Roux-en-Y Reconstruction Following Distal Gastrectomy Reduced Endoscopic Reflux Esophagitis in Older Adults: Propensity Score–Matching Analysis

We identified 421 patients who underwent DG in our database of gastric cancer from 2011 to 2018. Of these, we excluded 230 patients aged <75 years, 12 who had stage IV disease, 4 who underwent double tract reconstruction, and 1 with severe preoperative hiatal hernia and esophagitis. Of the remaining 174 patients, 103 and 71 had undergone RY and BI reconstructions, respectively.

Propensity score matching by a logistic regression model was conducted to minimize the bias between the RY and BI groups. Propensity scores were calculated with the following items: age, sex, body mass index, the Charlson Comorbidity Index, tumor location, and pathologic stage determined by the American Joint Committee on Cancer staging.²³  Patients from the 2 groups were individually matched using statistical software, as noted below. As shown in a flow chart (Fig. 1), 62 patients who underwent RY and BI reconstructions were each matched.

This study was conducted according to the Declaration of Helsinki principles and was approved by the Institutional Review Board of the Japanese Red Cross Musashino Hospital. The requirement for written informed consent was waived because of the retrospective design; however, the study's protocol was displayed for patients on our hospital website (https://www.musashino.jrc.or.jp/clinical/index.html; in Japanese).

We incised the stomach at about 3 cm length from the cardia, so that enough proximal margin was secured for the tumor localized in the middle or the lower portion. Surgeons selected BI reconstruction when they confirmed a remnant stomach of adequate size and no anastomotic tension. Otherwise, we selected RY reconstruction. In the case of distal subtotal gastrectomy, double tract reconstruction was performed. In BI reconstruction, we performed an extracorporeal gastroduodenostomy using a circular stapler of 28 mm in diameter. In RY reconstruction, we divided the jejunum at approximately 20 to 30 cm from the duodenojejunal flexure and made a Roux limb with a length of approximately 20 to 30 cm.¹¹  We performed gastrojejunostomy or jejunojejunostomy using 60-mm–long and 45-mm–long linear staplers, respectively. Antiperistaltic anastomosis was performed on both sides, and the Roux limb was anastomosed in the anticolic route. We excluded patients with severe hiatal hernias; therefore, no patients underwent hiatal hernia repair. A certified surgeon conducted all reconstructions as an operator or instructive assistant.

The following surgical outcomes were investigated: time of operation, estimated blood loss, postoperative recovery course (time to the first flatus, day until initiation of solid foods, and postoperative hospital stay), postoperative complications related to the remnant stomach (leakage, DGE, etc), mortality, reoperation, and readmission within 90 days. Although the initiation of liquid and solid foods was decided on postoperative days 1 and 4 in our clinical pathway, it was postponed in some patients because of poor bowel movements. The discharge from hospital was decided on postoperative day 7 or later in our clinical pathway; however, it was finally established according to the patient's requirement in clinical practice. RY and BI reconstructions were compared for the following clinical findings at 1 year postoperatively: (1) endoscopic findings by the Residue, Gastritis, and Bile (RGB) classification,²⁴  reflux esophagitis based on the Los Angeles classification (grade A or severe),²⁵  and the presence of hiatus hernia detected endoscopically after confirming the space in the gastric orifice in the reversed image¹⁹; (2) relative value to the preoperative level of the following factors: hemoglobin, albumin, vitamin B₁₂, and body weight; and (3) clinical symptoms due to gastroesophageal reflux disease (GERD) and early dumping syndrome via personal interviews with patients. GERD-related symptoms included regurgitation, thoracalgia, and heartburn. The symptoms of early dumping syndrome included perspiration, palpitations, hot flashes, and vertigo immediately postprandially. Additionally, preoperative endoscopic findings of hiatal hernia and reflux esophagitis were investigated to analyze risk factors for endoscopic reflux esophagitis after DG.

We analyzed categoric data using the χ² and Fisher exact test and analyzed continuous data using the Mann-Whitney U test. The item was considered statistically significant when the P value was <0.05. The nearest-neighbor matching method was used to match patients between the 2 groups. We used a binary logistic multiple regression model with significant dummy variables (P < 0.05) to evaluate a predictive factor and performed all analyses using SPSS 24 (SPSS Japan, Tokyo, Japan).

Intergroup differences were not observed in the background of patients after matching (Table 1). Although the RY group had a significantly longer operation time than the BI group, blood loss volume and postoperative recovery course were comparable between both groups (Table 2). The proximal margin was significantly longer in the BI group than in the RY group (Table 2). The total postoperative complications related to the remnant stomach were similar between the RY and BI groups (Table 2). However, DGE occurred only in the RY group, with 3 of 5 patients needing drainage with a nasogastric tube, whereas no patient had DGE in the BI group.

Clinical findings at 1 year postoperatively were evaluated in 48 and 43 patients in the RY and BI groups, respectively (Table 3). Thirty-three patients could not be followed up for 1 year because of recurrent disease (n = 12), treatment for other malignancies or severe diseases (n = 4), death from other diseases (n = 6) or from complications after gastrectomy (n = 1), and not visiting our hospital (n = 10). Regarding the RGB scores, remnant gastritis, bile reflux, and reflux esophagitis were significantly less frequent in the RY group than in the BI group. The incidence of endoscopic hiatus hernia was comparable between groups both preoperatively and postoperatively; however, the prevalence was significantly higher postoperatively. No patients developed preoperative reflux esophagitis. The incidence of GERD-related symptoms and dumping syndrome was not significantly different between the 2 groups, and the relative values of hemoglobin, serum albumin, vitamin B₁₂ levels, and body weight were comparable between both groups. Two patients who underwent RY took iron for anemia at 1 year postoperatively, whereas 6 received postoperative adjuvant chemotherapy in both groups.

Risk factors for postoperative endoscopic reflux esophagitis are shown in Table 4. Although both BI reconstruction and preoperative hiatus hernia were independent risk factors, postoperative hiatus hernia did not show a significant association with endoscopic reflux esophagitis. Moreover, bile reflux, remnant gastritis, and residual food showed no significant association with postoperative endoscopic reflux esophagitis.

We demonstrated that RY reconstruction reduced reflux esophagitis and remnant gastritis, which are probably caused by bile reflux in older patients following DG. The findings of this study are similar to those of previous reports showing that RY reduced endoscopic remnant gastritis and bile reflux.^1,2,4–10  Therefore, these findings are likely to be common in both older and young patients.

The incidence of grade A or higher reflux esophagitis was 0 to 17% in RY and 5% to 30% in BI reconstruction for patients of all ages at 1 year postoperatively.^1,8–10  Some reports have shown a significant decrease in reflux esophagitis after RY reconstruction.^1,5,7,9,10  Another study on BI reconstruction showed the involvement of a smaller remnant stomach with reflux esophagitis.²⁶  BI reconstruction was only performed when the remnant stomach was large enough to reach the duodenum without tension in this study. Therefore, the remnant stomach's size might not affect the increasing incidence of reflux esophagitis after BI reconstruction.

Postoperative reflux esophagitis was associated with both BI reconstruction and preoperative hiatus hernia. However, postoperative hiatus hernia showed no significant association with endoscopic reflux esophagitis. Hiatus hernia is involved in the relaxation of the lower esophageal sphincter and the delayed clearance of esophageal acid in patients who have not undergone gastrectomy.^27,28  Hiatus hernia is considered a risk factor for reflux esophagitis in patients without gastrectomy.^18,19  Severe chalasis of the lower esophageal sphincter after DG may occur in patients with preoperative hernia; therefore, RY reconstruction would be a good method in such patients to prevent reflux esophagitis. RY reconstruction may be required more frequently after DG because the incidence of preoperative hiatus hernia is likely to be greater in older patients than in the young population.^20,21 

A study on bilirubin monitoring in the esophagus after DG showed increased duration of bile reflux following BI reconstruction compared with that following RY reconstruction.⁴  Therefore, bile reflux is considered an important cause of reflux esophagitis after DG. However, its macroscopic detection in the esophagus is difficult in endoscopic examinations. Similarly, it causes mucosal inflammation in remnant gastritis¹³ ; however, bile reflux into the remnant stomach or remnant gastritis did not significantly relate to reflux esophagitis in the present study.

We found that the GERD-related symptoms at 1 year postoperatively were similar between the RY and BI groups, contrary to other studies that found these symptoms significantly reduced following RY.^4,5,7–9,16  In patients who did not undergo gastrectomy, GERD-related symptoms showed no correlation with the severity of reflux esophagitis.²⁹  Older patients with severe GERD commonly had atypical or mild symptoms, whereas those patients with mild reflux esophagitis complained of heartburn more frequently than young patients.^30,31  GERD-related complaints were less reliable than endoscopic findings, particularly in elderly patients. These might be reasons for the divergence between the endoscopic findings and GERD-related symptoms in this study.

Early postoperative complications were not significantly different in randomized controlled trials and some retrospective studies comparing RY and BI reconstruction.^1,2,6,8,32  However, 2 studies comparing the procedures revealed that RY reconstruction had a significantly higher incidence of DGE, bowel obstruction, internal herniation, and intra-abdominal abscess than BI reconstruction.^15,33  Although a previous study has reported a 21% incidence of DGE after RY,⁶  it was the most common complication in this study. We diagnosed DGE based on symptoms requiring fasting in patients without gastrointestinal stenosis. However, various definitions of DGE have been proposed, and many of them included the late removal of the nasogastric tube or a postponement of initiating diet.³⁴  Therefore, a comparison between studies is difficult. A retrospective study showed that isoperistaltic gastrojejunostomy in RY reconstruction, higher body mass index, and tumors located in the lower third of the stomach were risk factors for DGE after DG.¹⁴  Antiperistaltic gastrojejunostomy in RY did not prevent DGE in the present study. Patients with DGE recovered within 1 month after gastrectomy, and no patients needed temporary stent placement or reoperation in this study.

Our study has some limitations. (1) This study was retrospective, although data were collected prospectively. Moreover, we did not use internationally validated questionnaires to investigate clinical symptoms or quality of life. (2) BI reconstruction was selected only in cases with no anastomotic tension. This selection could have potentially created a bias and influenced our results, although the tumor location was matched to minimize the bias. As mentioned above, a larger remnant stomach was more effective to prevent reflux esophagitis in a published study.²⁶  However, 4 cases of distal subtotal gastrectomy were excluded from the analysis because of double tract reconstruction. Based on our operation records in the matched groups, the stomach was divided at about 3 cm length from the cardia, except in the cases of 3 patients, where the length was unrecorded. Therefore, the remnant stomach size did not differ considerably in the matched cases. In fact, the proximal margin was shorter in the RY group, including more tumors located in the middle portion than in the BI group. Additionally, the tumor location and the length of proximal margin were not associated with endoscopic reflux esophagitis in the present study. (3) Reflux into the esophagus was not assessed by pH monitoring. (4) The number of patients who developed postoperative reflux esophagitis was too low to demonstrate the significance of other risk factors.

In older patients who underwent DG for gastric cancer, RY was a more favorable reconstruction procedure than BI, resulting in decreased postoperative endoscopic reflux esophagitis, bile reflux, and remnant gastritis. Similarly, it was a better procedure for patients with hiatus hernia, which was a risk factor for endoscopic reflux esophagitis after DG. However, GERD-related symptoms were not different between the groups, and DGE was the most common postoperative complication in older patients following RY.

M. Inokuchi, T. Ogo, and T. Irie were responsible for drafting the manuscript. M. Inokuchi and T. Ogo contributed to analysis and interpretation of data. S. Kato, H. Nagano, and K. Kawachi contributed to data collection. The authors have no conflicts of interest to declare.