Comparison of 5‐year postoperative outcomes after Billroth I and Roux‐en‐Y reconstruction following distal gastrectomy for gastric cancer: Results from a multi‐institutional randomized controlled trial

Abstract Aim We previously reported in a randomized controlled trial that Billroth I and Roux‐en‐Y reconstructions were generally equivalent regarding body weight change and nutritional status 1 year after distal gastrectomy for gastric cancer. We describe the long‐term follow‐up data 5 years after distal gastrectomy. Methods We analyzed consecutive gastric cancer patients who were randomly assigned to undergo Billroth I or Roux‐en‐Y reconstruction after distal gastrectomy. We evaluated body weight change, nutritional status, late complications, quality of life (QOL) using the European Organization for Research and Treatment of Cancer Core QOL Questionnaire, and dysfunction using the Dysfunction After Upper Gastrointestinal Surgery for Cancer, 5 years after surgery. Results A total of 228 patients (Billroth I = 105; Roux‐en‐Y = 123) were eligible for efficacy analyses in this study. Body weight loss 5 years after surgery did not differ significantly between the Billroth I and Roux‐en‐Y groups (10.0% ± 7.9% and 9.6% ± 8.4%, respectively; P = .70). There were no significant differences in other aspects of nutritional status between the two groups. Reflux esophagitis occurred in 19.0% of the patients in the Billroth I group vs 4.9% in the Roux‐en‐Y group (P = .002). Regarding QOL, Billroth I was significantly inferior to Roux‐en‐Y on the diarrhea scale (Billroth I: 28.6, Roux‐en‐Y: 16.0; P = .047). Regarding dysfunction, no score differed significantly between the two groups. Conclusions Billroth I and Roux‐en‐Y reconstructions were generally equivalent regarding body weight change, nutritional status, and QOL 5 years after distal gastrectomy, although Roux‐en‐Y more effectively prevented reflux esophagitis and diarrhea.


| INTRODUC TI ON
Gastrectomy for gastric cancer is one of the most common gastroenterological operations in Japan. 1 Among gastrectomy techniques, distal gastrectomy is most frequently performed, and Billroth I (BI) or Roux-en-Y (RY) are the main reconstruction methods after distal gastrectomy. 2,3 In BI reconstruction, food flows physiologically through the esophagus, stomach, and duodenum, but duodenal juice flows backward into the residual stomach because pyloric function is lost. 4,5 Conversely, in RY reconstruction, although there is no reflux of duodenal juice, Roux stasis syndrome is a possible complication. [6][7][8] Each of these reconstruction methods has advantages and disadvantages, and the evaluation differs depending on the items being examined, such as body weight loss, nutritional status, gastritis in the remnant stomach, reflux esophagitis, and quality of life (QOL). [8][9][10][11][12][13] There is no clear answer to the question of which reconstruction method is better. Currently, the reconstruction method after distal gastrectomy is chosen according to the policies at each facility and the preference of the operator except in cases where the remnant stomach is very small or patients are at high risk. 3 We previously reported finding no difference in weight loss, nutritional status, and QOL 1 year after surgery, but BI was often associated with reflux symptoms associated with gastritis in the remnant stomach and reflux esophagitis, as shown in randomized controlled trials evaluating BI and RY. 14,15 However, many studies comparing reconstruction methods were retrospective studies, and only a small number were randomized studies. To our knowledge, there have been no prospective long-term comparisons of BI and RY with large samples. The aim of this study was to compare BI or RY as a reconstruction method after distal gastrectomy to evaluate longterm changes in body weight and nutritional status, dysfunction, and QOL, 5 years after surgery.

| Subjects
We examined patients registered in a phase II randomized controlled trial (RCT) who underwent distal gastrectomy for gastric adenocarcinoma at the participating hospitals. This trial was a multi-institutional RCT designed to compare the clinical effects of BI or RY reconstructive operations for gastric cancer resection. 14,15 The primary endpoint in the trial was postoperative body weight loss 1 year after surgery and the secondary endpoint was surgical morbidity. We also evaluated items related to nutritional status, such as serum albumin and lymphocyte count, as well as endoscopic examination findings of the remnant stomach and esophagus, postoperative QOL 1 year after surgery, and long-term outcomes 5 years after surgery. Long-term outcomes were set as a secondary endpoint in the protocol, which was amended during the case enrollment period. In the original study, we hypothesized that, compared with BI, RY may lead to 5% less body weight loss 1 year after surgery. The current study was conducted in consecutive patients recruited in the original trial. In this study, we provide the results of the final analysis of the 5-year follow-up data describing body weight loss, nutritional status, late complications, QOL, and dysfunction, 5 years after surgery.
Patients who required distal gastrectomy for gastric cancer with BI or RY reconstruction were eligible for this study. In other words, the eligibility criteria were that the tumor was localized in the middle or lower third of the stomach, with the expectation that a third of the stomach would remain after resection, and that the stomach could be reconstructed by either BI or RY. The outline of the trial is described below. 14,15 Patient eligibility criteria for the study were: Billroth I reconstruction, body weight, distal gastrectomy, gastric cancer, Roux-en-Y reconstruction Registration number: UMIN000000878

| Randomization in the original trial
During surgery, surgeons confirmed that the eligibility criteria were met and that both reconstruction procedures could be chosen after distal gastrectomy, considering the length of the residual stomach before intraoperative randomization. After that, they immediately phoned the data center to receive a randomly generated assignment.
Patients were intraoperatively randomized to either the BI group or the RY group. Randomization was performed using a minimization method according to body mass index and institutional preferences.
Of the 332 patients enrolled from 18 hospitals, 163 were assigned to the BI group, and 169 were assigned to the RY group between May 2004 and October 2009.

| Surgical procedure
In both groups, the surgeons performed standard distal gastrectomy with laparotomy or laparoscopic operations. Lymphadenectomy approaches were categorized as D1-D3, as defined by the Japanese Classification of gastric carcinoma. 16 The surgeons reconstructed by BI or RY according to the intraoperative allocation, and the reconstruction details were described previously. 14,15 Briefly, for BI reconstruction, the duodenum and remnant stomach were sutured, and for RY reconstruction, the jejunum was divided 20 cm distal to the ligament of Treitz, and gastrojejunostomy and jejunojejunostomy were performed. The oral portion of the jejunum was then anastomosed to the midjejunum, 30 cm distal to the gastrojejunostomy. There were no restrictions regarding an open or laparoscopic approach, hand-sewn or stapling anastomosis, and antecolic or retrocolic routes during the RY reconstruction. In this study, all surgical procedures were performed or supervised by surgeons who were board certified by the Japanese Society of Gastroenterological Surgery and who were members of the Japanese Gastric Cancer Association. Laparoscopic surgery was performed or supervised by a qualified surgeon approved by the Endoscopic Surgical Skill Qualification system for clinical T1 early gastric cancer.

| Follow-up and data collection
Patients were followed for 5 years from the date of random assignment, and for as long as possible, thereafter. Adjuvant therapy was not specified in the protocol. Patients came to the hospital for examination at least once every 3 or 6 months for the first year after surgery. From the second year onward, patients were re-evaluated at least every 6 or 12 months until 5 years postoperatively. Relapse was confirmed by imaging studies, including ultrasonography and/or computed tomography at least at 1-year intervals until 5 years after surgery. Endoscopic examination was performed 1, 3, and 5 years after surgery to observe reflux esophagus (Los Angeles classification) and residual food in the remnant stomach.
We recorded patients' percentage body weight change from their pre-surgical body weight to their weight 1, 2, 3, 4, and 5 years after surgery. Other nutritional status characteristics, such as serum albumin level, lymphocyte count, and prognostic nutritional index (PNI) were evaluated before and at 1 and 5 years after surgery. PNI was calculated as 10 × serum albumin level (g/dL) + 0.005 × lymphocyte count in peripheral blood (cells/ mm 3 ). 17 Postoperative late complications ≥grade 2 according to the Common Terminology Criteria for Adverse Events version 3.0, recurrence, and survival were assessed from 1 to 5 years after surgery. 18

| Statistical analysis
All statistical analyses were performed with JMP Pro version 13.1 (SAS Institute Japan). Differences were considered significant at P < .05. Data were expressed as means ± standard deviation (SD).
Fisher's exact test for categorical variables and the two-sample t test for numerical variables were used to assess differences between the two groups, as appropriate. Total scores for the EORTC QLQ-C30 and DAUGS 20 were compared between the two groups using the Mann-Whitney test.

| Patients' characteristics
A consort flowchart of the trial design is shown in Figure 1 Table 1. There were 105 patients in the BI group and 123 patients in the RY group. The operation time was significantly longer in the RY group than in the BI group. Other characteristics were well-balanced in both groups. R0 gastrectomy was performed in all cases.

| Body weight change and nutritional status
The annual percentage body weight change for both groups is shown in Figure 2, from preoperative to the 5th year postoperatively. The percentage body weight change 5 years after surgery was −10.0% ± 7.9% for BI and −9.6% ± 8.4% for RY (P = .65), and there was no significant difference between the two groups at any point up to 5 years after surgery ( Figure 2). Additionally, no significant body weight loss was observed in either group from the first year to the 5th year after surgery.
Serum albumin levels, lymphocyte counts, and PNI values did not differ after 5 years (Table 2), and there was no difference in nutritional status between 1 and 5 years after surgery in both groups.
There was no significant difference in long-term weight loss or nutritional status according to the approach, anastomotic procedure, or route of reconstruction (data not shown).
F I G U R E 1 Study flow diagram. BI, Billroth I; RY, Roux-en-Y

| Late complications
The incidence of late complications occurring by the 5th year after surgery did not differ overall, but BI was associated with more frequent reflux esophagitis compared with RY (Table 3).  Figure 3A). Regarding the functional scales, no significant difference was found in any of the five scales (physical, role, emotional, cognitive, and social functioning; Figure 3A). Regarding the symptom scales, BI was significantly inferior to RY on the diarrhea scale (BI: 28.6 ± 47.0, RY: 16.0 ± 24.4; P = .047; Figure 3B). There were no significant differences in the other eight symptom scales (fatigue, nausea  Figure 4). In the subclass analysis, there was no difference between the two groups for all items, 5 years after surgery.

| QOL and gastrointestinal dysfunction
There was no significant difference in QOL according to surgical techniques (data not shown).

| D ISCUSS I ON
This study, conducted as an adjunct to a multicenter randomized controlled trial evaluating the reconstruction method after distal gastrectomy, showed that there was no difference between BI and RY regarding long-term weight loss and nutritional status. This finding was present even though RY was not superior to BI in terms of body weight change 1 year after surgery, as the primary endpoint in the original study. However, BI was inferior to RY regarding reflux esophagitis and diarrhea 5 years after surgery. We also found no difference in weight loss between the two groups from the first year to the 5th year after surgery, and there was no significant change in that time in either group. Several retrospective studies have compared BI and RY after distal gastrectomy. 10,11 This study provides valuable data because it involved a 5-year long-term follow-up in a large number of prospectively randomized cases and because we investigated QOL in addition to long-term postoperative weight loss and complications.
It is important to note that this study has limitations. First, we could not evaluate all patients because of death, loss to follow-up, or lack of informed consent for the amended protocol. The transfer of several attending doctors who registered the case before the protocol revision was also considered a cause. Additionally, the questionnaire collection rate was low at 65.4%. Most patients who participated in this trial were followed in their respective hospitals, but it is possible that their interest in the trial diminished 5 years after surgery. Second, laparoscopic surgery is currently the standard surgical procedure for distal gastrectomy for gastric cancer, but at the time of case accumulation for this study, the proportion of patients undergoing laparoscopic surgery was low at less than 25%.
Despite these limitations, our study should be useful to assist surgeons with deciding between the two procedures.
In the long-term nutritional assessment, nutritional indicators such as serum albumin concentration did not differ between the two groups after 5 years as well as after the 1st year after surgery. 14 There was also no difference in the weight loss rate between 1 and 5 years after surgery. It is generally known that body weight after gastrectomy decreases significantly in the early postoperative period and stabilizes after 6 months to 1 year; our results were consistent with previous reports. 10,11 A retrospective study reported that BI had a lower weight loss rate than RY, but it should be noted that there was bias in the operative choice in the study. 13 Generally, RY is selected when the remnant stomach is small, and the size of the remnant stomach may affect the degree of weight loss. According to the results of an RCT by Nakamura et al, there was no significant difference in nutritional index 3 years after surgery, but weight loss following BI was significantly less than after RY. 9 In that report, delayed gastric emp-

ACK N OWLED G EM ENTS
The authors thank the participating institutions and the chief partici- com/ac) for editing a draft of this manuscript.