We observed patients enrolled in a previous double-blind randomized controlled trial evaluating the effects of H. pylori eradication on mucosal atrophy and intestinal metaplasia in the remnant stomach after distal gastrectomy [10]. Detailed methods were described previously. In brief, newly diagnosed gastric cancer patients aged 18 to 70 years were recruited and screened for eligibility at the National Cancer Center, Korea, between August 2003 and May 2006. Eligibility criteria were as follows: distal gastric adenocarcinoma of clinical stage IA, IB, II, or IIIA according to the 6th edition of the International Union Against Cancer/American Joint Committee on Cancer TNM classification system [12], and positive tests for H. pylori infection. Informed consent was obtained from all participants. The Institutional Review Board of the National Cancer Center, Korea, approved the study protocol (NCCCTS03-063) and the original trial was registered with ClinicalTrials.gov (NCT01002443) [10].

Both the rapid urea test (RUT) and histology before distal gastrectomy were used for assessment of H. pylori infection. On the initial esophagogastroduodenoscopy (EGD), six biopsy specimens were taken from the lesser curvature of the antrum and the lesser and greater curvatures of the corpus for evaluation of H. pylori status, mucosal atrophy, and intestinal metaplasia. The collected biopsy specimens were immediately fixed in neutral-buffered 10% formalin, embedded in paraffin blocks, and stained with hematoxylin-eosin and Wright-Giemsa for histological evaluation. A single pathologist (M.-C.K.) performed pathological evaluation. One additional biopsy specimen was taken from the greater curvature of the corpus for RUT (Proton Dry, Medical Instruments Corporation, Solothurn, Switzerland). Patients confirmed as H. pylori infection by both a positive RUT and histological evaluation were enrolled in the study.

Eligible patients were randomly assigned to the treatment group or the placebo group by computer-generated random number. All investigators and patients were blinded to the randomization results throughout the trial. The treatment group received proton pump inhibitor (PPI)–based standard triple therapy, consisting of rabeprazole 10 mg, clarithromycin 500 mg, and amoxicillin 1,000 mg twice daily for 7 days. The placebo group received the same schedule of rabeprazole 10 mg but received a placebo rather than clarithromycin and amoxicillin. These treatments were administered before distal gastrectomy. Then, all patients underwent an open or laparoscopic distal gastrectomy with D1+ or D2 lymph node dissection [13]. The reconstruction methods were Billroth I (gastroduodenostomy) or II (gastrojejunostomy).

Follow-up EGDs with biopsy, RUT, and urea breath test (UBT) were performed at 3 months after distal gastrectomy for assessment of H. pylori status. The biopsy specimens were taken from the lesser and greater curvatures of the corpus, but not the antrum. At the follow-up examination, H. pylori was considered negative if all three tests were negative.

The calculation of study sample size was described in detail previously [10]. In the current study, outcomes were OS, gastric cancer-specific death, and gastric cancer recurrence. OS and gastric cancer-specific survival were determined from the time of distal gastrectomy to death; OS included deaths from any cause, but gastric cancer-specific death included only death from gastric cancer. Gastric cancer recurrence included metachronous cancer in the remnant stomach, lymph node metastasis, and distant organ metastasis occurring after gastrectomy for distal gastrectomy.

Continuous variables were compared using the Mann-Whitney U test and categorical variables were compared using the chi-square test or Fisher exact test. Follow-up data on deaths and recurrences were obtained until December 2014. The Kaplan-Meier method was used for construction of survival curves, and comparisons were performed using the log-rank test. Comparisons of gastric cancer-specific death and gastric cancer recurrence rates were performed using Gray’s test because of the possible competing risk of death without recurrence. A multivariate Cox proportional hazards model analysis was performed for overall mortality, and a multivariate Fine-Gray regression model analysis was performed for cancer recurrence with competing risks. In model 1, multivariate analyses were performed according to the allocated groups, and in model 2, analyses were performed according to the H. pylori status at 3 months after gastrectomy. Covariates for multivariate analyses were age, sex, and variables having a p-value less than 0.1 in univariate analyses. Analyses were performed using STATA ver. 12.1 (Stata Corp., College Station, TX), except for the Gray’s test, which was analyzed using SAS ver. 9.3 (SAS Inc., Cary, NC). A p-value less than 0.05 was considered statistically significant.

During a median follow-up period of 9.4 years (IQR, 9.0 to 10.2 years; range, 0.7 to 11.3 years), 26 patients died: 14 in the treatment group and 12 in the placebo group. The 5-year OS rates were 98.9% in the treatment group and 91.5% in the placebo group, and results of Kaplan-Meier analysis showed no significant difference between groups (p=0.957 by log-rank test) (Fig. 2A). Overall, 13 patients had gastric cancer recurrence (6 [6.9%] in the treatment group and 7 [8.5%] in the placebo group). Of these, metachronous cancers occurred in 4 patients (3 patients in the treatment group and 1 patient in the placebo group), and the metachronous cancer rates were not different between the treatment group and the placebo group (3.5% vs. 1.5%, p=0.341). One patient died of gastric cancer metastasis at 5 months after total gastrectomy for metachronous cancer, while, the remaining nine patients had lymph node or other organ metastases (3 patients in the treatment group and 6 patients in the placebo group). Of the patients with gastric cancer recurrence, 10 patients died because of gastric cancer recurrence (4 patients in the treatment group and 6 patients in the placebo group). The 5-year incidence rates of gastric cancer recurrence were 4.6% in the treatment group and 8.5% in the placebo group, and Gray’s test showed no statistical significance (p=0.652) (Fig. 2B). No difference in the cumulative incidences of gastric cancer-specific death were observed between the treatment and placebo groups (5-year cumulative incidence rate, 1.1% vs. 6.1%, respectively; p=0.422 by Gray’s test) (Fig. 2C).

At 3 months after gastrectomy, 82 patients had negative H. pylori status and 85 had positive H. pylori status. After gastrectomy, OS did not differ between patients with negative H. pylori status and those with positive H. pylori status (5-year OS rates, 96.3% vs. 94.1%, respectively; p=0.997 by log-rank test) (Fig. 3A). In addition, no statistical differences in gastric cancer recurrence (5-year cumulative incidence rates, 6.1% for negative status vs. 7.1% for positive status; p=0.897 by Gray’s test) (Fig. 3B) and gastric cancer-specific death (5-year cumulative incidence rates, 3.7% for negative status vs. 3.5% for positive status; p=0.792 by Gray’s test) (Fig. 3C) were observed between patients with and without H. pylori at 3 months after distal gastrectomy.

In univariate analyses, overall mortality after distal gastrectomy was not affected by H. pylori treatment allocation (crude hazard ratio [cHR] for H. pylori treatment, 1.02; p=0.957) or positive H. pylori status at 3 months after distal gastrectomy (cHR for H. pylori–positive, 1.00; p=0.997). Male sex, Billroth II anastomosis after distal gastrectomy, and stage ≥ IIIa were significant factors associated with overall mortality. In model 1, multivariate analysis according to H. pylori treatment allocation showed that age (adjusted HR [aHR], 1.06; p=0.017), male sex (aHR, 8.38; p=0.005), and advanced stage ≥ IIIa (aHR, 9.18; p < 0.001) were independent risk factors for overall mortality. H. pylori treatment was not a significant factor for overall mortality (aHR for treatment group, 0.75; p=0.495). In model 2, positive H. pylori status at 3 months after gastrectomy did not show significant association with overall mortality (aHR, 1.16; p=0.715), whereas age, male sex, and advanced stage ≥ IIIa were independent risk factors for overall mortality (Table 2).

For cancer recurrence, neither H. pylori treatment (cHR, 0.78; p=0.649) nor positive H. pylori status after gastrectomy (cHR, 1.13; p=0.821) were significant factors in univariate analyses. Multivariate Fine-Gray regression analyses showed that only the advanced stage ≥ IIIa was a significant risk factor in both model 1 and 2. In these analyses, H. pylori treatment (aHR, 0.60; p=0.431) and positive H. pylori status at 3 months after distal gastrectomy (aHR, 1.05; p=0.921) did not affect cancer recurrence (Table 3).