Browse the corpus

What Is KnownMitochondrial liver disease is an important cause of infantile liver failure. The most effective way to diagnose mitochondrial liver disease in the setting of liver failure is unclear. What Is NewLow hepatic mitochondrial DNA copy number may be a consequence of liver disease rather than a cause of liver disease. Screening for known mutations causing mitochondrial liver disease may be the best diagnostic strategy. Acute liver failure (ALF) in infancy is a rare and devastating disease, which has a poor outcome without liver transplantation (LT). In approximately 20% of cases, infantile liver failure is caused by genetic mitochondrial liver disease (MLD) (1–3) with the commonest single entity being mitochondrial DNA (mtDNA) depletion syndrome (MDS). Mitochondria contain multiple copies of mtDNA. MDS is caused by mutations in nuclear genes involved in mtDNA replication or in the maintenance of the deoxynucleotide pools required for de novo mtDNA replication, resulting in a quantitative loss of mtDNA copy number (4). Pathogenic mutations causing hepatocerebral MDS have been described in a number of genes to date with the commonest reported being DGUOK(5), POLG(6), MPV17(7), and PEO1 (encoding the Twinkle helicase) (8).

ucleotide pools required for de novo mtDNA replication, resulting in a quantitative loss of mtDNA copy number (4). Pathogenic mutations causing hepatocerebral MDS have been described in a number of genes to date with the commonest reported being DGUOK(5), POLG(6), MPV17(7), and PEO1 (encoding the Twinkle helicase) (8). Normal mitochondrial function is contingent upon the expression of many other nuclear genes, which encode constituent proteins of the respiratory chain, proteins needed for assembly of the respiratory chain, or for translation of mtDNA-encoded proteins (9). Mutations in these genes can also cause MLD and in particular mutations in TRMU, which encodes an enzyme essential for post-transcriptional modification of mitochondrial tRNAs, can cause infantile ALF. Such cases are particularly important to recognize because there is a significant chance of spontaneous recovery with supportive treatment (10,11).

also cause MLD and in particular mutations in TRMU, which encodes an enzyme essential for post-transcriptional modification of mitochondrial tRNAs, can cause infantile ALF. Such cases are particularly important to recognize because there is a significant chance of spontaneous recovery with supportive treatment (10,11). Definitive diagnosis of most nuclear-encoded mitochondrial disorders is best established by recognizing 2 pathogenic mutations in known disease-causing genes. In the absence of an informative family history this is, however, time consuming. In the absence of a genetic diagnosis, laboratory diagnosis requires demonstrating abnormal respiratory chain function and/or loss of mtDNA copy number in clinically relevant tissue(s) (12). The most commonly sampled tissue is muscle as it is easily accessible with well-established normal ranges (12), although in multisystemic presentations of mitochondrial disease, muscle respiratory chain activities and mtDNA levels may be normal and muscle biopsy will a priori fail to detect isolated hepatic disease (6). Consequently liver biopsy is often necessary, if feasible. Abnormalities of respiratory chain function or of mtDNA copy number in damaged liver tissue, however, may not be due to genetic mitochondrial disease but may be a secondary change due to liver disease of other causes (13). The aim of the present study was to determine the incidence of genetic mitochondrial disease in a group of children younger than 2 years presenting with ALF and to determine the utility of routine investigations to detect mitochondrial disease.

Definitive diagnosis of most nuclear-encoded mitochondrial disorders is best established by recognizing 2 pathogenic mutations in known disease-causing genes. In the absence of an informative family history this is, however, time consuming. In the absence of a genetic diagnosis, laboratory diagnosis requires demonstrating abnormal respiratory chain function and/or loss of mtDNA copy number in clinically relevant tissue(s) (12). The most commonly sampled tissue is muscle as it is easily accessible with well-established normal ranges (12), although in multisystemic presentations of mitochondrial disease, muscle respiratory chain activities and mtDNA levels may be normal and muscle biopsy will a priori fail to detect isolated hepatic disease (6). Consequently liver biopsy is often necessary, if feasible. Abnormalities of respiratory chain function or of mtDNA copy number in damaged liver tissue, however, may not be due to genetic mitochondrial disease but may be a secondary change due to liver disease of other causes (13). The aim of the present study was to determine the incidence of genetic mitochondrial disease in a group of children younger than 2 years presenting with ALF and to determine the utility of routine investigations to detect mitochondrial disease. The present study was registered as an audit of clinical practice at Birmingham Children's Hospital NHS Trust. Children were investigated and managed according to an established in-house clinical protocol (see Supplemental Digital Content 1, Protocol). METHODS Methods are available online as Supplemental Digital Content 2.

The present study was registered as an audit of clinical practice at Birmingham Children's Hospital NHS Trust. Children were investigated and managed according to an established in-house clinical protocol (see Supplemental Digital Content 1, Protocol). METHODS Methods are available online as Supplemental Digital Content 2. RESULTS A total of 39 infants (20 girls, 19 boys) presented with ALF during the study period. Ethnicity was white (30), Asian (5), and black (4). Four were born prematurely and median birth weight was 2.7 kg (range 1.8–4.1). Median age at presentation was 17 days (1–689). There had been no affected siblings in these families before the study. There were 2 sets of siblings included, 1 being twins. Three children were from consanguineous families. Overall 10 infants died without LT during the acute illness and 18 survived without LT (Table 1). Eleven underwent LT of whom 3 died in the early postoperative period. There were 2 later deaths after transplantation; 1 from progressive multisystemic disease after 3 months and 1 from vascular complications 1 year later. One child with ornithine transcarbamylase deficiency underwent elective LT 1 year after presentation because of metabolic instability and remains well 3 years later. One child (subject 37) had recurrent episodes of ALF and he was subsequently shown to have mutations in neuroblastoma amplified sequence causing recurrent ALF syndrome (14).

thine transcarbamylase deficiency underwent elective LT 1 year after presentation because of metabolic instability and remains well 3 years later. One child (subject 37) had recurrent episodes of ALF and he was subsequently shown to have mutations in neuroblastoma amplified sequence causing recurrent ALF syndrome (14). The results of diagnostic investigations are summarized in Table 1. The largest etiological group was infection, accounting for 12 cases, including proven HSV in 8 cases, enterovirus in 3, and adenovirus in 1. Four had inborn errors of metabolism: galactosemia (2), ornithine transcarbamylase (1), recurrent ALF syndrome (1); 4 had neonatal hemochromatosis phenotype; and 12 cases were indeterminate despite an extensive diagnostic workup. Five children (13%) were found to have genetically confirmed MLD all of whom had MDS. All were born at full term after normal pregnancies. Three were born to consanguineous parents. Median age at presentation was 110 days (9 days to 23 months). The genetic causes were mutations in DGUOK (2), POLG (2), and MPV17 (1). Four of the 5 children with genetic MLD showed rapid deterioration and died within 3 weeks of presentation. One child who was homozygous for a p.(Leu304Arg) mutation in POLG presented at 18 months old, recovered with supportive treatment only and remains well without evidence of liver disease 6 years later.

Five children (13%) were found to have genetically confirmed MLD all of whom had MDS. All were born at full term after normal pregnancies. Three were born to consanguineous parents. Median age at presentation was 110 days (9 days to 23 months). The genetic causes were mutations in DGUOK (2), POLG (2), and MPV17 (1). Four of the 5 children with genetic MLD showed rapid deterioration and died within 3 weeks of presentation. One child who was homozygous for a p.(Leu304Arg) mutation in POLG presented at 18 months old, recovered with supportive treatment only and remains well without evidence of liver disease 6 years later. In addition, there was 1 unexplained case (subject 24) with some features of genetic MLD. This was a female infant who became jaundiced and unwell on the first day of life. She developed progressive encephalopathy and coagulopathy with peak INR of 3.5. Muscle biopsy showed steatosis and mtDNA depletion studies were borderline in both muscle (49%) and liver (39%). Cranial magnetic resonance imaging (MRI) showed features of cerebral edema only. She underwent LT at the age of 23 days. She made an initial smooth recovery but when aged 3 months developed evidence of cardiomyopathy and died of progressive systemic disease 2 months after LT. No evidence of a genetic cause of MLD was found.

al magnetic resonance imaging (MRI) showed features of cerebral edema only. She underwent LT at the age of 23 days. She made an initial smooth recovery but when aged 3 months developed evidence of cardiomyopathy and died of progressive systemic disease 2 months after LT. No evidence of a genetic cause of MLD was found. Clinical and laboratory features of the infants with genetically proven MLD compared to those with other causes of ALF are summarized in Table 2. Children with MLD tended to have lower birth weight and presented later but these differences were not significant. Similarly, there were no significant differences in the presenting laboratory values between the 2 groups. Although the median plasma lactate levels were similar between the groups, all infants with MLD had abnormal lactate values, whereas these were initially normal in 9 of 34 without MLD. Results of tissue studies and radiology are listed in Table 1 and summarized in Table 3. Liver histology was available in 21 cases. The dominant lesion was hepatocyte necrosis in 13 cases, and this was accompanied by microvesicular steatosis in 3 cases. Including these 3 cases, significant microvesicular steatosis was present in 8 cases overall. Three who had genetically confirmed MLD had liver histology available and all showed microvesicular steatosis. The remaining 4 biopsies showed established fibrosis/cirrhosis (3) and unexplained macrophage storage material, respectively.

these 3 cases, significant microvesicular steatosis was present in 8 cases overall. Three who had genetically confirmed MLD had liver histology available and all showed microvesicular steatosis. The remaining 4 biopsies showed established fibrosis/cirrhosis (3) and unexplained macrophage storage material, respectively. Liver mtDNA copy number results were available in 17 cases, 2 of whom had genetically proven MLD due to DGUOK. These 2 children had low (15%) and borderline (37%) liver mtDNA copy number. In 15 children without MLD, 7 had normal mtDNA copy number in liver and 8 had low levels of mtDNA: depletion (4) and borderline depletion (4). The causes of ALF in these 8 children with decreased mtDNA copy number without genetically proven MLD were indeterminate in 6 and 1 each of neonatal hemochromatosis and enterovirus infection. Two of these children died, 2 recovered without LT, and 4 underwent successful LT. One child, referred to earlier, underwent successful LT but died from apparent multisystemic disease 2 months later. None of the 5 survivors have shown evidence of multisystemic disease after up to 6 years of follow-up.

ovirus infection. Two of these children died, 2 recovered without LT, and 4 underwent successful LT. One child, referred to earlier, underwent successful LT but died from apparent multisystemic disease 2 months later. None of the 5 survivors have shown evidence of multisystemic disease after up to 6 years of follow-up. Muscle biopsies were available in 12 cases. None showed specific changes suggestive of mitochondrial involvement such as ragged-red fibers. Increased intrafiber lipid was found in 4 of 5 children with MLD who underwent muscle biopsy but was only found in 1 of 7 children without MLD. This latter child was the one who died of a multisystemic disease after LT. Muscle mtDNA copy number data were available in 11 cases, 4 of whom had MLD. Two children with MLD had low mtDNA copy number; 1 of these had complex IV deficiency in muscle tissue and 1 had normal enzyme activities. Two children with MLD had normal mtDNA copy number, and both also had normal respiratory chain activity. Six of 7 children without MLD had normal mtDNA copy number and, in the 4 cases in which these were measured, they also had normal respiratory chain enzyme activities. One had an ambiguous muscle mtDNA copy number (47%).

children with MLD had normal mtDNA copy number, and both also had normal respiratory chain activity. Six of 7 children without MLD had normal mtDNA copy number and, in the 4 cases in which these were measured, they also had normal respiratory chain enzyme activities. One had an ambiguous muscle mtDNA copy number (47%). A total of 15 children underwent cranial MRI with diffusion-weighted imaging and 10 had magnetic resonance spectroscopy (MRS). All 5 children with MLD had MRI, which in 1 case (who had POLG mutation) showed symmetrical posterior midbrain changes similar to those reported in mitochondrial disease (4). Three showed cerebral edema which had a cytotoxic or demyelination pattern in 2 cases and a vasogenic pattern in 1. Two children had an initial normal MRI, but in 1 case repeat MRI showed progression to vasogenic cerebral edema. Ten children without MLD had an MRI, which was normal in 3 and showed cerebral edema in 7, appearing cytotoxic in 2 and vasogenic in 5. Five children with MLD had MRS, which showed a lactate peak in 3. Five children without MDS had MRS, which showed a lactate peak in 2.

I showed progression to vasogenic cerebral edema. Ten children without MLD had an MRI, which was normal in 3 and showed cerebral edema in 7, appearing cytotoxic in 2 and vasogenic in 5. Five children with MLD had MRS, which showed a lactate peak in 3. Five children without MDS had MRS, which showed a lactate peak in 2. DISCUSSION Infantile ALF is a serious disorder with a variety of potential causes. A structured, rapid approach to diagnostic investigations in tandem with identifying and treating correctable disorders is necessary. We have confirmed that MLD is an important cause of infantile ALF and that genetically confirmed MDS is the commonest entity in this group. The outlook for affected infants is poor and early recognition is important to minimize unnecessary invasive investigations, to prevent inappropriate LT, and to facilitate family counseling. Ideally, diagnostic investigations should be available within days of presentation. The definitive method to diagnose MLD is by detection of 2 pathogenic mutations in recognized genes; hence, some attempt at targeted mutation detection should be initiated at the time of initial presentation. This could later be reassessed if other diagnostic information becomes available.

within days of presentation. The definitive method to diagnose MLD is by detection of 2 pathogenic mutations in recognized genes; hence, some attempt at targeted mutation detection should be initiated at the time of initial presentation. This could later be reassessed if other diagnostic information becomes available. In the absence of pathogenic disease-causing mutations, the diagnosis of MLD depends on tissue studies. It has been a sine qua non in the investigation of suspected mitochondrial disease that an affected tissue should be studied. Our findings cast doubt on this approach in the setting of ALF. We have found that reduced mtDNA copy number in affected liver tissue is not synonymous with genetically proven MDS. In 3 of the 8 cases reported here plausible alternative causes of ALF were found. In the 5 unexplained cases we cannot definitely exclude mitochondrial disease because undetected genetic disorders may yet be present. For at least 4 of these cases, primary mitochondrial disease, however, seems unlikely; no pathogenic mutations have been detected and no other evidence to support progressive mitochondrial disease has appeared even after prolonged follow-up. One of these children, who developed a multisystemic disease after LT, did have some features of systemic mitochondrial disease but no genetic cause was detected.

y; no pathogenic mutations have been detected and no other evidence to support progressive mitochondrial disease has appeared even after prolonged follow-up. One of these children, who developed a multisystemic disease after LT, did have some features of systemic mitochondrial disease but no genetic cause was detected. There have been few studies examining the accuracy of low hepatic mtDNA copy number to diagnose MLD in which the primary presentation is with clinical liver disease. In end-stage liver disease some studies have shown that low mtDNA copy number appeared to be specific for MDS (15), but in another study 10 of 45 unselected cases undergoing LT had low copy number (16). Low mtDNA copy number has also been reported in Mauriac syndrome in which the clinical findings are often reversible (17). In ALF low copy number appears to be common irrespective of the etiology. Helbling et al (15) found low mtDNA copy number in 29 of 44 patients with ALF and all 3 cases reported by Lane et al (16) had decreased number. Decreased copy numbers were found even where a plausible nonmitochondrial cause of ALF existed. In contrast, Al-Hussaini and colleagues (1) found hepatic mtDNA copy number to be specific for MDS, but only 4 children in whom liver disease did not have a mitochondrial cause were studied.

et al (16) had decreased number. Decreased copy numbers were found even where a plausible nonmitochondrial cause of ALF existed. In contrast, Al-Hussaini and colleagues (1) found hepatic mtDNA copy number to be specific for MDS, but only 4 children in whom liver disease did not have a mitochondrial cause were studied. Overall these reports are consistent with our findings and suggest that liver disease, and especially ALF, may cause a secondary lowering of mtDNA copy number as a consequence of the primary disease. We cannot exclude that as yet undetected mutations in other genes underlie these examples of mtDNA depletions. An important part of the present study is, however, the length of subsequent follow-up, which makes late sequela of unrecognized disease less likely. We also cannot comment as to whether the low mtDNA copy number contributes to the pathogenesis of ALF in these cases. What we can say is that clinical management decisions, including whether to proceed with transplantation, should not be influenced by hepatic mtDNA copy number in the absence of proven mutations. Rapid detection of pathogenic mutations in candidate genes remains the ideal method for diagnosis of MLD. The commonest causes of MLD are recessively-inherited mutations in DGUOK, POLG, MPV17, PEO1, and TRMU(1,7,18). Certainly, screening for mutations in these genes should be initiated at presentation with infantile ALF. The prioritization of genes to screen will depend on local experience and available facilities, while recognizing that this approach will only recognize a proportion of defects.

POLG, MPV17, PEO1, and TRMU(1,7,18). Certainly, screening for mutations in these genes should be initiated at presentation with infantile ALF. The prioritization of genes to screen will depend on local experience and available facilities, while recognizing that this approach will only recognize a proportion of defects. Up to 1300 nuclear genes encode mitochondrial-related proteins and the basis of many defects remain unknown (18). It is to be hoped that next-generation screening techniques, including custom captures of specific nuclear-mitochondrial genes or whole exome or whole genome sequencing, will transform this situation. For example, it is now possible to sequence the entire mitochondrial genome and all coding exons of the nuclear genes encoding mitochondrial proteins. Initial experience using this approach for children with suspected mitochondrial disease achieved a firm diagnosis in 24% of cases and a probable cause in a further 30% (19). The major future challenge will be to ensure next-generation screening results can be made available in a clinically relevant timescale, that is, within days if possible, and certainly within a fortnight, although this will vary according to local practice and laboratory diagnostic algorithms.

se in a further 30% (19). The major future challenge will be to ensure next-generation screening results can be made available in a clinically relevant timescale, that is, within days if possible, and certainly within a fortnight, although this will vary according to local practice and laboratory diagnostic algorithms. Even establishing a molecular diagnosis does not absolutely establish prognosis. Although 4 of the 5 cases showed rapid progression and death from systemic disease, 1 child with recessive POLG mutations recovered spontaneously; interestingly, she was homozygous for the p.(Leu304Arg) mutation that is usually associated with a late-onset POLG phenotype of sensory ataxic neuropathy with dysarthria and ophthalmoplegia rather than liver disease (20). This mutation has been reported to cause ALF in compound with a second (p.[Ala467Thr]) heterozygous POLG mutation (21), which supports the observation of Tzoulis et al (22) that compound heterozygosity often has a worse prognosis than homozygous POLG mutations. Recent work has suggested that the pattern of mtDNA when visualized by fluorescence microscopy in cultured fibroblasts may also provide further prognostic information (12). Spontaneous recovery from ALF has been previously recognized in at least 1 other child with POLG mutations (23) and emphasizes that, although LT is inappropriate in this group, these patients should not be denied appropriate supportive treatment.

ultured fibroblasts may also provide further prognostic information (12). Spontaneous recovery from ALF has been previously recognized in at least 1 other child with POLG mutations (23) and emphasizes that, although LT is inappropriate in this group, these patients should not be denied appropriate supportive treatment. Recognizing and defining central nervous system involvement in MLD is crucial to guide prognosis and management. In ALF from other causes central nervous system involvement with encephalopathy is common and is generally reversible after successful LT. In MLD such involvement, however, may be a contraindication to LT. MRI abnormalities are common, but not invariable, in MLD and range from widespread generalized white matter changes to cortical atrophy to specific involvement of deeper brain structures (1,4,24). These latter appear to be more specific for MLD but were found in only 1 of our cases. We found that generalized abnormalities were common in ALF irrespective of cause and that there was a similar distribution between cytotoxic and vasogenic cerebral edema whether or not liver failure was due to MLD. Similarly, MRS detection of a lactate peak did not provide useful discrimination between mitochondrial and nonmitochondrial causes. We did confirm that MRI changes may develop and evolve quickly and that serial evaluation may be necessary. In this group of ill infants MRI, however, only helped the decision on appropriateness of LT in a small proportion of cases.

id not provide useful discrimination between mitochondrial and nonmitochondrial causes. We did confirm that MRI changes may develop and evolve quickly and that serial evaluation may be necessary. In this group of ill infants MRI, however, only helped the decision on appropriateness of LT in a small proportion of cases. In conclusion, we have shown that MLD is an important cause of infantile ALF and that mutation detection is the most robust diagnostic method. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content The diagnostic laboratories in Oxford, London, and Newcastle upon Tyne are funded by the UK NHS Highly Specialized “Rare Mitochondrial Disorders of Adults and Children” Service. R.M. and R.W.T. are supported by a Wellcome Trust Strategic Award (096919Z/11/Z). S.R. is supported by Great Ormond Street Hospital Children's Charity. The authors have no conflicts of interest. TABLE 1 Investigations and clinical outcome of 39 infants with acute liver failure admitted to Birmingham Children's Hospital from 2009 to 2011

Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content The diagnostic laboratories in Oxford, London, and Newcastle upon Tyne are funded by the UK NHS Highly Specialized “Rare Mitochondrial Disorders of Adults and Children” Service. R.M. and R.W.T. are supported by a Wellcome Trust Strategic Award (096919Z/11/Z). S.R. is supported by Great Ormond Street Hospital Children's Charity. The authors have no conflicts of interest. TABLE 1 Investigations and clinical outcome of 39 infants with acute liver failure admitted to Birmingham Children's Hospital from 2009 to 2011 Subject no. Diagnosis Age at presentation, days Initial serum lactate, mmol/L Muscle histology Muscle respiratory chain activity Muscle mtDNA copy number, % Liver histology Liver mtDNA copy number, % Cranial MRI MR spectroscopy Mutations detected Outcome 1 MDS (DGUOK) 9 7.2 Increased lipid Decrease in complex IV 45.0 Micro- and macrovesicular steatosis 37.0 Vasogenic pattern cerebral edema Lactate peak p.[Glu44Lys]; [Glu44Lys] Died 2 MDS (DGUOK) 16 25.0 Increased lipid Normal 50.0 Hepatocellular necrosis and microvesicular steatosis 15.0 Vasogenic pattern cerebral edema Lactate peak p.[Arg39X]; [Arg39X] Died 3 MDS (MPV17) 110 9.9 Increased lipid Decrease in complexes I–IV ND ND ND Normal Normal p.[Gln44X]; [Gln44X] Died 4 MDS (POLG) 477 6.0 Normal Normal 55.0 ND ND Cytotoxic pattern cerebral edema Lactate peak p.[Thr914Pro]; [Ala467Thr] Died 5 MDS (POLG) 503 3.7 Increased lipid Normal 107.0 Micro- and macrovesicular steatosis ND Post-midbrain lesion in keeping with mitochondrial disease Normal p.[Leu304Arg]; [Leu304Arg] Recovered. Remains well 6 years later.

mal Normal 55.0 ND ND Cytotoxic pattern cerebral edema Lactate peak p.[Thr914Pro]; [Ala467Thr] Died 5 MDS (POLG) 503 3.7 Increased lipid Normal 107.0 Micro- and macrovesicular steatosis ND Post-midbrain lesion in keeping with mitochondrial disease Normal p.[Leu304Arg]; [Leu304Arg] Recovered. Remains well 6 years later. 6 HSV 8 8.5 ND ND ND Hemorrhagic necrosis 83 ND ND None Died following LT 7 HSV 11 16.0 ND ND ND Hemorrhagic necrosis ND ND ND None Died 8 HSV 8 ND ND ND ND ND ND ND None Recovered 9 HSV 8 ND ND ND ND ND ND ND None Recovered 10 HSV 11 ND ND ND ND ND ND ND None Recovered 11 HSV 11 ND ND ND ND ND ND ND None Died following LT 12 HSV 21 6.4 ND ND ND ND ND ND ND None Died 13 HSV 17 4.7 ND ND ND ND ND Vasogenic pattern cerebral edema ND None Recovered 14 Enterovirus 6 22.4 ND ND ND ND ND ND ND None Died 15 Enterovirus 12 Normal Normal 54.0 Hemorrhagic necrosis 34.0 ND ND None Died following LT 16 Enterovirus 9 9.6 ND ND ND ND ND Vasogenic pattern cerebral edema Lactate peak None Died 17 Adenovirus 112 ND Normal ND ND ND Cytotoxic pattern cerebral edema ND None Recovered 18 Indeterminate 498 5.6 ND ND ND Panacinar necrosis 29.0 ND ND None Successful LT 19 Indeterminate 228 3.3 ND ND ND Microvesicular steatosis 25.0 ND ND None Recovered 20 Indeterminate 52 3.8 ND ND ND Severe fibrosis and microvesicular steatosis 40.0 ND ND None Recovered 21 Indeterminate 349 4.7 ND ND ND Panacinar necrosis and macrovesicular steatosis 42.0 ND ND None Successful LT 22 Indeterminate 260 1.2 ND ND ND ND ND ND ND None Died 23 Indeterminate 15 4.3 Normal Normal 127.0 Hepatocyte necrosis and macrophage storage material 91.0 Normal ND None Died 24 Indeterminate 1 3.0 Increased lipid ND 47.0 Panacinar necrosis 39.0 Vasogenic pattern cerebral edema ND None Underwent LT but died of progressive multisystemic illness 2 months later 25 Indeterminate 689 3.8 ND ND ND Hepatocellular necrosis and microvesicular steatosis 60 ND ND None Recovered 26 Indeterminate 644 2.7 ND ND ND Panacinar necrosis 92 ND ND None Recovered 27 Indeterminate 1 2.1 Normal Normal 83.0 ND ND ND ND None Recovered 28 Indeterminate 37 Normal Normal 81.0 Moderate fibrosis ND Normal ND None Recovered 29 Indeterminate 216 4.4 ND ND ND Panacinar necrosis 25 Cytotoxic pattern cerebral edema Normal None Successful LT but died 1 year later from vascular complications 30 NH 8 4.7 ND ND ND ND ND ND ND None Recovered 31 NH 24 3.2 Normal ND ND Panacinar necrosis 65.0 ND ND None Successful LT 32 NH 21 0.7 ND ND ND Established cirrhosis 54.0

216 4.4 ND ND ND Panacinar necrosis 25 Cytotoxic pattern cerebral edema Normal None Successful LT but died 1 year later from vascular complications 30 NH 8 4.7 ND ND ND ND ND ND ND None Recovered 31 NH 24 3.2 Normal ND ND Panacinar necrosis 65.0 ND ND None Successful LT 32 NH 21 0.7 ND ND ND Established cirrhosis 54.0 ND ND None Successful LT 33 NH 13 1.0 Normal Normal 88.0 Panacinar necrosis 26.0 Cytotoxic pattern cerebral edema Normal None Successful LT 34 OTC deficiency 599 2.5 ND ND ND ND ND ND ND None Recovered. Underwent elective LT age 3. 35 Galactosemia 7 ND ND ND ND ND ND ND None Recovered 36 Galactosemia 9 ND ND ND ND ND ND ND None Recovered 37 RALF 539 5.8 ND Normal 80 Microvesicular steatosis 127 Normal Normal None Recovered and had subsequent recurrent bouts of liver failure 38 HLH 6 4.3 ND ND ND ND ND Vasogenic pattern cerebral edema Lactate peak None Died 39 Chemotherapy 121 6.6 ND ND ND Macro- and microvesicular steatosis ND ND ND None Recovered HLH = hemophagocytic lymphohistiocytosis; HSV = herpes simplex virus infection; LT = liver transplantation; MDS = mitochondrial depletion syndrome; MRI = magnetic resonance imaging; MRS = magnetic resonance spectroscopy; mtDNA = mitochondrial DNA; ND = not done; NH = neonatal hemochromatosis phenotype; OTC = ornithine transcarbamylase deficiency; RALF = recurrent acute liver failure. TABLE 2 Clinical and laboratory features in infant with and without genetically proven mitochondrial liver disease as a cause of liver failure

Underwent elective LT age 3. 35 Galactosemia 7 ND ND ND ND ND ND ND None Recovered 36 Galactosemia 9 ND ND ND ND ND ND ND None Recovered 37 RALF 539 5.8 ND Normal 80 Microvesicular steatosis 127 Normal Normal None Recovered and had subsequent recurrent bouts of liver failure 38 HLH 6 4.3 ND ND ND ND ND Vasogenic pattern cerebral edema Lactate peak None Died 39 Chemotherapy 121 6.6 ND ND ND Macro- and microvesicular steatosis ND ND ND None Recovered HLH = hemophagocytic lymphohistiocytosis; HSV = herpes simplex virus infection; LT = liver transplantation; MDS = mitochondrial depletion syndrome; MRI = magnetic resonance imaging; MRS = magnetic resonance spectroscopy; mtDNA = mitochondrial DNA; ND = not done; NH = neonatal hemochromatosis phenotype; OTC = ornithine transcarbamylase deficiency; RALF = recurrent acute liver failure. TABLE 2 Clinical and laboratory features in infant with and without genetically proven mitochondrial liver disease as a cause of liver failure Genetically proven MLD (n = 5) Other causes of ALF (n = 34) P Birth weight, kg 2.6 (2.3–2.9) 2.8 (1.8–4.1) 0.76 Age at presentation, days 110 (9–503) 16 (1–689) 0.33 Prothrombin time, s 26 (23–41) 34 (18–120) 0.5 Serum bilirubin, μmol/L 113 (34–335) 146 (5–492) 0.79 ALT, IU/L 113 (42–1875) 940 (19–6000) 0.11 Lactate, mmol/L 7.2 (3.7–25) 4.3 (0.7–22.4) 0.11 ALF = acute liver failure; ALT = alanine aminotransferase; MLD = mitochondrial liver disease. TABLE 3 Results of tissue studies and radiology undertaken in infants with and without genetically proven mitochondrial liver disease

Genetically proven MLD (n = 5) Other causes of ALF (n = 34) P Birth weight, kg 2.6 (2.3–2.9) 2.8 (1.8–4.1) 0.76 Age at presentation, days 110 (9–503) 16 (1–689) 0.33 Prothrombin time, s 26 (23–41) 34 (18–120) 0.5 Serum bilirubin, μmol/L 113 (34–335) 146 (5–492) 0.79 ALT, IU/L 113 (42–1875) 940 (19–6000) 0.11 Lactate, mmol/L 7.2 (3.7–25) 4.3 (0.7–22.4) 0.11 ALF = acute liver failure; ALT = alanine aminotransferase; MLD = mitochondrial liver disease. TABLE 3 Results of tissue studies and radiology undertaken in infants with and without genetically proven mitochondrial liver disease Genetically proven MLD (n = 5) Other causes of ALF (n = 34) Liver mtDNA depletion 2/2 8/15 Muscle mtDNA depletion 2/4 1/7 Abnormal muscle respiratory chain enzymes 2/5 0/4 Muscle steatosis 4/5 1/7 Hepatic steatosis 3/3 5/18 MRS lactate peak 3/5 2/5 ALF = acute liver failure; MLD = mitochondrial liver disease; MRS = magnetic resonance spectroscopy; mtDNA = mitochondrial DNA.

What Is KnownEnvironmental enteric dysfunction is associated with reduced linear growth/stunting. The cumbersome lactulose:mannitol test is used to assess environmental enteric dysfunction. A noninvasive biomarker to identify environmental enteric dysfunction is needed. What Is NewRisk factors for environmental enteric dysfunction in rural African children are identified. Quantification of fecal host mRNAs in a random forest model shows promise as a noninvasive method to identify children with environmental enteric dysfunction. Optimal gut health includes as the ability of the intestines to absorb all necessary dietary nutrients while mounting responses to limit dissemination of inflammatory microbes and their products from the lumen, so as to avert local and systemic inflammation. Environmental enteric dysfunction (EED) is the most pervasive condition associated with poor gut health worldwide, which is prevalent in rural African children and is associated with stunting (1–3).

dissemination of inflammatory microbes and their products from the lumen, so as to avert local and systemic inflammation. Environmental enteric dysfunction (EED) is the most pervasive condition associated with poor gut health worldwide, which is prevalent in rural African children and is associated with stunting (1–3). Tissue-dependent assessment of gut health involves visualization of the mucosa and biopsy (4). Endoscopy is, however, an invasive and expensive procedure that is poorly suited to mass screening, or to frequent reassessment in individuals. The dual sugar absorption test is a widely used alternative, in which lactulose and mannitol are ingested, and urines are collected during the subsequent several hours. Disrupted cell junctions allow lactulose, a disaccharide, and mannitol, a monosaccharide, to be passively absorbed, providing a measure of gut epithelial integrity. Mannitol is also absorbed across cell membranes, which is measure of the surface area of the intestinal tract. Once absorbed, these sugars are excreted intact in the urine. The ratio of urinary lactulose to mannitol (L:M) indicates the degree of epithelial disruption in the small bowel, and, by extension, poor gut health (5). The L:M test is theoretically sound and often used, but cumbersome to administer.

the intestinal tract. Once absorbed, these sugars are excreted intact in the urine. The ratio of urinary lactulose to mannitol (L:M) indicates the degree of epithelial disruption in the small bowel, and, by extension, poor gut health (5). The L:M test is theoretically sound and often used, but cumbersome to administer. Stool contains exfoliated enterocytes, representing gut mucosal tissue. Stool specimens are acquired noninvasively. Fecal extractions have been used to analyze expression of individual host transcripts by droplet digital PCR (ddPCR) (6,7). Measurement of host fecal transcripts is challenging because human mRNA is estimated to be <1% of total fecal RNA, which is predominantly of microbial and ribosomal origin. mRNA in feces is also relatively degraded, and quantification can be further hampered by coextracted inhibitors, but methods to overcome these limitations have recently been developed (7). Here we use the L:M test as the standard to assess gut health in rural Malawian children, determine the relation between L:M and linear growth, and compare a panel of host fecal transcripts and clinical characteristics to this standard.

Stool contains exfoliated enterocytes, representing gut mucosal tissue. Stool specimens are acquired noninvasively. Fecal extractions have been used to analyze expression of individual host transcripts by droplet digital PCR (ddPCR) (6,7). Measurement of host fecal transcripts is challenging because human mRNA is estimated to be <1% of total fecal RNA, which is predominantly of microbial and ribosomal origin. mRNA in feces is also relatively degraded, and quantification can be further hampered by coextracted inhibitors, but methods to overcome these limitations have recently been developed (7). Here we use the L:M test as the standard to assess gut health in rural Malawian children, determine the relation between L:M and linear growth, and compare a panel of host fecal transcripts and clinical characteristics to this standard. METHODS Subjects Among rural Malawian children 12 to 61 months of age who participated in 1 of 3 clinical studies, 798 did not receive an intervention and were therefore eligible for inclusion in this biomarker study (8–10). Children with severe acute malnutrition, diarrhea within the last 3 days, congenital abnormalities, and chronic debilitating illnesses were excluded. Enrolled children were from families of rural subsistence farmers, consumed water from wells or boreholes, lived in unelectrified mud huts, and were at high risk for EED. Ethical approval was obtained from the University of Malawi, Baylor College of Medicine, and Washington University in St Louis.

nesses were excluded. Enrolled children were from families of rural subsistence farmers, consumed water from wells or boreholes, lived in unelectrified mud huts, and were at high risk for EED. Ethical approval was obtained from the University of Malawi, Baylor College of Medicine, and Washington University in St Louis. Study Design This was an observational study to determine whether a single L:M predicts subsequent linear growth in these rural African children; which dietary, demographic, and household sanitation practices are associated with L:M; and the extent to which fecal host mRNAs predict L:M. The participants had originally been enrolled in 1 of 3 clinical studies, which included L:M testing using a uniform, controlled, and standardized method. One of the studies monitored the growth of twins in 4 villages monthly, with L:M testing conducted at a single point in time (8). The other 2 studies were randomized, double-blind, placebo-controlled clinical trials to ameliorate EED and from these studies only the data collected at the time of enrollment from children assigned to placebo group were included in these analyses (9,10). Thus the data do not represent a longitudinal cohort, and L:M testing was not available from >1 time for any given child.

-controlled clinical trials to ameliorate EED and from these studies only the data collected at the time of enrollment from children assigned to placebo group were included in these analyses (9,10). Thus the data do not represent a longitudinal cohort, and L:M testing was not available from >1 time for any given child. Using data from these 3 studies, the primary outcomes evaluated were the ability of L:M to predict subsequent linear growth in a linear regression model, identification of clinical and environmental risk factors for EED as defined by an abnormal L:M, and the sensitivity and specificity of random forest modeling of sets of fecal host transcripts to predict the severity of EED. Participation Information regarding the demographics, dietary intake, and household sanitation practices information was collected using a standard questionnaire from the child's primary caretaker. Length and weight were measured as previously described (9,10).

Using data from these 3 studies, the primary outcomes evaluated were the ability of L:M to predict subsequent linear growth in a linear regression model, identification of clinical and environmental risk factors for EED as defined by an abnormal L:M, and the sensitivity and specificity of random forest modeling of sets of fecal host transcripts to predict the severity of EED. Participation Information regarding the demographics, dietary intake, and household sanitation practices information was collected using a standard questionnaire from the child's primary caretaker. Length and weight were measured as previously described (9,10). All of the subjects underwent a carefully conducted L:M test with adequate urine collection and sugar excretion (11,12). Caregivers were directed not to feed the child for 12 hours before participation. Children were given lactulose (5 g) and mannitol (1 g) dissolved in 20 mL of water. All urine was collected from the time of sugar ingestion through at least 4 hours after sugar ingestion in containers with 10-mg merthiolate, to limit bacterial degradation of the sugars. Children were encouraged to drink water to facilitate urination and mothers were instructed not to breast-feed their children during this time. The total urine volume was measured and a 4-mL aliquot was transferred into cryovials, flash frozen on site, and transported frozen (−80°C) to Baylor College of Medicine.

gars. Children were encouraged to drink water to facilitate urination and mothers were instructed not to breast-feed their children during this time. The total urine volume was measured and a 4-mL aliquot was transferred into cryovials, flash frozen on site, and transported frozen (−80°C) to Baylor College of Medicine. Fresh stool specimens were collected on site before completion of the L:M testing using a small, clean nonabsorbent, plastic diaper. The stools were immediately transferred to cryovials and flash frozen in liquid nitrogen, without buffers, enzymes, or preservative solutions. Samples were transferred to a −80°C freezer and transported to Washington University (St Louis, MO) at −80°C, where they were then processed and analyzed for human fecal mRNA.

. The stools were immediately transferred to cryovials and flash frozen in liquid nitrogen, without buffers, enzymes, or preservative solutions. Samples were transferred to a −80°C freezer and transported to Washington University (St Louis, MO) at −80°C, where they were then processed and analyzed for human fecal mRNA. Laboratory Analyses Concentrations of lactulose and mannitol in the urine specimens were analyzed by high-performance liquid chromatography (HPLC) using a modification of the method of Catassi et al (13,14). The assays are sensitive to 1 μg/mL lactulose and mannitol and the coefficient of variation is ≤5% (15). Because of concerns raised in the literature of the accuracy of HPLC measurements of lactulose and mannitol (16), 115 samples were sent to the laboratory of Dr William Faubion at Mayo Clinic (Rochester, MN) and tested using liquid chromatography-tandem mass spectrometry (Applied Biosystems-MDS SCIEX, Foster City, CA) (16). Urines were mixed with 250 μL of internal standard, 13C labeled mannitol and lactulose, and chromatographic separation is achieved using a CARBOSep COREGEL 87C column from Transgenomic (Transgenomic, Omaha, NE) at 0.6 mL/min and 83oC. Intra-assay coefficients of variation for mannitol ranged from 2.6% to 4.1% for levels at 1.52, 35.8, 146, and 458 μg/mL and lactulose coefficients of variation ranged from 3.0% to 6.4% for levels at 0.6, 4.65, 38.9, and 143 μg/mL.

p COREGEL 87C column from Transgenomic (Transgenomic, Omaha, NE) at 0.6 mL/min and 83oC. Intra-assay coefficients of variation for mannitol ranged from 2.6% to 4.1% for levels at 1.52, 35.8, 146, and 458 μg/mL and lactulose coefficients of variation ranged from 3.0% to 6.4% for levels at 0.6, 4.65, 38.9, and 143 μg/mL. Host fecal mRNAs were analyzed using a conservative method for host mRNA isolation and ddPCR as previously described (6,7). This method uses glyceraldehyde 3-phosphate dehydrogenase as an internal standard to normalize all transcript measurements, which reduces variations in measurement values when samples are run in different settings or in different conditions. Eighteen transcripts correlated with L:M (7) were eligible for incorporation into the random forest model. Data Analysis Clinical and laboratory data were aggregated from the enrollment data from the 3 clinical studies. Height-for-age z score (HAZ) and weight-for-height z score (WHZ) were determined using the 2006 World Health Organization Multicenter Growth Reference Study child growth standards (17). L:M was calculated as the ratio of the lactulose to mannitol concentrations in the urine.

enrollment data from the 3 clinical studies. Height-for-age z score (HAZ) and weight-for-height z score (WHZ) were determined using the 2006 World Health Organization Multicenter Growth Reference Study child growth standards (17). L:M was calculated as the ratio of the lactulose to mannitol concentrations in the urine. To identify children at greatest risk for stunting and create a categorical random forest model to predict EED, 3 categories of L:M were designated; no EED as L:M ≤ 0.15, moderate EED as 0.15 < L:M < 0.45 and severe EED as L:M ≥ 0.45. The L:M value chosen to represent no EED is based on measurements made in healthy individuals in Europe and North America (18,19). Severe EED was designated to be L:M > 0.45 because this corresponds to L:M measurements in children with Crohn disease in remission and in celiac disease (20,21). No EED constituted 18% of the study population, moderate EED 66% of the study population, and severe EED 17% of the study population. The demographic, dietary, anthropometric, and sanitation practice characteristics were compared for these 3 categories of EED using a one-way ANOVA (SPSS22, Chicago, IL).

isease (20,21). No EED constituted 18% of the study population, moderate EED 66% of the study population, and severe EED 17% of the study population. The demographic, dietary, anthropometric, and sanitation practice characteristics were compared for these 3 categories of EED using a one-way ANOVA (SPSS22, Chicago, IL). Linear growth measurements over the 3 months subsequent to specimen acquisition were available from children in 2 of the 3 studies (8,10); change in HAZ (ΔHAZ) was calculated for each child from those 2 studies and these data were used to create a stepwise, backward linear regression model to predict ΔHAZ. L:M was the primary independent variable; other covariates were the child's age, sex, WHZ, HAZ, whether his/her mother was the primary caregiver, whether the father was alive, number of siblings, number of individuals that sleep in the same room as the child, roofing material, bicycle ownership, whether animals sleep in the house with the child, whether water is from a clean source, whether the family uses a pit latrine, household food insecurity score (22), dietary diversity score (23), number of times per day animal source foods are consumed, and diarrhea reported 4 to 7 days before L:M testing. Covariates were considered significant if P < 0.05. The third study in our dataset included children with L:M, collection of clinical and dietary information and fecal specimens; however, the duration of follow-up was only 7 weeks, so reliable linear growth data were not available (9).

d 4 to 7 days before L:M testing. Covariates were considered significant if P < 0.05. The third study in our dataset included children with L:M, collection of clinical and dietary information and fecal specimens; however, the duration of follow-up was only 7 weeks, so reliable linear growth data were not available (9). Modeling We chose random forest modeling as the machine-based learning method of modeling because it is the most appropriate and powerful for determining categorical outcomes, in our case no EED, moderate EED, and severe EED. Random forest modeling creates a set of computer-generated decision trees from a set independent variables, in our case the copy numbers of a set host transcripts, to designate the child into an EED category (10). Each tree is given a “vote” and the category with the most votes for a given child's matrix is the assigned EED category. Random forest modeling was performed using the rf package in R (www.r-project.org, Vienna, Austria) and default settings. Cross validation of models was done by removing 30 children from the testing and training sets, and testing the accuracy of the model in this naïve group. This method of random forest modeling was chosen because the desired output of the model was the categorical severity of EED, and fine-grained numerical L:M prediction was not felt to be clinically meaningful. RESULTS Study Subjects and L:M Testing Of the 798 children included in the study, 140 (18%), 524 (66%), and 134 (17%) had no, moderate, or severe EED, respectively (Table 1).

Modeling We chose random forest modeling as the machine-based learning method of modeling because it is the most appropriate and powerful for determining categorical outcomes, in our case no EED, moderate EED, and severe EED. Random forest modeling creates a set of computer-generated decision trees from a set independent variables, in our case the copy numbers of a set host transcripts, to designate the child into an EED category (10). Each tree is given a “vote” and the category with the most votes for a given child's matrix is the assigned EED category. Random forest modeling was performed using the rf package in R (www.r-project.org, Vienna, Austria) and default settings. Cross validation of models was done by removing 30 children from the testing and training sets, and testing the accuracy of the model in this naïve group. This method of random forest modeling was chosen because the desired output of the model was the categorical severity of EED, and fine-grained numerical L:M prediction was not felt to be clinically meaningful. RESULTS Study Subjects and L:M Testing Of the 798 children included in the study, 140 (18%), 524 (66%), and 134 (17%) had no, moderate, or severe EED, respectively (Table 1). TABLE 1 Characteristics of the rural Malawian children categorized by EED severity†

Modeling We chose random forest modeling as the machine-based learning method of modeling because it is the most appropriate and powerful for determining categorical outcomes, in our case no EED, moderate EED, and severe EED. Random forest modeling creates a set of computer-generated decision trees from a set independent variables, in our case the copy numbers of a set host transcripts, to designate the child into an EED category (10). Each tree is given a “vote” and the category with the most votes for a given child's matrix is the assigned EED category. Random forest modeling was performed using the rf package in R (www.r-project.org, Vienna, Austria) and default settings. Cross validation of models was done by removing 30 children from the testing and training sets, and testing the accuracy of the model in this naïve group. This method of random forest modeling was chosen because the desired output of the model was the categorical severity of EED, and fine-grained numerical L:M prediction was not felt to be clinically meaningful. RESULTS Study Subjects and L:M Testing Of the 798 children included in the study, 140 (18%), 524 (66%), and 134 (17%) had no, moderate, or severe EED, respectively (Table 1). TABLE 1 Characteristics of the rural Malawian children categorized by EED severity† Characteristic No EED L:M ≤ 0.15 (n = 140) Moderate EED 0.15 < L:M< 0.45 (n = 524) Severe EED L:M ≥ 0.45 (n = 134) P‡ Age, mo 37.7 ± 10.5 31.4 ± 11.3 25.4 ± 10.2 <0.001* Female, n (%) 65 (46%) 262 (50%) 57 (43%) 0.28 Mid-upper arm circumference, cm 15.2 ± 1.2 14.9 ± 1.2 14.4 ± 1.3 <0.001* Weight-for-height z score 0.28 ± 0.95 0.06 ± 0.91 −0.38 ± 1.05 <0.001* Height-for-age z score −2.5 ± 1.2 −2.3 ± 1.2 −2.1 ± 1.2 0.05 Stunted,§ n (%) 92 (66%) 307 (59%) 70 (52%) 0.08 Caregiver is mother, n (%) 131 (94%) 498 (95%) 127 (95%) 0.73 Father is alive, n (%) 132 (94%) 512 (98%) 132 (99%) 0.06 Siblings 3.7 ± 1.7 3.5 ± 1.8 2.8 ± 1.9 <0.001* Individuals that sleep in same room as child 3.1 ± 1.6 3.3 ± 1.4 3.2 ± 1.0 0.28 Home with a metal roof, n (%) 37 (26%) 114 (22%) 35 (26%) 0.36 Family owns bicycle, n (%) 83 (59%) 305 (58%) 68 (51%) 0.25 Animals sleep in house, n (%) 40 (29%) 219 (42%) 56 (42%) 0.014* Water from a clean source, n (%) 72 (51%) 376 (72%) 91 (68%) <0.001* Uses a pit latrine, n (%) 70 (50%) 314 (60%) 56 (42%) <0.001* Child does not use pit latrine or clean water, n (%) 31 (22%) 178 (34%) 67 (50%) <0.001* Household Food Insecurity Access Score|| 2.8 ± 3.9 3.2 ± 3.8 4.0 ± 3.7 0.03* Dietary diversity score¶ 4.6 ± 1.1 4.3 ± 1.2 4.0 ± 1.2 <0.001* Consumes animal source foods, times per day# 1.6 ± 1.4 1.6 ± 1.3 2.0 ± 1.8 0.012* Diarrhea in last days, n (%) 11 (8%) 101 (19%) 43 (32%) <0.001* Lactulose:mannitol 0.11 ± 0.04 0.27 ± 0.08 0.69 ± 0.25 <0.001* Lactulose, % excreted 0.2 ± 0.1 0.4 ± 0.2 0.7 ± 0.4 <0.001* Mannitol, % excreted 8.3 ± 4.9 6.8 ± 3.3 5.5 ± 2.6 <0.001* EED = environmental enteric dysfunction; SD = standard deviation.

8 0.012* Diarrhea in last days, n (%) 11 (8%) 101 (19%) 43 (32%) <0.001* Lactulose:mannitol 0.11 ± 0.04 0.27 ± 0.08 0.69 ± 0.25 <0.001* Lactulose, % excreted 0.2 ± 0.1 0.4 ± 0.2 0.7 ± 0.4 <0.001* Mannitol, % excreted 8.3 ± 4.9 6.8 ± 3.3 5.5 ± 2.6 <0.001* EED = environmental enteric dysfunction; SD = standard deviation. *P < 0.05. †Data are expressed as means ± SD for continuous measures or counts (percentages) for categorical measures. ‡For continuous characteristics P value calculated with one-way ANOVA with Tukey's correction and for categorical characteristics P value calculated using Chi-square test in a 2 × 3 table. Main effects and interactions were considered significant at P < 0.05. §Defined as height-for-age z score <−2. ||Household Food Insecurity Access Score measures the degree of food insecurity in the household in the past 4 weeks for the 9 food insecurity related conditions. Range 0 to 27 with higher scores representing more food insecurity (22). ¶Range 0 to 7, higher score more dietary diversity (23). #Animal source foods defined as foods containing meat, milk, fish, poultry, or eggs.

||Household Food Insecurity Access Score measures the degree of food insecurity in the household in the past 4 weeks for the 9 food insecurity related conditions. Range 0 to 27 with higher scores representing more food insecurity (22). ¶Range 0 to 7, higher score more dietary diversity (23). #Animal source foods defined as foods containing meat, milk, fish, poultry, or eggs. The HPLC analytical method used to determine lactulose and mannitol concentrations was validated in 115 urine samples using liquid chromatography-tandem mass spectrometry, 6 of 115 (5%) of the L:M values were discordant, defined as >25% difference in measurements using the 2 methods. Using all data points, the r values between the 2 methods were 0.88 for L:M and 0.92 for lactulose (P < 0.001). A Bland-Altman plot shows average differences between the 2 methods of 0.1% for L and 1% for M, which is excellent agreement (Supplemental Digital Content, Fig. 1).

erence in measurements using the 2 methods. Using all data points, the r values between the 2 methods were 0.88 for L:M and 0.92 for lactulose (P < 0.001). A Bland-Altman plot shows average differences between the 2 methods of 0.1% for L and 1% for M, which is excellent agreement (Supplemental Digital Content, Fig. 1). Clinical Associations With Categories of Environmental Enteric Dysfunction Age younger than 24 months, WHZ < 0, animals sleeping in the same room as the child, use of a potentially contaminated water source combined with the absence of a pit latrine in the household, and diarrhea 4 to 7 days before L:M testing are the characteristics associated with EED (Table 1). When 4 or more of these characteristics were present, only 3% of such children had no EED, whereas 46% had severe EED. These values contrast with the 18% and 17% background rates of no or severe EED observed in the study population, respectively (P = 0.0003). We could, however, find no combination of clinical characteristics that predict severe EED from either moderate EED or no EED with >65% sensitivity. L:M and Linear Growth L:M was correlated with ΔHAZ (Pearson correlation coefficient = −0.27 (P < 0.001) and Spearman correlation coefficient = −0.32 (P < 0.001)). Severe EED was associated with decreased ΔHAZ in the subsequent 3 months (Fig. 1). Linear regression modeling identified L:M as a significant predictor of ΔHAZ (Table 2).

r Growth L:M was correlated with ΔHAZ (Pearson correlation coefficient = −0.27 (P < 0.001) and Spearman correlation coefficient = −0.32 (P < 0.001)). Severe EED was associated with decreased ΔHAZ in the subsequent 3 months (Fig. 1). Linear regression modeling identified L:M as a significant predictor of ΔHAZ (Table 2). FIGURE 1 Change in height-for-age z scores in the 3 months after L:M testing. L:M values are categorized as no EED, moderate EED, or severe EED. Data expressed as mean ± SEM. Differences between any of the 3 categories were statistically significant, (P < 0.0001), using one-way ANOVA. ANOVA = analysis of variance; EED = environmental enteric dysfunction; HAZ = height-for-age z score; L:M = lactulose to mannitol; SEM = standard error mean. TABLE 2 Characteristics that predict change in height-for-age over subsequent 3 months∗ Model Standardized coefficients beta t P Constant −3.226 0.001 Lactulose:mannitol −0.140 −3.971 <0.001 Age, mo 0.147 3.913 <0.001 Height-for-age, z score −0.352 −10.429 <0.001 Weight-for-height, z score 0.197 5.705 <0.001 Household Food Insecurity Access Scale† −0.106 −3.009 0.003 *Determined by developing a stepwise, backward linear regression model (r = 0.564, P < 0.001), only covariates with P < 0.05 listed in table. All characteristics included in Table 1 were used in the prediction model. †Household Food Insecurity Access Score measures the degree of food insecurity in the household in the past 4 weeks for the 9 food insecurity related conditions. Range 0 to 27 with higher scores representing more food insecurity (22).

Model Standardized coefficients beta t P Constant −3.226 0.001 Lactulose:mannitol −0.140 −3.971 <0.001 Age, mo 0.147 3.913 <0.001 Height-for-age, z score −0.352 −10.429 <0.001 Weight-for-height, z score 0.197 5.705 <0.001 Household Food Insecurity Access Scale† −0.106 −3.009 0.003 *Determined by developing a stepwise, backward linear regression model (r = 0.564, P < 0.001), only covariates with P < 0.05 listed in table. All characteristics included in Table 1 were used in the prediction model. †Household Food Insecurity Access Score measures the degree of food insecurity in the household in the past 4 weeks for the 9 food insecurity related conditions. Range 0 to 27 with higher scores representing more food insecurity (22). Random Forest Models to Associate Host Fecal mRNA With L:M Eighteen transcripts of interest recently identified as being associated with L:M were evaluated for their association with EED (Table 3) (7). Random forest modeling identified 7 of these transcripts as important in models to predict EED. TABLE 3 Summary of 18 transcripts correlated with L:M, shaded are the 7 transcripts identified as significant predictors of L:M in random forest modeling

Random Forest Models to Associate Host Fecal mRNA With L:M Eighteen transcripts of interest recently identified as being associated with L:M were evaluated for their association with EED (Table 3) (7). Random forest modeling identified 7 of these transcripts as important in models to predict EED. TABLE 3 Summary of 18 transcripts correlated with L:M, shaded are the 7 transcripts identified as significant predictors of L:M in random forest modeling Gene symbol Gene description Primary function Mean ± SD (n) Transcript concentration*; median (25th, 75th percentiles) Spearman's r with L:M (P value of r) ACP1 Acts on tyrosine phosphorylated proteins, low-MW aryl and acyl phosphates. Isoform 3 does not possess phosphatase activity 0.009 ± 0.009 (378) 0.007 (0.005, 0.011) −0.104 (0.043) AQP9 Forms a transmembrane channel. Mediates passage of noncharged solutes including carbamides, polyols, purines, and pyrimidines Transporter activity 0.145 ± 0.318 (73) 0.065 (0.019, 0.150) 0.299 (0.01) BIRC3 Regulates caspases and apoptosis, modulates inflammatory signaling and immunity, mitogenic kinase signaling, and cell proliferation Inflammatory response 0.169 ± 0.177 (542) 0.120 (0.077, 0.196) −0.125 (0.004) CD53 Mediates signal transduction promoting cell development. Complexes with integrins. Mutations in this gene result in immunodeficiency Cell adhesion 0.076 ± 0.124 (324) 0.035 (0.013, 0.091) 0.168 (0.002) CDX1 Caudal type homeobox 1. Plays a role in the terminal differentiation of the intestine Intestinal differentiation 0.047 ± 0.346 (562) 0.027 (0.016, 0.042) −0.166 (<0.001) DECR1 Enzyme of β-oxidation. It participates in the metabolism of unsaturated fatty enoyl-CoA esters Fatty acid metabolism 0.041 ± 0.027 (85) 0.035 (0.022, 0.047) 0.220 (0.043) DEFA6 Has antimicrobial activity against Gram-negative and Gram-positive bacteria. Protects cells against infection with HIV-1 Viral response 0.090 ± 0.142 (443) 0.041 (0.016, 0.099) 0.118 (0.013) HLADRA Binds peptides from antigens that access the endocytic route of antigen presenting cells and presents them on the cell surface for T cells Adaptive immune response 0.220 ± 0.170 (567) 0.174 (0.103, 0.281) −0.125 (0.003) IFI30 Lysosomal thiol reductase reduces disulfide bonds, unfolds proteins destined for lysosomal degradation. Active in antigen processing Antigen processing 0.309 ± 0.611 (73) 0.168 (0.081, 0.288) 0.264 (0.024) LYZ Lysozymes have primarily a bacteriolytic function; those in tissues and body fluids are associated with the monocyte-macrophage system Bacterial response 0.080 ± 0.112 (324) 0.047 (0.022, 0.095) 0.218 (<0.001) MUC12 Mucin 12.

ntigen processing Antigen processing 0.309 ± 0.611 (73) 0.168 (0.081, 0.288) 0.264 (0.024) LYZ Lysozymes have primarily a bacteriolytic function; those in tissues and body fluids are associated with the monocyte-macrophage system Bacterial response 0.080 ± 0.112 (324) 0.047 (0.022, 0.095) 0.218 (<0.001) MUC12 Mucin 12. A protein in gastrointestinal mucous layer, involved in epithelial cell protection, cell adhesion, and epithelial cell growth. Stimulated by cytokines Epithelial barrier function 0.447 ± 0.499 (319) 0.294 (0.163, 0.539) −0.228 (<0.001) PIK3AP1 Signaling adapter in B-cell development. Links Toll-like receptor signaling to PI3K activation, reduces inflammatory cytokines Inflammatory response 0.199 ± 0.420 (77) 0.048 (0.022, 0.121) 0.249 (0.029) REG1A Acts as an inhibitor of spontaneous calcium carbonate precipitation. Associated with intestinal, brain, and pancreas regeneration Regeneration of epithelial cells 0.114 ± 0.268 (622) 0.042 (0.018, 0.107) 0.173 (<0.001) REG3A Bactericidal lectin that acts against Gram-positive bacteria and mediates bacterial killing by binding to surface-exposed carbohydrate moieties Bacterial response 0.123 ± 0.233 (344) 0.040 (0.015, 0.100) 0.183 (0.001) S100A8 Calprotectin, a cation-binding protein that regulates inflammation and immune response. Induces neutrophil chemotaxis and adhesion Innate immune response 0.979 ± 1.738 (550) 0.386 (0.154, 1.169) 0.096 (0.025) SELL Cell surface adhesion protein. Promotes initial tethering and rolling of leukocytes in endothelia Cell adhesion 0.052 ± 0.134 (80) 0.009 (0.003, 0.038) 0.295 (0.008) SI Sucrase isomaltase. A disaccharidase that plays an important role in carbohydrate digestion Carbohydrate digestion 0.114 ± 0.895 (753) 0.017 (0.008, 0.036) −0.104 (0.004) TNF Cytokine that binds to TNFRSF1A/TNFR1. Secreted by macrophages, potent pyrogen, promotes cell death. Induces IL-12 in dendritic cells Innate immune response 0.008 ± 0.015 (751) 0.004 (0.002, 0.008) −0.149 (<0.001) SD = standard deviation.

arbohydrate digestion 0.114 ± 0.895 (753) 0.017 (0.008, 0.036) −0.104 (0.004) TNF Cytokine that binds to TNFRSF1A/TNFR1. Secreted by macrophages, potent pyrogen, promotes cell death. Induces IL-12 in dendritic cells Innate immune response 0.008 ± 0.015 (751) 0.004 (0.002, 0.008) −0.149 (<0.001) SD = standard deviation. *Expressed as copies/copy GAPDH. A random forest model to identify children with severe EED was created using CDX1, HLA-DRA, MUC12, REG1A, S100A8, and TNF, and the model was 84% sensitive and 73% specific (n = 284, node size = 4, max node = 20, mtry = 3). Validation of model with 30 samples removed from the model creation exercises yielded 80% sensitivity and 72% specificity. A random forest model to discriminate children without EED from those with severe EED was created using TNF, HLA-DRA, MUC12, and CD53 and found to be 84% sensitive for severe EED and 83% sensitive for no EED (n = 284, node size = 4, max node = 20, mtry = 3). Validation of this model with 30 samples removed from the model creation exercises resulted in a prediction with 83% sensitivity for the identification of children with severe EED and 86% sensitivity for identification of children without EED. DISCUSSION In rural Malawians aged 12 to 61 months, increased gut permeability, as measured by L:M, is a predictor of linear growth faltering. Severe EED can be predicted by a small number of host fecal mRNAs using random forest modeling with 80% to 85% sensitivity.

A random forest model to identify children with severe EED was created using CDX1, HLA-DRA, MUC12, REG1A, S100A8, and TNF, and the model was 84% sensitive and 73% specific (n = 284, node size = 4, max node = 20, mtry = 3). Validation of model with 30 samples removed from the model creation exercises yielded 80% sensitivity and 72% specificity. A random forest model to discriminate children without EED from those with severe EED was created using TNF, HLA-DRA, MUC12, and CD53 and found to be 84% sensitive for severe EED and 83% sensitive for no EED (n = 284, node size = 4, max node = 20, mtry = 3). Validation of this model with 30 samples removed from the model creation exercises resulted in a prediction with 83% sensitivity for the identification of children with severe EED and 86% sensitivity for identification of children without EED. DISCUSSION In rural Malawians aged 12 to 61 months, increased gut permeability, as measured by L:M, is a predictor of linear growth faltering. Severe EED can be predicted by a small number of host fecal mRNAs using random forest modeling with 80% to 85% sensitivity. The primary limitation of the present study is that the use of the L:M test was not extended to younger children. Growth faltering in the first year of life is of great interest in the global health community, and there is often the assumption that EED plays a causal role infant stunting. Identification of a biomarker for EED in infants would be a powerful tool in elucidating the role of EED in stunting and identifying children who may benefit from an intervention to ameliorate stunting. L:M testing in such infants could be, however, problematic because lactulose can cause osmotic diarrhea and infants may be more compromised by the consequent to fluid loss. In addition, complete urine collections are more difficult in infants than toddlers. Another limitation is that the association of L:M and growth faltering is based on a single L:M measurement and a single linear growth measurement 3 months later, a more powerful methodology to implicate abnormal L:M in growth faltering would be to test a child every 3 months over 1 to 2 years and associate these multiple L:M measurements with change in length.

of L:M and growth faltering is based on a single L:M measurement and a single linear growth measurement 3 months later, a more powerful methodology to implicate abnormal L:M in growth faltering would be to test a child every 3 months over 1 to 2 years and associate these multiple L:M measurements with change in length. We are encouraged by the demonstration that host fecal transcripts can be used in a machine learning model to predict EED with >80% sensitivity. Stool collections are noninvasive. Our group has done extensive work investigating the transcriptome in EED, and we recognize that none transcript will serve as biomarker with sufficient sensitivity and specificity (24). This technology, built on iterative design using expert opinion and nonbiased high-density microarray nomination of candidate transcripts, however, enables adaptive mRNA biomarkers to be constructed on a cohort and geographic-specific basis. Moreover, these disease and population-tailored readouts can be determined on materials that are handled with minimal processing on site and uniform purification technology downstream.

nation of candidate transcripts, however, enables adaptive mRNA biomarkers to be constructed on a cohort and geographic-specific basis. Moreover, these disease and population-tailored readouts can be determined on materials that are handled with minimal processing on site and uniform purification technology downstream. A caveat regarding the diagnostic characterization of EED as used in the present study is that we focused on small bowel dysfunction as the canonical lesion in EED. The more general constellation of growth faltering and increased systemic inflammation in an asymptomatic child living in an unsanitary environment, which is sometimes described as environmental enteropathy (without the dysfunction component), certainly includes children with a spectrum range of L:M measurements. In such children the host fecal mRNAs identified in the present study may be less likely to constitute an adequate biomarker of the general environmental enteropathy syndrome. Their utility in identifying those with enteropathy who may progress to stunting, however, seems worthy of further pursuit.

of L:M measurements. In such children the host fecal mRNAs identified in the present study may be less likely to constitute an adequate biomarker of the general environmental enteropathy syndrome. Their utility in identifying those with enteropathy who may progress to stunting, however, seems worthy of further pursuit. The primary biomarker for EED is at present the L:M test. No other biomarker has been used as an outcome in a clinical trial to ameliorate EED or is associated with change in length (18,25–28). The L:M test stresses upper intestinal integrity; a large, inert osmotic load is added to the lumen and evidence of excessive paracellular leakage is sought. This theoretically sound test is compromised by physiologic and technical issues. The dual sugar load is likely to create fluid shifts between intestine tissue and gut lumen (reverse solute drag) (29), evoke an immunologic stress response (24), and temporarily alter gut microbial communities (30). For these reasons, repeated L:M tests performed on consecutive days may well yield different results. Moreover, the cumbersome nature of a several hour urine collection from a young child, which is required for a successful L:M test, is obviated with the use of host fecal transcripts. In addition, there is endogenous mannitol in human urine (31). Measurement of host fecal transcripts does not perturb gut biology, and for this reason, could be used to measure EED repeatedly and frequently to assess intestinal health.

uired for a successful L:M test, is obviated with the use of host fecal transcripts. In addition, there is endogenous mannitol in human urine (31). Measurement of host fecal transcripts does not perturb gut biology, and for this reason, could be used to measure EED repeatedly and frequently to assess intestinal health. In the present study, the cost of conducting an L:M test was about $90, whereas the host transcript test with 4 transcripts costs $35. We acknowledge that a normal L:M value in a variety of populations are not available, nor is an estimate of the effect size of a change in L:M on linear growth (4). In part this is because the L:M test is not conducted in a uniform manner, different doses of sugars are used, timing of urine collections vary, and assay technologies differ. These issues necessitate that our categories of no, moderate and severe EED be defined empirically. These categories are, however, similar to other studies that have used the L:M test (4), and are useful in understanding the causes of stunting in rural African children.

collections vary, and assay technologies differ. These issues necessitate that our categories of no, moderate and severe EED be defined empirically. These categories are, however, similar to other studies that have used the L:M test (4), and are useful in understanding the causes of stunting in rural African children. An alternative noninvasive approach for EED is to measure fecal host proteins (32). Although this has been described in the literature, it has not been used as an outcome measure in any clinical studies. Detection of a protein in feces is contingent upon significant quantities of the protein being secreted into the extracellular space and the secreted protein remaining largely intact as it passes along the intestinal tract, which is unusual given the abundance of proteases present in the intestinal lumen from the host and its microbiota. α1-Antitrypsin, calprotectin, myeloperoxidase, neopterin, and lithostathine 1β are proteins that are suitable for measurement in fecal samples and have been described as EED biomarkers (32–37). Although commercial enzyme immunoassays are available for each of these proteins, their use is limited by variable reliance on polyclonal antibodies (which reduce specificity and may contribute to batch to batch variability), cost, the mass of specimen material needed, and differing dilution and diluent/buffer conditions. ddPCR uses amplification of transcripts, which detects very few copy numbers of the target nucleic acid. Moreover, this technology permits normalization of transcript quantity to a housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase, which serves as an “internal standard” for every sample, thereby controlling for nonspecific mRNA degradation.

scripts, which detects very few copy numbers of the target nucleic acid. Moreover, this technology permits normalization of transcript quantity to a housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase, which serves as an “internal standard” for every sample, thereby controlling for nonspecific mRNA degradation. Additional research is needed to determine whether this panel of transcripts will predict EED in younger populations and populations different from rural African children. Analyses of fecal host transcripts should be included in clinical trials to ameliorate EED, as they may serve as a noninvasive biomarker. Supplementary Material Supplemental Digital Content M.I.O and N.S. are co-first authors. The present study was supported by the Bill and Melinda Gates Foundation. P.I.T. is also supported by the NIH Biobank Core of the Washington University DDRCC (P30DK052574). The sponsors played no role in the study design, data collection, data analyses, interpretation, or decision to submit for publication. P.I.T. is a co-inventor of a new method to detect increased gut permeability. The remaining authors report no conflicts of interest.

vitamin D status at 6 months of age. There were no differences in LAZ, WLZ, or WAZ trajectories by vitamin D status at 6 months of age in univariate and multivariate analyses (P values >0.05). There was also no association of vitamin D status at 6 months of age with incident stunting, underweight, or wasting (Table 2). FIGURE 3 Mean length-for-age z score (LAZ) (panel A), weight-for-length z score (WLZ) (panel B), and weight-for-age z score (WAZ) (panel C) growth curves stratified by 6 month 25(OH)D concentration. There was no significant difference in trajectory of LAZ, WLZ, or WAZ by 6-month vitamin D status categories (P values for difference in trajectory >0.05). 25(OH)D = 25-hydroxyvitamin D. Vitamin D and Morbidity Table 3 presents the relationship of vitamin D status at 6 weeks and 6 months of age with the incidence of diarrhea, upper respiratory tract infection (URI), acute lower respiratory tract infection (ALRI), and clinical malaria. In multivariate analyses, there was no statistically significant association between vitamin D status at 6 weeks or 6 months of age with the incidence of diarrhea, URI, or clinical malaria (P > 0.05). In multivariate models, infants with 25(OH)D concentrations of 20 to 29.9 ng/mL at 6 months had decreased risk of ALRI as compared with infants with concentrations ≥30 ng/mL (incidence rate ratio 0.67; 95% CI 0.50–0.91).

at 6 weeks or 6 months of age with the incidence of diarrhea, URI, or clinical malaria (P > 0.05). In multivariate models, infants with 25(OH)D concentrations of 20 to 29.9 ng/mL at 6 months had decreased risk of ALRI as compared with infants with concentrations ≥30 ng/mL (incidence rate ratio 0.67; 95% CI 0.50–0.91). DISCUSSION In this prospective cohort study of Tanzanian HIV-unexposed infant, we determined that vitamin D deficiency, defined by 25(OH)D levels <20 ng/mL, was present in >75% of infants at 6 weeks of age but declined to <25% by 6 months. Exclusively breastfed infants were at increased risk of vitamin D deficiency at 6 weeks of age and there was a consistent seasonal relationship at both 6 weeks and 6 months. There was no association of vitamin D status at 6 weeks or 6 months of age with LAZ and WLZ trajectory, nor with the incidence of stunting, wasting, or underweight. We also found no relationship of vitamin D status at 6 week or 6 months with incidence of diarrhea, URI, or clinical malaria, but infants with 25(OH)D concentrations ≥30 ng/mL at 6 months had increased risk of incident ALRI as compared with 20 to 29.9 ng/mL.

higher vitamin D concentrations during infancy to 3 years of age were associated with leaner body composition (28). Studies evaluating the association of child 25(OH)D status with body composition are warranted in LMICs given rapidly increasing rates of pediatric obesity and noncommunicable diseases in adulthood (29). This study has a few limitations. First, the moderately small sample size may have led to inadequate power to detect modest effect sizes on growth and incidence of morbidities. Nevertheless, the direction of the point estimates and bounds of 95% confidence intervals suggest vitamin D supplementation is not likely to provide a large beneficial effect on growth or morbidity in the study population. Due to the observational nature of the study, we also cannot rule out residual confounding by SES, child feeding method, and other factors. Of note, we did not have data on maternal vitamin D status at 6 weeks or 6 months postpartum.

What Is KnownHistorically, despite limited clinical evidence and knowledge on their mode of action, fermented formulae are used to promote digestibility and tolerance and are considered safe for use in infants (with respect to growth), although new studies are warranted. Infant formulae containing the specific prebiotic mixture of 90% short-chain galacto-oligosaccharides and 10% long-chain fructo-oligosaccharides (scGOS/lcFOS; 9:1) stimulate a closer to human milk gut microbiota, stool characteristics, have a beneficial effect on immune function and are considered safe for use in infants. To our knowledge, clinical evidence on the safety of infant formulae combining scGOS/lcFOS (9:1) and fermented formula is lacking. What Is NewPartly fermented infant formulae containing scGOS/lcFOS (9:1) can be considered as nutritionally adequate and safe for use in healthy, term infants. A beneficial effect of scGOS/lcFOS (9:1) on gut microbiota and stool characteristics was confirmed, irrespective of presence or dosage of fermented formula. Exclusive human milk is the preferred feeding for all term newborn infants and provides a complete supply of nutrients to support growth and development in early life. In addition, human milk contains bioactive components that beneficially affect intestinal health, gut microbial colonization, and immune maturation (1–3). Because human milk feeding may not always be possible, human milk substitutes should provide nutritional and functional properties as close as possible to those of human milk.

n, human milk contains bioactive components that beneficially affect intestinal health, gut microbial colonization, and immune maturation (1–3). Because human milk feeding may not always be possible, human milk substitutes should provide nutritional and functional properties as close as possible to those of human milk. Early nutrition plays an important role in the development and functioning of the gastrointestinal (GI) tract (4,5). Historically, fermented and/or acidified infant milk formulae have been used to promote digestibility and tolerance of formula. Only limited clinical evidence is, however, available to substantiate these suggested health benefits (6–10). Fermented formulae may reduce frequency and severity of gut discomfort (8), reduce the severity of infectious diarrhea, that is, fewer cases of dehydration or medical consultation (9) and stimulate the presence of intestinal bifidobacteria in healthy young infants (10). The ESPGHAN Committee on Nutrition, based on the limited evidence, cautiously reported that no safety concerns are expressed for the use of fermented formulae, although they emphasized more studies are warranted (7).

consultation (9) and stimulate the presence of intestinal bifidobacteria in healthy young infants (10). The ESPGHAN Committee on Nutrition, based on the limited evidence, cautiously reported that no safety concerns are expressed for the use of fermented formulae, although they emphasized more studies are warranted (7). The development and maturation of the GI tract is accompanied by the dynamic enteric process of microbiota development, considered crucial for healthy infant development (11). Early nutrition has a strong effect on microbiota development: breast-fed infants typically have a microbiota dominated by bifidobacteria, whereas nonprebiotic supplemented formula-fed infants have a more diverse microbiota (12). Inspired by the molecular size distribution of the nondigestible oligosaccharides present in human milk (13), a specific mixture of 90% short-chain galacto-oligosaccharides and 10% long-chain fructo-oligosaccharides (scGOS/lcFOS; 9:1) was developed (14,15). Formulae containing this specific prebiotic mixture were shown to stimulate intestinal colonization with bifidobacteria resulting in beneficial effects on immune function (16–18). The addition of this prebiotic mixture prevents constipation, which is known to be more common in formula-fed than breast-fed infants (19). Safety and tolerance of formulae containing scGOS/lcFOS (9:1) has been demonstrated in several studies, with respect to growth (20–22) and digestive function (22–25). None of the studies included the prebiotic mixture in combination with fermented infant formula.

more common in formula-fed than breast-fed infants (19). Safety and tolerance of formulae containing scGOS/lcFOS (9:1) has been demonstrated in several studies, with respect to growth (20–22) and digestive function (22–25). None of the studies included the prebiotic mixture in combination with fermented infant formula. Supplementation of fermented formula with the specific prebiotic mixture scGOS/lcFOS could have a complementary, beneficial effect on the GI function of infants. First, the nutritional safety and adequacy of this new nutritional concept should be demonstrated following stringent evaluation (26–28). In our study, safety was evaluated with daily weight gain as primary outcome parameter, and in addition, length gain, head circumference gain, and (serious) adverse events (AEs) were monitored. As an exploratory outcome, fecal physiological and microbial parameters were evaluated to confirm the beneficial effects of scGOS/lcFOS.

dy, safety was evaluated with daily weight gain as primary outcome parameter, and in addition, length gain, head circumference gain, and (serious) adverse events (AEs) were monitored. As an exploratory outcome, fecal physiological and microbial parameters were evaluated to confirm the beneficial effects of scGOS/lcFOS. SUBJECTS AND METHODS Subjects Subjects were healthy infants recruited from mothers who could not or had chosen not to (continue to) breast-feed. Inclusion criteria were gestational age between 37 weeks and 42 weeks, birth weight between 2.5 and 4.5 kg, and postnatal age ≤28 days. Infants were excluded from the study if they had a congenital condition or illness that could interfere with the study or if they had a known or increased risk for cow's milk allergy. Other exclusion criteria were maternal gestational diabetes, participation in another clinical trial or investigator's uncertainty about the willingness or ability of the parents to comply with the protocol requirements. The study was conducted according to ICH-GCP principles, and in compliance with the principles of the Declaration of Helsinki and with the local laws and regulations of the country where the study was performed. All participating centers obtained approval of their independent local Ethical Review Board. Written informed consent was obtained from all parent(s)/guardian, ages 18 years or older, before enrolment to the study. This trial was registered in the Dutch Trial Register (NTR 2521).

The study was conducted according to ICH-GCP principles, and in compliance with the principles of the Declaration of Helsinki and with the local laws and regulations of the country where the study was performed. All participating centers obtained approval of their independent local Ethical Review Board. Written informed consent was obtained from all parent(s)/guardian, ages 18 years or older, before enrolment to the study. This trial was registered in the Dutch Trial Register (NTR 2521). Participating Centers The present study was conducted in a total of 24 study centers in France (n = 7; CHU of Dijon, American Memorial Hospital in Reims, CHU La conception in Marseille, CHU of Angers, CHU of Besancon, and a private pediatrician clinic in Essey les Nancy and in Floirac), Belgium (n = 10; UZ Brussel, Jessa Ziekenhuis Hasselt, private practice in Asse, Sainte Elisabeth Clinique in Namur, Heilig Hart Hospital in Roeleare, Aalsters Stedelijk Ziekenhuis in Aalst, Imelda Hospital in Bonheiden, GZA Sint Augustinus in Wilrijk, AZ Sint Blasius in Dendermonde, and AZ Sint Vicentius in Antwerp), and Ireland (n = 7; 2 tertiary referral hospitals: Cork University Hospital, National Maternity Hospital, Dublin, and 5 primary care centers: Medical Center and The Palms surgeries in Gorey, Meadowcroft Surgery in Wicklow, Slaney Medical Center in Enniscorthy, and Town Hall Surgery in Bray).

Sint Vicentius in Antwerp), and Ireland (n = 7; 2 tertiary referral hospitals: Cork University Hospital, National Maternity Hospital, Dublin, and 5 primary care centers: Medical Center and The Palms surgeries in Gorey, Meadowcroft Surgery in Wicklow, Slaney Medical Center in Enniscorthy, and Town Hall Surgery in Bray). Trial Design The present study was a prospective, double-blind, controlled, randomized, parallel-group, multicenter controlled equivalence trial. Infants, whose parents intended to feed their infants formula, were assigned to 1 of the 4 groups, using a computerized random-number generator with sex and study site strata. Formulae were coded by the sponsor and both the investigators and the infant's parents were blinded to the formula. Baseline measurements were taken at enrolment and parents received the assigned infant formula with written preparation instructions, were advised to feed infants ad libitum, and were provided with recommendations for product intake volumes depending on the infant's weight.

infant's parents were blinded to the formula. Baseline measurements were taken at enrolment and parents received the assigned infant formula with written preparation instructions, were advised to feed infants ad libitum, and were provided with recommendations for product intake volumes depending on the infant's weight. During the intervention period infants were to be fed ad libitum exclusively with their allocated formula starting on the day of enrolment (0–28 days of age) until 17 weeks of age. The study consisted of a baseline visit and 4 subsequent hospital visits at 4, 8, 13, and 17 weeks of postnatal age. In addition, a follow-up phone call was performed 2 weeks after the last study visit. Parents were provided with diaries to record formula intake, GI symptoms, crying, sleeping, and stool characteristics every day during the week before the study visits. At each visit, investigators took anthropometric measures, reviewed compliance (based on evaluation of diaries), provided study product, and assessed the occurrence and outcome of (serious) AEs and use of concomitant medication and nutritional supplements by open questioning and/or diagnosis. Furthermore, fecal samples were collected by the parents either during or shortly after the baseline visit, and just before the final visit at their infant's age of 17 weeks for analysis of the fecal physiological parameters and microbiological composition.

During the intervention period infants were to be fed ad libitum exclusively with their allocated formula starting on the day of enrolment (0–28 days of age) until 17 weeks of age. The study consisted of a baseline visit and 4 subsequent hospital visits at 4, 8, 13, and 17 weeks of postnatal age. In addition, a follow-up phone call was performed 2 weeks after the last study visit. Parents were provided with diaries to record formula intake, GI symptoms, crying, sleeping, and stool characteristics every day during the week before the study visits. At each visit, investigators took anthropometric measures, reviewed compliance (based on evaluation of diaries), provided study product, and assessed the occurrence and outcome of (serious) AEs and use of concomitant medication and nutritional supplements by open questioning and/or diagnosis. Furthermore, fecal samples were collected by the parents either during or shortly after the baseline visit, and just before the final visit at their infant's age of 17 weeks for analysis of the fecal physiological parameters and microbiological composition. Study Formulae Each formula was powdered infant formula providing complete nutritional support for infants in the first 6 months of life. The formulae were isocaloric and contained per 100 mL a similar amount of 66 kcal energy, 1.35 g protein, 8.2–8.4 g carbohydrate, 3.0–3.1 g lipids, and vitamins and minerals according to Directive 2006/141/EC. The products varied only in the amount of formula produced by a specific fermentation process using Bifidobacterium breve and Streptococcus thermophiles (this process is called Lactofidus) and the presence of prebiotics, that is, 90% short-chain galacto-oligosaccharides and 10% long-chain fructo-oligosaccharides (0.8 g per 100 mL; Table 1). The fermentation process (Lactofidus) is followed by mild heat treatment. In total, 4 formulae were tested (i) a nonfermented commercially available formula with prebiotics (scGOS/lcFOS), (ii) a formula containing 15% fermented formula and prebiotics (scGOS/lcFOS + 15% fermented formulae [FERM]), (iii) a formula containing 50% fermented formula and prebiotics (scGOS/lcFOS + 50% FERM), and (iv) a commercially available formula containing 50% fermented formula without prebiotics (50% FERM). Hence, formula ii, iii, and iv are composed of a mixture of fermented and nonfermented infant formula in a ratio of 15:85 (ii) or 50:50 (iii, iv). The rationale to supplement prebiotics to a product with a level of 50% fermented formula was based on the previously observed beneficial effects on gut comfort (8). The level of 15% fermented formula was included to assess a potential dose dependency given the unknown effect on safety and tolerance of the newly developed formulae and has been shown to be well-tolerated (29). All formulae had a similar taste, smell, and color and were manufactured according to good manufacturing practices by Nutricia (Steenvoorde, France). The study products were to be stored at the sites at a secure and limited access storage area protected from extremes of light, temperature, and humidity.

tolerated (29). All formulae had a similar taste, smell, and color and were manufactured according to good manufacturing practices by Nutricia (Steenvoorde, France). The study products were to be stored at the sites at a secure and limited access storage area protected from extremes of light, temperature, and humidity. TABLE 1 Composition of the intervention products (per 100 mL) Per 100 mL scGOS/lcFOS scGOS/lcFOS +15% FERM scGOS/lcFOS +50% FERM 50% FERM scGOS/lcFOS 9:1, g 0.8 0.8 0.8 – Fermented formula* – 15% 50% 50% Energy, kcal 66 66 66 66 Fat, g 3.0 3.0 3.0 3.1 Protein, g 1.35 1.35 1.35 1.36 Carbohydrates, g 8.2 8.2 8.2 8.4 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. *Fermented by a specific combination of Bifidobacterium breve and Streptococcus thermophilus. Measurements The primary outcome was weight gain per day from study entry to 17 weeks of age, that is, from ≤28 to 119 days of age. The secondary outcomes were length, head circumference, and mid-upper arm circumference. For safety assessment, the number, type, and severity of AEs were monitored and the use of medication. In addition, fecal physiological parameters and microbial composition were studied.

f age, that is, from ≤28 to 119 days of age. The secondary outcomes were length, head circumference, and mid-upper arm circumference. For safety assessment, the number, type, and severity of AEs were monitored and the use of medication. In addition, fecal physiological parameters and microbial composition were studied. At each visit, the mean weight for each infant was registered by weighing them twice naked, on calibrated electronic scales, and in case of >100 g deviation an additional measurement was performed. The mean supine length of infants was registered by measuring them twice using a standard measuring board, and in case of >5 mm deviation an additional measurement was performed. A nonstretchable slotted insertion tape was used to measure head circumference in duplicate (or 3 times if deviation of >2 mm between measurements). The repeated measures in 1 infant were performed by the same investigator, if possible, using the same equipment. In case a third measurement was required, the 2 measures closest together were averaged as outcome measurement.

asure head circumference in duplicate (or 3 times if deviation of >2 mm between measurements). The repeated measures in 1 infant were performed by the same investigator, if possible, using the same equipment. In case a third measurement was required, the 2 measures closest together were averaged as outcome measurement. Diaries, filled in by the parents during the 7-day period before each visit, registered information on quantity of formula intake. In the feeding record, the number of feeds and the amount of formula prepared and left over per feed was registered. During the 17-week study period, (serious) AEs were documented by the investigators at each visit. Details recorded were start and stop date of the event, its severity, and actions that were taken. Moreover, the investigators documented (probability of) any relation with the study product. Fecal Parameters Stool samples were collected by parents and brought to the study site for storage at or below –18°C for a maximum of 4 months. Thereafter, the samples were transported to Nutricia Research in insulated containers containing solid CO2 (dry ice) and stored at –80°C. Fecal parameters were determined for samples of visit 1 (baseline) and visit 5 (end of intervention period) for each study arm. Only samples of a subgroup of subjects were analyzed, which had a complete set of stool samples (both visits) with sufficient amount of stool for all analyses. In addition, samples from infant that used any systemic antibiotics any time after birth or that used thickeners added to formula during the study were excluded from analyses.

a subgroup of subjects were analyzed, which had a complete set of stool samples (both visits) with sufficient amount of stool for all analyses. In addition, samples from infant that used any systemic antibiotics any time after birth or that used thickeners added to formula during the study were excluded from analyses. In the selected set of fecal samples the effect of the infant formulae was assessed on the following physiological and microbial parameters: pH, short-chain fatty acid (SCFA) levels (ie, acetate, propionate, butyrate, isobutyrate, valerate, and isovalerate), d- and l-lactate, secretory immunoglobulin A (SIgA), calprotectin, and presence of Clostridium difficile. The quantification methodology of these parameters have been described in more detail previously (25,30,31), except for calprotectin and presence of C. difficile. Human calprotectin levels in the stool samples were quantified with the Bühlmann Calprotectin ELISA kit (Bühlmann Laboratories AG, Schönenbuch, Switzerland) according to the manufacturer's protocol. The presence of C. difficile was detected with quantitative polymerase chain reaction utilizing primers 16S-Cldif -F (5′-GCA ACG CGA AGA ACC TTA CCT A-3′) and 16S-Cldif -R (5′-GAA GGG AAC TCT CCG ATT AAG GA-3′) in conjunction with probe 16S-Cldif (5′-VIC-TGA CAT CCC AAT GAC A-NFQ-MGB-3′), which is labeled with the reporter dye VIC and with the quencher NFQ-MGB (Applied Biosystems, Bleiswijk, The Netherlands). Genomic DNA of C. difficile LMG 21717 was used for the standard curve. The assay was performed with 1× TaqMan universal master mix (Applied Biosystems), 0.4 μmol/L of both primers, 0.2 μmol/L probe, and DNA extract in a final volume of 25 μL. The amplification program included an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of denaturation at 95°C for 15 seconds and primer annealing and extension at 60°C for 60 seconds.

(Applied Biosystems), 0.4 μmol/L of both primers, 0.2 μmol/L probe, and DNA extract in a final volume of 25 μL. The amplification program included an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of denaturation at 95°C for 15 seconds and primer annealing and extension at 60°C for 60 seconds. Statistics During statistical analysis 3 comparisons of interest have been investigated. The effect of fermented formula is assessed by comparing scGOS/lcFOS + 15% FERM and scGOS/lcFOS + 50% FERM groups with the scGOS/lcFOS group. The effect of prebiotic addition is assessed by comparing the scGOS/lcFOS + 50% FERM group versus the 50% FERM group. The primary objective of the study was to test for equivalence of weight gain using the predefined margins of equivalence of 0.5 standard deviation (SD) of the difference in weight gain per day from randomization until 17 weeks of age. We set a minimum of 3 g/day (28) and a maximum value of 5 g/day for the equivalence margin to be used as clinically relevant margins in case 0.5 SD was exceeding these margins. To conclude equivalence, the 2-sided 90% confidence intervals for the differences in mean weight gain should lie entirely between –0.5 SD and +0.5 SD margins. This margin was also used to conclude equivalence for length gain and head circumference gain. The required sample size for 2 one-sided statistical testing using α = 0.05 and a power = 0.80 was 70 infants per intervention group. Allowing for a drop-out rate of 35%, a total of 432 infants (108 per group) had to be enrolled. The equivalence analysis was performed using parametric growth curves, this model describes the development of growth parameters (ie, weight) over time by a second-order polynomial curve correcting for sex, site was taken into account as a random effect, and each subject's intercept and slope were taken into account as random effects. Sensitivity analyses to evaluate robustness of results were performed using analysis of covariance on the 17 weeks of age measurement with the randomization measurement as covariate and a mixed model using the age variable as a categorical variable. In addition, to compare the intervention groups with the WHO Child Growth Standards of breastfed infants, an analysis of growth parameter z scores using WHO growth trajectories (32) were performed by using a mixed model with adjustment of baseline z score.

a mixed model using the age variable as a categorical variable. In addition, to compare the intervention groups with the WHO Child Growth Standards of breastfed infants, an analysis of growth parameter z scores using WHO growth trajectories (32) were performed by using a mixed model with adjustment of baseline z score. In any other analyses, for continuous data 2 sample t tests were used and Wilcoxon rank sum tests (WR) were used in case of violation of normality assumption and/or presence of outliers. Categorical response parameters were analyzed by using Chi-square tests (C; Fisher exact tests [FEs] in case sparse cells occurred). The statistical analysis was performed by Nutricia Research using SAS (SAS Enterprise Guide 4.3 or higher) for Windows (SAS Institute Inc., Cary, NC). In the intention-to-treat (ITT) analysis, the data of all infants, except for those erroneously randomized were to be used. In the per-protocol (PP) analysis, eligibility of data was assessed on visit level, that is, a subject could be part of the PP population at baseline and visits 2 and 3, but excluded from the PP population at visits 4 and 5 if a protocol violation occurred after visit 3. In case of major protocol deviations, for example, if an inclusion criterion for birth weight was not met or lack of postbaseline visits, complete subjects were excluded from the PP dataset. Fecal composition data were evaluated using SigmaStat Statistical Analysis Software (Systat Software Inc., San Jose, CA). If <30% of the values were below limit of detection (BLD) or—in case of the quantitative polymerase chain reaction assays—not quantifiable (NQ), the value BLD was replaced by (detection limit/2) and the value NQ was replaced by (detection limit + limit of quantification)/2, after which WRs were performed. If >30% of the values were BLD or NQ, value BLD was replaced by “0” and the other values (including “NQ”) were replaced by “1,” after which P values were calculated using Chi-square test to assess differences in prevalence of the corresponding parameter.

mit + limit of quantification)/2, after which WRs were performed. If >30% of the values were BLD or NQ, value BLD was replaced by “0” and the other values (including “NQ”) were replaced by “1,” after which P values were calculated using Chi-square test to assess differences in prevalence of the corresponding parameter. RESULTS Subject Characteristics From the total of 432 randomized subjects, included between October 2010 and September 2012, 16 subjects were excluded from the all-subjects-treated (AST) population because they did not consume any study product. One subject was diagnosed with hypothyroidism and was considered as not healthy, erroneously randomized, and excluded from ITT analysis (Fig. 1). A total of 276 subjects completed the study until 17 weeks of age, whereas 155 (36%) subjects dropped-out from the study prematurely. The number of subjects that completed the study was not apparently different between groups, although the number tended to be higher in the group combining scGOS/lcFOS with 50% FERM compared to that with 50% FERM only (FE P = 0.091). The most common reason for drop-out was an (serious) AE (n = 72), other reasons were withdrawal of informed consent (n=54), lost to follow-up (n = 15), protocol violation (n = 1), or other reasons (n = 13). There were no differences in reasons for drop-out between groups. At baseline, no differences in subject characteristics, including sex ratio, of the PP intervention groups were observed, except for a higher number of first-born infants in the scGOS/lcFOS with 50% FERM compared with the 50% FERM only group (FE P = 0.014; Table 2), which is considered to be a statistically significant difference by chance and without clinical relevance. Baseline anthropometric measures were not statistically different between treatment groups (Table 2).

st-born infants in the scGOS/lcFOS with 50% FERM compared with the 50% FERM only group (FE P = 0.014; Table 2), which is considered to be a statistically significant difference by chance and without clinical relevance. Baseline anthropometric measures were not statistically different between treatment groups (Table 2). FIGURE 1 Disposition of study subjects per intervention group. ASR = all subjects randomized; FERM = fermented infant formula; ITT = intention-to-treat; PP = per protocol; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. TABLE 2 Demographics and baseline characteristics of infants per intervention group of the per protocol population (n = 298)

FIGURE 1 Disposition of study subjects per intervention group. ASR = all subjects randomized; FERM = fermented infant formula; ITT = intention-to-treat; PP = per protocol; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. TABLE 2 Demographics and baseline characteristics of infants per intervention group of the per protocol population (n = 298) Statistic scGOS/lcFOS (N = 75) scGOS/lcFOS +15% FERM (N = 79) scGOS/lcFOS +50% FERM (N = 79) 50% FERM (N = 65) Sex F Male n (%) 39 (52.0%) 37 (46.8%) 39 (49.4%) 33 (50.8%) Female n (%) 36 (48.0%) 42 (53.2%) 40 (50.6%) 32 (49.2%) Mode of birth F Natural n (%) 61 (81.3%) 60 (75.9%) 58 (73.4%) 52 (80.0%) C-section n (%) 14 (18.7%) 19 (24.1%) 21 (26.6%) 13 (20.0%) Ethnicity F Caucasian/White n (%) 67 (89.3%) 71 (89.9%) 72 (91.1%) 61 (93.8%) Black n (%) 2 (2.7%) 1 (1.3%) 3 (3.8%) 1 (1.5%) Asian n (%) 0 (0.0%) 1 (1.3%) 0 (0.0%) 0 (0.0%) Combination n (%) 3 (4.0%) 2 (2.5%) 3 (3.8%) 0 (0.0%) Other n (%) 3 (4.0%) 4 (5.1%) 1 (1.3%) 3 (4.6%) Country F Belgium n (%) 28 (37.3%) 24 (30.4%) 25 (31.6%) 22 (33.8%) France n (%) 26 (34.7%) 34 (43.0%) 33 (41.8%) 25 (38.5%) Ireland n (%) 21 (28.0%) 21 (26.6%) 21 (26.6%) 18 (27.7%) Order of birth F First n (%) 28 (37.3%) 29 (36.7%) 34 (43.0%)* 22 (33.8%)* Second n (%) 31 (41.3%) 34 (43.0%) 36 (45.6%) 27 (41.5%) Third n (%) 14 (18.7%) 12 (15.2%) 6 (7.6%) 16 (24.6%) >Third n (%) 2 (2.7%) 4 (5.1%) 3 (3.8%) 0 (0.0%) Age at first visit, day W Median (min-max) 4 (0–28) 4 (0–28) 6 (0–28) 5 (0–28) Gestational age, wk W Median (min-max) 39 (37–41) 40 (37–41) 40 (37–41) 39 (37–41) Weight at birth, g T Mean ± SD 3336 ± 429 3378 ± 358 3370 ± 368 3444 ± 377 Length at birth, cm W Median (min-max) 50 (45–54) 50 (46–54) 50 (45–53) 50 (46–54) Maternal age, yr T Mean ± SD 29.4 ± 4.9 30.3 ± 5.2 29.5 ± 5.1 29.5 ± 4.9 Maternal BMI, kg/m2 W Median (min-max) 24 (17–43) 24 (20–45) 25 (16–39) 25 (16–36) Paternal age, yr T Mean ± SD 32.2 ± 5.4 32.6 ± 6.2 32.1 ± 5.6 32.7 ± 5.3 Paternal BMI, kg/m2 W Median (min-max) 25 (18–40) 26 (18–34) 26 (19–36) 26 (19–36) BMI = body mass index; 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides; SD = standard deviation.

5 (18–40) 26 (18–34) 26 (19–36) 26 (19–36) BMI = body mass index; 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides; SD = standard deviation. *Values are significantly different. For the analysis of continuous data 2 sample t tests (T) were used and Wilcoxon rank sum tests (W) were used in case of violation of normality assumption and/or presence of outliers. Categorical response parameters were analyzed by using Fisher exact tests (F). Formula Consumption The feeding history of subjects before the baseline visit, (any) breast-feeding and/or type of formula feeding, was not significantly different between groups in the PP population. In general, all study formulae were well accepted by the infants indicated by the high parental evaluation score (median score of 9 on evaluation scale 1–10 for all formulae; data not shown). The study product intake was not significantly different between subjects of each intervention group throughout the study (Table 3). TABLE 3 Average daily product intake (mL/day) during the intervention period of the per protocol population (means ± standard deviation)

Formula Consumption The feeding history of subjects before the baseline visit, (any) breast-feeding and/or type of formula feeding, was not significantly different between groups in the PP population. In general, all study formulae were well accepted by the infants indicated by the high parental evaluation score (median score of 9 on evaluation scale 1–10 for all formulae; data not shown). The study product intake was not significantly different between subjects of each intervention group throughout the study (Table 3). TABLE 3 Average daily product intake (mL/day) during the intervention period of the per protocol population (means ± standard deviation) scGOS/lcFOS (n = 75) scGOS/lcFOS +15% FERM (n = 79) scGOS/lcFOS +50% FERM (n = 79) 50% FERM (n = 65) Visit 2 (n) 699 ± 124 (67) 685 ± 131 (72) 697 ± 111 (69) 728 ± 129 (60) Visit 3 (n) 772 ± 109 (65) 768 ± 140 (66) 782 ± 161 (70) 825 ± 155 (56) Visit 4 (n) 827 ± 151 (59) 817 ± 139 (65) 830 ± 129 (67) 867 ± 159 (54) Visit 5 (n) 852 ± 123 (55) 859 ± 128 (62) 878 ± 137 (62) 928 ± 159 (52) Diaries were included in the analysis if product intake was recorded for at least 3 days. The data were analyzed using a t test for each comparison of interest. 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides.

scGOS/lcFOS (n = 75) scGOS/lcFOS +15% FERM (n = 79) scGOS/lcFOS +50% FERM (n = 79) 50% FERM (n = 65) Visit 2 (n) 699 ± 124 (67) 685 ± 131 (72) 697 ± 111 (69) 728 ± 129 (60) Visit 3 (n) 772 ± 109 (65) 768 ± 140 (66) 782 ± 161 (70) 825 ± 155 (56) Visit 4 (n) 827 ± 151 (59) 817 ± 139 (65) 830 ± 129 (67) 867 ± 159 (54) Visit 5 (n) 852 ± 123 (55) 859 ± 128 (62) 878 ± 137 (62) 928 ± 159 (52) Diaries were included in the analysis if product intake was recorded for at least 3 days. The data were analyzed using a t test for each comparison of interest. 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. Infant Growth Patterns In all comparisons of interest, the 2-sided 90% confidence interval (CI) of the mean lay well within the predefined equivalent margins (Fig. 2), demonstrating equivalence in weight gain for the PP population. Equivalence of weight gain per day for the same comparison groups was also confirmed for the ITT population (data not shown). Length gain, head circumference gain, and mid-upper arm circumference gain during the study period were not different between intervention groups in these 3 comparisons of interest (Table 4). Moreover, in line with weight gain per day, equivalence in length and head circumference gain per day during the intervention period was demonstrated for these parameters in all but 1 comparison of interest. Equivalence could not be demonstrated for the length gain of scGOS/lcFOS versus scGOS/lcFOS + 50% FERM groups in the PP population, with an apparently higher gain in length in the latter group. In comparison with the WHO Growth Standards based on growth of exclusively breastfed infants, the mean z score values (including their 95% CI) for weight-for-age, length-for-age, and weight-for-length in all intervention groups lay within or close to the + 0.5 to –0.5 range (Fig. 3).

gain in length in the latter group. In comparison with the WHO Growth Standards based on growth of exclusively breastfed infants, the mean z score values (including their 95% CI) for weight-for-age, length-for-age, and weight-for-length in all intervention groups lay within or close to the + 0.5 to –0.5 range (Fig. 3). FIGURE 2 Graphical display of the difference in means of the daily weight gain (g/d) equivalence analysis for the intervention group comparisons of interest. The equivalence analysis was performed using parametric growth curves correcting for sex, site was taken into account as random effect, and each subject's intercept and slope taken into account as random effects. FERM = fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. TABLE 4 Average gain in infant growth, length, and head circumference (mean ± standard deviation) during the intervention period (baseline to 17 weeks) in the per protocol population

FIGURE 2 Graphical display of the difference in means of the daily weight gain (g/d) equivalence analysis for the intervention group comparisons of interest. The equivalence analysis was performed using parametric growth curves correcting for sex, site was taken into account as random effect, and each subject's intercept and slope taken into account as random effects. FERM = fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. TABLE 4 Average gain in infant growth, length, and head circumference (mean ± standard deviation) during the intervention period (baseline to 17 weeks) in the per protocol population scGOS/lcFOS (n = 75) scGOS/lcFOS +15% FERM (n = 79) scGOS/lcFOS +50% FERM (n = 79) 50% FERM (n = 65) Weight gain in g/day (n) 29.734 ± 6.121 (57) 28.538 ± 5.064 (65) 28.727 ± 5.903 (64) 28.235 ± 4.833 (52) Length gain in mm/day (n) 1.078 ± 0.195 (56) 1.078 ± 0.180 (64) 1.135 ± 0.127 (63) 1.092 ± 0.147 (52) Head circumference gain in mm/day (n) 0.572 ± 0.128 (47) 0.559 ± 0.122 (54) 0.566 ± 0.098 (52) 0.580 ± 0.083 (42) 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides.

in mm/day (n) 0.572 ± 0.128 (47) 0.559 ± 0.122 (54) 0.566 ± 0.098 (52) 0.580 ± 0.083 (42) 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. FIGURE 3 Weight-for-age (upper), length-for-age (middle), and weight-for-length (lower) WHO z score: plot of predicted value plus 95% confidence interval (CI) from mixed model analysis of z-score for the PP population per intervention group. FERM = fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides.

age (middle), and weight-for-length (lower) WHO z score: plot of predicted value plus 95% confidence interval (CI) from mixed model analysis of z-score for the PP population per intervention group. FERM = fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. (Serious) Adverse Events The safety analysis was performed on the all-subjects-treated population. In total, 28 serious AEs (SAE) were reported for 27 subjects (6.5%) during the entire study period. The prevalence of SAEs was apparently higher in the scGOS/lcFOS group, but was only significantly different from the scGOS/lcFOS + 15% FERM group (13.3% vs 2.7%, respectively; FE P = 0.007; Table 5). An initial higher than expected rate of pyelonephritis was recorded in the scGOS/lcFOS group. Assessing the medical records in more detail revealed inconsistencies for the use of this medical term; 1 case was found to be an urinary tract infection and 1 case a direct result of a congenital anatomical anomaly causing vesicoureteral reflux, reducing the incidence to the level of normal background incidence. Only 3 SAEs were reported to be possibly (n = 2; reflux in scGOS/lcFOS + 50% FERM group and abdominal pain in 50% FERM group) or probably (n = 1) related to the study products, the latter being a possible case of cow's milk allergy in the scGOS/lcFOS + 50% FERM intervention group. The rest of the SAEs were “unlikely to be related” or “not related” to the intake of the intervention products. Overall, there was no safety concern. No differences were observed in the frequency and type of AEs. The most commonly observed (37%) were GI disorders (n = 154), consisting mainly of abdominal pain, gastroesophageal reflux and vomiting, and respiratory system disorders (n = 110; 26%), of which rhinitis and upper respiratory tract infection were the most common.

ferences were observed in the frequency and type of AEs. The most commonly observed (37%) were GI disorders (n = 154), consisting mainly of abdominal pain, gastroesophageal reflux and vomiting, and respiratory system disorders (n = 110; 26%), of which rhinitis and upper respiratory tract infection were the most common. TABLE 5 Number and percentage of subjects with at least 1 serious adverse event presented by preferred term in the all subjects treated population n (%) scGOS/lcFOS (n = 98) scGOS/lcFOS +15% FERM (n = 110) scGOS/lcFOS +50% FERM (n = 106) 50% FERM (n = 102) Allergy 1 (0.9%) Fever 1 (1.0%) 1 (0.9%) 1 (1.0%) Pyloric stenosis 1 (1.0%) 1 (0.9%) Abdominal pain 1 (1.0%) Gastroesophageal reflux 1 (0.9%) 1 (1.0%) Vomiting 2 (2.0%) Fracture 1 (1.0%) Urinary tract infection 1 (0.9%) 1 (1.0%) Apnoea 1 (1.0%) Bronchitis 1 (0.9%) 1 (0.9%) Pneumonia 1 (0.9%) 1 (0.9%) 1 (1.0%) Upper respiratory tract infection 2 (2.0%) Surgery 1 (1.0%) Varicella 1 (1.0%) Pyelonephritis 3 (3.1%)* Total 13 (13.3%)† 3 (2.7%)† 7 (6.6%) 4 (3.9%) 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. *Although officially recorded as 3 cases of pyelonephritis, detailed assessment of the medical records revealed an inconsistency for the use of this medical term, 1 case was a urinary tract infection and 1 case a direct result of a congenital anatomical anomaly causing vesicoureteral reflux.

n (%) scGOS/lcFOS (n = 98) scGOS/lcFOS +15% FERM (n = 110) scGOS/lcFOS +50% FERM (n = 106) 50% FERM (n = 102) Allergy 1 (0.9%) Fever 1 (1.0%) 1 (0.9%) 1 (1.0%) Pyloric stenosis 1 (1.0%) 1 (0.9%) Abdominal pain 1 (1.0%) Gastroesophageal reflux 1 (0.9%) 1 (1.0%) Vomiting 2 (2.0%) Fracture 1 (1.0%) Urinary tract infection 1 (0.9%) 1 (1.0%) Apnoea 1 (1.0%) Bronchitis 1 (0.9%) 1 (0.9%) Pneumonia 1 (0.9%) 1 (0.9%) 1 (1.0%) Upper respiratory tract infection 2 (2.0%) Surgery 1 (1.0%) Varicella 1 (1.0%) Pyelonephritis 3 (3.1%)* Total 13 (13.3%)† 3 (2.7%)† 7 (6.6%) 4 (3.9%) 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides. *Although officially recorded as 3 cases of pyelonephritis, detailed assessment of the medical records revealed an inconsistency for the use of this medical term, 1 case was a urinary tract infection and 1 case a direct result of a congenital anatomical anomaly causing vesicoureteral reflux. †Values are significantly different, based on Fisher exact test (difference in number of subjects between groups).

*Although officially recorded as 3 cases of pyelonephritis, detailed assessment of the medical records revealed an inconsistency for the use of this medical term, 1 case was a urinary tract infection and 1 case a direct result of a congenital anatomical anomaly causing vesicoureteral reflux. †Values are significantly different, based on Fisher exact test (difference in number of subjects between groups). Fecal Physiological and Microbial Parameters After the intervention, that is, at 17 weeks of age, the measured fecal physiological and microbial parameters of the infants from the scGOS/lcFOS only arm confirmed its beneficial effect (33), that is, lower pH, high relative amounts of acetate, low relative levels of the other measured SCFAs, high occurrence of measurable d- and l-lactate, high levels of SIgA and low occurrence of C difficile, a taxonomic group known to be overrepresented in formula-fed infants as compared to breast-fed infants (34) (Table 6). No differences in fecal parameters were observed comparing both scGOS/lcFOS + 15% FERM and scGOS/lcFOS + 50% FERM to the scGOS/lcFOS only group, except for a higher pH in the scGOS/lcFOS + 50% FERM group (WR P = 0.028; Table 6). In contrast, the fecal parameters of the infants from the 50% FERM only arm were always significantly different compared with the scGOS/lcFOS + 50% FERM group, with the exception of the SCFAs acetate, propionate, and a tendency for higher butyrate levels in the 50% FERM group (C P = 0.066; Table 6). The levels of fecal calprotectin are in line with previously reported data from formula-fed infants and no differences between groups were observed in the comparisons of interest (Table 6; (35)).

on of the SCFAs acetate, propionate, and a tendency for higher butyrate levels in the 50% FERM group (C P = 0.066; Table 6). The levels of fecal calprotectin are in line with previously reported data from formula-fed infants and no differences between groups were observed in the comparisons of interest (Table 6; (35)). TABLE 6 Physiological parameters and presence of Clostridium difficile in fecal samples at 17 weeks of age

on of the SCFAs acetate, propionate, and a tendency for higher butyrate levels in the 50% FERM group (C P = 0.066; Table 6). The levels of fecal calprotectin are in line with previously reported data from formula-fed infants and no differences between groups were observed in the comparisons of interest (Table 6; (35)). TABLE 6 Physiological parameters and presence of Clostridium difficile in fecal samples at 17 weeks of age Parameter Statistic shown scGOS/lcFOS (N = 30) scGOS/lcFOS +15% FERM (N = 30) scGOS/lcFOS +50% FERM (N = 30) 50% FERM (N = 30) Acetate W (mmol/kg wet weight faeces) Median (min-max) 81.9 (65.7–93.7) 78.6 (63.4–103.2) 74.5 (59.0–93.2) 62.9 (50.3–74.2) Propionate W (mmol/kg wet weight faeces) Median (min-max) 8.0 (2.9–14.6) 8.1 (4.5–16.0) 15.3 (5.5–18.0) 14.3 (6.4–20.1) Butyrate C (mmol/kg wet weight faeces) Median (min-max) 0.9 (0.0–3.1) 0.8 (0.0–3.0) 1.2 (0.0–3.0) 5.3 (1.1–8.6) Isobutyrate C 0 n (%) 23 (79.3%) 23 (76.7%) 17 (58.6%)* 7 (23.3%)* 1 n (%) 6 (20.7%) 7 (23.3%) 12 (41.4%)* 23 (76.7%)* Valerate C 0 n (%) 27 (93.1%) 25 (83.3%) 29 (100.0%)* 26 (86.7%)* 1 n (%) 2 (6.9%) 5 (16.7%) 0 (0.0%)* 4 (13.3%)* Isovalerate C 0 n (%) 16 (55.2%) 14 (46.7%) 13 (44.8%)* 1 (3.3%)* 1 n (%) 13 (44.8%) 16 (53.3%) 16 (55.2%)* 29 (96.7%)* l-lactate C 0 n (%) 4 (13.8%) 5 (16.7%) 8 (26.7%)* 21 (70.0%)* 1 n (%) 25 (86.2%) 25 (83.3%) 22 (73.3%)* 9 (30.0%)* d-lactate C 0 n (%) 9 (31.0%) 14 (46.7%) 13 (43.3%)* 22 (73.3%)* 1 n (%) 20 (69.0%) 16 (53.3%) 17 (56.7%)* 8 (26.7%)* Costridium difficile C 0 n (%) 22 (73.3%) 24 (80.0%) 24 (80.0%)* 13 (43.3%)* 1 n (%) 8 (26.7%) 6 (20.0%) 6 (20.0%)* 17 (56.7%)* SIgA (μg/g wet weight feces) W Median (min-max) 1245.9 (144.7–3006.1) 1371.5 (57.0–4093.3) 1424.4* (84.9–3346.4) 499.7* (45.6–5145.4) Calprotectin (μg/g wet weight feces) W Median (min-max) 225.8 (48.6 – 1815.3) 240.0 (3.1–1377.2) 181.0 (8.6–4584.9) 207.1 (44.3–1926.3) pH W Median (min-max) 5.6† (5.0–7.3) 5.7 (5.1–7.6) 6.1*,† (5.2–7.6) 6.9* (5.6–8.2) WFor the analysis a Wilcoxon rank sum tests (W) was used due to violation of normality assumption and/or presence of outliers.

ight feces) W Median (min-max) 225.8 (48.6 – 1815.3) 240.0 (3.1–1377.2) 181.0 (8.6–4584.9) 207.1 (44.3–1926.3) pH W Median (min-max) 5.6† (5.0–7.3) 5.7 (5.1–7.6) 6.1*,† (5.2–7.6) 6.9* (5.6–8.2) WFor the analysis a Wilcoxon rank sum tests (W) was used due to violation of normality assumption and/or presence of outliers. CCategorical response parameters (0 = absent, 1 = present) were analyzed by using Chi-square tests (C). 15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides; SIgA = secretory immunoglobulin A. *Significantly different values between groups for the tested comparison of interest: scGOS/lcFOS + 50% FERM versus 50% FERM only. †Significantly different values between groups for the tested comparison of interest: scGOS/lcFOS + 15% FERM or scGOS/lcFOS + 50% FERM versus scGOS/lcFOS only.

15% FERM = infant formula consisting of 15% fermented infant formula; 50% FERM = infant formula consisting of 50% fermented infant formula; scGOS/lcFOS = short-chain galacto-oligosaccarides and long-chain fructo-oligosaccharides; SIgA = secretory immunoglobulin A. *Significantly different values between groups for the tested comparison of interest: scGOS/lcFOS + 50% FERM versus 50% FERM only. †Significantly different values between groups for the tested comparison of interest: scGOS/lcFOS + 15% FERM or scGOS/lcFOS + 50% FERM versus scGOS/lcFOS only. DISCUSSION To our knowledge, this is the first study evaluating the safety of a partly fermented infant formula supplemented with prebiotics. The primary outcome of the study is the demonstrated equivalence in infant weight gain per day during the intervention period analyzed according to both Intention to Treat and Per Protocol, fulfilling the safety criteria of the American Academy of Pediatrics (28). No differences in number and type of (serious) AEs were observed. The obtained anthropometric measures in all intervention groups were close to the WHO Growth Standards (mean z scores within ±0.5 SD), indicating that the formulae combining the specific prebiotic mixture (scGOS/lcFOS 9:1) with fermented infant formula support an adequate infant growth.

of (serious) AEs were observed. The obtained anthropometric measures in all intervention groups were close to the WHO Growth Standards (mean z scores within ±0.5 SD), indicating that the formulae combining the specific prebiotic mixture (scGOS/lcFOS 9:1) with fermented infant formula support an adequate infant growth. The current findings in the 50% FERM group confirm the previously assessed adequate infant growth and tolerance based on the long history of use (6) and based on the secondary anthropometric outcomes of several studies which were smaller-scaled (10), using a thickened fermented formula (8), or performed in older (9) or preterm infants (36). Likewise, the outcomes of the present study for the scGOS/lcFOS only intervention group confirm previous findings indicating a normal infant growth (20–22). A meta-analysis demonstrated a slightly increased weight gain (weight gain increase of 0.97 g/day, 95% CI 0.24–1.70) but not length or head circumference gain, in infants consuming a prebiotic-containing formula (37). In the present study, we confirmed a small increase (0.31 g/day) in mean weight gain per day during the intervention period when prebiotics were added to a partly fermented formula.

ease of 0.97 g/day, 95% CI 0.24–1.70) but not length or head circumference gain, in infants consuming a prebiotic-containing formula (37). In the present study, we confirmed a small increase (0.31 g/day) in mean weight gain per day during the intervention period when prebiotics were added to a partly fermented formula. All formulae were well accepted by parents and the presence of the prebiotics and/or fermented formula did not affect formula consumption. Only 1 SAE was found to be probably related to study product intake, being an allergic reaction to cow's milk experienced in the scGOS/lcFOS + 50% FERM group evidently relating to the presence of dairy protein rather than fermented formula or prebiotics. The same intervention group contained 1 case of symptomatic gastroesophageal reflux indicated by the investigators to be possibly related to the study product intake, as did the 50% FERM intervention group. The overall prevalence of symptomatic gastroesophageal reflux was, however, low and also similar between intervention groups. The percentage of infants with at least 1 SAE was similar in all intervention groups with partly fermented formula, but slightly higher in the scGOS/lcFOS only intervention group, reaching statistical significance compared with the scGOS/lcFOS + 15%FERM intervention group. The observed number of SAE in the scGOS/lcFOS group is, however, within the normal range for an infant population and quite similar to the values reported in comparable studies for (prebiotic) formula- or breast-fed infants (38,39). Moreover, none of the SAEs in the scGOS/lcFOS intervention group was indicated to be related to the product and some rather the result of chance, for example, 4 of the 13 SAEs were infants experiencing varicella, fracture, surgery, and pyloric stenosis. Overall, the reported types of (serious) AEs are typical for young infants and no difference in the frequency per type of (serious) AEs was observed between the different formula groups. Moreover, the drop-out rate and reasons for dropping out were similar between intervention groups. Hence, there are no safety concerns for use of products combining prebiotics and fermented formula in healthy, term infants.

ce in the frequency per type of (serious) AEs was observed between the different formula groups. Moreover, the drop-out rate and reasons for dropping out were similar between intervention groups. Hence, there are no safety concerns for use of products combining prebiotics and fermented formula in healthy, term infants. The presence of scGOS/lcFOS (9:1) in nonfermented infant formula was found to modulate the microbiota of infants toward a more “breast-fed like” composition with more bifidobacteria and less pathogenic bacteria such as clostridia-related species (16,18,20,31). This observed modulation in microbiota was found to be associated with differences in metabolic activity of the microbiota resulting in a reduction in stool pH, an SCFA pattern containing a higher proportion of acetate, a lower proportion of propionate or other SCFAs, and increased levels of fecal SIgA (16,20,31). The present study confirms this established bifidogenic effect of scGOS/lcFOS (9:1) when added to partly fermented formulae, indicated by the lower percentage of C difficile, the higher occurrence of measurable lactate levels, the decreased pH, and increased SIgA levels reported for the scGOS/lcFOS + 50% FERM group compared with the 50% FERM only group. Moreover, no differences in fecal parameters were demonstrated between the formulae containing scGOS/lcFOS (9:1), irrespective of the presence or dosage of fermented formula (scGOS/lcFOS + 15% FERM and scGOS/lcFOS + 50% FERM vs scGOS/lcFOS only). Hence, the present study reconfirms the beneficial effect of scGOS/lcFOS, which was not affected by the presence of fermented formula.

d between the formulae containing scGOS/lcFOS (9:1), irrespective of the presence or dosage of fermented formula (scGOS/lcFOS + 15% FERM and scGOS/lcFOS + 50% FERM vs scGOS/lcFOS only). Hence, the present study reconfirms the beneficial effect of scGOS/lcFOS, which was not affected by the presence of fermented formula. In conclusion, a partly fermented formula containing a specific mixture of prebiotics supports adequate growth in early life firstly based on the weight gain equivalence between intervention groups in the ITT and PP populations and, secondly, on the similarity in growth development compared to the WHO growth standards. In addition, these formulae were well-tolerated and safe based on growth and AE outcomes. As a next step, the potential effect of partly fermented formulae with scGOS/lcFOS on GI function is under investigation. Acknowledgments The authors would like to thank all the families participating in the study, and all participating pediatricians from Ireland (Dr Perry, Dr Harrington, Dr Forde, Dr Murphy, and Dr Belton), France (Dr Morville, Dr Thiriez, and Dr Logre) and Belgium (Dr Alliet, Dr Franckx, Dr Jespers, Dr Logghe, Dr Peeters, Dr Van Eldere, Dr Vandeputte, Dr Verlinde, and Dr Vertruyen) and their research staff for their contribution to the FIPS study. We would like to thank Stefanie Schoen from Nutricia Research for her contribution to the FIPS study. Nutricia Research, The Netherlands, provided the funding to conduct the study. The funder contributed to the design of the study, interpretation of the findings, and writing of the manuscript.

bution to the FIPS study. We would like to thank Stefanie Schoen from Nutricia Research for her contribution to the FIPS study. Nutricia Research, The Netherlands, provided the funding to conduct the study. The funder contributed to the design of the study, interpretation of the findings, and writing of the manuscript. This trial was registered in the Dutch Trial Register www.trialregister.nl registration number: NTR 2521. Nutricia Research provided financial support for the study. The authors have no conflicts of interest.

What Is KnownCross-sectional evidence suggests physical activity is inversely related to nonalcoholic fatty liver disease in children/adolescents. Prospective associations have not been examined. Thus, we are unable to establish whether lower physical activity causes nonalcoholic fatty liver disease or nonalcoholic fatty liver disease causes lower physical activity. What Is NewWe provide evidence that adolescents who are more active in late childhood have lower risk of ultrasound scan fatty liver and lower γ-glutamyl transferase levels. These findings are likely to be, in part, explained by adiposity. If replicated, our findings highlight the importance of maintaining healthy levels of physical activity throughout childhood for preventing nonalcoholic fatty liver disease.

What Is NewWe provide evidence that adolescents who are more active in late childhood have lower risk of ultrasound scan fatty liver and lower γ-glutamyl transferase levels. These findings are likely to be, in part, explained by adiposity. If replicated, our findings highlight the importance of maintaining healthy levels of physical activity throughout childhood for preventing nonalcoholic fatty liver disease. Greater adiposity is an important risk factor for nonalcoholic fatty liver disease (NAFLD) (1). Consequently, lifestyle modifications, including increasing physical activity levels with the aim of reducing adiposity and limiting fatty liver infiltration and necroinflammation, are recommended for both adult and child patients with NAFLD (2,3). Physical activity could influence NAFLD risk in several ways. Physical inactivity is associated with greater total adiposity (4), which may increase the risk of fat infiltration into hepatocytes by increasing free fatty acid (FFA) influx from adipose tissue to the liver (5). Physical activity has been shown to improve insulin sensitivity through decreasing fasting insulin and increasing lean mass (6). In an insulin-resistant state, the ability of insulin to suppress adipose tissue lipolysis is impaired, leading to an increased efflux of FFA from adipose tissue and an increased delivery of FFA to the liver (7). Physical activity may also reduce levels of inflammatory markers that are involved in the progression of steatosis to nonalcoholic steatohepatitis (8,9).

insulin to suppress adipose tissue lipolysis is impaired, leading to an increased efflux of FFA from adipose tissue and an increased delivery of FFA to the liver (7). Physical activity may also reduce levels of inflammatory markers that are involved in the progression of steatosis to nonalcoholic steatohepatitis (8,9). The existing evidence for an inverse association between physical activity and NAFLD comes from cross-sectional studies using questionnaires to assess physical activity (10–18), thus potentially underestimating associations in comparison with objectively measured physical activity (19). Furthermore, cross-sectional studies cannot rule out reverse causation (ie, greater adiposity and associated NAFLD causing people to be less active rather). Two recent studies support this possibility (20,21). We are unaware of any prospective studies assessing associations of physical activity with NAFLD in children and/or adolescents. We aimed to assess whether objectively measured physical activity at mean ages 12 and 14 years are associated with markers of NAFLD assessed at mean age 17.8 years, and to determine whether any observed associations are mediated through fat mass and insulin.

physical activity with NAFLD in children and/or adolescents. We aimed to assess whether objectively measured physical activity at mean ages 12 and 14 years are associated with markers of NAFLD assessed at mean age 17.8 years, and to determine whether any observed associations are mediated through fat mass and insulin. METHODS Study Population Avon Longitudinal Study of Parents and Children (ALSPAC) is a prospective birth cohort from southwest England (full details in online supplement) (22,23). The study website contains details of all available data through a fully searchable data dictionary (www.bris.ac.uk/alspac/researchers/data-access/data-dictionary). Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and the local research ethics committees. Briefly, ALSPAC recruited a cohort of 14,541 pregnancies with expected delivery dates between April 1, 1991, and December 31, 1992. A total of 13,678 singleton live-born infants resulted from these pregnancies. The cohort has been followed-up since birth, including repeat clinical assessment from age 7 years to capture information on a range of characteristics such as demographics, health-related behaviors, psychological and social wellbeing, and health factors. A total of 5081 participants attended the 17- to 18-year follow-up assessment at mean age 17.8 years, and of these, N = 3188 (62.7%) had data available for blood-based indicators of liver function including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and/or GGT. A liver ultrasound scan substudy was undertaken on a randomly selected subgroup of participants attending the 17- to 18-year follow-up (N = 1887, 37%, Fig. 1) (24).

7%) had data available for blood-based indicators of liver function including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and/or GGT. A liver ultrasound scan substudy was undertaken on a randomly selected subgroup of participants attending the 17- to 18-year follow-up (N = 1887, 37%, Fig. 1) (24). FIGURE 1 Participant flow through the study. ∗Participants were excluded if they had no measure of physical activity measure at 12 or 14 years, or they had harmful alcohol consumption. ALT = alanine aminotransferase; AST = aspartate aminotransferase; GGT = γ-glutamyl transferase; USS = ultrasound scan. None of the participants had a known history of jaundice or hepatitis, were taking medications or receiving treatment that would indicate they had hepatic disease, or were taking medication known to influence liver function. Participants’ alcohol consumptions were assessed at mean ages 16.7 and 17.8 years using the Alcohol Use Disorders Identification Test (AUDIT), which has been used to validate the assessment of alcohol consumption in adolescents (25). Participants answered 10 questions about their alcohol consumption, and from their responses, a score between 0 and 20 was derived. A score >16 is classified as harmful alcohol consumption. Ten participants who completed the ultrasound scan examination and 25 participants with blood-based liver data who were classified as harmful drinkers at both ages were excluded from this study.

on, and from their responses, a score between 0 and 20 was derived. A score >16 is classified as harmful alcohol consumption. Ten participants who completed the ultrasound scan examination and 25 participants with blood-based liver data who were classified as harmful drinkers at both ages were excluded from this study. Outcome Assessment Ultrasound Scan Assessment of Liver Fat and Stiffness Details of ultrasound scan assessment in ALSPAC have been published and are in the online supplement (24). Briefly, upper abdominal ultrasound scan was completed by 1 of 4 trained sonographers using an Acuson S2000 ultrasound scan system (Siemens, Erlangen, Germany) to assess echogenicity and liver stiffness (our main indicator of liver fibrosis). Echogenicity was assessed during deep inspiration and recorded as present, absent, or uncertain according to established protocols (26,27). Thus, the binary ultrasound scan liver fat variable in this study is coded as present or absent. Acoustic radiation force impulse (ARFI) imaging of the right lobe of the liver was used to measure liver stiffness using standard protocols (28,29). Levels of agreement in identifying echogenicity between the 4 sonographers was high, both immediately after training and at 6-month intervals during data collection (absolute agreement of 98% or greater).

(ARFI) imaging of the right lobe of the liver was used to measure liver stiffness using standard protocols (28,29). Levels of agreement in identifying echogenicity between the 4 sonographers was high, both immediately after training and at 6-month intervals during data collection (absolute agreement of 98% or greater). Assessment of Blood-Based Liver Outcomes Participants were instructed to fast overnight or for a minimum of 6 hours. Fasting blood samples were immediately spun and frozen at −80°C. Measurements were assayed within 3 to 9 months after samples were taken, with no previous freeze-thaw cycles. ALT, GGT, and AST were measured by automated analyzers with enzymatic methods. All of the inter coefficients and intracoefficients of variation for these blood-based assays were <5%.

y spun and frozen at −80°C. Measurements were assayed within 3 to 9 months after samples were taken, with no previous freeze-thaw cycles. ALT, GGT, and AST were measured by automated analyzers with enzymatic methods. All of the inter coefficients and intracoefficients of variation for these blood-based assays were <5%. Assessment of Physical Activity Detailed methods of physical activity assessment have been described (30) and are in the online supplement. Briefly, physical activity was objectively assessed at mean ages 12 and 14 years with a uniaxial ActiGraph accelerometer (AM7164 2.2; ActiGraph LLC, Fort Walton Beach, FL; http://www.theactigraph.com) for 7 days. The participants recorded the times at which the ActiGraph was worn and any times that they swam or cycled each day. Total physical activity (average counts per minute) during the valid measurement period and average time spent in moderate to vigorous physical activity (MVPA) in minutes per valid day were derived. The cutoff point used to define MVPA was a CPM >3600 (31). Counts are a result of summing postfiltered accelerometer values (raw data at 30 Hz) into epoch “chunks.” The value of the counts will vary based on the frequency and intensity of the raw acceleration. The filtering process by which counts are produced is proprietary to ActiGraph.

d to define MVPA was a CPM >3600 (31). Counts are a result of summing postfiltered accelerometer values (raw data at 30 Hz) into epoch “chunks.” The value of the counts will vary based on the frequency and intensity of the raw acceleration. The filtering process by which counts are produced is proprietary to ActiGraph. Covariables The assessment of all of the covariables is described in detail in the online supplement. The following were considered potential confounders: sex, maternal age at delivery, parity, maternal education, head of household, social class, maternal body mass index (BMI), ethnicity, energy intake at age 11 years, pubertal status at age 11 years, age at physical activity assessments and liver outcome assessment, and length of time accelerometer was worn (in minutes). Given that adiposity is known to be causally related to physical activity (20), and could possibly be causally related to NAFLD, we considered adiposity (measured by dual x-ray absorptiometry [DXA] total body fat mass) as a potential confounder. We also considered adiposity and insulin resistance (assessed by the homeostatic model assessment of insulin resistance [HOMA-IR]) at the time of liver outcome assessment as potential mediators (ie, potentially being on the causal pathway between physical activity and NAFLD).

l body fat mass) as a potential confounder. We also considered adiposity and insulin resistance (assessed by the homeostatic model assessment of insulin resistance [HOMA-IR]) at the time of liver outcome assessment as potential mediators (ie, potentially being on the causal pathway between physical activity and NAFLD). Eligibility Criteria For analysis of ultrasound scan liver outcomes, eligible participants had to have valid data for ultrasound scan liver fat and liver stiffness, and a measure of physical activity at mean age 12 and/or 14 years (n = 1292). For analyses of blood-based liver outcomes, eligible participants had to have data for ALT, AST, and GGT and a measure of physical activity at mean age 12 and/or 14 years (n = 2612). As described above, participants’ alcohol consumption was assessed using AUDIT (25). No eligible participants were classified as having harmful levels of alcohol consumption. Statistical Analysis All the analyses were conducted in Stata (StataCorp, College Station, TX). Total physical activity was divided by 100 and MVPA by 15; these values were chosen because they represent realistic intervention targets (32). ALT, AST, GGT, and liver stiffness were positively skewed, and their natural logged values were used in all of the linear regression analyses. Coefficients from regression models including these logged variables as outcomes were back transformed. Associations of total physical activity and MVPA with the binary ultrasound scan liver fat variable were assessed with logistic regression.

ir natural logged values were used in all of the linear regression analyses. Coefficients from regression models including these logged variables as outcomes were back transformed. Associations of total physical activity and MVPA with the binary ultrasound scan liver fat variable were assessed with logistic regression. Associations of total physical activity and MVPA with liver outcomes were examined using the following regression models: unadjusted; adjusted for, sex, maternal age at delivery, parity, maternal education, head of household, social class, maternal BMI, ethnicity, energy intake at age 11 years, pubertal status at age 11 years, age at physical activity and liver outcome assessments, and the length of time the accelerometer was worn; same as model 2 but additionally adjusted for DXA-determined fat mass, height, and height2 at the time physical activity was assessed as a potential confounder; same as model 2 but additionally adjusted for DXA-determined fat mass, height, and height2 at the time all of the liver outcomes were assessed as a potential mediator; same as model 4 but additionally adjusted for HOMA-IR at the time all the liver outcomes were assessed as another potential mediator. The inclusion of height and height-squared as covariables in all of the analyses including fat mass is to ensure adjustment for greater relative adiposity, rather than greater fat mass as a result of greater height. Likelihood ratio tests were used to check for sex interactions in all of the models.

potential mediator. The inclusion of height and height-squared as covariables in all of the analyses including fat mass is to ensure adjustment for greater relative adiposity, rather than greater fat mass as a result of greater height. Likelihood ratio tests were used to check for sex interactions in all of the models. Dealing With Missing Data and Additional Analyses Of the 1292 eligible participants with ultrasound scan data and a physical activity measure at ages 12 and/or 14 years, 7% of the participants were missing data for one of the physical activity measures at either age 12 or 14 years. Of the 2612 eligible participants with blood-based liver data and a physical activity measure at ages 12 and/or 14 years, 30% of the participants were missing data for one of the physical activity measures at either age 12 or 14 years. There was also missing data for potential confounders. Web Tables A and B, show the percentage of imputed data for each variable. To minimize selection bias and increase efficiency, multivariate multiple imputation was used to impute missing data for eligible participants. Full details of this procedure are in the online supplement. Details of a series of sensitivity analyses, conducted to verify model assumptions and to test the robustness of our findings, are also provided in the online supplement.

tivariate multiple imputation was used to impute missing data for eligible participants. Full details of this procedure are in the online supplement. Details of a series of sensitivity analyses, conducted to verify model assumptions and to test the robustness of our findings, are also provided in the online supplement. RESULTS Characteristics of the ultrasound scan study participants stratified by ultrasound scan–identified liver fat are described in Table 1. The sample consisted of 96% white participants. There was no evidence that associations of total physical activity and MVPA at ages 12 and 14 years with the liver outcomes differed between boys and girls (all interaction P values >0.05). Thus, results are presented with sexes combined, and sex is adjusted for in all the models. Prevalence of ultrasound scan fatty liver was 2.7% in boys (n = 14/524) and 2.1% in girls (n = 16/768). ALT was elevated (>40 U/L) in 4.2% of the boys (n = 53/1249) and 1.8% of the girls (n = 25/1363). Variable distributions were similar among the imputed and observed data sets (Web Tables C and D.

RESULTS Characteristics of the ultrasound scan study participants stratified by ultrasound scan–identified liver fat are described in Table 1. The sample consisted of 96% white participants. There was no evidence that associations of total physical activity and MVPA at ages 12 and 14 years with the liver outcomes differed between boys and girls (all interaction P values >0.05). Thus, results are presented with sexes combined, and sex is adjusted for in all the models. Prevalence of ultrasound scan fatty liver was 2.7% in boys (n = 14/524) and 2.1% in girls (n = 16/768). ALT was elevated (>40 U/L) in 4.2% of the boys (n = 53/1249) and 1.8% of the girls (n = 25/1363). Variable distributions were similar among the imputed and observed data sets (Web Tables C and D. Associations of Total Physical Activity and MVPA at Mean Ages 12 and 14 Years With the Ultrasound Scan Liver Outcomes at Mean Age 17.8 Years Confounder and mediator-adjusted associations (ie, models 2–5) of total physical activity and MVPA with the ultrasound scan liver outcomes are presented in Table 2. Unadjusted results are in Web Table E. There was evidence of an inverse association of total physical activity and MVPA at ages 12 and 14 years with the risk of ultrasound scan liver fat in the unadjusted model (model 1, Web Table E and after adjusting for potential confounders (model 2, Table 2). Associations attenuated toward the null after additional adjustment for fat mass at the time of physical activity assessment as a potential confounder (model 3, Table 2), and to a lesser extent after adjustment for fat mass at the time of liver outcome assessment as a potential mediator (model 4, Table 2). Associations did not further attenuate upon additional adjustment of HOMA-IR as a potential mediator (model 5, Table 2). Coefficient magnitudes were greater at age 12 years compared with age 14 years. There were no associations of total physical activity and MVPA with liver stiffness (Table 2).

ator (model 4, Table 2). Associations did not further attenuate upon additional adjustment of HOMA-IR as a potential mediator (model 5, Table 2). Coefficient magnitudes were greater at age 12 years compared with age 14 years. There were no associations of total physical activity and MVPA with liver stiffness (Table 2). Associations of Total Physical Activity and MVPA at Ages 12 and 14 Years With the Blood-Based Liver Outcomes at Mean Age 17.8 Years Confounder and mediator-adjusted associations (ie, models 2–5) of total physical activity and MVPA with the blood-based liver outcomes are presented in Table 3. Unadjusted results are in Web Table F. After adjusting for potential confounders (model 2, Table 3), there was no evidence of associations of total physical activity or MVPA at 12 or 14 years with ALT. There was evidence for small positive associations of total physical activity and MVPA at age 12 years with AST in all the models. There was also evidence of positive associations of total physical activity and MVPA at age 14 years with AST in the unadjusted model (model 1, Web Table F and the point estimate did not change after adjustment for potential confounders and mediators (models 2–5, Table 3). After adjusting for potential confounders (model 2, Table 3), there was strong evidence of small inverse associations of total physical activity and MVPA at age 12 years with GGT. Point estimates attenuated toward the null after adjusting for fat mass as a potential confounder (model 3, Table 3) or a potential mediator (model 4, Table 3), and associations did not change after additional adjustment for HOMA-IR as a potential mediator (model 5, Table 3).

activity and MVPA at age 12 years with GGT. Point estimates attenuated toward the null after adjusting for fat mass as a potential confounder (model 3, Table 3) or a potential mediator (model 4, Table 3), and associations did not change after additional adjustment for HOMA-IR as a potential mediator (model 5, Table 3). Additional Analyses Results were similar when restricted to participants with complete data for all the variables except confidence intervals were much wider, likely because of the large reduction in sample size (ultrasound scan data set reduced from n = 1292 in the imputed data to n = 506 in the complete case, and the biomarker data set reduced from n = 2612 to n = 1117, Web Tables G and H. The results were also similar when multiple imputation models were restricted to participants with data for physical activity at age 12 years (Web Tables I and J and participants with data for physical activity at age 14 (Web Tables K and L. There was no evidence that the odds of missing physical activity data at age 14 years were associated with physical activity at age 12 years, adjusted for all of the variables included in our multiple imputation models (Web Table M. Results were similar after additional adjustment for AUDIT scores in the year before assessing liver outcomes (Web Tables N and O.

ing physical activity data at age 14 years were associated with physical activity at age 12 years, adjusted for all of the variables included in our multiple imputation models (Web Table M. Results were similar after additional adjustment for AUDIT scores in the year before assessing liver outcomes (Web Tables N and O. DISCUSSION In this study, we assessed prospective associations of objectively measured physical activity at mean ages 12 and 14 years with measures of liver fat at mean age 17.8 years. Greater total physical activity and MVPA, at both ages 12 and 14 years, were prospectively associated with lower risk of ultrasound scan liver fat. These associations with ultrasound scan liver fat attenuated toward the null after adjusting for fat mass as a potential confounder, and to a lesser extent with adjustment for fat mass as a potential mediator. Greater total physical activity and MVPA at age 12 years were also prospectively associated with lower levels of GGT, which is associated with cardiovascular disease (33), diabetes (34), and insulin resistance (35). These associations attenuated toward null after adjusting for fat mass as a potential confounder and as a potential mediator. Associations did not further attenuate upon additional (to fat mass) adjustment of HOMA-IR as a potential mediator. Thus, the effect of physical activity on the liver outcomes may operate through adiposity mediating the relation of physical activity to NAFLD, and/or through greater adiposity leading to reduced physical activity (19,20) and greater NAFLD risk, thus confounding the association, or a combination of both.

tial mediator. Thus, the effect of physical activity on the liver outcomes may operate through adiposity mediating the relation of physical activity to NAFLD, and/or through greater adiposity leading to reduced physical activity (19,20) and greater NAFLD risk, thus confounding the association, or a combination of both. Cross-sectional studies have reported inverse associations with ultrasound scan liver fat (12,14,16) and biomarkers, including ALT (11–13) and GGT (11), with some of these studies reporting associations to be independent of adiposity measures (13,16). A small number of studies show no association of physical activity with ultrasound scan or biomarkers of liver fat in children, irrespective of adjustment for adiposity (10,17,18). Randomized controlled trials show beneficial effects of lifestyle interventions on markers of liver fat, including on levels of GGT and ALT providing some support for a possible causal effect (3,36–38). Those trials, however, have all combined physical activity interventions with nutrition programs or antioxidant therapy; thus, it is impossible to assess the relative contribution of physical activity to the observed improvements in NAFLD parameters.

nd ALT providing some support for a possible causal effect (3,36–38). Those trials, however, have all combined physical activity interventions with nutrition programs or antioxidant therapy; thus, it is impossible to assess the relative contribution of physical activity to the observed improvements in NAFLD parameters. We found no associations of physical activity with ALT in our study. As noted above, in cross-sectional studies of children, greater levels of physical activity have been related to lower ALT levels. Given the prospective association we observed with ultrasound scan fatty liver, which is a more direct measure of liver fat, the lack of association with ALT in our study is difficult to explain but could reflect random sampling variation. We also found that childhood physical activity was not associated with ultrasound scan–assessed liver stiffness. This may reflect earlier physical activity reducing the risk of fat in the liver but not being sufficient to influence fibrosis, if it has become established. Further examination in prospective studies with relevant data is important before making conclusions about these findings.

rasound scan–assessed liver stiffness. This may reflect earlier physical activity reducing the risk of fat in the liver but not being sufficient to influence fibrosis, if it has become established. Further examination in prospective studies with relevant data is important before making conclusions about these findings. We found that children with greater levels of physical activity in childhood had higher AST levels in adolescence. AST is abundant in muscle cells, and studies have reported greater physical activity, particularly vigorous activity, to be associated with higher circulating levels possibly because of muscle cell breakdown (39–41). Consistent with this and our findings, a recent cross-sectional study found that adolescents achieving daily recommended time spent in objectively measured MVPA (compared with those not achieving this recommendation) had lower GGT and ALT but higher AST (42).

ng levels possibly because of muscle cell breakdown (39–41). Consistent with this and our findings, a recent cross-sectional study found that adolescents achieving daily recommended time spent in objectively measured MVPA (compared with those not achieving this recommendation) had lower GGT and ALT but higher AST (42). Strengths and Limitations To our knowledge, this is the first study to prospectively assess associations of objectively measured physical activity in childhood with the risk of NAFLD in adolescence. Although power for the binary outcome of ultrasound scan–measured liver fat was limited because of the small number of cases, we looked at a range of related outcomes including ALT. A recent meta-analysis including 49 different studies found ultrasound scan to be reliable and accurate for the detection of moderate-severe hepatic steatosis compared with liver histology (43) but less able to distinguish mild fat infiltration. This means our NAFLD cases are likely to reflect the moderate/severe end of the disease spectrum, and the associations observed between physical activity and ultrasound scan liver fat may be underestimated. The ARFI measure of liver stiffness has been validated in a small number of clinical studies (44,45). Although there was a greater proportion of missing data for the physical activity measure at age 14 years compared with at age 12, results were similar for the complete case analyses and when restricting multiple imputation analyses to those with a measure of physical activity at ages 12 and 14 years (ie, only imputing missing confounder data). Furthermore, we found no evidence that missingness at age 14 years was associated with physical activity at age 12 years. Together, these findings suggest that missing data is unlikely to importantly introduce bias in these associations.

ical activity at ages 12 and 14 years (ie, only imputing missing confounder data). Furthermore, we found no evidence that missingness at age 14 years was associated with physical activity at age 12 years. Together, these findings suggest that missing data is unlikely to importantly introduce bias in these associations. CONCLUSIONS We have shown that adolescents who were more active in late childhood have lower fat mass in adolescence, which in turn is related to the lower risk of ultrasound scan fatty liver and lower levels of GGT. If our findings are replicated, they highlight the importance of maintaining healthy levels of physical activity through childhood to prevent the risk of subsequent NAFLD. Supplementary Material Supplemental Digital Content Acknowledgments The authors thank all of the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.

Content Acknowledgments The authors thank all of the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. The research leading to these results received funding from the UK Medical Research Council (G0801456), the British Heart Foundation (PG/11/33/28794), and the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. HEALTH-F2-2009-241762 for the project FLIP. The UK Medical Research Council and Wellcome Trust (092731), together with the University of Bristol, provide core support for the Avon Longitudinal Study of Parents and Children. Drs Lawlor, Fraser, and Howe work in a unit that receives funding from the UK Medical Research Council, and Emma L. Anderson's studentship is funded by that grant. Drs Fraser and Howe are funded by the UK Medical Research Council Postdoctoral research fellowships (G0701594 and G1002375). The other authors report no conflicts of interest. TABLE 1 Characteristics of participants included in the ultrasound scan study (N = 1292)

The research leading to these results received funding from the UK Medical Research Council (G0801456), the British Heart Foundation (PG/11/33/28794), and the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. HEALTH-F2-2009-241762 for the project FLIP. The UK Medical Research Council and Wellcome Trust (092731), together with the University of Bristol, provide core support for the Avon Longitudinal Study of Parents and Children. Drs Lawlor, Fraser, and Howe work in a unit that receives funding from the UK Medical Research Council, and Emma L. Anderson's studentship is funded by that grant. Drs Fraser and Howe are funded by the UK Medical Research Council Postdoctoral research fellowships (G0701594 and G1002375). The other authors report no conflicts of interest. TABLE 1 Characteristics of participants included in the ultrasound scan study (N = 1292) Participants without ultrasound scan liver fat (n = 1262) Participants with ultrasound scan liver fat (n = 30) n with available data Distribution n with available data Distribution P for difference Median liver stiffness (IQR) 1262 1.18 (1.07, 1.31) 30 1.41 (1.23, 1.98) <0.001 Median ALT, U/L (IQR) 866 15.5 (12.4, 19.6) 16 20.65 (17.1, 42.7) <0.001 Median AST, U/L (IQR) 866 19.7 (16.9, 23.3) 16 21.6 18.15 31.75 21.6 18.15 31.75 21.6 (18.15, 31.75) 0.06 Median GGT, U/L (IQR) 865 16.0 (13.0, 20.0) 16 28.5 (17.0, 41.0) <0.01 Median CPM at age 12 y (IQR) 1180 564 (464, 694) 28 481 (441, 613) 0.03 Median CPM at age 14 y (IQR) 882 500 (398, 650) 21 420 (397, 524) 0.08 Median MVPA at age 12 y, minutes per day (IQR) 1180 18.4 (11.07, 29.83) 28 14.51 (7.75, 18.57) 0.02 Median MVPA at age 14 y, minutes per day (IQR) 882 19.15 (9.80, 32.0) 21 15.67 (6.0, 22.0) 0.08 Mean maternal age, y (SD) 1220 29.58 (26.58, 32.83) 27 30 (27.58, 33.42) 0.60 Median maternal BMI, kg/m2 (IQR) 1109 22.18 (20.47, 24.38) 26 23.50 (21.33, 26.49) <0.01 Median predicted energy intake at age 12 y, kcal (IQR) 1258 2011 (1911, 2156) 30 2051 (1963, 2234) 0.11 Median fat mass at mean age 12 y, kg (IQR) 1208 10.1 (6.8, 15.0) 29 19.5 (16.1, 27.5) <0.001 Median fat mass at mean age 14 y, kg (IQR) 1115 12.2 (7.9, 18.1) 26 26.9 (18.1, 30.4) <0.001 Median fat mass at mean age 17.8 y, kg (IQR) 1224 16.6 (11.3, 23.3) 29 36.9 (28.9, 46.6) <0.001 Median HOMA-IR at mean age 17.8 y, kg (IQR) 865 1.49 (1.08, 2.09) 16 3.24 (2.24, 4.23) <0.001 Mean height at mean age 17.8 y, cm (SD) 1234 170.57 (9.40) 30 171.81 (8.98) 0.48 Median age at outcome assessment, mo (IQR) 1262 213 (212, 216) 30 214 (213, 217) 0.16 BMI category 1233 30 Obese 5.52 53.33 <0.001 Overweight 16.30 30.00 Normal weight 73.15 10.00 Underweight 5.03 6.67 Sex 1262 30 Males 40.41 46.67 0.49 Females 59.59 53.33 Parity 1165 26 0 49.61 34.62 0.30 1 34.59 42.31 2+ 15.79 23.08 Head of household, social class Manual 1134 11.99 24 16.67 0.49 Nonmanual 88.01 83.33 Mothers’ education 1148 24 ≤O level 51.39 41.67 0.55 A level 27.96 37.50 Degree or above 20.64 20.83 Tanner stage for pubic hair development at age 11 y 946 21 Prepubertal 62.58 76.19 0.65 Pubertal/p

0.30 1 34.59 42.31 2+ 15.79 23.08 Head of household, social class Manual 1134 11.99 24 16.67 0.49 Nonmanual 88.01 83.33 Mothers’ education 1148 24 ≤O level 51.39 41.67 0.55 A level 27.96 37.50 Degree or above 20.64 20.83 Tanner stage for pubic hair development at age 11 y 946 21 Prepubertal 62.58 76.19 0.65 Pubertal/p ostpubertal 37.42 23.81 Alcohol consumption in the year before outcome assessment 1166 29 Normal 63.72 62.07 0.95 Hazardous 31.73 31.03 Harmful 4.55 6.90 When the median and IQR is displayed, a Mann-Whitney U test was used to check for differences between those with and without ultrasound scan liver fat.

0.30 1 34.59 42.31 2+ 15.79 23.08 Head of household, social class Manual 1134 11.99 24 16.67 0.49 Nonmanual 88.01 83.33 Mothers’ education 1148 24 ≤O level 51.39 41.67 0.55 A level 27.96 37.50 Degree or above 20.64 20.83 Tanner stage for pubic hair development at age 11 y 946 21 Prepubertal 62.58 76.19 0.65 Pubertal/p ostpubertal 37.42 23.81 Alcohol consumption in the year before outcome assessment 1166 29 Normal 63.72 62.07 0.95 Hazardous 31.73 31.03 Harmful 4.55 6.90 When the median and IQR is displayed, a Mann-Whitney U test was used to check for differences between those with and without ultrasound scan liver fat. When the mean and SD are displayed, a 2-tailed t test was used to check for differences between those with and without ultrasound scan liver fat. The ‘N with available data’ columns relate to the number of participants included in the imputation data sets, who had data for each of the variables included in our analysis. The highest pubertal stage was established by taking the highest Tanner rating for either breast development or pubic hair. If there were missing data for breast development, pubic hair ratings were used when available and vice versa. BMI was categorized as follows: underweight (BMI ≤ 18), normal weight (BMI >18 to ≤25), overweight (BMI >25 to ≤30), and obese (BMI > 30). ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CPM = counts per minute; GGT = γ-glutamyl transferase; HOMA-IR = homeostatic model assessment of insulin resistance; IQR = interquartile range; MVPA = moderate to vigorous physical activity; SD = standard deviation.

ese (BMI > 30). ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CPM = counts per minute; GGT = γ-glutamyl transferase; HOMA-IR = homeostatic model assessment of insulin resistance; IQR = interquartile range; MVPA = moderate to vigorous physical activity; SD = standard deviation. TABLE 2 Associations of physical activity measures at ages 12 and 14 years with USS liver outcomes at mean age 17.8 years in the imputed data (n = 1292)

ese (BMI > 30). ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CPM = counts per minute; GGT = γ-glutamyl transferase; HOMA-IR = homeostatic model assessment of insulin resistance; IQR = interquartile range; MVPA = moderate to vigorous physical activity; SD = standard deviation. TABLE 2 Associations of physical activity measures at ages 12 and 14 years with USS liver outcomes at mean age 17.8 years in the imputed data (n = 1292) Adjusted for confounders* Adjusted for confounders plus fat mass, height, and height2 at the time physical activity was assessed† Adjusted for confounders plus fat mass, height, and height2 at the time of liver outcome assessment‡ Adjusted for confounders plus fat mass, height, height2, and HOMA-IR at the time of liver outcome assessment§ USS liver fat OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P Total physical activity 12 y 0.71 (0.54–0.94) 0.02 0.82 (0.60–1.12) 0.22 0.76 (0.56–1.03) 0.08 0.73 (0.52–1.01) 0.06 14 y 0.77 (0.58–1.02) 0.07 0.84 (0.61–1.15) 0.28 0.82 (0.60–1.10) 0.19 0.79 (0.57–1.08) 0.14 MVPA 12 y 0.47 (0.27–0.84) 0.01 0.65 (0.35–1.21) 0.17 0.59 (0.32–1.09) 0.09 0.53 (0.27–1.03) 0.06 14 y 0.63 (0.39–1.04) 0.07 0.72 (0.42–1.26) 0.26 0.70 (0.41–1.19) 0.19 0.67 (0.39–1.13) 0.13 USS liver stiffness % Change (95% CI) P % Change (95% CI) P % Change (95% CI) P % Change (95% CI) P Total physical activity 12 y 0 (−1 to 1) 0.96 0 (0–1) 0.35 0 (0–1) 0.67 0 (0–1) 0.66 14 y 0 (−1 to 0) 0.21 0 (−1 to 0) 0.44 0 (−1 to 0) 0.27 0 (−1 to 0) 0.26 MVPA 12 y 0 (−1 to 1) 0.83 1 (−1 to 2) 0.36 0 (−1 to 1) 0.72 0 (−1 to 1) 0.72 14 y −1 (−2 to 0) 0.23 0 (−1 to 1) 0.44 −1 (−2 to 0) 0.27 −1 (−2 to 0) 0.26 Coefficients are per increase of 100 CPM (total physical activity) or per 15-minute increase in MVPA. Unadjusted results (model 1) are provided in Table E of the online supplement. BMI = body mass index; CI = confidence interval; CPM = counts per minute; HOMA-IR = homeostatic model assessment of insulin resistance; MVPA = moderate to vigorous physical activity; OR = odds ratio; USS = ultrasound scan.

increase in MVPA. Unadjusted results (model 1) are provided in Table E of the online supplement. BMI = body mass index; CI = confidence interval; CPM = counts per minute; HOMA-IR = homeostatic model assessment of insulin resistance; MVPA = moderate to vigorous physical activity; OR = odds ratio; USS = ultrasound scan. *Model 2: adjusted for mother's age at delivery, parity, sex, ethnicity, mother's education, head of household, social class, mother's BMI, energy intake at age 11 years, pubertal status at age 11 years, age at physical activity assessment, length of time accelerometer was worn (in minutes), and age at the time of liver assessment. †Model 3: adjusted for all of the potential confounders listed above plus fat mass, height, and height2 at the time physical activity was assessed. ‡Model 4: adjusted for all of the potential confounders listed above plus fat mass, height, and height2 at the time of liver outcome assessment. §Model 5: adjusted for all potential confounders listed above plus fat mass, height, height2 and HOMA-IR at the time of liver outcome assessment. TABLE 3 Associations of physical activity at ages 12 and 14 years with blood-based liver outcomes at mean age 17.8 years in the imputed data (n = 2612)

‡Model 4: adjusted for all of the potential confounders listed above plus fat mass, height, and height2 at the time of liver outcome assessment. §Model 5: adjusted for all potential confounders listed above plus fat mass, height, height2 and HOMA-IR at the time of liver outcome assessment. TABLE 3 Associations of physical activity at ages 12 and 14 years with blood-based liver outcomes at mean age 17.8 years in the imputed data (n = 2612) Adjusted for confounders* Adjusted for confounders plus fat mass, height, and height2 at the time physical activity was assessed† Adjusted for confounders plus fat mass, height, and height2 at the time of liver outcome assessment‡ Adjusted for confounders plus fat mass, height, height2, and HOMA-IR at the time of liver outcome assessment§ % Change (95% CI) P % Change (95% CI) P % Change (95% CI) P % Change (95% CI) P ALT Total physical activity 12 y 0 (−1 to 1) 0.81 1 (0–2) 0.22 1 (0–2) 0.17 1 (0–2) 0.17 14 y 0 (−1 to 1) 0.74 0 (−1 to 1) 0.59 0 (−1 to 1) 0.45 0 (−1 to 1) 0.44 MVPA 12 y −1 (−3 to 1) 0.28 0 (−1 to 2) 0.68 1 (−1 to 3) 0.39 1 (−1 to 3) 0.40 14 y −1 (−3 to 1) 0.31 0 (−2 to 1) 0.75 0 (−2 to 2) 0.97 0 (−2 to 2) 0.98 AST Total physical activity 12 y 1 (0–1) 0.02 1 (0–1) 0.03 1 (0–2) 0.01 1 (0–2) 0.01 14 y 1 (0–1) 0.09 1 (0–1) 0.11 1 (0–1) 0.05 1 (0–1) 0.05 MVPA 12 y 1 (0–2) 0.05 1 (0–2) 0.05 2 (0–3) 0.01 2 (0–3) 0.01 14 y 1 (0–2) 0.13 1 (0–2) 0.13 1 (0–3) 0.07 1 (0–3) 0.07 GGT Total physical activity 12 y −1 (−2 to 0) <0.01 −1 (−2 to 0) <0.01 0 (−1 to 0) 0.25 0 (−1 to 0) 0.23 14 y −1 (−2 to 0) 0.21 −1 (−2 to 0) 0.21 0 (−1 to 1) 0.91 0 (−1 to 0) 0.23 MVPA 12 y −3 (−4 to −1) <0.01 −3 (−4 to −1) <0.01 −1 (−3 to 0) 0.15 0 (−3 to 0) 0.13 14 y −1 (−2 to 1) 0.31 −1 (−2 to 1) 0.31 0 (−1 to 2) 0.90 0 (−1 to 2) 0.94 Coefficients are per increase of 100 CPM (total physical activity) or per 15-minute increase in MVPA. Unadjusted results (model 1) are provided in Table F of the online supplement. ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CI = confidence interval; CPM = counts per minute; GGT = γ-glutamyl transferase; HOMA-IR = homeostatic model assessment of insulin resistance; MVPA = moderate to vigorous physical activity.

ided in Table F of the online supplement. ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CI = confidence interval; CPM = counts per minute; GGT = γ-glutamyl transferase; HOMA-IR = homeostatic model assessment of insulin resistance; MVPA = moderate to vigorous physical activity. *Model adjusts for mother's age at delivery, parity, sex, ethnicity, mother's education, head of household, social class, mother's BMI, energy intake at age 11 years, pubertal status at age 11 years, age at physical activity assessment, length of time accelerometer was worn (in minutes), and age at the time of liver assessment. †Model adjusts for all the potential confounders listed above plus fat mass, height, and height2 at the time physical activity was assessed. ‡Model adjusts for all the potential confounders listed above plus fat mass, height, and height2 at the time of liver outcome assessment. §Model adjusts for all the potential confounders listed above plus fat mass, height, height2, and HOMA-IR at the time of liver outcome assessment.

What Is KnownAmino acid-based formulae are well-tolerated, effective, and safe, and effectively promote growth in infants with cow's milk allergy. What Is NewThe present study shows that an amino acid-based formula is effective in providing adequate dietary mineral intake and maintaining mineral status in infants with cow's milk allergy. These data add to the understanding of the nutritional status of infants with cow's milk allergy consuming an amino acid-based formula.

What Is NewThe present study shows that an amino acid-based formula is effective in providing adequate dietary mineral intake and maintaining mineral status in infants with cow's milk allergy. These data add to the understanding of the nutritional status of infants with cow's milk allergy consuming an amino acid-based formula. Cow's milk allergy (CMA) is the most common food allergy in infancy, affecting up to 5% of infants in their first year of life (1). The symptoms of food allergy are gastrointestinal, cutaneous, and respiratory (1). Fundamental to CMA management is complete elimination of offending proteins. For infants who are breast-fed, this requires mothers eliminating all cow's milk protein from their diet. If breast-feeding is not possible, a suitable cow's milk-free formula is required. Hypoallergenic formulae such as extensively hydrolyzed formula (eHF) or amino acid-based formula (AAF) should be the formula of choice (2). An eHF is recommended for infants with mild to moderate CMA, whereas initiation with an AAF is recommended for more severe or complex allergy or when symptom-resolution with eHF fails (2). Dietary management of CMA with an AAF has been proven to be well-tolerated, effective, and safe; adequate infant growth on an AAF has widely been reported (3–5). Data on mineral status of infants consuming AAF have, to our knowledge, not, however, been published. Yet, eliminating dairy products in diets of CMA infants, as well as coexisting feeding problems, can result in removing a substantial source of minerals, including calcium and phosphorus, which are involved in essential body functions (6). The aim of this study was to assess mineral status (calcium, phosphorus, chloride, sodium, potassium, magnesium, and iron) of infants receiving an AAF for 16 weeks. In addition to this, dietary mineral intakes were estimated and evaluated against Adequate Intakes (AIs).

ich are involved in essential body functions (6). The aim of this study was to assess mineral status (calcium, phosphorus, chloride, sodium, potassium, magnesium, and iron) of infants receiving an AAF for 16 weeks. In addition to this, dietary mineral intakes were estimated and evaluated against Adequate Intakes (AIs). METHODS In a prospective, randomized, double-blind controlled study, full-term infants age 0 to 8 months with confirmed Immunoglobulin E (IgE) or non-IgE-mediated CMA were randomized to receive an AAF (n = 110) with or without synbiotics (short-chain and long-chain oligosaccharides [1.0 g/100 kcal], Bifidobacterium breve M-16 V [2.16 × 109 CFU/100 kcal]) for 16 weeks (3,4). Details about the study formulae (Neocate; SHS International Ltd., Nutricia Advanced Medical Nutrition, Liverpool, UK) can be found in Supplemental Digital Content, Table 1. CMA was diagnosed according to specified criteria listed in Supplemental Digital Content, Table 2. Infants could have received dietary treatment with an alternative commercially available hypoallergenic formula before study entry. Solid foods free of milk and other allergenic proteins dependent on documented allergies were part of the subject's diet as advised by their physician. Subjects were recruited between April 2008 and March 2012 within 29 participating sites in the United SA.

mercially available hypoallergenic formula before study entry. Solid foods free of milk and other allergenic proteins dependent on documented allergies were part of the subject's diet as advised by their physician. Subjects were recruited between April 2008 and March 2012 within 29 participating sites in the United SA. Primary outcomes were growth and formula tolerance and have been reported previously (3,4). The clinical trial was registered as NCT00664768 and medical ethical approval was obtained by the Chesapeake Institutional Review Board (Columbia, MD). All parents/guardians gave written informed consent for their children to participate. Mineral status was assessed by standard analyses of blood samples obtained at baseline and at 16 weeks, and included calcium, phosphorus, chloride, sodium, potassium, magnesium, and iron (ferritin). Furthermore, hemoglobin, albumin, and total protein were determined. These laboratory measurements are routinely used to assess the nutritional status of our target group (7). All blood samples were analyzed in 1 central laboratory (Covance Central, Laboratories Services Inc, Indianapolis, IN). Differences in blood chemistry parameters (at baseline, week 16 and change from baseline) between the study products have been analyzed before and were not found to be statistically significant or clinically relevant (3). Therefore, blood chemistry parameters are presented for the overall population.

Indianapolis, IN). Differences in blood chemistry parameters (at baseline, week 16 and change from baseline) between the study products have been analyzed before and were not found to be statistically significant or clinically relevant (3). Therefore, blood chemistry parameters are presented for the overall population. Average daily formula intake consumed by each subject was recorded in 3-day food diaries at weeks 2, 4, 8, and 16 of the study. From these intake data, individual energy and mineral intakes were calculated based on study product composition. Energy intakes were evaluated against Dietary Reference Values for energy of the UK Scientific Advisory Committee on Nutrition (8) and the US Institute of Medicine (IOM) (9). Due to the correlation between energy requirements and energy intake, it was not possible to estimate the prevalence of inadequate energy intake without information on individual energy requirements (9,10). Mineral intakes (calcium, phosphorus, chloride, sodium, potassium, magnesium, and iron) were evaluated against AIs defined by the European Food Safety Authority (EFSA) (11) and the IOM (12,13). The AI is the daily mean nutrient intake when healthy, full-term infants are consuming human milk, but cannot be used to estimate the proportion of infants with inadequate intakes (12). Therefore, intake was evaluated qualitatively. If median intake was above the AI, the prevalence of inadequate intake was stated as “low”. When this was not the case, the adequacy of the diet could not be evaluated (“no statement”) (10). To estimate total mineral intake, that is, intake from both formula and complementary foods, we used published intake data made by complementary foods (ie, all foods other than infant formula/human milk/cow's milk) from the UK Diet and Nutrition Survey of Infants and Young Children (14). In this survey, no estimates are given of the contribution of complementary foods to intakes of phosphorus, chloride, potassium, and magnesium, and therefore only intake data from formula are presented for these minerals.

a/human milk/cow's milk) from the UK Diet and Nutrition Survey of Infants and Young Children (14). In this survey, no estimates are given of the contribution of complementary foods to intakes of phosphorus, chloride, potassium, and magnesium, and therefore only intake data from formula are presented for these minerals. RESULTS Baseline characteristics of the enrolled subjects are presented in Supplemental Digital Content, Table 3. Average age of infants at inclusion was 4.6 ± 2.5 months (mean ± SD). Out of the 110 infants included in the study, 82 (75%) had blood parameters analyzed at baseline and 66 (60%) after 16 weeks on AAF. Reasons for dropout are presented in Supplemental Digital Content 4, Figure 1. At baseline and after 16 weeks on AAF, mean blood concentrations of all minerals were within the specified reference ranges set for the corresponding ages of the infants (Table 1). Among some minerals, there was a number of individual values at baseline that were below the reference ranges for age, that is, calcium (n = 1), phosphorus (n = 1), chloride (n = 1), sodium (n = 1), and ferritin (n = 6), whereas at week 16 only ferritin concentrations remained below the reference range for age for a number of individuals (n = 15) (Table 1). Mean values of hemoglobin, albumin and total protein were within reference ranges for age at baseline and after 16 weeks (Table 1).

de (n = 1), sodium (n = 1), and ferritin (n = 6), whereas at week 16 only ferritin concentrations remained below the reference range for age for a number of individuals (n = 15) (Table 1). Mean values of hemoglobin, albumin and total protein were within reference ranges for age at baseline and after 16 weeks (Table 1). The median daily intake of energy ranged between 467 and 588 kcal/day in boys and between 421 and 515 kcal/day in girls younger than 6 months of age, that is, below the age-group specific estimated energy requirement (EER) (Supplemental Digital Content 5, Table 4). For infants older than 6 months, the gap between energy intake from formula and EER was bigger. This was especially true for boys age 10 and 12 months who met approximately 60% of their EER from formula (vs 94% at 0–3 months), compared with 80% in girls of the same age (vs 85% at 0–3 months).

gital Content 5, Table 4). For infants older than 6 months, the gap between energy intake from formula and EER was bigger. This was especially true for boys age 10 and 12 months who met approximately 60% of their EER from formula (vs 94% at 0–3 months), compared with 80% in girls of the same age (vs 85% at 0–3 months). The median intakes of the minerals for subjects below 6 months are presented in Table 2. Full intake distributions for all ages are provided in Supplemental Digital Content 6, Table 5. For all minerals, median intakes of infants age 0 to 6 months was above the AI, indicating a low prevalence of inadequate intakes (Table 2). For calcium and phosphorus (Supplemental Digital Content 6, Table 5A and B), median intakes from formula in both boys and girls age 7 to 12 months was above the AI indicating a low prevalence of inadequate intakes. For iron (Supplemental Digital Content 6, Table 5C), median intakes from formula alone in boys and girls age 7 to 12 months was above or close to the AI defined by the EFSA (8 mg/day) and median intakes from formula and complementary foods was above or close to the AI defined by the IOM (11 mg/day). Overall, suggesting a low prevalence of inadequate iron intake.

ble 5C), median intakes from formula alone in boys and girls age 7 to 12 months was above or close to the AI defined by the EFSA (8 mg/day) and median intakes from formula and complementary foods was above or close to the AI defined by the IOM (11 mg/day). Overall, suggesting a low prevalence of inadequate iron intake. Median sodium intakes from formula alone ranged between 164 and 209 mg/day in 7 to 12-month-old boys and between 187 and 254 mg/day in 7 and 12-month-old girls. Taking intake from complementary foods into account resulted in an increase of the median intake to the range of 424 to 614 mg/day for boys and 432 to 659 mg/day for girls (Supplemental Digital Content 6, Table 5E), that is, a level above the AI of 170 to 370 mg/day, indicating a low prevalence of inadequate sodium intake. For chloride, potassium and magnesium (Supplemental Digital Content 6, Table 5D to G), median intakes of formula alone in 7 to 12-month-olds were below the AI and therefore no statement about the prevalence of inadequate intakes could be made. Due to the absence of data in the UK Diet and Nutrition Survey, intake from complementary foods could not be considered for these minerals. DISCUSSION The present study shows that AAF with or without synbiotics, which have been reported previously to be equally tolerated and support normal growth (3,4), as part of an age-adapted diet, are effective in providing an adequate mineral intake and status in CMA infants.

Median sodium intakes from formula alone ranged between 164 and 209 mg/day in 7 to 12-month-old boys and between 187 and 254 mg/day in 7 and 12-month-old girls. Taking intake from complementary foods into account resulted in an increase of the median intake to the range of 424 to 614 mg/day for boys and 432 to 659 mg/day for girls (Supplemental Digital Content 6, Table 5E), that is, a level above the AI of 170 to 370 mg/day, indicating a low prevalence of inadequate sodium intake. For chloride, potassium and magnesium (Supplemental Digital Content 6, Table 5D to G), median intakes of formula alone in 7 to 12-month-olds were below the AI and therefore no statement about the prevalence of inadequate intakes could be made. Due to the absence of data in the UK Diet and Nutrition Survey, intake from complementary foods could not be considered for these minerals. DISCUSSION The present study shows that AAF with or without synbiotics, which have been reported previously to be equally tolerated and support normal growth (3,4), as part of an age-adapted diet, are effective in providing an adequate mineral intake and status in CMA infants. The vast majority of infants age 0 to 6 months (formula only) and age 7 to 12 months (formula and complementary foods) had mineral intakes above the AI, indicating a high likelihood of achieving adequate intakes. For those infants with a mineral intake below the AI, this was related to a lower formula volume intake compared with what would have been expected based on the infant's EER (data not shown). Intake levels of calcium and phosphorus from formula alone were far above AIs set by the EFSA and IOM, especially in the first half year of infants’ life. The AAF used in this study is suitable for the whole first year of life, thus including the weaning period. The level of calcium and phosphorus in the AAF should therefore compensate for the low intakes of these nutrients in the merely dairy-free complementary diet of CMA infants. Moreover, compared with other countries (15) or Institutes (16), EFSA, and IOM set relatively low AIs for calcium and phosphorus. Neither the IOM (12,13) nor the EU Scientific Committee on Food (17) set a Tolerable Upper Intake Level (UL) of calcium and phosphorus for infants, but the P95 intake levels of both minerals remain far below the ULs of calcium and phosphorus for children age 1 to 3 years. Calcium and phosphorus levels in the study formulae comply with all compositional standards for infant formula applicable in the USA, EU, China, and Brazil.

of calcium and phosphorus for infants, but the P95 intake levels of both minerals remain far below the ULs of calcium and phosphorus for children age 1 to 3 years. Calcium and phosphorus levels in the study formulae comply with all compositional standards for infant formula applicable in the USA, EU, China, and Brazil. Mineral status was within reference ranges for the vast majority of infants. At baseline, a few infants had individual blood mineral concentrations below reference ranges, whereas at week 16 all mineral concentrations, with the exception of ferritin, were within the age-specific range. Serum ferritin concentrations decreased during the study, a phenomenon also seen in healthy infants as they grow older (18). Prevalence of depleted iron stores (serum ferritin <12 μg/L (19)) found after 16 weeks of study, that is, 22%, are similar to those found in healthy infants (20). Other indices for assessing iron status, such as hematocrit, mean corpuscular volume, and total iron binding capacity, were all within reference ranges for age (data not shown) (3). Serum phosphorus concentrations also dropped slightly over the time of the study, in line with documented declines in phosphorus concentrations from 6 to 12 months of age (21), but remained above age-specific reference ranges.

rmula-fed infants, the staple complementary food in Tanzania is maize-based porridge, which in itself is a poor source of vitamin D; however, cow's milk, condensed milk, meat, and other vitamin D containing foods are sometimes added or consumed separately (24). We did not have data available on sun exposure of infants. Vitamin D is a potent immunomodulator with effects on both adaptive and innate immune responses and as a result may alter the incidence and severity of viral and bacterial infections (25). In this study, we found no statistically significant relationship of infant vitamin D status at 6 weeks or 6 months of age with incidence of diarrhea, URIs, or clinical malaria. In our previous study of Tanzanian HIV-exposed uninfected infants, which was also conducted in Dar es Salaam, we found no significant association of vitamin D at 6 weeks of age with incidence of diarrhea and ALRI, but 25(OH)D concentrations ≥30 ng/mL were associated with increased risk of clinical malaria (9). It is possible that our previous study findings indicating greater levels of vitamin D may increase risk of malaria were due to confounding, occurred by chance, or are specific to HIV-exposed populations. Only one randomized trial has examined the effect of infant vitamin D supplementation on the incidence and severity of morbidities in LMICs (26). In this trial conducted among Afghani infants (1–11 months of age) who were hospital admitted with pneumonia, vitamin D supplements significantly increased the incidence of repeat episodes of pneumonia as compared with placebo (26). In this study, we also found infants with 25(OH)D concentrations ≥30 ng/mL at 6 months of age had increased risk of ALRI as compared to infants with concentrations of 20 to 29.9 ng/mL. Consequently, the existing, even though very limited, observational and trial evidence do not support vitamin D supplementation among the general population of infants in LMICs to reduce common morbidities.

d total iron binding capacity, were all within reference ranges for age (data not shown) (3). Serum phosphorus concentrations also dropped slightly over the time of the study, in line with documented declines in phosphorus concentrations from 6 to 12 months of age (21), but remained above age-specific reference ranges. This study has some limitations. First, the high number of subjects (40%) not available for analysis of the blood parameters. Baseline characteristics of the dropouts were comparable to the subjects included in the final analysis and the dropout rate was equal in both study arms. Therefore, it is not likely that this has influenced our findings. Secondly, in our study, some infants were already receiving an AAF at enrolment, and thus, baseline blood values cannot be strictly considered as “before intervention” values. Nevertheless, according to the study aim, the present analyses provide evidence that after 16 weeks of management with an AAF, mineral status of all participating infants was adequate. Finally, the use of the UK Diet and Nutrition Survey (14) has a limitation since it only presents data for a selected number of minerals. Moreover, estimates of nutrient intakes from complementary foods from this survey including healthy infants may be an overestimate for CMA infants given that food intakes overall may be poorer and consequences of excluding milk from the diet affects consumption of other food groups, such as breakfast cereals. This survey data have been used as a reasonable approximation in the absence of data on intakes of complementary foods by CMA infants.

imate for CMA infants given that food intakes overall may be poorer and consequences of excluding milk from the diet affects consumption of other food groups, such as breakfast cereals. This survey data have been used as a reasonable approximation in the absence of data on intakes of complementary foods by CMA infants. In conclusion, these data further contribute to the knowledge about nutritional status and disease management of CMA infants by showing that an AAF with or without synbiotics is effective in providing an adequate mineral status in these infants. To assure normal-for-age mineral status by all CMA infants, it is necessary to provide them with regular medical and nutritional care to ensure adequate formula intakes and, if needed, additional supplementation. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content This research was supported by Nutricia Research, Nutricia Advanced Medical Nutrition. S.R.B.M.E. and L.F.H. are employees of Nutricia Research. A.W.B. received grants from the National Institutes of Health (NIH; Bethesda, MD), Wallace Research Foundation, and Nutricia North America and is consultant for Merck and McNeill Nutritionals. B.M.H. declares no conflict of interest. TABLE 1 Blood chemistry parameters at baseline (n = 82) and after 16 weeks on amino acid-based formula (n = 66)

S.R.B.M.E. and L.F.H. are employees of Nutricia Research. A.W.B. received grants from the National Institutes of Health (NIH; Bethesda, MD), Wallace Research Foundation, and Nutricia North America and is consultant for Merck and McNeill Nutritionals. B.M.H. declares no conflict of interest. TABLE 1 Blood chemistry parameters at baseline (n = 82) and after 16 weeks on amino acid-based formula (n = 66) Baseline After 16 weeks on AAF Parameter Reference value Mean (SD) n (%) Mean (SD) n (%) Ca, mmol/L 2.25–2.74 2.67 (0.14) 1 (1%) 2.62 (0.11) 0 P, mmol/L 1.36–2.62 (<1 y) 2.05 (0.25) 1 (1%) 1.97 (0.20) 0 1.03–1.97 (≥1 y) — 1.86 (0.24) 0 Cl, mmol/L 94–112 104 (3.2) 1 (1%) 104 (2.3) 0 Na, mmol/L 132–147 140 (3.3) 1 (1%) 140 (2.3) 0 K, mmol/L 3.7–5.6 (<1 y) 5.0 (0.52) 0 4.6 (0.29) 0 3.4–5.4 (≥1 y) — 4.6 (0.48) 0 Mg, mmol/L 0.70–0.98 (<30 day) 0.89 (0.05) 0 0.92 (0.05) 0 0.66–1.03 (M; ≥30 day) 0.95 (0.06) 0 0.95 (0.07) 0 0.78–0.98 (F; ≥30 day) 0.95 (0.05) 0 0.96 (0.07) 0 Ferritin, μg/L ≥12 61 (55) 6 (7%) 24 (18) 15 (22%) Hemoglobin, g/L 100–200 (<30 day) 168 (4.6) 0 — — 100–140 (<5 mo) 115 (11) 4 (7%) 122 (8.5) 0 105–135 (≥5 mo) 119 (9.4) 2 (8%) 121 (9.0) 2 (4%) Albumin, g/L 24–48 (<30 day) 38 (2.3) 0 — 28–48 (≥30 day) 39 (5.2) 3 (4%) 41 (3.7) 0 Total protein, g/L 54–74 (<3 mo) 58 (3.4) 1 (4%) 55–71 (≥3 mo) 61 (6.5) 5 (9%) 64 (4.3) 0 Mean (SD) and number (%) of infants having a mineral status below the lowest range of the reference value. AAF = amino acid-based formula; Ca = calcium; Cl = chloride; K = potassium; Mg = magnesium; Na = sodium; P = phosphorus; y = years.

Baseline After 16 weeks on AAF Parameter Reference value Mean (SD) n (%) Mean (SD) n (%) Ca, mmol/L 2.25–2.74 2.67 (0.14) 1 (1%) 2.62 (0.11) 0 P, mmol/L 1.36–2.62 (<1 y) 2.05 (0.25) 1 (1%) 1.97 (0.20) 0 1.03–1.97 (≥1 y) — 1.86 (0.24) 0 Cl, mmol/L 94–112 104 (3.2) 1 (1%) 104 (2.3) 0 Na, mmol/L 132–147 140 (3.3) 1 (1%) 140 (2.3) 0 K, mmol/L 3.7–5.6 (<1 y) 5.0 (0.52) 0 4.6 (0.29) 0 3.4–5.4 (≥1 y) — 4.6 (0.48) 0 Mg, mmol/L 0.70–0.98 (<30 day) 0.89 (0.05) 0 0.92 (0.05) 0 0.66–1.03 (M; ≥30 day) 0.95 (0.06) 0 0.95 (0.07) 0 0.78–0.98 (F; ≥30 day) 0.95 (0.05) 0 0.96 (0.07) 0 Ferritin, μg/L ≥12 61 (55) 6 (7%) 24 (18) 15 (22%) Hemoglobin, g/L 100–200 (<30 day) 168 (4.6) 0 — — 100–140 (<5 mo) 115 (11) 4 (7%) 122 (8.5) 0 105–135 (≥5 mo) 119 (9.4) 2 (8%) 121 (9.0) 2 (4%) Albumin, g/L 24–48 (<30 day) 38 (2.3) 0 — 28–48 (≥30 day) 39 (5.2) 3 (4%) 41 (3.7) 0 Total protein, g/L 54–74 (<3 mo) 58 (3.4) 1 (4%) 55–71 (≥3 mo) 61 (6.5) 5 (9%) 64 (4.3) 0 Mean (SD) and number (%) of infants having a mineral status below the lowest range of the reference value. AAF = amino acid-based formula; Ca = calcium; Cl = chloride; K = potassium; Mg = magnesium; Na = sodium; P = phosphorus; y = years. TABLE 2 Median (p50) intakes of minerals from formula alone for 0 to 3-month-old boys, 4 to 6-month-old boys, 0 to 3-month-old girls and 4 to 6-month-old girls compared with Adequate Intakes Set by the European Food Safety Authority and US Institute of Medicine

AAF = amino acid-based formula; Ca = calcium; Cl = chloride; K = potassium; Mg = magnesium; Na = sodium; P = phosphorus; y = years. TABLE 2 Median (p50) intakes of minerals from formula alone for 0 to 3-month-old boys, 4 to 6-month-old boys, 0 to 3-month-old girls and 4 to 6-month-old girls compared with Adequate Intakes Set by the European Food Safety Authority and US Institute of Medicine Boys Girls 0–3-month olds 4–6-month olds 0–3-month olds 4–6-month olds EFSA AI US AI Prevalence of inadequate intake Study visit 2 4 8 2 4 8 16 2 4 8 2 4 8 16 Calcium, mg 551 616 694 618 589 678 694 513 497 507 560 608 581 585 200 200 Low Phosphorus, mg 390 436 491 438 417 480 491 363 352 359 396 431 411 414 100 100 Low Iron, mg 7.0 7.8 8.8 7.8 7.4 8.6 8.8 6.5 6.3 6.4 9.0 9.7 9.3 9.4 0.3 0.27 Low Chloride, mg 379 424 477 425 405 467 477 353 342 349 385 418 399 402 300 180 Low Sodium, mg 185 207 233 208 198 228 233 172 167 170 188 205 195 197 120 120 Low Potassium, mg 517 578 651 580 552 636 651 481 466 475 525 570 545 549 400 400 Low Magnesium, mg 50 56 63 56 54 62 63 47 45 46 51 55 53 53 25 30 Low AI = adequate intake; EFSA = European Food Safety Authority.

What Is KnownAntibiotics are growth-promoting in livestock, and early life exposure to antibiotics has been recently associated with obesity in several high-income countries. The effect of antibiotic use may be different among children in highly contaminated environments in low-income settings in which undernutrition is more common than obesity. What Is NewAntibiotic-associated growth promotion can also occur among children in low-resource settings who are often undernourished. Antibiotic use in the first 6 months of life was associated with increased weight, especially among children with 2 or more exposures to macrolides, metronidazole, and cephalosporins.

The effect of antibiotic use may be different among children in highly contaminated environments in low-income settings in which undernutrition is more common than obesity. What Is NewAntibiotic-associated growth promotion can also occur among children in low-resource settings who are often undernourished. Antibiotic use in the first 6 months of life was associated with increased weight, especially among children with 2 or more exposures to macrolides, metronidazole, and cephalosporins. After decades of antibiotic use to promote growth of livestock in the agricultural industry, researchers have begun to explore the potential role of antibiotic use in promoting growth in humans as well (1). Evidence of the potential for antibiotics to cause long-lasting perturbations of the gut microbiota (2,3) and the role of the microbiota in metabolism (4–6) has suggested that this exposure could be an important research target as a modifiable factor that may be contributing to the obesity epidemic in high-income countries. Antibiotic exposure early in life, during the maturation process of the gut microbiota and enteric immune system, has been hypothesized to have the largest potential impact on growth (6,7). Epidemiologic studies investigating the role of early life antibiotic exposures in high-income settings of Finland (8), United Kingdom (9,10), United States (11,12), The Netherlands (13), and Denmark (14) have shown associations between early antibiotic exposure and increased weight gain in children from 2 to 10 years of age, although some found mixed or negative results (14,15).

ibiotic exposures in high-income settings of Finland (8), United Kingdom (9,10), United States (11,12), The Netherlands (13), and Denmark (14) have shown associations between early antibiotic exposure and increased weight gain in children from 2 to 10 years of age, although some found mixed or negative results (14,15). In low-income settings, obesity among young children is less common, and poor growth and undernutrition is often a more pressing public health concern. In these settings, antibiotics are recommended by the WHO as part of the treatment for severe acute malnutrition (16), despite the mixed evidence for the impact of antibiotic treatment on recovery and mortality (17,18). Outside of clinical settings, it is not clear whether antibiotic use promotes growth in children who are not acutely malnourished but remain below the average growth curve and reside in environments with high pathogen exposure. In these settings, antibiotics may affect growth through clearance of infections or through modifications of the microbiota, the hypothesized mechanism in high-income settings (19,20). Unique to low-income settings, there is additional confounding in observational studies by the indications for treatment, because some of the most common illnesses resulting in antibiotic treatment (eg, diarrhea) can have long-term negative impacts on growth. An international cross-sectional study that found a positive association between antibiotic treatment during infancy and BMI among boys did not include these factors and was also limited by self-reported heights and weights (21).

ting in antibiotic treatment (eg, diarrhea) can have long-term negative impacts on growth. An international cross-sectional study that found a positive association between antibiotic treatment during infancy and BMI among boys did not include these factors and was also limited by self-reported heights and weights (21). To appropriately account for these factors, we leveraged the high-resolution illness and treatment data from The Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development Project (MAL-ED), a birth cohort study performed at 8 sites in South America, sub-Saharan Africa, and south Asia. Antibiotic use was highly prevalent in almost all sites (22). We aimed to assess whether early antibiotic use in the first 6 months of life was associated with physical growth from 6 months to 2 years of age and determine the impact of antibiotic class and duration of use.

h America, sub-Saharan Africa, and south Asia. Antibiotic use was highly prevalent in almost all sites (22). We aimed to assess whether early antibiotic use in the first 6 months of life was associated with physical growth from 6 months to 2 years of age and determine the impact of antibiotic class and duration of use. METHODS The MAL-ED study was conducted between November 2009 and February 2014 at 8 sites in Dhaka, Bangladesh, Fortaleza, Brazil, Vellore, India, Bhaktapur, Nepal, Naushahro Feroze, Pakistan, Loreto, Peru, Venda, South Africa, and Haydom, Tanzania. Study design and methods have been previously described (23). Children were enrolled within 17 days of birth if their enrollment weight was ≥1500 g, they were not hospitalized, and they did not have severe or chronic conditions. Follow-up was conducted twice per week at home visits until 2 years of age to document illnesses, breast-feeding practices, and antibiotic use. Caregivers reported all oral or injected antibiotics given to their child. Medication packaging and prescriber documentation were used to confirm antibiotic use and class. Non-diarrheal surveillance stool samples were collected monthly and tested for 40 enteropathogens (24). Weight and length were measured monthly, and weight-for-age (WAZ) and length-for-age (LAZ) z scores were calculated using the 2006 WHO child growth standards (25). Length measurements from Pakistan were excluded due to poor data quality. Socioeconomic status was assessed biannually and summarized using the child's average WAMI (Water, Assets, Maternal education, Income) score, which is based on monthly household income, maternal education, wealth measured by 8 assets, and access to improved water and sanitation (26). All sites received ethical approval from their respective governmental, local institutional, and collaborating institutional ethical review boards. Written, informed consent was obtained from the caregiver of each child.

maternal education, wealth measured by 8 assets, and access to improved water and sanitation (26). All sites received ethical approval from their respective governmental, local institutional, and collaborating institutional ethical review boards. Written, informed consent was obtained from the caregiver of each child. Data and Definitions Assessment of antibiotic use practices in the MAL-ED cohort has been previously described (22). A new antibiotic course was defined after 2 antibiotic-free days, assuming antibiotics were not received on the 2% of days with missing surveillance information. Diarrhea was defined as maternal report of 3 or more loose stools in 24 hours or at least 1 loose stool with visible blood (27). Respiratory illness was defined as cough or shortness of breath, and was considered an acute lower respiratory infection if accompanied by fieldworker-determined rapid respiratory rate (27). Fever and vomiting were caregiver reported.

3 or more loose stools in 24 hours or at least 1 loose stool with visible blood (27). Respiratory illness was defined as cough or shortness of breath, and was considered an acute lower respiratory infection if accompanied by fieldworker-determined rapid respiratory rate (27). Fever and vomiting were caregiver reported. Analysis We used multivariable linear regression to estimate the association between antibiotic use in the first 6 months of life and monthly WAZ and LAZ from 6 to 24 months of age. Antibiotic exposure was modeled as a continuous measure of duration in days from 0 to 5 months of age and as a categorical variable by number of courses received to assess the potential for a nonlinear dose-response. We also stratified antibiotic effects by sex and site. Generalized estimating equations with robust variance were used to account for correlation between anthropometric measurements within children across time points. Confounding variables for adjustment included baseline characteristics and indications for treatment, and were selected by causal diagram (28) based on expert opinion and a previous analysis of factors associated with antibiotic use in MAL-ED (22). All analyses were adjusted for site, child sex, enrollment WAZ, WAMI score, crowding (people/room in household), maternal height, maternal education, and characteristics of the child's first 6 months of life: percent days exclusively breast-fed; number of diarrhea episodes; days with fever, vomiting, and respiratory illness; and presence of acute lower respiratory infection, bloody stools, and hospitalization. Length models also included enrollment LAZ.

l education, and characteristics of the child's first 6 months of life: percent days exclusively breast-fed; number of diarrhea episodes; days with fever, vomiting, and respiratory illness; and presence of acute lower respiratory infection, bloody stools, and hospitalization. Length models also included enrollment LAZ. We further explored the effects of class-specific antibiotic exposure use by modeling class-specific exposure as dichotomous (exposed to a specific class on at least 1 day vs not) and as a categorical variable by number of class-specific courses received, using the models above and additionally adjusting for other antibiotic class exposures to isolate class-specific effects. We also explored effect measure modification by malnutrition (stunted and underweight) at 6 months and pathogen burden from 0 to 5 months (presence of Campylobacter, enteroaggregative Escherichia coli, and Giardia, and average number of bacterial pathogens detected (24)) by including interaction terms between the exposures and these variables and estimating subgroup-specific effects. To assess the period during which the effects of early antibiotic use (before 6 months of age) were manifested, we estimated these effects on anthropometry at different age periods, from 0 to 5, 6 to 11, 12 to 17, and 17 to 24 months, adjusting for the child's anthropometric z scores at the beginning of the age period.

To assess the period during which the effects of early antibiotic use (before 6 months of age) were manifested, we estimated these effects on anthropometry at different age periods, from 0 to 5, 6 to 11, 12 to 17, and 17 to 24 months, adjusting for the child's anthropometric z scores at the beginning of the age period. To compare early life exposures with later exposures, we used linear regression to estimate the effects of exposures from 6 to 24 months on cross-sectional WAZ and LAZ at 2 years. A child's measurement closest to 24 months, between 23 and 25 months was considered their anthropometry at 2 years. Adjustment variables included the same baseline characteristics as above, including enrollment WAZ, enrollment LAZ (for length models only), illness burden as characterized above over the whole 2 years of follow-up, and antibiotic use in the first 6 months of life. Using these models with cross-sectional outcomes of WAZ and LAZ at 2 years, we demonstrated that the early life effects were insensitive to further statistical adjustment for illnesses and antibiotic use after 6 months of age.

above over the whole 2 years of follow-up, and antibiotic use in the first 6 months of life. Using these models with cross-sectional outcomes of WAZ and LAZ at 2 years, we demonstrated that the early life effects were insensitive to further statistical adjustment for illnesses and antibiotic use after 6 months of age. RESULTS Across 8 sites in the MAL-ED cohort, 1954 children were followed until at least 6 months of age and had at least 1 subsequent anthropometric measurement. The majority of these children (n = 1736, 88.3%) remained under surveillance and had anthropometric measurements at 2 years. Baseline characteristics, early antibiotic exposure, and growth outcomes differed across sites (Table 1), with the highest antibiotic use in the South Asian sites (22). Mean enrollment weight and length within 17 days of birth across sites were near 1 standard deviation below the WHO standard. All sites except Brazil showed reductions in average WAZ and LAZ over the 2 years of follow-up with overall means at 2 years of −1.06 and −1.71, respectively.

use in the South Asian sites (22). Mean enrollment weight and length within 17 days of birth across sites were near 1 standard deviation below the WHO standard. All sites except Brazil showed reductions in average WAZ and LAZ over the 2 years of follow-up with overall means at 2 years of −1.06 and −1.71, respectively. A 7-day increase in duration of antibiotic exposure in the first 6 months of life was associated with an adjusted 0.03 (95% confidence interval [CI]: 0.00, 0.05) higher WAZ from 6 months to 2 years of age compared to unexposed children. There was no difference between boys and girls, and this association was largely consistent across sites, except in Brazil and South Africa, where antibiotic use was least common and the estimates are least precise (Fig. 1A). The largest differences in WAZ were associated with children who received >3 courses in the first 6 months (Supplemental Digital Content 2, Fig. 2A). Children who received >3 courses of antibiotics in the first 6 months of life had an average 0.18 (95% CI: 0.06, 0.30) higher WAZ from 6 months to 2 years compared to children receiving 3 courses or less. This effect was largest in Pakistan (WAZ difference: 0.46, 95% CI: 0.25, 0.67) and Tanzania (WAZ difference: 0.21, 95% CI: −0.01, 0.43), but these estimates translate to relatively small differences in weight (600 and 280 g at 2 years, respectively).

6 months to 2 years compared to children receiving 3 courses or less. This effect was largest in Pakistan (WAZ difference: 0.46, 95% CI: 0.25, 0.67) and Tanzania (WAZ difference: 0.21, 95% CI: −0.01, 0.43), but these estimates translate to relatively small differences in weight (600 and 280 g at 2 years, respectively). FIGURE 1 Adjusted weight-for-age (WAZ) (A) and length-for-age (LAZ) (B) z score differences associated with a linear 7-day increase in duration of antibiotic exposure in the first 6 months of life among 1954 children followed in the MAL-ED birth cohort until at least 6 months of age with subsequent anthropometry. Sites ordered from greatest to least proportion exposed to antibiotics <6 months. Pakistan: 261/265 (98.5%); Bangladesh: 236/240 (98.3%); Tanzania: 195/245 (79.6%); Peru: 211/269 (78.4%); India: 147/234 (62.8%); Nepal: 129/235 (54.9%); South Africa: 80/259 (13.3%); Brazil: 25/207 (12.1%); All sites: 1284/1954 (65.7%). The associations of duration of antibiotic exposure with LAZ were more varied across sites (Fig. 1B). There was no difference in LAZ between children exposed to between 0 and 4 courses of antibiotics. Children exposed to 5 or more courses of antibiotics before 6 months had a slightly higher but nonsignificant increase in LAZ compared to unexposed children (Fig., Supplemental Digital Content 2, Fig. 2B).

ites (Fig. 1B). There was no difference in LAZ between children exposed to between 0 and 4 courses of antibiotics. Children exposed to 5 or more courses of antibiotics before 6 months had a slightly higher but nonsignificant increase in LAZ compared to unexposed children (Fig., Supplemental Digital Content 2, Fig. 2B). In the assessment of class-specific exposure, macrolide and metronidazole use on at least 1 day during the first 6 months were associated with an adjusted 0.14 (95% CI: 0.02, 0.25) and 0.17 (95% CI: 0.04, 0.31) increase in WAZ, respectively. Cephalosporins, fluoroquinolones, and penicillins were associated with smaller and nonsignificant increases in WAZ. Associations with LAZ were generally smaller and not statistically significant for any antibiotic class (Table 2). Although the effects of macrolides and cephalosporins were close to null if only 1 course was received, 2 or more courses were associated with an adjusted increase of 0.23 (95% CI: 0.05, 0.42) and 0.19 (95% CI: 0.04, 0.35) in WAZ, respectively (Table 2). Metronidazole was associated with increases in WAZ both when children were exposed to only 1 course (WAZ difference: 0.14, 95% CI: −0.00, 0.29) and 2 or more courses (0.23, 95% CI: 0.05, 0.42). In contrast, a dose-response pattern was absent for the effects on LAZ, except for cephalosporins, in which 2 or more courses of cephalosporins were associated with a 0.19 (95% CI: 0.01, 0.37) increase in LAZ (Table 2). There was no statistical evidence for effect measure modification by malnutrition or pathogen burden in surveillance stools (not shown).

was absent for the effects on LAZ, except for cephalosporins, in which 2 or more courses of cephalosporins were associated with a 0.19 (95% CI: 0.01, 0.37) increase in LAZ (Table 2). There was no statistical evidence for effect measure modification by malnutrition or pathogen burden in surveillance stools (not shown). The weight and length increases associated with antibiotic use before 6 months of age occurred during the exposure window and up to 1 year later at 18 months of age. By 6 months, children who were exposed to >3 courses of antibiotics had a 0.09 (95% CI: −0.04, 0.22) greater WAZ compared to children receiving 3 courses or less (Supplemental Digital Content 3, Fig. 3A). These children had further gains of 0.08 (95% CI: −0.06, 0.06) and 0.04 (95% CI: −0.02, 0.05) in WAZ compared to children receiving 3 courses or less from 6 to 11 and 12 to 17 months, respectively. Adjusting for their WAZ at 18 months, there was no further difference in WAZ from 18 to 24 months associated with early antibiotic use (WAZ difference: 0.00, 95% CI: −0.04, 0.05). Children who did not, however, receive early antibiotics did not catch-up during this period such that the majority of the overall WAZ difference was maintained to 24 months. More than 3 courses of antibiotics was associated with a similar increase in LAZ by 6 months (0.08, 95% CI: −0.05, 0.21), but there were no associations in periods after 6 months (Supplemental Digital Content 3, Fig. 3B) or overall as shown above.

the majority of the overall WAZ difference was maintained to 24 months. More than 3 courses of antibiotics was associated with a similar increase in LAZ by 6 months (0.08, 95% CI: −0.05, 0.21), but there were no associations in periods after 6 months (Supplemental Digital Content 3, Fig. 3B) or overall as shown above. In contrast to the antibiotic effects in early infancy, duration of antibiotic exposure after 6 months of age were not associated with cross-sectional WAZ or LAZ at 2 years (Supplemental Digital Content 4, Table), adjusting for exposure before 6 months and illnesses across the first 2 years of life. Among the antibiotic classes, only fluoroquinolones were associated with increases in size at 2 years of age. Two or more courses of fluoroquinolones after 6 months of age were associated with an adjusted increase of 0.21 (95% CI: 0.05, 0.37) in WAZ at 2 years (Supplemental Digital Content 4, Table). The associations with other antibiotic classes were near the null and/or did not show a dose-response trend.

at 2 years of age. Two or more courses of fluoroquinolones after 6 months of age were associated with an adjusted increase of 0.21 (95% CI: 0.05, 0.37) in WAZ at 2 years (Supplemental Digital Content 4, Table). The associations with other antibiotic classes were near the null and/or did not show a dose-response trend. DISCUSSION We demonstrate that the growth-promoting phenomenon of early life antibiotic exposure among healthy children is not unique to high-resource settings and can also occur in populations with low average weights and lengths. Antibiotic exposure was associated with increases in weight in the MAL-ED cohort, and this effect was limited to early exposures in the first 6 months of life, which is consistent with previous studies (8,9,13,14). Associations of antibiotics with length were generally smaller and inconsistent across sites and drug classes. The greatest increases in WAZ were associated with >3 courses of exposure and when multiple treatment courses of macrolides, metronidazole, and cephalosporins were received.

with previous studies (8,9,13,14). Associations of antibiotics with length were generally smaller and inconsistent across sites and drug classes. The greatest increases in WAZ were associated with >3 courses of exposure and when multiple treatment courses of macrolides, metronidazole, and cephalosporins were received. A larger impact of macrolides compared to other antibiotics on BMI was similarly documented in a retrospective US study (12). Broad-spectrum antibiotics were associated with early childhood obesity in the United States, whereas narrow-spectrum antibiotics (penicillins) were not (11). Two previous trials of metronidazole treatment in the general pediatric population in Guatemala (1982) (29) and in malnourished children in Jamaica (1993) (30) found an association between metronidazole and improved growth. Although metronidazole is considered narrow-spectrum, its anaerobic activity may be particularly destructive to the gut microbiota. Antibiotics with anaerobic activity have been associated with growth in a dose-dependent manner in the United Kingdom, whereas antibiotics without anaerobic activity were not (10).

growth. Although metronidazole is considered narrow-spectrum, its anaerobic activity may be particularly destructive to the gut microbiota. Antibiotics with anaerobic activity have been associated with growth in a dose-dependent manner in the United Kingdom, whereas antibiotics without anaerobic activity were not (10). Our results provide support for the hypothesized mechanism through the gut microbiota. The influence of antibiotics on the diversity and composition of the gut microbiota can persist long after treatment is completed (2,31), especially among infants (32,33). Altered microbiota can modify metabolism and intestinal inflammation and immunity resulting in increased energy harvest from the diet (4,6,34). Murine models suggest a causal association between the microbiota and obesity; when antibiotic-altered microbiota from overweight mice were transferred to germ-free mice, these mice experienced the same growth and immune response phenotypes as the donor mice (6).

ting in increased energy harvest from the diet (4,6,34). Murine models suggest a causal association between the microbiota and obesity; when antibiotic-altered microbiota from overweight mice were transferred to germ-free mice, these mice experienced the same growth and immune response phenotypes as the donor mice (6). Because the effect of early life antibiotics did not, however, affect growth rates after 18 months of age, the impact of an altered microbiota may be short lived. This evidence that the microbiota is not necessarily permanently “reprogrammed” to cause increased growth rates long after exposure is supported by a recent large study in the United States that found that early antibiotic use was not associated with higher rates of weight gain after the exposure period (15). Their results may be explained by their analytic exclusion of antibiotic-induced weight gain during the exposure period. Taken together, the evidence suggests that short-term boosts in growth due to antibiotics may be more relevant than long-term changes in growth rates.

s of weight gain after the exposure period (15). Their results may be explained by their analytic exclusion of antibiotic-induced weight gain during the exposure period. Taken together, the evidence suggests that short-term boosts in growth due to antibiotics may be more relevant than long-term changes in growth rates. Still, because the unexposed children in the MAL-ED cohort did not complete catch-up growth, such that the weight difference associated with antibiotic exposure was still present at 24 months, these exposures may have sustained impact on a child's attained size. Short-term growth and attained size have been negatively associated with hospitalization and mortality (35), and positively with cognitive outcomes such as IQ, years of schooling, and income (36), respectively. The first 2 years of life are an especially critical period (37), and even modest improvements in weight gain during this time may be beneficial. Increased relative weight gain in the first 2 years of life, beyond what was expected from linear growth, was, however, not associated with improvements in human capital in Brazil (36), which suggests linear growth may ultimately be more important for long-term cognitive outcomes.

in during this time may be beneficial. Increased relative weight gain in the first 2 years of life, beyond what was expected from linear growth, was, however, not associated with improvements in human capital in Brazil (36), which suggests linear growth may ultimately be more important for long-term cognitive outcomes. The observed effects limited to exposure in the first 6 months suggest that antibiotic-related modifications of the microbiota, which are most detrimental early in microbiota development, may be more important than clearance of bacterial enteropathogens, which increase in prevalence across the first 2 years of life. If treatment of infections were the main mechanism, we would expect antibiotic exposure after 6 months of age to also have an impact. Furthermore, we did not find that children with more bacterial pathogens before 6 months of age had larger improvements in WAZ associated with antibiotics than those without bacterial pathogens. Pathogen prevalence in the MAL-ED study was, however, high, especially for Campylobacter and enteroaggregative E coli in the first 6 months of life (38,39), and clearance of enteropathogens may also be contributing to improved growth in these children. Direct analysis of the microbiome in antibiotic exposed and unexposed children would clarify growth-promoting mechanisms.

high, especially for Campylobacter and enteroaggregative E coli in the first 6 months of life (38,39), and clearance of enteropathogens may also be contributing to improved growth in these children. Direct analysis of the microbiome in antibiotic exposed and unexposed children would clarify growth-promoting mechanisms. This analysis improves on previous studies of the relationship between antibiotics and child growth in low-resource settings. The cross-sectional study conducted in 5 nonaffluent countries (Nigeria, India, Indonesia, Thailand, and Syria) relied on caregiver recall of antibiotic exposures at a minimum of 4 years after exposure and caregiver report of anthropometry (21). A previous analysis of an observational birth cohort in India was unable to assess antibiotic class and had partially missing information on antibiotics for nondiarrheal illnesses (40). A systematic review of randomized controlled trials in low- and middle-income countries only assessed antibiotic treatment for trial-specific indications and did not consider cumulative antibiotic exposure in early life (41). In contrast, MAL-ED included prospective follow-up for complete antibiotic information and monthly anthropometry measurement by trained fieldworkers.

low- and middle-income countries only assessed antibiotic treatment for trial-specific indications and did not consider cumulative antibiotic exposure in early life (41). In contrast, MAL-ED included prospective follow-up for complete antibiotic information and monthly anthropometry measurement by trained fieldworkers. This analysis was limited by the inability to assess long-term growth outcomes 5 to 10 years after exposure. This may explain why there was no observable effect of antibiotic exposure after 6 months of age, which may only become apparent later in childhood (9,12,14). This may also explain why the effects on length were smaller since length represents a longer-term growth process. In addition, the precision of estimates was highly dependent on frequency of use, which varied for different antibiotic classes across sites. Because MAL-ED was an observational study, we cannot eliminate the potential for unmeasured confounding, for example, by mode of delivery, which was not recorded. Our analyses, however, accounted for factors associated with antibiotic use previously identified in MAL-ED (22) and each child's detailed illness history. Because we would expect children with more illnesses to be exposed to more antibiotics and have poorer growth, residual confounding by illnesses would likely bias our estimates toward the null, such that our estimates would be conservative.

previously identified in MAL-ED (22) and each child's detailed illness history. Because we would expect children with more illnesses to be exposed to more antibiotics and have poorer growth, residual confounding by illnesses would likely bias our estimates toward the null, such that our estimates would be conservative. An association between antibiotics and weight gain in high-income settings is often discussed as a negative side effect of treatment since obesity is a growing public health problem and particularly pernicious among children. The cost-benefit analysis may, however, be different in low-resource settings where children may benefit from improvements in growth and are commonly infected with enteropathogens even in the absence of diarrhea (38). On the contrary, overuse of antibiotics is a major concern worldwide and can lead to adverse events, drug toxicity, and antimicrobial resistance. Antibiotic exposure can also cause antibiotic-associated diarrhea (42), alter intestinal immune function, increase intestinal permeability, and increase risk of systemic infections and subsequent diarrhea (32,43–46). Furthermore, increased weight gain may not be an unmitigated positive if antibiotic-induced changes to the microbiota lead to increased risk for obesity or metabolic syndrome later in life (6). It is unknown whether antibiotic-induced weight gain in these settings is equivalent in terms of developmental impact to similar gains achieved by appropriate nutrition and illness management. Therefore, the total impact of antibiotic exposure early in life among children in low-resource settings is unknown and may be mixed. Reduction of inappropriate antibiotic use must be a public health priority, although opportunities for rational and targeted antibiotic therapy may provide additional benefit by promoting weight gain in these children.

nor with the incidence of stunting, wasting, or underweight. We also found no relationship of vitamin D status at 6 week or 6 months with incidence of diarrhea, URI, or clinical malaria, but infants with 25(OH)D concentrations ≥30 ng/mL at 6 months had increased risk of incident ALRI as compared with 20 to 29.9 ng/mL. The high prevalence of vitamin D deficiency (76%) at 6 weeks of age among HIV-unexposed Tanzanian infants is consistent with our previous study in the same setting which found that 57% of HIV-infected and 60% of HIV-exposed uninfected Tanzanian infants were deficient at 6 weeks of age (9). Vitamin D deficiency is common during early infancy due to a combination of poor transplacental transfer of maternal vitamin D stores, low vitamin D content in human breastmilk, and limited sun exposure of young infants (22,23). Infants who were exclusively breastfed in our study had an increased prevalence of vitamin D deficiency at 6 weeks of age as compared to formula-fed infants and nonexclusively breastfed infants who did not receive formula. Infants who received formula at 6 weeks, which was likely vitamin D fortified, had the lowest risk of vitamin D deficiency. As for nonexclusively and non–formula-fed infants, the staple complementary food in Tanzania is maize-based porridge, which in itself is a poor source of vitamin D; however, cow's milk, condensed milk, meat, and other vitamin D containing foods are sometimes added or consumed separately (24). We did not have data available on sun exposure of infants.

ntibiotic exposure early in life among children in low-resource settings is unknown and may be mixed. Reduction of inappropriate antibiotic use must be a public health priority, although opportunities for rational and targeted antibiotic therapy may provide additional benefit by promoting weight gain in these children. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Acknowledgments The authors thank the staff and participants of the MAL-ED Network Project for their important contributions. The MAL-ED Network Investigators include representatives from the following organizations: A.B. PRISMA, Iquitos, Peru; Aga Khan University, Karachi, Pakistan; Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand; Christian Medical College, Vellore, India; Duke University, Durham, NC; Fogarty International Center/National Institutes of Health, Bethesda, MD; Foundation for the NIH, Bethesda, MD; Haydom Lutheran Hospital, Haydom, Tanzania; icddr,b, Dhaka, Bangladesh; Institute of Medicine, Tribhuvan University, Kathmandu, Nepal; Johns Hopkins University, Baltimore, MD; The Pennsylvania State University, University Park, PA; Temple University, Philadelphia, PA; Universidade Federal do Ceara, Fortaleza, Brazil; University of Bergen, Norway; University of Illinois at Chicago, IL; University of Venda, Thohoyandou, South Africa; University of Virginia, Charlottesville, VA; Walter Reed/AFRIMS Research Unit, Kathmandu, Nepal; Haukeland University Hospital, Bergen, Norway. A complete list of investigators is included in the Supplemental Digital Content 1, Text.

University of Illinois at Chicago, IL; University of Venda, Thohoyandou, South Africa; University of Virginia, Charlottesville, VA; Walter Reed/AFRIMS Research Unit, Kathmandu, Nepal; Haukeland University Hospital, Bergen, Norway. A complete list of investigators is included in the Supplemental Digital Content 1, Text. The Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development Project (MAL-ED) is carried out as a collaborative project supported by the Bill & Melinda Gates Foundation, the Foundation for the NIH, and the National Institutes of Health/Fogarty International Center. This work was supported by the Fogarty International Center, National Institutes of Health (D43-TW009359 to E.T.R.). The authors report no conflicts of interest. TABLE 1 Baseline characteristics and antibiotic use by site among 1954 children in the MAL-ED cohort who were followed until at least 6 months of age with subsequent anthropometry Dhaka, BangladeshNo. (%) Fortaleza, BrazilNo. (%) Vellore, IndiaNo. (%) Bhaktapur, NepalNo. (%) Loreto, PeruNo. (%) Naushahro Feroze, PakistanNo. (%) Venda, South AfricaNo. (%) Haydom, TanzaniaNo. (%) OverallNo.

TABLE 1 Baseline characteristics and antibiotic use by site among 1954 children in the MAL-ED cohort who were followed until at least 6 months of age with subsequent anthropometry Dhaka, BangladeshNo. (%) Fortaleza, BrazilNo. (%) Vellore, IndiaNo. (%) Bhaktapur, NepalNo. (%) Loreto, PeruNo. (%) Naushahro Feroze, PakistanNo. (%) Venda, South AfricaNo. (%) Haydom, TanzaniaNo. (%) OverallNo. (%) Children followed until 6 mo with subsequent anthropometry 240 207 234 235 269 265 259 245 1954 Children with anthropometry at 2 years 212 (87.6) 168 (80.4) 227 (96.2) 228 (96.6) 200 (74.1) 249 (94.0) 238 (91.2) 214 (86.3) 1736 (88.3) Female sex 122 (50.8) 101 (48.8) 128 (54.7) 109 (46.4) 123 (45.7) 136 (51.3) 129 (49.8) 124 (50.6) 972 (49.7) Crowding in the home (2+ people per room) 225 (93.8) 29 (14.0) 185 (79.0) 93 (39.6) 100 (37.2) 229 (86.4) 35 (13.5) 107 (43.7) 1003 (51.3) Maternal education < 6 years 152 (63.3) 27 (13.0) 82 (35.3) 60 (25.5) 60 (22.3) 218 (82.3) 6 (2.3) 94 (38.4) 699 (35.8) Monthly income <150 USD 157 (65.4) 7 (3.4) 216 (92.3) 114 (48.5) 188 (69.9) 145 (54.7) 51 (19.7) 242 (98.8) 1120 (57.3) Median months of exclusive breast-feeding (IQR) 5.0 (3.8, 5.7) 2.6 (1.3, 4.3) 3.5 (2.5, 4.6) 3.0 (1.5, 4.4) 2.7 (1.0, 4.3) 0.5 (0.3, 0.7) 1.0 (0.6, 1.7) 1.8 (1.1, 2.7) 2.2 (0.9, 4) Courses of antibiotics before 6 mo 0 4 (1.7) 182 (87.9) 87 (37.2) 106 (45.1) 58 (21.6) 4 (1.5) 179 (69.1) 50 (20.4) 670 (34.3) 1 18 (7.5) 25 (12.1) 64 (27.4) 90 (38.3) 71 (26.4) 14 (5.3) 55 (21.2) 76 (31.0) 413 (21.1) 2 36 (15.0) 0 (0) 36 (15.4) 26 (11.1) 71 (26.4) 25 (9.4) 21 (8.1) 54 (22.0) 269 (13.8) 3+ 182 (75.8) 0 (0) 47 (20.1) 13 (5.5) 69 (25.7) 222 (83.8) 4 (1.5) 65 (26.5) 602 (30.8) Median days of antibiotics before 6 mo (IQR) 26 (15, 38) 0 (0, 0) 4 (0, 9) 3 (0, 7) 7 (3, 13) 31 (16, 48) 0 (0, 2) 8 (3, 16) 6 (0, 18) At least 1 day of class-specific antibiotic exposure before 6 mo Penicillins 202 (84.2) 18 (8.7) 93 (39.7) 61 (26.0) 173 (64.3) 151 (57.0) 59 (22.8) 150 (61.2) 907 (46.4) Cephalosporins 158 (65.8) 5 (2.4) 62 (26.5) 27 (11.5) 33 (12.3) 205 (77.4) 1 (0.4) 2 (0.8) 493 (25.2) Macrolides 143 (59.6) 1 (0.5) 21 (9) 28 (11.9) 86 (32) 79 (29.8) 7 (2.7) 6 (2.5) 371 (19.0) Metronidazole 8 (3.3) 0 (0) 11 (4.7) 37 (15.7) 2 (0.7) 156 (58.9) 3 (1.2) 59 (24.1) 276 (14.1) Sulfonamides 3 (1.3) 1 (0.5) 22 (9.4) 19 (8.1) 45 (16.7) 73 (27.6) 10 (3.9) 55 (22.5) 228 (11.7) Fluoroquinolones 34 (14.2) 0 (0) 16 (6.8) 4 (1.7) 1 (0.4) 9 (3.4) 0 (0) 2 (0.8) 66 (3.4) Campylobacter detection before 6 mo 128 (53.3) 27 (13.0) 99 (42.3) 90 (38.3) 103 (38.3) 175 (66.0) 105 (40.5) 147 (6

(24.1) 276 (14.1) Sulfonamides 3 (1.3) 1 (0.5) 22 (9.4) 19 (8.1) 45 (16.7) 73 (27.6) 10 (3.9) 55 (22.5) 228 (11.7) Fluoroquinolones 34 (14.2) 0 (0) 16 (6.8) 4 (1.7) 1 (0.4) 9 (3.4) 0 (0) 2 (0.8) 66 (3.4) Campylobacter detection before 6 mo 128 (53.3) 27 (13.0) 99 (42.3) 90 (38.3) 103 (38.3) 175 (66.0) 105 (40.5) 147 (6 0.0) 874 (44.7) EAEC detection before 6 mo 166 (69.2) 150 (72.5) 178 (76.1) 147 (62.6) 107 (39.8) 219 (82.6) 161 (62.2) 223 (91.0) 1351 (69.1) Giardia detection before 6 mo 3 (1.3) 6 (2.9) 15 (6.4) 7 (3.0) 19 (7.1) 99 (37.4) 3 (1.2) 20 (8.2) 172 (8.8) Mean enrollment WAZ* −1.28 −0.16 −1.30 −0.91 −0.62 −1.42 −0.38 −0.13 −0.78 Mean enrollment LAZ* −1.03 −0.78 −1.03 −0.70 −0.96 −0.71 −1.01 −0.89 Mean WAZ at 2 years −1.61 0.37 −1.65 −0.93 −0.82 −1.65 −0.51 −1.33 −1.06 Mean LAZ at 2 years −2.03 −0.07 −1.92 −1.35 −1.89 −1.71 −2.66 −1.71 EAEC = enteroaggregative E coli; IQR = interquartile range; LAZ = length-for-age z score; USD = United States Dollars; WAZ = weight-for-age z score. *Within 17 days of birth. TABLE 2 Adjusted weight-for-age and length-for-age z score differences associated with class-specific antibiotic use in the first 6 months of life among 1954 children in the MAL-ED birth cohort

0.0) 874 (44.7) EAEC detection before 6 mo 166 (69.2) 150 (72.5) 178 (76.1) 147 (62.6) 107 (39.8) 219 (82.6) 161 (62.2) 223 (91.0) 1351 (69.1) Giardia detection before 6 mo 3 (1.3) 6 (2.9) 15 (6.4) 7 (3.0) 19 (7.1) 99 (37.4) 3 (1.2) 20 (8.2) 172 (8.8) Mean enrollment WAZ* −1.28 −0.16 −1.30 −0.91 −0.62 −1.42 −0.38 −0.13 −0.78 Mean enrollment LAZ* −1.03 −0.78 −1.03 −0.70 −0.96 −0.71 −1.01 −0.89 Mean WAZ at 2 years −1.61 0.37 −1.65 −0.93 −0.82 −1.65 −0.51 −1.33 −1.06 Mean LAZ at 2 years −2.03 −0.07 −1.92 −1.35 −1.89 −1.71 −2.66 −1.71 EAEC = enteroaggregative E coli; IQR = interquartile range; LAZ = length-for-age z score; USD = United States Dollars; WAZ = weight-for-age z score. *Within 17 days of birth. TABLE 2 Adjusted weight-for-age and length-for-age z score differences associated with class-specific antibiotic use in the first 6 months of life among 1954 children in the MAL-ED birth cohort Antibiotic class exposure in first 6 months of life Number exposed (%)(N = 1954†) Adjusted* WAZ difference(95% CI) Number exposed (%)(N = 1689†) Adjusted* LAZ difference(95% CI) Metronidazole 1 Course 170 (8.7) 0.14 (−0.01, 0.29) 109 (6.5) 0.00 (−0.15, 0.16) 2+ Courses 106 (5.4) 0.24 (0.04, 0.43) 11 (0.7) 0.02 (−0.35, 0.40) Macrolides 1 Course 258 (13.2) 0.09 (−0.03, 0.21) 206 (12.2) 0.07 (−0.05, 0.18) 2+ Courses 114 (5.8) 0.23 (0.05, 0.42) 87 (5.2) −0.00 (−0.20, 0.19) Cephalosporins 1 Course 244 (12.5) 0.03 (−0.09, 0.16) 182 (10.8) 0.02 (−0.11, 0.14) 2+ Courses 249 (12.7) 0.19 (0.04, 0.35) 106 (6.3) 0.19 (0.01, 0.37) Penicillins 1 Course 450 (23.0) 0.09 (−0.02, 0.19) 382 (22.6) −0.04 (−0.13, 0.06) 2+ Courses 457 (23.4) 0.07 (−0.04, 0.18) 373 (22.1) −0.00 (−0.11, 0.10) Fluoroquinolones (any) 66 (3.4) 0.08 (−0.14, 0.30) 57 (3.4) 0.05 (−0.16, 0.27) Sulfonamides (any) 228 (11.7) 0.03 (−0.09, 0.15) 155 (9.2) −0.02 (−0.14, 0.11) LAZ = length-for-age z score; WAZ = weight-for-age z score.

382 (22.6) −0.04 (−0.13, 0.06) 2+ Courses 457 (23.4) 0.07 (−0.04, 0.18) 373 (22.1) −0.00 (−0.11, 0.10) Fluoroquinolones (any) 66 (3.4) 0.08 (−0.14, 0.30) 57 (3.4) 0.05 (−0.16, 0.27) Sulfonamides (any) 228 (11.7) 0.03 (−0.09, 0.15) 155 (9.2) −0.02 (−0.14, 0.11) LAZ = length-for-age z score; WAZ = weight-for-age z score. *Adjusted for other antibiotic classes included in the table, site, child sex, enrollment WAZ, Water, Assets, Maternal education, Income (WAMI) score, crowding (people/room in household), maternal height, maternal education, and characteristics of the child's first 6 months of life: percent days exclusively breast-fed; number of diarrhea episodes; days with fever, vomiting, and respiratory illness; and presence of acute lower respiratory infection (ALRI), bloody stools, and hospitalization. LAZ difference is also adjusted for enrollment LAZ. †Children who were followed in the MAL-ED birth cohort until at least 6 months of age with subsequent anthropometry. LAZ difference estimates exclude Pakistan.

What Is KnownInfants are at high risk for vitamin D deficiency due to limited sun exposure and low dietary intake via human breastmilk. Dark skin pigmentation is a risk factor for vitamin D deficiency. Severe vitamin D deficiency can cause nutritional rickets. A previous cohort study conducted in Tanzania determined human immunodeficiency virus (HIV)-exposed uninfected infants with severe vitamin D deficiency at 6 weeks of age had increased risk of wasting. What Is NewVitamin D deficiency is common, particularly during the early weeks of life among HIV-unexposed Tanzanian infants. Vitamin D status among Tanzanian infants is primarily determined by breastfeeding status and season. Poor vitamin D status does not appear to be an important contributor to morbidity and growth among HIV-unexposed Tanzanian infants. See Commentary to Article Entitled “Vitamin D Deficiency Is Not Associated With Growth or the Incidence of Common Morbidities Among Tanzanian Infants” by Mager on page 357.

Vitamin D status among Tanzanian infants is primarily determined by breastfeeding status and season. Poor vitamin D status does not appear to be an important contributor to morbidity and growth among HIV-unexposed Tanzanian infants. See Commentary to Article Entitled “Vitamin D Deficiency Is Not Associated With Growth or the Incidence of Common Morbidities Among Tanzanian Infants” by Mager on page 357. Globally, it is estimated that 1 billion people are vitamin D deficient and nearly 50% of the global population is affected by vitamin D insufficiency (1,2). For most individuals, 50% to 90% of the body's vitamin D is produced by sun exposure of the skin while the remaining proportion is obtained from the diet (2). As a result, the risk of vitamin D deficiency is highly influenced by sun exposure, latitude, skin-covering clothing, and skin pigmentation (2,3). Infants are known to be at particularly high risk for vitamin D deficiency due to low rates of sun exposure and low dietary intake via human breastmilk, particularly if the mother is vitamin D deficient (4,5).

eficiency is highly influenced by sun exposure, latitude, skin-covering clothing, and skin pigmentation (2,3). Infants are known to be at particularly high risk for vitamin D deficiency due to low rates of sun exposure and low dietary intake via human breastmilk, particularly if the mother is vitamin D deficient (4,5). The link between severe vitamin D deficiency during childhood with nutritional rickets (which can cause significant growth faltering) has long been recognized; however, the effect of less severe forms of deficiency on linear growth and weight gain during infancy is not well established (1). To date, research on infant growth outcomes has primarily focused on the role of maternal vitamin D status in pregnancy, whereas less research has focused on infant vitamin D status (6,7). In particular, research is needed in low- and middle-income countries (LMICs) where both linear growth faltering during infancy and vitamin D deficiency are common (8). We previously published the first cohort study of infant vitamin D status and anthropometric growth in sub-Saharan Africa and found that among Tanzanian infants born to human immunodeficiency virus (HIV)-infected mothers, 58.6% of infants were vitamin D deficient (25-hydroxyvitamin D (25(OH)D) levels <20 ng/mL) at 6 weeks of age (9). HIV-exposed uninfected infants with severe vitamin D deficiency (25(OH)D) levels <10 ng/mL) at 6 weeks of age had significantly increased risk of incident wasting, but there was no association with other anthropometric outcomes. Nevertheless, it is unclear whether these growth findings among HIV-exposed infants are generalizable to the substantially larger population of infants born to HIV-uninfected mothers in sub-Saharan Africa.

significantly increased risk of incident wasting, but there was no association with other anthropometric outcomes. Nevertheless, it is unclear whether these growth findings among HIV-exposed infants are generalizable to the substantially larger population of infants born to HIV-uninfected mothers in sub-Saharan Africa. We therefore performed a prospective cohort study of Tanzanian HIV-unexposed infants to evaluate the relationship of vitamin D status at 6 weeks and 6 months of age with growth and common childhood morbidities. Our results are intended to inform whether infant vitamin D supplementation trials for growth improvement should be pursued among the general infant population in Tanzania and similar settings in sub-Saharan Africa. METHODS Parent Trial Design The parent trial for this prospective cohort study was a randomized, double-blind, factorial-designed, placebo-controlled trial of infant zinc and multivitamin supplementation conducted in Dar es Salaam, Tanzania from August 2007 to May 2011 (clinicaltrials.gov NCT 00421668) (10). Briefly, infants born to HIV-uninfected mothers were randomly assigned to one of four study arms (factorial design of zinc, multivitamins, and placebo) between 5 and 7 weeks of age. The multivitamin contained vitamin C, vitamin E, thiamine, riboflavin, niacin, vitamin B6, folate, and vitamin B12, and zinc capsules contained 5 mg of zinc. Infants were excluded from the trial if they were of multiple gestation or had a serious congenital anomaly.

ivitamins, and placebo) between 5 and 7 weeks of age. The multivitamin contained vitamin C, vitamin E, thiamine, riboflavin, niacin, vitamin B6, folate, and vitamin B12, and zinc capsules contained 5 mg of zinc. Infants were excluded from the trial if they were of multiple gestation or had a serious congenital anomaly. Study Population and Serum 25-Hydroxyvitamin D Quantification Serum samples included in this vitamin D study were selected from infants enrolled in a substudy of biomarkers of environmental enteric dysfunction (n = 590) (11). Infants were eligible for the environmental enteric dysfunction and subsequently the vitamin D study were randomly sampled if they had a blood sample available at 6 weeks and 6 months of age and were not stunted (length-for-age z score [LAZ] ≥ −2) at the baseline 6-week visit. A total of 581 children met these criteria and had serum samples with adequate volume available for the vitamin D substudy. We used high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) at Boston Children's Hospital using an API-5000 (AB Sciex, Foster City, CA) was used to quantify serum 25(OH)D concentration from 6-week and 6-month blood samples (12). Day-to-day coefficient of variation ranged from 5.6% to 8.5%.

udy. We used high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) at Boston Children's Hospital using an API-5000 (AB Sciex, Foster City, CA) was used to quantify serum 25(OH)D concentration from 6-week and 6-month blood samples (12). Day-to-day coefficient of variation ranged from 5.6% to 8.5%. Data Collection Mothers and infants were followed at monthly clinic visits for 18 months after randomization. At the baseline visit study nurses conducted a structured interview to collect information on demographic characteristics. A household asset score was calculated as the sum of asset items that included a sofa, television, radio, refrigerator, and fan (13). At all study visits nurses asked infant feeding practices in the previous week including breastfeeding status and frequency and introduction of other foods. Exclusive breastfeeding was defined as feeding a child with breast milk only without any additional foods.

d a sofa, television, radio, refrigerator, and fan (13). At all study visits nurses asked infant feeding practices in the previous week including breastfeeding status and frequency and introduction of other foods. Exclusive breastfeeding was defined as feeding a child with breast milk only without any additional foods. Study physicians performed a clinical exam every 3 months and/or when acute complaints of illness were noted by nurses at monthly visits. Mothers were also encouraged to bring their child to the study clinic for sick visits when the child was unwell. Diarrhea was defined as ≥3 loose or watery stools within a 24-hour period. Acute upper respiratory infection was defined as pharyngitis or rhinitis (both without fast breathing or chest indrawing). Acute lower respiratory infection was defined as cough or difficulty breathing, rapid respiratory rate (based on the same definition described previously), and either a fever of >38.3°C or chest retractions. Clinical malaria was symptomatically diagnosed with/without laboratory testing for parasitemia following integrated management of childhood illnesses guidelines to minimize malaria-related deaths (14).

piratory rate (based on the same definition described previously), and either a fever of >38.3°C or chest retractions. Clinical malaria was symptomatically diagnosed with/without laboratory testing for parasitemia following integrated management of childhood illnesses guidelines to minimize malaria-related deaths (14). Nurses collected infant length and weight measurements using a digital infant balance (Tanita, Japan) and a rigid length board with a movable foot piece at all study visits. LAZ, weight-for-length z score (WLZ), and weight-for-age z score (WAZ) were calculated using World Health Organization child growth standards (15). Stunting, wasting, and underweight were defined as a LAZ, WLZ, and WAZ of 2 or more standard deviations (SDs) below the World Health Organization population median, respectively (15). Statistical Methods Based on the distribution of 25(OH)D and commonly used clinical cutoffs, we categorized 6-week vitamin D into 3 categories <10 ng/mL,10–19.9 ng/mL, and ≥20 ng/mL (16). Similarly, we categorized 6-month vitamin D into 3 categories based on common clinical cutoffs <20 ng/mL, 20–30 ng/mL, and ≥30 ng/mL (16). We first determined the cross-sectional association of demographic and infant characteristics with severe vitamin D deficiency (<10 ng/mL) at 6 weeks of age and vitamin D deficiency (<20 ng/mL) at 6 months of age using log-binomial models to calculate relative risk estimates (17). Multivariate log-binomial models were fit including all variables with a P <0.20 in univariate analysis (18).

infant characteristics with severe vitamin D deficiency (<10 ng/mL) at 6 weeks of age and vitamin D deficiency (<20 ng/mL) at 6 months of age using log-binomial models to calculate relative risk estimates (17). Multivariate log-binomial models were fit including all variables with a P <0.20 in univariate analysis (18). Anthropometric growth curves stratified by 25(OH)D category were constructed using mixed effects models using restricted cubic splines with knots at 10 weeks, 2, 6, 9, and 16 months intervals after randomization and automatic knot selection using P < 0.05 to determine a parsimonious model (19). The association of 25(OH)D categories at 6 weeks and 6 months and change in anthropometric z scores (LAZ, WLZ, WAZ) were analyzed with generalized estimating equations. Change in anthropometric z scores between consecutive visits was treated as a longitudinal continuous outcome and 25(OH)D, infant age, along with other baseline covariates as explanatory variables. The potential nonlinear relationship of change in anthropometric z scores between consecutive visits over time was examined nonparametrically with restricted cubic splines for infant age (19). If a nonlinear relationship was found, we added the selected cubic spline terms to the above-specified model as covariates. To assess whether 25(OH)D status was associated with change in anthropometric z scores between consecutive visits over time, we included the interactions of 25(OH)D categories with infant age and its splines. The robust score test was then used to determine whether overall anthropometric z score trajectory differed by 25(OH)D categories at 6 weeks and 6 months of age.

th change in anthropometric z scores between consecutive visits over time, we included the interactions of 25(OH)D categories with infant age and its splines. The robust score test was then used to determine whether overall anthropometric z score trajectory differed by 25(OH)D categories at 6 weeks and 6 months of age. We also examined the association of 25(OH)D categories with stunting, wasting, and underweight using proportional hazard models. Infants with the outcome of interest at the time of 25(OH)D assessment were excluded from stunting, wasting, and underweight analyses. We examined the relationship of 25(OH)D with the incidence rate of physician diagnoses of morbidities using Poisson regression to take in account differences in timing of routine and sick visits (20). Potential confounders were selected a priori based on their potential association with infant vitamin D status and adverse child health outcomes baseline socioeconomic status, maternal health and nutrition indicators, and child nutrition factors. Confounders included maternal age, maternal education, parity, wealth tertile, child sex, study regimen, breast-feeding status, low birthweight (<2500 g), prematurity (<37 weeks), and season of 25(OH)D assessment. Missing data for covariates was retained in the analysis using the missing indicator method (21). All P values were 2-sided and P < 0.05 was considered statistically significant. Statistical analyses were performed using the SAS v 9.2 (SAS Institute Inc, Cary, NC).

aturity (<37 weeks), and season of 25(OH)D assessment. Missing data for covariates was retained in the analysis using the missing indicator method (21). All P values were 2-sided and P < 0.05 was considered statistically significant. Statistical analyses were performed using the SAS v 9.2 (SAS Institute Inc, Cary, NC). Ethics Written informed consent was obtained from all mothers in the parent trial. Institutional approval was granted by the Harvard School of Public Health Human Subjects Committee (12875-02), the Muhimbili University of Health and Allied Sciences Committee of Research and Publications (2014-10-29/AEC/Vol.IX/28) and the Tanzanian National Institute of Medical Research (HQ/R.8a/Vol.IX/482). RESULTS A total of 581 infants were enrolled in the vitamin D substudy. Maternal and child characteristics are presented in Table 1. Characteristics of infants included in the vitamin D study were similar to eligible infants (not stunted at 6 weeks) who did not have a vitamin D sample selected for analysis (n = 1661; Supplemental Digital Content, Table 1). The mean 25(OH)D concentration at 6 weeks of age was 14.5 ng/mL (SD 7.4) and 26.0 ng/mL (SD 8.7) at 6 months of age (Table 1). Figure 1A and B presents the distributions of 25(OH)D at 6 weeks and 6 months, respectively. The percentage of infants with vitamin D deficiency (<20 ng/mL) was 76.4% at 6 weeks and 21.2% at 6 months of age. The vast majority (92.1%) of infants had increased 25(OH)D levels from 6 weeks to 6 months of age with a mean increase of 12.2 ng/mL (SD 10.3).

stributions of 25(OH)D at 6 weeks and 6 months, respectively. The percentage of infants with vitamin D deficiency (<20 ng/mL) was 76.4% at 6 weeks and 21.2% at 6 months of age. The vast majority (92.1%) of infants had increased 25(OH)D levels from 6 weeks to 6 months of age with a mean increase of 12.2 ng/mL (SD 10.3). FIGURE 1 Distribution of 25(OH)D concentration (ng/mL) at 6 weeks (panel A) and 6 months (panel B) of age among 581 infants. 25(OH)D = 25-hydroxyvitamin D. Risk Factors for Vitamin D Deficiency Univariate and multivariate risk factor analyses for vitamin D deficiency (<20 ng/mL) at 6 weeks (Supplemental Digital Content, Table 2) and 6 months of age (Supplemental Digital Content, Table 3) are presented in the Supplemental Digital Content. In multivariate models, infants who were exclusively breastfed at 6 weeks had 2.05 (95% confidence interval [CI] 0.1.11–0.3.79; P = 0.02) times the risk of 25(OH)D levels <20 ng/mL as compared to formula-fed infants. Infants who were not exclusively breastfed and did not receive formula also appeared to be at increased risk for deficiency as compared to formula-fed infants (relative risk 1.78; 95% CI 0.96–3.30; P = 0.07). At 6 months of age, the postharvest season remained a significant risk factor for vitamin D deficiency (P < 0.01). Maternal age, maternal education, household wealth, child sex, birth order, low birth weight (<2500 g), and prematurity (<37 weeks) were not significantly associated with vitamin D deficiency at 6 weeks or 6 months of age.

e, the postharvest season remained a significant risk factor for vitamin D deficiency (P < 0.01). Maternal age, maternal education, household wealth, child sex, birth order, low birth weight (<2500 g), and prematurity (<37 weeks) were not significantly associated with vitamin D deficiency at 6 weeks or 6 months of age. Vitamin D and Anthropometric Growth Figure 2 presents LAZ, WLZ, and WAZ growth curves from 6 weeks to 18 months of age stratified by vitamin D status at 6 weeks of age. In multivariate analysis, the trajectory of WLZ significantly differed for infants with 25(OH)D concentrations of >20 ng/mL compared with those with concentrations 10 to 20 ng/mL (P value for difference in trajectory <0.01). There was a smaller increase in WLZ during the period of 6 weeks to 4 months for infants with vitamin D >20 ng/mL as compared to other groups, but all groups had similar WLZ at the end of the study period at 18 months of age. There were no differences in LAZ and WAZ trajectories by vitamin D status at 6 weeks of age (P values >0.05). There was also no association of vitamin D status at 6 weeks of age with incident stunting, underweight or wasting during follow-up (Table 2).

had similar WLZ at the end of the study period at 18 months of age. There were no differences in LAZ and WAZ trajectories by vitamin D status at 6 weeks of age (P values >0.05). There was also no association of vitamin D status at 6 weeks of age with incident stunting, underweight or wasting during follow-up (Table 2). FIGURE 2 Mean length-for-age z score (LAZ) (panel A), weight-for-length z score (WLZ) (panel B), and weight-for-age z score (WAZ) (panel C) growth curves stratified by 6-week 25(OH)D concentration. There was no significant difference in trajectory of LAZ or WAZ by 6-week vitamin D status categories (P values for difference in trajectory >0.05). There was a significant difference in WLZ trajectory for infants with 25(OH)D concentrations at 6 weeks of >20 ng/mL as compared with infants with concentrations 10–20 ng/mL (P value for difference in WLZ trajectory <0.01). 25(OH)D = 25-hydroxyvitamin D. Figure 3 presents LAZ, WLZ, and WAZ growth curves for 6 months to 18 months of age stratified by vitamin D status at 6 months of age. There were no differences in LAZ, WLZ, or WAZ trajectories by vitamin D status at 6 months of age in univariate and multivariate analyses (P values >0.05). There was also no association of vitamin D status at 6 months of age with incident stunting, underweight, or wasting (Table 2).

/mL at 6 months of age had increased risk of ALRI as compared to infants with concentrations of 20 to 29.9 ng/mL. Consequently, the existing, even though very limited, observational and trial evidence do not support vitamin D supplementation among the general population of infants in LMICs to reduce common morbidities. We also found no significant relationship of low vitamin D status at 6 weeks and 6 months with anthropometric growth, which is in contrast to our previous study where we found that HIV-exposed infants with 25(OH)D concentrations <10 ng/mL at 6 weeks of age had significantly poorer WLZ trajectory and increased risk of wasting during the first 2 years of life. Similarly in the current study, we found HIV-unexposed Tanzanian infants with 25(OH)D levels <20 ng/mL at 6 months had slightly increased risk of wasting but the results were not statistically significant. Of note, the incidence of wasting among HIV-unexposed infants in this study was less than half of that seen in our previous study of HIV-exposed uninfected infants and therefore the current study may have had inadequate power to detect a moderate difference in wasting. Limited data exist about the effect of vitamin D supplementation on weight gain and body composition. In 1 trial low birthweight (1.8–2.5 kg) Indian infants randomized to receive weekly vitamin D from birth had increased length, weight, and arm circumference at 6 months of age as compared to infants who received placebo (27). In addition, an observational study of Canadian infants found higher vitamin D concentrations during infancy to 3 years of age were associated with leaner body composition (28). Studies evaluating the association of child 25(OH)D status with body composition are warranted in LMICs given rapidly increasing rates of pediatric obesity and noncommunicable diseases in adulthood (29).

y to provide a large beneficial effect on growth or morbidity in the study population. Due to the observational nature of the study, we also cannot rule out residual confounding by SES, child feeding method, and other factors. Of note, we did not have data on maternal vitamin D status at 6 weeks or 6 months postpartum. Vitamin D deficiency is highly prevalent among Tanzanian HIV-unexposed infants at 6 weeks of age but declines significantly by 6 months. We found no relationship of low blood concentrations of vitamin D at 6 weeks or 6 months of age with anthropometric growth or morbidity to 18 months of age. Due to the observational nature the study the findings cannot be considered causal, but they do not provide support vitamin D supplementation among the general population of Tanzanian infants to improve growth or to reduce risk for morbid conditions such as diarrhea, upper respiratory infections, and ALRIs. Future vitamin D supplementation studies and trials may be warranted among subsets of high-risk infants, including HIV-exposed uninfected and/or low birthweight infants, in Tanzania and similar LMIC settings (27,30). Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content This study was funded by the Bill & Melinda Gates Foundation (OPP1066203) and the National Institutes of Health (R01 HD048969 and K24 DK104676). The authors report no conflicts of interest. TABLE 1 Characteristics of 581 infants assessed for serum 25-hydroxyvitamin D concentration 6 weeks and 6 months of age

Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content This study was funded by the Bill & Melinda Gates Foundation (OPP1066203) and the National Institutes of Health (R01 HD048969 and K24 DK104676). The authors report no conflicts of interest. TABLE 1 Characteristics of 581 infants assessed for serum 25-hydroxyvitamin D concentration 6 weeks and 6 months of age Mean ± SD or frequency (%) Maternal characteristics Maternal age, y 25.9 ± 4.9 Education None 10 (1.7) Primary 428 (73.7) Secondary or greater 140 (24.1) Married/living with partner 517 (89.0) Prior pregnancies None 195 (33.6) 1–3 372 (64.0) ≥4 11 (1.9) Household asset score 0–1 175 (30.1) 2–3 336 (57.8) ≥4 66 (11.4) Child characteristics Male 279 (48.0) Low birthweight (<2500 g) 16 (2.8) Prematurity (<37 weeks gestation) 59 (11.0) Exclusively breastfed at 6 weeks 390 (67.1) Formula fed at 6 weeks 15 (2.6) Exclusively breastfed at 6 months 10 (1.7) No breastfeeding at 6 months 11 (1.9) Formula fed at 6 months 14 (2.4) Mean duration exclusive breastfeeding, mo 2.0 ± 1.7 Mean duration breast-feeding, mo 15.0 ± 4.6 LAZ at 6 weeks −0.16 ± 0.99 WLZ at 6 weeks −0.13 ± 1.21 WAZ at 6 weeks −0.26 ± 0.88 Randomized regimen Placebo 149 (25.7) Zinc 143 (24.6) Multivitamins 146 (25.1) Multivitamins + zinc 143 (24.6) Season at 6 week 25(OH)D assessment Long rain (December–March) 187 (32.2) Harvest (April–May) 94 (16.2) Post-harvest (June–August) 201 (34.6) Short rain (September–November) 99 (17.0) Mean 25(OH)D concentration at 6 weeks, ng/mL 14.5 ± 7.4 Mean 25(OH)D concentration at 6 months, ng/mL 26.0 ± 8.7 LAZ = length-for-age z score; WAZ = weight-for-age z score; WLZ = weight-for-length z score; 25(OH)D = 25-hydroxyvitamin D.

.2) Post-harvest (June–August) 201 (34.6) Short rain (September–November) 99 (17.0) Mean 25(OH)D concentration at 6 weeks, ng/mL 14.5 ± 7.4 Mean 25(OH)D concentration at 6 months, ng/mL 26.0 ± 8.7 LAZ = length-for-age z score; WAZ = weight-for-age z score; WLZ = weight-for-length z score; 25(OH)D = 25-hydroxyvitamin D. TABLE 2 Prospective association of infant serum 25-hydroxyvitamin D at 6 weeks and 6 months of age with incident∗ stunting, wasting, and underweight

.2) Post-harvest (June–August) 201 (34.6) Short rain (September–November) 99 (17.0) Mean 25(OH)D concentration at 6 weeks, ng/mL 14.5 ± 7.4 Mean 25(OH)D concentration at 6 months, ng/mL 26.0 ± 8.7 LAZ = length-for-age z score; WAZ = weight-for-age z score; WLZ = weight-for-length z score; 25(OH)D = 25-hydroxyvitamin D. TABLE 2 Prospective association of infant serum 25-hydroxyvitamin D at 6 weeks and 6 months of age with incident∗ stunting, wasting, and underweight No. with event/no. in category (%) Unadjusted HR (95% CI) P Adjusted† HR (95% CI) P 25(OH)D at 6 weeks of age Stunting (LAZ < −2) <10 ng/mL 45/187 (24.1) 0.82 (0.53–1.26) 0.36 0.64 (0.41–1.01) 0.06 10–20 ng/mL 72/257 (28.0) 0.98 (0.67–1.45) 0.93 0.83 (0.55–1.23) 0.35 >20 ng/mL 39/137 (28.4) Ref. Ref. Wasting* (WLZ < −2) <10 ng/mL 25/172 (14.5) 0.90 (0.51–1.62) 0.73 1.00 (0.54–1.89) 0.98 10–20 ng/mL 35/238 (14.7) 0.94 (0.55–1.62) 0.83 0.99 (0.57–1.74) 0.98 >20 ng/mL 21/130 (16.2) Ref. Ref. Underweight* (WAZ < −2) <10 ng/mL 28/180 (15.6) 0.79 (0.46–1.36) 0.40 0.72 (0.40–1.28) 0.26 10–20 ng/mL 51/252 (20.2) 1.08 (0.67–1.75) 0.74 1.04 (0.63–1.70) 0.88 >20 ng/mL 25/134 (18.7) Ref. Ref. 25(OH)D at 6 months of age Stunting (LAZ < −2) <20 ng/mL 23/111 (20.7) 0.83 (0.49–1.39) 0.48 0.78 (0.44–1.36) 0.38 20–30 ng/mL 42/239 (17.6) 0.75 (0.49–1.17) 0.21 0.72 (0.46–1.14) 0.17 >30 ng/mL 37/160 (23.1) Ref. Ref. Wasting* (WLZ < −2) <20 ng/mL 15/108 (13.9) 1.23 (0.62–2.41) 0.63 1.27 (0.61–2.63) 0.53 20–30 ng/mL 27/242 (11.2) 1.05 (0.58–1.89) 0.87 1.03 (0.57–1.88) 0.92 >30 ng/mL 19/174 (10.9) Ref. Ref. Underweight* (WAZ < −2) <20 ng/mL 16/111 (14.4) 0.84 (0.45–1.55) 0.57 0.90 (0.47–1.74) 0.76 20–30 ng/mL 35/252 (13.9) 0.82 (0.50–1.35) 0.43 0.80 (0.48–1.33) 0.39 >30 ng/mL 28/168 (16.7) Ref. Ref. HR = hazard ratio; LAZ = length-for-age z score; WLZ = weight-for-length z score; 25(OH)D = 25-hydroxyvitamin D.

) Ref. Ref. Underweight* (WAZ < −2) <20 ng/mL 16/111 (14.4) 0.84 (0.45–1.55) 0.57 0.90 (0.47–1.74) 0.76 20–30 ng/mL 35/252 (13.9) 0.82 (0.50–1.35) 0.43 0.80 (0.48–1.33) 0.39 >30 ng/mL 28/168 (16.7) Ref. Ref. HR = hazard ratio; LAZ = length-for-age z score; WLZ = weight-for-length z score; 25(OH)D = 25-hydroxyvitamin D. *Analysis excludes infants with outcome of interest at the time of 25(OH)D assessment. †Multivariate models adjusted for maternal age, maternal education, parity, wealth tertile, child sex, randomized regimen, low birthweight (<2500 g), prematurity (<37 weeks), breastfeeding method at 6 weeks of age, and season of 25(OH)D assessment. TABLE 3 Association of 25-hydroxyvitamin D at 6 weeks and 6 months of age with incidence of diarrhea, upper respiratory tract infection, acute lower respiratory tract infection, and clinical malaria during follow-up

†Multivariate models adjusted for maternal age, maternal education, parity, wealth tertile, child sex, randomized regimen, low birthweight (<2500 g), prematurity (<37 weeks), breastfeeding method at 6 weeks of age, and season of 25(OH)D assessment. TABLE 3 Association of 25-hydroxyvitamin D at 6 weeks and 6 months of age with incidence of diarrhea, upper respiratory tract infection, acute lower respiratory tract infection, and clinical malaria during follow-up Vitamin D at 6 weeks of age <10 ng/mL P 10–19.9 ng/mL (Reference) P ≥20 ng/mL Diarrhea Mean diagnoses/year 1.12 1.01 1.30 Crude IRR (95% CI) 0.85 (0.67–1.08) 0.18 0.78 (0.63–0.97) 0.02 Ref. Adjusted* IRR (95% CI) 0.82 (0.63–1.07) 0.14 0.82 (0.65–1.03) 0.09 Ref. URI Mean diagnoses/year 3.98 3.47 4.02 Crude IRR (95% CI) 0.97 (0.83–1.13) 0.68 0.86 (0.75–1.00) 0.05 Ref. Adjusted* IRR (95% CI) 0.94 (0.79–1.11) 0.46 0.87 (0.75–1.01) 0.08 Ref. ALRI Mean diagnoses/year 0.74 0.73 0.85 Crude IRR (95% CI) 0.85 (0.64–1.14) 0.29 0.86 (0.65–1.14) 0.29 Ref. Adjusted* IRR (95% CI) 0.84 (0.61–1.17) 0.30 0.81 (0.60–1.08) 0.15 Ref. Clinical malaria Mean diagnoses/year 0.98 0.87 0.97 Crude IRR (95% CI) 1.00 (0.76–1.31) 0.97 0.91 (0.70–1.18) 0.48 Ref. Adjusted* IRR (95% CI) 1.09 (0.80–1.49) 0.58 0.96 (0.72–1.26) 0.75 Ref. Vitamin D at 6 months of age <20 ng/mL P 20–29.9 ng/mL (Reference) P ≥30 ng/mL Diarrhea Mean diagnoses/year 0.52 0.42 0.44 Crude IRR (95% CI) 1.21 (0.82–1.80) 0.34 1.01 (0.71–1.44) 0.96 Ref. Adjusted* IRR (95% CI) 1.20 (0.83–1.75) 0.33 1.03 (0.74–1.45) 0.85 Ref. URI Mean diagnoses/year 2.92 2.76 2.97 Crude IRR (95% CI) 0.99 (0.82–1.19) 0.91 0.97 (0.84–1.12) 0.70 Ref. Adjusted* IRR (95% CI) 0.99 (0.83–1.19) 0.93 0.98 (0.85–1.14) 0.81 Ref. ALRI Mean diagnoses/year 0.56 0.47 0.69 Crude IRR (95% CI) 0.82 (0.68–1.16) 0.27 0.72 (0.53–0.97) 0.03 Ref. Adjusted* IRR (95% CI) 0.81 (0.57–1.15) 0.23 0.67 (0.50–0.91) 0.01 Ref. Clinical malaria Mean diagnoses/year 0.97 0.86 0.75 Crude IRR (95% CI) 1.31 (0.96–1.78) 0.08 1.20 (0.92–1.56) 0.17 Ref. Adjusted* IRR (95% CI) 1.33 (0.97–1.82) 0.08 1.22 (0.93–1.59) 0.15 Ref. ALRI = acute lower respiratory tract infection; IRR = incidence rate ratio; URI = upper respiratory tract infection; 25(OH)D = 25-hydroxyvitamin D.

an diagnoses/year 0.97 0.86 0.75 Crude IRR (95% CI) 1.31 (0.96–1.78) 0.08 1.20 (0.92–1.56) 0.17 Ref. Adjusted* IRR (95% CI) 1.33 (0.97–1.82) 0.08 1.22 (0.93–1.59) 0.15 Ref. ALRI = acute lower respiratory tract infection; IRR = incidence rate ratio; URI = upper respiratory tract infection; 25(OH)D = 25-hydroxyvitamin D. *Multivariate models adjusted for maternal age, maternal education, parity, wealth tertile, child sex, randomized regimen, breastfeeding method, low birthweight (<2500 g), prematurity (<37 weeks), and season of 25(OH)D assessment.

What Is KnownApproximately 10% to 12% of patients with cystic fibrosis are not meeting their nutritional goals with dietary intake alone requiring them to turn to overnight enteral tube feeding. There are no long-term prospective trials in patients with cystic fibrosis to inform treatment regarding enteral nutrition. No recommendations can be made regarding the use of pancreatic enzyme replacement therapy with enteral nutrition due to the absence of well-controlled randomized clinical trials. What Is NewLong-term use of an in-line digestive enzyme cartridge (RELiZORB) with a regular overnight enteral nutrition regimen normalized plasma omega-3 levels and resulted in a 2-fold increase in omega-3 index, a long-term measure of fat absorption in patients with CF. RELiZORB use for 90-day period was associated with a lower ratio of omega-6/omega-3 fatty acids, a key marker of inflammation. Exocrine pancreatic insufficiency (EPI) occurs when pancreatic enzyme activity in the intestinal lumen is low enough that normal digestion cannot occur (1,2). EPI is an important clinical sequela of various developmental and acquired pancreatic conditions, including preterm birth, cystic fibrosis (CF), pancreatitis, pancreatic cancer, pancreatic surgery, and aging (2,3). Pancreatic lipase insufficiency results in maldigestion of dietary fat, causing most of the clinically important EPI symptoms, and complications, including malnutrition (1,3). Up to 90% of patients with CF exhibit some degree of EPI requiring the use of replacement digestive enzyme therapy to assist with the digestion of oral meals and snacks (4).

cy results in maldigestion of dietary fat, causing most of the clinically important EPI symptoms, and complications, including malnutrition (1,3). Up to 90% of patients with CF exhibit some degree of EPI requiring the use of replacement digestive enzyme therapy to assist with the digestion of oral meals and snacks (4). Beyond clinical gastrointestinal (GI) symptoms, fat malabsorption in EPI also causes deficiencies of long-chain polyunsaturated fatty acids (LCPUFAs), including docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) (5–8). LCPUFAs have well-established clinical benefits, including visual, cognitive, anti-inflammatory, and cardiovascular effects, and there is increasing evidence that LCPUFAs are important to the health of patients with CF. DHA and EPA counteract the inflammatory responses of CF and are essential to brain development and cognition (7,9–12). Low levels of DHA may contribute to increased inflammation in patients with CF (6,7,9,10). Evidence from small studies of patients with CF given DHA and EPA supplements indicate improved pulmonary function and decreased pulmonary exacerbations with increased DHA and EPA erythrocyte levels (10,13). Given the increasing recognition of their importance in normal tissue development and health as well as disease severity, LCPUFAs are being targeted as therapeutic interventions to improve patient outcomes (5–7,10).

ry function and decreased pulmonary exacerbations with increased DHA and EPA erythrocyte levels (10,13). Given the increasing recognition of their importance in normal tissue development and health as well as disease severity, LCPUFAs are being targeted as therapeutic interventions to improve patient outcomes (5–7,10). In patients with EPI, meals and snacks are typically supplemented with oral pancreatic enzyme replacement therapy (PERT) to improve fat absorption (4,14). Unfortunately, oral PERT is not formulated nor indicated for use with continuous enteral tube feedings (15). Because PERT is taken orally during waking hours and is not delivered continuously to the small intestine during tube feeding, it is not ideal for use with continuous enteral nutrition (EN), especially when given overnight. Furthermore, given an absence of evidence from clinical trials, guidelines for EN use in patients with CF make no specific recommendations for the use of PERT with enteral feedings (16). Although PERT remains critical for oral feedings, finding a way to improve fat absorption during enteral tube feedings is critical to improving the nutritional status of patients with CF and EPI.

, guidelines for EN use in patients with CF make no specific recommendations for the use of PERT with enteral feedings (16). Although PERT remains critical for oral feedings, finding a way to improve fat absorption during enteral tube feedings is critical to improving the nutritional status of patients with CF and EPI. RELiZORB (immobilized lipase) cartridge (Alcresta Therapeutics, Inc, Newton, MA) is a new therapeutic approach to improve fat absorption in patients who receive EN tube feedings (5,15). It is FDA-approved for pediatric (≥5 years of age) and adult patients to hydrolyze fats found in enteral formula. RELiZORB is a single-use digestive enzyme cartridge and connects in-line with EN feeding sets (17). Lipase is covalently bound to small polymer beads to form a complex known as iLipase that is retained within the cartridge. The fat in enteral formula is hydrolyzed as it comes in contact with iLipase in the cartridge. Because fats containing LCPUFAs must be hydrolyzed by pancreatic lipase, RELiZORB is particularly important for the intestinal absorption of dietary LCPUFAs. RELiZORB hydrolyzes triglycerides (TGs) in a wide range of commercially available enteral formulas that are used by individuals with EPI and fat malabsorptive disorder (18).

s containing LCPUFAs must be hydrolyzed by pancreatic lipase, RELiZORB is particularly important for the intestinal absorption of dietary LCPUFAs. RELiZORB hydrolyzes triglycerides (TGs) in a wide range of commercially available enteral formulas that are used by individuals with EPI and fat malabsorptive disorder (18). In a recently published randomized study involving 33 patients with CF and EPI receiving EN, the short-term use of RELiZORB was safe, well-tolerated, and significantly increased plasma omega-3 FA levels, systemic markers of fat absorption (5). Because this study involved only a single treatment with RELiZORB, a study to evaluate its longer-term use was needed. The Absorption and Safety with Sustained use of RELiZORB Evaluation (ASSURE) Study in Patients with Cystic Fibrosis Receiving Enteral Feeding was designed to evaluate safety, tolerability, and efficacy of sustained use of RELiZORB over a 90-day period in patients with CF and EPI using EN as part of their regular nutrition regimen. The omega-3 index used in this study provides a measure of tissue DHA and EPA levels and is a reliable estimation of LCPUFA absorption in patients with CF (15,19). This index was first used as a predictor of cardiovascular outcomes and has been applied to studies of patients with CF (19). As a measure of nutritional health, the omega-3 index has the advantage of being a measure of long-term tissue LCPUFA concentrations, being unchanged by patient fasting or fed state, and being easily measured through blood collection (15,20).

iovascular outcomes and has been applied to studies of patients with CF (19). As a measure of nutritional health, the omega-3 index has the advantage of being a measure of long-term tissue LCPUFA concentrations, being unchanged by patient fasting or fed state, and being easily measured through blood collection (15,20). METHODS The ASSURE trial was a prospective, single-arm, multicenter, open-label study of RELiZORB cartridge use with EN tube feedings in patients with CF over a 90-day study period (ClinicalTrials.gov Identifier: NCT02750501). The study was conducted between July 20, 2016, and March 30, 2017. Patients were eligible for the study if they were ≥4 years of age, had a confirmed diagnosis of CF, a documented history of EPI, were taking enteral formula a minimum of 4 times per week, were using PERT, and were consuming an unrestricted fat diet. Patients were excluded if they had uncontrolled diabetes mellitus, signs and symptoms of hepatic cirrhosis, portal hypertension, or significant liver disease (defined as liver transaminases greater than 3x the upper limit of normal or total bilirubin greater than 1.5x the upper limit of normal), received a lung or liver transplant, active cancer for which they were currently receiving treatment, Crohn disease, celiac disease, diarrheal illness unrelated to EPI (eg, infectious gastroenteritis, sprue, lactose intolerance, or inflammatory bowel disease), history of fibrosing colonopathy or recurrent distal intestinal obstructive syndrome. The study protocol and informed consent forms received institutional review board approval.

eliac disease, diarrheal illness unrelated to EPI (eg, infectious gastroenteritis, sprue, lactose intolerance, or inflammatory bowel disease), history of fibrosing colonopathy or recurrent distal intestinal obstructive syndrome. The study protocol and informed consent forms received institutional review board approval. The 90-day study treatment period was preceded by a 7-day observation period and a 7-day run-in period (Supplemental Digital Content 1). Throughout the observation period (day −14 to day −8), patients received their usual EN tube-feeding regimens, including PERT. The patients then entered a study run-in period (day −7 to day −1), during which they each received their usual volume of tube-fed enteral formula (minimum 500 mL and maximum 1000 mL per feeding) for at least 5 days before the treatment period. All patients used the same enteral formula, Peptamen 1.5; (Nestlé Nutrition Inc, Florham Park, NJ), in place of their usual formula. Peptamen 1.5, which contains 14 g of fat per 250 mL (70% medium-chain TGs [MCTs] and 30% long-chain TGs; LCTs but no DHA or EPA added) (21), was chosen because it is a commonly used semi-elemental formula in patients with CF and is similar to formula with a similar fat content used during the 90-day study period.

mula. Peptamen 1.5, which contains 14 g of fat per 250 mL (70% medium-chain TGs [MCTs] and 30% long-chain TGs; LCTs but no DHA or EPA added) (21), was chosen because it is a commonly used semi-elemental formula in patients with CF and is similar to formula with a similar fat content used during the 90-day study period. Throughout the study treatment period with RELiZORB, at least 5 days a week, participants received 500 to 1000 mL of Impact Peptide 1.5, (Nestlé Nutrition). Impact Peptide 1.5 contains 15.9 g of fat per 250 mL (50% MCTs and 50% LCTs), as well as 15.9 g/L of DHA and EPA (22). It is an enteral formula like Peptamen 1.5 but with a higher ratio of long-chain to medium-chain TG (21,22). A single RELiZORB cartridge was used for each overnight feed. During all phases of the study, participants continued their pre-study regimens of PERT use with oral meals and snacks. During the observation and run-in periods, participants used PERT with EN feedings, but during the 90-day treatment period, participants were not allowed to use PERT with overnight EN feedings.

feed. During all phases of the study, participants continued their pre-study regimens of PERT use with oral meals and snacks. During the observation and run-in periods, participants used PERT with EN feedings, but during the 90-day treatment period, participants were not allowed to use PERT with overnight EN feedings. All efficacy outcomes were measured at observation initiation (day –14), before RELiZORB initiation (day 0), and at 30-day intervals during the RELiZORB treatment period study (days 30, 60, and 90). The primary efficacy outcome measure was the change in omega-3 index, which is a measure of the percentage of total DHA plus EPA (DHA+EPA) relative to the total fatty acid (FA) composition present in erythrocyte membranes. Secondary efficacy outcomes included changes in plasma and erythrocyte membrane composition (%) of total EPA, total DHA, and omega-6 to omega-3 FAs (a key marker of inflammation) as well as plasma concentrations of total DHA+EPA. Plasma concentrations (μg/mL) of total DHA and total EPA were measured using a validated ultra-high-performance liquid chromatography (UHPLC) method by PPD Laboratories (Richmond, VA) adapted from protocols described Bowen et al (23). Total EPA+DHA was calculated as the sum of the concentrations of total DHA and total EPA. FA composition (%) in both erythrocytes and plasma was measured using gas chromatography-mass spectrometry (GCMS) methods by OmegaQuant (Sioux Falls, SD) as described by Potala et al (24). Exploratory efficacy outcomes included changes in plasma levels of fat-soluble vitamins A, D, and E, serum total protein, pre-albumin, albumin, and transferrin. Changes in weight gain and standardized body weight and body mass index (BMI) were also examined. Weight and BMI were normalized to age- and sex-specific z scores of healthy individuals aged 2 to 20 years using data from the Centers for Disease Control and Prevention.

tal protein, pre-albumin, albumin, and transferrin. Changes in weight gain and standardized body weight and body mass index (BMI) were also examined. Weight and BMI were normalized to age- and sex-specific z scores of healthy individuals aged 2 to 20 years using data from the Centers for Disease Control and Prevention. Safety and tolerability outcomes included the frequency and severity of adverse events (AEs) and unanticipated adverse device effects (UADEs), incidence of GI symptoms, clinical and laboratory findings, vital signs, and use of concomitant medications. AEs and UADEs were classified as related to RELiZORB if there was strong medical evidence of an association between the AE or UADE and RELiZORB use. Baseline measurements were defined for safety assessments as the last non-missing measurement taken before the administration of RELiZORB, for GI symptoms as symptoms recorded in the GI-symptom diary during the run-in period, and for efficacy assessments as those taken on day 0, or on day –14 if missing result for day 0.

seline measurements were defined for safety assessments as the last non-missing measurement taken before the administration of RELiZORB, for GI symptoms as symptoms recorded in the GI-symptom diary during the run-in period, and for efficacy assessments as those taken on day 0, or on day –14 if missing result for day 0. During the treatment phase of the study, from day 0 to 90, any UADEs were identified during scheduled study visits, through spontaneously reported medical complaints, and through investigations of non-scheduled healthcare visits. All AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA Version 18.0) and summarized based on the treatment at the time the event started. Participants recorded GI symptoms and concomitant medications in a study-specific symptom diary for 7 days before the day 0 visit and again before visits for days 30, 60, and 90 (Supplemental Digital Content 1).

latory Activities (MedDRA Version 18.0) and summarized based on the treatment at the time the event started. Participants recorded GI symptoms and concomitant medications in a study-specific symptom diary for 7 days before the day 0 visit and again before visits for days 30, 60, and 90 (Supplemental Digital Content 1). Statistical analysis was conducted using SAS software version 9.3 (SAS, Inc, Cary, NC). Sample size calculations estimated that 30 participants would be needed for 80% power to detect a minimum effect size of 0.53 based on a maximum DHA standard deviation (SD) of 1.5%. Assuming the same effect for DHA+EPA as for DHA alone with 30 participants, the study would have 80% power to detect a 0.8% absolute increase or 30% relative increase in DHA+EPA levels from baseline to study day 90. Study participants were included in the statistical analysis if they used RELiZORB at least once during the study. All efficacy endpoints were analyzed using mixed-model repeated-measures models, testing for least-square (LS) mean differences between baseline and post-baseline visits, where study visit is a categorical effect and study participant is a random effect.

al analysis if they used RELiZORB at least once during the study. All efficacy endpoints were analyzed using mixed-model repeated-measures models, testing for least-square (LS) mean differences between baseline and post-baseline visits, where study visit is a categorical effect and study participant is a random effect. RESULTS Of 49 patients screened, 44 enrolled as participants in the study, 5 patients discontinued before the RELiZORB treatment period (day 0), 39 used RELiZORB at least once, and 36 patients completed the study (Supplemental Digital Content 2). The 39 participants ranged in age from 5 to 33 years, had a mean duration of enteral tube feeding of 6.2 years (range 0.7–17.5 years), a mean BMI of 17.7 kg/m2, and 9 (23.1%) had cystic fibrosis-related diabetes (Table 1). Each feeding delivered a mean (SD) volume of 756 (186) mL at a rate of 120 (77) mL/h, and 5.8 PERT (3.1) capsules were used per EN feeding. Over the course of the 90-day experimental treatment period, participants used RELiZORB for a mean (SD) of 68.4 (18.43) days with a mean (SD) days per week of enteral tube feedings of 5.3 (1.2) days and a mean (SD) formula volume per day of 755.9 (145.16) mL.

mL/h, and 5.8 PERT (3.1) capsules were used per EN feeding. Over the course of the 90-day experimental treatment period, participants used RELiZORB for a mean (SD) of 68.4 (18.43) days with a mean (SD) days per week of enteral tube feedings of 5.3 (1.2) days and a mean (SD) formula volume per day of 755.9 (145.16) mL. The omega-3 index increased from a baseline value of 4.4%, which was below the target range of 8%, to 8.4% at 60 days and 9.4% at 90 days (P < 0.001 for each increase from baseline to 60 and 90 days). (Fig. 1). The magnitude and significance of omega-3 index increases were similar in the 2 younger age groups (≤12 years, 13–18 years) to the overall findings, but in adults (≥19 years) the differences were not statistically significant at day 60 (P = 0.051), likely because of the small sample size (n = 5) (Supplemental Digital Content 3). FIGURE 1 Changes in erythrocyte membrance fatty acid composition (%) for omega-3 index (ITT population)—composition of FA in erythrocyte membranes was measured by gas chromatography-mass spectrometry (GC/MS) (OmegaQuant, LLC, Sioux Falls, SD).

The omega-3 index increased from a baseline value of 4.4%, which was below the target range of 8%, to 8.4% at 60 days and 9.4% at 90 days (P < 0.001 for each increase from baseline to 60 and 90 days). (Fig. 1). The magnitude and significance of omega-3 index increases were similar in the 2 younger age groups (≤12 years, 13–18 years) to the overall findings, but in adults (≥19 years) the differences were not statistically significant at day 60 (P = 0.051), likely because of the small sample size (n = 5) (Supplemental Digital Content 3). FIGURE 1 Changes in erythrocyte membrance fatty acid composition (%) for omega-3 index (ITT population)—composition of FA in erythrocyte membranes was measured by gas chromatography-mass spectrometry (GC/MS) (OmegaQuant, LLC, Sioux Falls, SD). All secondary efficacy outcomes changed significantly from baseline to each post-baseline visit. Values increased for the composition (%) of erythrocyte membrane total DHA and total EPA, and plasma concentrations of total DHA, total EPA, and total DHA+EPA. The omega-6 to omega-3 FA ratios decreased in both erythrocyte membranes and plasma (Table 2). All exploratory efficacy outcomes including serum levels of fat-soluble vitamins A, D, and E in plasma total protein, pre-albumin, albumin, and transferrin were within normal ranges at study entry and remained so throughout the 90-day study treatment period. The notable exception among the exploratory outcomes examined was weight and age- and sex-specific z score means of BMI which were below target goal at study entry and did not change significantly over the course of the study period (data not shown).

y entry and remained so throughout the 90-day study treatment period. The notable exception among the exploratory outcomes examined was weight and age- and sex-specific z score means of BMI which were below target goal at study entry and did not change significantly over the course of the study period (data not shown). In the RELiZORB treatment period, at least 1 AE was reported by 29 (74%) patients with greatest incidence for respiratory (46%), infections (20%), and investigations (20%), which included decreased forced expiratory volume, pulmonary function test, and vitamin D levels. Only 1 AE reported as constipation was judged by the principal investigator (PI) to be possibly device related. No AEs resulted in discontinuation of enteral feeding. At least 1 UADE was reported by 10 of 39 (25.6%) participants, including 2 participants with infections and 8 with respiratory, thoracic, and mediastina disorders, but none were classified as being related to the RELiZORB use. Only 1 of these participants discontinued enteral feeds because of their UADE, which was an infective pulmonary exacerbation of CF requiring hospitalization unrelated to RELiZORB use. Using the 7-day GI symptom diary, a decreasing number of participants reported symptoms over the course of the study. Abdominal pain and gas were the most commonly reported symptoms (Table 3). There were no deaths in the study.

infective pulmonary exacerbation of CF requiring hospitalization unrelated to RELiZORB use. Using the 7-day GI symptom diary, a decreasing number of participants reported symptoms over the course of the study. Abdominal pain and gas were the most commonly reported symptoms (Table 3). There were no deaths in the study. DISCUSSION Findings from the ASSURE study demonstrate that 90 days of RELiZORB use with overnight EN feedings is safe, well-tolerated, and associated with improvements of LCPUFA levels in patients with CF and EPI. Despite having used overnight EN feedings for an average of 6 years, patients entering the study had low plasma and erythrocyte levels of DHA and EPA reflecting a poor nutritional status. During the study, participant DHA and EPA levels normalized in both the plasma and erythrocyte membranes indicating improvement in fat malabsorption with the use of RELiZORB. Serum protein and fat-soluble vitamin levels were normal at study baseline and did not change significantly with 90 days of RELiZORB suggesting that low FA levels as well as low BMI were the main nutritional deficiencies observed in the study population.

cating improvement in fat malabsorption with the use of RELiZORB. Serum protein and fat-soluble vitamin levels were normal at study baseline and did not change significantly with 90 days of RELiZORB suggesting that low FA levels as well as low BMI were the main nutritional deficiencies observed in the study population. As in the currently reported ASSURE study, participants in the previous RELiZORB study were deficient in DHA and EPA at study entry even though they had been taking supplemental EN feedings for over 6 years (5). In both studies, Impact Peptide 1.5 was used as the formula with RELiZORB because it has a relatively high percentage of fat as long-chain TGs (MCT:LCT 50:50), which are more difficult than medium-chain TGs to absorb in the absence of pancreatic lipase, making it a more rigorous test of RELiZORB than other enteral formulas (5,22). In the previous short-term study, plasma DHA and EPA levels increased significantly with levels 2.8 times higher than controls in the randomized cross-over phase (5). In the ASSURE study, omega-3 FA erythrocyte levels increased with longer-term RELiZORB use, supporting the role of RELiZORB in normalizing deficient DHA and EPA levels and maintaining them over longer periods when RELiZORB was utilized with overnight EN feeds.

igher than controls in the randomized cross-over phase (5). In the ASSURE study, omega-3 FA erythrocyte levels increased with longer-term RELiZORB use, supporting the role of RELiZORB in normalizing deficient DHA and EPA levels and maintaining them over longer periods when RELiZORB was utilized with overnight EN feeds. The findings of this longer-duration study also confirm the favorable safety and efficacy findings of the previous clinical RELiZORB studies (5). In a previous study, GI symptoms were reported less frequently and with less severity in participants using RELiZORB. Participant-reported appetite was also better when using RELiZORB than just oral PERT (5). In the ASSURE study, the frequency of RELiZORB-related UADEs was low, despite a high overall frequency of AEs, which were primarily respiratory events unrelated to the study treatment. In addition, the number of participants reporting GI symptoms decreased in the group using RELiZORB throughout the ASSURE study.

T (5). In the ASSURE study, the frequency of RELiZORB-related UADEs was low, despite a high overall frequency of AEs, which were primarily respiratory events unrelated to the study treatment. In addition, the number of participants reporting GI symptoms decreased in the group using RELiZORB throughout the ASSURE study. It is not possible to draw definitive conclusions from the current and previous RELiZORB studies about the influence of RELiZORB use on changes in patient anthropometric measurements. While weight, BMI z-scores and percentiles were not significantly different from baseline to 90 days, 61% (20/33) patients in the ASSURE study had improvements in weight z scores and percentiles over the course of the study. Furthermore, recent data based on retrospective case series of patients from CF treatment centers has reported that RELiZORB use was associated with improvements in weight gain, weight for age z scores, the ability to meet or exceed age appropriate weight goals while simultaneously reducing GI symptoms associated with fat malabsorption (25–27).

ased on retrospective case series of patients from CF treatment centers has reported that RELiZORB use was associated with improvements in weight gain, weight for age z scores, the ability to meet or exceed age appropriate weight goals while simultaneously reducing GI symptoms associated with fat malabsorption (25–27). While this study has several strengths, it also has some limitations. It is a relatively small study, but represents 1% of the estimated 3600 US patients with CF who receive supplemental enteral tube feedings (4). A strength of the study is its long duration compared to the previous 7-day study of RELiZORB use (5). In addition, the study was strictly open label and was not intended to compare outcomes between participants who did and did not use RELiZORB. Furthermore, other than EN, dietary intake was not recorded during the study, limiting the ability to assess whether oral caloric intake was adequate to lead to weight gain. Additionally, the study may not have been long enough to observe an increase in body weight or BMI for patients in a relatively fat-starved condition (13). Because the current study did not measure body tissue composition, it is unknown whether study participants improved their tissue composition without changing body weight or size. It may turn out that nutritional health may be more accurately measured using biomarkers other than the traditional body weight and BMI (28,29). Body tissue composition, including fat mass, fat-free mass, and lean body mass may be more important to overall health than body size or mass (23). It is likely that body tissue composition is an indirect measure of important cellular and molecular processes that may be directly measured using molecules such as LCPUFAs, which are essential building blocks of cell membranes and play important roles in cell and tissue function throughout the body (30–32).

mass (23). It is likely that body tissue composition is an indirect measure of important cellular and molecular processes that may be directly measured using molecules such as LCPUFAs, which are essential building blocks of cell membranes and play important roles in cell and tissue function throughout the body (30–32). Given the strengths of the ASSURE study and the finding of improved LCPUFA levels with RELiZORB use in patients with CF and EPI on EN tube feedings, RELiZORB is a safe and effective option for increasing fat absorption in this population. Patients with CF and EPI may have poor nutritional status despite long-term use of supplemental EN feedings. This study establishes the safety and tolerability of RELiZORB and shows its potential to normalize fat absorption, improve symptoms commonly associated with fat malabsorption and enhance nutritional status in patients with CF receiving EN feedings. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Acknowledgments The authors thank Chantal Caviness, MD PhD, of Eubio Medical Communications as well as Rita Pease and Archie Stone, PhD, of Alcresta Therapeutics, Inc. for providing medical writing support as well as Robert Gallotto of Alcresta Therapeutics, Inc. for his input into the design of this study. This study was funded by Alcresta Therapeutics, Inc. www.clinicaltrials.gov Identifier: NCT02750501.

Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Acknowledgments The authors thank Chantal Caviness, MD PhD, of Eubio Medical Communications as well as Rita Pease and Archie Stone, PhD, of Alcresta Therapeutics, Inc. for providing medical writing support as well as Robert Gallotto of Alcresta Therapeutics, Inc. for his input into the design of this study. This study was funded by Alcresta Therapeutics, Inc. www.clinicaltrials.gov Identifier: NCT02750501. D.G. is an employee of Alcresta Therapeutics, Inc; S.D.F. has received research support from Alcresta Therapeutics, Inc. The remaining authors report no conflicts of interest. TABLE 1 Absorption and Safety With Sustained Use of RELiZORB Evaluation study demographics and baseline characteristics (ITT population) Parameter ITT population (n = 39) Age, y Mean (SD) 13.8 (5.4) Min, Max 5, 33 Age category, y ≤12 17 (43.6%) 13–18 17 (43.6%) ≥19 5 (12.8%) Gender Male/female 24 (61.5%)/15 (38.5%) Weight, kg Mean (SD) 40.83 (12.3) Height, cm Mean (SD) 149.84 (18.9) BMI, kg/m2 Mean (SD) 17.68 (1.8) CFRD category CFRD 9 (23.1%) Non-CFRD 30 (76.9%) BMI = body mass index; CFRD = cystic fibrosis-related diabetes; ITT = intent-to-treat. TABLE 2 Changes in docosahexaenoic acid, eicosapentaenoic acid, docosahexaenoic acid+eicosapentaenoic acid. and ω-6/ω-3 ratio in plasma

Parameter ITT population (n = 39) Age, y Mean (SD) 13.8 (5.4) Min, Max 5, 33 Age category, y ≤12 17 (43.6%) 13–18 17 (43.6%) ≥19 5 (12.8%) Gender Male/female 24 (61.5%)/15 (38.5%) Weight, kg Mean (SD) 40.83 (12.3) Height, cm Mean (SD) 149.84 (18.9) BMI, kg/m2 Mean (SD) 17.68 (1.8) CFRD category CFRD 9 (23.1%) Non-CFRD 30 (76.9%) BMI = body mass index; CFRD = cystic fibrosis-related diabetes; ITT = intent-to-treat. TABLE 2 Changes in docosahexaenoic acid, eicosapentaenoic acid, docosahexaenoic acid+eicosapentaenoic acid. and ω-6/ω-3 ratio in plasma Baseline, n = 36 Day 30, n = 36 Day 60, n = 36 Day 90, n = 36 DHA, μg/mL 50.33 (34.1) 102.64 (39.3) 96.77 (36.4) 102.36 (36.3) EPA, μg/mL 22.41 (21.1) 65.34 (45.9) 64.01 (49.6) 64.09 (41.5) Total DHA+EPA, μg/mL 72.73 (52.1) 167.99 (82.5) 160.78 (83.2) 166.46 (73.7) ω-6/ω-3 ratio 11.52 (4.3) 5.31 (2.5) 5.63 (3.3) 5.23 (3.1) Plasma concentrations of total DHA and EPA were measured by ultra-high-performance liquid chromatography (UHPLC) (PPD LLC, Richmond, VA). Total DHA+EPA were calculated by adding the concentrations of total DHA and EPA. DHA = docosahexaenoic acid; EPA = eicosapentaenoic acid. TABLE 3 Gastrointestinal adverse events for ITT population (n = 39)

Baseline, n = 36 Day 30, n = 36 Day 60, n = 36 Day 90, n = 36 DHA, μg/mL 50.33 (34.1) 102.64 (39.3) 96.77 (36.4) 102.36 (36.3) EPA, μg/mL 22.41 (21.1) 65.34 (45.9) 64.01 (49.6) 64.09 (41.5) Total DHA+EPA, μg/mL 72.73 (52.1) 167.99 (82.5) 160.78 (83.2) 166.46 (73.7) ω-6/ω-3 ratio 11.52 (4.3) 5.31 (2.5) 5.63 (3.3) 5.23 (3.1) Plasma concentrations of total DHA and EPA were measured by ultra-high-performance liquid chromatography (UHPLC) (PPD LLC, Richmond, VA). Total DHA+EPA were calculated by adding the concentrations of total DHA and EPA. DHA = docosahexaenoic acid; EPA = eicosapentaenoic acid. TABLE 3 Gastrointestinal adverse events for ITT population (n = 39) Number of subjects with incidence of GI symptoms PERT + Usual enteral formula PERT + Peptamen 1.5 90-day open-label treatment period RELiZORB with Impact Peptide 1.5 Observation period Run-in period Day 30 Day 60 Day 90 Any symptom 23 (59.0%) 22 (56.4%) 16 (41.0%) 17 (43.6%) 12 (30.8%) Abdominal pain 12 (30.8%) 12 (30.8%) 11 (28.2%) 9 (23.1%) 7 (17.9%) Bloating 4 (10.3%) 2 (5.1%) 3 (7.7%) 2 (5.1%) 3 (7.7%) Constipation 2 (5.1%) 0 0 2 (5.1%) 2 (5.1%) Diarrhea 4 (10.3%) 6 (15.4%) 3 (7.7%) 4 (10.3%) 0 Gas 16 (41.0%) 16 (41.0%) 9 (23.1%) 11 (28.2%) 8 (20.5%) Indigestion/heartburn 3 (7.7%) 2 (5.1%) 2 (5.1%) 4 (10.3%) 3 (7.7%) Nausea 6 (15.4%) 5 (12.8%) 2 (5.1%) 4 (10.3%) 3 (7.7%) Steatorrhea (fatty stool) 2 (5.1%) 3 (7.7%) 2 (5.1%) 2 (5.1%) 0 Vomiting 4 (10.3%) 4 (10.3%) 2 (5.1%) 2 (5.1%) 0 GI = gastrointestinal; ITT = intent-to-treat; PERT = pancreatic enzyme replacement therapy.

What Is KnownPowdered formula has been associated with serious illness due to infections with Salmonella spp and Cronobacter sakazakii. International guidelines have been published to indicate how to prepare powdered formula, using water not below 70°C. What Is NewIn practical use, starting from water at 70°C, maximum temperatures registered in 200 mL of reconstituted formula were between 57.5 and 60°C. These conditions may not ensure inactivation of pathogens. Water at higher temperature must be considered to prepare powdered formula, to improve the food safety, but pros and cons need further evaluation.

What Is NewIn practical use, starting from water at 70°C, maximum temperatures registered in 200 mL of reconstituted formula were between 57.5 and 60°C. These conditions may not ensure inactivation of pathogens. Water at higher temperature must be considered to prepare powdered formula, to improve the food safety, but pros and cons need further evaluation. Salmonella spp. and Cronobacter sakazakii are ubiquitous Gram-negative, facultative anaerobic, motile, non-spore–forming bacteria that cause human disease (1). Since August 2017, an outbreak of Salmonella Agona linked to the consumption of infant formula (powdered formula) has been ongoing in France. As of January 11, 2018, the outbreak affected 39 infants (children <1 year of age): 37 in France, 1 in Spain and 1 in Greece, but new cases may be detected. Epidemiological investigations in humans and traceability investigations in food, identified seven different brands of infant formula from a single processing company in France as the vehicles of infection (2). No other common foods or drinks were identified among cases. A variety of types of water (tap and different brands of bottled water) were used to prepare these products, and no evidence was given that infections were related to inappropriate methods for preparing and handling powdered infant formula by the caregivers (3). Rather, some weaknesses could have been happened in the production process (2). Salmonella infections due to the consumption of contaminated infant formula can occur in healthy term infants, while C sakazakii mainly occur in preterm, unhealthy, immunocompromised infants, hospitalized in neonatal intensive care units (4,5).

er, some weaknesses could have been happened in the production process (2). Salmonella infections due to the consumption of contaminated infant formula can occur in healthy term infants, while C sakazakii mainly occur in preterm, unhealthy, immunocompromised infants, hospitalized in neonatal intensive care units (4,5). The current consolidated Reg. CE 2073/2005 (6) states for Salmonella spp and for Cronobacter spp (Enterobacter sakazakii) the complete absence in 30 samples of 25 and 10 g, respectively, during the shelf-life in infant formula to be distributed. Starting early 2000s, following an outbreak of C sakazakii(7) increasing attention has been addressed to the safety issue of reconstituted infant formulas, particularly focusing on products and recipients (8). In 2007, World Health Organization (WHO) in collaboration with the Food and Agriculture Organization of the United Nations (FAO), published the guidelines about the safe preparation, storage and handling of powdered infant formula. The recommendations in this guidance document were largely based on the quantitative risk assessment for C sakazakii and not evidence based, establishing that the inclusion of a pathogenic lethal step at preparation (eg, reconstitution of formula with water ≥70°C), and a decrease during holding and feeding time would effectively reduce the risk. No risk assessment was carried out for Salmonella spp, but the expert group reported that the basic risk control principles for C sakazakii would also hold for S enterica(9).

t preparation (eg, reconstitution of formula with water ≥70°C), and a decrease during holding and feeding time would effectively reduce the risk. No risk assessment was carried out for Salmonella spp, but the expert group reported that the basic risk control principles for C sakazakii would also hold for S enterica(9). In our study we monitored the time/temperature profile of formula and the survival of Salmonella spp and C sakazakii during the preparation of powdered infant formula following the procedure suggested by FAO/WHO (10) and an alternative procedure. METHODS Formula Contamination Three different Salmonella spp strains (Salmonella Typhimurium ATCC 14028, Salmonella Abortus ovis CIP 55132 from Institut Pasteur, Paris and Salmonella Agbeni, wild strain) and one C sakazakii strain (CIP 103183 from Institut Pasteur, Paris, France), were used to contaminate powdered formula (commercial formula). Before use, the individual Salmonella strains were serially diluted and combined in equal volumes to obtain a multistrain cocktail. C sakazakii strain was separately diluted. Sterile plastic 250 mL bottles containing 30 g of powdered formula were separately contaminated with 2 mL of each pathogen suspension (to reach a final inoculum level of about 1 to 1.5 log CFU/mL in reconstituted formula). Another bottle containing 30 g of powdered formula was prepared to check the time/temperature profile during the test.

L bottles containing 30 g of powdered formula were separately contaminated with 2 mL of each pathogen suspension (to reach a final inoculum level of about 1 to 1.5 log CFU/mL in reconstituted formula). Another bottle containing 30 g of powdered formula was prepared to check the time/temperature profile during the test. Preparation Procedures Formula was reconstituted with 200 mL (15% w/v) of hot water from commercial plastic bottles. To simulate the procedure suggested by FAO/WHO, water was previously brought to a boil (500 mL) and allowed to cool at room temperature up to 70°C; to evaluate an alternative procedure, water was previously brought to a boil (500 mL) and allowed to cool at room temperature for 10 minutes. In both cases, at the end of the reconstitution phase the product was mixed by stirring the bottle for 120 seconds and then it was stored at room temperature for 2 hours (maximum storage time suggested for reconstituted product). A positive control was prepared adding water at room temperature (∼24°C) to contaminated formula, to calculate the initial inoculum level for each pathogen. Three independent replicates were performed for each preparation procedure.

oom temperature for 2 hours (maximum storage time suggested for reconstituted product). A positive control was prepared adding water at room temperature (∼24°C) to contaminated formula, to calculate the initial inoculum level for each pathogen. Three independent replicates were performed for each preparation procedure. Sampling Times and Analyses Both pathogens were enumerated in positive control by plate count methods in Blood Agar Base (incubated at 37°C). In the other contaminated bottles, formula samples were analyzed after the formula mixing and after 2 hours of storage at room temperature. Salmonella spp and C sakazakii presence/absence was observed in 25 mL of product by ISO 6579-1 (11) and ISO 22964 (12), respectively, to evaluate their inactivation. The time/temperature profile in all the replicates was monitored by datalogger Thermo Button (Astori Tecnica s.n.c., Brescia, Italy). Data Analyses Bacterial counts were converted to log CFU/mL. The individual means and standard deviations were determined as the average result of 2 samples for 3 replicates. When the pathogens levels were below to the detection limit (1 log CFU/mL), results were reported as presence or absence in 25 mL.

Sampling Times and Analyses Both pathogens were enumerated in positive control by plate count methods in Blood Agar Base (incubated at 37°C). In the other contaminated bottles, formula samples were analyzed after the formula mixing and after 2 hours of storage at room temperature. Salmonella spp and C sakazakii presence/absence was observed in 25 mL of product by ISO 6579-1 (11) and ISO 22964 (12), respectively, to evaluate their inactivation. The time/temperature profile in all the replicates was monitored by datalogger Thermo Button (Astori Tecnica s.n.c., Brescia, Italy). Data Analyses Bacterial counts were converted to log CFU/mL. The individual means and standard deviations were determined as the average result of 2 samples for 3 replicates. When the pathogens levels were below to the detection limit (1 log CFU/mL), results were reported as presence or absence in 25 mL. RESULTS The initial concentrations of Salmonella spp. and C. sakazakii artificially inoculated in formula were 1.2 ± 0.1 log CFU/mL and 1.4 ± 0.2 log CFU/mL respectively. During the simulation of the procedure suggested by FAO/WHO, the formula was prepared using hot water at 70°C. Just after formula mixing (120 seconds) and after the storage time (2 hours at room temperature), pathogens levels were below the detection limit (<1 log CFU/mL) but the presence of viable cells was observed in 25 mL of product (Table 1). Only in one replicate (replicate 2), we observed the pathogens absence in 25 mL after the formula mixing, but results were positive (presence in 25 mL) after the storage time of replicate (2 hours), indicating the cells survival during the procedure of preparation.

able cells was observed in 25 mL of product (Table 1). Only in one replicate (replicate 2), we observed the pathogens absence in 25 mL after the formula mixing, but results were positive (presence in 25 mL) after the storage time of replicate (2 hours), indicating the cells survival during the procedure of preparation. The time/temperature profiles registered during the tests are reported in Figure 1. The results show that the maximum temperature registered in 200 mL of reconstituted formula ranged between 60 and 57.5°C during the bottles mixing (Fig. 1A). FIGURE 1 Time/temperature profiles monitored in formula reconstituted with water at 70°C (FAO/WHO) (A) and with water at 87 ± 2°C (alternative procedure) (B). Data are reported for all tests (3 replicates). For the pathogens inactivation with a different time/temperature profile, 500 mL of boiling water was cooled at room temperature for 10 minutes and then it was added to powdered formula. Starting from water temperature at 87 ± 2°C (range of 85–89°C), the maximum temperature registered in formula was 76°C (range of 73.5–76°C) (Fig. 1B). In these conditions, no pathogens survival was observed in formula sampled after the mixing and after the storing time in all investigated replicates (Table 1).

formula. Starting from water temperature at 87 ± 2°C (range of 85–89°C), the maximum temperature registered in formula was 76°C (range of 73.5–76°C) (Fig. 1B). In these conditions, no pathogens survival was observed in formula sampled after the mixing and after the storing time in all investigated replicates (Table 1). DISCUSSION The contamination of low-water activity foods (eg, formula powder, egg powder, chocolate, peanut butter) with Salmonella is a well-known problem. It has been observed that foodborne pathogens may not grow in these food matrices, but they can survive for long periods once the food is contaminated, thus representing a significant risk even at low levels (13). Therefore, also powdered infant formula may be intrinsically contaminated with pathogens that can cause serious illness in infants. An even more challenging situation was identified for C sakazakii within food factories potentially available to young infants and households (14). For these reasons, Salmonella spp and C sakazakii are of major concern in the infant-food industry, because ingestion of very few (10–100 CFU/mL) Salmonella spp cells by young children may cause severe illness, while on the other hand C sakazakii may be at the origin severe infections such as meningitis, bacteriemia, and necrotizing enterocolitis, particularly in preterm, immune-compromised infants, with a death rate up to 80% and an infectious dose estimated 1000 cells (15). The recent Salmonellosis outbreak in France, Spain, and Greece due to the consumption of contaminated formula powder highlights the need to improve in any case the safety of these products and to provide more information about their correct use. International guidelines, for instance, the ESPGHAN Committee on Nutrition in 2004 (8) have been developed to diminish the risks associated with the presence of both Salmonella spp and C sakazakii in powdered infant formula, while highlighting the role of good hygiene practices in the preparation of powdered formula, its safe storage, transport, and use. Reconstitution of powdered formula for infants with hot water (no less than 70°C) has been recommended by the FAO/WHO (10) to reduce the biological risk for Salmonella spp. and C sakazakii.

hile highlighting the role of good hygiene practices in the preparation of powdered formula, its safe storage, transport, and use. Reconstitution of powdered formula for infants with hot water (no less than 70°C) has been recommended by the FAO/WHO (10) to reduce the biological risk for Salmonella spp. and C sakazakii. This procedure may, however, be considered not safe enough to guarantee the product safety because Salmonella spp are reported to have an increased heat tolerance at temperatures above 70°C in foods at low water activity achieved by drying (16) and it may be difficult to replicate the procedure suggested by FAO/WHO in a domestic setting.

however, be considered not safe enough to guarantee the product safety because Salmonella spp are reported to have an increased heat tolerance at temperatures above 70°C in foods at low water activity achieved by drying (16) and it may be difficult to replicate the procedure suggested by FAO/WHO in a domestic setting. In our study, monitoring in real time the temperature profile during formula preparation shows that even if starting from water at 70°C a rapid cooling may occur, providing non-lethal conditions for some preexisting pathogens in the powdered formula. Accordingly, Salmonella spp and C sakazakii may survive in reconstituted formula and, as a consequence, cause severe illness in infants. Just using hot water (>85°C in our study) the formula reached temperatures lethal to pathogens to reduce the associated biological risks. These results are consistent with the progressive risk reductions described in the passage from water at 50°C to reconstitution at 70°C (9). Accordingly, higher temperatures are needed to reduce the biological risk within the “real-world” associated conditions in which these products are prepared and used. Indeed, among parents of formula-fed infants <12 weeks old, only 22% were reported using water heated at ≥70°C to dilute powdered formula (17).

70°C (9). Accordingly, higher temperatures are needed to reduce the biological risk within the “real-world” associated conditions in which these products are prepared and used. Indeed, among parents of formula-fed infants <12 weeks old, only 22% were reported using water heated at ≥70°C to dilute powdered formula (17). In conclusion, powdered infant formula could not be sterile. Industry and regulators play the primary critical role in minimizing the risk of illness from consumption of reconstituted powdered formula and ensuring that the product is safe (18). More information are needed to improve the safety of these products to propose a procedure able to guarantee the safety of the product by inactivating pathogenic bacteria and be easily feasible in actual domestic conditions. In maternity units where premature infants are looked after, sterile liquid formulae appear to be a suitable alternative to powder formulae to reduce the risk of formula-borne infections (8). Once a general agreement is reached on the optimal temperature of reconstitution, the effects on the nutrient bio-availability should be carefully evaluated. Acknowledgments The authors would like to thank Paola Monastero, Emanuela Bonometti and Franca Rossi (Food Microbiology Department, IZSLER, Brescia, Italy) for the technical assistance during the tests. The authors report no conflicts of interest. TABLE 1 Microbiological results about the Salmonella spp and C sakazakii survival in powdered formula using different procedures.

Acknowledgments The authors would like to thank Paola Monastero, Emanuela Bonometti and Franca Rossi (Food Microbiology Department, IZSLER, Brescia, Italy) for the technical assistance during the tests. The authors report no conflicts of interest. TABLE 1 Microbiological results about the Salmonella spp and C sakazakii survival in powdered formula using different procedures. Pathogen Inoculum level (log CFU/mL) After formula mixing After Storage time FAO/WHO* Alternative procedure FAO/WHO Alternative procedure Salmonella spp. 1.2 ± 0.1 + − + − − − + + + − − − C. sakazakii 1.4 ± 0.2 + − + − − − + + + − − − When the pathogen levels were below to the detection limit (1 log CFU/mL) the presence/absence in 25 mL of formula was determined by ISO 6579-1 (11) and ISO 22964 (12). *Data are reported as presence (+) or absence (−) of the pathogens in 25 mL of formula (3 replicates).

What Is KnownThe administration of nonsteroidal anti-inflammatory drugs, such as indomethacin and ibuprofen, is associated with several adverse effects in neonates. Indomethacin impairs key metabolic pathways in the immature human intestine. What Is NewKetoprofen generates fewer deleterious effects than other nonsteroidal anti-inflammatory drugs on the human mid-gestational intestine, possibly due to its role as double inhibitor of the enzymes involved in arachidonic acid metabolism. The intestinal response to at least some nonsteroidal anti-inflammatory drugs may vary from individual to individual. As compared to ketoprofen alone, an effect of the H2S-releasing derivative of ketoprofen could not be demonstrated in the organ culture set up.

What Is NewKetoprofen generates fewer deleterious effects than other nonsteroidal anti-inflammatory drugs on the human mid-gestational intestine, possibly due to its role as double inhibitor of the enzymes involved in arachidonic acid metabolism. The intestinal response to at least some nonsteroidal anti-inflammatory drugs may vary from individual to individual. As compared to ketoprofen alone, an effect of the H2S-releasing derivative of ketoprofen could not be demonstrated in the organ culture set up. Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most widely used drugs in the neonatal period. NSAIDs, such as indomethacin and ibuprofen, are used as tocolytic agents (1) as well as for the treatment of the most common cardiovascular abnormality among neonates, the patent ductus arteriosus (PDA) (2). It is well known that the administration of NSAIDs, in adults, is associated with several adverse effects, particularly gastrointestinal, such as gastropathies and enteropathies (3–6). It is therefore not surprising to see that those NSAIDs also cause several harmful side effects in neonates. In addition to being associated with a greater risk of miscarriage and congenital malformation, pre- and post-natal exposure to NSAIDs are also responsible for several disorders affecting many of the infant's organs, most notably the gastrointestinal tract (7). We have previously demonstrated that NSAIDs, such as indomethacin, can directly induce damaging effects on the human immature intestine (8).

malformation, pre- and post-natal exposure to NSAIDs are also responsible for several disorders affecting many of the infant's organs, most notably the gastrointestinal tract (7). We have previously demonstrated that NSAIDs, such as indomethacin, can directly induce damaging effects on the human immature intestine (8). A number of derivatives have been developed with the aim of minimizing the deleterious effects brought on by the administration of classic NSAIDs. Among these, some have exploited the beneficial cellular effects known for biologically active gases, such as hydrogen sulfide (H2S) (9). Indeed, several studies have shown that H2S could counterbalance some of the adverse effects brought on by the administration of NSAIDs by limiting certain elements of inflammation (10). The contribution of H2S occurs in 2 ways: by anti-inflammatory actions, such as the inhibition of the activation of NFκB, and by protective effects on the gastrointestinal tract, as the inhibition of leukocyte adhesion and the reduction of pro-inflammatory cytokine expression, such as TNFα, IL-1β and IFNγ (11–13). Recent studies have shown that the H2S-releasing derivatives are able to prevent the development of intestinal damage while being as effective as their classic counterparts (11,14,15). In this study, we tested another NSAID, ketoprofen and its H2S-releasing derivative on the immature small intestine in organ culture.

. Recent studies have shown that the H2S-releasing derivatives are able to prevent the development of intestinal damage while being as effective as their classic counterparts (11,14,15). In this study, we tested another NSAID, ketoprofen and its H2S-releasing derivative on the immature small intestine in organ culture. METHODS Tissues To evaluate the impact of ketoprofen and its H2S-releasing derivative ATB-352 on the human immature intestinal mucosa, a set of 6 specimens of small intestine (ileum) obtained from 6 fetuses following legal pregnancy interruption between 17 and 20 weeks of gestation were used. No tissues were collected from cases associated with known fetal abnormalities or intrauterine fetal demise. Studies were approved by the institutional review board for the use of human material from the “Centre Intégré Universitaire de Santé et de Services Sociaux de l’Estrie - Centre Hospitalier Universitaire de Sherbrooke.”

es were collected from cases associated with known fetal abnormalities or intrauterine fetal demise. Studies were approved by the institutional review board for the use of human material from the “Centre Intégré Universitaire de Santé et de Services Sociaux de l’Estrie - Centre Hospitalier Universitaire de Sherbrooke.” Serum-free Organ Culture Small intestinal tissues (ileum) obtained from 6 fetuses were prepared for organ culture as previously described (16,17). Briefly, each intestinal tissue was cut into several 5 × 5 mm explants which were maintained in organ culture dishes (Falcon Plastics, Fisher Scientific, Ottawa, ON, Canada) with Leibovitz L-15 serum-free culture medium containing 40 mg/mL amphotericin and 40 mg/mL mycostatin at the interface of a 95% air to 5% CO2 gas mixture at 37°C. Each culture dish contained 6 to 9 explants. Two dishes were used for each condition tested for each of the 6 small intestinal samples. Explants were maintained in culture for 48 hours and media was changed daily. The effects of ketoprofen and its derivative ATB-352 were tested at 1 and 10 μM. Prostaglandin E2 (PGE2) was measured as before (8). RNA Extraction, DNase Treatment, and Reverse Transcription RNA was extracted with TRIzol (Invitrogen, Burlington, ON, Canada) according to the manufacturer's protocol and stored at –80°C. RNA samples first underwent DNase treatment (Invitrogen) according to the manufacturer's protocol and then reverse transcription of the samples was performed with Superscript II (Invitrogen).

RNA was extracted with TRIzol (Invitrogen, Burlington, ON, Canada) according to the manufacturer's protocol and stored at –80°C. RNA samples first underwent DNase treatment (Invitrogen) according to the manufacturer's protocol and then reverse transcription of the samples was performed with Superscript II (Invitrogen). Real-time PCR All reactions were performed in an Mx3000P real-time PCR system (Stratagene, Cedar Creek, TX) using Brilliant II SYBR Green QPCR Master Mix (Stratagene) in duplicate as previously described (18,19). Briefly, the run started with 5 minutes of Taq activation at 95°C followed by 40 cycles of melting (95°C, 30 seconds), primer annealing (55°C, 45 seconds) and extension (72°C, 45 seconds) ending with a melting curve analysis to validate the PCR product's specificity. Fluorescence data were acquired after each annealing step. The genes investigated in this study were ATP synthase, H+ transporting, mitochondrial Fo complex, subunit C1 (subunit 9) (ATP5G1), claudin-1 (CLDN1), cyclooxygenase-2 (PTGS2), chemokine, C-X-C motif, ligand 14 (CXCL14), cytochrome P450, family 3, subfamily A, polypeptide 4 (CYP3A4), dual oxidase 2 (DUOX2), intercellular adhesion molecule 1 (ICAM-1), NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 9, 39 kDa (NDUFA9), nitric oxide synthase 2 (NOS2), occludin (OCLN), superoxide dismutase 2 (SOD2), and trefoil factor 1 (TFF1). Peptidylprolyl isomerase A (PPIA) and ribosomal protein S23 (RPS23) were used as reference genes. Primers are listed in Supplemental Table 1 (Supplemental Digital Content) and were generated by the primer formation software Primer3 (http://bioinfo.ut.ee/primer3).

eroxide dismutase 2 (SOD2), and trefoil factor 1 (TFF1). Peptidylprolyl isomerase A (PPIA) and ribosomal protein S23 (RPS23) were used as reference genes. Primers are listed in Supplemental Table 1 (Supplemental Digital Content) and were generated by the primer formation software Primer3 (http://bioinfo.ut.ee/primer3). Data Expression The differential expression of genes was estimated by comparing the expression of untreated (control) and treated explants (ketoprofen or ATB-352) using the equation R = (Etarget)ΔCttarget/(Ereference)ΔCtreference(20). Samples were normalized to a set of 2 reference genes, PPIA and RPS23. The ratio of ATB/ketoprofen was used to evaluate the additional effect of H2S on the human mid-gestational intestine compared with ketoprofen alone. The statistical analyses were made by the software GraphPad Prism 7 (GraphPad, San Diego, CA). The Wilcoxon test (when n = 6) and the one sample t test (when n = 3) were used to estimate the significance of the expression of a gene compared with its expression control (fixed to 1), while the Mann-Whitney test was used to compare the expression of a gene between the various treatments. P values lower than 0.05 (0.06 where specifically indicated) were considered significant.

n = 3) were used to estimate the significance of the expression of a gene compared with its expression control (fixed to 1), while the Mann-Whitney test was used to compare the expression of a gene between the various treatments. P values lower than 0.05 (0.06 where specifically indicated) were considered significant. RESULTS Direct Effects of Ketoprofen on the Human Immature Intestinal Mucosa As previously validated for indomethacin testing, 1 approach to estimate relevant concentrations for testing NSAIDs on small intestinal explants is to mimic circulating levels reported to be efficient for patent ductus arteriosus closure in the neonate and then validate it for the inhibition of PGE2 production (8). Unfortunately, such data were not available for ketoprofen but in 1 study where this NSAID was used as a tocolytic agent, ketoprofen concentrations estimated to be between 2 and 12 μM were measured in the plasma of neonates in the first hours of life (21). When tested on small intestinal explants, the 1 and 10 μM concentrations were found to be sufficient to inhibit 60% and 80% of PGE2 production over a 48-hour period, respectively (Fig. 1A). Experiments were thus performed with the 2 concentrations but only data with 10 μM were further considered.

st hours of life (21). When tested on small intestinal explants, the 1 and 10 μM concentrations were found to be sufficient to inhibit 60% and 80% of PGE2 production over a 48-hour period, respectively (Fig. 1A). Experiments were thus performed with the 2 concentrations but only data with 10 μM were further considered. FIGURE 1 Effects of ketoprofen on the immature intestinal mucosa. A, Inhibition of PGE2 production was evaluated in the presence of 2 concentrations of ketoprofen (mean ± SEM, ∗∗∗P < 0.005). B, qPCR analysis of transcript levels for representative markers of mucosal homeostasis categories after 48 hours of culture in the presence of 10 μM ketoprofen. C, For comparison, markers expressed relative to the same housekeeping genes in response to 1 μM indomethacin under similar conditions. qPCR data are expressed as ratios of treated over untreated segments (Log2 scale). Values from 6 independent biological samples (mean ± SEM). ∗P < 0.05 (#P < 0.06) versus corresponding untreated control segments.

xpressed relative to the same housekeeping genes in response to 1 μM indomethacin under similar conditions. qPCR data are expressed as ratios of treated over untreated segments (Log2 scale). Values from 6 independent biological samples (mean ± SEM). ∗P < 0.05 (#P < 0.06) versus corresponding untreated control segments. To investigate the influence of ketoprofen on the mid-gestation intestine, matched control and ketoprofen-treated explants were tested for the expression of markers for specific metabolic pathways previously identified as being significantly altered in response to indomethacin (8,19). Ketoprofen treatment was found to increase expression of CYP3A4, DUOX2, and SOD2, 3 markers of oxidoreductase activity, CXCL14, a marker of the inflammatory response, and OCLN and CLDN1, 2 structural components of the tight junctions while a reduction of the expression of ATP5G1, a marker of the oxidative phosphorylation pathway, was observed (Fig. 1B). Unexpectedly, these results were found to be different to those observed with indomethacin at comparable effective concentrations (8,19). To allow a more direct comparison, our previous data with indomethacin were recalculated using the same new set of housekeeping genes (PPIA and RPS23) chosen on the basis of their stability in intestinal tissue under stressful conditions (Thibault MP, unpublished data). Recalculation with the new reference genes had little influence on the overall results as shown in Fig. 1C, where only 2 of the 10 tested genes, CYP3A4 and CXCL14, appear to follow the same trend in response to ketoprofen and indomethacin treatments, confirming the differential intestinal response between these 2 NSAIDs.

calculation with the new reference genes had little influence on the overall results as shown in Fig. 1C, where only 2 of the 10 tested genes, CYP3A4 and CXCL14, appear to follow the same trend in response to ketoprofen and indomethacin treatments, confirming the differential intestinal response between these 2 NSAIDs. Two Patterns of Response: Ketoprofen Responders and Non-responders Surprisingly, analysis of PTGS2 (cyclooxygenase-2) in ketoprofen-treated explants revealed highly variable levels of expression of this gene from sample to sample. In fact, as shown on Figure 2A for 6 independent cultures, 3 were found to be unaltered (<50% variation; 0.97 ± 0.23, NS) while the 3 others were induced by more than 100% (2.40 ± 0.14, P = 0.0006) suggesting that some samples were “responders” to ketoprofen treatment at the transcript level while others were “non-responders.” Incidentally, PTGS2 analysis from the indomethacin treated samples showed that the responder/non-responder phenomenon was not elicited, PTGS2 expression being uninduced by indomethacin (0.68 ± 0.18, NS).

t some samples were “responders” to ketoprofen treatment at the transcript level while others were “non-responders.” Incidentally, PTGS2 analysis from the indomethacin treated samples showed that the responder/non-responder phenomenon was not elicited, PTGS2 expression being uninduced by indomethacin (0.68 ± 0.18, NS). FIGURE 2 Characterization of intestinal ketoprofen responders and non-responders. A, qPCR analysis of PTGS2 transcript levels in the 6 independent explant cultures. B, Based on data from (A) qPCR data for the markers of mucosal homeostasis are shown for the 3 responders and the 3 non-responders to ketoprofen. qPCR data are expressed as ratios of treated over untreated segments for each sample for (A) and for the mean of 3 independent biological samples for each responder group (Log2 scale, ∗P < 0.05, #P < 0.06) versus corresponding untreated control segments (mean ± SEM).

3 responders and the 3 non-responders to ketoprofen. qPCR data are expressed as ratios of treated over untreated segments for each sample for (A) and for the mean of 3 independent biological samples for each responder group (Log2 scale, ∗P < 0.05, #P < 0.06) versus corresponding untreated control segments (mean ± SEM). We thus reanalyzed the data taking into consideration this distinction. In this context, overall the responders showed fewer harmful effects that the non-responders (Fig. 2B). As expected, non-responders displayed a deleterious indomethacin-like phenotype (compare Fig. 2B with Fig. 1C). Indeed, while the expression of CYP3A4 tends to always increase, the expression of the other oxidoreductase activities tested varied largely between them, DUOX2, NOS2, and SOD2 being upregulated only in the responders (Fig. 2B). Similarly, the level of genes involved in oxidative phosphorylation and inflammatory response was not altered in the responders while the non-responders displayed an indomethacin-like response. Finally, expression of the intestinal permeability markers OCLN and CLDN1 increased in the responders but not in the non-responders (Fig. 2B).

evel of genes involved in oxidative phosphorylation and inflammatory response was not altered in the responders while the non-responders displayed an indomethacin-like response. Finally, expression of the intestinal permeability markers OCLN and CLDN1 increased in the responders but not in the non-responders (Fig. 2B). Impact of the Contribution of H2S by the Derivative ATB-352 on Gene Expression in the Human Immature Intestinal Mucosa The evaluation of the effects of H2S on the human immature intestinal mucosa was made via the treatment of explants maintained in organ culture with ATB-352, a H2S-releasing derivative of ketoprofen. The qPCR analysis allowed the determination of the effects of H2S release on the metabolic pathways previously analyzed with ketoprofen alone. This comparison was made using an ATB/ketoprofen ratio to estimate the additive effects of a potential release of H2S on the human immature intestinal mucosa. In addition to some minor exceptions, the ATB compound did not seem to significantly affect the expression of the selected genes compared to ketoprofen alone (Table 1). Analysis of representative genes for which expression was found to be attenuated by H2S donors such as ICAM-1 and TNF(12) confirmed the lack of effect of ATB in organ culture (Fig. 3).

ons, the ATB compound did not seem to significantly affect the expression of the selected genes compared to ketoprofen alone (Table 1). Analysis of representative genes for which expression was found to be attenuated by H2S donors such as ICAM-1 and TNF(12) confirmed the lack of effect of ATB in organ culture (Fig. 3). FIGURE 3 Analysis of expression of representative H2S-targeted genes in ATB-352 or ketoprofen treated intestine. qPCR analysis of TNFα and ICAM1 transcript levels in control (Ctrl), ATB-352 (ATB) or ketoprofen (Keto) treated responder and non-responder explants. Data are expressed as ratios of treated over control explants as the mean of 3 independent biological samples for each group (mean ± SEM). ∗Denotes a difference (P < 0.05) versus corresponding untreated control explants. NS = non-significant.

-352 (ATB) or ketoprofen (Keto) treated responder and non-responder explants. Data are expressed as ratios of treated over control explants as the mean of 3 independent biological samples for each group (mean ± SEM). ∗Denotes a difference (P < 0.05) versus corresponding untreated control explants. NS = non-significant. DISCUSSION The anti-inflammatory properties of ketoprofen, like other classical NSAIDs, pass through the nonspecific inhibition of the isoenzymes COX-1 and COX-2 responsible for triggering the inflammation cascade via the production of prostaglandins. This study was undertaken with the aim of estimating the direct effects of ketoprofen on the human immature intestinal mucosa to help in the understanding of the molecular mechanism responsible for NSAID enteropathies (3,7) and to evaluate the potential beneficial effects of exposure to H2S using ATB-352, a H2S-releasing derivative of ketoprofen (14). Overall, while our data showed a relative inefficiency of H2S release on the immature small intestine in the organ culture setup, it revealed a previously undisclosed phenomenon pertaining to a distinct intestinal response to ketoprofen among individuals referred to above as responders versus non-responders.

ofen (14). Overall, while our data showed a relative inefficiency of H2S release on the immature small intestine in the organ culture setup, it revealed a previously undisclosed phenomenon pertaining to a distinct intestinal response to ketoprofen among individuals referred to above as responders versus non-responders. The initial analysis of the effects of ketoprofen, which included 6 independent cultures, showed some trends in the modulation of pro-inflammatory-related genes such as an increase of CXCL14 expression, a pro-inflammatory cytokine (22). Increase of CYP3A4, an abundant form of the cytochrome P450 in the intestine responsible for the metabolism of drugs (23), was also observed but the overall response to ketoprofen was found to be quite distinct to what had been previously reported by our group under similar conditions with another NSAID, indomethacin (8). A more detailed analysis of the data revealed however that in contrast to COX activity, which was consistently inhibited by ketoprofen (as evaluated by PGE2 production), COX2 mRNA (PTGS2) expression levels were highly variable between the independent cultures. Incidentally, as for PTGS2, a large proportion of the tested markers used for monitoring the inflammatory response also displayed large standard deviations. Further analysis of the data revealed that the intestine of the 3 individuals showing a significant induction in the expression of PTGS2 were also exhibiting a typical protective anti-inflammatory gene expression profile similar to that reported by our group after epidermal growth factor treatment with increase in levels of anti-oxidative stress markers such as DUOX2, SOD2, NOS2 and the tight junction components CLDN1 and OCN (18,19). The intestines of these individuals were thus referred to as the “responders.” In contrast, in the intestine of the 3 non-responders, PTGS2 expression was not modulated and the tissues displayed a deleterious pro-inflammatory response gene expression signature similar to the one observed with indomethacin (8). It is noteworthy that this responder/non-responder phenomenon was not observed with indomethacin since, based on the lack of PTGS2 induction at the transcript level, all samples behaved as non-responders.

deleterious pro-inflammatory response gene expression signature similar to the one observed with indomethacin (8). It is noteworthy that this responder/non-responder phenomenon was not observed with indomethacin since, based on the lack of PTGS2 induction at the transcript level, all samples behaved as non-responders. A possible explanation for such a distinct mucosal response to ketoprofen versus indomethacin relies on a recent finding suggesting distinct mechanisms of action for these 2 NSAIDs. Indeed, as for the majority of NSAIDs, indomethacin exerts its anti-inflammatory properties by inhibiting the main arachidonic acid metabolism pathway, namely the production of prostaglandins, through the inhibition of COX activities (6,24,25). However, the inhibition of prostaglandin synthesis by indomethacin leads to higher arachidonic acid metabolism by the alternative 5-lipoxygenase (5-LO) pathway responsible for leukotriene production, including leukotriene B4 (LTB4), an important inflammatory mediator (6,24,26,27). In addition to increasing microvascular permeability and promoting neutrophil infiltration, LTB4 also stimulates the production of reactive oxygen species by the release of superoxide by neutrophils and hydrogen peroxide by macrophages (28,29) suggesting a potential source of inflammatory-related damage observed by indomethacin on the immature intestine. In contrast, ketoprofen appears to act as a dual COX and 5-LO inhibitor, blocking the synthesis of prostaglandins and leukotrienes (30), thus providing the basis for a possible explanation for the observed divergence between the 2 treatments on the immature intestine at least for a subset of individuals. Further work is nevertheless needed to better understand the mechanisms regulating this intestinal responder/non-responder phenomenon. No correlation with the sex of the individual or gestational age can explain the phenomenon. However, genetic susceptibilities associated to NSAID metabolism have been identified in relation to gastrointestinal bleeding (31), while identification of genetic variants in lipoxygenases and cyclooxygenases have been reported in association with the risk of digestive neoplasia (32–34) suggesting that the implication of such polymorphisms in the differential intestinal response to ketoprofen cannot be ruled out at this time.

stinal bleeding (31), while identification of genetic variants in lipoxygenases and cyclooxygenases have been reported in association with the risk of digestive neoplasia (32–34) suggesting that the implication of such polymorphisms in the differential intestinal response to ketoprofen cannot be ruled out at this time. Finally, the testing of ATB-352, a ketoprofen derivative bound to a 4-hydroxythiobenzamide group which allows the release of H2S (15), in an organ culture system enabled us to evaluate the direct effects of the compound on the immature intestinal mucosa. In vivo, several studies have shown that the administration of a H2S donor prevents the oxidative damage caused to the intestinal mucosa in addition to limiting certain elements of inflammation brought on by the administration of NSAIDs (11,13–15). For instance, H2S can inhibit leukocyte adhesion and migration to the sites of inflammation in addition to reducing the production of pro-inflammatory cytokines through a mechanism that appears to involve suppression of nuclear factor kB activity (12,15). Our data indicate that ATB-352 has similar effects as ketoprofen on the immature intestinal mucosa (including the responder/non-responder-type response) suggesting that the H2S release occurs rapidly in an open in vitro system. Incidentally, H2S is known to be released within minutes after administration in vivo (15). The limited influence on mucosal cells under these conditions was confirmed by the analysis of ICAM-1 and TNF, 2 specific H2S target genes (12). As recently suggested for nitric oxide, another potential protective molecule for the intestinal mucosa (19), future studies could include an adaptation of the organ culture system for analyzing the direct effects of specific gaseous forms such as H2S on the intestinal mucosa.

and TNF, 2 specific H2S target genes (12). As recently suggested for nitric oxide, another potential protective molecule for the intestinal mucosa (19), future studies could include an adaptation of the organ culture system for analyzing the direct effects of specific gaseous forms such as H2S on the intestinal mucosa. Taken together, these results show that ketoprofen induces fewer deleterious effects than the NSAID indomethacin on the immature small intestinal mucosa and that the difference is the result of a distinct individual intestinal response to ketoprofen, half of the samples exhibiting a protective response that included upregulation of anti-oxidative stress markers and tight junction components while the other half displayed a deleterious response comparable to the one observed with indomethacin. Future studies with a larger cohort should help to further document this phenomenon. The facts that in contrast to most other NSAIDs ketoprofen has been reported to inhibit both the COX and 5-LO pathways and the existence of polymorphisms susceptible to affecting these pathways are 2 aspects that would need to be considered. Supplementary Material Supplemental Digital Content This work was supported by grants from the Canadian Institutes of Health Research and the Star Foundation. J.F.B. was the recipient of the Canada Research Chair in Intestinal Physiopathology. J.L.W. is Chief Scientific Officer for Antibe Therapeutics Inc. J.L.W.'s research is supported by an operating grant from the Canadian Institutes of Health Research. The authors report no conflicts of interest.

Supplementary Material Supplemental Digital Content This work was supported by grants from the Canadian Institutes of Health Research and the Star Foundation. J.F.B. was the recipient of the Canada Research Chair in Intestinal Physiopathology. J.L.W. is Chief Scientific Officer for Antibe Therapeutics Inc. J.L.W.'s research is supported by an operating grant from the Canadian Institutes of Health Research. The authors report no conflicts of interest. TABLE 1 qPCR analysis of the effects of H2S released on the expression of representative markers for distinct mucosal homeostasis categories in ketoprofen responders and non-responders Responders Non-responders Genes Ratio ATB/Keto* P Ratio ATB/Keto* P Oxidoreductase activity CYP3A4 1.00 ± 0.44 0.9907 2.07 ± 2.19 0.4866 DUOX2 1.61 ± 0.34 0.0900 0.43 ± 0.46 0.1616 NOS2 1.60 ± 0.27 0.0596 0.33 ± 0.24 0.0418† SOD2 1.22 ± 0.16 0.1374 0.78 ± 0.29 0.3251 Oxidative phosphorylation ATP5G1 1.24 ± 0.40 0.4064 0.83 ± 0.10 0.0909 NDUFA9 1.21 ± 0.41 0.4644 0.93 ± 0.44 0.8155 Inflammatory response CXCL14 1.38 ± 0.07 0.0125† 1.21 ± 0.24 0.2636 TFF1 1.26 ± 0.13 0.0724 0.85 ± 0.53 0.6710 Permeability CLDN-1 1.13 ± 0.44 0.6663 0.73 ± 0.17 0.1134 OCLN 1.27 ± 0.61 0.5276 0.91 ± 0.43 0.7609 *Data are expressed as ratios of ATB-352 over ketoprofen (ATB/Keto)-treated immature intestinal explants in both ketoprofen responder and non-responder groups. Data are expressed as the mean ratio of 3 independent biological samples for each group. †Denote a statistical difference.

What Is KnownEnvironmental enteric dysfunction is heterogeneous and difficult to diagnose. Zinc absorption from foods is relatively low in children exhibiting characteristics of environmental enteric dysfunction. Endogenous fecal zinc losses have been reported to be positively associated with the lactulose to mannitol urine ratio. What Is NewIn young children with evidence of environmental enteric dysfunction, zinc absorption of an aqueous dose given in fasting state was positively associated with markers of inflammation and gut permeability. Endogenous fecal zinc losses decreased with increasing biomarkers of environmental enteric dysfunction. Impaired zinc absorption and low zinc intakes, rather than excessive endogenous zinc losses, may be greater factors predisposing young children with environmental enteric dysfunction to zinc deficiency. Impaired zinc homeostasis and status may synergistically interact with the complex condition environmental enteric dysfunction (EED). Zinc is critical for maintenance of a normal intestinal barrier, and it has a potent anti-inflammatory function. Both of these roles may be challenged by the pathology of EED. Zinc deficiency has thus been proposed to contribute to the poor growth and morbidity associated with EED (1–3). Zinc homeostasis is controlled through both absorption of dietary zinc and secretion and reabsorption of intestinal endogenous zinc excretion, the latter being most closely associated with chronic zinc status.

nc deficiency has thus been proposed to contribute to the poor growth and morbidity associated with EED (1–3). Zinc homeostasis is controlled through both absorption of dietary zinc and secretion and reabsorption of intestinal endogenous zinc excretion, the latter being most closely associated with chronic zinc status. Although historically EED has been difficult to diagnose and characterize (4), expert consensus holds that it is a chronic subclinical inflammatory condition of the proximal small intestine resulting in increased permeability and diffuse villous blunting resulting in malabsorption (4–6). With disruptions to the intestinal tract seen with EED, absorption of exogenous zinc and reabsorption of endogenous zinc may be impaired (1). Few investigations of zinc homeostasis have been undertaken in young children with EED. In 2 reports in 3- to 5-year old children in Malawi, daily absorbed zinc was apparently adequate and was related to dietary zinc intake as expected (7,8), but endogenous fecal zinc (EFZ) excretion was positively associated with the dual sugar urinary lactulose to mannitol ratio (L:M) and contributed to negative zinc balance (7). In contrast, we recently reported that absorption of dietary zinc was lower than predicted over a wide range of zinc intakes in Bangladeshi toddlers with evidence of EED (9). Reported high rates of zinc deficiency (10) and EED (11) in young Bangladeshi children support the need for a better understanding of the effects of marginal dietary zinc intake and EED on zinc homeostasis.

zinc was lower than predicted over a wide range of zinc intakes in Bangladeshi toddlers with evidence of EED (9). Reported high rates of zinc deficiency (10) and EED (11) in young Bangladeshi children support the need for a better understanding of the effects of marginal dietary zinc intake and EED on zinc homeostasis. The primary aims of the present study were 2-fold: to measure fractional absorption of zinc (FAZ) from a standardized aqueous zinc dose and to determine the habitual daily EFZ; both measurements were compared between groups of children with normal and abnormal L:M (9,11). A secondary aim was to examine relationships of biomarkers of inflammation and gut function with zinc homeostasis.

ional absorption of zinc (FAZ) from a standardized aqueous zinc dose and to determine the habitual daily EFZ; both measurements were compared between groups of children with normal and abnormal L:M (9,11). A secondary aim was to examine relationships of biomarkers of inflammation and gut function with zinc homeostasis. METHODS Study Design This experimental study aimed to measure FAZ of a standard dose of aqueous zinc administered in the postabsorptive state in 18–24-month-old children living in a peri-urban slum area of Dhaka, Bangladesh who were at risk for EED. In the same subjects, habitual daily losses of EFZ were measured while the children consumed a diet similar to their usual intake (Figure, Supplemental Digital Content 1, Study design, clinical protocol and associated outcomes). As previously described, the L:M was used to assign 20 participants per group to high L:M (≥0.09) and low L:M (<0.09) groups (9,11–13). Participants were admitted to the Clinical Trials Unit (CTU) at the International Center for Diarrheal Disease Research, Bangladesh (icddr,b) the evening before initiation of the isotope study procedures. On the day children arrived in the CTU, they also completed procedures for the lactulose to rhamnose urine ratio (L:R), obtained as an additional putative biomarker of EED. To measure FAZ and EFZ, procedures included oral administration of a zinc stable isotope-labeled aqueous dose of zinc; intravenous zinc isotope infusion; measurement of dietary intake; and collections of urine and stool (Figure, Supplemental Digital Content 1). Fractional absorption and EFZ were measured by previously described methods (14–16). Serum and fecal biomarkers of systemic and intestinal inflammation and gut function were measured to further characterize EED. The study was approved by Research Review Committee and Ethical Review Committee of icddr,b as well as the University of Colorado Multiple Institutional Review Board. The study was registered under ClinicalTrials.gov NCT02760095.

ic and intestinal inflammation and gut function were measured to further characterize EED. The study was approved by Research Review Committee and Ethical Review Committee of icddr,b as well as the University of Colorado Multiple Institutional Review Board. The study was registered under ClinicalTrials.gov NCT02760095. Participants Participants were recruited and completed the study procedures between October 2016 and April 2017 as previously described for a companion study in children from the same community (9). Inclusion criteria were willingness to comply with the study demands; length-for-age Z score (LAZ) between −1.25 and −3.0 (17); and hemoglobin (Hb) ≥8 g/dL. Exclusion criteria included chronic illness; diarrhea treated with zinc supplement within 2 weeks of screening; severe anemia; and severe acute malnutrition. After written informed consent, 4 days before the isotope studies, the child and guardian returned to the Mirpur health clinic for L:M testing. Field research assistants administered an oral solution of lactulose and mannitol to participants (11); urine was quantitatively collected over the subsequent 2 hours (18). The L:M was used as the primary EED screening tool, and children were grouped by high or low L:M.

rned to the Mirpur health clinic for L:M testing. Field research assistants administered an oral solution of lactulose and mannitol to participants (11); urine was quantitatively collected over the subsequent 2 hours (18). The L:M was used as the primary EED screening tool, and children were grouped by high or low L:M. Power and Sample Size Sample size was calculated to detect a 20% change in EFZ between high L:M and low L:M groups, assuming that the magnitude and variability of the EFZ data would be similar to that observed in a study in rural Malawi (8). Given standard deviation (SD) of 0.2 mg/day and specifying α = 0.5 and power of 0.8, it was determined that 17 children per group would be needed to detect a 0.2 mg/day difference in EFZ between groups. Description of Diets: Preparation and Administration Throughout the stay in the CTU, all meals were prepared on site in the icddr,b food laboratory using standard recipes developed for previous studies to reflect participants’ typical diets (19). For the metabolic period to determine EFZ, pre- and post-weights of all meals were measured, and duplicate diets were collected for analysis of daily total dietary zinc (TDZ) intake.

on site in the icddr,b food laboratory using standard recipes developed for previous studies to reflect participants’ typical diets (19). For the metabolic period to determine EFZ, pre- and post-weights of all meals were measured, and duplicate diets were collected for analysis of daily total dietary zinc (TDZ) intake. Preparation and Administration of Aqueous Zinc (Zn) Dose and Isotopes Doses of 67Zn and 70Zn stable isotopes (Trace Science International, Toronto, Canada) were prepared in the Pediatric Nutrition Laboratory at University of Colorado, Denver (UC) following standard methods (15,20). The isotopes were tested for sterility, fungal growth, and pyrogenicity at the beginning of study and at 6-month intervals. Approximately 3 mg of accurately measured aqueous zinc sulfate (American Regent, Shirley, NY) was added directly to accurately measured oral 70Zn isotope doses (∼250 μg) before transport to icddr,b. On Study Day 1, at 07:00–08:00 hours, the oral 70Zn-labelled aqueous zinc dose was administered in the fasted state, with no food or liquid (except water) consumed ≥ 2 hours before administration. Approximately 1 hour later, a sterile, accurately measured quantity of ∼400 μg 67Zn was intravenously administered (15,20). All dose losses from saliva and blood were collected and analyzed for isotope enrichment to adjust the administered doses.

o food or liquid (except water) consumed ≥ 2 hours before administration. Approximately 1 hour later, a sterile, accurately measured quantity of ∼400 μg 67Zn was intravenously administered (15,20). All dose losses from saliva and blood were collected and analyzed for isotope enrichment to adjust the administered doses. Sample Collections When participants were admitted to the CTU, spot urine and stool samples were collected as baseline specimens before isotope administration; stools were also used for fecal biomarkers of inflammation and permeability. Participants were also administered 1000 mg lactulose and 200 mg l-rhamnose in 10 mL of sterile water (Figure, Supplemental Digital Content 1). Urine was collected for 1 hour immediately following sugar administration for subsequent L:R determination (21). Three days after isotope administration (Study Days 4–8), complete 24-hour fecal collections were initiated for 4 consecutive days in the CTU. Precautions were taken to avoid cross-contamination of zinc isotope between urine and stool samples, including use of zinc-free urine bags (22). Two spot urine samples (approximately 30 mL at morning and evening) were collected daily. All urine, fecal, and dietary samples were stored at −20 °C until shipment to UC. On Study Day 8, approximately 5 mL of blood was collected; serum was separated after 30 minutes. Samples were stored at −80 °C at icddr,b until analysis for nutritional and inflammatory biomarkers.

Three days after isotope administration (Study Days 4–8), complete 24-hour fecal collections were initiated for 4 consecutive days in the CTU. Precautions were taken to avoid cross-contamination of zinc isotope between urine and stool samples, including use of zinc-free urine bags (22). Two spot urine samples (approximately 30 mL at morning and evening) were collected daily. All urine, fecal, and dietary samples were stored at −20 °C until shipment to UC. On Study Day 8, approximately 5 mL of blood was collected; serum was separated after 30 minutes. Samples were stored at −80 °C at icddr,b until analysis for nutritional and inflammatory biomarkers. Sample Analyses All total zinc and stable isotope enrichment analyses were completed at UC. Urine samples were digested and zinc was separated from other inorganic matter using chelation procedures (15,20). Zinc isotope ratios were measured by inductively coupled plasma mass spectrometry and converted to enrichment (15). Stool samples were ashed, then purified using column chromatography (22) for measurement of the intravenous isotope (67Zn) ratios. Serum inflammatory and nutritional biomarkers were analyzed by the icddr,b Nutrition Biochemistry Lab (9). Fecal inflammatory markers and pathogens, and the L:M and L:R were measured by the icddr,b Parasitology Lab (9,11).

Sample Analyses All total zinc and stable isotope enrichment analyses were completed at UC. Urine samples were digested and zinc was separated from other inorganic matter using chelation procedures (15,20). Zinc isotope ratios were measured by inductively coupled plasma mass spectrometry and converted to enrichment (15). Stool samples were ashed, then purified using column chromatography (22) for measurement of the intravenous isotope (67Zn) ratios. Serum inflammatory and nutritional biomarkers were analyzed by the icddr,b Nutrition Biochemistry Lab (9). Fecal inflammatory markers and pathogens, and the L:M and L:R were measured by the icddr,b Parasitology Lab (9,11). Data Calculations and Analyses FAZ was determined using the dual isotope tracer ratio method from urine enrichment of the oral and IV isotopes (14,15). As no data on absorption of a fasting aqueous zinc dose in young children were found in the literature, absorption values were predicted from data on absorption of a similar dose in adults based on 2 different assumptions: that absorption by children and adults will be proportional to small intestine length and to absorption of the same quantity of dietary zinc. In the first calculation an adult FAZ of 0.72 from a 3.2 mg aqueous dose (23) was multiplied by the ratio of estimated small intestine length in children with a height of 77 cm (24,25) and adults (26,27), ie, 0.72 × 360 cm/650 cm, yielding a predicted child FAZ value of 0.40. In the second calculation, the adult fasting FAZ of 0.72 was multiplied by the ratio of the FAZ of 3.2 mg of dietary zinc in children at 20 months (28) and in adults (29), ie, 0.72 × 0.25/0.53, generating a predicted FAZ of 0.34.

nd adults (26,27), ie, 0.72 × 360 cm/650 cm, yielding a predicted child FAZ value of 0.40. In the second calculation, the adult fasting FAZ of 0.72 was multiplied by the ratio of the FAZ of 3.2 mg of dietary zinc in children at 20 months (28) and in adults (29), ie, 0.72 × 0.25/0.53, generating a predicted FAZ of 0.34. Mean EFZ (mg/d) was calculated by dividing the endogenous zinc (total fecal zinc × fecal 67Zn enrichment) excreted by the product of average urine 67Zn enrichment and the number of collection days (8,16,22). For reference values, we calculated an EFZ of 0.65 mg/day using a published Eq. (30) and also considered a published EFZ value of 0.67 ± 0.23 mg/day observed in toddlers (22).

(total fecal zinc × fecal 67Zn enrichment) excreted by the product of average urine 67Zn enrichment and the number of collection days (8,16,22). For reference values, we calculated an EFZ of 0.65 mg/day using a published Eq. (30) and also considered a published EFZ value of 0.67 ± 0.23 mg/day observed in toddlers (22). Composite scores of EED were calculated from the alpha-1 antitrypsin, myeloperoxidase, and neopterin data (31). Summary statistics of FAZ, EFZ, and biomarkers of nutritional status and systemic and intestinal inflammation were calculated for low and high L:M groups, and distributions were examined. Comparisons between groups were performed with the Student's t-test or the Mann-Whitney nonparametric test. Associations of FAZ and EFZ with the biomarkers of inflammation and nutritional status were investigated using multiple linear regression analysis. Regression models were evaluated for adherence to regression assumptions and by quality criteria. Data analyses were performed using GraphPad Prism V.7.00 (GraphPad Software, La Jolla, CA) and R statistical software V.3.2.2 (32). A statistical difference or association was defined by a P value <0.05. Associations were also evaluated by their impact on the regression model selection criterion (corrected Akaike Information Criterion).

performed using GraphPad Prism V.7.00 (GraphPad Software, La Jolla, CA) and R statistical software V.3.2.2 (32). A statistical difference or association was defined by a P value <0.05. Associations were also evaluated by their impact on the regression model selection criterion (corrected Akaike Information Criterion). RESULTS Forty-six toddlers consented and enrolled in the study with complete sampling and analyses for 40 children. Six participants voluntarily exited the study because parents perceived study demands to be too great or they were no longer interested in study participation. There were no statistical differences for anthropometric and demographic comparisons between groups at baseline with the exception of the L:M, upon which the grouping was based (Table, Supplemental Digital Content 2, Baseline demographic and anthropometric data of Bangladeshi toddlers). The average age of participants was 20 months. Weight and length averaged at 9.2 ± 1.0 kg and 77.2 ± 2.2 cm, respectively, with the average LAZ −2.10 ± 0.43 and average hemoglobin concentration 10.6 ± 1.3 g/dL.

Table, Supplemental Digital Content 2, Baseline demographic and anthropometric data of Bangladeshi toddlers). The average age of participants was 20 months. Weight and length averaged at 9.2 ± 1.0 kg and 77.2 ± 2.2 cm, respectively, with the average LAZ −2.10 ± 0.43 and average hemoglobin concentration 10.6 ± 1.3 g/dL. Neither of the measures of zinc homeostasis differed statistically between low and high L:M groups (Table 1). The mean (±SD) ingested aqueous doses approximated the planned quantity and did not differ between the groups (3.2 ± 0.2 mg for both, P = 0.39). Average FAZs were similar to the predicted values (Table 1), but the variability of the FAZ data was 3 to 4 times larger than that observed for absorption of zinc consumed in meals in other children studied from this same community (9) (Fig. 1). Means for EFZ were also similar between the 2 L:M groups and were 10% to 15% higher than the predictions (Table 1). The TDZ intakes during the 4-day urine and fecal collection period for determination of EFZ were marginally higher for the low L:M group (P = 0.10) but were less than the estimated average requirement of 2.5 mg/day for this age (33) for both groups.

L:M groups and were 10% to 15% higher than the predictions (Table 1). The TDZ intakes during the 4-day urine and fecal collection period for determination of EFZ were marginally higher for the low L:M group (P = 0.10) but were less than the estimated average requirement of 2.5 mg/day for this age (33) for both groups. FIGURE 1 Fractional absorption of zinc versus quantity of ingested zinc shown for the aqueous fasting dose in the present study (circle symbols) and for a zinc-supplemented meal (x symbols) administered to a separate group of toddlers from the same Bangladeshi population (9). The greater variability in the fasting FAZ data is apparent, having a standard deviation of 0.19 compared with 0.06 for the FAZ from meals (taking into account the variation in FAZ with quantity of zinc in meal—dotted curve). ●—FAZ from aqueous zinc dose (high L:M, present study); ○—FAZ from aqueous zinc dose (low L:M, present study); x—FAZ from single meal supplemented with zinc. Data from (9). FAZ = fractional absorption of ingested zinc.

FAZ from meals (taking into account the variation in FAZ with quantity of zinc in meal—dotted curve). ●—FAZ from aqueous zinc dose (high L:M, present study); ○—FAZ from aqueous zinc dose (low L:M, present study); x—FAZ from single meal supplemented with zinc. Data from (9). FAZ = fractional absorption of ingested zinc. Biomarkers of nutritional status and systemic inflammation generally exhibited nonnormal distributions with positive skew, and thus were summarized by median and interquartile range (Table 2). Among these biomarkers, only soluble transferrin receptor (sTfR) and high sensitivity C-reactive protein (hsCRP) showed statistical differences between groups, with the medians for both being higher for the high L:M group. Evidence of iron deficiency was demonstrated by high sTfR and low mean ferritin, also most notable in the high L:M group. Elevated levels were prevalent among all fecal inflammatory biomarkers except myeloperoxidase; none differed by L:M group. The median EED score was slightly higher for the high L:M group: 5 versus 4 for the low L:M group, not a statistical difference.

sTfR and low mean ferritin, also most notable in the high L:M group. Elevated levels were prevalent among all fecal inflammatory biomarkers except myeloperoxidase; none differed by L:M group. The median EED score was slightly higher for the high L:M group: 5 versus 4 for the low L:M group, not a statistical difference. As the primary outcomes and most secondary outcomes were similar between groups, they were combined and regression analyses were performed on the complete dataset. Regression models demonstrated associations of FAZ and EFZ with multiple biomarkers (Table 3). Fractional absorption was directly associated with tumor necrosis factor-α, hsCRP, and L:R, and it was inversely associated with Hb. EFZ was directly associated with vitamin B12 and inversely associated with Hb, neopterin, myeloperoxidase, and L:R. Three dimensional graphs of the modeling of selected covariates exemplify the associations with FAZ and with EFZ (Figure, Supplemental Digital Content 3, 3-dimensional graphs of modeling of selected covariates of FAZ and EFZ). As dietary zinc intake and absorbed zinc are normally associated with EFZ (22,34), TDZ was added as a covariate to examine whether the associations of EFZ with the biomarkers were affected. Controlling for TDZ did not have an appreciable effect on the other relationships in any of the models. Given the expected elevations in biomarker values in the presence of EED, all associations with 1 exception were consistent with increasing FAZ and decreasing EFZ in EED. The sole exception was the inverse relation of EFZ to Hb. No associations of the L:M ratio with either FAZ or EFZ were observed, and the EED score was not found to be related to FAZ.

biomarker values in the presence of EED, all associations with 1 exception were consistent with increasing FAZ and decreasing EFZ in EED. The sole exception was the inverse relation of EFZ to Hb. No associations of the L:M ratio with either FAZ or EFZ were observed, and the EED score was not found to be related to FAZ. DISCUSSION The major findings from this study of young children with evidence of EED were that neither zinc absorption from a standard dose of aqueous zinc given in the fasting state, nor fecal excretion of endogenous zinc differed between L:M groups. Furthermore, systemic markers of inflammation were positively associated with zinc absorption, whereas markers of intestinal inflammation were inversely related to fecal excretion of endogenous zinc. The L:M did not clearly distinguish EED versus non-EED in that both groups exhibited evidence of systemic and intestinal inflammation, reflecting the heterogeneity of EED and present diagnostic challenges.

sorption, whereas markers of intestinal inflammation were inversely related to fecal excretion of endogenous zinc. The L:M did not clearly distinguish EED versus non-EED in that both groups exhibited evidence of systemic and intestinal inflammation, reflecting the heterogeneity of EED and present diagnostic challenges. Reduced zinc absorption and zinc deficiency have frequently been cited as potential links between EED and the growth failure commonly associated with this condition (1,3,35). Indeed, in a companion study, we reported markedly reduced absorption of zinc from meals in toddlers with evidence of EED (9). The positive association of FAZ with markers of systemic inflammation and intestinal permeability (L:R) observed here appears to contradict the earlier findings, but this is likely a consequence of the different study design. Although average absorption from the fasting dose was similar to that predicted from adult and child reference data, there was a great deal of variability in the FAZ data (with range exceeding an order of magnitude, Fig. 1) compared with either FAZ from meals in children from the same community or from a fasting dose FAZ in adults (23). This suggests that in EED, fasting FAZ of the aqueous dose is influenced to a greater extent by inflammation and permeability, possibly reflecting zinc uptake through nontransporter-mediated mechanisms. We surmise that this would be consistent with the impaired barrier function in the proximal small bowel documented in adults with EED (36,37). These findings may have implications for the efficacy of zinc administered as a liquid supplement or dispersible tablet compared with zinc fortification of foods.

ed mechanisms. We surmise that this would be consistent with the impaired barrier function in the proximal small bowel documented in adults with EED (36,37). These findings may have implications for the efficacy of zinc administered as a liquid supplement or dispersible tablet compared with zinc fortification of foods. Another factor observed to strongly influence FAZ was Hb concentration, which varied inversely with FAZ. Systemic inflammation, which was evident in many participants, is associated with low Hb, and our observation may reflect the degree of enteropathy. Contrary to our hypothesis that EFZ would be relatively elevated in these young children, our observed means for the 2 L:M groups were only 10% to 15% above the EFZ predicted by the equation proposed for the European Food Safety Authority (EFSA) (30). Furthermore, the results of our regression modeling indicated that EFZ decreased as biomarkers of intestinal inflammation and permeability increased.

hildren, our observed means for the 2 L:M groups were only 10% to 15% above the EFZ predicted by the equation proposed for the European Food Safety Authority (EFSA) (30). Furthermore, the results of our regression modeling indicated that EFZ decreased as biomarkers of intestinal inflammation and permeability increased. The most plausible explanation for the unexpected findings is chronic zinc deficiency, in which case conservation of endogenous zinc would be the normal homeostatic response. Although data are limited, studies in adults suggest that the site of reabsorption of endogenous zinc is the distal small bowel, and thus may be less impacted by EED compared with absorption of dietary zinc, which occurs primarily in the proximal small bowel where the characteristic pathology of EED is most prominent (3,4,38). The habitual dietary intake of the participants was well below recommended intakes (39), and our findings of impaired zinc absorption from virtually the same diets in similar aged children from the same community resulted in daily absorbed zinc well below estimated physiologic requirements (9). Survey data for children in the communities from which the participants were recruited also indicate high rates of zinc deficiency (10).

aired zinc absorption from virtually the same diets in similar aged children from the same community resulted in daily absorbed zinc well below estimated physiologic requirements (9). Survey data for children in the communities from which the participants were recruited also indicate high rates of zinc deficiency (10). Under conditions of normal gut health, EFZ generally varies with habitual daily intake and with absorbed zinc, and it is conserved when intake and absorption are low (22,38–40). In contrast, it has been proposed that young children with EED may experience excessive EFZ losses, representing a disruption of zinc homeostasis and predisposition to zinc deficiency (1,3). This is based in part on observations in other conditions of small bowel pathology with apparently excessive EFZ losses, eg, celiac disease (41). Although data on EFZ directly measured in children with EED are limited, our results contrast with the observations in Malawi of a direct relationship between EFZ and L:M (7).

his is based in part on observations in other conditions of small bowel pathology with apparently excessive EFZ losses, eg, celiac disease (41). Although data on EFZ directly measured in children with EED are limited, our results contrast with the observations in Malawi of a direct relationship between EFZ and L:M (7). The strengths of this study include the measurement of FAZ from a standard aqueous dose in fasting state, which allowed us to compare absorptive capacity without the confounding impact of food in the gut, potentially unmasking effects of gut permeability and inflammation. Measurement of EFZ over 4 days in young children with evidence of EED addressed a relative gap in the understanding of the effects of EED on zinc homeostasis. The application of multiple regression modeling to both homeostatic processes provided deeper insights into the influence of multiple covariates of inflammation and nutritional status than is possible from more traditional biomarker measurements.

ative gap in the understanding of the effects of EED on zinc homeostasis. The application of multiple regression modeling to both homeostatic processes provided deeper insights into the influence of multiple covariates of inflammation and nutritional status than is possible from more traditional biomarker measurements. We also acknowledge limitations of this study. Foremost of these was our inability to distinguish 2 distinct study groups with and without EED based on the L:M screening or subsequently by other potential EED markers. Thus, we did not have a true control group of children from the same environment and with similar diets. Our findings do, however, characterize relationships across a range of biomarkers. Determination of EFZ based on the isotope dilution method is dependent on complete fecal collections, and it is possible that the collections were incomplete, which would lead to underestimates of EFZ. However, the absolute EFZ quantity may be less important than the associations identified by the regression analyses.

Determination of EFZ based on the isotope dilution method is dependent on complete fecal collections, and it is possible that the collections were incomplete, which would lead to underestimates of EFZ. However, the absolute EFZ quantity may be less important than the associations identified by the regression analyses. The findings from this study confirm that the alterations in small bowel integrity associated with EED are likely to impact zinc homeostasis. Gut permeability and systemic inflammation were associated with enhanced absorption of an aqueous dose of elemental zinc, which contrasts to observations of impaired zinc absorption from food across a wide range of intakes in young children with EED (9). Distinct from others’ findings in older children, we did not observe dramatically increased EFZ losses in our population. We tentatively conclude that in the typical context of EED, the combination of low zinc intakes and compromised zinc absorption contribute more toward suboptimal zinc status than increased losses. Efforts to improve zinc status should thus focus on substantially increasing zinc intakes. Supplementary Material Supplemental Digital Content P.M. and J.M.L. shared first authorship. Source of Funding: This project was supported by the Bill and Melinda Gates Foundation OPP 1113134 and registered as ClinicalTrials.gov #NCT02760095 at https://clinicaltrials.gov/ct2/show/NCT02760095?term=eed&cntry=BD&rank=2. The authors report no conflicts of interest.

Supplementary Material Supplemental Digital Content P.M. and J.M.L. shared first authorship. Source of Funding: This project was supported by the Bill and Melinda Gates Foundation OPP 1113134 and registered as ClinicalTrials.gov #NCT02760095 at https://clinicaltrials.gov/ct2/show/NCT02760095?term=eed&cntry=BD&rank=2. The authors report no conflicts of interest. TABLE 1 Fractional absorption of ingested zinc from standardized aqueous zinc dose in fasting state; daily endogenous fecal zinc, and total dietary zinc by lactulose to mannitol ratio (L:M) group∗ during the 4-day metabolic period Estimated reference values† High L:M (n = 20) Low L:M (n = 20) P FAZ (0.34–0.4) 0.38 ± 0.19 0.31 ± 0.19 0.25 EFZ, mg/day (0.65–0.67) 0.73 ± 0.27 0.76 ± 0.20 0.67 TDZ, mg/day 1.78 ± 0.93 2.26 ± 0.85 0.10 Values presented as mean±SD. EFZ = endogenous fecal zinc; FAZ = fractional absorption of zinc; TDZ = total dietary zinc. *High L:M ≥0.09, low L:M <0.09. †Estimated reference values from previously reported data on adults and children (see text). TABLE 2 Biochemical data of Bangladeshi toddlers by lactulose to mannitol ratio (L:M) groups∗

Estimated reference values† High L:M (n = 20) Low L:M (n = 20) P FAZ (0.34–0.4) 0.38 ± 0.19 0.31 ± 0.19 0.25 EFZ, mg/day (0.65–0.67) 0.73 ± 0.27 0.76 ± 0.20 0.67 TDZ, mg/day 1.78 ± 0.93 2.26 ± 0.85 0.10 Values presented as mean±SD. EFZ = endogenous fecal zinc; FAZ = fractional absorption of zinc; TDZ = total dietary zinc. *High L:M ≥0.09, low L:M <0.09. †Estimated reference values from previously reported data on adults and children (see text). TABLE 2 Biochemical data of Bangladeshi toddlers by lactulose to mannitol ratio (L:M) groups∗ Normal range† high L:M (n = 18–20) low L:M (n = 17–20) P‡ Serum biomarkers of nutritional status Zinc, mg/L 0.65–1.18 0.83 (0.73, 1.35) 1.02 (0.74, 1.48) 0.61 Zinc, μmol/L 9.9–18.1 12.7 (11.2, 20.7) 15.6 (11.3, 22.6) Ferritin, ng/mL 20–200 15 (7.6, 22) 22 (9.6, 37) 0.27 Ferritin, pmol/L 44.9–449 33.7 (17.1, 49.4) 49.4 (21.6, 83.1) Soluble transferrin receptor, μg/mL 2.2–6.3 7.7 (5.8, 13) 5.0 (3.9, 8.7) 0.04 Soluble transferrin receptor, nmol/L 25.9–74.1 90.6 (68.2, 153) 58.8 (45.9, 102) Vitamin B12, pg/ML 264–1215 307 (222, 475) 364 (291, 496) 0.46 Vitamin B12, pmol/L 195–897 227 (164, 351) 269 (215, 366) Retinol, μg/dL ≥20 33 (26, 39) 33 (28, 38) 0.95 Retinol, μmol/L 0.70 1.15 (0.91, 1.36) 1.15 (0.98, 1.33) Serum biomarkers of inflammation α-1 acid glycoprotein, mg/dL 50–120 98 (84, 120) 83 (66, 108) 0.09 High sensitivity C-reactive protein, mg/L 0.1–2.8 2.0 (0.48, 5.8) 0.41 (0.14, 1.07) 0.005 Tumor necrosis factor-α, pg/mL <29.4 31 (29, 34) 29 (27, 32) 0.08 Urine and fecal biomarkers of intestinal function and inflammation Lactulose to rhamnose ratio n/a 0.16 (0.08, 0.23) 0.09 (0.04, 0.15) 0.10 % lactulose excretion, L:M n/a 0.20 (0.10, 0.49) 0.12 (0.06, 0.19) 0.04 % lactulose excretion, L:R n/a 0.20 (0.10, 0.38) 0.10 (0.0.04, 0.17) 0.02 Calprotectin, μg/g <50 115 (34, 337) 64 (33, 110) 0.21 Myeloperoxidase, ng/g <2000 1161 (421, 3383) 668 (144, 2635) 0.70 Neopterin, nmol/L <70 1026 (668, 1593) 954 (535, 1945) 0.67 α-1-antitrypsin, mg/g <0.27 0.40 (0.15, 0.79) 0.17 (0.12, 0.35) 0.14 Values are presented as median (interquartile values).

.04, 0.17) 0.02 Calprotectin, μg/g <50 115 (34, 337) 64 (33, 110) 0.21 Myeloperoxidase, ng/g <2000 1161 (421, 3383) 668 (144, 2635) 0.70 Neopterin, nmol/L <70 1026 (668, 1593) 954 (535, 1945) 0.67 α-1-antitrypsin, mg/g <0.27 0.40 (0.15, 0.79) 0.17 (0.12, 0.35) 0.14 Values are presented as median (interquartile values). *High L:M ≥ 0.09, low L:M < 0.09. †References for normal ranges provided in previous publication (9). ‡Mann-Whitney nonparametric test. TABLE 3 Regression models of fractional absorption of zinc and endogenous fecal zinc for Bangladeshi toddlers Covariate Parameter estimates P FAZ Model (R2 = 0.52) Hemoglobin −0.077 0.0006 Tumor necrosis factor-α 0.017 0.029 High sensitivity C-reactive protein 0.011 0.041 Lactulose to rhamnose ratio 0.43 0.035 EFZ Model (R2 = 0.66) Hemoglobin −0.095 0.0002 Vitamin B12 0.00049 0.0020 Myeloperoxidase* −0.000028 0.065 Neopterin* −0.000055 0.094 Lactulose to rhamnose ratio −0.54 0.019 FAZ −0.43 0.021 EFZ, endogenous fecal zinc; FAZ, fractional absorption of zinc. *If either myeloperoxidase or neopterin are removed from this model, P for the other drops to <0.004.

In many developed countries, many parents send their infants and young children to day care centres. It is well documented that day care attendance increases the risk of infections and that the risk of infections increases with the number of children in a group (1–3). Both respiratory and gastrointestinal infections (GIIs) are a major cause of disease in children attending day care centres (4). At least 89% of illness-related absence from a day care centre is the result of infectious diseases, of which 60% to 70% is because of respiratory infections. A study by Ponka et al (5) reported that the most common diagnoses were upper respiratory tract infections (URTIs) (46%), diarrhoea (17%), otitis media (12.9%), eye infection (4%), acute tonsillitis (3.2%), and bronchitis (3%). The first 3 years of life are important for the development of various organs, and essential for the development of the immune system. A specific mixture of short-chain galacto-oligosaccharides and long-chain fructo-oligosaccharides (scGOS/lcFOS; Immunofortis; Danone Research, Wageningen, the Netherlands) in a ratio of 9:1 has been shown to support immune development by significantly influencing the intestinal microbiota (6,7). Prebiotics are nondigestible food ingredients that beneficially affect the host by selectively stimulating the growth and/or activity of one or a limited number of bacteria in the colon that can improve host health (8). This mixture of scGOS/lcFOS mimics the molecular size distribution of human milk oligosaccharides and has been shown to positively affect the development of the immune system of infants; it has been demonstrated that the incidence and severity of URTIs were lower in healthy infants who received an infant formula with this mixture (9), and in another study, it was shown that the number of infectious episodes during the first 6 months of life was lower in infants receiving this mixture (10). The scGOS/lcFOS mixture has therefore been shown to be beneficial in infants; however, whether similar benefit can also be shown in young children needs further investigation.

r study, it was shown that the number of infectious episodes during the first 6 months of life was lower in infants receiving this mixture (10). The scGOS/lcFOS mixture has therefore been shown to be beneficial in infants; however, whether similar benefit can also be shown in young children needs further investigation. Another factor that may affect the immunological defence of healthy young children, and which is receiving increasing attention, is the long-chain polyunsaturated fatty acids (LCPUFAs) status. LCPUFAs are functionally the most important fatty acids for immune cells (11–15). The most important n-3 LCPUFAs are docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA). The most important n-6 LCPUFA is arachidonic acid (AA). Most regular Western diets have an imbalanced intake of n-6:n-3 LCPUFAs. This imbalance is associated with inflammatory/immune responses. A more balanced LCPUFAs profile in immune cells may result in a more balanced and optimal immune regulation, maturation, and response following a stimulus (16). Such a better-balanced LCPUFAs profile may be achieved by increasing dietary intake of n-3 LCPUFAs. This is particularly important in a period of rapid growth, such as during infancy or young childhood. Increasing intake of n-3 LCPUFAs will balance LCPUFAs ratios and may support an appropriate immune system response.

. Such a better-balanced LCPUFAs profile may be achieved by increasing dietary intake of n-3 LCPUFAs. This is particularly important in a period of rapid growth, such as during infancy or young childhood. Increasing intake of n-3 LCPUFAs will balance LCPUFAs ratios and may support an appropriate immune system response. Given the different mechanisms by which scGOS/lcFOS and LCPUFAs appear to exert their beneficial effect on the immune system, we hypothesised that a combination of scGOS/lcFOS and n-3 LCPUFAs may, in a complementary manner, be effective in reducing the risk of infections in young children. This study assessed the effect of a growing-up milk (GUM) with added scGOS/lcFOS/n-3 LCPUFAs on the occurrence of infections in young children attending day care centres in 5 countries in Europe and Asia. The potential effect of cow's milk in comparison with GUM in a small nonrandomised reference group was also assessed. METHODS Participants The study was a randomised double-blind controlled clinical study carried out in private practices, children's hospitals, university hospitals, or site management organisations (organisations that provide clinical trial–related services) located in 5 countries in Europe and Asia: Malaysia, the Netherlands, Poland, Portugal, and Thailand.

domised double-blind controlled clinical study carried out in private practices, children's hospitals, university hospitals, or site management organisations (organisations that provide clinical trial–related services) located in 5 countries in Europe and Asia: Malaysia, the Netherlands, Poland, Portugal, and Thailand. Healthy young children ages 11 to 29 months attending a day care centre at least 2 times per week were recruited into the study between September 2008 and May 2009 after informed written consent was obtained from the parents or legal guardians. Day care centres were defined as any private or governmental facility providing care or playgroups for at least 20 children. Exclusion criteria included atopic dermatitis according to the Hannifin criteria; disorders requiring a special diet; any relevant congenital abnormality, chromosomal disorder, or severe disease; preexisting pathology of severe respiratory or GI diseases; use of antiregurgitation, antireflux, or laxative medication at the time of inclusion; and use of immune modulators or of prescribed prophylactic antibiotics at the time of inclusion. The study was conducted following the principles of the Declaration of Helsinki and Good Clinical Practice guidelines. The protocol was approved by the ethics committees of the different institutes in the countries.

Healthy young children ages 11 to 29 months attending a day care centre at least 2 times per week were recruited into the study between September 2008 and May 2009 after informed written consent was obtained from the parents or legal guardians. Day care centres were defined as any private or governmental facility providing care or playgroups for at least 20 children. Exclusion criteria included atopic dermatitis according to the Hannifin criteria; disorders requiring a special diet; any relevant congenital abnormality, chromosomal disorder, or severe disease; preexisting pathology of severe respiratory or GI diseases; use of antiregurgitation, antireflux, or laxative medication at the time of inclusion; and use of immune modulators or of prescribed prophylactic antibiotics at the time of inclusion. The study was conducted following the principles of the Declaration of Helsinki and Good Clinical Practice guidelines. The protocol was approved by the ethics committees of the different institutes in the countries. Study Design/Intervention Study products were either the active study product, GUM with the addition of 1.2 g/100 mL of scGOS/ lcFOS (9:1) (Immunofortis) and 19.2 mg/100 mL of n-3 LCPUFAs (EPA + DHA, 4:6), the control study product, GUM without scGOS/lcFOS/n-3 LCPUFAs (Danone Research). GUM is specially formulated milk for young children ages 1 to 3 years, enriched with key nutrients such as vitamins A, C, and D, iron, and calcium. Both study products were packed in identical tins; they were of the same colour, weight, smell, and taste. Both the research staff and parents and subjects were unaware of the real nature of the product. The unblinding procedure was performed after the study was completed and after the statistical analyses were finalised.

h study products were packed in identical tins; they were of the same colour, weight, smell, and taste. Both the research staff and parents and subjects were unaware of the real nature of the product. The unblinding procedure was performed after the study was completed and after the statistical analyses were finalised. The study consisted of a 4-week run-in period and a 52-week intervention period. Subjects eligible for participation started with the run-in period, in which control product was consumed. The run-in period was needed to get the children used to the product and the parents used to the completion of the diaries. Subjects who had successfully completed the run-in period were randomly allocated to receive either the active study product or the control study product in a double-blind period of 52 weeks. To ensure a gradual transition during the first 2 weeks of this period, the run-in product was mixed with the active/control study product following a schedule provided by the centres. The requested intake of the study product was 400 to 750 mL daily for both the run-in period and the 52-week intervention period. This wide range was chosen because the different countries involved in the study had a wide variety of dosage recommendations. Every 2 to 4 weeks, there was a contact session with the parents, either by visit or by telephone, in which study diaries were discussed. A final contact session took place 2 weeks after intervention, at 54 weeks.

chosen because the different countries involved in the study had a wide variety of dosage recommendations. Every 2 to 4 weeks, there was a contact session with the parents, either by visit or by telephone, in which study diaries were discussed. A final contact session took place 2 weeks after intervention, at 54 weeks. In addition, there was a reference group of subjects receiving cow's milk already before the start of the study and which followed the same procedures as the intervention groups; however, this group was not randomised and consumed regular cow's milk for 52 weeks. Subjects from the reference group were only recruited in the Netherlands. Data Collection Background information was collected on the family, their environment, the child's nutritional habits, and illnesses. During the study, parents daily recorded the intake of study product as well as, in a symptom diary, any respiratory symptoms (cough, fever, blocked or runny nose, sore throat, wheeze, and/or ear discharge) and GI symptoms (vomiting or diarrhoea). They also reported absence from the day care centre, absence of parents from work, contacts with a physician, doctor's diagnoses, and prescription of medication. If parents suspected that their child had fever, body temperature of the subject was measured and recorded, with thermometers provided to all of the children.

They also reported absence from the day care centre, absence of parents from work, contacts with a physician, doctor's diagnoses, and prescription of medication. If parents suspected that their child had fever, body temperature of the subject was measured and recorded, with thermometers provided to all of the children. Furthermore, parents completed 3 times a 3-day diary, with questions about GI symptoms and stool characteristics, at the start of the intervention period, at 26 weeks, and at 52 weeks. The diary covered stool frequency and stool consistency and severity of GI tract characteristics (diarrhoea, constipation, and flatulence) with a 4-point scale (none, mild, moderate, severe). Sample Size and Randomisation Sample size was estimated based on the incidence of respiratory tract infections. Based on a previous study on the incidence of respiratory tract infections (17,18), the study was designed to have sufficient power (80%) to detect a 20% difference between the groups with a 0.05 significance level based on a 2-tailed test. The number of children necessary in each group was 300 subjects, including an estimated dropout rate of 25%. Each child, except for the children in the reference group, was randomly allocated to either the active study product or the control study product. The randomised subjects were distributed to the 2 study groups based on a randomisation code, developed using a computer random number generator. A block size of 6 was used, and stratification was applied for investigator site.

oup, was randomly allocated to either the active study product or the control study product. The randomised subjects were distributed to the 2 study groups based on a randomisation code, developed using a computer random number generator. A block size of 6 was used, and stratification was applied for investigator site. Outcome Measures The primary outcome was the number of episodes of URTI or GIIs based on a combination of subject's illness symptoms reported by the parents during the intervention period. URTI was defined as any combination of at least 2 of the following symptoms: cough, fever, blocked or runny nose, sore throat, wheeze, and/or ear discharge. GII was defined as the occurrence of vomiting or diarrhoea or both for at least 1 day, with or without fever. See Table 1 for the definition of symptoms. An episode of URTI and/or GII was defined as having ≥1 symptoms for at least 1 day. A child was considered as recovered from an episode when the subject showed no symptoms of the reported infection and had a normal level of activity. A new episode was considered to have occurred if the symptom(s) occurred at least 3 days after the stated time of recovery from a previous episode. The symptoms that were reported by the parents were always checked by the investigator. He or she could indicate whether he or she found the symptoms to be an infection.

An episode of URTI and/or GII was defined as having ≥1 symptoms for at least 1 day. A child was considered as recovered from an episode when the subject showed no symptoms of the reported infection and had a normal level of activity. A new episode was considered to have occurred if the symptom(s) occurred at least 3 days after the stated time of recovery from a previous episode. The symptoms that were reported by the parents were always checked by the investigator. He or she could indicate whether he or she found the symptoms to be an infection. Secondary outcomes were total number and duration of episodes of URTI and/or GII; parents’ absence from work because of illness of the child; number, duration, and season of subjects’ absence from day care centre; and number and type of all infections diagnosed by a physician or investigator.

The symptoms that were reported by the parents were always checked by the investigator. He or she could indicate whether he or she found the symptoms to be an infection. Secondary outcomes were total number and duration of episodes of URTI and/or GII; parents’ absence from work because of illness of the child; number, duration, and season of subjects’ absence from day care centre; and number and type of all infections diagnosed by a physician or investigator. Statistical Analysis Descriptive statistics were used to describe the basic features according to the age, sex, food intake, and family characteristics. Results on the primary parameter were obtained from both a logistic regression model and a zero-inflated negative binomial regression model. The latter model combines 2 separate parts: a (binomial) part, whether or not a subject would get an infection; and a (negative-binomial) part fitting the counts for infections for the subjects. All analyses were based on the all-subjects randomised group: all of the subjects who had been randomised and to whose parents the study product had been dispensed. For safety analysis, such as growth, height, and tolerance, the safety analysis population was used consisting of all children except the screen failure subjects for the intervention period. In an analysis of the normally distributed continuous data, a t test was performed, whereas a Wilcoxon rank-sum test was used in analysis of non-normality continuous data. In addition, a χ2 test (Fisher exact test in the case of <5 expected cell counts) was performed in analysis of nominal data. Furthermore, Kaplan-Meier (KM) curves together with corresponding log-rank tests were conducted in analysis of time to event type of data. All statistical analyses were performed with SAS (SAS Enterprise Guide 4.1 or higher) for Windows (SAS Institute Inc, Cary, NC).

l counts) was performed in analysis of nominal data. Furthermore, Kaplan-Meier (KM) curves together with corresponding log-rank tests were conducted in analysis of time to event type of data. All statistical analyses were performed with SAS (SAS Enterprise Guide 4.1 or higher) for Windows (SAS Institute Inc, Cary, NC). RESULTS Of the 907 subjects screened and enrolled in the run-in period, 767 were randomised: 388 received the active study product, and 379 received the control study product. In total, there were 697 completers: Malaysia 135, the Netherlands 199, Poland 126, Portugal 70, and Thailand 167. Thirty-seven subjects were included in the reference group, which was not randomised (Fig. 1). There were more subjects randomised than was actually calculated as sample size. This was because of logistical reasons, timing differences of start of inclusion periods at different locations and the run-in period, and the fact that it was uncertain how many subjects would remain in the intervention period. FIGURE 1 Flowchart of the study. ASR = all subjects randomised. No statistically significant differences were observed between the 2 groups in regard to age, sex, length, and weight at birth and any of the other baseline characteristics analysed (Table 2). Also, no differences were seen in the mother's and father's characteristics, with regard to the highest level of education or the professional status at baseline (Tables 3 and 4).

ed between the 2 groups in regard to age, sex, length, and weight at birth and any of the other baseline characteristics analysed (Table 2). Also, no differences were seen in the mother's and father's characteristics, with regard to the highest level of education or the professional status at baseline (Tables 3 and 4). Occurrence of Infections Logistic regression analysis was performed in which the chance of having ≥1 episodes was modelled against the chance of having 0 episodes; this yielded a P value of 0.03 (299/388 [77%] children active group vs 313/379 [83%] children control group) and a relative risk of 0.93 with a 95% confidence interval 0.87–1.00 and number needed to treat (NNT) 17.

performed in which the chance of having ≥1 episodes was modelled against the chance of having 0 episodes; this yielded a P value of 0.03 (299/388 [77%] children active group vs 313/379 [83%] children control group) and a relative risk of 0.93 with a 95% confidence interval 0.87–1.00 and number needed to treat (NNT) 17. The logistic regression model was based on a distinction between the 2 categories: did infections occur, yes or no; however, the data contained information on the number of reported infection episodes as well (Fig. 2). To analyse all of the available information in the data, a Poisson regression model is needed that can address the fact that a considerable proportion of children had the value 0 (ie, no infections, Fig. 3). The zero-inflated negative binomial regression model is such a model and showed a P value of 0.07 for the overall effect on occurrence of infections. The binomial part of the model analyses the occurrence of zero infections and showed a significant difference (89/388 children active group vs 66/379 children control group; P = 0.03; Fig. 3), whereas the negative binomial part of the model did not show a significant difference. This suggests that the intervention increased the likelihood of having no infections. FIGURE 2 Descriptive statistics for the number of reported episodes during intervention period per group.

The logistic regression model was based on a distinction between the 2 categories: did infections occur, yes or no; however, the data contained information on the number of reported infection episodes as well (Fig. 2). To analyse all of the available information in the data, a Poisson regression model is needed that can address the fact that a considerable proportion of children had the value 0 (ie, no infections, Fig. 3). The zero-inflated negative binomial regression model is such a model and showed a P value of 0.07 for the overall effect on occurrence of infections. The binomial part of the model analyses the occurrence of zero infections and showed a significant difference (89/388 children active group vs 66/379 children control group; P = 0.03; Fig. 3), whereas the negative binomial part of the model did not show a significant difference. This suggests that the intervention increased the likelihood of having no infections. FIGURE 2 Descriptive statistics for the number of reported episodes during intervention period per group. FIGURE 3 Number of subjects with number of infections for active and control group, based on the statistical model. There are more children with no reported episodes in active group (A). There are fewer children with small numbers of reported episodes in active group (1–5 number of reported infectious episodes) (B). No difference between groups in percentages of children with high number of reported episodes (≥6 number of reported infectious episodes) (C).

with no reported episodes in active group (A). There are fewer children with small numbers of reported episodes in active group (1–5 number of reported infectious episodes) (B). No difference between groups in percentages of children with high number of reported episodes (≥6 number of reported infectious episodes) (C). A post hoc analysis that focused on only those parent-reported symptoms that were confirmed by an investigator (based on their experience) showed a significant effect (P = 0.004 and relative risk 0.89 with a 95% confidence interval 0.82–0.97, NNT 12, Table 5) on the number of infectious episodes. Also, here this was mostly driven by the higher statistically significant effect on having no infectious episodes (P = 0.003). Furthermore, post hoc analyses were performed to check for differences when URTI and GII were analysed separately with or without fever. No effect was observed (Table 6). Secondary outcomes such as any type of medication use (which consisted of oral antibiotics, paracetamol-acetaminophen, cough mixture medication, nose drops/spray, ear drops, oral rehydration solution, homeopathic medication, and any other medication) and duration of absence from day care showed no significant differences between the 2 study groups (Table 7).

edication use (which consisted of oral antibiotics, paracetamol-acetaminophen, cough mixture medication, nose drops/spray, ear drops, oral rehydration solution, homeopathic medication, and any other medication) and duration of absence from day care showed no significant differences between the 2 study groups (Table 7). In total, 2217 adverse events, including 78 serious adverse events, were reported. The majority of adverse events (∼49%) were related to the respiratory system (eg, cough, fever, blocked or runny nose, sore throat, wheeze, ear discharge). A total of 29 adverse events were assessed by the investigator as possibly or probably related to the study product during the run-in or intervention period. The related adverse events were mild GI symptoms and occurred equally between the 2 groups. None of the reported serious adverse events were assessed as related to study product and none were unexpected. Comparison of GUM Groups With Cow's Milk Group In the study, the small nonrandomised cow's milk reference group, coming from the Netherlands only, was compared with the total active and control groups. Both study groups had fewer parent-reported episodes than the cow's milk group (cow's milk vs active: P = 0.008; and cow's milk vs control: P = 0.004; Table 8).

roup In the study, the small nonrandomised cow's milk reference group, coming from the Netherlands only, was compared with the total active and control groups. Both study groups had fewer parent-reported episodes than the cow's milk group (cow's milk vs active: P = 0.008; and cow's milk vs control: P = 0.004; Table 8). Time to occurrence of first reported episode was significantly different (P = 0.04) between active and cow's milk group in favour of the active group with a survival comparison with log-rank test. A Kaplan-Meier curve (Fig. 4) showed the same result when looking at deviations between the active versus the cow's milk group. FIGURE 4 The Kaplan-Meier curve. Active, control, and cow's milk group: time to occurrence of first reported episode.

Time to occurrence of first reported episode was significantly different (P = 0.04) between active and cow's milk group in favour of the active group with a survival comparison with log-rank test. A Kaplan-Meier curve (Fig. 4) showed the same result when looking at deviations between the active versus the cow's milk group. FIGURE 4 The Kaplan-Meier curve. Active, control, and cow's milk group: time to occurrence of first reported episode. GUM Intake, Weight, and Height The mean average intake of GUM products was 528 mL/day, and no significant differences were found between the 2 study groups regarding daily intake of the study products (active: 529 ± 128 vs control: 527 ± 139 mL/day; Table 9). The daily intake of scGOS/lcFOS (9:1) was at least 3.6 g/day, and for DHA + EPA, this was at least 57.6 mg/day. Subjects in the active arm weighed statistically significantly less at baseline and were smaller when compared with the control arm (Table 10); however, these differences are seen as normal variability of weight and height. It is not expected that this difference could influence the study outcome. The difference in weight was maintained throughout the intervention period; however, for body mass index and for the change in weight and height, no statistically significant difference between the 2 study arms was observed.

f weight and height. It is not expected that this difference could influence the study outcome. The difference in weight was maintained throughout the intervention period; however, for body mass index and for the change in weight and height, no statistically significant difference between the 2 study arms was observed. The weight change from baseline at week 52 of children from the reference group compared with the active or control group did not differ significantly; however, the length change and BMI change did differ significantly (except for length change when reference was compared with control). DISCUSSION This randomised double-blind controlled clinical study is the first to show a reduced risk of infections in young children following consumption of GUM supplemented with scGOS/lcFOS/n-3 LCPUFAs. It is remarkable that although the GUM is only a relatively small part of a young child's diet, it can apparently influence the risk of infections. The borderline significant effect we observed is mainly derived from the decreased risk of developing an infection. This does not mean that the child does not encounter any viral or bacterial microorganism, but that the child does not experience symptoms of the infection. This implies that the child is better equipped against infections.

ificant effect we observed is mainly derived from the decreased risk of developing an infection. This does not mean that the child does not encounter any viral or bacterial microorganism, but that the child does not experience symptoms of the infection. This implies that the child is better equipped against infections. This is the first study conducted in young children to demonstrate that the use of a prebiotic mixture can influence the risk of infections. Similar studies have evaluated the effect of consumption of milk with probiotics on incidence of infections, with different results. In a randomised double-blind placebo-controlled study (n = 281, age range 1–7 years), administration of Lactobacillus GG (LGG) resulted in a significantly reduced risk of URTI (19). Another study in which LGG was administered showed a reduction in the number of children with respiratory infections (20). Neither study showed any effect on GII. A more recent study with LGG could not show an effect on respiratory illness in children (21), but the daily dose of LGG in this study was lower compared with the other studies. In contrast, results from another study with other bacteria, Bifidobacterium lactis or Lactobacillus reuteri, did not show a beneficial probiotic effect on the duration of respiratory illness (18). In contrast to the above reports, in our study, we investigated the effect of GUM with prebiotics and LCPUFAs, not with the addition of probiotics. Our results indicate that the specific scGOS/lcFOS prebiotic mixture and LCPUFAs can have an effect only on the risk of infections. Moreover, the dropout rate was extremely low (only 10%) in this study, and also the compliance of product intake over both groups was better compared with other published studies. In our study, we evaluated the combination of URTI and GII together and not separately.

s can have an effect only on the risk of infections. Moreover, the dropout rate was extremely low (only 10%) in this study, and also the compliance of product intake over both groups was better compared with other published studies. In our study, we evaluated the combination of URTI and GII together and not separately. As for the mechanism responsible for the beneficial role of GUM supplemented with scGOS/lcFOS/n-3 LCPUFAs, studies in infants have documented that the mixture of scGOS/lcFOS supports immune development by significantly improving the intestinal microbiota. A healthy microbiota in turn is important for the development of the immune system. The positive effects of scGOS/lcFOS on the immune response have also been observed in experimental animal studies (22,23). Furthermore, several LCPUFA supplementation studies in preterm infants (24), term infants (25,26), and school-age children (27) suggest normalisation and improved immune function following dietary LCPUFAs supplementation. Supplementation of LCPUFAs to infant formula resulted in cytokine and T-cell population profiles that were more similar to those of breast-fed infants compared with those of unsupplemented formula-fed infants, and resulted in reduced incidence of respiratory illnesses (26). LCPUFA supplementation in children 5 to 7 years of age appeared to result in enhanced initiation of the immune response to B-cell mitogen and a reduced inflammatory response (27). Finally, dietary DHA enrichment reduced the incidence and severity of illnesses in Thai schoolchildren (28,29), and these effects were accompanied by a shift of cytokine profile toward a less inflammatory profile. Together, these studies support the concept that a dietary source of n-3 LCPUFAs can improve immune function in children, and thereby support disease resistance. Few studies have been done in which the effect of prebiotics on infections was examined. We recently reported the results of a study with healthy young children indicating that the use of a GUM with added scGOS/lcFOS/n-3 LCPUFAs leads to a higher level of bifidobacteria during the intake period compared with the control group (29a). The mechanisms of prebiotics and n-3 LCPUFAs apparently operated in a complementary way on infections in this study.

y with healthy young children indicating that the use of a GUM with added scGOS/lcFOS/n-3 LCPUFAs leads to a higher level of bifidobacteria during the intake period compared with the control group (29a). The mechanisms of prebiotics and n-3 LCPUFAs apparently operated in a complementary way on infections in this study. As for the clinical relevance of the results of this study, there was only a 6% lower risk of infectious episodes when using GUM supplemented with scGOS/lcFOS/n-3 LCPUFAs and an 8% lower risk when the data of the investigator-confirmed episodes analysis are used. The NNT to prevent one child from having an infectious episode is 17, and when the data of the investigator-confirmed episodes analysis are used, this decreases to 12. Knowing that the study product is a food product, which only accounts for a small part of a child's diet, the effect on NNT is large. Furthermore, this may have a positive effect on health economics. What the exact effect will be on health care cost reduction remains to be investigated.

are used, this decreases to 12. Knowing that the study product is a food product, which only accounts for a small part of a child's diet, the effect on NNT is large. Furthermore, this may have a positive effect on health economics. What the exact effect will be on health care cost reduction remains to be investigated. We are aware of several limitations to our study. First, the symptoms were based on parental reporting, but we did not actually determine whether the children with symptoms had infections by measuring immune responses. Second, a low frequency of infections was observed, which could be because of underreporting. Underreporting could also be a result of the parental reporting of symptoms. Study results were only measured as parent-reported symptoms, to limit the burden for the child. Third, because there are no clear definitions, the definitions for the respiratory and/or GI symptoms used throughout this study are not exactly the same as in other published studies, which makes it sometimes difficult to compare the results. In addition, there are no validated questionnaires available for the monitoring of respiratory symptoms, which would have made the results more consistent with other studies.

oms used throughout this study are not exactly the same as in other published studies, which makes it sometimes difficult to compare the results. In addition, there are no validated questionnaires available for the monitoring of respiratory symptoms, which would have made the results more consistent with other studies. The comparison of the cow's milk group with the GUM groups showed that more infectious episodes were observed in the cow's milk group. We realise that the cow's milk group was not randomised or powered. Therefore, the results obtained should be interpreted with caution; however, the results suggest that GUM as such has a beneficial effect on the occurrence of infections compared with cow's milk. This could be because of the fact that GUM is nutritionally enriched compared with cow's milk. GUM contains relevant vitamins and minerals required for that age range, which may contribute to a better support of growth and development, and may indirectly positively affect the immune system. Iron and vitamin D, for instance, are 2 nutrients included in GUM for which intake is often seen as insufficient in young children (30,31) (nutritional surveys in several countries, systematic surveys of nutrition and health status of young children and women of child-bearing age based on literature review and expert opinions conducted in several countries). An explanation for the significant difference in height change between reference group and active group could be that the children from the reference group were from the Netherlands only. It is well known that people from northern Europe, and especially the Netherlands, are tall compared with people from other European countries and Asia. The fact that there was no significant difference in weight change was to be expected because the ratio of height and weight should not differ. Further studies in which cow's milk is compared with GUM for effects on the occurrence of infections are needed to obtain conclusive results.

other European countries and Asia. The fact that there was no significant difference in weight change was to be expected because the ratio of height and weight should not differ. Further studies in which cow's milk is compared with GUM for effects on the occurrence of infections are needed to obtain conclusive results. In conclusion, this is the first study in children to show a reduced risk of infection following consumption of GUM supplemented with scGOS/lcFOS/n-3 LCPUFAs. Although the primary outcome was of borderline statistical significance, it is supported by the post hoc significantly decreased risk of infectious symptoms with investigator-confirmed episodes; therefore, this study supports the use of GUM supplemented with scGOS/lcFOS/n-3 LCPUFAs to reduce the risk of infection in children attending day care centres. Acknowledgments The authors thank the parents of the children in the GIANT study for their contribution and the site staff from all of the participating sites from the GIANT study group for their valuable assistance in the study. Drs Chatchatee and Lee participated equally in this study. www.trialregister.nl registration no.: NTR1451. Y.Y., B.S., and P.L. are employees of Nutricia Research. The other authors report no conflicts of interest. TABLE 1 Definition of symptoms

Acknowledgments The authors thank the parents of the children in the GIANT study for their contribution and the site staff from all of the participating sites from the GIANT study group for their valuable assistance in the study. Drs Chatchatee and Lee participated equally in this study. www.trialregister.nl registration no.: NTR1451. Y.Y., B.S., and P.L. are employees of Nutricia Research. The other authors report no conflicts of interest. TABLE 1 Definition of symptoms Symptom Definition Cough Sudden expulsion of air from the lungs with a noise that clears the air passages Blocked or runny nose Nasal discharge of any mucus-like material coming out of the nose (rhinorrhea) Sore throat Discomfort, pain, or scratchiness in the throat Wheeze Breathing with an abnormal (eg, husky, whistling) sound mostly caused by difficult respiration from narrowed airway passages Ear discharge Drainage of blood, ear wax, pus, or fluid from the ear Fever A rectal temperature of at least 38°C for at least 2 measurements with an interval of at least 2 hours Diarrhoea ≥3 loose or liquid stools per day Vomiting Forcing the contents of the stomach up through the oesophagus and out of the mouth TABLE 2 Demographic data and subjects’ characteristics at baseline in the study groups

tal temperature of at least 38°C for at least 2 measurements with an interval of at least 2 hours Diarrhoea ≥3 loose or liquid stools per day Vomiting Forcing the contents of the stomach up through the oesophagus and out of the mouth TABLE 2 Demographic data and subjects’ characteristics at baseline in the study groups Active group, N = 388 Control group, N = 379 P Sex* Female 175 (45.1) 157 (41.4) 0.304 Male 213 (54.9) 222 (58.6) Age at screening, mo† 19 (14–24) 19 (15–24) 0.283 Ethnicity* Asian 165 (42.5) 159 (42.0) 0.658 Black 4 (1.0) 6 (1.6) White 202 (52.1) 189 (49.9) Combination 16 (4.1) 23 (6.1) Other 1 (0.3) 2 (0.5) Season of birth* Autumn 95 (24.5) 87 (23.0) 0.874 Spring 88 (22.7) 94 (24.8) Summer 70 (18.0) 71 (18.7) Winter 135 (34.8) 127 (33.5) Type of delivery* Caesarean 109 (28.1) 106 (28.0) 0.969 Vaginal 279 (71.9) 273 (72.0) Gestational age, wk† 39 (38–40) 39 (38–40) 0.890 Length at birth, cm† 50 (48–53) 51 (49–53) 0.241 Weight at birth, kg† 3 (3–4) 3 (3–4) 0.546 Estimated total number of infection episodes in the last 6 mo‡ 3.4 (2.17) 3.4 (2.17) 0.942 Start age of breast-feeding, mo† 0 (0–0) 0 (0–0) 0.606 End age of breast-feeding, mo† 5 (2–9) 5 (3–9) 0.230 Start age of formula feeding, mo† 3 (1–6) 4 (1–6) 0.381 End age of formula feeding, mo† 18 (14–23) 18 (14–23) 0.955 Start age of cow's milk, mo† 12 (12–15) 12 (12–13) 0.972 End age of cow's milk† 30 (27–37) 33 (29–38) 0.082 Start age of weaning food, mo† 6 (4–6) 6 (4–6) 0.426 Living area of the family* Rural 37 (9.5) 36 (9.5) 0.897 Suburban 136 (35.1) 127 (33.5) Urban 215 (55.4) 216 (57.0) No. rooms in the family house (excl. kitchens, bathrooms, and toilets)* 1–3 183 (47.8) 192 (51.3) 0.345 4–6 190 (49.6) 168 (44.9) >6 10 (2.6) 14 (3.7) Missing 5 5 No. people living at the family house (excl. participating child)* 0 0 (0.0) 1 (0.3) 1–3 225 (58.7) 236 (63.1) 0.410 4–6 138 (36.0) 117 (31.3) >6 20 (5.2) 20 (5.3) Missing 5 5 No. siblings of the child (excl. participating child)* 0 136 (35.2) 162 (42.7) 0.571 1 166 (43.0) 145 (38.3) 2 55 (14.2) 45 (11.9) 3 21 (5.4) 20 (5.3) 4 4 (1.0) 3 (0.8) 5 3 (0.8) 3 (0.8) >5 1 (0.3) 1 (0.3) Missing 2 0 Smoking during pregnancy of child* No 364 (95.3) 358 (95.2) 0.961 Yes 18 (4.7) 18 (4.8) Missing 6 3 Age of father, y† 33 (30–37) 33 (30–37) 0.868 Age of mother, y† 31 (27–34) 31 (27–34) 0.466 Age of siblings, y† 5 (2–8) 4 (2–7) 0.328 *N (%), P from the χ2 test.

) 3 (0.8) 5 3 (0.8) 3 (0.8) >5 1 (0.3) 1 (0.3) Missing 2 0 Smoking during pregnancy of child* No 364 (95.3) 358 (95.2) 0.961 Yes 18 (4.7) 18 (4.8) Missing 6 3 Age of father, y† 33 (30–37) 33 (30–37) 0.868 Age of mother, y† 31 (27–34) 31 (27–34) 0.466 Age of siblings, y† 5 (2–8) 4 (2–7) 0.328 *N (%), P from the χ2 test. †Median (Q1–Q3), P from the Wilcoxon rank-sum test. TABLE 3 Demographic data and mothers’ characteristics at baseline in the study groups Active arm, N = 388 Control arm, N = 379 P Highest level of education at baseline* Basic minimum education 47 (12.2) 45 (11.9) 0.821 Additional education but not tertiary level 151 (39.2) 156 (41.2) Tertiary education 184 (47.8) 173 (45.6) Unknown 3 (0.8) 5 (1.3) Missing 3 0 Professional status at baseline* Working or self-employed 325 (84.6) 328 (87.0) 0.765 Unable to work 15 (3.9) 9 (2.4) In search of employment 7 (1.8) 5 (1.3) Housewife or houseman 28 (7.3) 29 (7.7) Unemployed 5 (1.3) 4 (1.1) Studying 4 (1.0) 2 (0.5) Missing 4 2 Mother smoking during pregnancy* No 364 (95.3) 358 (95.2) 0.961 Yes 18 (4.7) 18 (4.8) Missing 6 3 *N (%), P from the χ2 test. TABLE 4 Demographic data and fathers’ characteristics at baseline in the study groups

Active arm, N = 388 Control arm, N = 379 P Highest level of education at baseline* Basic minimum education 47 (12.2) 45 (11.9) 0.821 Additional education but not tertiary level 151 (39.2) 156 (41.2) Tertiary education 184 (47.8) 173 (45.6) Unknown 3 (0.8) 5 (1.3) Missing 3 0 Professional status at baseline* Working or self-employed 325 (84.6) 328 (87.0) 0.765 Unable to work 15 (3.9) 9 (2.4) In search of employment 7 (1.8) 5 (1.3) Housewife or houseman 28 (7.3) 29 (7.7) Unemployed 5 (1.3) 4 (1.1) Studying 4 (1.0) 2 (0.5) Missing 4 2 Mother smoking during pregnancy* No 364 (95.3) 358 (95.2) 0.961 Yes 18 (4.7) 18 (4.8) Missing 6 3 *N (%), P from the χ2 test. TABLE 4 Demographic data and fathers’ characteristics at baseline in the study groups Active arm, N = 388 Control arm, N = 379 P Highest level of education at baseline* Basic minimum education 66 (17.2) 59 (15.8) 0.125 Additional education but not tertiary level 167 (43.5) 170 (45.5) Tertiary education 143 (37.2) 144 (38.5) Unknown 8 (2.1) 1 (0.3) Missing 4 5 Professional status at baseline* Working or self-employed 367 (96.8) 361 (97.0) 0.612 Unable to work 6 (1.6) 2 (0.5) In search of employment 2 (0.5) 3 (0.8) Unemployed 3 (0.8) 5 (1.3) Studying 1 (0.3) 1 (0.3) Missing 9 7 *N (%), P value from the χ2 test. TABLE 5 Percentage of children with episodes with and without confirmation given by investigator

Active arm, N = 388 Control arm, N = 379 P Highest level of education at baseline* Basic minimum education 66 (17.2) 59 (15.8) 0.125 Additional education but not tertiary level 167 (43.5) 170 (45.5) Tertiary education 143 (37.2) 144 (38.5) Unknown 8 (2.1) 1 (0.3) Missing 4 5 Professional status at baseline* Working or self-employed 367 (96.8) 361 (97.0) 0.612 Unable to work 6 (1.6) 2 (0.5) In search of employment 2 (0.5) 3 (0.8) Unemployed 3 (0.8) 5 (1.3) Studying 1 (0.3) 1 (0.3) Missing 9 7 *N (%), P value from the χ2 test. TABLE 5 Percentage of children with episodes with and without confirmation given by investigator Active group, N = 388 Control group, N = 379 P RR (95% CI) NNT % Children with episodes 77 83 0.03 0.93 (0.87–1.00) 17 % Children with episodes confirmed by investigator 69 77 0.004 0.89 (0.82–0.97) 12 Post hoc analyses: arm comparison: (active vs control), P values are based on logistic regression. CI = confidence interval; NNT = number needed to treat. TABLE 6 Number of infectious episodes split for GII and URTI with or without fever Active vs control group GII or URTI (original) Only GII Only URTI With fever Without fever RR 0.93 0.98 0.91 0.91 1.00 95% CI 0.87–1.00 0.86–1.12 0.84–0.99 0.82–1.02 0.90–1.11 P 0.07 0.50 0.10 0.19 0.47 Post hoc analyses: arm comparison: RR and corresponding 95% CIs are calculated based on raw data, whereas P values are based on zero-inflated negative binomial regression model. CI = confidence interval; GII = gastrointestinal infections; NNT = number needed to treat; URTI = upper respiratory tract infection; RR = relative risk.

nalyses: arm comparison: RR and corresponding 95% CIs are calculated based on raw data, whereas P values are based on zero-inflated negative binomial regression model. CI = confidence interval; GII = gastrointestinal infections; NNT = number needed to treat; URTI = upper respiratory tract infection; RR = relative risk. TABLE 7 Medication use and duration of absence from day care Active group, N = 388 Control group, N = 379 P Use of any type of medication, median (IQR) 3.0 (1.0–6.0) 3.0 (1.0–6.0) 0.408 Duration of absence from day care, median (IQR) 2.0 (0.0–7.0) 3.0 (0.0–8.0) 0.536 Secondary outcomes: arm comparison: (active vs control), P values are based on Wilcoxon rank-sum test. IQR = interquartile range. TABLE 8 Comparison of cow's milk reference group vs GUM groups on the occurrence of infections Cow's milk group vs RR (95% CI) P Active group 1.19 (1.07–1.33) 0.008 Control group 1.11 (1.00–1.24) 0.004 Active and control groups 1.15 (1.04–1.28) 0.005 Post hoc analyses: RR and corresponding 95% CIs are calculated based on raw data, whereas P obtained from nonparametric test of the Mann-Whitney U test. CI = confidence interval; RR = relative risk. TABLE 9 Compliance with study product intake Intake of study product during intervention period, mL/day Active group, N = 388 Control group, N = 379 Overall No. infants missing 377 (11) 378 (1) 755 (12) Mean 528.8 527.2 528.0 Standard deviation 128.2 139.2 133.7 Arm comparison: active vs control, the t test P = 0.8715. TABLE 10 Weight, height, and BMI

TABLE 9 Compliance with study product intake Intake of study product during intervention period, mL/day Active group, N = 388 Control group, N = 379 Overall No. infants missing 377 (11) 378 (1) 755 (12) Mean 528.8 527.2 528.0 Standard deviation 128.2 139.2 133.7 Arm comparison: active vs control, the t test P = 0.8715. TABLE 10 Weight, height, and BMI Weight (kg), length (cm), and BMI (kg/m2) by visit (excluding screen failures)* Active group, N = 388 Control group, N = 379 Reference group, N = 33 P active vs control P active vs reference P control vs reference Weight of child at baseline 11.2 (1.64) 11.6 (1.77) 12.2 (1.16) 0.007 0.0001 0.003 Weight of child at wk 52 13.9 (2.05) 14.4 (2.27) 14.8 (1.30) 0.005 0.0007 0.087 Weight change of child from baseline at wk 52 2.7 (1.45) 2.8 (1.41) 2.5 (0.91) 0.266 0.196 0.051 Length of child at baseline 82.4 (5.80) 83.1 (6.06) 86.5 (5.15) 0.098 0.0001 0.001 Length of child at wk 52 92.3 (5.50) 93.3 (5.52) 97.8 (5.78) 0.010 0.0001 0.0001 Length change of child from baseline at wk 52 9.9 (3.29) 10.3 (3.20) 11.2 (3.62) 0.149 0.045 0.144 BMI of child at baseline 16.6 (1.66) 16.8 (1.89) 16.4 (1.77) 0.107 0.540 0.243 BMI of child at wk 52 16.4 (1.90) 16.5 (2.19) 15.6 (1.72) 0.321 0.023 0.012 BMI change of child from baseline at wk 52 −0.1 (1.92) −0.2 (1.88) −0.9 (1.37) 0.693 0.010 0.017 BMI = body mass index. *Mean (SD), P refers to the t test.

Inflammatory bowel disease (IBD) is often diagnosed in the pediatric age group, and these patients will eventually transfer from pediatric to adult health care. The prevailing culture of pediatric- and adult-centered care, however, differs tremendously (1,2). Although pediatric care values nurturance and often includes family members, the adult-centered care model values autonomy and respect for a patient's privacy, often with subsequent exclusion of family members. Adult-centered care requires a more active level of participation and self-management by the patient. Thus, the development of self-management skills is essential for a successful transition. Decision making, self-advocacy, communicative skills, and medication knowledge are all important aspects of self-management and need to be developed to facilitate a smooth transition to adult health care (3–5). Medication knowledge can be imparted as early as 10 to 12 years (6) and recommendations have been made for the gradual assumption of other self-management skills as patients mature. These skills include understanding the disease and the ability to schedule appointments and contact the provider.

adult health care (3–5). Medication knowledge can be imparted as early as 10 to 12 years (6) and recommendations have been made for the gradual assumption of other self-management skills as patients mature. These skills include understanding the disease and the ability to schedule appointments and contact the provider. Transition planning is increasingly recognized as an essential aspect of clinical care, yet <50% of children with special health care needs nationally are receiving these services (7). Although the need for transition planning is clear, the way to accomplish this is less certain. The majority of transition literature consists of expert opinion and recommendations rather than evidence-based conclusions (1,8–11). It is not yet known what constitutes the best way to accomplish efficient and effective transition planning (12). Some programs have transition coordinators who track all patients in the age range of interest or who attend visits with the patient in the adult setting (13). In some conditions, joint pediatric and adult clinics allow the providers to see patients together. Creating a formal comprehensive assessment and education program would be time and resource intensive. Transition clinics, whether for a single-disease entity or for all graduating patients, can add an institutional political element to the discussion. Without a clear reimbursement strategy, the approach to a formal structured transition seems expensive.

ssment and education program would be time and resource intensive. Transition clinics, whether for a single-disease entity or for all graduating patients, can add an institutional political element to the discussion. Without a clear reimbursement strategy, the approach to a formal structured transition seems expensive. Providers, however, impart tremendous amounts of information to patients in their typical interactions and patients view providers as the single best source of information. Providers presently report providing transition assessment and information on an informal basis (14). It seems plausible that increasing the awareness and knowledge of providers could have a large impact on their patients’ behavior. Before allocating extensive resources, we sought to examine the role of provider education and awareness in the patients’ self-management skill acquisition in an evidence-based manner.

14). It seems plausible that increasing the awareness and knowledge of providers could have a large impact on their patients’ behavior. Before allocating extensive resources, we sought to examine the role of provider education and awareness in the patients’ self-management skill acquisition in an evidence-based manner. METHODS Consecutive patients with IBD older than 10 years who presented to the outpatient setting were identified and administered a survey before seeing their provider. Patients given the diagnoses of Crohn disease, ulcerative colitis, or indeterminate colitis by histologic and endoscopic criteria were included. Patients were excluded if they had only recently been diagnosed as having IBD (<2 months) or if they were unable to fill out the survey (non–English-speaking patients or patients with severe developmental delays). The surveys were distributed consecutively at outpatient IBD clinic appointments during 2008 and 2011. A total of 358 patients were approached in 2008 and 156 patients in 2011. The survey was introduced by a letter given with the survey with parallel questionnaires for patients and parents. Participation was confidential, voluntary, and identified only by coded numbers. The survey was deidentified by putting it in a nameless sealed envelope, coded with serial numbers, and then collected by a member of the research staff before the appointment with the doctor to ensure confidentiality and anonymity. The providers were unaware of the patients’ answers or even whether they had completed the survey. Details of the survey can be found in previously published work (15).

th serial numbers, and then collected by a member of the research staff before the appointment with the doctor to ensure confidentiality and anonymity. The providers were unaware of the patients’ answers or even whether they had completed the survey. Details of the survey can be found in previously published work (15). Educational sessions on the topic of transition were held in 2009 and 2010 for all clinicians in the gastroenterology division, including attendings, fellows, nurses, dietitians, and social workers. All faculty attended at least 2 sessions and some attended all sessions. Two 60-minute sessions were factual, evidence-based lectures on transition. Two 60-minute sessions were case-based discussions designed to challenge and shift provider attitude. One of these cases is published and can be used publicly (16). The sessions raised awareness of the topic but did not prescribe or mandate any specific behavior. Informal discussions and hallway conversations among various providers on this topic once it was raised were noted and advice was sought from the educational session presenters for specific patient recommendations.

(16). The sessions raised awareness of the topic but did not prescribe or mandate any specific behavior. Informal discussions and hallway conversations among various providers on this topic once it was raised were noted and advice was sought from the educational session presenters for specific patient recommendations. Data Analysis The committee on clinical investigation deemed this study to be a quality improvement initiative not requiring formal review. Patient demographics and Likert scores were described by frequency and described as proportions. Analyses included descriptive statistics, cross-tabulations for categorical variables, and analyses of variance for continuous variables and logistic regressions. Significance was determined using the Fisher exact test for cross-tabulations and Wald test statistics for the coefficients of the logistic regressions. All analyses were performed using SPSS version 19 (IBM SPSS Statistics, Armonk, NY). RESULTS In 2008, 294 of 313 (94%) participants completed surveys. In 2011, 142 of 154 (92%) participants completed surveys. There were 40 patients who participated in both the 2008 and the 2011 cohorts. Demographics of both cohorts were similar (Table 1). The mean age in years was 16.7 (standard deviation 3.5) in 2008 and 16.5 (standard deviation 3.5) in 2011. There were no significant differences in the 2 groups, and previous exposure to the survey did not demonstrate reports of increased self-management skills.

rts. Demographics of both cohorts were similar (Table 1). The mean age in years was 16.7 (standard deviation 3.5) in 2008 and 16.5 (standard deviation 3.5) in 2011. There were no significant differences in the 2 groups, and previous exposure to the survey did not demonstrate reports of increased self-management skills. Independent self-management behaviors regarding medication-related tasks did not differ between the 2 groups (Table 2). There was a steady rise in reported participation in both groups. Patients showed the least independence in behaviors that occur outside the visit, such as scheduling appointments or contacting the provider if there was a problem that arose (Table 3). There was a trend toward independence over time, with 31% of 19- to 21-year-olds contacting the provider between visits in 2008, increasing to 56% in 2011, although this did not reach statistical significance. Patients showed the most independence with behaviors during the provider visits (Table 4). Preparing questions was the only category that reached statistical significance and only with those >21 years, increasing from 30% in 2008 to 67% in 2011. DISCUSSION We found that exposing providers to the concept and issues of transition did not result in significant changes in the reported self-management of patients in this evidence-based study. There remain extremely large gaps in the self-management skill sets of patients by age 18, an age at which many patients live apart from parents at college or are required to switch to adult-centered care.

on did not result in significant changes in the reported self-management of patients in this evidence-based study. There remain extremely large gaps in the self-management skill sets of patients by age 18, an age at which many patients live apart from parents at college or are required to switch to adult-centered care. In previous studies, we have shown that self-management skills often develop late, past the age at which many patients transfer to adult-centered care. This is in line with other studies that report psychosocial developmental milestones may be delayed in this population. Patients with IBD are reported to have fewer jobs in secondary school, vacation without adults less often, and fall in love later (17). Young adults with other pediatric-onset conditions tend to have similar delayed development (18).

s that report psychosocial developmental milestones may be delayed in this population. Patients with IBD are reported to have fewer jobs in secondary school, vacation without adults less often, and fall in love later (17). Young adults with other pediatric-onset conditions tend to have similar delayed development (18). The need for increased self-management skills in patients is clear because the health outcomes after transition often worsen, as has been documented in patients with diabetes mellitus, sickle cell disease, congenital heart disease, and liver transplantation (19–23). The added costs of teaching patients, coordination of care, and extra communication with accepting providers can, however, be problematic in this resource-conscious time. The American Academy of Pediatrics surveyed pediatricians and found that a low percentage followed transition guidelines because of limited staff training, lack of an identified staff member responsible for transition, and financial barriers (24). In Rhode Island, 1 multidisciplinary pilot transition clinic found that clinical billing did not cover the cost of care (25).

ians and found that a low percentage followed transition guidelines because of limited staff training, lack of an identified staff member responsible for transition, and financial barriers (24). In Rhode Island, 1 multidisciplinary pilot transition clinic found that clinical billing did not cover the cost of care (25). Many of the suggested transition interventions involve the expenditure of extra time or hiring of extra staff. A pilot study of liver transplant patient outcomes was improved by the addition of a transition coordinator (13). Some specialty clinics have joint clinics, with staff from both adult and pediatric providers (26), but insurance coverage may be problematic. Another recommendation is to have scheduled observations by each set of providers; however, that author notes “reimbursement for such activities may be challenging in systems of care dominated by insurance companies” (27). This was a single-center study. The survey documents reported behavior rather than observed behavior; thus, there may be a social desirability bias that causes overestimation of independence. We also did not collect disease severity, which has been shown to affect shared management (28), and transitional care is limited during a flare of the disease (29). This study may have been underpowered, because many categories showed a trend toward improved self-management skills but failed to reach statistical significance.

did not collect disease severity, which has been shown to affect shared management (28), and transitional care is limited during a flare of the disease (29). This study may have been underpowered, because many categories showed a trend toward improved self-management skills but failed to reach statistical significance. CONCLUSIONS Patients look to their providers for information about their condition and management. Although informal education of providers does seem to effect small shifts in patient behavior, it seems clear that a structured transition program would be needed to effectively move patients to more consistent self-management. Future studies are needed to assess the most cost-effective way to educate patients and providers, as well as to judge the outcomes of various transition strategies. This article has been developed as a Journal CME Activity by NASPGHAN. Visit http://www.naspghan.org/wmspage.cfm?parm1=742 to view instructions, documentation, and the complete necessary steps to receive CME credit for reading this article. The authors report no conflicts of interest. TABLE 1 Demographics of respondents

CONCLUSIONS Patients look to their providers for information about their condition and management. Although informal education of providers does seem to effect small shifts in patient behavior, it seems clear that a structured transition program would be needed to effectively move patients to more consistent self-management. Future studies are needed to assess the most cost-effective way to educate patients and providers, as well as to judge the outcomes of various transition strategies. This article has been developed as a Journal CME Activity by NASPGHAN. Visit http://www.naspghan.org/wmspage.cfm?parm1=742 to view instructions, documentation, and the complete necessary steps to receive CME credit for reading this article. The authors report no conflicts of interest. TABLE 1 Demographics of respondents Demographics 2008 2011 P N 294 142 Age, y (%) 10–12 50 (17) 19 (13) ns 13–15 82 (28) 36 (25) ns 16–18 77 (26) 50 (35) ns 19–21 51 (17) 25 (18) ns >21 34 (12) 12 (9) ns Sex (%) Male 148 (50) 62 (44) ns Female 146 (50) 80 (56) ns Diagnosis (%) CD 201 (68) 96 (68) ns UC 82 (28) 44 (31) ns IC 11 (3) 2 (1) ns Duration of disease, y (%) 3 or less 159 (54) 63 (44) ns >3 54 (46) 79 (56) ns Medication (%) On calcineurin inhibitors 6 (2) 3 (2) ns On biologics 84 (29) 55 (39) ns On immunomodulators 133 (45) 25 (18) ns On aminosalicylates 122 (41) 52 (37) ns For each cohort the number of respondents is listed with corresponding percentages. Patients could be taking >1 medication so the percentages do not add up to 100. There are no significant differences in the cohorts in terms of age, sex, diagnosis, or duration of disease, or in medication. CD = Crohn disease; IC = indeterminate colitis; ns = not significant; UC = ulcerative colitis.

nding percentages. Patients could be taking >1 medication so the percentages do not add up to 100. There are no significant differences in the cohorts in terms of age, sex, diagnosis, or duration of disease, or in medication. CD = Crohn disease; IC = indeterminate colitis; ns = not significant; UC = ulcerative colitis. TABLE 2 Independent behavior regarding medication among respondents Behavior by age, y 2008 2011 P Calls in refills (%) 10–12 0 0 ns 13–15 11 8 ns 16–18 13 27 ns 19–21 45 65 ns >21 76 70 ns Picks up medication at pharmacy (%) 10–12 0 0 ns 13–15 0 0 ns 16–18 10 15 ns 19–21 47 52 ns >21 73 60 ns Remembers to take medication (%) 10–12 26 11 ns 13–15 41 36 ns 16–18 59 57 ns 19–21 84 87 ns >21 94 100 ns Independent behavior is considered 4 (“mostly me”) or 5 (“I do it totally myself”) and represents the responsibility for doing the task rather than whether the task was always completed. Percentage is proportion of those who answered 4 or 5 over the total number in that age group that year. There were no significant differences seen in any group. ns = not significant. TABLE 3 Independent between-visit behaviors among respondents

Behavior by age, y 2008 2011 P Calls in refills (%) 10–12 0 0 ns 13–15 11 8 ns 16–18 13 27 ns 19–21 45 65 ns >21 76 70 ns Picks up medication at pharmacy (%) 10–12 0 0 ns 13–15 0 0 ns 16–18 10 15 ns 19–21 47 52 ns >21 73 60 ns Remembers to take medication (%) 10–12 26 11 ns 13–15 41 36 ns 16–18 59 57 ns 19–21 84 87 ns >21 94 100 ns Independent behavior is considered 4 (“mostly me”) or 5 (“I do it totally myself”) and represents the responsibility for doing the task rather than whether the task was always completed. Percentage is proportion of those who answered 4 or 5 over the total number in that age group that year. There were no significant differences seen in any group. ns = not significant. TABLE 3 Independent between-visit behaviors among respondents Behavior by age, y 2008 2011 P Contact MD if problem (%) 10–12 0 0 ns 13–15 1 0 ns 16–18 12 12 ns 19–21 31 56 ns >21 65 67 ns Schedules appointments (%) 10–12 2 0 ns 13–15 1 0 ns 16–18 9 6 ns 19–21 34 56 ns >21 67 75 ns Remembers appointments (%) 10–12 2 0 ns 13–15 3 0 ns 16–18 14 12 ns 19–21 41 52 ns >21 73 83 ns Independent behavior is considered 4 (“mostly me”) or 5 (“I do it totally myself”) and represents the responsibility for doing the task rather than whether the task was always completed. Percentage is proportion of those responding 4 or 5 compared with the total in that age group that year. There is no statistically significant difference between groups. ns = not significant. TABLE 4 Independent visit–related behaviors among respondents

Behavior by age, y 2008 2011 P Contact MD if problem (%) 10–12 0 0 ns 13–15 1 0 ns 16–18 12 12 ns 19–21 31 56 ns >21 65 67 ns Schedules appointments (%) 10–12 2 0 ns 13–15 1 0 ns 16–18 9 6 ns 19–21 34 56 ns >21 67 75 ns Remembers appointments (%) 10–12 2 0 ns 13–15 3 0 ns 16–18 14 12 ns 19–21 41 52 ns >21 73 83 ns Independent behavior is considered 4 (“mostly me”) or 5 (“I do it totally myself”) and represents the responsibility for doing the task rather than whether the task was always completed. Percentage is proportion of those responding 4 or 5 compared with the total in that age group that year. There is no statistically significant difference between groups. ns = not significant. TABLE 4 Independent visit–related behaviors among respondents Behavior by age, y 2008 2011 P Prepared questions (%) 10–12 12 11 ns 13–15 13 5 ns 16–18 9 16 ns 19–21 22 24 ns >21 30 66 0.041 Main role in talking (%) 10–12 8 17 ns 13–15 15 24 ns 16–18 36 45 ns 19–21 61 76 ns >21 85 92 ns Asks questions (%) 10–12 4 6 ns 13–15 10 11 ns 16–18 14 20 ns 19–21 54 52 ns >21 68 83 ns Answers questions (%) 10–12 16 22 ns 13–15 36 32 ns 16–18 56 64 ns 19–21 86 92 ns >21 100 100 ns Independent behavior is considered 4 (“mostly me”) or 5 (“I do it totally myself”) and represents the responsibility for doing the task rather than whether the task was always completed. Percentage is proportion of those responding 4 or 5 compared with the total in that age group that year.

Parenteral nutrition, or intravenous feeding, provides nutritional support for infants who do not have adequate gastrointestinal function. It can be a lifesaving therapy for newborn patients. Since the first documented report of an infant given parenteral nutrition was published in 1944, >30,000 neonates have survived through its use (1). The treatment is, however, not without its inherent risks, among which parenteral nutrition–associated liver disease (PNALD) is a common complication. Approximately 30% to 60% of the infants who require long-term parenteral nutrition develop PNALD, with abnormalities of liver function and hepatic damage (2). The latter can lead to cholestasis, especially in premature infants, and may result in life-threatening liver cirrhosis (3). Although the pathogenesis of parenteral nutrition–associated cholestasis is not entirely understood, in severe cases bile duct regeneration, portal inflammation, and fibrosis are contributing factors (4). Soybean oil is composed mainly of ω-6 polyunsaturated fatty acids (PUFAs). Recent evidence suggests that lipid emulsions that consist of soybean oil in parenteral nutrition mixtures may have an essential role in the onset of subsequent liver damage (1,5). On the contrary, fish oil is rich in ω-3 PUFAs, and fish oil–based lipid emulsions may be hepatoprotective and prevent PNALD (5,6). Moreover, the ω-3 PUFAs in fish oil are relatively safe and can be used in neonates (7) and preterm infants (8). Nevertheless, the mechanism associated with putative ω-3 PUFA–mediated hepatoprotection is unclear.

l is rich in ω-3 PUFAs, and fish oil–based lipid emulsions may be hepatoprotective and prevent PNALD (5,6). Moreover, the ω-3 PUFAs in fish oil are relatively safe and can be used in neonates (7) and preterm infants (8). Nevertheless, the mechanism associated with putative ω-3 PUFA–mediated hepatoprotection is unclear. The endoplasmic reticulum (ER) is the site of synthesis and folding of secretory proteins. Disturbances in ER function may cause unfolded protein response and ER stress (ER stress), eventually leading to cell death and many human diseases (9,10). Hepatocytes are secretory cells that are rich in ER, and ER stress in hepatocytes is closely associated with the pathogenesis of liver diseases (11,12). Glucose-regulated protein 94 (GRP94), a member of the heat shock protein 90 family, contributes to the regulation of protein folding in the ER and thus the control of ER stress (13,14). In the present study, we evaluated the effects of total parenteral nutrition (TPN) containing ω-3fish oil or ω-6 soybean oil on PNALD, by monitoring GRP94 levels in neonate rabbits.

The endoplasmic reticulum (ER) is the site of synthesis and folding of secretory proteins. Disturbances in ER function may cause unfolded protein response and ER stress (ER stress), eventually leading to cell death and many human diseases (9,10). Hepatocytes are secretory cells that are rich in ER, and ER stress in hepatocytes is closely associated with the pathogenesis of liver diseases (11,12). Glucose-regulated protein 94 (GRP94), a member of the heat shock protein 90 family, contributes to the regulation of protein folding in the ER and thus the control of ER stress (13,14). In the present study, we evaluated the effects of total parenteral nutrition (TPN) containing ω-3fish oil or ω-6 soybean oil on PNALD, by monitoring GRP94 levels in neonate rabbits. METHODS Experimental Assignment and Establishment of TPN Model The animal ethics committee of the Children's Hospital Affiliated to Soochow University granted approval for this study. Seven-day-old full-term New Zealand white rabbits (male and female, n = 24, weighing 100–120 g) were obtained from Wuxi Huishan Jiangnan Experimental Animal Centre (animal license number SCXK [Su] 2009–0005), Jiangsu, China. All of the rabbits were nursed from their mother before arrival. During the experimental period, the rabbits were maintained in an incubator at 26°C to 28°C and 40% to 60% humidity, under a 12 hour/12 hour light/dark cycle.

Jiangnan Experimental Animal Centre (animal license number SCXK [Su] 2009–0005), Jiangsu, China. All of the rabbits were nursed from their mother before arrival. During the experimental period, the rabbits were maintained in an incubator at 26°C to 28°C and 40% to 60% humidity, under a 12 hour/12 hour light/dark cycle. The rabbits were randomly and equally divided into 3 groups of 8, to be sustained for 1 week on total parenteral nutrition with soybean oil (TPN-soy) via infusion, TPN containing fish oil (TPN-FO) via infusion, or naturally nursed with rabbit milk only (control). In the TPN groups, animals were infused with nutrient mix; the total daily volume of intravenous nutrient solution for each rabbit in the TPN groups was 240 mL/kg, infused within 24 hours. The TPN regimen was sustained for 7 days, as previously described (15) (the components in the mix were purchased from Sino-Swed Pharmaceutical, Beijing, China [Tables 1 and 2].

nutrient mix; the total daily volume of intravenous nutrient solution for each rabbit in the TPN groups was 240 mL/kg, infused within 24 hours. The TPN regimen was sustained for 7 days, as previously described (15) (the components in the mix were purchased from Sino-Swed Pharmaceutical, Beijing, China [Tables 1 and 2]. Anesthesia was implemented with intraperitoneal injection of chloral hydrate (0.3 g/kg body weight). The rabbit was then placed in a horizontal dorsal decubitus position on the surgical table, and its legs were fixed to the extremities of the table. Skin sterilization was performed with benzalkonium bromide solution. For injection, the jugular vein was located, and a 10-gauge angiocatheter with a 1.2-mm silica gel tube was inserted approximately 1.5 cm into the superior vena cava. The tail end of the silica gel tube led out through a 0.5-cm incision in the dorsal scapular area, which was used as a subcutaneous tunnel exit. To avoid detachment, the end of the silica gel tube was connected to a rotating device.

a 1.2-mm silica gel tube was inserted approximately 1.5 cm into the superior vena cava. The tail end of the silica gel tube led out through a 0.5-cm incision in the dorsal scapular area, which was used as a subcutaneous tunnel exit. To avoid detachment, the end of the silica gel tube was connected to a rotating device. The components in 240 mL of the TPN given the TPN-soy group were 40 mL of a 20% medium- and long-chain fat emulsion (72 kcal, 34.3% of total calories), consisting of 2 g soybean oil, 2 g medium-chain triglyceride, and 0.24 g egg phospholipids; 80 mL of 11.4% pediatric compound amino acid injection-18AA-II (36.4 kcal, 17.3% of calories); 36 mL of 50% glucose (72 kcal, 48.4% of calories); 74 mL of 10% glucose (29.6 kcal); 4 mL of 10% NaCl; 3 mL of 10% KCl; 3 mL of 10% calcium gluconate; a half-piece of a water-soluble vitamin (0.3 mg vitamin B1, 0.36 mg vitamin B2, 4 mg nicotinamide, 0.4 mg vitamin B6, 1.5 mg pantothenic acid, 10 mg vitamin C, 6 μg biotin, 40 μg folic acid, and 0.5 μg vitamin B12); a half-piece fat-soluble vitamin (25 μg [82.5 IU] vitamin A, 10.125 μg [5 IU] vitamin D, 0.2275 mg [0.25 IU] vitamin E, and 3.75 μg vitamin K1); and a half-piece of trace elements (CaCl2·2H2O, 39.25 mg; MgCl2·6H2O, 15.21 mg; FeCl3·6H2O, 0.675 mg; ZnCl2, 0.135 mg; MnCl2·4H2O, 0.395 mg; CuCl2·2H2O, 42.5 μg; NaF, 0.105 mg; KI, 8.5 μg; Table 2). The TPN-FO group received the same as the TPN-soy group, but substituting for the 20% medium- and long-chain fat emulsion was 40 mL of 10% ω-3 fish oil–based lipid emulsion (containing 1.92–2.88 g soybean oil, 1.92–2.88 g medium-chain triglycerides, 1.6–2.4 g olive oil, 0.96–1.44 g fish oil, 4.32–6.48 mg dl-α-tocopherol, 0.384–0.576 g egg phosphatide, 0.9–1.1 g glycerol, 9.6–14.4 mg sodium oleate, and 0.72–0.88 mg sodium hydroxide). The ratio of ω-6 to ω-3 PUFAs in the ω-3 fish oil–based lipid emulsion was 3.0 to 2.2 to 1.

, 1.92–2.88 g medium-chain triglycerides, 1.6–2.4 g olive oil, 0.96–1.44 g fish oil, 4.32–6.48 mg dl-α-tocopherol, 0.384–0.576 g egg phosphatide, 0.9–1.1 g glycerol, 9.6–14.4 mg sodium oleate, and 0.72–0.88 mg sodium hydroxide). The ratio of ω-6 to ω-3 PUFAs in the ω-3 fish oil–based lipid emulsion was 3.0 to 2.2 to 1. Each 240-mL portion of TPN comprised 210 kcal, and the ratio of sugar to lipid was 1.4:1. Fat in each TPN group was given for 40 mL · kg−1 · day−1; components are listed in Table 1. Serological Evaluation To evaluate the relevant serological indicators, rabbits were anesthetized by intraperitoneal injection of 10% chloral hydrate. Two milliliters of blood were collected by cardiac puncture into lithium heparin anticoagulant tubes. After centrifuging at 3500 rpm, the serum was carefully separated and stored at −20°C until used. Before analysis, the serum samples were removed from the −20°C refrigerator and incubated overnight at 4°C. The total bilirubin, direct bilirubin, alanine aminotransferase, aspartate aminotransferase, total protein, albumin, γ-glutamyl transpeptidase (γ-GT), alkaline phosphatase (ALP), triglyceride, total cholesterol, and prealbumin levels were examined using a Hitachi 7600 automated chemistry analyzer (Hitachi, Tokyo, Japan).

The total bilirubin, direct bilirubin, alanine aminotransferase, aspartate aminotransferase, total protein, albumin, γ-glutamyl transpeptidase (γ-GT), alkaline phosphatase (ALP), triglyceride, total cholesterol, and prealbumin levels were examined using a Hitachi 7600 automated chemistry analyzer (Hitachi, Tokyo, Japan). Pathological Examination In addition to collecting blood samples (described above), liver tissues were collected after the animals were anesthetized. The abdomen was opened and the liver tissues were carefully removed. Some of the tissues were stored at −80°C for analysis of GRP94 messenger RNA (mRNA) and GRP94 protein (described below), whereas other portions were used for immunohistochemical (below) or histopathological analysis. After washing with normal saline tissue, samples were fixed in 10% paraformaldehyde, dehydrated through an alcohol series, cleared in xylene, and embedded in paraffin. Paraffin-embedded tissues were sectioned (5-μm thick) with a microtome. For histopathological comparisons, sections were dried, deparaffinized, and stained with hematoxylin and eosin. A pathologist who is experienced in liver disease reviewed the histology slides. Reverse-Transcriptase Polymerase Chain Reaction The mRNA levels of GRP94 were detected via reverse-transcriptase polymerase chain reaction. Liver tissues were removed from the −80°C refrigerator and crushed. Total RNA was extracted using Trizol reagent in accordance with the manufacturer's instructions (Invitrogen, Carlsbad, CA), and the amount and purification were evaluated with an ultraviolet spectrophotometer.

e-transcriptase polymerase chain reaction. Liver tissues were removed from the −80°C refrigerator and crushed. Total RNA was extracted using Trizol reagent in accordance with the manufacturer's instructions (Invitrogen, Carlsbad, CA), and the amount and purification were evaluated with an ultraviolet spectrophotometer. A total of 1 μg RNA was used to synthesize complementary DNA using a Reverse Transcriptase Kit (Promega, Madison, WI). PCR primers were designed using Primer 5.0 software, compared with the GenBank database for identification, and synthesized by Sangon Biotech (Shanghai, China). PCR amplification was performed on the GRP94 gene upstream primer (5′-AGGAAACACTCTGGGACG-3′) and downstream primer (5′-ATTCAGGTACTTAGGCATC-3′), producing an amplified fragment of 583 bp. Amplification of the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene upstream primer (5′-GTTTGTGATGGGCGTGAA-3′) and downstream primer (5′-CGAAGGTAGAGGAGTGGGTG-3′) produced an amplified fragment of 497 bp. The PCR reaction included 200 μmol/L dNTP, 2 U Taq DNA polymerase, 0.2 μmol/L of each of the upstream and downstream primers, 5 μL template complementary DNA, and ddH2O to reach a total volume of 50 μL. Reaction conditions were predenaturation at 94°C for 10 minutes; denaturation at 94°C for 45 seconds, annealing at 55°C for 30 seconds, and extension at 72°C for 30 seconds, for a total of 35 cycles; and extension at 72°C for 7 minutes.

imers, 5 μL template complementary DNA, and ddH2O to reach a total volume of 50 μL. Reaction conditions were predenaturation at 94°C for 10 minutes; denaturation at 94°C for 45 seconds, annealing at 55°C for 30 seconds, and extension at 72°C for 30 seconds, for a total of 35 cycles; and extension at 72°C for 7 minutes. After electrophoresis in a 1.5% agarose gel, ethidium bromide–stained bands were visualized by ultraviolet transillumination, and the fluorescence intensity was semiquantified using a Bio2239 gel analysis system (Bio-Print, Chicago, IL). Immunohistochemistry Some of the paraffin-embedded tissue sections (described above) were used for immunohistochemical analysis. The glass slides used for immunohistochemistry were precoated with poly-l-lysine (WuHanBoster Biological Technology, Wuhan, China). Immunostaining was carried out with a streptavidin-peroxidase kit obtained from Suzhou Enmaike Bio-Tech (Suzhou, China) in accordance with the manufacturer's instructions. Three nonoverlapping fields were randomly selected under ×400 magnification. The cells positively stained with anti-GRP94 primary antibody appeared with brown–yellow granules in the cytoplasm of hepatocytes. The intensity of immunohistochemical staining was analyzed using Image-Pro-Plus image analysis software. An average gray value supplied by the software was used to reference the intensity of GRP94 staining (ie, normalized to an internal control).

body appeared with brown–yellow granules in the cytoplasm of hepatocytes. The intensity of immunohistochemical staining was analyzed using Image-Pro-Plus image analysis software. An average gray value supplied by the software was used to reference the intensity of GRP94 staining (ie, normalized to an internal control). Western Blot Total protein was extracted from cells using lysis buffer. Protein concentrations were measured and equal amounts of protein extracts were resolved using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), which were then transferred to a polyvinylidene fluoride membrane (Millipore, Temecula, CA). Membranes were blocked with blocking buffer for 2 hours, and then incubated with primary antibody against GRP94 or β-actin (1:1000 dilution) at 4°C overnight. After washing, membranes were incubated with ALP-conjugated goat anti-rabbit secondary antibody (1:600 dilution) at room temperature. Immunobands were visualized using an ALP kit (WesternBreeze; Invitrogen). To quantify protein levels, the expression bands of target proteins were analyzed, and the densitometric values were used to conduct statistical analysis. The housekeeping protein β-actin was used as an internal control. Statistical Analyses Data were analyzed using SPSS 17.0 software (SPSS Inc, Chicago, IL) and are presented as mean ± standard deviation. Statistical significance was determined using 1-way analysis of variance. P < 0.05 was recognized as significantly different.

Western Blot Total protein was extracted from cells using lysis buffer. Protein concentrations were measured and equal amounts of protein extracts were resolved using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), which were then transferred to a polyvinylidene fluoride membrane (Millipore, Temecula, CA). Membranes were blocked with blocking buffer for 2 hours, and then incubated with primary antibody against GRP94 or β-actin (1:1000 dilution) at 4°C overnight. After washing, membranes were incubated with ALP-conjugated goat anti-rabbit secondary antibody (1:600 dilution) at room temperature. Immunobands were visualized using an ALP kit (WesternBreeze; Invitrogen). To quantify protein levels, the expression bands of target proteins were analyzed, and the densitometric values were used to conduct statistical analysis. The housekeeping protein β-actin was used as an internal control. Statistical Analyses Data were analyzed using SPSS 17.0 software (SPSS Inc, Chicago, IL) and are presented as mean ± standard deviation. Statistical significance was determined using 1-way analysis of variance. P < 0.05 was recognized as significantly different. RESULTS Serological Indicators No statistical significance was found in the levels of total protein, alanine aminotransferase, ALP, aspartate aminotransferase, triglyceride, total cholesterol, or prealbumin among the 3 groups (P > 0.05, Table 3). Compared with control animals, TPN-soy animals had significantly higher serum total bilirubin, direct bilirubin, and γ-GT levels, but lower albumin (P < 0.01). No statistical differences were detected in the levels of these indicators in TPN-FO animals compared with the controls (P > 0.05, each). Compared with the TPN-soy group, serum total bilirubin, direct bilirubin, and γ-GT were significantly lower in the TPN-FO group (F = 1247.40, 1037.94, 971.09, respectively; P < 0.01 each), and albumin was significantly higher (F = 70.31, P < 0.01).

dicators in TPN-FO animals compared with the controls (P > 0.05, each). Compared with the TPN-soy group, serum total bilirubin, direct bilirubin, and γ-GT were significantly lower in the TPN-FO group (F = 1247.40, 1037.94, 971.09, respectively; P < 0.01 each), and albumin was significantly higher (F = 70.31, P < 0.01). Liver Pathology Histological examination of the liver tissues obtained from the 3 experimental groups revealed that those of the control rabbits appeared normal, with intact hepatocytes (Fig. 1A) and without any signs of hepatocyte degeneration, necrosis, inflammatory cell infiltration, cholangiectasis, bile duct epithelial hyperplasia, or cholestasis. In the liver tissues of the TPN-soy group, inflammatory cell infiltration, diffuse hepatic steatosis, and disrupted hepatic cord structure were, however, evident (Fig. 1B), but there was no cholestasis or liver fibrosis, and the hepatic lobule was still visible. In the TPN-FO group, only mild hepatic steatosis and inflammatory cell infiltration were found (Fig. 1C); the morphology of hepatocytes was normal, and there was no cholangiectasis, bile duct epithelial hyperplasia, or cholestasis. FIGURE 1 Representative histologic images of liver tissues. A, Control; B, TPN-soy group; C, TPN-FO group. Original magnification ×400. TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil.

Liver Pathology Histological examination of the liver tissues obtained from the 3 experimental groups revealed that those of the control rabbits appeared normal, with intact hepatocytes (Fig. 1A) and without any signs of hepatocyte degeneration, necrosis, inflammatory cell infiltration, cholangiectasis, bile duct epithelial hyperplasia, or cholestasis. In the liver tissues of the TPN-soy group, inflammatory cell infiltration, diffuse hepatic steatosis, and disrupted hepatic cord structure were, however, evident (Fig. 1B), but there was no cholestasis or liver fibrosis, and the hepatic lobule was still visible. In the TPN-FO group, only mild hepatic steatosis and inflammatory cell infiltration were found (Fig. 1C); the morphology of hepatocytes was normal, and there was no cholangiectasis, bile duct epithelial hyperplasia, or cholestasis. FIGURE 1 Representative histologic images of liver tissues. A, Control; B, TPN-soy group; C, TPN-FO group. Original magnification ×400. TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. GRP94 mRNA Levels in Liver Tissues Hematoxylin and eosin staining of liver tissues revealed that rabbits given TPN containing fish oil sustained only mild hepatic steatosis and inflammatory cell infiltration compared with the animals infused with TPN containing soybean oil. Considering that ER stress is associated with liver pathology and GRP94 participates in the regulation of ER stress, we then assessed the mRNA levels of GRP94 in the liver tissues of the different groups to investigate the molecular mechanism underlying the seeming hepatic protection of TPN-FO against TPN-soy–induced liver damage.

that ER stress is associated with liver pathology and GRP94 participates in the regulation of ER stress, we then assessed the mRNA levels of GRP94 in the liver tissues of the different groups to investigate the molecular mechanism underlying the seeming hepatic protection of TPN-FO against TPN-soy–induced liver damage. We found that the GRP94 mRNA levels in liver tissues of the TPN-soy group (1.217 ± 0.113, referenced to the gray value standard) were significantly higher than these levels in the control (0.614 ± 0.034, P < 0.01; Fig. 2) and also significantly higher than the GRP94 mRNA levels of the TPN-FO group (0.661 ± 0.117). The GRP94 mRNA levels of the TPN-FO and controls were similar. FIGURE 2 GRP94 mRNA levels in liver tissues determined by RT-PCR. A, Representative RT-PCR results. MARK, DNA marker; lanes 1, 2, control; lanes 3, 4, TPN-soy; lanes 5, 6, TPN-FO. B, Quantitative data. #P < 0.01, compared with control; ∗P < 0.01, compared with TPN-soy. Glyceraldehyde 3-phosphate dehydrogenase was used as internal control. GRP94 = glucose-regulated protein 94; mRNA = messenger RNA; RT-PCR = reverse-transcriptase polymerase chain reaction; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. GRP94 Protein Levels in Liver Tissues To further our investigation of the mechanism underlying hepatic protection associated with TPN-FO, the protein levels of GRP94 in liver tissues were determined via immunohistochemistry and Western blot assays.

FIGURE 2 GRP94 mRNA levels in liver tissues determined by RT-PCR. A, Representative RT-PCR results. MARK, DNA marker; lanes 1, 2, control; lanes 3, 4, TPN-soy; lanes 5, 6, TPN-FO. B, Quantitative data. #P < 0.01, compared with control; ∗P < 0.01, compared with TPN-soy. Glyceraldehyde 3-phosphate dehydrogenase was used as internal control. GRP94 = glucose-regulated protein 94; mRNA = messenger RNA; RT-PCR = reverse-transcriptase polymerase chain reaction; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. GRP94 Protein Levels in Liver Tissues To further our investigation of the mechanism underlying hepatic protection associated with TPN-FO, the protein levels of GRP94 in liver tissues were determined via immunohistochemistry and Western blot assays. Immunostaining of these tissues showed that GRP94 protein levels in the liver tissues of the TPN-soy group (133.84 ± 13.66, referenced to the gray value standard) were significantly higher than those of the controls (78.14 ± 8.17, P < 0.01; Fig. 3) and also significantly higher than the GRP94 protein levels of the TPN-FO group (80.73 ± 9.36, P < 0.01), whereas the GRP94 protein levels of the TPN-FO and controls were similar.

133.84 ± 13.66, referenced to the gray value standard) were significantly higher than those of the controls (78.14 ± 8.17, P < 0.01; Fig. 3) and also significantly higher than the GRP94 protein levels of the TPN-FO group (80.73 ± 9.36, P < 0.01), whereas the GRP94 protein levels of the TPN-FO and controls were similar. FIGURE 3 Immunohistological analysis of GRP94 protein in liver tissues. Representative results of GRP94 staining in samples derived from (A) control, (B) TPN-soy group, and (C) TPN-FO group. Quantitative data (D). Original magnification ×400. #P < 0.01, compared with control; ∗P < 0.01, compared with TPN-soy. GRP94 = glucose-regulated protein 94; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. The results obtained by Western blot were in accord with those of the immunostaining (Fig. 4). That is, there was no associated upregulation in the GRP94 protein levels in the TPN-FO group (0.29 ± 0.03, relative optical density) as there was in the TPN-soy–treated animals (0.63 ± 0.04, P < 0.05), and GRP94 protein levels of the TPN-FO and controls (0.22 ± 0.01) did not differ significantly (P > 0.05). These data suggest that TPN-FO may prevent liver damage induced by TPN-soy, possibly by suppressing GRP94 upregulation and ER stress.

l density) as there was in the TPN-soy–treated animals (0.63 ± 0.04, P < 0.05), and GRP94 protein levels of the TPN-FO and controls (0.22 ± 0.01) did not differ significantly (P > 0.05). These data suggest that TPN-FO may prevent liver damage induced by TPN-soy, possibly by suppressing GRP94 upregulation and ER stress. FIGURE 4 Western blot of the GRP94 protein in liver tissues. A, Representative Western blot results. Lane 1, control; lane 2, TPN-FO; lane 3, TPN-soy. B, Quantitative data. #P < 0.05, compared with control; ∗P < 0.05, compared with TPN-soy. Actin was used as internal control. GRP94 = glucose-regulated protein 94; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil.

results. Lane 1, control; lane 2, TPN-FO; lane 3, TPN-soy. B, Quantitative data. #P < 0.05, compared with control; ∗P < 0.05, compared with TPN-soy. Actin was used as internal control. GRP94 = glucose-regulated protein 94; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. DISCUSSION PNALD is a serious complication of patients, especially infants, who require long-term parenteral nutrition therapy. Here, we sustained 7-day-old rabbit kits for 1 week with TPN containing either soybean oil or ω-3 fish oil, to examine the role of ω-3 fish oil in preventing parenteral nutrition–associated liver injury. The effects of both of these were compared with normally nursed controls with regard to signs of PNALD. We found that, compared with the control group, usual soybean oil parenteral nutrition was associated with significant liver dysfunction, as indicated by higher serum total bilirubin, direct bilirubin, and γ-GT levels and lower serum albumin. These effects were not observed in the TPN-FO group, which was similar to the control group. Moreover, histological examination of liver tissues revealed hepatic damage in the TPN-soy group not seen in the TPN-FO, including inflammatory cell infiltration, diffuse hepatic steatosis, and disrupted hepatic cord structure. These observations are consistent with previous reports (15).

similar to the control group. Moreover, histological examination of liver tissues revealed hepatic damage in the TPN-soy group not seen in the TPN-FO, including inflammatory cell infiltration, diffuse hepatic steatosis, and disrupted hepatic cord structure. These observations are consistent with previous reports (15). Rats given TPN were found to have higher levels of the ER stress marker protein CHOP (C/EBP homologous protein), and also higher levels of molecules that are proapoptotic under ER stress, c-Jun NH2-terminal kinase (JNK1/2), and p38 MAPK (16). This suggests that ER stress is induced by TPN therapy. GRP94 is also a marker for ER stress, and in the present study we detected higher levels of GRP94 mRNA and GRP94 protein in the liver tissues of the rabbits given TPN-soy. Therefore, ER stress may participate in TPN-mediated liver damage in both rats and rabbits. Furthermore, it was reported that in normal liver L02 cells cultured in vitro, ER stress contributed to the progression of PNALD (17). In that report, ERS was induced with palmitate, which led to the upregulation of tribbles homolog 3 (TRB3), a pseudokinase that is known to be involved in the pathogenesis of PNALD. Therefore, the ER stress seems be an important contributing factor in the pathogenesis and progression of PNALD.

rogression of PNALD (17). In that report, ERS was induced with palmitate, which led to the upregulation of tribbles homolog 3 (TRB3), a pseudokinase that is known to be involved in the pathogenesis of PNALD. Therefore, the ER stress seems be an important contributing factor in the pathogenesis and progression of PNALD. Although the etiology of PNALD is poorly understood, the soybean or combined soybean and safflower oils that are included in TPN are accepted as contributing factors (18). Both of these oils are rich in ω-6 fatty acids. It has been reported that ω-6 PUFAs generate proinflammatory mediators, which may contribute to the onset of liver diseases, whereas mediators derived from ω-3 PUFAs are largely anti-inflammatory (19). A randomized controlled trial conducted by Puder et al (20) showed that fish oil–based intravenous lipid emulsion was safe for infants with PNALD, and could reduce mortality and organ transplantation rates in children with short bowel syndrome. Consistent with their observations, Diamond et al (21) reported that ω-3 fatty acids may prevent PNALD by improving bile flow, inhibiting steatosis, and exerting immunomodulatory effects, although the molecular mechanism involved in this process remains unclear. In the present study, we found that substitution of ω-3 fish oil for soybean fat emulsion was associated with prevention of liver dysfunction indicated by serology results, and liver tissue damage observed through histology. This implies that ω-3 fish oil may protect against PNALD. Moreover, levels of GRP94 mRNA and GRP94 protein in the kits given TPN with ω-3 fish oil were comparable with the control rabbits, and significantly lower than those given TPN with soybean fat emulsion. These data indicate that TPN-induced liver injury was reduced in those given ω-3 fish oil, unlike those given soybean fat emulsion, and its mechanism may be associated with a reduction in ER stress by reducing the GRP94 expression associated with soybean fat emulsion. Our findings are consistent with a previous report that glycyrrhizin, an active component of licorice root that has been used to treat chronic hepatitis, represses TPN-associated acute liver injury in rats by suppressing ER stress (16).

in ER stress by reducing the GRP94 expression associated with soybean fat emulsion. Our findings are consistent with a previous report that glycyrrhizin, an active component of licorice root that has been used to treat chronic hepatitis, represses TPN-associated acute liver injury in rats by suppressing ER stress (16). There was a difference in the amount of α-tocopherol between the 2 TPN solutions. α-Tocopherol is a well-known lipophilic antioxidant that has the ability to scavenge peroxyl radicals (22). Nandivada et al (19) reported that the risk factors of cholestasis and hepatic injury observed in PNALD included elevated serum concentrations of phytosterols, an abundance of ω-6 PUFAs, and a relative paucity of α-tocopherol. Moreover, previous research indicated that α-tocopherol protected against CCl4-induced liver damage (22). Unfortunately, we did not investigate whether the α-tocopherols play a role in the prevention of PNALD in the TPN-FO group. In summary, our present study showed that substitution of ω-3 fish oil for soybean fat emulsion in TPN greatly prevented liver dysfunction and liver tissue damage in week-old rabbit kits, possibly by preventing ER stress. Our study may provide valuable evidence for the use of ω-3 fish oil for preventing PNALD in infants. Acknowledgments The authors gratefully acknowledge all members of the laboratory for sharing reagents and advice and thank the pathologist for reviewing the histology slides. The authors also thank Medjaden Bioscience for assisting in the preparation of this manuscript.

In summary, our present study showed that substitution of ω-3 fish oil for soybean fat emulsion in TPN greatly prevented liver dysfunction and liver tissue damage in week-old rabbit kits, possibly by preventing ER stress. Our study may provide valuable evidence for the use of ω-3 fish oil for preventing PNALD in infants. Acknowledgments The authors gratefully acknowledge all members of the laboratory for sharing reagents and advice and thank the pathologist for reviewing the histology slides. The authors also thank Medjaden Bioscience for assisting in the preparation of this manuscript. Drs Zhu and Wang contributed equally to the article. This research was supported by grants from the Suzhou Science and Technology Development Project (No. SYS201136, SYS201440), Natural Science Foundation Project of Jiangsu Province (No. BK20141183), Research Project of Department of Health of Jiangsu Province (No. H201316), 135 Project of Department of Health of Jiangsu Province (No. RC2007076), and the Research Project of the Suzhou Key Laboratory of Children's Developmental Brain Injury Prevention and Care (No. SZS201108). The authors report no conflicts of interest. TABLE 1 Fat emulsions of the TPN-soy and TPN-FO groups

This research was supported by grants from the Suzhou Science and Technology Development Project (No. SYS201136, SYS201440), Natural Science Foundation Project of Jiangsu Province (No. BK20141183), Research Project of Department of Health of Jiangsu Province (No. H201316), 135 Project of Department of Health of Jiangsu Province (No. RC2007076), and the Research Project of the Suzhou Key Laboratory of Children's Developmental Brain Injury Prevention and Care (No. SZS201108). The authors report no conflicts of interest. TABLE 1 Fat emulsions of the TPN-soy and TPN-FO groups Wt/40 mL* TPN-soy 20% medium- and long-chain fats Soybean oil 2 g Medium-chain triglyceride 2 g Egg phospholipids 0.24 g TPN-FO 10% ω-3 fish oil–based lipid emulsion† Soybean oil 1.92–2.88 g Medium-chain triglyceride 1.92–2.88 g Olive oil 1.6–2.4 g Fish oil 0.96–1.44 g dl-α-Tocopherol 4.32–6.48 mg Egg phosphatide 0.384–0.576 g Glycerol 0.9–1.1 g Sodium oleate 9.6–14.4 mg Sodium hydroxide 0.72–0.88 mg TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. *In each 240-mL portion of TPN, containing 40 mL fat and comprising 210 kcal. Ratio of sugar to lipid was 1.4:1. †The ratio of ω-6 to ω-3 PUFAs in the ω-3 fish oil-based lipid emulsion was 3.0–2.2 to 1. TABLE 2 Vitamins and trace elements per 240 mL TPN in both groups

Wt/40 mL* TPN-soy 20% medium- and long-chain fats Soybean oil 2 g Medium-chain triglyceride 2 g Egg phospholipids 0.24 g TPN-FO 10% ω-3 fish oil–based lipid emulsion† Soybean oil 1.92–2.88 g Medium-chain triglyceride 1.92–2.88 g Olive oil 1.6–2.4 g Fish oil 0.96–1.44 g dl-α-Tocopherol 4.32–6.48 mg Egg phosphatide 0.384–0.576 g Glycerol 0.9–1.1 g Sodium oleate 9.6–14.4 mg Sodium hydroxide 0.72–0.88 mg TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. *In each 240-mL portion of TPN, containing 40 mL fat and comprising 210 kcal. Ratio of sugar to lipid was 1.4:1. †The ratio of ω-6 to ω-3 PUFAs in the ω-3 fish oil-based lipid emulsion was 3.0–2.2 to 1. TABLE 2 Vitamins and trace elements per 240 mL TPN in both groups Vitamins Water-soluble B1 0.3 mg B2 0.36 mg Nicotinamide 4 mg B6 0.4 mg Pantothenic acid 1.5 mg C 10 mg Biotin 6 μg Folic acid 40 μg B12 0.5 μg Fat-soluble A 25 μg (82.5 IU) D 10.125 μg (5 IU) E 0.2275 mg (0.25 IU) K1 3.75 μg Trace elements CaCl2·2H2O 39.25 mg MgCl2·6H2O 15.21 mg FeCl3·6H2O 0.675 mg ZnCl2 0.135 mg MnCl2·4H2O 0.395 mg CuCl2·2H2O 42.5 μg NaF 0.105 mg KI 8.5 μg TPN = total parenteral nutrition. TABLE 3 Serological indicator levels of the 3 experimental groups

Vitamins Water-soluble B1 0.3 mg B2 0.36 mg Nicotinamide 4 mg B6 0.4 mg Pantothenic acid 1.5 mg C 10 mg Biotin 6 μg Folic acid 40 μg B12 0.5 μg Fat-soluble A 25 μg (82.5 IU) D 10.125 μg (5 IU) E 0.2275 mg (0.25 IU) K1 3.75 μg Trace elements CaCl2·2H2O 39.25 mg MgCl2·6H2O 15.21 mg FeCl3·6H2O 0.675 mg ZnCl2 0.135 mg MnCl2·4H2O 0.395 mg CuCl2·2H2O 42.5 μg NaF 0.105 mg KI 8.5 μg TPN = total parenteral nutrition. TABLE 3 Serological indicator levels of the 3 experimental groups TPN-FO TPN-soy Control F P Total bilirubin, μmol/L 3.28 ± 1.69* 50.2 ± 3.06† 2.28 ± 1.04 1247.40 <0.01 Direct bilirubin, μmol/L 1.76 ± 0.86* 47.2 ± 3.91† 1.05 ± 0.57 1037.94 <0.01 Total protein, g/L 25.6 ± 3.01 23.0 ± 1.94 24.2 ± 3.12 1.48 0.25 Albumin, g/L 26.3 ± 3.12* 11.7 ± 1.05† 26.8 ± 2.83 70.31 <0.01 AST, IU/L 77.1 ± 4.41 78.8 ± 2.70 72.7 ± 10.3 1.47 0.25 ALT, IU/L 24.9 ± 2.98 26.8 ± 6.57 21.9 ± 1.49 2.79 0.08 γ-GT, IU/L 14.3 ± 5.46* 130 ± 5.61† 11.2 ± 5.68 971.09 <0.01 ALP, IU/L 74.0 ± 4.30 72.3 ± 6.69 72.3 ± 13.6 0.08 0.92 Triglyceride, mmol/L 2.18 ± 0.89 2.86 ± 1.49 1.74 ± 1.29 0.14 0.87 Total cholesterol, mmol/L 0.69 ± 0.53 0.92 ± 0.62 0.41 ± 0.21 0.42 0.66 Prealbumin, mg/L 88.8 ± 27.2 86.5 ± 13.8 80.8 ± 29.9 0.64 0.54 ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; γ-GT, γ-glutamyl transpeptidase; TPN = total parenteral nutrition; TPN-FO = TPN with ω-3 fish oil; TPN-soy = TPN with soybean oil. *P < 0.01 compared with TPN-soy. †P < 0.01 compared with the control.

Ulcerative colitis is a type of inflammatory bowel disease characterized by chronic mucosal inflammation of the colon. With the exception of patients who have a cecal patch, the inflammatory response usually begins in the rectum and extends proximally with a diffuse, continuous pattern. Approximately 15% to 20% of patients with ulcerative colitis are children. In the United States, the incidence of pediatric ulcerative colitis varies between 1 and 4/100,000 individuals per year (1). Estimates of the average age-at-onset in children vary, although 80% to 90% of patients are age ≥9 years when symptoms develop (2,3). The incidence and disease pattern seen in the United States are similar to those observed in other developed countries (4). Evidence from the medical literature suggests that the clinical course and manifestations of ulcerative colitis are similar in children and adults (5,6); however, younger children tend to have increased colitis and more diffuse involvement with pancolitis compared with older children and adults (7). The most common symptoms of ulcerative colitis—diarrhea, abdominal pain, rectal bleeding, fever, and weight loss—are found in comparable proportions of both children and adults, and are more dependent on the disease activity than on age.

e diffuse involvement with pancolitis compared with older children and adults (7). The most common symptoms of ulcerative colitis—diarrhea, abdominal pain, rectal bleeding, fever, and weight loss—are found in comparable proportions of both children and adults, and are more dependent on the disease activity than on age. Oral mesalamine (Asacol; Warner Chilcott, Rockaway, NJ) is often used as maintenance treatment of ulcerative colitis in adults and children. Although some evidence points to a dose-response relation in adult patients with active ulcerative colitis treated with oral mesalamine (8), data are sparse to support the claim of a relation between dose and clinical efficacy in the pediatric population. Moreover, the safety of oral mesalamine, which is well established in the adult population, lacks confirmation in children. Among practicing pediatric gastroenterologists, the daily dose of oral mesalamine administered to children with active ulcerative colitis ranges from 30 to >100 mg · g−1 · day−1. The purpose of this study was to investigate the safety and efficacy of low- and high-dose oral, delayed-release mesalamine for the treatment of children with active, mild-to-moderate ulcerative colitis.

of oral mesalamine administered to children with active ulcerative colitis ranges from 30 to >100 mg · g−1 · day−1. The purpose of this study was to investigate the safety and efficacy of low- and high-dose oral, delayed-release mesalamine for the treatment of children with active, mild-to-moderate ulcerative colitis. METHODS Study Design This was a randomized, multicenter, double-blind, active control, parallel group study. It was conducted in accordance with the ethical principles of Good Clinical Practice and approved at all sites by the appropriate institutional review boards or independent ethics committees, as applicable. The study was conducted in 26 clinical practice centers across the United States, Canada, Romania, Croatia, and Poland.

conducted in accordance with the ethical principles of Good Clinical Practice and approved at all sites by the appropriate institutional review boards or independent ethics committees, as applicable. The study was conducted in 26 clinical practice centers across the United States, Canada, Romania, Croatia, and Poland. Patients Written informed consent was obtained from each patient's parent or legal guardian according to the US Code of Federal Regulations (US CFR; Title 21, Part 50, §55, 56), International Conference on Harmonisation harmonized tripartite guideline for good clinical practice, and ethical principles that have their origin in the Declaration of Helsinki. In addition, age-appropriate patient information sheets were provided, and patients who were >7 years of age were asked to sign a form indicating their assent to participate in the study. Both male and female patients ages 5 to 17 years with a history of biopsy- and endoscopy-confirmed ulcerative colitis were enrolled. Inclusion criteria included mild-to-moderately active ulcerative colitis (relapsed or newly diagnosed) as defined by Pediatric Ulcerative Colitis Activity Index (PUCAI) scores of ≥10 to ≤55 (9); baseline scores of ≥1 for the symptomatic components of the Mayo Score, rectal bleeding, and stool frequency (Table 1); and body weight ≥17 to ≤90 kg. Patients had to be able to swallow the study drug tablets. In addition, female patients were either premenarchal or had a negative urine pregnancy test. If sexually active, patients had to practice a reliable form of contraception. Exclusion criteria, including medical history and previous therapies, are listed in supplementary Table S1. Patients were prohibited from taking exclusionary drugs and any drugs that may interfere with the evaluation of the study medication during the study. Patients were required to stop their oral mesalamine at randomization.

criteria, including medical history and previous therapies, are listed in supplementary Table S1. Patients were prohibited from taking exclusionary drugs and any drugs that may interfere with the evaluation of the study medication during the study. Patients were required to stop their oral mesalamine at randomization. Protocol Patients were seen at screening, baseline (within 1 week of screening), week 3, and week 6/withdrawal visits. In addition, a follow-up telephone call was made 1 week after baseline. Screening and baseline visits were allowed to take place on the same day (supplementary Fig. S1). At screening, patient eligibility was determined by PUCAI score, physical examination findings, clinical laboratory tests (hematology, serum biochemistry, and urinalysis), and pregnancy test results. At baseline, samples were taken for clinical laboratory tests if the baseline visit occurred >7 days after screening, and for pregnancy testing. Patients underwent clinical assessments of disease activity at baseline and at week 6/withdrawal visit. These included all domains of the PUCAI and the truncated Mayo Score components (stool frequency and rectal bleeding). A PUCAI diary card was completed on each of the 2 days preceding the visit. Bristol stool charts were used to assist with the relevant PUCAI domain assessment. Questions were addressed to the patient initially, wherever appropriate, based on age and responsiveness, followed by the parent for an additional perspective or confirmation. In addition, at week 6/withdrawal visit the patient was asked, “How do you rate the change in disease activity since starting?” A 7-point scale was used for the global assessment of change of disease activity: significantly improved, moderately improved, mildly improved, no change, mildly worse, moderately worse, and significantly worse.

k 6/withdrawal visit the patient was asked, “How do you rate the change in disease activity since starting?” A 7-point scale was used for the global assessment of change of disease activity: significantly improved, moderately improved, mildly improved, no change, mildly worse, moderately worse, and significantly worse. Fecal calprotectin and lactoferrin samples were collected at baseline and visits at weeks 3 and 6/withdrawal. Endoscopy was not mandated by the study protocol. Rescue medication was not permitted during the study, and any patient requiring additional treatment for ulcerative colitis was removed from the trial. Compliance was assessed at visits at weeks 3 and 6/withdrawal by counting the unused pills.

weeks 3 and 6/withdrawal. Endoscopy was not mandated by the study protocol. Rescue medication was not permitted during the study, and any patient requiring additional treatment for ulcerative colitis was removed from the trial. Compliance was assessed at visits at weeks 3 and 6/withdrawal by counting the unused pills. Study Drug Assignment Patients were randomly assigned in a 1:1 ratio to either a high or low dose of oral, delayed-release mesalamine in a double-blind fashion. Randomization was stratified by body weight (17 to <33, 33 to <54, and 54–90 kg) and by disease severity (mild: defined as a baseline PUCAI score of 10–30; or moderate: a baseline PUCAI score of 31–55). Subjects received body weight–dependent doses of oral, delayed-release mesalamine for 6 weeks in a low- (27–71 mg · g−1 · day−1) or high-dose group (53–118 mg · g−1 · day−1). The high doses in each body weight category were approximately 1.67-, 1.8-, and 2-times the low doses, respectively, keeping within the mg · g−1 · day−1 limits described above, as well as the 400-mg tablet constraint (Table 2). Patients were given identical-looking placebo tablets to match the number of tablets in each body weight group and instructed to take their drug in 2 divided doses approximately 12 hours apart.

, respectively, keeping within the mg · g−1 · day−1 limits described above, as well as the 400-mg tablet constraint (Table 2). Patients were given identical-looking placebo tablets to match the number of tablets in each body weight group and instructed to take their drug in 2 divided doses approximately 12 hours apart. Primary Objective and Outcome The primary efficacy objective was to assess the proportion of patients in each dose group that achieved treatment success after 6 weeks of treatment with mesalamine using the validated PUCAI. Treatment success was defined as either a complete response (a PUCAI score of <10) or a partial response (defined by a reduction in the PUCAI score of ≥20 points from baseline to week 6/withdrawal, but with a week 6/withdrawal absolute PUCAI score of ≥10). Secondary Objectives and Outcomes Secondary efficacy objectives included assessment of the proportion of patients who achieved PUCAI-defined complete and partial responses, and truncated Mayo Score treatment success (defined as either a complete response with stool frequency and rectal bleeding scores = 0; or a partial response with an improvement from baseline of either stool frequency or rectal bleeding, with no worsening of the other parameter). In addition, efficacy was measured as the proportion of patients for whom the investigator declared improvement at week 6/withdrawal using the global assessment of change of disease activity question.

h an improvement from baseline of either stool frequency or rectal bleeding, with no worsening of the other parameter). In addition, efficacy was measured as the proportion of patients for whom the investigator declared improvement at week 6/withdrawal using the global assessment of change of disease activity question. Biomarker endpoints included mean change from baseline and the proportion of patients who had a reduction in the fecal lactoferrin and calprotectin from baseline to weeks 3 and 6/withdrawal. Safety was assessed by monitoring adverse events (AEs), compliance, and tolerability (eg, patient withdrawals, AEs); and changes in vital signs, clinical laboratory test results, and standardized and replicated body weight and height. Study Populations The modified intent-to-treat (MITT) population was used for all efficacy analyses and included all of the patients who were randomized and took ≥1 dose of study medication. Patients in the MITT population were analyzed based on the mesalamine dose level to which they were randomized (high vs low), regardless of the treatment that they actually received during the study. If a patient was withdrawn for safety or efficacy reasons, he or she was counted as a treatment failure for the MITT analysis. Voluntary withdrawals without outcome data were not included in the MITT analysis. The evaluation of safety was based on the safety population, that is, those who were randomized and received ≥1 dose of study medication. For safety assessments patients were reported according to the dose they actually received.

alysis. Voluntary withdrawals without outcome data were not included in the MITT analysis. The evaluation of safety was based on the safety population, that is, those who were randomized and received ≥1 dose of study medication. For safety assessments patients were reported according to the dose they actually received. Statistical Analyses The primary categorical efficacy endpoint was analyzed using the Cochran-Mantel-Haenszel test to compare mesalamine dose levels (high vs low), adjusting for body weight group and disease severity. The test was 2-sided at the α = 0.05 level. Continuous endpoints, that is, change from baseline to weeks 3 and 6/withdrawal in fecal lactoferrin and calprotectin levels, were analyzed using analysis of variance with mesalamine dose level (high vs low), body weight group, and disease severity as main effects. Interaction effects and the baseline value as a covariate were assessed for significance and included in the model as appropriate. No formal statistical analyses were carried out on any secondary endpoints except for the mean change in fecal biomarkers. Data for these endpoints were summarized appropriately for the evaluable population using descriptive statistics and frequency counts. Safety data were summarized for the study population.

opriate. No formal statistical analyses were carried out on any secondary endpoints except for the mean change in fecal biomarkers. Data for these endpoints were summarized appropriately for the evaluable population using descriptive statistics and frequency counts. Safety data were summarized for the study population. RESULTS Patient Demographics Of 100 patients assessed for eligibility, 83 were randomized from December 2008 to March 2011; 16 patients were excluded for not meeting inclusion criteria and 1 patient declined to participate. One patient in the mesalamine high-dose group withdrew consent before dosing. Of the remaining 82 patients, 41 were randomized to each dose group comprising the MITT and safety populations. Patient demographics were similar for both dose groups (supplementary Table S2). One patient in the high-dose group withdrew voluntarily from the study without collecting clinical efficacy outcome data (supplementary Fig. S2).

2 patients, 41 were randomized to each dose group comprising the MITT and safety populations. Patient demographics were similar for both dose groups (supplementary Table S2). One patient in the high-dose group withdrew voluntarily from the study without collecting clinical efficacy outcome data (supplementary Fig. S2). Overall, 40% of the patients in the high-dose group had pancolitis at baseline versus 24% in the low-dose group. Nevertheless, the median PUCAI scores were similar in the low- and high-dose groups at 30 and 35 points, respectively. Approximately two-thirds of patients in each group had an endoscopy performed within 6 weeks of entering the study. Approximately 60% of all patients were newly diagnosed as having ulcerative colitis, and the median time from ulcerative colitis diagnosis was 1.1 and 2.2 months in the low- and high-dose group, respectively. Approximately half of the patients in each group presented with mild disease. Overall, 95% and 98% of patients in the low- and high-dose groups, respectively, were ≥85% compliant with their study medication during the entire study period. Less than half of all of the patients in both groups reported exposure to oral mesalamine (or sulfasalazine) before the study entry, 18 and 16 patients in low- and high-dose groups, respectively.

the low- and high-dose groups, respectively, were ≥85% compliant with their study medication during the entire study period. Less than half of all of the patients in both groups reported exposure to oral mesalamine (or sulfasalazine) before the study entry, 18 and 16 patients in low- and high-dose groups, respectively. Efficacy Patients in the low-dose mesalamine group received 27 to 71 mg · g−1 · day−1 compared with 53 to 118 mg · g−1 · day−1 in the high-dose group. A total of 23 of 41 (56.1%) and 22 of 40 (55.0%) patients achieved PUCAI-defined treatment success in the low- and high-dose groups, respectively (95% CI for difference −22.7 to 20.5, P = 0.924). The vast majority of patients achieved treatment success as assessed by the truncated Mayo Score (Table 3). Approximately 78% of all of the patients had improvement (significantly, moderately, or mildly improved) in disease activity at week 6/withdrawal weeks, with approximately half demonstrating clinically significant improvement as assessed by the global assessment of change of disease activity (Table 4). The reduction in fecal biomarkers, especially lactoferrin, tended to be greater and occurred in a higher proportion of patients in the high-dose group, but this did not reach statistical significance (Table 5).

Efficacy Patients in the low-dose mesalamine group received 27 to 71 mg · g−1 · day−1 compared with 53 to 118 mg · g−1 · day−1 in the high-dose group. A total of 23 of 41 (56.1%) and 22 of 40 (55.0%) patients achieved PUCAI-defined treatment success in the low- and high-dose groups, respectively (95% CI for difference −22.7 to 20.5, P = 0.924). The vast majority of patients achieved treatment success as assessed by the truncated Mayo Score (Table 3). Approximately 78% of all of the patients had improvement (significantly, moderately, or mildly improved) in disease activity at week 6/withdrawal weeks, with approximately half demonstrating clinically significant improvement as assessed by the global assessment of change of disease activity (Table 4). The reduction in fecal biomarkers, especially lactoferrin, tended to be greater and occurred in a higher proportion of patients in the high-dose group, but this did not reach statistical significance (Table 5). Safety Treatment-emergent AEs (TEAEs) in the safety population were reported in 23 patients (56.1%) in the low-dose group and in 21 patients (51.2%) in the high-dose group (Table 6). TEAEs occurring in ≥5% of patients were exacerbation of ulcerative colitis, nasopharyngitis, headache, dizziness, and sinusitis in the low-dose group, and nasopharyngitis, fatigue, and pyrexia in the high-dose group. The number and percentage of patients who withdrew from the study because of AEs were 5 patients (12.2%) reporting 6 AEs in the low-dose group, and 2 patients (4.9%) reporting 3 AEs in the high-dose group (supplementary Table S3). In addition to the 3 cases of ulcerative colitis flare in the low-dose group, 3 other AEs were reported to result in early discontinuation in the low-dose group: adenovirus infection, sclerosing cholangitis, or pancreatitis. In the high-dose group, 1 patient withdrew because of increased serum amylase and lipase levels and 1 patient withdrew because of upper abdominal pain.

n the low-dose group, 3 other AEs were reported to result in early discontinuation in the low-dose group: adenovirus infection, sclerosing cholangitis, or pancreatitis. In the high-dose group, 1 patient withdrew because of increased serum amylase and lipase levels and 1 patient withdrew because of upper abdominal pain. The majority of AEs were classified as mild or moderate. Of the TEAEs reported, 2 (4.0%) were severe in the low-dose group and 7 (17.1%) were severe in the high-dose group. No deaths occurred during the study.

n the low-dose group, 3 other AEs were reported to result in early discontinuation in the low-dose group: adenovirus infection, sclerosing cholangitis, or pancreatitis. In the high-dose group, 1 patient withdrew because of increased serum amylase and lipase levels and 1 patient withdrew because of upper abdominal pain. The majority of AEs were classified as mild or moderate. Of the TEAEs reported, 2 (4.0%) were severe in the low-dose group and 7 (17.1%) were severe in the high-dose group. No deaths occurred during the study. AEs suggestive of salicylate toxicity, as well as events involving the kidneys, liver, heart, pancreas, stomach, and gallbladder, were carefully considered. No case of tinnitus was reported during the study. Pancreatitis, considered possibly related to the study drug, and sclerosing cholangitis, considered doubtfully related to the study drug, were reported in 1 patient each in the low-dose group; each of these events led to the patient's discontinuation from the study. Increased serum amylase and lipase levels were reported in 1 patient in the low-dose group who also had elevated lipase levels at screening; this patient was terminated from the study. Increased lipase was also reported in 1 patient in the high-dose group who had elevated lipase levels at screening. Increased alanine transaminase levels were reported in 1 patient in the high-dose group who also had elevated lipase levels at screening. Bilirubinuria, without elevated serum bilirubin levels, was reported in 1 patient in the high-dose group. No clinically relevant trends in changes in laboratory test values, including serum creatinine, were observed during the study that could point toward drug-related renal toxicity.

evated lipase levels at screening. Bilirubinuria, without elevated serum bilirubin levels, was reported in 1 patient in the high-dose group. No clinically relevant trends in changes in laboratory test values, including serum creatinine, were observed during the study that could point toward drug-related renal toxicity. DISCUSSION The study confirmed that oral, delayed-release mesalamine is an efficacious medication in children with mild-to-moderately active ulcerative colitis, and that it resulted in a clinically significant improvement in half of the patients treated for 6 weeks. The study, however, did not demonstrate a difference in efficacy or tolerability of therapy with either high- or low-dose mesalamine. The study was prospectively powered for efficacy assuming that the low-dose response rate resembled a placebo. A 2-sided α = 0.05 Fisher exact test had an estimated power of 0.74 to detect a clinically meaningful difference between P (response on low dose) = 0.20 and P (response on high dose) = 0.50, with 40 patients per dose level. In retrospect, the expected low-dose response may have been too great in patients with mild-to-moderate disease activity, and a larger sample size may be necessary to demonstrate a difference between low- and high-dose mesalamine in this population.

and P (response on high dose) = 0.50, with 40 patients per dose level. In retrospect, the expected low-dose response may have been too great in patients with mild-to-moderate disease activity, and a larger sample size may be necessary to demonstrate a difference between low- and high-dose mesalamine in this population. Owing to the limitations imposed by having only a 400-mg tablet of mesalamine available, overlap between the 2 groups was unavoidable. The recommended dose in the joint European Crohn's and Colitis Organization and European Society for Pediatric Gastroenterology, Hepatology, and Nutrition guideline was 60 to 80 mg · g−1 · day−1(1). The response of the low-dose group may have been favorably affected because some patients were receiving therapeutic doses. Conversely, response to treatment in the high-dose group may have been adversely affected because some patients were receiving doses below what is recommended. Having formulations specifically for children, which allow for more accurate body weight dosing in the recommended therapeutic range, may demonstrate better efficacy for delayed-release mesalamine. The present study is limited owing to the lack of a placebo control group and the overlap in dosing between the 2 groups. Placebo-controlled trials in children with active disease have more than a minimal risk and would not provide any benefit to patients. For these reasons, approval of a placebo-controlled trial by an institutional review board is highly unlikely.

of a placebo control group and the overlap in dosing between the 2 groups. Placebo-controlled trials in children with active disease have more than a minimal risk and would not provide any benefit to patients. For these reasons, approval of a placebo-controlled trial by an institutional review board is highly unlikely. Some may argue that the full Mayo Score should have been used in this study; however, the truncated Mayo Score accurately determined disease activity in 3 of the 4 clinimetric properties (9). The truncated Mayo Score was used as a secondary outcome measure because it provided a useful benchmark for comparison with clinical efficacy data collected in previous placebo-controlled studies in adult patients (10,11). In these studies, the placebo response rate using the same outcome measure was approximately 30% (unpublished data). In the present study, the overall truncated Mayo Score treatment success rate (partial and complete response) was observed in 73% and 70% of patients in the mesalamine low- and high-dose groups, respectively. This compares with overall response rates (complete and partial response using the same endpoint) of 55%, 57%, and 79% in adult patients treated with mesalamine 1.6 g/day, 2.4 g/day, and 4.8 g/day, respectively. Using these parameters, children appear to have a better response to mesalamine compared with adults. This may be because some adult patients could have more established inflammation with more fibrosis, which may not be as responsive to anti-inflammatory medication.

ne 1.6 g/day, 2.4 g/day, and 4.8 g/day, respectively. Using these parameters, children appear to have a better response to mesalamine compared with adults. This may be because some adult patients could have more established inflammation with more fibrosis, which may not be as responsive to anti-inflammatory medication. The present study suggests that the effective difference between the low and high doses of oral, delayed-release mesalamine in a pediatric population is small, if any. Although large dose-response studies resembling those in adults are unlikely to be carried out in the pediatric population, a dose-response outcome may become more apparent in certain subgroups of pediatric patients, similar to what was found in adult patient studies (8,12,13).

a pediatric population is small, if any. Although large dose-response studies resembling those in adults are unlikely to be carried out in the pediatric population, a dose-response outcome may become more apparent in certain subgroups of pediatric patients, similar to what was found in adult patient studies (8,12,13). In this study, fecal lactoferrin and calprotectin were investigated as biomarkers of inflammation. The biomarker data suggest a numerical trend toward higher efficacy at the high dose versus the low dose. The decrease in mean fecal calprotectin appears to support its utility as a noninterventional outcome measure in children with ulcerative colitis. Xiang et al (14) in their study used fecal markers to evaluate the response to treatment with promising results. Among 27 patients with ulcerative colitis and 11 with Crohn disease, 97% of patients had higher calprotectin levels, which normalized with treatment (14). Similarly Wagner et al (15) evaluated the utility of fecal calprotectin and reported significant decreases in levels with treatment. The data regarding the use of fecal lactoferrin and calprotectin as biomarkers should be interpreted with caution because the demonstrated trends could be driven by baseline differences and not reflect a response to medication. Additional studies should be powered to investigate the predictability of fecal lactoferrin and calprotectin in determining response to therapy.

calprotectin as biomarkers should be interpreted with caution because the demonstrated trends could be driven by baseline differences and not reflect a response to medication. Additional studies should be powered to investigate the predictability of fecal lactoferrin and calprotectin in determining response to therapy. The study identified no dose-related patterns of AEs. Pancreatitis is a known complication of mesalamine in adults and is a plausible drug-related adverse effect in children. Because this study allowed for mesalamine treatment for >7 days before study entry, the patients who had elevated pancreatic enzymes before enrollment could have had mesalamine-related pancreatitis before study entry. As with the other safety measures in this study cohort, the number of patients with pancreatitis or elevation of pancreatic enzymes was not higher in the high-dose group. This finding is consistent with the postulated idiosyncratic nature of mesalamine-induced pancreatitis (16). The safety of high-dose mesalamine, up to 117 mg · g−1 · day−1, in this limited pediatric patient population supports the safety of escalating doses in patients who may benefit by higher doses to achieve response. In conclusion, oral, delayed-release mesalamine is an effective treatment in children with mild-to-moderately active ulcerative colitis treated for 6 weeks. Low and high doses of delayed-release mesalamine were similarly effective, and both doses were generally well tolerated, with only 8.5% of patients discontinuing treatment owing to an AE.

ayed-release mesalamine is an effective treatment in children with mild-to-moderately active ulcerative colitis treated for 6 weeks. Low and high doses of delayed-release mesalamine were similarly effective, and both doses were generally well tolerated, with only 8.5% of patients discontinuing treatment owing to an AE. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Acknowledgments The authors thank Tam Vo, PhD, Marlene Knippenberg, PhD, and James O’Reilly, PhD, from Excerpta Medica for providing editorial and writing assistance in the preparation of this article. www.clinicaltrials.gov registration number: NCT00713310. Financial support for the study and for writing and editorial services was provided by Warner Chilcott (US) LLC. The opinions and information in this article are those of the authors, and do not represent the views and/or policies of the US Food and Drug Administration.

www.clinicaltrials.gov registration number: NCT00713310. Financial support for the study and for writing and editorial services was provided by Warner Chilcott (US) LLC. The opinions and information in this article are those of the authors, and do not represent the views and/or policies of the US Food and Drug Administration. H.S.W.: consultant for Janssen Pharmaceuticals, Prometheus Laboratories, Mead Johnson, Salix, Shire, and AstraZeneca; grant support from Janssen Pharmaceuticals, Prometheus Laboratories, UCB Pharmaceuticals, Autism Research Institute, Pediatric IBD Foundation, and Nutricia. P.K.: employee of Procter&Gamble Pharmaceuticals and Warner Chilcott at the time of the study. M.B.H.: grant support from Shire, Salix Pharmaceuticals, Procter&Gamble Pharmaceuticals, UCB Pharmaceuticals, Janssen Pharmaceuticals, CCFA, and NIH (DK060617). M.K.: scientific board member of Mead Johnson Nutrition research grant “ALERNI Education Programme.” S.K.: sponsored research for Chr. Hansen; lectures provided for Abbott, Arla Foods, BioGaia, JGL, Nestlé, Nutricia, and MSD; grant support from FALK, Abbott, BioGaia, and Nestlé. J.A.Q.: consultant for Sigma-Tau, Prometheus Laboratories, and Santarus. B.Y.: employee of Procter&Gamble Pharmaceuticals at time of study; speakers bureau for Optimer, Santarus, and Forest; consultant to N-8, GlaxoSmithKline, Sucampo, and Procter&Gamble Pharmaceuticals; and grant funding (ISP) from Merck. The remaining authors report no conflicts of interest. TABLE 1 Truncated Mayo Score for rectal bleeding and stool frequency

H.S.W.: consultant for Janssen Pharmaceuticals, Prometheus Laboratories, Mead Johnson, Salix, Shire, and AstraZeneca; grant support from Janssen Pharmaceuticals, Prometheus Laboratories, UCB Pharmaceuticals, Autism Research Institute, Pediatric IBD Foundation, and Nutricia. P.K.: employee of Procter&Gamble Pharmaceuticals and Warner Chilcott at the time of the study. M.B.H.: grant support from Shire, Salix Pharmaceuticals, Procter&Gamble Pharmaceuticals, UCB Pharmaceuticals, Janssen Pharmaceuticals, CCFA, and NIH (DK060617). M.K.: scientific board member of Mead Johnson Nutrition research grant “ALERNI Education Programme.” S.K.: sponsored research for Chr. Hansen; lectures provided for Abbott, Arla Foods, BioGaia, JGL, Nestlé, Nutricia, and MSD; grant support from FALK, Abbott, BioGaia, and Nestlé. J.A.Q.: consultant for Sigma-Tau, Prometheus Laboratories, and Santarus. B.Y.: employee of Procter&Gamble Pharmaceuticals at time of study; speakers bureau for Optimer, Santarus, and Forest; consultant to N-8, GlaxoSmithKline, Sucampo, and Procter&Gamble Pharmaceuticals; and grant funding (ISP) from Merck. The remaining authors report no conflicts of interest. TABLE 1 Truncated Mayo Score for rectal bleeding and stool frequency Rectal bleeding scale 0 No blood seen 1 Streaks of blood with stool less than half of the time 2 Obvious blood with stool most of the time 3 Blood alone passed Stool frequency scale 0 Normal stool frequency per day 1 1–2 stools greater than normal per day 2 3–4 stools greater than normal per day 3 ≥5 stools greater than normal per day TABLE 2 Mesalamine dose groups

of blood with stool less than half of the time 2 Obvious blood with stool most of the time 3 Blood alone passed Stool frequency scale 0 Normal stool frequency per day 1 1–2 stools greater than normal per day 2 3–4 stools greater than normal per day 3 ≥5 stools greater than normal per day TABLE 2 Mesalamine dose groups Body weight range, kg Mesalamine dose groups, g/day Dosing range, mg · g−1 · day−1 17 to <33 Low dose: 1.2 36–71 High dose: 2.0 61–118 33 to <54 Low dose: 2.0 37–61 High dose: 3.6 67–109 54–90 Low dose: 2.4 27–44 High dose: 4.8 53–89 TABLE 3 Efficacy outcomes Mesalamine dose groups, n (%) Low dose (n = 41) High dose (n = 40) PUCAI treatment success* 23 (56.1) 22 (55.0) PUCAI complete response 19 (46.3) 17 (42.5) PUCAI partial response 4 (9.8) 5 (12.5) Truncated Mayo Score treatment success 30 (73.2) 28 (70.0) Truncated Mayo Score complete response 14 (34.1) 17 (42.5) Truncated Mayo Score partial response 16 (39.0) 11 (27.5) *Treatment success was defined as either a complete response (PUCAI score <10) or a partial response (defined by a reduction in the PUCAI score of ≥20 points from baseline to week 6/withdrawal, but with a week 6/withdrawal absolute PUCAI score of ≥10). PUCAI = Pediatric Ulcerative Colitis Activity Index. TABLE 4 Global assessment of change of disease activity

Mesalamine dose groups, n (%) Low dose (n = 41) High dose (n = 40) PUCAI treatment success* 23 (56.1) 22 (55.0) PUCAI complete response 19 (46.3) 17 (42.5) PUCAI partial response 4 (9.8) 5 (12.5) Truncated Mayo Score treatment success 30 (73.2) 28 (70.0) Truncated Mayo Score complete response 14 (34.1) 17 (42.5) Truncated Mayo Score partial response 16 (39.0) 11 (27.5) *Treatment success was defined as either a complete response (PUCAI score <10) or a partial response (defined by a reduction in the PUCAI score of ≥20 points from baseline to week 6/withdrawal, but with a week 6/withdrawal absolute PUCAI score of ≥10). PUCAI = Pediatric Ulcerative Colitis Activity Index. TABLE 4 Global assessment of change of disease activity Mesalamine dose groups, n (%) Low dose (n = 41) High dose (n = 40) Significantly improved 19 (46.3) 22 (55.0) Moderately improved 8 (19.5) 4 (10.0) Mildly improved 4 (9.8) 6 (15.0) No change 2 (4.9) 4 (10.0) Mildly worse 4 (9.8) 1 (2.5) Moderately worse 2 (4.9) 1 (2.5) Significantly worse 2 (4.9) 2 (5.0) TABLE 5 Analysis of change from baseline in fecal biomarker levels by visit

= 40) Significantly improved 19 (46.3) 22 (55.0) Moderately improved 8 (19.5) 4 (10.0) Mildly improved 4 (9.8) 6 (15.0) No change 2 (4.9) 4 (10.0) Mildly worse 4 (9.8) 1 (2.5) Moderately worse 2 (4.9) 1 (2.5) Significantly worse 2 (4.9) 2 (5.0) TABLE 5 Analysis of change from baseline in fecal biomarker levels by visit Mesalamine dose groups Biomarker Low dose (n = 41) High dose (n = 41) P* Fecal lactoferrin, μg/g Week 3 n = 33 n = 33 0.0667 Baseline mean (±SE) 505 (± 132) 583 (± 179) Mean change (±SE) 22 (± 146) −224 (± 139) Patients with reduction, n (%) 20 (60.6) 20 (60.6) Week 6 n = 30 n = 30 0.1537 Baseline mean (±SE) 353 (± 85) 582 (± 192) Mean change (±SE) 105 (± 142) −176 (± 84) Patients with reduction, n (%) 17 (56.7) 21 (70.0) Fecal calprotectin, μg/g Week 3 n = 33 n = 32 0.8388 Baseline mean (±SE) 1215 (± 338) 1710 (± 409) Mean change (±SE) −234 (± 218) −275 (± 435) Patients with reduction, n (%) 20 (60.6) 22 (68.8) Week 6 n = 30 n = 29 0.8142 Baseline mean (±SE) 902 (± 199) 1699 (± 451) Mean change (±SE) −189 (± 216) −762 (± 483) Patients with reduction, n (%) 16 (53.3) 22 (75.9) *P values for mean change difference between the 2 dose groups based on Koch nonparametric analysis of covariance test with fixed effects for treatment, bodyweight group, and disease severity. SE = standard error of the mean. TABLE 6 Overall treatment-emergent AE profile of mesalamine (safety population; N = 82)

Mesalamine dose groups Biomarker Low dose (n = 41) High dose (n = 41) P* Fecal lactoferrin, μg/g Week 3 n = 33 n = 33 0.0667 Baseline mean (±SE) 505 (± 132) 583 (± 179) Mean change (±SE) 22 (± 146) −224 (± 139) Patients with reduction, n (%) 20 (60.6) 20 (60.6) Week 6 n = 30 n = 30 0.1537 Baseline mean (±SE) 353 (± 85) 582 (± 192) Mean change (±SE) 105 (± 142) −176 (± 84) Patients with reduction, n (%) 17 (56.7) 21 (70.0) Fecal calprotectin, μg/g Week 3 n = 33 n = 32 0.8388 Baseline mean (±SE) 1215 (± 338) 1710 (± 409) Mean change (±SE) −234 (± 218) −275 (± 435) Patients with reduction, n (%) 20 (60.6) 22 (68.8) Week 6 n = 30 n = 29 0.8142 Baseline mean (±SE) 902 (± 199) 1699 (± 451) Mean change (±SE) −189 (± 216) −762 (± 483) Patients with reduction, n (%) 16 (53.3) 22 (75.9) *P values for mean change difference between the 2 dose groups based on Koch nonparametric analysis of covariance test with fixed effects for treatment, bodyweight group, and disease severity. SE = standard error of the mean. TABLE 6 Overall treatment-emergent AE profile of mesalamine (safety population; N = 82) Mesalamine dose groups Category Low dose (n = 41) High dose (n = 41) AEs No. people with AE, n (%) 23 (56.1) 21 (51.2) No. AEs 50 41 Serious AEs No. people with AE, n (%) 5 (12.2) 2 (4.9) No. AEs 8 3 Withdrawn because of AEs No. people with AE, n (%) 5 (12.2) 2 (4.9) No. AEs 6 3 AE severity, n (%) n = 50 n = 41 Mild 34 (68.0) 25 (61.0) Moderate 14 (28.0) 9 (22.0) Severe 2 (4.0) 7 (17.1) AE causality, n (%) n = 50 n = 41 Doubtful 38 (76.0) 29 (70.7) Possible 11 (22.0) 10 (24.4) Probable 1 (2.0) 2 (4.9) AE = adverse event.

α-1-Antitrypsin (AAT) deficiency is a metabolic-genetic disease that, in its classical and most typical form, is caused by homozygosity for the AAT mutant Z gene (SERPINA1). These individuals, so-called ZZ or “PIZZ” in World Health Organization nomenclature, occur in 1 in 2000 to 3500 births in North American and European populations. AAT deficiency is one of the most common single-gene diseases in the United States, with approximately 100,000 individuals affected, although it is widely underrecognized and most patients are undiagnosed (1–3). There is no newborn screening for AAT deficiency, nor is there any other organized or widely accepted patient identification process outside of pilot screening studies and the testing of symptomatic individuals in the context of routine medical care. This is in spite of the fact that symptomatic infants with AAT deficiency are more common than infants affected by many other conditions already represented on the expanded newborn screen. These considerations have led to an analysis of the indications for instituting newborn screening for ZZ AAT deficiency. Extensive documentation of the genetics, gene frequency, natural history, and biochemistry of AAT is found in the monograph Standards for the Diagnosis and Management of Individuals with Alpha-1 Antitrypsin Deficiency and the other references noted, although we briefly review the disease's natural history, technical data, and rationale for newborn screening as discussed by the workshop participants (1,2).

ry of AAT is found in the monograph Standards for the Diagnosis and Management of Individuals with Alpha-1 Antitrypsin Deficiency and the other references noted, although we briefly review the disease's natural history, technical data, and rationale for newborn screening as discussed by the workshop participants (1,2). AAT is a protein produced in large amounts in the liver and then secreted into serum. Its physiologic function is to inhibit neutrophil proteases when these enzymes leak from leukocytes into extracellular fluid during inflammation (1). In this way, AAT is critical in protecting host tissues, especially the elastic fibers of the lung, from nonspecific damage during infection and inflammation. The Z mutant of the AAT gene encodes the synthesis of a mutant protein, which is retained and accumulates in the liver rather than being appropriately secreted into serum. Accumulation of the Z mutant AAT protein in the liver can cause chronic liver disease, including cirrhosis and liver failure, in infants, children, and adults, whereas the decreased circulating levels of AAT significantly increase the risk of emphysematous lung disease in adults (4,5). Individuals heterozygous for 1 normal M allele and 1 disease Z allele, so-called MZ, are generally considered asymptomatic carriers, although some data indicate a possible small increase in risk for some lung and liver conditions (1,6,7).

AT significantly increase the risk of emphysematous lung disease in adults (4,5). Individuals heterozygous for 1 normal M allele and 1 disease Z allele, so-called MZ, are generally considered asymptomatic carriers, although some data indicate a possible small increase in risk for some lung and liver conditions (1,6,7). The natural history of ZZ AAT deficiency is highly variable (1). Studies indicate that approximately 20% of homozygous ZZ newborns develop symptomatic cholestatic hepatitis, although as many as 50% of ZZ infants and children are likely to have some kind of hepatic abnormality, including elevated enzymes, hepatomegaly, or nutritional problems, at some point during childhood (8). The risk of life-threatening liver disease in childhood (liver failure leading to death or transplant) is approximately 5%, according to the only unbiased cohort identified in a newborn screening study undertaken in Sweden in the 1970s (1,2,8–11); however, it is unclear whether the results from this genetically homogeneous Swedish population are fully applicable to a population such as North America with a different and likely wider array of modifier genes. This is because there are a number of presentations and complications of liver disease reported from various single-center studies that are not represented in the Swedish newborn cohort (12–14). Despite incomplete data and a lack of exact numbers, it was shown in the Swedish study and in limited US screening that a significant proportion of ZZ children, likely the majority, are asymptomatic and are unlikely to develop any severe disease until adulthood (13). Autopsy studies in adults suggest that the lifetime risk of cirrhosis may be as high as 50% and appears to increase in incidence in late adulthood (15). The risk of hepatocellular carcinoma is increased in ZZ patients, although the magnitude of the risk is unclear. ZZ children may experience asthma or recurrent infections, although emphysematous lung disease does not develop until early or middle adulthood (1,16,17). The lifetime risk of serious lung disease may be 50%, but is dramatically increased by personal smoking and secondhand cigarette smoke exposure. Presently, there are no specific treatments available for ATT-deficiency liver disease, other than standard supportive therapy for liver failure and liver transplantation.

7). The lifetime risk of serious lung disease may be 50%, but is dramatically increased by personal smoking and secondhand cigarette smoke exposure. Presently, there are no specific treatments available for ATT-deficiency liver disease, other than standard supportive therapy for liver failure and liver transplantation. Intravenous protein replacement with human plasma–derived AAT has been used for >20 years as a US Food and Drug Administration–approved treatment for the associated lung disease in adults, but it has no effect on the progression of liver disease. Only testing of targeted populations and not newborn screening is used for the detection of AAT deficiency (9). Patients with obstructive airway diseases, liver disease of unknown etiology, or therapy-resistant asthma are considered candidates for testing as recommended in a consensus statement of the European Respiratory Society, the American Thoracic Society, and the World Health Organization (WHO) (1,2). The statement recommends that all individuals with chronic obstructive pulmonary disease be tested for AAT deficiency. The rationale for this recommendation, even in older adults, is that it could identify carriers who may have at-risk family members. In addition, adults with incompletely reversible asthma, unexplained bronchiectasis, and unexplained liver disease, as well as individuals who are relatives of known affected patients should be tested.

for this recommendation, even in older adults, is that it could identify carriers who may have at-risk family members. In addition, adults with incompletely reversible asthma, unexplained bronchiectasis, and unexplained liver disease, as well as individuals who are relatives of known affected patients should be tested. More than 100 other mutations in the AAT gene have been identified, but the Z mutant is associated with the vast majority of disease (1). Some patients carry the rare null/null and Z/null genes, which are associated with adult emphysema, but these individuals do not develop liver disease. The only other mutation reviewed was the S mutant. This is a mutation thought to be equally as common as Z, but not associated with disease as is MS or SS; however, some SZ individuals may develop emphysema and some SZ individuals have been described to have liver disease. The risk of disease in SZ is thought to be considerably lower than ZZ for both lung and liver. A total of 54 SZ individuals were identified in the Swedish cohort, but none has ever developed liver disease. After review of these data, the workshop participants focused recommendations on ZZ homozygous individuals, unless or until new data on other genotypes become available.

lower than ZZ for both lung and liver. A total of 54 SZ individuals were identified in the Swedish cohort, but none has ever developed liver disease. After review of these data, the workshop participants focused recommendations on ZZ homozygous individuals, unless or until new data on other genotypes become available. Significant changes have occurred in recent years in the US legal environment for individuals with genetic disease. The Genetic Information Nondiscrimination Act (GINA) bill was passed by Congress in 2008 and took effect in 2009. It protects individuals from discrimination in health care by prohibiting health insurance providers from requiring genetic information, or the genetic information of a family member, for eligibility, coverage, underwriting, or premium-setting decisions. It also prohibits health insurance providers from using genetic information to collect with intent to make enrollment or coverage decisions or requiring that an individual or an individual's family member undergo a genetic test. If genetic information is acquired during research, then it may not be used for underwriting purposes; however, GINA does not apply to members of the US military, veterans participating in Department of Veterans Affairs programs, small companies with fewer than 15 employees, or the Indian Health Service. GINA also does not include protections from genetic discrimination in life insurance, disability insurance, or long-term-care insurance. GINA covers only an individual's predictive, presymptomatic, genetic information, and does not cover an individual if he or she has been diagnosed or shows clinical signs of a particular condition. Another major change in the medicolegal environment is the Affordable Care Act. This removes denials of coverage for a preexisting condition, but still allows variable rates to apply based on health status. The Affordable Care Act also has no effect on life insurance denials. Other changes may follow. Understanding how these new laws influence the risk–benefit ratio of newborn screening will be a major focus of future pilot studies in the United States.

ion, but still allows variable rates to apply based on health status. The Affordable Care Act also has no effect on life insurance denials. Other changes may follow. Understanding how these new laws influence the risk–benefit ratio of newborn screening will be a major focus of future pilot studies in the United States. METHODS AND MEETING OBJECTIVES The Alpha-1 Foundation convened a focused workshop in Washington, DC, to investigate the risks, benefits, costs, and feasibility of newborn screening for AAT deficiency. The workshop was led by co-chairs David Mannino, MD, R. Rodney Howell, MD, and Richard Sharp, PhD, and included physicians, scientists, representatives of health advocacy groups, federal employees, patients, and patients’ families. Face-to-face meetings took place September 17 to 18, 2008; manuscript recommendations and impact of the Affordable Care Act were reviewed in June 2010; final assessment of recommendations with implementation of the Affordable Care Act was made in December 2011 and June 2012, and the final manuscript editing was performed in June 2013. This workshop was a follow-up to a 1999 screening and detection workshop, which recommended that screening for AAT deficiency should be limited to at-risk populations, such as patients with established liver disease, families with a positive history, and patients with chronic obstructive pulmonary disease, and should not progress to newborn screening. In part, the conservative screening recommendations produced by the previous workshop resulted from concerns about genetic discrimination and other unintended consequences of identifying asymptomatic and healthy patients with AAT deficiency. The recent passage of the GINA has lessened many of these concerns, suggesting a need to revisit the pros and cons of newborn testing for AAT deficiency. The increased protections to individuals with preexisting conditions in the Affordable Care Act further support this reassessment. During the 2-day workshop, experts from multiple government organizations and academia, as well as health care providers, patients, and patient advocacy groups, discussed relevant issues. The topics explored were as follows: Is there a sufficient scientific rationale for newborn screening for AAT deficiency to justify any possible negative consequences and to address cost issues (1,2,9–11)? What steps are necessary to add a condition to the newborn screening panels of individual states?

relevant issues. The topics explored were as follows: Is there a sufficient scientific rationale for newborn screening for AAT deficiency to justify any possible negative consequences and to address cost issues (1,2,9–11)? What steps are necessary to add a condition to the newborn screening panels of individual states? What public health infrastructure exists to accommodate the possible influx of newly diagnosed patients and carriers and what would be needed to build proper follow-up and educational programs? What advocacy efforts would be required to convince states to add AAT deficiency to their newborn screening panels? Does the required technology exist to implement newborn screening for AAT deficiency?

of newly diagnosed patients and carriers and what would be needed to build proper follow-up and educational programs? What advocacy efforts would be required to convince states to add AAT deficiency to their newborn screening panels? Does the required technology exist to implement newborn screening for AAT deficiency? WORKSHOP RESULTS Is There a Sufficient Scientific Rationale for Newborn Screening for AAT Deficiency? It was determined by the workshop participants, after extensive review of the available literature, that there is insufficient knowledge of the risk–benefit ratio of newborn testing for this disease at this time (3,9–11). There is no specific treatment for AAT deficiency liver disease, which is the primary source of morbidity and mortality in children, other than liver transplant. It is not clear whether improved general care provided earlier in life, which may be 1 result of presymptomatic detection of many ZZ patients, would result in reduced morbidity from liver disease or reduced transplantation. Avoidance of smoking has clear benefits to these patients, but those benefits are decades in the future for an individual diagnosed at birth. There is no disease detected by newborn screening in which the benefits are so distant in time. A single study of a cohort of patients identified at birth in Sweden in the 1970s suggested a greatly reduced rate of smoking in adulthood by individuals identified at birth, although these data have not been reproduced in other populations (10). Studies in this same Swedish cohort also found significant increases in psychosocial stress in the families of the patients, even when these patients were asymptomatic and healthy throughout childhood. In the United States, there have been significant risks of psychosocial stress negative consequences for the detection of asymptomatic individuals with any disease because of concerns for loss of health insurance, loss of employment, and other negative financial and social outcomes; however, the increased access to care for individuals with preexisting conditions, which are promised in the Affordable Care Act, are likely to have a major impact on these considerations. The combination of GINA and the Affordable Care Act, although advantageous when considering many aspects of medical care, still leave both presymptomatic and symptomatic genetic disease communities vulnerable to reduced access to life insurance, long-term care insurance, and other options available to undiagnosed people.

e combination of GINA and the Affordable Care Act, although advantageous when considering many aspects of medical care, still leave both presymptomatic and symptomatic genetic disease communities vulnerable to reduced access to life insurance, long-term care insurance, and other options available to undiagnosed people. It was recommended that pilot studies be conducted to determine whether early detection improves outcomes and what psychosocial risks may result (see specific recommendations below). Possible themes for pilot studies could include defining the risk–benefit ratio, and analyzing the psychosocial and financial costs of early detection in this new legal environment. A pilot study in different age cohorts could be done to analyze smoking prevention and parental smoking cessation correlating to prevention of lung disease (18). A pilot study analyzing liver disease management would also be beneficial, especially in children (1,3,19).

of early detection in this new legal environment. A pilot study in different age cohorts could be done to analyze smoking prevention and parental smoking cessation correlating to prevention of lung disease (18). A pilot study analyzing liver disease management would also be beneficial, especially in children (1,3,19). What Steps Are Necessary to Add a Condition to the Newborn Screening Panels of Individual States? It was agreed that the Secretary of Health and Human Services’ Advisory Committee on Heritable Disorders in Newborns and Children (SACHDNC) nomination is the most important first step in adding a condition to the screening panel. There are 3 nomination considerations. The first is the incidence and natural history of the condition in question. Presently, incidence data for ZZ AAT deficiency in North America are not certain but can be estimated to be 1 ZZ per 2000 to 3500 live births, which is similar to or greater than the incidence of cystic fibrosis and other conditions that are currently screened for (1). As noted above, the morbidity is highly variable and, with present knowledge, not predictable for an individual patient. Many ZZ individuals are asymptomatic in childhood.

to 3500 live births, which is similar to or greater than the incidence of cystic fibrosis and other conditions that are currently screened for (1). As noted above, the morbidity is highly variable and, with present knowledge, not predictable for an individual patient. Many ZZ individuals are asymptomatic in childhood. The second consideration is cost; a cost-efficient test must be readily available. The panel's analysis suggests that a reliable test could be developed, with costs similar to other ongoing screening practices. Current analytical methodologies are already being studied for use on dried blood spots, and analysis by participants suggested that economies of scale would likely bring the cost to the range of $5 per test. The third consideration is treatment. There is no specific treatment for pediatric AAT-deficiency liver disease, other than supportive care and transplantation. Newborn testing may be a new paradigm, shifting the purpose from a pediatric medical treatment focus to an overall lifelong health, smoking-avoidance focus. The treatment plan would focus on the noninvasive, preventive intervention of smoking avoidance in the index patient and smoking cessation in household contacts. It is vital that the efficacy of these medical interventions be investigated and proven in North America. There must be specific, evidence-based follow-up plans and advice to parents on preventive methods.

ve, preventive intervention of smoking avoidance in the index patient and smoking cessation in household contacts. It is vital that the efficacy of these medical interventions be investigated and proven in North America. There must be specific, evidence-based follow-up plans and advice to parents on preventive methods. What Public Health Infrastructure Exists to Accommodate the Possible Influx of Newly Diagnosed Patients and Carriers and What Would Be Needed to Build Proper Follow-up and Educational Programs? Infrastructure does exist for care and follow-up of the conditions presently identified on state newborn screens. In many cases, these are the same centers that and clinicians who deal with pediatric AAT-deficiency liver disease; however, there are additional needs that must be addressed, such as what evidence-based advice and resources should be made available to a group of newly identified but asymptomatic patients and families. There are numerous community-based ATT deficiency support groups, at least ≥1 in each state. There are 53 clinical resource centers available in the United States, mostly based at academic medical centers, which regularly accept referrals for AAT patient evaluations. They could be expanded to include more pediatric support, although many already include a pediatric gastroenterologist. The majority of these community resources and centers have been organized privately by the Alpha-1 Foundation. Rosters of practitioners knowledgeable about and interested in AAT deficiency at these sites are kept and publicized. Biannual meetings of site representatives are organized by the Foundation, but no direct financial support is given to the centers. It is unclear whether this loose infrastructure is adequate for an influx of newly diagnosed infants. For the infrastructure to be ready for α-1 newborn screening, obstetrics and gynecology practitioners, family practitioners, general pediatricians, nurse practitioners, physician assistants, genetic counselors, and other medical professionals who would commonly interact with newborns and their families would need more education to understand the screening results and to properly inform patients and families about the condition and the care of asymptomatic patients. Depending on the screening method used, many carriers may be identified and they would need to have a support system for education.

rns and their families would need more education to understand the screening results and to properly inform patients and families about the condition and the care of asymptomatic patients. Depending on the screening method used, many carriers may be identified and they would need to have a support system for education. The question would also arise as to the notification of carriers or of individuals with indications of low levels and rare genotypes. Regional or state-level clinical expertise needs to be developed for a standardized education plan to be implemented upon diagnosis. The potential benefits of carrier detection would be possible mitigation of the small risk of lung and liver symptoms that, some data suggest, is associated with the carrier state, as well as general reproductive information to families; however, the risks of identifying up to 2% of the US population as carriers could lead to significant costs and psychosocial stress compared with the small number of individuals with health risks.

oms that, some data suggest, is associated with the carrier state, as well as general reproductive information to families; however, the risks of identifying up to 2% of the US population as carriers could lead to significant costs and psychosocial stress compared with the small number of individuals with health risks. What Advocacy Efforts Would Be Required to Convince States to Add AAT Deficiency to Their Newborn Screening Panels? It was agreed that the first step is to achieve the support of the SACHDNC, which will increase the chances that states adopt screening for AAT deficiency. Using the Newborn Screening Saves Lives Act, states will be eligible for grant funds once they adopt the recommendations of the SACHDNC. Partnerships with professional associations, such as the American Academy of Pediatrics, the Association of Public Health Laboratories, the Genetic Alliance, the March of Dimes, and the American College of Medical Genetics, would help with legislation support.

t funds once they adopt the recommendations of the SACHDNC. Partnerships with professional associations, such as the American Academy of Pediatrics, the Association of Public Health Laboratories, the Genetic Alliance, the March of Dimes, and the American College of Medical Genetics, would help with legislation support. Does the Required Technology Exist to Implement Newborn Screening for AAT Deficiency? It was determined that appropriate technology is available (see diagnostic discussion in reference (1)). It was decided that the objective of newborn screening would be to identify ZZ individuals only and not carriers (and not SZ individuals unless new pilot data becomes available) because the majority of morbidity and mortality involves ZZ homozygotes. Available practical methods include protein assays, tandem mass spectrometry, and DNA-based testing. The protein assay (enzyme-linked immunosorbent assay) is a good option because it does not detect carriers and is easily automated for high throughput; however, there are no normal levels known for newborns, and levels of AAT in serum can vary with illness and age. A well-controlled, population-based study of AAT levels in normal newborns would be a benefit to the field. Tandem mass spectrometry testing is available in laboratories, but it is not available specifically for AAT deficiency. DNA-based testing is inexpensive and both specific and sensitive. Conversely, it will detect carriers, but it will not detect null genotypes.

levels in normal newborns would be a benefit to the field. Tandem mass spectrometry testing is available in laboratories, but it is not available specifically for AAT deficiency. DNA-based testing is inexpensive and both specific and sensitive. Conversely, it will detect carriers, but it will not detect null genotypes. Taking into account recent legislative and technological developments, as well as medical advances in the science of ATT deficiency and the work performed by the Health and Human Services–appointed SACHDNC, workshop participants concluded that there is insufficient evidence to support expanded newborn screening for this disease at the present time; however, it was recognized that the certainty to develop a final recommendation would require new data that has not yet been collected. To develop the knowledge base required to assess the appropriateness of adding AAT deficiency to state newborn screening panels, workshop participants recommended that a number of pilot studies be undertaken. For example:A pilot study to explore ethical issues related to newborn screening for AAT deficiency, such as best practices for informing families of test results, managing psychosocial and financial costs of early detection, returning ambiguous test results and information on carrier status, and questions of misattributed paternity

pilot study to explore ethical issues related to newborn screening for AAT deficiency, such as best practices for informing families of test results, managing psychosocial and financial costs of early detection, returning ambiguous test results and information on carrier status, and questions of misattributed paternity A pilot study to explore the following issues related to providing sufficient scientific rationale for newborn screening for AAT deficiency: whether early detection improves liver outcomes, defining risk–benefit ratio, smoking prevention and parental smoking cessation, and liver disease management A pilot study to identify the best methodologies to implement a newborn testing program for AAT deficiency, to develop a family support system, and to determine the impact on pediatric liver care Workshop participants also made the following general recommendations about the addition of AAT deficiency to state newborn screening panels:Obtaining official SACHDNC nomination is a critical step in adding a new disease to any newborn screening panel. Targeting SACHDNC for approval, partnerships should be pursued with other professional associations including but not limited to the American College of Medical Genetics. Screening of newborn for AAT should be targeted to find only ZZ individuals and not carriers or SZ at the present time.

Workshop participants also made the following general recommendations about the addition of AAT deficiency to state newborn screening panels:Obtaining official SACHDNC nomination is a critical step in adding a new disease to any newborn screening panel. Targeting SACHDNC for approval, partnerships should be pursued with other professional associations including but not limited to the American College of Medical Genetics. Screening of newborn for AAT should be targeted to find only ZZ individuals and not carriers or SZ at the present time. Testing program infrastructure should include expanded pediatric support in clinical resource centers, education on AAT deficiency for obstetrics and gynecology and other physicians, and development of regional or state-level clinical expertise for a standardized education plan. Producing a decision model paper should determine the total lifetime cost of diagnosing 1 person with AAT deficiency.

Testing program infrastructure should include expanded pediatric support in clinical resource centers, education on AAT deficiency for obstetrics and gynecology and other physicians, and development of regional or state-level clinical expertise for a standardized education plan. Producing a decision model paper should determine the total lifetime cost of diagnosing 1 person with AAT deficiency. WORKSHOP CONCLUSIONS Given the demonstrated and perceived benefits and risks of newborn testing for ZZ AAT deficiency summarized in this article, the majority of the workshop participants concurred that the potential for newborn screening should be further explored with appropriate pilot studies. The Alpha-1 Foundation will define its role in implementing such studies, possibly in partnerships with other interested entities. It is understood that a newborn screening system needs to encompass not only testing but also a comprehensive approach by building the necessary infrastructure and educating health care providers and affected families. Recommending newborn screening for ZZ AAT deficiency, in the absence of a treatment for the associated pediatric liver disease but with the justification that early diagnosis would reduce smoking and adult lung disease, would be a significant paradigm shift for the field of newborn screening. The detection of a large number of individuals who would be asymptomatic and healthy throughout childhood was also a concern. Recent passage of the GINA and the Affordable Care Act may have a major impact on reducing the psychosocial consequences of newborn screening. Study of the impact of the new US legal environment on the results of newborn screening should be an intense area of focus. The findings summarized in this report are of interest to the broader AAT-deficiency disease community, to other rare disease organizations, and to government agencies responsible for implementing and regulating newborn screening programs.

ironment on the results of newborn screening should be an intense area of focus. The findings summarized in this report are of interest to the broader AAT-deficiency disease community, to other rare disease organizations, and to government agencies responsible for implementing and regulating newborn screening programs. Supplementary Material Supplemental Digital Content Address correspondence and reprint requests to Jeffrey Teckman, MD, Saint Louis University School of Medicine, 1465 South Grand Blvd, St Louis, MO 63104 (e-mail: teckmanj@slu.edu). A list of the workshop attendees is available to view online at. The meeting was funded by the Alpha-1 Foundation. The authors report no conflicts of interest.

It is well established that early postnatal growth deficits may have adverse consequences on future growth, body composition, and neurodevelopmental, metabolic, and cardiovascular outcomes (1–3). Suboptimal growth during the preterm period may affect important phases of organ development and differentiation (4). Data suggest that higher growth velocity during neonatal intensive care unit (NICU) hospitalization is associated with a significant effect on later growth and neurodevelopment (1,5). Analysis of data from >24,000 preterm infants has highlighted that the proportion of infants whose weight is <10th percentile at discharge from the NICU increases with the degree of prematurity (5). In addition, in the short term, poor growth is associated with increased vulnerability to infection, and respiratory or intestinal disorders (6). Growth, and in particular sustained growth, is also a factor that positively impacts on the number of days the preterm infant is hospitalized in the NICU (7,8). A growth velocity for a preterm infant of 15 g · kg−1 · day−1 or more is generally regarded as acceptable growth, and is in line with intrauterine growth rate (9,10). Despite efforts to optimize parenteral and enteral feeding, growth in preterm infants is often inadequate (10–12). It is therefore important to explore interventions that can improve the quantity and quality of growth in the early neonatal period.

acceptable growth, and is in line with intrauterine growth rate (9,10). Despite efforts to optimize parenteral and enteral feeding, growth in preterm infants is often inadequate (10–12). It is therefore important to explore interventions that can improve the quantity and quality of growth in the early neonatal period. Studies of the composition of human milk have contributed to the recognition that the long-chain polyunsaturated fatty acids (LCPUFAs), docosahexaenoic acid (DHA; 22:6 n-3), and arachidonic acid (ARA; 20:4 n-6) play an important role for the growing infant. These fatty acids are components of membrane phospholipids and are required for normal growth, immune function, and visual and central nervous system development (13–16). Increasing awareness of the benefits of breast milk for premature infants have led neonatologists to use banked pasteurized breast milk (PBM; donor milk) if mother's own milk is unavailable or insufficient in quantity. Donor milk is routinely pasteurized to prevent transmission of pathogens. In many NICUs, pasteurization of mother's own milk is also present practice; however, pasteurization destroys important biologically active components, such as the heat-labile bile salt–stimulated lipase (BSSL), also known as carboxyl ester lipase or bile salt–dependent lipase (17,18).

revent transmission of pathogens. In many NICUs, pasteurization of mother's own milk is also present practice; however, pasteurization destroys important biologically active components, such as the heat-labile bile salt–stimulated lipase (BSSL), also known as carboxyl ester lipase or bile salt–dependent lipase (17,18). BSSL is a lipolytic enzyme expressed in all species examined to date. It is expressed in the exocrine pancreas and secreted into the intestinal lumen and is primarily acknowledged for its function in facilitating digestion and absorption of dietary fat. BSSL is also found in blood, but its role in the circulation is less clear (19). In the gastrointestinal (GI) tract, BSSL has broad specificity and hydrolyzes a variety of different substrates, for example, tri-, di-, and monoglycerides, cholesteryl and retinyl esters, phospholipids, and ceramides (20–22). The exocrine pancreatic function is not fully developed at birth, and the BSSL production is insufficient in supporting proper fat absorption. In some species, including humans, BSSL is secreted by the lactating mammary gland and provided to the infant through the breast milk, thereby compensating for the poor endogenous production (21,23). Breast milk–derived BSSL, once activated by endogenous bile salts in the upper small intestine, contributes significantly to the efficient use of milk fat in breast-fed infants (18,24). BSSL is therefore believed to be important for the healthy growth and development of the preterm infant (21,25). A recombinant human BSSL (rhBSSL) has been developed as an oral therapeutic strategy to improve growth and lipid absorption, including LCPUFAs, in preterm infants receiving PBM or preterm formula aiming at introducing lipase activity corresponding to that in fresh breast milk.

development of the preterm infant (21,25). A recombinant human BSSL (rhBSSL) has been developed as an oral therapeutic strategy to improve growth and lipid absorption, including LCPUFAs, in preterm infants receiving PBM or preterm formula aiming at introducing lipase activity corresponding to that in fresh breast milk. Two phase II multicenter clinical studies with identical design, except for the type of infant feeding, were performed to investigate the effect of treatment with rhBSSL in preterm infants not fed fresh breast milk. These randomized double-blind placebo-controlled studies compared 7 days’ treatment with rhBSSL and placebo using a crossover design, with one study using preterm infant formula study (PF study) and the other using PBM (PBM study). The primary objective in the respective studies was to investigate the coefficient of fat absorption (CFA) in preterm infants following treatment with rhBSSL compared with placebo. Because the individual studies were not designed to have sufficient power to demonstrate an improvement in growth, a plan was prespecified to conduct analyses of the combined data with the primary objective to demonstrate that rhBSSL improves growth. Hence, the primary endpoint in the analysis of combined data was not the same as that for the individual studies. This is the first presentation of the results of these studies, exploring the effect of rhBSSL intervention on growth and fat absorption in preterm infants fed PBM or formula.

e that rhBSSL improves growth. Hence, the primary endpoint in the analysis of combined data was not the same as that for the individual studies. This is the first presentation of the results of these studies, exploring the effect of rhBSSL intervention on growth and fat absorption in preterm infants fed PBM or formula. METHODS Subjects Infants eligible for these studies were born before week 32 of gestational age (GA) and were <33 weeks of GA at the time of enrollment. Each study planned to enroll 32 infants to obtain 26 evaluable infants. The sample size estimation for the individual studies was based on an anticipated 10% difference in CFA between treatment periods and a standard deviation of 15%, with a power of 90% and a significance level of 5% (18). It was anticipated that a 10% difference in CFA would result in a 2-g · kg−1 · day−1 difference in growth velocity. The studies were conducted between April 2008 and August 2009 (PF study) in 5 centers in Italy, and between April 2008 and March 2010 (PBM study) in 5 centers in France and 2 in Italy. Infants were not included if they were to receive any fresh mother's breast milk. Nor were they included if they received parenteral nutrition, mechanical ventilation or required ≥30% oxygen, were small for GA, had conditions that may affect growth and development, had hemodynamically significant persistent ductus arteriosus, sepsis or necrotizing enterocolitis, or were treated with corticosteroids, with the exception of hydrocortisone.

d parenteral nutrition, mechanical ventilation or required ≥30% oxygen, were small for GA, had conditions that may affect growth and development, had hemodynamically significant persistent ductus arteriosus, sepsis or necrotizing enterocolitis, or were treated with corticosteroids, with the exception of hydrocortisone. Both studies were conducted according to International Conference on Harmonisation-Good Clinical Practice guidelines and the Declaration of Helsinki. They were approved by the appropriate independent ethics committees, and written informed consent was provided by the guardians of each infant. Directive 2001/20/EC: Ethical Considerations for Clinical Trials Performed in Children was consulted. Feeding Regimens In the PF study, the infants were required to receive a single preterm formula (formulation developed by Ordesa as part of the Early Nutrition Programming project) (26). Key components in the composition of the formula are listed in Table 1; a more detailed list of components is included in the online-only table. Infants receiving an alternative preterm formula before enrolment were switched to the study formula on the day of enrolment.

of the Early Nutrition Programming project) (26). Key components in the composition of the formula are listed in Table 1; a more detailed list of components is included in the online-only table. Infants receiving an alternative preterm formula before enrolment were switched to the study formula on the day of enrolment. In the PBM study, 4 centers used banked milk and 3 centers’ pasteurized the mother's own milk. Fortification of the PBM was allowed using a single fortifier (Eoprotin, Milupa, Germany) at a constant concentration throughout the study. Other milk fortifiers had to be discontinued or switched to Eoprotin at least 2 days before randomization. The fortifier provided maltodextrin, protein, vitamins, and minerals; no additional lipid supplementation was permitted. For each infant, the investigator selected a target volume between 150 and 180 mL · kg−1 · day−1 and kept the feeding volume at this target level throughout the study. The amount of formula or milk given by bottle or feeding tube was recorded.

In the PBM study, 4 centers used banked milk and 3 centers’ pasteurized the mother's own milk. Fortification of the PBM was allowed using a single fortifier (Eoprotin, Milupa, Germany) at a constant concentration throughout the study. Other milk fortifiers had to be discontinued or switched to Eoprotin at least 2 days before randomization. The fortifier provided maltodextrin, protein, vitamins, and minerals; no additional lipid supplementation was permitted. For each infant, the investigator selected a target volume between 150 and 180 mL · kg−1 · day−1 and kept the feeding volume at this target level throughout the study. The amount of formula or milk given by bottle or feeding tube was recorded. Study Design and Treatment Details Within each study, treatment was administered in a double-blind manner according to a 2-period and 2-sequence randomization scheme. The infants were randomized to receive either rhBSSL or placebo added to their feed (PF or PBM) for the first 7 days. After a 2-day washout period, the infants crossed over to the other treatment and received another 7 days of treatment (Fig. 1). The primary objective of the individual studies was to compare the fat absorption following treatment with rhBSSL or placebo when administered in PBM in 1 study and PF in the other. Secondary objectives were to compare body weight and length following treatment with rhBSSL to that with placebo.

s of treatment (Fig. 1). The primary objective of the individual studies was to compare the fat absorption following treatment with rhBSSL or placebo when administered in PBM in 1 study and PF in the other. Secondary objectives were to compare body weight and length following treatment with rhBSSL to that with placebo. FIGURE 1 Design of the 2 phase II studies on rhBSSL in preterm infants: PF study and PBM study. PBM = pasteurized breast milk; PF = preterm formula; rhBSSL = recombinant human bile-salt–stimulated lipase. The investigational product, rhBSSL (Sobi, Stockholm, Sweden), is a recombinant glycosylated form of the human BSSL, manufactured in a Chinese hamster ovary cell. The rhBSSL concentration was determined by an optimized enzyme activity assay using 4-nitrophenyl butyrate as substrate. The final product was delivered as a frozen oral solution in glass vials, at a concentration of 15 g/L and a fill volume of 1.3 mL, and was stored frozen (−25°C to −15°C) at each study center. Before administration, the frozen solution was thawed and a 0.9-mL aliquot of the rhBSSL solution was transferred to 90 mL of the preterm formula (PF study) or PBM (PBM study) to give a final concentration in the feed of 0.15 g/L, representative of the physiological concentration found in breast milk (27). Placebo (sterile water) was stored and added to the feeds in the same way.

9-mL aliquot of the rhBSSL solution was transferred to 90 mL of the preterm formula (PF study) or PBM (PBM study) to give a final concentration in the feed of 0.15 g/L, representative of the physiological concentration found in breast milk (27). Placebo (sterile water) was stored and added to the feeds in the same way. Growth Assessments The infants’ weight in grams was recorded each day using a standardized scale with an accuracy of at least ±5 g. Whenever possible, body weight was measured at approximately the same time each day. The length of the infant's leg was measured from the knee to the heel using an infant knemometer. Head circumference was not assessed in this study, because the treatment periods (7 days) were considered too short to produce measurable differences between treatments.

as measured at approximately the same time each day. The length of the infant's leg was measured from the knee to the heel using an infant knemometer. Head circumference was not assessed in this study, because the treatment periods (7 days) were considered too short to produce measurable differences between treatments. Coefficient of Fat Absorption Methodology The determination of CFA was performed during a 72-hour period toward the end of each treatment period, using a fat-balance methodology. A carmine red tracer dye was given as a GI transit marker together with a feed on day 4 of each treatment period and collection of stool commenced with the first appearance of the marker in the stool. Collection continued until the appearance of a second marker, administered 72 hours after the first one. (The stool containing the second marker was not collected.) To avoid interference with the CFA evaluation, the use of ointments was prohibited during the stool collection period. The volume of all feeds administered between administration of the GI transit markers was recorded, as was loss owing to regurgitation, spitting, and the like.

ing the second marker was not collected.) To avoid interference with the CFA evaluation, the use of ointments was prohibited during the stool collection period. The volume of all feeds administered between administration of the GI transit markers was recorded, as was loss owing to regurgitation, spitting, and the like. Diapers were shipped frozen to a central laboratory (Avantage Nutrition, Marseille, France) for analysis. After addition of a fixed amount of heptadecanoic acid as internal standard, the lipids were extracted by the Folch method (28) and analyzed by gas chromatography. The method was validated for the purpose of the study, and individual fatty acids, including DHA and ARA, were quantified following separation using an Omegawax 250 column (Supelco, Sigma-Aldrich, St Louis, MO); however, because of coelution of DHA with nervonic acid (24:1 n-9), which was only present in the breast milk, samples from infants in the PBM study were also analyzed using a SP-2380 column (Supelco) to separate and quantify the DHA. The weight of total lipid was estimated from the sum of the weights of all individual fatty acids. The same analytical principle was used to determine the lipids present in samples of the PBM and in the preterm formula to estimate fat intake during the fat-balance period.

umn (Supelco) to separate and quantify the DHA. The weight of total lipid was estimated from the sum of the weights of all individual fatty acids. The same analytical principle was used to determine the lipids present in samples of the PBM and in the preterm formula to estimate fat intake during the fat-balance period. Adverse Events Any adverse events (AEs) that occurred during the period from administration of the first dose (day 1) to the follow-up visit (1 week ± 3 days after the last dose of study drug intake) were reported, whether or not the event was considered to be treatment related. No protocol-specific laboratory values were collected. Any routine laboratory values considered by the investigator to indicate a clinically significant deviation were reported as AEs. An independent Data and Safety Monitoring Board assessed unblinded safety data in regular intervals during the study. Methodology for Statistical Analysis The number of patients in the individual studies was chosen to ensure 90% power in demonstrating an improvement in total fat absorption assuming a true difference of 10 percentage points and a common standard deviation of 15 percentage points (18). Neither of the studies was expected to have sufficient power to demonstrate an improvement in growth, because of the relatively small number of infants in each study; therefore, a predefined analysis of combined data from the 2 studies was specified, ensuring 80% power in demonstrating improvement in growth assuming a true difference in growth velocity of 2 g · kg−1 · day−1 for rhBSSL treatment over placebo.

in growth, because of the relatively small number of infants in each study; therefore, a predefined analysis of combined data from the 2 studies was specified, ensuring 80% power in demonstrating improvement in growth assuming a true difference in growth velocity of 2 g · kg−1 · day−1 for rhBSSL treatment over placebo. The statistical analysis plan detailing the analysis of combined data was developed and finalized before database lock and unblinding of the clinical database for either of the 2 clinical studies. Growth velocity, the prespecified primary endpoint for the combined analysis, was defined as the change in weight during the period divided by the weight at the beginning of the period and the number of days exposed. Both the primary efficacy outcome and the secondary efficacy outcome, CFA for total fat, were analyzed by analysis of variance models with treatment, feeding regimen (preterm formula or PBM), treatment sequence, period, and infant nested within feeding regimen and treatment sequence as factors.

number of days exposed. Both the primary efficacy outcome and the secondary efficacy outcome, CFA for total fat, were analyzed by analysis of variance models with treatment, feeding regimen (preterm formula or PBM), treatment sequence, period, and infant nested within feeding regimen and treatment sequence as factors. Additional secondary outcomes assessed in the combined analysis were the coefficients of absorption (CAs) for ARA and DHA (exploratory efficacy variables in the original studies). A post hoc analysis was made to explore the effect of rhBSSL on reducing the risk of suboptimal growth. Suboptimal growth was defined as growth of <15 g · kg−1 · day−1 during each treatment period, respectively. A logistic regression model with explanatory variables of treatment, feeding regimen, treatment sequence, and infant nested within feeding regimen and treatment sequence was used to estimate the odds ratio (OR) with corresponding 95% confidence interval, and to test the null hypothesis that the OR equals 1 against the alternative that rhBSSL reduces the risk of suboptimal growth. Safety data were summarized by descriptive statistics. The respective populations used for assessment of safety, primary/secondary clinical efficacy outcomes and CFA are detailed under the Patient Disposition section and in Table 2. Each individual study was analyzed using similar methodology.

Additional secondary outcomes assessed in the combined analysis were the coefficients of absorption (CAs) for ARA and DHA (exploratory efficacy variables in the original studies). A post hoc analysis was made to explore the effect of rhBSSL on reducing the risk of suboptimal growth. Suboptimal growth was defined as growth of <15 g · kg−1 · day−1 during each treatment period, respectively. A logistic regression model with explanatory variables of treatment, feeding regimen, treatment sequence, and infant nested within feeding regimen and treatment sequence was used to estimate the odds ratio (OR) with corresponding 95% confidence interval, and to test the null hypothesis that the OR equals 1 against the alternative that rhBSSL reduces the risk of suboptimal growth. Safety data were summarized by descriptive statistics. The respective populations used for assessment of safety, primary/secondary clinical efficacy outcomes and CFA are detailed under the Patient Disposition section and in Table 2. Each individual study was analyzed using similar methodology. RESULTS Study Population Patient Disposition A total of 65 preterm infants were randomized across the 2 clinical studies: 33 infants in the PF study and 32 infants in the PBM study. The number of infants completing and discontinuing during the treatment periods, and the reasons for discontinuation, are shown in Figure 2. FIGURE 2 Flow diagram of patients included in the 2 phase II studies. AE = adverse event; PBM = pasteurized breast milk; PF = preterm formula; rhBSSL = recombinant human bile salt–stimulated lipase.

RESULTS Study Population Patient Disposition A total of 65 preterm infants were randomized across the 2 clinical studies: 33 infants in the PF study and 32 infants in the PBM study. The number of infants completing and discontinuing during the treatment periods, and the reasons for discontinuation, are shown in Figure 2. FIGURE 2 Flow diagram of patients included in the 2 phase II studies. AE = adverse event; PBM = pasteurized breast milk; PF = preterm formula; rhBSSL = recombinant human bile salt–stimulated lipase. In total, 63 infants received at least 1 dose of study treatment and were evaluated for safety (safety analysis set) (Table 2). Sixty infants, who received at least 1 dose of the randomized study medication and had 1 baseline and at least 1 postbaseline weight assessment in both treatment periods, were evaluated for clinical efficacy (full analysis set, FAS). The CFA analysis was conducted based on data from the 46 infants for whom complete data on food intake and full stool collections during the required period were available (per-protocol [PP] analysis set). The lower number in this cohort is because of the practical difficulties encountered when performing rigorous food intake measurements and stool collections at the study sites.

ts for whom complete data on food intake and full stool collections during the required period were available (per-protocol [PP] analysis set). The lower number in this cohort is because of the practical difficulties encountered when performing rigorous food intake measurements and stool collections at the study sites. Demographics The mean GA at birth was approximately 1 week higher in the PF study compared with the PBM study (29.18 and 28.13 weeks, respectively). The mean age at randomization (first dose) was lower in the PF study (3.26 weeks) compared with the PBM study (4.39 weeks). As a result, the mean postmenstrual age on the day of randomization was similar in the 2 studies (32.45 and 32.51 weeks, respectively). The sex distribution was similar across the 2 studies (Table 3). Efficacy Outcomes Growth Velocity The predefined analysis using combined data from both studies demonstrated a statistically significant improvement in growth velocity during rhBSSL treatment as compared with placebo (P < 0.001) (Table 4). The mean growth velocity was 2.93 g · kg−1 · day−1 higher during rhBSSL treatment compared with placebo treatment, corresponding to an improvement of approximately 20%.

studies demonstrated a statistically significant improvement in growth velocity during rhBSSL treatment as compared with placebo (P < 0.001) (Table 4). The mean growth velocity was 2.93 g · kg−1 · day−1 higher during rhBSSL treatment compared with placebo treatment, corresponding to an improvement of approximately 20%. In the PF study, there was a statistically significant improvement in growth velocity of 3.74 g · kg−1 · day−1 during the rhBSSL treatment period compared with placebo (P = 0.001) in the FAS population. In the PBM study, there was an improvement of 1.95 g · kg−1 · day−1 with rhBSSL, which did not reach statistical significance (P = 0.119) (Table 4). The fat content and energy intake for infants receiving preterm formula were estimated to be 5.5 g · kg−1 · day−1 and 130 kcal · kg−1 · day−1, respectively. For the infants receiving PBM, the equivalent estimated intakes were lower: 4.0 g · kg−1 · day−1 and 105 kcal · kg−1 · day−1, respectively.

(P = 0.119) (Table 4). The fat content and energy intake for infants receiving preterm formula were estimated to be 5.5 g · kg−1 · day−1 and 130 kcal · kg−1 · day−1, respectively. For the infants receiving PBM, the equivalent estimated intakes were lower: 4.0 g · kg−1 · day−1 and 105 kcal · kg−1 · day−1, respectively. A clinically important observation was that some infants lost weight over the 7-day period when they received placebo, whereas all infants gained weight while receiving rhBSSL (Table 4). In a post hoc analysis of the combined data, rhBSSL significantly decreased the risk of suboptimal growth as compared with placebo (OR 0.19, 95% CI, 0.04–0.66, P = 0.004), with 30% of infants having suboptimal growth during rhBSSL treatment and 52% during placebo treatment. The difference in the proportion of infants having suboptimal growth was consistent across both studies, with a decrease during rhBSSL treatment compared with placebo from 42% to 21% in the PF study and from 63% to 41% in the PBM study. There was a mean increase in knee-to-heel length of 2.36 mm during rhBSSL treatment and 2.19 mm during placebo. The difference was not statistically significant (P = 0.834). There was a large within-patient variability owing to difficulties in measuring length in these small infants.

A clinically important observation was that some infants lost weight over the 7-day period when they received placebo, whereas all infants gained weight while receiving rhBSSL (Table 4). In a post hoc analysis of the combined data, rhBSSL significantly decreased the risk of suboptimal growth as compared with placebo (OR 0.19, 95% CI, 0.04–0.66, P = 0.004), with 30% of infants having suboptimal growth during rhBSSL treatment and 52% during placebo treatment. The difference in the proportion of infants having suboptimal growth was consistent across both studies, with a decrease during rhBSSL treatment compared with placebo from 42% to 21% in the PF study and from 63% to 41% in the PBM study. There was a mean increase in knee-to-heel length of 2.36 mm during rhBSSL treatment and 2.19 mm during placebo. The difference was not statistically significant (P = 0.834). There was a large within-patient variability owing to difficulties in measuring length in these small infants. CFA Fat Intake As part of the CFA analysis, the total amount of fat ingested between GI transit markers was estimated, together with the total amount of fat in the stools during the same period. The amount of fat ingested was similar during the rhBSSL and placebo treatments; the mean fat intake in the 72-hour fat-balance period during rhBSSL treatment was 4.84 g · kg−1 · day−1 compared with 4.92 g · kg−1 · day−1 for placebo (Table 5); however, there was a pronounced difference in fat intake between infants receiving formula and PBM; those fed with formula received 5.5 g fat · kg−1 · day−1 on average, whereas those fed with PBM received an average of 3.9 to 4.2 g fat · kg−1 · day−1.

1 · day−1 compared with 4.92 g · kg−1 · day−1 for placebo (Table 5); however, there was a pronounced difference in fat intake between infants receiving formula and PBM; those fed with formula received 5.5 g fat · kg−1 · day−1 on average, whereas those fed with PBM received an average of 3.9 to 4.2 g fat · kg−1 · day−1. Analysis of CFA Complete and correct milk/formula intake and stool collection were required for the accurate determination of the CFA; therefore, only infants with reported complete collection were included in this analysis (PP set, n = 46). In the combined analysis of the CFA data, there was a trend toward an improvement in fat absorption after 7 days of treatment with rhBSSL compared with placebo (mean CFA 69.1% vs 65.6%, respectively), with the mean difference in total fat absorption being 3.51% (P = 0.071, 95% CI −0.31 to 7.34) (Table 6). In both the PF and PBM studies, numerically higher mean CFAs were observed during treatment with rhBSSL compared with placebo (PF 69.5% vs 67.4, PBM 68.6% vs 63.9%, respectively).

CFA 69.1% vs 65.6%, respectively), with the mean difference in total fat absorption being 3.51% (P = 0.071, 95% CI −0.31 to 7.34) (Table 6). In both the PF and PBM studies, numerically higher mean CFAs were observed during treatment with rhBSSL compared with placebo (PF 69.5% vs 67.4, PBM 68.6% vs 63.9%, respectively). CA for Key Fatty Acids The absorption of individual fatty acids in the combined analysis showed that treatment with rhBSSL resulted in significantly higher CAs for both DHA and ARA in favor of rhBSSL over placebo: 5.76% higher for DHA (P = 0.013, 95% CI 1.27–10.25) and 8.55% for ARA (P = 0.001, 95% CI 3.52–13.57) (Table 6). In the PBM study, statistically significant higher mean absorption of both DHA and ARA was observed during treatment with rhBSSL compared with placebo (DHA 78.47% vs 73.86%, P = 0.029; ARA 72.55% vs 61.97%, P = 0.012), with the mean difference of 4.62% (P = 0.029) for DHA and 10.58% (P = 0.012 for ARA. In the PF study, a trend toward higher mean absorption for DHA and ARA after rhBSSL treatment was observed that did not reach statistical significance (DHA 81.15% vs 74.31%, P = 0.079; ARA 82.96% vs 75.83%, P = 0.051). The mean difference was 6.85% (P = 0.079) for DHA and 7.14% (P = 0.051%) for ARA.

(P = 0.012 for ARA. In the PF study, a trend toward higher mean absorption for DHA and ARA after rhBSSL treatment was observed that did not reach statistical significance (DHA 81.15% vs 74.31%, P = 0.079; ARA 82.96% vs 75.83%, P = 0.051). The mean difference was 6.85% (P = 0.079) for DHA and 7.14% (P = 0.051%) for ARA. Safety and Tolerability The combined safety analysis set consisted of the 63 infants who received at least 1 dose of study medication, of whom 61 were exposed to rhBSSL. One infant discontinued treatment during rhBSSL treatment because of grade 2a necrotizing enterocolitis; the treatment was withdrawn after 3 days in the first treatment period of the PBM study. Three infants discontinued placebo owing to AEs (4.8%). The 3 AEs leading to treatment discontinuation during placebo treatment were reported by the investigator as ulcerative colitis (first period, discontinued after 7 days), septic shock (serious AE in the first period, discontinued after 7 days), and meningitis (serious AE resulting in death, occurring in the second period and treatment discontinued on the third day of the period). The AEs reported reflected those commonly observed in this patient population and had similar frequencies across treatments in individual studies as well as in the combined data. The most commonly reported events were diaper dermatitis and anemia (Table 7).

Safety and Tolerability The combined safety analysis set consisted of the 63 infants who received at least 1 dose of study medication, of whom 61 were exposed to rhBSSL. One infant discontinued treatment during rhBSSL treatment because of grade 2a necrotizing enterocolitis; the treatment was withdrawn after 3 days in the first treatment period of the PBM study. Three infants discontinued placebo owing to AEs (4.8%). The 3 AEs leading to treatment discontinuation during placebo treatment were reported by the investigator as ulcerative colitis (first period, discontinued after 7 days), septic shock (serious AE in the first period, discontinued after 7 days), and meningitis (serious AE resulting in death, occurring in the second period and treatment discontinued on the third day of the period). The AEs reported reflected those commonly observed in this patient population and had similar frequencies across treatments in individual studies as well as in the combined data. The most commonly reported events were diaper dermatitis and anemia (Table 7). DISCUSSION The 2 studies presented here described the first data investigating the use of rhBSSL treatment in preterm infants. A predefined combined analysis of the studies showed that treatment with rhBSSL in 60 preterm infants (FAS population) fed PF or PBM resulted in approximately 20% higher growth velocity compared with placebo.

2 studies presented here described the first data investigating the use of rhBSSL treatment in preterm infants. A predefined combined analysis of the studies showed that treatment with rhBSSL in 60 preterm infants (FAS population) fed PF or PBM resulted in approximately 20% higher growth velocity compared with placebo. It is believed that BSSL stimulates the growth velocity by improving lipid digestion and absorption through its ability to hydrolyze a broad variety of lipids and related substrates found in the PBM and the preterm formula. Although there was a significant increase in the absorption of ARA and DHA after rhBSSL treatment, the total fat absorption did not reach statistical significance. Additional analysis also revealed that there was no statistically significant correlation between the effect of rhBSSL on growth velocity and CFA, indicating that the mechanism of action of rhBSSL may be more complex than just causing an increase in total fat absorption. Emerging preclinical data on the effects of BSSL on overall digestive efficiency show that undigested lipids in the intestine may cause damage to the villus epithelium in the distal small intestine, which is prevented in the presence of BSSL (29). Thus, positive effects on lipid absorption and epithelial function related to BSSL may secondarily improve overall nutrient absorption. It should also be noted that determining CFA is challenging (30), especially in multicenter studies, which is reflected by the number of incomplete collections in these studies. For the accuracy of the results, it was crucial that all intakes and outputs were carefully collected and handled in a standardized way during the entire 72 hours that the balance study was ongoing. Extracting lipids for analysis from diapers was an additional difficulty.

ber of incomplete collections in these studies. For the accuracy of the results, it was crucial that all intakes and outputs were carefully collected and handled in a standardized way during the entire 72 hours that the balance study was ongoing. Extracting lipids for analysis from diapers was an additional difficulty. Growth velocity was higher in the rhBSSL-treated infants in the PF study than in the PBM study (mean 18.05 vs 15.58 g · kg−1 · day−1). A likely explanation for the greater growth velocity in infants who received preterm formula is the higher fat and energy intake from the preterm formula, with infants consuming approximately 1.5 g · kg−1 · day−1 more fat and 25 kcal · kg−1 · day−1 more energy compared with PBM. The preterm formula used for the study was a specially optimized formulation created as part of the Early Nutrition Programming project (26); it met the present European Society for Pediatric Gastroenterology, Hepatology, and Nutrition recommendation, at the time of the study, of 110–135 kcal · kg−1 · day−1(31). Despite the special formulation of the preterm formula, 42% of infants had suboptimal growth (<15 g · kg−1 · day−1) during placebo treatment; this percentage was reduced to 21% during 7 days of rhBSSL treatment.

ogy, Hepatology, and Nutrition recommendation, at the time of the study, of 110–135 kcal · kg−1 · day−1(31). Despite the special formulation of the preterm formula, 42% of infants had suboptimal growth (<15 g · kg−1 · day−1) during placebo treatment; this percentage was reduced to 21% during 7 days of rhBSSL treatment. Preterm infants have limited endogenous biosynthesis of ω-3 and ω-6 LCPUFAs (32,33), which are required for normal growth, immune function, and maturation of numerous organ systems, most important, the brain and eye (14,34,35). Low blood levels of DHA (an ω-3 fatty acid) appear to be associated with lower visual and neural maturation in both infancy and later childhood (36–38). In the present analysis of combined data, a trend toward better total fat absorption and significantly improved absorption of the particularly important LCPUFAs ARA (P = 0.001) and DHA (P = 0.013) were observed. This is in line with the observation that BSSL, compared with colipase-dependent pancreatic lipase, more efficiently releases these fatty acids from triglycerides (39). A detailed analysis on the absorption of other LCPUFAs, as well as other fatty acids, will be reported separately.

) and DHA (P = 0.013) were observed. This is in line with the observation that BSSL, compared with colipase-dependent pancreatic lipase, more efficiently releases these fatty acids from triglycerides (39). A detailed analysis on the absorption of other LCPUFAs, as well as other fatty acids, will be reported separately. rhBSSL is an enzyme (lipase), which was administered orally at physiological (breast milk) levels, mixed with either PBM or infant formula. The enzyme is delivered to the intestine in the same way as native BSSL, and exerts its effect locally in the lumen of the GI tract following activation by bile salts in the duodenum. rhBSSL is a large glycosylated protein and, as such, is not expected to be absorbed to any significant degree when administered orally, and it is protected from degradation by intestinal proteases by the bile salts (40). Based on this, no theoretical risks have been identified. The AEs reported in the 2 phase II trials reflected those commonly observed in this patient population and had similar frequencies across treatments. An increase in growth velocity of approximately 20% during the NICU stay may bring both short- and long-term clinical benefits. It may shorten the length of hospital stay (eg, the time for a 1.0-kg infant to reach 1.8 kg is reduced from 43 to 35 days if the growth velocity is increased from 14 to 17 g · kg−1 · day−1) and it may facilitate development because higher growth rates during the NICU stay have been associated with improved neurodevelopmental outcome (1,41).

th of hospital stay (eg, the time for a 1.0-kg infant to reach 1.8 kg is reduced from 43 to 35 days if the growth velocity is increased from 14 to 17 g · kg−1 · day−1) and it may facilitate development because higher growth rates during the NICU stay have been associated with improved neurodevelopmental outcome (1,41). Another important aspect in the management of preterm infants is avoidance of growth restriction. Short-term consequences may include increased vulnerability to infection and respiratory or intestinal disorders, and longer-term consequences of growth restriction in preterm infants have been reported to include impaired neurodevelopment, together with cardiovascular, metabolic, and renal concerns (6). A post hoc analysis was made to explore the effect of rhBSSL on decreasing the risk of suboptimal growth (defined as a growth of <15 g · kg−1 · day−1). A significantly lower proportion of patients had suboptimal growth during the rhBSSL treatment period (30%) compared with the placebo period (52%). The crossover study design was chosen to minimize interpatient variability as infants served as their own controls in the analysis, and to thereby reduce the number of infants required. A 2-day washout between treatment periods was judged to be sufficient to minimize any treatment carryover effects; however, because preterm infants grow quickly, there may well be a treatment sequence effect. This sequence bias was managed by the analysis of variance model analysis.

duce the number of infants required. A 2-day washout between treatment periods was judged to be sufficient to minimize any treatment carryover effects; however, because preterm infants grow quickly, there may well be a treatment sequence effect. This sequence bias was managed by the analysis of variance model analysis. In summary, results of the analyses of combined data demonstrated that 7-day treatment with rhBSSL significantly improves growth in preterm infants receiving PBM or preterm formula compared with placebo. In addition, rhBSSL improves the absorption of the clinically important LCPUFAs ARA and DHA, which are important factors for healthy infant development. rhBSSL had an AE profile similar to placebo. Because the encouraging phase II results of short-term treatment with rhBSSL presented here warrant confirmatory studies, a randomized placebo-controlled phase III study with 4-week treatment in preterm infants is ongoing (http://clinicaltrials.gov/ NCT01413581).

In summary, results of the analyses of combined data demonstrated that 7-day treatment with rhBSSL significantly improves growth in preterm infants receiving PBM or preterm formula compared with placebo. In addition, rhBSSL improves the absorption of the clinically important LCPUFAs ARA and DHA, which are important factors for healthy infant development. rhBSSL had an AE profile similar to placebo. Because the encouraging phase II results of short-term treatment with rhBSSL presented here warrant confirmatory studies, a randomized placebo-controlled phase III study with 4-week treatment in preterm infants is ongoing (http://clinicaltrials.gov/ NCT01413581). Supplementary Material Supplemental Digital Content Acknowledgments The authors thank the following investigators for their major contributions in the 2 double-blind clinical trials: Italy, Maria Paola Bellagamba (Ancona); Costantino Romagnoli, Simonetta Costa, Mario De Curtis, and Lucilla Managanozzi (Rome); Lino Chiandetti and Giovanna Verlato (Padua); Nicola Laforgia and Grazia Calderoni (Bari); France, Elsa Kermorvant (Paris); Umberto Simeoni, Véronique Millet, and Farid Boubred (Marseille); Elie Saliba and Amélie Favreau (Tours); Nathalie Montjaux (Toulouse). We also acknowledge Maria Rodríguez Palmero, Ordesa, Spain, for providing the preterm infant formula, and the assistance of a medical writer, Josie Saulter, for the initial drafting of the manuscript. Last but not least, we sincerely thank the families who consented to their preterm infants being included in these studies.

lso acknowledge Maria Rodríguez Palmero, Ordesa, Spain, for providing the preterm infant formula, and the assistance of a medical writer, Josie Saulter, for the initial drafting of the manuscript. Last but not least, we sincerely thank the families who consented to their preterm infants being included in these studies. www.clinicaltrials.gov registration no: NCT00658905 and NCT00659243. The preterm formula study was carried out partially with research funding from the European Community's Sixth Framework Program (the Early Nutrition Programming [EARNEST]) using a formula provided by Ordesa. Swedish Orphan Biovitrum (Sobi) provided the remainder of the funding for the 2 phase II studies and their analysis. O.H., A.L., C.C., V.P.C. are consultants to Sobi for the rhBSSL project. K.T., B.O., and M.V. are employed by Sobi and holders of company shares. The other authors report no conflicts of interest. TABLE 1 Composition for key components of the preterm infant formula used in the PF study Nutrients Units 100 mL 100 Kcal Energy kcal 81 kj 339 Protein g 2.3 2.8 Fat* g 4.1 5.1 MCT g — — Linoleic acid mg 689.0 851.0 α-Linolenic acid mg 62.0 77.0 ARA mg 20.5 25.3 DHA mg 14.4 17.8 Sodium mg 34.0 42.0 Potassium mg 94.0 116.0 Chloride mg 49.0 60.5 Calcium mg 105.0 129.6 Phosphorus mg 58.0 71.6 Vitamin A μg 195.0 240.0 Vitamin D μg 1.7 2.1 Vitamin E mg 3.6 4.4 Vitamin K μg 7.5 9.3 *Source: Martek oil; ARA = arachidonic acid; DHA = docosahexaenoic acid; MCT = medium-chain triglycerides; PF = preterm formula.

34.0 42.0 Potassium mg 94.0 116.0 Chloride mg 49.0 60.5 Calcium mg 105.0 129.6 Phosphorus mg 58.0 71.6 Vitamin A μg 195.0 240.0 Vitamin D μg 1.7 2.1 Vitamin E mg 3.6 4.4 Vitamin K μg 7.5 9.3 *Source: Martek oil; ARA = arachidonic acid; DHA = docosahexaenoic acid; MCT = medium-chain triglycerides; PF = preterm formula. TABLE 2 Infant disposition in combined analysis population by individual study and combined analysis [number (percentage)] PF study PBM study Total No. infants randomized 33 32 65 Safety analysis set* 33 (100.0) 30 (100.0) 63 (100.0) FAS† 33 (100.0) 27 (90.0) 60 (95.2) PP analysis set‡ 26 (78.8) 20 (66.7) 46 (73.0) FAS = full analysis set; PBM = pasteurized breast milk; PF = preterm formula; PP = per-protocol; rhBSSL = recombinant human bile-salt–stimulated lipase. *Infants who received at least 1 dose (rhBSSL/placebo). †Infants who received at least 1 dose (rhBSSL/placebo) and had a baseline and at least 1 postbaseline weight assessment in both treatment periods. ‡Infants for which complete data on food intake and stool collections were available. TABLE 3 Infant demographics in combined analysis population by individual study and combined analysis (safety analysis set)

†Infants who received at least 1 dose (rhBSSL/placebo) and had a baseline and at least 1 postbaseline weight assessment in both treatment periods. ‡Infants for which complete data on food intake and stool collections were available. TABLE 3 Infant demographics in combined analysis population by individual study and combined analysis (safety analysis set) PF study PBM study Combined analysis No. infants treated 33 30 63 GA at birth, wk Mean (SD) 29.18 (1.31) 28.13 (1.67) 28.68 (1.57) Median (range) 29.10 (25.4–31.4) 28.25 (25.1–31.0) 29.00 (25.1–31.4) Sex (%) Female 16 (51.5) 14 (46.7) 30 (47.6) Postmenstrual age at randomization, wk Mean (SD) 32.45 (0.46) 32.51 (0.54) 32.56 (0.49) Median (range) 32.60 (31.4–33.0) 32.70 (31.0–33.7) 32.70 (31.0–33.7) Age at randomization, wk Mean (SD) 3.26 (1.19) 4.39 (1.62) 3.87 (1.49) Median (range) 3.10 (1.3–7.3) 4.00 (1.9–7.6) 3.60 (1.3–7.6) Weight at randomization, g Mean (SD) 1493.5 (195.04) 1437.0 (189.61) 1466.6 (193.02) Median (range) 1450.0 (1190–1805) 1405.0 (1088–1915) 1440.0 (1088–1915) GA = gestational age; PF = preterm formula; PBM = pasteurized breast milk; SD = standard deviation. TABLE 4 Growth velocity (g · kg−1 · day−1) for individual studies and combined analysis (FAS)

PF study PBM study Combined analysis No. infants treated 33 30 63 GA at birth, wk Mean (SD) 29.18 (1.31) 28.13 (1.67) 28.68 (1.57) Median (range) 29.10 (25.4–31.4) 28.25 (25.1–31.0) 29.00 (25.1–31.4) Sex (%) Female 16 (51.5) 14 (46.7) 30 (47.6) Postmenstrual age at randomization, wk Mean (SD) 32.45 (0.46) 32.51 (0.54) 32.56 (0.49) Median (range) 32.60 (31.4–33.0) 32.70 (31.0–33.7) 32.70 (31.0–33.7) Age at randomization, wk Mean (SD) 3.26 (1.19) 4.39 (1.62) 3.87 (1.49) Median (range) 3.10 (1.3–7.3) 4.00 (1.9–7.6) 3.60 (1.3–7.6) Weight at randomization, g Mean (SD) 1493.5 (195.04) 1437.0 (189.61) 1466.6 (193.02) Median (range) 1450.0 (1190–1805) 1405.0 (1088–1915) 1440.0 (1088–1915) GA = gestational age; PF = preterm formula; PBM = pasteurized breast milk; SD = standard deviation. TABLE 4 Growth velocity (g · kg−1 · day−1) for individual studies and combined analysis (FAS) PF study (N = 33) PBM study (N = 27) Combined analysis (N = 60) Treatment rhBSSL Placebo rhBSSL Placebo rhBSSL Placebo LS mean (95% CI) 18.05 (16.5–19.6) 14.31 (12.8–15.8) 15.58 (13.8–17.3) 13.63 (11.9–15.4) 16.86 (15.7–18.0) 13.93 (12.8–15.1) Range 9.2–25.5 −4.5 to 23.3 7.5–26.5 −3.1 to 24.6 7.5–26.5 −4.5 to 24.6 Difference (95% CI) 3.74 (1.58–5.90) 1.95 (−0.54 to 4.43) 2.93 (1.35–4.51) P 0.001 0.119 <0.001 CI = confidence interval; FAS = full analysis set; LS mean = least squares mean; PF = preterm formula; PBM = pasteurized breast milk; rhBSSL = recombinant human bile salt–stimulated lipase.

−3.1 to 24.6 7.5–26.5 −4.5 to 24.6 Difference (95% CI) 3.74 (1.58–5.90) 1.95 (−0.54 to 4.43) 2.93 (1.35–4.51) P 0.001 0.119 <0.001 CI = confidence interval; FAS = full analysis set; LS mean = least squares mean; PF = preterm formula; PBM = pasteurized breast milk; rhBSSL = recombinant human bile salt–stimulated lipase. TABLE 5 Amount of fat intake and excretion in the stool during the 3-day fat-balance period, by individual study and combined analysis (PP analysis set) PF study PBM study Combined analysis No. infants 26 20 46 treatment rhBSSL Placebo rhBSSL Placebo rhBSSL Placebo Amount of fat ingested in food (formula or breast milk) during 3-day balance period (g ·kg−1 · day−1) Mean 5.56 5.49 3.90 4.17 4.84 4.92 Range 4.80–6.67 4.78–6.58 2.87–5.06 2.92–7.73 2.87–6.67 2.92–7.73 Total amount of fat in stool during 3-day balance period (g ·kg−1 · day−1) Mean 1.67 1.78 1.26 1.53 1.49 1.67 Range 0.71–3.27 0.37–3.55 0.21–2.72 0.36–3.19 0.21–3.27 0.36–3.55 PF = preterm formula; PBM = pasteurized breast milk; PP = per-protocol; rhBSSL = recombinant human bile salt–stimulated lipase. TABLE 6 Effect of rhBSSL and placebo on CA (%) for total fat, DHA, and ARA (PP analysis set)

PF study PBM study Combined analysis No. infants 26 20 46 treatment rhBSSL Placebo rhBSSL Placebo rhBSSL Placebo Amount of fat ingested in food (formula or breast milk) during 3-day balance period (g ·kg−1 · day−1) Mean 5.56 5.49 3.90 4.17 4.84 4.92 Range 4.80–6.67 4.78–6.58 2.87–5.06 2.92–7.73 2.87–6.67 2.92–7.73 Total amount of fat in stool during 3-day balance period (g ·kg−1 · day−1) Mean 1.67 1.78 1.26 1.53 1.49 1.67 Range 0.71–3.27 0.37–3.55 0.21–2.72 0.36–3.19 0.21–3.27 0.36–3.55 PF = preterm formula; PBM = pasteurized breast milk; PP = per-protocol; rhBSSL = recombinant human bile salt–stimulated lipase. TABLE 6 Effect of rhBSSL and placebo on CA (%) for total fat, DHA, and ARA (PP analysis set) PF study PBM study Combined data Statistics rhBSSL (N = 26) Placebo (N = 26) rhBSSL (N = 20) Placebo (N = 20) rhBSSL (N = 46) Placebo (N = 46) Total fat LS mean 69.46 67.38 68.61 63.87 69.09 65.57 LS mean difference 2.08 4.74 3.51 95% CI −3.67 to 7.84 −0.52 to 10.01 −0.31 to 7.34 P 0.462 0.075 0.071 DHA LS mean 81.15 74.31 78.47 73.86 79.83 74.07 LS mean difference 6.85 4.62 5.76 95% CI −0.84 to 14.54 0.53–8.71 1.27–10.25 P 0.079 0.029 0.013 ARA LS mean 82.96 75.83 72.55 61.97 77.60 69.06 LS mean difference 7.14 10.58 8.55 95% CI −0.03 to 14.30 2.63–18.53 3.52–13.57 P 0.051 0.012 0.001 ARA = arachidonic acid; CA = coefficient of absorption; CI = confidence interval; DHA = docosahexaenoic acid; LS mean = least squares mean; PP = per-protocol; rhBSSL = recombinant human bile salt–stimulated lipase.

69.06 LS mean difference 7.14 10.58 8.55 95% CI −0.03 to 14.30 2.63–18.53 3.52–13.57 P 0.051 0.012 0.001 ARA = arachidonic acid; CA = coefficient of absorption; CI = confidence interval; DHA = docosahexaenoic acid; LS mean = least squares mean; PP = per-protocol; rhBSSL = recombinant human bile salt–stimulated lipase. TABLE 7 Total treatment-emergent AEs occurring in >3% of infants in either treatment period (combined safety analysis set) rhBSSL, N = 61 (%) Placebo, N = 62 (%) Diaper dermatitis 13 (21.3) 13 (21.0) Anemia 3 (4.9) 6 (9.7) Bradycardia 1 (1.6) 5 (8.1) Cardiac murmur 4 (6.6) 2 (3.2) Hypokalemia 3 (4.9) 2 (3.2) Thrombocytopenia 0 3 (4.8) Anemia neonatal 2 (3.2) 1 (1.6) Urinary tract infection 1 (1.6) 2 (3.2) Anal fissure 2 (3.2) 0 Apnea 0 2 (3.2) Retinopathy of prematurity 0 2 (3.2) If an infant had >1 count for a particular preferred AE term, the infant was counted only once in any treatment period. AE = adverse event; rhBSSL = recombinant human bile salt–stimulated lipase.

Rapid growth, in particular rapid weight gain in infancy, is associated with later overweight and obesity (1–8). Although causality is not established in this association, it is nevertheless conceivable that by slowing down rapid weight gain in infancy, a reduction of the risk of later obesity may be achieved (9). Among measures that could slow down weight gain in infancy and potentially reduce the risk of later obesity, a reduction in protein intake appears promising.

this association, it is nevertheless conceivable that by slowing down rapid weight gain in infancy, a reduction of the risk of later obesity may be achieved (9). Among measures that could slow down weight gain in infancy and potentially reduce the risk of later obesity, a reduction in protein intake appears promising. Protein needs of infants decrease appreciably during the first year of life (10). During the first few months, breast milk alone and subsequently breast milk with complementary foods are presumed to meet the protein needs of infants. The protein content of human milk, which may be as high as 2.09 g/100 kcal in the first month after birth, is 1.28 g/100 kcal at 3 to 4 months (11) and approximately 1.24 g/100 kcal by 9 to 12 months. This suggests that formulas fed after 3 months should contain no less than 1.30 g/100 kcal of a high-quality protein. The lower regulatory limit for protein content of formulas (0–12 months) in the European Union (EU) and the United States is 1.8 g/100 kcal. In actuality, the protein content of formulas typically exceeds these levels, especially in the case of follow-on formulas. In the European Obesity Project (12), the protein concentration of the lower-protein formula (2.20 g/100 kcal) fed from 5 months on exceeded the lower limit in EU countries. Protein intakes in the later parts of infancy have been found in several localities to be high and to exceed required intakes (13–16). Epidemiologic evidence links high protein intakes in infancy to obesity in childhood (16–19). Also, high protein intakes have been shown in prospective studies to lead to increased weight gain and higher adiposity in infancy (12) and childhood (20). Lower protein intake from breast milk than from formula may be among the reasons why breast-fed (BF) infants are at lower risk for obesity later in life, as the preponderance of the evidence indicates (7,21–23). For all of these reasons, reducing the protein intake during infancy may reduce the risk of later obesity.

. Lower protein intake from breast milk than from formula may be among the reasons why breast-fed (BF) infants are at lower risk for obesity later in life, as the preponderance of the evidence indicates (7,21–23). For all of these reasons, reducing the protein intake during infancy may reduce the risk of later obesity. The present study tested a bovine whey-based formula with a protein content of 1.65 g/100 kcal, which is below the regulatory lower limit in Europe and the United States. Because protein levels <1.80 g/100 kcal have not been studied before, we reduced the protein level by only 9% below the regulatory level. The formula was fed after 3 months of age. The offspring of overweight and obese women are at increased risk for overweight later in life (7,24–31) and show accelerated growth already during infancy (32). Therefore, any measure that could reduce the risk of later obesity would be of particular importance for infants born to overweight mothers.

3 months of age. The offspring of overweight and obese women are at increased risk for overweight later in life (7,24–31) and show accelerated growth already during infancy (32). Therefore, any measure that could reduce the risk of later obesity would be of particular importance for infants born to overweight mothers. METHODS Study Design The study was designed to test the hypothesis that a formula with a protein content of 1.65 g/100 kcal and a caloric density of 62.8 kcal/100 mL that contains added probiotics (Lactobacillus PR and Bifidobacterium lactis) (EXPL [experimental] formula) leads to slower growth between 3 and 6 months of age than a formula with a protein content of 2.70 g/100 kcal and standard caloric density of 65.6 kcal/100 mL without probiotics (CTRL [control] formula). Furthermore, it was hypothesized that the lower-protein formula would lead to biomarkers of protein metabolism that were closer to those of BF infants. A secondary objective of the study was to establish that the formula with protein content 1.65 g/100 kcal supports normal growth. The hypothesis was tested in a double-blind trial in which formula-fed infants were at 3 months randomly assigned to 1 of the study formulas, which was to be fed exclusively until 6 months and with complementary foods until 12 months. A reference group of BF infants was studied in identical fashion. Growth was monitored until 24 months and formula consumption was determined periodically until 9 months. The study protocol was reviewed and approved by the ethical committee of Regional Health Service, IX Region de la Araucanía, Chile.

until 12 months. A reference group of BF infants was studied in identical fashion. Growth was monitored until 24 months and formula consumption was determined periodically until 9 months. The study protocol was reviewed and approved by the ethical committee of Regional Health Service, IX Region de la Araucanía, Chile. The age period from 3 to 6 months was treated as the primary study period, with growth during that period designated as the primary outcome because during that period the study formulas were the near-exclusive source of nutrients. Absence of complementary foods was considered important for an evaluation of the adequacy of the formulas. Sample Size Calculation Sample size was estimated assuming that a difference in weight gain of 2 g/day is clinically relevant. According to the World Health Organization (WHO) Child Growth Standards (33), weight gain between 3 and 6 months averages 17 g/day, with a standard deviation of approximately 4 g/day (boys and girls). To detect a difference of 2 g/day with a type I error (α) of 5% (2-sided test) and power of 80%, 64 infants were needed in each arm of the formula trial. With an expected dropout rate of approximately 30%, a total of 182 formula-fed infants needed to be enrolled. For the BF reference group similar considerations indicated that 91 infants needed to be enrolled. As the study progressed, it became clear that the actual dropout rate was <30% and enrollment was stopped when it could be expected that at least 64 infants per group would reach 6 months.

d infants needed to be enrolled. For the BF reference group similar considerations indicated that 91 infants needed to be enrolled. As the study progressed, it became clear that the actual dropout rate was <30% and enrollment was stopped when it could be expected that at least 64 infants per group would reach 6 months. Subjects Pregnant women attending the maternity facilities associated with the Universidad de La Frontera were, under the supervision of the principal investigator (J.I.), screened. If their (self-reported and/or medical records) prepregnancy body mass index (BMI) was >25 kg/m2, they were informed about the study. Medical and other information about the mothers was obtained from hospital records or from the mothers directly. At birth the infant was assessed and was offered study participation if he or she met the study inclusion criteria (birth weight >2500 g and <4800 g, gestational age ≥37 weeks and <42 weeks) and did not meet the exclusion criteria (weight <5th percentile for gestational age, maternal diabetes, including gestational-onset diabetes, >5 cigarettes per day during pregnancy, use of illicit drugs, or presence of chronic inflammatory condition). Infants with congenital illnesses or malformations that may affect growth were excluded, as were infants who required hospitalization for >2 days.

l age, maternal diabetes, including gestational-onset diabetes, >5 cigarettes per day during pregnancy, use of illicit drugs, or presence of chronic inflammatory condition). Infants with congenital illnesses or malformations that may affect growth were excluded, as were infants who required hospitalization for >2 days. Of 330 infants screened, 305 were enrolled at birth (N = 302) or between 5 and 31 days of age (N = 3). Mothers provided written informed consent at the time of enrollment. Most infants were BF when they left the hospital. All of the mothers received breast-feeding consultation during their visits in the study center. At 1.5 months of age, approximately 80% of infants were exclusively or partially BF. Whenever mothers chose to discontinue breast-feeding, a whey-based infant formula (NAN 1; protein 1.8 g/100 kcal, 67 kcal/100 mL [Nestlé Ltd, Vevey, Switzerland]) was provided. At 3 months of age, infants who were predominantly BF (no more than 1 formula feeding per day) were assigned to the BF reference group (n = 76). Predominantly formula-fed infants were randomly assigned to 1 of the study formulas, that is, to EXPL formula (n = 86) or CTRL formula (n = 86).

y, Switzerland]) was provided. At 3 months of age, infants who were predominantly BF (no more than 1 formula feeding per day) were assigned to the BF reference group (n = 76). Predominantly formula-fed infants were randomly assigned to 1 of the study formulas, that is, to EXPL formula (n = 86) or CTRL formula (n = 86). The flow of subjects in the intention-to-treat (ITT) population through 12 months is shown in Figure 1. Formula-fed infants were withdrawn from the study because mothers objected to blood draws (n = 2), infants did not accept study formulas (n = 2), and mothers wished to continue mixed feeding (formula + breast-feeding) (n = 14). An additional 12 formula-fed infants left the study for reasons unrelated to the study. Although somewhat more EXPL infants (n = 20) left the study before 6 months than CTRL infants (n = 10), there was not a single cause that explained the difference. Four BF infants were withdrawn because mothers wished to continue mixed feeding. A total of 170 infants (56 BF, 50 EXPL, and 64 CTRL) in the ITT sample were studied up to 24 months. In the per-protocol (PP) sample there were at 6 months 55 infants in EXPL and 68 in CTRL; at 12 months there were 47 in EXPL and 60 in CTRL. FIGURE 1 Flow of subjects in the intention-to-treat (ITT) population.

The flow of subjects in the intention-to-treat (ITT) population through 12 months is shown in Figure 1. Formula-fed infants were withdrawn from the study because mothers objected to blood draws (n = 2), infants did not accept study formulas (n = 2), and mothers wished to continue mixed feeding (formula + breast-feeding) (n = 14). An additional 12 formula-fed infants left the study for reasons unrelated to the study. Although somewhat more EXPL infants (n = 20) left the study before 6 months than CTRL infants (n = 10), there was not a single cause that explained the difference. Four BF infants were withdrawn because mothers wished to continue mixed feeding. A total of 170 infants (56 BF, 50 EXPL, and 64 CTRL) in the ITT sample were studied up to 24 months. In the per-protocol (PP) sample there were at 6 months 55 infants in EXPL and 68 in CTRL; at 12 months there were 47 in EXPL and 60 in CTRL. FIGURE 1 Flow of subjects in the intention-to-treat (ITT) population. Study Feedings Composition of the study formulas is indicated in supplementary Table S1. Both formulas were produced especially for the present trial. The formulas differed in protein content, which was 1.65 g/100 kcal in the EXPL formula and 2.70 g/100 kcal in the CTRL formula. Protein was provided by intact bovine milk proteins with a whey-to-casein ratio of 60:40. Details of the protein composition in the EXPL formula are presented in Table S1. Because of reported beneficial effects that probiotics confer on infants (34–36), the EXPL formula contained 2 × 107 cfu/g formula powder of each of the probiotic strains Lactobacillus PR and B lactis (Bb12); the CTRL formula did not contain probiotics. Levels of minerals, vitamins, and trace elements in the study formulas corresponded to the recommendations of the Codex Alimentarius (37) for infant and follow-up formulas. The formulas were provided in powder form. Parents were requested not to feed any other formula and not to start complementary foods until 6 months. The study formulas were provided free of charge until 12 months. The tins were labeled with color codes (2 colors for each formula); the identity of the formulas was unknown to the parents, study staff, and investigators, and was known only to the manufacturer. BF infants could at study entry (3 months) receive up to 1 formula feeding per day. Parents were requested to withhold complementary feedings until 6 months. After 6 months, whenever the mother chose to discontinue breast-feeding, a commercially available follow-up formula (NAN 2) was provided until 12 months of age that had a protein content of 2.4 g/100 kcal and a caloric density of 67 kcal/100 mL.

Parents were requested to withhold complementary feedings until 6 months. After 6 months, whenever the mother chose to discontinue breast-feeding, a commercially available follow-up formula (NAN 2) was provided until 12 months of age that had a protein content of 2.4 g/100 kcal and a caloric density of 67 kcal/100 mL. Study Procedures After enrollment infants were seen within 5 days of age 1.5 months and subsequent study visits took place within 7 days of ages 3, 4, 6, 9, and 12 months and within 14 days of age 24 months. At birth, anthropometric data were obtained from hospital records. At study visits, anthropometry was performed and an interval medical history was obtained. At the 3-month visit, formula-fed infants were randomly assigned to EXPL formula or CTRL formula using the Internet-based randomization system, TrialSys. Stratification factors were sex, ethnicity (white/other), prepregnancy BMI of the mother (25–30, >30 kg/m2), and type of feeding between 1.5 and 3 months (formula exclusively or formula and breast). At study visits, a supply of study formula was dispensed that was expected to last until the next study visit.

TrialSys. Stratification factors were sex, ethnicity (white/other), prepregnancy BMI of the mother (25–30, >30 kg/m2), and type of feeding between 1.5 and 3 months (formula exclusively or formula and breast). At study visits, a supply of study formula was dispensed that was expected to last until the next study visit. Anthropometry: During study visits, weight without clothes was determined to the nearest 10 g using calibrated electronic scales (Sartorius, Gottingen, Germany). Recumbent length was measured by 2 measurers using a measuring board with fixed head board and movable foot board (infant stadiometer) and was recorded to the nearest 1 mm. Head circumference was measured using a nonstretchable measuring tape to the nearest 1 mm. All measurements were obtained in duplicate and the average was used. Body composition measurements at 12 to 13 months were performed using a DXA Lunar Prodigy Advance with software EnCore version 13.6 (both GE Healthcare, Madison, WI). Dietary intake: Intake of formula (quantitative) and consumption of complementary foods (semiquantitatively) were recorded by the parents on hand-held diaries for 3 days before the visits at 3, 4, 6, and 9 months of age. At study visits, records were checked for completeness and ambiguities were resolved.

Anthropometry: During study visits, weight without clothes was determined to the nearest 10 g using calibrated electronic scales (Sartorius, Gottingen, Germany). Recumbent length was measured by 2 measurers using a measuring board with fixed head board and movable foot board (infant stadiometer) and was recorded to the nearest 1 mm. Head circumference was measured using a nonstretchable measuring tape to the nearest 1 mm. All measurements were obtained in duplicate and the average was used. Body composition measurements at 12 to 13 months were performed using a DXA Lunar Prodigy Advance with software EnCore version 13.6 (both GE Healthcare, Madison, WI). Dietary intake: Intake of formula (quantitative) and consumption of complementary foods (semiquantitatively) were recorded by the parents on hand-held diaries for 3 days before the visits at 3, 4, 6, and 9 months of age. At study visits, records were checked for completeness and ambiguities were resolved. Serum (plasma) biomarkers: Venous blood was obtained during visits at 3, 6, and 12 months. Blood was drawn >1.5 hours after the previous meal. Blood was drawn into heparinized (for amino acid determinations) and EDTA-containing (for ghrelin determinations) tubes or plain tubes (for all other determinations). Determinations were performed by the Central Laboratory in Temuco unless noted otherwise. Determinations of insulin growth factor-1, C-peptide, and leptin were performed by ELISA using kits from ALPCO (ALPCO Diagnostics, Salem, NH). Determination of insulin was performed by Microplate Enzymatic Immuno-Assay (Insulin Assay; Axsym, Abbott, Abbott Park, IL). Ghrelin was determined by enzyme immunometric assay (ALPCO) at the Nestlé Research Center, Lausanne, Switzerland. Amino acids were determined using the EZ:Faast-Free Amino Acid kit (Phenomenex, Torrance, CA) and gas chromatography at Fleury Laboratory, São Paulo, Brazil.

(Insulin Assay; Axsym, Abbott, Abbott Park, IL). Ghrelin was determined by enzyme immunometric assay (ALPCO) at the Nestlé Research Center, Lausanne, Switzerland. Amino acids were determined using the EZ:Faast-Free Amino Acid kit (Phenomenex, Torrance, CA) and gas chromatography at Fleury Laboratory, São Paulo, Brazil. Data Processing and Statistical Analysis Data were analyzed on an ITT basis and also on a PP basis. Results based on ITT analysis are presented unless otherwise specified. Inclusion in the PP sample required that infants consumed study formula from 3 to 6 months exclusively, which was defined as no breast-feeding indicated in the diaries at 4 and 6 months, nonstudy formula <1 bottle/week or fewer than 3 full consecutive days, and consumption of solid foods <4 teaspoons/day until 4 months and <1 serving/day thereafter. Subjects were also excluded from PP evaluation if they were hospitalized for more than 3 days or received treatment with oral corticosteroids for >15 days.

nonstudy formula <1 bottle/week or fewer than 3 full consecutive days, and consumption of solid foods <4 teaspoons/day until 4 months and <1 serving/day thereafter. Subjects were also excluded from PP evaluation if they were hospitalized for more than 3 days or received treatment with oral corticosteroids for >15 days. Weight gain (g/day) between 3 and 6 months (primary outcome variable), 6 and 12 months, 12 and 24 months, 3 and 12 months, and 3 and 24 months was calculated as the difference in weight divided by the exact number of days between measurements. Weight gain between 3 and 6 months of formula-fed infants was also calculated separately for infants with weight >75th percentile at 3 months, for infants whose mother had a BMI > 30 kg/m2, and for infants with both of these characteristics. Differences between groups were compared by t test (primary outcome only) and also by ANCOVA correcting for infant weight at 3 months, maternal BMI (kg/m2), sex (male/female), ethnicity (white/other), antibiotic use, and complementary food before 6 months. Decisions on covariates were made before the code was broken.

. Differences between groups were compared by t test (primary outcome only) and also by ANCOVA correcting for infant weight at 3 months, maternal BMI (kg/m2), sex (male/female), ethnicity (white/other), antibiotic use, and complementary food before 6 months. Decisions on covariates were made before the code was broken. z scores for weight, length, BMI, and head circumference at 3 to 24 months were derived based on the WHO 2006 Child Growth Standards (38). Baseline values (3 months) of z scores were compared using ANCOVA correcting for maternal BMI (kg/m2), sex (male/female), and ethnicity. Longitudinal comparison of weight, length, BMI, and head circumference and of the corresponding z scores (4–24 months) were performed by a mixed model to test for different time trends (39). Fixed effects were the correction factors used in the ANCOVA along with visits, treatment, and visit × treatment. Random effect was subject. The percentage of infants in EXPL and CTRL with weight >90th percentile of the WHO Standards at 3, 6, and 12 months was compared by calculating odds ratios (ORs) and 95% confidence intervals (CIs) by logistic regression.

used in the ANCOVA along with visits, treatment, and visit × treatment. Random effect was subject. The percentage of infants in EXPL and CTRL with weight >90th percentile of the WHO Standards at 3, 6, and 12 months was compared by calculating odds ratios (ORs) and 95% confidence intervals (CIs) by logistic regression. Serum biomarkers that showed lognormal distribution were analyzed after logarithmic transformation, but data are presented as arithmetic means and standard deviations. Only PP data are presented because biochemical parameters are strongly influenced by nutritional intake. Plasma amino acid concentrations were compared by ANCOVA correcting for maternal BMI. Comparisons of biomarkers were performed by ANCOVA correcting for values at 3 months. Body composition data were compared by the 2-sample t test. Intakes (volume, energy) were compared by ANCOVA correcting for intakes at 3 months. RESULTS Characteristics of mothers and infants are presented in supplementary Table S2. Study infants were born between October 2007 and September 2009. Differences between groups were small and not statistically significant, except that mothers of BF infants were less likely to smoke during pregnancy but were more likely to consume alcohol than mothers of formula-fed infants. Infant anthropometric data at birth were similar for all 3 groups.

r 2007 and September 2009. Differences between groups were small and not statistically significant, except that mothers of BF infants were less likely to smoke during pregnancy but were more likely to consume alcohol than mothers of formula-fed infants. Infant anthropometric data at birth were similar for all 3 groups. Weight gain between 3 and 6 months, the primary study outcome, was significantly lower in EXPL than in CTRL. The difference was −1.77 g/day (95% CI −3.29 to −0.24, P = 0.024, t test) in the ITT sample (Table 1, top panel). In the PP sample (data not shown) the difference was −1.85 g/day (95% CI −3.46 to −0.24, P = 0.025). After correcting for covariates by ANCOVA, the difference remained significant for the period 3 to 6 months (lower panel of Table 1). The difference in weight gain between 3 and 6 months was almost entirely explained by the strong effect in infants whose weight was >75th percentile at 3 months. In that subgroup (EXPL N = 22, CTRL N = 28), the difference was −3.70 g/day (95% CI −6.69 to −0.70, P = 0.016). In the subgroup of infants whose mothers had a BMI > 30 kg/m2 (EXPL N = 21, CTRL N = 21), the difference was −4.21 g/day (95% CI −7.75 to −0.81, P = 0.017). In infants with both characteristics (EXPL N = 6; CTRL N = 7), the difference was as large as −8.34 g/day (CI −16.05 to −0.63, P = 0.035). Weight gain of EXPL was not significantly different from that of BF throughout the study, but weight gain of CTRL was higher than that of BF for most time intervals (Table 1).

017). In infants with both characteristics (EXPL N = 6; CTRL N = 7), the difference was as large as −8.34 g/day (CI −16.05 to −0.63, P = 0.035). Weight gain of EXPL was not significantly different from that of BF throughout the study, but weight gain of CTRL was higher than that of BF for most time intervals (Table 1). Weight predicted by the longitudinal analysis is shown in Figure 2. Overall, in the longitudinal analysis the difference in weight between EXPL and CTRL was statistically significant in both the ITT and the PP samples (P = 0.022). Weight-for-age z scores (Table 2) of EXPL and CTRL were not significantly different at 3 months, but z scores of BF were significantly higher. After 3 months weight-for-age z scores began to differ between EXPL and CTRL; the difference was statistically significant at 9 months and continued to increase until 24 months. The differences in z scores at 12 and 24 months corresponded to weight differences of 316 and 446 g, respectively. Differences between EXPL and BF between 6 and 24 months were not significant. On the contrary, differences between CTRL and BF were statistically significant between 9 and 24 months of age. The percentage of infants whose weight was >90th percentile of the WHO standards was similar in EXPL and CTRL at 3 months (OR 1.0, 95% CI 0.4–2.5) but at 6 and 9 months it was significantly lower in EXPL than in CTRL. The percentages at 6 months were 10.6% in EXPL and 22.4% in CTRL (OR 5.3, 95% CI 1.2–23.5) and at 12 months were 18.5% and 31.8% (OR 3.6, 95% CI 1.1–11.2). The number needed to treat was 8, meaning that 8 infants would need to be fed the EXPL formula to prevent 1 formula-fed infant from exceeding the 90th percentile at 6 or 12 months.

at 6 months were 10.6% in EXPL and 22.4% in CTRL (OR 5.3, 95% CI 1.2–23.5) and at 12 months were 18.5% and 31.8% (OR 3.6, 95% CI 1.1–11.2). The number needed to treat was 8, meaning that 8 infants would need to be fed the EXPL formula to prevent 1 formula-fed infant from exceeding the 90th percentile at 6 or 12 months. FIGURE 2 Weight at 4 to 24 months (ITT population) as predicted by the longitudinal analysis using the mixed model (39). Vertical bars indicate SE. EXPL vs CTRL P = 0.011; BF vs CTRL P < 0.001; BF vs EXPL nonsignificant. CTRL = control; EXPL = experimental; ITT = intention-to-treat; SE = standard error. Length-for-age z scores (supplementary Table S3) were at 3 months significantly lower in the 2 formula groups compared with the BF group. Subsequently, with correction for length at 3 months, differences were not significant between 6 and 24 months. Mean head circumference z scores (data not shown) were above the corresponding WHO standards at all ages for all groups and standard deviations were <1. Differences in head circumference z scores between EXPL and BF were not significant. Length and head circumference, as predicted by the longitudinal analysis, showed no differences between the 3 groups (data not shown).

re above the corresponding WHO standards at all ages for all groups and standard deviations were <1. Differences in head circumference z scores between EXPL and BF were not significant. Length and head circumference, as predicted by the longitudinal analysis, showed no differences between the 3 groups (data not shown). BMI predicted by the longitudinal analysis (Fig. 3) was significantly lower in EXPL and BF than in CTRL (EXPL vs CTRL P = 0.027, PP). BMI of EXPL and BF did not differ significantly and, at 24 months, were almost identical. BMI-for-age z scores at 3 months (supplementary Table S4) were not significantly different in EXPL compared with CTRL; however, at subsequent ages, BMI z scores were lower in EXPL than in CTRL; and at 12 and 24 months, the difference reached statistical significance. Also, at 12 and 24 months, BMI z scores of BF were significantly lower than those of CTRL. Notably, mean BMI z scores of all 3 groups were greater than zero and increased between 3 and 24 months, indicating progressively greater BMI compared with the WHO standards. FIGURE 3 BMI at 4 to 24 months (ITT population) as predicted by the longitudinal analysis using the mixed model (39). Vertical bars indicate SE. EXPL vs CTRL P = 0.027; BF vs CTRL P < 0.001; BF vs EXPL nonsignificant. BMI = body mass index; CTRL = control; EXPL = experimental; ITT = intention-to-treat; SE = standard error.

RE 3 BMI at 4 to 24 months (ITT population) as predicted by the longitudinal analysis using the mixed model (39). Vertical bars indicate SE. EXPL vs CTRL P = 0.027; BF vs CTRL P < 0.001; BF vs EXPL nonsignificant. BMI = body mass index; CTRL = control; EXPL = experimental; ITT = intention-to-treat; SE = standard error. Results for serum biomarkers are summarized in Table 3. Blood urea nitrogen showed significant differences only at 6 months when concentration was lowest in BF and highest in CTRL, with EXPL showing intermediate levels. At 12 months, blood urea nitrogen concentrations were no longer significantly different. Insulin growth factor-1 concentration at 6 months was significantly lower in EXPL and BF compared with CTRL. Plasma concentrations of essential amino acids and tyrosine at 6 months of age are presented in supplementary Table S5. Concentrations of the insulinogenic amino acids leucine, isoleucine, and valine were significantly lower in EXPL than in CTRL, but none of the other amino acids showed significant differences between EXPL and CTRL. In BF infants concentrations of most amino acids were appreciably lower than in formula-fed infants. The exceptions were histidine and threonine, where concentrations were similar, and glutamine and serine, where concentrations were higher in BF than in the formula groups.

significant differences between EXPL and CTRL. In BF infants concentrations of most amino acids were appreciably lower than in formula-fed infants. The exceptions were histidine and threonine, where concentrations were similar, and glutamine and serine, where concentrations were higher in BF than in the formula groups. Body composition was determined in a subset of subjects at 12 months of age. Data are summarized in supplementary Table S6. Differences in fat mass and in lean mass were not statistically significant. Bone mineral content was lower in EXPL than in CTRL, but when expressed per unit of lean body mass, the difference was not statistically significant. Formula intake (milliliters per day) (supplementary Table S7) as recorded by parents was not significantly different in EXPL compared with CTRL between 4 and 9 months; however, calculated energy intakes were almost identical at 4 and 6 months. Protein intakes showed large and highly significant differences between EXPL and CTRL. DISCUSSION The dual purpose of the present study was to determine whether in infants of overweight mothers a whey-based formula with lower protein and energy content decreases weight gain, and whether the lower protein intake still meets the needs of infants. The study was conducted in infants of overweight mothers because these infants are at increased risk for overweight later in life and may particularly benefit from slowed growth.

d formula with lower protein and energy content decreases weight gain, and whether the lower protein intake still meets the needs of infants. The study was conducted in infants of overweight mothers because these infants are at increased risk for overweight later in life and may particularly benefit from slowed growth. The experimental formula used in this randomized double-blind trial had a protein content of 1.65 g/100 kcal, which was lower than that of currently available formulas and lower than the regulatory limits in the EU and the United States. The protein content was still well above 1.30 g/100 kcal, which is the protein content considered adequate for infants beyond 3 months of age (10,11) and the amino acid profile of the formula corresponded to that of human milk. Formulas with protein content <1.8 g/100 kcal have not been evaluated in infants of this age.

he protein content was still well above 1.30 g/100 kcal, which is the protein content considered adequate for infants beyond 3 months of age (10,11) and the amino acid profile of the formula corresponded to that of human milk. Formulas with protein content <1.8 g/100 kcal have not been evaluated in infants of this age. The main finding of the present study was that feeding the EXPL formula had the effect of slowing down rapid weight gain between 3 and 6 months compared with the CTRL formula. The finding could also be interpreted as showing that the higher protein intake provided by the CTRL formula was accelerating weight gain. Either way, the EXPL formula prevented accelerated weight gain. It is likely that the growth-slowing effect of the EXPL formula was because of its lower protein content. The presence of the 2 probiotic strains (B lactis and Lactobacillus PR) was highly unlikely to have had an effect on weight gain. Other studies and committee statements have clearly shown that the 2 selected probiotic strains did not affect growth (40–42). Probiotics were added to the EXPL formula because of the reported beneficial effects they confer on infants (34–36). Similarly, the small difference in energy density of 2.8 kcal/100 mL was highly unlikely to have affected growth as recorded energy intakes were nearly identical.

c strains did not affect growth (40–42). Probiotics were added to the EXPL formula because of the reported beneficial effects they confer on infants (34–36). Similarly, the small difference in energy density of 2.8 kcal/100 mL was highly unlikely to have affected growth as recorded energy intakes were nearly identical. It is of note that the preventive effect of the EXPL formula occurred mainly in infants of obese women and in infants with weight >75th percentile. These are the groups of infants presumed to be at particularly high risk of later obesity. Infants born to overweight mothers are prone to show accelerated weight gain during infancy (32). Growth of infants fed EXPL was similar to that of BF infants. Biomarkers of protein metabolism indicated that the amount of protein provided by the CTRL formula exceeded the needs of the infants by a larger margin than the EXPL formula. In this regard our results are consistent with the findings of the European Obesity Project (43).

nts fed EXPL was similar to that of BF infants. Biomarkers of protein metabolism indicated that the amount of protein provided by the CTRL formula exceeded the needs of the infants by a larger margin than the EXPL formula. In this regard our results are consistent with the findings of the European Obesity Project (43). The other main finding of the present study was that the EXPL formula with a protein concentration 9% below the regulatory limits supported normal growth of infants from 3 months on. Given that the protein content of the formula was higher than that of breast milk, the finding was not surprising. Nevertheless, it is an important finding because it establishes that a formula with this concentration of a high-quality protein provides an intake that appears to be adequate in every respect. The judgment of adequacy rests on the observed growth performance, which showed that weight-for-age z scores of all 3 groups were at all ages significantly greater than zero, meaning that the average weight of study infants was greater than the average weight of the WHO standards. EXPL infants were never at risk for having weight-for-age z scores <−2.0, which is the WHO definition of malnutrition. Negative values for mean z scores for length indicated that average length of EXPL and CTRL was lower than the WHO mean length, whereas the length of BF infants tended to be greater than the WHO mean. The length of infants in EXPL tracked the respective WHO z score channel (38) until 24 months. Standard deviations of length z scores in EXPL infants at 6, 9, and 12 months were <1. The study therefore showed that EXPL infants were not at risk for inadequate length gain.

infants tended to be greater than the WHO mean. The length of infants in EXPL tracked the respective WHO z score channel (38) until 24 months. Standard deviations of length z scores in EXPL infants at 6, 9, and 12 months were <1. The study therefore showed that EXPL infants were not at risk for inadequate length gain. The judgment of adequacy also rests on metabolic parameters, which were close to those of the BF group and were consistent with the notion that protein intakes were slightly above those of BF infants. In addition, body composition measurements at 12 months of age indicated similar fat mass and fat-free mass in infants fed the EXPL formula and BF infants.

on metabolic parameters, which were close to those of the BF group and were consistent with the notion that protein intakes were slightly above those of BF infants. In addition, body composition measurements at 12 months of age indicated similar fat mass and fat-free mass in infants fed the EXPL formula and BF infants. High intakes of protein that are commonly observed among older infants (13–16) have been linked to increased risk of obesity in childhood in epidemiologic studies (16–19). Direct evidence for the growth-stimulating effect of high-protein feedings has been provided by the prospective, randomized trial conducted in the European Obesity Project (12). The feeding of a formula with high protein content (4.60 g/100 kcal) led to increased weight with evidence of increased adiposity. The protein content of the high-protein formula used in that trial (12) was representative of the traditional high protein content of formulas for older infants (follow-on formulas). Protein concentrations of formulas need to be somewhat higher than the protein concentration of human milk because the quality of formula protein may be somewhat less than that of human milk protein. The concentration of 1.65 g/100 kcal was still well above the required level of 1.30 g/100 kcal and could thus be expected to support normal growth. The protein content of infant formulas has successfully been lowered to 1.80 to 1.90 g/100 kcal; this has been shown to support normal growth in the first 4 months of life (44–46).

ntration of 1.65 g/100 kcal was still well above the required level of 1.30 g/100 kcal and could thus be expected to support normal growth. The protein content of infant formulas has successfully been lowered to 1.80 to 1.90 g/100 kcal; this has been shown to support normal growth in the first 4 months of life (44–46). The present study was conducted among infants at increased risk of obesity later in life because of having an overweight or obese mother (7,24–32). It is therefore necessary to exercise caution before generalizing the present findings to all infants regardless of the weight status of their mothers; however, the fact that our findings agree closely with those of Koletzko et al (12) suggests that the present findings may be generalizable. Nevertheless, to establish safety and efficacy of the low-protein formula for the general infant population, studies are needed of infants of nonobese mothers conducted in countries other than Chile. The study had some limitations. It would have been desirable for both formulas to have the exact same caloric density and for each formula to contain probiotics. As mentioned above, however, calorie intakes were nearly identical, and the presence of the 2 probiotic strains was highly unlikely to have affected growth, the primary outcome.

in the milk of every woman, only 9 (Streptococcus, Staphylococcus, Serratia, Pseudomonas, Corynebacterium, Ralstonia, Propionibacterium, Sphingomonas, and Bradyrhizobiaceae) were present in every sample from every woman. On the contrary, milk bacterial community was generally stable over time within an individual (9). In contrast to staphylococci, streptococci, corynebacteria, or propionibacteria, which seem to be widespread in human milk, the presence of lactobacilli and bifidobacteria seems to be more variable among women (9,11,14,15). Such variability may be the consequence of isolation difficulties, owing to fastidious growth and incubation requirements, or may be the result of the technical bias associated to molecular studies. It could also be because of host peculiarities; it has been suggested that the human milk microbiome is influenced by several factors that significantly skew its composition (16). In this context, the objective of this study was to assess whether demographic or clinical factors, such as country and date of birth, infant age, delivery mode, or antibiotherapy during pregnancy and lactation, may exert an influence on the bifidobacterial and lactobacillic population present in the breast milk of healthy women.

limitations. It would have been desirable for both formulas to have the exact same caloric density and for each formula to contain probiotics. As mentioned above, however, calorie intakes were nearly identical, and the presence of the 2 probiotic strains was highly unlikely to have affected growth, the primary outcome. There clearly is a need for effective interventions capable of decreasing obesity in childhood (47). Educational interventions have so far been only moderately successful (48). The promotion of breast-feeding undoubtedly is an important weapon in efforts to curtail obesity. The use of low-protein feedings after 3 months, like the one used in the present study, expands the scope of possibilities for intervention. Whether the slowing of growth in the short term translates to a reduction of obesity risk in the long term remains to be determined. Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Supplementary Material Supplemental Digital Content Acknowledgments The authors acknowledge the substantial contributions that were made to the conduct of the study by Jorge Sapunar, MD; Ruth Prieto, MW; Ana M. Vinet, MD; Jacqueline Fierro, MD; Andres Roman, MD; and Eduardo Hebel, MD. The authors also thank Françoise Chauffard and Sara Colombo-Mottaz for managing the study and Emilie Ba for data management.

e the substantial contributions that were made to the conduct of the study by Jorge Sapunar, MD; Ruth Prieto, MW; Ana M. Vinet, MD; Jacqueline Fierro, MD; Andres Roman, MD; and Eduardo Hebel, MD. The authors also thank Françoise Chauffard and Sara Colombo-Mottaz for managing the study and Emilie Ba for data management. www.ClinicalTrials.gov registration number: NCT00820833. This study was supported by Nestec SA, Vevey, Switzerland. J.I. receives grant support from Nestlé. F.H. is the chairman of the Nestlé Nutrition Institute, which receives educational and scientific grants from Nestec Ltd. P.S. and D.G. are employees of Nestlé R&D. E.E.Z. and S.E.N. receive grant support from Nestlé. E.E.Z. serves as a paid consultant and is a member of the speakers’ bureaus of Nestlé, Abbott Nutrition, and Mead Johnson. TABLE 1 Weight gain (g/day) and comparison between groups by ANCOVA

J.I. receives grant support from Nestlé. F.H. is the chairman of the Nestlé Nutrition Institute, which receives educational and scientific grants from Nestec Ltd. P.S. and D.G. are employees of Nestlé R&D. E.E.Z. and S.E.N. receive grant support from Nestlé. E.E.Z. serves as a paid consultant and is a member of the speakers’ bureaus of Nestlé, Abbott Nutrition, and Mead Johnson. TABLE 1 Weight gain (g/day) and comparison between groups by ANCOVA Age, mo EXPL CTRL BF 3–6 18.97 ± 4.19* 20.74 ± 5.01 20.07 ± 5.79 6–12 10.97 ± 3.05 12.13 ± 3.03 10.18 ± 3.85 12–24 7.65 ± 2.23 8.02 ± 2.64 7.89 ± 2.99 3–12 13.64 ± 2.65 15.01 ± 2.74 13.44 ± 3.43 3–24 10.19 ± 1.68 11.00 ± 1.74 10.29 ± 2.58 EXPL vs CTRL EXPL vs BF CTRL vs BF Difference† (95% CI) P Difference† (95% CI) P Difference† (95% CI) P 3–6 −2.26 (−3.88 to −0.64) 0.006 −0.72 (−2.46 to 1.01) 0.411 1.54 (−0.13 to 3.21) 0.071 6–12 −0.88 (−2.10 to 0.35) 0.159 0.77 (−0.50 to 2.05) 0.233 1.65 (0.45–2.85) 0.007 12–24 −0.40 (−1.42 to 0.62) 0.435 −0.27 (−1.34 to 0.80) 0.621 0.13 (−0.87 to 1.14) 0.791 3–12 −1.34 (−2.42 to −0.26) 0.015 0.33 (−0.80 to 1.45) 0.569 1.66 (0.60–2.72) 0.002 3–24 −0.86 (−1.64 to −0.08) 0.031 −0.11 (−0.93 to 0.71) 0.798 0.75 (−0.01 to 1.52) 0.054 ANCOVA = analysis of covariance; BF = breast-fed; BMI = body mass index; CI = confidence interval; CTRL = control; EXPL = experimental. *Mean ± standard deviation. †Differences between groups estimated by ANCOVA (correction for maternal BMI, weight z score at 3 months, infant sex, ethnicity, antibiotic intake, and complementary food intake <6 months).

Age, mo EXPL CTRL BF 3–6 18.97 ± 4.19* 20.74 ± 5.01 20.07 ± 5.79 6–12 10.97 ± 3.05 12.13 ± 3.03 10.18 ± 3.85 12–24 7.65 ± 2.23 8.02 ± 2.64 7.89 ± 2.99 3–12 13.64 ± 2.65 15.01 ± 2.74 13.44 ± 3.43 3–24 10.19 ± 1.68 11.00 ± 1.74 10.29 ± 2.58 EXPL vs CTRL EXPL vs BF CTRL vs BF Difference† (95% CI) P Difference† (95% CI) P Difference† (95% CI) P 3–6 −2.26 (−3.88 to −0.64) 0.006 −0.72 (−2.46 to 1.01) 0.411 1.54 (−0.13 to 3.21) 0.071 6–12 −0.88 (−2.10 to 0.35) 0.159 0.77 (−0.50 to 2.05) 0.233 1.65 (0.45–2.85) 0.007 12–24 −0.40 (−1.42 to 0.62) 0.435 −0.27 (−1.34 to 0.80) 0.621 0.13 (−0.87 to 1.14) 0.791 3–12 −1.34 (−2.42 to −0.26) 0.015 0.33 (−0.80 to 1.45) 0.569 1.66 (0.60–2.72) 0.002 3–24 −0.86 (−1.64 to −0.08) 0.031 −0.11 (−0.93 to 0.71) 0.798 0.75 (−0.01 to 1.52) 0.054 ANCOVA = analysis of covariance; BF = breast-fed; BMI = body mass index; CI = confidence interval; CTRL = control; EXPL = experimental. *Mean ± standard deviation. †Differences between groups estimated by ANCOVA (correction for maternal BMI, weight z score at 3 months, infant sex, ethnicity, antibiotic intake, and complementary food intake <6 months). TABLE 2 Weight-for-age z scores and comparison between groups by ANCOVA at age 3 months and a mixed longitudinal model at other ages

oper storage of the samples and, together with the low SCC level of milk samples, indicates that participating women did not experience mastitis. Higher counts are usually related to non-hygienic sampling, improper storage of the samples, and/or use of contaminated milk pumps for sampling collection or mastitis (26,27) Although the main target of this study were lactobacilli and bifidobacteria, coagulase-negative staphylococci and viridans streptococci could be isolated from 51 (77.27%) and 40 (60.61%), respectively, of the 66 cultured samples. Staphylococci and streptococci constitute the dominant culturable bacteria in human milk, and related DNA sequences of both genera are also the prevailing ones in this biological fluid, albeit with substantial interindividual differences (4,8,9,11,14,16,18,28). In spite of this, staphylococci and streptococci have received a marginal attention regarding their role in the human mammary gland and in the early colonization of the infant gut, although they could be useful to reduce the acquisition of undesired pathogens, particularly in infants exposed to hospital environments (7,28–30).

†Differences between groups estimated by ANCOVA (correction for maternal BMI, weight z score at 3 months, infant sex, ethnicity, antibiotic intake, and complementary food intake <6 months). TABLE 2 Weight-for-age z scores and comparison between groups by ANCOVA at age 3 months and a mixed longitudinal model at other ages Age, mo EXPL CTRL BF 3 0.24 ± 0.91* 0.22 ± 0.89 0.67 ± 0.82 6 0.37 ± 0.73 0.54 ± 0.95 0.89 ± 1.03 9 0.59 ± 0.78 0.80 ± 0.99 1.00 ± 0.99 12 0.58 ± 0.81 0.88 ± 0.88 0.90 ± 0.96 24 0.62 ± 0.79 0.91 ± 0.93 0.89 ± 1.06 EXPL vs CTRL EXPL vs BF CTRL vs BF Difference† (95% CI) P Difference† (95% CI) P Difference† (95% CI) P 3 0.03 (−0.23 to 0.29) 0.845 −0.39 (−0.67 to −0.12) 0.005 −0.42 (−0.69 to −0.15) 0.003 6 −0.16 (−0.35 to 0.02) 0.078 −0.07 (−0.26 to 0.13) 0.508 0.10 (−0.09 to 0.29) 0.302 9 −0.22 (−0.41 to −0.03) 0.022 0.04 (−0.16 to 0.24) 0.665 0.27 (0.07–0.46) 0.006 12 −0.31 (−0.50 to −0.11) 0.002 0.16 (−0.05 to 0.36) 0.128 0.46 (0.27–0.66) <0.001 24 −0.33 (−0.53 to −0.13) 0.001 0.17 (−0.04 to 0.38) 0.104 0.50 (0.30–0.70) <0.001 ANCOVA = analysis of covariance; BF = breast-fed; CI = confidence interval; CTRL = control; EXPL = experimental. *Mean ± standard deviation. †Differences between groups estimated by ANCOVA at age 3 months and by a mixed model (39) at other ages. TABLE 3 Serum concentrations of biomarkers (PP data set); comparisons by ANCOVA correcting for values at age 3 months

Age, mo EXPL CTRL BF 3 0.24 ± 0.91* 0.22 ± 0.89 0.67 ± 0.82 6 0.37 ± 0.73 0.54 ± 0.95 0.89 ± 1.03 9 0.59 ± 0.78 0.80 ± 0.99 1.00 ± 0.99 12 0.58 ± 0.81 0.88 ± 0.88 0.90 ± 0.96 24 0.62 ± 0.79 0.91 ± 0.93 0.89 ± 1.06 EXPL vs CTRL EXPL vs BF CTRL vs BF Difference† (95% CI) P Difference† (95% CI) P Difference† (95% CI) P 3 0.03 (−0.23 to 0.29) 0.845 −0.39 (−0.67 to −0.12) 0.005 −0.42 (−0.69 to −0.15) 0.003 6 −0.16 (−0.35 to 0.02) 0.078 −0.07 (−0.26 to 0.13) 0.508 0.10 (−0.09 to 0.29) 0.302 9 −0.22 (−0.41 to −0.03) 0.022 0.04 (−0.16 to 0.24) 0.665 0.27 (0.07–0.46) 0.006 12 −0.31 (−0.50 to −0.11) 0.002 0.16 (−0.05 to 0.36) 0.128 0.46 (0.27–0.66) <0.001 24 −0.33 (−0.53 to −0.13) 0.001 0.17 (−0.04 to 0.38) 0.104 0.50 (0.30–0.70) <0.001 ANCOVA = analysis of covariance; BF = breast-fed; CI = confidence interval; CTRL = control; EXPL = experimental. *Mean ± standard deviation. †Differences between groups estimated by ANCOVA at age 3 months and by a mixed model (39) at other ages. TABLE 3 Serum concentrations of biomarkers (PP data set); comparisons by ANCOVA correcting for values at age 3 months P Age, mo EXPL CTRL BF EXPL vs CTRL EXPL vs BF CTRL vs BF BUN, mg/dL 3 5.05 ± 2.36* 5.96 ± 4.10 3.65 ± 1.55 — — — 6 5.24 ± 2.84 9.27 ± 3.10 3.57 ± 2.05 <0.001 <0.001 <0.001 12 10.58 ± 15.14 9.60 ± 4.11 8.84 ± 4.82 0.333 0.833 0.243 IGF-1, μg/L 3 88.1 ± 52.1 87.5 ± 52.4 80.5 ± 51.1 — — — 6 76.5 ± 60.0 87.0 ± 51.8 59.1 ± 30.6 0.010 0.728 0.005 12 70.5 ± 47.0 77.5 ± 38.1 66.5 ± 40.3 0.050 0.657 0.130 Insulin, mU/L 3 14.81 ± 16.07 14.13 ± 10.87 11.38 ± 8.32 — — — 6 9.05 ± 7.60 9.49 ± 7.03 7.91 ± 6.54 0.596 0.479 0.212 12 10.70 ± 11.14 11.28 ± 10.66 7.94 ± 7.19 0.521 0.585 0.220 C-peptide, pmol/L 3 312.8 ± 199.3 301.1 ± 170.3 245.1 ± 155.0 — — — 6 285.3 ± 154.4 242.6 ± 134.0 214.0 ± 105.1 0.070 0.071 0.864 12 365.2 ± 190.6 368.5 ± 237.5 308.7 ± 197.3 0.746 0.115 0.172 Ghrelin (plasma), ng/L 3 528.1 ± 120.2 605.4 ± 206.2 539.9 ± 180.7 — — — 6 688.2 ± 220.2 851.9 ± 517.6 765.4 ± 305.7 0.260 0.319 0.960 Leptin, μg/L 3 6.77 ± 4.28 7.10 ± 5.54 8.51 ± 5.53 — — — 6 5.61 ± 4.70 5.39 ± 4.01 7.68 ± 7.99 0.914 0.270 0.292 12 2.35 ± 1.15 3.01 ± 2.77 3.48 ± 1.76 0.095 0.009 0.254 ANCOVA = analysis of covariance; BF = breast-fed; BUN = blood urea nitrogen; CI = confidence interval; CTRL = control; EXPL = experimental; IGF-1 = insulin growth factor-1; PP = per protocol.

51 ± 5.53 — — — 6 5.61 ± 4.70 5.39 ± 4.01 7.68 ± 7.99 0.914 0.270 0.292 12 2.35 ± 1.15 3.01 ± 2.77 3.48 ± 1.76 0.095 0.009 0.254 ANCOVA = analysis of covariance; BF = breast-fed; BUN = blood urea nitrogen; CI = confidence interval; CTRL = control; EXPL = experimental; IGF-1 = insulin growth factor-1; PP = per protocol. *Mean ± standard deviation.

Bacterial gut colonization in early life is a process that exerts a short-, medium-, and long-term influence on the health status of a host, and that involves bacteria arising from different sources (1); among them, culture-dependent studies have revealed that human milk is a source of live staphylococci, streptococci, lactic acid bacteria, bifidobacteria, propionibacteria, corynebacteria, and closely related Gram-positive bacteria to the infant gut (2). Several studies have shown that there is a mother-to-infant transfer of bacterial strains belonging, at least, to the genera Lactobacillus, Staphylococcus, Enterococcus, and Bifidobacterium through breast-feeding (3–7). In fact, human milk constitutes 1 of the main sources of bacteria to the breast-fed infant gut since a baby consuming approximately 800 mL/day of milk would ingest between 1 × 105 and 1 × 107 bacteria daily (8). It has been suggested that exposure of the breast-fed infant to such a wealth of bacterial phylotypes through breast-feeding may exert beneficial effects against several diseases (9). Breast-feeding has been shown to improve infant health outcomes lowering the risk of respiratory and gastrointestinal tract infections, necrotizing enterocolitis, otitis media, and allergic disease and to prevent later health problems such as inflammatory bowel disease, obesity, and type 2 diabetes mellitus (10).

(9). Breast-feeding has been shown to improve infant health outcomes lowering the risk of respiratory and gastrointestinal tract infections, necrotizing enterocolitis, otitis media, and allergic disease and to prevent later health problems such as inflammatory bowel disease, obesity, and type 2 diabetes mellitus (10). The application of culture-independent molecular techniques, and particularly those based on 16S rRNA genes, allowed a complementary biodiversity assessment of the human milk microbiome. The use of such techniques confirmed the dominance of staphylococci and streptococci, the relatively frequent presence of lactic acid bacteria and bifidobacteria, and the existence of DNA belonging to other bacterial groups, such as some Gram-negative bacteria (5,11–13). Recently, the first microbiome study focused on human milk was published and the results indicated that milk bacterial communities were generally complex (9). Among the hundreds of operational taxonomic units detected in the milk of every woman, only 9 (Streptococcus, Staphylococcus, Serratia, Pseudomonas, Corynebacterium, Ralstonia, Propionibacterium, Sphingomonas, and Bradyrhizobiaceae) were present in every sample from every woman. On the contrary, milk bacterial community was generally stable over time within an individual (9).

text, the objective of this study was to assess whether demographic or clinical factors, such as country and date of birth, infant age, delivery mode, or antibiotherapy during pregnancy and lactation, may exert an influence on the bifidobacterial and lactobacillic population present in the breast milk of healthy women. METHODS Subjects and Sampling A total of 160 healthy women participated in the study and provided a sample of breast milk. Women were recruited to cover a moderately wide area of central Europe from randomly chosen regions in Germany and Austria to represent southern and eastern (Germany) and western and eastern (Austria) parts of both countries, which included both rural and urban settings. Recruitment was carried out by midwives, who were contacted initially via the HiPP Scientific sales force (Pfaffenhofen, Germany). All of the volunteers gave written informed consent to the protocol, which was approved by the ethical committee of Hospital Clínico (Madrid, Spain). The milk samples were collected in a sterile tube by manual expression using sterile gloves. Previously, nipples and mammary areola were cleaned with soap and sterile water and soaked in chlorhexidine. The first drops (∼500 μL) were discarded. All of the samples were kept frozen until delivery to the laboratory. All of the women filled a questionnaire designed to collect information on demographic characteristics and some other factors, such as mode of delivery, anesthesia during labor, or antibiotherapy during pregnancy and lactation (Table 1).

e discarded. All of the samples were kept frozen until delivery to the laboratory. All of the women filled a questionnaire designed to collect information on demographic characteristics and some other factors, such as mode of delivery, anesthesia during labor, or antibiotherapy during pregnancy and lactation (Table 1). Count and Identification of Bacteria in the Samples Adequate dilutions of 66 randomly selected milk samples were spread onto agar plates of Man, Rogosa, and Sharpe (Oxoid, Basingstoke, UK) supplemented with l-cysteine (0.5 g/L) (MRS-Cys) for isolation of lactobacilli and bifidobacteria. The plates were incubated for 48 hours at 37°C anaerobically (85% nitrogen, 10% hydrogen, 5% carbon dioxide) in an anaerobic workstation (MINI-MACS; DW Scientific, Shipley, UK).

Rogosa, and Sharpe (Oxoid, Basingstoke, UK) supplemented with l-cysteine (0.5 g/L) (MRS-Cys) for isolation of lactobacilli and bifidobacteria. The plates were incubated for 48 hours at 37°C anaerobically (85% nitrogen, 10% hydrogen, 5% carbon dioxide) in an anaerobic workstation (MINI-MACS; DW Scientific, Shipley, UK). After incubation and counting, 10 isolates from each culture medium were randomly selected and identified at the species level by classical morphological and biochemical tests. In addition, all the Gram-positive isolates with morphology compatible with that of lactobacilli or bifidobacteria were selected and identified at the genus level by classical morphological and biochemical tests and by demonstration of fructose-6-phosphate phosphoketolase activity in cellular extracts. Identification at the species level was performed by MALDI-TOF (Vitek MS; BioMerieux, Marcy l’Etoile, France) or by polymerase chain reaction (PCR) sequencing of a 470-bp fragment of the 16S rRNA gene using primers pbl16 (5′-AGAGTTTGATCCTGGCTCAG-3′) and mlb16 (5′-GGCTGCTGGCACGTAGTTAG-3′) (17). The PCR conditions were as follows: 96°C for 30 seconds, 48°C for 30 seconds, and 72°C for 45 seconds (40 cycles) and a final extension at 72°C for 4 minutes. The amplicons were purified using the Nucleospin Extract II kit (Macherey-Nagel, Düren, Germany) and sequenced at the Genomics Unit of the Universidad Complutense de Madrid, Spain. The resulting sequences were used to search sequences deposited in the EMBL database using the BLAST algorithm, and the identity of the isolates was determined on the basis of the highest scores (≥98%).

(Macherey-Nagel, Düren, Germany) and sequenced at the Genomics Unit of the Universidad Complutense de Madrid, Spain. The resulting sequences were used to search sequences deposited in the EMBL database using the BLAST algorithm, and the identity of the isolates was determined on the basis of the highest scores (≥98%). Somatic Cell Count (SCC) in the Samples SCC was performed with a DeLaval cell counter DCC (DeLaval International AB, Tumba, Sweden), using single-use cell counter cassettes and instructions provided by the manufacturer. The cassette that contains small amounts of a DNA-specific stain (propidium iodide) is used to collect the sample. A piston carries the milk sample toward a counting window that is exposed to an LED light source. The fluorescence signal given by the cell nuclei is recorded as a digital image that is subjected to automated image analysis. Bacterial DNA Isolation From the Milk Samples Initially, a fraction of the breast milk samples (1 mL) was centrifuged at 7150 g for 20 minutes. Then, total DNA was isolated from the pellets using the QIAamp DNA Stool Mini Kit (QIAgen, Hilden, Germany) following a protocol described previously (11). DNA was eluted in 20 μL of buffer AE (provided in the kit), and the purified DNA extracts were stored at −20°C.

les (1 mL) was centrifuged at 7150 g for 20 minutes. Then, total DNA was isolated from the pellets using the QIAamp DNA Stool Mini Kit (QIAgen, Hilden, Germany) following a protocol described previously (11). DNA was eluted in 20 μL of buffer AE (provided in the kit), and the purified DNA extracts were stored at −20°C. Qualitative PCR Assays Genus-specific detection of DNA from the genera Lactobacillus or Bifidobacterium was accomplished using the primers and PCR conditions reported by Collado et al (18). At the species level, a 2-step multiplex PCR assay was used to detect DNA from L fermentum, L gasseri, L plantarum, L reuteri, L rhamnosus, and L salivarius, using species-specific primers and PCR conditions previously reported (19). In parallel, the presence of DNA from L casei/L paracasei was assessed using the primers and PCR conditions described by Chagnaud et al (20). PCR detection of DNA from B longum, B infantis, B dentium, and B gallicum was carried using the primers and PCR conditions described by Matsuki et al (21), whereas the presence of DNA from B adolescentis, B angulatum, B bifidum, B breve, B catenulatum, and B pseudocatenulatum was assessed using those reported by Matsuki et al (22). Finally, PCR detection of DNA from B lactis was performed using the species-specific PCR assay developed by Ventura et al (23).

ki et al (21), whereas the presence of DNA from B adolescentis, B angulatum, B bifidum, B breve, B catenulatum, and B pseudocatenulatum was assessed using those reported by Matsuki et al (22). Finally, PCR detection of DNA from B lactis was performed using the species-specific PCR assay developed by Ventura et al (23). Each PCR assay included DNA extracted from a reference strain of each targeted species (positive control). Amplicons were analyzed by electrophoresis (90 V, 1 hour) on 1% agarose gels. Subsequently, the gels were stained and bands were visualized in a Gel-Doc system (Bio-Rad, Hercules, CA). Statistical Analysis Quantitative data were expressed as the mean and 95% confidence interval (CI) of the mean or, when they were not normally distributed, as the median and interquartile range. A correlation analysis was performed to test the relation between bacterial counts in MRS-Cys and SCCs. Proportions were compared using χ2 statistics, including the Fisher exact test and the Freeman-Halton test for contingency tables greater than 2 × 2. Differences were considered significant at P < 0.050. SAS version 9.2 (SAS Institute Inc, Cary, NC) was used to carry out the analyses cited above.

s in MRS-Cys and SCCs. Proportions were compared using χ2 statistics, including the Fisher exact test and the Freeman-Halton test for contingency tables greater than 2 × 2. Differences were considered significant at P < 0.050. SAS version 9.2 (SAS Institute Inc, Cary, NC) was used to carry out the analyses cited above. RESULTS Characteristics of the Lactating Women Participating in the Study The 160 women enrolled in this study had a mean age of 31.82 years (95% CI 31.10–32.54 years) and their breast-fed baby had a gestational age of 39.72 weeks (ranging from 29 to 44 weeks). The median number of children was 1 (n = 104, 65.00% of participants), and only 11 women (6.88%) had 3 or more children. Moreover, 21.87% of the infants were born by cesarean section, more than one-third of the women (35.22%) received anesthesia during delivery, and 40.62% of the women had received antibiotherapy during pregnancy and/or lactation (Table 1). Most of the participating women were exclusively breast-feeding their babies whereas 32.08% did it partially. Regarding the time of sampling, most of the breast milk samples (73.75%) were collected from women during the second to fourth week of lactation, whereas 10.00% were obtained during the first week and the remaining 15.63% after the first month of lactation. Most of the women who gave samples during the first month of lactation were exclusively breast-feeding their infants (75.37%), whereas this percentage descended notably in samples obtained after the first month of lactation (28.00%). Other demographic and clinical characteristics, such as the month of delivery, nationality and origin of the mother (urban or rural), and place where the samples were obtained, are summarized in Table 1.

(75.37%), whereas this percentage descended notably in samples obtained after the first month of lactation (28.00%). Other demographic and clinical characteristics, such as the month of delivery, nationality and origin of the mother (urban or rural), and place where the samples were obtained, are summarized in Table 1. Bacterial and SCCs Bacterial growth was observed in 58 of the 66 milk samples inoculated on MRS-Cys agar plates. The mean (95% CI) value of bacterial counts obtained in such medium was 1.63 (1.49–1.77) log10 colony-forming units (CFU)/mL and ranged between 1.0 and 2.7 log10 CFU/mL (Fig. 1). In those breast milk samples where bacterial growth was observed, the mean (95% CI) value of SCCs was 36.67 (35.96–41.37) cells per microliter and ranged between 23 and 68 cells per microliter, whereas in the rest of the samples (n = 8) where bacterial growth was undetectable, lower SSCs values (mean value of 27.16 cells/μL) were observed. A weak but statistically significant correlation was noted between bacterial counts in MRS-Cys plates and SSCs (r = 0.395, P = 0.002), as shown in Figure 1. Overall, the bacterial count and SCC values found in these breast milk samples indicated that none of the women were experiencing mastitis when the samples were collected.

ut statistically significant correlation was noted between bacterial counts in MRS-Cys plates and SSCs (r = 0.395, P = 0.002), as shown in Figure 1. Overall, the bacterial count and SCC values found in these breast milk samples indicated that none of the women were experiencing mastitis when the samples were collected. FIGURE 1 Bacterial counts in MRS-Cys and SCCs in breast milk samples (n = 66). Correlation between both parameters is shown as a solid line and follows the model: SCC (cells/μL) = 27.22 + 8.00 ∗ Bacterial counts in MRS-Cys (log10 CFU/mL), r = 0.395, P = 0.002; 95% CI for the mean value of SSC as a function of bacterial counts in MRS-Cys is shown as solid gray lines and 95% prediction intervals for new observations of bacterial counts in MRS-Cys as a function of SCC are shown as the outer dotted gray lines. Black dots = breast milk sample; ND = breast milk samples in which bacterial growth was not detected; they were not included in the correlation analysis. Bacterial Identification In relation to Gram-positive cocci, Staphylococcus spp were isolated from 51 samples (77.27%); Staphylococcus epidermidis was detected in all 51 samples, whereas other coagulase-negative staphylococci were isolated from <25% of the samples (data not shown). Staphylococcus aureus could not be detected in any sample. Streptococci were also isolated from 40 samples (60.61%) and most of them belonged to the species Streptococcus mitis, Streptococcus salivarius, or Streptococcus parasanguinis (data not shown).

phylococci were isolated from <25% of the samples (data not shown). Staphylococcus aureus could not be detected in any sample. Streptococci were also isolated from 40 samples (60.61%) and most of them belonged to the species Streptococcus mitis, Streptococcus salivarius, or Streptococcus parasanguinis (data not shown). Lactobacilli and bifidobacteria could be isolated from 27 (40.91%) and 7 (10.61%) samples of breast milk, respectively (Fig. 2). The mean (95% CI) values of bacterial counts for Lactobacillus and Bifidobacterium were 1.11 (0.99–1.23) and 0.96 (0.76–1.16) log10 CFU/mL, respectively. Identification of the isolates at the species level revealed that they belonged to the following species: B breve, B longum, L casei, L fermentum, L gasseri, L gastricus, L plantarum, L reuteri, L rhamnosus, L salivarius, and L vaginalis. The species most frequently found was L salivarius that was isolated from 9 milk samples (13.64% of the 66 cultured samples), followed by L fermentum (7 samples, 10.61%), L gasseri (6 samples, 9.09%), and B breve (5 samples, 7.58%) (Table 2). Usually, only 1 species of either lactobacillus or bifidobacteria was present in an individual breast milk sample, although 2 different species were isolated from 7 samples, and B longum, L gastricus, and L reuteri were detected simultaneously in 1 sample (Table 2).

es, 9.09%), and B breve (5 samples, 7.58%) (Table 2). Usually, only 1 species of either lactobacillus or bifidobacteria was present in an individual breast milk sample, although 2 different species were isolated from 7 samples, and B longum, L gastricus, and L reuteri were detected simultaneously in 1 sample (Table 2). FIGURE 2 Total, lactobacilli, and bifidobacteria viable cells in breast milk samples after culturing in MRS-Cys. Bacterial growth was undetectable in 8 samples out of 66 analyzed. Mean values of bacterial counts are indicated with a “+” within each box-and-whisper plot. Median values are indicated by the line within the box plot. The box extends from the 25th to 75th percentiles, and the whiskers indicate the minimum and maximum values obtained. Outliers are represented as dots. PCR Assessment of Lactobacilli and Bifidobacteria Diversity in the Milk Samples The presence of lactobacilli and bifidobacterial DNA alone or in combination was analyzed by PCR using species-specific primers in 160 breast milk samples, and it was confirmed in 113 (70.60%) samples (Fig. 3A). More specifically, Lactobacillus sequences (alone or in combination with Bifidobacterium sequences) and Bifidobacterium sequences (alone or in combination with Lactobacillus sequences) were detected in 108 and 41 samples (67.50% and 25.62% of total samples), respectively, whereas both genera were present simultaneously in 36 samples (22.50% of total samples) (Fig. 3A).

r in combination with Bifidobacterium sequences) and Bifidobacterium sequences (alone or in combination with Lactobacillus sequences) were detected in 108 and 41 samples (67.50% and 25.62% of total samples), respectively, whereas both genera were present simultaneously in 36 samples (22.50% of total samples) (Fig. 3A). FIGURE 3 Presence of lactobacilli and/or bifidobacteria DNA in breast milk samples as determined by species-specific PCR (n = 160). Frequency of detection of DNA from genus Lactobacillus and/or Bifidobacterium (A) and lactobacilli and bifidobacterial species (B) in breast milk samples identified by species-specific PCR. The percentages are expressed over the total 160 samples analyzed. PCR = polymerase chain reaction. The Lactobacillus species most frequently found was L salivarius (56 samples, 35.00% of total samples), followed by L fermentum (40 samples, 25.00%) and L gasseri (35 samples, 21.88%) (Fig. 3B). Other lactobacilli species detected using this approach were L reuteri (11.88%), L plantarum (10.63%), L rhamnosus (8.13%), and L casei (4.38%). Regarding bifidobacteria, B breve DNA was the most frequently found and it was present in 21 (13.75%) of the analyzed samples. B longum and B lactis were detected only in 7 samples (4.38%), each. Another 7 breast milk samples contained other species of bifidobacteria (Fig. 3B).

L rhamnosus (8.13%), and L casei (4.38%). Regarding bifidobacteria, B breve DNA was the most frequently found and it was present in 21 (13.75%) of the analyzed samples. B longum and B lactis were detected only in 7 samples (4.38%), each. Another 7 breast milk samples contained other species of bifidobacteria (Fig. 3B). Globally, there was great interindividual variability regarding the lactobacilli and bifidobacterial PCR profile. In fact, up to 52 different species combinations were found among the 113 breast milk samples in which lactobacilli and/or bifidobacteria could be detected (Fig. 4). A total of 31 women displayed a profile that was not shared with any other recruited woman. The profiles of most of the samples comprised only 1 or 2 different species (38 and 43 samples, respectively) (Fig. 4). The profile comprising only L salivarius DNA was shared by 13 of the analyzed samples (representing 8% of total samples), whereas the combination of L fermentum with either L salivarius or L gasseri was present in 9 and 8 human milk samples, respectively (∼5% of total samples). In contrast, 8 breast milk samples contained DNA from 4 different Lactobacillus and Bifidobacterium species. A detailed analysis of all the combinations of lactobacilli and bifidobacterial species that were detected by PCR is presented in Figure 4.

ent in 9 and 8 human milk samples, respectively (∼5% of total samples). In contrast, 8 breast milk samples contained DNA from 4 different Lactobacillus and Bifidobacterium species. A detailed analysis of all the combinations of lactobacilli and bifidobacterial species that were detected by PCR is presented in Figure 4. FIGURE 4 Lactobacilli and bifidobacterial diversity in breast milk samples (n = 113) as determined by PCR using species-specific primers. The 52 different combinations of DNA from lactobacilli and bifidobacterial species detected by PCR in individual breast milk samples are shown in the box grid in the middle of the figure; each file represents 1 unique combination and a gray box indicates the presence of DNA from a particular species. The bar graph at the top of the figure indicates the number of milk samples in which each specific combination of DNA from lactobacilli and bifidobacterial species was found. At the bottom of the box grid, the number of stars represents the total number of different species (lactobacilli and bifidobacteria) in each DNA profile. PCR = polymerase chain reaction.

icates the number of milk samples in which each specific combination of DNA from lactobacilli and bifidobacterial species was found. At the bottom of the box grid, the number of stars represents the total number of different species (lactobacilli and bifidobacteria) in each DNA profile. PCR = polymerase chain reaction. Taken as a whole, there was a good correspondence between isolation by culture technique and PCR detection of lactobacilli or bifidobacteria, both at the genus and the species level in the 66 breast milk samples that were seeded on MRS-Cys plates (Table 3). Lactobacilli DNA was detected in 50 samples, and viable lactobacilli were isolated from approximately half of them (26 samples). The proportion of samples in which bifidobacteria were isolated by culture and in which their DNA was detected by PCR was lower (6 samples of 18 positives for bifidobacteria by PCR; Table 3). B lactis was detected only using molecular techniques (in 2 milk samples).