdr1 SNPs) were included in the analysis. According to the Mann–Whitney rank sum test, there was a significant difference between N86 and 86Y (P < .001), 184F and Y184 (P < .001), and D1246 and 1246Y (P = .006; Table 2). Black lines, median values; grey lines, interquartile ranges. P. falciparum with the N86/184F/D1246 haplotype was statistically significantly less sensitive than P. falciparum with the alternative haplotypes 86Y/Y184/1246Y (31.4 nM vs 2.16 nM [a 14.5-fold difference]; P < .001) and 86Y/Y184/D1246 (31.4 nM vs 0.678 nM [a 46.3-fold difference]; P < .001; Figure 2 and Table 2). There was no significant correlation between pfmdr1 haplotypes and D0 parasitemia (as a proxy marker of fitness) Figure 2. Estimated lumefantrine concentrations for reinfecting Plasmodium falciparum carrying different pfmdr1 haplotypes at codons 86, 184, and 1246. Each open circle represents a reinfection. Only haplotypes with ≥3 observations were considered for analysis. Median values were 31.4 nM (interquartile range [IQR], 10.5–76.1 nM) for NFD, 15.8 nM (IQR, 2.53–46.0 nM) for NYD, 2.16 nM (IQR, 0.293–3.77 nM) for YYY, and 0.678 nM (IQR, 0.108–3.87 nM) for YYD (Table 2). Black lines, median values; grey lines, interquartile ranges.

s with ≥3 observations were considered for analysis. Median values were 31.4 nM (interquartile range [IQR], 10.5–76.1 nM) for NFD, 15.8 nM (IQR, 2.53–46.0 nM) for NYD, 2.16 nM (IQR, 0.293–3.77 nM) for YYY, and 0.678 nM (IQR, 0.108–3.87 nM) for YYD (Table 2). Black lines, median values; grey lines, interquartile ranges. The highest estimated LUM concentration that reinfecting parasites carrying N86 versus those carrying 86Y could withstand differed by a factor of 35 (1184.3 nM and 34.3 nM, respectively; Figure 1). The influence of Y184F and D1246Y SNPs on drug susceptibility is less clear-cut, with the “sensitive” Y184 and 1246Y parasites able to withstand the highest drug levels (1184.3 nM and 1081.5 nM, respectively. There was a distinct subset of 8 parasites that were able to grow at estimated LUM concentrations of >550 nM, whereas no other parasites grew at concentrations >300 nM (Table 3). The subset represented 4.57% (8 of 175; 95% confidence interval, 1.99%–8.81%) of the reinfections occurring up to 35 days after treatment initiation. These least susceptible parasites all carried the pfmdr1 N86 allele and had LUM D7 levels 1.8–10.3-fold higher than 328 nM, confirming adequate treatment and bioavailability. Table 3. Reinfecting Plasmodium falciparum Able to Grow at Estimated Lumefantrine (LUM) Concentrations of >550 nM

ays after treatment initiation. These least susceptible parasites all carried the pfmdr1 N86 allele and had LUM D7 levels 1.8–10.3-fold higher than 328 nM, confirming adequate treatment and bioavailability. Table 3. Reinfecting Plasmodium falciparum Able to Grow at Estimated Lumefantrine (LUM) Concentrations of >550 nM Parasite Code Study LUM CEST, nM LUM CD7, nM pfmdr1 N86Y pfmdr1 Y184F pfmdr1 D1246Y F147 I 1184 1184 N Y D 11129 II 1081 1081 N F Y F26 I 706 1070 N F D F63 I 678 678 … … D 11066 II 581 581 N Y D 9096 II 794 3397 N F … F202 I 565 2416 N Y D F13 I 558 2388 N F D Polymorphisms in pfmdr1 at codon N86Y, Y184F, and D1246Y were analyzed on the day of recurrent parasitaemia. Abbreviations: CD7, measured LUM concentration 7 days after treatment initiation; CEST, estimated LUM concentrations; D, aspartic acid; F, phenylalanine; N, asparagine; Y, tyrosine; –, unsuccessful polymerase chain reaction analysis. There were no significant differences in D7 concentration between patients who were reinfected with one of these “least susceptible parasites” and those who were adequately treated and then experienced recrudescence (P = .113; Supplementary Table 2). DISCUSSION The present study allowed an estimate of in vivo susceptibility to LUM by P. falciparum and its association with pfmdr1 alleles. The data provide evidence that the observed post-AL treatment selection of pfmdr1 alleles is associated with a significant decrease in LUM susceptibility. This reinforces the hypothesis that pfmdr1 is a central player in P. falciparum resistance to LUM.

y to LUM by P. falciparum and its association with pfmdr1 alleles. The data provide evidence that the observed post-AL treatment selection of pfmdr1 alleles is associated with a significant decrease in LUM susceptibility. This reinforces the hypothesis that pfmdr1 is a central player in P. falciparum resistance to LUM. We found that pfmdr1 N86, 184F, and 1246D were associated with reduced LUM susceptibility, compared with the alternative 86Y, Y184, and Y1246 alleles. When analyzed as haplotypes, the NFD haplotype were able to withstand estimated LUM concentration 15-fold higher than those with the YYY haplotype. This is in line with previous in vivo and in vitro work [1, 2, 16]. Clinical isolates from Kenya with pfmdr1 N86 had a 2.9-fold higher median LUM median inhibitory concentration than the 86Y allele in vitro [16]. In line with the Kenyan results, we found a 12.2-fold difference between these 2 SNPs in vivo. Haplotype analysis showed a trend of decreased LUM susceptibility, in the order of NFD, NYD, YYY, and YYD. This suggests a gradually acquired tolerance, starting with N86, followed by the combination of N86 + D1246 and, thereafter, the combination of N86 + 184F + D1246. This might be comparable with the selection of SNPs in P. falciparum dihydrofolate reductase (pfdhfr) associated with a stepwise decrease in susceptibility to sulfadoxine-pyrimethamine [17]. The D0 parasitemias were similar irrespective of haplotype, supporting the hypothesis that differences in drug tolerability is a measure of reduced drug susceptibility as opposed to fitness.

ciparum dihydrofolate reductase (pfdhfr) associated with a stepwise decrease in susceptibility to sulfadoxine-pyrimethamine [17]. The D0 parasitemias were similar irrespective of haplotype, supporting the hypothesis that differences in drug tolerability is a measure of reduced drug susceptibility as opposed to fitness. An important question is whether we have identified LUM-resistant parasites. We identified P. falciparum parasites able to survive at levels near or above blood drug concentrations of 1 µM. Such capacity to withstand drug pressure means that these parasites are able to start proliferating just 2 days after completion of the AL treatment, when the LUM blood levels drop to these levels [18]. Irrespective of exact concentrations, taking into account sampling errors, such a collapse of protection capacity is probably paving the way for the emergence of fully resistant parasites (if they are not already present). This is in agreement with a previously proposed model describing AL-driven pfmdr1 SNP selection in Africa [18]: the parasite is developing its way of “climbing” the pharmacokinetic curve. As discussed above, the N86 allele seems to be fundamental to—albeit not sufficient for—this process, to which other pfmdr1 SNPs (184F and D1246) and additional, as-yet-unveiled genetic changes add.

L-driven pfmdr1 SNP selection in Africa [18]: the parasite is developing its way of “climbing” the pharmacokinetic curve. As discussed above, the N86 allele seems to be fundamental to—albeit not sufficient for—this process, to which other pfmdr1 SNPs (184F and D1246) and additional, as-yet-unveiled genetic changes add. The proposed in vivo genotype/phenotype association method has the advantage of being independent of constraints associated with the definitions of recrudescence and clinical failure [6, 19], drug bioavailability, and other issues not directly related to parasitological drug sensitivity. The method provides an in vivo estimate of the capacity of parasites to evade drug action. It has an in-built high specificity, since no sensitive parasites are expected to be detected at high blood drug levels. This high specificity is expected to be an important factor for the unambiguous identification of tolerant/resistant P. falciparum. The method will underestimate the proportion of parasites with reduced drug susceptibility, because such parasites, as well as fully susceptible parasites, will thrive when drug concentrations are low. The method will thereby be prone to false-negative findings. Further meta-analysis of clinical trials using this method will increase the possibility to reliable identify resistant parasites and resistance-associated SNPs.

parasites, as well as fully susceptible parasites, will thrive when drug concentrations are low. The method will thereby be prone to false-negative findings. Further meta-analysis of clinical trials using this method will increase the possibility to reliable identify resistant parasites and resistance-associated SNPs. In this work, we have benefited from a previous detailed pharmacokinetic study performed in the same area on a similar study population [12]. This provided the necessary data for inferring LUM concentrations beyond D7. Such pharmacokinetic data from the specific target populations is generally not available in malaria settings. We suggest that for the application of the proposed concept, future ACT efficacy/effectiveness trials should preferably include at least 2 drug level assessment points (eg, at D7 and D14), thereby defining an individual slope of drug elimination for each patient. This would overcome the limitation in this study of using the mean population terminal elimination half-life rather than individual values to extrapolate individual D7 levels to the time of hepatocyte burst. Because of the pharmacokinetics of LUM, we expect a significant number of patients to show quantifiable concentrations of this antimalarial 2 weeks after treatment initiation [20]. Additionally, recent developments in liquid chromatography–mass spectrometry [21] promise improvements of least 1 order of magnitude (ie, down to 1–10 nM) in lower limits of detection. This will make the determination of the D14 LUM concentrations feasible in a large majority of patients.

ter treatment initiation [20]. Additionally, recent developments in liquid chromatography–mass spectrometry [21] promise improvements of least 1 order of magnitude (ie, down to 1–10 nM) in lower limits of detection. This will make the determination of the D14 LUM concentrations feasible in a large majority of patients. Our proof of concept was applied to artemether-lumefantrine clinical trials. This method is, however, equally valuable for studying emerging resistance toward all ACT long-acting partner drugs. The use of drug concentrations can help in accurate interpretation of clinical trial outcomes and will also give an improved definition of the phenotype associated with reduced susceptibility. The concept can also be used to identify genotypes associated with reduced susceptibility to Plasmodium vivax, where ex vivo/in vitro work is limited. In conclusion, we present a new concept using D7 drug concentrations and pharmacokinetic data to estimate the drug concentrations that parasites withstand in vivo. We found that reinfecting parasites with the pfmdr1 N86/184F/D1246 haplotype were able to withstand LUM blood concentrations 15-fold higher than parasites with the pfmdr1 86Y/Y184/1246Y, supporting the role of pfmdr1 in LUM susceptibility. Our method to correlate drug concentrations and genotypes is applicable to all antimalarial drugs, can contribute to the early detection of reduced drug susceptibility, and represents a novel way for unveiling molecular markers of antimalarial drug resistance.

1246Y, supporting the role of pfmdr1 in LUM susceptibility. Our method to correlate drug concentrations and genotypes is applicable to all antimalarial drugs, can contribute to the early detection of reduced drug susceptibility, and represents a novel way for unveiling molecular markers of antimalarial drug resistance. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. This report is dedicated to the late Niklas Lindegårdh. This work was only possible because of his key role in the development of methods for determination of drug levels in filter-paper-preserved blood samples. His comments on the present report were highly valuable. Financial support. This work was supported by the Swedish Development Cooperation Agency-Department for Research Cooperation (SIDA-SAREC; SWE 2004–3850, Bil-Tz 16/9875007059 and SWE-2009-165), the World Health Organization MIM-TDR (protocol ID: [A60100] MAL IRM 06 03), the Goljes Foundation, and the Swedish medical research council (K2010-56X-21457-01-3). The Mahidol-Oxford Tropical Medicine Research Unit is supported by the Wellcome Trust of Great Britain. Potential conflicts of interest. All authors: No reported conflicts.

Financial support. This work was supported by the Swedish Development Cooperation Agency-Department for Research Cooperation (SIDA-SAREC; SWE 2004–3850, Bil-Tz 16/9875007059 and SWE-2009-165), the World Health Organization MIM-TDR (protocol ID: [A60100] MAL IRM 06 03), the Goljes Foundation, and the Swedish medical research council (K2010-56X-21457-01-3). The Mahidol-Oxford Tropical Medicine Research Unit is supported by the Wellcome Trust of Great Britain. Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed

Dengue is an acute illness caused by 1 of 4 single-stranded positive-sense RNA viruses and is the commonest arboviral infection of humans. In countries where dengue is endemic, the case burden strains already fragile healthcare systems and has an economic cost [1, 2]. There are currently no licensed vaccines for dengue (although late-stage trials are in progress), and mosquito vector control has been mostly unsuccessful or unsustainable. Clinically apparent dengue manifests with a spectrum of symptoms. High fever, erythema, headache, and myalgia are common symptoms, and laboratory findings of leukopenia and mild thrombocytopenia are typical. The critical phase occurs around the time of defervescence, typically on days 4–6 of illness, during which a transient capillary permeability syndrome manifests in some patients. In children particularly, capillary permeability can be significant enough to precipitate life-threatening circulatory shock, called dengue shock syndrome (DSS). Treatment is supportive, and the mortality rate for DSS in experienced hospital settings is <1% [2].

pillary permeability syndrome manifests in some patients. In children particularly, capillary permeability can be significant enough to precipitate life-threatening circulatory shock, called dengue shock syndrome (DSS). Treatment is supportive, and the mortality rate for DSS in experienced hospital settings is <1% [2]. The magnitude of the early dengue virus (DENV) burden in patients with dengue has been associated with overall clinical outcome. For example, the early plasma viremia and/or NS1 antigenemia levels in pediatric dengue patients who develop clinically significant capillary permeability are higher than in patients without this complication [3–6]. The higher antigenic burden in these patients is believed to trigger a cascade of immunological events that promotes capillary permeability [7]. The association between high viral burdens in the first few days of illness and more severe outcomes has encouraged antiviral discovery efforts for dengue [8, 9], with the rationale that a reduction of the viral burden should result in a reduced incidence of severe complications and a lessening of symptoms and illness duration.

iation between high viral burdens in the first few days of illness and more severe outcomes has encouraged antiviral discovery efforts for dengue [8, 9], with the rationale that a reduction of the viral burden should result in a reduced incidence of severe complications and a lessening of symptoms and illness duration. Balapiravir is a prodrug of a nucleoside analogue (4′-azidocytidine) called R1479 and was developed for the treatment of chronic hepatitis C Virus (HCV) infection by Hoffmann-La Roche. [10–12]. Monotherapy twice per day for 14 days reduced plasma HCV levels in a dose- and time-dependent manner and was well-tolerated at doses up to 3000 mg in adult male patients [13]. However, the clinical development of balapiravir for HCV infection was stopped when clinical safety signals were detected in patients receiving extended courses (2–3 months) of balapiravir therapy in conjunction with pegylated interferon and ribavirin. Because HCV and DENV possess RNA-dependent RNA polymerases that share a similar overall architecture [14], we explored a new indication for balapiravir by testing the in vitro activity of R1479 against DENV. Subsequently, the safety, tolerability, and antiviral efficacy of balapirivir in adult dengue patients were investigated in a clinical trial.

RNA-dependent RNA polymerases that share a similar overall architecture [14], we explored a new indication for balapiravir by testing the in vitro activity of R1479 against DENV. Subsequently, the safety, tolerability, and antiviral efficacy of balapirivir in adult dengue patients were investigated in a clinical trial. METHODS Ethics Statement The study was conducted according to International Conference on Harmonisation Good Clinical Practice guidelines. All patients provided written informed consent. The trial protocol was approved by the Oxford University Tropical Research Ethical Committee and the Scientific and Ethical Committee of the Ministry of Health, Vietnam. The trial was registered at http://www.clinicaltrials.gov (NCT01096576). Clinical Studies Patient Enrollment Adult men attending the outpatient department of the Hospital for Tropical Diseases, Ho Chi Minh City, Vietnam, who presented with clinical suspicion of dengue confirmed by positive NS1 rapid test (NS1 STRIP, Bio-Rad) were invited to participate in the study. The patient inclusion criteria for study enrollment were (1) male patient aged 18–65 years, (2) history or presence of fever (temperature, ≥38°C) coinciding with a clinical suspicion of DENV infection and a positive NS1 rapid test, (3) onset of symptoms <48 hours prior to initial dosing, (4) and body mass index of 18–35. Two forms of contraception were required for male patients and their partners of childbearing potential, and written informed consent was obtained before any study-specific procedures were performed.

nd a positive NS1 rapid test, (3) onset of symptoms <48 hours prior to initial dosing, (4) and body mass index of 18–35. Two forms of contraception were required for male patients and their partners of childbearing potential, and written informed consent was obtained before any study-specific procedures were performed. The patient exclusion criteria included (1) clinically significant abnormal laboratory test results which were deemed to be unassociated with dengue infection, or alternatively were diagnostic of dengue shock syndrome; (2) clinical evidence or a history of clinically significant respiratory, metabolic, cardiac, renal, hepatic, gastrointestinal, hematological, neurological, psychiatric, or chronic disease; (3) positive test result for human immunodeficiency virus (HIV) at screening; (4) history of autoimmune or immune dysfunction disease; (5) neutrophil count of <1500 cells/mm3, Haemoglobin (Hgb) concentration of <13 g/dL, or platelet count of <90 000 cells/mm3 at screening; (6) calculated creatinine clearance of <80 mL/minutes; (7) hypertension or hypotension; (8) pulse pressure of <20 mm Hg; (9) essential concomitant medication with the exception of paracetamol; (10) positive test result for drugs of abuse or alcohol; (11) recent participation in an investigational drug or device study; and (12) not being a suitable candidate for enrollment in the opinion of the investigator or sponsor.

ssure of <20 mm Hg; (9) essential concomitant medication with the exception of paracetamol; (10) positive test result for drugs of abuse or alcohol; (11) recent participation in an investigational drug or device study; and (12) not being a suitable candidate for enrollment in the opinion of the investigator or sponsor. Randomization, Masking, Dosing Schedule, and Dose Escalation Cohorts Randomization to treatment group was by computer-generated randomization sequencing in blocks of 2. Balapiravir was formulated in a 500-mg film-coated tablet. The placebo consisted of an identical looking tablet containing excipient only. All balapiravir and placebo tablets were provided by F. Hoffmann-La Roche. Patients in cohort 1 (n = 20) received 1500 mg of balapiravir or an identical placebo orally every 12 hours for 5 days (10 doses). The decision to dose escalate (to 3000 mg) was made after review of the partially unblinded group mean clinical and laboratory data acquired from patients in cohort 1. Patients in cohort 2 (n = 44) received 3000 mg of balapiravir or placebo orally twice a day for 5 days (10 doses).

every 12 hours for 5 days (10 doses). The decision to dose escalate (to 3000 mg) was made after review of the partially unblinded group mean clinical and laboratory data acquired from patients in cohort 1. Patients in cohort 2 (n = 44) received 3000 mg of balapiravir or placebo orally twice a day for 5 days (10 doses). Safety Assessments Eligible patients remained in clinic during a 7-day in-patient period then returned for follow-up visits on study days 14, 28, and 84. During the inpatient period, patient vital signs were assessed every 6 hours and routine laboratory investigations were performed daily or as clinically indicated. Safety and tolerability were monitored at regular intervals and included physical examinations, vital signs, electrocardiograms, clinical laboratory assessments, incidence of clinical adverse events, and concomitant medications. Any clinically significant abnormal laboratory test results were followed up until resolved or stabilized. A quality of life assessment, using a questionnaire and a visual analog scale, was implemented on study days 1, 3, 5, 7, 14, 28, and 84. Clinical Laboratory Investigations Pharmacokinetics Serum samples for pharmacokinetic investigations were collected at time 0 (predose) and 2, 4, 8, and 12 hours postdose on days 1 and 5. Pharmacokinetics parameters were calculated using noncompartmental methods.

Safety Assessments Eligible patients remained in clinic during a 7-day in-patient period then returned for follow-up visits on study days 14, 28, and 84. During the inpatient period, patient vital signs were assessed every 6 hours and routine laboratory investigations were performed daily or as clinically indicated. Safety and tolerability were monitored at regular intervals and included physical examinations, vital signs, electrocardiograms, clinical laboratory assessments, incidence of clinical adverse events, and concomitant medications. Any clinically significant abnormal laboratory test results were followed up until resolved or stabilized. A quality of life assessment, using a questionnaire and a visual analog scale, was implemented on study days 1, 3, 5, 7, 14, 28, and 84. Clinical Laboratory Investigations Pharmacokinetics Serum samples for pharmacokinetic investigations were collected at time 0 (predose) and 2, 4, 8, and 12 hours postdose on days 1 and 5. Pharmacokinetics parameters were calculated using noncompartmental methods. Virological and Immunological Measurements Plasma samples were collected for virological and immunological investigations every 12 hours, beginning immediately before commencement of treatment, during the next 6 study days and on one occasion on study day 7 and day 14 (virological markers) or day 7 and day 28 (immunological markers). Viremia was measured using a validated, internally controlled reverse-transcription polymerase chain reaction assay in a Good Clinical Laboratory Practice environment [15]. The limit of detection was 357 copies/mL for DENV-1, 72 copies/mL for DENV-2, 357 copies/mL for DENV-3, and 720 copies/mL for DENV-4. The presence of NS1 in plasma was determined using the Platelia NS1 assay (Bio-Rad) and was performed according to the manufacturer's instructions. Plasma cytokine levels (interleukin 1β [IL-1β], IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, IL-13, interferon γ [IFN-γ], and tumor necrosis factor α [TNF-α]) were measured using a Bio-Plex human cytokine assay (Bio-Rad) and a multiplex array reader (Luminex Systems, Bio-Plex workstation from Bio-Rad Laboratories) according to the manufacturer's instructions. Briefly, 50-μL plasma samples were incubated with monoclonal antibody coupled beads. Complexes were washed twice, then incubated with biotinylated detection antibodies and, finally, labeled with streptavidin-phycoerythrin prior to analysis. Cytokine concentrations in samples were calculated by use of recombinant cytokines as standards and software provided by the manufacturer (Bio-Plex Manager). All research samples (for pharmacology, virology, and immunology) were collected and processed in the laboratory within 1 hour of venupuncture. All virological and immunological measurements and analyses were performed by analysts who were blind to the treatment assignment

ovided by the manufacturer (Bio-Plex Manager). All research samples (for pharmacology, virology, and immunology) were collected and processed in the laboratory within 1 hour of venupuncture. All virological and immunological measurements and analyses were performed by analysts who were blind to the treatment assignment Statistical Methods All randomized patients in the study were analyzed according to the intention-to-treat principle with 3 treatment groups: 1500 mg of balapiravir, 3000 mg of balapirivir, and placebo (combining the patients in the placebo arms of both cohorts). Key viremia endpoints of the study were as follows: area under the log-transformed viremia curve (AUC) from first dose to the end of study day 7 (study hour 168), calculated on the basis of the trapezoidal rule with values below the limit of detection replaced by half of the detection limit; time to first viremia level of <1000 copies/mL until study day 7; and time to the first negative NS1 test result. Other predefined key endpoints included fever clearance time, defined as the time from the start of treatment to the start of the first 48-hour period during which axillary temperature remained <37.5°C; maximum hematocrit level; maximum percentage increase of hematocrit level from baseline; platelet count nadir; and lowest recorded quality of life score.

ded fever clearance time, defined as the time from the start of treatment to the start of the first 48-hour period during which axillary temperature remained <37.5°C; maximum hematocrit level; maximum percentage increase of hematocrit level from baseline; platelet count nadir; and lowest recorded quality of life score. Time to event endpoints were compared between treatment groups on the basis of Cox regression and continuous endpoints on the basis of linear regression. Analyses were adjusted for the predose value of the respective endpoint; viremia and NS1 endpoints were additionally adjusted for dengue serotype. For all endpoints, we report P values of a trend test for a dose-response relationship with treatment entered as a continuous variable with values 0 (placebo), 1 (low-dose), and 2 (high-dose) into the regression model.

of the respective endpoint; viremia and NS1 endpoints were additionally adjusted for dengue serotype. For all endpoints, we report P values of a trend test for a dose-response relationship with treatment entered as a continuous variable with values 0 (placebo), 1 (low-dose), and 2 (high-dose) into the regression model. The sample size of the study was determined by practical and clinical considerations and not based on formal statistical power calculations. However, we used simulation to post hoc estimate the power of this trial to detect different effect sizes, expressed as log10 viremia reductions per day due to active drug. We assumed that the effect of high-dose drug is twice as large as the low-dose effect and that the 32 recruited placebo patients are representative of the entire target population. We simulated drug effects of varying size on top of the observed viremia profiles for these 32 patients and then used bootstrap simulation to assess the power of an overall comparison of the AUC of log10 viremia adjusted for serotype and baseline log10 viremia between study arms (trend test) and of a pairwise comparison of high dose versus placebo. According to these simulations, the study would have had 80% power to detect a true reduction of 0.25 log10 viremia per day (ie, one log10 reduction over 4 days) in the high-dose group by both the trend test and the pairwise comparison at the 2-sided 5% significance level. All analyses were performed with the statistical software R, version 2.11.1 (R Foundation for Statistical Computing).

The sample size of the study was determined by practical and clinical considerations and not based on formal statistical power calculations. However, we used simulation to post hoc estimate the power of this trial to detect different effect sizes, expressed as log10 viremia reductions per day due to active drug. We assumed that the effect of high-dose drug is twice as large as the low-dose effect and that the 32 recruited placebo patients are representative of the entire target population. We simulated drug effects of varying size on top of the observed viremia profiles for these 32 patients and then used bootstrap simulation to assess the power of an overall comparison of the AUC of log10 viremia adjusted for serotype and baseline log10 viremia between study arms (trend test) and of a pairwise comparison of high dose versus placebo. According to these simulations, the study would have had 80% power to detect a true reduction of 0.25 log10 viremia per day (ie, one log10 reduction over 4 days) in the high-dose group by both the trend test and the pairwise comparison at the 2-sided 5% significance level. All analyses were performed with the statistical software R, version 2.11.1 (R Foundation for Statistical Computing). RESULTS In Vitro Inhibition of DENV Replication by R1479 in Huh-7 Cells R1479, the active nucleoside released from balapiravir, inhibited replication of DENV reference strains and clinical isolates in Huh-7 cells with mean half maximal effective concentration (EC50) values of 1.9–11 μM (Supplementary Table 1). Mean values of median inhibitory concentration (IC50) for NITD008, a previously characterized nucleoside analogue polymerase inhibitor of DENV [16], were 0.1–0.6 μM (Supplementary Table 1). R1479 was also active against DENV-1, DENV-2, and DENV-4 (DENV-3 was not tested) in primary human macrophages (mean EC50 range, 1.3–3.2 μM) and dendritic cells (mean EC50 range, 5.2–6.0 μM; data not shown). These data suggest R1479 has activity against DENV in vitro at EC50 concentrations pharmacologically attainable in adult humans receiving ≥1500 mg of balapiravir twice per day [13].

e score, measured on a visual analogue scale, was also similar between the treatment groups (Table 4). Collectively, these results indicate a lack of measurable effect of balapiravir on these virological, clinical, and routine laboratory markers. Table 4. Analysis of Predefined Clinical and Routine Laboratory Endpoints Key Endpoint Placebo (n = 32) Low-dose Balapiravir (n = 10) High-dose Balapiravir (n = 22) P Median fever clearance time, d (IQR)a 3 (3–4) 3 (2–4) 3 (2–4) .286 Maximum recorded hematocrit level, % Mean 46.0 46.7 44.6 .102 Median (IQR) 45.9 (44.0–47.3) 45.8 (43.8–49.3) 45.1 (41.8–46.1) Maximum percentage increase of hematocrit level, % Mean 7.0 11.5 2.6 .131 Median (IQR) 6.7 (0.8–10.0) 11.8 (3.1–15.3) 1.2 (−4.1–9.2) Minimum platelet count, 109 cells/L Mean 52.6 26.3 54.4 .886 Median (IQR) 47.5 (28.5–64.0) 23 (16.5–33.0) 52 (25.5–64.8) Maximum INR Mean 1.23 1.17 1.27 .819 Median (IQR) 1.15 (1.08–1.26) 1.16 (1.05–1.26) 1.17 (1.09–1.22) Maximum aPTT, s Mean 42.2 41.3 41.8 .962 Median (IQR) 40.7 (39.0, 44.5) 40.6 (39.6–43.1) 41.8 (38.3–44.1) Minimum fibrinogen level, g/L Mean 1.92 1.94 1.88 .902 Median (IQR) 1.99 (1.63–2.14) 2.08 (1.68–2.25) 1.86 (1.60–2.03) Maximum AST level, IU/L Mean 134.0 143.6 118.4 .646 Median (IQR) 97.0 (70.0–191.5) 157.5 (116.5–171.5) 108.0 (45.0–151.5) Maximum ALT level, IU/L Mean 111.6 96.3 75.2 .125 Median (IQR) 87.5 (53.3–149.0) 97.5 (58.0–128.8) 68.5 (33.3–104.5) Minimum quality of life scorea Mean 76.5 78.3 74.45 .478 Median (IQR) 84.5 (69.8–89.3) 77.0 (69.8–89.0) 77.5 (67.8–82.0) Abbreviations: ALT, alanine aminotransferase; aPTT, activated partial thromboplastin time; AST, aspartate aminotransferase; IQR, interquartile range; INR, international normalized ratio.

not tested) in primary human macrophages (mean EC50 range, 1.3–3.2 μM) and dendritic cells (mean EC50 range, 5.2–6.0 μM; data not shown). These data suggest R1479 has activity against DENV in vitro at EC50 concentrations pharmacologically attainable in adult humans receiving ≥1500 mg of balapiravir twice per day [13]. Randomized, Double-Blind Placebo Controlled Trial of Balapiravir in Adult Male Dengue Patients A total of 64 adult patients with dengue were randomly assigned to receive either balapiravir or placebo within 48 hours of symptom onset. Patients in cohort 1 were enrolled from 15 July 2010 through 10 September 2010 and received 1500 mg of balapiravir (n = 10) or placebo (n = 10) orally twice a day for 5 days (Figure 1). Patients in cohort 2 were enrolled from 1 October 2010 through 16 January 2011 and received 3000 mg of balapiravir (n = 22) or placebo (n = 22) orally twice a day for 5 days (Figure 1). All enrolled patients completed their schedule of study drug doses. Figure 1. Study enrollment and follow-up. The study enrolled 120 NS1-positive dengue patients who consented and were eligible to undergo screening against the inclusion and exclusion criteria; of these, 64 were eligible to continue into the study. The most common reasons for exclusion were low creatinine clearance rates (53.6% of excluded cases), hepatitis B surface antigen positivity (23.2%), and abnormal serum creatinine level (17.9%). All patients completed their dosage schedule of study drug.

n criteria; of these, 64 were eligible to continue into the study. The most common reasons for exclusion were low creatinine clearance rates (53.6% of excluded cases), hepatitis B surface antigen positivity (23.2%), and abnormal serum creatinine level (17.9%). All patients completed their dosage schedule of study drug. Baseline Characteristics of the Patients The baseline characteristics of the intention-to-treat patient population (n = 64 total) were similar in the balapiravir and placebo groups (Table 1). The median duration of illness at the time of commencing treatment was <40 hours for patients in each arm of the study. DENV-1 and DENV-2 accounted for the majority of infections (Table 1). Phylogenetic analyses of genome-length virus sequences determined directly from plasma samples from 40 (62.5%) of 64 patients identified the DENV-1 viruses (n = 24 patients) as belonging to the genotype 1 lineage, the DENV-2 viruses (n = 11 patients) as belonging to the Asian 1 lineage, and DENV-3 viruses (n = 5 patients) as belonging to the genotype 2 lineage (data not shown). Table 1. Baseline Characteristics of the Patient Population

ients identified the DENV-1 viruses (n = 24 patients) as belonging to the genotype 1 lineage, the DENV-2 viruses (n = 11 patients) as belonging to the Asian 1 lineage, and DENV-3 viruses (n = 5 patients) as belonging to the genotype 2 lineage (data not shown). Table 1. Baseline Characteristics of the Patient Population Variable Placebo (n = 32) Balapiravir 1500 mg (n = 10) Balapiravir 3000 mg (n = 22) Age, years 23 (21.0–28.0) 29.5 (23.5–32.0) 21 (20.0–31.3) Weight, kg 60 (54.8–65.0) 64.5 (54.6–74.3) 58.5 (53.3–66.4) Duration of illness at dosing, h 39.25 (36.11–44.89) 33.63 (27.5–45.13) 37.75 (30.85–43.38) Oral temperature, °C 38.2 (37.6–39.1) 38.8 (38.1–39.3) 38.7 (38.2–39.0) Serotype, no. of patients/plasma viremia level (range), log10 copies/mL DENV-1 18/9.07 (4.63–10.52) 5/8.91 (7.50–10.63) 9/9.34 (8.05–9.86) DENV-2 9/7.89 (6.51–9.58) 4/8.23 (7.57–9.14) 8/8.61 (8.03–9.46) DENV-3 2/9.79 (9.62–9.96) 0 3/9.07 (8.79–9.17) DENV-4 3/8.12 (6.12–8.37) 1/5.43 (5.43–5.43) 2/7.84 (7.58–8.10) Serology, no. (%) of patients Primary 4 (12) 1 (10) 0 (0) Secondary 28 (88) 9 (90) 22 (100) Data are median (interquartile range), unless otherwise specified.

(6.51–9.58) 4/8.23 (7.57–9.14) 8/8.61 (8.03–9.46) DENV-3 2/9.79 (9.62–9.96) 0 3/9.07 (8.79–9.17) DENV-4 3/8.12 (6.12–8.37) 1/5.43 (5.43–5.43) 2/7.84 (7.58–8.10) Serology, no. (%) of patients Primary 4 (12) 1 (10) 0 (0) Secondary 28 (88) 9 (90) 22 (100) Data are median (interquartile range), unless otherwise specified. Intention-to-Treat Analysis of the Safety and Tolerability of Balapiravir A summary of the clinically significant adverse events by treatment arm is shown in Supplementary Table 2. No major clinical or laboratory safety signals or significant differences in adverse event profiles between treatment groups were observed. The range of adverse events reported was consistent with the known clinical and laboratory features of dengue. There were 4 reported serious adverse events (SAEs), 2 in the placebo arm and 1 each in the 1500 mg and 3000 mg balapiravir arms (Supplementary Table 2). These SAEs were typical of dengue (2 patients with prolonged thrombocytopenia, 1 patient with transient loss of visual acuity, and 1 patient with narrowed pulse pressure). No SAEs were considered to be dose-related to balapiravir, nor was the dose of study drug altered in response to these SAEs. All SAEs resolved.

2). These SAEs were typical of dengue (2 patients with prolonged thrombocytopenia, 1 patient with transient loss of visual acuity, and 1 patient with narrowed pulse pressure). No SAEs were considered to be dose-related to balapiravir, nor was the dose of study drug altered in response to these SAEs. All SAEs resolved. Pharmacological Profile and Antiviral Activity of Balapiravir in Dengue In the first 12 hours of treatment, 95% of patients receiving 3000 mg of balapiravir had plasma maximum concentration (Cmax) values of R1479 that were >6 μM (Table 2), a concentration that was inhibitory to DENV in vitro (Supplementary Table 1). Values of Cmax were lower in patients receiving 1500 mg of balapiravir, but nonetheless were >6 μM in most patients (Table 2). The C12 h values were 2.7–4.9 μM for 1500 mg and 2.1–15.7 μM for 3000 mg on day 1 of treatment. Similar dose-dependent pharmacological findings were observed on study day 5 (results not shown). Despite the pharmacological evidence that balapiravir treatment elicited dose-dependent levels of R1479 in vivo, there was no measurable effect on the predefined virological endpoints of time to clearance of viremia (Table 3; Figure 2), time to clearance of NS1 antigenemia (Table 3; Figure 3), or AUC (Table 3). Table 2. Pharmacokinetics of R1479 on Study Day 1

balapiravir treatment elicited dose-dependent levels of R1479 in vivo, there was no measurable effect on the predefined virological endpoints of time to clearance of viremia (Table 3; Figure 2), time to clearance of NS1 antigenemia (Table 3; Figure 3), or AUC (Table 3). Table 2. Pharmacokinetics of R1479 on Study Day 1 Parameter Tmax, ha Cmax, μMb Cmin, μMb Tlast, hc AUClast, h × μM Cohort 1 (1500 mg twice daily) Minimum 2 5.46 2.71 12 42.63 Median 4 16.54 3.56 12 110.56 Maximum 8 19.76 4.93 12 130.70 CV, % 49.1 30.5 21.3 … 55.81 Cohort 2 (3000 mg twice daily) Minimum 2 8.78 2.14 12 70.05 Median 4 23.86 5.82 12 167.04 Maximum 12 90.60 15.73 12 608.09 CV, % 55.3 60 45.5 … 116.89 Abbreviations: AUClast, area under the log-transformed viremia curve; Cmax, maximum plasma concentration of R1479 on day 1; Cmin, minimum plasma concentration of R1479 on day 1; CV, coefficient of variation Tlast, time since treatment when the last sample for pharmacokinetic measurement was collected; Tmax, time after treatment when the maximum plasma concentration of R1479 was reached. Table 3. Analysis of Predefined Virological Endpoints in Patients Treated With Balapiravir or Placebo

Parameter Tmax, ha Cmax, μMb Cmin, μMb Tlast, hc AUClast, h × μM Cohort 1 (1500 mg twice daily) Minimum 2 5.46 2.71 12 42.63 Median 4 16.54 3.56 12 110.56 Maximum 8 19.76 4.93 12 130.70 CV, % 49.1 30.5 21.3 … 55.81 Cohort 2 (3000 mg twice daily) Minimum 2 8.78 2.14 12 70.05 Median 4 23.86 5.82 12 167.04 Maximum 12 90.60 15.73 12 608.09 CV, % 55.3 60 45.5 … 116.89 Abbreviations: AUClast, area under the log-transformed viremia curve; Cmax, maximum plasma concentration of R1479 on day 1; Cmin, minimum plasma concentration of R1479 on day 1; CV, coefficient of variation Tlast, time since treatment when the last sample for pharmacokinetic measurement was collected; Tmax, time after treatment when the maximum plasma concentration of R1479 was reached. Table 3. Analysis of Predefined Virological Endpoints in Patients Treated With Balapiravir or Placebo Endpoint Placebo (n = 32) Low-dose Balapiravir (n = 10) High-dose Balapiravir (n = 22) P AUC viremia, log10 copies/mL × da .623 Mean 32.78 34.49 32.56 Median (IQR) 32.19 (26.63–39.24) 29.63 (27.41–39.86) 31.98 (27.60–34.98) Median time to first viremia level of <1000 copies/mL, d (IQR)b 4 (3–6) 5 (4, NA) 4 (3–5) .476 Median time to first negative NS1 test result, d (IQR)b 4 (3–13) 3 (3–14) 4 (3–6) .852 a Values of area under the curve (AUC) of log10 viremia from day 1 first dose (hour 0) to the end of day 7 (hour 168) calculated with the trapezoidal rule. IQR, interquartile range. b Kaplan-Meier estimates based on data from the inpatient period only.

Endpoint Placebo (n = 32) Low-dose Balapiravir (n = 10) High-dose Balapiravir (n = 22) P AUC viremia, log10 copies/mL × da .623 Mean 32.78 34.49 32.56 Median (IQR) 32.19 (26.63–39.24) 29.63 (27.41–39.86) 31.98 (27.60–34.98) Median time to first viremia level of <1000 copies/mL, d (IQR)b 4 (3–6) 5 (4, NA) 4 (3–5) .476 Median time to first negative NS1 test result, d (IQR)b 4 (3–13) 3 (3–14) 4 (3–6) .852 a Values of area under the curve (AUC) of log10 viremia from day 1 first dose (hour 0) to the end of day 7 (hour 168) calculated with the trapezoidal rule. IQR, interquartile range. b Kaplan-Meier estimates based on data from the inpatient period only. Figure 2. Viremia levels in balapiravir- and placebo-treated patients. Shown are serotype-stratified viremia levels, measured by reverse-transcription polymerase chain reaction, in 12-hour spaced plasma samples, in patients treated with placebo, low-dose balapiravir, or high-dose balapiravir. The colored lines in each graph represent smoothing lines derived from local polynomial regression fitting to data from each treatment arm. The gray background lines represent individual patient data. Abbreviations: DENV, dengue virus; R.arm, randomistation arm.

placebo, low-dose balapiravir, or high-dose balapiravir. The colored lines in each graph represent smoothing lines derived from local polynomial regression fitting to data from each treatment arm. The gray background lines represent individual patient data. Abbreviations: DENV, dengue virus; R.arm, randomistation arm. Figure 3. Kaplan-Meier plot of NS1 antigenemia over time in balapiravir- and placebo-treated patients. Shown are the proportions of patients over time that tested NS1-positive in serial plasma samples collected daily from baseline (pretreatment) to study day 7, and again on study day 14. There was no significant difference in time to clearance of NS1 antigenemia between treatment groups. Abbreviation: R.arm, randomistation arm.

e the proportions of patients over time that tested NS1-positive in serial plasma samples collected daily from baseline (pretreatment) to study day 7, and again on study day 14. There was no significant difference in time to clearance of NS1 antigenemia between treatment groups. Abbreviation: R.arm, randomistation arm. Balapiravir Effects on Clinical Signs and Routine Laboratory Markers Consistent with a lack of measurable antiviral activity, balapiravir treatment did not affect clinical signs or routine laboratory findings (summary endpoint findings in Table 4). Thus, fever clearance times (Supplementary Figure 1; Table 4) and changes in hematological markers such as platelet count (Table 4) or hematocrit level (Table 4) were not affected by balapiravir treatment. Similarly, the kinetics of biochemical changes (liver transaminase and coagulation marker level) were similar in each of the treatment arms (Table 4). A quality of life score, measured on a visual analogue scale, was also similar between the treatment groups (Table 4). Collectively, these results indicate a lack of measurable effect of balapiravir on these virological, clinical, and routine laboratory markers. Table 4. Analysis of Predefined Clinical and Routine Laboratory Endpoints

(33.3–104.5) Minimum quality of life scorea Mean 76.5 78.3 74.45 .478 Median (IQR) 84.5 (69.8–89.3) 77.0 (69.8–89.0) 77.5 (67.8–82.0) Abbreviations: ALT, alanine aminotransferase; aPTT, activated partial thromboplastin time; AST, aspartate aminotransferase; IQR, interquartile range; INR, international normalized ratio. a Visual analogue scale: 0 is the worst imaginable health state the patient can think of, 100 is the best imaginable health state.

(33.3–104.5) Minimum quality of life scorea Mean 76.5 78.3 74.45 .478 Median (IQR) 84.5 (69.8–89.3) 77.0 (69.8–89.0) 77.5 (67.8–82.0) Abbreviations: ALT, alanine aminotransferase; aPTT, activated partial thromboplastin time; AST, aspartate aminotransferase; IQR, interquartile range; INR, international normalized ratio. a Visual analogue scale: 0 is the worst imaginable health state the patient can think of, 100 is the best imaginable health state. Balapiravir Effects on the Host Immune Response and Virus Sequence Diversity Plasma concentrations of TNF-α, IFN-γ, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, IL-13, and IL-1β were measured in serial (daily) plasma samples during the inpatient period and at follow-up (day 28). Concentrations of IFN-γ and IL-10 were transiently elevated during the 48–96 hours after patient enrollment, but this was independent of the treatment assignment. Concentrations of TNF-α, IL-2, IL-4, IL-5, IL-6, IL-12p70, IL-13, and IL-1β were low and did not increase by >2-fold during the treatment period (data not shown). Changes in whole blood gene transcript abundance in serial samples collected at baseline and on study days 4 and 28 were also monitored, but there was no evidence of treatment-related effects on gene transcript abundance in subsequent samples (data not shown). Mutation rates in virus genome sequences in serial plasma samples from patients in the balapiravir and placebo arms were also not significantly different (data not shown). Collectively, these exploratory analyses are consistent with the predefined trial clinical and virological endpoints that suggest balapiravir did not measurably alter the disease evolution of dengue.

plasma samples from patients in the balapiravir and placebo arms were also not significantly different (data not shown). Collectively, these exploratory analyses are consistent with the predefined trial clinical and virological endpoints that suggest balapiravir did not measurably alter the disease evolution of dengue. DISCUSSION The development of antiviral therapies for dengue is, alongside vaccine development and vector control, a rational approach to reducing morbidity and preventing transmission. To this end, here we show that the nucleoside analogue R1479 reduces dengue virus replication in human cells in vitro. Balapiravir, a prodrug of R1479, was also safe and well-tolerated in adult dengue patients who received doses of 1500 mg or 3000 mg twice daily for 5 days. However, balapiravir treatment did not improve virological, immunological, or clinical measures of disease in patients who commenced treatment within the first 48 hours of their illness.

of R1479, was also safe and well-tolerated in adult dengue patients who received doses of 1500 mg or 3000 mg twice daily for 5 days. However, balapiravir treatment did not improve virological, immunological, or clinical measures of disease in patients who commenced treatment within the first 48 hours of their illness. Balapiravir was well tolerated at both 1500 mg and 3000 mg dosage schedules. The range of clinical and laboratory adverse events observed in the balapiravir treatment arms were typical in Vietnamese adults with dengue and not different from those seen in the placebo arm. The rationale for stopping this exploratory study after a total enrollment of 64 patients was that the dengue transmission season was at an end and the number of patients eligible for enrollment had dropped significantly. Second, a planned joint review of the clinical and laboratory data by the sponsor and investigators led to the conclusion that balapiravir was insufficiently potent to warrant further clinical investigation.

transmission season was at an end and the number of patients eligible for enrollment had dropped significantly. Second, a planned joint review of the clinical and laboratory data by the sponsor and investigators led to the conclusion that balapiravir was insufficiently potent to warrant further clinical investigation. Balapiravir treatment in this study was not associated with measurable changes in a range of virological, clinical, or immunological endpoints, despite evidence that the active moiety R1479 had anti-DENV potency in vitro. Several possibilities may explain this. First, the size of the patient cohorts meant that only large effects could have been detected (see the Methods section for further information on effect size). Second, concentrations of R1479 in plasma may have been insufficient to inhibit DENV replication. Pharmacokinetic and pharmacodynamic analyses of balapiravir in hepatitis C showed a good correlation between exposure (measured as AUC0–12 h or Cmin) and antiviral effect (measured as reduction in plasma HCV RNA level) [11]. In that study, dosing of 1500, 3000, or 4500 mg of balapiravir twice daily was associated with a dose-dependent increase in R1479 plasma exposure and antiviral effect. A mean plasma trough level 3.5-fold above the human serum adjusted HCV replicon EC50 was achieved at the 1500 mg twice daily dose, and this ratio increased to 4.9 and 7.8 in the 2 higher dose groups. In contrast, no significant antiviral effect was apparent at the lower dose of 500 mg twice daily, for which the mean plasma trough level was similar to the HCV replicon EC50, even though Cmax was >5-fold above the HCV replicon EC50 [11]. With a similar relationship between exposure and antiviral effect in DENV infection, mean Cmin concentrations exceeding 6.7–39 µM or Cmax concentrations exceeding 25–145 µM may be needed for the observation of antiviral effects of R1479 on DENV. This would argue for higher or more frequent doses of balapiravir. Dosing of 4500 mg twice daily of balapiravir achieved mean Cmin of 21 µM and mean Cmax of 88 µM in HCV-infected persons [11]. However, this more aggressive strategy might also be accompanied by increased likelihood of drug-related adverse events as seen previously with short-course monotherapy (4500 mg of balapiravir twice a day in HCV-infected patients) [13]. Third, the antiviral effect of R1479 is dependent on phosphorylation to the DENV polymerase inhibitor R1479 triphosphate.

might also be accompanied by increased likelihood of drug-related adverse events as seen previously with short-course monotherapy (4500 mg of balapiravir twice a day in HCV-infected patients) [13]. Third, the antiviral effect of R1479 is dependent on phosphorylation to the DENV polymerase inhibitor R1479 triphosphate. Phosphorylation efficiency can differ between cell types, and R1479 may not be efficiently phosphorylated in the primary DENV target cells in humans. Finally, although the median duration of illness history was <40 hours in all treatment groups, it is possible that the timing of treatment was too late in this patient population to demonstrate a marked antiviral effect above the level that the host adaptive immune response achieves. Indeed, it is striking that decreases in viremia were observed in the majority of patients within 24 hours of enrollment irrespective of the treatment assignment (Figure 2). This underlines the imperative of early diagnosis and treatment with antiviral candidates.

level that the host adaptive immune response achieves. Indeed, it is striking that decreases in viremia were observed in the majority of patients within 24 hours of enrollment irrespective of the treatment assignment (Figure 2). This underlines the imperative of early diagnosis and treatment with antiviral candidates. New therapies may require certain properties to be successful for the treatment of an acute infection such as dengue. Antiviral agents must be effective immediately to rapidly control viral replication. Problems of distribution, active metabolite formation, or protein binding may limit the effective response time. Consideration of prophylactic treatment at the first indication of symptoms may be needed to successfully impede viremia, but it will require a compound with a robust safety profile. Although the current study did not furnish evidence that balapiravir at the doses tested is a candidate drug for dengue, it is hoped that this marks the beginning of an intensive period of clinical research on dengue therapeutics. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

rdjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank the doctors and nurses at the Hospital for Tropical Diseases for their outstanding assistance and the patients for their participation. Financial support. This work was supported by the Wellcome Trust (grant 084368/Z/07/Z to C. P. S.) and Hoffmann-La Roche. Role of the sponsor. Hoffmann-La Roche sponsored and funded the clinical and laboratory studies described here. The sponsor conducted the in vitro laboratory testing of balapiravir, and the results reported in this paper reflect the sponsor's analyses of these experiments. In collaboration with the corresponding author (C. P. S.), Hoffmann-La Roche, the sponsor, helped design the clinical trial and associated laboratory studies. The sponsor also provided clinical, laboratory, and logistical support to the conduct of the clinical trial. The sponsor was not involved in the collection or analysis of the clinical and laboratory data reported in this paper. Representatives of the sponsor (R. P., K. K., and H. J.) provided comments on the final manuscript. The corresponding author (C. P. S.) had full access to all the data reported in this study and takes responsibility for the content of this paper and the decision to submit for publication.

y data reported in this paper. Representatives of the sponsor (R. P., K. K., and H. J.) provided comments on the final manuscript. The corresponding author (C. P. S.) had full access to all the data reported in this study and takes responsibility for the content of this paper and the decision to submit for publication. Potential conflicts of interest. C. P. S. is a consultant to Unither Virology, a company with an interest in the development of dengue therapeutics. H. J., K. K., J. H., and R. P. are employees of Hoffman-La Roche. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

In Europe around 10% of antiretroviral-naive patients are infected with drug-resistant human immunodeficiency virus type 1 (HIV-1), that is, transmitted drug resistance (TDR) [1]. Because HIV infection is thought to be characterized by a single or narrow spectrum of viruses from the donor, wild-type viral variants are unlikely to coexist with drug-resistant variants, unlike the selection of resistance during treatment. Therefore, the observed rate at which TDR mutations become undetectable (“lost”) is likely to be multifactorial, depending on the number of back mutations required, the relative fitness of mutant and back-mutated viruses, the rate of viral turnover, the presence of compensatory mutations, and the sensitivity of the sequencing assay for detecting low level variants [2–5]. Several studies have reported data on the loss and persistence of TDR mutations; however, the number of patients included in these studies have been small [3, 6]. One larger study (75 patients) quantified the rate of loss of TDR mutations for groups of mutations and found that non-nucleoside reverse transcriptase inhibitor (NNRTI) and protease inhibitor (PI) mutations were lost at a similar slow rate, with a statistically nonsignificant trend toward a higher rate of loss of thymidine analogue mutations (TAMs) and T215 revertants [2]. However, no study has systematically examined and compared the persistence of individual TDR mutations

or (NNRTI) and protease inhibitor (PI) mutations were lost at a similar slow rate, with a statistically nonsignificant trend toward a higher rate of loss of thymidine analogue mutations (TAMs) and T215 revertants [2]. However, no study has systematically examined and compared the persistence of individual TDR mutations METHODS Study Population and Definitions ART-naive patients (both acute/early infection and unknown duration of infection), aged 16 years or older, with TDR mutation(s) detected at their first resistance test (performed between 04/1997 and 09/2009) and who had subsequent resistance test(s) while ART-naive, were identified from the UK HIV Drug Resistance Database [7]. Population sequenced (which detects viral variants above a frequency of 15%–25%) genotypic resistance tests of the pol gene were analyzed. The genetic similarity of the sequences from the initial and subsequent resistance tests were compared to exclude super-infection and to check that the samples derived from the same patient. TDR was defined as the presence of ≥1 mutations from the surveillance drug resistance mutations list [8]. Viral subtype was assigned using the REGA algorithm. Demographic and clinical information was acquired by linkage to the UK Collaborative HIV Cohort Study and the UK Register of HIV Seroconverters [7].

ame patient. TDR was defined as the presence of ≥1 mutations from the surveillance drug resistance mutations list [8]. Viral subtype was assigned using the REGA algorithm. Demographic and clinical information was acquired by linkage to the UK Collaborative HIV Cohort Study and the UK Register of HIV Seroconverters [7]. Data Analysis All analyses were carried out in Stata version 12.0 (StataCorp, College Station, Texas). The rate at which mutations became undetectable (“lost”) was examined using survival models accounting for interval-censored censoring, that is, the exact time the mutation is lost is known only to occur between the last resistance test that detected the mutation and the first test without the mutation (intcens command in Stata). Although a Weibull model indicated a decreasing hazard (results not shown), the parameters from this model lack direct interpretation without knowledge of individuals' dates of infection [2]. We therefore present estimates from the exponential (constant hazard) model; although the data contradict the constant hazard assumption, the estimates can be interpreted as the average rate of loss of mutations following their identification in ART-naive patients during chronic infection.

uals' dates of infection [2]. We therefore present estimates from the exponential (constant hazard) model; although the data contradict the constant hazard assumption, the estimates can be interpreted as the average rate of loss of mutations following their identification in ART-naive patients during chronic infection. Mutations with individual frequencies ≥10 (and T215F) were analyzed individually, and those with lower frequencies were grouped by drug class, with the exception of T215 revertants, which were grouped together. An additional analysis examined the effect of patient-level factors on the rate of loss of mutations (accounting for individual mutations), including CD4 cell count and viral load at the first resistance test, viral subtype, first test within 18 months of infection, the number of mutations detected at first test, and whether the mutation was “pure” or part of a mixture. All analyses accounted for multiple mutations at the first resistance test by allowing for within-individual correlation. Finally, we conducted sensitivity analyses removing patients with M184V, those with non-B subtype, and CD4 <200 cells/mm3 at the first resistance test, as these factors increase the likelihood that a patient had prior unrecorded ART exposure.

s at the first resistance test by allowing for within-individual correlation. Finally, we conducted sensitivity analyses removing patients with M184V, those with non-B subtype, and CD4 <200 cells/mm3 at the first resistance test, as these factors increase the likelihood that a patient had prior unrecorded ART exposure. RESULTS A total of 313 patients were included in the analysis. Subjects were mainly, but not exclusively, homo/bisexual men infected with a subtype B virus (Table 1). For only a few patients (47; 15%) was the first resistance test known to have been conducted within 18 months of infection. 59% of patients had a single mutation detected at their first test; 27% and 6% had mutations conferring resistance to two and three ART classes, respectively. Of the total 717 TDR mutations detected at the first resistance test, 147 (21%) were present as a mixture (92 with wild-type amino acid alone, 37 with a non-TDR mutation alone, 18 with both). Most patients (279; 89%) had only one resistance test following the initial test which detected TDR mutations and before starting ART; the median (interquartile range [IQR]) interval between tests was 40 (10–96) weeks. Table 1. Description of Study Population and Initial Resistance Test

R mutation alone, 18 with both). Most patients (279; 89%) had only one resistance test following the initial test which detected TDR mutations and before starting ART; the median (interquartile range [IQR]) interval between tests was 40 (10–96) weeks. Table 1. Description of Study Population and Initial Resistance Test N (%) or Median (IQR) No. of patients 313 Gender Male 220 (70) Female 22 (7) Unknown 71 (23) Exposure source Homo/bisexual 187 (60) Heterosexual 24 (8) Other (including 1 injecting drug user) 13 (4) Unknown 89 (28) CD4 at first test (cells/mm3)a 427 (268, 545) Viral load at first test (log10 copies/mL)b 4.6 (4.0, 5.1) Subtype B 248 (79) Non-B 42 (13) Not classified 23 (7) First test within 18 mo of infectionc No or unknown 266 (85) Yes 47 (15) No. of mutations in first test 1 185 (59) 2 59 (19) 3 23 (7) ≥4 46 (15) No. of patients with ≥1 NRTI mutation 204 (65) ≥1 NNRTI mutation 120 (38) ≥1 PI mutation 74 (24) No. of patients with resistance to 1 class 212 (68) 2 classes 83 (27) 3 classes 18 (6) Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range. a Within 90 days before/after resistance test, N = 217. b Within 90 days before/after resistance test, N = 238. c First resistance test within 18 months of HIV-negative test in patients with ≤18 months between HIV-negative and HIV-positive tests.

N (%) or Median (IQR) No. of patients 313 Gender Male 220 (70) Female 22 (7) Unknown 71 (23) Exposure source Homo/bisexual 187 (60) Heterosexual 24 (8) Other (including 1 injecting drug user) 13 (4) Unknown 89 (28) CD4 at first test (cells/mm3)a 427 (268, 545) Viral load at first test (log10 copies/mL)b 4.6 (4.0, 5.1) Subtype B 248 (79) Non-B 42 (13) Not classified 23 (7) First test within 18 mo of infectionc No or unknown 266 (85) Yes 47 (15) No. of mutations in first test 1 185 (59) 2 59 (19) 3 23 (7) ≥4 46 (15) No. of patients with ≥1 NRTI mutation 204 (65) ≥1 NNRTI mutation 120 (38) ≥1 PI mutation 74 (24) No. of patients with resistance to 1 class 212 (68) 2 classes 83 (27) 3 classes 18 (6) Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range. a Within 90 days before/after resistance test, N = 217. b Within 90 days before/after resistance test, N = 238. c First resistance test within 18 months of HIV-negative test in patients with ≤18 months between HIV-negative and HIV-positive tests. Rate of Loss of Individual TDR Mutations The overall rate of loss of mutations was 18 (95% confidence interval [CI], 14–23) per 100 person-years of follow-up (PYFU), although the rate varied considerably for individual mutations (Table 2). Within drug class, NRTI mutations showed the most variation in persistence (heterogeneity P < .001). As expected, M184V was lost rapidly at a rate of 71 (95% CI, 34–149) per 100 PYFU. M41L was commonly observed and highly persistent (rate of loss 8 (95% CI, 4–15) per 100 PYFU), and a similar low rate of loss was seen for other TAMs (D67N, L210W, and K219Q/N); however, K70R appeared to be lost more quickly. There was also a rapid transition of T215F and T215Y to one of the T215 revertants, but the revertants themselves were highly stable with a rate of loss of 5 (95% CI, 3–11) mutations per 100 PYFU. Consequently, there was a large number of T215 revertants at the initial resistance test. Table 2. Rate of Loss of TDR Mutations

was also a rapid transition of T215F and T215Y to one of the T215 revertants, but the revertants themselves were highly stable with a rate of loss of 5 (95% CI, 3–11) mutations per 100 PYFU. Consequently, there was a large number of T215 revertants at the initial resistance test. Table 2. Rate of Loss of TDR Mutations Mutation No. of mutations at first resistance test No. (%) of mutations which became undetectable Rate of loss (95% CI) (per 100 PYFU) Median time to loss (years) (95% CI) All 717 171 (24) 18 (14–23) 3.9 (3.0–5.0) Any NRTI 401 90 (22) 15 (11–21) 4.6 (3.3–6.4) M41L 77 11 (14) 8 (4–15) 8.6 (4.6–16.0) D67N 27 4 (15) 12 (4–33) 6.0 (2.1–16.9) K70R 14 7 (50) 38 (17–83) 1.8 (.8–4.0) M184V 34 16 (47) 71 (34–149) 1.0 (.5–2.0) L210W 25 6 (24) 14 (6–33) 4.8 (2.1–11.2) T215Y 25 13 (52) 41 (20–84) 1.7 (.8–3.4) T215F 9 4 (44) 58 (15–224) 1.2 (.3–4.6) T215 revertants 106 9 (8) 5 (3–11) 13.0 (6.6–25.7) K219Q 25 2 (8) 4 (1–19) 15.8 (3.6–70.0) K219N 12 2 (17) 15 (3–72) 4.6 (1.0–22.4) All other NRTIa 47 16 (34) 22 (12–38) 3.2 (1.8–5.6) Any NNRTI 154 37 (24) 25 (17–38) 2.7 (1.8–4.1) K103N 73 12 (16) 18 (10–34) 3.7 (2.0–6.8) Y181C 20 10 (50) 54 (26–113) 1.3 (.6–2.7) G190A 17 4 (24) 19 (6–56) 3.6 (1.2–15.5) All other NNRTIb 44 11 (25) 27 (13–54) 2.6 (1.3–5.3) Any PI 162 44 (27) 21 (14–31) 3.3 (2.2–4.9) M46L 16 5 (31) 22 (8–59) 3.1 (1.2–8.4) I54V 16 5 (31) 21 (8–50) 3.3 (1.4–7.8) V82A 16 3 (19) 13 (5–39) 5.1 (1.8–14.8) I84V 10 3 (30) 20 (5–76) 3.4 (.9–12.9) L90M 32 5 (16) 12 (5–31) 5.8 (2.2–15.3) All other PIc 72 23 (32) 28 (17–46) 2.5 (1.5–4.1) Abbreviations: CI, confidence interval; NRTI, nucleoside reverse transcriptase inhibitors; NNRTI, non-NRTI; PYFU, person-years of follow-up; PI, protease inhibitors; TDR, transmitted drug resistance.

3 (30) 20 (5–76) 3.4 (.9–12.9) L90M 32 5 (16) 12 (5–31) 5.8 (2.2–15.3) All other PIc 72 23 (32) 28 (17–46) 2.5 (1.5–4.1) Abbreviations: CI, confidence interval; NRTI, nucleoside reverse transcriptase inhibitors; NNRTI, non-NRTI; PYFU, person-years of follow-up; PI, protease inhibitors; TDR, transmitted drug resistance. a K65R(3), D67E(1), D67G(6), T69D(7), 69 insertion(T)(1), K70E(1), L74I(3), L74 V(3), V75A(2), V75M(2), V75 T(2), Y115F(1), Q151M(1), M184I(2), K219E(6), K219R(6). b L100I(3), K101E(9), K101P(3), K103S(4), V106A(2), V106M(4), Y181 V(1), Y188L(8), G190E(2), P225H(5), M230L(3). c L24I(2), D30N(2), V32I(3), M46I(8), I47A(1), I47 V(1), G48 V(4), I50 V(2), F53L(6), I54A(2), I54L(3), I54 T(2), G73S(6), G73 T(2), V82F(2), V82L(9), V82S(1), V82 T(4), N83D(2), I85 V(6), N88D(3), N88S(1).

a K65R(3), D67E(1), D67G(6), T69D(7), 69 insertion(T)(1), K70E(1), L74I(3), L74 V(3), V75A(2), V75M(2), V75 T(2), Y115F(1), Q151M(1), M184I(2), K219E(6), K219R(6). b L100I(3), K101E(9), K101P(3), K103S(4), V106A(2), V106M(4), Y181 V(1), Y188L(8), G190E(2), P225H(5), M230L(3). c L24I(2), D30N(2), V32I(3), M46I(8), I47A(1), I47 V(1), G48 V(4), I50 V(2), F53L(6), I54A(2), I54L(3), I54 T(2), G73S(6), G73 T(2), V82F(2), V82L(9), V82S(1), V82 T(4), N83D(2), I85 V(6), N88D(3), N88S(1). There was no statistically significant difference in the rate of loss of NNRTI mutations (heterogeneity P = .1); K103N was the most common NNRTI mutation, with a rate of loss of 18 (95% CI, 10–34) mutations per 100 PYFU. NNRTI mutations appeared to be lost more quickly than most TAMs (M41L, D67N, L210W, and K219Q/N) and the 215 revertants (P < .001 for both comparisons). L90M was the most common PI mutation, with a rate of loss of 12 (95% CI, 5–31) mutations per 100 PYFU. However, there was little variation in the rate of loss across PI mutations (heterogeneity P = .6), with a rate of loss similar to that of most of the NNRTI mutations. Sensitivity analyses removing patients with M184V (n = 34), those with non-B subtype (n = 42), or patients with CD4 < 200 cells/mm3 at the first resistance test (n = 29) resulted in slightly lower absolute rates of TDR mutation loss but did not materially affect comparisons within and between drug classes.

I mutations. Sensitivity analyses removing patients with M184V (n = 34), those with non-B subtype (n = 42), or patients with CD4 < 200 cells/mm3 at the first resistance test (n = 29) resulted in slightly lower absolute rates of TDR mutation loss but did not materially affect comparisons within and between drug classes. Predictors of the Rate of Loss of TDR Mutations In multivariate analysis, there was no clear effect on the rate of loss of TDR mutations of CD4 cell count (P = .5) or viral load (P = .2) at the first resistance test, recent infection (P = .3), or the number of mutations detected at the first test (P = 1.0). A statistically significant higher rate of loss was seen with non-B subtype infection than subtype B infection (adjusted hazard ratio = 2.8, 95% CI, 1.2–6.3, P = .01), and also, as expected, if the TDR mutation at the first resistance test was present as a mixture (adjusted hazard ratio = 6.8, 95% CI, 4.2–11.2, P < .001).

tatistically significant higher rate of loss was seen with non-B subtype infection than subtype B infection (adjusted hazard ratio = 2.8, 95% CI, 1.2–6.3, P = .01), and also, as expected, if the TDR mutation at the first resistance test was present as a mixture (adjusted hazard ratio = 6.8, 95% CI, 4.2–11.2, P < .001). DISCUSSION By including patients with unknown duration of infection as well as those identified during acute/early infection, this is the largest study to date to provide quantitative estimates of the persistence of individual TDR mutations. Wide variability in persistence was observed for NRTI mutations in particular, highlighting the need to be careful when grouping mutations for the purposes of analysis. In a recent study of patients with acute/early infection all TAMs were combined for analysis, but we found marked variation within this group of mutations, with T215F/Y and K70R being lost more rapidly than other TAMs [2]. However, T215 revertants were highly persistent, consistent with the fitness advantages associated with this evolutionary pathway [9]. M184V was lost rapidly, although at a lower rate than reported by Jain et al [2], possibly reflecting a selection bias in our analysis. Lesser heterogeneity was observed for NNRTI and PI mutations, and mutations from these classes were lost more rapidly than the T215 revertants and the more stable TAMs, such as M41L. This is in contrast to previous smaller studies, which have generally observed NNRTI mutations to be relatively stable [3, 10], and also to the study by Jain et al, which reported a trend toward a higher rate of loss of TAMs and T215 revertants compared to NNRTI mutations [2].

ts and the more stable TAMs, such as M41L. This is in contrast to previous smaller studies, which have generally observed NNRTI mutations to be relatively stable [3, 10], and also to the study by Jain et al, which reported a trend toward a higher rate of loss of TAMs and T215 revertants compared to NNRTI mutations [2]. Transmission of TDR occurs both from ART-experienced patients with acquired resistance as well as onward transmission from ART-naive individuals. Our finding that certain mutations are highly stable and not replaced by wild-type virus, along with high levels of viral suppression among patients receiving ART, suggests that TDR may increasingly stem from the ART-naive population. HIV transmission models are critical for predicting future levels and patterns of TDR, and TDR persistence among ART-naive patients is a key component of these [11, 12]. Because of the lack of epidemiological data, Wagner et al [12] used estimates of fitness costs from viral competition experiments [13] to calibrate their models. They reported that at least 2 mutations (K70R and Y181C) could form self-sustaining transmission chains. However, there is a discord between our empirical estimates of persistence of individual TDR mutations with in vitro fitness cost estimates. For example, certain TAMs and PI mutations were more stable than would be expected, given their highly impaired replicative impairment [13, 14]. The determinants of persistence of specific viral species in vivo will not only include complex genetic interactions (eg, compensatory mutations [4, 5]) but also other aspects of host-pathogen biology, such as immune responses.

stable than would be expected, given their highly impaired replicative impairment [13, 14]. The determinants of persistence of specific viral species in vivo will not only include complex genetic interactions (eg, compensatory mutations [4, 5]) but also other aspects of host-pathogen biology, such as immune responses. Our finding that some TDR mutations may persist for several years supports the continued use of baseline genotypic resistance testing in chronically infected patients. It is also important to note that the marked variability in the persistence of individual TDR mutations indicates that the detection of ≥1 mutations may signal that viruses harboring other undetected mutations could have been archived in latent cells and thus affect response to subsequent ART. We found no effect of CD4 cell count or viral load on the persistence of TDR mutations, in agreement with Bezemer et al [6], although the rate of loss was higher in patients with non-B subtype infection than subtype B. Although there is no obvious virological explanation for this finding, one possibility is differential ART misclassification by patient characteristics linked to viral subtype. The rate of loss of mutations was similar if the mutation detected at the first test was present in isolation or accompanied by other mutations; further analyses are planned to look at the role of compensatory mutations [4, 5].

ty is differential ART misclassification by patient characteristics linked to viral subtype. The rate of loss of mutations was similar if the mutation detected at the first test was present in isolation or accompanied by other mutations; further analyses are planned to look at the role of compensatory mutations [4, 5]. To maximize the information available we included patients with unknown duration of infection, as well as patients identified during acute/early infection. This introduces a selection effect because some potentially eligible patients will have lost TDR mutations before their first resistance test. Only including patients identified during acute infection would minimize this effect, although, even then, some highly unfit mutations such as M184V could still be missed. Another limitation of the analysis is that population sequencing was used to detect mutations rather than more sensitive methods capable of detecting minor variants, and therefore we may be overestimating the rate of loss [15]. In summary, this is the first study to our knowledge to provide estimates of the persistence of individual TDR mutations. The disconnect with in vitro estimates of resistance-associated viral fitness costs underlines the key role of epidemiological data in calibrating HIV transmission models, which are critical for predicting the future course of the TDR epidemic. Notes Acknowledgments. We thank the UK Collaborative HIV Cohort Study and the UK Register of HIV Seroconverters for providing demographic and clinical information for this study.

In summary, this is the first study to our knowledge to provide estimates of the persistence of individual TDR mutations. The disconnect with in vitro estimates of resistance-associated viral fitness costs underlines the key role of epidemiological data in calibrating HIV transmission models, which are critical for predicting the future course of the TDR epidemic. Notes Acknowledgments. We thank the UK Collaborative HIV Cohort Study and the UK Register of HIV Seroconverters for providing demographic and clinical information for this study. UK HIV Drug Resistance Database Steering Committee:

In summary, this is the first study to our knowledge to provide estimates of the persistence of individual TDR mutations. The disconnect with in vitro estimates of resistance-associated viral fitness costs underlines the key role of epidemiological data in calibrating HIV transmission models, which are critical for predicting the future course of the TDR epidemic. Notes Acknowledgments. We thank the UK Collaborative HIV Cohort Study and the UK Register of HIV Seroconverters for providing demographic and clinical information for this study. UK HIV Drug Resistance Database Steering Committee: Celia Aitken, Gartnavel General Hospital, Glasgow; David Asboe, Anton Pozniak, Chelsea & Westminster Hospital, London; Daniel Webster, Royal Free NHS Trust, London; Patricia Cane, Health Protection Agency, Porton Down; Hannah Castro, David Dunn, David Dolling, Esther Fearnhill, Kholoud Porter, MRC Clinical Trials Unit, London; David Chadwick, South Tees Hospitals NHS Trust, Middlesbrough; Duncan Churchill, Brighton and Sussex University Hospitals NHS Trust; Duncan Clark, St Bartholomew's and The London NHS Trust; Simon Collins, HIV i-Base, London; Valerie Delpech, Health Protection Agency, Centre for Infections, London; Anna Maria Geretti, University of Liverpool; David Goldberg, Health Protection Scotland, Glasgow; Antony Hale, Leeds Teaching Hospitals NHS Trust; Stéphane Hué, University College London; Steve Kaye, Imperial College London; Paul Kellam, Wellcome Trust Sanger Institute and UCL Medical School; Linda Lazarus, Expert Advisory Group on AIDS Secretariat, Health Protection Agency, London; Andrew Leigh-Brown, University of Edinburgh; Nicola Mackie, Imperial NHS Trust; Chloe Orkin, St. Bartholomew's Hospital, London; Philip Rice, St George's Healthcare Trust, London; Deenan Pillay, Andrew Phillips, Caroline Sabin, University College London Medical School; Erasmus Smit, Health Protection Agency, Birmingham Heartlands Hospital; Kate Templeton, Royal Infirmary of Edinburgh; Peter Tilston, Manchester Royal Infirmary; William Tong, Guy's and St. Thomas' NHS Foundation Trust, London; Ian Williams, Mortimer Market Centre, London; Hongyi Zhang, Addenbrooke's Hospital, Cambridge; Mark Zuckerman, King's College Hospital, London.

Heartlands Hospital; Kate Templeton, Royal Infirmary of Edinburgh; Peter Tilston, Manchester Royal Infirmary; William Tong, Guy's and St. Thomas' NHS Foundation Trust, London; Ian Williams, Mortimer Market Centre, London; Hongyi Zhang, Addenbrooke's Hospital, Cambridge; Mark Zuckerman, King's College Hospital, London. Centres contributing data to UK HIV Drug Resistance Database:

Heartlands Hospital; Kate Templeton, Royal Infirmary of Edinburgh; Peter Tilston, Manchester Royal Infirmary; William Tong, Guy's and St. Thomas' NHS Foundation Trust, London; Ian Williams, Mortimer Market Centre, London; Hongyi Zhang, Addenbrooke's Hospital, Cambridge; Mark Zuckerman, King's College Hospital, London. Centres contributing data to UK HIV Drug Resistance Database: Clinical Microbiology and Public Health Laboratory, Addenbrooke's Hospital, Cambridge (Jane Greatorex); HIV/GUM Research Laboratory, Chelsea and Westminster Hospital, London (Adrian Wildfire); Guy's and St. Thomas' NHS Foundation Trust, London (Siobhan O'Shea, Jane Mullen); HPA – Public Health Laboratory, Birmingham Heartlands Hospital, Birmingham (Erasmus Smit); HPA London (Tamyo Mbisa); Imperial College Health NHS Trust, London (Alison Cox); King's College Hospital, London (Richard Tandy); Medical Microbiology Laboratory, Leeds Teaching Hospitals NHS Trust (Tony Hale, Tracy Fawcett); Specialist Virology Centre, Liverpool (Mark Hopkins, Lynn Ashton); Department of Clinical Virology, Manchester Royal Infirmary, Manchester (Peter Tilston); Department of Virology, Royal Free Hospital, London (Daniel Webster, Ana Garcia-Diaz); Edinburgh Specialist Virology Centre, Royal Infirmary of Edinburgh (Jill Shepherd); Department of Infection and Tropical Medicine, Royal Victoria Infirmary, Newcastle (Matthias L Schmid, Brendan Payne); South Tees Hospitals NHS Trust, Middlesbrough (David Chadwick); St George's Hospital, London (Phillip Hay, Phillip Rice, Mary Paynter); Department of Virology, St Bartholomew's and The London NHS Trust (Duncan Clark, David Bibby); Molecular Diagnostic Unit, Imperial College, London (Steve Kaye); University College London Hospitals (Stuart Kirk); West of Scotland Specialist Virology Lab Gartnavel, Glasgow (Alasdair MacLean, Celia Aitken, Rory Gunson).

; Department of Virology, St Bartholomew's and The London NHS Trust (Duncan Clark, David Bibby); Molecular Diagnostic Unit, Imperial College, London (Steve Kaye); University College London Hospitals (Stuart Kirk); West of Scotland Specialist Virology Lab Gartnavel, Glasgow (Alasdair MacLean, Celia Aitken, Rory Gunson). Financial support. This work was supported by the UK Medical Research Council (grant G0900274) and the European Community's 7th framework programme (FP7/2007–2013) under the Collaborative HIV and Anti-HIV Drug Resistance Network (CHAIN; project 223131). Potential conflicts of interest. A. P. has had agreements to provide modelling analysis reports for ViiV Healthcare, Gilead Sciences, Inc., Johnson & Johnson, Bristol-Myers Squibb, and GlaxoSmithKline Biologicals, and is a coinvestigator on a grant for research from Bristol-Myers Squibb. H. C., D. P., P. C., D. A., V. C., and D. T. D.: no conflict. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Identifying characteristics that distinguish different encephalopathies occurring in children in malaria-endemic regions can be difficult, especially in resource-poor settings. Plasmodium falciparum is often assumed to be the main cause, but there are many other causes of encephalopathy, including bacterial meningitis or viral meningitis [1]. Frequently, no evidence of an infectious agent is found [2], and between 2007 and 2011, 51% of comatose children admitted to Kilifi District Hospital on the Kenyan coast had coma with no cause identified. In addition, an earlier study of Kenyan children with acute encephalopathy found that a significant proportion who fulfilled the World Health Organization definition of cerebral malaria had viruses detected in the cerebrospinal fluid (CSF) [3]. Whereas bacterial meningitis can be excluded by the examination and culture of CSF [4], exclusion of other encephalopathies remains a significant challenge. This seriously confounds studies on the pathophysiology of cerebral malaria and also delays critical decisions on appropriate clinical management. Therefore, we need better ways of identifying children who have or may develop cerebral malaria to facilitate early clinical decisions on management.

significant challenge. This seriously confounds studies on the pathophysiology of cerebral malaria and also delays critical decisions on appropriate clinical management. Therefore, we need better ways of identifying children who have or may develop cerebral malaria to facilitate early clinical decisions on management. We have previously demonstrated the measurable presence of differentially expressed proteins in plasma from a mouse model of cerebral malaria, compared with noninfected mice [5]. To determine whether differentially expressed proteins could be identified in body fluids from patients with cerebral malaria, we undertook a similar proteomic study, using CSF and plasma from children with cerebral malaria, and compared the protein profiles in these biological matrices with those in samples from children with confirmed acute bacterial meningitis and other nonspecific encephalopathies.

uids from patients with cerebral malaria, we undertook a similar proteomic study, using CSF and plasma from children with cerebral malaria, and compared the protein profiles in these biological matrices with those in samples from children with confirmed acute bacterial meningitis and other nonspecific encephalopathies. MATERIALS AND METHODS Subjects The study used archived plasma and CSF samples collected from clinically well-characterized children attending Kilifi District Hospital between 2001 and 2002. The Kenya Medical Research Institute Ethics Review Committee approved all studies. The children came from a geographic region described in detail elsewhere [6]. Children were grouped according to the results of a malaria slide, CSF leukocyte count, and microbiological findings [6]. Cerebral malaria (n = 12) was defined according to World Health Organization criteria: (1) a Blantyre coma score of <3 in the presence of peripheral asexual malarial parasites on the blood film, and (2) negative results of CSF or blood cultures and a CSF leukocyte count of ≤10 cells/μL. In addition, we selected patients with a parasitemia of >2500 parasites/µL, since this cutoff has the most specificity in this area [6]. Acute bacterial meningitis (n = 12) was defined as the absence of asexual malaria parasites in 3 slides of blood specimens obtained over 24 hours; the presence of a CSF leukocyte count of >10 cells/μL or a positive CSF culture result, a positive blood culture result, or detection of bacterial antigen in CSF [1]. Nonspecific encephalopathy (n = 12) was defined as impaired consciousness, no detection of asexual-stage parasites in 3 slides of blood specimens obtained over 24 hours, and no growth on CSF or blood cultures. Matched CSF and plasma samples and complete medical records were available from each patient.

igen in CSF [1]. Nonspecific encephalopathy (n = 12) was defined as impaired consciousness, no detection of asexual-stage parasites in 3 slides of blood specimens obtained over 24 hours, and no growth on CSF or blood cultures. Matched CSF and plasma samples and complete medical records were available from each patient. Sample Preparation Archived CSF samples that had been stored at −80°C for 5–6 years were thawed at 4°C and desalted using Micorocon YM-3 centrifugal units (Millipore, United States). Archived plasma samples stored at −80°C were also thawed at 4°C. Protein concentrations for the plasma (1:100 v/v dilution with water) and desalted CSF sample were determined using the Bradford assay as previously described [5]. Before storage, CSF samples were centrifuged at 450 ×g. All but the bottom 0.5 mL of the supernatant was transferred to new container and stored at −80°C. For plasma, heparinized blood was centrifuged at 450 ×g and the plasma was removed, aliquoted, and stored at −80°C.

g the Bradford assay as previously described [5]. Before storage, CSF samples were centrifuged at 450 ×g. All but the bottom 0.5 mL of the supernatant was transferred to new container and stored at −80°C. For plasma, heparinized blood was centrifuged at 450 ×g and the plasma was removed, aliquoted, and stored at −80°C. 2-Dimensional Gel Electrophoresis Proteins were separated using 2-dimensional gel electrophoresis and analyzed as previously described [5]. Master gels were prepared by analyzing duplicate gels of samples from 12 individual patients. Spots matched in >75% of the gels were included in the master gel, using PDQuest 2-dimensional software, with semiquantitative analysis performed using Progenesis 200 software. For plasma samples, protein spots of interest were excised from Coomassie-stained gels and digested as previously described [5]. Because of the low protein content in CSF samples, Coomassie-stained gels were not prepared, and spots were cut directly from silver-stained gels and digested using a modified method previously described by Terry et al [7]. Mass spectra of tryptic digests were obtained using a MALDI-ToF mass spectrometer (MS; Shimadzu CFR Plus, Manchester, United Kingdom) as previously described [5]. When definitive protein identification could not be formally made by MALDI-ToF, the tryptic digests of the spots of interest were further separated by reverse-phase–high-performance liquid chromatography performed on an UltiMate 3000 LC system (Dionex, United Kingdom). A total of 1 µL of the concentrated sample was diluted with 4 µL of 2.5% v/v acetonitrile in water containing 0.1% formic acid and injected onto a monolithic capillary column (200-µm internal diameter × 5 cm; Dionex). Peptides were eluted at a flow rate of 1.5 µL/minute, using a solvent gradient of solvent A (2.5% v/v acetonitrile in water with 0.1% formic acid) and solvent B (90% v/v acetonitrile in water with 0.1% formic acid), starting at 5% solvent B, linearly ramped to 40% solvent B over 12 minutes, and then to 90% solvent B for a further 2 minutes. Solvent B was then decreased to 5%, and this was maintained to the end of the run at 27 minutes. Resulting ions were eluted into a LCQ Deca XP Plus ion trap MS (ThermoFinnigan, United States) equipped with a nanospray source connected to a PicoTip column. Further details of the methods used can be found in the Supplementary Materials.

en decreased to 5%, and this was maintained to the end of the run at 27 minutes. Resulting ions were eluted into a LCQ Deca XP Plus ion trap MS (ThermoFinnigan, United States) equipped with a nanospray source connected to a PicoTip column. Further details of the methods used can be found in the Supplementary Materials. Two-Dimensional Liquid Chromatography Tandem MS (LC-MS/MS) The protein separation on gels was subsequently replaced by a 2-dimensional LC-MS/MS–based approach. An equivalent of 100 µg of CSF protein or 200 µg of plasma protein was injected onto a ProSwift RP-1S monolith column (4.6 × 50 mm, Dionex). The high-performance liquid chromatography was performed on an UltiMate 3000 LC system (Dionex UK). Samples were eluted with Solvent A, 2.5% acetonitrile in water with 0.1% trifluoroacetic acid and solvent B, and 90% acetonitrile in water with 0.1% trifluoroacetic acid. The flow rate was maintained at 200 µL/minute, and the separated proteins were eluted into a 96-well plate. Forty-eight fractions per sample were collected. The fractions were then dried down overnight in an oven set at 50°C. A total of 25 µL of 100 mM ammonium bicarbonate was added to the sample, followed by 5 µL of a 20-µg/mL solution of trypsin in 25 mM ammonium bicarbonate. Samples were thoroughly mixed and incubated overnight at 37°C to achieve complete digestion. The resulting digest was subjected to reverse-phase–high-performance liquid chromatography MS as described above.

bonate was added to the sample, followed by 5 µL of a 20-µg/mL solution of trypsin in 25 mM ammonium bicarbonate. Samples were thoroughly mixed and incubated overnight at 37°C to achieve complete digestion. The resulting digest was subjected to reverse-phase–high-performance liquid chromatography MS as described above. P. falciparum Histidine-Rich Protein 2 (HRP2) Double-Site Antigen-Capture Enzyme-Linked Immunosorbent Assay (ELISA) ELISA was used to determine the presence of pHRP2 in frozen plasma and CSF samples. Plates were coated with 100 µL/well of 1.0 µg/mL immunoglobulin M monoclonal anti-HRP2 antibody (MPFM-55A, Immunology Consultants Laboratories, Newberg, OR) diluted in phosphate-buffered saline (PBS) and incubated overnight at 4°C. Plates were saturated for 2 hours at room temperature with 200 µL of 3% skimmed milk (Marvel; catalog no. UKFF 005M EC) in PBS. Plates were then washed 3 times in PBS/Tween (Sigma; catalog no. P1379; 500 mL; 0.05%) washing solution. A total of 100 µL of diluted plasma samples (1:64) or 100 µL of CSF samples was added to the plates, which were sealed and incubated at room temperature in a humid chamber for 2 hours and then washed 5 times. A total of 100 µL of secondary antibody conjugated with horseradish peroxidase (MPFG-55P, Immunology Consultants Laboratories; 0.2 μg/mL diluted in 2% bovine serum albumin, 1% Tween 20, and PBS) was added to the wells and incubated for 1 hour at room temperature. After incubation, substrate (Sigmafast OPD; catalog no. P9187-50SET) was added and incubated for 30 minutes at room temperature, and the reaction was stopped by adding 50 µL of 2N sulfuric acid. Plates were read at an optical density of 490 nm. Standards were made by serially diluting plasma samples of known parasitemia, with a high parasitemia of 0.2% and a low of 0.003125% (this gave a 0.1 absorbance value above baseline). A cutoff of 0.025 in plasma and 0.004 in CSF had been determined to separate children with cerebral malaria from all other children with impaired consciousness in a separate group of children.

Studying precisely defined patient groups enabled us to dissect out the influence of mycobacterial load and HIV coinfection on M. tuberculosis-specific cellular immunity to tease out which functional and phenotypic subsets could serve as markers of mycobacterial pathogen burden independently of HIV coinfection status. METHODS Participants were prospectively enrolled from 3 clinical centers in London, during the period January 2008–February 2011 under National Research Ethics Service approval (07/H0712/85). Participants were ≥18 years, provided written, informed consent, and were eligible if under clinical investigation for active tuberculosis, undergoing latent tuberculosis infection screening, or had recognized tuberculosis risk factors (eg, known tuberculosis contact). Suspected active tuberculosis was confirmed microbiologically by the clinical diagnostic laboratory. Latent tuberculosis infection was defined as a positive response to RD-1 antigens in either T-SPOT.TB (carried out in routine clinical work up) or M. tuberculosis IFN-γ ELISpot (carried out for the current study) in the absence of symptomatic, microbiological, or radiological evidence of active tuberculosis.

of known parasitemia, with a high parasitemia of 0.2% and a low of 0.003125% (this gave a 0.1 absorbance value above baseline). A cutoff of 0.025 in plasma and 0.004 in CSF had been determined to separate children with cerebral malaria from all other children with impaired consciousness in a separate group of children. Data Management and Statistical Analysis Differences in clinical characteristics of the 3 disease phenotypes were evaluated using Stata, version 11.2. Medians were calculated, and Kruskal–Wallis P values reported. Spectra obtained from the MALDI-ToF were used to search through the NCBInr database, using the Mascot Peptide Mass Fingerprinting software [8]. Protein scores were considered significant in accordance with cutoff scores recommended by Mascot for Homo sapiens 65 P < .05 or for P. falciparum 55 P < .05. MS/MS spectra were evaluated using the TurboSEQUEST algorithm in BioWorks v 3.1 software provided by ThermoFinnigan and were searched against the human and P. falciparum subsets of the NCBInr database and the P. falciparum database (PlasmoDB, version 4.4) downloaded from the Sanger Institute. All searches were performed according to search parameters described in the Supplementary Materials. Proteins identified were stored in Stata, version 11.2, and a heat map showing the presence or absence of a protein in a patient was generated using Stata, version 11.2. Proteins were included in the phenotype database if they were identified in samples from at least 6 of the 12 patients in the group.

entary Materials. Proteins identified were stored in Stata, version 11.2, and a heat map showing the presence or absence of a protein in a patient was generated using Stata, version 11.2. Proteins were included in the phenotype database if they were identified in samples from at least 6 of the 12 patients in the group. Functional cataloging of proteins was performed as described elsewhere [5]. Further protein and pathway analysis was undertaken using tools available at the Universal Protein Resource [9]. RESULTS Patient Characteristics Samples from 36 children were used in this study. Table 1 reports clinical features of the children. As expected, children with acute bacterial meningitis had significantly higher levels of white blood cell counts in the CSF and lower levels of glucose in the CSF. Children with cerebral malaria had significantly fewer platelet counts. Table 1. Clinical Characteristics of Patients

rts clinical features of the children. As expected, children with acute bacterial meningitis had significantly higher levels of white blood cell counts in the CSF and lower levels of glucose in the CSF. Children with cerebral malaria had significantly fewer platelet counts. Table 1. Clinical Characteristics of Patients Characteristic Acute Bacterial Meningitis Cerebral Malaria Nonspecific Encephalopathy Pa Age, months 32.75 (6.07–86.03) 31.83 (19.98–39.47) 25.55 (13.47–35.50) .7730 Parasite density, iRBCs ×103/μL 0 (0–0) 566 000 (22 434–1 289 500) 0 (0–56.5) .0001 Hemoglobin level, g/dL 9.7 (9.05–11.5) 6.3 (5.3–7.7) 9.6 (8–10.15) .0002 Platelet level, platelets/μL 494 (320.5–655.5) 93 (76–184) 419 (273–605) .0021 CSF WBC count, cells/μL 228 (34–1000) 1 (0–2) 2 (0–2) .0003 CSF protein level, mg/dL 1.865 (0.875–2.645) 0.24 (0.205–0.375) 0.195 (0.165–0.24) .0001 Blood glucose, mg/dL 5.9 (3.8–7.1) 4.6 (3.75–5.05) 4.5 (3.85–6.65) .4281 CSF glucose, mg/dL 0.95 (0.6–2.15) 3.4 (3–3.7) 3.3 (2.85–3.9) .0004 Ratio of CSF to blood glucose 0.19 (0.12–0.41) 0.72 (0.65–0.96) 0.70 (0.61–0.77) .0011 Plasma HRP2 level 0.0025 (0–0.006) 6.1915 (0.4845–13.3285) 0 (0–0.2235) .0001 CSF HRP2 level 0.002 (0–0.0035) 0.0265 (0.0085–0.108) 0.003 (0.001–0.01) .0002 Hospitalization duration, d 9.5 (8.5–14) 3 (2.5–3.5) 4 (2.5–6) .0003 Data are median (interquartile range). Abbreviations: CSF, cerebrospinal fluid; HRP2, histidine-rich protein 2; iRBC, infected red blood cell; WBC, white blood cell. a By the Kruskal-Wallis test.

Characteristic Acute Bacterial Meningitis Cerebral Malaria Nonspecific Encephalopathy Pa Age, months 32.75 (6.07–86.03) 31.83 (19.98–39.47) 25.55 (13.47–35.50) .7730 Parasite density, iRBCs ×103/μL 0 (0–0) 566 000 (22 434–1 289 500) 0 (0–56.5) .0001 Hemoglobin level, g/dL 9.7 (9.05–11.5) 6.3 (5.3–7.7) 9.6 (8–10.15) .0002 Platelet level, platelets/μL 494 (320.5–655.5) 93 (76–184) 419 (273–605) .0021 CSF WBC count, cells/μL 228 (34–1000) 1 (0–2) 2 (0–2) .0003 CSF protein level, mg/dL 1.865 (0.875–2.645) 0.24 (0.205–0.375) 0.195 (0.165–0.24) .0001 Blood glucose, mg/dL 5.9 (3.8–7.1) 4.6 (3.75–5.05) 4.5 (3.85–6.65) .4281 CSF glucose, mg/dL 0.95 (0.6–2.15) 3.4 (3–3.7) 3.3 (2.85–3.9) .0004 Ratio of CSF to blood glucose 0.19 (0.12–0.41) 0.72 (0.65–0.96) 0.70 (0.61–0.77) .0011 Plasma HRP2 level 0.0025 (0–0.006) 6.1915 (0.4845–13.3285) 0 (0–0.2235) .0001 CSF HRP2 level 0.002 (0–0.0035) 0.0265 (0.0085–0.108) 0.003 (0.001–0.01) .0002 Hospitalization duration, d 9.5 (8.5–14) 3 (2.5–3.5) 4 (2.5–6) .0003 Data are median (interquartile range). Abbreviations: CSF, cerebrospinal fluid; HRP2, histidine-rich protein 2; iRBC, infected red blood cell; WBC, white blood cell. a By the Kruskal-Wallis test. Protein Separation by 2-Dimensional Gel Electrophoresis Reference gels for each phenotype were created using PDQuest. For plasma, averages of 200, 194, and 71 spots were included in the cerebral malaria, nonspecific encephalopathies, and acute bacterial meningitis gels, respectively. For CSF, averages of 150 spots were included in the cerebral malaria and nonspecific encephalopathies reference gels, and 80 spots were included in the acute bacterial meningitis gel. The 2-dimensional gel electrophoresis patterns of both plasma and CSF samples showed significant differences between the 3 clinical phenotypes studied (Figure 1). Proteins of interest that were definitively identified using mass spectrometry are listed in Table 2 (plasma) and Table 3 (CSF). Plasma proteins identified were mainly involved in platelet activation and aggregation, protein transport, endocytosis and cell communication, lipid metabolism, and binding and protein/antigen/nucleotide binding. CSF proteins of interest were mainly involved in apoptosis and proteolysis. Table 2. Host Proteins Identified as Differentially Expressed When Gels of Plasma From Individuals With Cerebral Malaria Were Compared to Other Phenotypes

cell communication, lipid metabolism, and binding and protein/antigen/nucleotide binding. CSF proteins of interest were mainly involved in apoptosis and proteolysis. Table 2. Host Proteins Identified as Differentially Expressed When Gels of Plasma From Individuals With Cerebral Malaria Were Compared to Other Phenotypes Biological Process, Accession No. Description Spot Difference Acute Bacterial Meningitis Nonspecific Encephalopathy Platelet activation/aggregation P01009 Alpha-1-antitrypsin Unique Unique P02647 Apolipoprotein A … Down P02787 Serotransferrin precursor Unique Missing P02787 C chain of human serum transferrin … Missing P02679 F chain of fragment D of fibrinogen … Missing P52735 Guanine nucleotide exchange factor VAV2 Unique Unique D3DP16 Fibrinogen γ chain, isoform CRA_a … Missing Protein transport/endocytosis/cell communication O60493 Sorting nexin-3 (protein SDP3) Down Down P19652 Alpha-1-acid glycoprotein, type 2 Up Up P02763 Alpha-1-acid glycoprotein, type 1 Up Up P02768 Albumin, isoform CRA_g … Unique Lipid metabolism and binding P02647 A chain of human apolipoprotein A-I … Down Protein/antigen/nucleotide binding Q96Q89 M-phase phosphoprotein 1 Up Unique P16871 Interleukin 7 receptor, isoform CRA_b … Missing Table 3. Host Proteins Identified as Differentially Expressed When Gels of Cerebrospinal Fluid From Individuals With Cerebral Malaria Were Compared to Other Phenotypes

Biological Process, Accession No. Description Spot Difference Acute Bacterial Meningitis Nonspecific Encephalopathy Platelet activation/aggregation P01009 Alpha-1-antitrypsin Unique Unique P02647 Apolipoprotein A … Down P02787 Serotransferrin precursor Unique Missing P02787 C chain of human serum transferrin … Missing P02679 F chain of fragment D of fibrinogen … Missing P52735 Guanine nucleotide exchange factor VAV2 Unique Unique D3DP16 Fibrinogen γ chain, isoform CRA_a … Missing Protein transport/endocytosis/cell communication O60493 Sorting nexin-3 (protein SDP3) Down Down P19652 Alpha-1-acid glycoprotein, type 2 Up Up P02763 Alpha-1-acid glycoprotein, type 1 Up Up P02768 Albumin, isoform CRA_g … Unique Lipid metabolism and binding P02647 A chain of human apolipoprotein A-I … Down Protein/antigen/nucleotide binding Q96Q89 M-phase phosphoprotein 1 Up Unique P16871 Interleukin 7 receptor, isoform CRA_b … Missing Table 3. Host Proteins Identified as Differentially Expressed When Gels of Cerebrospinal Fluid From Individuals With Cerebral Malaria Were Compared to Other Phenotypes UniProt KB Accession No. Description Spot Difference Gene Ontology Acute Bacterial Meningitis Nonspecific Encephalopathy P02766 Transthyretin precursor … Unique Protein transport/hormone activity and binding P62745 ρ-related GTP-binding protein Down … Angiogenesis/apoptosis/cell differentiations P50213 Isocitrate dehydrogenase [NAD] subunit α, mitochondrial precursor Unique … Carbohydrate metabolic processing Q9NRR2 Tryptase γ preproprotein Down … Proteolysis O60423 Probable phospholipid-transporting ATPase IK Unique … Cation transport A4D1V4 39S ribosomal protein L32 Unique … Translation

cell differentiations P50213 Isocitrate dehydrogenase [NAD] subunit α, mitochondrial precursor Unique … Carbohydrate metabolic processing Q9NRR2 Tryptase γ preproprotein Down … Proteolysis O60423 Probable phospholipid-transporting ATPase IK Unique … Cation transport A4D1V4 39S ribosomal protein L32 Unique … Translation Figure 1. Composite gel showing the differences between gels in (A) plasma collected from children with a diagnosis of cerebral malaria (CM) and acute bacterial meningitis (ABM; A), plasma collected from children with a diagnosis of CM and nonspecific encephalopathy (B), cerebrospinal fluid (CSF) collected from children with a diagnosis of CM and ABM (C), and CSF collected from children with a diagnosis of CM and ABM (D). Composite gel maps were created using the comparison tool in PDQuest. The gel maps were created by analyzing duplicate gels for 12 patients’ biological replicates (2 gels per patient sample). The extent of the correlation of protein spots between the individual gels was >0.77 (a coefficient of 1.00 indicates that the replicate gels are perfectly similar). Red spots are spots found in the CM gel map, blue spots are unique to the comparator, and green spots depict spots found in both.

els per patient sample). The extent of the correlation of protein spots between the individual gels was >0.77 (a coefficient of 1.00 indicates that the replicate gels are perfectly similar). Red spots are spots found in the CM gel map, blue spots are unique to the comparator, and green spots depict spots found in both. Protein Separation by 2-Dimensional LC-MS/MS Analysis of plasma samples using 2-dimensional LC-MS/MS revealed a total of 339 host proteins and 573 falciparum proteins. In CSF, we identified 113 host proteins and 254 falciparum proteins. Heat maps of all proteins identified in all 36 patients are shown in Figure 2. Figure 2. Heat maps showing distribution of proteins expressed using a 2-dimensional liquid chromatography tandem mass spectrometry strategy. A, Host proteins in plasma and cerebrospinal fluid (CSF). B, Plasmodium falciparum proteins in plasma and CSF. C, Graph showing breakdown of the gene ontology categorization of host proteins identified in plasma.

bution of proteins expressed using a 2-dimensional liquid chromatography tandem mass spectrometry strategy. A, Host proteins in plasma and cerebrospinal fluid (CSF). B, Plasmodium falciparum proteins in plasma and CSF. C, Graph showing breakdown of the gene ontology categorization of host proteins identified in plasma. We selected 259 host proteins of interest (Figure 3 and Supplementary Table 1). Selection was based on whether the protein was of known function and unique to the cerebral malaria phenotype in plasma (n = 1) or CSF (n = 7) or whether it was identified in the other 2 phenotypes but not in cerebral malaria. Seventy-six proteins were found in CSF but not in plasma, and 29 were found in both plasma and CSF. Of particular interest were 3 brain-specific proteins found in plasma namely brain-specific angiogenesis inhibitor 2, calcium/calmodulin-dependent protein kinase IV in acute bacterial meningitis, and spectrin nonerythroid β chain 3 in cerebral malaria. In addition, we also identified 14 other proteins that have been implicated in cerebral malaria or other brain injury (Table 4). Table 4. Host Proteins of Interest Implicated in Brain Injury Identified Using a 2-Dimensional Liquid Chromatography Tandem Mass Spectrometry Strategy

erythroid β chain 3 in cerebral malaria. In addition, we also identified 14 other proteins that have been implicated in cerebral malaria or other brain injury (Table 4). Table 4. Host Proteins of Interest Implicated in Brain Injury Identified Using a 2-Dimensional Liquid Chromatography Tandem Mass Spectrometry Strategy Protein UniProt KB Accession No. CSF Plasma Reference(s) CM; NE; ABM CM; NE; ABM Cytoplasmic protein NCK2 O43639 Yes; No; Yes No; No; Yes [29] Brain-specific angiogenesis inhibitor 2 O60241 No; No; No No; No; Yes [30] Spectrin beta chain, erythrocyte P11277 No; No; No No; No; Yes [31, 32] Secretogranin-2 precursor P13521 Yes; No; No No; No; Yes [33] Sodium/glucose cotransporter 1 P13866 Yes; No; No No; No; No [34, 35] Vinculin P18206 No; Yes; No No; No; No [36] Neurofibromin P21359 No; No; Yes No; No; No [37, 38] Alanine glyoxylate aminotransferase P21549 Yes; No; No No; No; No [39] Retinol-binding protein 2 P50120 No; No; No No; Yes; yes [5, 40] Retinal guanylyl cyclase 2 precursor P51841 Yes; No; No Yes; No; Yes [19, 41] Neurexophilin-1 precursor P58417 No; No; No No; No; Yes [42] Calcium/calmodulin-dependent protein kinase IV; brain Ca Q16566 No; No; No No; No; Yes [41, 43] phospholipase C, β 2 Q59F77 Yes; No; No No; No; No [44] Spectrin β chain, brain 3 Q9H254 Yes; No; No Yes; No; No [15] Reticulon 4 Q9NQC3 No; No; No No; No; Yes [45, 46] Bromodomain adjacent to zinc finger domain protein 2B Q9UIF8 Yes; No; Yes No; No; Yes [47] NDRG2 protein Q9UN36 No; Yes; No Yes; No; No [48, 49] Endothelial lipase precursor Q9Y5X9 Yes; No; No No; No; No [50] Abbreviations: ABM, acute bacterial meningitis; CM, cerebral malaria; CSF, cerebrospinal fluid; NE, nonspecific encephalopathies.

nt to zinc finger domain protein 2B Q9UIF8 Yes; No; Yes No; No; Yes [47] NDRG2 protein Q9UN36 No; Yes; No Yes; No; No [48, 49] Endothelial lipase precursor Q9Y5X9 Yes; No; No No; No; No [50] Abbreviations: ABM, acute bacterial meningitis; CM, cerebral malaria; CSF, cerebrospinal fluid; NE, nonspecific encephalopathies. Figure 3. Venn diagram showing distribution of proteins of interest expressed using a 2-dimensional liquid chromatography tandem mass spectrometry strategy. The red circles represent cerebral malaria, the green circles represent acute bacterial meningitis, and the blue circles represent nonspecific encephalopathy. Abbreviation: CSF, cerebrospinal fluid. In total, we compiled a list of 107 nonhypothetical P. falciparum proteins of interest (Figure 3 and Supplementary Table 2) that were found in plasma and CSF. This list included proteins involved in host cell modification, such as heat shock protein 40; in antigenic variation and host cell interaction, such as 17 variants of erythrocyte membrane protein 1; and 7 rifins.

P. falciparum proteins of interest (Figure 3 and Supplementary Table 2) that were found in plasma and CSF. This list included proteins involved in host cell modification, such as heat shock protein 40; in antigenic variation and host cell interaction, such as 17 variants of erythrocyte membrane protein 1; and 7 rifins. We wondered whether the presence of parasite proteins in slide-negative children could be due to a recent infection, and we therefore measured levels of pHRP2, a parasite protein that has been shown to have a half-life of about 2 weeks after infection [10]. All children with cerebral malaria had pHRP2 levels in plasma and CSF that were above set cutoff levels. One child with acute bacterial meningitis and 3 children with nonspecific encephalopathy had high levels in plasma. The 3 children with nonspecific encephalopathy also had levels of pHRP2 in CSF that were equal to or above the set cutoff level. Interestingly, 3 children with acute bacterial meningitis had levels of pHRP2 in CSF equal to or above the set cutoff level but did not have any pHRP2 detected in the plasma.

evels in plasma. The 3 children with nonspecific encephalopathy also had levels of pHRP2 in CSF that were equal to or above the set cutoff level. Interestingly, 3 children with acute bacterial meningitis had levels of pHRP2 in CSF equal to or above the set cutoff level but did not have any pHRP2 detected in the plasma. DISCUSSION Pathogenic states in children with impaired consciousness in malaria-endemic areas could be reflected by changes in protein biomarkers in both plasma and CSF. Proteomic approaches allow for the analysis of large numbers of proteins at the same time, and this may help elucidate pathways that can be targeted for therapy. Additionally, proteomic platforms allow for the measurement of low-abundance proteins in complex materials such as plasma. This article describes the use of proteomic platforms to elucidate differences in plasma and CSF proteomes collected from children in a malaria-endemic area presenting with impaired consciousness. In this analysis, by use of 2 different proteomic strategies, 259 host proteins and 107 parasite (ie, P. falciparum) proteins were identified as differentially expressed in both plasma and CSF, and these could help elaborate mechanistic differences in the different encephalopathies describe here. Importantly, some of the proteins could help point to the cause of disease in the group of patients with nondefined encephalopathy.

iparum) proteins were identified as differentially expressed in both plasma and CSF, and these could help elaborate mechanistic differences in the different encephalopathies describe here. Importantly, some of the proteins could help point to the cause of disease in the group of patients with nondefined encephalopathy. The first striking observation was the presence of P. falciparum proteins in both plasma and CSF of slide-negative children with acute bacterial meningitis. This result raises 2 possibilities: (1) the children have had a recent episode of malaria, which could predispose them to acute bacterial meningitis; and (2) children in malaria-endemic areas have persistently low levels of parasite proteins in their sera, although one wonders why fewer proteins were identified in the slide-negative children without acute bacterial meningitis. To confirm the second point, we would need to analyze samples from healthy community controls. The presence of pathogen proteins in CSF could be due to a leaky blood brain barrier. However, not all proteins found in the CSF were found in plasma, and additionally we did not always find a correlation with plasma pHRP2 levels and CSF pHRP2 levels. We therefore cannot rule out that proteins in the CSF are there as a result of sequestration in the brain and/or leukocyte migration into the brain.

rier. However, not all proteins found in the CSF were found in plasma, and additionally we did not always find a correlation with plasma pHRP2 levels and CSF pHRP2 levels. We therefore cannot rule out that proteins in the CSF are there as a result of sequestration in the brain and/or leukocyte migration into the brain. We found 14 spots differentially regulated based on gels from the cerebral malaria group, compared with gels from the other 2 groups. Seven of these were identified as proteins that play a role in platelet activation and aggregation (Table 2). In addition, using the 2-dimensional LC-MS/MS strategy, we identified 13 proteins that play a role in coagulation and that were missing in plasma from the cerebral malaria group (Supplementary Table 1) or were found in CSF from acute bacterial meningitis group and not in the other groups. This is not surprising because thrombocytopenia is associated with cerebral malaria [11, 12], and in fact children in this study had lower levels of platelets. Parasite platelet complexes could be a cause of reduced platelet counts, and a proteomic study of platelet microparticles could help clarify mechanisms leading to thrombocytopenia.

ng because thrombocytopenia is associated with cerebral malaria [11, 12], and in fact children in this study had lower levels of platelets. Parasite platelet complexes could be a cause of reduced platelet counts, and a proteomic study of platelet microparticles could help clarify mechanisms leading to thrombocytopenia. Differentially expressed proteins identified in this study, such as α 1 acid glycoprotein type 1 and type 2 (orosomucoid 2), reticulon 4, and retinol binding protein 2 could be markers of an uncontrolled and harmful inflammatory response. An increase in orosomucoid 2, differentially expressed in both plasma and CSF of patients with cerebral malaria, together with an increased production of ceruloplasmin and glutathione, would enhance antioxidant defenses and limit the stimulatory effects of oxidant molecules on cytokine production. This acute-phase protein has previously been studied in connection with malaria [13], and it has been suggested that its production may reflect the severity of the acute phase response [14].

nd glutathione, would enhance antioxidant defenses and limit the stimulatory effects of oxidant molecules on cytokine production. This acute-phase protein has previously been studied in connection with malaria [13], and it has been suggested that its production may reflect the severity of the acute phase response [14]. Spectrin β chain brain 3, a protein that belongs to a family of spectrin proteins, was differentially expressed in both plasma and CSF of children with cerebral malaria. In our previous mouse study [5], this protein was differentially expressed in plasma of mice infected with Plasmodium berghei, compared with expression in noninfected mice. In the brain, this protein is enriched in myelinated neurons, where it colocalizes with ankyrin at axon initial segments and nodes of Ranvier and participates in the clustering of voltage-gated Na+ channels and cell-adhesion molecules at initial segments and nodes of Ranvier [15]. Additionally, spectrin proteins have also been identified as binding partners for various P. falciparum proteins. In particular, spectrins have been shown to associate with heat shock protein 40 [16], and in ring-stage infected red blood cells (RBCs), ring-infected erythrocyte surface antigen associates with spectrin and stabilizes the membrane skeleton. In mature-stage parasitized RBCs, knob-associated His-rich protein molecules self-associate to form conical structures that interact with spectrin. Pf332 and mature-parasite-infected erythrocyte surface antigen bind to the junction complex, while P. falciparum erythrocyte membrane protein 3, identified in this study in plasma of children with cerebral malaria and acute bacterial meningitis, binds to spectrin, further compromising RBC membrane deformability [17]. These interactions stabilize spectrin tetramers in the infected RBC, increasing resistance to further parasite invasion of the cell by increasing infected RBC rigidity, and could facilitate sequestration and inhibit splenic clearance [18]. We hypothesize that the difference in spectrin expression in our study could be linked to these interactions with the infected RBC.

the infected RBC, increasing resistance to further parasite invasion of the cell by increasing infected RBC rigidity, and could facilitate sequestration and inhibit splenic clearance [18]. We hypothesize that the difference in spectrin expression in our study could be linked to these interactions with the infected RBC. Excitotoxic cell death in CNS disorders is partly due to dysfunction of the sodium/potassium pump, resulting in an increased uptake of water, which could as a consequence lead to an increased influx of calcium. Retinal guanylyl cyclase 2 precursor is a gene that displays calcium-dependent regulation [19]. In addition, sodium-dependent glucose transport could also be affected, and this could result in the differential expression of sodium/glucose cotransporter 1 (SGLT1) seen in the CSF of patients with cerebral malaria. SGLT1 has been shown to be expressed in neurons and is upregulated during metabolic stress when there is a decrease in D-glucose content [20]. The sorting nexin family of proteins, which contain a Phox homology domain, play crucial roles in regulating the intracellular membrane trafficking of the endocytic pathway [21]. SNX3, which was differentially expressed in this study, is associated with the early endosome through the PX domain, which is capable of interaction with phosphatidylinositol-3-phosphate. Overexpression of SNX3 alters endosomal morphology and delays transport to the lysosome.

trafficking of the endocytic pathway [21]. SNX3, which was differentially expressed in this study, is associated with the early endosome through the PX domain, which is capable of interaction with phosphatidylinositol-3-phosphate. Overexpression of SNX3 alters endosomal morphology and delays transport to the lysosome. It is important to consider potential limitations of our study design and analysis strategy. First, proteomic studies require careful sample collection and storage. Prolonged contact of CSF and plasma with cellular components has been shown to affect protein and peptide quality because of the presence of proteolytic enzymes in plasma [22, 23] and CSF [24, 25]. Samples analyzed in this study were centrifuged within an hour of collection to remove all cellular components, although protease inhibitors, which can minimize degradation, were not added. These inhibitors can interfere with peptide and amino acid MS signals [26], and some studies suggest that there are no differences in CSF and plasma proteomes determined in the presence or absence of protease inhibitors [27, 28]. Second, a “normal” control could not be established for this study because CSF can only be ethically obtained from children with impaired consciousness. However, future studies could include a nonfebrile group of children with impaired consciousness, such as those with epileptic seizures or poisoning or those who have a lumbar puncture to rule out meningitis. Third, the samples were obtained at admission, and therefore the effects of antimalarial treatment and other effects of disease progression, such as the inflammatory responses that may influence the protein levels, could not be fully assessed. Fourth, cytokines and chemokines that have previously been identified as playing a role in cerebral malaria were not differentially expressed in our study. Two possible reasons for this finding are that (1) semiquantitative 2-dimensional gel analysis may not be sensitive enough to detect such low abundant proteins and that (2) the 2-dimensional LC-MS/MS method described in this study was not quantitative and therefore only reported the presence or absence of proteins in a patient group. Most cytokines and chemokines implicated in malaria were present in all 3 groups.

y not be sensitive enough to detect such low abundant proteins and that (2) the 2-dimensional LC-MS/MS method described in this study was not quantitative and therefore only reported the presence or absence of proteins in a patient group. Most cytokines and chemokines implicated in malaria were present in all 3 groups. Despite the limitations mentioned above, the results from this study show that novel disease specific biomarkers can be identified using proteomic strategies. In addition to providing insight into underlying pathophysiological mechanisms, they have the potential after thorough validation to be used as biomarker panels, which may be of value in the early diagnosis of disease and in monitoring responses to therapies. In particular, this study has shown that there are proteins uniquely expressed in cerebral malaria. The next phase of this work will be to rigorously assess and validate the diagnostic value of these differentially expressed proteins before transfer to a suitable platform that can generate an affordable point-of-care diagnostic. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

rdjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. We thank the director of the Kenya Medical Research Institute, for permission to publish this work, and Prof Kevin Marsh and Dr Britta Urban, for critical discussion and review of the manuscript. E. G. performed the proteomic experiments and drafted the manuscript. C. J. R. C. N., S. A. W., and G. K. participated in the design of the study and helped draft the manuscript. All authors read and approved the final manuscript. Financial support. This work was supported by the World Health Organization (TDR/MIM grant 980074 to G. O. K.), the KEMRI /Wellcome Trust Research Programme Kenya (to E. N. G.), and the Wellcome Trust (senior fellowship 050533 to C. R. J. C. N.). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

The high proportion of the global tuberculosis burden among those with human immunodeficiency virus (HIV) coinfection results in poor individual and population level outcomes due to increased susceptibility, morbidity, and mortality [1]. Mycobacterium tuberculosis (M. tuberculosis) induces several conventional and unconventional T-cell subsets, but the predominant response is mediated by classically restricted, peptide-specific Th1 type CD4+ T cells and CD8+ cytotoxic T lymphocytes (CTLs), which are essential for protective immunity in murine models of tuberculosis [2]. HIV and active tuberculosis both impact M. tuberculosis-specific T-cell immunity, as evidenced by the phenomenon of skin test anergy and impaired cellular immunity in those with active tuberculosis without HIV coinfection [3] and HIV coinfected individuals [4]. Dissecting out the effects of tuberculosis stage and HIV infection is thus necessary to delineate the potential roles of distinct T-cell subsets as biomarkers of active tuberculosis and latent tuberculosis infection.

unity in those with active tuberculosis without HIV coinfection [3] and HIV coinfected individuals [4]. Dissecting out the effects of tuberculosis stage and HIV infection is thus necessary to delineate the potential roles of distinct T-cell subsets as biomarkers of active tuberculosis and latent tuberculosis infection. Functional CD4+ and CD8+ T-cell subsets have been defined based on single-cell cytokine (interferon γ [IFN-γ]/interleukin 2 [IL-2]/tumor necrosis factor α [TNF-α]) signatures. These are differentially impacted by disease stage, mycobacterial load, and treatment [5–7], suggesting that certain subsets may serve as biomarkers of disease activity, pathogen burden, or treatment response. However, evidence to date is limited by a paucity of data on the cell surface marker phenotype of these subsets in M. tuberculosis infection, which is key to characterization of T cells, denoting memory status, disease site homing, survival, and activation [8]. Changes in dominant functionally defined memory response have been associated with varying antigen load in other disease models [9, 10], and studies suggest that M. tuberculosis-specific cells present in active tuberculosis are predominantly of effector-memory phenotype [11–13]. However, further work is needed to confirm and further understand how memory and activation phenotype relates to tuberculosis disease stage [6] and HIV coinfection.

els [9, 10], and studies suggest that M. tuberculosis-specific cells present in active tuberculosis are predominantly of effector-memory phenotype [11–13]. However, further work is needed to confirm and further understand how memory and activation phenotype relates to tuberculosis disease stage [6] and HIV coinfection. Previous data indicated that measurement of CD4+ M. tuberculosis-specific TNF-α-only secreting cells might serve as an accurate biomarker of active tuberculosis [14]. We hypothesized that measuring both M. tuberculosis-specific T-cell function and phenotype would refine this approach and reveal more discriminatory biomarker(s). Therefore, we performed multiparameter flow cytometry for 3 canonical cytokines and key markers of memory and activation in subjects distinguished by mycobacterial load (active tuberculosis vs latent tuberculosis infection) and HIV status. This enabled simultaneous definition of functional and phenotypic M. tuberculosis-specific T-cell profiles at the single-cell level. Studying precisely defined patient groups enabled us to dissect out the influence of mycobacterial load and HIV coinfection on M. tuberculosis-specific cellular immunity to tease out which functional and phenotypic subsets could serve as markers of mycobacterial pathogen burden independently of HIV coinfection status.

aboratory. Latent tuberculosis infection was defined as a positive response to RD-1 antigens in either T-SPOT.TB (carried out in routine clinical work up) or M. tuberculosis IFN-γ ELISpot (carried out for the current study) in the absence of symptomatic, microbiological, or radiological evidence of active tuberculosis. Presence of HIV infection was confirmed by third or fourth generation sero-assay performed by the clinical diagnostic laboratory and using HIV-1 type specific enzyme immunoassay (EIA), according to national standards. HIV viral load (VL) and CD4 T-lymphocyte counts were assayed in the local Clinical Pathology Association-accredited diagnostic laboratories at the time of study recruitment. HIV diagnostics were available for all patients with active tuberculosis (in line with the national screening policy) and the majority of those with latent tuberculosis infection; the remainder had no risk factors for HIV and normal CD4:CD8 lymphocyte ratios and were classified as HIV-uninfected. IFN-γ M. tuberculosis ELISpot Fresh or frozen peripheral blood mononuclear cells (PBMCs), 2.5 × 105 per well, were stimulated overnight (37°C, 5% CO2, 16–20 hours) in an IFN-γ ELISpot plate (Mabtech) with phytohemagglutinin (PHA; positive control; Sigma-Aldrich), Tuberculin Purified Protein Derivative (PPD; Statens Serum Institute) or pools of M. tuberculosis-specific 15-mer overlapping peptides covering each of ESAT-6, CFP-10, EspC, TB7.7, Rv3879c, Rv3873, and Rv3878. Unstimulated cells were used as a negative control. The IFN-γ ELISpot assay was performed as described elsewhere [15].

in Purified Protein Derivative (PPD; Statens Serum Institute) or pools of M. tuberculosis-specific 15-mer overlapping peptides covering each of ESAT-6, CFP-10, EspC, TB7.7, Rv3879c, Rv3873, and Rv3878. Unstimulated cells were used as a negative control. The IFN-γ ELISpot assay was performed as described elsewhere [15]. Intracellular Cytokine Staining and Polychromatic Flow Cytometry Thawed PBMCs (3–5 × 106 per well) were cultured for 16 hours (37°C, 5% CO2) in 10% human serum (Sigma–Aldrich) in RPMI-1640 (Sigma-Aldrich) at a concentration of 1 × 107 cells/mL. Cells were stimulated with PMA-Ionomycin (positive control) (Sigma–Aldrich; final concentration of 5ng/ml for PMA and 500ng/ml for Ionomycin), PPD (16.7 μg/mL final concentration), or a cocktail of peptides spanning the length of 3 highly immunodominant M. tuberculosis-specific RD1-associated antigens, ESAT-6, CFP-10, and EspC (10 µg/mL final concentration per peptide) [16]. Unstimulated cells were used as a negative control. After 2 hours, monensin (2 μM final concentration) was added. Following stimulation, cells were washed and stained with a dead cell marker (LIVE/DEAD Fixable Dead Cell Stain Kits, aqua, Invitrogen) for 30 minutes at 4°C in phosphate-buffered saline (PBS). Cells were then washed in PBS and placed in FC block buffer (10% human serum in filtered FACS solution [0.5% bovine serum albumin and 2 mM EDTA in PBS]) for 20 minutes at 4°C before staining with a pre-titrated and optimized antibody cocktail with fluorochrome-conjugated antibodies against CD3-APC-Alexa Fluor750, CD4-Qdot605, CD45RA-Qdot655 (Invitrogen), CD8-APC, CCR7-PE-Cy7, CD127-FITC (BD Biosciences), and PD-1-PerCP/Cy5.5 (Biolegend). After washing, the cells were fixed and permeabilized using BD Cytofix/Cytoperm Fixation/Permeabilization kit (BD Biosciences) for 20 minutes at 4°C. The cells were washed twice with Perm/Wash solution (BD Biosciences) then stained with pre-titrated fluorochrome-conjugated antibodies in Perm/Wash solution with IFN-γ-V450, IL-2-PE, and TNF-α-AlexaFluor 700 (BD Biosciences) for 30 minutes at 4°C. 1 × 106 events (where possible) were acquired straightaway using a BD LSR-II flow cytometer. Anti-Rat and Anti-Mouse Ig compensation beads (BD Biosciences) were used to set compensation parameters. Fluorescence minus one (FMO) controls were used in each experiment to set gates.

0 (BD Biosciences) for 30 minutes at 4°C. 1 × 106 events (where possible) were acquired straightaway using a BD LSR-II flow cytometer. Anti-Rat and Anti-Mouse Ig compensation beads (BD Biosciences) were used to set compensation parameters. Fluorescence minus one (FMO) controls were used in each experiment to set gates. Data Analysis and Thresholds The data were analyzed on FlowJo version 9.4.4,TreeStar, Inc. Events were gated on live cells, singlets, and lymphocytes using forward and side scatter properties. CD3+CD4+ and CD3+ CD8+ subsets were defined. Gating controls were used to define IFN-γ, IL-2, and TNF-α responses and surface marker expression. For phenotypic analysis of M. tuberculosis-specific cells, only participants with a positive response were included. Positive responders were defined as those with a response that was ≥2 times the background (in unstimulated but fully stained samples) and >0.001% of CD3+CD4+ or CD3+CD8+ cells. This cutoff was used because we did not use costimulation to enhance responses, and we normalized to background (unstimulated) data before applying the cutoff (rather than classifying background as negligible). A strict cutoff meant only antigen-specific cells were included in the phenotypic analysis.

r CD3+CD8+ cells. This cutoff was used because we did not use costimulation to enhance responses, and we normalized to background (unstimulated) data before applying the cutoff (rather than classifying background as negligible). A strict cutoff meant only antigen-specific cells were included in the phenotypic analysis. Statistical Analysis Statistical analysis was conducted using IBM SPSS Statistics version 20 and GraphPad Prism version 5.00 for Mac OS X, GraphPad Software, California. Tuberculosis disease stage compared all tuberculosis (n = 13) vs all latent tuberculosis infection (n = 21) regardless of HIV status, the impact of HIV compared all HIV-infected (n = 17) vs uninfected (n = 17) regardless of tuberculosis disease stage and across all 4 subgroups. The 2-tailed Mann–Whitney U test was used for nonparametric 2-sample comparisons. Spearman rank correlation coefficients were used to test correlations. Receiver operator characteristic (ROC) curve defined the sensitivity and specificity of the diagnostic approach. RESULTS Study Population Demographic characteristics of the study population are shown in Table 1. Table 1. Demographics and Clinical Test Results of Participants

Statistical Analysis Statistical analysis was conducted using IBM SPSS Statistics version 20 and GraphPad Prism version 5.00 for Mac OS X, GraphPad Software, California. Tuberculosis disease stage compared all tuberculosis (n = 13) vs all latent tuberculosis infection (n = 21) regardless of HIV status, the impact of HIV compared all HIV-infected (n = 17) vs uninfected (n = 17) regardless of tuberculosis disease stage and across all 4 subgroups. The 2-tailed Mann–Whitney U test was used for nonparametric 2-sample comparisons. Spearman rank correlation coefficients were used to test correlations. Receiver operator characteristic (ROC) curve defined the sensitivity and specificity of the diagnostic approach. RESULTS Study Population Demographic characteristics of the study population are shown in Table 1. Table 1. Demographics and Clinical Test Results of Participants HIV/Tuberculosis Tuberculosis HIV/Latent Tuberculosis Infection Latent Tuberculosis Infection Total 7 (%) 6 (%) 10 (%) 11 (%) 34 (%) Median (IQR) age 43 (40.5–52.4) 34.5 (28.0–56.0) 36 (24.0–39.0) 33 (31.0–35.5) 35.5 (31.3–40.8) Sex Male 4 (57.1) 3 (50.0) 6 (60.0) 4 (36.4) 17 (50.0) Female 3 (42.9) 3 (50.0) 4 (40.0) 7 (63.6) 17 (50.0) Ethnicity Black 5 (71.4) 1 (16.7) 8 (80.0) 6 (54.5) 20 (58.8) Asian 1 (14.3) 3 (50.0) 0 (0.0) 3 (27.3) 7 (20.6) White 1 (14.3) 2 (33.3) 2 (20.0) 2 (18.2) 7 (20.6) BCG vaccination yes 5 (71.4) 3 (50.0) 9 (90.0) 8 (72.7) 25 (73.5) no 1 (14.3) 2 (33.3) 0 (0.0) 2 (18.2) 5 (14.7) unknown 1 (14.3) 1 (16.7) 1 (10.0) 1 (9.1) 4 (11.8) Microbiological (smear/culture) confirmation positive 7a (100.0) 6a (100.0) NA NA NA NA 13 (100.0) negative 0 (0.0) 0 (0.0) NA NA NA NA 0 (0.0) HIV test positive 7 (100.0) 0 (0.0) 10 (100.0) 0 (0.0) 17 (50.0) negative 0 (0.0) 6 (100.0) 0 (0.0) 6 (54.5) 12 (35.3) not done 0 (0.0) 0 (0.0) 0 (0.0) 5b (45.5) 5 (14.7) Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range; NA, not applicable.

(100.0) negative 0 (0.0) 0 (0.0) NA NA NA NA 0 (0.0) HIV test positive 7 (100.0) 0 (0.0) 10 (100.0) 0 (0.0) 17 (50.0) negative 0 (0.0) 6 (100.0) 0 (0.0) 6 (54.5) 12 (35.3) not done 0 (0.0) 0 (0.0) 0 (0.0) 5b (45.5) 5 (14.7) Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range; NA, not applicable. All subjects tested positive in one or more of the tuberculin skin test, TSPOT.TB, QuantiFERON-TB Gold In-Tube or M. tuberculosis IFN-γ ELISpot, performed clinically or for the current study. a 10 patients with tuberculosis had not started treatment at the point of recruitment, 2 had received <14 days treatment, and 1 had received ≥14 days treatment. b All participants without clinical need for HIV testing had normal CD4:CD8 ratios.

All subjects tested positive in one or more of the tuberculin skin test, TSPOT.TB, QuantiFERON-TB Gold In-Tube or M. tuberculosis IFN-γ ELISpot, performed clinically or for the current study. a 10 patients with tuberculosis had not started treatment at the point of recruitment, 2 had received <14 days treatment, and 1 had received ≥14 days treatment. b All participants without clinical need for HIV testing had normal CD4:CD8 ratios. The Frequency of CD4+ and CD8+ Cells With an IFN-γ+ and TNF-α+ Response Was Increased in Active Tuberculosis We first examined the frequency of CD4+ and CD8+ functional effector cell subsets. Boolean gating was used to create individual nonoverlapping subsets by combining data in 3 dimensions (Figure 1A). The frequency of PPD-specific CD4+ IFN-γ only, TNF-α only, and IFN-γ/TNF-α-dual-secreting cells was higher in active tuberculosis compared with latent tuberculosis infection (P = .003, .002, and .002, respectively; Figure 1B). A similar relationship was seen for RD1-peptide-specifc CD4+ IFN-γ-only-secreting cells (not significant; data not shown), and IFN-γ/TNF-α-dual-secreting cells (P = .017; Figure 1B). The presence of HIV infection was not associated with an altered frequency of these cell subsets. We observed no difference in the frequency of trifunctional cells in patients with active tuberculosis compared with those with latent tuberculosis infection (Supplementary Figure 1). Figure 1. Frequency of IFN-γ and TNF-α secreting CD4+ and CD8+ cell subsets are increased in active tuberculosis. A, Example gating strategy for the CD4+ TNF-α-only-secreting subset using representative plots from an individual with active tuberculosis whose cells were stimulated overnight with PPD is shown. Cells were gated on live singlets (not shown) and CD3 + CD4+ cells (top row), then according to IFN-γ, IL-2, and TNF-α expression using FMOs (middle row). Boolean gating was used to define individual nonoverlapping functional subsets, eg, the TNF-α-only subset which did not express IFN-γ or IL-2 (bottom row). Graphs show frequency and median of CD4+ (B) and CD8+ (C) cells secreting IFN-γ and TNF-α in response to overnight stimulation with PPD or RD1-peptides in participants with active tuberculosis vs latent tuberculosis infection. Those with HIV coinfection (filled circles) and without HIV coinfection (open circles) are indicated. Results were analyzed by Mann–Whitney U test; and P values of < .05 were considered significant. Abbreviations: HIV, human immunodeficiency virus; IFN, interferon; IL, interleukin; PPD, purified protein derivative; TNF, tumor necrosis factor.

on (filled circles) and without HIV coinfection (open circles) are indicated. Results were analyzed by Mann–Whitney U test; and P values of < .05 were considered significant. Abbreviations: HIV, human immunodeficiency virus; IFN, interferon; IL, interleukin; PPD, purified protein derivative; TNF, tumor necrosis factor. The majority of participants with active tuberculosis, but not latent tuberculosis infection, had a CD8+ IFN-γ response (PPD: 12/13 for active tuberculosis vs 6/21 for latent tuberculosis infection, RD1-peptides: 10/13 for active tuberculosis vs 11/21 for latent tuberculosis infection). The frequencies of PPD- and RD1-peptide-specific CD8+ IFN-γ-only producing cells were significantly higher in active tuberculosis than in latent tuberculosis infection (P = .017 and .016, respectively), as was the frequency of CD8+ PPD-specific cells secreting both IFN-γ and TNF-α (P = .013; Figure 1C). HIV coinfection was (nonsignificantly) associated with a reduced frequency of PPD-specific IFN-γ/TNF-α-dual and TNF-α-only responses in active tuberculosis compared with HIV negativity (P = .051 for both). In the HIV-uninfected tuberculosis group the percentages of these PPD-specific CD8+ cells were significantly higher than in latent tuberculosis infection (P = .008 and .022, respectively). Similarly, the frequency of cells secreting IFN-γ-only were significantly higher in active tuberculosis compared with latent tuberculosis infection in HIV-uninfected (P = .023). These CD8+ effector functional subsets were therefore related to mycobacterial load analogously to equivalent CD4+ subsets, but the impact of HIV coinfection was more profound.

secreting IFN-γ-only were significantly higher in active tuberculosis compared with latent tuberculosis infection in HIV-uninfected (P = .023). These CD8+ effector functional subsets were therefore related to mycobacterial load analogously to equivalent CD4+ subsets, but the impact of HIV coinfection was more profound. PPD-specific and RD1-peptide-specific CD4+ Cellular Differentiation Was Increased in Active Tuberculosis vs Latent Tuberculosis Infection We analyzed the memory phenotype of the CD4+ functional subsets as putative correlates of mycobacterial pathogen load. Nonresponders were excluded. Each T-cell functional subset was gated for expression of CD45RA and CCR7 (Figure 2A). Memory phenotypes of functional subsets were defined as naive CD45RA + CCR7+, central memory (TCM) CD45RA− CCR7+, effector memory (TEM) CD45RA−CCR7− and CD45RA+ effectors (TEMRA) CD45RA+ CCR7−. This last subset was mainly evident in CD8+ cells. Figure 2. Cell surface phenotype of CD4+ cell functional subsets is influenced by tuberculosis disease stage: CD4+ cell functional subsets were examined for CD45RA and CCR7 expression in active vs latent tuberculosis infection in those with a positive response. A, An example gating strategy for a PPD-specific CD4+ IFN-γ-only secreting subset is demonstrated using representative plots from an individual with active tuberculosis. Each CD3 + CD4+ (top row) functional subset, eg, IFN-γ-only-secreting cells (middle row) was analyzed for expression of CD45RA and CCR7 (bottom row). B, Graphs show the percentage (and median percentage) of PPD-stimulated CD4+ IFN-γ-only (top row), TNF-α-only (middle row), and IL-2-only (bottom row) cells that were CD45RA-CCR7+ (TCM) (left column) and CD45RA-CCR7- (TEM) (right column) in patients with active tuberculosis and latent tuberculosis infection. The top 2 rows are representative of changes observed in active vs latent tuberculosis infection in all M. tuberculosis-specific (responding to PPD and RD-1-peptides) CD4+ functional subsets except IL-2-only-secreting cells. Those with HIV coinfection (filled circles) and without HIV coinfection (open circles) are indicated. Results were analyzed by Mann Whitney U test; and P values of < .05 were considered significant. Abbreviations: HIV, human immunodeficiency virus; IFN, interferon; IL, interleukin; M. tuberculosis, Mycobacterium tuberculosis; PPD, purified protein derivative; TNF, tumor necrosis factor.

on (open circles) are indicated. Results were analyzed by Mann Whitney U test; and P values of < .05 were considered significant. Abbreviations: HIV, human immunodeficiency virus; IFN, interferon; IL, interleukin; M. tuberculosis, Mycobacterium tuberculosis; PPD, purified protein derivative; TNF, tumor necrosis factor. PPD- and RD1-peptide-specific CD4+ cell effector functional subsets were principally TCM in latent tuberculosis infection compared to TEM in active tuberculosis, for example, fewer PPD-specific CD4+ IFN-γ-only secreting cells were TCM in active tuberculosis compared with latent tuberculosis infection (P = .005; Figure 2B). Comparisons of HIV-infected with uninfected patients were nonsignificant except for the CD4+ RD1-peptide-specific IFN-γ-only- and TNF-α/IL-2-dual-secreting subsets, fewer of which were TCM in HIV-infected than uninfected subjects (P = .030 and .006; Supplementary Figure 2). In contrast to the CD4+ effector functional subsets, CD4+ IL-2-only PPD- (Figure 2B) and RD1-peptide-specific cells (data not shown) were principally TCM in both active tuberculosis and latent tuberculosis infection and were therefore unaffected by tuberculosis disease stage. CD8+ IFN-γ-only cellular responses to PPD and RD1-peptides were mostly TEM and TEMRA (data not shown).

L-2-only PPD- (Figure 2B) and RD1-peptide-specific cells (data not shown) were principally TCM in both active tuberculosis and latent tuberculosis infection and were therefore unaffected by tuberculosis disease stage. CD8+ IFN-γ-only cellular responses to PPD and RD1-peptides were mostly TEM and TEMRA (data not shown). CD127 (IL7Rα) expression is reduced following antigen stimulation in effector T cells [17] and defines CD4+ and CD8+ subsets of differentiated murine T effector cells distinct from effector memory cells [18, 19]. We therefore measured CD127 expression on antigen-specific CD4+ functional T-cell subsets (Figure 3A). A smaller percentage of PPD- and RD1-peptide-specific CD4+ cells expressed CD127 in active tuberculosis compared with latent tuberculosis infection, for example, a lower percentage of PPD-specific CD4+ IFN-γ-only- and TNF-α-only-secreting cells expressed CD127 in active tuberculosis compared with latent tuberculosis infection (P < .001 and P = .003, respectively; Figure 3B). Expression of CD127 on antigen-specific cells was unaffected by HIV status (Figure 3B). However, in patients with HIV coinfection, frequencies of several subsets of PPD-specific T cells expressing CD127 correlated with CD4 count (Figure 3C). Similarly, for RD-1-peptide-specific cells, IFN-γ/IL-2-dual-producing cells expressing CD127 correlated with CD4 count (ρ = 0.647; P = .047) and HIV VL correlated inversely with IFN-γ-only (ρ = −0.780; P = .013) and IFN-γ/TNF-α-dual-secreting CD127-expressing cells (ρ = −0.727; P = .005; data not shown). Figure 3. Percentage of M. tuberculosis-specific CD4+ functional T-cell subsets expressing CD127 is influenced by stage of tuberculosis infection and CD4 count: CD3 + CD4+ functional cell subsets were examined for CD127 expression. A, A representative gating strategy is shown. PBMCs from an individual with latent tuberculosis infection infection were stimulated overnight with PPD, and CD3 + CD4+ cells (top row) were gated for cytokine secretion, eg, IFN-γ-only-secreting subset (middle row) and analyzed for expression of CD127 (bottom row).

ssion. A, A representative gating strategy is shown. PBMCs from an individual with latent tuberculosis infection infection were stimulated overnight with PPD, and CD3 + CD4+ cells (top row) were gated for cytokine secretion, eg, IFN-γ-only-secreting subset (middle row) and analyzed for expression of CD127 (bottom row). Expansion of Differentiated CD4+ Functional Effector T Cells in Active Tuberculosis vs Latent Tuberculosis Infection We next investigated the potential of combined phenotypic and functional measurement as a clinical biomarker. We determined the percentage of PPD- and RD1-peptide-specific CD4+ cells secreting IFN-γ-only or TNF-α-only that were differentiated effector cells (TEFF; CD45RA-CCR7-CD127-; Figure 4A [active tuberculosis] and B [latent tuberculosis infection]). In active tuberculosis, compared with latent tuberculosis infection, a higher percentage of CD4+ cells secreting IFN-γ-only or TNF-α-only in response to PPD and RD1-peptides were TEFF. This was most significant for CD4+ PPD-specific CD4+ cells secreting TNF-α-only (P < .0001) and IFN-γ-only (P < .0001). A cutoff of 17.3% of TNF-α-only cells of TEFF phenotype distinguished active tuberculosis from latent tuberculosis infection with 100% sensitivity (95% CI, 73.5–100.0) and 92.9% specificity (95% CI, 66.1–99.8; Table 2). In ROC analysis, area under the curve was 0.99 (95% CI, .97–1.01; P < .0001; Figure 4C). Similar, although slightly less discriminatory ROC curves were generated for PPD-specific IFN-γ-only cells and for RD-1-specific cells (Figure 4C). Table 2. Clinical and Radiological Characteristics of Cases Sorted by Percentage of TNF-α-only-Secreting Cells That Were TEFF (CD45RA−CCR7−CD127−)

1; P < .0001; Figure 4C). Similar, although slightly less discriminatory ROC curves were generated for PPD-specific IFN-γ-only cells and for RD-1-specific cells (Figure 4C). Table 2. Clinical and Radiological Characteristics of Cases Sorted by Percentage of TNF-α-only-Secreting Cells That Were TEFF (CD45RA−CCR7−CD127−) No. %TEFF/TNF-α only HIV CD4 VL Sputum smear M.

1; P < .0001; Figure 4C). Similar, although slightly less discriminatory ROC curves were generated for PPD-specific IFN-γ-only cells and for RD-1-specific cells (Figure 4C). Table 2. Clinical and Radiological Characteristics of Cases Sorted by Percentage of TNF-α-only-Secreting Cells That Were TEFF (CD45RA−CCR7−CD127−) No. %TEFF/TNF-α only HIV CD4 VL Sputum smear M. tuberculosis culture Culture site Radiology (CXR or CT) Tuberculosis final diagnosis S135 78.9 Pos 190 281 671 Neg Pos BAL and pleural fluid Bilateral pleural effusions, lung and splenic nodules, peritoneal thickening Pulmonary S126 75.0 Pos 69 601 000 Pos Pos Sputum Azygos lobe focal consolidation in cavity, pleural effusions, no lymphadenopathy Pulmonary S221 72.4 Neg NA NA NT Pos Lymph node Enlarged low density lymph nodes in mediastinum and left axilla Extra pulmonary S059 46.9 Pos 140 <100 Pos Pos Sputum and BAL Effusion, thickened pleura, loss of volume left lung, ground glass change Pulmonary S184 40.5 Neg NA NA NT Pos Lymph node Multiple mediastinal, coeliac axis lymph nodes with nodules in spleen and breast Extra pulmonary S193 37.6 Pos 136 28 958 NT Pos Lymph node Axillary, para-aortic, and abdominal lymphadenopathy, subpleural nodules, liver lesions Extra pulmonary S083 30.0 Neg NA NA Neg Pos Lymph node, BAL, peritoneal Right pleural collection and right paratracheal lymphadenopathy Pulmonary S076 29.2 Pos 200 <50 Pos Pos Left upper lobe, BAL, sputum Consolidation and cavitation upper lobe, interstitial opacities, linear atelectasis Pulmonary S115 28.6 Neg NA NA NT Pos Lymph node Mediastianal lymphadenopathy Extra pulmonary S146 20.1 Pos 250 52 205 NT Pos Lymph node Supraclavicular, mediastinal, and abdominal lymphadenopathy, nodular infiltrates Extra pulmonary S195 18.0 Neg NA NA Pos Pos BAL and lymph node Mediastinal, hilar, and supraclavicular lymph nodes, patchy consolidation Pulmonary S153 17.8 Neg NA NA NT NT NT Nil of note Latent tuberculosis infection S082 17.5 Neg NA NA Neg Pos Lymph node Nil of note Extra pulmonary S074 17.1 Neg NA NA Neg M fortuitum BAL Opacification right upper lobe, small volume axillary, and mediastinal lymph nodes Latent tuberculosis infection S050 16.7 Pos 480 12 479 NT NT NT Nil of note Latent tuberculosis infection S052 13.3 Pos 520 45 719 Neg Neg Sputum Nil of note Latent tuberculosis infection S177 12.8 Neg NA Na NT NT NT Nil of note Latent tuberculosis infection S094 10.9 Pos 660 <50 NT NT NT Heavily calcified nodule and small lymph nodes right upper lobe Latent tuberculosis infection S092 10.3 NT NA NA NT NT NT N

is infection S052 13.3 Pos 520 45 719 Neg Neg Sputum Nil of note Latent tuberculosis infection S177 12.8 Neg NA Na NT NT NT Nil of note Latent tuberculosis infection S094 10.9 Pos 660 <50 NT NT NT Heavily calcified nodule and small lymph nodes right upper lobe Latent tuberculosis infection S092 10.3 NT NA NA NT NT NT N il of note Latent tuberculosis infection S145 9.4 Neg NA NA NT NT NT NT Latent tuberculosis infection S098 8.6 Neg NA NA NT NT NT (Fractured ribs T4–9 posteriorly) Latent tuberculosis infection S047 8.5 Pos 360 <50 NT NT NT Nil of note Latent tuberculosis infection S079 6.2 NT NA NA NT NT NT Nil of note Latent tuberculosis infection S001 5.9 Pos 430 775 NT NT NT Nil of note Latent tuberculosis infection S099 5.8 NT NA NA NT NT NT Nil of note Latent tuberculosis infection S120 4.7 NT NA NA NT NT NT Fibrosis both apices, hilar lymphadenopathy, pleural thickening Latent tuberculosis infection S097 NA Pos 177 19 906 Pos Neg Sputum Subcarinal and axillary lymph nodes Pulmonary S029 NA Pos 530 <50 NT NT NT Nil of note Latent tuberculosis infection S025 NA Pos 330 <50 NT NT NT Nil of note Latent tuberculosis infection S197 NA Pos 210 <50 NT NT NT Nil of note Latent tuberculosis infection S201 NA Neg NA NA NT NT NT Nil of note Latent tuberculosis infection S171 NA Pos 365 <50 NT NT NT Nil of note Latent tuberculosis infection S191 NA Pos 530 13 328 NT NT NT NT Latent tuberculosis infection S121 NA Neg NA NA NT NT NT Nil of note Latent tuberculosis infection Abbreviations: BAL, bronchoalveolar lavage; CT, computerized tomography; CXR, chest x-ray; HIV, human immunodeficiency virus; NA, not applicable; Neg, negative; NT, not tested; Pos, positive; TNF, tumor necrosis factor; VL, viral load.

losis infection S121 NA Neg NA NA NT NT NT Nil of note Latent tuberculosis infection Abbreviations: BAL, bronchoalveolar lavage; CT, computerized tomography; CXR, chest x-ray; HIV, human immunodeficiency virus; NA, not applicable; Neg, negative; NT, not tested; Pos, positive; TNF, tumor necrosis factor; VL, viral load. Those with >17.3% TEFF/TNF-alpha only cells would be predicted to have active tuberculosis in the original analysis and those below it to have latent tuberculosis infection. Those below the second bold dividing line did not have a positive TNF-α-only response to purified protein derivative. Figure 4. Combining functional subset analysis with memory phenotype reveals a potentially powerful biomarker to distinguish active and latent tuberculosis infection. Boolean gating was used to analyze the percentage of PPD-specific CD4+ TNF-α-only-secreting cells that had the phenotype TEFF (CD45RA-CCR7-CD127-) in active and latent tuberculosis infection. A representative gating strategy is shown for individuals with active tuberculosis (A) and latent tuberculosis infection (B). To test whether this approach was robust to differences in disease site, we compared individuals with active pulmonary and extrapulmonary disease. We found no significant difference in the proportion of PPD-specific cells secreting IFN-γ-only, or TNF-α-only that were TEFF (Figure 4D) when stratified by site of disease or HIV coinfection. There was also no difference in the proportion of RD-1 peptide-specific cells that were TEFF when stratified by disease site (data not shown).

nificant difference in the proportion of PPD-specific cells secreting IFN-γ-only, or TNF-α-only that were TEFF (Figure 4D) when stratified by site of disease or HIV coinfection. There was also no difference in the proportion of RD-1 peptide-specific cells that were TEFF when stratified by disease site (data not shown). The operator conducting analyses (KMP) was also involved in participant recruitment and therefore not blinded to patient categorization. To demonstrate integrity and reproducibility of the results, data for all study participants were regated and reanalysed by a second independent operator (HSW) who was blinded to patient diagnoses. Correlations between results obtained by operators 1 and 2 for the percentage of CD3 + CD4+ cells secreting TNF-α-only (ρ = 0.97, P < .0001), the percentage of CD3 + CD4+ TNF-α-only-secreting cells that are CD127+ (ρ = 0.96, P < .0001) and the percentage of CD3 + CD4+ TNF-α-only-secreting cells that are CD45RA−CCR7− (ρ = 0.88, P < .0001) were very strong. Using the 17.3% cutoff for TNF-α-only-secreting cells of TEFF phenotype to distinguish tuberculosis from latent tuberculosis infection, operator 2 misclassified 1 case of tuberculosis as latent tuberculosis infection, and 1 further case of latent tuberculosis infection as tuberculosis (data not shown).

ere very strong. Using the 17.3% cutoff for TNF-α-only-secreting cells of TEFF phenotype to distinguish tuberculosis from latent tuberculosis infection, operator 2 misclassified 1 case of tuberculosis as latent tuberculosis infection, and 1 further case of latent tuberculosis infection as tuberculosis (data not shown). DISCUSSION Our detailed interrogation of antigen-specific T-cell phenotype and function has delineated the association of tuberculosis disease stage with M. tuberculosis-specific cellular immunity. active tuberculosis was associated with an increased frequency of mono- or dual-functional CD4+ and CD8+ M. tuberculosis-specific T cells that secrete IFN-γ and/or TNF-α making these subsets potential biomarkers of disease activity.

ociation of tuberculosis disease stage with M. tuberculosis-specific cellular immunity. active tuberculosis was associated with an increased frequency of mono- or dual-functional CD4+ and CD8+ M. tuberculosis-specific T cells that secrete IFN-γ and/or TNF-α making these subsets potential biomarkers of disease activity. Measurement of the proportion of TNF-α-only secreting M. tuberculosis-responsive CD4+ cells has proved promising to distinguish between active tuberculosis and latent tuberculosis infection [14] but neither earlier data nor our findings have clinically sufficient discriminatory power. The slight difference in our findings compared with those previously published may be because we stratified for HIV infection and tuberculosis disease stage and we did not use costimulation in our assays. The use of costimulation may reflect the cells’ ability to react to certain stimuli not necessarily present in vivo, instead of measuring the ability of these cells to respond to antigen. Studies have suggested that the presence of trifunctional cells secreting IFN-γ, IL-2, and TNF-α correlates with active tuberculosis [20, 21], but our data did not support this association. Our functional data therefore partially corroborates and extends earlier observations [6, 14] that measurement of the frequency of M. tuberculosis-specific CD4+ effector-like cells secreting IFN-γ and/or TNF-α can distinguish active tuberculosis from latent tuberculosis infection.

data did not support this association. Our functional data therefore partially corroborates and extends earlier observations [6, 14] that measurement of the frequency of M. tuberculosis-specific CD4+ effector-like cells secreting IFN-γ and/or TNF-α can distinguish active tuberculosis from latent tuberculosis infection. Simultaneous evaluation of memory phenotype of responding cells provided a more sensitive and specific surrogate than CD4+ functional profile alone. Expression of CD45RA, CCR7, and CD127 on M. tuberculosis-specific T cells secreting only IFN-γ or TNF-α was lowest in those with active tuberculosis. Interestingly, memory phenotype was not exclusively linked to the functional profile (except for IL-2-only cells which were mainly TCM) but was closely related to underlying tuberculosis stage. These markers might therefore serve as indicators of tuberculosis activation. Combined measurement of both functional profile and differentiation phenotype provided a highly discriminatory immunological read-out for active tuberculosis and latent tuberculosis infection. In those with active tuberculosis, >17.3% of PPD-specific CD4+ TNF-α-only-secreting cells were CD45RA−CCR7−CD127−, and this phenotype was therefore strongly associated with activated infection. Given that responses to PPD are less specific for M. tuberculosis infection than responses to RD-1 antigens, this PPD-specific T-cell signature may be used to distinguish tuberculosis from latent tuberculosis infection in the second step of a 2-step diagnostic testing strategy, where M. tuberculosis infection has been ruled in at step one by a positive result in an RD-1-based immunodiagnostic test (eg, IGRA).

o RD-1 antigens, this PPD-specific T-cell signature may be used to distinguish tuberculosis from latent tuberculosis infection in the second step of a 2-step diagnostic testing strategy, where M. tuberculosis infection has been ruled in at step one by a positive result in an RD-1-based immunodiagnostic test (eg, IGRA). Memory phenotype of responding cells and association with tuberculosis stage has previously been evaluated qualitatively and to some extent quantitatively. Studies have suggested that changes in the predominant memory phenotype of the responding CD4+ and CD8+ cells might be dependent on severe vs nonsevere, or active vs latent tuberculosis in adults and children [22–25]. None were able to simultaneously measure the antigen-specific function and phenotype of multiple CD4+ and CD8+ T-cell subsets. Equally these studies did not specifically measure cells with the TEFF phenotype, which had lost CD127 expression. Therefore, although the principle was explored, none identified a subset that could serve as a highly sensitive and specific clinical biomarker.

tion and phenotype of multiple CD4+ and CD8+ T-cell subsets. Equally these studies did not specifically measure cells with the TEFF phenotype, which had lost CD127 expression. Therefore, although the principle was explored, none identified a subset that could serve as a highly sensitive and specific clinical biomarker. The frequency of both PPD and M. tuberculosis-specific CD8+ functional effectors increased in active tuberculosis and might therefore also be responding to increased mycobacterial load, consistent with the observation that these cells respond more effectively to heavily infected dendritic cells [26]. The frequency of M. tuberculosis-specific CD8+ cells was increased in adults with sputum smear-positive pulmonary tuberculosis [6] and in children with active tuberculosis compared with contacts [27]. Similarly our data showed that the frequency of CD8+ M. tuberculosis-specific cells, and therefore proportion of responders, was increased in active tuberculosis and this approach holds promise for the discrimination of tuberculosis disease stage especially in HIV coinfection where all participants had a positive response to M. tuberculosis peptides. This association, shown by us and others [6], precludes the comparison of combined function and phenotype of M. tuberculosis-specific CD8+ cells, however, because nonresponders were by default mainly in the latent tuberculosis infection group. Measurement of CD8+ functional subsets in active tuberculosis and latent tuberculosis infection was therefore not sufficiently discriminatory for active and latent tuberculosis infection in our cohort.

specific CD8+ cells, however, because nonresponders were by default mainly in the latent tuberculosis infection group. Measurement of CD8+ functional subsets in active tuberculosis and latent tuberculosis infection was therefore not sufficiently discriminatory for active and latent tuberculosis infection in our cohort. In this study we included individuals with HIV coinfection to compare and distinguish the impact of active tuberculosis on M. tuberculosis-specific cellular immunity with the impact of HIV coinfection. Where CD4+ M. tuberculosis-specific effector-like cells were influenced by tuberculosis disease stage, the impact of HIV coinfection per se was rarely significantly associated with these changes. This may have been partially due to the inclusion of patients who were treated for HIV infection. However, in the case of CD127, stratification by CD4 count showed that the stage of HIV disease influenced expression of this marker on M. tuberculosis-specific T cells. Reduced CD127 expression on HIV-specific CD8+ T cells (reviewed in [28]) and CD4+ T cells [29] is observed with HIV disease progression, but a relationship between HIV disease progression and CD127 expression on M. tuberculosis-specific T cells has not previously been noted. Our finding indicates that in HIV coinfection, M. tuberculosis-antigen-specific CD4+ cells lose CD127 expression with advancing HIV disease and are therefore potentially more differentiated. This effect could be directly virus-induced or secondary to increasing subclinical mycobacterial burden with advancing HIV infection.

finding indicates that in HIV coinfection, M. tuberculosis-antigen-specific CD4+ cells lose CD127 expression with advancing HIV disease and are therefore potentially more differentiated. This effect could be directly virus-induced or secondary to increasing subclinical mycobacterial burden with advancing HIV infection. Our cohort included individuals with both pulmonary and extra-pulmonary infection. HIV infection is more commonly associated with tuberculosis dissemination as evidenced by the widespread involvement in some of these individuals. Despite some variation in clinical phenotype of those with active tuberculosis, our biomarker reliably distinguished tuberculosis stage, regardless of site of disease suggesting that it may remain robust across the clinical tuberculosis disease spectrum and therefore have wide applicability. HIV coinfection has recently been shown to be associated with subclinical active tuberculosis infection in a tuberculosis endemic area [30]. No individuals with latent tuberculosis infection developed active tuberculosis during 12 months of follow-up suggesting that, in our cohort (recruited in a tuberculosis nonendemic area), subclinical tuberculosis was not present in those classified as latent tuberculosis infection. The lack of continuous exogenous priming or restimulation due to tuberculosis exposure distinguishes our cohort from others recruited from tuberculosis endemic areas and removes this is as a possible confounding effect on the immunological changes we observed.

esent in those classified as latent tuberculosis infection. The lack of continuous exogenous priming or restimulation due to tuberculosis exposure distinguishes our cohort from others recruited from tuberculosis endemic areas and removes this is as a possible confounding effect on the immunological changes we observed. It is unknown whether the changes in function and phenotype we demonstrated are consequent upon, or causal of, increased disease activity. In a study of tuberculosis contacts there was no percentage difference in CD4+ or CD8+ memory phenotype at baseline in progressors and nonprogressors [31] arguing against a causal role. Rather, mycobacterial burden might be driving the dominant functional and phenotypic changes, strengthening the potential of these markers to serve as a surrogate of mycobacterial burden. The loss of CCR7, allowing T-cell accumulation at the site of disease [32], and the loss of CD127, limiting the number and life span of these potentially tissue-destructive cells, may reflect an active adaptive response to replicating mycobacteria. However, given that immune containment has, by definition, been lost in active tuberculosis, it is equally possible that these T-cell subsets contribute to tissue pathology, cavitation, and bacillary dissemination.

otentially tissue-destructive cells, may reflect an active adaptive response to replicating mycobacteria. However, given that immune containment has, by definition, been lost in active tuberculosis, it is equally possible that these T-cell subsets contribute to tissue pathology, cavitation, and bacillary dissemination. Limitations of this study included the modest cohort size and some heterogeneity. Our cohort was sufficient to reveal powerful positive associations with tuberculosis stage; but a larger study would be required to confirm the absence of associations, for example, between HIV coinfection and frequencies of certain T-cell subsets. Moreover, because most of our HIV-infected participants had relatively high CD4 cell counts, our findings may be less applicable to more immunosuppressed HIV-infected patients. Similarly, while our data suggested that our biomarker was independent of tuberculosis disease site (ie, pulmonary vs extrapulmonary tuberculosis), a larger sample size is needed to corroborate this. Such studies could include other antigens, for example, heparin-binding hemagglutinin adhesin, that have shown promise in distinguishing active tuberculosis from latent tuberculosis infection and risk stratification of those with infection [33, 34]

nary tuberculosis), a larger sample size is needed to corroborate this. Such studies could include other antigens, for example, heparin-binding hemagglutinin adhesin, that have shown promise in distinguishing active tuberculosis from latent tuberculosis infection and risk stratification of those with infection [33, 34] Through dissection of the impact of varying tuberculosis stage as a surrogate for mycobacterial pathogen burden, with and without HIV coinfection, we have identified cellular changes that are highly sensitive to tuberculosis activity. Future work will require prospective evaluation of our findings in an independent validation cohort, which if corroborative will have significant clinical impact. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. Dr Lee Potiphar, Mr Samuel Bremang, the staff at the Jefferiss Wing Clinic, St Mary's Hospital, London, and Northwick Park Hospital Genitourinary and HIV Medicine clinic, Harrow.

nd group D (BCG–MVA85A; P = .02, Mann–Whitney U test; Figure 2). Using PCR, there was a significant 0.5- to 1-log reduction in estimated BCG copy number between the BCG-naive (A or B) and BCG-vaccinated groups (C or D; Mann–Whitney U test). No further reduction in BCG numbers was detected after vaccination with MVA85A. Ex Vivo IFN-γ ELISpot Responses Were as Expected for Vaccination Schedule Received Ex vivo IFN-γ ELISpot responses to PPD and a single pool of Ag85A peptides are shown in Figure 3. There were no significant differences in baseline Ag85A responses among the 4 groups (data not shown for groups A and C); however, baseline PPD responses were significantly higher (P = .008, Mann–Whitney U test) in the previously BCG-vaccinated groups. Those in group D who received MVA85A as a boost to BCG had significantly higher responses to Ag85A 7 days post vaccination than those in group B who were BCG naive, as previously reported [24]. However, Ag85A responses between the 2 groups were not significantly different on the day of challenge or on the day of biopsy. Responses to PPD on the day of challenge were significantly lower in group A compared with each of the other 3 groups; however, there was no significant difference in responses among groups B, C, and D at this time point. PPD responses for those in groups A and C were significantly higher 14 days post challenge than for those measured on the day of challenge (P = .002 and .0002, respectively; Wilcoxon matched-pairs). This significant increase was not observed in groups B and D due to the confounding effect of recent vaccination with MVA85A. Figure 3. Ex vivo interferon-gamma enzyme-linked immunosorbent spot (ELISpot) assay responses to purified protein derivative from Mycobacterium tuberculosis for all groups (A) and to a single pool of Ag85A peptides for groups B and D (B). A value of 1667 spot-forming cells (SFCs)/1 × 106 peripheral blood mononuclear cells (PBMCs) represents a blackout in the ELISpot well. ** P < .01. Abbreviation: BCG, bacille Calmette-Guérin.

Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data Notes Acknowledgments. Dr Lee Potiphar, Mr Samuel Bremang, the staff at the Jefferiss Wing Clinic, St Mary's Hospital, London, and Northwick Park Hospital Genitourinary and HIV Medicine clinic, Harrow. Financial support. This work was supported by a TB Funds Grant (TB08/02) from the British Lung Foundation. A. L. is a Wellcome Senior Research Fellow in Clinical Science and a National Institute of Health Research Senior Investigator. Potential conflicts of interest. A. L. is inventor of several patents underpinning T cell-based diagnosis. The ESAT-6/CFP-10 IFN-γ ELISpot was commercialized by an Oxford University spin-out company (T-SPOT.TB, Oxford Immunotec Ltd, Abingdon, UK) in which Oxford University and A. L. have a minority share of equity and royalty entitlements. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Plasmodium vivax threatens almost half of the world's population and is associated with significant, relapsing morbidity [1–3]. It is set to become the dominant malaria species in the Asia–Pacific region [2]. Transmission of P. vivax is dependent upon development of sufficient densities of mature, infectious gametocytes in the peripheral circulation and their subsequent ingestion by competent Anopheles mosquito vectors. Designing effective intervention strategies that will reduce the chance that transmission occurs requires a comprehensive understanding of the biological and epidemiological attributes of P. vivax gametocytes. We analyzed data from 3 large randomized controlled trials (1 on the Thai–Myanmar border [4] and 2 in Papua, Indonesia [5, 6]) to determine and compare the demographic and clinical factors associated with patent gametocytemia on presentation with vivax malaria and during the 6–9 weeks following treatment with 1 of 4 antimalarial regimens: artemether + lumefantrine (AM + LUM), dihydroartemisinin + piperaquine (DHA + PIP), artesunate + amodiaquine (AS + AQ), and chloroquine monotherapy (CQ). METHODS Study Sites Thailand The single Thai study in this analysis was conducted at Shoklo Malaria Research Unit clinics along the northwestern border of Thailand [4]. This region has low, seasonal malaria transmission with an incidence of approximately 1 episode per person-year, 53% due to P. vivax and 37% due to Plasmodium falciparum [7, 8]. P. vivax relapses occur approximately 3–4 weeks following administration of rapidly eliminated antimalarials [9].

tern border of Thailand [4]. This region has low, seasonal malaria transmission with an incidence of approximately 1 episode per person-year, 53% due to P. vivax and 37% due to Plasmodium falciparum [7, 8]. P. vivax relapses occur approximately 3–4 weeks following administration of rapidly eliminated antimalarials [9]. Papua, Indonesia The 2 Indonesian studies included in this analysis took place at the same 2 clinics in the municipality of Timika in south–central Indonesia, eastern Indonesia [5, 6]. The demographics and geography of this region have been described previously [3, 10]. In 2005, the prevalence of asexual parasitemia was 6.4% for P. vivax and 7.5% for P. falciparum [10]. Local P. vivax strains relapse at intervals of approximately 3 to 4 weeks [5, 6]. Study Designs The Thai study was carried out between January 2007 and December 2008 and compared DHA + PIP with CQ for slide-confirmed uncomplicated P. vivax monoinfections [4]. Primaquine was not given. Pregnant or lactating women; patients aged <1 year or <5 kg in weight; and those with known hypersensitivity to the study medications, intercurrent illness, or a hematocrit <20% were excluded.

er 2008 and compared DHA + PIP with CQ for slide-confirmed uncomplicated P. vivax monoinfections [4]. Primaquine was not given. Pregnant or lactating women; patients aged <1 year or <5 kg in weight; and those with known hypersensitivity to the study medications, intercurrent illness, or a hematocrit <20% were excluded. The first of the 2 Indonesian studies was carried out between July 2004 and June 2005 and compared DHA + PIP with AM + LUM for slide-confirmed uncomplicated malaria due to P. falciparum, P. vivax, or mixed species infection [5]. Unsupervised primaquine at a dose of 0.3 mg base/kg per day for 14 days was prescribed for patients with P. vivax and mixed species infections at day 28 if they did not have glucose-6-phosphate dehydrogenase (G6PD) deficiency.

nfirmed uncomplicated malaria due to P. falciparum, P. vivax, or mixed species infection [5]. Unsupervised primaquine at a dose of 0.3 mg base/kg per day for 14 days was prescribed for patients with P. vivax and mixed species infections at day 28 if they did not have glucose-6-phosphate dehydrogenase (G6PD) deficiency. The second of the 2 Indonesian studies was carried out between July 2005 and December 2005 and compared DHA + PIP with AS + AQ for the treatment of slide-confirmed uncomplicated P. falciparum, P. vivax, or mixed species malaria [6]. Unsupervised primaquine was offered to G6PD-normal individuals with P. vivax or mixed species malaria immediately after completion of the study regimens. Patients who were pregnant or lactating were excluded from both studies as were those who had a parasitemia of >4% or who fulfilled World Health Organization criteria for severe malaria [11]. The study comparing DHA + PIP with AL + LUM excluded individuals who weighed <10 kg, whereas the study of DHA + PIP vs AS + AQ excluded individuals who weighed <5 kg or were aged <1 year. Details of the drug regimens can be found in the respective study publications [4–6].

Organization criteria for severe malaria [11]. The study comparing DHA + PIP with AL + LUM excluded individuals who weighed <10 kg, whereas the study of DHA + PIP vs AS + AQ excluded individuals who weighed <5 kg or were aged <1 year. Details of the drug regimens can be found in the respective study publications [4–6]. In all studies, patients were followed with daily symptom enquiry and physical examination as well as blood smears until aparasitemic and afebrile. Thereafter, patients were followed weekly for 6 weeks (42 days) in Indonesia and 9 weeks (63 days) in Thailand. Block randomization and allocation concealment using sealed opaque envelopes were used in all studies. Drug administration was open label, but microscopists at both sites were unaware of patient allocation. Laboratory Methods In Thailand, sexual and asexual parasite counts, including the individual densities of trophozoites and schizonts, were expressed per 500 white blood cells (WBCs); if parasitemia was >1%, densities were expressed per 1000 red blood cells. Slides were declared negative after examination of at least 100 high-power fields. Hematocrit was measured using a microcentrifuge (Hawksley) and, in this analysis, converted to a hemoglobin concentration in g/dL by multiplying the percentage by a factor of 0.34 [12].

densities were expressed per 1000 red blood cells. Slides were declared negative after examination of at least 100 high-power fields. Hematocrit was measured using a microcentrifuge (Hawksley) and, in this analysis, converted to a hemoglobin concentration in g/dL by multiplying the percentage by a factor of 0.34 [12]. In Indonesia, asexual and sexual parasite counts were measured on Giemsa-stained thick films and reported per 200 WBCs. Slides were declared negative after examination of at least 100 high-power microscope fields. A thin film was also examined to confirm parasite species and for quantification per 1000 red blood cells if the parasitemia was >200 per 200 WBCs. Hemoglobin was measured using a portable photometer (HemoCue Hb201+, Angelholm, Sweden). G6PD status was tested using the fluorescent spot test in both locations.

e fields. A thin film was also examined to confirm parasite species and for quantification per 1000 red blood cells if the parasitemia was >200 per 200 WBCs. Hemoglobin was measured using a portable photometer (HemoCue Hb201+, Angelholm, Sweden). G6PD status was tested using the fluorescent spot test in both locations. Statistical Analysis The primary outcomes of interest were the presence or absence of P. vivax gametocytemia at enrollment, time to clearance of P. vivax gametocytes, and appearance of P. vivax gametocytes up to 63 days following enrollment. All analyses were stratified by country because there were several sources of intercountry heterogeneity, including differences in slide-reading protocols and likely differences in preexisting immunity to P. vivax malaria. The following were defined a priori as potential risk factors for gametocytemia at enrollment: sex, age (<5 years, 5 to <15 years, ≥15 years), G6PD status (normal or abnormal), asexual P. vivax parasite density (loge [parasites/µL]), anemia (hemoglobin <9 g/dL), fever (temperature >37.5°C), species of infection at enrollment (pure P. vivax vs mixed P. vivax/P. falciparum infection [Indonesia only]), and stage of infection at enrollment (presence or absence of schizonts [Thailand only]). Risk factors for the appearance of P. vivax gametocytes during follow-up were as above plus clearance of asexual parasitemia by day 1 (yes, no) and antimalarial regimen.

vs mixed P. vivax/P. falciparum infection [Indonesia only]), and stage of infection at enrollment (presence or absence of schizonts [Thailand only]). Risk factors for the appearance of P. vivax gametocytes during follow-up were as above plus clearance of asexual parasitemia by day 1 (yes, no) and antimalarial regimen. Analyses of gametocytemia at enrollment were done using separate univariable logistic regression models for the 2 locations. All patient factors were subsequently included in separate multivariable logistic regression models for each location. In Indonesia, there was no difference in gametocyte carriage following DHA + PIP in the first study (in which unsupervised primaquine was prescribed at day 28) compared with the second study (in which unsupervised primaquine was prescribed at day 3). Results from the 2 Indonesian studies were therefore pooled in all analyses.

nesia, there was no difference in gametocyte carriage following DHA + PIP in the first study (in which unsupervised primaquine was prescribed at day 28) compared with the second study (in which unsupervised primaquine was prescribed at day 3). Results from the 2 Indonesian studies were therefore pooled in all analyses. The cumulative incidence of P. vivax gametocytemia in each location between day 7 and the end of follow-up was assessed for each antimalarial regimen using the Kaplan-Meier method and compared using the log-rank test. Clinical and demographic risk factors for recurrent gametocytemia were examined using univariable and multivariable Cox regression models for each location (the latter stratified by treatment group). Fulfillment of the proportional hazards assumption was assessed by comparing log (cumulative hazard) by time of follow-up curves and subsequently by examination of Schoenfeld residuals. Patients who had recurrent asexual P. vivax infection without concurrent gametocytemia were retreated with antimalarial medication and were therefore censored at the time of recurrence. The proportions of individuals who had cleared their gametocytes by day 1 and day 2 were examined for each regimen stratified by location and compared using the χ² test or Fisher exact test. Comparisons of nonnormal distributions were made using the Mann–Whitney U test or the Wilcoxon signed rank test for matched data. We explored the association between asexual and sexual P. vivax parasite density (loge transformed) using Pearson correlation coefficient. All data merging and analyses were done using STATA version 10.1 (College Station, Texas).

were made using the Mann–Whitney U test or the Wilcoxon signed rank test for matched data. We explored the association between asexual and sexual P. vivax parasite density (loge transformed) using Pearson correlation coefficient. All data merging and analyses were done using STATA version 10.1 (College Station, Texas). Ethical Clearance The ethics committees of the Faculty of Tropical Medicine, Mahidol University, Thailand, and the Oxford Tropical Research, United Kingdom, approved the Thai study. The ethics committees of the National Institute of Health Research and Development, Indonesia, and the Menzies School of Health Research, Australia, approved the Indonesian studies. RESULTS A total of 492 patients with P. vivax monoinfections were evaluable in the Thai dataset and 476 patients with P. vivax infections (314 with monoinfections and 162 with concurrent P. falciparum infections) were evaluable in the Indonesian dataset (Table 1). Patients enrolled in the Thai study were slightly older and less anemic than their Indonesian counterparts and had higher asexual parasitemias (median = 6565/µL vs 2595/µL, P < .001). Results relevant to the patients with mixed infection in Indonesia are presented in a separate subsection below. Table 1. Characteristics of Evaluable Patients in the Thai and Indonesian Studies

d less anemic than their Indonesian counterparts and had higher asexual parasitemias (median = 6565/µL vs 2595/µL, P < .001). Results relevant to the patients with mixed infection in Indonesia are presented in a separate subsection below. Table 1. Characteristics of Evaluable Patients in the Thai and Indonesian Studies Characteristic Thailand Indonesia (monoinfections) Indonesia (mixed infections) N 492 314 162 Sex Male 328 (66.7%) 169 (53.8%) 103 (63.6%) Age <5 y 66 (13.4%) 86 (27.4%) 26 (16.0%) 5 to <15 y 135 (27.4%) 78 (24.8%) 44 (27.2%) >15 y 291 (59.1%) 150 (47.8%) 92 (56.8%) G6PD status Normal 463 (94.1%) 231 (73.6%) 104 (64.2%) Abnormal 28 (5.7%) 45 (14.3%) 12 (7.4%) Febrile (>37.5°C) 166 (33.7%) 52 (16.6%) 62 (38.3%) Asexual Plasmodium vivax parasitemia (/µL) 6565 (193–30 551)a 2595 (140–27 500)a 606 (38–14 036)a P. vivax gametocytes detected 415 (84.3%) 209 (66.6%) 92 (56.8%) P. vivax gametocytemia (per µL) 266 (33–2158)a 113 (35–727)a 98 (21–1205)a Hemoglobin (g/dL) 12.6 (9.9–15.6)a 10.3 (6.3–14.5)a 9.8 (5.6–14.4)a Anemia (Hb <9 g/dL) 5 (1.0%) 94 (29.9%) 65 (40.1%) Stage of infection Trophozoites alone 339 (68.9%) (  …  ) (  …  ) Trophozoites and schizonts 153 (31.1%) (  …  ) (  …  ) Treatment Artemether + lumefantrine 0 (0%) 84 (26.8%) 58 (35.8%) Dihydroartemisinin + piperaquine 248 (50.4%) 169 (53.8%) 86 (53.1%) Artesunate + amodiaquine 0 (0) 61 (19.4%) 18 (11.1%) Chloroquine 244 (49.6%) 0 (0%) 0 (0%) Abbreviation: G6PD, glucose-6-phosphate dehydrogenase. a Median (5th–95th percentiles).

Characteristic Thailand Indonesia (monoinfections) Indonesia (mixed infections) N 492 314 162 Sex Male 328 (66.7%) 169 (53.8%) 103 (63.6%) Age <5 y 66 (13.4%) 86 (27.4%) 26 (16.0%) 5 to <15 y 135 (27.4%) 78 (24.8%) 44 (27.2%) >15 y 291 (59.1%) 150 (47.8%) 92 (56.8%) G6PD status Normal 463 (94.1%) 231 (73.6%) 104 (64.2%) Abnormal 28 (5.7%) 45 (14.3%) 12 (7.4%) Febrile (>37.5°C) 166 (33.7%) 52 (16.6%) 62 (38.3%) Asexual Plasmodium vivax parasitemia (/µL) 6565 (193–30 551)a 2595 (140–27 500)a 606 (38–14 036)a P. vivax gametocytes detected 415 (84.3%) 209 (66.6%) 92 (56.8%) P. vivax gametocytemia (per µL) 266 (33–2158)a 113 (35–727)a 98 (21–1205)a Hemoglobin (g/dL) 12.6 (9.9–15.6)a 10.3 (6.3–14.5)a 9.8 (5.6–14.4)a Anemia (Hb <9 g/dL) 5 (1.0%) 94 (29.9%) 65 (40.1%) Stage of infection Trophozoites alone 339 (68.9%) (  …  ) (  …  ) Trophozoites and schizonts 153 (31.1%) (  …  ) (  …  ) Treatment Artemether + lumefantrine 0 (0%) 84 (26.8%) 58 (35.8%) Dihydroartemisinin + piperaquine 248 (50.4%) 169 (53.8%) 86 (53.1%) Artesunate + amodiaquine 0 (0) 61 (19.4%) 18 (11.1%) Chloroquine 244 (49.6%) 0 (0%) 0 (0%) Abbreviation: G6PD, glucose-6-phosphate dehydrogenase. a Median (5th–95th percentiles). Gametocytemia on Enrollment Gametocytes were detected on enrollment in 84.3% (415/492) of patients in Thailand and 66.6% (209/314) of patients with P. vivax monoinfections in Indonesia (P < .001). Thai patients also had a higher median gametocyte density than Indonesian patients (P < .001; Table 1 and Figure 1). Figure 1. Frequency distribution of loge gametocyte density for those with Plasmodium vivax monoinfections on presentation for treatment and at the time of recurrence.

infections in Indonesia (P < .001). Thai patients also had a higher median gametocyte density than Indonesian patients (P < .001; Table 1 and Figure 1). Figure 1. Frequency distribution of loge gametocyte density for those with Plasmodium vivax monoinfections on presentation for treatment and at the time of recurrence. In univariable analyses, higher loge asexual parasite density was associated with a highly statistically significant increase in the risk of gametocytemia on presentation in both locations (Table 2). Presence of schizonts on the admission blood film was associated with a 14-fold increased risk of gametocytemia in Thailand (P < .001). In multivariable analyses, the only independent predictors of gametocytemia on presentation were higher asexual parasitemia (adjusted odds ratio [AOR] per loge order increase = 2.31; 95% confidence interval [CI], 1.86–2.86; P < .001 in Thailand and AOR = 1.61; 95% CI, 1.39–1.87; P < .001 in Indonesia) and schizontemia at enrollment (AOR = 6.31; 95% CI, 1.78–22.4; P = .004 [Thailand only]). Table 2. Risk Factors for Plasmodium vivax Gametocytemia at Presentation in Patients Enrolled in the Thai and Indonesian Trials

terval [CI], 1.86–2.86; P < .001 in Thailand and AOR = 1.61; 95% CI, 1.39–1.87; P < .001 in Indonesia) and schizontemia at enrollment (AOR = 6.31; 95% CI, 1.78–22.4; P = .004 [Thailand only]). Table 2. Risk Factors for Plasmodium vivax Gametocytemia at Presentation in Patients Enrolled in the Thai and Indonesian Trials Univariable Models Multivariable Models Thailand Indonesia (monoinfection) Indonesia (mixed infection) Thailand Indonesia Risk Factor OR (95% CI) P Value OR (95% CI) P Value OR (95% CI) P Value AOR (95% CI) P Value AOR (95% CI) P Value Sex Male 1.00 1.00 1.00 1.00 1.00 Female 0.85 (.51–1.42) .54 1.03 (.64–1.66) .90 1.20 (.62–2.31) .58 0.97 (.50–1.88) .93 0.91 (.58–1.44) .70 Age <5 y 1.16 (.51–2.60) .73 0.86 (.48–1.54) .62 0.76 (.32–1.84) .55 1.22 (.44–3.45) .70 0.82 (.45–1.49) .51 5 to <15 y 0.58 (.34–.99) .05 0.54 (.30–.95) .03 0.75 (.36–1.56) .45 0.72 (.36–1.46) .36 0.59 (.34–1.02) .06 ≥15 y 1.00 1.00 1.00 1.00 1.00 G6PD status Normal 1.00 1.00 1.00 1.00 1.00 Abnormal 0.85 (.31–2.30) .75 1.51 (.74–3.07) .26 1.11 (.33–3.73) .87 1.11 (.30–4.03) .88 1.31 (.68–2.53) .42 Loge asexual parasite density (per loge order increase) 2.36 (1.95–2.85) <.001 1.49 (1.27–1.74) <.001 2.01 (1.57–2.58) <.001 2.31 (1.86–2.86) <.001 1.61 (1.39–1.87) <.001 Anemia (Hb <9 g/dL) 0.75 (.08–6.76) .79 1.89 (1.08–3.28) .03 1.35 (.71–2.57) .36 1.26 (.08–19.6) .87 1.39 (.83–2.34) .21 Fever (>37.5°C) 1.10 (.64–1.87) .74 1.59 (.81–3.13) .18 1.11 (.58–2.11) .76 0.53 (.27–1.05) .07 0.89 (.50–1.58) .69 Stage of infection Trophozoites alone 1.00 (  …  ) (  …  ) 1.00 (  …  ) Trophozoites and schizonts 14.0 (4.33–45.1) <.001 (  …  ) (  …  ) 6.31 (1.78–22.4) .004 (  …  ) Species of infection P. vivax monoinfection (  …  ) (  …  ) (  …  ) (  …  ) 1.00 Mixed P. vivax / P. falciparum (  …  ) (  …  ) (  …  ) (  …  ) 1.03 (.86–1.24) .74 Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; G6PD, glucose-6-phosphate dehydrogenase; OR, odds ratio.

) 6.31 (1.78–22.4) .004 (  …  ) Species of infection P. vivax monoinfection (  …  ) (  …  ) (  …  ) (  …  ) 1.00 Mixed P. vivax / P. falciparum (  …  ) (  …  ) (  …  ) (  …  ) 1.03 (.86–1.24) .74 Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; G6PD, glucose-6-phosphate dehydrogenase; OR, odds ratio. Gametocyte Clearance Overall, 42.5% (207/487) of patients in Thailand had cleared their asexual parasitemia by day 1 vs 90.7% (262/289) of patients with P. vivax monoinfections in Indonesia (P < .001). Of those with gametocytemia on enrollment, 58.4% (240/411) had cleared their gametocytemia by day 1 in Thailand vs 96.4% (270/280) in Indonesia (P < .001). If gametocytemia had been established against 200 rather than 500 white cells in Thailand, an estimated 78 additional patients with gametocytemia on day 1 would have been classified as agametocytemic and the proportion of patients who had cleared gametocytemia by day 1 would have been 77.4% as opposed to 58.4%. The proportions of patients who had cleared their gametocytemia by day 1 for the individual drugs in Thailand were 73.4% (152/207) after DHA + PIP vs 43.1% (88/204) after CQ, P < .001. By day 2, 93.5% (245/262) of patients had cleared their gametocytes in Thailand and by this time there was no difference between treatment arms. In Indonesia, the proportions of patients who had cleared their gametocytemia by day 1 were 90.2% (37/41) following AM + LUM, 98.1% (101/103) following DHA + PIP, and 93.9% (46/49) following AS + AQ (P = .03 for AM + LUM vs DHA + PIP; P = .52 for AM + LUM vs AS + AQ; and P = .18 for DHA + PIP vs AS + AQ; Figure 2). No individuals at either site had persistent P. vivax gametocytes at day 7. In Thailand, 22.1% (17/77) of individuals without gametocytemia on admission developed P. vivax gametocytemia between day 1 and day 4; the risk being nonsignificantly greater following treatment with chloroquine (30.8% 12/39) than with DHA + PIP (13.2% 5/38); P = .06. In Indonesia, no patients (0/103) without gametocytemia on enrollment subsequently developed gametocytemia between day 1 and day 4 (P < .001). Figure 2. Proportion of individuals examined with sexual and/or asexual forms of Plasmodium vivax from presentation through to end of follow-up in Thailand and Indonesia (excludes patients with mixed infection on presentation in Indonesia).

llment subsequently developed gametocytemia between day 1 and day 4 (P < .001). Figure 2. Proportion of individuals examined with sexual and/or asexual forms of Plasmodium vivax from presentation through to end of follow-up in Thailand and Indonesia (excludes patients with mixed infection on presentation in Indonesia). Abbreviations: AM + LUM, artemether + lumefantrine; AS + AQ, artesunate + amodiaquine; CQ, chloroquine; DHA + PIP, monotherapy dihydroartemisinin + piperaquine. Gametocytemia During Follow-up Overall, 146 of 492 (29.7%) participants had appearance of P. vivax gametocytemia between day 7 and day 63 in Thailand (67 [13.6%] of whom failed by day 42) and 28 of 314 (8.92%) participants with P. vivax monoinfections had appearance of P. vivax gametocytemia between day 7 and day 42 in Indonesia (see Table 3). Of the 174 appearances of gametocytemia during follow-up, only 2 (1.15%) were not associated with concurrent asexual-stage infection; both individuals had been treated with AS + AQ. In Thailand, 54.2% (147/271) of patients had patent gametocytemia at the time of P. vivax asexual recurrence compared with 33.8% (26/77) following P. vivax monoinfection in Indonesia (P = .002). Table 3. Cumulative Percentage Gametocyte Carriage by Treatment

ge infection; both individuals had been treated with AS + AQ. In Thailand, 54.2% (147/271) of patients had patent gametocytemia at the time of P. vivax asexual recurrence compared with 33.8% (26/77) following P. vivax monoinfection in Indonesia (P = .002). Table 3. Cumulative Percentage Gametocyte Carriage by Treatment Cumulative Percentage Gametocyte Carriage (95% confidence interval) P Value AM + LUM DHA + PIP AS + AQ CQ All AM + LUM v DHA + PIP AM + LUM v AS + AQ DHA + PIP v AS + AQ DHA + PIP v CQ Day 7–42 Thailand (  …  ) 6.92 (4.23–11.2) (  …  ) 29.1 (23.0–36.5) 16.9 (13.5–21.0) (  …  ) (  …  ) (  …  ) <.001 Indonesia (pure) 7.42 (3.14–17.0) 6.80 (3.46–13.2) 33.6 (21.6–49.8) (  …  ) 12.1 (8.50–17.2) .39 <.001 <.001 (  …  ) Indonesia (mixed) 17.5 (9.12–32.1) 4.76 (1.54–14.2) 34.7 (14.4–68.7) (  …  ) 12.3 (7.55–19.7) .01 .39 .001 (  …  ) Day 7–63 Thailand (  …  ) 32.9 (26.5–40.4) (  …  ) 57.9 (49.4–66.6) 43.7 (38.4–49.4) (  …  ) (  …  ) (  …  ) <.001 Abbreviations: AM + LUM, artemether + lumefantrine; AS + AQ, artesunate + amodiaquine; CQ, chloroquine; DHA + PIP, dihydroartemisinin + piperaquine.

14.4–68.7) (  …  ) 12.3 (7.55–19.7) .01 .39 .001 (  …  ) Day 7–63 Thailand (  …  ) 32.9 (26.5–40.4) (  …  ) 57.9 (49.4–66.6) 43.7 (38.4–49.4) (  …  ) (  …  ) (  …  ) <.001 Abbreviations: AM + LUM, artemether + lumefantrine; AS + AQ, artesunate + amodiaquine; CQ, chloroquine; DHA + PIP, dihydroartemisinin + piperaquine. In Thailand, the day 42 cumulative risk of gametocyte carriage was lower following DHA + PIP (6.92%; 95% CI, 4.23%–11.2%) than following CQ (29.1%; 95% CI, 23.0%–36.5%; P < .001). The cumulative risk of gametocyte carriage by day 42 following P. vivax monoinfections in Indonesia was greatest for AS + AQ (33.6%; 95% CI, 21.6%–49.8%) and lowest for DHA + PIP (6.80%; 95% CI, 3.46%–13.2%; P < .001; Table 3). There was no difference in the day 42 cumulative risk of gametocytemia following DHA + PIP between the Thai and Indonesian studies (Table 3).

42 following P. vivax monoinfections in Indonesia was greatest for AS + AQ (33.6%; 95% CI, 21.6%–49.8%) and lowest for DHA + PIP (6.80%; 95% CI, 3.46%–13.2%; P < .001; Table 3). There was no difference in the day 42 cumulative risk of gametocytemia following DHA + PIP between the Thai and Indonesian studies (Table 3). In univariable models, risk factors for appearance of gametocytes during follow-up included higher initial asexual parasite density in both locations and presence of gametocytemia on enrollment, as well as persistence of asexual parasitemia on day 1 in Thailand (Table 3). Persistent asexual parasitemia on day 2 was rare in Indonesia; in Thailand, it was not associated with recurrent gametocytemia in a univariable model (hazard ratio = 1.47; 95% CI, .87–2.49; P = .15). After adjusting for confounding factors, higher asexual parasite density on enrollment was associated with a greater chance of recurrent gametocytemia in both Thailand and Indonesia (adjusted hazard ratio [AHR] = 1.18; 95% CI, 1.02–1.35; P = .02 in Thailand and AHR = 1.58; 95% CI, 1.25–1.98; P < .001 in Indonesia; Table 4). Table 4. Risk Factors for Gametocytemia During Follow-Up

te density on enrollment was associated with a greater chance of recurrent gametocytemia in both Thailand and Indonesia (adjusted hazard ratio [AHR] = 1.18; 95% CI, 1.02–1.35; P = .02 in Thailand and AHR = 1.58; 95% CI, 1.25–1.98; P < .001 in Indonesia; Table 4). Table 4. Risk Factors for Gametocytemia During Follow-Up Risk Factor Univariable Models Multivariable Models Thailand, day 7–63 Indonesia (monoinfection), day 7–42 Indonesia (mixed infection), day 7–42 Thailand, day 7–63 Indonesia, day 7–42 HR (95% CI) P Value HR (95% CI) P Value HR (95% CI) P Value AHR (95% CI) P Value AHR (95% CI) P Value Female sex 0.93 (.66–1.31) .69 2.21 (1.02–4.78) .05 1.16 (.41–3.27) .77 0.91 (.62–1.32) .62 1.62 (.83–3.18) .16 Age <5 y 1.53 (.98–2.39) .06 2.28 (.96–5.40) .06 0.93 (.20–4.36) .92 1.54 (.95–2.49) .08 1.41 (.58–3.42) .45 5 to <15 y 0.78 (.53–1.16) .23 1.39 (.52–3.74) .51 1.29 (.42–3.95) .65 0.89 (.58–1.35) .57 1.92 (.82–4.52) .13 ≥15 y 1.00 1.00 1.00 1.00 1.00 G6PD status Normal 1.00 1.00 a 1.00 1.00 Abnormal 1.05 (.49–2.25) .90 1.56 (.66–3.67) .31 a 0.95 (.43–2.09) .90 1.84 (.74–4.53) .19 Enrollment loge asexual parasite density (per loge order increase) 1.21 (1.08–1.36) .001 1.48 (1.14–1.91) .003 1.72 (1.30–2.27) <.001 1.18 (1.02–1.35) .02 1.58 (1.25–1.98) <.001 Gametocytes on enrollment 1.82 (1.07–3.11) .03 2.15 (.87–5.31) .10 2.22 (.71–6.98) .17 1.22 (.68–2.22) .50 1.31 (.54–3.16) .55 Persistent asexual parasitemia on day 1 1.45 (1.04–2.03) .03 2.12 (.73–6.17) .17 1.78 (.39–8.01) .46 0.84 (.57–1.25) .39 1.13 (.42–3.01) .81 Anemia on enrollment (Hb <9 g/dL) 0.54 (.08–3.86) .54 1.46 (.68–3.17) .33 1.68 (.59–4.81) .33 0.85 (.11–6.45) .88 1.26 (.60–2.62) .55 Fever on enrollment (>37.5°C) 1.33 (.94–1.87) .11 0.40 (.09–1.67) .21 1.12 (.40–3.15) .83 1.22 (.84–1.77) .29 0.45 (.17–1.20) .11 Schizonts on admission blood film 1.23 (.87–1.73) .24 (  …  ) (  …  ) 1.05 (.70–1.57) .81 (  …  ) Species at enrollment Plasmodium vivax monoinfection (  …  ) (  …  ) (  …  ) (  …  ) 1.00 Mixed P. vivax / Plasmodium falciparum (  …  ) (  …  ) (  …  ) (  …  ) 2.76 (1.26–6.04) .01 Abbreviations: AHR, adjusted hazard ratio; CI, confidence interval; G6PD, glucose-6-phosphate dehydrogenase; HR, hazard ratio.

.81 (  …  ) Species at enrollment Plasmodium vivax monoinfection (  …  ) (  …  ) (  …  ) (  …  ) 1.00 Mixed P. vivax / Plasmodium falciparum (  …  ) (  …  ) (  …  ) (  …  ) 2.76 (1.26–6.04) .01 Abbreviations: AHR, adjusted hazard ratio; CI, confidence interval; G6PD, glucose-6-phosphate dehydrogenase; HR, hazard ratio. Multivariable models stratified by treatment group. a No patients with mixed infection and an abnormal G6PD status had a recurrence of P. vivax gametocytemia between 7 and 42 days. The relationship between loge transformed asexual- and sexual-stage density at enrollment and during follow-up is presented in Figure 3. In Thailand, the median ratio of gametocytes to asexuals did not differ between enrollment and recurrence (median ratio at enrollment = 0.04, interquartile range [IQR] = 0.009–0.08; median ratio at recurrence = 0.01, IQR = 0–0.19, P = .49). The same was true in Indonesia (median ratio at enrollment = 0.015, IQR = 0–0.076; median ratio at recurrence = 0, IQR = 0–0.03, P = .08). Figure 3. Correlation between the loge density of asexual and sexual stages of Plasmodium vivax at presentation for treatment and at the time of recurrence after treatment (analyses limited to those with P. vivax monoinfections at enrollment).

.015, IQR = 0–0.076; median ratio at recurrence = 0, IQR = 0–0.03, P = .08). Figure 3. Correlation between the loge density of asexual and sexual stages of Plasmodium vivax at presentation for treatment and at the time of recurrence after treatment (analyses limited to those with P. vivax monoinfections at enrollment). Mixed P. vivax/P. falciparum Infections In Indonesia, patients with mixed infections on enrollment were less likely to have patent P. vivax gametocytemia than patients with P. vivax monoinfection (56.8% [92/162] vs 66.6% [209/314], P = .04). However, those with mixed infection were at significantly greater risk of recurrent gametocytemia between day 7 and day 42 compared with patients with P. vivax monoinfection (AHR = 2.76; 95% CI, 1.26–6.04; P = .01).

ax gametocytemia than patients with P. vivax monoinfection (56.8% [92/162] vs 66.6% [209/314], P = .04). However, those with mixed infection were at significantly greater risk of recurrent gametocytemia between day 7 and day 42 compared with patients with P. vivax monoinfection (AHR = 2.76; 95% CI, 1.26–6.04; P = .01). DISCUSSION Our analysis of 3 large clinical drug trials from Thailand and Indonesia highlights several fundamental properties of P. vivax transmission dynamics, some of which have been given little consideration since early studies of neurosyphilitics and military personnel in the first half of the 20th century [13–21]. First, patent P. vivax gametocytemia is present in the majority of patients by the time they seek treatment. Second, P. vivax gametocytes do not persist after asexual parasite clearance (no patient in either country had persistent gametocytemia at day 7). Third, the relationship between asexual- and sexual-stage parasitemia does not differ substantially between initial and recurrent infections. Fourth, there are significant differences in the effects of artemisinin combination regimens on the risk of recurrent parasitemia and therefore the short-term transmissibility of P. vivax infections.

ip between asexual- and sexual-stage parasitemia does not differ substantially between initial and recurrent infections. Fourth, there are significant differences in the effects of artemisinin combination regimens on the risk of recurrent parasitemia and therefore the short-term transmissibility of P. vivax infections. In symptomatic falciparum malaria, patent gametocytemia occurs after the onset of symptoms and usually during convalescence [22, 23]. Rapidly effective blood schizontocidal drugs can therefore have a profound impact on overall gametocyte carriage and transmission potential [24]. The artemisinin derivatives are highly potent antimalarials that reduce the biomass of asexual parasites rapidly while also exerting strong gametocytocidal activity against early-stage sexual forms [24–28]. When combined with a slowly eliminated partner drug, the artemisinin derivatives minimize the risk of recrudescence and reduce P. falciparum transmissibility [29].

malarials that reduce the biomass of asexual parasites rapidly while also exerting strong gametocytocidal activity against early-stage sexual forms [24–28]. When combined with a slowly eliminated partner drug, the artemisinin derivatives minimize the risk of recrudescence and reduce P. falciparum transmissibility [29]. The dynamics of gametocyte carriage in vivax malaria are notably different [30]. Sexual stages appear early in the course of infection [13–16, 19, 23, 31] together with the rise in asexual parasitemia; thus, transmission often occurs before antimalarial treatment. Unlike P. falciparum gametocytes, P. vivax sexual forms are susceptible to all blood schizontocidal medications [32]. The relative transmission-blocking benefit of drugs with greater potency, such as the artemisinin derivatives that reduce P. vivax parasitemia more rapidly than others, is likely to be minimal. There are 2 reasons for this. First, gametocytes are often present for several days before presentation and are therefore likely to be transmitted prior to treatment [33, 34]. Assuming complete parasitological cure, administration of highly potent artemisinin-based therapy instead of chloroquine will truncate gametocyte carriage associated with the initial episode by at best 24–48 hours. Second, P. vivax infection is associated with multiple relapses, each associated with gametocytemia and thus transmissible for several days prior to clinical detection. Preventing recurrence, in particular due to relapse, is thus more important for reducing transmission of vivax malaria than rapid removal of gametocytes at each clinical presentation.

sociated with multiple relapses, each associated with gametocytemia and thus transmissible for several days prior to clinical detection. Preventing recurrence, in particular due to relapse, is thus more important for reducing transmission of vivax malaria than rapid removal of gametocytes at each clinical presentation. The antimalarial regimen with the greatest potential to block transmission of P. vivax will include a highly active blood schizontocidal regimen that completely eradicates blood stages and thus prevents recrudescence in combination with a hypnozoitocidal medication for preventing future relapses. Unfortunately, toxicity concerns and poor adherence to 2-week regimens continue to hamper the safe and effective use of primaquine, the only currently licensed hypnozoitocidal drug [35]. Where primaquine is not used or has been shown to be ineffective, slowly eliminated blood schizontocides that suppress the first relapse may have benefits over regimens with shorter elimination half-lives [36], though whether they reduce the total number of relapses and overall transmission potential is unknown.

[35]. Where primaquine is not used or has been shown to be ineffective, slowly eliminated blood schizontocides that suppress the first relapse may have benefits over regimens with shorter elimination half-lives [36], though whether they reduce the total number of relapses and overall transmission potential is unknown. AS + AQ has consistently been associated with higher P. falciparum recrudescence and, as shown in this analysis, higher P. vivax recurrence rates than either AM + LUM or DHA + PIP [6, 37]. This is likely to be attributable to the relatively short elimination half-life of amodiaquine and declining parasite susceptibility to this drug [6]. Gametocyte carriage was higher following AM + LUM (half-life approximately 4 days) than DHP + PIP, although this only reached significance in patients treated for mixed infections (Table 3). Chloroquine is potent against susceptible P. vivax strains and has an elimination half-life of 1–2 months [38]. It therefore has the potential to limit recrudescence and suppress the first and possibly even second P. vivax relapse. In Thailand, chloroquine was associated with greater gametocyte carriage during follow-up compared with DHA + PIP (elimination half-life approximately 28 days), suggesting declining chloroquine susceptibility of local strains. This scenario is likely to be mirrored in other regions where chloroquine has been used as the mainstay of vivax malaria treatment for many years [36].

ocyte carriage during follow-up compared with DHA + PIP (elimination half-life approximately 28 days), suggesting declining chloroquine susceptibility of local strains. This scenario is likely to be mirrored in other regions where chloroquine has been used as the mainstay of vivax malaria treatment for many years [36]. High asexual parasite density was shown to be a strong risk factor for gametocyte carriage during follow-up, independent of age and other potential confounders. There are 2 likely explanations for this finding. First, high asexual parasitemia is associated with an increased risk of parasite recrudescence (as shown in falciparum malaria) [29, 39–41]. Second, high parasite density reflects poor immunity, which has been associated with a greater risk of patent relapse [42]. Since P. vivax gametocytemia mirrors asexual infection, a higher risk of recurrent asexual infection, whether due to recrudescence or relapse, will result in a higher risk of gametocyte carriage. In other words, the factors that determine P. vivax transmissibility are those that determine asexual-stage parasite dynamics.

nce P. vivax gametocytemia mirrors asexual infection, a higher risk of recurrent asexual infection, whether due to recrudescence or relapse, will result in a higher risk of gametocyte carriage. In other words, the factors that determine P. vivax transmissibility are those that determine asexual-stage parasite dynamics. Patients with mixed P. vivax/P. falciparum infections in Indonesia were at greater risk of recurrent gametocytemia than those with vivax monoinfections. This contrasts with mixed species asexual infections in Thailand, which have been associated with a lower risk of P. falciparum gametocytemia [29]. In Indonesia, mixed infections are more severe than monoinfections with either species [3]. Malarial illness has been hypothesized to precipitate P. vivax relapses [9, 43, 44]. Therefore, the increased risk of recurrent gametocytemia following mixed infection may relate to greater pathophysiological derangement and hence greater stimulation of dormant liver-stage parasites. Alternatively, mixed infection in Indonesia may reflect poor immunity, which in turn is associated with a greater risk of relapse. As we made multiple comparisons, the possibility of a chance finding must also be considered.

pathophysiological derangement and hence greater stimulation of dormant liver-stage parasites. Alternatively, mixed infection in Indonesia may reflect poor immunity, which in turn is associated with a greater risk of relapse. As we made multiple comparisons, the possibility of a chance finding must also be considered. Our analysis has limitations. Follow-up was 3 weeks longer in Thailand than in Indonesia and thus conclusions drawn for the 42- to 63-day periods were based on Thai data only. A 42-day follow-up is insufficient to capture first relapses that follow administration of slowly eliminated antimalarial drugs. Parasite counts were done against 200 WBCs in Indonesia whereas in Thailand they were done against 500 WBCs. This will have increased the likelihood of gametocyte detection in Thailand relative to Indonesia and may partially explain the shorter gametocyte clearance times in Indonesia. Preexisting immunity to P. vivax is likely to have been greater in Indonesia than in Thailand due to more intense parasite exposure. This may have contributed to the slower gametocyte clearance times in Thailand. Unsupervised primaquine was prescribed for individuals with normal G6PD activity at day 28 in the first Indonesian study and at day 3 in the second Indonesian study. Exploratory analyses revealed that this difference did not have any substantial effect on subsequent gametocyte carriage. Residual minor effects will have been controlled for by inclusion of G6PD status in the multivariable models.

ity at day 28 in the first Indonesian study and at day 3 in the second Indonesian study. Exploratory analyses revealed that this difference did not have any substantial effect on subsequent gametocyte carriage. Residual minor effects will have been controlled for by inclusion of G6PD status in the multivariable models. Both P. vivax and P. falciparum can be transmitted at subpatent gametocyte densities [19, 45–50]. Thus, microscopic quantification may have underestimated the total period of infectiousness following therapy. Without treatment, P. vivax gametocytes are reported to persist in the peripheral circulation for a maximum of 3 days [48]. Since the study drugs reduced parasitemia rapidly, any persisting period of infectiousness from subpatent gametocytemia will be short-lived in the absence of recrudescence. In conclusion, we have shown that P. vivax gametocytemia closely mirrors asexual-stage carriage. Persistence of patent gametocytemia following eradication of asexual stages does not occur. Our results indicate that the most important means of blocking P. vivax transmission is likely to be prevention of future relapses, especially in patients with high asexual parasite density and mixed infections. Optimal prescribing practices that maximize patient adherence to primaquine are needed and, given the limitations of this drug, very high priority must be given to the development of novel antirelapse strategies.

vention of future relapses, especially in patients with high asexual parasite density and mixed infections. Optimal prescribing practices that maximize patient adherence to primaquine are needed and, given the limitations of this drug, very high priority must be given to the development of novel antirelapse strategies. Notes Acknowledgments. We thank the staff of the Shoklo Malaria Research Unit and the Timika Research Facility for their work and we thank all the patients who participated in the studies. We also acknowledge the Indonesian National Institute of Health Research and Development and Menzies School of Health Research for their support of the original trial in Timika. Financial support. N. M. D. is funded by the Rhodes Trust. R. N. P. is a Wellcome Trust Senior Research Fellow in Clinical Science (091625). F. N. and N. J. W. are supported by the Wellcome Trust. N. A. is a National Health and Medical Research Council (NHMRC) Practitioner Fellow. The clinical studies in Indonesia were funded by the Wellcome Trust and NHRMC (Wellcome Trust International Collaborative Research Grant GR071614MA-NHMRC ICRG ID 283321). The research carried out at the Shoklo Malaria Research Unit is part of the Wellcome Trust Mahidol University of Oxford Tropical Medicine Research Programme supported by the Wellcome Trust of Great Britain. The Thai trial was supported by Holleypharm (R. P. C.). The Timika Translational Research Facility is supported by AusAID. Potential conflicts of interest. All authors: No reported conflicts.

Financial support. N. M. D. is funded by the Rhodes Trust. R. N. P. is a Wellcome Trust Senior Research Fellow in Clinical Science (091625). F. N. and N. J. W. are supported by the Wellcome Trust. N. A. is a National Health and Medical Research Council (NHMRC) Practitioner Fellow. The clinical studies in Indonesia were funded by the Wellcome Trust and NHRMC (Wellcome Trust International Collaborative Research Grant GR071614MA-NHMRC ICRG ID 283321). The research carried out at the Shoklo Malaria Research Unit is part of the Wellcome Trust Mahidol University of Oxford Tropical Medicine Research Programme supported by the Wellcome Trust of Great Britain. The Thai trial was supported by Holleypharm (R. P. C.). The Timika Translational Research Facility is supported by AusAID. Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Disease caused by Mycobacterium tuberculosis continues to be a major global health problem. In 2011, 8.7 million new cases of tuberculosis were diagnosed worldwide and 1.4 million people died from the disease [1]. With tuberculosis causing a quarter of the deaths in people living with human immunodeficiency virus (HIV) [1] and with the emergence of increasingly drug-resistant strains of M. tuberculosis, an effective vaccine is urgently needed now in order to reduce the burden of this disease.

on people died from the disease [1]. With tuberculosis causing a quarter of the deaths in people living with human immunodeficiency virus (HIV) [1] and with the emergence of increasingly drug-resistant strains of M. tuberculosis, an effective vaccine is urgently needed now in order to reduce the burden of this disease. Since 2002, more than a dozen candidate vaccines have been entered into clinical testing [2]. However, it is difficult to determine which of these candidates will progress from relatively small-scale safety and immunogenicity studies through to large-scale, expensive efficacy trials because an immune correlate of vaccine-induced protection against infection or disease does not exist. Preclinical animal challenge models of M. tuberculosis infection [3–6] and in vitro mycobacterial killing assays [7–9] are used to assess vaccine efficacy. However, it is not clear whether either of these reliably predict what occurs in vivo in humans. Thus, the evaluation of vaccine efficacy currently relies on large, expensive, and time-consuming efficacy trials. A human mycobacterial challenge model that could be used to assess the efficacy of candidate tuberculosis vaccines at an early stage would be a great advancement to the field. Human challenge models are routinely used in vaccine development for pathogens such as malaria, influenza, dengue fever, and typhoid [10–13]; however, the deliberate infection of humans with M. tuberculosis would not be ethically acceptable. Previously, we demonstrated that a novel human challenge model that uses bacille Calmette-Guérin (BCG) as a surrogate for M. tuberculosis infection can detect differences in antimycobacterial immunity induced by previous BCG vaccination [14]. In this earlier trial, healthy volunteers were challenged with intradermal BCG. The BCG load was then quantified from a skin biopsy at the challenge site at 1, 2, and 4 weeks post challenge by culture on solid agar and quantitative polymerase chain reaction (qPCR). It was found that optimum recovery of BCG was achieved in 2 weeks; this time period was chosen for future challenge studies. Here we use this BCG challenge model to evaluate the reduction in mycobacterial load induced by BCG alone; a candidate tuberculosis vaccine, MVA85A; and a BCG prime–MVA85A boost vaccine regimen.

It was found that optimum recovery of BCG was achieved in 2 weeks; this time period was chosen for future challenge studies. Here we use this BCG challenge model to evaluate the reduction in mycobacterial load induced by BCG alone; a candidate tuberculosis vaccine, MVA85A; and a BCG prime–MVA85A boost vaccine regimen. METHODS Trial Design This phase 1 trial (ClinicalTrial.gov registry NCT01194180) was approved by the Medicines and Healthcare Products Regulatory Agency (EudraCT 2010-018425-19) and the Oxfordshire Research Ethics Committee A (reference 10/H0505/31). Twenty-six BCG-vaccinated and 23 BCG-naive healthy volunteers aged 18–55 years were enrolled between March 2011 and November 2011 at the Centre for Clinical Vaccinology & Tropical Medicine, Churchill Hospital, Oxford, United Kingdom (see Figure 1). All participants gave written informed consent, and the trial was conducted according to the principles of the Declaration of Helsinki and Good Clinical Practice. Figure 1. Consort diagram showing participant recruitment and follow-up. *One volunteer withdrew from group B for personal reasons after MVA85A vaccination but before bacille Calmette-Guérin challenge.

nt, and the trial was conducted according to the principles of the Declaration of Helsinki and Good Clinical Practice. Figure 1. Consort diagram showing participant recruitment and follow-up. *One volunteer withdrew from group B for personal reasons after MVA85A vaccination but before bacille Calmette-Guérin challenge. Those enrolled were in good health, had normal baseline hematology and biochemistry, and were serologically negative for hepatitis B, hepatitis C, and HIV. Latent infection with M. tuberculosis was excluded by a negative ex vivo enzyme-linked immunosorbent spot (ELISpot) assay response to ESAT-6 and CFP-10 peptides. The reasons for exclusion of 9 participants are shown in Figure 1. One individual in group B had to withdraw from follow-up for personal reasons after receiving MVA85A but before BCG challenge and was therefore replaced. Treatment Groups Participants were assigned to group A (BCG-naive; no vaccine received), group B (BCG-naive at baseline; received intradermal MVA85A, dose 1 × 108 pfu), group C (BCG-vaccinated at baseline; median time since vaccination 10 years), or group D (BCG-vaccinated at baseline; median time since vaccination 10.5 years; received intradermal MVA85A, dose 1 × 108 pfu) based on their prior BCG vaccination status and meeting inclusion criteria. One volunteer who was enrolled into group A on the basis of negative BCG status was later reassigned to group C after discovering that he/she had, in fact, received BCG as an infant (Figure 1).

rs; received intradermal MVA85A, dose 1 × 108 pfu) based on their prior BCG vaccination status and meeting inclusion criteria. One volunteer who was enrolled into group A on the basis of negative BCG status was later reassigned to group C after discovering that he/she had, in fact, received BCG as an infant (Figure 1). Vaccine Clinical-grade MVA85A was constructed as previously described [15] and produced following good manufacturing practices by IDT Biologika GmbH (Dessau-Rosslau, Germany). Challenge All participants were challenged with a standard vaccine dose of intradermal BCG (SSI (Statens Serum Institut); 0.1 mL containing 2 to 8 × 105 CFU). Those in groups B and D were challenged 4 weeks after MVA85A vaccination. To minimize variation between BCG vials, as many volunteers as possible were challenged from the same vial of BCG within 2 hours of reconstitution (15 different BCG vaccine vials were used over the course of the trial to challenge 48 volunteers). The challenge dose was verified by plating serial dilutions of a 100-µL aliquot onto solid Middlebrook 7H10 agar (Sigma). Skin Biopsies Skin biopsies were performed on the BCG challenge site of all 48 volunteers by a single operator 2 weeks post challenge as previously described [14]. The 4-mm punch biopsy specimen was taken from the center of the BCG vaccination site, transferred to a sterile Cryovial, snap frozen on dry ice, and stored in liquid nitrogen until the day of processing.

he BCG challenge site of all 48 volunteers by a single operator 2 weeks post challenge as previously described [14]. The 4-mm punch biopsy specimen was taken from the center of the BCG vaccination site, transferred to a sterile Cryovial, snap frozen on dry ice, and stored in liquid nitrogen until the day of processing. Biopsy Homogenization and Culture All 48 biopsies were processed on the same day. Samples were thawed in a 37°C water bath and transferred to a Dispomix tube (Miltenyl Biotech) that contained 1 mL sterile phosphate-buffered saline (PBS). Tubes were loaded onto a Dispomix machine (Thistle Scientific) and homogenized as previously described [16]. Next, 100 µL of neat homogenate and 100 µL of a 10−1 and 10−2 dilution were plated in triplicate onto Middlebrook 7H10 agar and incubated at 37°C for 5 weeks. A BCG SSI vaccine vial was reconstituted in PBS and 100 µL of a 10−2, 10−3, and 10−4 dilution were plated in triplicate as positive controls. The remaining biopsy homogenate was stored at −20°C for later DNA extraction. DNA Extraction Homogenate was thawed and BCG DNA from 200 µL homogenate was released using the tough microorganism lysing kit (Precellys) in a Precellys 24 machine at 6500 rpm for 3 × 30 seconds. Homogenate was transferred to a separate tube, and 50 µL PBS was used to wash the remaining homogenate from the beads. Next, 180 µL animal tissue lysis buffer and 20 µL proteinase K (Qiagen) were added, vortexed, and incubated at 56°C for 4 hours. From this point, the extractions were carried out as previously described [16].

ate was transferred to a separate tube, and 50 µL PBS was used to wash the remaining homogenate from the beads. Next, 180 µL animal tissue lysis buffer and 20 µL proteinase K (Qiagen) were added, vortexed, and incubated at 56°C for 4 hours. From this point, the extractions were carried out as previously described [16]. qPCR Primers ET 1 and ET 3 were used for detection of BCG DNA. These are complementary to regions that flank the BCG deletion RD1 sequence and amplify a 196-bp fragment [17]. These sequences were modified by Minassian et al [16], and the modified sequences were used for this work (Table 1). PCR reactions were carried out as previously described [16] using BCG-naive macaque tissue homogenate as a negative control. A standard curve was obtained by extracting BCG DNA from 1 in 10 serial dilutions of 5 pooled vaccine vials in PBS and correcting for live BCG from the corresponding colony-forming unit counts on solid agar. Table 1. Primer Sequences Used to Detect Bacille Calmette-Guérin by Quantitative Polymerase Chain Reaction Primer Primer Sequence ET 1/3 forward 5′ -CCG CCG ACC GAC CTG ACG AC- 3′ ET 1/3 reverse 5′ -GGC GAT CTG GCG GTT TGG GG- 3′

qPCR Primers ET 1 and ET 3 were used for detection of BCG DNA. These are complementary to regions that flank the BCG deletion RD1 sequence and amplify a 196-bp fragment [17]. These sequences were modified by Minassian et al [16], and the modified sequences were used for this work (Table 1). PCR reactions were carried out as previously described [16] using BCG-naive macaque tissue homogenate as a negative control. A standard curve was obtained by extracting BCG DNA from 1 in 10 serial dilutions of 5 pooled vaccine vials in PBS and correcting for live BCG from the corresponding colony-forming unit counts on solid agar. Table 1. Primer Sequences Used to Detect Bacille Calmette-Guérin by Quantitative Polymerase Chain Reaction Primer Primer Sequence ET 1/3 forward 5′ -CCG CCG ACC GAC CTG ACG AC- 3′ ET 1/3 reverse 5′ -GGC GAT CTG GCG GTT TGG GG- 3′ Ex vivo Interferon-gamma ELISpot Assay Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood and ELISpots were performed as previously described [18]. This occurred for all volunteers on the day of screening, the day of BCG challenge, and the day of skin biopsy. Groups B and D also had an ELISpot performed 7 days after MVA85A vaccination, which occurred 21 days prechallenge. Responses to purified protein derivative (PPD) from M. tuberculosis (SSI; 20 µg/mL) and a single pool of 66 Ag85A peptides (Peptide Protein Research; 2 µg/mL each peptide) were assessed for all volunteers at each time point. Staphylococcal enterotoxin B (Sigma) was used as a positive control (10 µg/mL). Unstimulated PBMCs were used as a measure of background interferon-gamma (IFN-γ) production. Results are reported as spot-forming cells (SFC) per million PBMC, calculated by subtracting the mean count of the unstimulated PBMCs from the mean count of duplicate antigen wells and correcting for the number of PBMCs in the well.

mulated PBMCs were used as a measure of background interferon-gamma (IFN-γ) production. Results are reported as spot-forming cells (SFC) per million PBMC, calculated by subtracting the mean count of the unstimulated PBMCs from the mean count of duplicate antigen wells and correcting for the number of PBMCs in the well. Whole Blood Growth Inhibition Assay The whole blood growth inhibition assay was performed on heparinized whole blood on the day of BCG challenge using the BACTEC mycobacteria growth indicator tube (MGIT) system (Becton Dickinson) as previously described [19], with the exception that whole blood was incubated with BCG (Pasteur) for 96 hours instead of 72 hours. Growth inhibition was determined by calculating time to positivity (TTP) in the sample and TTP in the control and converting to colony-forming units using a standard curve. The growth ratio (GR) was calculated by GR = CFU sample (96 hours)/CFU control (0 hours).

ated with BCG (Pasteur) for 96 hours instead of 72 hours. Growth inhibition was determined by calculating time to positivity (TTP) in the sample and TTP in the control and converting to colony-forming units using a standard curve. The growth ratio (GR) was calculated by GR = CFU sample (96 hours)/CFU control (0 hours). Swabbing of BCG Vaccination Site The feasibility of swab-based quantification of BCG from the surface of the BCG challenge vaccination site, as an alternative or complementary technique to the biopsy technique, was investigated in a separate cohort of healthy volunteers (study approved by the University of Oxford Central University Research Ethics Committee, reference MSD/IDREC/C1/2012/7). Seven BCG-naive, adult healthcare workers who were due to receive BCG vaccination for employment reasons were recruited and gave written informed consent. Cotton-tipped swabs (Transwab MW171, Medical Wire & Equipment) were used to swab the surface of the BCG site at 2, 7, 14, and 21 days following vaccination. The ability of this method to recover BCG from the swabs was tested before the study began. Serial dilutions of a BCG vaccine vial (SSI) were made and 100 µL of each dilution was plated onto Middlebrook 7H10 agar. Twenty microliters of each dilution was used to spike the swabs, which were then immersed in 500 µL Middlebrook 7H9 broth (Sigma), left for 1 hour, and sonicated for 30 seconds. 300 µL of broth was then plated onto Middlebrook 7H10 agar and incubated at 37°C for 3 weeks. The number of colonies recovered was compared with the inoculum in 20 µL. Swabs from the BCG vaccination site were processed in the same way.

500 µL Middlebrook 7H9 broth (Sigma), left for 1 hour, and sonicated for 30 seconds. 300 µL of broth was then plated onto Middlebrook 7H10 agar and incubated at 37°C for 3 weeks. The number of colonies recovered was compared with the inoculum in 20 µL. Swabs from the BCG vaccination site were processed in the same way. Statistical Analysis Statistical analyses were performed using GraphPad Prism. One-way analysis of variance (Kruskal-Wallis) and Mann–Whitney U tests were used to determine significant differences between groups. The Wilcoxon matched pairs test was used to determine differences between time points in the same group. The Spearman rank correlation test was used to determine correlations between numbers of BCG recovered from biopsies and ex vivo IFN-γ ELISpot responses.

ere used to determine significant differences between groups. The Wilcoxon matched pairs test was used to determine differences between time points in the same group. The Spearman rank correlation test was used to determine correlations between numbers of BCG recovered from biopsies and ex vivo IFN-γ ELISpot responses. RESULTS BCG Challenge Was Safe and Well Tolerated by All Groups Other than BCG vaccination status, the volunteers’ baseline characteristics did not significantly differ among the 4 groups (Table 2). BCG challenge was well tolerated, with all volunteers developing an expected local inflammatory reaction to BCG. It was noted that previously BCG-vaccinated volunteers (groups C and D) experienced significantly more frequent local adverse events and significantly greater diameters of erythema and swelling at the challenge vaccination site during the first 2 weeks than did BCG-naive volunteers (groups A and B; data not shown), which is consistent with previous studies [20, 21]. Intradermal administration of candidate vaccine MVA85A at a dose of 1 × 108 pfu was safe and well tolerated, with an adverse event profile consistent with previous experience [18, 22]. No serious adverse events occurred. Vaccination with MVA85A 4 weeks before challenge had no effect on the reactogenicity of the subsequent BCG challenge. Table 2. Demographics of Enrolled Participants

of 1 × 108 pfu was safe and well tolerated, with an adverse event profile consistent with previous experience [18, 22]. No serious adverse events occurred. Vaccination with MVA85A 4 weeks before challenge had no effect on the reactogenicity of the subsequent BCG challenge. Table 2. Demographics of Enrolled Participants Characteristic Group A (n = 11) Group B (n = 12) Group C (n = 13) Group D (n = 12) P Value Prior BCG No No Yes Yes MVA85A No Yes No Yes Female, n (%) 8 (73) 7 (58) 6 (46) 6 (50) 0.58 Median age, years (range) 23 (18–41) 23 (19–30) 23 (21–41) 22 (19–33) 0.63 Median time interval since BCG in years (range) n/a n/a 10 (8–38) 10.5 (6–33) 0.48 Continent of birth Europe 9 8 13 10 Africa 0 0 0 1 Asia 0 0 0 1 Americas 1 3 0 0 Australasia 1 1 0 0 Abbreviations: BCG, bacille Calmette-Guérin; n/a, not applicable.

an age, years (range) 23 (18–41) 23 (19–30) 23 (21–41) 22 (19–33) 0.63 Median time interval since BCG in years (range) n/a n/a 10 (8–38) 10.5 (6–33) 0.48 Continent of birth Europe 9 8 13 10 Africa 0 0 0 1 Asia 0 0 0 1 Americas 1 3 0 0 Australasia 1 1 0 0 Abbreviations: BCG, bacille Calmette-Guérin; n/a, not applicable. BCG Was Detected by Both qPCR and Culture BCG was detected in all 48 biopsy samples by qPCR and in 45 of 48 samples by culture on solid agar (Figure 2). Estimated copy numbers per biopsy using PCR were 1–2 logs higher than the corresponding colony-forming unit counts by culture. A positive correlation was observed between the 2 methods of detection (Spearman R = 0.36, P = .01; Figure 2). Figure 2. Quantification of bacterial load from punch biopsies 14 days post bacille Calmette-Guérin (BCG) challenge by culture on solid agar (A) and quantitative polymerase chain reaction (qPCR) (B). Individual values are shown for each volunteer. Horizontal bars indicate median values in each group. Significant differences between groups are as follows: *P ≤ .05, **P ≤ .01, ***P ≤ .001; Mann–Whitney U test. A significant positive correlation was observed between the culture and qPCR results (C). BCG Challenge Dose Administered Was Comparable for All Volunteers Quantification of BCG from each vaccine vial used in this trial showed that the range in challenge dose was small (1.85 × 105 to 3.15 × 105 cfu, median = 2.35 × 105 cfu) and at the lower end of that stated by the manufacturer [23].

BCG Was Detected by Both qPCR and Culture BCG was detected in all 48 biopsy samples by qPCR and in 45 of 48 samples by culture on solid agar (Figure 2). Estimated copy numbers per biopsy using PCR were 1–2 logs higher than the corresponding colony-forming unit counts by culture. A positive correlation was observed between the 2 methods of detection (Spearman R = 0.36, P = .01; Figure 2). Figure 2. Quantification of bacterial load from punch biopsies 14 days post bacille Calmette-Guérin (BCG) challenge by culture on solid agar (A) and quantitative polymerase chain reaction (qPCR) (B). Individual values are shown for each volunteer. Horizontal bars indicate median values in each group. Significant differences between groups are as follows: *P ≤ .05, **P ≤ .01, ***P ≤ .001; Mann–Whitney U test. A significant positive correlation was observed between the culture and qPCR results (C). BCG Challenge Dose Administered Was Comparable for All Volunteers Quantification of BCG from each vaccine vial used in this trial showed that the range in challenge dose was small (1.85 × 105 to 3.15 × 105 cfu, median = 2.35 × 105 cfu) and at the lower end of that stated by the manufacturer [23]. Levels of BCG Recovered Were Lower in Groups With Previous BCG Vaccination Enumeration of BCG by solid culture showed a trend toward a lower median colony-forming unit count in the previously BCG-vaccinated groups, with a statistically significant 0.5-log reduction in colony-forming unit count between group A (naive) and group D (BCG–MVA85A; P = .02, Mann–Whitney U test; Figure 2). Using PCR, there was a significant 0.5- to 1-log reduction in estimated BCG copy number between the BCG-naive (A or B) and BCG-vaccinated groups (C or D; Mann–Whitney U test). No further reduction in BCG numbers was detected after vaccination with MVA85A.

ein derivative from Mycobacterium tuberculosis for all groups (A) and to a single pool of Ag85A peptides for groups B and D (B). A value of 1667 spot-forming cells (SFCs)/1 × 106 peripheral blood mononuclear cells (PBMCs) represents a blackout in the ELISpot well. ** P < .01. Abbreviation: BCG, bacille Calmette-Guérin. Ex Vivo IFN-γ ELISpot Responses Correlate With Number of BCG Detected The correlation between ex vivo IFN-γ ELISpot responses and BCG colony number detected by PCR is shown in Figure 4. An inverse correlation was observed at all time points for responses to PPD and Ag85A, and this correlation is significant for both antigens 7 days after receipt of MVA85A (P = .005, Spearman) and for PPD 14 days post challenge (P = <.0001, Spearman). The same trend was observed when ELISpot responses were correlated with colony-forming unit counts by culture but did not reach statistical significance at any time point (data not shown). Figure 4. Correlation between ex vivo interferon-gamma enzyme-linked immunosorbent spot assay responses to purified protein derivative or Ag85A and estimated bacille Calmette-Guérin (BCG) copy number by polymerase chain reaction (PCR). Spearman R values are shown with asterisks indicating P values as follows: *P ≤ .05, **P ≤ .01, ***P ≤ .001. Abbreviations: PBMC, peripheral blood mononuclear cell; SFC, spot-forming cell.

to purified protein derivative or Ag85A and estimated bacille Calmette-Guérin (BCG) copy number by polymerase chain reaction (PCR). Spearman R values are shown with asterisks indicating P values as follows: *P ≤ .05, **P ≤ .01, ***P ≤ .001. Abbreviations: PBMC, peripheral blood mononuclear cell; SFC, spot-forming cell. In Vitro Mycobacterial Growth Inhibition Did Not Differ Among Groups on Day of Challenge Figure 5 shows the growth ratios obtained from incubating whole blood taken on the day of challenge with BCG in the MGIT assay. This assay detected no significant differences among the 4 treatment groups in the ability of whole blood to reduce growth of BCG during a 96-hour incubation period (P = .13, Kruskal-Wallis). A nonsignificant positive correlation was observed between growth ratio and BCG copy number by PCR (Figure 5B) and also between growth ratio and colony-forming unit count (data not shown). Figure 5. Growth ratios obtained from the mycobacteria growth indicator tube assay (A) and correlation with estimated bacille Calmette-Guérin (BCG) copy number by quantitative polymerase chain reaction (PCR) (B). BCG Was Not Detected From Swabbing of the BCG Vaccination Site BCG could not be detected by swabbing of the vaccination site in any of the volunteers at any of the time points investigated, by either culture on solid agar or qPCR. This is despite an average recovery of 88% by culture when swabs were spiked with serial dilutions of a BCG vaccine vial.

ng of the BCG Vaccination Site BCG could not be detected by swabbing of the vaccination site in any of the volunteers at any of the time points investigated, by either culture on solid agar or qPCR. This is despite an average recovery of 88% by culture when swabs were spiked with serial dilutions of a BCG vaccine vial. DISCUSSION Here we present a proof-of-concept clinical trial to evaluate a novel BCG challenge model in BCG- and/or MVA85A-vaccinated adults. In this trial, 48 volunteers were challenged with a standard vaccine dose of BCG. A punch biopsy of the vaccination site was taken 14 days later. BCG was detected in all 48 biopsies by qPCR and in 45 of 48 biopsies by culture on solid agar. It has been shown that estimated copy numbers using PCR were 1–2 logs higher than the corresponding colony-forming unit counts [14], even though a positive correlation was observed between the 2 methods of detection. The discrepancy between the 2 methods of quantification is most likely due to the fact that PCR does not distinguish between live and dead BCG, whereas culture only detects viable bacteria. The challenge dose received by each volunteer was similar and no correlation was observed between challenge dose and BCG recovery. BCG could not be detected by either culture or qPCR after swabbing of the BCG vaccination site in a separate cohort.

between live and dead BCG, whereas culture only detects viable bacteria. The challenge dose received by each volunteer was similar and no correlation was observed between challenge dose and BCG recovery. BCG could not be detected by either culture or qPCR after swabbing of the BCG vaccination site in a separate cohort. A significant reduction in BCG was detected by solid culture in the BCG–MVA85A group compared with the naive group. Using PCR, there was a significant 0.5- to 1-log reduction in BCG copy number in the 2 groups that had been BCG vaccinated when compared with the 2 BCG-naive groups. These findings suggest that prior BCG vaccination gives some protection against a subsequent challenge dose. Administration of MVA85A 4 weeks prior to BCG challenge had no added effect on the reduction of numbers of BCG detected. This finding is consistent with data recently published on the efficacy of MVA85A in a phase 2b trial in BCG-vaccinated infants in South Africa where boosting with MVA85A conferred no significant efficacy over BCG alone [25]. Vaccination with MVA85A induced a range of ex vivo IFN-γ ELISpot responses to Ag85A 7 days post vaccination (83–1667 sfc/million PBMC, median 1649), which inversely correlated with the number of colony-forming units recovered from the punch biopsies. The same inverse correlation was observed with PPD responses from PBMC isolated 14 days post challenge, suggesting that IFN- γ produced from antigen-specific CD4+ effector T cells is important for bacterial clearance from the challenge site.

ted with the number of colony-forming units recovered from the punch biopsies. The same inverse correlation was observed with PPD responses from PBMC isolated 14 days post challenge, suggesting that IFN- γ produced from antigen-specific CD4+ effector T cells is important for bacterial clearance from the challenge site. The MGIT assay did not detect any differences between the groups’ ability to reduce growth of BCG when incubated with whole blood on the day of challenge. The lack of detectable difference between the BCG-naive and BCG-vaccinated groups may be due to the length of time between BCG vaccination and when the MGIT assay was performed, that is, a median of 10 and 10.5 years for groups C and D, respectively. Other in vitro studies have shown enhanced mycobacterial growth inhibition due to BCG vaccination involving shorter time intervals (2–12 months) between vaccination and performance of the assays [26–29]. Comparison of this human BCG challenge model to animal M. tuberculosis challenge models shows a comparable effect of prior BCG vaccination. The same is not true of MVA85A, which has been shown to improve efficacy over BCG alone when given as a boost in preclinical animal models [5, 6, 30, 31]. However, in the animal challenge models, a high dose of M. tuberculosis is given by the aerosol route, while a standard vaccine dose of BCG is given by the intradermal route in the human model. Therefore, the challenge dose of BCG may be too low to detect further improvement over BCG alone. Further data are needed to truly compare the 2 models.

llenge models, a high dose of M. tuberculosis is given by the aerosol route, while a standard vaccine dose of BCG is given by the intradermal route in the human model. Therefore, the challenge dose of BCG may be too low to detect further improvement over BCG alone. Further data are needed to truly compare the 2 models. Data from this study support previous findings that this novel BCG challenge model can detect differences in antimycobacterial immunity induced by vaccination. In this trial, a difference could be detected between prior BCG vaccination and no prior BCG vaccination. MVA85A vaccination 4 weeks before challenge did not appear to further inhibit BCG growth. However, this study was performed in a population in which BCG had been demonstrated to be extremely effective [32]. Therefore, it might not be possible to see a additional effect of MVA85A vaccination with such small group sizes and numbers of BCG recovered. Model sensitivity needs to be improved, and further challenge trials are planned to address this issue. This will be done by varying the BCG challenge strain and the dose. The optimum time interval between vaccination and challenge also needs to be considered. After these parameters have been optimized, the study must be repeated with larger groups. The model also merits evaluation in populations where the BCG vaccine has a lower efficacy. Also, further work is needed to determine the target population and type of vaccine candidate that this BCG challenge model has utility for.

er these parameters have been optimized, the study must be repeated with larger groups. The model also merits evaluation in populations where the BCG vaccine has a lower efficacy. Also, further work is needed to determine the target population and type of vaccine candidate that this BCG challenge model has utility for. In the absence of a correlate of protection against M. tuberculosis, human BCG challenge provides a useful model to complement preclinical animal testing and immunological assessment to allow optimal selection of vaccines that will progress to field efficacy testing. Notes Acknowledgments. The authors thank Raquel Lopez Ramen, Mary Smith, Laura Dinsmore, Natalie Lella, Magali Matsumiya, and all the trial participants. Financial support. This work was supported by the Wellcome Trust (Senior Clinical Research Fellowship held by H. M.) and the National Institute for Health Research, Oxford Biomedical Research Centre, Oxford Universities Hospital. Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

An understanding of the adaptation of pathogens to changing environments and specific control efforts is important but challenging; however, developments in genomic methods offer approaches that may become widely effective. When new alleles are strongly selected within a population, associated haplotypes in flanking regions of the genome affect local linkage disequilibrium and the allele frequency spectrum [1]. However, evaluation of such statistical signatures against evidence of historical selection has rarely been performed directly, as estimates obtained by measuring changes in allele frequencies over time are rarely possible for natural populations.

fect local linkage disequilibrium and the allele frequency spectrum [1]. However, evaluation of such statistical signatures against evidence of historical selection has rarely been performed directly, as estimates obtained by measuring changes in allele frequencies over time are rarely possible for natural populations. In the human malaria parasite Plasmodium falciparum, alleles have been widely surveyed for genes that determine resistance to chloroquine (crt on chromosome 7 encoding the chloroquine resistance transporter and mdr1 on chromosome 5 encoding a multidrug resistance transporter) or antifolates (dhfr on chromosome 4 encoding the target of pyrimethamine and dhps on chromosome 8 encoding the target of sulfadoxine). Particular haplotypes at each of these chromosomal loci have undergone selective sweeps [2–7], some of these spreading from Asia into Africa [2, 8–10]. Several studies have compared proportions of alleles in endemic populations at different times, for example, in Mozambique over 5 years [11], Tanzania over 6 years [12], western Kenya over 8 years [13], Malawi over 8 years [14] and 9 years [15], Papua New Guinea over 12 years [16], Gabon over 14 years [17], and eastern Kenya over 14 years [18]. In each case, resistance alleles were already common at the start of the survey period. In addition to positive selection by drugs, there has also been a decline in prevalence of chloroquine resistance alleles after use of the drug ended in particular populations [14, 19], highlighting the existence of fitness costs and suggesting the potential for future reintroduction of chloroquine [20]. However, each of these temporal comparisons covered only part of the period during which local changes in drug resistance occurred, and none of them were directly related to attempts to detect or interpret signatures of selection from genome-wide analyses.

ng the potential for future reintroduction of chloroquine [20]. However, each of these temporal comparisons covered only part of the period during which local changes in drug resistance occurred, and none of them were directly related to attempts to detect or interpret signatures of selection from genome-wide analyses. Here, we report the longest temporal survey of drug resistance genes within an endemic population, covering different phases of antimalarial drug use and focusing on the 4 gene loci in P. falciparum that are responsible for chloroquine resistance (crt, mdr1) and antifolate resistance (dhfr, dhps). Parasite samples were collected in The Gambia over a 25-year period from 1984, when resistance was unknown locally, through the subsequent gradual failure of chloroquine and antifolate therapy, until the eventual switch to artemisinin combination therapy in 2008. The changes in allele frequencies are related to therapeutic policies and practices at different times during the period. We then surveyed whole parasite genome sequences from an independent sample of clinical isolates at the end of this period and examined the statistical impact of selection on these loci compared with the rest of the genome that encodes more than 5000 other genes. Results highlight the accuracy of genomic signatures of selection, as well as their transient nature, and identify additional loci under recent strong selection, indicating the need to apply these approaches more intensively in Africa where the public health implications of the spread of drug resistance are most profound.

ults highlight the accuracy of genomic signatures of selection, as well as their transient nature, and identify additional loci under recent strong selection, indicating the need to apply these approaches more intensively in Africa where the public health implications of the spread of drug resistance are most profound. METHODS Ethics Statement The Joint Ethics Committee (JEC) of the Gambia Government (JEC) and Medical Research Council Gambia Unit (MRC) approved this study. All subjects gave informed consent for sample collection, and procedures were conducted in accordance with the principles expressed in the Declaration of Helsinki.