ARTICLE IN PRESS
Long-term outcomes after prophylactic bursectomy in patients with resectable gastric cancer: Final analysis of a multicenter randomized controlled trial Motohiro Hirao, MD,a Yukinori Kurokawa, MD,b Junya Fujita, MD,c Hiroshi Imamura, MD,d Yoshiyuki Fujiwara, MD,e Yutaka Kimura, MD,f Shuji Takiguchi, MD,b Masaki Mori, MD,b Yuichiro Doki, MD,b and Osaka University Clinical Research Group for Gastroenterological Study, Osaka, Japan
Background. Bursectomy, a traditional operative procedure to remove the peritoneal lining covering the pancreas and the anterior plane of the transverse mesocolon, has been performed for serosa-positive gastric cancer in Japan and Eastern Asia. We conducted a multicenter, randomized, controlled trial to demonstrate the noninferiority of the omission of bursectomy. Methods. Between July 2002 and January 2007, 210 patients with cT2-3 gastric adenocarcinoma were randomized intraoperatively to D2 gastrectomy with or without bursectomy. The primary endpoint was overall survival (OS). We provide the results of the final analysis of the complete 5-year follow-up data. Results. After the median follow-up of 80 months, 5-year OS was 77.5% for the bursectomy group and 71.3% for the nonbursectomy group (2-sided P = .16 for superiority; 1-sided P = .99 for noninferiority). The hazard ratio for death in the nonbursectomy group was 1.40 (95% CI, 0.87–2.25). The 5-year recurrence-free survivals were 73.7% and 66.6% in the bursectomy and nonbursectomy groups, respectively (2-sided P = .33 for superiority; 1-sided P = .99 for noninferiority). Cox multivariate analysis revealed that bursectomy was an independent prognostic factor of good OS (P = .033). Subgroup analysis showed a trend toward improved survival after bursectomy for tumors in the middle or lower third of the stomach and for pathologically serosa-positive tumors. Conclusion. The final analysis could not demonstrate the noninferiority of the omission of bursectomy. Bursectomy should not be abandoned as a futile procedure. (Surgery 2015;j:j-j.) From the Department of Surgery,a National Hospital Organization Osaka National Hospital; the Department of Gastroenterological Surgery,b Osaka University Graduate School of Medicine; the Department of Surgery,c NTT West Osaka Hospital; the Department of Surgery,d Toyonaka Municipal Hospital; the Department of Surgery,e Osaka Medical Center for Cancer and Cardiovascular Diseases; and the Department of Surgery,f Sakai Municipal Hospital, Osaka, Japan
SINCE THE 1960S IN JAPAN, bursectomy has been performed in conjunction with radical gastrectomy and extended lymphadenectomy for advanced Conflict of Interest Disclosures: All authors have no conflicts of interest or financial ties to disclose. Funded by the Supporting Center for Clinical Research and Education (SCCRE). Accepted for publication December 9, 2014. Reprint requests: Yukinori Kurokawa, MD, Department of Gastroenterological Surgery, Osaka University Graduate School of Medicine, 2-2-E2, Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2014.12.024
gastric cancers.1,2 Bursectomy consists of surgically removing the anterior membrane of the transverse mesocolon and also the pancreatic capsule with omentectomy. The rationale for this procedure is that en bloc resection of the posterior gastric cavity lining, which may include free cancer cells or micrometastasis, may reduce the incidence of cancer recurrence.3-5 According to the previous version of the Gastric Cancer Treatment Guidelines of the Japanese Gastric Cancer Association, bursectomy was recommended for tumors with serosal invasion,6 so it has been performed for serosapositive gastric cancer in Eastern Asia. However, prophylactic bursectomy has not been performed globally, because several studies reported unclear SURGERY 1
ARTICLE IN PRESS 2 Hirao et al
effects on overall survival in patients with gastric cancer.7-9 Our multicenter, randomized, controlled trial to evaluate prophylactic bursectomy in patients with resectable gastric cancer was launched in July 2002. Although not planned initially, an interim analysis was proposed by the study’s steering committee to examine survival and provide early release of the results, because the time to final and definitive analysis (5 years) was very long. After meeting the criteria of Korn et al10 for preliminary data release in randomized clinical trials of noninferiority, we reported the results of the interim analysis performed in August 2010.11 The current report presents the results of the final analysis, performed in November 2013. METHODS Patients. Patients eligibility criteria for the study included (1) histologically proven primary adenocarcinoma of the stomach, (2) preoperative and intraoperative classification of T2 (MP/SS) N0, T3 (SE) N0, T2 (MP/SS) N1, or T3 (SE) N1 according to the 13th edition of Japanese classification of gastric carcinoma (MP, tumor invades the muscularis propria; SS, tumor invades the subserosa; SE, tumor penetrates the serosa),12 (3) lack of noncurative surgical factors except for positive lavage cytology, (4) no Borrmann type 4 (linitis plastica) cases, (5) no prior chemotherapy or radiation therapy, (6) no history of gastrectomy or other malignant diseases during the last 5 years, (7) age 20–80 years, (8) performance status of 0–2 according to the Eastern Cooperative Oncology Group scale, and (9) written informed consent. Patients were enrolled from 11 institutions belonging to the Osaka University Clinical Research Group for Gastroenterological Surgery. More than 50 gastrectomies were performed every year in these 11 hospitals. All patients gave written, informed consent before undergoing randomization. The institutional review boards of all participating institutions approved the study protocol. Randomization. The surgeons confirmed the eligibility criteria during surgery and immediately phoned the data center to receive a randomly generated assignment. Patients were randomized intraoperatively to either the bursectomy group (a D2 gastrectomy with bursectomy) or the nonbursectomy group (a D2 gastrectomy without bursectomy), using the minimization method, according to sex, clinical T stage (cT2 vs cT3), and gastrectomy (total vs distal gastrectomy). The data center of the Osaka University Clinical Research Group for Gastroenterological Surgery
Surgery j 2015
was located in Osaka University. The data center independently performed data management, central monitoring, and statistical analysis. None of the surgeons who performed the operations were involved in data analysis. Survival analyses were conducted only twice (August 2010 for interim analysis and November 2013 for final analysis) to avoid multiplicity. Operative procedures. In both groups, the surgeons performed a total or distal gastrectomy with a D2 lymph node dissection as a standard operation. In total gastrectomy for T2 or deeper tumors in the proximal one-third of the stomach, the spleen was removed for splenic hilar lymphadenectomy in principle. Pancreatectomy was confined to those patients whose pancreas was involved by the primary tumor. The prophylactic cholecystectomy and the type of reconstruction after gastrectomy were not specified in the protocol. The details of the bursectomy operative procedure were described previously.13 In brief, the bursectomy group required the en bloc resection of the peritoneal lining of the bursa omentalis, to as great an extent as possible, from the anterior plane of the transverse mesocolon and the pancreas. On the left side of the transverse mesocolon, the peritoneum was removed up to the left gastroepiploic artery. For distal gastrectomy cases, the pancreatic capsule was removed up to the proximal one-half of the splenic artery. In the nonbursectomy group, only a minimal amount of peritoneum could be removed for lymph node dissection. Omentectomy was mandatory for both groups. All operations were performed or supervised by senior surgeons who were members of the Japanese Gastric Cancer Association. Statistical analysis. The primary endpoint was overall survival (OS), defined as the time from randomization to death. Secondary endpoints were (1) recurrence-free survival (RFS), (2) operative morbidity and mortality, and (3) levels of amylase in drainage liuid on postoperative day 1. RFS was defined as the time from randomization to either the first recurrence or death from any cause. Operative methods and pathology results were recorded according to the 13th edition of Japanese classification of gastric carcinoma.12 Hospital mortality was defined as postoperative death from any cause within 30 days, or death during the same hospitalization as the operation. Patients were followed every 3 months with blood examination including tumor markers and every 6 months with image examination for 5 years postoperatively.
ARTICLE IN PRESS Surgery Volume j, Number j
We initially planned to recruit 200 patients, with an error of 0.1 and statistical power of 80%. This would allow us to detect a hazard ratio (HR) with a noninferiority margin of 1.56 in the nonbursectomy group and an estimated 60% 5-year OS in the bursectomy group. The projected accrual period and follow-up period were 3 and 5 years, respectively. After the registration of 204 patients, we amended the sample size and analysis to reduce errors and to achieve a revised estimated 5-year OS in the bursectomy group of 75%. The amended sample size was 464, with an error of 0.05, statistical power of 80%, and HR noninferiority margin of 1.50, with an 8-year accrual period (in total) and 5year follow-up. In January 2007, a large-scale, randomized, controlled trial evaluating the efficacy of adjuvant S-1 chemotherapy for stage II/III gastric cancer patients reported positive results. Since then, adjuvant S-1 chemotherapy has been the new standard treatment for stage II/III gastric cancer in Japan. Because our study did not permit adjuvant treatment, including S-1 chemotherapy, we decided to close accrual of our study in January 2007. Differences in proportions between the 2 groups were evaluated using Fisher’s exact test or the Chi-square test. Differences in continuous variables, including age and tumor size, between the 2 groups were tested with the Mann–Whitney U test. Data from all eligible patients were analyzed for OS and RFS on an intention-to-treat basis. Survival curves were estimated by the Kaplan–Meier method and compared using the log-rank test. HRs were calculated by Cox regression analysis without adjustment for stratification factors. Twosided P values were used for testing superiority, and 1-sided P values were used for testing noninferiority. All noninferiority tests were conducted using the handicap log-rank test with the HR noninferiority margin set at 1.50. All P values were reported as statistically significant if P < .05, to provide conventional interpretation of results. Statistical analysis was performed using SPSS Statistics software, version 20.0 (SPSS, Chicago, IL) and the R programming language. RESULTS From July 2002 to January 2007, 210 patients were randomized to either bursectomy (104 patients) or nonbursectomy (106 patients). Although 1 patient in the bursectomy group and 1 in the nonbursectomy group did not undergo allocated surgery, we analyzed based on the intent-to-treat principle (Fig 1). Most patient characteristics were well balanced between the 2 groups (Table I). The
Hirao et al 3
Fig 1. CONSORT flow chart.
bursectomy group showed trends toward older patients (P = .060) and more patients with pathologically positive nodes (P = .12) than the nonbursectomy group. Only 7 patients (3.3%) underwent R1 resection. The reasons for R1 resection were positive lavage cytology in all 6 patients (3 patients in each group), and positive proximal margin in 1 nonbursectomy patient. As reported previously, bursectomy required a longer operative time, with a median added time of 27 minutes in patients with combined resection and 26 minutes in patients without combined resection.13 Intraoperative blood loss was also greater in the bursectomy group (median, 475 mL) than in the nonbursectomy group (median, 350 mL; P = .047); other surgical factors did not vary significantly. The overall morbidity rate was 14.3% for both groups. The median amylase levels in drainage fluid on postoperative day 1 were similar in the 2 groups (P = .54). The hospital mortality rate was 0.95%, with 1 patient death in each group. At the time of the final analysis in November 2013, the median patient follow-up was 80 months. There had been 30 deaths in the bursectomy group and 40 deaths in the nonbursectomy group; 21 bursectomy patients and 30 nonbursectomy patients died from gastric cancer. The 5-year OS was 77.5% for the bursectomy group and 71.3% for the nonbursectomy group (2-sided P = .16 for superiority; 1-sided P = .99 for noninferiority; Fig 2, A). The HR for death in the nonbursectomy group was 1.40 (95% CI, 0.87–2.25). To adjust for possible confounding baseline factors, Cox multivariate analysis was performed (Table II). The bursectomy procedure was an independent prognostic factor for good OS, with statistical significance (P = .034). When we additionally conducted the Cox multivariate analysis using pathological T/N stages instead of clinical T/N stages, this sensitivity analysis also showed significance for the bursectomy procedure (P = .048).
ARTICLE IN PRESS 4 Hirao et al
Surgery j 2015
Table I. Patient characteristics
Characteristic
Bursectomy group (n = 104)
Nonbursectomy group P (n = 106) value
Age (y), 65 (31–80) 64 (34–79) median (range) Sex Male 73 77 Female 31 29 Tumor size (cm), 4.3 (0.9–11.0) 4.5 (1.5–12.0) median (range) Longitudinal tumor location U 15 16 M/L 89 90 Circumferential tumor location Posterior 19 15 Not posterior 85 91 Gastrectomy Distal 82 79 Total 22 27 Histology Differentiated 47 50 Undifferentiated* 57 56 cT stage T2 (MP/SS) 61 67 T3 (SE) 43 39 cN stage N0 59 61 N1-3 45 45 pT stage T1 (M/SM) 17 19 T2 (MP/SS) 62 64 T3-4 (SE/SI) 25 23 pN stage N0 49 60 N1 37 24 N2-3 18 22 Residual tumor R0 101 102 R1 3 4
.060 .69
.31 .89
.42
.46
.77
.50
.91
.90
.12
.72
*Undifferentiated type included 1 case of neuroendocrine carcinoma in the nonbursectomy group. cT, cN, pT, and pN stages were according to the 13th edition of Japanese classification of gastric carcinoma. M, Tumor confined to the mucosa; M/L, middle or lower third of the stomach; MP, tumor invades the muscularis propria; SE, tumor penetrates the serosa; SI, tumor invades adjacent structures; SM, tumor confined to the submucosa; SS, tumor invaded the subserosa; U, upper third of the stomach.
The 5-year RFS were 73.7% and 66.6% in the bursectomy and nonbursectomy groups, respectively (2-sided P = .33 for superiority; 1-sided P = .99 for noninferiority; Fig 2, B). The HR for recurrence in the nonbursectomy group was 1.25 (95% CI, 0.80–1.97). The most frequent site of the first tumor recurrence was the peritoneum,
as seen in 9 patients in the bursectomy group (8.7%) and 15 patients in the nonbursectomy group (14.2%). The proportions of other recurrence sites---nodal and hepatic---were similar in the 2 groups. We also assessed statistical interactions between the treatment effects and baseline factors of age, sex, longitudinal tumor location, circumferential tumor location, histologic type, cT, and cN (Fig 3). Although there were no interactions between treatment effect and any patient characteristics, there was a trend toward interaction between treatment effect and longitudinal tumor location. Among the 31 patients with tumors in the upper onethird of the stomach, 5-year OS was 57.8% for the bursectomy group versus 75.0% for the nonbursectomy group (P = .23; HR = 0.50; 95% CI, 0.16–1.60). In contrast, among the 179 patients with tumors in the middle or lower third of the stomach, OS in the bursectomy group was significantly better than that in the nonbursectomy group (P = .048; HR = 1.69; 95% CI, 1.00–2.86). The 5-year OS was 80.7% for the bursectomy group and 70.7% for the nonbursectomy group. We also performed a subgroup analysis regarding pT stage. Among the 162 pT1 or T2 (MP/SS) patients, 5-year OS were 84.5% and 81.6% for the bursectomy and nonbursectomy groups, respectively (P = .36; Fig 4, A). The HR in the nonbursectomy group was 1.34 (95% CI, 0.72–2.52). In contrast, among the 48 pT3 (SE) or T4 patients, 5-year OS was 55.5% for the bursectomy group, in comparison to 34.8% for the nonbursectomy group (P = .096; Fig 4, B). The HR in the nonbursectomy group was 1.84 (95% CI, 0.89–3.79). DISCUSSION In the present study, we compared the longterm survival of patients with gastric cancer who did or did not undergo prophylactic bursectomy. The final analysis could not demonstrate the noninferiority of the omission of bursectomy. The bursectomy group had a trend toward improved survival. Although these differences were not statistically significant, Cox multivariate analysis revealed that bursectomy was a significant prognostic factor for good OS. Experienced surgeons in our study group safely performed D2 gastrectomy with bursectomy without increasing major surgical complications, despite longer operation time and increased intraoperative blood loss. Recently, it was also reported that D2 gastrectomy with additional bursectomy could be performed in Western patients.14
ARTICLE IN PRESS Hirao et al 5
Surgery Volume j, Number j
Fig 2. Kaplan–Meier overall survival (A) and recurrence-free survival (B) in all randomized patients.
Table II. Cox multivariate analysis of overall survival Variable
Category
Hazard ratio (95% CI)
Age Sex Longitudinal tumor location Circumferential tumor location Histology cT stage cN stage Procedure
$65 y Male U Not posterior Undifferentiated* T3 (SE) N1-3 Nonbursectomy
1.44 1.88 1.07 1.34 1.83 3.52 0.97 1.71
(0.88–2.36) (1.06–3.34) (0.56–2.04) (0.64–2.81) (1.10–3.04) (2.12–5.84) (0.59–1.59) (1.04–2.82)
P value .15 .032 .83 .44 .021
Long-term outcomes after prophylactic bursectomy in patients with resectable gastric cancer: Final analysis of a multicenter randomized controlled trial Motohiro Hirao, MD,a Yukinori Kurokawa, MD,b Junya Fujita, MD,c Hiroshi Imamura, MD,d Yoshiyuki Fujiwara, MD,e Yutaka Kimura, MD,f Shuji Takiguchi, MD,b Masaki Mori, MD,b Yuichiro Doki, MD,b and Osaka University Clinical Research Group for Gastroenterological Study, Osaka, Japan
Background. Bursectomy, a traditional operative procedure to remove the peritoneal lining covering the pancreas and the anterior plane of the transverse mesocolon, has been performed for serosa-positive gastric cancer in Japan and Eastern Asia. We conducted a multicenter, randomized, controlled trial to demonstrate the noninferiority of the omission of bursectomy. Methods. Between July 2002 and January 2007, 210 patients with cT2-3 gastric adenocarcinoma were randomized intraoperatively to D2 gastrectomy with or without bursectomy. The primary endpoint was overall survival (OS). We provide the results of the final analysis of the complete 5-year follow-up data. Results. After the median follow-up of 80 months, 5-year OS was 77.5% for the bursectomy group and 71.3% for the nonbursectomy group (2-sided P = .16 for superiority; 1-sided P = .99 for noninferiority). The hazard ratio for death in the nonbursectomy group was 1.40 (95% CI, 0.87–2.25). The 5-year recurrence-free survivals were 73.7% and 66.6% in the bursectomy and nonbursectomy groups, respectively (2-sided P = .33 for superiority; 1-sided P = .99 for noninferiority). Cox multivariate analysis revealed that bursectomy was an independent prognostic factor of good OS (P = .033). Subgroup analysis showed a trend toward improved survival after bursectomy for tumors in the middle or lower third of the stomach and for pathologically serosa-positive tumors. Conclusion. The final analysis could not demonstrate the noninferiority of the omission of bursectomy. Bursectomy should not be abandoned as a futile procedure. (Surgery 2015;j:j-j.) From the Department of Surgery,a National Hospital Organization Osaka National Hospital; the Department of Gastroenterological Surgery,b Osaka University Graduate School of Medicine; the Department of Surgery,c NTT West Osaka Hospital; the Department of Surgery,d Toyonaka Municipal Hospital; the Department of Surgery,e Osaka Medical Center for Cancer and Cardiovascular Diseases; and the Department of Surgery,f Sakai Municipal Hospital, Osaka, Japan
SINCE THE 1960S IN JAPAN, bursectomy has been performed in conjunction with radical gastrectomy and extended lymphadenectomy for advanced Conflict of Interest Disclosures: All authors have no conflicts of interest or financial ties to disclose. Funded by the Supporting Center for Clinical Research and Education (SCCRE). Accepted for publication December 9, 2014. Reprint requests: Yukinori Kurokawa, MD, Department of Gastroenterological Surgery, Osaka University Graduate School of Medicine, 2-2-E2, Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2014.12.024
gastric cancers.1,2 Bursectomy consists of surgically removing the anterior membrane of the transverse mesocolon and also the pancreatic capsule with omentectomy. The rationale for this procedure is that en bloc resection of the posterior gastric cavity lining, which may include free cancer cells or micrometastasis, may reduce the incidence of cancer recurrence.3-5 According to the previous version of the Gastric Cancer Treatment Guidelines of the Japanese Gastric Cancer Association, bursectomy was recommended for tumors with serosal invasion,6 so it has been performed for serosapositive gastric cancer in Eastern Asia. However, prophylactic bursectomy has not been performed globally, because several studies reported unclear SURGERY 1
ARTICLE IN PRESS 2 Hirao et al
effects on overall survival in patients with gastric cancer.7-9 Our multicenter, randomized, controlled trial to evaluate prophylactic bursectomy in patients with resectable gastric cancer was launched in July 2002. Although not planned initially, an interim analysis was proposed by the study’s steering committee to examine survival and provide early release of the results, because the time to final and definitive analysis (5 years) was very long. After meeting the criteria of Korn et al10 for preliminary data release in randomized clinical trials of noninferiority, we reported the results of the interim analysis performed in August 2010.11 The current report presents the results of the final analysis, performed in November 2013. METHODS Patients. Patients eligibility criteria for the study included (1) histologically proven primary adenocarcinoma of the stomach, (2) preoperative and intraoperative classification of T2 (MP/SS) N0, T3 (SE) N0, T2 (MP/SS) N1, or T3 (SE) N1 according to the 13th edition of Japanese classification of gastric carcinoma (MP, tumor invades the muscularis propria; SS, tumor invades the subserosa; SE, tumor penetrates the serosa),12 (3) lack of noncurative surgical factors except for positive lavage cytology, (4) no Borrmann type 4 (linitis plastica) cases, (5) no prior chemotherapy or radiation therapy, (6) no history of gastrectomy or other malignant diseases during the last 5 years, (7) age 20–80 years, (8) performance status of 0–2 according to the Eastern Cooperative Oncology Group scale, and (9) written informed consent. Patients were enrolled from 11 institutions belonging to the Osaka University Clinical Research Group for Gastroenterological Surgery. More than 50 gastrectomies were performed every year in these 11 hospitals. All patients gave written, informed consent before undergoing randomization. The institutional review boards of all participating institutions approved the study protocol. Randomization. The surgeons confirmed the eligibility criteria during surgery and immediately phoned the data center to receive a randomly generated assignment. Patients were randomized intraoperatively to either the bursectomy group (a D2 gastrectomy with bursectomy) or the nonbursectomy group (a D2 gastrectomy without bursectomy), using the minimization method, according to sex, clinical T stage (cT2 vs cT3), and gastrectomy (total vs distal gastrectomy). The data center of the Osaka University Clinical Research Group for Gastroenterological Surgery
Surgery j 2015
was located in Osaka University. The data center independently performed data management, central monitoring, and statistical analysis. None of the surgeons who performed the operations were involved in data analysis. Survival analyses were conducted only twice (August 2010 for interim analysis and November 2013 for final analysis) to avoid multiplicity. Operative procedures. In both groups, the surgeons performed a total or distal gastrectomy with a D2 lymph node dissection as a standard operation. In total gastrectomy for T2 or deeper tumors in the proximal one-third of the stomach, the spleen was removed for splenic hilar lymphadenectomy in principle. Pancreatectomy was confined to those patients whose pancreas was involved by the primary tumor. The prophylactic cholecystectomy and the type of reconstruction after gastrectomy were not specified in the protocol. The details of the bursectomy operative procedure were described previously.13 In brief, the bursectomy group required the en bloc resection of the peritoneal lining of the bursa omentalis, to as great an extent as possible, from the anterior plane of the transverse mesocolon and the pancreas. On the left side of the transverse mesocolon, the peritoneum was removed up to the left gastroepiploic artery. For distal gastrectomy cases, the pancreatic capsule was removed up to the proximal one-half of the splenic artery. In the nonbursectomy group, only a minimal amount of peritoneum could be removed for lymph node dissection. Omentectomy was mandatory for both groups. All operations were performed or supervised by senior surgeons who were members of the Japanese Gastric Cancer Association. Statistical analysis. The primary endpoint was overall survival (OS), defined as the time from randomization to death. Secondary endpoints were (1) recurrence-free survival (RFS), (2) operative morbidity and mortality, and (3) levels of amylase in drainage liuid on postoperative day 1. RFS was defined as the time from randomization to either the first recurrence or death from any cause. Operative methods and pathology results were recorded according to the 13th edition of Japanese classification of gastric carcinoma.12 Hospital mortality was defined as postoperative death from any cause within 30 days, or death during the same hospitalization as the operation. Patients were followed every 3 months with blood examination including tumor markers and every 6 months with image examination for 5 years postoperatively.
ARTICLE IN PRESS Surgery Volume j, Number j
We initially planned to recruit 200 patients, with an error of 0.1 and statistical power of 80%. This would allow us to detect a hazard ratio (HR) with a noninferiority margin of 1.56 in the nonbursectomy group and an estimated 60% 5-year OS in the bursectomy group. The projected accrual period and follow-up period were 3 and 5 years, respectively. After the registration of 204 patients, we amended the sample size and analysis to reduce errors and to achieve a revised estimated 5-year OS in the bursectomy group of 75%. The amended sample size was 464, with an error of 0.05, statistical power of 80%, and HR noninferiority margin of 1.50, with an 8-year accrual period (in total) and 5year follow-up. In January 2007, a large-scale, randomized, controlled trial evaluating the efficacy of adjuvant S-1 chemotherapy for stage II/III gastric cancer patients reported positive results. Since then, adjuvant S-1 chemotherapy has been the new standard treatment for stage II/III gastric cancer in Japan. Because our study did not permit adjuvant treatment, including S-1 chemotherapy, we decided to close accrual of our study in January 2007. Differences in proportions between the 2 groups were evaluated using Fisher’s exact test or the Chi-square test. Differences in continuous variables, including age and tumor size, between the 2 groups were tested with the Mann–Whitney U test. Data from all eligible patients were analyzed for OS and RFS on an intention-to-treat basis. Survival curves were estimated by the Kaplan–Meier method and compared using the log-rank test. HRs were calculated by Cox regression analysis without adjustment for stratification factors. Twosided P values were used for testing superiority, and 1-sided P values were used for testing noninferiority. All noninferiority tests were conducted using the handicap log-rank test with the HR noninferiority margin set at 1.50. All P values were reported as statistically significant if P < .05, to provide conventional interpretation of results. Statistical analysis was performed using SPSS Statistics software, version 20.0 (SPSS, Chicago, IL) and the R programming language. RESULTS From July 2002 to January 2007, 210 patients were randomized to either bursectomy (104 patients) or nonbursectomy (106 patients). Although 1 patient in the bursectomy group and 1 in the nonbursectomy group did not undergo allocated surgery, we analyzed based on the intent-to-treat principle (Fig 1). Most patient characteristics were well balanced between the 2 groups (Table I). The
Hirao et al 3
Fig 1. CONSORT flow chart.
bursectomy group showed trends toward older patients (P = .060) and more patients with pathologically positive nodes (P = .12) than the nonbursectomy group. Only 7 patients (3.3%) underwent R1 resection. The reasons for R1 resection were positive lavage cytology in all 6 patients (3 patients in each group), and positive proximal margin in 1 nonbursectomy patient. As reported previously, bursectomy required a longer operative time, with a median added time of 27 minutes in patients with combined resection and 26 minutes in patients without combined resection.13 Intraoperative blood loss was also greater in the bursectomy group (median, 475 mL) than in the nonbursectomy group (median, 350 mL; P = .047); other surgical factors did not vary significantly. The overall morbidity rate was 14.3% for both groups. The median amylase levels in drainage fluid on postoperative day 1 were similar in the 2 groups (P = .54). The hospital mortality rate was 0.95%, with 1 patient death in each group. At the time of the final analysis in November 2013, the median patient follow-up was 80 months. There had been 30 deaths in the bursectomy group and 40 deaths in the nonbursectomy group; 21 bursectomy patients and 30 nonbursectomy patients died from gastric cancer. The 5-year OS was 77.5% for the bursectomy group and 71.3% for the nonbursectomy group (2-sided P = .16 for superiority; 1-sided P = .99 for noninferiority; Fig 2, A). The HR for death in the nonbursectomy group was 1.40 (95% CI, 0.87–2.25). To adjust for possible confounding baseline factors, Cox multivariate analysis was performed (Table II). The bursectomy procedure was an independent prognostic factor for good OS, with statistical significance (P = .034). When we additionally conducted the Cox multivariate analysis using pathological T/N stages instead of clinical T/N stages, this sensitivity analysis also showed significance for the bursectomy procedure (P = .048).
ARTICLE IN PRESS 4 Hirao et al
Surgery j 2015
Table I. Patient characteristics
Characteristic
Bursectomy group (n = 104)
Nonbursectomy group P (n = 106) value
Age (y), 65 (31–80) 64 (34–79) median (range) Sex Male 73 77 Female 31 29 Tumor size (cm), 4.3 (0.9–11.0) 4.5 (1.5–12.0) median (range) Longitudinal tumor location U 15 16 M/L 89 90 Circumferential tumor location Posterior 19 15 Not posterior 85 91 Gastrectomy Distal 82 79 Total 22 27 Histology Differentiated 47 50 Undifferentiated* 57 56 cT stage T2 (MP/SS) 61 67 T3 (SE) 43 39 cN stage N0 59 61 N1-3 45 45 pT stage T1 (M/SM) 17 19 T2 (MP/SS) 62 64 T3-4 (SE/SI) 25 23 pN stage N0 49 60 N1 37 24 N2-3 18 22 Residual tumor R0 101 102 R1 3 4
.060 .69
.31 .89
.42
.46
.77
.50
.91
.90
.12
.72
*Undifferentiated type included 1 case of neuroendocrine carcinoma in the nonbursectomy group. cT, cN, pT, and pN stages were according to the 13th edition of Japanese classification of gastric carcinoma. M, Tumor confined to the mucosa; M/L, middle or lower third of the stomach; MP, tumor invades the muscularis propria; SE, tumor penetrates the serosa; SI, tumor invades adjacent structures; SM, tumor confined to the submucosa; SS, tumor invaded the subserosa; U, upper third of the stomach.
The 5-year RFS were 73.7% and 66.6% in the bursectomy and nonbursectomy groups, respectively (2-sided P = .33 for superiority; 1-sided P = .99 for noninferiority; Fig 2, B). The HR for recurrence in the nonbursectomy group was 1.25 (95% CI, 0.80–1.97). The most frequent site of the first tumor recurrence was the peritoneum,
as seen in 9 patients in the bursectomy group (8.7%) and 15 patients in the nonbursectomy group (14.2%). The proportions of other recurrence sites---nodal and hepatic---were similar in the 2 groups. We also assessed statistical interactions between the treatment effects and baseline factors of age, sex, longitudinal tumor location, circumferential tumor location, histologic type, cT, and cN (Fig 3). Although there were no interactions between treatment effect and any patient characteristics, there was a trend toward interaction between treatment effect and longitudinal tumor location. Among the 31 patients with tumors in the upper onethird of the stomach, 5-year OS was 57.8% for the bursectomy group versus 75.0% for the nonbursectomy group (P = .23; HR = 0.50; 95% CI, 0.16–1.60). In contrast, among the 179 patients with tumors in the middle or lower third of the stomach, OS in the bursectomy group was significantly better than that in the nonbursectomy group (P = .048; HR = 1.69; 95% CI, 1.00–2.86). The 5-year OS was 80.7% for the bursectomy group and 70.7% for the nonbursectomy group. We also performed a subgroup analysis regarding pT stage. Among the 162 pT1 or T2 (MP/SS) patients, 5-year OS were 84.5% and 81.6% for the bursectomy and nonbursectomy groups, respectively (P = .36; Fig 4, A). The HR in the nonbursectomy group was 1.34 (95% CI, 0.72–2.52). In contrast, among the 48 pT3 (SE) or T4 patients, 5-year OS was 55.5% for the bursectomy group, in comparison to 34.8% for the nonbursectomy group (P = .096; Fig 4, B). The HR in the nonbursectomy group was 1.84 (95% CI, 0.89–3.79). DISCUSSION In the present study, we compared the longterm survival of patients with gastric cancer who did or did not undergo prophylactic bursectomy. The final analysis could not demonstrate the noninferiority of the omission of bursectomy. The bursectomy group had a trend toward improved survival. Although these differences were not statistically significant, Cox multivariate analysis revealed that bursectomy was a significant prognostic factor for good OS. Experienced surgeons in our study group safely performed D2 gastrectomy with bursectomy without increasing major surgical complications, despite longer operation time and increased intraoperative blood loss. Recently, it was also reported that D2 gastrectomy with additional bursectomy could be performed in Western patients.14
ARTICLE IN PRESS Hirao et al 5
Surgery Volume j, Number j
Fig 2. Kaplan–Meier overall survival (A) and recurrence-free survival (B) in all randomized patients.
Table II. Cox multivariate analysis of overall survival Variable
Category
Hazard ratio (95% CI)
Age Sex Longitudinal tumor location Circumferential tumor location Histology cT stage cN stage Procedure
$65 y Male U Not posterior Undifferentiated* T3 (SE) N1-3 Nonbursectomy
1.44 1.88 1.07 1.34 1.83 3.52 0.97 1.71
(0.88–2.36) (1.06–3.34) (0.56–2.04) (0.64–2.81) (1.10–3.04) (2.12–5.84) (0.59–1.59) (1.04–2.82)
P value .15 .032 .83 .44 .021
