Browse the corpus

Complications after esophageal resection present a serious problem: 26–41 % of patients develop major adverse events; the majority of which are cardiopulmonary and infectious complications or anastomotic leakage. Mortality rates are 4–10 %.1 Therefore, it is imperative to develop tools that predict unfavorable postoperative outcome adequately. The assessment of sarcopenia may indicate whether complications are likely to arise. Sarcopenia is defined as a syndrome characterized by progressive and generalized loss of skeletal muscle mass and strength, which increases the risk of adverse health-related outcomes, such as impaired physical ability, decreased quality of life, and mortality.2,3 We showed before that in colorectal surgery, sarcopenia as an important element of functional compromise is associated with postoperative mortality.4 Neoadjuvant chemoradiotherapy (CRT) induces significant sarcopenia in esophageal cancer patients.5 Moreover, esophageal cancer can further deteriorate skeletal muscle mass by causing cachexia. Sarcopenia is associated with complications following surgery.6,7 Furthermore, sarcopenic obesity, defined as high body mass index and low muscle tissue on CT scan, is hypothesized to be a predictor of unfavorable oncologic and surgical outcomes.8,9 It is known that both sarcopenia and obesity are associated with chronic inflammation, which could explain the high incidence of postoperative complications in sarcopenic obese patients.10,11 In addition, these patients are often not seen as “at risk,” because normal body mass index due to high fat mass camouflages sarcopenia.

own that both sarcopenia and obesity are associated with chronic inflammation, which could explain the high incidence of postoperative complications in sarcopenic obese patients.10,11 In addition, these patients are often not seen as “at risk,” because normal body mass index due to high fat mass camouflages sarcopenia. A readily available instrument to quantify skeletal muscle mass is CT-based muscle area measurement at the level of the lumbar three vertebral landmark, which can be accomplished by using open-source software.12,13 Although low preoperative muscle mass measured on CT scans could not be associated with mortality and length of hospital stay in patients undergoing esophageal surgery, the amount of muscle mass loss induced by neoadjuvant CRT may be a more dynamic predictor of postoperative outcome.5 Increased muscle loss before surgery presumably reflects high metabolic stress due to malnutrition, cachexia, tumor load, or inflammation, negatively affecting the metabolic response to major gastrointestinal surgery. In this study, it was hypothesized that the amount of muscle mass loss during neoadjuvant CRT predicts postoperative mortality after esophageal resection. The primary purpose of the study was to measure whether the amount of muscle mass lost during neoadjuvant CRT in patients undergoing esophageal resection was associated with postoperative mortality. The secondary goal of the study was to measure whether sarcopenic obesity and fat loss over the course of neoadjuvant CRT, using a pre- and post-CRT time point, was associated with postoperative mortality.

st during neoadjuvant CRT in patients undergoing esophageal resection was associated with postoperative mortality. The secondary goal of the study was to measure whether sarcopenic obesity and fat loss over the course of neoadjuvant CRT, using a pre- and post-CRT time point, was associated with postoperative mortality. Methods Patients All patients who underwent resection of esophageal cancer in a single teaching hospital from January 2008 until January 2012 were enrolled in a prospectively maintained, digital database. Data included characteristics of the primary tumor and oncologic staging, specifications of surgical treatment, chemotherapy, radiotherapy, and postoperative complications. All patients who received neoadjuvant CRT were included in this study. Neoadjuvant CRT consisted of three cycles of cisplatin/5-fluorouracil (CF) or five cycles of paclitaxel/carboplatin (PC), all with concurrent radiotherapy, or consisted of three cycles of epirubicin/cisplatin/capecitabine (ECC), also with concurrent radiotherapy in some patients. A routine total body PET/CT scan before neoadjuvant CRT (preCRT) and after neoadjuvant CRT (postCRT) was performed, as standard care for all patients diagnosed with esophageal cancer. After a multidisciplinary meeting, which took place weekly, patients were admitted to the neoadjuvant CRT protocol, generally meaning that treatment started within 2 weeks after PET/CT. The second PET/CT was performed directly after neoadjuvant CRT, following the same protocol in all patients. The average time between the first and second PET/CT was 111 days [standard deviation (SD), 13 days]. The variation was mostly due to neoadjuvant CRT taking longer because of side-effects. Patients were closely monitored by a dietician from the moment of diagnosis until hospital discharge after surgery, and malnutrition was screened for using the Short Nutritional Assessment Questionnaire (SNAQ). Adequate oral intake was verbally encouraged, and if nutritional depletion was present, patients received protein-rich and lipid-rich drinks. A gastric or jejunal feeding tube was placed in patients with esophageal stenosis. Furthermore, physical activity was stimulated by counseling or by active preconditioning by an expert physiotherapist. The study was conducted with approval from local medical ethical committees and according to the revised version of the Declaration of Helsinki (October 2008, Seoul). The principles of good clinical practice (GCP) were followed during this study.

by counseling or by active preconditioning by an expert physiotherapist. The study was conducted with approval from local medical ethical committees and according to the revised version of the Declaration of Helsinki (October 2008, Seoul). The principles of good clinical practice (GCP) were followed during this study. Postoperative Mortality The primary clinical outcome of the study was postoperative mortality. This was defined as mortality within 30 days or within hospital admission.

by counseling or by active preconditioning by an expert physiotherapist. The study was conducted with approval from local medical ethical committees and according to the revised version of the Declaration of Helsinki (October 2008, Seoul). The principles of good clinical practice (GCP) were followed during this study. Postoperative Mortality The primary clinical outcome of the study was postoperative mortality. This was defined as mortality within 30 days or within hospital admission. CT-Based Muscle and Fat Measurements CT-based measurements were performed using Osirix® Version 3.3 (32-bit; http://www.osirix-viewer.com) by a researcher who was blinded from outcome. The cross-sectional skeletal muscle surface (cm2) as measure of sarcopenia was measured at the level of the third lumbar vertebra (L3) on two consecutive transversal slices on which both vertebral spines were visible.13 The “Grow Region (2D/3D Segmentation)” tool in the menu of the program facilitated to select automatically all skeletal muscle mass in one slice. In the same manner, L3 total adipose tissue was composed of all adipose tissue on the L3 level, including subcutaneous adipose tissue, visceral adipose tissue, and intramuscular adipose tissue. Visceral and subcutaneous fat were measured after manual adjustment to ensure no excessive tissue was included in the measurements. The average surface (cm2) of the two consecutive slices was used for the analyses. The distinction between different tissues was based on Hounsfield units (HU) with a threshold range of −29 to +150 HU used for skeletal muscle and a range of −190 to −30 HU for fat tissue.7,12 Muscles measured were: psoas, quadratus lumborum, paraspinal, transverse abdominal, external oblique, internal oblique, and rectus abdominis muscles. Hand-adjustment of the selected areas was performed if necessary and the values of muscle or fat area were automatically calculated by Osirix®13

30 HU for fat tissue.7,12 Muscles measured were: psoas, quadratus lumborum, paraspinal, transverse abdominal, external oblique, internal oblique, and rectus abdominis muscles. Hand-adjustment of the selected areas was performed if necessary and the values of muscle or fat area were automatically calculated by Osirix®13 Sarcopenia and Muscle Loss Index The L3 muscle area surfaces were normalized for patient height to calculate the L3 muscle index and expressed in cm2/m2. Patient height and weight were measured by a nurse at admission to the hospital at the time of diagnosis. The cutoff values used for sarcopenia were 52.4 cm2/m2 for men and 38.5 cm2/m2 for women, based on the method of Prado et al.12 Previous data showed high interobserver agreement with kappa of sarcopenia assessment by CT image analysis using Osirix® of 0.87 (95 % confidence interval 0.82–0.93) and Bland–Altman analysis produced 95 % limits of agreement: −5.9 to 8.7 %.14 The muscle loss index (MLI), as a measure of muscle mass loss during chemoradiotherapy, was calculated as follows: (postCRT L3 index—preCRT L3 index)/preCRT L3 index × 100 %.

Sarcopenia and Muscle Loss Index The L3 muscle area surfaces were normalized for patient height to calculate the L3 muscle index and expressed in cm2/m2. Patient height and weight were measured by a nurse at admission to the hospital at the time of diagnosis. The cutoff values used for sarcopenia were 52.4 cm2/m2 for men and 38.5 cm2/m2 for women, based on the method of Prado et al.12 Previous data showed high interobserver agreement with kappa of sarcopenia assessment by CT image analysis using Osirix® of 0.87 (95 % confidence interval 0.82–0.93) and Bland–Altman analysis produced 95 % limits of agreement: −5.9 to 8.7 %.14 The muscle loss index (MLI), as a measure of muscle mass loss during chemoradiotherapy, was calculated as follows: (postCRT L3 index—preCRT L3 index)/preCRT L3 index × 100 %. Visceral Obesity and Sarcopenic Obesity Sex-specific cutoff values for visceral obesity were 163.8 cm2 for men and 80.1 cm2 for women.15 For sarcopenic obesity, two separate definitions were used. The first definition was the combination of sarcopenia and visceral obesity (sarcopenic visceral obesity); the second definition was the combination of sarcopenia and obesity based on BMI > 30 kg/m2 (sarcopenic obesity). Visceral obesity and sarcopenic obesity were calculated at both time points (preCRT and postCRT), using height and weight values recorded at the time of diagnosis.

ral obesity (sarcopenic visceral obesity); the second definition was the combination of sarcopenia and obesity based on BMI > 30 kg/m2 (sarcopenic obesity). Visceral obesity and sarcopenic obesity were calculated at both time points (preCRT and postCRT), using height and weight values recorded at the time of diagnosis. Statistical Analysis Normality was tested using Kolmogorov–Smirnov. Frequencies are presented as absolute numbers and percentages. Continuous data are presented as mean and standard deviation (SD). Differences between groups were analyzed with the Pearson χ2 test for dichotomous parameters. Paired t test was used for between group comparison of preCRT and postCRT continuous data. Independent samples t test was used for between group comparison of continuous data on a single time point (preCRT or postCRT) and for between-group comparison of MLI and fat loss. Receiver operating characteristic (ROC) curves were used to calculate accuracy of the relative L3 index decrease to predict postoperative mortality. The ideal cutoff value for predicting mortality was defined as the cutoff value with maximum sum of sensitivity and specificity. Overall diagnostic accuracy was represented by the area under the curve (AUC). Two-tailed p values <0.05 were considered significant. All statistical analyses were performed using SPSS® 20.0 (SPSS Inc, Chicago, IL).

or predicting mortality was defined as the cutoff value with maximum sum of sensitivity and specificity. Overall diagnostic accuracy was represented by the area under the curve (AUC). Two-tailed p values <0.05 were considered significant. All statistical analyses were performed using SPSS® 20.0 (SPSS Inc, Chicago, IL). Results Patients and L3 Index During Chemoradiotherapy A total of 123 patients were included, of whom 101 (82 %) were males; 114 patients received neoadjuvant CRT. Both a preCRT and postCRT CT scan were available in 96 patients. Patient, tumor, and operation characteristics are presented in Table 1. An explanatory flowchart of patients, including those receiving CRT and availability of CT scans, is presented in Diagram 1. Sarcopenia was present in 60 of 108 patients (56 %) before CRT and in 74 of 111 patients (67 %) after CRT. Concomitantly, the mean L3-index decreased significantly during CRT, which was measured in patients with both a preCRT and a postCRT CT scan [50.9 (SD, 8.5) cm2/m2 to 48.4 (8.5) cm2/m2, p < 0.001; Fig. 1a]. In male subjects, the L3-index decreased from 53.4 (7.8) to 49.5 (7.9) cm2/m2, p < 0.001, and in female subjects the L3-index decreased from 42.7 (5.4) to 39.7 (4.4) cm2/m2, p = 0.02. Mortality within 30 days or within hospital admission was 11 of 123 (9 %) in the total cohort and 6 of 62 (10 %) in patients with stage III-IV tumors. Of note, only stage IV patients who responded well to CRT and were classified as stage III after CRT were eligible for surgery. Of patients with both a preCRT and postCRT CT scan, mortality rates were 9 of 96 (9 %) and 5 of 52 (10 %), respectively. Causes of mortality are listed in Table 2.Table 1 Patient, tumor, and operation characteristics

atients who responded well to CRT and were classified as stage III after CRT were eligible for surgery. Of patients with both a preCRT and postCRT CT scan, mortality rates were 9 of 96 (9 %) and 5 of 52 (10 %), respectively. Causes of mortality are listed in Table 2.Table 1 Patient, tumor, and operation characteristics Complete cohort (n = 123) Patients with preCRT and postCRT CT scans (n = 96) Number of patients Mean (SD) Number of patients Mean (SD) Sex Male 101 (82.1 %) 80 (83.3 %) Female 22 (17.9 %) 16 (16.7 %) Age (year) 63 (10) 63 (9.2) >70 32 (26 %) 23 (24 %) BMI (kg/m2) 24.7 (4.2) 24.7 (4.3) <18.5 12 (9.7 %) 10 (10.4 %) 18.5–24.9 54 (43.9 %) 42 (43.8 %) 25.0–29.9 44 (35.8 %) 32 (33.3 %) ≥30.0 13 (10.6 %) 12 (12.5 %) Length of hospital stay (days) 26 (33) 26 (35) ASA I 30 (25.4 %) 28 (29.2 %) II 52 (44.1 %) 44 (45.8 %) III 36 (30.5 %) 24 (25.0 %) Stage I–II 61 (49.6 %) 44 (45.8 %) III–IV 62 (50.4 %) 52 (54.2 %) Time between preCRT CT and postCRT CT (days) 111 (17) 111 (17) Time between postCRT CT and operation (days) 33 (19) 32 (19) Type of chemotherapy CF 99 (86.8 %) 84 (87.5 %) ECC 6 (5.2 %) 3 (3.1 %) PC 9 (7.9 %) 9 (9.4 %) Tumor type Adenocarcinoma 100 (81.3 %) 82 (85.4 %) Squamous cell carcinoma 23 (18.7 %) 14 (14.6 %) Tumor location Proximal esophageal 1 (0.9 %) 16 (16.7 %) Mid esophageal 20 (17.9 %) 53 (55.3 %) Distal esophageal 57 (48.7 %) 21 (21.9 %) Junctional 29 (24.8 %) 6 (6.3 %) Gastric 10 (8.5 %) 44 (45.8 %) Type of surgery Transthoracic 58 (47.2 %) 52 (54.2 %) Transhiatal 65 (52.8 %) 664 (1,085) Blood loss (mL) 664 (970)

%) Tumor location Proximal esophageal 1 (0.9 %) 16 (16.7 %) Mid esophageal 20 (17.9 %) 53 (55.3 %) Distal esophageal 57 (48.7 %) 21 (21.9 %) Junctional 29 (24.8 %) 6 (6.3 %) Gastric 10 (8.5 %) 44 (45.8 %) Type of surgery Transthoracic 58 (47.2 %) 52 (54.2 %) Transhiatal 65 (52.8 %) 664 (1,085) Blood loss (mL) 664 (970) ASA American Society of Anesthesiologists, CRT chemoradiotherapy, CF cisplatin/5-fluorouracil, ECC epirubicin/cisplatin/capecitabine, PC paclitaxel/carboplatin, BMI body mass index, calculated at time of diagnosis Fig. 1 L3 muscle index before and after neoadjuvant CRT in patients undergoing esophageal surgery. a Patients with both a preCRT and a postCRT CT scan (n = 96). b By tumor stage: stage III–IV (n = 52) versus I–II (n = 44) Table 2 Causes of mortality Number of patients Anastomotic leakage 5 Pneumonia and respiratory failure 3 Intestinal ischemia 1 Cardiac ischemia 1 Postoperative pulmonary bleeding 1

Fig. 1 L3 muscle index before and after neoadjuvant CRT in patients undergoing esophageal surgery. a Patients with both a preCRT and a postCRT CT scan (n = 96). b By tumor stage: stage III–IV (n = 52) versus I–II (n = 44) Table 2 Causes of mortality Number of patients Anastomotic leakage 5 Pneumonia and respiratory failure 3 Intestinal ischemia 1 Cardiac ischemia 1 Postoperative pulmonary bleeding 1 Influence of Tumor Characteristics on the L3 Index The mean L3 index was significantly lower before CRT began in patients with advanced (stage III–IV) tumors compared with localized tumors [49.1 (8.4) cm2/m2 and 52.9 (8.4) cm2/m2, p = 0.02; Fig. 1b]. (Male subjects: 54.3 (7.8) compared with 51.5 (7.6) cm2/m2, p = 0.02; female subjects: 44.8 (7.5) compared with 40.0 (3.3) cm2/m2, p = 0.03), respectively.) In both stage III–IV and stage I–II tumors, the L3 index decreased significantly during CRT. L3 indices were not different between patients with adenocarcinoma and patients with squamous cell carcinoma, and there was no difference between the amount of sarcopenic patients between carcinoma types.

0.03), respectively.) In both stage III–IV and stage I–II tumors, the L3 index decreased significantly during CRT. L3 indices were not different between patients with adenocarcinoma and patients with squamous cell carcinoma, and there was no difference between the amount of sarcopenic patients between carcinoma types. L3 Index Decrease as a Predictor of Mortality The MLI was not predictive for postoperative mortality directly related to complications in the total study population; −8.6 % (7.6 %) in patients who died postoperatively compared with −7.0 % (8.0 %; p = 0.57) in surviving patients. Because L3 indices differed significantly between stage III–IV and stage I–II disease, these groups were analyzed separately. In patients with stage I-II tumors, MLI was not different between patients with and those without postoperative mortality, respectively: −2.5 % (3.6 %) and −9.4 % (7.1 %), p = 0.07. In patients with stage III–IV tumors, patients who died postoperatively showed significantly higher decrease in MLI [−13.5 % (6.2 %)] compared with surviving patients [−5.0 % (8.3 %), p = 0.02; Figs. 2a, b]. An analysis for female patients could not be done, because there were no stage III–IV female patients with postoperative mortality in this cohort. Analysis of only male patients yielded comparable results.Fig. 2 Loss of muscle mass measured as decrease of L3 muscle index in patients with advanced (stage III–IV) tumors. a L3 indices before and after CRT in surviving patients (n = 47) and patients who die from postoperative complications (n = 5). b Relative L3 index decrease in surviving patients and patients who die from postoperative complications. c ROC curve of relative L3 index decrease to predict postoperative mortality

) tumors. a L3 indices before and after CRT in surviving patients (n = 47) and patients who die from postoperative complications (n = 5). b Relative L3 index decrease in surviving patients and patients who die from postoperative complications. c ROC curve of relative L3 index decrease to predict postoperative mortality Diagram 1 Proportion of patients having preCRT and postCRT scans MLI was not associated with incidence of specific complications (anastomotic leakage, abscess, empyema, sepsis, postoperative ileus, surgical site infections, pneumonia, and pneumothorax) or with length of hospital stay. To determine diagnostic accuracy of the MLI to detect postoperative mortality in patients with stage III–IV disease, ROC curve analysis was done (Fig. 2c). An optimal cutoff point of −13.2 % was found with sensitivity 80 %, specificity 85 %, positive likelihood ratio 5.33 [95 % confidence interval (CI), 2.5–11.0], and negative likelihood ratio 0.24 (95 % CI 0.04–1.4; p = 0.03). The area under the ROC curve was 0.80 (95 % CI 0.64–0.96).

C curve analysis was done (Fig. 2c). An optimal cutoff point of −13.2 % was found with sensitivity 80 %, specificity 85 %, positive likelihood ratio 5.33 [95 % confidence interval (CI), 2.5–11.0], and negative likelihood ratio 0.24 (95 % CI 0.04–1.4; p = 0.03). The area under the ROC curve was 0.80 (95 % CI 0.64–0.96). Sarcopenic Obesity as a Predictor of Postoperative Mortality At the preCRT time point, 16 patients were defined as having sarcopenic visceral obesity (sarcopenia and visceral obesity) and two patients were defined as having sarcopenic obesity (sarcopenia and high BMI; Table 3). At the postCRT time point, these were 20 and 2 patients, respectively. Sarcopenic visceral obesity was not associated with postoperative mortality (preCRT, p = 0.79; postCRT, p = 0.46). Furthermore, sarcopenic obesity was not associated with mortality (preCRT, p = 0.66; postCRT, p = 0.66). When patients with stage I–II tumors and those with stage III–IV tumors were analyzed separately, comparable results were found.Table 3 Body composition measures Number of patients Mean (SD) L3 index (cm2/m2) preCRT 50.9 (8.5) postCRT 48.4 (8.5) Visceral fat (cm2) preCRT 159.0 (98.5) postCRT 148.0 (89.5) Subcutaneous fat (cm2) preCRT 159.8 (79.5) postCRT 153.0 (80.5) Sarcopenic obesity preCRT 2 (2 %) postCRT 2 (2 %) Sarcopenic visceral obesity preCRT 16 (17 %) postCRT 20 (21 %) Height and weight values used for all calculations were recorded at time of diagnosis

sceral fat (cm2) preCRT 159.0 (98.5) postCRT 148.0 (89.5) Subcutaneous fat (cm2) preCRT 159.8 (79.5) postCRT 153.0 (80.5) Sarcopenic obesity preCRT 2 (2 %) postCRT 2 (2 %) Sarcopenic visceral obesity preCRT 16 (17 %) postCRT 20 (21 %) Height and weight values used for all calculations were recorded at time of diagnosis Fat Loss as a Predictor of Postoperative Mortality In contrast to loss of muscle mass, total fat loss was not associated with postoperative mortality in the complete study cohort, nor when stage I–II disease and stage III-IV disease were analyzed separately (stage I–II: −9.1 % (20.0 %) in patients who died postoperatively compared with −4.8 % (24.2 %) in surviving patients, p = 0.74; stage III–IV: −21.8 % (23.4 %) compared with −3.0 % (38.2 %), respectively, p = 0.29). Comparable results were found for subcutaneous fat and visceral fat.

IV disease were analyzed separately (stage I–II: −9.1 % (20.0 %) in patients who died postoperatively compared with −4.8 % (24.2 %) in surviving patients, p = 0.74; stage III–IV: −21.8 % (23.4 %) compared with −3.0 % (38.2 %), respectively, p = 0.29). Comparable results were found for subcutaneous fat and visceral fat. Discussion This study confirms the loss of muscle mass during neoadjuvant CRT in patients who undergo resection of esophageal malignancies. This is the first study that shows that loss of muscle mass, as measured on routinely obtained CT scans, is associated with postoperative mortality in patients with advanced tumors. This may provide a readily available and inexpensive assessment to identify patients at risk for developing unfavorable postoperative outcome. Intriguingly, none of the anthropometric parameters at a single time point were significantly associated with postoperative outcome, but rather the dynamic loss during neoadjuvant CRT. It is therefore important to calculate the MLI, as a difference between preCRT and postCRT. Although hypothesized, sarcopenic obesity was not associated with postoperative mortality in this study. The MLI seemed to be of more importance than the combination of sarcopenia and high BMI. No association could be found between loss of fat mass of any type and postoperative mortality. The mortality rate in this study is comparable to reported rates in recent literature, where percentages vary from 5 to 10 %.16,17

udy. The MLI seemed to be of more importance than the combination of sarcopenia and high BMI. No association could be found between loss of fat mass of any type and postoperative mortality. The mortality rate in this study is comparable to reported rates in recent literature, where percentages vary from 5 to 10 %.16,17 The application of patient-tailored approaches is upcoming in oncologic health care.18 In this context, measurements of the physical ability to recover adequately from therapeutic hits, such as surgery, is vital, particularly in the tumor-bearing patient in whom metabolic and nutritional resources are generally depleted. Considering the striking impact of metastasized disease on physiological depletion, patients with stage III and IV tumors are more prone to loss of reserves, manifesting as cachexia, which includes loss of muscle mass. The current data reflect this effect by lower L3 muscle index in patients with stage III–IV disease. This loss in muscle tissue is the result of both increased muscle protein degradation and reduced muscle synthesis. Tumors, especially in more advanced stages, may increase host’s resting energy expenditure (REE). Ravasco et al. found an elevated REE in patients with stage III–IV colorectal tumors.19 Furthermore, Bachmann et al. showed that pancreatic cancer patients with cachexia had a higher rate of more progressed tumor stages and a worse nutritional status.20 Selection of patients with increased risk for poor postoperative outcome within this vulnerable group therefore is important. Delaying the moment of surgery and with application of prehabilitation programs to regain physical resources or preventive postoperative ICU admissions may offer advantages for patients who are identified as high risk for poor outcome. During this study, patients who were diagnosed with and treated for esophageal cancer received standard care. Regarding physical activity, this means that they have received physical therapy during their hospital admission, but no additional physical intervention was provided.

identified as high risk for poor outcome. During this study, patients who were diagnosed with and treated for esophageal cancer received standard care. Regarding physical activity, this means that they have received physical therapy during their hospital admission, but no additional physical intervention was provided. The inability of a hospital to let patients with severe complications survive is known as failure to rescue (FTR). Instead of higher complication rates, higher FTR rates are the main determinant of postoperative mortality.21 FTR variability between hospitals is largely contributed to hospital-related factors, such as the available level of ICU care.22 On the other hand, variability of patient characteristics influences FTR rates as well.23 The amount of muscle mass lost during neoadjuvant CRT could be used as a predictor of FTR rates in future studies. Our data are in line with several studies. Deans and colleagues showed that weight loss, among others, was independently predictive of death after esophageal resections.24 This study focused on long-term mortality (disease prognosis); therefore, weight loss may actually be a reflection of tumor progression rather than functional depletion. In their recent study, Awad et al. showed a decrease in muscle mass during CRT as measured on the L3 level on CT scans, although a correlation with adverse postoperative outcome could not be found.5 Patients with advanced tumors were not analyzed separately from those with local tumors, as in the current study.

letion. In their recent study, Awad et al. showed a decrease in muscle mass during CRT as measured on the L3 level on CT scans, although a correlation with adverse postoperative outcome could not be found.5 Patients with advanced tumors were not analyzed separately from those with local tumors, as in the current study. As with every observational study, the current data should be taken with caution. Although the studied cohort was relatively large, numbers of complication-related deaths are small. Therefore, the observed effect should be validated in a larger, preferably multicenter or nationwide cohort. Second, sarcopenia rates were relatively high in this study using the predefined cutoff values that remain subject to debate. It is desirable to define cutoff values based on different cancer type populations. Finally, muscle and fat loss are only one aspect of functional depletion. Other parameters, such as loss of muscle function and strength, were not measured in this study. This is unfortunate, because a recent report showed that function tests may assess muscle quality, making their relationship with prognosis potentially better than that of muscle mass.25

one aspect of functional depletion. Other parameters, such as loss of muscle function and strength, were not measured in this study. This is unfortunate, because a recent report showed that function tests may assess muscle quality, making their relationship with prognosis potentially better than that of muscle mass.25 It should be addressed whether attempts to attenuate loss of muscle mass during neoadjuvant CRT are effective and whether they result in decreased FTR rates. In the studied population, efforts to counteract exhaustion of physiologic reserves by nutritional support were already done, indicating that nutrition alone is not the key to success. Several systematic reviews and meta-analyses have concluded that physical exercise may improve physical functioning and overall quality of life in cancer patients.26,27 Strength training is the most effective exercise to slow down the rate of loss of muscle mass and maintaining and improving muscle strength. The combination of physical exercise with essential amino acid ingestion elicits the greatest anabolic response.28,29 Because physical exercise is only effective in terms of improving muscle strength when performed 2–3 days a week, the feasibility of this approach in combination with the physical and mental side-effects of neoadjuvant CRT should be determined.30 Finally, eicosapentaenoic acid (EPA) can preserve lean body mass after esophageal surgery and therefore may be an interesting option to preserve muscle mass during CRT.31 Kostan W. Reisinger and Joanna W. A. M. Bosmans have contributed equally to the manuscript.

Even with neoadjuvant chemoradiotherapy (nCRT), the overall 5-year survival rate after esophagectomy remains relatively low at 47 % in patients with locally advanced esophageal cancer (EC).1 A strong prognostic indicator after a curative intended esophagectomy is the circumferential resection margin (CRM), rendered as microscopic tumor-free (R0) or tumor-positive (R1).2–8 Commonly used definitions of a circumferentially R0 resection are those of the College of American Pathologist (CAP; CRM >0 mm) and the Royal College of Pathologists (RCP; CRM >1 mm).9,10 After nCRT, the optimal CRM may be influenced by tumor downsizing, which facilitates a R0 resection.1 The optimal CRM cutoff point after nCRT has not been defined yet. Recently, two meta-analyses showed a significant association of a positive CRM according to both definitions with poor outcome, which was even worse in patients with stage T3 disease or after nCRT.11,12 However, these studies did not assess which CRM definition was more powerful after nCRT, while contradictory results after nCRT were reported in three other studies without a surgery-alone control group.13–15 Two studies, with only squamous cell carcinoma, showed a significant better survival rate in patients with a CRM > 1 mm, whereas no survival benefit was observed in R0 resections according to the CAP and RCP in a study with only T3 stage adenocarcinomas.13–15 We assessed the optimal CRM cutoff point and the prognostic value of R0 resections according to the CAP and RCP criteria in EC patients treated either with nCRT or surgery alone.

The optimal CRM cutoff point after nCRT has not been defined yet. Recently, two meta-analyses showed a significant association of a positive CRM according to both definitions with poor outcome, which was even worse in patients with stage T3 disease or after nCRT.11,12 However, these studies did not assess which CRM definition was more powerful after nCRT, while contradictory results after nCRT were reported in three other studies without a surgery-alone control group.13–15 Two studies, with only squamous cell carcinoma, showed a significant better survival rate in patients with a CRM > 1 mm, whereas no survival benefit was observed in R0 resections according to the CAP and RCP in a study with only T3 stage adenocarcinomas.13–15 We assessed the optimal CRM cutoff point and the prognostic value of R0 resections according to the CAP and RCP criteria in EC patients treated either with nCRT or surgery alone. Patients and Methods Data collection of this explorative retrospective study was provided from a prospective maintained database of EC patients according to the national guidelines and the rules approved by the local ethical commission (www.ccmo.nl). We included only patients with a locally advanced curatively resectable EC (stage II-III) treated between 1997 and 2013, in whom the CRM was adequately assessed by our expert pathologists. Of the patients treated with nCRT (n = 127) between 2005 and 2013, 23 were excluded because of the following criteria: incomplete medical records (n = 0), postoperative mortality (death within 90 days or in-hospital, n = 10), progressive disease within 3 months after surgery or microscopic irradical (R1; tumor cells <1 mm) longitudinal margins (n = 0) or follow-up <24 months (n = 13). Based on these exclusion criteria, a reference group of surgery-alone treated patients (n = 105) was constructed. Patients and tumor-related factors were matched and were equally distributed between both groups (Table 1).Table 1 Patient characteristics in the surgery-alone and neoadjuvant chemoradiotherapy (nCRT) groups

ased on these exclusion criteria, a reference group of surgery-alone treated patients (n = 105) was constructed. Patients and tumor-related factors were matched and were equally distributed between both groups (Table 1).Table 1 Patient characteristics in the surgery-alone and neoadjuvant chemoradiotherapy (nCRT) groups nCRT (n = 104) Surgery alone (n = 105) P value Male 79 (76.0 %) 82 (78.1 %) 0.714a Age (year), median (IQR) 63 (56–67) 64 (57–69) 0.228b Histology 0.382a Adenocarcinoma 88 (84.6 %) 84 (80 %) Squamous cell carcinoma 16 (15.4 %) 21 (20 %) Tumor location 0.654a Middle esophagus 8 (7.7 %) 12 (11.4 %) Distal esophagus 50 (48.1 %) 49 (46.7 %) GEJ 46 (44.2 %) 44 (41.9 %) Tumor length >5 cm 59 (56.7 %) 60 (57.1 %) 0.695b cT-stage 0.221a T2 16 (15.4 %) 9 (8.6 %) T3 83 (79.8 %) 93 (88.6 %) T4a 5 (4.8 %) 3 (2.9 %) cN-stage 0.176a N0 27 (26 %) 41 (39 %) N1 50 (48.1 %) 44 (41.9 %) N2 22 (21.2 %) 18 (17.1 %) N3 5 (4.8 %) 2 (1.9 %) pT-stage <0.001c Tx 1 (1 %) T0 21 (20.2 %) T1 22 (21.2 %) T2 14 (13.5 %) 20 (19 %) T3 46 (44.2 %) 82 (78.1 %) T4a 0 (0 %) 3 (2.9 %) pN-stage <0.001a N0 62 (59.6 %) 28 (26.7 %) N1 26 (25.0 %) 34 (32.4 %) N2 11 (10.6 %) 25 (23.8 %) N3 5 (4.8 %) 18 (17.1 %) Perineural growth 22 (21.2 %) 33 (31.4 %) 0.084a Angioinvasion 22 (21.2 %) 51 (48.6 %) <0.001a Number of LN (>4 LN+) 10 (9.6 %) 32 (30.5 %) <0.001a Lymph node ratio (>0.2) 18 (17.3 %) 50 (47.6 %) <0.001a Follow-up mo, median (IQR) 27.5 (15.0–42.0) 29 (15.5–56.0) 0.241b Tumor recurrence 63 (60.6 %) 75 (71.4 %) 0.098a Local recurrence 17 (16.3 %) 35 (33.3 %) 0.005a Death 60 (57.7 %) 83 (79 %) 0.001a

Angioinvasion 22 (21.2 %) 51 (48.6 %) <0.001a Number of LN (>4 LN+) 10 (9.6 %) 32 (30.5 %) <0.001a Lymph node ratio (>0.2) 18 (17.3 %) 50 (47.6 %) <0.001a Follow-up mo, median (IQR) 27.5 (15.0–42.0) 29 (15.5–56.0) 0.241b Tumor recurrence 63 (60.6 %) 75 (71.4 %) 0.098a Local recurrence 17 (16.3 %) 35 (33.3 %) 0.005a Death 60 (57.7 %) 83 (79 %) 0.001a Tumor-related death 54 (51.9 %) 73 (69.5 %) 0.009a CRM (mm), median (IQR) 3.3 (1.0–5.0) 0.5 (0–1.4) <0.001b 0 9 (8.7 %) 27 (25.7 %) <0.001a 0–1 13 (12.5 %) 40 (38.1 %) <0.001a >1 82 (78.8 %) 38 (36.2 %) <0.001a nCRT neoadjuvant chemoradiotherapy, cT clinical T stage, cN clinical lymph node stage, pT pathological T stage, pN pathologic lymph node stage, LN lymph node, CRM circumferential resection margin, CAP College of American Pathologists, RCP Royal College of Pathologists, IQR interquartile range a χ 2 test bMann–Whitney test cFisher exact test Tumors staged according to the 6th TNM edition were recoded into the 7th edition.16,17 Before 2000 (n = 11), staging consisted of endoscopic ultrasonography (EUS) with fine-needle aspiration (FNA), computed tomography (CT) of the neck, thorax, and abdomen and occasionally 18-F-fluorodeoxyglucose positron emission tomography (FDG-PET, n = 8). After 2000, a standard FDG-PET was added, which was replaced by FDG-PET/CT after 2009. Two weeks after nCRT, patients were restaged with a CT thorax and abdomen.

ation (FNA), computed tomography (CT) of the neck, thorax, and abdomen and occasionally 18-F-fluorodeoxyglucose positron emission tomography (FDG-PET, n = 8). After 2000, a standard FDG-PET was added, which was replaced by FDG-PET/CT after 2009. Two weeks after nCRT, patients were restaged with a CT thorax and abdomen. Treatment All patients underwent a transthoracic esophagectomy with en bloc dissection of regional mediastinal and abdominal (including the celiac trunk region) lymph nodes. Patients with nCRT were treated according to the Dutch Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study (CROSS) regimen, consisting of intravenous paclitaxel (50 mg/m2) and carboplatin (AUC: 2 ml/min), administered five times during a 5-week concurrent radiation period (41.4 Gy/23 fractions of 1.8 Gy).1 Before 2009, patients received nCRT based on their participation in the CROSS trial, from 2009 onwards nCRT became standard of care for locally advanced EC patients (T1-4aN1-3, T2-4aN0-3; n = 75).

tin (AUC: 2 ml/min), administered five times during a 5-week concurrent radiation period (41.4 Gy/23 fractions of 1.8 Gy).1 Before 2009, patients received nCRT based on their participation in the CROSS trial, from 2009 onwards nCRT became standard of care for locally advanced EC patients (T1-4aN1-3, T2-4aN0-3; n = 75). Pathology Resected specimens were examined according to a standardized protocol by two specialized gastrointestinal pathologists. The resected specimen was pinned on a Styrofoam plate by the surgeon, enabling accurate pathological assessment of the marked Clinical Tumor Volume and Gross Tumor Volume areas in patients treated with nCRT.18 CRM was measured according to the method of Quirke; the specimens were inked with Indian ink and fixed in formalin during 24 h.6 The specimens were sliced into transverse cross-sections of 0.5 cm for macroscopic assessment and sampling of at least two sections with the smallest CRM.2 The CRM was microscopically assessed on hematoxylin and eosin stained samples in tenths of millimetres. Furthermore, the pT-stage, pN-stage, the lymph-node ratio (>0.2 metastatic lymph node ratio), number of positive lymph nodes (>4), histological tumor type, tumor grade, angioinvasion, and perineural tumor growth were assessed.

s microscopically assessed on hematoxylin and eosin stained samples in tenths of millimetres. Furthermore, the pT-stage, pN-stage, the lymph-node ratio (>0.2 metastatic lymph node ratio), number of positive lymph nodes (>4), histological tumor type, tumor grade, angioinvasion, and perineural tumor growth were assessed. Follow-up Patients were followed for at least 2 years or until death, every 3 months during the first year after surgery, every 6 months in the second year, and every year thereafter for the next 10 years. Tumor recurrence was defined as histo/cytologically proven, suspected radiological imaging, or clinically evident recurrence. Local recurrence included recurrent disease at the anastomotic site or in the original tumor/mediastinal bed.

every 6 months in the second year, and every year thereafter for the next 10 years. Tumor recurrence was defined as histo/cytologically proven, suspected radiological imaging, or clinically evident recurrence. Local recurrence included recurrent disease at the anastomotic site or in the original tumor/mediastinal bed. Statistical Analysis Distribution of continuous patient characteristics was reported as median [interquartile range] and categorical variables were reported in numbers and percentages. The patients groups were compared with the Mann–Whitney test for continuous variables and χ2 or Fisher exact test for categorical response variables. Kaplan–Meier curves and log-rank test determine the 5-year disease-free survival (DFS) and local recurrence-free survival (LRFS) of both CRM definitions. Prognostic values of all variables for 2-year DFS were assessed with univariate Cox regression analysis. Factors within the univariate analysis were: age, tumor type, and grade (G1–2 vs. G3–4), clinical T and N stage, tumor length (>5 cm, measured endoscopic or with CT), treatment type (nCRT or surgery alone), and pathologic outcome: T and N stage, number of LN metastases (>4), and metastatic lymph node ratio (>0.2), perineural growth, and angioinvasion. Multivariate Cox regression was performed by incorporating all variables with a P value <0.1 on univariate analysis. Both, the CAP (CRM >0 mm) and RCP (CRM >1 mm) definition entered the multivariate analysis separately. The prognostic value of R0 resections according to the RCP and CAP for the 2-year DFS and 2-year LRFS was assessed with multivariate Cox regression analyses in both treatment groups. To assess the optimal cutoff value of the CRM on 2-year DFS, an explorative analysis was performed in both groups. Univariate analyses were undertaken to assess the prognostic value of all cutoff values (from 0.0 to 1.0 mm). The observed interval is based on the assumption that the expected optimal CRM cutoff should be between 0.0 and 1.0 mm. The Akaike Information Criterion (AIC), which quantifies the quality of a statistical model for a set of data was used to indirectly compare the prognostic value of the CAP and RCP model.19 It penalizes the number of explanatory variables by adding twice the number of variables in the model to the −2 log likelihood; in a formula AIC = −2 log likelihood +2 k, in which k is the number of explanatory variables in the model. The model with the lowest AIC was considered to be most prognostic.

and RCP model.19 It penalizes the number of explanatory variables by adding twice the number of variables in the model to the −2 log likelihood; in a formula AIC = −2 log likelihood +2 k, in which k is the number of explanatory variables in the model. The model with the lowest AIC was considered to be most prognostic. The backwards likelihood ratio method was used in the Cox regression analysis. Analyses were performed with SPSS version 22. Results Patient characteristics are summarized in Table 1. All nCRT patients with CAP-R1 resections (n = 9; 8.7 %) had stage pT3. Of the 27 (25.7 %) R1 resections in patients treated with surgery alone, 24 had stage pT3, 1 had pT2, and 2 had stage pT4a disease. The median CRM differed significantly with 3.3 [interquartile range (IQR) 1.0–5.0] mm versus 0.5 (IQR 0–1.4) mm for the nCRT and surgery-alone group, respectively. The median follow up was 29.0 (IQR 15.5–56.0) months and 27.5 (IQR 15.0–42.0) in the surgery-alone and nCRT groups, respectively.

2 had stage pT4a disease. The median CRM differed significantly with 3.3 [interquartile range (IQR) 1.0–5.0] mm versus 0.5 (IQR 0–1.4) mm for the nCRT and surgery-alone group, respectively. The median follow up was 29.0 (IQR 15.5–56.0) months and 27.5 (IQR 15.0–42.0) in the surgery-alone and nCRT groups, respectively. Prognostic Value of the CAP and RCP Criteria Figure 1 displays the DFS of both treatment groups, with a R0 resection or involved CRM (R1 resection) according to CAP (Fig. 1a) and RCP (Fig. 1b). With the log-rank test, the CAP definition was prognostic for 5-year DFS in both the surgery (P = 0.008) and nCRT group (P < 0.001) and the RCP definition was prognostic in the nCRT group (P < 0.001) but not in the surgery group (P = 0.071). The 5-year DFS was not different (P = 0.131) between CAP R1 patients treated with or without nCRT but differed (P = 0.031) between patients with an RCP R1 resection in both groups.Fig. 1 Disease-free survival in patients treated with neoadjuvant chemoradiotherapy and surgery-alone, with circumferential microscopic tumor-free (R0) or involved resection margins (R1), according to a CAP (0 mm) and b RCP (1 mm)

ut nCRT but differed (P = 0.031) between patients with an RCP R1 resection in both groups.Fig. 1 Disease-free survival in patients treated with neoadjuvant chemoradiotherapy and surgery-alone, with circumferential microscopic tumor-free (R0) or involved resection margins (R1), according to a CAP (0 mm) and b RCP (1 mm) Table 2 displays all prognostic factors with a P < 0.1 on univariate analysis and Table 3 shows the multivariate Cox regression models containing either the CAP or RCP for 2-year DFS and LRFS in both groups. Independent prognostic factors for 2-year DFS in the surgery-alone group were tumor length [P = 0.006, hazard ratio (HR) 2.68, CI 1.33–5.43], lymph node ratio (P = 0.047, HR 2.57, CI 1.01–6.51), and CAP (P = 0.012, HR 0.41, CI 0.21–0.83). Independent prognostic factors for 2-year LRFS were lymph node ratio (P = 0.020, HR 3.11, CI 1.20–8.09), tumor length (P = 0.002, HR 10.99, CI 2.49–48.43), and CAP (P = 0.004, HR 0.27, CI 0.11–0.658). Both for 2-year DFS and LRFS, the model containing the CAP had a lower AIC than the RCP model and therefore was more prognostic.Table 2 Prognostic factors with P < 0.1 on univariate analysis for disease-free and local recurrence-free survival in the surgery-alone and neoadjuvant chemoradiotherapy groups Surgery-alone group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value pN0 1.00 0.001a 1.00 0.007a

Table 2 displays all prognostic factors with a P < 0.1 on univariate analysis and Table 3 shows the multivariate Cox regression models containing either the CAP or RCP for 2-year DFS and LRFS in both groups. Independent prognostic factors for 2-year DFS in the surgery-alone group were tumor length [P = 0.006, hazard ratio (HR) 2.68, CI 1.33–5.43], lymph node ratio (P = 0.047, HR 2.57, CI 1.01–6.51), and CAP (P = 0.012, HR 0.41, CI 0.21–0.83). Independent prognostic factors for 2-year LRFS were lymph node ratio (P = 0.020, HR 3.11, CI 1.20–8.09), tumor length (P = 0.002, HR 10.99, CI 2.49–48.43), and CAP (P = 0.004, HR 0.27, CI 0.11–0.658). Both for 2-year DFS and LRFS, the model containing the CAP had a lower AIC than the RCP model and therefore was more prognostic.Table 2 Prognostic factors with P < 0.1 on univariate analysis for disease-free and local recurrence-free survival in the surgery-alone and neoadjuvant chemoradiotherapy groups Surgery-alone group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value pN0 1.00 0.001a 1.00 0.007a pN1 3.36 1.10–10.32 0.034 4.08 0.87–19.26 0.076 pN2 7.36 2.47–21.92 0.000 3.39 0.62–18.56 0.160 pN3 8.05 2.58–25.05 0.000 11.54 2.48–53.79 0.002 Tumor length 2.23 1.14–4.34 0.019 8.45 1.97–36.21 0.004 Perineural growth 2.00 1.11–3.60 0.021 NS Angioinvasion 2.90 1.49–5.65 0.002 NS Number of LN 4.01 2.21–7.26 <0.001 3.41 1.46–7.97 0.005 Lymph node ratio 3.96 2.08–7.55 <0.001 3.51 1.44–8.57 0.006 CAP R0 0.45 0.25–0.82 0.010 0.42 0.18–0.98 0.044 RCP R0 0.46 0.23–0.91 0.025 0.52 0.20–1.32 0.168 nCRT group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value cT2 1.00 0.013a NS cT3 5.13 1.24–21.22 0.024 cT4a 12.98 2.36–71.45 0.003 pT0 1.00 0.054a NS pT1 1.13 0.36–3.49 0.837 pT2 1.83 0.61–5.44 0.279 pT3 2.74 1.13–6.64 0.025 pN0 1.00 <0.001a 1.00 0.047a

044 RCP R0 0.46 0.23–0.91 0.025 0.52 0.20–1.32 0.168 nCRT group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value cT2 1.00 0.013a NS cT3 5.13 1.24–21.22 0.024 cT4a 12.98 2.36–71.45 0.003 pT0 1.00 0.054a NS pT1 1.13 0.36–3.49 0.837 pT2 1.83 0.61–5.44 0.279 pT3 2.74 1.13–6.64 0.025 pN0 1.00 <0.001a 1.00 0.047a pN1 2.58 1.30–5.13 0.007 0.56 0.07–4.65 0.590 pN2 4.69 2.14–10.30 0.000 5.37 1.28–22.55 0.022 pN3 6.87 2.47–19.11 0.000 7.31 0.76–70.88 0.086 Perineural growth 1.87 0.97–3.38 0.062 NS Angioinvasion 1.81 0.98–3.53 0.055 NS Number of LN 3.90 1.90–7.98 <0.001 8.00 1.92–33.25 0.004 Lymph node ratio 2.78 1.48–5.20 0.001 4.31 1.25–14.95 0.021 CAP R0 0.28 0.13–0.61 0.001 0.42 0.18–0.98 <0.001 RCP R0 0.40 0.22–0.74 0.003 0.30 0.09–1.06 0.061 DFS disease-free survival, LRFS local recurrence free survival, CI confidence interval, cT clinical T stage, cN clinical lymph node stage, pT pathological T stage, pN pathologic lymph node stage, LN lymph node, CRM circumferential resection margin, R0 tumor-free resection margin, CAP College of American Pathologists, RCP Royal College of Pathologists, NS not significant aOverall P value of the categorical variables Table 3 Multivariate analysis of models containing the CRM definition according to the CAP (CRM 0 mm) or the RCP (CRM 1 mm), in the surgery-alone and neoadjuvant chemoradiotherapy groups Surgery-alone group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value CAP model (AIC = 317.0) CAP model (AIC = 168.2) CAP 0.41 0.21–0.83 0.012a CAP 0.27 0.11–0.658 0.004a LN ratio 2.57 1.01–6.51 0.047a LN ratio 3.11 1.20–8.09 0.020a

Table 3 Multivariate analysis of models containing the CRM definition according to the CAP (CRM 0 mm) or the RCP (CRM 1 mm), in the surgery-alone and neoadjuvant chemoradiotherapy groups Surgery-alone group 2-year DFS 2-year LRFS HR 95 % CI P value HR 95 % CI P value CAP model (AIC = 317.0) CAP model (AIC = 168.2) CAP 0.41 0.21–0.83 0.012a CAP 0.27 0.11–0.658 0.004a LN ratio 2.57 1.01–6.51 0.047a LN ratio 3.11 1.20–8.09 0.020a Tumor length 2.68 1.33–5.43 0.006a Tumor length 10.99 2.49–48.43 0.002a Angioinvasion 1.90 0.94–3.85 0.075 No. of LN+ 2.13 0.92–4.95 0.078 RCP model (AIC = 320.9) RCP model (AIC = 174.2) RCP 0.83 0.38–1.78 0.627  RCP 0.57 0.22–1.51 0.258 LN ratio 2.68 1.05–6.81 0.039a LN ratio 3.13 1.12–8.24 0.021a Angioinvasion 1.95 0.94–4.03 0.072 Perineural growth 1.89 0.99–3.59 0.053 Tumor length 2.52 1.25–5.09 0.010a Tumor length 8.59 1.99–37.08 0.004a No. of LN 1.92 0.84–4.39 0.123 nCRT group 2-year DFS 2-year LR HR 95 % CI P value HR 95 % CI P value CAP model (AIC = 349.9) CAP model (AIC = 73.5) CAP 0.47 0.18–1.23 0.124  CAP 0.06 0.01–0.31 0.001a cT 3.20 0.76–13.49 0.114 pN0 1.00 0.004a,b Grade 16.91 2.12–135.05 0.008a pN1 2.70 1.31–5.59 0.007 pN2-3 3.39 1.43–8.03 0.005 RCP model (AIC = 350.3) RCP model (AIC = 80.0) RCP 0.69 0.31–1.52 0.359  RCP 1.01 0.08–13.50 0.995 cT 1.00 0.275  pN0–1 1.00 0.203 pN1–2 8.81 0.31–252.23 2.32 0.51–10.51 pN0 1.00 0.014a,b Grade 30.07 2.79–324.60 0.005a pN1 2.51 1.22–5.21 0.014 No. of LN 0.73 0.06–8.94 0.804 pN2–3 2.99 1.23–7.32 0.016 pT0–1 1.00 0.359b LN ratio 2.60 0.44–15.42 0.294 pT2 1.96 0.70–5.49 0.202 pT3–4a 1.84 0.71–4.73 0.209

pN1 2.70 1.31–5.59 0.007 pN2-3 3.39 1.43–8.03 0.005 RCP model (AIC = 350.3) RCP model (AIC = 80.0) RCP 0.69 0.31–1.52 0.359  RCP 1.01 0.08–13.50 0.995 cT 1.00 0.275  pN0–1 1.00 0.203 pN1–2 8.81 0.31–252.23 2.32 0.51–10.51 pN0 1.00 0.014a,b Grade 30.07 2.79–324.60 0.005a pN1 2.51 1.22–5.21 0.014 No. of LN 0.73 0.06–8.94 0.804 pN2–3 2.99 1.23–7.32 0.016 pT0–1 1.00 0.359b LN ratio 2.60 0.44–15.42 0.294 pT2 1.96 0.70–5.49 0.202 pT3–4a 1.84 0.71–4.73 0.209 DFS disease-free survival, LRFS local recurrence-free survival, CI confidence interval, SCC squamous cell carcinoma, cT clinical T stage, cN clinical lymph node stage, pT pathological T stage, pN pathologic lymph node stage, LN lymph node, CRM circumferential margin, R0 tumor-free resection margin, CAP College of American Pathologists, RCP Royal College of Pathologists aSignificant (P < 0.05) bOverall P value of the categorical variables The only independent prognostic factors for 2-year DFS in the nCRT group was the pN-stage (overall P = 0.004), pN1 (P = 0.007, HR 2.70, CI 1.31–5.59), and pN2–3 (P = 0.005, HR 3.39, CI 1.43–8.03). Both CAP (P = 0.001, HR 0.06, CI 0.01–0.31) and tumor grade (P = 0.008, HR 16.91, CI 2.12–135.05) were prognostic for 2-year LRFS. For both 2-year DFS and LRFS, the multivariate regression model containing the CAP definition had a lower AIC and therefore was more prognostic.

nd pN2–3 (P = 0.005, HR 3.39, CI 1.43–8.03). Both CAP (P = 0.001, HR 0.06, CI 0.01–0.31) and tumor grade (P = 0.008, HR 16.91, CI 2.12–135.05) were prognostic for 2-year LRFS. For both 2-year DFS and LRFS, the multivariate regression model containing the CAP definition had a lower AIC and therefore was more prognostic. Optimal CRM after Surgery Alone and after nCRT CRM cutoff values of 0.0 (P = 0.012, HR = 0.41, CI 0.21–0.83, AIC = 317.0), 0.1 (P = 0.045, HR = 0.50, CI 0.25–0.98, AIC = 320.0), and 0.2 mm (P = 0.028, HR = 0.48, CI 0.25–0.92, AIC = 318.8) were independent prognostic factors for 2-year DFS in the surgery-alone group. Based on the AIC, the 0.0-mm cutoff value (CAP) was the most prognostic. However, in the nCRT group, the optimal cutoff value for 2-year DFS was 0.3 mm (P = 0.045, HR = 0.35, CI 0.13–0.98, AIC = 348.1). Discussion The prognostic value of the circumferential margin (CRM) has been proven in EC patients after surgery alone, but its significance after neoadjuvant treatment is not well defined yet. This study conducted in stage II-III EC patients showed that both definitions of a free CRM were not prognostic for 2-year DFS in patients treated with nCRT. The CAP definition (>0 mm), however, was an independent prognostic factor for 2-year DFS in the surgery-alone group and for LRFS in the nCRT and surgery-alone group. The optimal CRM cutoff value for 2-year DFS was 0.3 and between 0.0 and 0.2 mm in the nCRT and surgery-alone group, respectively.

ients treated with nCRT. The CAP definition (>0 mm), however, was an independent prognostic factor for 2-year DFS in the surgery-alone group and for LRFS in the nCRT and surgery-alone group. The optimal CRM cutoff value for 2-year DFS was 0.3 and between 0.0 and 0.2 mm in the nCRT and surgery-alone group, respectively. This study is one of the first to assess the optimal cutoff value of the CRM after nCRT; previously published studies used either the RCP or CAP criteria of a free CRM. Although neoadjuvant treatment decreases the rate of R1 resection by transversal and sagittal tumor reduction, the induced fibrosis may contain different amounts of undetectable viable tumor cells.1 Therefore, the CRM assessment depends upon accurate histological examination of residual tumor, which might be related to tumor heterogeneity. CRM >1 mm showed to be prognostic, but several studies reported conflicting results in patients treated with nCRT (Table 4). Chao et al. described a significantly better disease-free and disease-specific survival, whereas Liu et al. noted a significantly better overall survival (OS).13,14 However, Harvin et al. failed to prove a survival benefit after nCRT with respect to both CAP and RCP–CRM resections.15 This difference might be explained by the inclusion of different pathologic tumor types; Harvin et al. only included ypT3 or higher adenocarcinomas, whereas Chao et al. and Liu et al. included only patients with squamous cell carcinomas.13–15 In our study, histologic tumor type did not to affect the prognostic value of the CRMs for DFS and LRFS, although the number of squamous cell carcinomas in the nCRT group was rather small (n = 16). Inclusion of pathologic T3 tumors in determining the optimal CRM seems comprehensible as circumferential R1 resections in pT2 tumors are generally considered to be caused by inadequate surgery.7,20,21 Moreover, Rao et al. stated that CRM involvement in the EC specimen is related to advanced disease rather than being an indicator of completeness of resection.4 In our study, only one patient staged as ypT2 disease had a R1 resection, due to extensive angioinvasive tumor growth within the CRM, which depends more on biologic aggressiveness rather than poor surgery. Another factor that might influence the CRM is the used surgical method; Suttie et al.

completeness of resection.4 In our study, only one patient staged as ypT2 disease had a R1 resection, due to extensive angioinvasive tumor growth within the CRM, which depends more on biologic aggressiveness rather than poor surgery. Another factor that might influence the CRM is the used surgical method; Suttie et al. noted that the transhiatal approach resulted in significantly more CRM involvement compared with the transthoracic approach.22 Because the transthoracic approach is our standard method, we could disregard this potential confounding effect.Table 4 Studies regarding prognostic value of the circumferential resection margin after neoadjuvant chemoradiotherapy Study (year) Histology Stage Patients (n) nCRT (%) Outcome CRM definition P valuea Thompson et al.23 AC, SCC cT1–4 240 124 (52 %) 5-year survival RCP NS Chao et al.13 SCC ypT3 151 151 (100 %) LRFS RCP <0.05 DFS RCP <0.05 DSS RCP <0.05 Harvin et al.15 AC ypT3 160 160 (100 %) OS, DFS, LRFS CAP NS OS, DFS, LRFS RCP NS Reid et al.25 AC SCC cT1–4 269 42 (16 %) DFS RCP <0.01 OS RCP 0.05 O’Farrell et al.24 AC, SCC, others cT3 157 82 (52 %) OS RCP NS OS CAP 0.02 Liu et al.14 SCC cT1–4 94 94 (100 %) OS RCP <0.01 SCC squamous cell carcinoma, AC adenocarcinoma, nCRT neoadjuvant chemoradiotherapy, cT clinical T stage, ypT pathologic T stage after nCRT, DFS disease-free survival, LRFS local recurrence-free survival, DSS disease-specific survival, CRM circumferential resection margin, CAP College of American Pathologists, RCP Royal College of Pathologists aMultivariate analysis

SCC squamous cell carcinoma, AC adenocarcinoma, nCRT neoadjuvant chemoradiotherapy, cT clinical T stage, ypT pathologic T stage after nCRT, DFS disease-free survival, LRFS local recurrence-free survival, DSS disease-specific survival, CRM circumferential resection margin, CAP College of American Pathologists, RCP Royal College of Pathologists aMultivariate analysis Three other studies assessed the value of the CRM in which only a part of the included patients received nCRT, again with conflicting results.23–25 Thompson et al. (n = 240, 52 % nCRT) did not find a survival benefit, whereas Reid et al. (n = 269, 15,6 % nCRT) found a significantly better DFS and OS in patients with a RCP R0 resection.23,25 Farrell et al. (n = 157, 52 % nCRT) found the CAP definition (P = 0.02) more prognostic for the OS than the RCP definition.24 As in patients treated with nCRT, the optimal CRM definition in surgically treated patients also is unclear. Two recent meta-analyses showed that both CRM definitions were associated with a poor survival, although the CAP criteria differentiated higher-risk groups.11,12 Moreover Chan et al. found that the CAP definition, based on the hazard ratio and subgroup analysis, had a prognostic advantage over the RCP criteria.12 Concordant to these results, we found that the optimal CRM cutoff value in the surgery-alone group, analyzed with the Akaike Information Criterion, was the CAP.

.11,12 Moreover Chan et al. found that the CAP definition, based on the hazard ratio and subgroup analysis, had a prognostic advantage over the RCP criteria.12 Concordant to these results, we found that the optimal CRM cutoff value in the surgery-alone group, analyzed with the Akaike Information Criterion, was the CAP. Beside the CRM, lymph node metastasis associated variables were important prognostic factors in this study; lymph node ratio >0.2 was independent prognostic for both 2-year DFS and LRFS in the surgery-alone group and pN-stage was the only prognostic factor for 2-year DFS in the nCRT group. One meta-analysis, which underlined the importance of lymph node metastasis, indicated that nodal metastases appeared to negate the prognostic value of the CRM.12 Moreover, the presence of lymph node metastases and an involved CRM indicated a more advanced-staged disease.26 Another prognostic factor in surgery-alone patients was the tumor length, which is in correspondence with previously published data.27 Pultrum et al. assessed the optimal CRM in surgically treated patients using the area under the curve (AUC) analysis on receiver operating curves (ROC, which does not incorporate the time factor.2 A method that includes the time factor is the more complex time-dependent ROC method according to Heagerty et al.28 For our limited explorative study, however, we prefer to use multivariate Cox regression analysis and suggest validating the results in a larger cohort.

(ROC, which does not incorporate the time factor.2 A method that includes the time factor is the more complex time-dependent ROC method according to Heagerty et al.28 For our limited explorative study, however, we prefer to use multivariate Cox regression analysis and suggest validating the results in a larger cohort. Conclusions This study showed that both definitions of a tumor-free CRM (CAP > 0 mm, RCP > 1 mm) were not prognostic for DFS in patients treated with nCRT. A free CRM according the CAP definition was prognostic for 2-year DFS in the surgery-alone group and an optimal CRM cutoff between 0.0 and 0.2 and at 0.3 mm in the surgery-alone and nCRT groups, respectively. These findings should be validated in a large, prospective study. J. B. Hulshoff has controlled over data. Disclosure The authors have nothing to disclose.

Skin metastases from breast cancer (BC) are often symptomatic for ulceration, bleeding, and pain, and they may represent a challenge for clinicians, particularly in heavily pretreated patients. Surgical resection, radiotherapy, and systemic therapies can be variously combined according to individual patient characteristics, tumor features, and physician choice.1 When surgical excision is not possible, radiotherapy ensures sustained local control, even if this is not feasible in preirradiated areas.2 Systemic therapies, such as endocrine treatment, chemotherapy, and targeted agents, represent valuable options, depending on the molecular subtype of BC and prior therapies.3 Application of topic chemotherapy and laser ablation is limited to cancers confined to the top layer of skin.4 Electrochemotherapy (ECT) combines the administration of a poorly permeant cytotoxic agent, such as bleomycin (BLM), with the local application of electric pulses that induce reversible electroporation, thus improving drug diffusion into cells.5 ECT was introduced in 2006, demonstrating a high rate of efficacy and favorable toxicity profile in a European multicenter study on skin metastases from different tumor histotypes.6 In this study, the objective response (OR) rate on treated tumor nodules was 89.0 % with complete regression in 73.3 % of cases. A recently published meta-analysis including 47 prospective studies comparing five skin-directed therapies (ECT, radiation, photodynamic therapy, intralesional therapy, and topical therapy), ECT demonstrated an OR rate of 75.4 % (CR rate, 47.5 %) with a low toxicity profile (grade 3 in less than 6 % of patients).7 In this analysis, melanoma and BC comprised 96.8 % of all cutaneous metastases, with similar response rates.

es (ECT, radiation, photodynamic therapy, intralesional therapy, and topical therapy), ECT demonstrated an OR rate of 75.4 % (CR rate, 47.5 %) with a low toxicity profile (grade 3 in less than 6 % of patients).7 In this analysis, melanoma and BC comprised 96.8 % of all cutaneous metastases, with similar response rates. To our knowledge, published data on ECT in BC patients with cutaneous skin metastases are based on small, single-center, heterogeneous series. Consequently, these series do not allow for identification of clinical and/or biologic factors that are reliably predictive of ECT response.8 The aim of our study was to provide a systemic analysis on a large series of BC patients treated with ECT, evaluating potential predictive factors of response to treatment.

ries. Consequently, these series do not allow for identification of clinical and/or biologic factors that are reliably predictive of ECT response.8 The aim of our study was to provide a systemic analysis on a large series of BC patients treated with ECT, evaluating potential predictive factors of response to treatment. Patients and Methods Patients Between January 2010 and June 2013, the Italian Senological Group for Electrochemotherapy (GISEL), involving 13 Italian institutions, performed this multicenter retrospective cohort study. Inclusion criteria for ECT included BC patients with cutaneous and/or subcutaneous histologically confirmed metastases. Exclusion criteria for ECT included tumors in close proximity to a cardiac pacemaker; allergy to BLM; prior cumulative dose of BLM exceeding 250,000 IU/m2; serum creatinine >150 μmol/L; lung fibrosis; and pregnancy or lactation. Patients were enrolled regardless of the presence of other metastases. The respective institutional review boards of the participating institutions approved the study. All patients gave informed consent for the procedure and for utilization of their data for scientific purposes. Clinical records were anonymously entered into a dedicated encrypted online database.

ence of other metastases. The respective institutional review boards of the participating institutions approved the study. All patients gave informed consent for the procedure and for utilization of their data for scientific purposes. Clinical records were anonymously entered into a dedicated encrypted online database. Evaluation of Estrogen Receptor (ER), Progesterone Receptor (PgR), HER2 Status, and Ki-67 Index Histologic diagnosis, immunohistochemical analysis, and fluorescence in situ hybridization for HER2 gene amplification (in case of inconclusive results on HER2 status) were performed according to international guidelines. The cutoff for ER and PgR positivity was 1 % of cells with positive nuclear staining.9 Positivity for HER2 was determined by either immunohistochemistry 3+ or fluorescence in situ hybridization amplification. The cutoff point for the Ki-67 labeling index was 14 %.10 Surrogate subtypes were defined according to the criteria established by the St. Gallen International Breast Cancer Conference.11 Treatment The European Standard Operative Procedures of Electrochemotherapy (ESOPE) were used for all patients.12 Accordingly, the dose and route of BLM administration were adapted to the number and size of tumors in case of intratumoral injection, and to the patient’s body surface area in case of intravenous infusion. The procedure was scheduled in a day-hospital regimen, and patients were usually discharged after an observation period of 24 h.

e dose and route of BLM administration were adapted to the number and size of tumors in case of intratumoral injection, and to the patient’s body surface area in case of intravenous infusion. The procedure was scheduled in a day-hospital regimen, and patients were usually discharged after an observation period of 24 h. Patient Assessment Patients were evaluated after 1 and 2 weeks for acute toxicity and at 4 and 8 weeks for late toxicity and tumor response; subsequent follow-up visits were planned every 3–4 months. Among 125 patients, 12 (9.6 %) were followed for less than 2 months after ECT and were not considered for assessment of response. For each patient, up to a maximum of five measurable tumors were registered as target lesions. The sum of their maximum diameters represented the baseline measurement for assessment of tumor response, which was clinically performed by Response Evaluation Criteria in Solid Tumors 1.1.13 In case of the presence of many confluent nodules, when it was impossible to count their exact number, they were considered as a single entity and measured as a single area of treatment. Treatment toxicity and adverse events were graded according to the Common Terminology Criteria for Adverse Events 4.14 Pain was graded according to a 0–10 numeric pain intensity scale (0 = no pain, 10 = maximum pain).15 Statistical Analysis In descriptive analyses, continuous variables are reported as median value and interquartile range and categorical variables are reported as absolute number and percentage.

For each patient, up to a maximum of five measurable tumors were registered as target lesions. The sum of their maximum diameters represented the baseline measurement for assessment of tumor response, which was clinically performed by Response Evaluation Criteria in Solid Tumors 1.1.13 In case of the presence of many confluent nodules, when it was impossible to count their exact number, they were considered as a single entity and measured as a single area of treatment. Treatment toxicity and adverse events were graded according to the Common Terminology Criteria for Adverse Events 4.14 Pain was graded according to a 0–10 numeric pain intensity scale (0 = no pain, 10 = maximum pain).15 Statistical Analysis In descriptive analyses, continuous variables are reported as median value and interquartile range and categorical variables are reported as absolute number and percentage. Evaluation of tumor response was performed by contingency tables and Pearson’s χ2 test. Survival curves were estimated by the Kaplan–Meier method and compared to the log rank test. Hazard ratios were calculated by a Cox proportional risk model, after proportional hazard assumption confirmation with Schoenfeld residuals. Local progression-free survival (LPFS) was calculated from achievement of response in the treated area to local progression of disease, including the appearance of new nodules in the same area, or last follow-up. Statistical analyses were performed by SPSS 22.0 (IBM) software.

on confirmation with Schoenfeld residuals. Local progression-free survival (LPFS) was calculated from achievement of response in the treated area to local progression of disease, including the appearance of new nodules in the same area, or last follow-up. Statistical analyses were performed by SPSS 22.0 (IBM) software. Results Patient and Disease Characteristics Baseline patient and disease characteristics are reported in Table 1. The prevalent tumor histotype was infiltrating ductal carcinoma (76.6 %). Tumor, node, metastasis classification of primary BC was T1–T2 in 48 % patients and T3–T4 in 52 %; 67.2 % patients had lymph node involvement, and 22.4 % had distant metastases.Table 1 Patient characteristics (n = 125) Characteristic Median (range) or n (%) Age (years) 63 (54–72) Histology IDC 97 (76.6) Non-IDC 28 (23.4) Time since occurrence of skin metastases (mo) 32 (9–109) Skin metastases (n = 239) No. per patient 1 (1–3) Size (mm) 21 (15–45) Location Chest 222 (92.9) Other site 17 (7.1) Skin condition Previous radiotherapy 92 (38.5) Lymphedema 30 (12.6) Ulceration 64 (26.8) Immunohistochemistry ER positive 72 (57.6) PgR positive 72 (57.6) HER2 overexpression 35 (28.0) Ki-67 < 14 % 63 (50.4) Surrogate subtypesa Luminal A-like 23 (18.4) Luminal B-like (HER2 negative) 22 (17.6) Luminal B-like (HER2 positive) 18 (14.4) Triple negative 35 (28) HER2 11 (8.8) Previous treatmentsb Radiotherapy 68 (54.4) Chemotherapy 92 (73.6) Endocrine therapy 71 (56.8) Targeted therapy 14 (11.2) Surgery for skin metastases 89 (71.2)

Characteristic Median (range) or n (%) Age (years) 63 (54–72) Histology IDC 97 (76.6) Non-IDC 28 (23.4) Time since occurrence of skin metastases (mo) 32 (9–109) Skin metastases (n = 239) No. per patient 1 (1–3) Size (mm) 21 (15–45) Location Chest 222 (92.9) Other site 17 (7.1) Skin condition Previous radiotherapy 92 (38.5) Lymphedema 30 (12.6) Ulceration 64 (26.8) Immunohistochemistry ER positive 72 (57.6) PgR positive 72 (57.6) HER2 overexpression 35 (28.0) Ki-67 < 14 % 63 (50.4) Surrogate subtypesa Luminal A-like 23 (18.4) Luminal B-like (HER2 negative) 22 (17.6) Luminal B-like (HER2 positive) 18 (14.4) Triple negative 35 (28) HER2 11 (8.8) Previous treatmentsb Radiotherapy 68 (54.4) Chemotherapy 92 (73.6) Endocrine therapy 71 (56.8) Targeted therapy 14 (11.2) Surgery for skin metastases 89 (71.2) IDC invasive ductal carcinoma, ER estrogen receptor, PR progesterone receptor aAssessed on 113 patients, according to St. Gallen consensus11 bAny setting The median number of target lesions was 1 (range 1–3), with a median size of 21 mm (range 15–45 mm). The overwhelming majority of lesions—222 (92.9 %) of 239—were localized on the chest wall.

IDC invasive ductal carcinoma, ER estrogen receptor, PR progesterone receptor aAssessed on 113 patients, according to St. Gallen consensus11 bAny setting The median number of target lesions was 1 (range 1–3), with a median size of 21 mm (range 15–45 mm). The overwhelming majority of lesions—222 (92.9 %) of 239—were localized on the chest wall. Forty-one patients (32.8 %) received chemotherapy in a neoadjuvant setting at the time of primary BC, while 62 patients (49.6 %) underwent chemotherapy in an adjuvant setting. Seventy-one patients (56.8 %) received adjuvant endocrine treatment. All patients had received at least one previous systemic treatment for metastatic disease. Specifically, 39 patients (31.2 %) received chemotherapy (median of two lines of treatment, range 1–6) and 69 patients received endocrine therapy (median two lines of treatment, range 1–3). Fifty-three patients (42.4 %) underwent adjuvant radiotherapy and 15 patients (12.0 %) were irradiated for the presence of skin metastases. As a result, 92 (38.5 %) of 239 target lesions in the present study were located in preirradiated skin. There were more previous systemic treatments in patients with triple negative (median 3, range 1–7) and HER2 positive (median 3, range 2–6) BC than in patients with luminal A-like (median 1, range 0–6), luminal B-like (median 2, range 0–6), and luminal B-like, HER2-positive tumors (median 2, range 0–5, P = 0.042).

kin. There were more previous systemic treatments in patients with triple negative (median 3, range 1–7) and HER2 positive (median 3, range 2–6) BC than in patients with luminal A-like (median 1, range 0–6), luminal B-like (median 2, range 0–6), and luminal B-like, HER2-positive tumors (median 2, range 0–5, P = 0.042). Treatment In 92 (73.6 %) of 125 patients, ECT was administered under general anesthesia or sedation, while local anesthesia was used in the remaining 33 patients (26.4 %). BLM was administered intravenously in 100 patients (80.0 %) and intratumorally in 25 (20.0 %). Of 239 tumors, 207 (86.6 %) were electroporated with a hexagonal-array needle electrode, 10 (4.2 %) with a linear-row needle electrode, 13 (5.4 %) with a plate electrode, and 9 (3.8 %) with multiple electrode types. Toxicity No serious adverse events were reported during the procedure. Toxicity data reported within the first 2 months are presented in Table 2. Paracetamol and nonsteroidal anti-inflammatory agents were effective in controlling postprocedural pain in all but four patients, who required narcotics, although as a single administration. The incidence of skin ulceration did not differ significantly depending on previous radiation (41.3 % of previous skin radiations vs. 30 % no previous skin radiation, P = 0.436). After the first ECT, 96 patients were asked if they would agree to receive another course of treatment, if required, and 96.9 % declared that they were potentially favorable.Table 2 Toxicity within 2 months after electrochemotherapy (n = 125)

of previous skin radiations vs. 30 % no previous skin radiation, P = 0.436). After the first ECT, 96 patients were asked if they would agree to receive another course of treatment, if required, and 96.9 % declared that they were potentially favorable.Table 2 Toxicity within 2 months after electrochemotherapy (n = 125) Toxicity Any grade, n (%) Grade 1, n (%) Grade 2, n (%) Grade 3, n (%) Skin pain 79 (63.2) 28 (22.4) 38 (30.4) 13 (10.4) Skin ulceration 41 (32.8) 17 (13.6) 14 (11.2) 10 (8.0) Skin hyperpigmentation 34 (27.2) 23 (18.4) 11 (8.8) – Body odor 10 (8) 4 (3.2) 6 (4.8) – Nausea 10 (8) 10 (8) 0 (0) 0 (0) Skin infection 9 (7.2) 6 (4.8) 2 (1.6) 1 (0.8) Flulike symptoms 8 (6.4) 8 (6.4) 0 (0) 0 (0) Fever 7 (5.6) 7 (5.6) 0 (0) 0 (0) Rash 5 (4) 1 (0.8) 4 (3.2) 0 (0) Soft tissue infection 2 (1.6) 0 (0) 2 (1.6) 0 (0) Vomiting 2 (1.6) 2 (1.6) 0 (0) 0 (0) Localized edema 1 (0.8) 0 (0) 1 (0.8) 0 (0) Postoperative hemorrhage 1 (0.8) 0 (0) 1 (0.8) 0 (0)

6 (4.8) 2 (1.6) 1 (0.8) Flulike symptoms 8 (6.4) 8 (6.4) 0 (0) 0 (0) Fever 7 (5.6) 7 (5.6) 0 (0) 0 (0) Rash 5 (4) 1 (0.8) 4 (3.2) 0 (0) Soft tissue infection 2 (1.6) 0 (0) 2 (1.6) 0 (0) Vomiting 2 (1.6) 2 (1.6) 0 (0) 0 (0) Localized edema 1 (0.8) 0 (0) 1 (0.8) 0 (0) Postoperative hemorrhage 1 (0.8) 0 (0) 1 (0.8) 0 (0) Tumor Response Among 125 patients, the follow-up of 12 (9.6 %) was less than 2 months after ECT; these subjects were not evaluated for response. Therefore, 113 (90.4 %) of 125 patients and 214 (89.5 %) of 239 target lesions were evaluated for tumor response. Two months after ECT, per-tumor response was as follows: CR 68.5 %, partial response (PR) 23.5 %, stable disease 6.6 %, progressive disease 0.9 %, and not evaluable 0.5 % as a result of inflammatory reaction and crust formation. Accordingly, the OR rate was 92 % (Fig. 1).Fig. 1 Skin metastases from BC treated with ECT in two patients. Baseline presentation (a, c) and 1-year follow-up (b, d). Arrows contour tumor spread or indicate skin metastases Sixty-six patients (58.4 %) experienced CR, 36 (31.8 %) PR, 8 (7.1 %) stable disease, and 2 (1.8 %) had progressive disease; in 1 patient (0.9 %), tumor response was not evaluable as a result of local skin conditions. Overall, the per-patient OR rate was 90.2 %.

Tumor Response Among 125 patients, the follow-up of 12 (9.6 %) was less than 2 months after ECT; these subjects were not evaluated for response. Therefore, 113 (90.4 %) of 125 patients and 214 (89.5 %) of 239 target lesions were evaluated for tumor response. Two months after ECT, per-tumor response was as follows: CR 68.5 %, partial response (PR) 23.5 %, stable disease 6.6 %, progressive disease 0.9 %, and not evaluable 0.5 % as a result of inflammatory reaction and crust formation. Accordingly, the OR rate was 92 % (Fig. 1).Fig. 1 Skin metastases from BC treated with ECT in two patients. Baseline presentation (a, c) and 1-year follow-up (b, d). Arrows contour tumor spread or indicate skin metastases Sixty-six patients (58.4 %) experienced CR, 36 (31.8 %) PR, 8 (7.1 %) stable disease, and 2 (1.8 %) had progressive disease; in 1 patient (0.9 %), tumor response was not evaluable as a result of local skin conditions. Overall, the per-patient OR rate was 90.2 %. The variables associated with response are shown in Table 3. The CR rate was higher in small (<3 cm) rather than large (≥3 cm) tumors (80.3 vs. 46.1 %, P < 0.0001) and in patients without visceral metastases rather than in those with visceral involvement (80.5 vs. 55.0 %, P < 0.001). The CR rate was also higher among ER-positive (77.2 vs. 59.8 % in ER-negative, P = 0.006) and low proliferating tumors (Ki-67 < 14 %, 79.5 % vs. Ki-67 > 14 %, 58.8 %; P < 0.001). In multivariate analysis, tumor size <3 cm was confirmed to be the most powerful predictor of CR (P < 0.001), followed by the absence of visceral metastases (P = 0.001), ER-positive status (P = 0.016), and low Ki-67 (P = 0.024).Table 3 OR for CR after electrochemotherapy

%, 79.5 % vs. Ki-67 > 14 %, 58.8 %; P < 0.001). In multivariate analysis, tumor size <3 cm was confirmed to be the most powerful predictor of CR (P < 0.001), followed by the absence of visceral metastases (P = 0.001), ER-positive status (P = 0.016), and low Ki-67 (P = 0.024).Table 3 OR for CR after electrochemotherapy Variable CR (%) Univariate analysis Multivariate analysis 1-year LPFS % 95 % CI Multivariate analysis OR 95 % CI P OR 95 % CI P HR 95 % CI P Size <3 cm (n = 55) 80.3 1.966 1.432–2.699 <0.001 7.22 3.35–15.57 <0.001 97.4 92.6–100 5.88 1.59–21.67 0.008 ≥3 cm (n = 58) 46.1 75.6 63.9–89.4 Ulceration No (n = 72) 73.7 1.285 1.035–1.595 0.008 1.59 0.74–3.41 0.234 85.6 76.5–95.8 1.21 0.44–3.30 0.712 Yes (n = 41) 55.7 87.3 77.3–98.6 Receptor status ER positive (n = 65) 77.2 1.590 1.149–2.199 0.006 2.45 1.19–5.07 0.016 84.8 75.4–95.3 0.85 0.37–3.33 0.852 ER negative (n = 48) 59.8 88.8 80.1–98.6 Ki-67 <14 % (n = 58) 79.5 1.641 1.175–2.293 <0.001 2.38 1.12–5.07 0.024 90.2 82.2–98.9 0.85 0.41–3.29 0.786 >14 % (n = 55) 58.0 80.7 69.0–94.5 Visceral metastases No (n = 67) 80.5 1.783 1.364–2.330 <0.001 3.60 1.66–7.79 0.001 86.5 77.4–96.7 1.98 0.72–5.50 0.185 Yes (n = 46) 55.0 86.0 76.0–97.3 CR complete response, OR odds ratio, CI confidence interval, LPFS local progression-free survival, HR hazard ratio, ER estrogen receptor

Size <3 cm (n = 55) 80.3 1.966 1.432–2.699 <0.001 7.22 3.35–15.57 <0.001 97.4 92.6–100 5.88 1.59–21.67 0.008 ≥3 cm (n = 58) 46.1 75.6 63.9–89.4 Ulceration No (n = 72) 73.7 1.285 1.035–1.595 0.008 1.59 0.74–3.41 0.234 85.6 76.5–95.8 1.21 0.44–3.30 0.712 Yes (n = 41) 55.7 87.3 77.3–98.6 Receptor status ER positive (n = 65) 77.2 1.590 1.149–2.199 0.006 2.45 1.19–5.07 0.016 84.8 75.4–95.3 0.85 0.37–3.33 0.852 ER negative (n = 48) 59.8 88.8 80.1–98.6 Ki-67 <14 % (n = 58) 79.5 1.641 1.175–2.293 <0.001 2.38 1.12–5.07 0.024 90.2 82.2–98.9 0.85 0.41–3.29 0.786 >14 % (n = 55) 58.0 80.7 69.0–94.5 Visceral metastases No (n = 67) 80.5 1.783 1.364–2.330 <0.001 3.60 1.66–7.79 0.001 86.5 77.4–96.7 1.98 0.72–5.50 0.185 Yes (n = 46) 55.0 86.0 76.0–97.3 CR complete response, OR odds ratio, CI confidence interval, LPFS local progression-free survival, HR hazard ratio, ER estrogen receptor The distribution of tumor response according to the BC intrinsic subtypes is presented in Table 4. The CR rate in patients with luminal A-like disease was significantly higher compared to all other subgroups (73.9 vs. 54.7 %, P = 0.02). There was no significant difference in tumor size among BC subtypes (P = 0.262).Table 4 Tumor response to electrochemotherapy according to surrogate definition of breast cancer intrinsic subtypes

n patients with luminal A-like disease was significantly higher compared to all other subgroups (73.9 vs. 54.7 %, P = 0.02). There was no significant difference in tumor size among BC subtypes (P = 0.262).Table 4 Tumor response to electrochemotherapy according to surrogate definition of breast cancer intrinsic subtypes Response Luminal A-like (n = 23), n (%) Luminal B-like (HER2 negative) (n = 22), n (%) Luminal B-like (HER2 positive) (n = 18), n (%) Triple negative (n = 35), n (%) HER2 positive (n = 11), n (%) CR 17 (73.9) 11 (50.0) 10 (55.6) 20 (57.1) 6 (54.5) PR 4 (17.4) 9 (40.9) 5 (27.8) 11 (31.4) 5 (45.5) SD 1 (4.3) 2 (9.1) 1 (5.6) 4 (11.4) 0 (0) PD 1 (4.3) 0 (0) 1 (5.6) 0 (0) 0 (0) NA 0 (0) 0 (0) 1 (5.6) 0 (0) 0 (0) According to Goldhirsh et al.11; n = 109 (in four patients, there was no reliable pathologic information). Luminal A-like tumors (ER and PgR positive, HER2 negative, Ki-67 low); luminal B-like, HER2-negative tumors (ER positive, HER2 negative, Ki-67 high and/or PgR low or negative); luminal B-like, HER2 positive tumors (ER positive, HER2 overexpressed or amplified); HER2 positive, nonluminal tumors (HER2 overexpressed or amplified, ER and PgR negative); triple negative tumors (ER, PgR, and HER2 negative) CR complete response, PR partial response, SD stable disease, PD progressive disease, NA not assessable, ER estrogen receptor, PR progesterone receptor

Response Luminal A-like (n = 23), n (%) Luminal B-like (HER2 negative) (n = 22), n (%) Luminal B-like (HER2 positive) (n = 18), n (%) Triple negative (n = 35), n (%) HER2 positive (n = 11), n (%) CR 17 (73.9) 11 (50.0) 10 (55.6) 20 (57.1) 6 (54.5) PR 4 (17.4) 9 (40.9) 5 (27.8) 11 (31.4) 5 (45.5) SD 1 (4.3) 2 (9.1) 1 (5.6) 4 (11.4) 0 (0) PD 1 (4.3) 0 (0) 1 (5.6) 0 (0) 0 (0) NA 0 (0) 0 (0) 1 (5.6) 0 (0) 0 (0) According to Goldhirsh et al.11; n = 109 (in four patients, there was no reliable pathologic information). Luminal A-like tumors (ER and PgR positive, HER2 negative, Ki-67 low); luminal B-like, HER2-negative tumors (ER positive, HER2 negative, Ki-67 high and/or PgR low or negative); luminal B-like, HER2 positive tumors (ER positive, HER2 overexpressed or amplified); HER2 positive, nonluminal tumors (HER2 overexpressed or amplified, ER and PgR negative); triple negative tumors (ER, PgR, and HER2 negative) CR complete response, PR partial response, SD stable disease, PD progressive disease, NA not assessable, ER estrogen receptor, PR progesterone receptor There was no significant association between response and several clinical (patient age, P = 1.00; type of surgery on primary BC, P = 0.070; time from primary BC to recurrence, P = 0.269; presence of lymphedema, P = 0.636; previous radiation, P = 1.00) and procedural (anesthesiology technique, P = 0.377; electrode type, P = 0.799; route of BLM administration, P = 0.606; number of electric pulses, P = 0.842) parameters.

e of surgery on primary BC, P = 0.070; time from primary BC to recurrence, P = 0.269; presence of lymphedema, P = 0.636; previous radiation, P = 1.00) and procedural (anesthesiology technique, P = 0.377; electrode type, P = 0.799; route of BLM administration, P = 0.606; number of electric pulses, P = 0.842) parameters. Local Tumor Control Median follow-up time was 5.9 months (range 3–58 months). Median LPFS was not reached. One-year LPFS was 86.2 % (95 % confidence interval [CI] 79.3–93.8) (Fig. 2).Fig. 2 Tumor control after ECT. Kaplan–Meier curves for LPFS in a whole cohort and b subgroups of patients with lesions <3 cm (blue line) and ≥ 3 cm (yellow line) One-year LPFS in patients who experienced a CR was 96.4 % (95 % CI 91.6–100). In multiple Cox regression analysis, tumor size was the only significant prognostic factor for LPFS (Table 3). One-year LPFS survival in patients with small (<3 cm) tumors was 97.4 % (95 % CI 92.6–100), whereas in those with larger tumors (≥3 cm), it was 75.6 % (95 % CI 63.9–83.4 P = 0.005). Discussion This study showed for the first time that a subgroup of BC patients, identified by routinely used immunohistochemical markers, was particularly sensitive to ECT with BLM. To our knowledge, this cohort analysis was based on the largest series of BC patients treated by ECT to date.

One-year LPFS survival in patients with small (<3 cm) tumors was 97.4 % (95 % CI 92.6–100), whereas in those with larger tumors (≥3 cm), it was 75.6 % (95 % CI 63.9–83.4 P = 0.005). Discussion This study showed for the first time that a subgroup of BC patients, identified by routinely used immunohistochemical markers, was particularly sensitive to ECT with BLM. To our knowledge, this cohort analysis was based on the largest series of BC patients treated by ECT to date. ECT was mainly administered under general sedation or general anesthesia (74 % of patients), while the preferential route for BLM administration was intravenous infusion (80 % of patients). In most cases (89 %), treated tumors were managed using a hexagonal-array, 20 mm long needle electrode. The most frequently reported adverse effects were transient pain and dermatologic toxicity.16–22 Generally, treated skin develops a transient inflammatory reaction. Occasionally, erosions or ulcerations may occur, followed by crust formation. In case of tumor regression, the skin may appear slightly less pigmented, while in some patients it is possible to observe local skin hyperpigmentation, which is a well-known effect of BLM. In patients with locally advanced disease, tumor shrinkage after ECT may cause tissue ulceration requiring specialist wound care.22 Nevertheless, previous experience has demonstrated that effective management of cutaneous metastases provided symptomatic relief and better quality of life to patients.20

wn effect of BLM. In patients with locally advanced disease, tumor shrinkage after ECT may cause tissue ulceration requiring specialist wound care.22 Nevertheless, previous experience has demonstrated that effective management of cutaneous metastases provided symptomatic relief and better quality of life to patients.20 With the present study, we confirm the absence of systemic adverse effects of ECT, as well as a favorable toxicity profile (grade 3 ulceration in 8 % of patients, according to the meta-analysis of Spratt et al., grade 2 hyperpigmentation in 8.8 %), and a high level of acceptance.7 Patients experienced minimal discomfort and needed small amounts of postprocedural analgesics; further, only 10 % of adverse effects were severe, with the exception of transient pain within the first 48 h. As a result, 97 % of the 96 patients who were asked if they would agree to receive further treatment responded favorably. Our results are in line with the ESOPE study, where more than 90 % of patients declared that they were potentially amenable to treatment.6

with the exception of transient pain within the first 48 h. As a result, 97 % of the 96 patients who were asked if they would agree to receive further treatment responded favorably. Our results are in line with the ESOPE study, where more than 90 % of patients declared that they were potentially amenable to treatment.6 Melanoma and BC represented more than 95 % of tumors included in two recently published meta-analyses where the indicated CR rates after ECT were 59 and 57.5 %, respectively.7,23 In the present study, the OR rate was 90.2 %, with a CR rate of 58.4 %, in agreement with the ESOPE study which reported an OR rate of 90.4 %, with 64.3 % of patients experiencing CR.6 A recently published clinical trial on 55 patients, representing the largest published retrospective experience with ECT in BC, showed a CR response rate of 40 % as the most favorable outcome among elderly patients.22 Consistent with a recent meta-analysis, in the present study the response to treatment in small tumors (<3 cm) was higher, similar to that seen in ER-positive, low-proliferating tumors (representing the luminal A-like BC subtype) and in patients without ulcerated lesions or visceral metastases.24

ng elderly patients.22 Consistent with a recent meta-analysis, in the present study the response to treatment in small tumors (<3 cm) was higher, similar to that seen in ER-positive, low-proliferating tumors (representing the luminal A-like BC subtype) and in patients without ulcerated lesions or visceral metastases.24 In particular, the CR rate in the luminal A-like BC subtype was 73.9 %, which was significantly higher than in triple-negative and HER2 positive BC patients (57.1 and 54.5 %, respectively), independent of tumor size. However, although ECT in triple-negative BC in our series was used after failure of several lines of treatment and in conditioning a highly refractory disease, the CR rate in this group nonetheless exceeded 50 %. We are aware that clinical evaluation may be a subjective assessment of tumor response. A pilot study including 11 patients with chest wall recurrence from BC investigated the role of 18F-fludeoxyglucose positron emission tomography (FDG-PET). This study indicates that not only FDG-PET/computed tomography (CT) but also dual time point imaging FDG-PET/CT is promising for evaluation and planning of ECT and could be useful for other localized anticancer treatments as well.25 On the other hand, this imaging technique, which is not widely available and which has nonnegligible costs, has a low sensitivity for small tumor deposits, limiting its application in cutaneous oncology.26

mising for evaluation and planning of ECT and could be useful for other localized anticancer treatments as well.25 On the other hand, this imaging technique, which is not widely available and which has nonnegligible costs, has a low sensitivity for small tumor deposits, limiting its application in cutaneous oncology.26 In our patients, data on local control indicated a 1-year LPFS of 86.2 % within the ECT field (Fig. 2), increasing to 96.4 % in those with CR. In our experience, small (<3 cm) tumor size represented the main predictor of local control compared to large (≥3 cm) tumor size (97.4 vs. 75.6 % at 1 year, respectively).

mising for evaluation and planning of ECT and could be useful for other localized anticancer treatments as well.25 On the other hand, this imaging technique, which is not widely available and which has nonnegligible costs, has a low sensitivity for small tumor deposits, limiting its application in cutaneous oncology.26 In our patients, data on local control indicated a 1-year LPFS of 86.2 % within the ECT field (Fig. 2), increasing to 96.4 % in those with CR. In our experience, small (<3 cm) tumor size represented the main predictor of local control compared to large (≥3 cm) tumor size (97.4 vs. 75.6 % at 1 year, respectively). Skin involvement represents a less frequent but not uncommon event in the metastatic pattern of BC, accounting for 5–30 % of advanced cases in different series.1,2 In addition to their association with unfavorable prognosis, skin metastases cause strong psychologic distress.16 Surgical resection with a radical intent can only be offered to a limited number of patients as a result of multifocality and clinically occult lymphangitic spread.27 In these cases, radiotherapy is generally the best option, but it is often unfeasible on previously irradiated tissues and on lesions that have spread on a wide area. Lack of capillary distribution of radiologic facilities in the territory and the long duration of the entire cycle on multiple sessions may represent further criticisms. Conversely, ECT is applicable on preirradiated areas with the possibility to treat many lesions in a single session, without systemic side effects and a favorable toxicity profile. At any rate, ECT is repeatable and can even be performed in an outpatient setting.

e on multiple sessions may represent further criticisms. Conversely, ECT is applicable on preirradiated areas with the possibility to treat many lesions in a single session, without systemic side effects and a favorable toxicity profile. At any rate, ECT is repeatable and can even be performed in an outpatient setting. Undoubtedly, our findings need broader and prospective confirmation. Furthermore, it will be necessary to clarify whether delaying progression of cutaneous metastases by ECT may provide clinically meaningful benefit to patients, such as delay of disease-related symptoms or preservation of quality of life. In general, the value of progression-free survival, as a surrogate marker for patient benefit, has recently been subjected to critical reappraisal.28 In fact, patient-centered outcomes will a crucial issue in future studies on ECT.29,30 C. Cabula and L. G. Campana have contributed equally to this article, and both should be considered first author. Acknowledgment We are indebted to Francesca de Terlizzi for her fundamental help with the statistical analysis; we also thank Luigi Corti and Gaetano Castiglione for patient care and Christina Drace for editorial help. Disclosure The authors declare no conflict of interest.

Hilar cholangiocarcinoma, also known as Klatskin tumor, is a malignancy originating from the biliary epithelium at the confluence of the left and right hepatic duct.1 Cure can only be achieved by complete surgical resection of the tumor, consisting of extrahepatic bile duct resection with a partial hepatectomy, lymphadenectomy of the hepatoduodenal ligament, and occasionally portal vein or hepatic artery resection.2–4 Despite recent improvements in surgical techniques, the prognosis after curative resection remains poor, with reported 5-year survival rates below 40 %.2,4–6 Regional lymph node status is known to be an important predictor of survival after resection.2,4,5,7,8 However, even patients without lymph node metastases at postoperative pathologic examination frequently develop tumor recurrence. One possible explanation is the presence of micrometastases in the lymph nodes, resulting in understaging of the disease.

n to be an important predictor of survival after resection.2,4,5,7,8 However, even patients without lymph node metastases at postoperative pathologic examination frequently develop tumor recurrence. One possible explanation is the presence of micrometastases in the lymph nodes, resulting in understaging of the disease. Conventional routine histologic examination of the resected specimen includes single sectioning of the lymph nodes with hematoxylin and eosin (H&E) staining. This procedure may underestimate the incidence of nodal metastases. Additional immunohistochemical staining with cytokeratin antibodies may facilitate the detection of small tumor deposits in lymph nodes (micrometastases). In recent years, a number of studies have shown the clinical significance of immunohistochemically detected lymph node micrometastases in a variety of tumors, including those of breast, lung, esophagus, stomach, colon, and gallbladder.9–21 Regarding hilar cholangiocarcinoma, contradicting reports have been published.22–24 This study was undertaken to further investigate the significance of multiple lymph node sectioning and additional immunohistochemical detection of micrometastases on the survival of patients who were initially classified as having lymph node-negative (pN0) hilar cholangiocarcinoma on the basis of conventional histologic examination.

undertaken to further investigate the significance of multiple lymph node sectioning and additional immunohistochemical detection of micrometastases on the survival of patients who were initially classified as having lymph node-negative (pN0) hilar cholangiocarcinoma on the basis of conventional histologic examination. Methods Patients and Operation Between January 1990 and July 2010, a total of 146 patients underwent a curative-intent resection and systematic lymph node dissection of the hepatoduodenal ligament for hilar cholangiocarcinoma at two university medical centers in the Netherlands. Fifty-six patients were treated at the University Medical Center Groningen (UMCG), and 90 were treated at the Academic Medical Center Amsterdam (AMC). In 91 (62 %) of these patients, no lymph node metastases were detected by conventional histologic examination (H&E staining) of the surgical specimen. There were 49 male and 42 female patients. Mean age was 62 ± 9 years (range 36–78 years).

gen (UMCG), and 90 were treated at the Academic Medical Center Amsterdam (AMC). In 91 (62 %) of these patients, no lymph node metastases were detected by conventional histologic examination (H&E staining) of the surgical specimen. There were 49 male and 42 female patients. Mean age was 62 ± 9 years (range 36–78 years). All patients underwent extrahepatic bile duct resection with lymphadenectomy of the liver hilum, in most cases in combination with (extended) hemihepatectomy. Standard regional lymph node dissection consisted of an exploration of the hepatoduodenal ligament and skeletonization of the portal vein and hepatic artery after the distal common bile duct was cut at the level of the pancreatic head and dissected free. In accordance with the American Joint Committee on Cancer (AJCC; 7th edition), lymph nodes at the following locations were defined as regional: along the cystic duct, common bile duct, proper hepatic artery, and portal vein.25 Routine dissection of more distant lymph nodes was not performed. However, suspected distant lymph nodes (N2) were sampled via biopsy, and the operation was aborted if intraoperative frozen section analysis of these samples showed tumor cells.

the cystic duct, common bile duct, proper hepatic artery, and portal vein.25 Routine dissection of more distant lymph nodes was not performed. However, suspected distant lymph nodes (N2) were sampled via biopsy, and the operation was aborted if intraoperative frozen section analysis of these samples showed tumor cells. The majority of patients (91 %) underwent a concomitant partial hepatectomy. Left hemihepatectomy (either extended or not) was performed in 38 patients (42 %) and (extended) right hemihepatectomy was performed in 35 patients (38 %). Ten patients (11 %) underwent only concomitant segment 4 resection. Portal vein resection and reconstruction was undertaken in 18 patients (20 %), and arterial resection and reconstruction was undertaken in 2 patients (2 %). One patient underwent combined arterial and portal venous reconstruction. Tumors were staged according to the tumor, node, metastasis classification system as proposed by the AJCC.25 A total of 324 lymph nodes were retrieved from 91 surgical specimens. On all lymph nodes, additional immunohistochemical staining of cytokeratin 19 (K19) was performed. None of the patients in either institute received postoperative chemotherapy or radiotherapy because this is not recommended in the Dutch guidelines for bile duct carcinoma.

The majority of patients (91 %) underwent a concomitant partial hepatectomy. Left hemihepatectomy (either extended or not) was performed in 38 patients (42 %) and (extended) right hemihepatectomy was performed in 35 patients (38 %). Ten patients (11 %) underwent only concomitant segment 4 resection. Portal vein resection and reconstruction was undertaken in 18 patients (20 %), and arterial resection and reconstruction was undertaken in 2 patients (2 %). One patient underwent combined arterial and portal venous reconstruction. Tumors were staged according to the tumor, node, metastasis classification system as proposed by the AJCC.25 A total of 324 lymph nodes were retrieved from 91 surgical specimens. On all lymph nodes, additional immunohistochemical staining of cytokeratin 19 (K19) was performed. None of the patients in either institute received postoperative chemotherapy or radiotherapy because this is not recommended in the Dutch guidelines for bile duct carcinoma. Immunohistochemistry Formalin-fixed, paraffin-embedded tissue blocks of the lymph nodes were retrieved from the tissue archives of the departments of pathology in the AMC and UMCG. Four new levels of each lymph node with a distance of 250 µm between the levels were investigated. Two 4 µm thick sections of each level were cut serially. One section was stained with H&E and one with an antibody against K19 (dilation 1:100, clone RCK108; Dako, Glostrup, Denmark). All staining procedures were performed in the UMCG. The K19 staining was performed using a Ventana Benchmark Ultra automated stainer (Ventana Medical Systems, Tucson, AZ, USA) after pretreatment with protease.

stained with H&E and one with an antibody against K19 (dilation 1:100, clone RCK108; Dako, Glostrup, Denmark). All staining procedures were performed in the UMCG. The K19 staining was performed using a Ventana Benchmark Ultra automated stainer (Ventana Medical Systems, Tucson, AZ, USA) after pretreatment with protease. Micrometastases were defined as cells detected by immunostaining with morphologic features of adenocarcinoma. H&E and K19 immunolabeled slides were investigated by two experienced pathologists (ASHG and JJD) blinded to the demographic and clinicopathologic features of patients. End Points The effect of micrometastases was assessed in terms of patient survival and time to recurrence. Patients who died during postoperative hospital admission were excluded from the analyses because they were unlikely to have died from recurrent disease. Patient survival was determined from the time of surgery to the time of death or most recent follow-up (March 17, 2014). No patient was lost to follow-up. The median follow-up among survivors was 52 months (range 8 months to 20 years).

ed from the analyses because they were unlikely to have died from recurrent disease. Patient survival was determined from the time of surgery to the time of death or most recent follow-up (March 17, 2014). No patient was lost to follow-up. The median follow-up among survivors was 52 months (range 8 months to 20 years). Recurrences were defined as any new lesion on imaging that was highly suspicious for recurrence of hilar cholangiocarcinoma. Pathologic confirmation was often obtained but was not required. Recurrences at the liver resection margin, distal bile duct remnant, hepaticojejunostomy, or elsewhere in the liver hilum were classified as local recurrences. All other recurrences were classified as distant. Time to recurrence was measured from the time of surgery to the time of the first recurrence. Patients who had no observed recurrence were censored at the time of last follow-up, and patients who died from other causes before developing a recurrence were censored at the time of death.

rences were classified as distant. Time to recurrence was measured from the time of surgery to the time of the first recurrence. Patients who had no observed recurrence were censored at the time of last follow-up, and patients who died from other causes before developing a recurrence were censored at the time of death. Statistical Analysis Patient data and baseline characteristics were retrospectively collected in a database, and statistical analyses were carried out by SPSS Statistics software (IBM, Armonk, New York, USA). Continuous variables were expressed as mean ± SD. Categorical variables were expressed as numbers and percentages. Comparison of means was performed with Student’s t test for independent samples. Comparison of categorical variables was performed with the χ2 test or Fisher’s exact probability test. Univariable analyses were conducted for patient survival and time to recurrence by Kaplan–Meier estimates of survival probabilities and the log-rank test for comparisons. A Cox proportional hazard regression model was used to analyze associations with patient survival in multivariable analysis, including all factors with a P value of less than 0.10 in univariable analysis. Because AJCC pT1 stage was considered a potential confounder, we a priori included this variable in the Cox model. P values were two sided, and values of less than 0.05 were considered statistically significant.

l in multivariable analysis, including all factors with a P value of less than 0.10 in univariable analysis. Because AJCC pT1 stage was considered a potential confounder, we a priori included this variable in the Cox model. P values were two sided, and values of less than 0.05 were considered statistically significant. Results Micrometastases were detected in 16 (5 %) of 324 lymph nodes and in 11 (12 %) of 91 patients who were initially considered lymph node negative by conventional histologic examination (Fig. 1). The K19 labeling clearly highlighted the adenocarcinoma foci and confirmed their biliary origin.Fig. 1 Cytokeratin 19-positive metastatic tumor cells in regional lymph nodes from resected hilar cholangiocarcinoma specimens. a Micrometastases consisting of two small tumor glands (original magnification, ×4). b Micrometastases consisting of single tumor cells (arrow; original magnification, ×4)

their biliary origin.Fig. 1 Cytokeratin 19-positive metastatic tumor cells in regional lymph nodes from resected hilar cholangiocarcinoma specimens. a Micrometastases consisting of two small tumor glands (original magnification, ×4). b Micrometastases consisting of single tumor cells (arrow; original magnification, ×4) Postoperative death during hospital admission (in-hospital mortality) occurred in 11 (12 %) of 91 patients, who were subsequently excluded from the survival analyses. In-hospital mortality occurred only in the group without micrometastases (range 8–80 days). Clinicopathologic details of the remaining 80 patients with and without lymph node micrometastases are shown in Table 1. There were no statistically significant differences between the two groups. During postoperative follow-up, 49 (61 %) of 80 patients died. At the end of follow-up, 31 patients were alive, two of whom were diagnosed with disease recurrence.Table 1 Clinicopathologic characteristics of patients after curative-intent resection for hilar cholangiocarcinoma with and without lymph node micrometastases Characteristic Lymph node micrometastases P Absent (n = 69) Present (n = 11) Mean age in years (±SD) 61 ± 10 59 ± 6 0.40 Gender male/female 37/32 5/6 0.75 Type of hepatectomy Left 25 4 0.90 Extended left 6 0 Right 5 1 Extended right 17 4 Only segment 4 resection 9 1 No liver resection 7 1 pT stagea

Postoperative death during hospital admission (in-hospital mortality) occurred in 11 (12 %) of 91 patients, who were subsequently excluded from the survival analyses. In-hospital mortality occurred only in the group without micrometastases (range 8–80 days). Clinicopathologic details of the remaining 80 patients with and without lymph node micrometastases are shown in Table 1. There were no statistically significant differences between the two groups. During postoperative follow-up, 49 (61 %) of 80 patients died. At the end of follow-up, 31 patients were alive, two of whom were diagnosed with disease recurrence.Table 1 Clinicopathologic characteristics of patients after curative-intent resection for hilar cholangiocarcinoma with and without lymph node micrometastases Characteristic Lymph node micrometastases P Absent (n = 69) Present (n = 11) Mean age in years (±SD) 61 ± 10 59 ± 6 0.40 Gender male/female 37/32 5/6 0.75 Type of hepatectomy Left 25 4 0.90 Extended left 6 0 Right 5 1 Extended right 17 4 Only segment 4 resection 9 1 No liver resection 7 1 pT stagea pT1 14 0 0.48 pT2a 32 5 pT2b 12 3 pT3 3 1 pT4 8 2 Microscopic resection margin positive 20 (29 %) 4 (36 %) 0.73 Perineural invasion positive 39 (57 %) 8 (73 %) 0.31 Complication rate grade III or IVb 24 (35 %) 4 (36 %) 0.92 Mean no. of dissected lymph nodes per patient 4 5 0.18 Patients with in-hospital mortality (n = 11) were excluded from analysis aAccording to American Joint Committee on Cancer staging manual, 7th edition

pT1 14 0 0.48 pT2a 32 5 pT2b 12 3 pT3 3 1 pT4 8 2 Microscopic resection margin positive 20 (29 %) 4 (36 %) 0.73 Perineural invasion positive 39 (57 %) 8 (73 %) 0.31 Complication rate grade III or IVb 24 (35 %) 4 (36 %) 0.92 Mean no. of dissected lymph nodes per patient 4 5 0.18 Patients with in-hospital mortality (n = 11) were excluded from analysis aAccording to American Joint Committee on Cancer staging manual, 7th edition bAccording to Clavien-Dindo classification of surgical complications. Grade V (in-hospital mortality) was excluded

pT1 14 0 0.48 pT2a 32 5 pT2b 12 3 pT3 3 1 pT4 8 2 Microscopic resection margin positive 20 (29 %) 4 (36 %) 0.73 Perineural invasion positive 39 (57 %) 8 (73 %) 0.31 Complication rate grade III or IVb 24 (35 %) 4 (36 %) 0.92 Mean no. of dissected lymph nodes per patient 4 5 0.18 Patients with in-hospital mortality (n = 11) were excluded from analysis aAccording to American Joint Committee on Cancer staging manual, 7th edition bAccording to Clavien-Dindo classification of surgical complications. Grade V (in-hospital mortality) was excluded Survival Analysis The survival rates for patients were calculated according to lymph node status. We defined three groups: patients with lymph node (macro)metastases detected at routine H&E examination (pN1); patients with lymph node micrometastases detected with multiple sectioning and K19 staining (pN0 with micrometastases); and patients without lymph node micrometastases (pN0 without micrometastases). Five-year survival rates in patients without lymph node micrometastases were significantly higher compared to the other groups (P < 0.001) (Fig. 2). There was a significant difference in 5-year survival between patients with and without micrometastases (27 vs. 54 %, P = 0.01), but not between patients with micro- and macrometastases (27 vs. 15 %, P = 0.54).Fig. 2 Survival after resection for hilar cholangiocarcinoma according to lymph node status. Patients with in-hospital mortality were excluded from analysis. pN0 without micrometastases versus pN0 with micrometastases: P = 0.01 (log-rank test). pN0 with micrometastases versus pN1: P = 0.54 (log-rank test)

%, P = 0.54).Fig. 2 Survival after resection for hilar cholangiocarcinoma according to lymph node status. Patients with in-hospital mortality were excluded from analysis. pN0 without micrometastases versus pN0 with micrometastases: P = 0.01 (log-rank test). pN0 with micrometastases versus pN1: P = 0.54 (log-rank test) Recurrence During the study period, 30 (38 %) of 80 patients developed disease recurrence (local recurrence in 21 patients and distant metastases in 9 patients). Figure 3 presents the estimated cumulative probability of recurrence over time according to the presence or absence of micrometastases. At 5-year follow-up, the estimated probability of recurrence was 65 % in the group with micrometastases, versus 33 % in patients without micrometastases (P = 0.06).Fig. 3 Cumulative probability of recurrence after resection for hilar cholangiocarcinoma in patients classified as pN0 based on routine histologic examination, with or without lymph node micrometastases on subsequent immunohistochemistry. Patients with in-hospital mortality (n = 11) were excluded from analysis; P = 0.06 (log-rank test) Next, the relationship between lymph node micrometastases and pattern of recurrence was analyzed. There was a twofold higher percentage of distant site recurrences in the group with micrometastases compared to the group without micrometastases (18 vs. 9 %), but this did not reach statistical significance (P = 0.40).

Recurrence During the study period, 30 (38 %) of 80 patients developed disease recurrence (local recurrence in 21 patients and distant metastases in 9 patients). Figure 3 presents the estimated cumulative probability of recurrence over time according to the presence or absence of micrometastases. At 5-year follow-up, the estimated probability of recurrence was 65 % in the group with micrometastases, versus 33 % in patients without micrometastases (P = 0.06).Fig. 3 Cumulative probability of recurrence after resection for hilar cholangiocarcinoma in patients classified as pN0 based on routine histologic examination, with or without lymph node micrometastases on subsequent immunohistochemistry. Patients with in-hospital mortality (n = 11) were excluded from analysis; P = 0.06 (log-rank test) Next, the relationship between lymph node micrometastases and pattern of recurrence was analyzed. There was a twofold higher percentage of distant site recurrences in the group with micrometastases compared to the group without micrometastases (18 vs. 9 %), but this did not reach statistical significance (P = 0.40). Analysis of Prognostic Factors in Patients with pN0 Disease Prognostic factors for survival were identified using univariable analyses (Table 2). Two variables—microscopic resection margin status (P = 0.09) and lymph node micrometastases (P = 0.01)—were identified as predictors for survival. These two variables were included in a multivariable Cox regression analysis, together with AJCC pT stage, because this was a priori considered a possible confounding variable. Only the presence of micrometastases was identified as an independent predictive factor for survival in patients with pN0 hilar cholangiocarcinoma (Hazard ratio 2.43, 95 % confidence interval 1.16–5.10) (Tables 2).Table 2 Univariable analysis of survival in patients with hilar cholangiocarcinoma classified as pN0 based on routine histologic examination

es was identified as an independent predictive factor for survival in patients with pN0 hilar cholangiocarcinoma (Hazard ratio 2.43, 95 % confidence interval 1.16–5.10) (Tables 2).Table 2 Univariable analysis of survival in patients with hilar cholangiocarcinoma classified as pN0 based on routine histologic examination Variable No. patients 5-year survival (%) P Univariable analysis Age <60 year 30 (38 %) 43 0.38 ≥60 year 50 (62 %) 55 Sex Male 42 (53 %) 49 0.71 Female 38 (47 %) 52 Type of hepatectomya Left 45 (63 %) 50 0.40 Right 27 (37 %) 40 pT stageb T1 14 (17 %) 71 0.37 T2a and 2b 52 (66 %) 47 T3 and 4 14 (17 %) 41 Perineural invasion Negative 33 (41 %) 47 0.60 Positive 47 (59 %) 53 Lymph node micrometastases Negative 69 (86 %) 54 0.01 Positive 11 (14 %) 27 Microscopic resection margin status Negative 56 (70 %) 56 0.09 Positive 24 (30 %) 38 Variable Hazard ratio 95 % confidence interval P Multivariable analysis pT stageb T1 Reference category T2a and T2b 1.55 0.64–3.75 0.33 T3 and T4 1.25 0.42–3.76 0.69 Lymph node micrometastases 2.43 1.16–5.10 0.02 Microscopic resection margin status 1.54 0.85–2.80 0.16 Patients with in-hospital mortality were excluded from analysis aPatients without concomitant liver resection (n = 8) were excluded from analysis bAccording to American Joint Committee on Cancer staging manual, 7th edition

T1 Reference category T2a and T2b 1.55 0.64–3.75 0.33 T3 and T4 1.25 0.42–3.76 0.69 Lymph node micrometastases 2.43 1.16–5.10 0.02 Microscopic resection margin status 1.54 0.85–2.80 0.16 Patients with in-hospital mortality were excluded from analysis aPatients without concomitant liver resection (n = 8) were excluded from analysis bAccording to American Joint Committee on Cancer staging manual, 7th edition Discussion Lymph node metastases have been repeatedly identified as an important prognostic factor for survival after resection of hilar cholangiocarcinoma.2,5,26 This has also been the experience of the two centers participating in this study.4,7 The current study was undertaken to improve tumor staging by identifying lymph node micrometastases in the pN0 group. To this end, we applied rigorous multiple sectioning of the lymph node tissue blocks to achieve deeper levels, investigated several levels of each lymph node, and increased the sensitivity of tumor cell identification by K19 immunolabeling. This method is a slight modification of the histopathologic investigation of sentinel nodes in other types of cancer (e.g., breast cancer and melanoma). Using this technique, micrometastases were detected in 5 % of the collected lymph nodes, which related to 12 % of the 91 patients who were initially characterized as pN0 on the basis of conventional histologic examination of the surgical resection specimen with H&E staining. Survival in the group with lymph node micrometastases was significantly worse compared to patients without micrometastases and comparable to patients with pN1 disease. There was a trend toward earlier tumor recurrence in patients with micrometastases, although this did not reach statistical significance (Fig. 3). Remarkably, the presence of lymph node micrometastases was found to be a stronger predictor than the level of tumor invasion (T stage) and microscopic resection margin status (Table 2). The same phenomenon was previously observed in a study by Yonemori et al.24 A possible explanation is that lymph node micrometastases represent a more aggressive biologic behavior of the tumor, similar to lymph node metastases found on routine histology.

sion (T stage) and microscopic resection margin status (Table 2). The same phenomenon was previously observed in a study by Yonemori et al.24 A possible explanation is that lymph node micrometastases represent a more aggressive biologic behavior of the tumor, similar to lymph node metastases found on routine histology. Only three previous studies, all from Japan, have addressed the effect of lymph node micrometastases on survival in hilar cholangiocarcinoma, and these studies have shown contradictory results.22–24 Yonemori et al. found lymph node micrometastases to be of influence on the survival in a heterogeneous group of 151 patients with biliary cancer (including gallbladder cancer, intrahepatic cholangiocarcinoma, and ampullary cancer).24 A subgroup analysis of patients with hilar bile duct cancer only revealed lymph node micrometastases in 4 % of the investigated nodes and in 22 (27 %) of 83 patients, but no significant impact on survival was found. Similarly, Tojima et al. detected lymph node micrometastases in 13 (1.4 %) of 954 examined lymph nodes, corresponding to 11 (24 %) of 45 patients; they also concluded that lymph node micrometastases had no effect on postoperative survival.23 In contrast, Taniguchi et al. found lymph node micrometastases in 14 (3.3 %) of 423 lymph nodes and 11 (39 %) of 28 patients; these authors reported a significantly lower survival rate in patients with lymph node micrometastases.22

lso concluded that lymph node micrometastases had no effect on postoperative survival.23 In contrast, Taniguchi et al. found lymph node micrometastases in 14 (3.3 %) of 423 lymph nodes and 11 (39 %) of 28 patients; these authors reported a significantly lower survival rate in patients with lymph node micrometastases.22 The current study is the first Western series in which the incidence of lymph node micrometastases and its influence on survival was investigated and presents the largest cohort published to date, consisting of 91 pN0 patients. There are some differences between our series and those reported in literature.

lso concluded that lymph node micrometastases had no effect on postoperative survival.23 In contrast, Taniguchi et al. found lymph node micrometastases in 14 (3.3 %) of 423 lymph nodes and 11 (39 %) of 28 patients; these authors reported a significantly lower survival rate in patients with lymph node micrometastases.22 The current study is the first Western series in which the incidence of lymph node micrometastases and its influence on survival was investigated and presents the largest cohort published to date, consisting of 91 pN0 patients. There are some differences between our series and those reported in literature. First, we found micrometastases in 12 % of the patients who were initially considered node negative on routine histologic examination. This is slightly less than the 24–39 % reported in the three Japanese series.22–24 The number of serial sections that were cut from the archival tissue blocks can be an explanation for this difference. In our study, four additional sections were performed, with a distance of 250 µm between consecutive levels, compared to five to eight serial sections reported in the three Japanese studies.22–24 We believe that with our technique all tumor cell clusters in the lymph nodes were detected. Some authors state that 200 µm should be the maximum distance between consecutive levels.27 When adopting this strategy, we should have investigated five instead of four levels, but it is questionable whether this extra level would have significantly increased the sensitivity of the procedure. To illustrate this, in the Japanese series, the percentage of patients in whom micrometastases were detected did not linearly rise with increased sectioning (24 % in the study with five serial sections, vs. 27 % in the study with eight serial sections).23,24 Recent studies of cancer of the breast, lung, stomach, and colorectum have classified metastatic tumor cells by size into micrometastases (>0.2 mm) and isolated tumor cells (≤0.2 mm).23,24,28–31 In our series, this type of further differentiation was not possible because in all cases metastatic tumor cells were larger than 0.2 mm in size (between 0.2 and 0.4 mm). Increased sectioning would probably have resulted in increased detection of isolated tumor cells. However, the significance of isolated tumor cells remains a subject of debate, even in breast cancer, where this issue has been studied more than in any other type of cancer.32

than 0.2 mm in size (between 0.2 and 0.4 mm). Increased sectioning would probably have resulted in increased detection of isolated tumor cells. However, the significance of isolated tumor cells remains a subject of debate, even in breast cancer, where this issue has been studied more than in any other type of cancer.32 Second, lymphadenectomy in our population resulted in a mean of 4 lymph nodes per patient (range 1–16), which is less than the 10–20 nodes per patient reported in the Japanese series.22–24 However, this is in line with other Western series and in accordance with the AJCC/International Union Against Cancer guidelines, which recommend that at least 3 lymph nodes should be collected for adequate staging of hilar cholangiocarcinoma.33,34 The difference in number of collected lymph nodes between our study and the Japanese series can be explained by the variation in extent of the lymphadenectomy. In contrast to the Japanese experience, it is not routine surgical practice in the Netherlands to perform standard removal of N2 nodes (para-aortic, caval vein, and superior mesenteric). An N2 node is only removed when macroscopically suspicious and is then sent for intraoperative frozen sections analysis. The surgical procedure is aborted when frozen section analysis shows metastatic tumor cells.

he Netherlands to perform standard removal of N2 nodes (para-aortic, caval vein, and superior mesenteric). An N2 node is only removed when macroscopically suspicious and is then sent for intraoperative frozen sections analysis. The surgical procedure is aborted when frozen section analysis shows metastatic tumor cells. The in-hospital mortality rate in the present study was 15 % for the entire group of patients and 12 % in the group with pN0 disease. Some series report a lower mortality rate, but in these series, patients who only underwent extrahepatic bile duct resection, without concomitant partial liver resection, were also included.5 Most of our patients (>90 %) underwent partial hepatectomy in addition to extrahepatic bile duct resection. Indeed, the postoperative mortality rate in our population is comparable to the mortality rate of 10–11 % described in other series in which partial hepatectomy was part of the surgical procedure.34–36 In the current study, micrometastases were detected by multiple sectioning of the lymph nodes together with K19 immunohistochemistry. In our opinion, application of both methods is important to detect micrometastases. In some cases, micrometastases were evident even without K19 staining (Fig. 1a) but would not have been found if additional sectioning had been omitted. In other cases, K19 labeling was necessary to highlight the small micrometastases.

In our opinion, application of both methods is important to detect micrometastases. In some cases, micrometastases were evident even without K19 staining (Fig. 1a) but would not have been found if additional sectioning had been omitted. In other cases, K19 labeling was necessary to highlight the small micrometastases. In conclusion, this study shows that the presence of lymph node micrometastases in patients with otherwise node-negative hilar cholangiocarcinoma has a negative effect on survival. The technique of multiple lymph node sectioning together with K19 immunostaining results in improved staging of patients with hilar cholangiocarcinoma. Hendrik T. J. Mantel, Jim K. Wiggers, Annette S. H. Gouw and Robert J. Porte have contributed equally to this work. Conflict of interest The authors declare no conflict of interest.

Pseudomyxoma peritonei (PMP) is a rare tumor disease characterized by disseminated mucus and mucinous tumor tissue implants on the peritoneal surfaces, now considered to originate from the appendix.1 By use of cytoreductive surgery (CRS) combined with intraperitoneal chemotherapy (IPC), as introduced by Sugarbaker, with the IPC now mostly being hyperthermic (HIPEC), the prognosis has improved.2 Thus, experienced centres report a 5-year overall survival (OS) in the range of 70–95 % compared with 30–40 % often reported for the strategy of debulking surgery.3–5 The main factors associated with favorable prognosis are complete CRS, low tumor load, and low histological grade.5 Systemic chemotherapy alone as treatment of PMP has not been extensively investigated but seems poorly active in this disease.6 However, the role of the chemotherapy part of the CRS and IPC treatment package in PMP is unclear. Thus, the effect of CRS with or without IPC has not been directly compared and long-term survival has been reported with CRS alone.5,7,8 In contrast, IPC provides clinical benefit as adjunct to CRS in peritoneal carcinomatosis from gastric and ovarian cancer.8,9 Furthermore, IPC in PMP differs between treatment centres in terms of drug selection, dosing, and timing of the IPC.8,10 Thus, IPC as part of treatment of PMP is in need of further investigation to define its role and provide a basis for how to optimize this resource demanding treatment step.

ric and ovarian cancer.8,9 Furthermore, IPC in PMP differs between treatment centres in terms of drug selection, dosing, and timing of the IPC.8,10 Thus, IPC as part of treatment of PMP is in need of further investigation to define its role and provide a basis for how to optimize this resource demanding treatment step. We used a short-term ex vivo assay to evaluate the tumor cell sensitivity to cytotoxic drugs in samples from PMP patients undergoing CRS and IPC. The goals were to provide information on the pattern of drug activity in PMP and to correlate the ex vivo drug sensitivity pattern to the clinical outcome. Methods Patients and Tumor Samples A total of 133 patients scheduled for CRS and HIPEC for PMP at the Department of Surgery, Uppsala University Hospital between May 2006 and December 2011, and from which a tumor sample for ex vivo assessment of drug activity was obtained, formed the basis for the study. Tumor sampling was performed intraoperatively prior to HIPEC, which consisted of 30–35 mg/m2 of mitomycin C, 100 mg/m2 of cisplatin combined with 15 mg/m2 of doxorubicin or 360 mg/m2 of both irinotecan and oxaliplatin.4 Tumor sampling and data collection was based on patient informed consent and approved by the Regional Ethical Review Board in Uppsala (Dnr 2007/237). None of the patients had adjuvant systemic chemotherapy following CRS and HIPEC.

n combined with 15 mg/m2 of doxorubicin or 360 mg/m2 of both irinotecan and oxaliplatin.4 Tumor sampling and data collection was based on patient informed consent and approved by the Regional Ethical Review Board in Uppsala (Dnr 2007/237). None of the patients had adjuvant systemic chemotherapy following CRS and HIPEC. Tumor histopathology was classified as disseminated peritoneal adenomucinosis (DPAM), peritoneal mucinous carcinomatosis (PMCA), or PMCA with intermediate features.11 Tumor load was assessed as the Peritoneal Cancer Index (PCI) at time of surgery.12 Residual disease after a maximal surgical effort was quantified according to the completeness of cytoreduction score (CC). CC scores 0 (no macroscopic tumor left) and 1 (residual tumor <0.25 cm) were considered as complete cytoreduction.13 Ex Vivo Assessment of Drug Sensitivity The tumor specimen was kept in buffer at 6 °C until preparation. Tumor cells were prepared by collagenase digestion as described.14 The cells obtained were mostly single cells or small cell clusters with ≥90 % viability and with <30 % contaminating nonmalignant cells, as judged by morphological examinations of May-Grünwald-Giemsa-stained cytocentrifugate preparations.

l preparation. Tumor cells were prepared by collagenase digestion as described.14 The cells obtained were mostly single cells or small cell clusters with ≥90 % viability and with <30 % contaminating nonmalignant cells, as judged by morphological examinations of May-Grünwald-Giemsa-stained cytocentrifugate preparations. The drugs used for HIPEC (see above) were tested ex vivo. In addition, 5FU, an established drug in gastrointestinal cancer treatment, was included. All drugs were from commercially available clinical preparations. The drugs were tested at three tenfold dilutions from the maximal concentration (μM) of 100 for cisplatin, 100 for oxaliplatin, 10 for doxorubicin, 1000 for 5FU, 100 for mitomycin C, and 1000 for irinotecan. The drug concentrations used ex vivo are chosen empirically to produce concentration—response curves allowing for extraction of 50 % inhibitory concentrations (IC50), i.e., the drug concentration producing a cell survival of 50 % compared with an unexposed control. The maximal concentrations used ex vivo are close to Cmax achievable during IPC for most drugs.15 384-well microplates (Nunc) were prepared with 5-μl drug solution at 10× the final drug concentration using the pipetting robot BioMek 2000 (Beckman Coulter). The plates were then stored at −70 °C until further use.

ntrol. The maximal concentrations used ex vivo are close to Cmax achievable during IPC for most drugs.15 384-well microplates (Nunc) were prepared with 5-μl drug solution at 10× the final drug concentration using the pipetting robot BioMek 2000 (Beckman Coulter). The plates were then stored at −70 °C until further use. The semiautomated fluorometric microculture cytotoxicity assay (FMCA) was used to assess drug sensitivity.16 Briefly, tumor cells from patient samples (5000 cells/well in 45 μl culture medium RPMI 1640 (supplemented with 10 % foetal calf serum, glutamine and antibiotics) were seeded in the drug-prepared 384-well plates using the pipetting robot Precision 2000 (Bio-Tek Instruments Inc., Winooski, VT). Three columns without drugs served as controls and one column with medium only served as blank. The culture plates were incubated at 37 °C in humidified atmosphere containing 95 % air and 5 % CO2. After 72 h incubation, the culture medium was washed away and 50 μl/well of a physiological buffer containing 10 μg/ml of the vital dye fluorescein diacetate (FDA) were added to control, experimental, and blank wells. After incubation for 30–45 min at 37 °C, the fluorescence from each well was read in a FluoroScan 2 (Labsystems OY, Helsinki, Finland).

e medium was washed away and 50 μl/well of a physiological buffer containing 10 μg/ml of the vital dye fluorescein diacetate (FDA) were added to control, experimental, and blank wells. After incubation for 30–45 min at 37 °C, the fluorescence from each well was read in a FluoroScan 2 (Labsystems OY, Helsinki, Finland). Quality criteria for a successful assay were: ≥70 % tumor cells in the cell preparation before incubation and/or on the assay day, a fluorescence signal in control cultures of ≥5 x mean blank values, and a coefficient of variation of cell survival in control cultures of ≤30 %. The results obtained by the viability indicator FDA are calculated as survival index (SI), defined as the fluorescence of the test expressed as a percentage of control cultures, with blank values subtracted.

ultures of ≥5 x mean blank values, and a coefficient of variation of cell survival in control cultures of ≤30 %. The results obtained by the viability indicator FDA are calculated as survival index (SI), defined as the fluorescence of the test expressed as a percentage of control cultures, with blank values subtracted. Patient Data and Follow-Up Clinical data relevant for the study were retrieved from the patient files. Patients with complete cytoreduction were followed for progression-free survival (PFS) by assessment of serum tumor markers (CEA, CA19-9, CA 125, and CA 72.3) every 3 months and with CT scan of abdomen and thorax every 6 months for 3 years and then every 12 months, for another 2 years. An increase in a tumor marker ≥25 % triggered a CT scan for verification of new lesions consistent with PMP relapse. Overall survival (OS) was assessed from registry data up to February 2014. Data on treatment following relapse was incomplete and indicated individualized approaches used. This is expected to affect the OS observed probably making this endpoint poorly associated to the IPC (see “Results” section).

ent with PMP relapse. Overall survival (OS) was assessed from registry data up to February 2014. Data on treatment following relapse was incomplete and indicated individualized approaches used. This is expected to affect the OS observed probably making this endpoint poorly associated to the IPC (see “Results” section). Data Evaluation and Statistics IC50 was calculated using non-linear regression to a standard sigmoidal dose–response model in GraphPad Prism version 5 for Mac (GraphPad Software, San Diego, CA). Alternatively, sample sensitivity was scored according to the SI at the highest cytotoxic drug concentration used ex vivo. In this case, low drug resistance (LDR) was defined as a SI below the median, intermediate drug resistance (IDR) as a SI between the median and median plus two standard deviations (SDs), and extreme drug resistance (EDR) as a SI above median plus two SDs based on all samples investigated ex vivo.16,17 Statistical inferences between several means were performed by one-way ANOVA with Tukey HSD post-hoc tests. The prognostic importance of clinicopathological variables and ex vivo drug sensitivity for OS and PFS was assessed in a Cox regression model. In the model on OS only univariate results with p < 0.2 were included in the final multivariable analysis. Analyses on PFS were adjusted for WHO performance status, histopathological subtype, and tumor load. The level of significance for all statistical tests was set to p < 0.05. Data are presented as mean ± SD unless otherwise stated.

S only univariate results with p < 0.2 were included in the final multivariable analysis. Analyses on PFS were adjusted for WHO performance status, histopathological subtype, and tumor load. The level of significance for all statistical tests was set to p < 0.05. Data are presented as mean ± SD unless otherwise stated. Results A successful ex vivo assay fulfilling the quality criteria was obtained from 92 tumor samples (69 %) and data from these patients were included for analysis in the study. Mucin-rich tumor samples, often of the DPAM subtype, dominated among samples not possible to run in the assay due to difficulties to recover a sufficient number of epithelial cells when a lot of mucin was present during cell preparation. The majority of patients had a histopathology of DPAM (n = 57), whereas 24 had PMCA and 11 patients had a PMCA intermediate histology (Table 1). A majority of patients, 64 %, had previously been treated systemically and/or locally with chemotherapy for PMP. In 61 patients (66 %) CC 0–1, i.e., complete cytoreduction was achieved at CRS. Eighty patients, including 59 of the 61 patients with complete cytoreduction, received HIPEC, most commonly single drug mitomycin C (n = 56), combinations of cisplatin and doxorubicin (n = 17), or irinotecan and oxaliplatin (n = 7).Table 1 Clinical characteristics of the pseudomyxoma peritonei samples successfully analyzed ex vivo (n = 92)

uding 59 of the 61 patients with complete cytoreduction, received HIPEC, most commonly single drug mitomycin C (n = 56), combinations of cisplatin and doxorubicin (n = 17), or irinotecan and oxaliplatin (n = 7).Table 1 Clinical characteristics of the pseudomyxoma peritonei samples successfully analyzed ex vivo (n = 92) Age, year, mean (range) 56 (24–78) BMI, kg/m2, mean (range) 25 (19–38) Male/female 47/45 Histopathology DPAM 57 (62 %) PMCA intermediate 11 (12 %) PMCA 24 (26 %) Prior chemotherapy No 59 (64 %) Yes 33 (36 %) PCI scorea 1–10 9 (10 %) 11–20 13 (14 %) 21–39 69 (76 %) WHO performance status 0 79 (86 %) 1–2 13 (14 %) Complete cytoreductive surgeryb 61 (66 %) Hyperthermic intraperitoneal chemotherapy 80 (87 %) DPAM disseminated peritoneal adenomucinosis, PMCA peritoneal mucinous carcinomatosis, PCI peritoneal carcinoma index, WHO World Health Organization aInformation on PCI score unavailable in one patient bCC score 0–1

1–10 9 (10 %) 11–20 13 (14 %) 21–39 69 (76 %) WHO performance status 0 79 (86 %) 1–2 13 (14 %) Complete cytoreductive surgeryb 61 (66 %) Hyperthermic intraperitoneal chemotherapy 80 (87 %) DPAM disseminated peritoneal adenomucinosis, PMCA peritoneal mucinous carcinomatosis, PCI peritoneal carcinoma index, WHO World Health Organization aInformation on PCI score unavailable in one patient bCC score 0–1 Drug sensitivity varied considerably between patient samples as indicated by the high SDs observed for the IC50 values for all drugs (Table 2). Samples obtained from patients previously exposed to cytotoxic drugs were statistically significantly more resistant to all drugs tested except irinotecan. There were statistically significant differences in drug sensitivity to oxaliplatin, 5-FU, and cisplatin between the histopathological subtypes of PMP (Table 2); PMCA showed higher IC50 values for oxaliplatin, 5-FU, and cisplatin compared with samples of hybrid histology. Similarly, PMCA samples had higher IC50 for oxaliplatin than DPAM. There were no statistically significant differences in drug sensitivity between samples divided into low and high grade according to Bradley (Supplementary Table 1).18Table 2 IC50 values (μM, mean ± standard deviation) for the indicated drugs in the pseudomyxoma peritonei samples (n = 92; IC50 values available in 88–92 cases depending on cytotoxic drug) according to previous chemotherapy and histopathological subtype Previous chemotherapy Histopathological subtype All PMP Yes n = 59 No n = 33 DPAM n = 57 PMCA intermediate n = 11 PMCA

Drug sensitivity varied considerably between patient samples as indicated by the high SDs observed for the IC50 values for all drugs (Table 2). Samples obtained from patients previously exposed to cytotoxic drugs were statistically significantly more resistant to all drugs tested except irinotecan. There were statistically significant differences in drug sensitivity to oxaliplatin, 5-FU, and cisplatin between the histopathological subtypes of PMP (Table 2); PMCA showed higher IC50 values for oxaliplatin, 5-FU, and cisplatin compared with samples of hybrid histology. Similarly, PMCA samples had higher IC50 for oxaliplatin than DPAM. There were no statistically significant differences in drug sensitivity between samples divided into low and high grade according to Bradley (Supplementary Table 1).18Table 2 IC50 values (μM, mean ± standard deviation) for the indicated drugs in the pseudomyxoma peritonei samples (n = 92; IC50 values available in 88–92 cases depending on cytotoxic drug) according to previous chemotherapy and histopathological subtype Previous chemotherapy Histopathological subtype All PMP Yes n = 59 No n = 33 DPAM n = 57 PMCA intermediate n = 11 PMCA n = 24 n = 92 Oxaliplatin 47.2 ± 36.1 25.9 ± 29.4a 30.9 ± 32.0 16.3 ± 12.0 47.9 ± 38.8b,c 33.6 ± 33.5 5FU 708 ± 354 517 ± 431a 591 ± 436 327 ± 256 692 ± 376c 586 ± 414 Mitomycin C 35.0 ± 75.6 12.1 ± 16.8a 22.4 ± 59.6 11.4 ± 19.6 20.0 ± 20.5 20.4 ± 48.3 Doxorubicin 3.3 ± 5.6 1.5 ± 3.0a 2.6 ± 4.9 1.1 ± 2.0 1.5 ± 2.7 2.1 ± 4.2 Irinotecan 410 ± 755 223 ± 311 347 ± 631 1126 ± 78 244 ± 290 291 ± 522 Cisplatin 41.0 ± 35.8 25.6 ± 33.4a 29.8 ± 31.8 13.9 ± 11.7 42.4 ± 45.1c 31.1 ± 34.9

6 ± 414 Mitomycin C 35.0 ± 75.6 12.1 ± 16.8a 22.4 ± 59.6 11.4 ± 19.6 20.0 ± 20.5 20.4 ± 48.3 Doxorubicin 3.3 ± 5.6 1.5 ± 3.0a 2.6 ± 4.9 1.1 ± 2.0 1.5 ± 2.7 2.1 ± 4.2 Irinotecan 410 ± 755 223 ± 311 347 ± 631 1126 ± 78 244 ± 290 291 ± 522 Cisplatin 41.0 ± 35.8 25.6 ± 33.4a 29.8 ± 31.8 13.9 ± 11.7 42.4 ± 45.1c 31.1 ± 34.9 aStatistically significant difference from patients who received preoperative cytotoxic drug treatment, p < 0.05 by Mann–Whitney U test bStatistically significant difference from DPAM, p < 0.05, Kruskal–Wallis followed by Mann–Whitney U test cStatistically significant difference from PMCA intermediate, p < 0.05–0.01, Kruskal–Wallis followed by Mann–Whitney U test Analysis of OS according to clinical variables, histopathology, and drug sensitivity were performed by uni- and multivariable Cox regression for all patients with successful ex vivo assay (n = 92; Supplementary Table 2). Drug sensitivity was not statistically significantly associated with OS (univariate hazard ratio range, 0.59–1.20), whereas, as expected, impaired WHO performance status and completeness of cytoreduction (no vs. yes) were associated with shorter OS (hazard ratio, 7.93 and 11.73, respectively; p < 0.001).

tary Table 2). Drug sensitivity was not statistically significantly associated with OS (univariate hazard ratio range, 0.59–1.20), whereas, as expected, impaired WHO performance status and completeness of cytoreduction (no vs. yes) were associated with shorter OS (hazard ratio, 7.93 and 11.73, respectively; p < 0.001). Because of the strong prognostic value of complete cytoreductive surgery, subsequent analyses on prognostic impact of ex vivo drug sensitivity were performed in patients with complete cytoreduction (n = 61) with PFS as the clinical endpoint. Because only five patients in this group died during follow-up, analysis on OS was not considered. Following adjustment for performance status, PCI score, and histopathologic subtype, a strong trend towards longer PFS was observed for individuals with tumors sensitive to mitomycin C and cisplatin (p = 0.063 and 0.062, respectively; Table 3).Table 3 Univariate and multivariate Cox regression model for progression-free survival according to dichotomized drug sensitivity values (below vs. above the median IC50) and clinicopathological variables in pseudomyxoma peritonei patients with complete cytoreductive surgery (n = 61) Univariate hazard ratio p Multivariate hazard ratioa p Oxaliplatin 1.33 0.6 0.96 1.0 5-FU 1.24 0.7 0.98 1.0 Mitomycin C 0.66 0.4 0.36 0.063 Doxorubicin 1.49 0.5 1.21 0.8 Irinotecan 0.95 1.0 0.72 0.6 Cisplatin 0.54 0.3 0.36 0.062 aAdjusted for histopathological subtype, PCI score, and WHO performance status

Because of the strong prognostic value of complete cytoreductive surgery, subsequent analyses on prognostic impact of ex vivo drug sensitivity were performed in patients with complete cytoreduction (n = 61) with PFS as the clinical endpoint. Because only five patients in this group died during follow-up, analysis on OS was not considered. Following adjustment for performance status, PCI score, and histopathologic subtype, a strong trend towards longer PFS was observed for individuals with tumors sensitive to mitomycin C and cisplatin (p = 0.063 and 0.062, respectively; Table 3).Table 3 Univariate and multivariate Cox regression model for progression-free survival according to dichotomized drug sensitivity values (below vs. above the median IC50) and clinicopathological variables in pseudomyxoma peritonei patients with complete cytoreductive surgery (n = 61) Univariate hazard ratio p Multivariate hazard ratioa p Oxaliplatin 1.33 0.6 0.96 1.0 5-FU 1.24 0.7 0.98 1.0 Mitomycin C 0.66 0.4 0.36 0.063 Doxorubicin 1.49 0.5 1.21 0.8 Irinotecan 0.95 1.0 0.72 0.6 Cisplatin 0.54 0.3 0.36 0.062 aAdjusted for histopathological subtype, PCI score, and WHO performance status Because very high concentrations of cytotoxic drugs are obtained locally when subjects are treated with IPC, additional analyses on drug sensitivity in relation to PFS were conducted based on the drug activity, categorized as LDR, IDR, and EDR, at the highest drug concentration used ex vivo.15 Following adjustment for patient performance status, histopathological subtype, and PCI score, the general pattern observed was that of a stepwise increase in risk for disease progression when going from LDR to IDR and EDR ex vivo sensitivity scores (Table 4). This was statistically significant for cisplatin and 5FU and marginally so for mitomycin C. The stepwise decrease in PFS when comparing LDR to IDR and EDR for mitomycin C and cisplatin is illustrated in Fig. 1.Table 4 Univariate and multivariable Cox regression model for progression-free survival according to drug sensitivity at the highest cytotoxic drug concentration used ex vivo in pseudomyxoma patients with complete cytoreductive surgery (n = 61)

g LDR to IDR and EDR for mitomycin C and cisplatin is illustrated in Fig. 1.Table 4 Univariate and multivariable Cox regression model for progression-free survival according to drug sensitivity at the highest cytotoxic drug concentration used ex vivo in pseudomyxoma patients with complete cytoreductive surgery (n = 61) n Univariate hazard ratio p Multivariate hazard ratioa p Mitomycin C LDR 35 1 1 IDR 22 2.32 0.2 3.38 0.05 EDR 4 5.19 0.05 6.00 0.05 Cisplatin LDR 35 1 1 IDR 20 1.86 0.3 3.00 0.064 EDR 4 5.16 0.05 14.35 0.001 Irinotecan LDR 30 1 1 IDR 26 1.38 0.6 1.53 0.5 EDR 5 1.93 0.5 1.68 0.6 5FU LDR 30 1 1 IDR 26 0.52 0.3 0.55 0.4 EDR 4 3.83 0.05 4.91 0.05 Oxaliplatin LDR 33 1 1 IDR 24 0.68 0.7 2.26 0.2 EDR 3 1.28 0.9 3.52 0.3 Doxorubicin LDR 32 1 1 IDR 20 1.05 1.0 1.03 1.0 EDR 7 1.8 0.4 1.76 0.5 aAdjusted for histopathological subtype, PCI score, and WHO performance status Fig. 1 Progression-free survival in patients with complete cytoreductive surgery according to ex vivo sensitivity to mitomycin C and cisplatin categorized into low drug resistance (LDR), intermediate drug resistance (IDR), and extreme drug resistance (EDR) at the highest drug concentration tested ex vivo. Adjusted for patient performance status, histopathological subtype, and PCI score in a Cox regression model. For details on number of patients and statistical significance, see Table 4

sistance (LDR), intermediate drug resistance (IDR), and extreme drug resistance (EDR) at the highest drug concentration tested ex vivo. Adjusted for patient performance status, histopathological subtype, and PCI score in a Cox regression model. For details on number of patients and statistical significance, see Table 4 Discussion This study investigated the poorly defined role of IPC in PMP by ex vivo assessment of tumor cell sensitivity in an assay shown to reflect clinically relevant drug sensitivity in diagnostic groups as well as in individual patients.14,17,19,20 The number of patients included was quite large for this uncommon tumor type and OS was strongly associated with completeness of the CRS and patient performance status, as expected, indicating that our material and the findings are representative for PMP.4,5 The fact that the extent of CRS was strongly associated with OS, whereas ex vivo drug sensitivity was not, points to the importance of qualified surgery to achieve long-term survival in PMP. However, ex vivo drug sensitivity provided prognostic information for PFS and points to a possible impact of IPC on PFS when added to CRS. These observations are largely in line with previous findings pointing to a benefit from HIPEC for PFS but not OS as well as the possibility to achieve long-term OS with surgery alone.4,5,21–23 Still, a prolongation of PFS from IPC is considered clinically relevant, because it is reasonably associated with less disease-related symptoms and, thus, improved quality of life.

indings pointing to a benefit from HIPEC for PFS but not OS as well as the possibility to achieve long-term OS with surgery alone.4,5,21–23 Still, a prolongation of PFS from IPC is considered clinically relevant, because it is reasonably associated with less disease-related symptoms and, thus, improved quality of life. The observations indicating that IPC has limited effect and require good CRS make sense from a tumor biology point of view, because the penetrance of cytotoxic drugs into tumor tissue is very limited. Thus, no substantial effect of IPC on macroscopic tumor lesions is to be expected.8 Still, some effect from IPC in the presence of remaining macroscopic disease following CRS cannot be excluded. In our series, patients without complete CRS who had HIPEC (n = 21) had a OS of 63 versus 7 months for those who had not (n = 10), a difference that was statistically significant also after adjustment for performance status, histopathology, and PCI (p = 0.007; not shown). However, patient selection based on other prognostic factors reasonably explains most of this difference. Our finding that ex vivo drug sensitivity provided prognostic information for PFS points to a possible impact of IPC on PFS when added to CRS. Still, it cannot be excluded that ex vivo drug sensitivity is only a prognostic marker reflecting tumor behavior unrelated to a therapeutic effect from IPC. The only way to differentiate between a purely prognostic vs predictive impact from ex vivo drug sensitivity assessment would be a clinical trial in which IPC is guided by ex vivo drug sensitivity data.

vo drug sensitivity is only a prognostic marker reflecting tumor behavior unrelated to a therapeutic effect from IPC. The only way to differentiate between a purely prognostic vs predictive impact from ex vivo drug sensitivity assessment would be a clinical trial in which IPC is guided by ex vivo drug sensitivity data. This would be a way to try to improve the effect of IPC by individualized selection of active drug(s) but also to decrease treatment related toxicity by avoiding IPC if no drugs seem active. The large interindividual sample differences in drug sensitivity that we observed clearly point to the potential for IPC individualization. Such trial would be technically feasible, because tumor tissue for ex vivo analysis could be obtained by laparoscopy before the CRS and ex vivo drug sensitivity data can be obtained within a few days. Such study is presently under discussion at our center. Given that standard protocol IPC is currently part of the standard treatment for PMP, which conclusions can be drawn from the current study? The PMP samples in this study were essentially equally drug sensitive as peritoneal metastasis samples of colorectal cancer analyzed ex vivo with the same technique.15 Because several of the drugs analyzed are active in the treatment of colorectal cancer, similar drug activity also could be expected in PMP, provided similar drug exposure as in IPC.

were essentially equally drug sensitive as peritoneal metastasis samples of colorectal cancer analyzed ex vivo with the same technique.15 Because several of the drugs analyzed are active in the treatment of colorectal cancer, similar drug activity also could be expected in PMP, provided similar drug exposure as in IPC. The IC50 values for cisplatin and oxaliplatin were almost identical in the PMP subgroups. Given that a considerably higher dose of oxaliplatin compared with cisplatin can be given in IPC, the platinum of choice for IPC in PMP to achieve maximum effect is suggested to be oxaliplatin. There were no major differences in drug sensitivity between the histopathological PMP subtypes, indicating that IPC protocols for PMP do not need to consider histopathological subtype. Furthermore, the frequently observed poor prognosis of PMCA compared with DPAM seems related to tumor biological factors other than tumor cell drug sensitivity. The large interindividual differences in sensitivity to drugs used in IPC among the PMP samples were substantial. A reasonable interpretation is that IPC may be a more or less futile treatment step for patients with drug resistant tumor cells and that these patients would be better off with CRS alone.

There were no major differences in drug sensitivity between the histopathological PMP subtypes, indicating that IPC protocols for PMP do not need to consider histopathological subtype. Furthermore, the frequently observed poor prognosis of PMCA compared with DPAM seems related to tumor biological factors other than tumor cell drug sensitivity. The large interindividual differences in sensitivity to drugs used in IPC among the PMP samples were substantial. A reasonable interpretation is that IPC may be a more or less futile treatment step for patients with drug resistant tumor cells and that these patients would be better off with CRS alone. Finally, tumor cells from patients previously exposed to cytotoxic drugs were generally more drug resistant than those previously unexposed. This is in line with the clinical observation that prior chemotherapy was associated with impaired prognosis in PMP.5 Because systemic chemotherapy seems not very active in PMP, it might be argued that PMP patients should go directly to CRS and HIPEC rather than be started on systemic chemotherapy with the risk for disease progression and development of drug resistance.6 Conclusions Ex vivo assessment of tumor cell sensitivity to cytotoxic drugs provides prognostic information in PMP and may be useful for sparing the most resistant patients from IPC expected to be futile. However, whether selection of drugs for IPC in PMP based on ex vivo assessment also is predictive for a treatment effect, and thus could be used for treatment individualization, needs to be investigated in a prospective, clinical trial.

y be useful for sparing the most resistant patients from IPC expected to be futile. However, whether selection of drugs for IPC in PMP based on ex vivo assessment also is predictive for a treatment effect, and thus could be used for treatment individualization, needs to be investigated in a prospective, clinical trial. Electronic Supplementary Material Supplementary material 1 (DOCX 28 kb) Supplementary material 2 (DOCX 69 kb) Acknowledgment The skillful technical assistance of Kristin Blom, Annika Jonasson, and Anna-Karin Lannergård for the ex vivo assessment of drug sensitivity is gratefully acknowledged. This study was supported by grants from the Swedish Cancer Society and Lions Cancer Research Fund. Disclosure None of the authors have any conflicts of interest to declare.

In approximately 15–25 % of patients with rectal cancer who are treated with chemoradiotherapy (CRT), no residual tumor is found in the resection specimen, indicating a pathologic complete response (CR; ypT0N0).1 The increasing interest in organ-saving treatment through local excision or even a nonoperative treatment (a watch-and-wait strategy) demands a reliable method to identify patients with CR.2,3 Digital rectal examination (DRE) and endoscopy have been the main assessment tools to evaluate the response when the aim was to avoid surgery in specific indications, such as after contact radiotherapy in small rectal cancers.4 In the studies by Habr-Gama et al., who explored nonoperative treatment for CR in a wider group of patients, DRE and endoscopy also served as main selection tools.2,5 A drawback of endoscopy is that it only provides information on the luminal side and not on the deeper layers and the mesorectum. MRI can provide this additional information, which can be critical for decision making.6 Although MRI has been widely adopted for the primary staging of rectal cancer, restaging after CRT with standard T2-weighted (T2W) MRI is hampered by the difficulty of distinguishing fibrosis from viable tumor, often leading to incorrectly classifying fibrosis as residual tumor.6–8

can be critical for decision making.6 Although MRI has been widely adopted for the primary staging of rectal cancer, restaging after CRT with standard T2-weighted (T2W) MRI is hampered by the difficulty of distinguishing fibrosis from viable tumor, often leading to incorrectly classifying fibrosis as residual tumor.6–8 Recently, diffusion-weighted MRI (DWI) has been shown to provide more accuracy than T2W-MRI.9 Initially in our center we relied on MRI as the first restaging method and used endoscopy for further evaluation when MRI was suggestive of a CR.3,10 With this selection strategy, a substantial part of those with CR was missed. Therefore, we changed the restaging strategy to routinely include DRE and endoscopy in all patients. The aim of this study was to evaluate the respective value of clinical examination, consisting of DRE and endoscopy, with T2W-MRI and DWI for the detection of CR after CRT. Methods Patients Fifty consecutive patients were prospectively included within 3 years in a study on disease restaging after CRT. Patients provided written informed consent for this restaging study. CRT was indicated for a (1) very distal tumor or (2) T4 tumor or (3) T3 tumor with involved mesorectal fascia and/or N1 disease with distal or midrectal location or (4) N2 status. CRT consisted of 28 fractions of 1.8 Gy radiation with capecitabine 825 mg/m2. Restaging was scheduled 6–8 weeks after completion of CRT.

y. CRT was indicated for a (1) very distal tumor or (2) T4 tumor or (3) T3 tumor with involved mesorectal fascia and/or N1 disease with distal or midrectal location or (4) N2 status. CRT consisted of 28 fractions of 1.8 Gy radiation with capecitabine 825 mg/m2. Restaging was scheduled 6–8 weeks after completion of CRT. Clinical Assessment: DRE and Endoscopy The patients were examined by one of three colorectal surgeons (GB, SB, LS). At DRE, findings were classified as: (1) normal bowel wall, (2) subtle residual abnormality of the bowel wall, and (3) obvious residual tumor. All patients underwent flexible endoscopy (Pentax Medical Netherlands, Uithoorn, The Netherlands) of the rectum after a rectal phosphate enema. Only white light imaging was used with HDTV, and the images of the tumor area were digitally stored. CR was defined as the absence of residual tumor with only a flat, white scar with or without teleangiectasia (Fig. 1). A small, flat ulcer with smooth edges without signs of residual polypoid tissue was considered to be a potential CR (Fig. 1). Every other type of ulcer or mass was considered as definite residual tumor (Fig. 1). A biopsy was only performed in equivocal cases, as judged by the surgeon during the endoscopy. Biopsy results that indicated tumor or high-grade dysplasia were considered proof of residual tumor. Absence of tumor or high-grade dysplasia in biopsy samples was not considered definite proof of CR because of the risk of sampling error. For the purpose of this study, two experienced clinicians (GB and MM, blinded to the MRI results and further clinical outcome), in consensus, rated the combination of the DRE and endoscopy findings and assigned a confidence level score for the overall clinical assessment (Table 1).Fig. 1 Response assessment with T2W-MRI (a–c) and with endoscopy (d–f). Pre- and post-CRT MR images are shown. T indicates tumor; arrows indicate scar or residual tumor after CRT. a Typical CR at T2W-MRI, b equivocal image at T2W-MRI, and c obvious residual tumor at T2W-MRI. d Typical endoluminal image of CR with white scar with teleangiectasia. e Small ulcer with smooth edges (arrows) but without residual polypoid tissue. Patients imaged in (d) and (e) experienced sustained clinical CR at follow-up. f Example of large ulcer that was deemed residual tumor after CRT

sidual tumor at T2W-MRI. d Typical endoluminal image of CR with white scar with teleangiectasia. e Small ulcer with smooth edges (arrows) but without residual polypoid tissue. Patients imaged in (d) and (e) experienced sustained clinical CR at follow-up. f Example of large ulcer that was deemed residual tumor after CRT Table 1 Definitions of confidence level scores for assessment of complete response for every modality

sidual tumor at T2W-MRI. d Typical endoluminal image of CR with white scar with teleangiectasia. e Small ulcer with smooth edges (arrows) but without residual polypoid tissue. Patients imaged in (d) and (e) experienced sustained clinical CR at follow-up. f Example of large ulcer that was deemed residual tumor after CRT Table 1 Definitions of confidence level scores for assessment of complete response for every modality CL Clinical assessment T2W-MRI findings DWI findings CL 0 Positive biopsy result or gross residual tumor at endoscopy with or without palpable mass at DRE Gross residual isointense mass and/or involved nodes Marked hyperintense signal at former tumor location on b1000 images with low ADC CL 1 Visible (with or without palpable) mass or polypoid tissue with negative biopsy Small residual isointense mass and/or involved nodes Small but obvious area of hyperintense signal at former tumor location on b1000 images with low ADC CL 2 Ulcer with irregular borders and small palpable ridge, ulcer or wall thickening with negative biopsy Irregular wall thickening with both hypointense and isointense signal Possible foci of hyperintense signal on b1000 images at former tumor location with low ADC in an area of irregular wall thickening CL 3 Small nonpalpable ulcer with regular borders and negative biopsy Pronounced hypointense wall thickening without isointense signal and no involved nodes No clear areas of residual hyperintense signal on b1000 images at former tumor location CL 4 White scar with teleangiectasia, no palpable lesions and negative biopsy Normalized rectal wall or only subtle wall hypointense wall thickening and no involved nodes No residual hyperintense signal on b1000 images or low ADC at former tumor location

dual hyperintense signal on b1000 images at former tumor location CL 4 White scar with teleangiectasia, no palpable lesions and negative biopsy Normalized rectal wall or only subtle wall hypointense wall thickening and no involved nodes No residual hyperintense signal on b1000 images or low ADC at former tumor location CL confidence level, T2W-MRI T2-weighted MRI, DWI diffusion-weighted imaging, DRE digital rectal examination, ADC apparent diffusion coefficient

dual hyperintense signal on b1000 images at former tumor location CL 4 White scar with teleangiectasia, no palpable lesions and negative biopsy Normalized rectal wall or only subtle wall hypointense wall thickening and no involved nodes No residual hyperintense signal on b1000 images or low ADC at former tumor location CL confidence level, T2W-MRI T2-weighted MRI, DWI diffusion-weighted imaging, DRE digital rectal examination, ADC apparent diffusion coefficient MRI All MRI examinations were performed at 1.5 T using a phased array body coil [Intera (Achieva) or Ingenia, Philips Medical Systems, Best, The Netherlands] and included T2W-MRI in three orthogonal directions (axial, sagittal, and coronal). Additional axial diffusion-weighted images were obtained with b0 as the lowest and b1000 as the highest b value. The sequence details are shown in Appendix. An intravenous bolus injection of 20 mg of butylscopolamine (Buscopan; Boehringer Ingelheim, Ingelheim, Germany) was administered to reduce peristaltic movement; patients did not receive bowel preparation. An apparent diffusion coefficient (ADC) map was automatically calculated. The T2W-MRI and DWI axial images were angled in identical planes. A reader with 5-year specific experience in rectal cancer MRI (DL) scored the T2W-MRI images together with the DWI (b1000 and ADC) images for the presence of CR with confidence level scores (Table 1). ycN0 was assessed on the basis of size and morphology criteria.11 The reader had the pre-CRT MRI at her disposal and was blinded to the endoscopy results and histopathology (if available). Figure 1 shows examples of a CR, equivocal score, and obvious residual tumor by T2W-MRI. Figure 2 illustrates an example of DWI being decisive in determining a CR when clinical assessment and T2W-MRI show equivocal results.Fig. 2 Example of patient with a CR where T2W-MRI (a) revealed marked hypointense residual wall thickening resulting with an equivocal (confidence level 2) score. Clinical assessment (b) revealed a white scar with some stenosis and distortion, and small superficial ulceration, also resulting in an equivocal score. DWI (c) revealed absence of diffusion restriction indicating CR

led marked hypointense residual wall thickening resulting with an equivocal (confidence level 2) score. Clinical assessment (b) revealed a white scar with some stenosis and distortion, and small superficial ulceration, also resulting in an equivocal score. DWI (c) revealed absence of diffusion restriction indicating CR Reference Standards Histopathology of the total mesorectal excision (TME) resection specimen was used as the reference standard, with both high-grade dysplasia and carcinoma considered as residual tumor. CR was defined as ypT0N0. Surgical specimens were evaluated according to the method of Quirke and Dixon.12 Some Patients underwent clinical exams and endoscopy + DWI-MRI in the first year of follow-up every 3 months and from the second year this was performed every 6 months.3 For these patients, a local recurrence-free follow-up time of ≥12 months was used as a surrogate end point for a CR. For patients who underwent a local excision of the remaining scar [transanal endoscopic microsurgery (TEM)], the reference standard consisted of histopathology of the specimen with ≥12 months’ follow-up by MRI and endoscopy.

local recurrence-free follow-up time of ≥12 months was used as a surrogate end point for a CR. For patients who underwent a local excision of the remaining scar [transanal endoscopic microsurgery (TEM)], the reference standard consisted of histopathology of the specimen with ≥12 months’ follow-up by MRI and endoscopy. Statistical Analysis Statistical analyses were performed with SPSS Statistics 20 (IBM, Armonk, NY) and Stata 11.0 (StataCorp, College Station, TX). Receiver operator characteristics (ROC) curves were constructed with confidence levels to assess the diagnostic performance of clinical assessment and MRI. The areas under the ROC curve (AUC) with corresponding sensitivities and specificities were calculated for all modalities. The cutoff for sensitivity and specificity was set between confidence level 2 and 3 at the start of the study for both clinical assessment and MRI. AUCs were compared between modalities by the method of Hanley and McNeil.13 With logistic regression analyses, predicted probabilities were calculated for the diagnostic performance of the combination of MRI with clinical assessment. With these predicted probabilities, a ROC curve was constructed. The positive outcome measure was the presence of a CR. In addition to the diagnostic performance of the modalities, the positive and negative likelihood ratios were calculated for the following: (1) clinical assessment, (2) T2W-MRI with DWI, and (3) both modalities combined.14 These likelihood ratios were used to calculate posttest probabilities for a CR when the modalities are combined by the multiplying pretest odds with the likelihood ratios. P values of <0.05 were considered statistically significant.

ing: (1) clinical assessment, (2) T2W-MRI with DWI, and (3) both modalities combined.14 These likelihood ratios were used to calculate posttest probabilities for a CR when the modalities are combined by the multiplying pretest odds with the likelihood ratios. P values of <0.05 were considered statistically significant. Results Patients Of the 50 included patients, 33 were men (66 %). The median age was 67.5 years (range 34–88 years). Thirty-four patients underwent a TME, and six underwent a TEM as part of a study.

ing: (1) clinical assessment, (2) T2W-MRI with DWI, and (3) both modalities combined.14 These likelihood ratios were used to calculate posttest probabilities for a CR when the modalities are combined by the multiplying pretest odds with the likelihood ratios. P values of <0.05 were considered statistically significant. Results Patients Of the 50 included patients, 33 were men (66 %). The median age was 67.5 years (range 34–88 years). Thirty-four patients underwent a TME, and six underwent a TEM as part of a study. Seventeen patients experienced a CR (34 %): eight after surgery (two after TEM, six after TME), and nine had a clinical CR and were followed with a watch-and-wait policy, with a median follow-up of 17 months (range 12–20 months). One patient with residual tumor had ypT0N1 disease. At primary staging, 72 % of patients had a cT3 tumor (36 of 50), 20 % (10 of 50) had a cT2 tumor, and 8 % (4 of 50) had a cT4 tumor. At primary presentation, 38 (76 %) of 50 tumors were palpable at DRE. The median interval between the last radiation dose and the restaging was 8 weeks (range 3–35 weeks), and between restaging MRI and histopathology 9.5 days (range 0–74 days). The median time between clinical assessment and restaging MRI was 0 days (range 0–56 days). At endoscopy, biopsies were performed in 29 patients; findings were benign in 20 patients, eight of which turned out to be false CRs after surgery. In three patients the biopsy results revealed adenocarcinoma, and high-grade dysplasia was found in six. In this small sample, the sensitivity of a biopsy for persistent tumor was 9 (53 %) of 17, and the negative predictive value for persistent tumor was 12 (60 %) of 20.

which turned out to be false CRs after surgery. In three patients the biopsy results revealed adenocarcinoma, and high-grade dysplasia was found in six. In this small sample, the sensitivity of a biopsy for persistent tumor was 9 (53 %) of 17, and the negative predictive value for persistent tumor was 12 (60 %) of 20. Diagnostic Performance Figure 3 shows the ROC curves for MRI and clinical assessment. Table 2 shows the diagnostic parameters for the modalities. For clinical assessment, the AUC was 0.88 (95 % confidence interval 0.78–0.99), and sensitivity and specificity were 53 and 97 %, respectively. For T2W-MRI and DWI, the AUC was 0.79 (95 % confidence interval 0.66–0.92), with a sensitivity of 35 % and a specificity of 94 %. The difference between clinical assessment and T2W-MRI and DWI was not statistically significant (P = 0.17).Fig. 3 ROC curves for modalities. Clinical assessment consists of endoscopy, DRE, and biopsy result (if available) Table 2 Diagnostic parameters for clinical assessment, T2W-MRI and DWI, and all assessment modalities

Diagnostic Performance Figure 3 shows the ROC curves for MRI and clinical assessment. Table 2 shows the diagnostic parameters for the modalities. For clinical assessment, the AUC was 0.88 (95 % confidence interval 0.78–0.99), and sensitivity and specificity were 53 and 97 %, respectively. For T2W-MRI and DWI, the AUC was 0.79 (95 % confidence interval 0.66–0.92), with a sensitivity of 35 % and a specificity of 94 %. The difference between clinical assessment and T2W-MRI and DWI was not statistically significant (P = 0.17).Fig. 3 ROC curves for modalities. Clinical assessment consists of endoscopy, DRE, and biopsy result (if available) Table 2 Diagnostic parameters for clinical assessment, T2W-MRI and DWI, and all assessment modalities Parameter Clinical assessment T2W-MRI and DWI All Sensitivity 53 % 35 % 71 % Specificity 97 % 94 % 97 % PPV 90 % 75 % NA NPV 80 % 74 % NA AUC 0.88 (0.78–0.99) 0.79 (0.66–0.92) 0.89 (0.79–0.99) LR positive 17.67 5.83 – LR negative 0.48 0.69 – Positive posttest probability 90 % 75 % 98 % Negative posttest probability 20 % 26 % 15 % Positive posttest probability is the probability of CR when both tests have positive results (indicate CR) and negative posttest probability is the probability of CR when both tests have negative results (indicate residual tumor). Diagnostic parameters were calculated on the basis of predefined cutoff in confidence levels between 2 and 3

robability is the probability of CR when both tests have positive results (indicate CR) and negative posttest probability is the probability of CR when both tests have negative results (indicate residual tumor). Diagnostic parameters were calculated on the basis of predefined cutoff in confidence levels between 2 and 3 T2W-MRI T2-weighted MRI, DWI diffusion-weighted MRI, NA not applicable, PPV positive predictive value, NPV negative predictive value, AUC area under the receiver operator characteristic curve, LR likelihood ratio Probability for CR with Combination of Methods The positive likelihood ratio for a CR for clinical assessment was 17.67 and for T2W-MRI and DWI 5.83. The posttest probability (calculated with the positive likelihood ratios) for the presence of a CR for clinical assessment was 90 % and for T2W-MRI and DWI MRI was 75 %. When all three modalities were combined, the posttest probability for a CR was 98 %, indicating that when all three modalities predict a CR, this is correct in 98 % of the cases, with only a 2 % risk of missing residual tumor. The negative likelihood ratio was 0.48 for clinical assessment and 0.69 for T2W-MRI with DWI. These likelihood ratios led to a posttest probability of a CR of 20 % for clinical assessment and 26 % for T2W-MRI and DWI when either of the modalities indicates residual tumor. When combining all modalities, this decreases to 15 %, meaning that when all three modalities indicate residual tumor, there still is a 15 % chance for a CR.

ratios led to a posttest probability of a CR of 20 % for clinical assessment and 26 % for T2W-MRI and DWI when either of the modalities indicates residual tumor. When combining all modalities, this decreases to 15 %, meaning that when all three modalities indicate residual tumor, there still is a 15 % chance for a CR. Figures 4 and 5 show how the modalities complement each other in assessment of response after CRT.Fig. 4 a Tumor (asterisks) before CRT. After CRT at T2W-MRI (b), fibrosis (arrows) is found with absence of high signal on DWI (c), suggestive of a CR. At endoscopy (d), a residual ulcer (arrows) is found, indicating residual tumor. Patient refused surgery and has been followed up for 3.5 years with stable MR image and a healed ulcer (e, arrows), so is classified as having experienced CR Fig. 5 a, b Distal tumor (asterisks) before CRT at T2W-MRI and c DWI. After CRT at T2W-MRI (d) and DWI (e), residual tumor was suspected (arrows). At endoscopy (f), CR (arrows) was determined, and the patient was treated with wait-and-see policy. After 3 months, DWI became normal; patient remained free of recurrent disease at 3.8 years of follow-up

risks) before CRT at T2W-MRI and c DWI. After CRT at T2W-MRI (d) and DWI (e), residual tumor was suspected (arrows). At endoscopy (f), CR (arrows) was determined, and the patient was treated with wait-and-see policy. After 3 months, DWI became normal; patient remained free of recurrent disease at 3.8 years of follow-up Discussion In this study, clinical assessment including DRE and endoscopy proved to be the most accurate strategy to select patients who may experience CR. The addition of MRI to DWI, however, increases the identification CR rate to a level that is reliable for clinical decision making. When clinical assessment, T2W-MRI, and DWI all indicate a CR, this is correct in 98 % of the cases, missing residual tumor in only 2 %. When all modalities indicate residual tumor, in 15 % of the cases, there is actually is a CR.

eases the identification CR rate to a level that is reliable for clinical decision making. When clinical assessment, T2W-MRI, and DWI all indicate a CR, this is correct in 98 % of the cases, missing residual tumor in only 2 %. When all modalities indicate residual tumor, in 15 % of the cases, there is actually is a CR. Rigid endoscopy and DRE have been the standard of response assessment in the past treatment of rectal cancer with radiotherapy alone.15 A continuing decrease in size and disappearance of the tumor with healing of the mucosa were generally considered signs of a clinical CR. In later series, rigid endoscopy was often replaced by flexible sigmoidoscopy and imaging with endorectal ultrasound with or without MRI was added.16 Habr-Gama et al. showed that whitening of the mucosa (with or without teleangiectasia) or a complete normalization of the tumor bed should be considered a CR, a finding that is confirmed in the current study.5 The literature has shown that residual tumor can be found in any layer of the bowel wall, regardless of tumor stage or presence of ulcer.17–20 Therefore, a major concern of clinical and endoscopic assessment of response is the risk of missing such scattered tumor deposits, leading to the cautious strategy to perform a major resection whenever potential residual tumor is suspected. This approach and the degree of subjectivity of clinical assessment are illustrated in a study where DRE only detected 3 of 14 patients with a CR, while a CR was never falsely predicted in the 80 patients with residual tumor.21 Given the fact that sampling errors occur regularly in case of residual tumor, biopsies have only limited clinical value for ruling out residual cancer.22 This variability of tumor scatter could explain the 10–30 % early and late regrowths in series of watchful waiting and underscores the need for imaging methods that evaluate the deeper layers of the bowel wall and the mesorectum.3,23–25

tumor, biopsies have only limited clinical value for ruling out residual cancer.22 This variability of tumor scatter could explain the 10–30 % early and late regrowths in series of watchful waiting and underscores the need for imaging methods that evaluate the deeper layers of the bowel wall and the mesorectum.3,23–25 Endorectal ultrasound, FGD-PET-CT, and T2W-MRI all have shown insufficient diagnostic performance to detect residual tumor in fibrosis after CRT, and the strategy to err on the safe side leads to overestimation of residual tumor.9,26–28 The accuracy of T2W-MRI can be improved by adding a DWI sequence, generating qualitative and quantitative information on the cellular architecture on the basis of differences in movement (diffusion) of water protons within the various tissues. Malignant tissues, with a high cellular density, show restricted proton movement leading to an increased signal. A meta-analysis on response assessment in rectal cancer has shown that DWI improves the diagnostic performance, mainly through increasing the detection rate of response up to 84 %, along with a very low risk of missing residual tumor.9 In the present study, combined prediction of a CR on clinical assessment as well as MRI including DWI resulted in a very high predictive value for a CR of 98 %. With this strategy, however, about one in three CRs is missed.

he detection rate of response up to 84 %, along with a very low risk of missing residual tumor.9 In the present study, combined prediction of a CR on clinical assessment as well as MRI including DWI resulted in a very high predictive value for a CR of 98 %. With this strategy, however, about one in three CRs is missed. A clinically relevant question is whether it is necessary to err so much on the safe side. A transanal excision of the scar can provide histologic proof when there is an equivocal clinical and radiologic picture. The disadvantage is that follow-up is somewhat more difficult, and in the event of a recurrence, the ideal surgical plane may have been violated. Another alternative is to extend the observation interval for an additional 1–2 months, as it can take several months before the full effect of the CRT becomes evident.29 The two approaches of local excision and extending the observation interval will increase the number of patients who can be offered organ preservation.

other alternative is to extend the observation interval for an additional 1–2 months, as it can take several months before the full effect of the CRT becomes evident.29 The two approaches of local excision and extending the observation interval will increase the number of patients who can be offered organ preservation. The most practical and cost-efficient strategy to identify patients likely to experience clinical CR also depends on local logistics and expertise. Currently, experience with clinical assessment after CRT is limited and lacks standardization. Additionally, clinical assessment has a high degree of observer variability. When restaging MRI is part of the routine, it could serve as a first selection tool and avoid unnecessary endoscopies in patients with obvious residual tumor. When restaging MRI is not part of the routine, DRE is by far the most cost-efficient way to determine gross residual tumor. Regardless of the first screening method, it is prudent in patients considered candidates for organ preservation to use all methods: DRE, endoscopy, and MRI. MRI provides information on the presence of tumor in the deeper layers of the rectal wall, the mesorectum, and the lymph nodes, and it provides detailed images that can be used for serial follow-up.10

od, it is prudent in patients considered candidates for organ preservation to use all methods: DRE, endoscopy, and MRI. MRI provides information on the presence of tumor in the deeper layers of the rectal wall, the mesorectum, and the lymph nodes, and it provides detailed images that can be used for serial follow-up.10 The most important limitation of the present study is the relatively small sample size, and thus some caution in the interpretation of the results is required. Second, the prevalence of CR after CRT (34 %) is higher than generally reported in the literature (15–25 %) as a result of the referral pattern to our center of patients with a good response. Another limitation is that in some patients, the reference standard was a lasting clinical CR at follow-up of at least 1 year, with a median of 16.5 months. Although most regrowths occur within the first year of follow-up, it cannot be excluded that some will occur later. Additionally, the range in interval between last radiation dose and response evaluation and surgery is wide, which could have an influence on our results. In conclusion, clinical assessment with DRE and endoscopy is the most accurate strategy to identify patients likely to experience CR, and it should be incorporated in a post-CRT restaging strategy when organ preservation is considered. Addition of MRI (including DWI) further improves diagnostic performance, and the combination of the two can be recommended as a strategy for a safe and accurate selection of CR after CRT.

ients likely to experience CR, and it should be incorporated in a post-CRT restaging strategy when organ preservation is considered. Addition of MRI (including DWI) further improves diagnostic performance, and the combination of the two can be recommended as a strategy for a safe and accurate selection of CR after CRT. Appendix The standard rectal imaging protocol consisted of T2W fast-spin echo sequences in three orthogonal planes (TR 8456–9558 ms, TE 130 ms, 25 echotrain length, 2–6 NSA, 0.78 × 1.14 × 3.00 mm voxel size, 30 slices, 4.37–6.03 min acquisition time). Diffusion-weighted imaging (TR/TE 4829/70 ms, EPI factor 53–61, 5 NSA, 1.8 × 2.3 × 5 mm acquisition voxel size, 24–50 slices, 5.33–10.37 min’ acquisition time) was performed with DWIBS in half of the patients and with SPIR/SPAIR in the remaining patients. Disclosure The authors declare no conflict of interest.

Defining Breast Cancer Risk Defining breast cancer risk incorporates knowledge of individual risk factors known to be associated with increased risk. These risk factors are included in various available risk-calculation models to provide a numeric risk that can be used to help quantify the level of individual risk.1 Breast cancer risk factors have historically been described as modifiable versus nonmodifiable factors. Modifiable risk factors in general are associated with lifestyle behaviors and exogenous hormone exposure. These include physical inactivity, increased alcohol consumption, obesity, and use of estrogen and progestin therapies, all of which are associated with increasing breast cancer risk.2–5 Physicians have an important role in counseling women on the effectiveness of lifestyle modification and avoidance of long-term postmenopausal hormone therapy in the primary prevention of breast cancer. Nonmodifiable risk factors include increasing age, family history, precancerous breast lesions, and reproductive factors (early menarche, late-onset menopause, first live birth after age 30 years, or nulliparity). These risk factors are independently associated with a higher risk of developing breast cancer but it is not known if they are additive for an individual when estimating breast cancer risk.

ast lesions, and reproductive factors (early menarche, late-onset menopause, first live birth after age 30 years, or nulliparity). These risk factors are independently associated with a higher risk of developing breast cancer but it is not known if they are additive for an individual when estimating breast cancer risk. Breast cancer risk can be categorized as average, high, and very high risk.6 In general, a woman having no family history of breast cancer or prior history of a precancerous breast biopsy would be considered at average risk. The lifetime risk for developing breast cancer for an average-risk woman is 12 %. The following criteria are most often used to identify women at high risk: (i) first-degree relative with a breast cancer diagnosis before age 50 years; (ii) history of atypical hyperplasia (AH); (iii) 5-year Gail model risk of ≥1.7 %; (iv) history of lobular carcinoma in situ (LCIS); (v) having received chest radiation between the ages of 10 and 30 years; (vi) increased mammographic breast density; and (vii) International Breast Cancer Intervention Study (IBIS) model (Tyrer–Cuzick) lifetime risk of ≥20 %.7–12 Breast cancer risk factors and the respective absolute or attributable risk of developing breast cancer are described in Table 1.Table 1 Definition of high risk

s; (vi) increased mammographic breast density; and (vii) International Breast Cancer Intervention Study (IBIS) model (Tyrer–Cuzick) lifetime risk of ≥20 %.7–12 Breast cancer risk factors and the respective absolute or attributable risk of developing breast cancer are described in Table 1.Table 1 Definition of high risk Risk factor Defining high risk First-degree family member diagnosed at <50 years of age Twofold risk Atypical hyperplasia Cumulative absolute risk is 30 % at 25-year follow-up Chest radiation between 10 and 30 years of age 40 % lifetime risk Gail model 5-year risk Five-year risk ≥1.7 % Breast density (BI-RADS, D3 or D4) Women with extremely dense breasts have a twofold increased risk compared with average women Lobular carcinoma in situ 25 % lifetime risk International Breast Cancer Intervention Study model (Tyrer–Cuzick model) life-time risk ≥20 % lifetime risk BI-RADS Breast Imaging Reporting and Data System, D3 the breast tissue is heterogeneously dense, D4 the breast tissue is extremely dense

Risk factor Defining high risk First-degree family member diagnosed at <50 years of age Twofold risk Atypical hyperplasia Cumulative absolute risk is 30 % at 25-year follow-up Chest radiation between 10 and 30 years of age 40 % lifetime risk Gail model 5-year risk Five-year risk ≥1.7 % Breast density (BI-RADS, D3 or D4) Women with extremely dense breasts have a twofold increased risk compared with average women Lobular carcinoma in situ 25 % lifetime risk International Breast Cancer Intervention Study model (Tyrer–Cuzick model) life-time risk ≥20 % lifetime risk BI-RADS Breast Imaging Reporting and Data System, D3 the breast tissue is heterogeneously dense, D4 the breast tissue is extremely dense Women presenting with a strong hereditary predisposition, or known BRCA1 or 2 mutation carriers, are, by definition, considered at very high risk for developing breast cancer. A family history that entails multiple affected relatives with early-onset breast or ovarian cancer over several generations would be an indication to refer to a genetic counselor to discuss the options of genetic testing. The lifetime risk of developing invasive breast cancer for a BRCA mutation carrier is estimated at 40–85 %.13 Women with a BRCA mutation should be offered bilateral prophylactic mastectomy (BPM) and risk-reducing salpingo-oophorectomy as these are the only risk-reducing strategies shown to be effective in this population. Those not interested in BPM should have enhanced surveillance with annual mammogram and magnetic resonance imaging, and be offered preventive therapy. The evidence of efficacy of preventive therapy in this population is less compelling.14,15 Although there is no evidence to support BPM in women who have had thoracic radiation, there is preclinical evidence that tamoxifen decreases the incidence of radiation-induced breast cancer.16,17

d be offered preventive therapy. The evidence of efficacy of preventive therapy in this population is less compelling.14,15 Although there is no evidence to support BPM in women who have had thoracic radiation, there is preclinical evidence that tamoxifen decreases the incidence of radiation-induced breast cancer.16,17 Several complementary risk assessment and calculation tools are available to assist physicians with making decisions regarding preventive therapy, and individualizing risks. These tools incorporate most of the breast cancer risk factors described above and are easily available to the physician at the point of care. When counseling women about preventive therapy, it is recommended that physicians use a shared decision-making approach with women at high or very high risk as they are most likely to benefit from risk-reduction options.18,19 Women with a history of prior chest-wall radiation age <30 years, or women with a history of LCIS, are considered to be high enough risk to be considered for preventive therapy [National Comprehensive Cancer Network (NCCN) guidelines version 1.2014 Breast Cancer Risk Reduction]. Other women can be assessed for suitability by using a risk assessment tool.

wall radiation age <30 years, or women with a history of LCIS, are considered to be high enough risk to be considered for preventive therapy [National Comprehensive Cancer Network (NCCN) guidelines version 1.2014 Breast Cancer Risk Reduction]. Other women can be assessed for suitability by using a risk assessment tool. Determining Eligibility for Preventive Therapy/Risk Assessment Tools The American Society of Clinical Oncology, the NCCN, the Canadian Task Force on Preventive Health Care, and the US Preventive Services Task Force (USPSTF) advise counseling women ≥35 years of age who are at increased risk for breast cancer regarding available medications to reduce their risk and to offer medication to women at low risk of medication-related side effects (USPSTF B recommendation).20–23 The Gail model risk calculator is the most widely utilized tool to identify candidates suitable for chemoprevention.9,24–26 The original validated Gail model was updated and modified to become the Breast Cancer Risk Assessment Tool (BCRAT) by the National Cancer Institute and the National Surgical Adjuvant Breast and Bowel Project (NSABP) Biostatistics Center.27 The BCRAT includes the following breast cancer risk factors: current age, reproductive history (age at menarche, age at first live birth), history of prior breast disease (number of previous breast biopsies and history of AH), and family history (number of first-degree relatives with breast cancer), with age being the most heavily weighted risk factor.27

ast cancer risk factors: current age, reproductive history (age at menarche, age at first live birth), history of prior breast disease (number of previous breast biopsies and history of AH), and family history (number of first-degree relatives with breast cancer), with age being the most heavily weighted risk factor.27 This model does not include the age of onset of breast cancer in family members, paternal family history, or any family history of ovarian cancer. It is suitable for women ≥35 years of age with no history of ductal carcinoma in situ (DCIS) or LCIS, no prior history of thoracic radiation, and without a strong family history of breast cancer or ovarian cancer suggestive of a genetic predisposition. The model was updated in 2008 to provide adjusted estimates for African American women derived from the Women’s Contraceptive and Reproductive Experiences (CARE) Study and from Surveillance, Epidemiology and End Results (SEER) data, and in 2011 to include Asian and Pacific Islander women using data from the Asian American Breast Cancer Study (AABCS) combined with the SEER database.25,28

an American women derived from the Women’s Contraceptive and Reproductive Experiences (CARE) Study and from Surveillance, Epidemiology and End Results (SEER) data, and in 2011 to include Asian and Pacific Islander women using data from the Asian American Breast Cancer Study (AABCS) combined with the SEER database.25,28 Any woman with a 5-year risk of ≥1.7 % determined by using the Gail model can be considered for preventive therapy. This is the risk estimate utilized for the major breast cancer prevention trials and supported by NCCN guidelines.21 Based on risk–benefit tables developed by Freedman et al.29 the USPSTF concludes that, in general, women with an estimated 5-year breast cancer risk of ≥3 % are likely to have more benefit than harm from using a selective estrogen receptor modulator (SERM) as chemoprevention, although the balance depends on age, ethnicity, the medication used, and whether or not the patient has a uterus.23 In general, women with a history of AH, or women under the age of 50 years, are more likely to benefit from preventive therapy. This is based on the Breast Cancer Prevention Trial (BCPT) data subgroup analysis that demonstrated a significant 86 % risk reduction for women with AH. Furthermore, the evidence supports that women under the age of 50 years are far less likely to incur the harms of therapy seen in women 50 years of age or older.30,31 Conversely, in many older women the harms of preventive therapy far outweigh the benefits as their risk of adverse effects is greater.

for women with AH. Furthermore, the evidence supports that women under the age of 50 years are far less likely to incur the harms of therapy seen in women 50 years of age or older.30,31 Conversely, in many older women the harms of preventive therapy far outweigh the benefits as their risk of adverse effects is greater. The NCCN Breast Cancer Risk Reduction Panel has adopted the 1.7 % or greater 5-year actuarial breast cancer risk defined by the modified Gail model as the risk threshold for discussion of chemoprevention. This is consistent with eligibility criteria utilized in the NSABP BCPT and the Study of Tamoxifen and Raloxifene (STAR).30–33Another risk calculation model commonly used is the IBIS or Tyrer–Cuzick model.11 It includes BRCA status, height, weight, hormone replacement therapy (HRT) use, age at first live birth, age of onset of cancers in relatives, the presence of ovarian cancer, and second- and third-generation family history on the maternal and paternal side. It is more complex, less accessible to primary care providers, and currently utilized mainly to determine eligibility for enhanced screening with MRI, in addition to mammography, in women with a lifetime risk of breast cancer ≥20 %. The recently updated American Society of Clinical Oncology guideline on the use of pharmacological interventions for breast cancer risk reduction states that the risk for breast cancer may be determined by the aforementioned BCRAT tool “or other validated models including Tyrer–Cuzick”.20

The NCCN Breast Cancer Risk Reduction Panel has adopted the 1.7 % or greater 5-year actuarial breast cancer risk defined by the modified Gail model as the risk threshold for discussion of chemoprevention. This is consistent with eligibility criteria utilized in the NSABP BCPT and the Study of Tamoxifen and Raloxifene (STAR).30–33Another risk calculation model commonly used is the IBIS or Tyrer–Cuzick model.11 It includes BRCA status, height, weight, hormone replacement therapy (HRT) use, age at first live birth, age of onset of cancers in relatives, the presence of ovarian cancer, and second- and third-generation family history on the maternal and paternal side. It is more complex, less accessible to primary care providers, and currently utilized mainly to determine eligibility for enhanced screening with MRI, in addition to mammography, in women with a lifetime risk of breast cancer ≥20 %. The recently updated American Society of Clinical Oncology guideline on the use of pharmacological interventions for breast cancer risk reduction states that the risk for breast cancer may be determined by the aforementioned BCRAT tool “or other validated models including Tyrer–Cuzick”.20 In a head-to-head comparison of the BCRAT and the IBIS model looking at the absolute 10-year risk of breast cancer, the IBIS model showed better discrimination (area under the curve [AUC] for IBIS 69.5 %, 95 % CI 63.8–75.2 versus AUC for BRCAT 63.2 %, 95 % CI 57.6–68.9).34

The recently updated American Society of Clinical Oncology guideline on the use of pharmacological interventions for breast cancer risk reduction states that the risk for breast cancer may be determined by the aforementioned BCRAT tool “or other validated models including Tyrer–Cuzick”.20 In a head-to-head comparison of the BCRAT and the IBIS model looking at the absolute 10-year risk of breast cancer, the IBIS model showed better discrimination (area under the curve [AUC] for IBIS 69.5 %, 95 % CI 63.8–75.2 versus AUC for BRCAT 63.2 %, 95 % CI 57.6–68.9).34 There is no validated model that accounts for breast density, yet it is hoped that one might be developed in the future that will include breast density and be capable of effectively identifying women suitable for both enhanced screening and chemoprevention.35 Preventive Therapy Tamoxifen and raloxifene, both SERMs, as well as two aromatase inhibitors (AIs), exemestane and anastrozole, have been shown in randomized controlled trials to significantly reduce breast cancer incidence in women at increased risk of the disease.30–33,36,37 The SERMs are US FDA approved for this indication in postmenopausal women, although only tamoxifen has been studied and received an indication for breast cancer risk reduction in premenopausal women. The FDA has not approved either of these two AIs for breast cancer risk reduction, and their use in the US is considered off-label. There are a paucity of data on the effectiveness of preventive therapy in women with a history of chest-wall radiation.38 Tamoxifen and Raloxifene

Preventive Therapy Tamoxifen and raloxifene, both SERMs, as well as two aromatase inhibitors (AIs), exemestane and anastrozole, have been shown in randomized controlled trials to significantly reduce breast cancer incidence in women at increased risk of the disease.30–33,36,37 The SERMs are US FDA approved for this indication in postmenopausal women, although only tamoxifen has been studied and received an indication for breast cancer risk reduction in premenopausal women. The FDA has not approved either of these two AIs for breast cancer risk reduction, and their use in the US is considered off-label. There are a paucity of data on the effectiveness of preventive therapy in women with a history of chest-wall radiation.38 Tamoxifen and Raloxifene In the landmark BCPT, tamoxifen reduced the risk of breast cancer in both pre- and postmenopausal women at increased risk of the disease by approximately one-half (relative risk [RR] 0.51; 95 % CI 0.39–0.66). Women with AH had a highly significant 86 % breast cancer risk reduction (RR 0.14; 95 % CI 0.03–0.47), whereas women with LCIS, due to the small sample size, had a nonstatistically significant reduction of 56 % (RR 0.44; 95 % CI 0.16–1.06).30 Women under the age of 50 years obtained comparable breast cancer risk reduction to women 50 years of age and older. In the 7-year follow-up analysis, the benefits of tamoxifen were shown to persist in women at increased risk of the disease, even after stopping therapy, with a reduction in breast cancer risk of 43 % (RR 0.57; 95 % CI 0.46–0.70). Risk remained decreased by 75 % (RR 0.25; 95 % CI 0.10–0.52) in women with AH, while women with LCIS continued to have a nonstatistically significant risk reduction, now 46 % (RR 0.54; 95 % CI 0.27–1.02).31 An updated analysis of the European IBIS-I trial has demonstrated that tamoxifen continues to reduce breast cancer risk at a median of 16 years of follow-up (HR 0.71; 95 % CI 0.60–0.83). The risk of developing breast cancer was similar between years 0–10 (HR 0.72; 95 % CI 0.59–0.88) and after 10 years (HR 0.69; 95 % CI 0.53–0.91).14

lysis of the European IBIS-I trial has demonstrated that tamoxifen continues to reduce breast cancer risk at a median of 16 years of follow-up (HR 0.71; 95 % CI 0.60–0.83). The risk of developing breast cancer was similar between years 0–10 (HR 0.72; 95 % CI 0.59–0.88) and after 10 years (HR 0.69; 95 % CI 0.53–0.91).14 Tamoxifen is associated with an increased risk of endometrial cancer (RR 2.53; 95 % CI 1.35–4.97; absolute annual risk per 1000: placebo 0.91 vs. tamoxifen 2.30), venous thromboembolic events, including stroke (RR 1.59; 95 % CI 0.93–2.77; absolute annual risk per 1000: placebo 0.92 vs. tamoxifen 1.45), pulmonary embolus (RR 3.01; 95 % CI 1.15–9.27; absolute annual risk per 1000: placebo 0.23 vs. tamoxifen 0.69), deep vein thrombosis (RR 1.60; 95 % CI 0.91–2.86; absolute annual risk per 1000: placebo 0.84 vs. tamoxifen 1.34), cataract development (RR 1.14; 95 % CI 1.01–1.29; absolute annual risk per 1000: placebo 21.72 vs. tamoxifen 24.82), and the need for cataract surgeries (RR 1.57; 95 % CI 1.16–2.14; absolute annual risk per 1000: placebo 3.00 vs. tamoxifen 4.72).30 These risks were not significantly different in the 2010 analysis. The serious risks were not significantly increased in women under the age of 50 years, thus identifying a population of women who obtain significant risk reduction benefits without incurring serious harm. Common side effects reported included bothersome hot flashes and vaginal discharge.

different in the 2010 analysis. The serious risks were not significantly increased in women under the age of 50 years, thus identifying a population of women who obtain significant risk reduction benefits without incurring serious harm. Common side effects reported included bothersome hot flashes and vaginal discharge. The STAR demonstrated that raloxifene was equivalent to tamoxifen in reducing breast cancer risk for postmenopausal women at increased risk of the disease while on therapy.32 In the 2010 updated analysis, with a median follow-up of 81 months, benefits with tamoxifen were greater, while the risks were lower with raloxifene. Raloxifene retained 76 % of the effectiveness of tamoxifen in preventing invasive disease. Raloxifene was not associated with an increased risk of uterine cancer risk and has a slightly lower risk of venous thromboembolic events than tamoxifen.33 Raloxifene is associated with hot flashes, night sweats, vaginal dryness, and weight gain.

ed 76 % of the effectiveness of tamoxifen in preventing invasive disease. Raloxifene was not associated with an increased risk of uterine cancer risk and has a slightly lower risk of venous thromboembolic events than tamoxifen.33 Raloxifene is associated with hot flashes, night sweats, vaginal dryness, and weight gain. Aromatase Inhibitors In the National Cancer Institute of Canada (NCIC) Mammary Prevention 3 (MAP.3) trial, after 35 months of follow-up, exemestane reduced breast cancer risk by 65 % (hazard ratio [HR] 0.35; 95 % CI 0.18–0.70) in high-risk postmenopausal women.36 A 53 % reduction in breast cancer risk was seen with anastrozole in the European IBIS-II trial in women at increased risk of breast cancer (HR 0.47; 95 % CI 0.32–0.68).37 Data on AIs in women with AH or LCIS are limited. In this subgroup, anastrozole reduced breast cancer risk by 69 % (HR 0.31; 95 % CI 0.12–0.84),37 whereas exemestane produced a nonsignificant reduction in the risk of breast cancer by 64 % (HR 0.36; 95 % CI 0.11–1.12).36 It is important to note that these analyses are in a very small number of women, limiting the ability to assess the effectiveness of therapies in women with LCIS or AH.

0.31; 95 % CI 0.12–0.84),37 whereas exemestane produced a nonsignificant reduction in the risk of breast cancer by 64 % (HR 0.36; 95 % CI 0.11–1.12).36 It is important to note that these analyses are in a very small number of women, limiting the ability to assess the effectiveness of therapies in women with LCIS or AH. Neither exemestane nor anastrozole were associated with an increased risk of thromboembolic or cardiovascular events, or other cancers. In the MAP.3 trial, although short-term use of exemestane was shown to worsen age-related bone loss in spite of calcium and vitamin D supplementation, long-term follow-up will be needed to assess the effect on fracture risk in a prevention population.39 The side effects of exemestane, including vasomotor, sexual, and musculoskeletal symptoms, had limited impact on quality of life.40 In addition to vasomotor symptoms, musculoskeletal events (arthralgias, joint stiffness, carpal tunnel syndrome) were more common in the anastrozole arm.37

sk in a prevention population.39 The side effects of exemestane, including vasomotor, sexual, and musculoskeletal symptoms, had limited impact on quality of life.40 In addition to vasomotor symptoms, musculoskeletal events (arthralgias, joint stiffness, carpal tunnel syndrome) were more common in the anastrozole arm.37 Although it is important that high-risk women be considered for chemoprevention, several barriers have been identified that impact uptake, compliance, and adherence. These include fear of possible side effects of the anti-estrogen therapies, specifically thromboembolic events and an endometrial cancer risk, which may be perceived as outweighing the potential benefits of the pharmacologic therapy on reducing the incidence of breast cancer.41–44 Furthermore, it is becoming increasingly evident that physicians are encountering barriers to prescribing pharmacologic therapies, including lack of time to effectively counsel patients about available options, knowledge gaps about the risks and benefits of the medications, and challenges with identifying eligible women with a favorable risk-to-benefit ratio who will benefit from the pharmacologic therapy to reduce breast cancer risk.45,46

rapies, including lack of time to effectively counsel patients about available options, knowledge gaps about the risks and benefits of the medications, and challenges with identifying eligible women with a favorable risk-to-benefit ratio who will benefit from the pharmacologic therapy to reduce breast cancer risk.45,46 Conclusions Physicians are strongly encouraged to assess breast cancer risk and appropriately identify high-risk women with a positive risk–benefit ratio eligible for chemoprevention. Communication of the risks and benefits of SERMs and AIs as preventive therapies and shared decision-making approaches are critical to patient uptake and adherence. More widespread utilization of these agents can reduce the incidence of estrogen receptor (ER)-positive breast cancer but will have no impact on ER-negative breast cancer. Future opportunities for breast cancer risk reduction should target hormone-negative, especially triple-negative, breast cancer. Disclosure Sandhya Pruthi, Ruth E. Heisey, and Therese B. Bevers declare they have nothing to disclose.

A gap in quality of healthcare exists whenever variability of care coexists with evidence that high performance is achievable.1,2 Multiple, recent reports have documented significant variability of care for oncologic reoperation after initial lumpectomy for breast cancer.3–6 Rates of reoperation vary from less than 10 % to more than 50 %. This variability is not accounted for by patient or disease characteristics. Therefore, the American Society of Breast Surgeons (ASBrS) convened a multidisciplinary consensus conference entitled a “Collaborative Attempt to Lower Lumpectomy Reoperation rates” (CALLER). The CALLER conference mission statement was defined as: “Reduce the national reoperation rate in patients undergoing breast-conserving surgery for cancer, without increasing mastectomy rates or adversely affecting cosmetic outcome, thereby improving value of care.” The purpose of the consensus conference was to develop a practical toolbox of recommendations to help providers reduce lumpectomy reoperations to the best achievable level based on available evidence and expert opinion. The target goal is not zero, and to attempt this would be expected to impact cosmetic outcome and lower the breast-conserving therapy rate. The group identified and considered concurrent efforts to reduce reoperation variability, including the meta-analysis that resulted in the SSO-ASTRO margin statement and an updated systematic review of the literature performed by the American College of Surgeons for their new “Operative Standards for Cancer” manual.7–9

The group identified and considered concurrent efforts to reduce reoperation variability, including the meta-analysis that resulted in the SSO-ASTRO margin statement and an updated systematic review of the literature performed by the American College of Surgeons for their new “Operative Standards for Cancer” manual.7–9 Methods Consensus conference participants included experts in breast cancer care from multiple disciplines (surgery, radiology, pathology, plastic surgery, and radiation and medical oncology). A statistician and a patient representative with patient advocacy experience were included. Participants with expertise in quality measurement, patient-reported outcomes, guideline development, and clinical trials were present. There was diversity across breast surgeon practice type, including community and academic surgeons. Toolbox development followed to the extent possible the standards of the Institute of Medicine for guideline development.10 Multiple recent systematic literature reviews were referenced by participants.7,9,11–21 Before the conference, all participants were provided with key topics, references, speaker presentations, and potential “tools” for the toolbox. After topic presentation, an interactive discussion occurred followed by voting. Conference participants and the ASBrS Board of Directors approved toolbox recommendations.

–21 Before the conference, all participants were provided with key topics, references, speaker presentations, and potential “tools” for the toolbox. After topic presentation, an interactive discussion occurred followed by voting. Conference participants and the ASBrS Board of Directors approved toolbox recommendations. Results The proposed conference tools, references, level of evidence, consensus, and strength of recommendation are described in Tables 1 and 2. Recognizing the impact of reoperations on patient care, cost, and outcomes, the conference participants had uniform agreement to set a 5-year target goal for a national average reoperation rate in the year 2020. However, there was lack of uniformity for the actual target number. Two-thirds (10/15) of participants recommended a goal of less than 20 %.Table 1 CALLER Toolbox to reduce reoperation and improve cosmetic outcomes Tool % CALLER participants recommending Level of evidence/consensus Strength of recommendation References SSO-ASTROa guideline 94 % High 2A nonuniform Strong-moderate 7,8,14 Minimally invasive breast biopsy 94 % High 1 nonuniform Strong 12,15,49,50 Complete diagnostic mammography and US as needed 94 % Lower 2B nonuniform Strong-moderate 11,16,51–54 Oncoplastic lumpectomy 100 % Lower 2A uniform Strong-moderate 17,43–48,55,56 Lesion localization 94 % Lower 2A nonuniform Strong 9,18–20,49,50,53,54,57–86 Specimen orientation 95 % Lower 2A nonuniform Strong 49,50,87,88 Cavity shaves 75 % Lower 2A nonuniform Strong-moderate 25,89–97 Specimen imaging and surgeon review 100 % Lower 2A uniform Strong 50,98–106

Oncoplastic lumpectomy 100 % Lower 2A uniform Strong-moderate 17,43–48,55,56 Lesion localization 94 % Lower 2A nonuniform Strong 9,18–20,49,50,53,54,57–86 Specimen orientation 95 % Lower 2A nonuniform Strong 49,50,87,88 Cavity shaves 75 % Lower 2A nonuniform Strong-moderate 25,89–97 Specimen imaging and surgeon review 100 % Lower 2A uniform Strong 50,98–106 Intraoperative pathology 89 % Lower 2A–2B nonuniform Strong-moderate 13,21,27,107–124 Preoperative multidisciplinary planning 100 % Lower 2A uniform Strong-moderate 49,50,125,126 Patient-reported outcome measurement 57 % Lower 2B nonuniform Moderate-weak 127–135 aSSO-ASTRO guideline only applicable for invasive cancer Table 2 Level of evidence/consensus and strength of recommendation categories Strength of recommendation Level of evidence/consensus 1. Strong 2. Strong-moderate 3. Moderate 4. Moderate to weak 5. Weak 6. Insufficient evidence 1. (1) High-level evidence; uniform CALLER consensus that intervention is appropriate 2. (2A) Lower-level evidence; uniform CALLER consensus that intervention is appropriate 3. (2B) Lower-level evidence; CALLER majority consensus that intervention is appropriate 4. (3) Based on any level evidence; major CALLER disagreement that intervention is appropriate Level of evidence and consensus scale is adapted from NCCN guidelines

vidence; uniform CALLER consensus that intervention is appropriate 3. (2B) Lower-level evidence; CALLER majority consensus that intervention is appropriate 4. (3) Based on any level evidence; major CALLER disagreement that intervention is appropriate Level of evidence and consensus scale is adapted from NCCN guidelines Tool 1: Preoperative Diagnostic Imaging Should Include Full-Field Digital Mammography and Supplementary Imaging to Include Ultrasound as Needed All participants agreed that high-quality, meticulous, preoperative, diagnostic mammography was necessary preoperatively. “Selective” use of ipsilateral ultrasound (US) was recommended. US may be of less benefit when screening mammography identifies calcifications without mass. Despite near routine actual use of US by conference participants, they concluded that the level of evidence did not support a recommendation for “routine” US. Breast tomography was discussed and judged to have future applications but was not yet included in the toolbox due to insufficient evidence. Routine use of MRI was not recommended based on meta-analyses that show its use does not affect the rate of reexcision or local recurrence. Selective use of MRI is described in position statements from other groups.22–24

d to have future applications but was not yet included in the toolbox due to insufficient evidence. Routine use of MRI was not recommended based on meta-analyses that show its use does not affect the rate of reexcision or local recurrence. Selective use of MRI is described in position statements from other groups.22–24 Tool 2: Minimally Invasive Breast Biopsy (MIBB) for Breast Cancer Diagnosis Some studies demonstrate lower reoperation rates when a diagnosis of malignancy is known before surgical excision. MIBB provides opportunity for preoperative treatment planning to include genetic risk assessment, medical oncology, and plastic surgery consultation and axillary evaluation. Tool 3: Multidisciplinary Discussions to Include Radiology, Pathology, Surgery, and Radiation and Medical Oncology Optimizing reoperation rates requires preoperative collaboration between radiologists, surgeons, and pathologists. In patients considered for neoadjuvant therapy, medical oncology consultation also is necessary. Preoperative knowledge of number of lesions, geometry, distance to skin and chest wall, and possible extension towards the nipple may all facilitate negative margins. Information technology that enhances communication and provides intraoperative archived images can aid lesion review and communication. Postoperative discussion with all specialties aids decision making regarding reoperation.

and chest wall, and possible extension towards the nipple may all facilitate negative margins. Information technology that enhances communication and provides intraoperative archived images can aid lesion review and communication. Postoperative discussion with all specialties aids decision making regarding reoperation. Tool 4: For Nonpalpable Breast Lesions, the Use of Radioactive Seeds, Intraoperative US, or Wire Localization to Direct Lesion Excision is Recommended A localization method should be used for resection of all nonpalpable cancers. Although some studies have indicated superiority of one technique compared with another, the conference concluded that evidence to recommend a single technique was not definitive. Surgeon use of US also can be used to aid targeting and decide volume of resection in both palpable and nonpalpable lesions. Placement of multiple localizing wires or seeds (bracketing) may be useful for larger lesions, multifocal tumors, or extensive ductal carcinoma in situ (DCIS).

ingle technique was not definitive. Surgeon use of US also can be used to aid targeting and decide volume of resection in both palpable and nonpalpable lesions. Placement of multiple localizing wires or seeds (bracketing) may be useful for larger lesions, multifocal tumors, or extensive ductal carcinoma in situ (DCIS). Tool 5: Oncoplastic Techniques can Reduce the Need for Reoperation in Anatomically Suitable Patients Oncoplastic techniques have the potential to decrease positive margins at initial lumpectomy by allowing resection of a larger volume of tissue. They also may improve ipsilateral breast appearance and contralateral breast symmetry. There was uniform agreement for their potential benefit. The conference recommends applying these techniques only in a selective group of patients. Small primary cancers can be excised with acceptable cosmetic results without oncoplastic techniques. For all procedures, marker clips or other marking modality should be considered for application to cavity side walls to aid radiation planning.

ecommends applying these techniques only in a selective group of patients. Small primary cancers can be excised with acceptable cosmetic results without oncoplastic techniques. For all procedures, marker clips or other marking modality should be considered for application to cavity side walls to aid radiation planning. Tool 6: Specimen Orientation of 3 or More Margins When the breast cancer is excised, markers or ink should be placed on the specimen for orientation to ensure which margin edge(s) is/are positive to guide focused reexcision of the correct tissue, if necessary. There are limited data linking orientation directly to reoperation rates, but the conference concluded the benefit/burden ratio of orientation was high. All excisions should be oriented. Orientation is associated with better cosmetic outcomes by avoiding “entire cavity” reexcision in patients with nonoriented positive margins. The consensus was that orientation of at least three sides was superior to two sides. Some participants favored intraoperative six-sided inking as best practice, but there was no consensus on orientation methodology beyond labeling at least three margins.

avity” reexcision in patients with nonoriented positive margins. The consensus was that orientation of at least three sides was superior to two sides. Some participants favored intraoperative six-sided inking as best practice, but there was no consensus on orientation methodology beyond labeling at least three margins. Tool 7: Specimen Radiograph with Surgeon Intraoperative Review The primary role of specimen imaging is to document removal of the targeted nonpalpable lesion before the patient leaves the operating room. Lower-level evidence supports specimen radiography as a method to assess distance of lesion to margin and therefore direct and potentially reduce reoperation. Specimens should not undergo compression during imaging, because it may cause specimen fracture that allows ink to enter the crevasse and a false-positive margin. Some participants supplement specimen radiography with US. Surgeons should review the specimen imaging before the operation has been completed, ideally with surgeon-radiology communication. Real-time review may avoid a complete “miss” of the lesion or direct the surgeon to perform an additional cavity shave for a “close” margin. Specimen imaging may not be universally available. If not, the conference strongly encourages systems to develop necessary resources for specimen imaging with immediate image review. Two views at orthogonal angles may identify close or positive margins not seen on a single view. Intraoperative imaging with other modalities to include tomograms, MRI, CT, and other imaging are being investigated.

ongly encourages systems to develop necessary resources for specimen imaging with immediate image review. Two views at orthogonal angles may identify close or positive margins not seen on a single view. Intraoperative imaging with other modalities to include tomograms, MRI, CT, and other imaging are being investigated. Tool 8: Consider Cavity Shave Margins in Patients with T2 or Greater Tumor Size or TI with Extensive Intraductal Carcinoma (EIC) There are moderate levels of evidence that cavity side wall excisions correlate with lower reoperation rate. Shave size should provide adequate sampling of the residual wall. “Tiny shaves” representing only a small portion of a “wall” were discouraged. If shaves are performed, the “final” edge should be marked; i.e., nonoriented shave with even a small amount of tumor on the surface would constitute a final ink positive margin status requiring reexcision. Some surgeons routinely perform shaves of all cavity side walls regardless of tumor type or size. Others perform selective shaves directed by palpation, imaging, or pathologic specimen examination. There has been one recently published, randomized, controlled trial of cavity shave versus no-shave margins, which demonstrated a statistically significant decrease in the reoperation rate for patients undergoing breast conservation surgery.25

e shaves directed by palpation, imaging, or pathologic specimen examination. There has been one recently published, randomized, controlled trial of cavity shave versus no-shave margins, which demonstrated a statistically significant decrease in the reoperation rate for patients undergoing breast conservation surgery.25 Tool 9: Intraoperative Pathology Assessment of Lumpectomy Margins may Help Decrease Reexcision When Feasible A systematic literature review demonstrates that intraoperative margin assessment with frozen histologic section or imprint cytology are associated with lower reoperation rates by allowing intraoperative reexcision of positive margins.13 There is lower-level evidence to support only gross specimen examination. Resources and expertise may limit the feasibility of routine intraoperative pathology assessment. Several institutions report low reoperation rates without intraoperative margin assessment.

llowing intraoperative reexcision of positive margins.13 There is lower-level evidence to support only gross specimen examination. Resources and expertise may limit the feasibility of routine intraoperative pathology assessment. Several institutions report low reoperation rates without intraoperative margin assessment. Tool 10: Compliance with the SSO-ASTRO Margin Guideline to Not Routinely Reoperate for Close Margins with no Tumor on Ink in Patients with Invasive Cancer Compliance with this guideline has the potential to reduce reoperations by 40 %.6 The remaining tools are targeted towards reducing ink positive margins at the initial lumpectomy. By meta-analysis, recurrence risk doubles when ink positive margins are not excised. Recurrence is not improved by reoperation if the margin is negative. If ink positive margins occur, the need for reoperation should be evaluated by the treating team in collaboration with the patient (“shared decision making”), providing patients with recurrence risks in absolute percentages for the choices of reoperation or not. As a consequence, some patients may choose not to have reoperation. The margin guideline is applicable to subsets of patients with “bad tumor biology” (triple negative, Her 2 positive, high grade), young age, lobular cancer, EIC, or not receiving systemic treatment. There is no proven benefit for reoperation in these patients if they have ink negative margins. Some patients with negative margins may still be considered for reoperation, if clinical and/or imaging findings suggest residual persistent adjacent disease. The margin meta-analysis did not include patients with neoadjuvant therapy or pure DCIS. Given the lack of consensus regarding acceptable margin width for DCIS, decisions regarding reoperation in these patients optimally involves multidisciplinary input and shared decision making with the patient. Until new evidence is available for DCIS, the conference supports NCCN guidelines for reoperation if the margin is ink positive or <1 mm.26

egarding acceptable margin width for DCIS, decisions regarding reoperation in these patients optimally involves multidisciplinary input and shared decision making with the patient. Until new evidence is available for DCIS, the conference supports NCCN guidelines for reoperation if the margin is ink positive or <1 mm.26 Tool 11: Routine Breast-Specific Patient Reported Outcome (PRO) Measurement may Help to Assess Cosmetic Outcomes When Feasible There is limited reporting in the literature of cosmetic and functional outcomes from the patient perspective. Validated PRO tools, such as BREAST-Q©, should be more widely adopted and may aid improvement. New tools need to be developed that decrease the burdens for both providers and patients for reporting.

n Feasible There is limited reporting in the literature of cosmetic and functional outcomes from the patient perspective. Validated PRO tools, such as BREAST-Q©, should be more widely adopted and may aid improvement. New tools need to be developed that decrease the burdens for both providers and patients for reporting. Discussion The goal of the consensus conference was to provide practitioners with a variety of tools that can be adapted to help lower rates of reoperation following lumpectomy. While these recommendations are not meant to serve as guidelines or standard of care, conference leaders complied with most principles for guideline development as defined by the IOM.10 Updated systematic reviews were referenced and the group included multiple disciplines and stakeholders.7,9,11–21 The group did not provide a period for public comment, request for other society endorsement, or commission new systematic literature reviews. For expediency, recommendations were provided that could be implemented into clinical practice quickly. “Standard of care” is a legal term, and our toolbox does not establish a new legal “standard of care.” It also is important to recognize that performing reoperation does not mean poor quality care. Particularly, omission of reoperation for positive margins is not recommended. Reoperation of a positive margin is good quality care and results in lower risk of cancer recurrence. All tools in the toolbox earned endorsement by a majority vote. It does not follow that all tools are recommended for every patient.

care. Particularly, omission of reoperation for positive margins is not recommended. Reoperation of a positive margin is good quality care and results in lower risk of cancer recurrence. All tools in the toolbox earned endorsement by a majority vote. It does not follow that all tools are recommended for every patient. At least three factors should be considered for selection. The first is resource availability. For example, one tool is the use of intraoperative frozen section (FS) for margin assessment, a tool associated with very low rates of reoperation.27 This service may not be available in all settings, and there should be no inference of “poor quality” for lack of access to it. In contrast, multidisciplinary preoperative planning—in person or virtual—can be implemented widely. The second consideration for tool selection is baseline reoperation rate. The average reoperation rate in four national databases ranges from 20 to 24 %.3–6 For surgeons and institutions with average or higher rates, a trial of previously unused or underutilized tools should be considered, followed by tracking of outcomes. For those with rates already in the best tiers of performance, there can be attempts to improve even further by testing different or additional tools, but performance tracking will still be necessary.

gher rates, a trial of previously unused or underutilized tools should be considered, followed by tracking of outcomes. For those with rates already in the best tiers of performance, there can be attempts to improve even further by testing different or additional tools, but performance tracking will still be necessary. The last consideration for number of tools is “redundancy.” For example, if circumferential lumpectomy FS is used and negative, then the benefit of additional shaving of cavity side walls is low. If complete cavity side wall shavings are performed, then the benefit of lumpectomy margin FS is low too. Some participants recommended using more tools when operating on patients with known factors associated with positive margins, such as larger size, invasive lobular type, low-grade noncalcified DCIS, and EIC status. All tools in the toolbox can be applied for patients with DCIS and invasive cancer except the SSO-ASTRO margin statement, which was specific for invasive cancer and did not include patients with pure DCIS. Intraoperative devices to assess margin status were discussed as potential tools to decrease reoperation. A recent, randomized trial concluded that the MarginProbe™ device was associated with fewer reoperations.28 The conference majority vote was to omit these devices from the toolbox until further investigation.28–36

The last consideration for number of tools is “redundancy.” For example, if circumferential lumpectomy FS is used and negative, then the benefit of additional shaving of cavity side walls is low. If complete cavity side wall shavings are performed, then the benefit of lumpectomy margin FS is low too. Some participants recommended using more tools when operating on patients with known factors associated with positive margins, such as larger size, invasive lobular type, low-grade noncalcified DCIS, and EIC status. All tools in the toolbox can be applied for patients with DCIS and invasive cancer except the SSO-ASTRO margin statement, which was specific for invasive cancer and did not include patients with pure DCIS. Intraoperative devices to assess margin status were discussed as potential tools to decrease reoperation. A recent, randomized trial concluded that the MarginProbe™ device was associated with fewer reoperations.28 The conference majority vote was to omit these devices from the toolbox until further investigation.28–36 Measurement of both individual surgeon and institutional outcomes are essential prerequisites during attempts to reduce reoperation after initial lumpectomy. Measurement assesses the impact of these initiatives. If resources are available, a comprehensive audit that tracks intended and unintended outcomes is recommended (Table 3). If resources are limited, then minimal tracking would include reoperation, positive margin, and breast-conserving therapy (BCT) rates. Reoperation rates and BCT rates can be reported in the ASBrS Mastery database, the National Consortium of Breast Centers Quality Measurement Program, and “in-house” registries.37,38 All breast cancer quality-measurement programs were recently summarized.39Table 3 Performance tracking options during initiatives to reduce reoperation and improve cosmetic outcome

orted in the ASBrS Mastery database, the National Consortium of Breast Centers Quality Measurement Program, and “in-house” registries.37,38 All breast cancer quality-measurement programs were recently summarized.39Table 3 Performance tracking options during initiatives to reduce reoperation and improve cosmetic outcome 1. Core needle biopsy rate for cancer diagnosis* 2. Specimen imaging rate* 3. Specimen orientation rate* 4. Rate of ink positive margins at initial lumpectomy 5. Compliance rate with SSO-ASTRO margin statement 6. Reoperation rate after initial lumpectomy for breast cancer* 7. Breast conserving therapy rate 8. Cost/charges per episode of care 9. Patient reported outcomes to include cosmetic outcome after lumpectomy* 10. Ipsilateral breast tumor recurrence rate * ASBrS endorsed Quality Measure audited in Mastery Program37 Increased mastectomy rates and poor cosmetic outcomes are potential unintended adverse outcomes of efforts to lower reoperation rates and therefore should be monitored.40–42 These risks were recognized but were felt to be balanced by the potential to improve overall patient care by following conference recommendations. There is evidence that both reoperation rate and cosmetic outcome can improve by adoption of oncoplastic techniques.43–48

ation rates and therefore should be monitored.40–42 These risks were recognized but were felt to be balanced by the potential to improve overall patient care by following conference recommendations. There is evidence that both reoperation rate and cosmetic outcome can improve by adoption of oncoplastic techniques.43–48 The conference process and work product is not without limitations. We did not follow strict guideline development standards and did not use a formal Delphi process in arriving at consensus. Furthermore, most of the tools are not based on high-level evidence. The strength of the conference is its recognition that unacceptable variability occurs in the care of patients undergoing lumpectomy. As a consequence, multiple stakeholders accepted ownership and then developed recommendations to improve care, cost, and outcomes by using “best available” evidence and expert opinion.

. The strength of the conference is its recognition that unacceptable variability occurs in the care of patients undergoing lumpectomy. As a consequence, multiple stakeholders accepted ownership and then developed recommendations to improve care, cost, and outcomes by using “best available” evidence and expert opinion. Conclusions Recognition of the gap between actual and achievable care led to development of a toolbox of recommendations to reduce the proven variability of reoperation and the suspected variability of cosmetic outcome after initial lumpectomy for breast cancer. A list of other ASBrS initiatives to reduce reoperation and improve cosmesis is described in Table 4. Tracking of outcomes is recommended for all initiatives. Next steps include: (1) dissemination and implementation strategies; (2) comparative effectiveness research to determine which tools or collection of tools are most strongly associated with reoperation rates, cosmetic outcome, and value; and (3) collaboration with industry, payer, and government stakeholders to provide better support for performance reporting that is funded, incentivized, and less burdensome for providers.Table 4 American Society of Breast Surgeon efforts to reduce variability of reoperation rates after initial lumpectomy for cancer 1. Orlando Consensus Conference April 30, 2015 2. Auditing and peer performance comparison of re-excision rates and reasons for re-operation available in the ASBrS Mastery Program6,37

Conclusions Recognition of the gap between actual and achievable care led to development of a toolbox of recommendations to reduce the proven variability of reoperation and the suspected variability of cosmetic outcome after initial lumpectomy for breast cancer. A list of other ASBrS initiatives to reduce reoperation and improve cosmesis is described in Table 4. Tracking of outcomes is recommended for all initiatives. Next steps include: (1) dissemination and implementation strategies; (2) comparative effectiveness research to determine which tools or collection of tools are most strongly associated with reoperation rates, cosmetic outcome, and value; and (3) collaboration with industry, payer, and government stakeholders to provide better support for performance reporting that is funded, incentivized, and less burdensome for providers.Table 4 American Society of Breast Surgeon efforts to reduce variability of reoperation rates after initial lumpectomy for cancer 1. Orlando Consensus Conference April 30, 2015 2. Auditing and peer performance comparison of re-excision rates and reasons for re-operation available in the ASBrS Mastery Program6,37 3. Development of formal specifications for a reexcision lumpectomy rate quality measure in 2014 4. Development of a patient reported cosmetic outcome measure in the Mastery patient survey 5. Development of a guideline for the technique of “breast-conserving surgery” available on the ASBrS website 6. Education emphasizing compliance with the SSO-ASTRO margin statement during the 2014 and 2015 annual meetings 7. Quality and Research committees of the ASBrS to begin a prospective, observational study of members to search for associations between reoperation rates and the CALLER conference tools in 2015. This effort is intended to aid the design of subsequent comparative effectiveness research

he 2014 and 2015 annual meetings 7. Quality and Research committees of the ASBrS to begin a prospective, observational study of members to search for associations between reoperation rates and the CALLER conference tools in 2015. This effort is intended to aid the design of subsequent comparative effectiveness research The authors thank Sharon Grutman for conference planning; Choua Vang for assistance in manuscript preparation; Sarah Blair for assistance with bibliography; Gundersen Medical Foundation for unrestricted dollars to support the conference; and Dune Medical Devices for unrestricted dollars to support the conference Conflict of Interest Statement Unrestricted funds from Dune Medical Devices but no industry representative at CALLER conference and all participants and authors disclose no financial relationship with Dune Medical Devices.

Papillary thyroid carcinoma (PTC) is the most common well-differentiated thyroid carcinoma and is characterized by follicular differentiation and distinct atypical nuclear features. PTC accounts for approximately 85–90 % of differentiated thyroid carcinomas.1 Furthermore, the incidental detection of papillary thyroid microcarcinoma (PTMC), with a maximum diameter of 10 mm or less, has recently increased as a consequence of the frequent use of ultrasonography in regular health checkup.2–5

features. PTC accounts for approximately 85–90 % of differentiated thyroid carcinomas.1 Furthermore, the incidental detection of papillary thyroid microcarcinoma (PTMC), with a maximum diameter of 10 mm or less, has recently increased as a consequence of the frequent use of ultrasonography in regular health checkup.2–5 The presence of minimal extrathyroid extension (ETE) in PTCs is defined as when tumor cells extend to the sternothyroid muscle or perithyroid soft tissue.6 Most cases of PTCs with minimal ETE exhibit an extension to the perithyroid soft tissue rather than to the sternothyroid muscle. The presence of a minimal ETE is classified as the T3 category in PTCs, irrespective of tumor size, by the American Joint Commission on Cancer (AJCC) cancer staging system (7th edition).6 Because clinical recommendations for proper treatment can vary according to the TNM stage, accurate T categorization is very important.7–9 However, histological confirmation of the presence of minimal ETE in PTCs can be both controversial and subjective among endocrine pathologists because of the absence of a well-defined true capsule in the thyroid gland.10–12 The thyroid capsule is usually composed of inconspicuous thin fibrous tissues with a variable amount of adipose tissue, blood vessels, and skeletal muscle. It is deficient in the anterior midline of the isthmus, and skeletal muscle can even be mixed with thyroid follicles within the thyroid parenchyma of this area.10–12

he thyroid capsule is usually composed of inconspicuous thin fibrous tissues with a variable amount of adipose tissue, blood vessels, and skeletal muscle. It is deficient in the anterior midline of the isthmus, and skeletal muscle can even be mixed with thyroid follicles within the thyroid parenchyma of this area.10–12 We classified the extent of minimal ETE and investigated the clinicopathological significance of the presence of minimal ETE to identify any effect on recurrence-free survival (RFS) in patients with solitary PTCs and PTMCs.

he thyroid capsule is usually composed of inconspicuous thin fibrous tissues with a variable amount of adipose tissue, blood vessels, and skeletal muscle. It is deficient in the anterior midline of the isthmus, and skeletal muscle can even be mixed with thyroid follicles within the thyroid parenchyma of this area.10–12 We classified the extent of minimal ETE and investigated the clinicopathological significance of the presence of minimal ETE to identify any effect on recurrence-free survival (RFS) in patients with solitary PTCs and PTMCs. Materials and Methods Patients We retrospectively reviewed 546 patients (484 female, 62 male) with a solitary PTC without evidence of extensive ETE (T4 category). Distant metastases were initially detected according to the current AJCC cancer staging system in patients who underwent total thyroidectomy or hemithyroidectomy, irrespective of cervical lymph node (LN) dissection, at Asan Medical Center, Seoul, Korea, from 1998 to 2003. Patient ages ranged from 8 to 87 years (average 44 years). LN dissection was performed when a LN metastasis was clinically suspicious by ultrasonography (US) and computed tomography (CT) or was histologically confirmed by previous fine needle aspiration biopsy prior to surgery. We reviewed the patient clinical and pathological parameters, including age, gender, procedures, mean tumor diameter, extent of tumor extension, cervical LN metastasis status, and the presence of tumor recurrence (Table 1). This study was approved by the institutional review board of Asan Medical Center.Table 1 Demographic and clinicopathologic characteristics of 546 patients with solitary papillary thyroid carcinoma

n tumor diameter, extent of tumor extension, cervical LN metastasis status, and the presence of tumor recurrence (Table 1). This study was approved by the institutional review board of Asan Medical Center.Table 1 Demographic and clinicopathologic characteristics of 546 patients with solitary papillary thyroid carcinoma Characteristics Number (%) Age (year) Mean at diagnosis (range) 44 (8–87) <45 275 (50.4) ≥45 271 (49.6) Sex Female 484 (88.6) Male 62 (11.4) Procedures Lobectomy/less than total resection 95 (17.4) Total thyroidectomy 451 (82.6) No lymph node dissection 5 (0.9) Central lymph node dissection 466 (85.3) Modified radical lymph node dissection 75 (13.7) Tumor size (mm) Mean (range) 20 (1–100) ≤10 144 (26.4) >10 402 (73.6) Extension Confinement to thyroid parenchyma 196 (35.9) Minimal extrathyroid extension 350 (64.1) Perithyroid soft tissue 259 (47.4) Sternothyroid muscle 91 (16.7) Cervical lymph node metastasis Present 334 (61.2) Absent 212 (38.8) Recurrence Present 61 (11.2) Median duration (mo, interquartile range) 39 (16–72) Site Neck level I lymph node 1 (1.6) Neck level II, III, IV lymph nodes 50 (81.9) Neck level V lymph node 11 (18.0) Neck level VI lymph node 3 (4.9) Previous resection sites 12 (19.7) Lung (distant metastasis) 4 (6.6) Inguinal lymph node (distant metastasis) 1 (1.9)

Median duration (mo, interquartile range) 39 (16–72) Site Neck level I lymph node 1 (1.6) Neck level II, III, IV lymph nodes 50 (81.9) Neck level V lymph node 11 (18.0) Neck level VI lymph node 3 (4.9) Previous resection sites 12 (19.7) Lung (distant metastasis) 4 (6.6) Inguinal lymph node (distant metastasis) 1 (1.9) Pathological Evaluation All slides from 546 patients were reviewed by two pathologists, including one experienced endocrine pathologist (D.E.S.). We subclassified the extent of tumor extension into the following three categories: confinement to the thyroid parenchyma (E0), extension to the perithyroid soft tissue (E1), and extension to the sternothyroid muscle (E2). Among 546 patients, 196 (35.9 %) were classified as E0, 259 (47.4 %) were classified as E1, and 91 (16.7 %) were classified as E2 (Table 1). Among 546 patients, 334 patients revealed cervical lymph node metastasis at the first operation. We analyzed the extent of the lymph node metastasis at the first operation, which included maximum metastatic tumor size, N stage, and presence of extranodal extension (Supplementary Table S1).

e classified as E2 (Table 1). Among 546 patients, 334 patients revealed cervical lymph node metastasis at the first operation. We analyzed the extent of the lymph node metastasis at the first operation, which included maximum metastatic tumor size, N stage, and presence of extranodal extension (Supplementary Table S1). Follow-up All patients with PTCs were regularly followed up every 6–12 months with physical examinations, serum thyroglobulin and anti-thyroglobulin antibody measurements, and US neck examinations. A diagnostic whole body scan was performed after total thyroidectomy and remnant RAI ablation, as previously described.13 Following hemithyroidectomy, a CT scan and/or 18F-FDG PET were used to detect recurrence or distant metastasis if clinically suspected. The median follow-up period for patients was 113 (range 1–168) months. Recurrence was defined as structural disease recurrence, such as the reappearance of a pathologically confirmed malignant tissue and/or the appearance of a metastatic lesion in other organs by imaging studies during follow-up. Statistical Analyses The Chi squared test and Fisher’s exact test for univariate analysis were used along with binary logistic regression and Cox–Hazard regression models for multivariate analysis. To estimate survival rates, survival curves were generated using the Kaplan–Meier method and log-rank test, respectively. We used SPSS software version 18.0 and R statistical software for these analyses. Any p value <0.05 was considered to indicate a statistically significant difference.

dels for multivariate analysis. To estimate survival rates, survival curves were generated using the Kaplan–Meier method and log-rank test, respectively. We used SPSS software version 18.0 and R statistical software for these analyses. Any p value <0.05 was considered to indicate a statistically significant difference. Results Recurrence-Free Survival Outcomes Associated with the Subclassification of an Extrathyroid Extension Recurrence (median duration, 39 months; interquartile range, 16–72 months) occurred in 61 (11.2 %) of our patients. Recurrence was identified in diverse sites, including one or more sites and occurred in 3 central cervical LNs (level VI), 62 lateral cervical LNs (level I, II, III, IV, or V), 5 distant metastasis (inguinal LN, 1 case; lung, 4 cases), and 12 previous resection sites (Table 1). Regarding previous resection sites, there were 11 ipsilateral soft tissue (operation bed) after total thyroidectomy, which included 6 clear resection margins and 5 involved margins, and 1 contralateral remnant thyroid parenchyma after lobectomy (Supplementary Table S2).

ases), and 12 previous resection sites (Table 1). Regarding previous resection sites, there were 11 ipsilateral soft tissue (operation bed) after total thyroidectomy, which included 6 clear resection margins and 5 involved margins, and 1 contralateral remnant thyroid parenchyma after lobectomy (Supplementary Table S2). In the primary PTCs of 61 patients with recurrence, 12 (6.1 %) were confined to the thyroid parenchyma (E0), 38 (14.7 %) extended to the perithyroid soft tissue (E1), and 11 (12.1 %) extended to the sternothyroid muscle (E2). There was a more statistically significant difference in the RFS outcomes when two groups (E0 versus E1 + E2) were analyzed than when three groups were analyzed (E0 vs. E1 vs. E2) using the log-rank test with Cox-proportional hazard analysis (Table 2; Supplementary Fig. S1). In another two-group comparison (E0 + E1 vs. E2), the presence of extension to the sternothyroid muscle had no significant effect on RFS. Therefore, we used a two-tier classification scheme of minimal ETE (E0 vs. E1 + E2) based on the AJCC cancer staging system in the subsequent evaluations of the clinicopathological significance of minimal ETE in patients with solitary PTCs.Table 2 Recurrence according to subclassification of invasion in solitary papillary thyroid carcinoma

two-tier classification scheme of minimal ETE (E0 vs. E1 + E2) based on the AJCC cancer staging system in the subsequent evaluations of the clinicopathological significance of minimal ETE in patients with solitary PTCs.Table 2 Recurrence according to subclassification of invasion in solitary papillary thyroid carcinoma Classification Recurrence, no. (%) HR 95 % CI p value I E0 12 (6.1) 1 (ref) 0.015 E1 38 (14.7) 2.444 1.277–4.676 0.007 E2 11 (12.1) 2.076 0.916–4.786 0.080 II (current T stage) E0 12 (6.1) 1 (ref) E1 + E2 49 (14.0) 2.350 1.250–4.419 0.006 III E0 + E1 50 (11.0) 1 (ref) E2 11 (12.1) 1.144 0.595–2.198 0.686 E0 confinment to thyroid parenchyma, E1 extension to perithyroid soft tissue, E2 extension to sternothyroid muscle Clinicopathological Parameters Associated with the Presence of a Minimal ETE In 546 patients with solitary PTCs, there was no significant difference in the presence of minimal ETE (E1 + E2) according to gender (p = 0.416). Older age (odds ratio [OR] 1.688; 95 % confidence interval [CI] 1.154–2.469; p = 0.007), larger tumor size (OR 3.078; 95 % CI 2.040–4.645; p < 0.001), and the presence of a LN metastasis (OR 2.071; 95 % CI 1.407–3.048; p < 0.001) were significantly correlated with the presence of a minimal ETE (Supplementary Table S3).

Older age (odds ratio [OR] 1.688; 95 % confidence interval [CI] 1.154–2.469; p = 0.007), larger tumor size (OR 3.078; 95 % CI 2.040–4.645; p < 0.001), and the presence of a LN metastasis (OR 2.071; 95 % CI 1.407–3.048; p < 0.001) were significantly correlated with the presence of a minimal ETE (Supplementary Table S3). Clinicopathological Parameters that Affect Recurrence-Free Survival In 546 patients with solitary PTCs, females exhibited a lower rate of recurrence (hazard ratio [HR] 0.536; 95 % CI 0.288–0.997; p = 0.049). A larger tumor diameter was significantly correlated with a higher recurrence rate (p = 0.019) in univariate analysis but not in multivariate analysis (HR 1.616; 95 % CI 0.757–3.451; p = 0.215). There was a significant difference between types of surgery in univariate analysis (p = 0.020) but not in multivariate analysis (p = 0.356). The presence of cervical LN metastasis was significantly associated with recurrence (HR 3.173; 95 % CI 1.536–6.553; p = 0.002). Among the extent of LN metastasis, N stage, presence of extranodal extension, and maximum metastatic tumor size were significantly associated with recurrence in univariate analysis, but N stage had no effect in multivariate analysis (Supplementary Table S4). Additionally, the presence of minimal ETE was significantly correlated with recurrence in univariate analysis (p = 0.008); however, there was no significance in multivariate analysis (HR 1.879; 95 % CI 0.992–3.560; p = 0.053; Table 3). Minimal ETE according to types of surgery had also no effect on recurrence (Supplementary Table S5).Table 3 Clinicopathologic parameters associated with recurrence of solitary papillary thyroid carcinomas

); however, there was no significance in multivariate analysis (HR 1.879; 95 % CI 0.992–3.560; p = 0.053; Table 3). Minimal ETE according to types of surgery had also no effect on recurrence (Supplementary Table S5).Table 3 Clinicopathologic parameters associated with recurrence of solitary papillary thyroid carcinomas Parameters HR Univariate HR Multivariate 95 % CI p value 95 % CI p value Age (year) <45 1 (ref) 1 (ref) ≥45 0.626 0.373–1.050 0.076 0.670 0.397–1.131 0.134 Sex Male 1 (ref) 1 (ref) Female 0.409 0.221–0.754 0.004 0.536 0.288–0.997 0.049 Procedures Lobectomy/less than total resection 1 (ref) 1 (ref) Total thyroidectomy 3.973 1.244–12.685 0.020 1.798 0.517–6.259 0.356 Size (mm) ≤10 1 (ref) 1 (ref) >10 2.348 1.159–5.128 0.019 1.445 0.663–3.151 0.355 Cervical lymph node No metastasis 1 (ref) 1 (ref) Metastasis 3.932 1.938–7.978 <0.001 3.173 1.536–6.553 0.002 Invasion Confinement 1 (ref) 1 (ref) Minimal extrathyroid extension 2.350 1.250–4.419 0.008 1.879 0.992–3.560 0.053

–6.259 0.356 Size (mm) ≤10 1 (ref) 1 (ref) >10 2.348 1.159–5.128 0.019 1.445 0.663–3.151 0.355 Cervical lymph node No metastasis 1 (ref) 1 (ref) Metastasis 3.932 1.938–7.978 <0.001 3.173 1.536–6.553 0.002 Invasion Confinement 1 (ref) 1 (ref) Minimal extrathyroid extension 2.350 1.250–4.419 0.008 1.879 0.992–3.560 0.053 In 144 patients with solitary PTMCs, the presence of minimal ETE was not associated with tumor recurrence in univariate analysis (p = 0.254) and multivariate analysis (HR 1.215; 95 % CI 0.235–6.278; p = 0.816; Table 4). Additionally, there was no significant difference in RFS outcomes between the absence and presence of minimal ETE (p = 0.240; Fig. 1). The presence of cervical LN metastasis was significantly associated with recurrence in univariate analysis (p = 0.034), but there was no significant difference in multivariate analysis (HR 7.353; 95 % CI 0.831–65.051; p = 0.073; Table 4).Table 4 Clinicopathologic parameters associated with recurrence of solitary papillary thyroid microcarcinomas

metastasis was significantly associated with recurrence in univariate analysis (p = 0.034), but there was no significant difference in multivariate analysis (HR 7.353; 95 % CI 0.831–65.051; p = 0.073; Table 4).Table 4 Clinicopathologic parameters associated with recurrence of solitary papillary thyroid microcarcinomas Parameters HR Univariate HR Multivariate 95 % CI p value 95 % CI p value Age (year) <45 1 (ref) 1 (ref) ≥45 0.250 0.051–1.241 0.090 0.324 0.062–1.702 0.183 Sex Male 1 (ref) 1 (ref) Female 0.417 0.051–3.404 0.414 0.905 0.108–7.594 0.927 Procedures Lobectomy/less than total resection 1 (ref) 1 (ref) Total thyroidectomy 2.410 0.486–11.947 0.281 1.277 0.213–7.673 0.789 Cervical lymph node No metastasis 1 (ref) 1 (ref) Metastasis 9.599 1.181–78.042 0.034 7.353 0.831–65.051 0.073 Invasion Confinement 1 (ref) 1 (ref) Minimal extrathyroid extension 2.303 0.550–9.643 0.254 1.215 0.235–6.278 0.816 Fig. 1 Recurrence-free survival (RFS) according to presence of minimal ETE in solitary PTMCs. There was no significant difference in the RFS between the absence and presence of minimal ETE in solitary PTMCs (p = 0.240)

nement 1 (ref) 1 (ref) Minimal extrathyroid extension 2.303 0.550–9.643 0.254 1.215 0.235–6.278 0.816 Fig. 1 Recurrence-free survival (RFS) according to presence of minimal ETE in solitary PTMCs. There was no significant difference in the RFS between the absence and presence of minimal ETE in solitary PTMCs (p = 0.240) Discussion The presence of minimal ETE in PTC has been associated with LN metastasis, local recurrence, and poor prognosis.14–16 However, recent studies have called into question the prognostic significance of the presence of minimal ETE for tumor recurrence in PTC.17–19 Tumors with minimal ETE can be classified as category T3, irrespective of tumor diameter, according to the AJCC TNM cancer staging system.2 Thus, most cases of PTMCs can reveal the presence of minimal ETE after microscopic evaluation, especially when tumors are peripherally located despite a small diameter (accounting for 43 % of cases in the present study). Moreover, the benefit of evaluating true capsular invasion to confirm minimal ETE has been debated among endocrine pathologists because the anatomy of the thyroid gland demonstrates no definite fibrous capsule. The adipose tissue and fibers of the skeletal muscle, which are usually observed in the extrathyroid area, are normally also present within the thyroid parenchyma.10,11 Although the involvement of skeletal muscle is a more reliable feature for confirming the presence of minimal ETE than involvement of the perithyroid soft tissue, skeletal muscle is not uniformly distributed throughout the thyroid gland and can even be mixed with thyroid follicles in the isthmus.12 Identification of thick extrathyroid arteries might be helpful for confirming the presence of minimal ETE, but there is no reliable anatomical landmark that can serve as a criterion for minimal ETE.20 In our present study, we microscopically subclassified the extent of minimal ETE to estimate and compare the effect on RFS based on extension to the perithyroid soft tissue or the sternothyroid muscle, respectively. We found that invasion to the sternothyroid muscle had no dominant effect on RFS (Table 2). Furthermore, our current analyses suggested that the current definition for minimal ETE based on the AJCC TNM cancer staging system (E0 vs. E1 + E2) revealed the most statistically significant difference in RFS.2

, respectively. We found that invasion to the sternothyroid muscle had no dominant effect on RFS (Table 2). Furthermore, our current analyses suggested that the current definition for minimal ETE based on the AJCC TNM cancer staging system (E0 vs. E1 + E2) revealed the most statistically significant difference in RFS.2 Among the various clinicopathological parameters that we tested, older age, larger tumor size, and the presence of LN metastasis were each found to be associated with the presence of minimal ETE (Table S3), but there was no significant difference in terms of gender. Although male patients exhibited a significantly higher recurrence rate (Table 3) for PTC, these estimates were limited by the relatively small number of male patients (11.4 %, 62/546). Moreover, there was no significant difference in PTMC found to be related to gender. The only independent prognostic factor that we found to be associated with recurrence was the presence of LN metastasis in our solitary PTC but not in solitary PTMC (Tables 3, 4). In cases with LN metastasis, presence of extranodal extension and macrometastasis was significantly associated with a higher recurrence rate. The reason why there was a significance in N stage in univariate analysis but no effect in multivariate analysis might be caused by an effect that the presence of extranodal extension and larger metastatic tumor size displayed tendency of higher N stage. The presence of minimal ETE had no effect on RFS in solitary PTC and PTMC regardless of types of surgery (Tables 3, 4; Fig. 1; Tables S1 and S5). This finding is consistent with the results of previous studies of RFS or relapse-free survival in both PTC and PTMC patient groups.17–19,21

layed tendency of higher N stage. The presence of minimal ETE had no effect on RFS in solitary PTC and PTMC regardless of types of surgery (Tables 3, 4; Fig. 1; Tables S1 and S5). This finding is consistent with the results of previous studies of RFS or relapse-free survival in both PTC and PTMC patient groups.17–19,21 Our current retrospective study had several limitations of note, including a small sample size (546 PTCs, 144 PTMCs). A larger study of RFS in patients with PTMCs needs to be performed in the future. Indeed, a larger multicenter study will be needed for the proper validation of our current findings. Various effects of different clinical treatment modalities, such as RAI, were not analyzed for RFS in our present study. We attempted to eliminate the effects of multifocal PTCs in RFS, which limited our analysis to cases with solitary PTCs. Conclusions The presence of LN metastasis is the only independent prognostic factor associated with RFS in solitary PTC patients. The presence or absence of minimal ETE has no significant effect on RFS in solitary PTC and PTMC. Because the microscopic diagnostic criteria for the presence of minimal ETE remain the subject of debate, there is risk of over assigning patients with PTMC to the T3 category based on the present AJCC TNM cancer staging system. Electronic supplementary material Supplementary material 1 (DOCX 24 kb) Supplementary material 2 (PPTX 90 kb)

Giant cell tumor of bone (GCTB) is an aggressive, bone lytic, osteoclastogenic stromal tumor that mainly occurs in young adults.1,2 It commonly presents as an epiphyseal, monostotic lytic lesion most often found in the distal femur, proximal tibia, and distal radius.1 It is characterized by progressive growth and geographic bone lysis, leading to cortical bone expansion or dissolution with or without soft tissue extension. Symptoms generally include pain, swelling, and impaired mobility and function.1 Local mechanical load and joint function compromise are common in untreated disease. Rarely, GCTB can undergo malignant transformation. In addition, 1–4 % of GCTB cases give rise to pulmonary metastases even when the histologic appearance remains benign.3

lly include pain, swelling, and impaired mobility and function.1 Local mechanical load and joint function compromise are common in untreated disease. Rarely, GCTB can undergo malignant transformation. In addition, 1–4 % of GCTB cases give rise to pulmonary metastases even when the histologic appearance remains benign.3 Currently, surgical removal of the lesion remains the only curative intent treatment for GCTB;4 however, local recurrence or metastasis can still occur following curative intent surgery with modern imaging and high-speed burring.5,6 The most common form of surgical treatment for GCTB is aggressive local curettage with or without packing of the defect with bone cement or bone graft and internal fixation. The aim of this approach is to remove the tumor while preserving the local functional anatomy, including the articular joint surface. Varying rates of local recurrence have been reported after intralesional surgical therapy, and have led to the suggestion that the use of local adjuvants such as phenol, peroxide, water, or liquid nitrogen may further improve local control.7–11 More aggressive surgical approaches employing wide resection of the involved bone may be chosen to achieve tumor removal and potentially decrease the risk of local recurrence, at the cost of greater functional compromise.7 Major excision and resection of the involved bone (e.g. amputation, joint resection, or hemipelvectomy) for advanced GCTB,3 even if some form of bone or joint reconstruction is possible, is associated with significant functional deficit or morbidity.12

of local recurrence, at the cost of greater functional compromise.7 Major excision and resection of the involved bone (e.g. amputation, joint resection, or hemipelvectomy) for advanced GCTB,3 even if some form of bone or joint reconstruction is possible, is associated with significant functional deficit or morbidity.12 Denosumab, a monoclonal antibody directed against the receptor activator of nuclear factor-kappa β ligand (RANKL), has recently been approved in the United States, Europe, and Japan for the treatment of adults and skeletally mature adolescents with GCTB that is unresectable or when surgical resection is likely to result in severe morbidity.13–15 GCTB has been shown to be pathogenetically driven by pervasive expression of osteoclastic differentiation signals by tumor mononuclear stromal cells.16–19 Immunohistochemical and molecular probes have shown that stromal cell elements of GCTB strongly produce and express RANKL.17 RANKL appears to play an autocrine role in lesion development fostering and maintaining osteoclast formation, activation, and survival,18 resulting in continuous bone resorption19,20 via activating RANK receptor-positive osteoclast-like giant cells and their precursors.16,21

f GCTB strongly produce and express RANKL.17 RANKL appears to play an autocrine role in lesion development fostering and maintaining osteoclast formation, activation, and survival,18 resulting in continuous bone resorption19,20 via activating RANK receptor-positive osteoclast-like giant cells and their precursors.16,21 Previous results from an open-label, single-arm, phase II study demonstrated sustained denosumab-induced tumor responses in patients with GCTB (based on assessment of histologic or radiologic response).22 Denosumab treatment produced rapid and substantial suppression of bone turnover and significant reduction in the numbers of multinucleated giant cells seen in post-treatment resection specimens, as well as a marked reduction in the number and cross-sectional area of residual mononuclear stromal cells.17,22 There was a consistent finding of complete or near complete elimination of RANKL-producing stromal cells and disappearance of original RANK-positive multinuclear giant cells, along with the concomitant production of osteoid and new woven bone.17,22 These histopathologic changes correlated with an increase in radiographic density on computed tomography scanning.22 An initial planned interim analysis of the first 100 patients treated with denosumab therapy whose planned surgery was associated with severe morbidity found that 74 % had not undergone surgery for GCTB and that 16 % had a surgical procedure associated with less morbidity.23 At a median follow-up of 9.2 months (interquartile range [IQR] 4.2–12.9 months), 61 % of patients derived clinical benefit from denosumab, including pain reduction and improved mobility and function.23 In this study, we confirm and extend the results from the initial interim downstaging analysis23 and report detailed results from an unplanned interim analysis, performed at regulatory agency request, in 222 denosumab-treated patients with evaluable, resectable GCTB whose initially planned curative intent surgery was expected to result in severe morbidity.

he results from the initial interim downstaging analysis23 and report detailed results from an unplanned interim analysis, performed at regulatory agency request, in 222 denosumab-treated patients with evaluable, resectable GCTB whose initially planned curative intent surgery was expected to result in severe morbidity. Methods Patients and Procedures The study design and inclusion/exclusion criteria for this open-label, phase II study were previously reported (ClinicalTrials.gov identifier NCT00680992).23 Briefly, adults or skeletally mature adolescents (≥12 years of age) weighing ≥45 kg with radiologic evidence of ≥1 mature long bone, histologically confirmed GCTB, radiographically measurable active disease within 1 year before study enrollment, and Karnofsky performance status ≥50 % were enrolled. Exclusion criteria included concurrent use of alternative treatments for GCTB; known or suspected diagnosis of sarcoma, non-GCTB, giant cell–rich tumors, brown cell bone tumor of hyperparathyroidism, or Paget disease; diagnosis of a second malignancy in the past 5 years; history or current evidence of osteonecrosis or osteomyelitis of the jaw, active dental or jaw problems necessitating oral surgery, or nonhealed dental or oral surgery; or pregnancy.

t cell–rich tumors, brown cell bone tumor of hyperparathyroidism, or Paget disease; diagnosis of a second malignancy in the past 5 years; history or current evidence of osteonecrosis or osteomyelitis of the jaw, active dental or jaw problems necessitating oral surgery, or nonhealed dental or oral surgery; or pregnancy. Enrolled patients were separated into three cohorts.23 Patients from cohort 2 who were evaluable for surgical downstaging were included in this analysis. These patients had planned GCTB surgery that was associated with functional compromise or severe morbidity based on either the planned procedure, such as joint resection, limb amputation, or hemipelvectomy, or the extent or location of the lesion. The study was approved by the independent ethics committee or institutional review board for each study center. All patients provided written informed consent. The cutoff date for the data analysis was 30 August 2013.

such as joint resection, limb amputation, or hemipelvectomy, or the extent or location of the lesion. The study was approved by the independent ethics committee or institutional review board for each study center. All patients provided written informed consent. The cutoff date for the data analysis was 30 August 2013. Procedures Patients received open-label subcutaneous denosumab 120 mg every 4 weeks, with additional doses administered on days 8 and 15 during the first month of therapy only. For patients who had complete tumor resection, denosumab therapy continued for six additional doses after resection. In all other cases, denosumab therapy continued per protocol until either disease progression, recommendation of discontinuation by the investigator or sponsor, absence of clinical benefit according to the investigator’s judgment, withdrawal of patient consent, pregnancy, or use of any proscribed treatments. All patients were strongly advised to take daily supplements of ≥500 mg calcium and ≥400 IU vitamin D.

n, recommendation of discontinuation by the investigator or sponsor, absence of clinical benefit according to the investigator’s judgment, withdrawal of patient consent, pregnancy, or use of any proscribed treatments. All patients were strongly advised to take daily supplements of ≥500 mg calcium and ≥400 IU vitamin D. Curative intent surgical procedures planned at study entry were recorded prospectively, and actual surgical procedures performed after denosumab treatment were reported by investigators. Procedure selection and timing were based on serial review of radiographic imaging and clinical response by the treating physician. Disease status and clinical benefit (investigator-determined, every 4 weeks) were based on physical examination, patient report of symptoms, and serial radiologic imaging assessment per local standard practice. Serial radiographic assessments24,25 of GCTB lesions were performed per local practice guidelines, and the recommended surgical intervention was provided; the procedure was ranked using an invasiveness and postsurgical functional deficit scale.12,24 The initially recommended surgeries ranged from curettage to hemipelvectomy (invasiveness/postoperative functional impairment scale detailed in electronic supplementary Table S1). Safety Assessment Adverse events (AEs) and serious AEs (SAEs) were recorded and graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 3.0.23.26

Curative intent surgical procedures planned at study entry were recorded prospectively, and actual surgical procedures performed after denosumab treatment were reported by investigators. Procedure selection and timing were based on serial review of radiographic imaging and clinical response by the treating physician. Disease status and clinical benefit (investigator-determined, every 4 weeks) were based on physical examination, patient report of symptoms, and serial radiologic imaging assessment per local standard practice. Serial radiographic assessments24,25 of GCTB lesions were performed per local practice guidelines, and the recommended surgical intervention was provided; the procedure was ranked using an invasiveness and postsurgical functional deficit scale.12,24 The initially recommended surgeries ranged from curettage to hemipelvectomy (invasiveness/postoperative functional impairment scale detailed in electronic supplementary Table S1). Safety Assessment Adverse events (AEs) and serious AEs (SAEs) were recorded and graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 3.0.23.26 Statistical Analysis Statistical analyses were descriptive in nature, and only summary statistics were presented. Efficacy and safety analyses included patients who enrolled, received at least one dose of denosumab, and were evaluable for surgical downstaging. No formal sample size calculations were undertaken. Descriptive statistics included median (IQR) as appropriate for continuous variables, and frequency (%) for categorical variables.

and safety analyses included patients who enrolled, received at least one dose of denosumab, and were evaluable for surgical downstaging. No formal sample size calculations were undertaken. Descriptive statistics included median (IQR) as appropriate for continuous variables, and frequency (%) for categorical variables. Results Baseline demographics and disease characteristics of patients in our cohort are shown in Table 1 and Fig. 1. Of the 222 patients enrolled and evaluable for surgical downstaging, 54.1 % (n = 120) were female and 80.2 % (n = 178) were white. The median (IQR) age was 34 (25–44) years. The lesions were in the lower (52.7 %; n = 117) and upper (27.9 %; n = 62) extremities or axial skeleton (14.9 %; n = 33). The majority (66.7 %; n = 148) of patients presented with primary GCTB, and 33.3 % (n = 74) of patients had a recurrent tumor following a previous curative intent surgical procedure.Table 1 Baseline demographics and disease characteristics Demographics/characteristics Primary GCTB (n = 148) Recurrent GCTB (n = 74) All patientsa (N = 222) Sex, n (%) Female 80 (54.1) 40 (54.0) 120 (54.1) Male 68 (45.9) 34 (46.0) 102 (45.9) Race/ethnicity, n (%) White 117 (79.1) 61 (82.4) 178 (80.2) Asian 10 (6.8) 4 (5.4) 14 (6.3) Hispanic 10 (6.8) 3 (4.1) 13 (5.9) Black 8 (5.4) 4 (5.4) 12 (5.4) Other 3 (2.0) 2 (2.7) 5 (2.3) Age, years, median (Q1, Q3) 34 (26, 43) 35 (25, 46) 34 (25, 44) GCTB presentation status, n (%) Primary 148 – 148 (66.7) Recurrent – 74 74 (33.3) Planned surgery at presentation, n (%)b

(80.2) Asian 10 (6.8) 4 (5.4) 14 (6.3) Hispanic 10 (6.8) 3 (4.1) 13 (5.9) Black 8 (5.4) 4 (5.4) 12 (5.4) Other 3 (2.0) 2 (2.7) 5 (2.3) Age, years, median (Q1, Q3) 34 (26, 43) 35 (25, 46) 34 (25, 44) GCTB presentation status, n (%) Primary 148 – 148 (66.7) Recurrent – 74 74 (33.3) Planned surgery at presentation, n (%)b Hemipelvectomy 10 (6.8) 0 10 (4.5) Amputation 21 (14.2) 17 (23.0) 38 (17.1) Joint/prosthesis replacement 17 (11.5) 8 (10.8) 25 (11.3) Joint resection/fusion 22 (14.9) 11 (14.9) 33 (14.9) En bloc resection 57 (38.5) 26 (35.1) 83 (37.4) En bloc excision 4 (2.7) 4 (5.4) 8 (3.6) Marginal excision 1 (0.7) 0 1 (<1.0) Curettage 9 (6.1) 8 (10.8) 17 (7.7) Other 7 (4.7) 0 7 (3.2) GCTB giant cell tumor of bone, Q1, Q3 quartile 1, quartile 3 aPatients evaluable for surgical downstaging bPercentages may not add up to 100 due to rounding Fig. 1 Giant cell tumor of bone lesion location at baseline and operative status. Lesion locations highlighted in blue show sites where ≥50 % of patients remain on denosumab without curative intent surgery

GCTB giant cell tumor of bone, Q1, Q3 quartile 1, quartile 3 aPatients evaluable for surgical downstaging bPercentages may not add up to 100 due to rounding Fig. 1 Giant cell tumor of bone lesion location at baseline and operative status. Lesion locations highlighted in blue show sites where ≥50 % of patients remain on denosumab without curative intent surgery Exposure and Treatment Duration As of the cutoff date for this analysis, the 222 patients enrolled in this surgical downstaging cohort were treated with denosumab for a median (IQR) duration of 15.3 (12.1–23.6) months. In the 106 patients who had not yet had surgery and continued on monthly denosumab per protocol, a median (IQR) of 22.5 (15.0–34.0) doses of denosumab were administered for a median of 19.5 (12.4–28.6) months (electronic supplementary Fig. S1, Panel A). In the 116 patients who underwent surgery, the median (IQR) duration of denosumab treatment was 14.2 (12.0–17.7) months (electronic supplementary Fig. S1, Panel B). Treatment with denosumab resulted in radiologic evidence of an arrest in bone lysis and the interval development of new intralesional calcification (measured as increasing density [average Hounsfield unit density] on computed tomography), increases in cortical bone thickness (including the reappearance of cortical integrity), and an overall reduction in GCTB lesion size (measured in terms of longest measured lesion diameter) [example radiographs shown in Fig. 2].Fig. 2 Example of radiographic images of giant cell tumor of bone of the proximal humerus and distal femur before (a, c) and after (b, d) denosumab therapy. The initial lesions were expansile with a thin peripheral calcified shell and primarily soft tissue density centrally (a) and showed extensive soft tissue displacement with progression following radiotherapy 2 years previously (c). After 4 months of treatment with denosumab, the peripheral calcification was thicker, the central lesion more heavily mineralized, and the overall size was slightly decreased (b, d)

tissue density centrally (a) and showed extensive soft tissue displacement with progression following radiotherapy 2 years previously (c). After 4 months of treatment with denosumab, the peripheral calcification was thicker, the central lesion more heavily mineralized, and the overall size was slightly decreased (b, d) Planned Versus Performed Surgery In this cohort of patients, most had either not yet undergone surgery (48 %; n = 106/222) and remained on denosumab therapy or had undergone a less morbid procedure than originally planned (38 %; n = 84/222; Table 2). High morbidity procedures were avoided in 80 % of patients with either a planned hemipelvectomy (n = 8/10) or planned amputation (n = 32/40). Eighty-eight percent (n = 7/8) of patients with a planned en bloc excision and 37 % (n = 31/85) of patients with a planned en bloc resection were managed without surgical intervention in the reported follow-up period. Of the 85 patients with a planned en bloc resection, 85 % (n = 71) were able to have a less invasive or bone excision–sparing procedure or no surgery at all. The native joint preservation rate was 96 % (n = 24/25) in patients with a planned joint/prosthesis replacement and 86 % (n = 30/35) in patients with a planned joint resection/fusion. Of the 18 patients with planned curettage at baseline, 44 % (n = 8) required no surgery, 39 % (n = 7) underwent curettage as planned, and 17 % (n = 3) required en bloc resection.Table 2 Planned versus actual surgery in the study cohort (N = 222)

nt and 86 % (n = 30/35) in patients with a planned joint resection/fusion. Of the 18 patients with planned curettage at baseline, 44 % (n = 8) required no surgery, 39 % (n = 7) underwent curettage as planned, and 17 % (n = 3) required en bloc resection.Table 2 Planned versus actual surgery in the study cohort (N = 222) Procedures associated with a higher surgical morbidity were performed in six cases on study that were not planned at study entry. There were three cases in which curettage was planned and an en bloc resection was performed: two cases with lesions located in vertebral bodies, associated with significant soft tissue extension and bony destruction involving adjoining ribs with significant spinal cord compression, and one case with a rapidly growing 7-cm mass that originated in the posterior iliac spine but displayed evidence of cortical break and had invaded the paravertebral and psoas muscles extensively. For the remaining three cases, there was one case each in which an en bloc excision was planned and an en bloc resection was performed (proximal tibia lesion that had recurred twice before referral for trial enrollment), an en bloc resection was planned and a joint/prosthesis replacement was performed (recurrent proximal tibia lesion that had been resected 13 months previously with placement of hardware), and a marginal excision was planned and an amputation performed (proximal first phalanx lesion with radiographic evidence of involvement of the articulating distal metacarpal head).

thesis replacement was performed (recurrent proximal tibia lesion that had been resected 13 months previously with placement of hardware), and a marginal excision was planned and an amputation performed (proximal first phalanx lesion with radiographic evidence of involvement of the articulating distal metacarpal head). This cohort study is remarkable in that 33 % of patients who enrolled had already had one or more local recurrences after failed primary curative intent surgery. The patients with locally recurrent GCTB had very similar results to those seen in the primary GCTB population. Specifically, 45 (61 %) had not yet undergone surgery at the data cutoff date, 17 (23 %) underwent a less morbid procedure, 10 (14 %) underwent surgery as planned, and only 2 (3 %) underwent a more invasive morbid procedure. Importantly, of 17 initially recommended amputations in the locally recurrent population, none were required to date (see outcomes in patients following local recurrence in electronic supplementary Table S2).

procedure, 10 (14 %) underwent surgery as planned, and only 2 (3 %) underwent a more invasive morbid procedure. Importantly, of 17 initially recommended amputations in the locally recurrent population, none were required to date (see outcomes in patients following local recurrence in electronic supplementary Table S2). Median (IQR) duration of postoperative follow-up for all patients (n = 116) who underwent curative intent surgery was 13.0 (8.5–17.9) months. Local recurrence was reported in 15 % (n = 17/116) of patients who had surgery. The median duration of postoperative time until recurrence in the 17 patients who experienced local recurrence was 13.6 (10.5–15.7) months. In the 99 patients who underwent surgery but had not experienced recurrence by the time of data cutoff, the median postoperative follow-up time was 12.9 (7.8–18.0) months (see electronic supplementary Fig. S2). Of the 17 patients with local recurrence following denosumab therapy, 14 underwent curettage as their initial on-study GCTB surgery, 2 underwent en bloc resection, and 1 had a joint resection. The median number of doses of denosumab given in the adjuvant setting per protocol was 6.0 (IQR 3.0–6.0). Local recurrence was reported in 5 of the 29 patients (17 %) with recurrent GCTB at enrollment who underwent surgery after on-study treatment with denosumab.

nderwent en bloc resection, and 1 had a joint resection. The median number of doses of denosumab given in the adjuvant setting per protocol was 6.0 (IQR 3.0–6.0). Local recurrence was reported in 5 of the 29 patients (17 %) with recurrent GCTB at enrollment who underwent surgery after on-study treatment with denosumab. Adverse Events AEs of any grade occurring with >10 % frequency were as follows: arthralgia (24.8 %), fatigue (20.7 %), pain in extremity (19.4 %), headache (18.9 %), nausea (18.0 %), and back pain (10.8 %). Grade 3 or 4 AEs occurring with a ≥1 % frequency were hypophosphatemia (2.7 %) and pain in extremity (1.4 %). Twenty-one (9.5 %) patients experienced SAEs, and nine (4.1 %) experienced AEs that resulted in treatment discontinuation (Table 3). Of the 21 SAEs reported by investigators, only two occurred more frequently than once (appendicitis and osteitis; both n = 2, 0.9 %). There was one case each of osteonecrosis, nondisplaced tibia fracture, back pain, other neoplasm, and myeloproliferative disorder. Four (1.8 %) patients were reported with malignant GCTB transformation on study: two within-field, radiation-associated sarcomatous transformations at 4 and 6 years after radiotherapy, respectively, and two with pelvic or sacral GCTB lesions that progressed on denosumab by 257 days of exposure. In each of these latter two cases, a diagnosis of primary malignant GCTB was felt by the investigator to have been missed by sampling error at initial core biopsy. Nonserious occurrences of hypocalcemia were observed in 3.2 % of patients; no serious occurrences were reported. Only one patient reported osteonecrosis of the jaw (CTCAE grade 1), which resolved by the data cutoff date.Table 3 Patients with adverse eventsa

to have been missed by sampling error at initial core biopsy. Nonserious occurrences of hypocalcemia were observed in 3.2 % of patients; no serious occurrences were reported. Only one patient reported osteonecrosis of the jaw (CTCAE grade 1), which resolved by the data cutoff date.Table 3 Patients with adverse eventsa Patients with AEs Study cohort (N = 222) [n (%)] Overall safety summary 193 (86.9) AEs occurring with >10 % frequency Arthralgia 55 (24.8) Fatigue 46 (20.7) Pain in extremity 43 (19.4) Headache 42 (18.9) Nausea 40 (18.0) Back pain 24 (10.8) Grade 3 or 4 AEs 33 (14.9) Hypophosphatemiab 6 (2.7) Pain in extremityb 3 (1.4) Serious AEs 21 (9.5) AEs leading to treatment discontinuation 9 (4.1) AEs of interest Hypocalcemia (nonserious) 7 (3.2) Serious infections 6 (2.7) Adjudicated positive osteonecrosis of the jawc 1 (<1) AE adverse event aBased on Medical Dictionary for Regulatory Activities, version 14.1, and Common Terminology Criteria for Adverse Events, version 3.0 bHypophosphatemia and pain in extremity were the only grade 3 or 4 AEs occurring with a frequency ≥1 % cOne case of osteonecrosis of the jaw resolved by the cutoff date

Patients with AEs Study cohort (N = 222) [n (%)] Overall safety summary 193 (86.9) AEs occurring with >10 % frequency Arthralgia 55 (24.8) Fatigue 46 (20.7) Pain in extremity 43 (19.4) Headache 42 (18.9) Nausea 40 (18.0) Back pain 24 (10.8) Grade 3 or 4 AEs 33 (14.9) Hypophosphatemiab 6 (2.7) Pain in extremityb 3 (1.4) Serious AEs 21 (9.5) AEs leading to treatment discontinuation 9 (4.1) AEs of interest Hypocalcemia (nonserious) 7 (3.2) Serious infections 6 (2.7) Adjudicated positive osteonecrosis of the jawc 1 (<1) AE adverse event aBased on Medical Dictionary for Regulatory Activities, version 14.1, and Common Terminology Criteria for Adverse Events, version 3.0 bHypophosphatemia and pain in extremity were the only grade 3 or 4 AEs occurring with a frequency ≥1 % cOne case of osteonecrosis of the jaw resolved by the cutoff date Discussion Among patients with resectable GCTB treated with denosumab and for whom curative intent surgery was planned and believed to be associated with significant morbidity before enrollment, 48 % had not yet undergone surgery altogether and remained on monthly denosumab treatments at the time of the data cutoff. Another 38 % of patients were treated with denosumab and underwent a less invasive surgical procedure than was planned at study entry. The patients who underwent a curative intent procedure while on study have not yet experienced an increased local recurrence rate (15 %, at a median postoperative follow-up of 13.0 months for the 116 patients who underwent surgery) or rebound effect following discontinuation of denosumab treatment. These results support the conclusion that denosumab therapy may represent an important option for patients with resectable GCTB to avoid immediate surgery, control disease, or achieve equivalent surgical outcomes with less morbid procedures.

t surgery) or rebound effect following discontinuation of denosumab treatment. These results support the conclusion that denosumab therapy may represent an important option for patients with resectable GCTB to avoid immediate surgery, control disease, or achieve equivalent surgical outcomes with less morbid procedures. For patients with resectable GCTB tumors, disease control can be achieved with wide surgical excision or less invasive intralesional curettage. GCTB is usually surgically treated with intralesional curettage combined with high-speed burring, which is the least invasive surgical option, improving the thoroughness of tumor removal and allowing preservation of the joint adjacent to the tumor. Recurrence rates associated with intralesional curettage using bone graft as void filler and no additional adjuvants (such as cryotherapy or phenol) are reported to be between 12 and 65 %.25,27–33 Although wide excision is associated with a lower risk of local recurrence (up to 12 %),25,27,28,30,34 it is necessarily associated with poorer long-term functional consequences due to greater bone loss and the limitation of joint motion due to resection reconstruction. In view of these risks, deferring surgery or downstaging the surgical procedure needed to treat GCTB may offer substantial clinical benefits to patients.

t is necessarily associated with poorer long-term functional consequences due to greater bone loss and the limitation of joint motion due to resection reconstruction. In view of these risks, deferring surgery or downstaging the surgical procedure needed to treat GCTB may offer substantial clinical benefits to patients. Denosumab may permit less invasive procedures for patients with GCTB without deleterious outcomes, possibly serving as a contrast to previous reports indicating that highly morbid procedures are required to limit disease progression and recurrence.25,33,35,36 The native joint preservation was >85 % in patients with planned joint/prosthesis replacement or joint resection/fusion with denosumab treatment. In addition, even in cases where prosthetic replacement was performed, reduction in the size of the tumor mass and bone healing around the periphery of the tumor can facilitate complete en bloc tumor resection. Furthermore, there are several patients in the study in whom highly invasive surgery (e.g. amputation, hemipelvectomy, or axial skeleton surgery) was planned who remained on treatment with denosumab after achieving disease control, thus far without the need for highly invasive surgical intervention.

oc tumor resection. Furthermore, there are several patients in the study in whom highly invasive surgery (e.g. amputation, hemipelvectomy, or axial skeleton surgery) was planned who remained on treatment with denosumab after achieving disease control, thus far without the need for highly invasive surgical intervention. Recurrence rates in this study following surgical resection were similar to published experience (between 12 and 65 %25,27–33), which is particularly notable given the location of the tumors in our cohort, as well as the number of patients (n = 74; 33.3 %) with recurrent disease. These findings suggest that downstaging of the surgical invasiveness in patients treated with denosumab has not given rise to an increase in recurrence rate at a median postoperative follow-up of 13.0 months. Although these data must be interpreted with some caution given the follow-up time, previous collaborative group studies34 and longitudinal institutional case series9,31 have shown that local recurrence following surgery tends to occur predominantly within the first postoperative 12–18 months. No new safety risks were observed in this population of patients with GCTB receiving denosumab therapy. Osteonecrosis of the jaw, as well as hypocalcemia, were observed at low rates, consistent with previous studies of denosumab.23 Additional protocol-directed follow-up time of these patients (for 60 months total following surgery) will continue to reveal whether surgical downstaging modifies the long-term risk of postsurgical local recurrences in this population.

cemia, were observed at low rates, consistent with previous studies of denosumab.23 Additional protocol-directed follow-up time of these patients (for 60 months total following surgery) will continue to reveal whether surgical downstaging modifies the long-term risk of postsurgical local recurrences in this population. We report six cases in which procedures associated with a higher surgical morbidity that were not planned at study entry were performed on study. In each of these cases, the patient experienced radiographic response (defined as a reduction in size and/or increase in calcification), clinical benefit (defined as a reduction in pain and/or improvement in function or mobility), or both. Aside from a grade 3 wound infection in the patient who underwent resection of his iliac lesion, there were no reported intraoperative or postoperative surgical complications, and none of these six patients had experienced local or distant recurrence at the time of the data cutoff. Conclusions As of the cutoff date for this interim analysis, treatment with denosumab decreased the need for surgical intervention and reduced surgical morbidity in patients with GCTB who underwent surgery with curative intent. These findings support the use of denosumab in a preoperative setting to defer or downstage the planned surgical procedure in patients with GCTB when surgical resection is likely to result in severe morbidity. Electronic supplementary material Supplementary material 1 (DOCX 262 kb)

Conclusions As of the cutoff date for this interim analysis, treatment with denosumab decreased the need for surgical intervention and reduced surgical morbidity in patients with GCTB who underwent surgery with curative intent. These findings support the use of denosumab in a preoperative setting to defer or downstage the planned surgical procedure in patients with GCTB when surgical resection is likely to result in severe morbidity. Electronic supplementary material Supplementary material 1 (DOCX 262 kb) This work was funded by Amgen Inc. Medical writing support was provided by Rick Davis, MS, Meghan Johnson, PhD (on behalf of Amgen Inc.), and Albert Rhee, PhD (Amgen Inc.). Disclosure Piotr Rutkowski and Leanne L. Seeger have received honoraria from Amgen Inc.; Stefano Ferrari has served on advisory boards for Amgen Inc. and GlaxoSmithKline, has received research funding from MolMed, Pharmar, Morphotek, and Amgen Inc., and has received honoraria from Takeda; Robert J. Grimer has served on advisory boards for Amgen Inc.; Sander P.D. Dijkstra has served on advisory boards for Implantcast GmbH; and Amy Feng, Zachary J. Roberts, and Bruce A. Bach are employed by and own stock in Amgen Inc. Paul D. Stalley, Andrzej Pienkowski, Gualter Vaz, and Jay S. Wunder have no disclosures to report.

Patients diagnosed with perihilar cholangiocarcinoma (PHC) typically present with obstructive jaundice, which impairs liver function and is a risk factor for mortality after hepatobiliary surgery.1 Preoperative biliary drainage can resolve jaundice before surgery and may help reduce perioperative morbidity in patients submitted to en bloc partial hepatectomy.2 In Western centers, patients are preoperatively treated with endoscopic biliary drainage (EBD), percutaneous transhepatic biliary drainage (PTBD), or both.

Preoperative biliary drainage can resolve jaundice before surgery and may help reduce perioperative morbidity in patients submitted to en bloc partial hepatectomy.2 In Western centers, patients are preoperatively treated with endoscopic biliary drainage (EBD), percutaneous transhepatic biliary drainage (PTBD), or both. PTBD has been the preferred preoperative drainage method in Asian centers for decades, with favorable perioperative morbidity and mortality rates, but recent studies have focused on seeding metastases after preoperative PTBD and resection.3–5 These seeding metastases presumably result from exfoliated tumor cells in bile that drains along the percutaneous catheter. The incidence of catheter tract recurrences in those studies ranged from 2 to 5 %, and the incidence of laparotomy scar recurrences was 1.3 %.6–9 A low rate of seeding metastases has been reported since the early years of preoperative PTBD, but it is only recently that this low rate has been used to advocate for an exclusively endobiliary strategy.10,11 On the basis of the above data, many Eastern authors recently suggested that preoperative PTBD should be avoided and that endoscopic nasobiliary drainage should be preferred.12–14 From an oncologic perspective, however, only recurrences that affect overall survival (OS) are clinically relevant recurrences. It remains unclear if the reported seeding metastases were solitary recurrences, if they coincided with other recurrences, or if they developed after recurrent metastatic disease. Moreover, none of the above studies have assessed the effect of preoperative PTBD on OS.

survival (OS) are clinically relevant recurrences. It remains unclear if the reported seeding metastases were solitary recurrences, if they coincided with other recurrences, or if they developed after recurrent metastatic disease. Moreover, none of the above studies have assessed the effect of preoperative PTBD on OS. The present study was designed to assess OS after resection of PHC in patients with preoperative PTBD compared to patients with preoperative EBD. Additionally, we assessed the incidence of seeding metastases developing as initial recurrence after resection because we assumed that these initial recurrences would potentially influence OS. The broader objective was to establish the role that PTBD should have in preoperative management of PHC: either as a drainage method that can safely be used or only as a salvage procedure when other methods have failed.

ce after resection because we assumed that these initial recurrences would potentially influence OS. The broader objective was to establish the role that PTBD should have in preoperative management of PHC: either as a drainage method that can safely be used or only as a salvage procedure when other methods have failed. Methods Study Population Consecutive patients who underwent a resection with curative intent for PHC were identified from prospectively maintained databases at the Academic Medical Center (AMC) in Amsterdam, the Netherlands, and the Memorial Sloan Kettering Cancer Center (MSKCC), New York. PHC was defined according to the 7th edition of the American Joint Committee on Cancer staging manual.15 Patients were included from 1991 to 2012 if they had undergone preoperative biliary drainage before resection of PHC using extrahepatic bile duct resection and reconstruction with or without concomitant liver resection. Exclusion criteria were R2 resection, repeat resection after initial resection at another hospital, and 90 day postoperative mortality. Additional data were collected through retrospective chart review. The institutional review board at both institutions approved this study.

with or without concomitant liver resection. Exclusion criteria were R2 resection, repeat resection after initial resection at another hospital, and 90 day postoperative mortality. Additional data were collected through retrospective chart review. The institutional review board at both institutions approved this study. Patient selection for resection and preoperative management was similar between the 2 centers, as described previously.16 Biliary drainage was initiated in either a regional center before referral, or after referral to AMC or MSKCC. Patients were treated with initial EBD or initial PTBD according to the treating physician’s preference. Additional preoperative PTBD was performed when biliary decompression was inadequate after EBD or if EBD was associated with complications, such as cholangitis. The PTBD group in this study included patients treated with initial PTBD and patients treated with additional PTBD after inadequate EBD. The control group consisted of patients treated with preoperative EBD without previous or subsequent PTBD. All patients in the AMC in Amsterdam were routinely treated with a preoperative low-dose irradiation protocol (3 × 3.5 Gy in the 3 days before the resection) with the aim of preventing seeding metastases.17 Entry sites of percutaneous drain tracts in the abdomen were not routinely excised after resection.

Patient selection for resection and preoperative management was similar between the 2 centers, as described previously.16 Biliary drainage was initiated in either a regional center before referral, or after referral to AMC or MSKCC. Patients were treated with initial EBD or initial PTBD according to the treating physician’s preference. Additional preoperative PTBD was performed when biliary decompression was inadequate after EBD or if EBD was associated with complications, such as cholangitis. The PTBD group in this study included patients treated with initial PTBD and patients treated with additional PTBD after inadequate EBD. The control group consisted of patients treated with preoperative EBD without previous or subsequent PTBD. All patients in the AMC in Amsterdam were routinely treated with a preoperative low-dose irradiation protocol (3 × 3.5 Gy in the 3 days before the resection) with the aim of preventing seeding metastases.17 Entry sites of percutaneous drain tracts in the abdomen were not routinely excised after resection. Follow-up After Resection All patients were followed with CT imaging at 3 and 6 months after resection to detect early recurrence. Thereafter, patients at AMC were followed with clinical assessments until 5 years after resection; imaging was performed when indicated by rising tumor markers, symptoms, or findings at physical examination. At MSKCC, follow-up included CT or MRI imaging every 4 to 6 months. Pathologic confirmation of recurrences was often obtained, but it was not required if imaging unambiguously demonstrated recurrent disease in patients who were unfit to undergo further treatment. Suspect lesions at the laparotomy scar or in the prior PTBD drainage tract were always confirmed with a biopsy.

to 6 months. Pathologic confirmation of recurrences was often obtained, but it was not required if imaging unambiguously demonstrated recurrent disease in patients who were unfit to undergo further treatment. Suspect lesions at the laparotomy scar or in the prior PTBD drainage tract were always confirmed with a biopsy. Outcomes The primary end point in this study was OS, measured from the date of surgery to the date of death. Patients alive at follow-up were censored at the date of last contact before April 1, 2014. We used propensity score method rather than traditional multivariable analysis because this method is considered superior in reducing confounding and bias, especially when analyzing relatively small observational data sets.18,19 Secondary end points were directed toward the incidence of seeding metastases after resection. Analysis of recurrences focused on the pattern of initial recurrences based on the assumption that prognosis was unlikely to be affected by seeding metastases that arose after recurrence at another site. Seeding metastases were defined as recurrences either in the percutaneous catheter tract (i.e., any recurrence along the catheter tract from skin to the intrahepatic bile duct) or in the laparotomy scar (i.e., any recurrence in the abdominal wall at the laparotomy scar).6,8 In addition, the incidence of peritoneal recurrences (i.e., intra-abdominal recurrence in the peritoneum or ascites with malignant cells) was assessed, although these recurrences were not necessarily regarded as seeding metastases.

laparotomy scar (i.e., any recurrence in the abdominal wall at the laparotomy scar).6,8 In addition, the incidence of peritoneal recurrences (i.e., intra-abdominal recurrence in the peritoneum or ascites with malignant cells) was assessed, although these recurrences were not necessarily regarded as seeding metastases. Statistical Analyses Our method of propensity score adjustment was straightforward. First, we estimated propensity scores for the probability of PTBD assignment on the basis of all observed baseline characteristics. Second, we analyzed OS with a Cox proportional hazards model including 2 variables: drainage method (PTBD vs. EBD only) and propensity score (continuous variable). This model adjusts the survival analysis conditional on the propensity score. Thus, the model calculates the effect of PTBD compared to EBD only given that the propensity scores (i.e., the observed baseline characteristics) are hold equal.

drainage method (PTBD vs. EBD only) and propensity score (continuous variable). This model adjusts the survival analysis conditional on the propensity score. Thus, the model calculates the effect of PTBD compared to EBD only given that the propensity scores (i.e., the observed baseline characteristics) are hold equal. In more detail, we calculated propensity scores using multivariable logistic regression with preoperative PTBD as the outcome of interest and with adjustment for observed baseline characteristics, including demographics, comorbidities, total bilirubin level at referral, the level of bile duct involvement (Bismuth class), preoperative imaging variables (Blumgart T stage), cholangitis, extended hepatectomy, and treating center. Three baseline characteristics had missing data, including bilirubin level at presentation (27.8 % missing), Blumgart T stage (6.9 % missing), and preoperative cholangitis (5.3 % missing). To avoid bias, multiple imputation with 10 imputed data sets was performed for these missing data before estimation of the propensity scores, using a regression model that included all baseline characteristics. To evaluate residual bias after adjustment for propensity score, logistic regressions with drainage method as outcome were performed for each of the baseline characteristics with and without adjustment for propensity scores. We then estimated OS using the Kaplan–Meier method, and compared the groups with the log-rank test in univariable analysis. Finally, a multivariable Cox proportional hazards model was used to compare OS between the PTBD and EBD-only groups after adjustment for the propensity score as a continuous variable. To assess the proportional hazards assumption, we inspected the hazard ratio plots and found no violation.

test in univariable analysis. Finally, a multivariable Cox proportional hazards model was used to compare OS between the PTBD and EBD-only groups after adjustment for the propensity score as a continuous variable. To assess the proportional hazards assumption, we inspected the hazard ratio plots and found no violation. Analysis of secondary end points was performed by χ2 tests, and 95 % confidence intervals (CIs) were determined using the standard deviation of the mean. The type of liver resection and pathologic characteristics were also compared by χ2 tests. All analyses were performed in SPSS v22 (IBM, Armonk, NY). Results Patients A total of 344 consecutive patients underwent resection of PHC during the study period, of whom 66 (19.2 %) were excluded because they had not undergone preoperative biliary drainage. Of the remaining 278 patients, 33 (11.9 %) were excluded for 90 day postoperative mortality: 3 (8.1 %) of 37 patients treated with preoperative PTBD; 17 (9.8 %) of 147 patients treated with preoperative EBD; and 13 (19.4 %) of 67 patients treated with both. As a result, 245 patients were included, comprising 128 treated at MSKCC and 117 at AMC. Patient characteristics were not different between MSKCC and AMC, except for older age at MSKCC (mean age 65 vs. 61, respectively). The policy to use preoperative PTBD was different between the centers: PTBD was more often used in MSKCC than in AMC (43.8 vs. 27.4 %, respectively; P = 0.008).

ed at MSKCC and 117 at AMC. Patient characteristics were not different between MSKCC and AMC, except for older age at MSKCC (mean age 65 vs. 61, respectively). The policy to use preoperative PTBD was different between the centers: PTBD was more often used in MSKCC than in AMC (43.8 vs. 27.4 %, respectively; P = 0.008). The PTBD group consisted of 88 patients (36 %) who were treated with preoperative PTBD, including 54 patients who underwent PTBD after inadequate EBD. Patients in the PTBD group had undergone a median of 2 preoperative PTBD procedures (range 1–5). The median time between the first PTBD drainage procedure and surgery was 38 days (range 3–262); 17 patients (19.3 %) had a percutaneous catheter in situ more than 60 days. The EBD-only group (i.e., the control group) consisted of 157 patients (64 %) who were treated with preoperative EBD without PTBD. The distribution of patients between the PTBD and EBD-only groups was equal throughout the study period. The percentage of PTBD procedures between 1991 and 1996 was 32.3 %; 1997 and 2001, 37.0 %; 2002 to 2006, 32.7 %; and 2007 to 2012, 38.2 % (P = 0.88).

ients (64 %) who were treated with preoperative EBD without PTBD. The distribution of patients between the PTBD and EBD-only groups was equal throughout the study period. The percentage of PTBD procedures between 1991 and 1996 was 32.3 %; 1997 and 2001, 37.0 %; 2002 to 2006, 32.7 %; and 2007 to 2012, 38.2 % (P = 0.88). Baseline characteristics of the study groups are presented in Table 1. Nearly all variables were different between groups, indicating severe bias at baseline. The mean ± standard deviation propensity scores for patients in the PTBD and EBD-only groups were 0.53 ± 0.24 and 0.27 ± 0.19, respectively, with an area under the curve of 0.79. Only minimal differences in baseline characteristics remained after adjustment for propensity score, as indicated by the adjusted P values in Table 1. Type of resection and pathologic characteristics of both study groups are shown in Table 2.Table 1 Baseline characteristics Variable PTBD (n = 88) EBD only (n = 157) P

Baseline characteristics of the study groups are presented in Table 1. Nearly all variables were different between groups, indicating severe bias at baseline. The mean ± standard deviation propensity scores for patients in the PTBD and EBD-only groups were 0.53 ± 0.24 and 0.27 ± 0.19, respectively, with an area under the curve of 0.79. Only minimal differences in baseline characteristics remained after adjustment for propensity score, as indicated by the adjusted P values in Table 1. Type of resection and pathologic characteristics of both study groups are shown in Table 2.Table 1 Baseline characteristics Variable PTBD (n = 88) EBD only (n = 157) P Imputed Imputed adjusted for propensity score Male 53 (60.2) 100 (63.7) 0.59 0.94 Age, y, median (IQR) 61 (16) 65 (13) 0.01 0.99 Comorbidity, Charlson score ,27 median (IQR) 0 (1) 0 (1) 0.92 0.81 Total bilirubin at referral, μmol/L, median (IQR) 135 (231) 39 (67) 0.02 0.92 Bismuth class on imaging 0.001 0.95 Type 1 8 (9.1) 41 (26.1) Type 2 11 (12.5) 23 (14.6) Type 3a 30 (34.1) 44 (28.0) Type 3b 18 (20.5) 28 (17.8) Type 4 19 (21.6) 16 (10.2) Left or right hepatic duct 2 (2.3) 5 (3.2) Blumgart T stage on imaging 0.004 0.96 1 38 (43.7) 86 (61.0) 2 28 (32.2) 40 (28.4) 3 21 (24.1) 15 (10.6) Preoperative cholangitis 25 (29.4) 13 (8.8) <0.001 0.89 Extended hepatectomy 43 (48.9) 44 (28.0) 0.001 0.99 Treating center 0.008 0.98 MSKCC 56 (43.8) 72 (56.3) AMC 32 (27.4) 85 (72.6) Data are presented as n (%) unless otherwise specified. Logistic regressions with drainage method as outcome (PTBD or EBD only) were performed for each of the baseline variables to evaluate residual bias after adjustment for propensity scores

, 21 of 55 eligible patients underwent liposuction and were at least 3 months postsurgery. An additional seven patients underwent surgery (four did not consent to research and three had less than 3 months of follow-up), eight were booked in, and 19 patients decided against surgery or found the costs prohibitive (22 %). Measures Limb Volume Volume was calculated using 4-cm truncated cone circumferential measurements.28 A measuring board was used. Arms were measured seated, with the arm in horizontal abduction, hand pronated, and commencing at the ulnar styloid. Leg measurements were taken in the supine position with legs slightly abducted, commencing at the ankle between the lateral and medial malleolus. Individual limb volume, difference between limbs, and percentage difference were calculated, comparing the affected limb with the unaffected limb.28,29 Bioimpedance Spectroscopy (L-Dex) Measurements were taken supine using the ImpediMed L-Dex® machine (ImpediMed, Carlsbad, CA, USA) that assesses extracellular fluid in a unilateral limb using a low-voltage electrical current. L-Dex readings are an impedance ratio comparing the unaffected limb with the affected limb, with the unaffected limb acting as a patient-specific internal control.30 Skin was cleaned with alcohol swabs, and electrodes placed according to the manufacturer’s recommendations. Normal (no lymphedema) L-Dex readings ranged between −10 and 10.31

Treating center 0.008 0.98 MSKCC 56 (43.8) 72 (56.3) AMC 32 (27.4) 85 (72.6) Data are presented as n (%) unless otherwise specified. Logistic regressions with drainage method as outcome (PTBD or EBD only) were performed for each of the baseline variables to evaluate residual bias after adjustment for propensity scores PTBD percutaneous transhepatic biliary drainage, EBD endoscopic biliary drainage, IQR interquartile range, MSKCC Memorial Sloan Kettering Cancer Center, AMC Academic Medical Center Table 2 Type of liver resection and pathologic characteristics Characteristic PTBD (n = 88) EBD only (n = 157) P Preoperative cytology assessment, n (%) 0.40 Positive or suspicious 49 (55.7) 93 (59.2) Negative 18 (20.4) 39 (24.8) Not performed 21 (23.9) 25 (15.9) Type of liver resection, n (%) 0.003 Extrahepatic bile duct resection only 10 (11.4) 38 (24.2) Segment 4/5 wedge resection 3 (3.4) 17 (10.8) Mesohepatectomy 2 (2.3) 0 (0) Left hemihepatectomy 22 (25.0) 34 (21.7) Left extended hemihepatectomy 8 (9.1) 12 (7.6) Right hemihepatectomy 5 (5.7) 17 (10.8) Right extended hemihepatectomy 38 (43.2) 39 (24.8) Resection including caudate lobea 46 (61.3) 63 (61.8) 0.08 Resection specimen, n (%) T3 or T4 tumor (AJCC 7th edition) 34 (38.6) 31 (19.7) 0.002 R1 resection 21 (23.9) 49 (31.2) 0.24 Moderate/poor differentiation 22 (25.0) 33 (21.0) 0.52 Perineural invasion 70 (79.5) 108 (68.8) 0.08 Resected lymph nodes Total lymph node count, median (range) 3 (1–22) 4 (1–20) 0.15 N1 lymph node metastasis, n (%) 30 (34.1) 35 (22.3) 0.05 Mean lymph node ratio (positive/negative) 0.14 (1/7) 0.09 (1/11) 0.03

1.2) 0.24 Moderate/poor differentiation 22 (25.0) 33 (21.0) 0.52 Perineural invasion 70 (79.5) 108 (68.8) 0.08 Resected lymph nodes Total lymph node count, median (range) 3 (1–22) 4 (1–20) 0.15 N1 lymph node metastasis, n (%) 30 (34.1) 35 (22.3) 0.05 Mean lymph node ratio (positive/negative) 0.14 (1/7) 0.09 (1/11) 0.03 PTBD percutaneous transhepatic biliary drainage, EBD endoscopic biliary drainage, AJCC American Joint Committee on Cancer a Percentage of caudate resections only concerns patients who underwent meso- or hemihepatectomy Overall Survival Among the 245 included patients, 173 patients (71 %) died during follow-up. The median OS was 38 months (95 % CI 32–44), and 5-year survival was 32 %. Median follow-up among survivors was 52 months (range 6–251 months). The unadjusted OS was comparable between the PTBD group (36 months) and the EBD-only group (41 months; P = 0.25; Fig. 1a). Stratifying patients in the PTBD group between those who underwent EBD plus PTBD and those who underwent PTBD only did not reveal a difference when these 2 groups were compared to patients who underwent EBD only (P = 0.44; Fig. 1b). After using propensity score adjustment to account for potential confounders, OS between the PTBD group and EBD-only group was similar (adjusted hazard ratio, 1.05; 95 % CI 0.74–1.49; P = 0.80; Fig. 2).Fig. 1 Unadjusted Kaplan–Meier survival plots. a Patients in PTBD and EBD-only groups had comparable survival (P = 0.26). b Stratifying patients in PTBD group between those who underwent PTBD only and those who underwent PTBD plus EBD did not reveal difference (P = 0.45)

atio, 1.05; 95 % CI 0.74–1.49; P = 0.80; Fig. 2).Fig. 1 Unadjusted Kaplan–Meier survival plots. a Patients in PTBD and EBD-only groups had comparable survival (P = 0.26). b Stratifying patients in PTBD group between those who underwent PTBD only and those who underwent PTBD plus EBD did not reveal difference (P = 0.45) Fig. 2 Survival plot after adjustment for propensity score in Cox regression analysis showing similar OS in PTBD and EBD-only groups (P = 0.80)

atio, 1.05; 95 % CI 0.74–1.49; P = 0.80; Fig. 2).Fig. 1 Unadjusted Kaplan–Meier survival plots. a Patients in PTBD and EBD-only groups had comparable survival (P = 0.26). b Stratifying patients in PTBD group between those who underwent PTBD only and those who underwent PTBD plus EBD did not reveal difference (P = 0.45) Fig. 2 Survival plot after adjustment for propensity score in Cox regression analysis showing similar OS in PTBD and EBD-only groups (P = 0.80) Seeding Metastases Developing as Initial Recurrence A total of 87 patients in the PTBD group and 147 patients in the EBD-only group were available for recurrence analysis (1 patient missing in the PTBD group and 10 patients missing in the EBD-only group; total 4 %). Seeding metastases occurred as initial recurrence in 3 (3.4 %) of 87 patients in the PTBD group (95 % CI 0–7.3), and in 4 (2.7 %) of 147 patients in the EBD-only group (95 % CI 0–5.3; P = 0.71). Among the total 7 patients who developed a seeding metastasis, 3 had developed a concurrent local recurrence. Time to diagnosis of the seeding metastases was 8, 13, 14, 17, 20, 21, and 66 months (median, 17 months), and OS in these 7 patients was 13, 30, 20, 21, 21, 27, and 142 months (median, 21 months), respectively. All 7 seeding metastases developing as initial recurrence were abdominal wall recurrences at the site of the laparotomy scar. No initial recurrences were observed in a percutaneous catheter tract. The incidence of seeding metastases was not significantly different between both centers: 5 (3.9 %) of 128 patients at MSKCC (95 % CI 0.5–7.3) and 2 (1.9 %) of 106 patients at AMC (95 % CI 0–4.5; P = 0.46).

n impedance ratio comparing the unaffected limb with the affected limb, with the unaffected limb acting as a patient-specific internal control.30 Skin was cleaned with alcohol swabs, and electrodes placed according to the manufacturer’s recommendations. Normal (no lymphedema) L-Dex readings ranged between −10 and 10.31 Function/Emotions Functional impairment was assessed using the Patient-Specific Functional Scale (PSFS).32 The PSFS is reliable and valid across contexts,33 and sensitive to change in breast cancer survivors,34 but not previously validated for lymphedema. Patients listed three personally important activities impaired by lymphedema (e.g. ‘brushing my hair’), then rated the extent to which lymphedema impaired each activity (‘0’, not able to perform, to ‘10’, able). Therefore, although the specific activities nominated by individuals differed substantially, impairment ratings were standardized and thus comparable across activities, patients, and time.32 Impairment ratings were summed for each patient, forming an individualized index of functional impairment (range 0–30), with higher numbers indicating less impairment. It was not appropriate to calculate internal consistency (e.g. Cronbach’s α) because activities were idiosyncratic across patients.

rences at the site of the laparotomy scar. No initial recurrences were observed in a percutaneous catheter tract. The incidence of seeding metastases was not significantly different between both centers: 5 (3.9 %) of 128 patients at MSKCC (95 % CI 0.5–7.3) and 2 (1.9 %) of 106 patients at AMC (95 % CI 0–4.5; P = 0.46). Peritoneal Recurrences Initial peritoneal recurrences were observed in 32 (13.7 %) of 234 patients with available recurrence status (95 % CI 9.2–18.1). These included 11 (12.6 %) of 87 patients in the PTBD group and 21 (14.3 %) of 147 patients in the EBD-only group, which was not significantly different between groups (P = 0.85). Concomitant peritoneal recurrence was observed in only 1 of 7 patients with a seeding metastasis as initial recurrence. Discussion Biliary drainage has become an important component of the preoperative preparation of patients with PHC, as multiple studies have shown that it decreases postoperative liver failure and mortality.2 Nonetheless, several controversies have evolved. Eastern centers reported catheter tract recurrences after preoperative PTBD and resection of PHC and promoted the use of alternative drainage methods. In the present study, however, we showed that preoperative PTBD is not associated with survival after resection of PHC compared to patients who underwent preoperative endoscopic drainage. Moreover, PTBD was not associated with an increase in seeding metastases developing as initial recurrence that would potentially affect survival.

t study, however, we showed that preoperative PTBD is not associated with survival after resection of PHC compared to patients who underwent preoperative endoscopic drainage. Moreover, PTBD was not associated with an increase in seeding metastases developing as initial recurrence that would potentially affect survival. One previous study has assessed survival after preoperative PTBD: in a study of 141 patients with resected PHC, Hirano et al. found a median OS of 31 months after preoperative PTBD compared to 59 months after preoperative EBD.20 The authors attributed this difference to an increase in peritoneal metastases after preoperative PTBD. However, the PTBD group in that study had more advanced disease as evidenced by more patients with Bismuth type 4 tumors (30 vs. 14 %), more perioperative blood transfusions (31 vs. 9 %), and more frequent hepatic artery resections (22 vs. 9 %). Although a survival difference was confirmed in multivariable analysis, the statistical model in that study may have been at risk to a false-positive finding due to overfitting because it was adjusted for 9 other covariates. Moreover, statistical criteria, like adjusting for all significant variables from univariable analysis, as used in the study by Hirano et al., are considered insufficient to characterize confounding or selection bias.21

risk to a false-positive finding due to overfitting because it was adjusted for 9 other covariates. Moreover, statistical criteria, like adjusting for all significant variables from univariable analysis, as used in the study by Hirano et al., are considered insufficient to characterize confounding or selection bias.21 Catheter tract recurrences have been reported in a range of 2 to 5 % after preoperative PTBD, but our study found no catheter tract recurrence developing as the initial recurrence after resection.6–9 This discrepancy may be partly explained by the behavior of seeding metastases. Apparently, catheter tract recurrences, if they occur at all, have a tendency to grow slowly and not to manifest before other recurrences have been diagnosed. Alternatively, differences in management and patient selection between centers in the present study and Eastern centers could explain the discrepancy. The duration of PTBD has been identified as a risk factor for catheter tract recurrences: preoperative PTBD longer than 60 days was associated with an increased risk in the study by Takahashi et al., and more than 25 % of the patients in that study reached the cutoff, compared to only 19 % in the present study.6 It is uncertain whether preoperative low-dose radiotherapy, which was standard treatment in the AMC and not in MSKCC, has prevented catheter tract recurrences or other seeding metastases. There was no difference in the incidence of seeding metastases between the 2 study centers, so the current data does not support routine use of preoperative radiotherapy.

ow-dose radiotherapy, which was standard treatment in the AMC and not in MSKCC, has prevented catheter tract recurrences or other seeding metastases. There was no difference in the incidence of seeding metastases between the 2 study centers, so the current data does not support routine use of preoperative radiotherapy. Normally, PHC spreads to the liver and through lymph nodes to the abdomen or extra-abdominal sites.22 In analogy to previous studies, we named recurrences in the laparotomy scar or in the percutaneous catheter tract “seeding metastases” because these recurrences show a deviating pattern from normally observed recurrences and are likely the result of tumor seeding. Nonetheless, clear evidence for a role of seeding tumor cells has not been demonstrated in these kinds of recurrences. Alternatively, local inflammation after surgical trauma might naturally attract circulating tumor cells.23 To a larger extent, peritoneal recurrences are doubtfully the result of tumor seeding. Although some peritoneal metastases may be caused by perioperative tumor seeding and could thus be preventable, most peritoneal metastases will reflect extensive disseminated disease. In the present study, preoperative PTBD did not increase the incidence of peritoneal recurrences.

oubtfully the result of tumor seeding. Although some peritoneal metastases may be caused by perioperative tumor seeding and could thus be preventable, most peritoneal metastases will reflect extensive disseminated disease. In the present study, preoperative PTBD did not increase the incidence of peritoneal recurrences. This retrospective study has several limitations. The sample size was likely insufficient to definitively exclude an adverse effect of preoperative PTBD on OS after resection of PHC. Although no statistically significant difference was found when comparing unadjusted OS and propensity-adjusted OS, the 95 % CI of the propensity-adjusted hazard ratio was still relatively wide. However, not a single catheter tract recurrence was found during follow-up of initial recurrences, and the incidence of initial abdominal wall recurrences was also similar between the PTBD and EBD-only groups.

OS and propensity-adjusted OS, the 95 % CI of the propensity-adjusted hazard ratio was still relatively wide. However, not a single catheter tract recurrence was found during follow-up of initial recurrences, and the incidence of initial abdominal wall recurrences was also similar between the PTBD and EBD-only groups. The analysis of seeding metastases as initial recurrence requires 3 comments. First, follow-up was not standardized, so it is possible that some initial seeding metastases were missed after patients were lost to follow-up. Second, we only recorded initial recurrences, and we may have missed seeding metastases that occurred after initial diagnosis of recurrent disease. This follow-up approach may not provide the true incidence of seeding metastases, but it is based on clinical meaningfulness: management or OS is unlikely to be affected by seeding metastases if they occur late in the course of the disease, after the initial diagnosis of recurrence. Third, the present study included only patients who underwent a potentially curative resection, so patients with inoperable disease due to (extra)hepatic or N2 lymph node metastases were excluded. On the basis of these data, we cannot be sure whether some patients had seeding metastases at the time of surgery. Nonetheless, in our experience with management of PHC, we have never observed any seeding metastases during exploratory laparotomy.

disease due to (extra)hepatic or N2 lymph node metastases were excluded. On the basis of these data, we cannot be sure whether some patients had seeding metastases at the time of surgery. Nonetheless, in our experience with management of PHC, we have never observed any seeding metastases during exploratory laparotomy. Regarding management of PHC, the use of preoperative drainage before smaller liver resections (e.g., left hemihepatectomy) may not be necessary because of the large liver remnant. Moreover, preoperative drainage could even be harmful in these patients, as a recent study showed that preoperative drainage might increase perioperative morbidity due to infection-related complications.24 In the present study, preoperative drainage was often used before small liver resections because many patients present to our centers with drains already in place or with badly placed drains that are associated with infection and require revision. Of note, 48 extrahepatic bile duct resections without liver resection were performed for Bismuth type 1 or 2 tumors during the early years of the study cohort. Since approximately 2000, a liver resection is part of a potentially curative resection for PHC, particularly for Bismuth type 2.

nfection and require revision. Of note, 48 extrahepatic bile duct resections without liver resection were performed for Bismuth type 1 or 2 tumors during the early years of the study cohort. Since approximately 2000, a liver resection is part of a potentially curative resection for PHC, particularly for Bismuth type 2. Preoperative PTBD for PHC is currently being used in most Western surgical specialty centers when EBD fails to obtain adequate preoperative biliary drainage. Some centers even prefer to use preoperative PTBD as primary drainage method instead of EBD. There are many reasons to use PTBD: it has been associated with fewer preoperative complications than EBD; percutaneous catheters provide direct access to bile ducts perioperatively; and percutaneous catheters can be used as stents to protect hepaticojejunostomies from leaking postoperatively.25 No definitive data are currently available for any of the suggested advantages, but a randomized controlled trial is being conducted to assess differences in perioperative complications between EBD and PTBD.26 In conclusion, these data suggest that PTBD can safely be used in preoperative management of PHC. The present study found no effect of PTBD on survival compared to patients who underwent preoperative EBD and no increase in seeding metastases that develop as initial recurrence. The decision to use preoperative PTBD should not be influenced by concerns about catheter tract recurrences; they are very rare, and they probably do not affect OS. Bas Groot Koerkamp and Robert J. Coelen have contributed equally to this article.

In conclusion, these data suggest that PTBD can safely be used in preoperative management of PHC. The present study found no effect of PTBD on survival compared to patients who underwent preoperative EBD and no increase in seeding metastases that develop as initial recurrence. The decision to use preoperative PTBD should not be influenced by concerns about catheter tract recurrences; they are very rare, and they probably do not affect OS. Bas Groot Koerkamp and Robert J. Coelen have contributed equally to this article. Acknowledgment JW was funded by the Academic Medical Center Young Talent Fund; BG was funded by the Dutch Cancer Society (DCS), Grant UVA 2011-4973; AD received research fellowship Grants from the French Association of Hepatobiliary Surgery and Transplantation (ACHBT) and from Université de Bourgogne. Conflict of interest The authors declare no conflict of interest. Disclosure None.

“I was always alone and at my lowest point I wanted to have my leg cut off. I remember going to my General Practitioner and asking if that were possible”, first leg liposuction patient (Fig. 1).Fig. 1 First arm and leg liposuction patients, both right-side affected. Left panel pre-liposuction, middle panel 6 months post-liposuction, right panel 12 months post-liposuction The negative impact of advanced lymphedema on the physical and emotional health of cancer survivors cannot be understated.1–9 Symptomatic management, rather than cure, is available for the large proportion of breast (21 %), gynecologic (20 %), melanoma (16 %), and genitourinary (10 %) cancer survivors (among others) who develop lymphedema.10,11 Early lymphedema swelling can usually be managed with decongestive lymphatic therapy (DLT): lymphatic drainage massage, compression garments, exercises, and skin care.2 However, until recently no further treatments existed for advanced lymphedema resistant to DLT. This report presents volume, bioimpedance, functional, and emotional outcomes from the first Australian clinic conducting liposuction for advanced lymphedema.

rainage massage, compression garments, exercises, and skin care.2 However, until recently no further treatments existed for advanced lymphedema resistant to DLT. This report presents volume, bioimpedance, functional, and emotional outcomes from the first Australian clinic conducting liposuction for advanced lymphedema. European teams first developed liposuction protocols for advanced lymphedema,12–14 the rationale being that swelling in advanced lymphedema is not only due solely to lymphatic fluid but also to accumulating adipose tissue15,16 and sometimes fibrosis.17 Liposuction significantly reduces excess limb volume12,17,18 for patients with advanced arm or leg lymphedema,12,13,19,20 with ongoing reduction maintained by continuous compression garment use.17,18,21 Furthermore, Swedish data suggest that liposuction reduces episodes of cellulitis (often requiring hospitalization) from an annual incidence of 40 % preoperatively to 10 % postoperatively.22 The best outcomes post-liposuction are achieved by patients who return frequently for monitoring and garment renewal.14,18 However, this very need for ongoing monitoring and expensive garments raises questions about whether liposuction will generalize out of the universal healthcare contexts of Sweden and the UK to countries such as Australia and the US where the cost of lymphedema treatment is paid for by the individual.23 Furthermore, temperatures in Sweden and the UK are comparatively low, making it more comfortable to wear compression garments. Nevertheless, liposuction remains the best contemporary surgical option to reduce swelling for patients with advanced lymphedema because alternatives, such as lymph node transfer (LNT) microsurgery,7,24–26 focus only on restoring lymphatic function in the affected area. Although perhaps stopping the progression of advanced lymphedema, such techniques cannot foreseeably reduce the fatty and fibrotic swelling already present.

h advanced lymphedema because alternatives, such as lymph node transfer (LNT) microsurgery,7,24–26 focus only on restoring lymphatic function in the affected area. Although perhaps stopping the progression of advanced lymphedema, such techniques cannot foreseeably reduce the fatty and fibrotic swelling already present. Two additional topics are not adequately addressed in liposuction research. First, although liposuction substantially improves function,3 studies disagree as to whether liposuction improves emotional outcomes.3,13 Second, no research has measured extracellular lymphatic fluid pre- and post-liposuction using bioimpedance, an assessment that will soon become standard-of-care for lymphedema.27 Therefore, in 2012, clinicians and researchers at Macquarie University convened a multidisciplinary Advanced Lymphedema Assessment Clinic (ALAC) for privately funded patients to carefully implement and evaluate liposuction for lymphedema. This report presents volume, bioimpedance, functional, and emotional outcomes after 2 years of this program (median postsurgery follow-up, 9 months; range 3–18).

a multidisciplinary Advanced Lymphedema Assessment Clinic (ALAC) for privately funded patients to carefully implement and evaluate liposuction for lymphedema. This report presents volume, bioimpedance, functional, and emotional outcomes after 2 years of this program (median postsurgery follow-up, 9 months; range 3–18). Methods Setting Specialists in rehabilitation (HM), plastic surgery (TL, QN), imaging (JM), oncology (JB), and allied health (LK, AH-W) established ALAC in May 2012. TL was trained in surgical technique by AM. TL, HM, and LK received further instruction in technique and follow-up protocol from HB in Sweden. HB also provided analysis tools to ensure comparability with previous studies. At ALAC, a rehabilitation specialist with expertise in lymphedema (HM) took a patient’s medical history and assessed their eligibility for surgery. A lymphedema-trained physiotherapist (AH-W) measured limb volumes, took bioimpedance spectroscopy (L-Dex), and assessed function. Eligible patients discussed liposuction with the plastic surgeon (TL). Preoperative magnetic resonance imaging (MRI) assessed the location and extent of lymphatic fluid and fat for surgical planning. If there was evidence of pitting edema or substantial lymphatic fluid on MRI, patients completed modified intensive DLT consisting of 1–2 weeks of presurgery bandaging at a private rehabilitation hospital. Patients returned to ALAC 2–6 weeks postsurgery and at 3, 6, 9, and 12 months to measure limb volume and order new compression garments, and for reviews 6-monthly thereafter. If progress was good and new garments were not required, 3- and/or 9-month assessment was omitted. MRI was repeated at 6 months to reassess fluid and fat distribution, and will be analyzed in future research to improve muscle, fat, and fluid differentiation and measurement. All cases were reviewed at monthly multidisciplinary team meetings.

nd new garments were not required, 3- and/or 9-month assessment was omitted. MRI was repeated at 6 months to reassess fluid and fat distribution, and will be analyzed in future research to improve muscle, fat, and fluid differentiation and measurement. All cases were reviewed at monthly multidisciplinary team meetings. Patients’ motivation to wear, and their ability to pay for, continuous garments was assessed. Patients without private health insurance self-funded a 5-day hospital stay, theater room hire, physiotherapy, and inpatient DLT. Patients and Eligibility Between May 2012 and May 2014, a total of 104 patients were assessed. Eligibility criteria for liposuction were (i) unilateral nonpitting primary or secondary advanced International Society of Lymphology (ISL) stage II or III lymphedema;12,13 (ii) limb volume difference greater than 25 % (calculated using the truncated-cone method);12,13 and (iii) DLT provided no further volume reductions. Patients were excluded if they had not undertaken maximum DLT (n = 3), had active recurrent cancer (n = 2), bilateral lymphedema (n = 5), frailty (n = 3), or were reluctant to wear compression garments continuously (n = 5).17,18,21 By May 2014, 21 of 55 eligible patients underwent liposuction and were at least 3 months postsurgery. An additional seven patients underwent surgery (four did not consent to research and three had less than 3 months of follow-up), eight were booked in, and 19 patients decided against surgery or found the costs prohibitive (22 %).

atients, and time.32 Impairment ratings were summed for each patient, forming an individualized index of functional impairment (range 0–30), with higher numbers indicating less impairment. It was not appropriate to calculate internal consistency (e.g. Cronbach’s α) because activities were idiosyncratic across patients. To complement the PSFS, patients also rated the impact of lymphedema on six functional/emotional domains drawn from previous research35,36 (see the Appendix): pain, heaviness, extent of swelling, degree of self-consciousness, anxiety, and negative emotions (‘0’, not at all, to ‘10’, extremely so). Although the domains were correlated (Cronbach’s α = 0.76), they were reported individually because they reflect diverse content.

wn from previous research35,36 (see the Appendix): pain, heaviness, extent of swelling, degree of self-consciousness, anxiety, and negative emotions (‘0’, not at all, to ‘10’, extremely so). Although the domains were correlated (Cronbach’s α = 0.76), they were reported individually because they reflect diverse content. Surgical Technique and Compression Therapy The surgical procedure was identical to that described in Sweden.12,13,19,20 Liposuction was performed under general anesthesia following limb exsanguination and tourniquet application. Using specialized Helixed Tri Port III cannulas (22 and 30 cm long, 4–5 mm wide) connected to a vacuum pump, subcutaneous tissue was removed through multiple small incisions along the limb. Presurgical limb volume determined how much tissue to remove to equalize volume relative to the unaffected limb. Compression garments were applied to the affected limb immediately postsurgery prior to tourniquet release—custom-made 30 mmHg JOBST® Elvarex® for arms, or Ready Wraps (Solaris) for legs. From 1-week postsurgery, all leg patients wore JOBST® Elvarex® custom-made compression garments 50–80 mmHg. Initial postsurgical garments were measured using the circumference of the unaffected limb; subsequent measurements were obtained from the operated limb by a trained garment fitter. Every order consisted of two garments, allowing one to be worn while the other was washed. Throughout follow-up, compression garments alone were used in areas where liposuction was performed. However, DLT was used, where indicated, in areas where liposuction was not performed (hands or feet) or areas that cannot be adequately compressed (shoulder or hip). These areas are not included in limb volume measurements.28,29

t follow-up, compression garments alone were used in areas where liposuction was performed. However, DLT was used, where indicated, in areas where liposuction was not performed (hands or feet) or areas that cannot be adequately compressed (shoulder or hip). These areas are not included in limb volume measurements.28,29 Statistical Analysis Volume, L-Dex, and function were compared with paired samples t tests. As analysis entailed five pairwise comparisons between succeeding timepoints (pre vs. post, pre vs. 3-month, etc.), the Bonferroni correction was applied, (i.e. p value was considered significant at 0.05/5 = 0.0137). With 21 participants, we calculated 80 % power at p = 0.01 to detect a d = 0.8 decrease in limb volume and L-Dex. However, as fewer participants had longer follow-up, less power was available at 12-month follow-up. Hence, further clinically meaningful volume reductions may not attain statistical significance. Statistical tests were performed using SPSS version 21 for Windows (IBM Corporation, Armonk, NY, USA). Data were collected with patients’ consent and Macquarie University Human Research Ethics Committee approval.

follow-up. Hence, further clinically meaningful volume reductions may not attain statistical significance. Statistical tests were performed using SPSS version 21 for Windows (IBM Corporation, Armonk, NY, USA). Data were collected with patients’ consent and Macquarie University Human Research Ethics Committee approval. Results Table 1 details characteristics of 15 arm and 6 leg liposuction patients. Cancer-related secondary lymphedema was a more common reason for liposuction (85.7 %) than primary (congenital) lymphedema (14.3 %), with breast cancer treatment being the most common underling cause (66.7 %). Patients with arm lymphedema were older (mean(arm) 57.8 years, range 25–69; mean(leg) 50.7, range 18–66) but had less longstanding lymphedema (mean(arm) 9.1 years, range 2–29), than patients with leg lymphedema (mean(leg) 15.5, range 3–42). All patients were female, reflecting the sex disparity in lymphedema prevalence due to breast cancer.Table 1 Participant characteristics Upper Limb Lower Limb Total (%) n n Sex Female 15 6 21 100 Age (years) <50 2 1 3 14.3 ≥50 13 5 18 85.7 Mean (SD) 57.8 (12.2) 50.7 (16.9) Range 25–69 18–66 BMI (kg/m2) Mean (SD) 28.3 (4.1) 29.69 (2.7) Low to normal (<25) 3 0 3 14.3 Overweight (25–30) 7 4 11 52.4 Obese (>30) 5 2 7 33.3 Cancer diagnosis Breast 14 0 14 66.7 Gynecological 0 4 4 19.0 Non-cancer 1 2 3 14.3 Nodal surgery Nil 1 5 6 28.6 Sentinel node biopsy 1 0 1 4.8 Nodal dissection 13 1 14 66.7 Radiotherapy Yes 11 1 12 57.1 No 4 5 9 42.9 Chemotherapy Yes 11 2 13 61.9 No 4 4 8 38.1 SD standard deviation, BMI body mass index

Sex Female 15 6 21 100 Age (years) <50 2 1 3 14.3 ≥50 13 5 18 85.7 Mean (SD) 57.8 (12.2) 50.7 (16.9) Range 25–69 18–66 BMI (kg/m2) Mean (SD) 28.3 (4.1) 29.69 (2.7) Low to normal (<25) 3 0 3 14.3 Overweight (25–30) 7 4 11 52.4 Obese (>30) 5 2 7 33.3 Cancer diagnosis Breast 14 0 14 66.7 Gynecological 0 4 4 19.0 Non-cancer 1 2 3 14.3 Nodal surgery Nil 1 5 6 28.6 Sentinel node biopsy 1 0 1 4.8 Nodal dissection 13 1 14 66.7 Radiotherapy Yes 11 1 12 57.1 No 4 5 9 42.9 Chemotherapy Yes 11 2 13 61.9 No 4 4 8 38.1 SD standard deviation, BMI body mass index Significant post-liposuction reduction in limb volume was achieved for all patients (Table 2). Mean preoperative limb difference was 45.1 % (range 23–83), decreasing between 2 and 6 weeks postsurgery to 13.2 % (range −2 to 24), a significant 68.2 % reduction (range 35–104, t(20) = 9.66; p < 0.001). Limb volume difference further reduced to 3.8 % by 6 months postsurgery, an 89.6 % (range 38–149) reduction of presurgical volume (t(18) = 9.17; p < 0.001). This near-complete reduction was maintained to 12 months (n = 8), a 97.7 % reduction (range 73–123, t(8) = 5.73; p < 0.001).Table 2 Excess volume and L-Dex value pre- and postoperatively Preoperative 6 months 12 months 18 months Upper limb (n) 15 12 7 1 L-Dex [mean (range)] 41.2 (18–75) 35.3 (14–49) 25.1 (13–45) 27 Significance – p = 0.068 p = 0.018 – Mean excess volume [ml (range)] 1139.5 (645–1755) 67.9 (−697 to 422) 18.7 (−244 to 218) −339a Mean excess volume [% (range)] 44.2 (27–67) 3.6 (−21 to 21) 1.3 (−5 to 8) −11a

Preoperative 6 months 12 months 18 months Upper limb (n) 15 12 7 1 L-Dex [mean (range)] 41.2 (18–75) 35.3 (14–49) 25.1 (13–45) 27 Significance – p = 0.068 p = 0.018 – Mean excess volume [ml (range)] 1139.5 (645–1755) 67.9 (−697 to 422) 18.7 (−244 to 218) −339a Mean excess volume [% (range)] 44.2 (27–67) 3.6 (−21 to 21) 1.3 (−5 to 8) −11a Significance – p < 0.001 p = 0.001 – Lower limb (n) 6 5 1 0 L-Dex [mean (range)] 46.9 (12–97) 49.3 (33–71) 39.0 – Significance – p = 0.746 – – Mean excess volume [ml (range)] 4058 (2068–8294) 400 (−112 to 867) −103 – Mean excess volume [% (range)] 47.3 (23–83) 4.3 (−1 to 11) −1.0 – Significance – p = 0.018 – – Significance assessed using paired samples t tests compared with preoperative value aThis patient gained weight overall but increased in fat volume in the unaffected arm only, i.e. the affected arm is now smaller than the unaffected arm

Significance – p < 0.001 p = 0.001 – Lower limb (n) 6 5 1 0 L-Dex [mean (range)] 46.9 (12–97) 49.3 (33–71) 39.0 – Significance – p = 0.746 – – Mean excess volume [ml (range)] 4058 (2068–8294) 400 (−112 to 867) −103 – Mean excess volume [% (range)] 47.3 (23–83) 4.3 (−1 to 11) −1.0 – Significance – p = 0.018 – – Significance assessed using paired samples t tests compared with preoperative value aThis patient gained weight overall but increased in fat volume in the unaffected arm only, i.e. the affected arm is now smaller than the unaffected arm Mean preoperative L-Dex was 42.9 (range 12–97) for all patients. L-Dex increased 4 weeks postsurgery to 55.0 (range 32–73), reflecting the extracellular fluid associated with postsurgical swelling (t(18) = −2.51; p = 0.02). L-Dex was at presurgical values 6 months postsurgery (mean 38.1, range 14–71, t(14) = 1.68; p = 0.12) and reduced below presurgical values at 12 months (mean 27.1, range 13–45, t(7) = 3.38; p = 0.02). However, L-Dex values remained elevated above the ‘normal’ range (0 ± 10), likely indicating ongoing lymphatic pathology. Although these comparisons were not statistically significant (applying the Bonferroni correction), change from timepoint to timepoint always exceeded the 10 points considered clinically significant for L-Dex.

Dex values remained elevated above the ‘normal’ range (0 ± 10), likely indicating ongoing lymphatic pathology. Although these comparisons were not statistically significant (applying the Bonferroni correction), change from timepoint to timepoint always exceeded the 10 points considered clinically significant for L-Dex. Functionally, all patients reported improvements on the PSFS index of personally important activities (Table 3) by 6 months postsurgery (p < 0.01). Improvements were also evident in the standardized domains of pain, heaviness, self-consciousness, levels of anxiety, perceived degree of swelling, and emotional impact; such improvements were statistically significant, with the exception of pain in the lower limb and anxiety about the upper limb. There have been no surgical complications; one patient had poor compression garment compliance.Table 3 Functional and emotional impact of lymphedema before and after liposuction Preoperative 6 months Effect at 6-month follow-up Mean (range) Mean (range) N t p value PSFS functional impairmenta Upper limb 11.1 (4–21) 22.1 (9–30) 7 3.86 0.008 Lower limb 7.4 (4–9) 28.0 (27–29) 5 23.6 <0.001 Painb Upper limb 3.9 (0–8) 0.8 (0–3) 9 3.60 0.007 Lower limb 3.7 (0–8) 0.2 (0–1) 5 2.50 0.07 Heavinessb Upper limb 6.7 (3–10) 0.3 (0–2) 9 9.71 <0.001 Lower limb 8.2 (6–10) 0.4 (0–2) 5 7.65 0.002 Self-consciousnessb Upper limb 6.9 (2–10) 0.6 (0–3) 9 5.94 <0.001 Lower limb 8.2 (4–10) 0 5 24.59 <0.001 Anxiousb Upper limb 5.1 (0–10) 0.2 (0–2) 9 3.31 0.11 Lower limb 7.2 (5–10) 0 5 9.36 <0.001 Swollenb

Upper limb 3.9 (0–8) 0.8 (0–3) 9 3.60 0.007 Lower limb 3.7 (0–8) 0.2 (0–1) 5 2.50 0.07 Heavinessb Upper limb 6.7 (3–10) 0.3 (0–2) 9 9.71 <0.001 Lower limb 8.2 (6–10) 0.4 (0–2) 5 7.65 0.002 Self-consciousnessb Upper limb 6.9 (2–10) 0.6 (0–3) 9 5.94 <0.001 Lower limb 8.2 (4–10) 0 5 24.59 <0.001 Anxiousb Upper limb 5.1 (0–10) 0.2 (0–2) 9 3.31 0.11 Lower limb 7.2 (5–10) 0 5 9.36 <0.001 Swollenb Upper limb 6.9 (2–10) 1.8 (0–4) 9 5.49 <0.001 Lower limb 9.0 (8–10) 1.6 (0–2) 5 9.89 <0.001 Impact on emotionsb Upper limb 6.0 (0–10) 1.0 (0–4) 9 4.07 0.004 Lower limb 7.8 (2–10) 0.6 (0–3) 5 12.37 <0.001 Significance assessed using paired samples t tests compared with preoperative value PSFS Patient-Specific Functional Scale32 aScores ranged from ‘0’ (not able to perform three activities at all) to ‘30’ (able to perform three activities perfectly) bScores ranged from ‘0’ (not at all) to ‘10’ (extremely so) Discussion “The most emotional day was when I got my first garment on – it was mind blowing to see my ankle and calf so much the same size as my other leg. Everything in my life has now changed – for the better”, first leg liposuction patient (Fig. 1).

aScores ranged from ‘0’ (not able to perform three activities at all) to ‘30’ (able to perform three activities perfectly) bScores ranged from ‘0’ (not at all) to ‘10’ (extremely so) Discussion “The most emotional day was when I got my first garment on – it was mind blowing to see my ankle and calf so much the same size as my other leg. Everything in my life has now changed – for the better”, first leg liposuction patient (Fig. 1). A multidisciplinary team convened Australia’s first clinic (ALAC) providing liposuction for lymphedema. European protocols12,14,18,20,21 were applied within a unique multidisciplinary context, where the surgeon and occupational therapist were joined by an oncologist and physiotherapist, and led by a rehabilitation specialist. ALAC conducted surgery on cancer-related lymphedema and primary (congenital) lymphedema. Liposuction was effective in this privately-funded context, eliminating excess volume on average and improving symptoms and function in the affected limb. Patients maintained reductions to 12 months by continuously wearing compression garments.

d surgery on cancer-related lymphedema and primary (congenital) lymphedema. Liposuction was effective in this privately-funded context, eliminating excess volume on average and improving symptoms and function in the affected limb. Patients maintained reductions to 12 months by continuously wearing compression garments. These results are comparable with international standards. Arm liposuction achieved mean volume reductions of 90 % at 6 months and 97 % at 12 months postsurgery compared with 103 and 111 % in Sweden,38 and 92 and 101 % (range 69–148) in the UK.13 Results 12 months postsurgery were 116 % (range 75–233) in The Netherlands14 and 111 % (range 90–130) in the US.24 Leg surgery achieved volume reductions of 88 % at 6 months and 101 % (one patient) at 12 months postsurgery compared with 84 % at 3 months and 105 % at 12 months postsurgery in Sweden,20 and 86 % (range 81–97) 12 months postsurgery in the US.24 Previous liposuction research demonstrated improved quality of life3 and overall wellbeing,13 but disagreed about improvements in emotional wellbeing.3,13 In ALAC, patients reported substantial and statistically significant reduction in lymphedema impact on important activities, improved limb function, and reduced lymphedema-specific emotional distress. Thus, liposuction has substantial functional and psychological benefit.

about improvements in emotional wellbeing.3,13 In ALAC, patients reported substantial and statistically significant reduction in lymphedema impact on important activities, improved limb function, and reduced lymphedema-specific emotional distress. Thus, liposuction has substantial functional and psychological benefit. These results are preliminary owing to the small number of patients. Nevertheless, they are clinically and statistically significant despite the expense associated with self-funding compression garments and the discomfort of wearing garments continuously in Australia’s hot climate. Liposuction addresses physical swelling but not the underlying lymphatic dysfunction; therefore, patients must maintain garment use to continue to benefit from the surgery. However, we have anecdotes that some patients are wearing compression intermittently yet continuing to maintain reduction. ALAC is planning to explore whether less burdensome garment requirements are possible through a randomized controlled trial of durations and levels of compression after liposuction for upper limb lymphedema.

e have anecdotes that some patients are wearing compression intermittently yet continuing to maintain reduction. ALAC is planning to explore whether less burdensome garment requirements are possible through a randomized controlled trial of durations and levels of compression after liposuction for upper limb lymphedema. All patients experienced benefit and there were no adverse events. This is consistent with lymphoscintigraphy studies demonstrating that liposuction is not associated with further damage to lymphatic transport in arms with lymphedema,39 and with a cadaveric study which demonstrated that longitudinal liposuction does not damage the epifascial lymph vessels.40 These results are early indications that liposuction is safe and will not additionally compromise lymphatic drainage, either in primary or secondary lymphedema. Although the treatment protocol at ALAC was similar to others,12,14,18,20,21 the clinic composition was expanded. In addition to a surgeon and occupational therapist/physiotherapist, the lead clinician was a rehabilitative specialist, i.e. liposuction was seen as the first step in ongoing nonsurgical management, including compression garment use requiring regular reordering.13,21,24 In addition, an oncologist was present to balance cosmesis, function, and quality of life against prognosis, if necessary. ALAC holds the conviction that the highest standard of care is achieved within this multidisciplinary environment.

anagement, including compression garment use requiring regular reordering.13,21,24 In addition, an oncologist was present to balance cosmesis, function, and quality of life against prognosis, if necessary. ALAC holds the conviction that the highest standard of care is achieved within this multidisciplinary environment. Future research should determine selection criteria and sequencing for other surgical procedures. Our clinic is evaluating LNT7,41 for patients with ISL stage I or II lymphedema where pitting edema rather than fat is the clinical presentation. LNT+ delayed autologous breast reconstruction is another potential approach for patients undergoing a mastectomy.24–26 Furthermore, surgeries might be combined, such as liposuction to remove fatty lymphedema followed by LNT± autologous breast reconstruction to eliminate the need for compression garments. The appropriate surgical approach can be defined using lymphedema clinical characteristics;24 ALAC holds the additional belief that patient factors such as motivation and the ability to pay for compression garments must be assessed.

y LNT± autologous breast reconstruction to eliminate the need for compression garments. The appropriate surgical approach can be defined using lymphedema clinical characteristics;24 ALAC holds the additional belief that patient factors such as motivation and the ability to pay for compression garments must be assessed. Conclusions With continued compression garment use, ALAC expects patients to maintain limb reductions as reported in 5- to 15-year follow-up in Europe.13,20,38 Only one patient has failed to maintain compliance for garment use, for unknown reasons. However, these Australian patients were carefully selected with regard to their long-term ability to pay for garments. Costs prevented 12 eligible patients from undergoing surgery. Considering the significant functional gains observed in this and other studies,3,13,24 and reduced hospitalizations due to infection,8,9,22 governments and insurance companies should consider the economic value of funding liposuction for advanced lymphedema. Disclosures John Boyages, Katrina Kastanias, Louise A. Koelmeyer, Caleb J .Winch, Thomas C. Lam, Kerry A. Sherman, David Alex Munnoch, Håkan Brorson, Quan D. Ngo, Asha Heydon-White, John S. Magnussen, and Helen Mackie have no commercial interests in the subject of the study.

Aggressive fibromatosis (AF; or desmoid-type fibromatosis) is a rare soft-tissue tumor that lacks the capacity to metastasize but may behave in a locally aggressive fashion. Knowledge on its epidemiology and etiology is limited. The Wingless/Wnt-pathway is involved although the mechanism is not fully understood.1–3 Three different subtypes are recognized as entities in the WHO-classification of desmoid-type fibromatose: extra-abdominal, abdominal, and intra-abdominal tumors.4 The first two mostly occur sporadic, whereas the latter has a correlation with familiar adenomatous polyposis (FAP).5

mechanism is not fully understood.1–3 Three different subtypes are recognized as entities in the WHO-classification of desmoid-type fibromatose: extra-abdominal, abdominal, and intra-abdominal tumors.4 The first two mostly occur sporadic, whereas the latter has a correlation with familiar adenomatous polyposis (FAP).5 The incidence of AF was reported previously by Reitamo et al. in 1982, estimated at 2.4–4.3 per million people per year.6 Their studies on the etiology and epidemiology often are referred to in the current literature.6–8 The correlation of intra-abdominal AF with FAP has been subject of more recent studies.9–11 Current research on AF mainly focuses on treatment strategies. Surgery has until recently been the primary treatment modality. Data regarding the prognostic value of surgical margins and adjuvant radiotherapy is conflicting.12–15 New insights suggest that asymptomatic patients can be carefully watched without active treatment, and this is suggested by international (NCCN and ESMO) guidelines.16,17 Symptomatic patients with tumors that can be resected completely with acceptable morbidity should be offered surgery. In patients with symptomatic and “unresectable” disease, radiotherapy may be considered.18 Isolated limb perfusion can be considered for irresectable AF of the extremities.19 Systemic treatment also can be considered, although response rates are rather low.20–22 We evaluated time trends of the incidence and treatment of extra-abdominal and abdominal wall AF within the Dutch population. Methods Data Collection

The incidence of AF was reported previously by Reitamo et al. in 1982, estimated at 2.4–4.3 per million people per year.6 Their studies on the etiology and epidemiology often are referred to in the current literature.6–8 The correlation of intra-abdominal AF with FAP has been subject of more recent studies.9–11 Current research on AF mainly focuses on treatment strategies. Surgery has until recently been the primary treatment modality. Data regarding the prognostic value of surgical margins and adjuvant radiotherapy is conflicting.12–15 New insights suggest that asymptomatic patients can be carefully watched without active treatment, and this is suggested by international (NCCN and ESMO) guidelines.16,17 Symptomatic patients with tumors that can be resected completely with acceptable morbidity should be offered surgery. In patients with symptomatic and “unresectable” disease, radiotherapy may be considered.18 Isolated limb perfusion can be considered for irresectable AF of the extremities.19 Systemic treatment also can be considered, although response rates are rather low.20–22 We evaluated time trends of the incidence and treatment of extra-abdominal and abdominal wall AF within the Dutch population. Methods Data Collection The Dutch Pathology Registry PALGA was searched for patients with extra-abdominal or abdominal AF, whereas patients with intra-abdominal tumors were excluded.23 The epidemiology and treatment of intra-abdominal tumors are linked to FAP and are considered a different entity. Data on this entity in the Dutch population have been analyzed recently.9 The PALGA database contains encoded excerpts of all pathology examinations obtained by a diagnostic procedure, including tissue biopsy or resection since 1979 in selected laboratories and expanded to nationwide inclusion in 1991. The conclusion sections of all pathology reports were queried for available information concerning patient, tumor, and treatment characteristics. Age was categorized as <20, 20–44, 45–64, 65–79, and >80 years old. Tumor localization was categorized as head/neck, trunk (including breast, thoracic aperture and back), abdominal wall, extremity, and others. Reports were scored based on the encoding of procedures and details in the report as biopsy, resection or re-resection and on manifestation of the tumor (primary or recurrence). All patients undergoing re-resection were considered to have had a prior resection, even when pathology reports of the resection were missing. In case of patient records documenting recurrent disease, an attempt was made to retrieve details on the primary tumor. Due to incomplete data registration, patients with disease presentation before 1993 were excluded. The years of diagnoses were categorized as 1993–1998, 1999–2003, 2004–2008, and 2009–2013.

e missing. In case of patient records documenting recurrent disease, an attempt was made to retrieve details on the primary tumor. Due to incomplete data registration, patients with disease presentation before 1993 were excluded. The years of diagnoses were categorized as 1993–1998, 1999–2003, 2004–2008, and 2009–2013. The primary objective was to analyze time trends in the incidence of AF. Trends of clinicopathological factors were analyzed as well as possible associations between the factors. The secondary objective was to analyze time trends in type of treatment, to which end the rate of resection was evaluated. Due to constrains in the pathology database structure, only data on pathology specimens, such as biopsy of resection were available. Information on other treatment strategies or outcome was not available. In order to compare the patient cohort with the Dutch population, data from Statistics Netherlands were obtained. This is a registry for all general population data. We used information on demographics to calculate annual incidence rates and information on surgical treatments, hormonal drugs, and newborns to analyze possible etiological correlations.

cohort with the Dutch population, data from Statistics Netherlands were obtained. This is a registry for all general population data. We used information on demographics to calculate annual incidence rates and information on surgical treatments, hormonal drugs, and newborns to analyze possible etiological correlations. Statistical Analysis Statistical analysis was performed using IBM SPSS Statistics 21. Continuous variables are shown as median and interquartile range (IQR), and categorical variables as numbers with percentages. Associations between clinicopathological variables were determined by χ2 analysis. Univariate logistic and linear regression analysis was performed to analyze trends over time. Results are shown as odds ratios (OR) or regression coefficient B (B) and with 95 % confidence intervals (CI). For all analyses, two-sided P < 0.050 was considered statistically significant. Results A total of 1134 patients were diagnosed with extra-abdominal or abdominal wall AF between January 1993 and December 2013; there were 326 men and 808 women. Median age was 37 years [interquartile range (IQR) 30–50]. The distribution of demographic factors is shown in Table 1.Table 1 Distribution of epidemiologic factors 1993–1998 1999–2003 2004–2008 2009–2013

Results A total of 1134 patients were diagnosed with extra-abdominal or abdominal wall AF between January 1993 and December 2013; there were 326 men and 808 women. Median age was 37 years [interquartile range (IQR) 30–50]. The distribution of demographic factors is shown in Table 1.Table 1 Distribution of epidemiologic factors 1993–1998 1999–2003 2004–2008 2009–2013 N % N % N % N % Gender Male 56 31.1 50 27.0 105 31.7 115 26.3 Female 124 68.9 135 73.0 226 68.3 323 73.7 Age (year) <20 18 10.0 14 7.6 29 8.8 39 8.9 20–44 115 63.9 124 67.0 170 51.4 239 54.6 45–64 37 20.6 33 17.8 85 25.7 112 25.6 65–79 10 5.6 11 5.9 39 11.8 43 9.8 80+ 0 0 3 1.6 8 2.4 5 1.1 Localization Head/neck 14 8.0 13 7.1 20 6.1 27 6.2 Trunk 29 16.7 39 21.4 102 30.9 152 34.8 Abdominal wall 77 44.3 88 48.4 113 34.2 151 34.6 Extremity 45 25.9 32 17.6 68 20.6 85 19.5 Other 7 4.0 6 3.3 22 6.7 22 5.0 Unknown 2 1.1 4 2.2 5 1.5 0 0 Pathology reports Biopsy 13 7.2 39 21.1 69 20.8 130 29.7 Biopsy + resection 39 21.7 40 21.6 98 29.6 161 36.8 Resection 114 63.3 101 54.6 163 49.2 147 33.6 Unknown 14 7.8 5 2.7 1 0.3 0 0 In addition to the 1134 patients diagnosed as having AF, an uncertain diagnosis of AF was stated in the pathology excerpt in 213 patients. This latter group of patients did not change significantly over the years (P = 0.730). These patients were not included in the analyses for the present series.

N % N % N % N % Gender Male 56 31.1 50 27.0 105 31.7 115 26.3 Female 124 68.9 135 73.0 226 68.3 323 73.7 Age (year) <20 18 10.0 14 7.6 29 8.8 39 8.9 20–44 115 63.9 124 67.0 170 51.4 239 54.6 45–64 37 20.6 33 17.8 85 25.7 112 25.6 65–79 10 5.6 11 5.9 39 11.8 43 9.8 80+ 0 0 3 1.6 8 2.4 5 1.1 Localization Head/neck 14 8.0 13 7.1 20 6.1 27 6.2 Trunk 29 16.7 39 21.4 102 30.9 152 34.8 Abdominal wall 77 44.3 88 48.4 113 34.2 151 34.6 Extremity 45 25.9 32 17.6 68 20.6 85 19.5 Other 7 4.0 6 3.3 22 6.7 22 5.0 Unknown 2 1.1 4 2.2 5 1.5 0 0 Pathology reports Biopsy 13 7.2 39 21.1 69 20.8 130 29.7 Biopsy + resection 39 21.7 40 21.6 98 29.6 161 36.8 Resection 114 63.3 101 54.6 163 49.2 147 33.6 Unknown 14 7.8 5 2.7 1 0.3 0 0 In addition to the 1134 patients diagnosed as having AF, an uncertain diagnosis of AF was stated in the pathology excerpt in 213 patients. This latter group of patients did not change significantly over the years (P = 0.730). These patients were not included in the analyses for the present series. Epidemiologic Factors The incidence of extra-abdominal and abdominal wall AF increased over the study period, from 2.10 to 5.36 per one million people (P < 0.001; Fig. 1).Fig. 1 Incidence of aggressive fibromatosis, per million people

In addition to the 1134 patients diagnosed as having AF, an uncertain diagnosis of AF was stated in the pathology excerpt in 213 patients. This latter group of patients did not change significantly over the years (P = 0.730). These patients were not included in the analyses for the present series. Epidemiologic Factors The incidence of extra-abdominal and abdominal wall AF increased over the study period, from 2.10 to 5.36 per one million people (P < 0.001; Fig. 1).Fig. 1 Incidence of aggressive fibromatosis, per million people Age The median age increased annually by B 0.285 (95 % CI 0.114–0.455; P = 0.001). The median age in 1993–1998 was 34 years (range 27–45) and was 39 years (range 30–51) in 2009–2013. The absolute numbers increased in all age groups over time (Fig. 2a). However, the percentage of patients per age groups changed, mostly in patients aged 20–79 years (Fig. 2b). Analysis of the distribution among age groups showed a significant annual decrease in the percentage of patients aged 20–45 years (OR 0.977; 95 % CI 0.957–0.997; P = 0.027) and a trend towards an annual increase in the percentage of patients aged 45–65 years and 65–80 years (OR 1.017; 95 % CI 0.993–1.042; P = 0.173 and OR 1.035; 95 % CI 0.997–1.074; P = 0.069 respectively).Fig. 2 Distribution among age and localization. a Distribution among age during study period. b Percentage of age distribution. c Distribution among localization during study period. d Distribution of localization per age group

7; 95 % CI 0.993–1.042; P = 0.173 and OR 1.035; 95 % CI 0.997–1.074; P = 0.069 respectively).Fig. 2 Distribution among age and localization. a Distribution among age during study period. b Percentage of age distribution. c Distribution among localization during study period. d Distribution of localization per age group Gender The absolute numbers of both male and female patients increased over the years. The male–female ratio showed an increasing female predominance, ranging from 68.6 % in 1993–1998 to 73.6 % in 2009–2013. Anatomic Tumor Localization Tumor localization was distributed as: 6.7 % head/neck, 29.0 % trunk, 38.6 % abdominal wall, 20.7 % extremity, and 5.1 % other (localization details were missing for 22 patients). Over the years, the absolute incidence in all groups increased (Fig. 2c). Analysis of the distribution of tumor localization showed a significant proportional increase in the percentage of patients with truncal localization (OR 1.057; 95 % CI 1.032–1.083; P < 0.001), whereas the percentage of patients with tumors in the abdominal wall decreased (OR 0.972; 95 % CI 0.952–0.993; P = 0.008).

c). Analysis of the distribution of tumor localization showed a significant proportional increase in the percentage of patients with truncal localization (OR 1.057; 95 % CI 1.032–1.083; P < 0.001), whereas the percentage of patients with tumors in the abdominal wall decreased (OR 0.972; 95 % CI 0.952–0.993; P = 0.008). Associations Between Clinicopathological Factors The distribution of tumor localization varied per age group (Fig. 2d). Extremity-based tumors were most common in patients younger than 20 years of age (45.0 %), whereas patients between 20 and 45 years most commonly harbored abdominal wall tumors (52.6 %); truncal tumors were predominantly seen in patients between 45 and 80 years of age (41.5 %). For patients older than 80 years of age, no dominant localization could be identified. The distribution of age groups and localization changed over the study period.

5 years most commonly harbored abdominal wall tumors (52.6 %); truncal tumors were predominantly seen in patients between 45 and 80 years of age (41.5 %). For patients older than 80 years of age, no dominant localization could be identified. The distribution of age groups and localization changed over the study period. Workup and Treatment In 251 patients (22.1 %) solely a biopsy report was retrieved; for 338 patients (29.8 %) a biopsy report and a pathology resection specimen report was retrieved, and for 525 patients (46.3 %) solely a pathology resection specimen report was retrieved. For 20 patients, the type of report was unknown (Fig. 3). From 1993–1998 to 2008–2013, the biopsy rate increased more than twofold: from 31.1 to 66.4 % (OR 1.096; 95 % CI 1.072–1.121, P < 0.001). The proportion of patients who underwent surgical resection decreased annually (OR 0.928; 95 % CI 0.902–0.954, P < 0.001). It was not known what treatment was offered to the patients who did not undergo surgery due to the nature of the database. Over time, surgical resection was increasingly preceded by biopsy. If a resection was preceded by biopsy, the resection margin status improved significantly (49.8 % R0-resection vs. 30.7 % in patients without biopsy; P < 0.001). Pathology reports did not discriminate between diagnostic or therapeutic resections. Median time between biopsy and resection was 1.6 months (IQR 0.9–2.7). The date of either biopsy or resection was missing for two patients. A substantial number of patients (210; 18.5 %) had a history of surgery in the same area where AF subsequently developed.Fig. 3 Type of pathology records per patient

ic resections. Median time between biopsy and resection was 1.6 months (IQR 0.9–2.7). The date of either biopsy or resection was missing for two patients. A substantial number of patients (210; 18.5 %) had a history of surgery in the same area where AF subsequently developed.Fig. 3 Type of pathology records per patient Dutch Population Since the abdominal wall was the most common tumor localization, we analyzed surgical trends in the Netherlands for the most common surgeries in this area (caesarean section, cholecystectomy, appendectomy, and colectomy).24 During the study period, surgical trauma to the abdominal wall increased (Figs. 4, 5). Due to minimal invasive techniques for many surgical interventions, the rate of laparotomy decreased and the rate of laparoscopic surgery increased.Fig. 4 Absolute number of most common abdominal wall surgery, in relation to the absolute number of patients with abdominal wall AF (on secondary axis) Fig. 5 Abdominal surgery in the Netherlands Data on hormonal drugs was available for the period 2006–2012. During this period, the overall use of hormonal medication in the Netherlands remained stable. The number of pregnancies of any gestational age was not available. The number of newborns per year was used as a surrogate, and during the study period this number decreased from 195.748 in 1993 to 171.341 in 2013. Discussion

Data on hormonal drugs was available for the period 2006–2012. During this period, the overall use of hormonal medication in the Netherlands remained stable. The number of pregnancies of any gestational age was not available. The number of newborns per year was used as a surrogate, and during the study period this number decreased from 195.748 in 1993 to 171.341 in 2013. Discussion The reference standard on the incidence and epidemiology of AF are Finnish studies by Reitamo et al.6–8 An incidence of 2.4–4.3 per million people was reported in those studies, using three methods of estimation (local, regional, and national). Distribution of disease was reported with a dominance of abdominal wall tumors (49 %) with variations per age groups. In the present population based study, a rising incidence of extra-abdominal and abdominal wall AF was observed from 2.1 to 5.36 per million people during the period 1993–2013. The distribution among age groups was similar to the Finnish studies, with a predominance of abdominal wall tumors in females aged 20–44 years. Remarkably, median age and female predominance increased over the years and the distribution of tumor localization shifted. The driving factor for these observed changes is unclear.

istribution among age groups was similar to the Finnish studies, with a predominance of abdominal wall tumors in females aged 20–44 years. Remarkably, median age and female predominance increased over the years and the distribution of tumor localization shifted. The driving factor for these observed changes is unclear. The PALGA database provided an elaborate overview of AF in the Netherlands. The nationwide coverage enabled epidemiological research on this rare disease. Then again, the available information was limited to the date and conclusion of the pathology reports. Although there was information on biopsy and resection, no information was available for nonsurgical treatments, which is a limitation of the present study. Still, important information could be extracted. Time Trends in Incidence Explanations for the observed rising incidence of AF are not evident. If an increase in incidence occurs, this can be due to improved diagnostic modalities (i.e., for instance detection of previously unrecognized tumors by improved imaging, improved recognition of the disease by pathologists, or the start of a screening program) or due to a true increase in the incidence of the disease. Improved registration and diagnostic tools are likely to have influenced the incidence figures to some extent. The changes in distribution of tumor localization might be an indication for a true change in disease. However, there are possible biases: other reasons could be an increased frequency of trunk computer-tomography scan or higher awareness due to screening programs.

have influenced the incidence figures to some extent. The changes in distribution of tumor localization might be an indication for a true change in disease. However, there are possible biases: other reasons could be an increased frequency of trunk computer-tomography scan or higher awareness due to screening programs. Dutch guidelines on registration of neoplasms have changed over the years. The introduction of the third edition of the WHO Classification for Soft Tissue and Bone Tumours stimulated improvement of coding, enabling a better pathology registration.25 Due to the benign nature, this neoplasm is not registered among soft tissue tumors in the national cancer registries precluding verification of our data. The overall incidence of sarcomas has remained stable over the years at approximately 30–35 patients per million people, with a slight increase to around 40 patients per million people over the past 5 years.26 Knowledge on β-catenin and its application in the diagnostic setting around 2005 aided the pathologist in diagnosing AF with more confidence.27–29 Nevertheless, the percentage of uncertain diagnoses has not changed significantly over the years, indicating that some difficulty to distinguish AF from low-grade and reactive spindle cell proliferations remains. Awareness of the presence of AF and the realization of the importance of a correct diagnosis have improved. In addition, the association with FAP is better understood. Lastly, screening programs may have influenced the stage of diagnosis, such as the breast cancer screening program in asymptomatic people.

ions remains. Awareness of the presence of AF and the realization of the importance of a correct diagnosis have improved. In addition, the association with FAP is better understood. Lastly, screening programs may have influenced the stage of diagnosis, such as the breast cancer screening program in asymptomatic people. Documented etiological factors are surgical trauma, hormonal influences, and pregnancy.6–8 National data on these factors was obtained to provide some context for the study data. A hypothesis could be that the increased rate of surgical trauma would lead to an increase in AF. On the contrary, a limitation of surgical trauma by means of minimal invasive techniques could possibly decrease the risk of AF. The analyses of abdominal surgery and abdominal AF both showed increasing rates over the study period, which might be supportive of the first hypothesis. The peak in occurrence of AF among fertile females is supportive of hormonal influences as an etiological factor. To test the hypothesis that a rise in hormonal levels would lead to an increase in AF, we compared data on hormonal drug use from Statistics Netherlands with the data from PALGA. Although the information on drug use was from a small period (2006–2012), the incidence of AF was rising during this period while the rate of hormonal drug use remained stable.

in hormonal levels would lead to an increase in AF, we compared data on hormonal drug use from Statistics Netherlands with the data from PALGA. Although the information on drug use was from a small period (2006–2012), the incidence of AF was rising during this period while the rate of hormonal drug use remained stable. Pregnancy is seen as an etiological factor within the hormonal influences. Because no data on pregnancies in the patient cohort were available, we obtained the rate of newborns in the Netherlands during the study period. The rate of pregnancies of any gestational age was not available. The hypothesis that an increase in pregnancies (represented by the number of newborns) would lead to an increase in AF was not supported, as the rate of newborns was decreasing. A more sensitive approach to test hormonal influences on AF, like analyzing hormonal receptors on the tumor, could provide more information but was not possible for the current study. We would like to emphasize that the presented comparisons between data from PALGA and Statistics Netherlands are all based on hypotheses. Direct correlations for these etiological factors could not be explored and possible biases should be taken into consideration. Time Trends in Diagnosis and Treatment Despite the aforementioned advances in diagnostic tools, the diagnosis of AF poses remaining challenges to the treating physicians. Although the rising incidence is most likely biased by diagnostic modalities and improved registration, the presented results showed an increasing number of patients being treated for AF.

Despite the aforementioned advances in diagnostic tools, the diagnosis of AF poses remaining challenges to the treating physicians. Although the rising incidence is most likely biased by diagnostic modalities and improved registration, the presented results showed an increasing number of patients being treated for AF. The presented results suggest an improved workup procedure of patients as histological biopsies were more often obtained. Surgical resection following a biopsy diagnosis resulted in a significant higher rate of negative resection margins, underscoring the importance of the diagnostic process. Treatment strategies changed in recent years and this is reflected in the present data. There has been a paradigm shift in the surgical treatment for AF patients. Before 2000, surgery with negative margins had been considered the standard of care for patients affected by AF, reflecting the same approach to extremity soft-tissue sarcomas. A reassessment has taken place by several groups, advocating a more conservative approach.30,31 The European consensus is currently set at an initial wait-and-see approach.32 The increasing number of patients undergoing nonsurgical treatment in the presented study indicated a tendency to adhere to this policy in the Netherlands. The growing knowledge and understanding of the etiology and involvement of CTNNB1-mutations will improve the diagnostic process.

an initial wait-and-see approach.32 The increasing number of patients undergoing nonsurgical treatment in the presented study indicated a tendency to adhere to this policy in the Netherlands. The growing knowledge and understanding of the etiology and involvement of CTNNB1-mutations will improve the diagnostic process. During the past 25 years, developments in the available diagnostic modalities and changing treatment insights had an impact on the workup and treatment of extra-abdominal and abdominal wall AF. More insight in current epidemiologic trends and treatment-related trends was imperative. This population-based study reflected these changes and showed an overall incidence rise of AF. The reasons for the changing incidence, age distribution, and anatomic localization distribution remain to be further elucidated.

Colorectal cancer is one of the leading causes of cancer death worldwide as a result of its considerable risk of development of metastases.1 When metastatic disease is confined to the liver, partial liver resection is the only curative therapeutic option, with 5-year overall survival (OS) percentages between 20 and 60 %, depending on patient and tumor characteristics.2–4 In order to explain these varying survival rates, different clinicopathologic risk scores have been developed. In many of these risk scores, nodal status of the primary tumor, size and number of the colorectal liver metastases (CRLM), disease-free interval from treatment of the primary until detection of the CRLM, and preoperative level of carcinoembryonic antigen (CEA) are combined to predict long-term survival.5–9 These scoring systems are relevant with respect to prediction of survival, but to our knowledge, they have not been used for risk stratification in controversial areas such as the administration of neoadjuvant or adjuvant systemic therapy or surveillance.

nic antigen (CEA) are combined to predict long-term survival.5–9 These scoring systems are relevant with respect to prediction of survival, but to our knowledge, they have not been used for risk stratification in controversial areas such as the administration of neoadjuvant or adjuvant systemic therapy or surveillance. In primary colorectal cancer histologic factors such as extramural venous invasion, perineural growth, lymphatic invasion, angioinvasion, and diffuse growth pattern have been associated with poorer survival outcomes.10,11 Extramural venous invasion in particular is considered a poor prognostic factor, and as a result, patients with extramural venous invasion in stage II colon cancer are considered candidates for adjuvant systemic treatment.12 Very little is known about the impact of histologic features of colorectal liver metastases on OS, as described in a recent review.13

ar is considered a poor prognostic factor, and as a result, patients with extramural venous invasion in stage II colon cancer are considered candidates for adjuvant systemic treatment.12 Very little is known about the impact of histologic features of colorectal liver metastases on OS, as described in a recent review.13 Vascular invasion, bile duct invasion, or lymphatic invasion by tumor cells in CRLM have all been suggested as prognostic factors for long-term survival.5,14–23 Perineural growth, the presence of a fibrous capsule, and tumor thickness at the tumor–normal interface have also been linked to survival in patients with CRLM.14,15,19,24–26 Variations in definitions and selection of patients have limited the impact of these studies. Furthermore, none of these previous studies has evaluated multiple histologic factors of the liver resection specimens, in combination with established risk scores in a homogenous group of patients. Most studies included patients who underwent neoadjuvant therapy as well as chemotherapy-naive patients, patients with multiple liver metastases, or patients with extrahepatic disease.5,14–21,23,24 The results of these previous studies might be biased because of the known changes in histologic features observed in liver metastases after systemic therapy, and the possible heterogeneous nature of multiple metastases.27–30

, patients with multiple liver metastases, or patients with extrahepatic disease.5,14–21,23,24 The results of these previous studies might be biased because of the known changes in histologic features observed in liver metastases after systemic therapy, and the possible heterogeneous nature of multiple metastases.27–30 The objective of the current study was to assess possible prognostic histologic factors for long-term survival in patients with solitary colorectal liver metastases who underwent a complete (R0) liver without neoadjuvant systemic therapy.

, patients with multiple liver metastases, or patients with extrahepatic disease.5,14–21,23,24 The results of these previous studies might be biased because of the known changes in histologic features observed in liver metastases after systemic therapy, and the possible heterogeneous nature of multiple metastases.27–30 The objective of the current study was to assess possible prognostic histologic factors for long-term survival in patients with solitary colorectal liver metastases who underwent a complete (R0) liver without neoadjuvant systemic therapy. Materials and Methods Patients Patients were identified who underwent complete (R0) liver resection for a solitary CRLM between 1992 and 2011 in a tertiary referral hospital. R0 resections were defined as liver resections with clear resection margins in patients who did not have evidence of disease in any other locations. Demographics and clinicopathologic factors with regard to the primary tumor, as well as the liver metastasis, were collected per patient. Special attention was given to the four different items from the clinical risk score according to Fong et al.: nodal status of the primary tumor; preoperative CEA level and size of the metastasis, and interval between resection of the primary tumor and diagnosis of CRLM.9 It is unknown whether systemic treatment influences the presence of certain histopathology factors and therefore patients who were treated with neoadjuvant systemic therapy were excluded from the current study. Patients who died from postoperative complications, defined as within 30 days after liver resection, were also excluded. Patients underwent follow-up according to our current Dutch follow-up guidelines, with regular outpatient visits, CEA testing and computed tomographic scans of chest and abdomen.

m the current study. Patients who died from postoperative complications, defined as within 30 days after liver resection, were also excluded. Patients underwent follow-up according to our current Dutch follow-up guidelines, with regular outpatient visits, CEA testing and computed tomographic scans of chest and abdomen. Histopathology R0 liver resection specimens with a solitary CRLM were selected from the archive. Routine workup consisted of sampling of macroscopically normal liver tissue, invasive front of the metastasis, and additional tumor blocks, depending on the size of metastasis. Slide revision was performed independently by two investigators (JdR, NK). Discrepancies were resolved by simultaneous reexamination of the slides by both investigators using a two-headed microscope. In case of discrepancy, the senior pathologist (IN) made the final call.

or blocks, depending on the size of metastasis. Slide revision was performed independently by two investigators (JdR, NK). Discrepancies were resolved by simultaneous reexamination of the slides by both investigators using a two-headed microscope. In case of discrepancy, the senior pathologist (IN) made the final call. Tumor thickness at the tumor–normal interface was determined in routine slides. Tumor–normal interface was defined as the interface between tumor and normal liver tissue, as described by Maru et al. and validated by others.26,31,32 In all tumors, tumor thickness was measured with a ruler at multiple foci, and maximum tumor thickness was used and defined as uninterrupted layers of tumor cells without admixed fibrotic stroma, acellular mucin, or nonneoplastic liver parenchyma. The median tumor thickness at tumor–normal interface was used to divide the patient group in a group with a larger and a smaller layer of vital tumor cells (Fig. 1).Fig. 1 Tumor thickness at tumor–normal interface; arrow indicates correct measurement with uninterrupted layer of tumor cells. Original magnification, ×10

The median tumor thickness at tumor–normal interface was used to divide the patient group in a group with a larger and a smaller layer of vital tumor cells (Fig. 1).Fig. 1 Tumor thickness at tumor–normal interface; arrow indicates correct measurement with uninterrupted layer of tumor cells. Original magnification, ×10 The presence of a fibrotic capsule around the metastasis was evaluated in routine slides. The fibrous tissue between tumors and liver parenchyma was classified as absent (no fibrous tissue observed) or present: tumor was separated from the liver parenchyma by several layers of collagen bundles in histologic sections (Fig. 2).Fig. 2 a Colorectal liver metastasis without fibrous capsule. Original magnification, ×20. b Colorectal liver metastasis with fibrous capsule (arrow). Original magnification, ×20 Immunohistochemistry and Scoring Methods Immunohistochemistry was performed as previously described.33 Antibodies, clones, dilution, and retrieval methods are summarized in Supplementary Table 1. Perineural growth was defined as a nerve, identified by S-100 staining, being surrounded by tumor cells for at least three quarters of the circumference and was scored as being present or absent (Fig. 3a).Fig. 3 Different forms of intrahepatic invasion by tumor cells. a Perineural growth showing S-100 reactivity. b Lymphatic invasion showing D2-40 reactivity. c Vascular invasion showing CD-31 reactivity. d Bile duct invasion showing CK-7 reactivity. Original magnification, ×20

scored as being present or absent (Fig. 3a).Fig. 3 Different forms of intrahepatic invasion by tumor cells. a Perineural growth showing S-100 reactivity. b Lymphatic invasion showing D2-40 reactivity. c Vascular invasion showing CD-31 reactivity. d Bile duct invasion showing CK-7 reactivity. Original magnification, ×20 Lymphatic invasion was defined as single tumor cells or cell clusters visible within vessels that showed immunoreactivity for D2-40 but not for CD31. Lymphatic invasion was scored as being present or absent (Fig. 3b). Vascular invasion was defined as single tumor cells or cell clusters visible within vessels that showed immunoreactivity for CD31 but not for D2-40. It was scored as being present or absent (Fig. 3c). Bile duct invasion was defined as single tumor cells or cell clusters (CK7 negative) visible within bile ducts that showed immunoreactivity for CK7. It was also scored as being present or absent (Fig. 3d). Outcome Primary outcomes were disease-free survival (DFS) and OS. DFS was defined as the interval in months between liver resection and disease recurrence, death, or last follow-up. OS was defined as the interval in months between liver resection and death or date of last follow-up.

Bile duct invasion was defined as single tumor cells or cell clusters (CK7 negative) visible within bile ducts that showed immunoreactivity for CK7. It was also scored as being present or absent (Fig. 3d). Outcome Primary outcomes were disease-free survival (DFS) and OS. DFS was defined as the interval in months between liver resection and disease recurrence, death, or last follow-up. OS was defined as the interval in months between liver resection and death or date of last follow-up. Statistical Analysis Pearson’s Chi square test was used to calculate correlations between the various histologic features. Survival curves were estimated by the Kaplan–Meier method and compared by log rank testing. Multivariate analysis was performed using Cox proportional hazard model, and variables were included that were associated with survival in univariate analysis with a p value of <0.10. SPSS statistical software, version 18.0 (IBM, Armonk, NY, USA) was used for all statistical analysis. A p value of <0.05 was considered statistically significant.

performed using Cox proportional hazard model, and variables were included that were associated with survival in univariate analysis with a p value of <0.10. SPSS statistical software, version 18.0 (IBM, Armonk, NY, USA) was used for all statistical analysis. A p value of <0.05 was considered statistically significant. Results Patients Between January 1992 and March 2011, a total of 383 patients underwent liver resection for metastatic disease. After excluding patients with multiple metastases, 135 patients remained who were surgically treated (R0) for solitary CRLM. Eleven patients were excluded because they received neoadjuvant chemotherapy (n = 5), were lost to follow-up (n = 2), or died within 30 days after liver resection (n = 4). A total of 124 patients were eligible to be included in the current study, 76 men (61.3 %) and 48 women (38.7 %). Median age at time of resection was 64 years (range 40–80 years). Liver metastasis were detected at a median of 8.8 months (range 0–82 months) after resection of the primary tumor. Median size of the metastasis was 35 mm (range 10–130 mm). Median follow-up was 41 months (range 1–232 months). In the complete study population, median DFS was 28 months (range 1–228 months) with a median OS of 57 months (range 1–232 months) and a 5-year survival of 48.1 %.

onths) after resection of the primary tumor. Median size of the metastasis was 35 mm (range 10–130 mm). Median follow-up was 41 months (range 1–232 months). In the complete study population, median DFS was 28 months (range 1–228 months) with a median OS of 57 months (range 1–232 months) and a 5-year survival of 48.1 %. Histopathologic Tumor Features Fibrous Capsule and Tumor Thickness In 34.4 % of patients (n = 43), the liver metastasis was surrounded by a fibrous capsule. Presence of a fibrous capsule was not associated with DFS, but it was associated with an improved OS of 109.3 months, versus 56.7 months in patients without a fibrous capsule (p = 0.037). In multivariate analysis, presence of a fibrous capsule was not an independent risk factor for OS (Tables 1, 2).Table 1 Relation of clinical and histologic factors with DFS after liver resection in patients with solitary CRLM

oved OS of 109.3 months, versus 56.7 months in patients without a fibrous capsule (p = 0.037). In multivariate analysis, presence of a fibrous capsule was not an independent risk factor for OS (Tables 1, 2).Table 1 Relation of clinical and histologic factors with DFS after liver resection in patients with solitary CRLM n % Median DFS UV p value MV p value Size (mm) ≤50 93 75 50.1 0.002* 0.020* >50 31 25 14.5 CEA (ng/ml) ≤200 121 97.6 27.5 0.508 – >200 3 2.4 40.6 DFI (months) ≤12 72 58.1 27.8 0.232 – >12 52 41.9 25.4 Nodal state primary N0 54 43.5 35.7 0.446 N+ 70 56.5 27.5 – Adjuvant therapy No 106 85.5 20.2 0.013* 0.025* Yes 18 14.5 >50 Tumor thickness at TNI (mm) ≤3 60 48.4 >51 0.023* 0.118 >3 64 51.6 19.4 Fibrous capsule Present 43 34.4 27.8 0.468 – Absent 81 65.6 25.8 Perineural growth Present 11 8.9 50.2 0.539 – Absent 113 91.1 27.5 Vascular invasion Present 46 37.1 18.0 0.055 0.287 Absent 78 62.9 40.8 Lymphatic invasion Present 33 26.6 19.4 0.280 – Absent 91 73.4 29.2 Bile duct invasion Present 11 8.8 27.8 0.624 – Absent 113 91.2 27.5 DFS disease-free survival, CRLM colorectal liver metastases, UV univariate, MV multivariate, CEA carcinoembryonic antigen, DFI disease-free interval between treatment of primary tumor and detection of the CRLM, TNI tumor–normal interface * p < 0.05 was considered statistically significant Table 2 Relation of clinical and histologic factors with OS after liver resection in patients with solitary CRLM

DFS disease-free survival, CRLM colorectal liver metastases, UV univariate, MV multivariate, CEA carcinoembryonic antigen, DFI disease-free interval between treatment of primary tumor and detection of the CRLM, TNI tumor–normal interface * p < 0.05 was considered statistically significant Table 2 Relation of clinical and histologic factors with OS after liver resection in patients with solitary CRLM n % Median OS UV p value MV p value Size (mm) ≤50 93 75 65.9 0.050* 0.004* >50 31 25 29.2 CEA (ng/ml) ≤200 121 97.6 57.3 0.912 – >200 3 2.4 28.9 DFI ≤12 72 58.1 49.0 0.059 0.019* >12 52 41.9 91.5 Nodal state primary N0 54 43.5 61.0 0.231 – N+ 70 56.5 44.6 Adjuvant therapy No 106 85.5 57.2 0.955 – Yes 18 14.5 29.2 Tumor thickness at TNI (mm) ≤3 60 48.4 95.3 0.043* 0.068 >3 64 51.6 48.8 Fibrous capsule Present 43 34.4 109.3 0.037* 0.240 Absent 81 65.6 56.7 Perineural growth Present 11 8.9 109.3 0.652 – Absent 113 91.1 55.9 Vascular invasion Present 46 37.1 48.8 0.483 – Absent 78 62.9 58.2 Lymphatic invasion Present 33 26.6 41.9 0.013* 0.041* Absent 91 73.4 62.2 Bile duct invasion Present 11 8.8 76.7 0.048* 0.094 Absent 113 91.2 55.9 OS overall survival, CRLM colorectal liver metastases, UV univariate, MV multivariate, CEA carcinoembryonic antigen, DFI disease-free interval between treatment of primary tumor and detection of the CRLM, TNI tumor–normal interface * p < 0.05 was considered statistically significant

n % Median OS UV p value MV p value Size (mm) ≤50 93 75 65.9 0.050* 0.004* >50 31 25 29.2 CEA (ng/ml) ≤200 121 97.6 57.3 0.912 – >200 3 2.4 28.9 DFI ≤12 72 58.1 49.0 0.059 0.019* >12 52 41.9 91.5 Nodal state primary N0 54 43.5 61.0 0.231 – N+ 70 56.5 44.6 Adjuvant therapy No 106 85.5 57.2 0.955 – Yes 18 14.5 29.2 Tumor thickness at TNI (mm) ≤3 60 48.4 95.3 0.043* 0.068 >3 64 51.6 48.8 Fibrous capsule Present 43 34.4 109.3 0.037* 0.240 Absent 81 65.6 56.7 Perineural growth Present 11 8.9 109.3 0.652 – Absent 113 91.1 55.9 Vascular invasion Present 46 37.1 48.8 0.483 – Absent 78 62.9 58.2 Lymphatic invasion Present 33 26.6 41.9 0.013* 0.041* Absent 91 73.4 62.2 Bile duct invasion Present 11 8.8 76.7 0.048* 0.094 Absent 113 91.2 55.9 OS overall survival, CRLM colorectal liver metastases, UV univariate, MV multivariate, CEA carcinoembryonic antigen, DFI disease-free interval between treatment of primary tumor and detection of the CRLM, TNI tumor–normal interface * p < 0.05 was considered statistically significant Tumor thickness at tumor–normal interface varied between 0.1 and 7.2 mm, with a median of 3 mm, and was not correlated with the size of the liver metastases (p = 0.213). Although there was a significant association of increased thickness with decreased outcome (both DFS and OS) in univariate analysis, it was no longer significant in multivariate analysis (Tables 1, 2).

tween 0.1 and 7.2 mm, with a median of 3 mm, and was not correlated with the size of the liver metastases (p = 0.213). Although there was a significant association of increased thickness with decreased outcome (both DFS and OS) in univariate analysis, it was no longer significant in multivariate analysis (Tables 1, 2). Intrahepatic Spread Frequency of different forms of intrahepatic invasion varied; perineural growth (n = 11; 8.9 %) and bile duct invasion (n = 11; 8.8 %) were both relatively uncommon, whereas vascular and lymphatic invasion were seen more frequently (n = 46; 37.1 %, respectively n = 33; 26.6 %). In univariate analysis, presence of bile duct invasion was associated with improved OS (76.7 vs. 55.9 months; p = 0.048), but this was not the case in multivariate analysis (p = 0.094). Presence of intrahepatic lymphatic invasion was correlated with a decreased median OS (41.9 vs. 62.2 months, p = 0.013), which remained significant in multivariate analysis (p = 0.041) (Supplementary Fig. 1a). In the current study, no correlation between different forms of intrahepatic spread or between any of the histologic features and the various items of the clinical risk score was observed. However, there was a correlation between presence of a fibrous capsule and absence of intrahepatic vascular invasion (p = 0.014) and between presence of a fibrous capsule and presence of intrahepatic bile duct invasion (p = 0.013).

f the histologic features and the various items of the clinical risk score was observed. However, there was a correlation between presence of a fibrous capsule and absence of intrahepatic vascular invasion (p = 0.014) and between presence of a fibrous capsule and presence of intrahepatic bile duct invasion (p = 0.013). In 15 patients, a combination of intrahepatic lymphatic invasion and intrahepatic vascular invasion was present, and this combination was associated with a decreased OS (median 28.1 vs. 62.2 months) in univariate and multivariate analysis (p < 0.0001) (Supplementary Fig. 1b). Discussion The current study describes the association between multiple histologic features in combination with clinical factors and survival in 124 patients who underwent liver resection for CRLM. A homogenous group of patients was evaluated because all patients underwent a complete resection (R0), for a solitary metastasis without neoadjuvant systemic treatment. The only significant histologic factor associated with decreased survival in multivariate analysis was presence of intrahepatic lymphatic invasion, especially in combination with intrahepatic vascular invasion.

tients underwent a complete resection (R0), for a solitary metastasis without neoadjuvant systemic treatment. The only significant histologic factor associated with decreased survival in multivariate analysis was presence of intrahepatic lymphatic invasion, especially in combination with intrahepatic vascular invasion. Other authors also described lymphatic invasion as a negative predictor for survival.13,18,20 In the current study, we observed a relative high frequency of lymphatic invasion (26.6 %) compared to earlier studies (12–15 %).18,20 This might be due to the use of immunohistochemistry, which is supported by a recently published study with the same methodology and a similar frequency of lymphatic invasion (29 %).18,20,34–36 Presence of lymphatic invasion has been associated with spread to hepatic lymph nodes, which often leads to incurable disease.20,37 In the current study, the worse prognosis was demonstrated in patients with a combination of vascular and lymphatic invasion. This unfavorable combination has been observed before and might reflect a tumor with aggressive behavior.23

iated with spread to hepatic lymph nodes, which often leads to incurable disease.20,37 In the current study, the worse prognosis was demonstrated in patients with a combination of vascular and lymphatic invasion. This unfavorable combination has been observed before and might reflect a tumor with aggressive behavior.23 Another interesting finding from the current study was that the median tumor thickness at tumor–normal interface in patients who were not treated with neoadjuvant systemic therapy was 3.0 mm. This was only slightly higher than the tumor thickness of 2.8 mm described in patients treated with neoadjuvant chemotherapy.26 This raises the question whether tumor thickness at tumor–normal interface reflects chemotherapy response or tumor biology; this would be an interesting subject for further research.

s 3.0 mm. This was only slightly higher than the tumor thickness of 2.8 mm described in patients treated with neoadjuvant chemotherapy.26 This raises the question whether tumor thickness at tumor–normal interface reflects chemotherapy response or tumor biology; this would be an interesting subject for further research. A major strength of the present study is the inclusion of patients with solitary CRLM only, who were operated with complete margins (R0) to create an homogenous group of patients. Previous studies on histologic prognostic factors included patients with multiple CRLM and R1 resections as well, which might lead to significant bias of the results.18,20,36 First, heterogeneity of histologic features between the different liver metastases might exist and could lead to bias studying prognostic factors for survival. Second, patients who undergo R1 resection usually have a higher risk of local recurrences and have an impaired survival.38,39 Third, patients with multiple metastases have a significantly decreased survival, and number of metastases is the most important factor in the Fong classification for survival.9 By excluding these potential biases in the present study, the assessment of the prognostic histologic factors are more reliable.

urvival.38,39 Third, patients with multiple metastases have a significantly decreased survival, and number of metastases is the most important factor in the Fong classification for survival.9 By excluding these potential biases in the present study, the assessment of the prognostic histologic factors are more reliable. Another strength is that this homogenous group of patients with solitary metastasis were not treated with neoadjuvant systemic therapy. In recent studies, patients with and without neoadjuvant systemic therapy were mixed, and conclusions were drawn from a population highly susceptible to bias.25,36,40 Neoadjuvant systemic therapy has a significant impact on tumor histology, and even prognostic factors such as resection margins might be less important.27,28,41 Because the detection of histologic prognostic factors in metastatic disease is still in its infancy and the effects of neoadjuvant systemic therapy on lymphatic invasion are unknown, a study with an homogeneous population should be a first step. However, there seems to be an increasing preference to utilize neoadjuvant systemic therapy for high risk patients, despite a lack of convincing evidence on survival benefit in patients with limited metastases.42–44 Therefore, a limitation of the present study is that the impact of lymphatic invasion on survival has to be confirmed in patients treated with neoadjuvant systemic therapy. In the total group of patients treated in our institution only 5 patients (3.8 %) with solitary metastasis were treated with neoadjuvant chemotherapy, which made it impossible to compare, but this should be the goal for future research.

survival has to be confirmed in patients treated with neoadjuvant systemic therapy. In the total group of patients treated in our institution only 5 patients (3.8 %) with solitary metastasis were treated with neoadjuvant chemotherapy, which made it impossible to compare, but this should be the goal for future research. In conclusion, intrahepatic lymphatic invasion, based on immunohistochemical detection of lymphatic vessels, is an adverse prognostic factor for OS in patients with a solitary CRLM. Therefore, we recommend evaluating the presence or absence of intrahepatic lymphatic and vascular invasion in the histologic assessment of CRLM. Future research is needed to determine whether adjuvant treatment strategies should be based on these adverse prognostic histologic factors. Electronic supplementary material Supplementary material 1 (DOCX 171 kb) Acknowledgment Supported in part by an unrestricted grant of the VALAMO Foundation. Disclosure The authors declare no conflict of interest.

Patients with clinically palpable nodal metastases of cutaneous melanoma in the groin have a poor prognosis. Balch et al. reported a 5 year overall survival (OS) rate of 59 % for stage IIIB melanoma in the 2009 American Joint Committee on Cancer (AJCC) melanoma staging system analysis.1 Reported 5 year OS rates for the subgroup of patients with palpable groin metastases ranged from 52 % for superficial involvement to 12 % for deep involvement.2–7 Standard of care for these patients consists of therapeutic lymph node dissection (TLND),2,8–10 and there is ongoing debate as to whether this should consist of either a combined superficial and deep groin dissection (CGD) or whether a superficial groin dissection (SGD) would suffice. Several cohort studies indicate no difference in survival between these two procedures, and patients may benefit from SGD alone if no positive deep pelvic nodes are present on preoperative imaging.2,8,10–12 Since the estimated prevalence of positive deep pelvic nodes in patients with palpable inguinal lymph nodes is 30 %, the majority of patients undergoing CGD may not benefit from deep groin dissection (DGD).6,12 As CGD is a more extensive procedure than SGD, the risk of morbidity is potentially higher.6 A clear need exists to select those patients who can be safely spared a DGD in the absence of deep pelvic nodal involvement.10,11,13–15

e majority of patients undergoing CGD may not benefit from deep groin dissection (DGD).6,12 As CGD is a more extensive procedure than SGD, the risk of morbidity is potentially higher.6 A clear need exists to select those patients who can be safely spared a DGD in the absence of deep pelvic nodal involvement.10,11,13–15 Preoperative imaging techniques such as computed tomography (CT) and positron emission tomography (PET) form a valuable adjunct to staging. Up to 27 % of patients presenting with palpable lymph node metastases have synchronous distant metastases at preoperative PET/CT, which changes the indication for surgery into palliative resection and/or systemic therapy.16 Additionally, imaging provides assessment of suspicious deep pelvic nodes prior to surgery. High positive (PPV) and negative predictive value (NPV) have been achieved by Allan et al. (100 and 86 %, respectively).3 Other series reported PPVs and NPVs of 40–60 %, which is too low to confirm or reject the presence of positive deep pelvic nodes based on preoperative imaging alone.2,17,18 Once suspicious deep pelvic nodes are detected on preoperative imaging, one cannot ignore their presence and CGD is highly recommended. The absence of suspicious deep pelvic nodes on imaging does not rule out deep pelvic nodal involvement. Once imaging has been performed, the focus should be on identification of further risk factors for positive deep pelvic nodes.2,7,11,15,17–21

e imaging, one cannot ignore their presence and CGD is highly recommended. The absence of suspicious deep pelvic nodes on imaging does not rule out deep pelvic nodal involvement. Once imaging has been performed, the focus should be on identification of further risk factors for positive deep pelvic nodes.2,7,11,15,17–21 Patients and Methods Patients This retrospective, multicenter cohort study described 209 therapeutic CGDs performed at four tertiary melanoma centers in The Netherlands between 1992 and 2013. Patient selection was based on the presence of a palpable nodal metastasis to the groin, complete pathology reports of the performed CGD (i.e. clearly describing the dissected lymph nodes as inguinal or iliac, including obturator area), and preoperative imaging (CT, PET, or PET/CT). Patients without imaging, with prior lymph node dissections in the groin area, or with isolated limb perfusion or positive sentinel node(s) as an indication for CGD were excluded. Analyzed preoperative imaging modalities were CT scan, PET, and combined PET with low-dose CT (PET/CT). All patient characteristics were obtained from medical records and collected in a database for the current study, according to local Institutional Review Committee guidelines and national legislation. Surgical Procedure CGD was performed either via two separate transverse incisions or via an inguinal ellipse-shaped incision extending cranially according to local preferences per center, as described in detail elsewhere.6,22

All patient characteristics were obtained from medical records and collected in a database for the current study, according to local Institutional Review Committee guidelines and national legislation. Surgical Procedure CGD was performed either via two separate transverse incisions or via an inguinal ellipse-shaped incision extending cranially according to local preferences per center, as described in detail elsewhere.6,22 Pathology CGD pathology reports were considered adequate when a clear description was given of the total number of inguinal nodes as well as the number of tumor-positive inguinal nodes, and a similar description was given of the number of dissected deep pelvic nodes (iliac nodes and obturator nodes) and the number of tumor-positive deep pelvic nodes.

were considered adequate when a clear description was given of the total number of inguinal nodes as well as the number of tumor-positive inguinal nodes, and a similar description was given of the number of dissected deep pelvic nodes (iliac nodes and obturator nodes) and the number of tumor-positive deep pelvic nodes. Statistics/Data Analysis Patients were divided into two categories based on deep pelvic nodal status—positive or negative. Univariable χ2 tests were performed to test for significant differences in prevalence of sex, primary tumor located on the trunk, primary tumor stage (T1–T4), ulceration, and inguinal extracapsular extension (ECE). Nonparametric tests were performed to test for differences in age, median Breslow thickness, total number of inguinal nodes and number of positive inguinal nodes, total number of excised nodes and number of positive nodes, total number of deep pelvic nodes, number of positive deep pelvic nodes, and LNR. Sensitivity, specificity, PPV, NPV, and accuracy were calculated for all imaging modalities using the number of true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN). Differences in baseline characteristics were tested using univariable logistic regression analysis, multivariable models were calculated using variables significant at univariable analysis, and binary logistic multivariable regression analyses were performed to test for independent predictors of deep pelvic nodal involvement.

nces in baseline characteristics were tested using univariable logistic regression analysis, multivariable models were calculated using variables significant at univariable analysis, and binary logistic multivariable regression analyses were performed to test for independent predictors of deep pelvic nodal involvement. Ridge regression analysis was performed to exclude the influence of multicollinearity in a prediction model based on independent predictive variables. An area under the receiver operating characteristic curve (AUC) was calculated for the model. The AUC indicates the probability that patients with observed positive deep pelvic nodes had a higher predicted probability than patients with observed negative deep pelvic nodes, providing information about the predictive value of the model. All statistical analyses, with the exception of Ridge regression, were performed using SPSS version 21.0 (released 2012; IBM Corporation, Armonk, NY, USA). Ridge regression was performed using RStudio (RStudio Inc., Boston, MA, USA). An α < 0.05 was considered significant. Results Patients Table 1 provides an overview of baseline characteristics. The majority of patients (n = 201, 96 %) had palpable stage IIIB disease, and eight patients (4 %) had stage IV disease. Median Breslow thickness was 2.10 mm (interquartile range [IQR] 1.4–3.4 mm), 12 patients had a history of negative sentinel node, and median follow-up was 21 months (IQR 11–46 months).Table 1 Baseline characteristics

atients (n = 201, 96 %) had palpable stage IIIB disease, and eight patients (4 %) had stage IV disease. Median Breslow thickness was 2.10 mm (interquartile range [IQR] 1.4–3.4 mm), 12 patients had a history of negative sentinel node, and median follow-up was 21 months (IQR 11–46 months).Table 1 Baseline characteristics Characteristic Total [n = 209 (100 %)] Negative pelvic nodes [n = 135 (65 %)] Positive pelvic nodes [n = 74 (35 %)] p value Sex Female 114 (54) 76 (56) 38 (51) 0.49 Male 95 (46) 59 (44) 36 (49) Age [years; median (IQR)] 57 (45–65) 55 (46–65) 59 (44–65) 0.63 Center 1 60 (29) 42 (31) 18 (24) 0.21 2 57 (27) 38 (28) 19 (26) 3 24 (12) 11 (8) 13 (18) 4 68 (32) 44 (33) 24 (32) Tumor stage T1 22 (11) 9 (7) 13 (18) 0.16 T2 57 (27) 36 (27) 21 (28) T3 60 (29) 39 (29) 21 (28) T4 30 (14) 22 (16) 8 (11) Unknown primary 10 (5) 8 (6) 2 (3) Missing 30 (14) 21 (15) 9 (12) Ulceration Absent 125 (60) 74 (55) 51 (69) 0.16 Present 54 (26) 38 (28) 16 (22) Missing 30 (14) 23 (17) 7 (9) Clark levela II 1 (0.5) 0 (–) 1 (1) 0.29 III 35 (17) 22 (16) 13 (17) IV 84 (40) 54 (40) 30 (41) V 13 (6) 11 (8) 2 (3) Missing 76 (36.5) 48 (36) 28 (38) Location Leg 166 (80) 106 (79) 60 (81) 0.62 Trunk 28 (13) 17 (13) 11 (15) Unknown primary 10 (5) 8 (6) 2 (3) Missing 5 (2) 4 (3) 1 (1) Histology SSM 67 (32) 44 (32) 23 (31) 0.24 NM 37 (18) 28 (21) 9 (12) Other 15 (7) 9 (7) 6 (8) Unknown primary 10 (5) 8 (6) 2 (3) Missing 80 (38) 46 (34) 34 (46) No. of nodes [median (IQR)] Inguinal 10 (7–13) 10 (7–13) 9 (7–12) 0.54 Deep 6 (4–10) 6 (4–9) 8 (5–11) 0.039b

imary 10 (5) 8 (6) 2 (3) Missing 5 (2) 4 (3) 1 (1) Histology SSM 67 (32) 44 (32) 23 (31) 0.24 NM 37 (18) 28 (21) 9 (12) Other 15 (7) 9 (7) 6 (8) Unknown primary 10 (5) 8 (6) 2 (3) Missing 80 (38) 46 (34) 34 (46) No. of nodes [median (IQR)] Inguinal 10 (7–13) 10 (7–13) 9 (7–12) 0.54 Deep 6 (4–10) 6 (4–9) 8 (5–11) 0.039b Total 17 (13–22) 17 (13–21) 17 (14–22) 0.39 No. of positive nodes [median (IQR)] Inguinal 2 (1–3) 1 (1–2) 3 (1–4) <0.001b Deep 0 (0–1) 0 (0) 2 (1–3) <0.001b Total 2 (1–4) 1 (1–2) 5 (3–7) <0.001b LNR [median (IQR)] 0.20 (0.11–0.33) 0.15 (0.10–0.25) 0.33 (0.14–0.54) <0.001b Inguinal ECE No 134 (64) 94 (70) 40 (54) 0.025b Yes 75 (36) 41 (30) 34 (46) Data are expressed as n (%) unless otherwise specified IQR interquartile range, T1 Breslow < 1.00 mm, T2 Breslow 1.01–2.00 mm, T3 Breslow 2.01–4.00 mm, T4 Breslow > 4.00 mm, LNR inguinal lymph node ratio, ECE extracapsular extension aClark levels II and III were combined for the χ 2 test bSignificant, p < 0.05, calculated using χ 2 and non-parametric tests Imaging and Pathology Four patients underwent both CT and PET/CT; they were scored as PET/CT since the additional information obtained from PET/CT was used for the final determination of clinical node status. Predictive accuracy per imaging modality is shown in Table 2. The different imaging modalities were used equally between the two groups (i.e. positive or negative deep pelvic nodes).Table 2 Identification of positive deep pelvic lymph nodes using preoperative imaging techniques (n = 209) CT (%) [n = 67] CT and/or PET (%) [n = 57]a PET/CT (%) [n = 85]b

Imaging and Pathology Four patients underwent both CT and PET/CT; they were scored as PET/CT since the additional information obtained from PET/CT was used for the final determination of clinical node status. Predictive accuracy per imaging modality is shown in Table 2. The different imaging modalities were used equally between the two groups (i.e. positive or negative deep pelvic nodes).Table 2 Identification of positive deep pelvic lymph nodes using preoperative imaging techniques (n = 209) CT (%) [n = 67] CT and/or PET (%) [n = 57]a PET/CT (%) [n = 85]b Sensitivity 57 36 61 Specificity 93 94 83 PPV 80 73 68 NPV 83 70 79 Accuracy 82 70 75 CT computed tomography, PET position emission tomography, PET/CT combined PET and low-dose CT, PPV positive predictive value, NPV negative predictive value aThirteen patients underwent PET alone bFour patients also underwent separate CT Logistic Regression Analysis Variables significantly different on univariable analysis (Table 1) were included in multivariable binary logistic regression analyses. LNR and number of positive inguinal lymph nodes were assessed in separate models due to evident multicollinearity. The remaining significant independent predictors were suspicious deep pelvic nodes on imaging (odds ratio [OR] 9.64, 95 % CI 4.35–21.3, p < 0.001), increasing LNR (OR 34.2, 95 % CI 5.47–214, p < 0.001), the presence of ECE (OR 2.13, 95 % CI 1.01–4.48, p = 0.046) and, in a separate multivariable model without LNR, increasing number of positive inguinal lymph nodes (OR 1.27, 95 % CI 1.06–1.53, p = 0.010).

imaging (odds ratio [OR] 9.64, 95 % CI 4.35–21.3, p < 0.001), increasing LNR (OR 34.2, 95 % CI 5.47–214, p < 0.001), the presence of ECE (OR 2.13, 95 % CI 1.01–4.48, p = 0.046) and, in a separate multivariable model without LNR, increasing number of positive inguinal lymph nodes (OR 1.27, 95 % CI 1.06–1.53, p = 0.010). Subgroup Analysis Negative Imaging Suspicious deep pelvic nodes on imaging were highly predictive for positive deep pelvic nodes. A subgroup of 155 patients without suspicious deep pelvic nodes on imaging was selected for further analysis of additional risk factors for positive deep pelvic nodes. Thirty-five of these patients (23 %) had positive deep pelvic nodes at histopathological examination with hematoxylin and eosin (H&E) staining, i.e. imaging was FN. Univariable analysis results are displayed in Table 3. Multivariable analysis was performed, including all significant variables assumed to be predictive for deep pelvic nodal status: number of positive inguinal nodes, LNR, and ECE status. Evident multicollinearity was observed.Table 3 Baseline characteristics for patients with negative preoperative imaging

yed in Table 3. Multivariable analysis was performed, including all significant variables assumed to be predictive for deep pelvic nodal status: number of positive inguinal nodes, LNR, and ECE status. Evident multicollinearity was observed.Table 3 Baseline characteristics for patients with negative preoperative imaging Characteristic Total (n = 155) Pelvic nodes− (n = 120) Pelvic nodes+ (n = 35) p value Sex Female 83 (54) 67 (56) 16 (46) Male 72 (46) 53 (44) 19 (54) 0.29 Age [years; median (IQR)] 56 (45–64) 55 (46–65) 57 (44–64) 0.99 Center 1 44 (28) 38 (32) 6 (17) 2 48 (31) 35 (29) 13 (37) 3 18 (12) 10 (8) 8 (23) 4 45 (29) 37 (31) 8 (23) 0.17 Breslow [median (IQR)] 2.10 (1.40–3.25) 2.20 (1.45–3.55) 1.90 (1.15–2.80) 0.11 Tumor stage T1 14 (9) 8 (7) 6 (17) T2 45 (29) 33 (28) 12 (34) T3 44 (28) 37 (31) 7 (20) T4 23 (15) 19 (16) 4 (11) Unknown primary 9 (6) 7 (6) 2 (6) Missing 20 (13) 16 (13) 4 (11) 0.38 Ulceration Absent 90 (58) 67 (56) 23 (66) Present 44 (28) 36 (31) 8 (23) Missing 21 (14) 17 (13) 4 (11) 0.34 Clark levela

.15–2.80) 0.11 Tumor stage T1 14 (9) 8 (7) 6 (17) T2 45 (29) 33 (28) 12 (34) T3 44 (28) 37 (31) 7 (20) T4 23 (15) 19 (16) 4 (11) Unknown primary 9 (6) 7 (6) 2 (6) Missing 20 (13) 16 (13) 4 (11) 0.38 Ulceration Absent 90 (58) 67 (56) 23 (66) Present 44 (28) 36 (31) 8 (23) Missing 21 (14) 17 (13) 4 (11) 0.34 Clark levela II 1 (0.6) 0 (–) 1 (3) III 25 (16) 20 (17) 5 (14) IV 65 (42) 49 (41) 16 (46) V 11 (7) 11 (9) 0 (–) Missing 53 (34) 40 (33) 13 (37) 0.070 Location Leg 118 (76) 93 (78) 25 (71) Trunk 23 (15) 16 (13) 7 (20) Unknown primary 9 (6) 7 (6) 2 (6) Missing 5 (3) 4 (3) 1 (3) 0.81 Histology SSM 52 (34) 40 (33) 5 (14) NM 31 (20) 26 (22) 12 (34) Other 10 (6) 8 (7) 2 (6) Unknown primary 9 (6) 7 (6) 2 (6) Missing 53 (34) 39 (32) 14 (40) 0.86 No. of nodes [median (IQR)] Total 17 (13–21) 17 (13–21) 17 (14–22) 0.42 Inguinal 10 (8–12) 10 (8–13) 9 (8–12) 0.69 Deep 6 (4–9) 6 (4–9) 7 (4–11) 0.15 No. of positive nodes [median (IQR)] Total 2 (1–4) 1 (1–2) 5 (3–6) <0.001b Inguinal 1 (1–3) 1 (1–2) 3 (1–4) <0.001b Deep 0 (0) 0 (0) 2 (1–2) <0.001b LNR [median (IQR)] 0.17 (0.11–0.31) 0.21 (0.10–0.25) 0.33 (0.13–0.50) 0.001b ECE inguinal No 96 (62) 79 (66) 17 (49) Yes 59 (38) 41 (34) 18 (51) 0.075 Data are expressed as n (%) unless otherwise specified IQR interquartile range, T1 Breslow < 1.00 mm, T2 Breslow 1.01–2.00 mm, T3 Breslow 2.01–4.00 mm, T4 Breslow > 4.00 mm, LNR inguinal lymph node ratio, ECE extracapsular extension aFor the χ 2 test, Clark II and III were combined bSignificant (p < 0.05)

ECE inguinal No 96 (62) 79 (66) 17 (49) Yes 59 (38) 41 (34) 18 (51) 0.075 Data are expressed as n (%) unless otherwise specified IQR interquartile range, T1 Breslow < 1.00 mm, T2 Breslow 1.01–2.00 mm, T3 Breslow 2.01–4.00 mm, T4 Breslow > 4.00 mm, LNR inguinal lymph node ratio, ECE extracapsular extension aFor the χ 2 test, Clark II and III were combined bSignificant (p < 0.05) To overcome this problem, a predictive Ridge logistic regression analysis was performed. Only LNR remained as a significant independent predictor for positive deep pelvic nodes (p = 0.014). The number of positive inguinal lymph nodes and ECE were chosen to remain in the model as contributing covariates as these were thought to be of substantial additional clinical relevance. A receiver operating characteristic (ROC) curve of the predicted probabilities for positive deep pelvic nodes was created, displaying a fair AUC of 0.72 (AUC values ranged between 0 and 1, where high scores are indicative of high accuracy) (Fig. 1). The optimum cut-off value for the predicted probability of the model (i.e. the probability at which the model outcome correctly identifies an observed positive patient as positive) was chosen based on high specificity in order to minimize FN outcomes. Corresponding probability cut-off value and sensitivity were deduced from the ROC curve. For a specificity of 90 %, the cut-off value for a positive test outcome was a probability for positive deep pelvic nodes of 32 % or more. Sensitivity was 43 %, PPV 50 %, NPV 84 %, and overall accuracy of this model was 77 %.Fig. 1 ROC curve for prediction model probability positive deep nodes. ROC receiver operator characteristics

city of 90 %, the cut-off value for a positive test outcome was a probability for positive deep pelvic nodes of 32 % or more. Sensitivity was 43 %, PPV 50 %, NPV 84 %, and overall accuracy of this model was 77 %.Fig. 1 ROC curve for prediction model probability positive deep nodes. ROC receiver operator characteristics Discussion In this CGD cohort, 35 % of all patients had deep pelvic nodal involvement, which is in line with the literature.10,11,13–15 This study analyses risk factors to identify deep pelvic nodal involvement, with imaging being a strong predictor. Our prediction model might lower the rate of CGD without positive pelvic nodes, and minimizes the number of FN outcomes after imaging. Imaging The imaging modalities used in this study are fair in correctly predicting positive deep pelvic nodes; however, a considerable number of patients have FP imaging (20–32 %), and we can only speculate on the possible causes of FP imaging. This might be partially explained by a small group of patients undergoing diagnostic excision biopsy of the palpable lymph node prior to imaging, which might cause lymph node enlargement in the pelvic area. Another cause may be the inevitable interobserver variability in radiology. Improvement of imaging techniques over time may have altered the number of FP lymph nodes detected during the present study period.

biopsy of the palpable lymph node prior to imaging, which might cause lymph node enlargement in the pelvic area. Another cause may be the inevitable interobserver variability in radiology. Improvement of imaging techniques over time may have altered the number of FP lymph nodes detected during the present study period. NPVs of the preoperative imaging techniques performed in the current study range between 70 and 83 %, leaving a substantial proportion of 23 % (17–30 %) of patients to be falsely diagnosed with negative deep pelvic nodes. Several studies have reported on the NPV of CT, and although high NPVs have been described by Allan et al. and Van der Ploeg et al. overall reported values ranged considerably.2,3,6,17,18 Ongoing development of the newest imaging techniques, such as the use of a melanoma-specific PET tracer ([18F]ICF01006), may enhance the accuracy of imaging and subsequently decrease the FN rate.23

NPVs have been described by Allan et al. and Van der Ploeg et al. overall reported values ranged considerably.2,3,6,17,18 Ongoing development of the newest imaging techniques, such as the use of a melanoma-specific PET tracer ([18F]ICF01006), may enhance the accuracy of imaging and subsequently decrease the FN rate.23 Predictive Factors Predictive factors for deep pelvic nodal involvement found in the current study are inguinal nodal status as defined by the number of positive inguinal nodes and LNR, inguinal ECE, and suspicious deep pelvic nodes on preoperative imaging, which is concordant with the literature.2,7,11,15,17–21 These risk factors may be applied to select patients for SGD, in addition to imaging without suspicious deep pelvic nodes. A hypothetical two-stage approach would be when preoperative imaging is negative, patients first solely undergo an SGD. The pathology results can then be used to determine the risk of occult positive deep pelvic nodes, and a decision can be made on whether to perform an additional DGD or not. The fact that patients must undergo two separate operations is a drawback, but this way a DGD can be spared in 126 of all patients (60 %). Patient Selection Standard CGD for palpable stage III melanoma shows that 135 of 209 deep pelvic groin dissections (65 %) have been performed in the absence of pelvic nodal metastases.

Predictive Factors Predictive factors for deep pelvic nodal involvement found in the current study are inguinal nodal status as defined by the number of positive inguinal nodes and LNR, inguinal ECE, and suspicious deep pelvic nodes on preoperative imaging, which is concordant with the literature.2,7,11,15,17–21 These risk factors may be applied to select patients for SGD, in addition to imaging without suspicious deep pelvic nodes. A hypothetical two-stage approach would be when preoperative imaging is negative, patients first solely undergo an SGD. The pathology results can then be used to determine the risk of occult positive deep pelvic nodes, and a decision can be made on whether to perform an additional DGD or not. The fact that patients must undergo two separate operations is a drawback, but this way a DGD can be spared in 126 of all patients (60 %). Patient Selection Standard CGD for palpable stage III melanoma shows that 135 of 209 deep pelvic groin dissections (65 %) have been performed in the absence of pelvic nodal metastases. Use of preoperative imaging alone for selection between CGD and SGD would reduce the number of CGDs from 209 to 54. The remaining 155 patients would undergo SGD alone. Thirty-five of these 155 patients undergoing SGD alone were FN (FN rate 23 %) and would possibly be undertreated (i.e. undergoing no DGD).

Patient Selection Standard CGD for palpable stage III melanoma shows that 135 of 209 deep pelvic groin dissections (65 %) have been performed in the absence of pelvic nodal metastases. Use of preoperative imaging alone for selection between CGD and SGD would reduce the number of CGDs from 209 to 54. The remaining 155 patients would undergo SGD alone. Thirty-five of these 155 patients undergoing SGD alone were FN (FN rate 23 %) and would possibly be undertreated (i.e. undergoing no DGD). Better patient selection is necessary in the negative imaging group as a potential decrease in the number of FNs will make patient selection safer. This formed the rationale for the prediction model, which is based on 153 patients (155—2 patients with missing data) with negative imaging. Using this model, 124 of 153 patients would undergo SGD alone, and FN rates would be reduced to 20 of 124 patients (FN rate 16 %). Concluding, this model forms an adjunct to the use of preoperative imaging as a selection tool for SGD or CGD, both drastically minimizing the number of patients without affected pelvic nodes undergoing a DGD, and controlling the number of patients with affected pelvic nodes potentially being undertreated by not undergoing a DGD.

this model forms an adjunct to the use of preoperative imaging as a selection tool for SGD or CGD, both drastically minimizing the number of patients without affected pelvic nodes undergoing a DGD, and controlling the number of patients with affected pelvic nodes potentially being undertreated by not undergoing a DGD. The 16 % FN rate of this model is still considerable. Although surgery forms the cornerstone of melanoma treatment, one may question the role of DGD in the current era of upcoming effective systemic treatments. On one hand, the majority of patients undergoing standard CGD for palpable groin metastases have negative deep pelvic nodes, while on the other hand there is evidence to assume that positive deep pelvic nodes may merely be a biomarker for stage IV disease as survival rates depend on deep pelvic nodal status rather than extent of surgery.6,8,11,12,15 Khosrotehrani et al. presented a nomogram for the prediction of prognosis in stage III B/C melanoma patients, using pathology results and age.24 Application of this nomogram could further aid in selecting patients for SGD alone. Another preoperative aid in addition to the presented model could be use of the biomarker S-100B. As Kruijff et al. have shown, high serum levels of S-100B are associated with a significantly lower disease-free survival and a trend towards worse melanoma-specific survival (MSS), indicating its potential as a biomarker for clinically occult stage IV disease.25,26 Patients with a low risk of deep pelvic nodal involvement and low S-100B could then undergo SGD alone, with regular control visits to detect early signs of deep pelvic nodal involvement (suspicious nodes on imaging/elevated S-100B). Bearing this in mind, the 16 % FN rate of the presented prediction model may be allowable.

ients with a low risk of deep pelvic nodal involvement and low S-100B could then undergo SGD alone, with regular control visits to detect early signs of deep pelvic nodal involvement (suspicious nodes on imaging/elevated S-100B). Bearing this in mind, the 16 % FN rate of the presented prediction model may be allowable. Limitations This study was retrospective and was spread over a long timeframe. This entails inevitable alterations and improvement of imaging techniques and clinical practice over time, affecting our results. The prediction model designed for the current study has not been validated internally due to a small sample of patients with positive deep pelvic nodes. It has to be pointed out that this model in its current state is not suited for clinical use as there is still much to be gained from further development and testing. A prospective, multicenter registration study is planned, enabling adequate data collection on all patients undergoing CGD for palpable groin metastases within a relatively small timeframe. Cross-validation of the presented prediction model will be performed and its role in future clinical practice will be further defined. With the proposed prospective study, accuracy of imaging techniques can be determined more adequately.

undergoing CGD for palpable groin metastases within a relatively small timeframe. Cross-validation of the presented prediction model will be performed and its role in future clinical practice will be further defined. With the proposed prospective study, accuracy of imaging techniques can be determined more adequately. Regarding the possible additional morbidity of a DGD, although to date no prospective randomized controlled trial (RCT) has been performed to address this, evidence exists that the additional morbidity of DGD in a CGD might be more limited than has been described in the past.6,22 The recently opened Australia and New Zealand Melanoma Trials Group 01.12 Evaluation of Groin Lymphadenectomy Extent For Metastatic Melanoma (EAGLE FM) trial (clinicaltrials.gov identifier NCT02166788) will hopefully provide an answer to this question. This multicenter RCT compares SGD and CGD for melanoma patients with groin metastases and no suspicious PET/CT scan. As operating time is generally longer in a CGD, there is a potentially higher risk of surgical site infections. In the large, retrospective series of Glarner et al. the number of surgical site infections is indeed significantly higher for CGDs, with an adjusted OR of 2.6.27 Once again, to gain more insight into the actual differences in morbidity between SGD and CGD, we will have to await results from the EAGLE FM Trial.

ctions. In the large, retrospective series of Glarner et al. the number of surgical site infections is indeed significantly higher for CGDs, with an adjusted OR of 2.6.27 Once again, to gain more insight into the actual differences in morbidity between SGD and CGD, we will have to await results from the EAGLE FM Trial. Conclusions High LNR, high number of positive inguinal nodes, and inguinal ECE are risk factors for positive deep pelvic nodes in patients with palpable groin metastases of cutaneous melanoma. To date, accurate prediction of deep pelvic nodal status is still suboptimal, hence reliable selection of patients who can be spared a DGD remains difficult. Combined use of preoperative imaging and a preliminary prediction model based on histopathology results of the inguinal (superficial) part of CGD could accurately predict negative deep pelvic nodes in up to 84 % of patients, thereby potentially identifying a group of low-risk patients in whom the extent of surgery might safely be minimized. The risk factors and the prediction model will be further investigated in a prospective, multicenter registry trial for CGDs. Disclosures Charlotte M.C. Oude Ophuis, Alexander C.J. van Akkooi, Harald J. Hoekstra, Johannes J. Bonenkamp, Julia van Wissen, Maarten G. Niebling, Johannes H.W. de Wilt, Bernies van der Hiel, Bart van de Wiel, Senada Koljenović, Dirk J. Grünhagen, and Cornelis Verhoef have declared no conflicts of interest.

In 2008, an estimated 482,300 people were diagnosed with esophageal cancer, and 406,800 patients died of the disease worldwide.1 Radical esophagolymphadenectomy is the cornerstone of the multimodality treatment with curative intent.2–5 Worldwide, open transthoracic esophagectomy is the preferred surgical approach for esophageal cancer, allowing en bloc resection of the tumor with the surrounding paratracheal, subcarinal and paraesophageal lymph nodes.6,7 However, the percentage of cardiopulmonary complications associated with the open transthoracic approach is high (50–70 %).6 Minimally invasive esophagectomy (MIE) was designed to reduce surgical trauma, resulting in lower rates of morbidity and mortality. With regard to MIE, a review of the literature shows a substantial decrease in blood loss, postoperative complications, and days of hospital stay, with comparable short-term oncologic results.8–13 These results were confirmed in a recently published randomized controlled trial where MIE was compared with open transthoracic esophagectomy.14 However, open transthoracic esophagectomy remains the gold standard worldwide for the treatment of esophageal cancer.7 In 2003, robot-assisted minimally invasive thoraco-laparoscopic esophagectomy (RAMIE) was developed at the University Medical Center Utrecht (UMC Utrecht), Utrecht, The Netherlands.15 Robot-assisted thoraco-laparoscopic esophagectomy facilitates complex minimally invasive procedures with an enlarged, three-dimensional (3D) field of view. The articulated instruments facilitate dissection with seven degrees of freedom.13,15–18

niversity Medical Center Utrecht (UMC Utrecht), Utrecht, The Netherlands.15 Robot-assisted thoraco-laparoscopic esophagectomy facilitates complex minimally invasive procedures with an enlarged, three-dimensional (3D) field of view. The articulated instruments facilitate dissection with seven degrees of freedom.13,15–18 From our first experience, reported in 2006 and 2009, it was concluded that RAMIE is a feasible and safe technique, associated with reduced blood loss, shorter intensive care unit (ICU) stay, and a lower percentage of cardiopulmonary complications compared with literature reports of open transthoracic esophagectomy.6,15,16 Following these initial reports of RAMIE, the current article presents our subsequent series with a focus on long-term oncologic results. Methods Patients Between June 2007 and September 2011, consecutive patients with potentially curative resectable esophageal cancer were operated on in the UMC Utrecht. In our institute, transthoracic esophagectomy is the standard treatment for patients with esophageal cancer. The standard neoadjuvant treatment for patients with esophageal adenocarcinoma was preoperative chemotherapy [epirubicin, cisplatin and capecitabine (ECC)].19 Patients with esophageal squamous cell carcinoma underwent preoperative chemoradiotherapy (carboplatin and taxol + 41.4 Gy).20 Data on surgical procedures were registered prospectively in the operating room. All complications and follow-up were registered in a prospective surgical database.

n and capecitabine (ECC)].19 Patients with esophageal squamous cell carcinoma underwent preoperative chemoradiotherapy (carboplatin and taxol + 41.4 Gy).20 Data on surgical procedures were registered prospectively in the operating room. All complications and follow-up were registered in a prospective surgical database. We prospectively recorded baseline characteristics and routine diagnostic work-up, including use and results of upper endoscopy, endoscopic ultrasound, computed tomography (CT) of the thorax and abdomen, and ultrasound of the neck region. Positron emission tomography (PET) scanning with fine-needle aspiration of the suspected lymph nodes was used at indication and recorded prospectively. All patients were discussed at a multidisciplinary oncology board meeting. Patients received standard postoperative follow-up at the outpatient department. Patients visited the outpatient department at 6 weeks and 3, 6, 9, and 12 months in the first year, and in the second, third, fourth, and fifth year postoperatively. Patients received follow-up every 6 months. In case symptoms of tumor recurrence occurred, patients underwent a CT of the thorax and abdomen. All patients had at least 29 months of follow-up and were followed for 5 years postoperatively.

the first year, and in the second, third, fourth, and fifth year postoperatively. Patients received follow-up every 6 months. In case symptoms of tumor recurrence occurred, patients underwent a CT of the thorax and abdomen. All patients had at least 29 months of follow-up and were followed for 5 years postoperatively. Operative Procedure The operative technique of thoraco-laparoscopic esophagectomy with two-field lymphadenectomy has been previously described.15,16 For the thoracic phase, the patient is positioned in the left lateral decubitus position, and tilted 45° towards the prone position. The trocar arrangement during the robot-assisted thoracoscopic and laparoscopic phases is shown in electronic supplementary Fig. S1.15 Robot-assisted esophagectomy included a thoracic lymphadenectomy, which included the right-sided paratracheal (lymph node station 2R), tracheobronchial (station 4), aortopulmonary window (lymph nodes in the window dorsal to the aortic arch, cranially to the left main bronchus up until the pulmonary artery, station 5), carinal (station 7), and perioesophageal (station 8) lymph nodes.15 The patient was placed in the supine position thereafter to facilitate a laparoscopic gastric mobilization, truncal lymph node dissection, and gastric tube formation with cervical hand-sewn end-to-side esophagogastrostomy.21

Operative Procedure The operative technique of thoraco-laparoscopic esophagectomy with two-field lymphadenectomy has been previously described.15,16 For the thoracic phase, the patient is positioned in the left lateral decubitus position, and tilted 45° towards the prone position. The trocar arrangement during the robot-assisted thoracoscopic and laparoscopic phases is shown in electronic supplementary Fig. S1.15 Robot-assisted esophagectomy included a thoracic lymphadenectomy, which included the right-sided paratracheal (lymph node station 2R), tracheobronchial (station 4), aortopulmonary window (lymph nodes in the window dorsal to the aortic arch, cranially to the left main bronchus up until the pulmonary artery, station 5), carinal (station 7), and perioesophageal (station 8) lymph nodes.15 The patient was placed in the supine position thereafter to facilitate a laparoscopic gastric mobilization, truncal lymph node dissection, and gastric tube formation with cervical hand-sewn end-to-side esophagogastrostomy.21 Postoperative Management Mechanical ventilation was continued until patients were transferred to the ICU, where they were extubated 2–3 h after ending the operation. After day 1, patients were transferred to the medium care unit (MCU) and then to the surgical ward on postoperative day 2.

The patient was placed in the supine position thereafter to facilitate a laparoscopic gastric mobilization, truncal lymph node dissection, and gastric tube formation with cervical hand-sewn end-to-side esophagogastrostomy.21 Postoperative Management Mechanical ventilation was continued until patients were transferred to the ICU, where they were extubated 2–3 h after ending the operation. After day 1, patients were transferred to the medium care unit (MCU) and then to the surgical ward on postoperative day 2. All patients were placed on a nil-by-mouth routine with enteral tube feeding by a needle-catheter jejunostomy on the first 7 days postoperatively. Nasogastric tubes were routinely placed. No postoperative swallow tests were performed as the sensitivity rate of detecting leakage was considered to be too low to change postoperative decision making.22 In the absence of signs of anastomotic dehiscence, patients started with sips of water and the oral intake was gradually increased to solid food. There was no enhanced recovery program. Postoperative Complications All complications were graded using the modified Clavien–Dindo classification (MCDC) of surgical complications. All reported complications were grade 2 and higher.23

All patients were placed on a nil-by-mouth routine with enteral tube feeding by a needle-catheter jejunostomy on the first 7 days postoperatively. Nasogastric tubes were routinely placed. No postoperative swallow tests were performed as the sensitivity rate of detecting leakage was considered to be too low to change postoperative decision making.22 In the absence of signs of anastomotic dehiscence, patients started with sips of water and the oral intake was gradually increased to solid food. There was no enhanced recovery program. Postoperative Complications All complications were graded using the modified Clavien–Dindo classification (MCDC) of surgical complications. All reported complications were grade 2 and higher.23 Pathological Analysis The resected specimen was evaluated using a standard protocol, with emphasis on resection margins, tumor type, extension of the tumor, and the presence of lymph nodes. The 7th edition of the Union for International Cancer Control (UICC) was used for TNM classification, tumor grade, and stage grouping.24 The (circumferential) resection margins were evaluated using the College of American Pathologists (CAP) criteria.25

umor type, extension of the tumor, and the presence of lymph nodes. The 7th edition of the Union for International Cancer Control (UICC) was used for TNM classification, tumor grade, and stage grouping.24 The (circumferential) resection margins were evaluated using the College of American Pathologists (CAP) criteria.25 Statistical Analysis Statistical analysis was performed using SPSS version 20.0 (IBM Corporation, Armonk, NY, USA). A p value of <0.05 was considered to be statistically significant. All skewed continuous data were presented as medians and ranges. Survival time was calculated as the duration from the day of surgery to the date of death or date of last follow-up. Disease-free interval was calculated from the day of surgery to the day of definitive diagnosis of recurrent tumor. Results Between June 2007 and September 2011, a total of 123 consecutive patients with potentially curative resectable esophageal cancer were eligible for transthoracic esophagectomy. In seven patients with locally advanced T4 tumors, an indication for open transthoracic esophagectomy was made preoperatively. Intraoperatively, irresectable disease was observed in 8 patients, leaving 108 patients eligible for RAMIE.

curative resectable esophageal cancer were eligible for transthoracic esophagectomy. In seven patients with locally advanced T4 tumors, an indication for open transthoracic esophagectomy was made preoperatively. Intraoperatively, irresectable disease was observed in 8 patients, leaving 108 patients eligible for RAMIE. The baseline characteristics of patients are summarized in electronic supplementary Table S1. The patients included 76 men and 32 women, with a median age of 62 years (range 42–78) and a body mass index (BMI) of 26 (range 16–36 kg/m2 ). The majority of patients (78 %) were clinically staged as cT3 and higher, and 68 % of patients had clinically positive nodal disease (cN1–N3). Co-morbidity, consisting of a history of vascular, cardiac, pulmonary, and oncologic disease, was observed frequently within this cohort.

index (BMI) of 26 (range 16–36 kg/m2 ). The majority of patients (78 %) were clinically staged as cT3 and higher, and 68 % of patients had clinically positive nodal disease (cN1–N3). Co-morbidity, consisting of a history of vascular, cardiac, pulmonary, and oncologic disease, was observed frequently within this cohort. In 20 patients (19 %), conversion to an open transthoracic or open transhiatal procedure was needed. Conversion to thoracotomy (n = 11) was necessary due to bulky adhesive tumor in the mediastinum (n = 4), insufficient collapse of the right lung (n = 2), or inadequate thoracoscopic trocar position (n = 1). Four patients had bleeding that could not be controlled thoracoscopically (n = 4). One patient had bleeding from the bronchial artery, two patients had bleeding from the azygos vein, and one patient had an iatrogenic lung bleed. Conversion to a transhiatal procedure (n = 9) was necessary due to insufficient collapse of the right lung (n = 6), inadequate thoracoscopic port position (n = 1), pleural adhesions (n = 1), or enlarged right cardiac atrium (unusual anatomy) (n = 1).

g from the azygos vein, and one patient had an iatrogenic lung bleed. Conversion to a transhiatal procedure (n = 9) was necessary due to insufficient collapse of the right lung (n = 6), inadequate thoracoscopic port position (n = 1), pleural adhesions (n = 1), or enlarged right cardiac atrium (unusual anatomy) (n = 1). Conversion of the laparoscopic abdominal phase to laparotomy was required in three patients due to bleeding that could not be controlled laparoscopically (n = 1), locally advanced tumor requiring total gastrectomy with colonic interposition (n = 1), or very low position of the greater curvature (n = 1). Patients who underwent intraoperative conversion did not statistically differ in baseline characteristics from patients who underwent a full RAMIE. There was a significant decrease in the percentage of conversions between the first group of 54 patients and the second group of 54 patients (13 [24 %] vs. 7 [13 %], respectively; p < 0.001). Operative Results The operative data of 108 patients are shown in Table 1. The median duration of the total procedure was 381 min (range 264–550), and the thoracoscopic phase (88 patients) had a median duration of 175 min (range 108–241). There was a significant decrease in thoracoscopic operative time between the first group of 44 patients and the second group of 44 patients who completed the thoracic phase thoracoscopically (199 min vs. 166 min, respectively; p < 0.001).Table 1 Patient operative data (n = 108)

dian duration of 175 min (range 108–241). There was a significant decrease in thoracoscopic operative time between the first group of 44 patients and the second group of 44 patients who completed the thoracic phase thoracoscopically (199 min vs. 166 min, respectively; p < 0.001).Table 1 Patient operative data (n = 108) Total operating room time [min; median (range)] 381 (264–636) Thoracoscopic phase [median (range)] 175 (108–281) Total blood loss [ml; median (range)] 340 (50–3800) Conversion thoracoscopy 20 (19) Reason for conversion Respiratory problems 8 (7) Bleeding 4 (4) Bulky tumor 4 (4) Trocar problems 2 (2) Pleural adhesions 1 (1) Unusual anatomy 1 (1) Conversion laparoscopy 3 (3) Reason for conversion Advanced tumor 1 (1) Bleeding 1 (1) Unusual anatomy 1 (1) Data are expressed as n (%) unless otherwise specified Postoperative Results Postoperative data are summarized in Table 2. An uncomplicated postoperative course was observed in 37 (34 %) patients, and pulmonary complications were most common. Pneumonia was diagnosed and treated in 36 (33 %) patients, and anastomotic leakage of the esophagogastrostomy was seen in 20 (19 %) patients, of whom 6 (6 %) also had intrathoracic manifestation. Chylothorax was seen in 19 (18 %) patients; in 15 of these patients the leakages were low-volume and could be treated conservatively, showing that the leakage was only from small side branches of the thoracic duct.Table 2 Postoperative data (n = 108)

19 %) patients, of whom 6 (6 %) also had intrathoracic manifestation. Chylothorax was seen in 19 (18 %) patients; in 15 of these patients the leakages were low-volume and could be treated conservatively, showing that the leakage was only from small side branches of the thoracic duct.Table 2 Postoperative data (n = 108) Uncomplicated procedures 37 (34) Complications 71 (66) Pulmonary 36 (33) Pneumonia 36 (33) Atelectasis 6 (6) Anastomotic leakage 20 (19) Intrathoracic manifestations 6 (6) Chylothorax 19 (18) Vocal cord paralysisa 10 (9) Cardiac 10 (9) Atrial fibrillation 9 (8) Myocardial infarction 1 (1) Wound infection 7 (6) Thromboembolic event 6 (6) Pneumothorax 6 (6) Otherb 3 (3) In-hospital death 5 (5) ICU stay [days; median (range)] 1 (1–76) Hospital stay [days; median (range)] 16 (9–123) Data are expressed as n (%) unless otherwise specified ICU intensive care unit a 8 temporary, 2 permanent b 1 omentum necrosis, 1 tracheoesophageal fistula, 1 bleeding Vocal-cord paralysis occurred in ten (9 %) patients, and paralysis was temporary in eight of these ten patients. The permanent recurrence paralysis rate was 2 %. Wound infections were seen in seven (6 %) patients; five patients were diagnosed with a cervical wound infection, of whom one patient also had a thoracic wound infection. The remaining two patients had abdominal wound infections. Postoperative pneumothorax requiring additional chest tube placement was seen in six (6 %) patients, and thromboembolic complications were seen in 6 % of patients.

agnosed with a cervical wound infection, of whom one patient also had a thoracic wound infection. The remaining two patients had abdominal wound infections. Postoperative pneumothorax requiring additional chest tube placement was seen in six (6 %) patients, and thromboembolic complications were seen in 6 % of patients. Patients were ventilated at the ICU for a median of 0 days (range 0–64). Median ICU stay was 1 day (range 1–76) and overall postoperative hospital stay was 16 days (range 9–123). In-hospital mortality was 5 % (four patients). One patient died from a myocardial infarction, one from a tracheo–neo-esophageal fistula, one from anastomotic leakage with respiratory insufficiency, and one from a mediastinal septic bleed following anastomotic leakage.

toperative hospital stay was 16 days (range 9–123). In-hospital mortality was 5 % (four patients). One patient died from a myocardial infarction, one from a tracheo–neo-esophageal fistula, one from anastomotic leakage with respiratory insufficiency, and one from a mediastinal septic bleed following anastomotic leakage. Histopathological Results An overview of the histopathological results is shown in Table 3. The majority of tumors were adenocarcinomas (78 %). In ten (9 %) patients, no viable tumor cells were detected in the resected specimen, corresponding to a pathological complete response (pCR) rate to neoadjuvant therapy of 14 %. The majority of tumors were located in the distal esophagus or at the gastroesophageal junction (GEJ) (85 %). In 102 (94 %) patients a radical resection (R0) was achieved. No gross irradical resections (R2 resections) were performed. In 108 operations, 2794 lymph nodes were retrieved, and the median number of lymph nodes was 26 (range 5–53). In total, 264 positive lymph nodes were dissected, with a median of one positive lymph node (range 0–22). The distribution of dissected lymph nodes is shown in electronic supplementary Fig. S2. In total, 15 % of all patients had lymph node metastases located at the subcarinal level and higher.Table 3 Histopathological data (n = 108)

positive lymph nodes were dissected, with a median of one positive lymph node (range 0–22). The distribution of dissected lymph nodes is shown in electronic supplementary Fig. S2. In total, 15 % of all patients had lymph node metastases located at the subcarinal level and higher.Table 3 Histopathological data (n = 108) Histological type Adenocarcinoma 78 (72) Squamous cell carcinoma 20 (19) No viable tumor cells 10 (9) Site of tumor Mid or upper esophageal 16 (15) Lower esophageal and GEJ 92 (85) Radicality R0 103 (95) R1 5 (5) No. of retrieved LNs [median (range)] 2794 [26 (5–57)] No. of positive LNs [median (range)] 264 [1 (0–22)] Pathological T stage (%) pT0 10 (9) pT1 20 (19) pT2 11 (10) pT3 65 (60) pT4a 2 (2) Pathological N stage (%) pN0 48 (44) pN1 30 (28) pN2 20 (19) pN3 10 (9) Data are expressed as n (%) unless otherwise specified GEJ gastroesophageal junction, LNs lymph nodes Recurrence and Outcome At the time of analysis, a median of 58 months after surgery, all patients had undergone esophagectomy at least 29 months previously. No patients were lost to follow-up, and median overall survival was 29 months. Kaplan–Meier curves for overall survival are shown in Fig. 1. Overall 5-year survival was 42 %.Fig. 1 Kaplan–Meier curves for overall survival (months)

hs after surgery, all patients had undergone esophagectomy at least 29 months previously. No patients were lost to follow-up, and median overall survival was 29 months. Kaplan–Meier curves for overall survival are shown in Fig. 1. Overall 5-year survival was 42 %.Fig. 1 Kaplan–Meier curves for overall survival (months) Of 108 patients, 5 died postoperatively; therefore, 103 patients were included in the recurrence analysis. Median disease-free survival was 21 months. In 42 patients (52 %), no signs of recurrent disease were observed after a median follow-up of 34 months. The remaining 39 patients developed symptomatic recurrent disease. In 52 of 103 patients (51 %), no signs of recurrent disease were observed after a median follow-up of 34 months. The remaining 51 patients developed symptomatic recurrent disease. The first site of symptomatic tumor recurrence was locoregional only in 6 (6 %) patients, systemic only in 31 (30 %) patients, and combined in 14 (14 %) patients (electronic supplementary Table S2). Kaplan–Meier curves for disease-free survival are shown in Fig. 2.Fig. 2 Kaplan–Meier curves for disease-free survival (months)

rst site of symptomatic tumor recurrence was locoregional only in 6 (6 %) patients, systemic only in 31 (30 %) patients, and combined in 14 (14 %) patients (electronic supplementary Table S2). Kaplan–Meier curves for disease-free survival are shown in Fig. 2.Fig. 2 Kaplan–Meier curves for disease-free survival (months) Discussion This article presents our experience with RAMIE, using a new cohort, following our initial reports in 2006 and 2009 which showed this technique to be feasible and safe.15,16 In the current group of consecutive patients we focused on oncologic long-term follow-up. RAMIE was shown to be effective, with a high percentage of R0 radical resections (95 %) and adequate lymphadenectomy. RAMIE provided local control, with a low percentage of local recurrence. The high percentage of radical resections in our cohort with a majority of locally advanced T3 tumors (60 %) may be the result of the robotic surgical approach. Mainly, the 3D, magnified surgical view combined with the high degree of freedom of the articulating surgical instruments, facilitates precise dissection in a confined operating space.18

cal resections in our cohort with a majority of locally advanced T3 tumors (60 %) may be the result of the robotic surgical approach. Mainly, the 3D, magnified surgical view combined with the high degree of freedom of the articulating surgical instruments, facilitates precise dissection in a confined operating space.18 Nodal-positive disease (pN+) was observed in 56 % of all patients. A proper mediastinal lymphadenectomy was performed, including the right-sided paratracheal (lymph node station 2R), tracheobronchial (station 4), aortopulmonary window (station 5), carinal (station 7), and perioesophageal (station 8) lymph nodes, with a median of 26 dissected lymph nodes. This number is comparable to a series of open transthoracic esophagectomies from the literature.6 For conventional MIE, the median number of dissected lymph nodes was 21. Overall survival of patients who underwent RAMIE was comparable to the results following MIE.26,27

nodes, with a median of 26 dissected lymph nodes. This number is comparable to a series of open transthoracic esophagectomies from the literature.6 For conventional MIE, the median number of dissected lymph nodes was 21. Overall survival of patients who underwent RAMIE was comparable to the results following MIE.26,27 For recurrence, in this study the results following RAMIE with 65 % neoadjuvant treatment were comparable with the results reported for open esophagectomy, in which all patients received neoadjuvant chemoradiotherapy.28 The first site of symptomatic tumor recurrence was locoregional, or in the locoregional lymph nodes, in only 6 % of all cases. This is comparable with results after chemoradiotherapy, where locoregional recurrence was observed in 7 % of all cases.28 Distant metastases were observed in 30 % of all patients compared with 28 % for patients who underwent neoadjuvant chemoradiotherapy.28 The percentage of patients who had simultaneous locoregional recurrence and systemic metastases was 14 % in our cohort and 13 % after neoadjuvant chemoradiotherapy.28

% of all cases.28 Distant metastases were observed in 30 % of all patients compared with 28 % for patients who underwent neoadjuvant chemoradiotherapy.28 The percentage of patients who had simultaneous locoregional recurrence and systemic metastases was 14 % in our cohort and 13 % after neoadjuvant chemoradiotherapy.28 Pneumonia was the most observed complication following RAMIE (34 % of patients). We compared our results with a recent randomized controlled trial where patients with resectable esophageal cancer were randomized between neoadjuvant chemoradiation and surgery alone. In this trial, only open esophagectomies were included, showing a pneumonia rate of 44 %.20 Another recent randomized controlled trial compared conventional MIE with open transthoracic esophagectomy.14 Results from this trial showed a reduced pulmonary complication rate in the MIE group compared with the open group.14 The percentage of in-hospital pulmonary infections after MIE in that trial was lower (12 %) than in our study;14 however, different definitions of postoperative pneumonia were used. Our definition of pneumonia was defined as the decision to treat suspected pneumonia (MCDC, grade II),23 while the definition of pneumonia used in the randomized controlled trial was more strict (infiltrate on pulmonary radiography combined with a positive sputum culture), leading to a lower percentage of pneumonia. Applying this definition to our cohort yields a pneumonia rate of 18 %, which is comparable to MIE.14 Reporting of postoperative pneumonia and postoperative outcomes after esophagectomy in general are heterogeneous and inconsistent. This makes comparison between different studies difficult and a consensus approach to reporting clinical outcomes should be considered.29,30

onia rate of 18 %, which is comparable to MIE.14 Reporting of postoperative pneumonia and postoperative outcomes after esophagectomy in general are heterogeneous and inconsistent. This makes comparison between different studies difficult and a consensus approach to reporting clinical outcomes should be considered.29,30 In addition to the aforementioned advantages of RAMIE, there were also disadvantages of RAMIE, such as the high costs of acquisition of the Da Vinci surgical system, disposable instruments, and a prolonged operative time compared with open esophagectomy.18 The introduction of RAMIE in a hospital needs careful proctoring by surgeons skilled and trained in RAMIE to reduce postoperative complications and to facilitate a steep learning curve.15 Centralization of robotic surgery in high-volume centers leads to a lower rate of postoperative complications and more efficient use of operating time.31 In this article we describe a decrease in thoracoscopic operative time between the first group of 43 patients and the second group of 42 patients (199 min vs. 166 min, respectively; p < 0.001), emphasizing the learning curve. The median duration of the full procedure is 381 min. We are currently performing the RAMIE procedure within 6 h. Furthermore, a significant decrease in the percentage of conversions was observed between the first group of 54 patients and the second group of 54 patients (13 [24 %] vs. 7 [13 %], respectively; p < 0.001). Currently our RAMIE conversion RATE is 4 %.

381 min. We are currently performing the RAMIE procedure within 6 h. Furthermore, a significant decrease in the percentage of conversions was observed between the first group of 54 patients and the second group of 54 patients (13 [24 %] vs. 7 [13 %], respectively; p < 0.001). Currently our RAMIE conversion RATE is 4 %. Our results from robot-assisted esophagectomy are in concordance with a recently published systematic review,18 which included nine articles (130 cases) describing robot-assisted esophagectomy. The level of evidence for RAMIE was suboptimal and was based on case series or expert opinions only (level 4 or 5).18 The aforementioned systematic review strongly emphasized the need for well-conducted randomized controlled trials and long-term survival studies within a framework of measured and comparable outcomes to prove the superiority of RAMIE over the worldwide current standard of open transthoracic esophagectomy.18 Therefore, we initiated the ROBOT trial (ClinicalTrial.gov identifier: NCT01544790) in January 2012 to compare RAMIE with open transthoracic esophagectomy.32 Conclusions In a cohort of Western European patients with advanced esophageal cancer, RAMIE with two-field lymphadenectomy was shown to be feasible and safe. Furthermore, RAMIE was shown to be oncologically effective, with a high percentage of R0 radical resections with adequate lymphadenectomy. RAMIE provided adequate local control, with a low percentage of local recurrence. Electronic Supplementary Material Supplementary material 1 (DOCX 141 kb)

Esophageal carcinoma is the sixth leading cause of cancer-related mortality worldwide and the incidence is rapidly increasing.1,2 Multimodality treatment combining neoadjuvant chemo(radio)therapy and surgical resection has improved the prognosis for resectable nonmetastatic disease;3 however, more than half of the patients develop recurrence within 3 years after treatment with curative intent.4–7 The prognosis of recurrent esophageal cancer is poor, with a median survival of 3–10 months after developing a recurrence.4,8–10 Therefore, detecting prognostic factors affecting post-recurrence survival and determining effectiveness of treatment strategies for recurrence are of high importance. Treatment can be attempted in a fair number of patients with recurrent disease and may include chemotherapy, radiotherapy, surgery, or a combination.9,11,12 However, the optimal treatment strategy for esophageal cancer patients with recurrent disease is not yet established and patients respond differently to treatment, with a wide range in long-term survival.12 The main aim of this study was to investigate prognostic factors that affect survival in patients diagnosed with recurrent disease after prior esophagectomy with curative intent for esophageal carcinoma. In addition, a second aim was to evaluate the different treatment strategies applied.

Esophageal carcinoma is the sixth leading cause of cancer-related mortality worldwide and the incidence is rapidly increasing.1,2 Multimodality treatment combining neoadjuvant chemo(radio)therapy and surgical resection has improved the prognosis for resectable nonmetastatic disease;3 however, more than half of the patients develop recurrence within 3 years after treatment with curative intent.4–7 The prognosis of recurrent esophageal cancer is poor, with a median survival of 3–10 months after developing a recurrence.4,8–10 Therefore, detecting prognostic factors affecting post-recurrence survival and determining effectiveness of treatment strategies for recurrence are of high importance. Treatment can be attempted in a fair number of patients with recurrent disease and may include chemotherapy, radiotherapy, surgery, or a combination.9,11,12 However, the optimal treatment strategy for esophageal cancer patients with recurrent disease is not yet established and patients respond differently to treatment, with a wide range in long-term survival.12 The main aim of this study was to investigate prognostic factors that affect survival in patients diagnosed with recurrent disease after prior esophagectomy with curative intent for esophageal carcinoma. In addition, a second aim was to evaluate the different treatment strategies applied. Methods Patients In this single-center cohort study, patients were selected from a prospectively assembled database at the University Medical Center Utrecht, Utrecht, The Netherlands. Between October 2003 and December 2013, a total of 379 consecutive patients underwent esophagectomy with curative intent for esophageal carcinoma. Patients with an unresectable tumor (cT4b) or metastatic disease (M1) detected intraoperatively were excluded (n = 22), as were patients deceased within 90 days after surgery or during hospitalization (n = 22). Of the remaining 335 patients, 171 were diagnosed with recurrent disease and were included in the current study. All patients were discussed at a multidisciplinary tumor board meeting preoperatively, postoperatively, and after developing recurrent disease. Institutional Review Board approval was obtained, and the informed consent requirement was waived for this study.

with recurrent disease and were included in the current study. All patients were discussed at a multidisciplinary tumor board meeting preoperatively, postoperatively, and after developing recurrent disease. Institutional Review Board approval was obtained, and the informed consent requirement was waived for this study. Treatment Eligible patients with locally advanced disease (cT ≥2 or cN+) and without clinical evidence of metastatic disease (cM0) received either perioperative chemotherapy or neoadjuvant chemoradiation according to the Dutch guidelines. Eligible patients were >18 years of age, had a WHO performance status ≤2, and did not lose >10 % of their body weight. Before 1 June 2012, the standard treatment for patients with esophageal carcinoma consisted of perioperative chemotherapy (epirubicin, cisplatinum, and 5-fluorouracil),14 and after that patients underwent neoadjuvant chemoradiation (carboplatin AUC2 and paclitaxel 50 mg/m2 weekly during 5 weeks concomitant with 41.4 Gy (23 × 1.8 Gy).3 Before 2008, neoadjuvant therapy was not part of the standard protocol and most patients were operated on without this treatment. Patients not eligible for neoadjuvant treatment were treated with esophageal resection alone. After esophagectomy with en bloc lymphadenectomy, all patients underwent gastric tube reconstruction with a left-sided cervical anastomosis.

t of the standard protocol and most patients were operated on without this treatment. Patients not eligible for neoadjuvant treatment were treated with esophageal resection alone. After esophagectomy with en bloc lymphadenectomy, all patients underwent gastric tube reconstruction with a left-sided cervical anastomosis. Histopathological Analysis The resected specimens were reviewed by experienced pathologists in accordance with the TNM-7 staging system of the AJCC.13 Resection margins were evaluated using the definitions of the College of American Pathologists.15,16

t of the standard protocol and most patients were operated on without this treatment. Patients not eligible for neoadjuvant treatment were treated with esophageal resection alone. After esophagectomy with en bloc lymphadenectomy, all patients underwent gastric tube reconstruction with a left-sided cervical anastomosis. Histopathological Analysis The resected specimens were reviewed by experienced pathologists in accordance with the TNM-7 staging system of the AJCC.13 Resection margins were evaluated using the definitions of the College of American Pathologists.15,16 Follow-Up and Definition of Recurrence After esophagectomy, patients were followed at the outpatient clinic with an interval of 3 months in the first year, 6 months in the second year, and 12 months thereafter until discharge after 5 years of follow-up, which consisted of medical history and physical examination. In case of clinical suspicion of tumor recurrence, diagnostic imaging was performed. Recurrence was confirmed by histopathological biopsy or clinical follow-up, and only the initial number and sites of recurrences were evaluated. The pattern of recurrence was classified as locoregional, distant, or a combination of both. Recurrences at the anastomotic site or within the area of previous resection and nodal clearance in the mediastinum or upper abdomen were classified as locoregional recurrence, while distant recurrence was defined as recurrence in distant organs, pleura or peritoneal cavity, or distant lymph nodes. Disease-free survival was defined as the time between the day of surgery and day of recurrent disease, and post-recurrence survival was defined as the time between the first recurrence and death or last follow-up.

currence was defined as recurrence in distant organs, pleura or peritoneal cavity, or distant lymph nodes. Disease-free survival was defined as the time between the day of surgery and day of recurrent disease, and post-recurrence survival was defined as the time between the first recurrence and death or last follow-up. Treatment of Recurrence Treatment for recurrent disease was discussed at a multidisciplinary tumor board meeting and was recommended if the patient was eligible. General considerations regarding eligibility included patient condition, location of recurrences, prior toxicity from chemotherapy or radiotherapy, and patient’s wish. Treatment consisted of chemotherapy, radiotherapy, and/or surgery focused on tumor reduction. Radiotherapy focused on tumor reduction was defined as radiotherapy with a radiation dose >30 Gy, excluding palliative radiotherapy for bone metastases. In all other cases, patients were treated with best supportive care.

ish. Treatment consisted of chemotherapy, radiotherapy, and/or surgery focused on tumor reduction. Radiotherapy focused on tumor reduction was defined as radiotherapy with a radiation dose >30 Gy, excluding palliative radiotherapy for bone metastases. In all other cases, patients were treated with best supportive care. Statistical Analysis To assess prognostic factors for post-recurrence survival, univariable and multivariable analyses by means of Cox proportional hazard models were used, providing hazard ratios (HRs) with 95 % confidence intervals (CIs). All variables with a p value <0.20 in univariable analysis were entered in a multivariable analysis. Kaplan–Meier survival curves were constructed for the prognostic factors that remained significantly associated with post-recurrence survival in multivariable analysis. A p value <0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS version 21 for Windows (IBM Corporation, Armonk, NY, USA).

l curves were constructed for the prognostic factors that remained significantly associated with post-recurrence survival in multivariable analysis. A p value <0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS version 21 for Windows (IBM Corporation, Armonk, NY, USA). Results Patient Characteristics Median follow-up of the 335 consecutive patients treated with esophagectomy during the study period was 22.0 months (range 2–135). Of all patients, 171 (51 %) developed recurrent disease, and these patients were included in the current study. The clinical and histopathological characteristics of these 171 patients are shown in Table 1. Mean age was 63 years (standard deviation 8.8) and most patients were male (n = 131, 77 %). Perioperative chemotherapy was administered in 63 patients (37 %) and neoadjuvant chemoradiation in 35 patients (21 %). The surgical procedure consisted of a transthoracic approach in 132 patients (77 %) and a transhiatal approach in the remaining 39 patients (23 %). Tumor histology was adenocarcinoma in 136 patients (80 %), and histopathology revealed ≥pT3 (n = 129, 75 %) and pN+ disease (n = 123, 72 %) in the majority of patients. Of all patients who developed a recurrence, 139 (81 %) underwent a microscopically radical (R0) resection.Table 1 Clinical and histopathological characteristics of 171 patients with recurrent disease after esophagectomy with curative intent Recurrence (Total = 171)

Results Patient Characteristics Median follow-up of the 335 consecutive patients treated with esophagectomy during the study period was 22.0 months (range 2–135). Of all patients, 171 (51 %) developed recurrent disease, and these patients were included in the current study. The clinical and histopathological characteristics of these 171 patients are shown in Table 1. Mean age was 63 years (standard deviation 8.8) and most patients were male (n = 131, 77 %). Perioperative chemotherapy was administered in 63 patients (37 %) and neoadjuvant chemoradiation in 35 patients (21 %). The surgical procedure consisted of a transthoracic approach in 132 patients (77 %) and a transhiatal approach in the remaining 39 patients (23 %). Tumor histology was adenocarcinoma in 136 patients (80 %), and histopathology revealed ≥pT3 (n = 129, 75 %) and pN+ disease (n = 123, 72 %) in the majority of patients. Of all patients who developed a recurrence, 139 (81 %) underwent a microscopically radical (R0) resection.Table 1 Clinical and histopathological characteristics of 171 patients with recurrent disease after esophagectomy with curative intent Recurrence (Total = 171) n (%) Gender Male 131 (77) Female 40 (23) Age, years (mean ± SD) 63 ± 8.8 ASA score 1 49 (29) 2 95 (56) ≥3 27 (16) Neoadjuvant therapy No neoadjuvant therapy 72 (42) Chemotherapy 63 (37) Radiotherapy 1 (1) Chemoradiation 35 (21) Surgical approach Transthoracic 132 (77) Transhiatal 39 (23) Adjuvant therapy No adjuvant therapy 137 (80) Chemotherapy 34 (20) Histological type Adenocarcinoma 136 (80) Squamous cell carcinoma 34 (20) Other 1 (<1) pT stage T0 9 (5) T1 16 (9) T2 17 (10) T3 121 (71) T4a 8 (5) pN stage N0 48 (28) N1 49 (29) N2 47 (28) N3 27 (16) Number of harvested lymph nodes (median [range]) 20 [2–80] Radicality R0 139 (81) R1 32 (19)

(80) Chemotherapy 34 (20) Histological type Adenocarcinoma 136 (80) Squamous cell carcinoma 34 (20) Other 1 (<1) pT stage T0 9 (5) T1 16 (9) T2 17 (10) T3 121 (71) T4a 8 (5) pN stage N0 48 (28) N1 49 (29) N2 47 (28) N3 27 (16) Number of harvested lymph nodes (median [range]) 20 [2–80] Radicality R0 139 (81) R1 32 (19) ASA American Society of Anesthesiologists, SD standard deviation Pattern of Recurrence Median time to recurrence was 9.0 months (range 1–86) and 164 patients (96 %) developed recurrence within 3 years after surgery. The most common presenting symptoms were pain (n = 38, 22 %), malaise (n = 23, 14 %), dysphagia (n = 21, 12 %), and anorexia (n = 21, 12 %). The diagnosis of recurrent disease was based on computed tomography (CT) findings in 118 patients (69 %), whereas in other patients the diagnosis was made with either endoscopic ultrasound (EUS), upper endoscopy, positron emission tomography (PET), or magnetic resonance imaging (MRI). The type of recurrence and the number of locations are presented in Table 2. Distant recurrence was the most common type of recurrent disease (n = 76, 44 %), and the liver was the most commonly affected site (n = 50, 15 %).Table 2 Location and treatment recurrence of 171 patients with recurrent disease after esophagectomy with curative intent Recurrence (Total = 171)

Pattern of Recurrence Median time to recurrence was 9.0 months (range 1–86) and 164 patients (96 %) developed recurrence within 3 years after surgery. The most common presenting symptoms were pain (n = 38, 22 %), malaise (n = 23, 14 %), dysphagia (n = 21, 12 %), and anorexia (n = 21, 12 %). The diagnosis of recurrent disease was based on computed tomography (CT) findings in 118 patients (69 %), whereas in other patients the diagnosis was made with either endoscopic ultrasound (EUS), upper endoscopy, positron emission tomography (PET), or magnetic resonance imaging (MRI). The type of recurrence and the number of locations are presented in Table 2. Distant recurrence was the most common type of recurrent disease (n = 76, 44 %), and the liver was the most commonly affected site (n = 50, 15 %).Table 2 Location and treatment recurrence of 171 patients with recurrent disease after esophagectomy with curative intent Recurrence (Total = 171) n (%) Type of recurrence Locoregional 27 (16) Distant 76 (44) Combined 68 (40) Location distant recurrence Liver 50 (15) Lung 41 (13) Abdominal lymph nodes 40 (12) Retroperitoneal 40 (12) Bone 30 (9) Other 123 (38) Number of locations with recurrence 1 49 (29) 2–3 62 (36) >3 60 (35) Type of management Treatment focused on tumor reduction 62 (37) Chemotherapy 24 (14) Radiotherapy 11 (6) Chemoradiation 13 (8) Surgery 5 (3) Surgery + chemotherapy 4 (2) Surgery + radiotherapy 4 (2) Other 1 (1) Best supportive care 109 (63) Condition 63 (37) Patient wish 29 (17) Toxicity 4 (2) Location 4 (2) Other 6 (4) Unknown 3 (2)

ent Treatment focused on tumor reduction 62 (37) Chemotherapy 24 (14) Radiotherapy 11 (6) Chemoradiation 13 (8) Surgery 5 (3) Surgery + chemotherapy 4 (2) Surgery + radiotherapy 4 (2) Other 1 (1) Best supportive care 109 (63) Condition 63 (37) Patient wish 29 (17) Toxicity 4 (2) Location 4 (2) Other 6 (4) Unknown 3 (2) Factors Affecting Post-recurrence Survival Median post-recurrence survival was 3.0 months (range 0–112), and the overall 1- and 2-year post-recurrence survival rates were 17 and 7 %, respectively. Nodal status, type of recurrence, number of locations, time to recurrence, and treatment of recurrence were significantly associated with post-recurrence survival in univariable analysis (Table 3; Fig. 1). In multivariable analysis, distant recurrence (HR 2.15, 95 % CI 1.27–3.65; p = 0.005), more than three recurrent tumor locations (HR 2.42, 95 % CI 1.34–4.34; p = 0.003), and treatment (HR 0.29, 95 % CI 0.20–0.44; p < 0.001) were identified as independent prognostic factors associated with post-recurrence survival (Table 3). The median post-recurrence survival of patients with distant and locoregional recurrence was 2.0 months and 12.0 months respectively. This was respectively 2.0 and 6.0 months for patients with more than three recurrent tumor locations and a solitary recurrence. Patients who received treatment focused on tumor reduction had a median post-recurrence survival of 9.0 months compared with 2.0 months in patients treated with best supportive care. Primary tumor characteristics, including neoadjuvant therapy, histological type, pTN stage, and radicality of resection, did not independently influence post-recurrence survival in multivariable analysis.Table 3 Univariable and multivariable analysis of potential prognostic factors for survival after diagnosis of recurrent esophageal carcinoma

tics, including neoadjuvant therapy, histological type, pTN stage, and radicality of resection, did not independently influence post-recurrence survival in multivariable analysis.Table 3 Univariable and multivariable analysis of potential prognostic factors for survival after diagnosis of recurrent esophageal carcinoma HR 95 % CI p-Valuea HR 95 % CI p-Valueb Age (years) 1.02 1.00–1.04 0.055 1.00 0.99–1.02 0.670 Neoadjuvant therapy None Reference – – Reference – – Chemotherapy 1.39 0.98–1.99 0.067 1.02 0.70–1.49 0.936 Radiotherapy 3.45 0.47–25.23 0.222 7.85 0.99–62.54 0.052 Chemoradiation 1.26 0.82–1.94 0.297 0.84 0.50–1.41 0.512 Histological type Adenocarcinoma Reference – – Squamous cell carcinoma 1.24 0.84–1.84 0.272 Other 1.10 0.15–7.93 0.922 pT stage T0 Reference – – Reference – – T1–2 0.47 0.21–1.06 0.067 0.60 0.25–1.41 0.243 T3–4 0.70 0.34–1.45 0.341 0.78 0.34–1.76 0.545 pN stage N0 Reference – – Reference – – N1 1.80 1.18–2.75 0.007 1.50 0.95–2.37 0.080 N2–3 1.35 0.91–1.99 0.131 1.10 0.70–1.73 0.689 Radicality R0 Reference – – R1 1.20 0.81–1.77 0.363 Type of recurrence Locoregional Reference – – Reference – – Distant 2.10 1.30–3.41 0.003 2.15 1.27–3.65 0.005 Combined 2.54 1.55–4.16 <0.001 1.58 0.89–2.81 0.120 Number of locations 1 Reference – – Reference – – 2–3 1.21 0.81–1.79 0.357 1.30 0.83–2.00 0.250 >3 2.20 1.46–3.32 <0.001 2.42 1.34–4.34 0.003 Time to recurrence (months) 0.98 0.96–1.00 0.013 0.99 0.98–1.01 0.263 Treatment of recurrence Best supportive care Reference – – Reference – – Treatment focused on tumor reduction 0.27 0.19–0.38 <0.001 0.29 0.20–0.44 <0.001

Combined 2.54 1.55–4.16 <0.001 1.58 0.89–2.81 0.120 Number of locations 1 Reference – – Reference – – 2–3 1.21 0.81–1.79 0.357 1.30 0.83–2.00 0.250 >3 2.20 1.46–3.32 <0.001 2.42 1.34–4.34 0.003 Time to recurrence (months) 0.98 0.96–1.00 0.013 0.99 0.98–1.01 0.263 Treatment of recurrence Best supportive care Reference – – Reference – – Treatment focused on tumor reduction 0.27 0.19–0.38 <0.001 0.29 0.20–0.44 <0.001 Analysis was performed using a Cox regression model Bold values indicate statistically significant (e.g. p < 0.05). All variables with a p value <0.2 from univariable analysis were used for multivariable analysis HR hazard ratio, CI confidence interval aUnivariable analysis bMultivariable analysis Fig. 1 a Type of recurrence, b number of tumor locations, and c type of management were identified as independent prognostic variables for post-recurrence survival in 171 patients with recurrent disease after curative esophagectomy. Survival curves were plotted using the Kaplan–Meier method

aUnivariable analysis bMultivariable analysis Fig. 1 a Type of recurrence, b number of tumor locations, and c type of management were identified as independent prognostic variables for post-recurrence survival in 171 patients with recurrent disease after curative esophagectomy. Survival curves were plotted using the Kaplan–Meier method Treatment of Recurrence Patients receiving best supportive care (n = 109, 63 %) were mainly either not eligible for treatment due to a poor performance status (n = 63, 37 %) or refused treatment (n = 29, 17 %). Some patients were not eligible due to prior toxicity of the neoadjuvant treatment regimen (n = 4, 4 %) or tumor location (n = 4, 4 %). Treatment focused on tumor reduction was applied in 62 patients (37 %) (Table 2). Patients with locoregional recurrence (n = 19, 70 %) and solitary recurrence (n = 24, 49 %) more often received treatment focused on reduction compared with those with distant recurrence (26, 34 %) and more than three recurrent tumor locations (n = 14, 23 %). Different chemotherapy regimens were administered in 41 patients, with most patients receiving a combination of epirubicin, cisplatin, and capecitabine (n = 20, 48 %). After treatment with chemotherapy only, two patients (5 %) showed a clinically complete tumor regression—one patient had a solitary metastasis in the liver, and the other had a solitary locoregional recurrence in the gastric conduit and truncal node. Both patients were alive at last follow-up (35 and 112 months after diagnosis of recurrence).

motherapy only, two patients (5 %) showed a clinically complete tumor regression—one patient had a solitary metastasis in the liver, and the other had a solitary locoregional recurrence in the gastric conduit and truncal node. Both patients were alive at last follow-up (35 and 112 months after diagnosis of recurrence). In 13 of 171 patients (8 %), surgical resection of the recurrence was performed (Table 4), with most of these patients having a solitary recurrence (n = 9, 69 %) at a distant location (n = 11, 85 %). Surgical resections are outlined in Table 4; five patients (38 %) underwent metastasectomy of a brain lesion. Median post-recurrence survival in patients who underwent resection was 11 months (95 % CI 4.5–17.5), and in 11 of 13 patients (85 %) the resection was performed with curative intent. Of these patients, 4 of 11 (36 %) were still alive at last follow-up, with a follow-up of 5, 46, 53, and 87 months after the diagnosis of their recurrence, whereas the remaining seven patients (64 %) deceased due to disease progression.Table 4 Characteristics, treatment, and survival of 13 patients treated with surgical resection for recurrent esophageal carcinoma

live at last follow-up, with a follow-up of 5, 46, 53, and 87 months after the diagnosis of their recurrence, whereas the remaining seven patients (64 %) deceased due to disease progression.Table 4 Characteristics, treatment, and survival of 13 patients treated with surgical resection for recurrent esophageal carcinoma Case Age, years Sex pTNM stage Time to recurrence (months) Type of recurrence Location recurrence Surgical intervention Other treatments Curative intent Status Survival after recurrence (months) CT RT 1 56 Male T3N2M0 11 Distant Abdominal LN LN resection No No Yes Dead 53 2 44 Male T3N2M0 3 Distant Abdominal wall Inguinal cutane Tumor resection Tumor resection Yes No Yes Dead 9 3 74 Female T4aN2M0 2 Distant Upper leg subcutane Inguinal LN Abdominal wall Abdominal LN Tumor resection LN resection No Yes No Dead 4 4 67 Male T3N2M0 8 Distant Brain, lung, liver Metastasectomy brain lesion No Yes No Dead 5 5 53 Male T0N0M0 21 Distant Brain Metastasectomy brain lesion No Yes Yes Dead 7 6 77 Female T3N0M0 14 Distant Brain Metastasectomy brain lesion No No Yes Dead 1 7 75 Male T1bN0M0 31 Distant Lung Partial pulmonary resection No No Yes Dead 18 8 62 Female T3N0M0 12 Distant Brain Metastasectomy brain lesion No No Yes Dead 4 9 50 Male T3N0M0 32 Distant Vesiculae seminales Excision vesiculae seminales No No Yes Dead 11 10 65 Male T3N3M0 8 Combined Quadriceps muscles Paraesophageal LN Metastasectomy quadriceps muscles Yes No Yes Alive 87 11 56 Male T2N0M0 13 Locoregional Gastric conduit Resection gastric conduit with jejunal reconstruction No No Yes Alive 46 12 65 Male T1aN0M0 10 Distant Liver Hemihepatectomy Yes No Yes Alive 53 13 62 Male T3N0M0 20 Distant Brain Metastasectomy brain lesion No Yes Yes Alive 5

tomy quadriceps muscles Yes No Yes Alive 87 11 56 Male T2N0M0 13 Locoregional Gastric conduit Resection gastric conduit with jejunal reconstruction No No Yes Alive 46 12 65 Male T1aN0M0 10 Distant Liver Hemihepatectomy Yes No Yes Alive 53 13 62 Male T3N0M0 20 Distant Brain Metastasectomy brain lesion No Yes Yes Alive 5 CT chemotherapy, RT radiotherapy, LN lymph node Discussion In this single-center cohort study, 171 patients with recurrent disease after treatment with curative intent for esophageal carcinoma were analyzed and factors affecting post-recurrence survival were evaluated. Distant recurrence and more than three recurrent locations were identified as independent prognostic factors associated with a worse post-recurrence survival, irrespective of primary tumor characteristics. Furthermore, treatment focused on tumor reduction, as opposed to best supportive care, prolonged survival in eligible patients and a selected group of patients were treated curatively.

fied as independent prognostic factors associated with a worse post-recurrence survival, irrespective of primary tumor characteristics. Furthermore, treatment focused on tumor reduction, as opposed to best supportive care, prolonged survival in eligible patients and a selected group of patients were treated curatively. This study confirms the poor prognosis of recurrent esophageal cancer reported in other series4,8,9,10 with a median post-recurrence survival of 3.0 months and a 2-year survival rate of only 7 %. Hence, understanding of the prognostic factors influencing survival is important in identifying patients who could have an improved post-recurrence survival by selecting them for the appropriate treatment. In accordance with the literature, distant recurrence was associated with a worse survival in this study, reflecting aggressive tumor biology.6,12,17 Furthermore, this study showed that patients with more than three recurrent tumor locations had a worse post-recurrence survival compared with those with less involved locations, which could also be explained by the more aggressive behavior of multiple recurrences. The survival of patients with more than three recurrent locations was extremely poor, with a median survival of 2.0 months after the diagnosis of recurrence compared with 6.0 months in patients with a solitary recurrence. The majority of patients had a poor clinical condition at the time of diagnosis of recurrence and were therefore considered ineligible for treatment focused on tumor reduction. The patients who underwent treatment had a significantly prolonged survival, which is likely explained by a combination of appropriate patient selection and treatment effectiveness.

inical condition at the time of diagnosis of recurrence and were therefore considered ineligible for treatment focused on tumor reduction. The patients who underwent treatment had a significantly prolonged survival, which is likely explained by a combination of appropriate patient selection and treatment effectiveness. As has been reported in previous studies,4,9,18 all different treatment strategies resulted in a prolonged survival in the current study. This finding suggests that all patients with recurrent disease should be stimulated to undergo treatment if the condition of patients allows it. Median post-recurrence survival in the treated group was 9.0 months compared with 2.0 months for those who were treated with best supportive care. It needs to be acknowledged that the majority of patients who received best supportive care were not eligible for therapy, causing bias through selection-by-indication in this comparison. Nonetheless, most patients who were not eligible had advanced disease (i.e. distant recurrence or more than three recurrent locations), which reflects high dependency of the patient’s condition on the site and number of recurrent tumors.

ible for therapy, causing bias through selection-by-indication in this comparison. Nonetheless, most patients who were not eligible had advanced disease (i.e. distant recurrence or more than three recurrent locations), which reflects high dependency of the patient’s condition on the site and number of recurrent tumors. Patients were treated with various therapies, of which chemotherapy was the most commonly applied. The benefit of a surgical resection of recurrent esophageal carcinoma is not yet completely elucidated. A few reports showed improved survival after surgical resection; 11,19,20 however, in most studies the resection was combined with either chemotherapy or radiotherapy and was performed in only a small number of patients. Also in this study, a small group of patients (n = 13) underwent resection of their recurrence, the majority (n = 9) of whom had an oligometastasis. Patients with oligometastases represent a special tumor behavior that is likely to gain from local control. In other types of cancer, the current literature also shows a survival benefit with long disease-free survival from local control with surgery for patients with oligometastases.21,22 Importantly, four patients had complete tumor remission after the resection and were still alive at last follow-up. Other studies also reported long-term survival after treatment of recurrent disease for esophageal carcinoma. 11,23–25 These findings suggest that a favorable outcome can be expected after surgical resection in a selected patient group, especially for those with solitary or localized recurrence of esophageal cancer.

p. Other studies also reported long-term survival after treatment of recurrent disease for esophageal carcinoma. 11,23–25 These findings suggest that a favorable outcome can be expected after surgical resection in a selected patient group, especially for those with solitary or localized recurrence of esophageal cancer. Although treatment of recurrence resulted in prolonged survival, the majority of patients (63 %) received best supportive care. This is in contrast with some other studies where the proportion of patients receiving best supportive care ranged from 12 to 44 %.9,11,17,26,27An explanation for the high percentage of best supportive care in this cohort could lie in the follow-up strategy; the current follow-up strategy is based on the existing literature showing that routine diagnostic imaging is of no benefit with regard to survival and costs.28 Furthermore, the consensus-based guidelines from the National Comprehensive Cancer Network also suggest that diagnostic imaging should only be performed when clinically indicated.29 Hence, this follow-up strategy is widely performed in The Netherlands; however, it could have resulted in more advanced recurrent tumor stages at the moment of diagnosis. Since the patient’s condition is largely determined by the number and site of recurrences, patients with multiple metastases are often not eligible for therapy; therefore, the follow-up strategy may need revision according to the findings of the current study. In light of new insights into the concept of oligometastases and the new combined treatment options, we suggest routinely performing a follow-up of patients with PET CT in the first 6–12 months following primary treatment.30 Another explanation for the high ‘best supportive care’ rate could be the large proportion of patients (27 %) who refused any form of treatment. In most other studies, only a fraction of patients did not receive treatment based on patient’s choice.17,26,27According to the results of the current study, eligible patients might be encouraged to have treatment focused on tumor reduction to improve their survival. Unfortunately, no information on quality of life, which is of paramount importance in patients being treated with palliative intent, was obtained from patients who were treated for recurrence.

he current study, eligible patients might be encouraged to have treatment focused on tumor reduction to improve their survival. Unfortunately, no information on quality of life, which is of paramount importance in patients being treated with palliative intent, was obtained from patients who were treated for recurrence. Conclusions Survival after developing a recurrence after esophagectomy with curative intent is poor. Distant recurrence and more than three recurrent locations were identified as independent factors associated with a worse survival, irrespective of primary tumor characteristics. Treatment focused on tumor reduction using various strategies contributed to a prolonged survival in all patients. Hence, stronger focus is needed to improve patient selection for treatment in recurrent esophageal carcinoma. Additionally, in a small group of patients (4 %), curative treatment of recurrent esophageal carcinoma appears possible. K. Parry and E. Visser share first authorship.

Global disparities in healthcare arise from a complex interplay of economic, educational, social, and cultural factors. At the Lancet Commission on Global Surgery 2030 meeting on 17 January 2014 in Boston, Jim King, President of the World Bank, stated that “surgery is an indivisible, indispensable part of healthcare and can help millions of people lead healthier, more productive lives.” In 2015, The Lancet Commission published an extensive report with evidence and solutions for achieving health, welfare, and economic development,1 while another report was published in regard to the delivery of safe, affordable, and timely cancer surgery.2 These two extensive papers were discussed shortly afterwards in editorials in Annals of Surgical Oncology and the European Journal of Surgical Oncology.2,3 Since 2013, cancer has been the second leading cause of death globally after cardiovascular disease, and is expected to increase in all countries due to population growth, aging, and increasing prevalence of risk factors.4 An estimated 20 % of ‘global surgery’ has been denoted as ‘cancer surgery’. In high-income countries (HICs), the standards of evidence-based multimodality cancer treatment (surgery, radiation, and systemic treatment) have been well defined, whereas the standards of surgical and anesthesia care in low- and middle-income countries (LMICs) are stagnating or regressing.1 This disparity in cancer care requires an accurate analysis to improve overall (surgical) cancer care.5

y cancer treatment (surgery, radiation, and systemic treatment) have been well defined, whereas the standards of surgical and anesthesia care in low- and middle-income countries (LMICs) are stagnating or regressing.1 This disparity in cancer care requires an accurate analysis to improve overall (surgical) cancer care.5 In this paper, the difference between cancer surgery, training, and quality assurance in HICs, and the common lack of such provisions in LMICs, are addressed. The potential role of community- and university-based surgical oncologists and surgical oncology societies in fighting the global disparity in cancer care is discussed. Finally, suggestions are made as to the use of new electronic technologies in teaching healthcare workers in LMICs and creating a general awareness of (malignant) diseases with lifestyle risk factors.

ased surgical oncologists and surgical oncology societies in fighting the global disparity in cancer care is discussed. Finally, suggestions are made as to the use of new electronic technologies in teaching healthcare workers in LMICs and creating a general awareness of (malignant) diseases with lifestyle risk factors. The Evolution of Cancer Surgery, Training, and Quality Assurance In most Western countries, long-term survival has doubled for some cancers, such as breast and colon cancer, over the past 40 years through early detection, the use of effective combined treatment modalities, radiation, and/or systemic treatment, reducing the 30-day postoperative mortality due to improved surgical techniques, surgical equipment with less intraoperative and postoperative bleeding, and, last but not least, anesthesia and intensive care facilities. Today, personalized cancer surgery has become a reality with the conservation of the integrity and function of the body and the preservation of quality of life. Compared with LMICs, surgeons in HICs have the ability to stage cancer patients using computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), and/or sentinel lymph node biopsy (SLNB), and to use effective combined treatment modalities with the various radiation techniques, treatment planning, radiation doses, and systemic approaches ranging from hormonal, chemotherapy, and immunotherapy to targeted drug therapy, as well as the various surgical interventions, from conventional surgery and laparoscopic procedures to robotic and image-guided surgery. For many types of cancer, surgery has plateaued as a treatment in regard to morbidity and mortality, local control, and long-term survival, whereas, in palliative care, the role of surgery is explored continuously. New developments are still being made in minimally invasive surgery, such as single-incision laparoscopic surgery, robot-assisted laparoscopic surgery, transanal endoscopic microsurgery, and natural orifice transluminal endoscopic surgery.6 All these procedures are accompanied by steep learning curves and require centralization of the surgical care of these cancer patients.

ive surgery, such as single-incision laparoscopic surgery, robot-assisted laparoscopic surgery, transanal endoscopic microsurgery, and natural orifice transluminal endoscopic surgery.6 All these procedures are accompanied by steep learning curves and require centralization of the surgical care of these cancer patients. There is significant diversity in surgical (oncology) training programs and certification requirements, with an inverse correlation to a country’s income. Almost all HIC and LMIC surgical (sub)specializations are organ-based rather than focused on oncology, with significant variability in the training of surgical oncologists worldwide.7 Although today most LMICs do not have these structured surgical oncology training pathways, at the same time, a few LMICs have had structured surgical oncology fellowships in place for a number of decades now. In HICs, surgeons have taken the lead with regard to further specializing in cancer surgery at high-volume centers, with centralization of cancer surgery in accredited cancer centers.

training pathways, at the same time, a few LMICs have had structured surgical oncology fellowships in place for a number of decades now. In HICs, surgeons have taken the lead with regard to further specializing in cancer surgery at high-volume centers, with centralization of cancer surgery in accredited cancer centers. Thirty-five national surgical oncology societies currently comprise the World Federation of Surgical Oncology Societies (WFSOS), but no international, equally acknowledged surgical oncology curriculum exists.8 The Society of Surgical Oncology (SSO) and European Society of Surgical Oncology (ESSO) have well-defined surgical oncology curricula and, in Mexico, a program with certification was recently started under the auspices of the Consejo Mexicano de Oncología (CMO).9–11 Training programs are also running in other LMICs, but a great diversity exists globally, and training and certification should therefore be streamlined.7 In Europe, the Union Européenne des Médecins Spécialiste (UEMS) represents more than 50 medical disciplines in 34 countries, and the Division of Surgical Oncology was established in 2003 to promote excellence in cancer surgery across Europe, along with European board examinations and certifications.12 Board certification for cancer surgery was introduced in the USA and The Netherlands in 2011 and 2014, respectively.13,14

nes in 34 countries, and the Division of Surgical Oncology was established in 2003 to promote excellence in cancer surgery across Europe, along with European board examinations and certifications.12 Board certification for cancer surgery was introduced in the USA and The Netherlands in 2011 and 2014, respectively.13,14 These board certifications and quality assurance programs improved the surgical oncology outcome in HICs, but LMICs still face the burden and need for more surgical care. The incidence of cancer has increased tremendously in LMICs, leading to a growing discrepancy between the demands and opportunities of surgical care. Too few surgeons are trained in basic surgical oncological procedures.15

rgical oncology outcome in HICs, but LMICs still face the burden and need for more surgical care. The incidence of cancer has increased tremendously in LMICs, leading to a growing discrepancy between the demands and opportunities of surgical care. Too few surgeons are trained in basic surgical oncological procedures.15 Burden of Surgical Cancer Care Advances in (cancer) surgery have been ignored in LMICs. Well-equipped hospitals are present in most capitals of LMICs, but a majority of the surgeons work with limited resources and a majority of their cancer patients are diagnosed at an advanced stage, with limited opportunities for effective cancer treatment. Even if patients in LMICs are treated, they are at high risk of recurrence and disease-related morbidity and mortality. The financial condition of each patient and financial support from family are generally limited as adequate health insurance is obsolete and bureaucratic. With the exception of the big cities, limited advanced diagnostic radiology capabilities are available in LMICs and, in general, there is a lack of (updated) radiation facilities. Furthermore, the Lancet Oncology Commission suggests that investment in and expansion of global access to radiotherapy could save lives and may have positive economic benefits; however, first and foremost, this requires billions of dollars of investments. Second, it also requires education and licensing programs for LMIC physicians specializing in radiation oncology at foreign cancer institutions in HICs.16 Systemic cancer treatment has been inhibited by the lack of well-equipped pharmacies, and effective chemotherapy protocols with targeted therapy are generally not affordable. Complications associated with systemic treatment may add to the burden of care due to the lack of effective antibiotics and/or injectable growth factors. The opportunities that most HIC surgeons have for maintaining the integrity of the patient’s body are not available in LMICs, especially when the patients have advanced cancer. The gap between the healthcare system in HICs and LMICs is not expected to close soon.

effective antibiotics and/or injectable growth factors. The opportunities that most HIC surgeons have for maintaining the integrity of the patient’s body are not available in LMICs, especially when the patients have advanced cancer. The gap between the healthcare system in HICs and LMICs is not expected to close soon. Because surgeons are the primary caregivers for cancer in LMICs, competent surgeons should be trained in a shorter period of time to become ‘basic surgical oncologists’ for the most commonly diagnosed cancers. This training should also include the basic principles of systemic anticancer treatment, as well as palliative care. How can Western surgeons and (surgical) oncologists play a major role in addressing and tackling the burden of surgical cancer care with disproportionate increases in that burden in LMICs? To achieve this role there are, among others, some important conditions to be fulfilled: (i) acknowledge the global surgical oncology disease burden; (ii) commence a global movement through the World Health Organization (WHO), making efficient use of the International Agency for Research on Cancer (IARC) programs, the US National Cancer Institute Center for Global Health, the Lancet Global Surgery Commission, and global cancer consortiums (GCC); (iii) train surgeons to be surgical oncologists through partnership programs with foreign institutions and/or through collaboration with international (surgical) cancer society programs focusing on LMICs, or through Comprehensive Cancer Centers; and (iv) HICs can be involved in cancer research in LMICs.5,17–20 Qualitative and quantitative research should be encouraged in LMICs to support decision making and patient access to healthcare facilities, or to obtain data to make policy makers in LMICs aware of the effects of the collaborations.21 Implementation of these points is an important starting point to provide a structural solution to the anticipated lack of knowledge and practical capabilities in LMICs, rather than the present practice of Western surgeons temporarily providing voluntary support to surgical clinics individually or through welfare organizations. The goal should be to create a situation in LMICs, as summarized in the following proverb: “Give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime.”

e of Western surgeons temporarily providing voluntary support to surgical clinics individually or through welfare organizations. The goal should be to create a situation in LMICs, as summarized in the following proverb: “Give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime.” Opportunities for Education Through Collaboration The WHO has just launched the Global Cancer Country Profiles, which include reference data on cancer mortality and incidence, risk factors, availability of cancer country plans, monitoring and surveillance, primary prevention policies, screening, treatment, and palliative care.22 Programs are related to cancer control, prevention, early detection, and palliative care. National smoking bans and programs promoting obesity rate reduction through lifestyle changes will have a dramatic effect on decreasing the incidence of cancer, cardiovascular diseases, and type II diabetes in HICs and LMICs. More than 168 countries have signed the WHO Framework Convention on Tobacco Control.23 In The Netherlands, for example, an initiative has been launched by physicians and healthcare entrepreneurs to legislate a ban on tobacco purchase and use by any individual born from 2016 onwards. Even smokers agree with this ban for their future children. The WHO also has a global nutrition program with the implementation of the Global database on the Implementation of Nutrition Action (GINA).24 In both programs, HICs can support LMICs in achieving the goals of national campaigns to reduce smoking and decrease the overweight population. The Western world can support these campaigns with the development of specific social media campaigns (see the eHealth and mHealth section).

ation of Nutrition Action (GINA).24 In both programs, HICs can support LMICs in achieving the goals of national campaigns to reduce smoking and decrease the overweight population. The Western world can support these campaigns with the development of specific social media campaigns (see the eHealth and mHealth section). In a recent paper, Brennan discussed the possibility of sophisticated third-year clinical rotation fellowships within a community cancer program for residents from LMICs.5 Regular postgraduate (surgical) cancer courses in LMICs will eventually have an impact on the provision of healthcare. Surgical oncologists or their institution may adopt the program or collaborate with an overseas university or regional hospital and start, through a memorandum of understanding, a collaborative clinical surgical oncology/research initiative. Over several decades, more than 20 Indonesian surgeons, with the support of the Department of Surgery of the University Medical Center Groningen (UMCG) and Dutch Cancer Society, received specific training in the basic principles of surgical (oncologic) procedures and palliative care at the Department of Surgical Oncology, Department of Urology, Department of Neurosurgery, and Department of Plastic Surgery of the UMCG. Some received board certifications prior to their return to Indonesia, but they all later became surgical (oncology) leaders in Indonesia and distributed their knowledge through their institutional surgical training programs and surgical societies (e.g. oncology societies such as Perhimpunan Ahli Bedah Onkologi Indonesia; PERABOI, the Indonesian Society of Surgical Oncology). Recently, a similar residents training program was initiated between the Department of Surgery of the UMCG and the Department of Surgery of the University Hospital Paramaribo in Suriname. This UMCG training program can easily be copied by other institutions. Another possibility is to start a foundation with the goal of improving (surgical) cancer care in a hospital, region, country, or continent.25

nt of Surgery of the UMCG and the Department of Surgery of the University Hospital Paramaribo in Suriname. This UMCG training program can easily be copied by other institutions. Another possibility is to start a foundation with the goal of improving (surgical) cancer care in a hospital, region, country, or continent.25 What are the opportunities for the two largest surgical oncology societies in the fight against cancer, nationally and globally?9,10 The SSO has 2721 members (52 members in LMICs, 2 %) and publishes the Annals of Surgical Oncology (impact factor [IF] 3.943), while the ESSO has 3822 members (49 members in LMICs, 1 %) and publishes the European Journal of Surgical Oncology (IF 3.009). Both societies have reduced their fees for membership and annual meetings for participants from LMICs. Why are they still unable to grant free electronic subscriptions for their journals to surgeons, residents, and medical students in LMICs? This is the way both societies could support our current and future colleagues in LMICs as the time of studying from old-fashioned textbooks is over.

gs for participants from LMICs. Why are they still unable to grant free electronic subscriptions for their journals to surgeons, residents, and medical students in LMICs? This is the way both societies could support our current and future colleagues in LMICs as the time of studying from old-fashioned textbooks is over. eHEALTH and mHEALTH An important instrument for collaboration through education is available today, and there is an opportunity to use modern electronic technology in teaching medical students, physicians, specialists, healthcare workers, and nurses, as well as in providing healthcare. eHealth is the application of information and communication technology in the service of health. Mobile health (mHealth) is the practice of medicine and public health supported by mobile devices. The boost from high-speed internet on smartphones and tablets has contributed to the rapid development of eHealth and mHealth technology, such as Short Message Service (SMS), Multimedia Message Service (MMS), telemonitoring, telecoaching, telecare, teleconsultation, telediagnosis, teleradiology, telesurgery (robotic surgery), teleconferencing (regionally, nationally, internationally) and e-consults. Built-in cameras and video recorders can be used as sensors to measure or track vital signs, such as heart rate and respiration. Today, we have great variation in medical tricorders, portable handheld scanning devices used by consumers, patients, caregivers, nurses, or physicians to (self)diagnose medical conditions within percentages, or show and summarize a person’s health status, and the data can be electronically transferred to anywhere in the world, even from rural areas. FaceTime provides the possibility of rapidly exchanging information between clinicians, such as when facing an urgent intraoperative problem. For quick communication among staff members, WhatsApp is available, a multimedia smartphone application for patient care and academic endorsement. There are no boundaries or disparities with eHealth and mHealth technology.

exchanging information between clinicians, such as when facing an urgent intraoperative problem. For quick communication among staff members, WhatsApp is available, a multimedia smartphone application for patient care and academic endorsement. There are no boundaries or disparities with eHealth and mHealth technology. The implementation of these techniques in LMICs will not be without difficulties. The application of eHealth facilities may be distinguished two ways: support of primary healthcare professionals, including medical students and residents, or education for people living in rural and remote areas. Most LMICs currently have an extensive cellular phone network and availability of Internet technology, but they are facing the challenge of receiving continuous high-quality, affordable, and universally accessible health education. The application of eHealth facilities for both purposes may be limited through a variety of problems. In rural and remote areas in certain parts of the world, these technologies are probably not working as optimally as predicted. Although technological problems may have solutions, there are human factors, such as behavioral changes, system constraints, and privacy limitations, which may set barriers in efforts to provide adequate healthcare information and administer care.26 A recent Cochrane analysis of studies on enhancing the effects of primary health interventions by mobile phone applications, such as SMS and MMS, demonstrated that much is not yet known about the long-term effects or potential negative consequences, although one study showed short-term effects of smoking cessation.27 A study on the effects of a telehealth system in rural and remote areas of Brazil indicated that most of the professionals were satisfied and that the system was cost effective and led to access to specialized healthcare. The main lessons learned from this study were that the system requires a broad collaborative network, must be simple, should have face-to-face components, and should be applied to address the problems for which there is a high service demand; however, the long-term effects of the system require further study.28 In a recent systematic review of the implementation of mobile communication techniques in Africa, the conclusion was that these techniques pose a potential to becoming an important part of the health sector to establish innovative approaches to the delivery of care.

g-term effects of the system require further study.28 In a recent systematic review of the implementation of mobile communication techniques in Africa, the conclusion was that these techniques pose a potential to becoming an important part of the health sector to establish innovative approaches to the delivery of care. The benefits have also been highly recommended, but it is clear that the projects are not a solution to the challenges that health systems face in many African countries. More evidence-based research is necessary in the field of mobile communication technique implementation, especially on a large scale and for a long time.29 The potential of eHealth and mHealth technology in education towards cancer prevention, diagnosis and treatment, and the (surgical) fight against cancer in LMICs, have not been properly investigated.

essary in the field of mobile communication technique implementation, especially on a large scale and for a long time.29 The potential of eHealth and mHealth technology in education towards cancer prevention, diagnosis and treatment, and the (surgical) fight against cancer in LMICs, have not been properly investigated. Prospects and Suggestions No major changes are foreseen in the current global disparities in cancer care or global gross domestic products. Future world population growth will come from the LMICs, i.e. countries with a high economic vulnerability, low life expectancy at birth, low per capita income, low levels of education, and negative effects of urbanization. These countries will also have a tremendous increase in cancer burden by a limited healthcare system. In 2030, 70 % of all cancer deaths will occur in LMICs, and there will be a shift in the distribution of all types of cancer due to increased cancer incidence rates with non-infectious etiology via Western lifestyle changes attributable to economic development. The ratio of cancer incidence to mortality is low in HICs (46 %) and high in LMICs (75 %);30 however, there have been developments that give us a reason to be optimistic. On 4 February 2015, the Health Minister of the Republic of Indonesia, Nila Farid Moeloek, decreed the Commitment to Cancer Management in Indonesia. One of the commitments was to support public regulation for prevention via a healthy lifestyle against cancer.31 Although this is a good national initiative, the impact on overall (surgical) cancer care and oncological outcome will be limited.

ila Farid Moeloek, decreed the Commitment to Cancer Management in Indonesia. One of the commitments was to support public regulation for prevention via a healthy lifestyle against cancer.31 Although this is a good national initiative, the impact on overall (surgical) cancer care and oncological outcome will be limited. To start with international collaboration between countries and societies through the WHO, IARC, US National Cancer Institute Center for Global Health, the Lancet Global Surgery Commission, GCC, or surgical oncology societies (e.g. SSO2 and ESSO3) might be the first step on the way to fighting this cancer burden by way of (i) educational exchange programs to enhance the curriculum content of undergraduate, postgraduate, surgical resident, and fellowship training programs in the basic principles of surgical oncology and multidisciplinary cancer conferences, as well as virtual training; (ii) the implementation of eHealth and mHealth technology and education programs at schools and to the general public, for education in health behavior, cancer prevention, diagnosis, and (surgical) cancer treatment; and (iii) simple tricorders for cancer diagnosis with mHealth technology.

, as well as virtual training; (ii) the implementation of eHealth and mHealth technology and education programs at schools and to the general public, for education in health behavior, cancer prevention, diagnosis, and (surgical) cancer treatment; and (iii) simple tricorders for cancer diagnosis with mHealth technology. Digital technology is transforming health and social care in the HICs, but also expanding to LMICs.32 The Department of Health and Human Services is providing funding opportunities through the National Cancer Institute RFA-CA-15-024 Cancer Detection, Diagnosis, and Treatment Technologies for Global Health, e.g. LMICs, the UG3/UH3 mechanism.33 Education with eHealth and mHealth technology, as well as collaboration and support from community and university-based surgical leaders and surgical oncology societies, is essential to successfully decrease the global cancer burden. According to Murray Brennan, ‘Western surgical oncology clinics’ can provide life-changing experiences for surgical residents to achieve a better understanding of cancer and the surgical options for fighting cancer through an educational exchange instead of a sophisticated fellowship.5

Colorectal cancer (CRC) is the second most common cancer worldwide.1 Although numerous improvements in treatment modalities have been achieved, approximately 40 % of patients will still die from recurrent or metastatic disease within 5 years.2 Consequently, conventional therapeutic strategies are unable to eliminate all cancer cells. CRC is a stem-cell-driven malignancy in which only a small population of cells, simplified as tumor-initiating cells (TICs), are able to initiate and sustain tumor growth.3 TICs are undifferentiated tumor cells with the exclusive ability to self-renew and to generate the cellular heterogeneity of a tumor. TICs are more resistant to conventional anticancer therapy and therefore may be the main cause of treatment escape and tumor relapse.4–6 Initially, the TIC population in CRC was identified by the presence of the surface marker CD133, which showed an increased tumorigenic potential in xenografts of immunodeficient mice.7 Despite the description of some surface markers, only an insufficient purity of TICs can be achieved so far and their biology remains undefined.8 Hence, identifying the regulatory mechanisms and signaling pathways involved in TICs, and developing targeted therapy, might raise promising strategies in the treatment of CRC.

spite the description of some surface markers, only an insufficient purity of TICs can be achieved so far and their biology remains undefined.8 Hence, identifying the regulatory mechanisms and signaling pathways involved in TICs, and developing targeted therapy, might raise promising strategies in the treatment of CRC. Emerging data revealed PI3K/AKT/mTOR signaling implicated in the progression of CRC and that components of the mTOR pathway were overexpressed in CRC.9 In recent studies, a new oral-specific AKT1/2/3 inhibitor, MK-2206, provided in vitro and in vivo antitumor activity as a single agent, as well as enhanced activity in combination with conventional chemotherapeutics.10–13 In addition, MK-2206 has been shown to be safe in humans, with early evidence of antitumor activity in clinical trials.14,15 The present study aimed to determine the phenotypic and molecular differences between colonic TICs and their normal colon stem cell counterparts. Transcriptome analyses revealed that genes involved in AKT signaling are enriched in the TIC cultures. Functional testing implicated the selective AKT inhibitor MK-2206 as a potential therapeutic for TIC-directed therapy in CRC.

d molecular differences between colonic TICs and their normal colon stem cell counterparts. Transcriptome analyses revealed that genes involved in AKT signaling are enriched in the TIC cultures. Functional testing implicated the selective AKT inhibitor MK-2206 as a potential therapeutic for TIC-directed therapy in CRC. Methods Patient Material Human colon cancer and adjacent normal mucosa tissue were obtained after surgical resection and characterization by a pathologist. Tissue collection was approved by the Ethics Committee of the University Hospital Frankfurt, and after written consent had been received from all patients involved in the study. Solid tissues were minced and dissociated with 200 U/ml Collagenase type III, 100 U/ml Dispase, and 100 U/ml DNase I (all Worthingtorn, USA) in HBSS for 60–90 min at 37 °C. Every 30 min the cell suspension was subjected to MACS tissue dissociator for 40 s. Cells were filtered through sterile 70 µm nylon mesh [Becton Dickinson (BD), Heidelberg, Germany], and contaminated red blood cells were removed by osmotic lysis. Sphere Formation Assay Isolated cells were suspended in serum-free DMEM/F12 (Gibco, Germany) supplemented with 20 ng/ml epidermal growth factor and fibroblast growth factor, 2 % N2 supplement (Life Technologies, Germany), 20 mmol/l HEPES, and 50 U/ml penicillin/streptomycin at a density of 50,000 cells (tumor) and 100,000 cells (normal) per well in ultra-low-attachment 24-well plates (Corning, Germany), as described by Kreso and O’Brien.16 Plates were scored microscopically after 7 and 14 days.

2 supplement (Life Technologies, Germany), 20 mmol/l HEPES, and 50 U/ml penicillin/streptomycin at a density of 50,000 cells (tumor) and 100,000 cells (normal) per well in ultra-low-attachment 24-well plates (Corning, Germany), as described by Kreso and O’Brien.16 Plates were scored microscopically after 7 and 14 days. Microarray Analysis Expression analysis was performed using Genechip Human Exon 1.0 ST. Array (Affymetrix, Santa Clara, CA, USA). RNA was extracted from 14-day tumorspheres and corresponding colonospheres from normal tissue using an RNeasy Midi kit according to the manufacturer’s instructions. RNA quantity and quality were assessed using Nanovue (GE Life Sciences, USA) and 2100 Bioanalyzer (Agilent, USA), respectively. Only samples with a high RNA integrity number (RIN: 8–10) were used for the profiling. Genes with a twofold cut-off were then further subdivided into functional categories and pathways with the bioinformatics analysis resource DAVID (Database for Annotation, Visualization and Integrated Discovery) of the Laboratory of Immunopathogenesis and Bioinformatics. Cell Culture Human CRC cell lines SW480 and hematocrit (HCT)-116 were cultured in McCoy’s 5a and CX-1 in MEM-Earls containing 10 % FCS, 200 mM HEPES, 2 mmol l-glutamine, 50 units/ml penicillin/streptomycin in 37 °C humidified atmosphere with 5 % CO2. All cell lines were purchased from CLS Cell Lines Service GmbH, Eppelheim, Germany.

uman CRC cell lines SW480 and hematocrit (HCT)-116 were cultured in McCoy’s 5a and CX-1 in MEM-Earls containing 10 % FCS, 200 mM HEPES, 2 mmol l-glutamine, 50 units/ml penicillin/streptomycin in 37 °C humidified atmosphere with 5 % CO2. All cell lines were purchased from CLS Cell Lines Service GmbH, Eppelheim, Germany. In Vitro Drug Treatment MK-2206 (Selleckchem, Munich, Germany) and 5-fluorouracil (Sigma-Aldrich, Munich, Germany) were dissolved in dimethyl sulfoxide (DMSO) and stored at −20 °C. 5000 SW480, HCT-116 and CX-1 cells per well in 96-well plates were cultured with or without 5-fluorouracil (10 µM) and combination treatment with or without MK2206 (5 µM) for 24–72 h. Proliferation was assessed using a 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) assay according to the manufacturer’s instructions in pentaplicate. For apoptosis and cell cycle assays, 25,000 cells per well were cultured in 24-well plates under the same treatment as previously described. Apoptosis was determined 24 and 72 h after treatment using AnnexinV/7AAD staining (BD, Germany), according to the manufacturer’s instructions, after cells were stained with anti CD133-PE. To determine the cell cycle distribution, 48 and 72 h after treatment cells were pulsed with bromodeoxyuridine (BrdU) for 8 h, stained with anti-CD133-PE followed by fixation/permeabilization and anti-BrdU and DNA content (7AAD) staining using an FITC BrdU flow kit (BD, Germany), according to the manufacturer’s instructions. The cells were analyzed on an FACSCanto II (BD, Germany).

tment cells were pulsed with bromodeoxyuridine (BrdU) for 8 h, stained with anti-CD133-PE followed by fixation/permeabilization and anti-BrdU and DNA content (7AAD) staining using an FITC BrdU flow kit (BD, Germany), according to the manufacturer’s instructions. The cells were analyzed on an FACSCanto II (BD, Germany). The effect of treatment on tumorsphere formation was investigated by sphere formation assay as described above. Flow Cytometry and Cell Sorting SW480 cells were detached with Accutase™ (Sigma-Aldrich, Germany) and incubated with a PE-conjugated antibody to human CD133 (clone AC133/1 Miltenyi Biotec, Bergisch Gladbach, Germany) or a mouse isotype control. Stained cells were analyzed via a FACSCanto or sorted via a FACSAria (BD, Germany). In Vivo Xenograft Experiments All animal experiments were approved by the local authorities. NOD.Cg-PrkdcscidI12rgtm1Wjl/SzJ (NSG) or NOD.CB17-Prkdcscid/J (non-obese diabetic/severe combined immunodeficiency [NOD/SCID]) mice (Jackson Laboratory, USA) were used at 6–8 weeks of age. Fluorescence-activated cell sorting (FACS)-sorted SW480 cells (CD133high, CD133+, CD133−) were resuspended 1:1 in Matrigel (BD, Germany) and serum-free medium, and were injected subcutaneously into the flank at 5 × 104 cells per mouse. For the in vivo treatment, MK-2206 (100 mg/kg) was administered orally three times a week when tumor size reached 0.2–0.3 cm in diameter. Tumor growth was measured twice weekly using a caliper, and mice were sacrificed when tumor size reached a diameter of 1 cm.

subcutaneously into the flank at 5 × 104 cells per mouse. For the in vivo treatment, MK-2206 (100 mg/kg) was administered orally three times a week when tumor size reached 0.2–0.3 cm in diameter. Tumor growth was measured twice weekly using a caliper, and mice were sacrificed when tumor size reached a diameter of 1 cm. Western Blot Heat-denatured protein samples were separated using SDS-PAGE and transferred to a PVDF membrane (GE Healthcare, Germany) by electroblotting (Bio-Rad, Germany). Blots were blocked with 10 % milk for 1 h and incubated with primary antibodies against total Akt, phosphoAkt, or phosphoRPS6 (all Cell Signaling Technology, Danvers, MA, USA) overnight at 4 °C. Proteins were detected with peroxidase-goat anti-mouse antibody and ECL (GE Healthcare, Germany) on a Fusion FX7 imaging system (Vilber Lourmat, France). β-Actin (Sigma-Aldrich, Germany) was used as the internal loading control. Statistical Analysis Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA). The microarray data were statistically analyzed using a t test, by ATLAS Biolabs GmbH. Results Gene expression revealed upregulated AKT, P53, and Wnt signaling in primary tumorspheres.

Western Blot Heat-denatured protein samples were separated using SDS-PAGE and transferred to a PVDF membrane (GE Healthcare, Germany) by electroblotting (Bio-Rad, Germany). Blots were blocked with 10 % milk for 1 h and incubated with primary antibodies against total Akt, phosphoAkt, or phosphoRPS6 (all Cell Signaling Technology, Danvers, MA, USA) overnight at 4 °C. Proteins were detected with peroxidase-goat anti-mouse antibody and ECL (GE Healthcare, Germany) on a Fusion FX7 imaging system (Vilber Lourmat, France). β-Actin (Sigma-Aldrich, Germany) was used as the internal loading control. Statistical Analysis Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA). The microarray data were statistically analyzed using a t test, by ATLAS Biolabs GmbH. Results Gene expression revealed upregulated AKT, P53, and Wnt signaling in primary tumorspheres. First, we aimed to determine changes in gene expression patterns and regulated pathways in TICs. Therefore, we compared the transcriptome of five paired samples consisting of tumorsphere-initiating cells and corresponding non-malignant colonosphere-forming cells from five different CRC patients. The spheres were derived from freshly resected cancer tissue and corresponding normal colon mucosa after a 14-day culture. FACS analysis showed a strong enrichment of CD133+ cells in spheres (82.7  ± 7.5 %) in comparison to monolayer culture (16.9 ± 11 %; n = 6, data not shown). The differential expression analysis revealed 79 induced and 32 repressed genes in tumorspheres in comparison to normal colonospheres (Fig. 1a, b, and Electronic Supplementary Table 1). Well-described genes that are involved in CRC progression were upregulated in tumorspheres, such as MET and MACC1.17 Next, we included the differentially expressed genes in a functional annotation analysis using the DAVID algorithm. When classified according to function, genes regulating ‘immune response’, ‘duplication’ and ‘protease’ were most frequently identified (Fig. 1c). Importantly, well-described pathways associated with CRC progression, such as ‘Wnt signaling’ (DKK, Axin, CCND1), ‘P53 signaling’ (PERP, SERPINB5), and ‘colorectal cancer’, including AKT (CCND1, TSC, Axin, MET) and transforming growth factor (TGF)-β signaling, were also overrepresented in tumorspheres (Fig. 1d) and were thus potential targets for a TIC-directed therapy.Fig. 1 Transcriptome profiling revealed upregulated AKT, P53, and Wnt signaling pathways in CRC spheres. Exon-array expression analyses of matched CRC tumorspheres/corresponding normal colonospheres from five different patients. a Heat map showing a two-dimensional clustering of the top 100 regulated genes. b Venn diagram of regulated genes, tumor versus normal (p < 0.05, fold change >2). c DAVID functional annotation. d Pathway analysis of differentially expressed genes in tumorspheres. The top functions (c) and main canonical pathways (d) with their corresponding p values are shown. CRC colorectal cancer

egulated genes. b Venn diagram of regulated genes, tumor versus normal (p < 0.05, fold change >2). c DAVID functional annotation. d Pathway analysis of differentially expressed genes in tumorspheres. The top functions (c) and main canonical pathways (d) with their corresponding p values are shown. CRC colorectal cancer High CD133 expression demarcates an aggressive TIC subset in colorectal cancer SW480 cells.

egulated genes. b Venn diagram of regulated genes, tumor versus normal (p < 0.05, fold change >2). c DAVID functional annotation. d Pathway analysis of differentially expressed genes in tumorspheres. The top functions (c) and main canonical pathways (d) with their corresponding p values are shown. CRC colorectal cancer High CD133 expression demarcates an aggressive TIC subset in colorectal cancer SW480 cells. To further investigate the role of TIC biology, we employed SW480 CRC cells, given that they are genetically similar to most sporadic CRCs. FACS analysis of the CRC cell line SW480 revealed a differential surface expression of the described TIC marker CD1337,18 that allowed us to prospectively separate three subsets of cells, CD133− (3 %), CD133+ (95.5 %), and CD133high (1.5 %) cells (Fig. 2a).Fig. 2 Selective AKT inhibition by MK-2206 diminishes TIC activity and survival in vitro. a Three subpopulations in SW480 cells based on CD133 expression. Representative FACS blot. b Tumor growth of 50,000 prospectively isolated CD133 subpopulations injected subcutaneously in NSG (n = 4). c–g SW480 cells were treated with 1, 5, or 10 µM MK-2206 for the indicated times. c Western blot analysis after 72 h of treatment. d Cell proliferation (MTT assay). e Fold change of CD133high-expressing cells detected via FACS. f Early apoptotic cells (AnnexinV+/7-AAD−) in CD133 subpopulations determined via FACS. (g) Cell cycle phase distribution in CD133+ bulk cells detected via BrdU incorporation. h Tumorsphere formation in CRC cell lines. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. TIC tumor-initiating cell, FACS fluorescence-activated cell sorting, DMSO dimethyl sulfoxide, MTT 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide, BrdU bromodeoxyuridine, CRC colorectal cancer, HCT hematocrit

lines. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. TIC tumor-initiating cell, FACS fluorescence-activated cell sorting, DMSO dimethyl sulfoxide, MTT 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide, BrdU bromodeoxyuridine, CRC colorectal cancer, HCT hematocrit To confirm that CD133 is a marker of tumor initiation in SW480 cells, all three subpopulations were prospectively isolated via FACS based on their CD133 expression, and subcutaneously inoculated in NSG mice to monitor tumor formation. Indeed, CD133high cells caused the most robust and rapid establishment of tumors, and the tumors progressed faster than tumors from the CD133+ or CD133− populations (Fig. 2b). These results identify an aggressive TIC population in SW480 cells. Selective AKT Inhibition Diminishes Colonic Tumor-Initiating Cells In Vitro Next, we assessed the effect of AKT inhibition on TIC activity and survival. To this aim, SW480 cells were treated with MK-2206, a selective AKT1/2/3 inhibitor, for 72 h. MK-2206 treatment for 24 h inhibited AKT kinase activity (Fig. 2c). To evaluate the consequences of AKT inhibition on tumor cell growth, CRC cells were treated with MK-2206, and their cell expansion was analyzed. At 10 µM, MK-2206 significantly reduced cell proliferation after 72 h of treatment (Fig. 2d). Next, we were able to demonstrate that MK-2206 significantly decreased the amount of the CD133high subset after 72 h of treatment (Fig. 2e), indicating that MK-2206 treatment specifically eradicated the TIC subset in CRC cells.

t 10 µM, MK-2206 significantly reduced cell proliferation after 72 h of treatment (Fig. 2d). Next, we were able to demonstrate that MK-2206 significantly decreased the amount of the CD133high subset after 72 h of treatment (Fig. 2e), indicating that MK-2206 treatment specifically eradicated the TIC subset in CRC cells. To further investigate whether the effects induced by MK-2206 were evoked by apoptosis induction or cell cycle inhibition, we determined AnnexinV/7AAD staining and BrdU incorporation in SW480 cells treated for 72 h via FACS, to simultaneously discriminate the various CD133-expressing cell subsets. While MK-2206 did not induce apoptosis within 72 h in the CD133+ bulk population of cancer cells, we observed enhanced apoptosis in the TIC-enriched fraction after 24 and 72 h of treatment (Fig. 2f). In addition to a selective effect on apoptosis in the TIC fraction, we determined a dose-dependent cell cycle G1 arrest in the bulk of SW480 cells treated with MK-2206, as assayed by the incorporation of BrdU during the S-phase of the cell cycle (Fig. 2g). To ensure that our findings were not restricted to one cell line, we performed the tumorsphere formation assay in two additional CRC cell lines to functionally confirm the effect of AKT inhibition on TIC activity. After 7 days in culture, the number of tumorspheres were significantly diminished by MK-2206, even at low concentrations in all three cell lines (Fig. 2h). Taken together, AKT inhibition with MK-2206 reduces CRC cell proliferation and directly inhibits TIC survival and function in vitro.

AKT inhibition on TIC activity. After 7 days in culture, the number of tumorspheres were significantly diminished by MK-2206, even at low concentrations in all three cell lines (Fig. 2h). Taken together, AKT inhibition with MK-2206 reduces CRC cell proliferation and directly inhibits TIC survival and function in vitro. MK-2206 Potentiates the Effect of 5-Fluorouracil Chemotherapy To elucidate the antitumoral effect of MK-2206, we analyzed its impact on CRC cell proliferation and tumorsphere formation when co-treated with the standard chemotherapeutic drug 5-fluorouracil in CRC treatment. The combined treatment of MK-2206 and 5-fluorouracil caused a significant inhibition of cell proliferation compared with 5-fluorouracil alone, indicating a synergistic anticancer effect of both drugs on CRC cells (Fig. 3a).Fig. 3 Synergistic antitumoral effects of 5-fluorouracil and MK-2206 in CRC. a–c SW480, HCT and Cx-1 cells were treated with either 5 µM MK-2206, 10 µM 5-fluorouracil, or in combination. a Cell proliferation was determined after 48 h (MTT assay). b SW480 tumorspheres scored after 14 days of treatment. c Percentage of CD133high-SW480 cells in G1 phase after 72 h of treatment detected via an 8 h BrdU pulse. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. CRC colorectal cancer, DMSO dimethyl sulfoxide, HCT hematocrit, 5-FU 5-fluorouracil, MTT 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide, BrdU bromodeoxyuridine

fter 72 h of treatment detected via an 8 h BrdU pulse. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. CRC colorectal cancer, DMSO dimethyl sulfoxide, HCT hematocrit, 5-FU 5-fluorouracil, MTT 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide, BrdU bromodeoxyuridine We then investigated whether the combination therapy would be effective in inhibiting TIC proliferation. MK-2206 alone was highly effective in inhibiting tumorsphere formation in SW480 cells, while a combined treatment with 5-fluorouracil did not further increase this effect (Fig. 3b). Furthermore, a single treatment with 5-fluorouracil did not cause cell cycle arrest in TICs, whereas MK-2206 alone and combined treatment with 5-fluorouracil significantly induced G1 cell cycle arrest in TIC-enriched fraction after 72 h (Fig. 3c). Although MK-2206 showed a synergistic antitumor effect when combined with 5-fluorouracil, MK-2206 alone was highly effective in suppressing colon TIC activity.

TICs, whereas MK-2206 alone and combined treatment with 5-fluorouracil significantly induced G1 cell cycle arrest in TIC-enriched fraction after 72 h (Fig. 3c). Although MK-2206 showed a synergistic antitumor effect when combined with 5-fluorouracil, MK-2206 alone was highly effective in suppressing colon TIC activity. Selective AKT Inhibition Reduces Tumor Initiation and Growth In Vivo and Patient-Derived Tumorsphere Formation Since TIC activity can only be reliably determined by its capacity to initiate tumor growth in vivo, we assessed the potency of MK-2206 in a xenograft transplantation model. Therefore, we pretreated SW480 cells with MK-2206 or DMSO for 72 h in culture before we subcutaneously injected equal numbers of living cells into NOD/SCID mice (Fig. 4a). The treatment with MK-2206 significantly delayed tumor formation and significantly reduced tumor growth (mean tumor weight 0.32 g vs. 0.12 g; p < 0.05) (Fig. 4b, c) induced by the remaining cells after the pretreatment, suggesting that MK-2206 selectively eliminates TICs.Fig. 4 AKT inhibition by MK-2206 reduces tumor initiation and growth in xenografts and tumorsphere formation in primary patient material. a SW480 cells were pretreated with 10 µM MK2206 for 72 h in vitro and, subsequently, 50,000 living cells were injected subcutaneously (n = 8). b Tumor volumes over time. c Tumor weight after 4 weeks. d Equal numbers of SW480 cells were injected subcutaneously. Once the tumor reached 25 mm3, mice were randomized into two groups and treated with MK-2206 (100 mg/kg, three times/week) or DMSO (control) [n = 6]. e Tumor volumes under treatment over time. f Tumor weight after 3 weeks of treatment. g Tumor cells from primary CRC patient material were treated with DMSO or MK-2206, and sphere formation was determined (n = 5 CRC patients). h Representative images of sphere formation. Scale bar 200 µm. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. DMSO dimethyl sulfoxide, CRC colorectal cancer, NOD non-obese diabetic, SCID severe combined immunodeficiency

sphere formation was determined (n = 5 CRC patients). h Representative images of sphere formation. Scale bar 200 µm. Data are expressed as mean and standard deviation. *p < 0.05; **p < 0.01 compared with DMSO. DMSO dimethyl sulfoxide, CRC colorectal cancer, NOD non-obese diabetic, SCID severe combined immunodeficiency To further demonstrate the anticancer effects of MK-2206 in vivo, NOD/SCID mice with established SW480 tumor cell xenografts were treated with MK-2206 or DMSO (control) (Fig. 4d). While the DMSO-treated mice showed a successive increase in tumor growth, MK-2206 treatment resulted in an almost complete stagnation of tumor size, leading to small tumors that were as equal in size as before the start of treatment.(Fig. 4e, f). In order to translate our results on MK-2206 activity against TICs to primary CRC patients, tumorsphere cultures derived from freshly isolated patient CRC tissue were treated with MK-2206. After 7 and 14 days of treatment, the number and diameter of formed tumorspheres were significantly reduced by MK-2206 compared with the DMSO control (Fig. 4g, h). These results suggest that AKT inhibition effectively impacts on TIC activity and serves as a treatment strategy against colorectal TICs.

ated with MK-2206. After 7 and 14 days of treatment, the number and diameter of formed tumorspheres were significantly reduced by MK-2206 compared with the DMSO control (Fig. 4g, h). These results suggest that AKT inhibition effectively impacts on TIC activity and serves as a treatment strategy against colorectal TICs. Discussion In this study, we report the effectiveness of selective AKT inhibition by MK-2206 on diminishing the tumorigenic potential of CRC-intiating cells. Although already highly effective alone, MK-2206 showed synergistic anticancer effects on CRC cell proliferation with conventional chemotherapeutics such as 5-fluorouracil. Of note, the effects of AKT inhibition on TIC activity could be determined in both mismatch repair proficient (SW480) and deficient (HCT-116) CRC cells. Importantly, MK-2206 reduced tumorsphere formation in cancer cells derived from primary patient specimens. Similar to recent results, treatment with MK-2206 generated a significant deceleration in tumor progression in vivo.19 mTOR/AKT signaling is essential in the progression of CRC, and mTOR inhibition leads to a reduction of CRC cell growth.8,9,20 Cai et al. identified mTOR high expression to be an independent risk factor for the prognosis of CRC patients.21 An activation of mTOR signaling has been detected in colon TICs.22,23 In concordance with these reports, the present study detected an increased expression of components of the AKT/mTOR signaling in colon TICs.

et al. identified mTOR high expression to be an independent risk factor for the prognosis of CRC patients.21 An activation of mTOR signaling has been detected in colon TICs.22,23 In concordance with these reports, the present study detected an increased expression of components of the AKT/mTOR signaling in colon TICs. Few studies have investigated the effects of mTOR inhibitors on colon TICs. In one study, treatment with rapamycin and PP242 diminished sphere formation as well as aldehyde dehydrogenase activity in vitro.21 However, the authors did not perform sufficient functional experiments to investigate the effect of mTOR inhibitors on the tumorigenic potential of colon TICs. A second study proposed the mTOR kinase inhibitor Torin-1 as a drug candidate for CRC therapy as it inhibited survival of colon TICs in vitro and slowed tumor progression in vivo.22 Similarly, Todaro et al. showed, that PI3K-inhibition by BKM-120 reduced metastatic growth of CRC cells.23 Nevertheless, no further studies have been reported to elucidate the effectiveness of Torin-1 in cancer or its safety in patients. Furthermore, the authors did not assess the effects of Torin-1 or BKM-120 on the tumor initiation ability of TICs.

that PI3K-inhibition by BKM-120 reduced metastatic growth of CRC cells.23 Nevertheless, no further studies have been reported to elucidate the effectiveness of Torin-1 in cancer or its safety in patients. Furthermore, the authors did not assess the effects of Torin-1 or BKM-120 on the tumor initiation ability of TICs. Conclusions We achieved a significant reduction of the tumor growth originating from TICs, both in vitro and in vivo, by selective AKT inhibition with MK-2206. Given that AKT is both an upstream activator of the mTOR complex 1 and a downstream effector of the mTOR complex2,24 and its overexpression is associated with resistance to chemotherapy,25 it represents a critical target of this signaling cascade. Therefore, selective AKT inhibition may be superior to mTOR inhibitors in the treatment of CRC, and represents a promising agent to prevent tumor relapse by eliminating the TIC subset. Our data strongly encourage further clinical testing of MK-2206 either alone or in combination with conventional chemotherapeutics in CRC patients. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary material 1 (XLSX 18 kb)

Conclusions We achieved a significant reduction of the tumor growth originating from TICs, both in vitro and in vivo, by selective AKT inhibition with MK-2206. Given that AKT is both an upstream activator of the mTOR complex 1 and a downstream effector of the mTOR complex2,24 and its overexpression is associated with resistance to chemotherapy,25 it represents a critical target of this signaling cascade. Therefore, selective AKT inhibition may be superior to mTOR inhibitors in the treatment of CRC, and represents a promising agent to prevent tumor relapse by eliminating the TIC subset. Our data strongly encourage further clinical testing of MK-2206 either alone or in combination with conventional chemotherapeutics in CRC patients. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary material 1 (XLSX 18 kb) Acknowledgment The authors thank Prof. Martin-Leo Hansmann, Chief Pathologist, University of Frankfurt, and his co-workers for their cooperation and support; Dr. Schuler, Bethanien Hospital, Frankfurt, for kindly providing us with primary patient material; and Sabrina Bothur and Claudia Jourdan for their excellent technical support. Patrizia Malkomes receives support from the Frankfurter Förderung ‘Nachwuchswissenschaftler’ and M.A. Rieger is supported by the LOEWE Center for Cell and Gene Therapy Frankfurt, Hessian Ministry for Science and the Arts (Hessisches Ministerium für Wissenschaft und Kunst) [III L 4-518/17.004 (2014)].

technical support. Patrizia Malkomes receives support from the Frankfurter Förderung ‘Nachwuchswissenschaftler’ and M.A. Rieger is supported by the LOEWE Center for Cell and Gene Therapy Frankfurt, Hessian Ministry for Science and the Arts (Hessisches Ministerium für Wissenschaft und Kunst) [III L 4-518/17.004 (2014)]. Author Contributions Conception and design of the study: PM and MAR; acquisition, analysis and interpretation of data: PM, IL, AL, EO, SB, NH; writing the manuscript: PM and MAR; study advice and critical manuscript revision: HS, KH, WOB. Disclosure The authors declare no conflict of interest.

Neoadjuvant chemoradiation therapy (CRT) and radical surgery including total mesorectal excision (TME) reduces the risk of local recurrence and is considered the standard of care for patients with locally advanced (T3–4 or any N1–2) mid-distal rectal cancer (LARC).1 – 4 A pathologic complete response (ypCR) shown in the surgical specimen of LARC patients treated by CRT is observed in up to one-third of the cases.5 In ypCR cases, a favorable long-term oncologic outcome has been observed,6 – 9 and organ preservation strategies including transanal full-thickness local excision and/or close observation are being explored in patients displaying clinical or pathologic complete response to CRT. This would lead to a reduction in surgery-related morbidity and mortality and to quality-of-life improvement.10 – 23 The potential presence of metastatic mesorectal lymph nodes, with the related risk of local and distant recurrences, represents a key limiting factor for the application of organ preservation strategies. The reported rate of metastatic mesorectal lymph nodes in the surgical specimen of LARC patients achieving a complete pathologic response (ypT0) in the primary tumor is variable,24 – 33 and the accuracy of lymph node status restaging after CRT is low.34 – 36

factor for the application of organ preservation strategies. The reported rate of metastatic mesorectal lymph nodes in the surgical specimen of LARC patients achieving a complete pathologic response (ypT0) in the primary tumor is variable,24 – 33 and the accuracy of lymph node status restaging after CRT is low.34 – 36 Because a priori knowledge of pathologic and oncologic outcome risks is an important issue for protocol design and for clinician–patient communication at clinical study enrollment, we specifically focused this study on patients with rectal cancer staged by endorectal ultrasonography (EUS), pelvic magnetic resonance imaging (MRI), or both as having metastatic mesorectal lymph nodes at their initial diagnosis (cN+). To evaluate whether cN+ patients could be reasonably eligible for treatment strategies aimed at organ preservation, we analyzed the pathologic and long-term oncologic outcomes for LARC patients treated by neoadjuvant CRT at our institution during a 17-year period. Methods All consecutive informed-consent patients treated by neoadjuvant CRT and surgery for LARC between January 1996 and October 2013 were identified from our institutional, prospectively maintained, rectal cancer database. Patients with synchronous distant metastasis were excluded from the study.

To evaluate whether cN+ patients could be reasonably eligible for treatment strategies aimed at organ preservation, we analyzed the pathologic and long-term oncologic outcomes for LARC patients treated by neoadjuvant CRT at our institution during a 17-year period. Methods All consecutive informed-consent patients treated by neoadjuvant CRT and surgery for LARC between January 1996 and October 2013 were identified from our institutional, prospectively maintained, rectal cancer database. Patients with synchronous distant metastasis were excluded from the study. All the patients had biopsy-proven adenocarcinoma of the rectum. The distance of the tumor from the anal verge was measured by rigid rectoscopy. Pre- and post-CRT primary tumor and nodal stagings were evaluated by EUS, pelvic MRI, or both. Lymph nodes 5 mm or larger were considered positive. In cases with discrepancy between the two imaging techniques, the higher stage was considered. Distant metastases were ruled out by thoracoabdominal and pelvic CT scan.

py. Pre- and post-CRT primary tumor and nodal stagings were evaluated by EUS, pelvic MRI, or both. Lymph nodes 5 mm or larger were considered positive. In cases with discrepancy between the two imaging techniques, the higher stage was considered. Distant metastases were ruled out by thoracoabdominal and pelvic CT scan. Treatment Preoperative CRT Preoperative CRT was administered according to several preoperative sequential treatment protocols developed at our institution, including a 5-fluorouracil (5-FU) bolus + leucovorin (LV) and 45 Gy with or withut adjuvant 5-FU/LV, raltitrexed and 50.4 + 10 Gy of intraoperative radiation therapy (IORT), capecitabine and 50.4 Gy, continuous infusion 5-FU + gefitinib and 50.4 + 10 Gy IORT, and capecitabine ± oxaliplatin and 50.4 Gy. The radiotherapy (RT) clinical target volume (CTV2) included the primary tumor, the mesorectum, and internal iliac lymph nodes. A second clinical target volume (CTV1) included the mesorectum corresponding to the primary tumor with a 2-cm radial margin. The RT fractionation was 180 cGy/day, 5 fractions per week. More details on RT technique and dose prescription have been reported previously.8

mor, the mesorectum, and internal iliac lymph nodes. A second clinical target volume (CTV1) included the mesorectum corresponding to the primary tumor with a 2-cm radial margin. The RT fractionation was 180 cGy/day, 5 fractions per week. More details on RT technique and dose prescription have been reported previously.8 Surgery The patients underwent surgery 6–8 weeks after completion of neoadjuvant CRT. The surgical procedures included abdominoperineal resection (APR), low anterior resection (LAR), and full-thickness transanal local excision (LE). Radical resection was performed according to TME principles. Reasons for the use of LE included medical comorbidity and patient refusal of APR for low-lying tumors not eligible for coloanal reconstruction due to anticipation of poor sphincter function. In more recent years, patients with a major clinical response to CRT were offered the option of LE in a prospective clinical study investigating the outcome of LE after a complete clinical and pathologic response. In these cases, LE was used to assess the pathologic response in the primary tumor. Medically fit patients showing no complete or almost complete pathologic response in the primary tumor (TRG1 and TRG2 according to Mandard tumor response grading)37 underwent subsequent TME surgery. After surgical resection, IORT to a high risk area (presacral region) was administered according to study protocols, as mentioned earlier.

ients showing no complete or almost complete pathologic response in the primary tumor (TRG1 and TRG2 according to Mandard tumor response grading)37 underwent subsequent TME surgery. After surgical resection, IORT to a high risk area (presacral region) was administered according to study protocols, as mentioned earlier. Postoperative Chemotherapy Adjuvant 5-FU-based chemotherapy was administered according to the study protocol, or in selected cases included patients with metastatic lymph nodes. Pathology Pathologic tumor staging was performed according to the guidelines of the American Joint Committee on Cancer and the College of American Pathologists.38 Patients with no residual cancer cells in the surgical specimen were considered pathologic complete responders (ypCR). Follow-up Evaluation Postoperatively, the patients were examined at follow-up visits every 3 months for the first 2 years and half-yearly thereafter. At each follow-up control visit, the CEA level was determined. Abdominal and pelvic computed tomography (CT) scan or liver ultrasound and chest x-ray were performed alternatively every 3–6 months. Colonoscopy was performed yearly.

ined at follow-up visits every 3 months for the first 2 years and half-yearly thereafter. At each follow-up control visit, the CEA level was determined. Abdominal and pelvic computed tomography (CT) scan or liver ultrasound and chest x-ray were performed alternatively every 3–6 months. Colonoscopy was performed yearly. Statistical Analysis The Chi square test or Fisher’s exact test was used to compare percentages between complete responders and non-complete responders, and the Wilcoxon rank test was used for median age comparison. Cumulative probabilities of overall survival (OS), disease-specific survival (DSS), disease-free survival (DFS), distant metastasis-free survival (DMFS), and local recurrence-free survival (LRFS) were estimated by Kaplan–Meier survival methods,39 and differences between subgroups were assessed using the log-rank test. The duration of follow-up evaluation was calculated as the time from surgery to the event of interest. Patients without event were censored at the date of the last follow-up visit. In cases with local and distant metastasis, both events were recorded and computed at any time of occurrence. For better assessment of the oncologic implications of ypCR, the Cox proportional hazards model was used to adjust the hazard ratios (HRs) and corresponding 95 % confidence intervals (CIs).40 Due to the limitation of sample size and number of events, only three variables were entered into the multivariate model: cNstage (cN0 vs cN1), type of surgery (TME vs LE), and ypT0 (yes vs no). Collinearity between variables was excluded by means of the Chi square test. A P value of 0.05 or lower was considered statistically significant (two-tailed). The SAS System 9.2 (SAS, Cary, NC, USA) was used as the statistical software for data analysis.

(cN0 vs cN1), type of surgery (TME vs LE), and ypT0 (yes vs no). Collinearity between variables was excluded by means of the Chi square test. A P value of 0.05 or lower was considered statistically significant (two-tailed). The SAS System 9.2 (SAS, Cary, NC, USA) was used as the statistical software for data analysis. Results Patients and Treatment Characteristics The study population comprised 226 consecutive patients (142 men and 84 women; median age, 64 years; range, 25–87 years) with mid-distal LARC and no distant metastasis treated by neoadjuvant CRT followed by surgery at our institution between January 1996 and October 2013.

(cN0 vs cN1), type of surgery (TME vs LE), and ypT0 (yes vs no). Collinearity between variables was excluded by means of the Chi square test. A P value of 0.05 or lower was considered statistically significant (two-tailed). The SAS System 9.2 (SAS, Cary, NC, USA) was used as the statistical software for data analysis. Results Patients and Treatment Characteristics The study population comprised 226 consecutive patients (142 men and 84 women; median age, 64 years; range, 25–87 years) with mid-distal LARC and no distant metastasis treated by neoadjuvant CRT followed by surgery at our institution between January 1996 and October 2013. At the initial evaluation, 226 patients were staged as follows: 5 cT2N1 (2.2 %), 79 cT3N0 (34.9 %), 104 cT3N1 (46 %), 12 cT4N0 (5.3 %), 13 cT4N1 (5.7 %), 2 cTxN0 (0.8 %), and 1 cTxN1 (0.4 %). In addition, 10 very low-lying cT2N0 tumors (4.4 %) were considered at high risk for recurrence, treated by neoadjuvant CRT, and included in this study. The median distance of the tumor from the anal verge was 5 cm (range, 1–12 cm). The total RT dose was 45 Gy for 42 patients (18.5 %), 50.4 Gy for 180 patients (79.6 %), and 25 Gy for 4 patients (1.7 %). Total mesorectal excision was performed for 179 patients (79 %) (142 LAR and 37 APR), whereas LE was performed for 47 patients (21 %). The documented reasons for the use of LE were preference after a major clinical response in 22 cases, patient absolute refusal of APR in 4 cases, and medical comorbidity in 3 cases. The remaining 18 patients were enrolled in the prospective clinical study investigating the outcome for LE after complete clinical and pathologic response. All patients restaged as ycN+ (n = 24) underwent TME surgery. Intraoperative radiation therapy was applied in the context of clinical studies. Postoperative chemotherapy was administered to all 33 ypN+ patients (14 %).

tive clinical study investigating the outcome for LE after complete clinical and pathologic response. All patients restaged as ycN+ (n = 24) underwent TME surgery. Intraoperative radiation therapy was applied in the context of clinical studies. Postoperative chemotherapy was administered to all 33 ypN+ patients (14 %). Clinical and Pathologic Response In the entire patient population, a complete pathologic response in the primary tumor (ypT0) was observed in 65 cases (28.7 %). For the 179 patients who underwent TME, the ypCR rate (ypT0N0) was 20.1 % (n = 36). The median number of examined lymph nodes was 13 (range, 2–37). Metastatic lymph nodes (ypN+) were found in 47 (26.2 %) of the surgical specimens: in 4 (10 %) of 40 ypT0 cases, in 1 (9 %) of 11 ypT1 cases, in 12 (21 %) of 57 ypT2 cases, in 28 (43 %) of 65 ypT3 cases, and in 2 (33.3 %) of 6 ypT4 cases. Among the patients who underwent TME surgery, metastatic lymph nodes (ypN+) were detected in 45 (41.6 %) of 108 cN+ patients compared with 2 (2.8 %) of 71 cN0 patients (P < 0.01). In the subgroup of cN+ tumors with ypT0 treated by TME surgery, 4 (16 %) of 25 (all restaged as ycN0) were ypN+ compared with 43 (51.8 %) of 83 cases that had no ypT0 (P < 0.01). In the subgroup of cN0 tumors with ypT0 treated by TME surgery, 0 of 15 were ypN+ compared with 2 (3.5 %) of 56 no-ypT0 cases. At restaging after CRT, comparing ycN status with ypN status, metastatic lymph nodes at pathology were detected in 9 of 23 ycN+ cases and in 10 of 25 ycN0 cases (sensitivity, 0.47; specificity, 0.51).

In the subgroup of cN0 tumors with ypT0 treated by TME surgery, 0 of 15 were ypN+ compared with 2 (3.5 %) of 56 no-ypT0 cases. At restaging after CRT, comparing ycN status with ypN status, metastatic lymph nodes at pathology were detected in 9 of 23 ycN+ cases and in 10 of 25 ycN0 cases (sensitivity, 0.47; specificity, 0.51). Recurrence and Survival No postoperative mortality occurred. During a median follow-up period of 48 months, 20 patients (8.84 %) experienced local recurrence only, 14 (6.19 %) experienced local recurrence and distant metastasis (9 liver, 4 lung and 1 other site cases), and 34 (15.04 %) experienced distant metastasis only (15 liver, 9 lung, 7 liver and lung, and 3 multiple-site cases). In the comparison of ypCR-patients (ypT0N0) and no-ypCR patients who underwent TME surgery, the 5-year DSS was respectively 91.0 and 79.4 % (P = 0.029), and the 5-year DFS was 84.9 and 61.7 % (P = 0.011). In the entire patient population, the 5-year survival rates were 79.2 % for OS, 83.0 % for DSS, 66.9 % for DFS, 77.1 % for DMFS, and 82.6 % for LRFS. In the subset of 65 ypT0 patients, 2 (3.1 %) experienced local recurrence only, 3 (4.61 %) experienced local recurrence and distant metastasis, and 3 (4.61 %) experienced distant metastasis only (1 liver, and 2 liver and lung cases).

e 79.2 % for OS, 83.0 % for DSS, 66.9 % for DFS, 77.1 % for DMFS, and 82.6 % for LRFS. In the subset of 65 ypT0 patients, 2 (3.1 %) experienced local recurrence only, 3 (4.61 %) experienced local recurrence and distant metastasis, and 3 (4.61 %) experienced distant metastasis only (1 liver, and 2 liver and lung cases). In the comparison of ypT0 patients (n = 65) with no-ypT0 patients (n = 161), the 5-year OS was respectively 89.4 versus 75.3 % (P = 0.005), the 5-year DSS was 94.5 versus 78.4 % (P = 0.005), the 5-year DFS was 87.3 versus 58.8 % (P < 0.001), the 5-year DMFS was 93.0 versus 70.6 % (P = 0.002), and the 5-year LRFS was 90.5 versus 79.3 % (P = 0.034). According to the clinical lymph node status at initial diagnosis, the 5-year DSS and DFS were respectively 80.5 and 64.2 % in cN+ cases compared with 86.3 and 69.7 % in cN0 cases (nonsignificant difference) (Table 1).Table 1 Long-term oncologic outcome according to clinocopathologic characteristics in locally advanced rectal cancer patients treated by neoadjuvant chemoradiation Variable Total 5-yrs DSS 5-yrs DFS 5-yrs LRFS

In the comparison of ypT0 patients (n = 65) with no-ypT0 patients (n = 161), the 5-year OS was respectively 89.4 versus 75.3 % (P = 0.005), the 5-year DSS was 94.5 versus 78.4 % (P = 0.005), the 5-year DFS was 87.3 versus 58.8 % (P < 0.001), the 5-year DMFS was 93.0 versus 70.6 % (P = 0.002), and the 5-year LRFS was 90.5 versus 79.3 % (P = 0.034). According to the clinical lymph node status at initial diagnosis, the 5-year DSS and DFS were respectively 80.5 and 64.2 % in cN+ cases compared with 86.3 and 69.7 % in cN0 cases (nonsignificant difference) (Table 1).Table 1 Long-term oncologic outcome according to clinocopathologic characteristics in locally advanced rectal cancer patients treated by neoadjuvant chemoradiation Variable Total 5-yrs DSS 5-yrs DFS 5-yrs LRFS n (%) % P Value % P Value % P Value Sex Female 84 (37.2) 80.2 0.638 69.9 0.461 89.7 0.042 Male 142 (62.8) 84.4 65.1 78.0 Age (years) ≤65 128 (56.6) 82.2 0.863 67.5 0.896 82.8 0.874 >65 98 (43.4) 84.0 65.9 82.1 cN status cN0 103 (45.6) 86.3 0.537 69.7 0.537 83.0 0.872 cN+ 123 (54.4) 80.5 64.2 82.2 Type of surgery TME 179 (79.2) 81.8 0.201 66.4 0.700 84.0 0.312 LE 47 (20.8) 87.2 68.2 76.8 ypT status ypT0 65 (28.7) 94.5 0.005 87.3 <0.001 90.5 0.034 No ypT0 161 (71.3) 78.4 58.8 79.3 ypN status 66.2 80.9 ypN0 132 (73.8) 88.0 0.002 72.2 0.008 85.4 0.309 ypN+ 47 (26.2) 71.7 55.7 85.7

.872 cN+ 123 (54.4) 80.5 64.2 82.2 Type of surgery TME 179 (79.2) 81.8 0.201 66.4 0.700 84.0 0.312 LE 47 (20.8) 87.2 68.2 76.8 ypT status ypT0 65 (28.7) 94.5 0.005 87.3 <0.001 90.5 0.034 No ypT0 161 (71.3) 78.4 58.8 79.3 ypN status 66.2 80.9 ypN0 132 (73.8) 88.0 0.002 72.2 0.008 85.4 0.309 ypN+ 47 (26.2) 71.7 55.7 85.7 DSS disease-specific survival, DFS disease-free survival, LRFS local-recurrence-free survival, cN clinical lymph node, TME total mesorectal excision, LE local excision, ypT0 complete pathologic response in the primary tumor, ypN pathologic lymph node status (only TME patients) Among the cN+ patients (n = 123) the 5-year DSS and DFS were respectively 100 and 91.6 % for the ypT0 patients compared with 71.2 and 58.0 % for the no-ypT0 patients (P < 0.01; Fig. 1). The 5-year DSS and DFS were both 100 % for the 4 ypT0N+ patients compared with 59.1 and 43.3 % respectively for the 43 no-ypTN+ patients (nonsignificant difference; Fig. 2).Fig. 1 Kaplan-Meier estimates for disease-specific survival (a) and disease-free survival (b) according to a complete pathologic response of the primary tumor (ypT0) in 123 cN+ rectal cancer patients treated by neoadjuvant chemoradiation followed by total mesorectal excision (TME) or full-thickness local excision (LE) surgery Fig. 2 Kaplan-Meier estimates for disease-specific survival (a) and disease-free survival (b) according to a complete pathologic response of the primary tumor (ypT0) in 47 ypN+ rectal cancer patients treated by neoadjuvant chemoradiation followed by total mesorectal excision (TME) surgery

Among the cN+ patients (n = 123) the 5-year DSS and DFS were respectively 100 and 91.6 % for the ypT0 patients compared with 71.2 and 58.0 % for the no-ypT0 patients (P < 0.01; Fig. 1). The 5-year DSS and DFS were both 100 % for the 4 ypT0N+ patients compared with 59.1 and 43.3 % respectively for the 43 no-ypTN+ patients (nonsignificant difference; Fig. 2).Fig. 1 Kaplan-Meier estimates for disease-specific survival (a) and disease-free survival (b) according to a complete pathologic response of the primary tumor (ypT0) in 123 cN+ rectal cancer patients treated by neoadjuvant chemoradiation followed by total mesorectal excision (TME) or full-thickness local excision (LE) surgery Fig. 2 Kaplan-Meier estimates for disease-specific survival (a) and disease-free survival (b) according to a complete pathologic response of the primary tumor (ypT0) in 47 ypN+ rectal cancer patients treated by neoadjuvant chemoradiation followed by total mesorectal excision (TME) surgery Among the cN+ patients who achieved ypT0, the 5-year DSS and DFS were respectively 100 and 85.7 % for the TME patients (n = 108) compared with 100 and 91.6 % for the LE patients (n = 15) (nonsignificant difference). In the multivariate analysis, ypT0 was the only independent prognostic factor for DSS (HR, 0.13; 95 % CI, 0.03–0.58; P = 0.007) and for DFS (HR, 0.25; 95 % CI, 0.12–0.54; P < 0.001).

vely 100 and 85.7 % for the TME patients (n = 108) compared with 100 and 91.6 % for the LE patients (n = 15) (nonsignificant difference). In the multivariate analysis, ypT0 was the only independent prognostic factor for DSS (HR, 0.13; 95 % CI, 0.03–0.58; P = 0.007) and for DFS (HR, 0.25; 95 % CI, 0.12–0.54; P < 0.001). Discussion The current study investigated whether LARC patients initially staged as cN+ and achieving ypT0 after neoadjuvant CRT are potential candidates for organ-preserving surgical strategies. To this end, the rate of ypT0, the incidence of metastatic lymph nodes, and the long-term oncologic outcome were analyzed in relation to cN status in LARC patients treated by neoadjuvant CRT and prospectively followed up at a single institution.

ant CRT are potential candidates for organ-preserving surgical strategies. To this end, the rate of ypT0, the incidence of metastatic lymph nodes, and the long-term oncologic outcome were analyzed in relation to cN status in LARC patients treated by neoadjuvant CRT and prospectively followed up at a single institution. For our patients treated with TME surgery after CRT, ypCR was achieved in 20.1 % of the cases, which is in line with the majority of studies previously reported in the literature.5 Our survival analysis supported the evidence of a favorable long-term oncologic outcome for patients displaying ypCR. In our series comparing ypCR patients with no-ypCR patients, the 5-year DFS rates were respectively 84.9 and 61.7 %. This is in line with the data reported by Maas et al.9 from a pooled analysis of 3105 LARC patients treated by preoperative CRT who showed a 5-year DFS of 83.3 % for ypCR patients compared with 65.6 % for no-ypCR patients. Similarly in a meta-analysis by Zorcolo et al.41 of 12 studies including 1913 LARC patients, the 5-year DFS was 86.9 % for ypCR patients compared with 63.9 % for no-ypCR patients. Recently Wasmooth et al.42 reported a 5-year DFS of 81 % for patients with ypCR and 50 % for patients without ypCR among 1384 patients enrolled in the national population-based colorectal cancer registry of Norway who had advanced T3 and T4 rectal cancer with N0-2,M0 managed by neoadjuvant long-course (chemo)radiation. Interestingly, ypCR was associated with a low risk of metastasizing.

with ypCR and 50 % for patients without ypCR among 1384 patients enrolled in the national population-based colorectal cancer registry of Norway who had advanced T3 and T4 rectal cancer with N0-2,M0 managed by neoadjuvant long-course (chemo)radiation. Interestingly, ypCR was associated with a low risk of metastasizing. In our subset of ypT0 patients treated by LE surgery, the local and distant recurrence rates were very low and similar to those for ypT0 patients treated by TME surgery. This finding is consistent with data reported by Borshitz et al.,13 who analyzed seven studies reporting oncologic outcome of LE after neoadjuvant CRT for cT2–3 tumors (n = 237). In their study, ypT0 was noted in 22 % of the cases, and the 5-year LRFS and DMFS were respectively 100 and 96 %. Similarly, Pucciarelli et al.18 reported that the 3-year LRFS was 96.9 % for 43 cT3 or low-lying cT2 rectal cancer patients treated with CRT followed by LE and observation for the ypT0-1 patients.

n = 237). In their study, ypT0 was noted in 22 % of the cases, and the 5-year LRFS and DMFS were respectively 100 and 96 %. Similarly, Pucciarelli et al.18 reported that the 3-year LRFS was 96.9 % for 43 cT3 or low-lying cT2 rectal cancer patients treated with CRT followed by LE and observation for the ypT0-1 patients. In our patients initially staged as cN+ and treated with CRT followed by TME surgery, metastatic lymph nodes at pathology were detected in 42.2 % of the cases. However, in the subgroup of patients with ypT0, metastatic lymph nodes were detected in 16 % of the surgical specimens. The rate of metastatic lymph nodes in LARC achieving ypT0 after CRT has been reported to vary between 2 and 17 %, which is in line with our findings of a 10 % rate (Table 2).24 – 33 Table 2 Metastatic lymph node rates in locally advanced rectal cancer with complete pathologic response in the primary tumor (ypT0) achieved after neoadjuvant chemoradiation No. of patients Examined lymph nodes median (range) YpT0 YpN+ N % N % Read et al.24 644 13 ± 8 42 6.52 1 2.38 Pucciarelli et al.25 235 9 (0–38) 56 23.83 1 1.79 Hughes et al.26 130 6 (0–21) 23 17.69 4 17.39 Guillem and Minsky27 188 9 (0–38) 37 19.68 1 2.70 Berho et al.28 86 13,1 (1–59) 18 20.93 2 11.11 Yeo et al.29 333 (all ypT0) 10 (0–78) 333 100 29 8.70 Jang et al.30 830 11 91 10.96 6 6.59 Tranchart et al.31 245 24 (3–60) 26 10.61 2 7.69 Park et al.32 725 11 (6–15) 143 19.72 13 9.09 Sprenger et al.33 398 28.0 ± 13.7 40 10.0 4 10.00 Current study 179 13 (2–37) 40 22.34 4 10.00

t al.28 86 13,1 (1–59) 18 20.93 2 11.11 Yeo et al.29 333 (all ypT0) 10 (0–78) 333 100 29 8.70 Jang et al.30 830 11 91 10.96 6 6.59 Tranchart et al.31 245 24 (3–60) 26 10.61 2 7.69 Park et al.32 725 11 (6–15) 143 19.72 13 9.09 Sprenger et al.33 398 28.0 ± 13.7 40 10.0 4 10.00 Current study 179 13 (2–37) 40 22.34 4 10.00 ypT0 complete pathologic response in the primary tumor, ypN pathologic lymph node status The assessment of response to treatment is becoming increasingly important in view of a personalized surgical approach. Among our patients, restaging accuracy using standard MRI and endorectal ultrasound was very low. This is in line with two recent meta-analysis leading to the conclusion that overall accuracy of restaging is not sufficiently consistent for clinical application.34,35 In addition, a nomogram using clinicopathologic parameters to predict ypN status after CRT developed in a training cohort of 891 LARC patients has been shown to achieve an accuracy of 0.77 in an external validation cohort of 258 patients.36

racy of restaging is not sufficiently consistent for clinical application.34,35 In addition, a nomogram using clinicopathologic parameters to predict ypN status after CRT developed in a training cohort of 891 LARC patients has been shown to achieve an accuracy of 0.77 in an external validation cohort of 258 patients.36 In view of the aforementioned limitations, even if surgical complications, including suture dehiscence and endoanal pain, are not uncommon among patients undergoing LE after CRT, as previously reported by us and others,18,43 this remains a procedure of investigational interest to confirm potential ypT0 status of patients with a major clinical response. On the other hand, a more conservative approach such as the “wait and see” option might be considered for patients with a complete clinical response.44 Hopefully, new techniques such as the fluorodeoxyglucose (FDG)-positron emission tomography (PET) scan and perfusion MRI might lead to a precise assessment of response.45–48 This study was limited by its single-center retrospective design and its small number of ypT0 patients displaying metastatic lymph nodes at pathology. In addition, the large time frame considered might have accounted for our relatively high local recurrence rate compared with the results of prospective clinical studies.

This study was limited by its single-center retrospective design and its small number of ypT0 patients displaying metastatic lymph nodes at pathology. In addition, the large time frame considered might have accounted for our relatively high local recurrence rate compared with the results of prospective clinical studies. In conclusion, our findings indicate that treatment protocols aimed at organ preservation in rectal cancer achieving ypT0 after CRT can be offered also to patients with clinically positive mesorectal lymph nodes at their initial diagnosis. The favorable long-term outcome for ypT0 tumors and the risk of metastatic mesorectal lymph nodes should be discussed in patient–clinician communication. Disclosure All protocols were approved by our institutional review committee and all patients were informed consent.

Neoadjuvant chemotherapy (NCT) was initially shown to downsize many large or locally advanced breast cancers, thus increasing the likelihood of clear margins with a mastectomy or lumpectomy. For triple-negative and HER2 (human epidermal growth factor receptor 2) tumors, pathologic complete response (pCR) correlates with improved survival.1–5 The Neoadjuvant BReast Symphony Trial (NBRST) found that MammaPrint/BluePrint molecular subtyping reclassifies 22 % (94 of 426) of tumors. MammaPrint/BluePrint molecular subtyping reassigned patients with more responsive disease to the HER2 and Basal categories while reassigning patients with less responsive disease to the Luminal categories. These findings suggest that compared to immunohistochemistry (IHC)/fluorescence in situ hybridization (FISH), MammaPrint/BluePrint more accurately identifies patients with disease likely to respond or not respond to NCT.6

asal categories while reassigning patients with less responsive disease to the Luminal categories. These findings suggest that compared to immunohistochemistry (IHC)/fluorescence in situ hybridization (FISH), MammaPrint/BluePrint more accurately identifies patients with disease likely to respond or not respond to NCT.6 NCT is increasingly being adopted in the clinical management of patients with more locally advanced disease and/or more aggressive tumor types. Several identified reasons for this trend include the following: an opportunity to observe tumor response to chemotherapy; improvement in operability; and association of improved survival for more aggressive subtypes in patients who experience pCR.1–5 In light of improvements in NCT regimens, there has been an increased use of this approach in patients with smaller tumors, who in the past would have more commonly gone straight to surgical lumpectomy. This has generally been seen in patients with either triple-negative or HER2 tumors with the hope that it would lead to a pCR and presumably better long-term survival. Intuitively, it seems that patients with smaller tumors would more often experience a pCR than those with larger tumors, which would in turn lead to an additional survival benefit. Thus, the purpose of this unplanned substudy was to determine if the pCR rate is also affected by tumor size and if the tumor size effect is modified by molecular subtype as determined by BluePrint molecular subtyping.

ence a pCR than those with larger tumors, which would in turn lead to an additional survival benefit. Thus, the purpose of this unplanned substudy was to determine if the pCR rate is also affected by tumor size and if the tumor size effect is modified by molecular subtype as determined by BluePrint molecular subtyping. Patients and Methods Patients Patients with histologically proven breast cancer, who had started or were scheduled to start NCT therapy or neoadjuvant hormone therapy, after successful MammaPrint/BluePrint assay were enrolled onto the prospective NBRST registry trial between June 2011 and November 2014 from 62 U.S. institutions. The trial was approved by institutional review boards in all participating centers and registered with ClinicalTrials.gov (NCT01479101). Before registration, all patients provided signed informed consent for the trial and for research on their tumor samples. Excluded from the study were patients who had an excisional biopsy or axillary dissection; patients with confirmed distant metastatic disease; patients with any prior chemotherapy, radiotherapy, or endocrine therapy for the treatment of breast cancer; and any serious uncontrolled intercurrent infections or other serious uncontrolled comorbid disease. Treatment was at the discretion of the physician adhering to either National Comprehensive Cancer Network—approved or other peer-reviewed, established regimens. No specific recommendations were given for the selection to treat patients with neoadjuvant treatment. The NBRST registry is a unique, large, real-world database of U.S. patients treated in high-volume breast programs that provides insight into physician choices for this neoadjuvant treatment—eligible population. For the current substudy, patients treated with neoadjuvant endocrine therapy were excluded; only patients with invasive ductal carcinoma were included.

world database of U.S. patients treated in high-volume breast programs that provides insight into physician choices for this neoadjuvant treatment—eligible population. For the current substudy, patients treated with neoadjuvant endocrine therapy were excluded; only patients with invasive ductal carcinoma were included. Molecular and Clinical Characteristics The 70-gene expression profile MammaPrint and the 80-gene molecular subtyping profile BluePrint were assessed from the fresh or formalin-fixed, paraffin-embedded core needle biopsy at the centralized Agendia Laboratory blinded for clinical and pathologic data. Microarray analysis (RNA labeling, microarray hybridization, and scanning) was performed on the RNA, which was cohybridized with a standard reference to the custom-designed diagnostic chip, each containing oligonucleotide probes for the profiles in triplicate or more.7,8 Four distinct molecular subgroups—Luminal A, Luminal B, HER2, and Basal—were identified and used for further analysis. In this study, we defined Luminal A-type tumors as Luminal type by BluePrint with a Low Risk score by MammaPrint, and Luminal B-type tumors as BluePrint Luminal type with a MammaPrint High Risk score.

Molecular and Clinical Characteristics The 70-gene expression profile MammaPrint and the 80-gene molecular subtyping profile BluePrint were assessed from the fresh or formalin-fixed, paraffin-embedded core needle biopsy at the centralized Agendia Laboratory blinded for clinical and pathologic data. Microarray analysis (RNA labeling, microarray hybridization, and scanning) was performed on the RNA, which was cohybridized with a standard reference to the custom-designed diagnostic chip, each containing oligonucleotide probes for the profiles in triplicate or more.7,8 Four distinct molecular subgroups—Luminal A, Luminal B, HER2, and Basal—were identified and used for further analysis. In this study, we defined Luminal A-type tumors as Luminal type by BluePrint with a Low Risk score by MammaPrint, and Luminal B-type tumors as BluePrint Luminal type with a MammaPrint High Risk score. Hormone receptor status [estrogen receptor (ER) and progesterone receptor (PR) status] and HER2 status were determined locally on pretreatment core biopsy samples. Both ER and PR status were determined by IHC and were considered positive if there was ≥1 % positive staining. HER2 status was determined by IHC and/or FISH assays locally. HER2 status was regarded as positive if there was 3+ staining and/or FISH positivity according to American Society of Clinical Oncology/College of American Pathologists HER2 testing guidelines at the time of diagnosis.

ve if there was ≥1 % positive staining. HER2 status was determined by IHC and/or FISH assays locally. HER2 status was regarded as positive if there was 3+ staining and/or FISH positivity according to American Society of Clinical Oncology/College of American Pathologists HER2 testing guidelines at the time of diagnosis. Objectives and Endpoints The primary study endpoint was pCR, defined as the absence of invasive carcinoma in both the breast and axilla at microscopic examination of the resection specimen, regardless of the presence of carcinoma-in situ (ypT0/isN0). All patients underwent pretreatment imaging of their primary tumor performed. The largest pre-NCT tumor size measurement from diagnostically used mammography, ultrasound, magnetic resonance imaging, positron emission tomography combined with computed tomography, positron emission mammography, or computed tomography was used. T stage was determined by the treating physician according to the American Joint Committee on Cancer (AJCC) 7th edition breast cancer staging.6 To determine whether lymph node status was relevant in the current analysis, we also analyzed the data using a definition of pCR in which lymph node status was not included (ypTis/0). Statistical Analysis Rates of pCR were calculated for each MammaPrint/BluePrint molecular subtype, tumor size subgroup, and tumor size by molecular subtype; the pCR rates are presented as proportions of the indicated subgroup.

To determine whether lymph node status was relevant in the current analysis, we also analyzed the data using a definition of pCR in which lymph node status was not included (ypTis/0). Statistical Analysis Rates of pCR were calculated for each MammaPrint/BluePrint molecular subtype, tumor size subgroup, and tumor size by molecular subtype; the pCR rates are presented as proportions of the indicated subgroup. Logistic regression was used to model the probability of pCR as a function of tumor size. This was modeled for the entire cohort. The odds ratio (OR) estimated from the logistic regression analysis was associated with a 2.1 cm change in tumor size (approximately equal to 1 standard deviation). In an effort to make an easily interpretable and useful tool for clinicians, we sought to establish a dichotomous variable for tumor size in order to evaluate differential pCR rates with respect to tumor size in specific molecular subtypes. Regarding the AJCC staging system, description of the primary tumor involves both size and extent of the tumor; the largest tumor diameter of T1 and T2 staged tumors is necessarily less than or equal to 5 cm. Logistic regression was therefore used to model the probability of pCR as a function of a tumor size variable dichotomized at 5 cm in the entire cohort and in each molecular subtype separately. Univariate logistic regression analyses of pCR were evaluated to identify individual patient and tumor prognostic factors. Significant factors from the univariate analyses were included in a multivariate modeling procedure in the overall cohort. Backward elimination followed by forward selection was performed to identify independently prognostic factors. The dichotomized tumor size variable was then included in the multivariate model to estimate an adjusted tumor size ORs for each of the three BluePrint molecular subtypes.

ling procedure in the overall cohort. Backward elimination followed by forward selection was performed to identify independently prognostic factors. The dichotomized tumor size variable was then included in the multivariate model to estimate an adjusted tumor size ORs for each of the three BluePrint molecular subtypes. Fisher’s exact test was used to compare pCR rates by molecular subgroup and by T stage. All calculations were performed by SAS 9.3 (SAS Institute, Cary, NC). Results Table 1 lists the pretreatment patient and tumor characteristics for the 608 evaluable patients per MammaPrint/BluePrint Molecular Subtyping group. The median age of the patients was 52 years.Table 1 Baseline patient characteristics by MammaPrint/BluePrint molecular subtyping group

All calculations were performed by SAS 9.3 (SAS Institute, Cary, NC). Results Table 1 lists the pretreatment patient and tumor characteristics for the 608 evaluable patients per MammaPrint/BluePrint Molecular Subtyping group. The median age of the patients was 52 years.Table 1 Baseline patient characteristics by MammaPrint/BluePrint molecular subtyping group Characteristic All patients (n = 608) Patients by subtype Luminal A (n = 66) Luminal B (n = 183) HER2 (n = 111) Basal (n = 248) Median age (range) 52 (18–89) 51 (33–69) 54 (22–79) 49 (23–81) 52 (18–89) Clinical tumor size (mm) Median (range) 33 (7–122) 32 (7–110) 35 (10–120) 31 (9–100) 31 (7–122) Tumor stage cT1 78 (13 %) 4 (6 %) 22 (12 %) 14 (13 %) 38 (15 %) cT2 374 (62 %) 42 (64 %) 116 (63 %) 65 (59 %) 151 (61 %) cT3 133 (22 %) 17 (26 %) 37 (20 %) 26 (23 %) 53 (21 %) cT4 23 (4 %) 3 (5 %) 8 (4 %) 6 (5 %) 6 (2 %) Nodal stage cN0 228 (38 %) 27 (41 %) 48 (26 %) 38 (34 %) 115 (46 %) cN1 308 (51 %) 33 (50 %) 109 (60 %) 62 (56 %) 104 (42 %) cN2 31 (5 %) 2 (3 %) 12 (7 %) 4 (4 %) 13 (5 %) cN3 18 (3 %) 2 (3 %) 3 (2 %) 2 (2 %) 11 (4 %) Missing 23 (4 %) 2 (3 %) 11 (6 %) 5 (5 %) 5 (2 %) Tumor grade 1 25 (4 %) 12 (18 %) 10 (5 %) 2 (2 %) 1 (< 1 %) 2 188 (31 %) 41 (62 %) 73 (40 %) 40 (36 %) 34 (14 %) 3 384 (63 %) 12 (18 %) 94 (51 %) 66 (59 %) 212 (85 %) Missing 11 (2 %) 1 (2 %) 6 (3 %) 3 (3 %) 1 (< 1 %) ER status (IHC) Negative 242 (40 %) 1 (2 %) 2 (1 %) 55 (50 %) 184 (74 %) Positive 364 (60 %) 65 (98 %) 181 (99 %) 56 (50 %) 62 (25 %) Missing 2 (<1 %) 0 0 0 2 (1 %) PR status (IHC) Negative 316 (52 %) 0 25 (14 %) 75 (68 %) 216 (87 %) Positive 290 (48 %) 66 (100 %) 158 (86 %) 36 (32 %) 30 (12 %) Missing 2 (<1 %) 0 0 0 2 (1 %) HER2 status (IHC/FISH) Negative 404 (66 %) 54 (82 %) 132 (72 %) 1 (1 %) 217 (88 %) Positive 201 (33 %) 12 (18 %) 51 (28 %) 110 (99 %) 28 (11 %) Missing 3 (< 1 %) 0 0 0 3 (1 %) MammaPrint Low Risk 68 (11 %) 66 (100 %) 0 2 (2 %) 0 High Risk 540 (89 %) 0 183 (100 %) 109 (98 %) 248 (100 %) Neoadjuvant treatment AC > T 259 (43 %) 39 (59 %) 94 (51 %) 1 (1 %) 125 (50 %) TC 59 (10 %) 9 (14 %) 24 (13 %) 0 (0 %) 26 (10 %) TAC 48 (8 %) 4 (6 %) 13 (7 %) 2 (2 %) 27 (11 %) TCH 95 (16 %) 2 (3 %) 25 (14 %) 53 (48 %) 15 (6 %) AC > TH 43 (7 %) 5 (8 %) 8 (4 %) 55 (50 %) 8 (3 %) THCP 34 (6 %) 4 (7 %) 10 (5 %) 17 (15 %) 3 (1 %) Other 73 (12 %) 6 (9 %) 9 (5 %) 16 (14 %) 44 (18 %) Surgery Mastectomy 346 (57 %) 40 (61 %) 102 (56 %) 65 (59 %) 139 (56 %) Lumpectomy 262 (43 %) 26 (39 %) 81 (44 %) 46 (41 %) 109 (44 %) Imaging (before NCT) MRI 287 (47 %) 27 (41 %) 95 (52 %) 51 (46 %)

55 (50 %) 8 (3 %) THCP 34 (6 %) 4 (7 %) 10 (5 %) 17 (15 %) 3 (1 %) Other 73 (12 %) 6 (9 %) 9 (5 %) 16 (14 %) 44 (18 %) Surgery Mastectomy 346 (57 %) 40 (61 %) 102 (56 %) 65 (59 %) 139 (56 %) Lumpectomy 262 (43 %) 26 (39 %) 81 (44 %) 46 (41 %) 109 (44 %) Imaging (before NCT) MRI 287 (47 %) 27 (41 %) 95 (52 %) 51 (46 %) 114 (46 %) Mammogram 151 (25 %) 23 (35 %) 47 (26 %) 32 (29 %) 49 (20 %) Ultrasound 135 (22 %) 16 (24 %) 33 (18 %) 22 (20 %) 64 (26 %) PET 18 (3 %) 0 6 (3 %) 2 (2 %) 10 (4 %) Other 17 (3 %) 0 2 (1 %) 4 (4 %) 11 (4 %) ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2, IHC immunohistochemistry, FISH fluorescence in situ hybridization, A doxorubicin, T taxane, C cyclophosphamide, H trastuzumab, P pertuzumab, THCP docetaxel–carboplatin–trastuzumab–pertuzumab, NCT neoadjuvant chemotherapy, MRI magnetic resonance imaging, PET positron emission tomography Tumor ER and PR status at diagnosis was determined in tumor samples from 606 patients; 60.1 % were ER positive and 47.9 % were PR positive. HER2 status at diagnosis was determined in tumor samples from 605 patients, and 33.2 % were HER2 positive. The median largest diameter of the primary tumor was 33 mm, ranging from 7 to 122 mm. A total of 25.7 % of patients presented with T3 or T4 tumors, and 61.0 % of the patients presented with clinically node-positive disease. Data analysis revealed that 95.8 % had tumors of intermediate or high histologic grade.

ositive. The median largest diameter of the primary tumor was 33 mm, ranging from 7 to 122 mm. A total of 25.7 % of patients presented with T3 or T4 tumors, and 61.0 % of the patients presented with clinically node-positive disease. Data analysis revealed that 95.8 % had tumors of intermediate or high histologic grade. Review of the chemotherapy regimens showed that the most commonly used regimens in clinical HER2-negative patients were: 59 % AC (doxorubicin/cyclophosphamide) followed by a taxane, 15 % TC (docetaxel/cyclophosphamide), and 11 % TAC (docetaxel/doxorubicin/cyclophosphamide). Of the HER2-enriched patients, 96 % received trastuzumab simultaneously with NCT, and 26 % of these patients received trastuzumab and pertuzumab. Before U.S. Food and Drug Administration approval of pertuzumab in the neoadjuvant setting (September 2013), 57 % received TCH (docetaxel/carboplatin/trastuzumab), and 28 % received AC followed by TH (doxorubicin/cyclophosphamide followed by docetaxel/trastuzumab). After September 2013, the following regimen were mostly used: 50 % TCHP (docetaxel/carboplatin/trastuzumab/pertuzumab), 20 % TCH (docetaxel/carboplatin/trastuzumab), and 10 % THP (docetaxel/trastuzumab/pertuzumab). A total of 89 % of patients completed their planned NCT without any modifications. Eight percent had dose modifications because of toxicities, 2 % stopped NCT early because of tumor progression, and no reason was specified for the remaining 1 %.

Review of the chemotherapy regimens showed that the most commonly used regimens in clinical HER2-negative patients were: 59 % AC (doxorubicin/cyclophosphamide) followed by a taxane, 15 % TC (docetaxel/cyclophosphamide), and 11 % TAC (docetaxel/doxorubicin/cyclophosphamide). Of the HER2-enriched patients, 96 % received trastuzumab simultaneously with NCT, and 26 % of these patients received trastuzumab and pertuzumab. Before U.S. Food and Drug Administration approval of pertuzumab in the neoadjuvant setting (September 2013), 57 % received TCH (docetaxel/carboplatin/trastuzumab), and 28 % received AC followed by TH (doxorubicin/cyclophosphamide followed by docetaxel/trastuzumab). After September 2013, the following regimen were mostly used: 50 % TCHP (docetaxel/carboplatin/trastuzumab/pertuzumab), 20 % TCH (docetaxel/carboplatin/trastuzumab), and 10 % THP (docetaxel/trastuzumab/pertuzumab). A total of 89 % of patients completed their planned NCT without any modifications. Eight percent had dose modifications because of toxicities, 2 % stopped NCT early because of tumor progression, and no reason was specified for the remaining 1 %. The overall pCR (ypT0/isN0) rate was 28.5 %. Comparison of pCR rates across the four MammaPrint/BluePrint molecular subgroups (on 3 degrees of freedom) was highly significant (p < 0.001). Luminal A tumors had 6.1 % pCR rate, which was not statistically significantly different (p = 0.604) from the pCR rate of 8.7 % in Luminal B tumors. The pCR rates for Luminal A and B subtypes were statistically significantly less than the pCR rates for Basal (p < 0.001 and p < 0.001, respectively) and HER2 subtypes (p < 0.001 and p < 0.001, respectively). The pCR rates for Basal (37.1 %) and HER2 (55.0 %) tumors also differed significantly (p = 0.002).

al B tumors. The pCR rates for Luminal A and B subtypes were statistically significantly less than the pCR rates for Basal (p < 0.001 and p < 0.001, respectively) and HER2 subtypes (p < 0.001 and p < 0.001, respectively). The pCR rates for Basal (37.1 %) and HER2 (55.0 %) tumors also differed significantly (p = 0.002). The rates for experiencing pCR by clinical T stage were as follows: T1 28.2 % (22 of 78), T2 31.8 % (119 of 374), T3 18.8 % (25 of 133), and T4 30.4 % (7 of 23) (Fisher’s exact test, p = 0.035). Moreover, the pCR rate for T1 and T2 stage tumors combined (31.2 %) was statistically significantly higher (p = 0.006) than the observed pCR rate in T3 stage tumors (18.8 %). Figure 1 shows the pCR rate according by clinical T stage and molecular subtype (excluding T4).Fig. 1 pCR (ypT0/isN0) rate according by clinical T stage and MammaPrint/BluePrint molecular subtyping group (excluding T4)

tatistically significantly higher (p = 0.006) than the observed pCR rate in T3 stage tumors (18.8 %). Figure 1 shows the pCR rate according by clinical T stage and molecular subtype (excluding T4).Fig. 1 pCR (ypT0/isN0) rate according by clinical T stage and MammaPrint/BluePrint molecular subtyping group (excluding T4) There were 602 patients with known tumor size included in the tumor size analyses. Because the pCR rate failed to be statistically significantly different and was low among the Luminal A and B subgroups, we decided to pool Luminal A and B in the analysis of tumor size. Subsequent analyses of tumor size were across three BluePrint molecular subgroups: pooled Luminal, HER2, and Basal. The probability of pCR significantly decreased with increasing tumor size as a continuous measure (p = 0.027). The OR of pCR associated with a 2.1 cm difference in tumor size (approximately 1 standard deviation) was 0.80 [95 % confidence interval (CI) 0.66, 0.98] (Fig. 2; Table 2). Analysis of the tumor size variable dichotomized at ≤5 versus >5 cm indicated the probability of pCR was significantly decreased in tumors >5 cm relative to smaller tumors (p = 0.022, OR 0.58, 95 % CI 0.36, 0.93).Fig. 2 pCR (ypT0/isN0) rate according to tumor size and BluePrint subtype Table 2 pCR (ypT0/isN0) rate by tumor size and MammaPrint/BluePrint molecular subtyping group

There were 602 patients with known tumor size included in the tumor size analyses. Because the pCR rate failed to be statistically significantly different and was low among the Luminal A and B subgroups, we decided to pool Luminal A and B in the analysis of tumor size. Subsequent analyses of tumor size were across three BluePrint molecular subgroups: pooled Luminal, HER2, and Basal. The probability of pCR significantly decreased with increasing tumor size as a continuous measure (p = 0.027). The OR of pCR associated with a 2.1 cm difference in tumor size (approximately 1 standard deviation) was 0.80 [95 % confidence interval (CI) 0.66, 0.98] (Fig. 2; Table 2). Analysis of the tumor size variable dichotomized at ≤5 versus >5 cm indicated the probability of pCR was significantly decreased in tumors >5 cm relative to smaller tumors (p = 0.022, OR 0.58, 95 % CI 0.36, 0.93).Fig. 2 pCR (ypT0/isN0) rate according to tumor size and BluePrint subtype Table 2 pCR (ypT0/isN0) rate by tumor size and MammaPrint/BluePrint molecular subtyping group Tumor size No. pCR/total (%) No. pCR/total (%) per MammaPrint/BluePrint subtyping group Luminal A Luminal B HER2 Basal ≤2 cm 25/89 (28 %) 0/8 (0 %) 2/22 (9 %) 9/15 (60 %) 14/44 (32 %) 2.1–3 cm 66/188 (35 %) 4/22 (18 %) 3/52 (6 %) 20/38 (53 %) 39/76 (51 %) 3.1–4 cm 34/116 (29 %) 0/14 (0 %) 3/42 (7 %) 10/14 (71 %) 21/46 (46 %) 4.1–5 cm 18/76 (24 %) 0/7 (0 %) 2/25 (8 %) 11/17 (65 %) 5/27 (19 %) >5.0 cm 27/133 (20 %) 0/15 (0 %) 6/41 (15 %) 8/23 (35 %) 13/54 (24 %) Total 170/602 (28 %) 4/66 (6 %) 16/182 (9 %) 58/107 (54 %) 92/247 (37 %) Odds ratio for pCRa (95 % CI) 0.58 (0.36, 0.93) 1.53 (0.56, 4.17) 0.36 (0.14, 0.95) 0.46 (0.23, 0.91)

(46 %) 4.1–5 cm 18/76 (24 %) 0/7 (0 %) 2/25 (8 %) 11/17 (65 %) 5/27 (19 %) >5.0 cm 27/133 (20 %) 0/15 (0 %) 6/41 (15 %) 8/23 (35 %) 13/54 (24 %) Total 170/602 (28 %) 4/66 (6 %) 16/182 (9 %) 58/107 (54 %) 92/247 (37 %) Odds ratio for pCRa (95 % CI) 0.58 (0.36, 0.93) 1.53 (0.56, 4.17) 0.36 (0.14, 0.95) 0.46 (0.23, 0.91) pCR pathologic complete response, CI confidence interval aOdds ratio for pCR associated with tumors with size >5 cm relative to ≤5 cm With significant difference in overall pCR rate for tumor size >5 versus ≤5 cm, we investigated the relationship of tumor size dichotomized at 5 cm with probability of experiencing pCR in subset analyses of molecular subgroups. When analyzed by BluePrint molecular subgroup, this relationship was statistically significant in the Basal subgroup (p = 0.026, OR 0.46, 95 % CI 0.23, 0.91) and the HER2 subgroup (p = 0.039, OR 0.36, 95 % CI 0.14, 0.95). In comparison, the dichotomized tumor size variable did not correlate with pCR rate in the pooled Luminal subgroup (p = 0.411).

lecular subgroup, this relationship was statistically significant in the Basal subgroup (p = 0.026, OR 0.46, 95 % CI 0.23, 0.91) and the HER2 subgroup (p = 0.039, OR 0.36, 95 % CI 0.14, 0.95). In comparison, the dichotomized tumor size variable did not correlate with pCR rate in the pooled Luminal subgroup (p = 0.411). The following factors were found to be significantly (p < 0.05) associated with the odds of experiencing pCR based on univariate logistic regression analyses (Table 3): clinical lymph node status, clinical tumor stage, tumor grade, ER status, PR status, HER2 status, MammaPrint result, and BluePrint result. The following factors were independently associated with the odds of experiencing pCR based on multivariate logistic regression modeling: clinical lymph node status, tumor grade, PR status, HER2 status, and BluePrint result. When the dichotomized tumor size variable was added to the multivariate base model and assessed within each BluePrint molecular subtype, the adjusted ORs for tumor size >5 versus ≤5 cm tumor were not significant in any of the BluePrint molecular subgroups (Basal subgroup, OR 0.56, 95 % CI 0.27, 1.17, p = 0.123; HER2 subgroup, OR 0.44, 95 % CI 0.15, 1.24, p = 0.119; Luminal subgroup, OR 1.9, 95 % CI 0.59, 6.17, p = 0.286).Table 3 Univariate analysis of patient and tumor characteristics associated with pCR (ypT0/isN0) versus incomplete pathologic primary tumor and axillary lymph node response to NCT

% CI 0.27, 1.17, p = 0.123; HER2 subgroup, OR 0.44, 95 % CI 0.15, 1.24, p = 0.119; Luminal subgroup, OR 1.9, 95 % CI 0.59, 6.17, p = 0.286).Table 3 Univariate analysis of patient and tumor characteristics associated with pCR (ypT0/isN0) versus incomplete pathologic primary tumor and axillary lymph node response to NCT Characteristic pCR Incomplete response Univariate p value Multivariate p value Multivariate OR (95 % CI) All patients 173 (28 %) 435 (72 %) Patient age ≤50 years 81 (29 %) 199 (71 %) 0.811 >50 years 92 (28 %) 236 (72 %) IHC ER status at diagnosisa Positive 65 (18 %) 299 (82 %) <0.001 Negative 107 (44 %) 135 (56 %) PR status at diagnosisa Positive 38 (13 %) 252 (87 %) <0.001 0.025 0.51 (0.28, 0.92) Negative 134 (42 %) 182 (58 %) Ref. HER2 status at diagnosisa Positive 84 (42 %) 117 (58 %) <0.001 <0.001 2.70 (1.72, 4.21) Negative 88 (22 %) 316 (78 %) Ref. Grade at diagnosisa 1 1 (4 %) 24 (96 %) <0.001 0.025 0.22 (0.03, 1.80) 2 34 (18 %) 154 (82 %) 0.52 (0.30, 0.88) 3 134 (35 %) 250 (65 %) Ref. T stage T1 22 (28 %) 56 (72 %) 0.046 T2 119 (32 %) 255 (68 %) T3 25 (19 %) 108 (81 %) T4 7 (30 %) 16 (70 %) Initial lymph node statusa Negative 90 (39 %) 138 (61 %) <0.001 <0.001 2.08 (1.36, 3.21) Positive 78 (22 %) 279 (78 %) Ref. BluePrint-subtype status Non-Luminal 153 (43 %) 206 (57 %) <0.001 <0.001 4.21 (2.10, 8.42) Luminal 20 (8 %) 229 (92 %) Ref. MammaPrint Low Risk 5 (7 %) 63 (93 %) <0.001 High Risk 168 (31 %) 372 (69 %)

1 1 (4 %) 24 (96 %) <0.001 0.025 0.22 (0.03, 1.80) 2 34 (18 %) 154 (82 %) 0.52 (0.30, 0.88) 3 134 (35 %) 250 (65 %) Ref. T stage T1 22 (28 %) 56 (72 %) 0.046 T2 119 (32 %) 255 (68 %) T3 25 (19 %) 108 (81 %) T4 7 (30 %) 16 (70 %) Initial lymph node statusa Negative 90 (39 %) 138 (61 %) <0.001 <0.001 2.08 (1.36, 3.21) Positive 78 (22 %) 279 (78 %) Ref. BluePrint-subtype status Non-Luminal 153 (43 %) 206 (57 %) <0.001 <0.001 4.21 (2.10, 8.42) Luminal 20 (8 %) 229 (92 %) Ref. MammaPrint Low Risk 5 (7 %) 63 (93 %) <0.001 High Risk 168 (31 %) 372 (69 %) pCR pathologic complete response, NCT neoadjuvant chemotherapy, OR odds ratio, CI confidence interval, IHC immunohistochemistry, PR progesterone receptor, ER estrogen receptor aIncluded in multivariate modeling of dichotomized tumor size and molecular subtype With the use of the pCR definition excluding nodal status (ypT0/is), the pCR rates were slightly higher; however, there was no difference in statistically significant versus nonsignificant analyses (data not shown). Discussion The NBRST registry provides a unique opportunity to study whether smaller tumors are likely to demonstrate pCR. This would imply an associated improved survival. In this current analysis of the NBRST study, we found that although size was correlated inversely with the frequency of pCR in BluePrint HER2 and Basal subtypes, this was not demonstrated in the multivariate logistic regression analyses, where BluePrint molecular subtype was the strongest of the characteristics associated with pCR (grade, HER2, PR, and lymph node status).

that although size was correlated inversely with the frequency of pCR in BluePrint HER2 and Basal subtypes, this was not demonstrated in the multivariate logistic regression analyses, where BluePrint molecular subtype was the strongest of the characteristics associated with pCR (grade, HER2, PR, and lymph node status). Patients who experience pCR defined as ypT0 ypN0 or ypT0/is ypN0 have improved survival.1,3 In our prospective registry study, we used the recommended definition of pCR (ypT0/isN0). All pCRs were verified with a deidentified copy of the surgical pathology report. The overall pCR rate to NCT of 28 % in our study is higher than the pCR rate of 20 % using the same definition in recently published meta-analyses.1,3 This may be because in our study almost all clinical HER2-positive patients received trastuzumab, while this was not yet the case in the pooled meta-analyses. In the meta-analyses of Cortazar et al., patients with clinical T1 or T2 tumors had nonstatistically significant higher pCR rates than patients with T3 and T4 tumors. Furthermore, clinical tumor size was not significantly correlated with an increase in overall survival.3 Another study investigating clinical characteristics and pCR association found an association of tumor size and pCR in univariate analysis, which disappeared in multivariate analysis.9

atients with T3 and T4 tumors. Furthermore, clinical tumor size was not significantly correlated with an increase in overall survival.3 Another study investigating clinical characteristics and pCR association found an association of tumor size and pCR in univariate analysis, which disappeared in multivariate analysis.9 Tumor size and lymph node status are the two most important determinants in the AJCC staging manual. Tumor size correlates with long-term survival, as patients with smaller tumors have a lower T stage and generally a better long-term prognosis from their breast cancer compared to those with larger tumors and a higher T stage.10 Whitworth et al. have previously reported in the NBRST study on the impact of MammaPrint/BluePrint molecular subtypes and the frequency of pCR in contrast to clinical subtypes (derived by ER, PR, and HER2). Luminal A and B tumors do not usually respond with a pCR to NCT, while pCR rates are significantly higher in HER2 and Basal subtypes.6 Some ER-positive highly proliferative tumors do respond to chemotherapy; indeed, identification of patients who are best treated by endocrine therapy versus cytotoxic chemotherapy is an area of great current interest and active investigation in clinical trials. It also has been shown that patients with HER2-positive or triple-negative cancers are more likely to experience improved long-term survival if they have a pCR after NCT.1–5

Whitworth et al. have previously reported in the NBRST study on the impact of MammaPrint/BluePrint molecular subtypes and the frequency of pCR in contrast to clinical subtypes (derived by ER, PR, and HER2). Luminal A and B tumors do not usually respond with a pCR to NCT, while pCR rates are significantly higher in HER2 and Basal subtypes.6 Some ER-positive highly proliferative tumors do respond to chemotherapy; indeed, identification of patients who are best treated by endocrine therapy versus cytotoxic chemotherapy is an area of great current interest and active investigation in clinical trials. It also has been shown that patients with HER2-positive or triple-negative cancers are more likely to experience improved long-term survival if they have a pCR after NCT.1–5 Even though tumor size would intuitively be a clinical determinant of pCR, the current unplanned subanalysis in the prospective neoadjuvant NBRST study showed that the adjusted OR for tumor size was not statistically significant in any of the BluePrint molecular subgroups. Factors significantly associated with pCR were PR status, HER2 status, grade, lymph node status, and BluePrint molecular subtyping, which had the strongest correlation. Acknowledgment We are grateful to all women participating in this study, as well as all the investigators, surgeons, pathologists, and research nurses. We also thank Tina Treece (Agendia Inc.) for bioinformatics support and Erin Yoder (Agendia Inc.) for data management support. Disclosure Lisette Stork-Sloots and Femke A de Snoo are consultants for Agendia NV.

Colorectal cancer (CRC) is one of the most prevalent carcinomas throughout the world.1 The liver is the most common organ for distant CRC metastasis, and liver metastasis is the leading cause of cancer-related death in CRC patients.2 Approximately 30–50 % of CRC patients develop local tumor recurrence or distant metastasis after curative resection of the primary lesion.3,4 New diagnostic markers that can detect early metastasis or predict the risk of metastasis in CRC are in urgent demand.5 Micro RNAs (miRNAs) are small, noncoding RNA molecules of ~19–25 nucleotides that potentially regulate 20–30 % of gene expression.6 Each miRNA has numerous targets, and depending on its target genes, a miRNA can play significant roles in tumor metastasis.7 In recent years, over 100 miRNAs have been implicated in CRC.8 However, more influential research is required to reveal how miRNAs act within the context of the molecular mechanisms for disease recurrence of CRC. In this study, we conducted a microarray-based analysis to recognize differentially expressed miRNAs in CRC by comparing miRNA profiles among primary CRC tissues from patients without liver metastasis, primary tissues with liver metastasis, and liver metastatic lesions. After the miRNA array analysis, we evaluated the role of miR-132 in human CRC.

microarray-based analysis to recognize differentially expressed miRNAs in CRC by comparing miRNA profiles among primary CRC tissues from patients without liver metastasis, primary tissues with liver metastasis, and liver metastatic lesions. After the miRNA array analysis, we evaluated the role of miR-132 in human CRC. Materials and Methods Collection of Human Tissue Specimens The testing cohort consisted of 28 primary CRC lesions and eight metastatic liver tumors that arose from CRC. Sixteen primary CRC lesion samples were collected from patients with stage II and III disease without liver metastasis during the follow-up period (median 1898 days, range 27–2155 days) after operation. Twelve primary CRC lesion samples were from patients with concomitant liver metastasis. The extended independent validation cohort consisted of tissue samples from 151 patients [135 primary lesions (109 without liver metastasis and 26 with liver metastasis), and 16 liver metastatic lesions arising from CRC]. These 135 patients included patients with stage I (n = 30), stage II (n = 42), and stage III (n = 37) CRC. The median follow-up period for these 135 patients was 1653 days (range 27–2228 days). The clinical characteristics of the testing cohort and the validation cohort are summarized in Supplementary Table S1. To compare the expression of miRNA in tumor versus normal tissues, the primary CRC tissues (n = 21) and their adjacent normal tissues (n = 21) were assessed using quantitative real-time PCR (qRT-PCR). All patients were resected with curative intent between 2003 and 2013 at Osaka University Hospital and its three related hospitals. The samples were stored at −80 °C as a fresh frozen samples with RNAlater (Ambion, Austin, TX, USA) until RNA extraction. All patients provided written informed consent, in accordance with the guidelines approved by the institutional research board of each institute.

University Hospital and its three related hospitals. The samples were stored at −80 °C as a fresh frozen samples with RNAlater (Ambion, Austin, TX, USA) until RNA extraction. All patients provided written informed consent, in accordance with the guidelines approved by the institutional research board of each institute. RNA Extraction Total RNA from tissues and cells were isolated using the miRNeasy Mini Kit (Qiagen, Hilden, Germany) and Trizol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. Microarray Analysis Microarray analysis (Sureprint G3 Humans miRNA 8 × 60 K; Agilent Technologies, Santa Clara, CA, USA) was performed for the testing cohort (n = 36) at Hokkaido System Science Corporation (Hokkaido, Japan) using miRNA Complete Labeling Reagent and Hyb Kit (Agilent Technologies). The microarray raw data are available in Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) database with accession code GSE72199. Reverse Transcription PCR and TaqMan miRNA Assay In the TaqMan microRNA Assay (Applied Biosystems, Foster City, CA, USA), hsa-miR-132 ID 000457 and RNU6B ID 001093 were used to measure miRNA levels. The TaqMan MicroRNA Reverse Transcription kit and TaqMan 2× universal PCR Master Mix, No AmpErase UNG (Applied Biosystems), were used according to the manufacturer’s protocol. The 7900HT Sequence Detection System 2.3 (Applied Biosystems) software was used to compute the relative change in RNA expression by the 2-ΔΔCt method.

qMan MicroRNA Reverse Transcription kit and TaqMan 2× universal PCR Master Mix, No AmpErase UNG (Applied Biosystems), were used according to the manufacturer’s protocol. The 7900HT Sequence Detection System 2.3 (Applied Biosystems) software was used to compute the relative change in RNA expression by the 2-ΔΔCt method. qRT-PCR Total RNA was reverse transcribed using the High Capacity RNA-to-cDNA Kit (Applied Biosystems). We performed qRT-PCR using the TaqMan Gene Expression Assay (Applied Biosystems) following the manufacturer’s protocol. Cell Lines and Cell Culture All human CRC cell lines (DLD-1 and HCT116) were obtained from the American Type Culture Collection in 2001. Cell lines were cultured in Dulbecco’s modified Eagle medium (DMEM D6046; Sigma Aldrich, St. Louis, MO, USA) containing 10 % fetal bovine serum in a humidified incubator under 5 % CO2 at 37 °C. Cell Transfection To perform transient transfections, the cells were transfected using Lipofectamine RNAiMAX (Invitrogen), with 25 nmol/L of miRVana miRNA mimic miR-132–3p (#4464066) and miRVana miRNA inhibitor miR-132–3p (#4464064), according to the manufacturer’s instructions. The control miRNAs [Negative Control #1 mimic (#4464058) and Negative Control #1 (#4464074)] were used as a control for nonspecific effects. Proliferation Assays Cells were seeded at a density of 2.5–3 × 104 per well in 24-well dishes and cultured for 72 h to determine proliferation. Cells were counted with a Celltac automatic hematology analyzer (Nihon Kohden, Tokyo, Japan).

Cell Transfection To perform transient transfections, the cells were transfected using Lipofectamine RNAiMAX (Invitrogen), with 25 nmol/L of miRVana miRNA mimic miR-132–3p (#4464066) and miRVana miRNA inhibitor miR-132–3p (#4464064), according to the manufacturer’s instructions. The control miRNAs [Negative Control #1 mimic (#4464058) and Negative Control #1 (#4464074)] were used as a control for nonspecific effects. Proliferation Assays Cells were seeded at a density of 2.5–3 × 104 per well in 24-well dishes and cultured for 72 h to determine proliferation. Cells were counted with a Celltac automatic hematology analyzer (Nihon Kohden, Tokyo, Japan). Colony Formation For colony formation assays, 500 transfected cells were plated in six-well plates. After incubation at 37 °C for 11 days, visible colonies were fixed with formalin and stained with Giemsa solution, and the numbers of colonies were counted using a microscope in 10 random visual fields (×4 magnification, n = 4). Invasion Assay The cell invasion assay was performed using transwell inserts with 8 μm pores (BD Biosciences, San Jose, CA, USA), in accordance with the manufacturer’s protocol. The invading cells were counted using a microscope in three random visual fields (×200 magnification).

Colony Formation For colony formation assays, 500 transfected cells were plated in six-well plates. After incubation at 37 °C for 11 days, visible colonies were fixed with formalin and stained with Giemsa solution, and the numbers of colonies were counted using a microscope in 10 random visual fields (×4 magnification, n = 4). Invasion Assay The cell invasion assay was performed using transwell inserts with 8 μm pores (BD Biosciences, San Jose, CA, USA), in accordance with the manufacturer’s protocol. The invading cells were counted using a microscope in three random visual fields (×200 magnification). Luciferase Reporter Assay We amplified the ANO1 3′ untranslated region (3′ UTR) containing the putative miR-132 binding sites by PCR using following primers: ANO1 (forward) 5′-GCTCGCTAGCCTCGAGGGGGCGTGGGAGCATCC-3′, (reverse) 5′-ATGCCTGCAGGTCGATGGCAGATTAAAGGGAATGT-3′. The resulting DNA fragment was inserted at restriction sites immediately downstream of the luciferase gene in the pmirGLO vector (Promega, Madison, WI, USA). Luciferase activity was measured using the Dual Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized against Renilla luciferase activity for each transfected well.

rted at restriction sites immediately downstream of the luciferase gene in the pmirGLO vector (Promega, Madison, WI, USA). Luciferase activity was measured using the Dual Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized against Renilla luciferase activity for each transfected well. Statistical Analysis Data were expressed as means ± standard deviations. Statistical analysis was performed using JMP Pro 10 software (SAS Institute, Cary, NC, USA). Statistical significance was compared using the Chi square test, and continuous variables were compared using Student’s t test or one-way analysis of variance, as appropriate. Survival curves were generated using the Kaplan–Meier method and assessed using the log-rank test. The Cox proportional hazard regression model was performed to identify independent prognostic factors. A value of P < 0.05 was considered statistically significant. Results Expression of miR-132 in Colorectal and Hepatic Tissue Samples

Statistical Analysis Data were expressed as means ± standard deviations. Statistical analysis was performed using JMP Pro 10 software (SAS Institute, Cary, NC, USA). Statistical significance was compared using the Chi square test, and continuous variables were compared using Student’s t test or one-way analysis of variance, as appropriate. Survival curves were generated using the Kaplan–Meier method and assessed using the log-rank test. The Cox proportional hazard regression model was performed to identify independent prognostic factors. A value of P < 0.05 was considered statistically significant. Results Expression of miR-132 in Colorectal and Hepatic Tissue Samples To explore the possible role of miRNA in CRC, miRNA expression was profiled in the testing cohort, which consisted of primary CRC tumor tissue from patients without liver metastasis (n = 16), tumor tissue from patients with liver metastasis (n = 12), and liver metastatic lesions (n = 8). MiRNA array analysis identified 39 miRNAs (Supplementary Table S2) whose expression levels differed between primary CRC lesions without and with liver metastasis (>2-fold change). Among them, we found that a tumor suppressor miR-132 expression was significantly lower in CRC liver metastatic lesions than in primary CRC lesions without liver metastasis (Fig. 1a).9 The microarray results of miR-132 expression were validated by qRT-PCR in the same testing cohort of 36 CRC tissues. The expression of miR-132 was significantly higher in primary CRC lesions without and with liver metastasis than in liver metastatic lesions (Supplementary Fig. S1a). A statistically significant correlation was observed between microarray data and qRT-PCR miR-132 expression data (r = 0.549, P = 0.0005; Supplementary Fig. S1b).Fig. 1 Expression of miR-132 in colorectal and hepatic tissue samples. a Microarray data showed that miR-132 expression was significantly lower in liver metastatic lesions than in primary CRC tumors from patients without liver metastasis. b In extended validation cohort, qRT-PCR showed that miR-132 expression was significantly down-regulated in primary CRC lesions with liver metastasis and in liver metastatic lesions compared to primary CRC lesions without liver metastasis. c MiR-132 expression of seven pairs of primary CRC was significantly down-regulated in corresponding synchronous liver metastases. d MiR-132 expression was significantly increased in tumor tissues (n = 21) compared to their pair-matched adjacent normal colon tissues (P = 0.0268)

ry CRC lesions without liver metastasis. c MiR-132 expression of seven pairs of primary CRC was significantly down-regulated in corresponding synchronous liver metastases. d MiR-132 expression was significantly increased in tumor tissues (n = 21) compared to their pair-matched adjacent normal colon tissues (P = 0.0268) On the basis of the preliminary data, we then examined miR-132 expression by qRT-PCR in the independent and extended validation cohort of 151 patients (109 primary CRC lesions without liver metastasis, 26 primary lesions with liver metastasis, and 16 liver metastatic lesions). As results, miR-132 expression was significantly down-regulated in primary CRC lesions with liver metastasis and in liver metastatic lesions compared to primary CRC lesions without liver metastasis (Fig. 1b, P = 0.0019 and P = 0.0002, respectively). In this study, we analyzed seven pairs of primary CRC and corresponding synchronous liver metastases that were collected from both microarray and validation cohort. MiR-132 expression was significantly down-regulated in corresponding liver metastases compared to primary CRC lesions (Fig. 1c, P = 0.0462). When we measured the difference expression between tumor and normal tissues, miR-132 expression was significantly increased in tumor tissues (n = 21) compared to their pair-matched adjacent normal colonic tissues (Fig. 1d, P = 0.0268).

orresponding liver metastases compared to primary CRC lesions (Fig. 1c, P = 0.0462). When we measured the difference expression between tumor and normal tissues, miR-132 expression was significantly increased in tumor tissues (n = 21) compared to their pair-matched adjacent normal colonic tissues (Fig. 1d, P = 0.0268). Overexpression of miR-132 Suppresses CRC Cell Growth and Invasion In Vitro CRC cells were transfected with miR-132 mimics, and the transfection efficiency was measured through real-time PCR. There was significantly higher miR-132 expression in miR-132-transfected cells at every time point compared to negative control (NC) miR-transfected cells (Supplementary Fig. S2). MiR-132-transfected CRC cells demonstrated a significantly slower growth rate than NC (Fig. 2a). Results of the colony-formation assay indicated that miR-132 overexpression significantly inhibited tumor growth (Fig. 2b). Because cell invasion is an initial step of metastasis, we examined the effect of miR-132 on the invasive capacity of CRC cell lines. MiR-132 overexpression markedly reduced the invasion ability compared to NC (Fig. 2c). In contrast, inhibiting miR-132 expression (inhibitor miR-132) showed increased invasion ability compared to NC (Fig. 2c).Fig. 2 In vitro transduction of miR-132 in CRC cells. a MiR-132-transfected cells demonstrated significantly slower growth rate than negative control-transfected cells. b Colony formation assay showed that miR-132 overexpression resulted in significant inhibition of tumor growth. c MiR-132 overexpression markedly reduced invasion ability compared to negative control. In contrast, miR-132 inhibitor increased invasion ability. Data are presented as mean ± SD. NC negative control-transfected cells, mimic miR-132 miR-132-transfected cells, inhibitor miR-132 inhibitor miR-132-transfected cells. *P < 0.05

-132 overexpression markedly reduced invasion ability compared to negative control. In contrast, miR-132 inhibitor increased invasion ability. Data are presented as mean ± SD. NC negative control-transfected cells, mimic miR-132 miR-132-transfected cells, inhibitor miR-132 inhibitor miR-132-transfected cells. *P < 0.05 MiR-132 Directly Targets 3′ UTR of ANO1 in CRC Cells To reveal the biological potential role of miR-132 in CRC, we explored the targets of miR-132. Using 390 expression arrays of CRC in the GEO database (GSE41258) and the target prediction tool TargetScan 6.2, we predicted the potential target genes. We identified 54 genes (Supplementary Table S3) and focused ANO1 as a potential target of miR-132. MiR-132 overexpression markedly suppressed ANO1 mRNA expression levels (Fig. 3a), while miR-132 inhibitor resulted in the up-regulation of ANO1 mRNA expression levels (Fig. 3b). The 3′ UTR of ANO1 mRNA contains a complementary site for the seed region of miR-132 (Fig. 3c). More specifically, in order to show that ANO1 is a direct target of miR-132, we cotransfected miR-132 expression vector along with the ANO1 3′ UTR sequence containing luciferase reporter constructs. The activity of a luciferase reporter containing the predicted miR-132 binding sequence of ANO1 3′ UTR was significantly repressed by the ectopic expression of miR-132 (Fig. 3d).Fig. 3 ANO1 as possible target of miR-132. a MiR-132 expression markedly suppressed ANO1 mRNA levels. b MiR-132 inhibitor resulted in up-regulation of ANO1 mRNA levels. c The 3′ UTR of ANO1 mRNA contains site for seed region of miR-132. d Activity of luciferase reporter containing predicted miR-132 binding sequence of ANO1 3′ UTR was significantly repressed by ectopic expression of miR-132. e ANO1 expression was higher in primary CRC with liver metastasis and in liver metastases than in primary tumors without metastasis (P = 0.0059 and P = 0.0001, respectively). f Significant inverse correlation was observed between miR-132 expression and ANO1 expression in validation cohort. Data are presented as mean ± SD. NC negative control-transfected cells, mimic miR-132 miR-132-transfected cells, inhibitor miR-132 inhibitor miR-132-transfected cells. *P < 0.05

P = 0.0001, respectively). f Significant inverse correlation was observed between miR-132 expression and ANO1 expression in validation cohort. Data are presented as mean ± SD. NC negative control-transfected cells, mimic miR-132 miR-132-transfected cells, inhibitor miR-132 inhibitor miR-132-transfected cells. *P < 0.05 Up-regulation of ANO1 Is Inversely Associated with Down-regulation of miR-132 in CRC Clinical Samples The above results prompted us to investigate whether miR-132 suppresses CRC growth and metastasis through ANO1 suppression. To further confirm the clinical relationship between miR-132 and ANO1 in CRC patients, we utilized qRT-PCR to examine ANO1 expression in the same validation cohort of 151 tissue samples that were used to examine miR-132 expression. ANO1 expression was higher in primary CRC with liver metastasis and in liver metastases than in primary tumors without metastasis (P = 0.0059, and P = 0.0001, respectively) (Fig. 3e). There was no significant difference in expression between tumors with liver metastasis and liver metastasis. Moreover, we found a significant inverse correlation between miR-132 expression levels and ANO1 expression levels in the validation cohort (r = −0.225 P = 0.0055; Fig. 3f). MiR-132 Acts as Prognostic Marker of CRC

Up-regulation of ANO1 Is Inversely Associated with Down-regulation of miR-132 in CRC Clinical Samples The above results prompted us to investigate whether miR-132 suppresses CRC growth and metastasis through ANO1 suppression. To further confirm the clinical relationship between miR-132 and ANO1 in CRC patients, we utilized qRT-PCR to examine ANO1 expression in the same validation cohort of 151 tissue samples that were used to examine miR-132 expression. ANO1 expression was higher in primary CRC with liver metastasis and in liver metastases than in primary tumors without metastasis (P = 0.0059, and P = 0.0001, respectively) (Fig. 3e). There was no significant difference in expression between tumors with liver metastasis and liver metastasis. Moreover, we found a significant inverse correlation between miR-132 expression levels and ANO1 expression levels in the validation cohort (r = −0.225 P = 0.0055; Fig. 3f). MiR-132 Acts as Prognostic Marker of CRC The clinicopathologic implications of miR-132 expression were assessed in 135 CRC patients from the validation cohort of primary CRC tissue samples. To clarify the correlation of miR-132 expression and postoperative survival of patients, we divided the patients into two groups according to the median value (1.299) of the miR-132 expression. Low miR-132 expression was positively associated with tumor size, depth, lymph node metastasis, venous permeation, and clinical disease stage (Supplementary Table S4). Kaplan–Meier survival curves showed that patients with low miR-132 expression demonstrated significantly worse clinical outcome [overall survival (OS): log-rank test P = 0.0021; median follow-up: 1653 days; Fig. 4a]. Univariate analysis for OS revealed that tumor depth, lymph node metastasis, lymphatic permeation, venous permeation, clinical stage, and miR-132 expression were significantly associated with OS (Table 1). Multivariate analysis for OS indicated that miR-132 expression (relative risk 3.838, 95 % confidence interval 1.054–24.683, P = 0.040), lymphatic permeation, and clinical stage were independent prognostic factors for CRC patients (Table 2).Fig. 4 MiR-132 acts as prognostic marker for CRC. a Kaplan–Meier survival curves showed that patients with low miR-132 expression demonstrated poorer clinical outcome (OS, P = 0.0021, median follow-up 1653 days). b Patients with low ANO1 levels had more favorable clinical outcome (OS) than patients with high ANO1 levels. c DFS rate were significantly lower in patients with low miR-132 expression than in patients with high miR-132 expression (P = 0.0220)

rated poorer clinical outcome (OS, P = 0.0021, median follow-up 1653 days). b Patients with low ANO1 levels had more favorable clinical outcome (OS) than patients with high ANO1 levels. c DFS rate were significantly lower in patients with low miR-132 expression than in patients with high miR-132 expression (P = 0.0220) Table 1 Univariate and multivariate analyses for overall survival Characteristic Univariate analysis P Multivariate analysis P RR 95 % CI RR 95 % CI Sex Male/female 0.586 0.215–1.593 0.288 Lesion Colon/rectum 0.783 0.288–2.129 0.626 Differentiation tub1, tub2/muc, por 0.302 0.084–1.933 0.173 Tumor size ≤35 mm/>35 mm 0.566 0.130–1.756 0.347 Depth T1, T2/T3, T4 <0.0001 0.386–0.386 0.003* <0.0001 <0.0001–2.574 0.192 Lymph node metastasis Negative/positive 0.156 0.036–0.483 0.001* 1.852 0.409–6.243 0.383 Lymphatic permeation Negative/positive 0.064 0.004–0.315 <0.0001* 0.103 0.006–0.537 0.004* Venous permeation Negative/positive 0.277 0.064–0.861 0.025* 0.890 0.196–2.999 0.860 Stage I, II/III, IV <0.0001 0.0968–0.0968 <0.0001* <0.0001 <0.0001–0.218 0.002* miR-132 expression Low/high 7.255 2.026–46.207 0.001* 3.838 1.054–24.683 0.040* RR relative risk, CI confidence interval * Statistically significant Table 2 Univariate and multivariate analyses for disease-free survival Characteristic Univariate analysis Multivariate analysis RR 95 % CI P RR 95 % CI P

RR 95 % CI RR 95 % CI Sex Male/female 0.586 0.215–1.593 0.288 Lesion Colon/rectum 0.783 0.288–2.129 0.626 Differentiation tub1, tub2/muc, por 0.302 0.084–1.933 0.173 Tumor size ≤35 mm/>35 mm 0.566 0.130–1.756 0.347 Depth T1, T2/T3, T4 <0.0001 0.386–0.386 0.003* <0.0001 <0.0001–2.574 0.192 Lymph node metastasis Negative/positive 0.156 0.036–0.483 0.001* 1.852 0.409–6.243 0.383 Lymphatic permeation Negative/positive 0.064 0.004–0.315 <0.0001* 0.103 0.006–0.537 0.004* Venous permeation Negative/positive 0.277 0.064–0.861 0.025* 0.890 0.196–2.999 0.860 Stage I, II/III, IV <0.0001 0.0968–0.0968 <0.0001* <0.0001 <0.0001–0.218 0.002* miR-132 expression Low/high 7.255 2.026–46.207 0.001* 3.838 1.054–24.683 0.040* RR relative risk, CI confidence interval * Statistically significant Table 2 Univariate and multivariate analyses for disease-free survival Characteristic Univariate analysis Multivariate analysis RR 95 % CI P RR 95 % CI P Sex Male/female 0.978 0.295–3.733 0.972 Lesion Colon/rectum 0.697 0.201–2.314 0.550 Differentiation tub1, tub2/muc, por 0.148 0.038–0.971 0.047* 0.099 0.016–0.765 0.030* Tumor size ≤35 mm/>35 mm 0.235 0.013–1.227 0.094 Depth T1, T2/T3, T4 <0.0001 0.429–0.429 0.005* <0.0001 0.269–1.201 0.071 Lymph node metastasis Negative/positive 0.179 0.039–0.620 0.006* 0.331 0.071–1.183 0.090 Lymphatic permeation Negative/positive 0.286 0.063–0.990 0.048* 0.277 0.060–0.978 0.046* Venous permeation Negative/positive 0.767 0.221–2.547 0.661 miR-132 expression Low/high 5.01 1.290–32.834 0.018* 5.838 1.374–42.770 0.015* RR relative risk, CI confidence interval * Statistically significant

Sex Male/female 0.978 0.295–3.733 0.972 Lesion Colon/rectum 0.697 0.201–2.314 0.550 Differentiation tub1, tub2/muc, por 0.148 0.038–0.971 0.047* 0.099 0.016–0.765 0.030* Tumor size ≤35 mm/>35 mm 0.235 0.013–1.227 0.094 Depth T1, T2/T3, T4 <0.0001 0.429–0.429 0.005* <0.0001 0.269–1.201 0.071 Lymph node metastasis Negative/positive 0.179 0.039–0.620 0.006* 0.331 0.071–1.183 0.090 Lymphatic permeation Negative/positive 0.286 0.063–0.990 0.048* 0.277 0.060–0.978 0.046* Venous permeation Negative/positive 0.767 0.221–2.547 0.661 miR-132 expression Low/high 5.01 1.290–32.834 0.018* 5.838 1.374–42.770 0.015* RR relative risk, CI confidence interval * Statistically significant To examine the association of ANO1 expression with clinical outcome, we divided the patients into two groups according to the median value (0.0055) of the ANO1 expression. Kaplan–Meier survival curves revealed that patients with low ANO1 expression had a more favorable clinical outcome (OS) than patients with high ANO1 expression (log-rank test P = 0.0344, Fig. 4b). These results suggest that ANO1 up-regulation through miR-132 suppression might affect the clinical outcome (OS) of patients with CRC.

r survival curves revealed that patients with low ANO1 expression had a more favorable clinical outcome (OS) than patients with high ANO1 expression (log-rank test P = 0.0344, Fig. 4b). These results suggest that ANO1 up-regulation through miR-132 suppression might affect the clinical outcome (OS) of patients with CRC. In 109 samples from patients with CRC (excluding 26 stage IV CRCs with liver metastasis), disease-free survival (DFS) rate were significantly lower in patients with low miR-132 expression than in patients with high miR-132 expression (log-rank test P = 0.0220, Fig. 4c). Low miR-132 expression was positively associated with large tumor size and deep invasion (Supplementary Table S5). In addition, univariate and multivariate analysis showed that miR-132 expression was an independent risk factor for DFS (P = 0.015; median follow-up: 1659 days; Table 2) in CRC patients. On the other hand, ANO1 expression was not a significant parameter for DFS (data not shown). Discussion Recent miRNA studies have highlighted cancer invasion and metastasis. Following microarray study, miR-214 and miR-181a were already suggested as regulators of CRC liver metastasis.10,11 In the present study, we examined miRNA profiles in primary CRC lesions with and without liver metastasis and in liver metastatic lesions to identify the key miRNA related to CRC.

invasion and metastasis. Following microarray study, miR-214 and miR-181a were already suggested as regulators of CRC liver metastasis.10,11 In the present study, we examined miRNA profiles in primary CRC lesions with and without liver metastasis and in liver metastatic lesions to identify the key miRNA related to CRC. On the basis of the microarray data, we validated the miR-132 expression in a larger independent validation cohort. MiR-132 was definitely down-regulated in primary CRC lesions with liver metastasis and also in liver metastatic lesions. Furthermore, down-regulation of miR-132 was associated with poorer OS and DFS in patients with CRC.

the microarray data, we validated the miR-132 expression in a larger independent validation cohort. MiR-132 was definitely down-regulated in primary CRC lesions with liver metastasis and also in liver metastatic lesions. Furthermore, down-regulation of miR-132 was associated with poorer OS and DFS in patients with CRC. MiR-132 has been reported as a tumor suppressor in a series of cancers.9,12–15 Zheng et al. showed that miR-132 inhibits CRC invasion and metastasis via directly targeting ZEB2 using 62 CRC samples.15 Our study extended their findings with a possible target ANO1 using a large scale of CRC samples (n = 163) as well as their metastatic lesions to liver (n = 24). Of interest was that low miR-132 expression was still maintained in hepatic metastatic lesions (Fig. 1b). Moreover, direct comparison of seven paired samples between primary CRC tumor and its corresponding synchronous hepatic metastases showed that hepatic metastases had even smaller amount of miR-132 compared to primary CRC tumors (Fig. 1c), although the paired number was limited in this study. From a therapeutic point of view, therefore, it is worth exploring whether mimic miR-132 would inhibit liver metastasis in animal models. Further study would clarify more about the close link of miR-132 with liver metastasis.

ompared to primary CRC tumors (Fig. 1c), although the paired number was limited in this study. From a therapeutic point of view, therefore, it is worth exploring whether mimic miR-132 would inhibit liver metastasis in animal models. Further study would clarify more about the close link of miR-132 with liver metastasis. When we compared miR-132 levels between normal and tumor tissues, we found that miR-132 expression was even higher in tumor tissues than in normal tissues. This result appears to be paradoxical to our finding of down-regulation of miR-132 in advanced or metastatic stage tumors. In this regard, Kara et al. showed a supportive result that miR-132 in CRC tissues had approximately threefold increase of that in normal mucosa.16 Therefore, we postulate that miR-132 might have differential roles in carcinogenesis and in tumor progression, although further studies are required to reach a definitive conclusion.

gard, Kara et al. showed a supportive result that miR-132 in CRC tissues had approximately threefold increase of that in normal mucosa.16 Therefore, we postulate that miR-132 might have differential roles in carcinogenesis and in tumor progression, although further studies are required to reach a definitive conclusion. Each miRNA can potentially down-regulate many target genes by binding their 3′ UTRs. Our results showed that ANO1 is a target of miR-132 that has a crucial role in CRC progression. ANO1 is also known as discovered on gastrointestinal stromal tumor protein (DOG1) and tumor-amplified and overexpressed sequence 2 (TMEM16A), and it is highly deregulated in different human cancers.17–19 These studies showed that ANO1 was a critical oncogenic factor that contributed to cell motility, invasion, and adhesion. Furthermore, the overexpression of ANO1 has a significant effect on both distant metastasis and poor prognosis. Our results suggest a potential role for ANO1 in CRC through miR-132. ANO1, which consists of 26 exons and encodes a protein that contains eight transmembrane regions, acts as a calcium-activated chloride channel.20–22 Calcium-activated chloride channels have recently become notable as a new drug target for anticancer therapy.23 ANO1 may contribute to this potential new approach to cancer therapy.

h consists of 26 exons and encodes a protein that contains eight transmembrane regions, acts as a calcium-activated chloride channel.20–22 Calcium-activated chloride channels have recently become notable as a new drug target for anticancer therapy.23 ANO1 may contribute to this potential new approach to cancer therapy. In conclusion, we demonstrated that lower expression of miR-132 in clinical CRC specimens was associated with CRC progression and poor survival. The tumor-suppressing function of miR-132 may in part be realized through targeting of the downstream gene ANO1. Our data suggest that further study is essentially important to reveal the relation between miR-132 and liver metastasis as well as the potential of miR-132 as a therapeutic target of CRC. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Fig. S1 Validation of microarray data of the preliminary cohort by qRT-PCR. (a) qRT-PCR showed that the relative expression of miR-132 was significantly higher in primary CRC lesions without and with liver metastasis than in liver metastatic lesions. (b) There is a significant correlation between microarray data and qRT-PCR expression data. Supplementary material 1 (TIFF 1429 kb) Supplementary Fig. S2 Transfection efficiency data. The data are presented as the mean ± SD. (NC; negative control-transfected cells, mimic miR-132; miR-132-transfected cells, *P<0.05). Supplementary material 2 (TIFF 7302 kb) Supplementary material 3 (DOCX 27 kb)

Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Fig. S1 Validation of microarray data of the preliminary cohort by qRT-PCR. (a) qRT-PCR showed that the relative expression of miR-132 was significantly higher in primary CRC lesions without and with liver metastasis than in liver metastatic lesions. (b) There is a significant correlation between microarray data and qRT-PCR expression data. Supplementary material 1 (TIFF 1429 kb) Supplementary Fig. S2 Transfection efficiency data. The data are presented as the mean ± SD. (NC; negative control-transfected cells, mimic miR-132; miR-132-transfected cells, *P<0.05). Supplementary material 2 (TIFF 7302 kb) Supplementary material 3 (DOCX 27 kb) Acknowledgment Supported in part by a Grant-in-Aid for Scientific Research (KAKENHI) to H.Y. (Grant 24390315). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Disclosure Yukako Mokutani, Mamoru Uemura, Koji Munakata, Daisuke Okuzaki, Naotsugu Haraguchi, Hidekazu Takahashi, Junichi Nishimura, Taishi Hata, Kohei Murata, Ichiro Takemasa, Tsunekazu Mizushima, Yuichiro Doki, Masaki Mori and Hirofumi Yamamoto declare no conflict of interest.

Research on tumor nonresponse to therapy has focused on the molecular mechanisms of chemoresistance, with drug distribution being comparatively neglected.1 For cytostatic drugs to be successful, they must fully penetrate the tissue of interest, reaching within all the cancer cells at a concentration sufficient to exert a therapeutic effect. Intraperitoneal tumor dissemination and metastasis is common in several forms of abdominal cancer.2 Numerous studies have investigated the potential role of intraperitoneal drug delivery as an adjunct to systemic chemotherapy in this situation.3 The rationale of intraperitoneal administration is to improve the therapeutic index by increasing the exposure of cancer cells within the peritoneal cavity to the drug while minimizing toxic effects to other organs. Prior studies have documented the limitations of intraperitoneal chemotherapy including the limited direct penetration of drugs into the tumor tissue and the unequal drug distribution throughout the peritoneal cavity.4

cancer cells within the peritoneal cavity to the drug while minimizing toxic effects to other organs. Prior studies have documented the limitations of intraperitoneal chemotherapy including the limited direct penetration of drugs into the tumor tissue and the unequal drug distribution throughout the peritoneal cavity.4 Pressurized intraperitoneal aerosol chemotherapy (PIPAC) is an innovative drug delivery system that takes advantage of the physical properties of the combination of gas and pressure in order to overcome these pharmacologic limitations.5 There is substantial in vitro, in vivo, and ex vivo evidence as well as evidence in human patients that PIPAC has superior pharmacologic properties.6–9 Because the therapeutic ratio between local and systemic drug concentration is increased by PIPAC, enhanced local efficacy together with low systemic toxicity was expected and has been demonstrated clinically.10 Retrospective analysis of the patient cohorts in ovarian gastric and colorectal cancer have shown encouraging results of repeated PIPAC in the palliative situation.11–13 A prospective phase 2 trial with low-dose doxorubicin and cisplatin in recurrent, platin-resistant ovarian cancer applied as a pressurized aerosol showed a clinical benefit rate of 62 % and an objective histologic regression rate of 76 %, coupled with a low incidence of severe adverse events (15 % Common Terminology Criteria for Adverse Events [CTCAE] grade 3, no CTCAE grades 4 or 5).14,15 So far, the role of PIPAC in combination with advanced cytoreductive surgery has not been determined.

ate of 62 % and an objective histologic regression rate of 76 %, coupled with a low incidence of severe adverse events (15 % Common Terminology Criteria for Adverse Events [CTCAE] grade 3, no CTCAE grades 4 or 5).14,15 So far, the role of PIPAC in combination with advanced cytoreductive surgery has not been determined. We hypothesized that electrostatic precipitation may further enhance the pharmacologic properties of PIPAC as so-called electrostatic precipitation pressurized intraperitoneal aerosol chemotherapy (ePIPAC). For electrostatic precipitation, we used a commercially available, CE-certified technology developed for clearing surgical smoke from the operative field of view during laparoscopy (Ultravision, Alesi Surgical Ltd., UK). The performance and safety of Ultravision has been demonstrated in bench studies, preclinical testing, and clinical testing, including a clinical study on 30 patients undergoing laparoscopic cholecystectomy.16,17 In particular, no adverse events such as cardiac arrhythmia, modification of ECG, bowel perforations, or skin burning were reported. The aims of this study were to assess the technical feasibility of ePIPAC, to compare the homogeneity of intraperitoneal distribution between PIPAC and ePIPAC, and to determine possible improvement of tissue uptake after ePIPAC. Materials and Methods Study Design This was an exploratory experimental in vivo study in a large animal model comparing the effect of ePIPAC (3 animals) versus PIPAC (3 animals) versus 1 control animal (ePIPAC, no stain).

The aims of this study were to assess the technical feasibility of ePIPAC, to compare the homogeneity of intraperitoneal distribution between PIPAC and ePIPAC, and to determine possible improvement of tissue uptake after ePIPAC. Materials and Methods Study Design This was an exploratory experimental in vivo study in a large animal model comparing the effect of ePIPAC (3 animals) versus PIPAC (3 animals) versus 1 control animal (ePIPAC, no stain). Animal Model The experiment was performed in compliance with the German Animal Protection Law (TierSchG 2006) and was authorized by the competent authority, State of Thuringia. ARRIVE guidelines were implemented. Seven German nonsyngeneic landrace pigs (5 females, 2 males) weighing 31.5 ± 4.5 kg were operated on by qualified surgeons under the supervision of a veterinarian. The sample size was determined in order to obtain experimental results in triplicate plus a negative control. Animals were randomly assigned to the experimental groups. Procedures were performed under general anesthesia adhering to strict protocols. The animals were euthanized under narcosis at the end of the procedure and immediately necropsied. The primary outcome measure was DT01 concentration in tissue and peritoneal fluid at the end of the procedures (quantitative measurement). Secondary outcomes were the homogeneity of blue staining within the peritoneal cavity (qualitative measurement) and the toluidine blue concentration in the peritoneal fluid at the end of the procedure (semiquantitative measurement).

in tissue and peritoneal fluid at the end of the procedures (quantitative measurement). Secondary outcomes were the homogeneity of blue staining within the peritoneal cavity (qualitative measurement) and the toluidine blue concentration in the peritoneal fluid at the end of the procedure (semiquantitative measurement). Staining Substances DT01 are noncoding small DNA fragments designed to bait and hijack the enzyme complexes that repair DNA double-strand breaks, diverting them away from their primary objective, the double-strand breaks on chromosomes.15,18 DT01 administration with PIPAC has previously been validated.8 In this study, 30 mg toluidine blue and 6 mg Cy5-labeled DT01 were diluted in 1000 ml NaCl 0.9 % solution. A volume of 150 ml solution was administered via an aerosolizer (Capnopen, Capnomed GmbH, Germany) to each animal (n = 6). The negative control animal received 150 ml NaCl 0.9 %.

AC has previously been validated.8 In this study, 30 mg toluidine blue and 6 mg Cy5-labeled DT01 were diluted in 1000 ml NaCl 0.9 % solution. A volume of 150 ml solution was administered via an aerosolizer (Capnopen, Capnomed GmbH, Germany) to each animal (n = 6). The negative control animal received 150 ml NaCl 0.9 %. Experimental Protocol PIPAC was applied as described previously.9 After insufflation of a 12 mm Hg capnoperitoneum with a Veress needle, 2 balloon safety trocars (Kii 5 and 12 mm; Applied Medical, Germany) were inserted into the abdominal wall. The Capnopen was connected to an intravenous high-pressure injector (Arterion 7; Medrad, Germany) and inserted into the abdomen. The pressurized aerosol was applied via aerosolizer and injector. Flow rate was 30 ml/min, and maximal upstream pressure was 200 psi. The therapeutic capnoperitoneum was maintained for 30 min. The aerosol was exsufflated and the trocars removed. Identical conditions were used for the ePIPAC subjects (n = 3), with the additional use of the Ultravision technology. The system was activated at the point of completion of aerosol generation and the electric current was maintained for 30 min. This negative control animal received NaCl 0.9 % through ePIPAC under the above conditions.

ions were used for the ePIPAC subjects (n = 3), with the additional use of the Ultravision technology. The system was activated at the point of completion of aerosol generation and the electric current was maintained for 30 min. This negative control animal received NaCl 0.9 % through ePIPAC under the above conditions. Electrostatic Precipitation The Ultravision system integrates the following components: a generator unit (voltage 7500–9500 V, current ≤10 µA), an active cable terminating in an atraumatic stainless steel brush electrode (Ionwand) that is responsible for the electrostatic charging of aerosol particles, and a return electrode with a solid patient return plate (Fig. 1). The Ionwand emits a stream of electrons, resulting in the creation of negative gas ions. The gas ions collide with particulate matter, passing on the negative charge. The return electrode confers a weak positive charge on the subject, which results in the electrostatic attraction of the negatively charged aerosol particles to the tissue surfaces of the contained space—that is, the peritoneum.Fig. 1 Principle of electrostatic precipitation ePIPAC. a Technical setting for ePIPAC, including high-pressure injector containing therapeutic solution micropump generating pressurized intraperitoneal aerosol, brush electrode for electrostatic loading of therapeutic aerosol, and return electrode (solid plate). b Intraoperative view of abdomen showing micropump producing aerosol and electrode actively loading this aerosol with electrostatic charges, leading to precipitation of aerosol particles

urized intraperitoneal aerosol, brush electrode for electrostatic loading of therapeutic aerosol, and return electrode (solid plate). b Intraoperative view of abdomen showing micropump producing aerosol and electrode actively loading this aerosol with electrostatic charges, leading to precipitation of aerosol particles Analysis Toluidine blue distribution was assessed qualitatively as described previously.7 Immediately after the procedure, peritoneum was sampled via biopsy; peritoneal fluid was sampled, and droplets were distributed onto filter paper to visualize the intensity of the blue stain. The tissue and peritoneal fluid samples were snap-frozen in liquid nitrogen and processed at the Institut Curie/Orsay for blinded analysis. The quantitation of DT01 in the peritoneal fluid was performed by a hybridization enzyme-linked immunosorbent assay (ELISA) using a biotin-conjugated capture oligonucleotide probe (300 μl) and a digoxigenin-conjugated detection probe (300 μl), with sequences complementary to the DT01 sequence (Exiqon, USA) in 96-well plates. Then samples were incubated with an anti-digoxigenin horseradish peroxidase–conjugated antibody (1:10,000; Roche, USA), and detection was performed by the addition of 3,3′,5,5′-tetramethylbenzidine substrate (100 μl). Absorbance was measured at 450 and 560 nm, and quantity of DT01 was calculated from calibration standards over a working range of 25–1000 ng/ml using a 4-parameter logistic curve. Because the ELISA failed to produce reliable quantification in tissues, we used fluorescent quantification, a reliable technique for assessing molecule distribution.19 Peritoneal tissue samples were defrosted, then crushed in phosphate-buffered saline–ethylenediaminetetraacetic acid. Fifty microliters of each extract was formed into aliquots in a 96-well plate, and fluorescence was measured with a Typhoon scanner (GE Healthcare, USA). The quantity of DT01 was calculated from a standard curve performed using peritoneal extract from untreated groups, with Cy5-labeled DT01 concentrations ranging from 0 to 1 µg/ml.

ters of each extract was formed into aliquots in a 96-well plate, and fluorescence was measured with a Typhoon scanner (GE Healthcare, USA). The quantity of DT01 was calculated from a standard curve performed using peritoneal extract from untreated groups, with Cy5-labeled DT01 concentrations ranging from 0 to 1 µg/ml. Statistics Size sample was determined and limited by the decision of the regulatory authority: experiments in triplicate (2 × 3), plus 1 control, for a total of 7 animals. Descriptive statistics including mean and standard deviation are provided. The null hypothesis was that DT01 concentration was equal in the peritoneal fluid and in peritoneal biopsy samples of PIPAC and ePIPAC animals. In spite of the small sample size, an exploratory comparative analysis was performed by a nonparametric test (Kruskal–Wallis test for independent samples). A p value of <0.05 was considered significant. Results There were no technical difficulties or intraoperative complications in any of the cases. In both the PIPAC and ePIPAC groups, rapid nebulization of the toluidine blue solution within the tightly closed abdomen was observed. Videoscopic control showed immediate staining of the complete abdominal cavity in both PIPAC and ePIPAC animals, including all exposed peritoneal surfaces. Intra-abdominal organs were not mobilized.

IPAC and ePIPAC groups, rapid nebulization of the toluidine blue solution within the tightly closed abdomen was observed. Videoscopic control showed immediate staining of the complete abdominal cavity in both PIPAC and ePIPAC animals, including all exposed peritoneal surfaces. Intra-abdominal organs were not mobilized. In the ePIPAC group, electrostatic loading of the saline aerosol was technically feasible without significant aberrant conduction. The maximal allowed current intensity was not reached in the ePIPAC group, denoted by the absence of an alarm signal. After activation of the electrode in the ePIPAC group, the aerosol completely cleared from the field of view within 15 s, as documented by real-time videoendoscopy. In contrast, in the PIPAC animals, aerosol particles remained in suspension until the end of the procedure (after 30 min of steady state).

an alarm signal. After activation of the electrode in the ePIPAC group, the aerosol completely cleared from the field of view within 15 s, as documented by real-time videoendoscopy. In contrast, in the PIPAC animals, aerosol particles remained in suspension until the end of the procedure (after 30 min of steady state). At necropsy, macroscopic stain distribution throughout the entire peritoneal cavity was homogeneous in the PIPAC group, including the small bowel and anterior abdominal wall and hidden surfaces such as the inferior aspect of the liver and the liver hilum. Comparable results were obtained in the ePIPAC group. In particular, no staining gradient toward or from the brush electrode was observed in the ePIPAC animals (Fig. 2). No bowel lesion or perforation was noted.Fig. 2 Adequacy of toluidine blue distribution. Autopsy findings in PIPAC (a1, a2) and ePIPAC (b1, b2) animals after aerosolization of low-dose toluidine blue. Staining of serosal surfaces is homogeneous in both groups. Importantly, inferior aspect of liver, including hilum and gallbladder, are stained After 30 min, peritoneal fluid still demonstrated the blue staining in the PIPAC group (Supplementary material 1, arrows), whereas it was greatly reduced in the ePIPAC group. This qualitative impression was confirmed by the absence of color on filter paper with peritoneal fluid after ePIPAC. In contrast, blue staining of the peritoneal fluid was still present after PIPAC.

blue staining in the PIPAC group (Supplementary material 1, arrows), whereas it was greatly reduced in the ePIPAC group. This qualitative impression was confirmed by the absence of color on filter paper with peritoneal fluid after ePIPAC. In contrast, blue staining of the peritoneal fluid was still present after PIPAC. The concentration of Cy5-labeled DT01 in the peritoneal fluid was lower after ePIPAC compared to PIPAC (Fig. 3a), confirming the results obtained from the toluidine blue assessment. The mean initial DT01 concentration in the aerosolized solution was 8.93 ± 0.72 µg/ml. After PIPAC, this concentration diminished to 1.46 ± 0.21 µg/ml, indirectly documenting a tissue uptake of 83.6 % in this closed system. After ePIPAC, the concentration further dropped to 0.11 ± 0.02 µg/ml, suggesting an almost complete tissue uptake of DT01 (98.7 %). The null hypothesis could therefore be rejected (p = 0.01). Superior DT01 uptake after ePIPAC versus PIPAC was confirmed by tissue measurement (Fig. 3b): DT01 concentration in tissue increased from 0.05 µg/ml (background noise) before therapy to 0.41 ± 0.17 µg/ml after PIPAC application and to 0.57 ± 0.20 µg/ml after ePIPAC application (p = 0.06).Fig. 3 a Peritoneal fluid DT01 concentration showing 15 % remaining concentration after PIPAC compared to initial concentration in aerosolized solution versus 1.5 % after ePIPAC (p = 0.01). Whereas PIPAC allows 85 % tissue uptake, ePIPAC achieves another order of magnitude with 98.5 % absorption. b Tissue DT01 concentration after PIPAC vs. ePIPAC application, confirming superior uptake after ePIPAC (p = 0.06)

C compared to initial concentration in aerosolized solution versus 1.5 % after ePIPAC (p = 0.01). Whereas PIPAC allows 85 % tissue uptake, ePIPAC achieves another order of magnitude with 98.5 % absorption. b Tissue DT01 concentration after PIPAC vs. ePIPAC application, confirming superior uptake after ePIPAC (p = 0.06) Discussion In this experimental study, we combined 3 physical principles (electrostatic precipitation, aerosol nature, and hydrostatic pressure) with the aim of further improving tissue uptake after intraperitoneal delivery, thus developing the concept of ePIPAC.

C compared to initial concentration in aerosolized solution versus 1.5 % after ePIPAC (p = 0.01). Whereas PIPAC allows 85 % tissue uptake, ePIPAC achieves another order of magnitude with 98.5 % absorption. b Tissue DT01 concentration after PIPAC vs. ePIPAC application, confirming superior uptake after ePIPAC (p = 0.06) Discussion In this experimental study, we combined 3 physical principles (electrostatic precipitation, aerosol nature, and hydrostatic pressure) with the aim of further improving tissue uptake after intraperitoneal delivery, thus developing the concept of ePIPAC. Electrostatic loading of a therapeutic aerosol was technically feasible in all ePIPAC animals in an environment highly saturated with saline solution without creating significant erratic electric currents. The aerosol was cleared much more quickly in ePIPAC animals compared to PIPAC animals. No major macroscopic differences were noted in dye distribution; in particular, there was no optical tissue staining gradient toward or from the active electrode in the ePIPAC group. At the end of the procedure, the peritoneal fluid was colorless in the ePIPAC group while the blue color was maintained in the PIPAC group, suggesting a more effective clearance of toluidine blue from the aerosol after electrostatic precipitation. Subsequent semiquantitative analysis with filter paper confirmed this clinical impression. Quantitative results obtained with the second tracer (DT01) demonstrated that transfer of the tracer from the aqueous solution to tissue surfaces was improved after ePIPAC compared to PIPAC. After PIPAC, approximately 1/10 of the DT01 was present within the peritoneal cavity, while with ePIPAC only about 1/102 remained. DT01 is a much larger molecule than toluidine blue, and therefore such efficient uptake was not anticipated. This superior uptake after ePIPAC was confirmed by a higher tissue concentration than after PIPAC.

ximately 1/10 of the DT01 was present within the peritoneal cavity, while with ePIPAC only about 1/102 remained. DT01 is a much larger molecule than toluidine blue, and therefore such efficient uptake was not anticipated. This superior uptake after ePIPAC was confirmed by a higher tissue concentration than after PIPAC. This difference in uptake represents a further improvement in the context of an earlier study in which tissue concentration of doxorubicin after PIPAC was found to be up to 200 times higher than reported after hyperthermic intraperitoneal chemotherapy (HIPEC), with only 10 % of the dose.9 Application of intraperitoneal chemotherapy with ePIPAC may have several potential advantages over existing techniques:First, if it can be shown that increased deposition efficiency translates to an increased tissue uptake, then it may increase drug uptake into tumor nodes and therefore achieve cytotoxic dose in larger nodules. This could in turn reduce the need for aggressive cytoreductive surgery and allow therapy of diffuse small bowel involvement, a contraindication for cytoreductive surgery and HIPEC.20

to an increased tissue uptake, then it may increase drug uptake into tumor nodes and therefore achieve cytotoxic dose in larger nodules. This could in turn reduce the need for aggressive cytoreductive surgery and allow therapy of diffuse small bowel involvement, a contraindication for cytoreductive surgery and HIPEC.20 Second, it may allow a further reduction of the dose needed to be effective. In ovarian and gastric cancer, PIPAC has been shown to be effective with 10 % of the usual systemic dose of cisplatin and doxorubicin.12,14 In colorectal cancer, PIPAC was effective with 20 % of the doses generally administered for HIPEC.13 Dose reduction allows not only a reduction in organ toxicity and systemic side effects but also a limit to local toxicity on the bowel and the normal peritoneum.10,14 The clinical significance of this is the possibility to use PIPAC earlier in the course of the disease as a secondary prevention of peritoneal metastasis, analogous to HIPEC, and may contribute to a lower incidence of peritoneal sclerosis.21–25 Third, ePIPAC may allow a significant reduction of the time needed for application. Finally, it may simplify the occupational safety aspects of PIPAC by reducing time of potential exposure and by minimizing any residual drug evacuated at the end of the procedure.

Second, it may allow a further reduction of the dose needed to be effective. In ovarian and gastric cancer, PIPAC has been shown to be effective with 10 % of the usual systemic dose of cisplatin and doxorubicin.12,14 In colorectal cancer, PIPAC was effective with 20 % of the doses generally administered for HIPEC.13 Dose reduction allows not only a reduction in organ toxicity and systemic side effects but also a limit to local toxicity on the bowel and the normal peritoneum.10,14 The clinical significance of this is the possibility to use PIPAC earlier in the course of the disease as a secondary prevention of peritoneal metastasis, analogous to HIPEC, and may contribute to a lower incidence of peritoneal sclerosis.21–25 Third, ePIPAC may allow a significant reduction of the time needed for application. Finally, it may simplify the occupational safety aspects of PIPAC by reducing time of potential exposure and by minimizing any residual drug evacuated at the end of the procedure. There are several limitations to this early work. These data have been obtained in an experimental model and cannot be extrapolated to human patients without further validation. This study was not performed in an animal model of peritoneal metastasis but in healthy pigs because such a model is not available. Therefore, tracer penetration into tumor tissue could not be assessed.

have been obtained in an experimental model and cannot be extrapolated to human patients without further validation. This study was not performed in an animal model of peritoneal metastasis but in healthy pigs because such a model is not available. Therefore, tracer penetration into tumor tissue could not be assessed. It was not possible to apply toxic agents such as cytotoxic drugs within the experimental operating room as a result of the absence of high-flow ventilation. Although DT01 is a validated marker for determining tissue drug uptake, the results presented are only valid for the substances tested.8,26 It is likely that the ability to improve tissue deposition of drug substances using electrostatic precipitation will be affected by the physical characteristics of these molecules. Thus, it is not possible to extrapolate the clearance of toluidine blue and DT01 to the clearance of chemotherapy drugs. Additional pharmacologic studies are required for each drug used for ePIPAC. Conclusions The therapeutic effect of ePIPAC is through a combination of aerosolization of the drug, applying a pressure across it and application of an electrostatic gradient. ePIPAC is technically feasible and improves tissue uptake of 2 tracer substances compared to PIPAC. ePIPAC has the potential to allow more efficient drug uptake, to permit further dose reduction, to significantly shorten the time required for PIPAC application, and to improve health and safety in the operating room when undertaking such procedures.

and improves tissue uptake of 2 tracer substances compared to PIPAC. ePIPAC has the potential to allow more efficient drug uptake, to permit further dose reduction, to significantly shorten the time required for PIPAC application, and to improve health and safety in the operating room when undertaking such procedures. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary material 1 (TIFF 1524 kb) Tinatin Kakchekeeva and Cedric Demtröder have contributed equally to this article, and both should be considered first author. Acknowledgment Supported in part by institutional funds and by an educational Grant and provision of the Ultravision system by Alesi Surgical Ltd. The authors thank Frank Pölzing, MD, for animal care, and Céline Agrario and Aurélie Herbette for technical assistance. Disclosure MAR has an equity interest in CAPNOMED GmbH, Villingendorf, Germany. JT is a medical advisor and shareholder of Alesi Surgical Ltd., Cardiff, UK. DG is an employee of Alesi Surgical Ltd. CD received research funding from CAPNOMED. NH is an employee of DNA Therapeutics, Evry, France. MD holds an equity interest in DNA Therapeutics. The other authors declare no conflicts of interest.

Adenocarcinoma has long been an independent histological class of lung cancer and has been broadly studied for therapeutic efficacy and toxicities.1–5 In 2011, a new classification system of subtypes of lung adenocarcinoma was recommended by the International Association for the Study of Lung Cancer (IASLC), American Thoracic Society (ATS), and European Respiratory Society (ERS) to further study and advance the field.6 Since then, a number of investigations have provided evidence for the relationship between different subtypes and treatment outcomes.7–11 Several studies have reported that micropapillary- and solid-predominant subtypes of lung adenocarcinoma were associated with poor prognoses;12–16 however, lung adenocarcinomas usually contain complex mixtures of different subtypes.17 Whether minor components of each subtype are associated with lymph node metastasis and poor prognosis still remains unknown and needs further clarification. In this study, we comprehensively analyzed 1244 consecutive patients who were diagnosed with stage I–IV invasive lung adenocarcinoma and who underwent surgical resection between August 2006 and May 2013. Our aim was to provide clinical evidence for the predictive and prognostic value of minor components of lung adenocarcinoma.

study, we comprehensively analyzed 1244 consecutive patients who were diagnosed with stage I–IV invasive lung adenocarcinoma and who underwent surgical resection between August 2006 and May 2013. Our aim was to provide clinical evidence for the predictive and prognostic value of minor components of lung adenocarcinoma. Patients and Methods Patients and Tissues Overall, 1244 consecutive patients who were diagnosed with invasive lung adenocarcinoma and who underwent surgical resection between August 2006 and May 2013 were included in this study. R0 resection was achieved in 1240 of the 1244 patients. Patients with no or insufficient archived tumor specimens were excluded, and no patient underwent neoadjuvant therapy. To ensure an accurate assessment, tumors were reclassified by three pathologists (XS, LS, and YL) and categorized into the following subtypes: lepidic (L), acinar (A), papillary (P), micropapillary (M), and solid (S) predominant subtypes, as well as invasive mucinous adenocarcinoma (IMA), according to the newly announced IASLC/ATS/ERS lung adenocarcinoma classification system.6 Each of the 1244 slides was reviewed by these three pathologists separately. Discordant results were reconsidered together by the three pathologists until a consensus was reached. Specimens that did not belong to any one of these categories were marked as ‘others’. For those specimens that were mixed by more than one subtype, the subtype that occupied most of the tumor (even if <50 %) was defined as the predominant subtype, and subtype(s) that occupied no less than 5 % but were not predominant were defined as minor components. We put them in a sequence from the largest to the smallest amount.