Patients
This retrospective study consisted of 156 patients diagnosed with gastric adenocarcinoma in the First Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China, from 2008 to 2016. All patients included in the study met the following criteria: underwent curative gastrectomy followed by adjuvant chemotherapy; with detailed postoperative pathological report and with stage Ⅱ or Ⅲ disease according to the American Joint Committee on Cancer (AJCC) seventh staging system; with definite medical records of recurrence or metastasis after surgical resection and available details on systemic adjuvant chemotherapy; received at least one cycle of systemic adjuvant chemotherapy; adjuvant chemotherapeutic regimens should be as doublet combination of platinum and fluoropyrimidines or triple combination of taxanes, platinum, and fluoropyrimidines. For platinum drugs, it could be cisplatin, oxaliplatin, or lobaplatin; for fluoropyrimidine compounds, intravenous 5-fluorouracil injection, oral capecitabine tablets, or tegafur-gimeracil-oteracil potassium capsules were all acceptable; taxanes could either be docetaxel or paclitaxel. In consideration of tolerance to toxicity, drug accessibility, etc., patients whose adjuvant chemotherapeutic drugs were switched from one platinum drug to another, or one fluoropyrimidine compound to another, or docetaxel to paclitaxel (vice versa) in different cycles were considered eligible. Patients who met any of following criteria were excluded from the study: did not receive curative surgery; diagnosed as having a very early phase of disease (stage Ⅰ); with no adjuvant chemotherapy apart from the abovementioned regimens; with missing information on surgery, pathology, adjuvant treatment, or recurrence and metastasis.
Data collection and follow-up
We retrieved the medical records of the patients from the Hospital Information System (HIS), and reviewed all related medical files. The following variables were collected: demographic variables including age and sex as well as pathological variables including tumor location, subtype, grade, tumor infiltration (T category), regional lymph node involvement (N category), vascular or perineuronal invasion by tumors in pathological specimens, and perigastric tumor deposits. Treatment information, including laparotomic approaches, scope of gastrectomy, type of lymphadenectomy, adjuvant chemotherapeutic drugs, date of the first and last dose of chemotherapeutic drug, and cycles of adjuvant chemotherapy, was obtained. Staging groups were derived from the T and N categories according to the AJCC seventh staging system. The time to chemotherapy after surgery was defined as interval days from primary surgery to the first dose of adjuvant chemotherapy. The duration of adjuvant chemotherapy was defined as interval days from the first to the last dose of chemotherapeutic drug. Recurrence and metastasis were defined as tumor recurrence in situ (anastomotic stoma or gastric remnant) or metastasis to distant organs, lymph node(s), or intraperitoneal implantation after gastrectomy, with evidence of either imaging, cytology, or histopathology. Survival status was obtained from death records in the HIS or telephone follow-up on the patients or their relatives.
Statistical analysis
DFS was defined as interval months between gastrectomy and first evidence of recurrence and metastasis. Overall survival (OS) was defined as interval months between gastrectomy and death date or last follow-up date. For categorical variables, the Chi-square test was used to examine the differences between the two groups; for numerical variables and nonparametric independent samples the Mann-Whitney U test was used. Survival curves were calculated using the Kaplan-Meier method, and differences were compared using the log-rank test. Hazard ratios (HRs) with 95% confidence intervals (CIs) and two-sided P values were reported. HRs in subgroups according to baseline characteristics and two-tailed 95% CIs were calculated using the Cox proportional hazards model. All statistical analyses were performed using SPSS 22.0 (SPSS Inc., Chicago, IL, USA). Two-sided P < 0.05 was considered to be statistically significant. GraphPad Prism 5.01 software (GraphPad Software Inc., San Diego, CA, USA) was used to draw survival curves and forest maps of the subgroup analysis.