Introduction

Gastric cancer (GC) is the six most common type of cancer worldwide and ranks fourth after liver cancer among all causes of death due to malignant disease [1]. Locally advanced GC can be cured by complete resection with or without perioperative adjuvant chemotherapy. In east Asia, the standard treatments of locally advanced GC are gastrectomy with extended (D2) lymph-node dissection followed by post-operative adjuvant chemotherapy. S-1 monotherapy, S-1 plus oxaliplatin, capecitabine plus oxaliplatin, and S-1plus docetaxel have been recognized as the standard postoperative adjuvant treatment after curative gastrectomy in patients with stage II or III GC in Japan [2,3,4,5].

Palliative chemotherapy is the main treatment approach for stage IV GC. According to the results of REGATTA trial, surgical resection of the primary lesion in stage IV gastric cancer patients has been discouraged [6]. Thus, stage IV GC is generally outside the indication of surgical resection, except for the patients complicated with severe symptoms by primary tumors such as bleeding and impaired passage of the stomach.

In contrast, resection of distant metastasis is a standard treatment with curative intention for stage IV or recurrent colorectal cancer, if possible. Recently, several studies have suggested favorable clinical outcomes of stage IV gastric cancer patients who underwent curative resection of liver metastasis and distant lymph node metastasis, reporting 5-year survival rates of 27–37% and 13–65%, respectively [7,8,9,10,11,12,13]. Based on these studies, multidisciplinary treatment combining surgery and post-operative chemotherapy for GC associated with oligo-metastasis (O-meta) is now regarded as a promising treatment option in Japan, although the benefit of postoperative chemotherapy following resection of O-meta of gastric cancer has not been clarified.

In Japan, cisplatin plus S-1 combination (CS) regimen has been one of the standard first line treatment in patients with stage IV GC based on the results of phase III trials in the SPIRITS and JCOG 9912 studies [14, 15], while S-1 monotherapy has been recognized as standard adjuvant chemotherapy after curative resection of stage II/III gastric cancer for more than 10 years [2]. Moreover, among gastric cancer patients with positive laparotomic cytology positive (CY1) as a single non-curative factor, there was no difference in overall survival after primary tumor resection with lymph node disection followed by post-operative chemotherapy between S-1 monotherapy and S-1 plus cisplatin [16]. Thus, it is also unclear which of these two regimen is optimal as post-operative chemotherapy afte resection of O-meta.

The aim of this retrospective study was to evaluate the efficacy of post-operative chemotherapy and explore the optimal chemotherapy regimen, S-1 or CS for adjuvant chemotherapy after curative resection in patients with GC with O-meta.

Patients and methods

Patients

We retrospectively reviewed the medical records of GC patients with synchronous O-meta who received curative surgical resection at 20 Japanese institutions between 2007 and 2012. The selection criteria were: age > 20 years, Eastern Cooperative Oncology Group (ECOG) performance status (PS) after surgery of 0 to 2, histologically confirmed gastric adenocarcinoma, R0 resection including O-meta limited to a single organ such as liver or distant lymph node diagnosed after surgery, and no prior chemotherapy for GC before surgery.

Treatment and procedure

Diagnosis and treatment approach for curative resection including synchronous O-meta were entrusted to each institution. Post-operative chemotherapy regimens were determined by each attending physician. In general, the standard dose and schedule of S-1 monotherapy and CS recommended in the Japanese Gastric Cancer Treatment Guideline were used. S-1 monothrerapy comprized S-1 (40–60 mg/body according to the body surface area), orally administered twice daily for 4 weeks followed by a 2-week rest period, repeated in a 6-week cycle. CS comprised oral administration of S-1 40–60 mg/body twice daily for the first 3 weeks and intravenous infusion of cisplatin 60 mg/m2 on day 8 of each cycle, repeated every 5 weeks. Treatment duration, dose redution, dose delay, schedule altration were discreted by each physician.

Evaluation and statistical analysis

We compared overall survival (OS) and relapse-free survival (RFS) between pateitns with (Post-Cx group) and those without adjuvant chemotherapy (Non-Cx group), and among the following post-operative chemotherapy regimens: S-1 monotherapy (S-1 group), S-1 plus cisplatin (CS group), other regimens (Others group) and no chemotherapy (Non-Cx group). OS was calculated from the date of surgery to the date of death from any cause or censored at the last follow-up of survivors. RFS was calculated from the date of surgery to the date of relapse or death from any cause. Surviving patients without relapse were censored at the last follow-up. OS and RFS were estimated using the Kaplan–Meier method and compared with the log-rank test. The median duration of follow-up and associated 95% confidence interval (CI) was calculated using the reverse Kaplan–Meier method. Proportions for categorical data and medians with ranges for continuous variables were calculated for patient characteristics. The multivariable Cox proportional hazard model was used to adjust confounding, evaluate survival differences among treatment groups, and present hazard ratios (HR) and 95% CI. Covariates for the multivarable analysis included treatment (Post-Cx or Non-Cx, and chemotherapy regimens, respectively), liver metastasis (present vs. absent), distant lymph node metastasis (present vs. absent), age (≤ 70 vs. > 70 years), ECOG PS (0 vs. ≥ 1), histological type (intestinal vs. diffuse), and post-operative complication Clavien–Dindo classification ≥ Grade 3 (present vs. absent). Statistical analyses were performed using the SAS software version 9.4 (SAS Institute, Cary, NC, USA). P values < 0.05 were considered as statistically significant differences.

Results

Patient characteristics

Figure 1 presents the consort diagram. Clinical data of totally 110 patients were colltected in the study. Sixteen patients (non-curative resection n = 12, other distant metastasis than liver or lymph node n = 2, and incomplete data n = 2) were excluded from the analysis. The characteristics of the remainig 94 patients are shown in Table 1. The median age was 66 (23–88); 71% of the patients were male; 84% had a PS 0; 39% had a diffuse type histologically; 41% and 59% had iver metastasis and distant lymph node metastasis, respectively. All patients underwent open surgerys. Incidences of post-operative complication (Clavien–Dindo classification: ≥ Grade 3) were 13% in all patients. Eighty-four (89.4%) patients underwent post-operative chemotherapy (the Post-Cx group) and 10 (10.6%) patients (the Non-Cx group) did not. The post-operative chemotherapy regimens were S-1 monotherapy (the S-1 group: n = 55, 58.5%); S-1 plus cisplatin (the CS group: n = 22, 23.5%); and others (the Others group: n = 7, 7.4%). The median age in the S-1, CS, Others, and Non-Cx groups were 66, 60, 61, and 79 years old, and proportions of liver/distant lymph node metastasis were 38%/72%, 41%/59%, 28%/72%, and 70%/30%, respectively. The median follow-up time was 64.9 months in the S-1, 94.0 months in the CS, 66.0 months in the Others and 60.1 months in the Non-Cx groups.

Fig. 1
figure 1

Consort diagram

Full size image
Table1 Patient characteristics
Full size table

Drug delivery

The median time from surgery to initiation of post-operative chemotherapy was 47 days in the S-1, 43 days in the CS and 49 days in the Others groups. The median treatment duration of post-operative chemotherapy was 203 days (range: 3–1230) in the S-1, 201 days (range: 12–461) in the CS and 70 days (range: 21–383) in the Others groups. The median number of cycles of cisplatin administration in the CS group was four.

Impact of post-operative chemotherapy

In all patietns, the median OS was 24.0 months (95% confidence interval [CI] 21.7–26.3) and the 3- and 5-year OS rates were 45.6% (95% CI 35.2–55.4) and 31.4% (95% CI 22.0–41.2) (Fig. 2a). The median RFS was 11.4 months (95% CI 8.4–15.4) and the 3- and 5-year RFS rates were 24.5% (95% CI 16.3–33.5) and 21.3% (95% CI 13.7–30.0) (Fig. 2b).

Fig. 2
figure 2

Kaplan–Meier analysis of overall survival and relapse-free survival

Full size image

The median OS was 35.2 and 11.1 months in the Post-Cx and Non-Cx groups (hazard ratio [HR] 3.56; 95% CI 1.74–7.30; p < 0.001) (Fig. 3a). The median RFS was 12.9 and 4.1 months in the Post-Cx and Non-Cx groups (HR 2.69; 95% CI 1.33–5.45; p < 0.004), respectively (Fig. 3b). Table 2 shows the results of multivariable analysis comparing Post-Cx and Non-Cx. The Non-Cx group (HR 3.15, 95% CI 1.32–7.51, p = 0.009), age > 70 years (HR 1.89, 95% CI 1.03–3.46, p = 0.038) and liver metastasis (HR 0.53, 95%CI 0.29–0.99, p = 0.047) were identified as significant prognostic factors.

Fig. 3
figure 3

Kaplan–Meier analysis of overall survival and relapse-free survival in the Post-Cx and Non-Cx groups

Full size image
Table 2 Univariable and multivariable analyses for OS with Cox proportional hazards models
Full size table

Comparison amonng the treatment groups

According to the chemotherapy regimens, the median OS was 28.5 months (95% CI 25.7–51.3) in the S-1, 36.8 months (95% CI 16.8–70.5) in the CS, and 54.0 months (95% CI 5.4–N.A) in the Others (Fig. 4a). The median RFS was 13.4 months (95% CI 8.4–21.7) in the S-1, 12.7 months (95% CI 4.9–25.6) in the SP, and 9.3 months (95% CI 2.5–45.1) in the Others (Fig. 4b).

Fig. 4
figure 4

Internal analysis in the post-operative chemotherapy group; Kaplan–Meier analysis of overall survival and relapse-free survival

Full size image

In the univariable analysis, there were no statistically significant differences in OS between the CS and S-1 groups (CS vs. S-1: HR 1.0, 95% CI 0.54–1.82, p = 1.00), and between the Other and S-1 groups (Others vs. S-1: HR 1.12, 95% CI 0.39–3.16, p = 0.82). There were no statistically significant differences in RFS between the CS and S-1 groups (CS vs. S-1: HR 1.21, 95% CI 0.71–2.09; p = 0.48), and between the Others and S-1groups (Others vs. S-1: HR 1.18, 95% CI 0.50–2.77, p = 0.75) (Table 3). In the multivariate analysis, there were no significant differences in OS among the three chemotherapy groups, identifying age as the significant prognostic factor (Table 3).

Table 3 Univariable and multivariable analyses for OS with Cox proportional hazards models in the treatment groups
Full size table

Subgroup analysis according to site of O-meta

The median OS in patients with liver metastasis and distant lymph node metastasis were 30.6 (95% CI 21.6–54.0) and 27.0 months (95% CI 17.1–39.2) (HR 1.12; 95% CI 0.68–1.86; p = 0.65), respectively. The 3-/5-year survival rates in patients with liver metastasis and distant lymph node metastasis were 48.7/43.4% and 31.0/31.5%, respectively (Suppl. Fig. S1a). The median RFS in patients with liver metastasis and distant lymph node metastasis was 12.5 (95% CI 5.3–25.9) and 10.5 months (95% CI 7.6–15.4) (HR 1.09; 95% CI 0.69–1.73; p = 0.71), respectively. The 3-/5-year RFS rates in patients with liver metastasis or distant lymph node metastasis were 28.2/23.1% and 21.8/20.0%, respectively (Suppl. Fig. S1b). Although there were no statistically significant differences in OS and RFS regardless metastatic sites in the univaiate analysis, multivariable analysis in all 94 patients showed that liver metastasis was an independent prognostic factor for longer OS than lymph node metastasis.

Liver metastasis

Liver metastasis was synchronous in all patients. Single and multiple (≧2) liver metastases were found in 23 and 16 patients, respectively. The median OS was 44.8 (95%CI 17.5–NR) and 25.3 months (95%CI 11.4–54.0), in the patients with a single liver metastasis and those with multiple liver metastasis, respectively (HR 1.79; 95% CI 0.81–3.96; P = 0.146). The 3-/5-year OS rates in patients with single liver metastasis and multiple liver metastasis were 59.1/39.8% and 31.3/20.8%, respectively (Suppl. Fig. S2a). The median RFS was 14.1 (95%CI 4.5–67.3) and 11.3 months (95%CI 4.1–25.9), in the patients with a single liver metastasis and those multiple liver metastasis (HR 1.41; 95% CI 0.68–2.89; p = 0.350), respectively. The 3-/5-year RFS rates in patients with single liver metastasis and those with multiple liver metastasis were 36.4/31.8% and 18.8/12.5%, respectively (Suppl. Fig. S2b).

Distant lymph node metastasis

In terms of distant lymph metastasis, para-aortic lymph node (PALN) and other extra regional lymph node (ERLN) metastases were found in 38 and 17 patients, respectively. In the present study, ERLN metastatic site were No.13 (n = 8), No.14a (n = 3), No. 8p (n = 3), No.17 (n = 2) and No. 108 (n = 1). The median OS was 27.1 (95%CI 16.8–39.2) and 26.0 months (95%CI12.5–NR), in the patients with PALN metastasis and those with ERLN metastasis (HR 0.77; 95% CI 0.35–1.54; p = 0.472), respectively. The 3-/5-year OS rates in the patients with PALN metastasis and those with ERLN metastasis were 41.4%/26.6% and 41.1/41.1%, respectively (Suppl. Fig. S3a). The median RFS was 11.0 months (95%CI 8.2–17.9) and 9.7 months (95%CI 3.4–16.7), in the patients with PALN metastasis and those with ERLN metastasis (HR 1.06; 95% CI 0.53–1.99; p = 0.84), respectively. The 3-/5-year RFS rates in the patients with PALN metastasis and those with ERLN metastasis were 21.0/18.2% and 23.5/23.5%, respectively (Suppl. Fig. S3b).

Discussion

Liver metastasis and distant lymph node metastasis are defined as M1 (metastatic) factors for staging in the 15th edition of the Japanese Classification of Gastric Carcinoma [17]. Although systemic chemotherapy is recognized as the standard treatment for GC patients with M1 disease, surgical approach is occassionally applied for patietns with O-meta. In the previous Spanish registry study, which registered 2549 patients with unresectable advanced gastric or gastro-oesophageal junction cancer registered from 32 centers, 1792 patients were analyzed. Among them, 92 patients (5%) underwent surgery for Oligo-meta. The most common O-meta sites were peritoneal (29%), liver (24%), and distant lymph nodes (11%)[18]. The peritoneum, liver and distant lymph nodes are common metastatic sites of unresectable advanced GC as well as O-meta. In the present study, patients with peritoneal metastasis were not included to focus on the treatment strategy for O-meta detectable by image diagnosis before surgery, because peritonal metastasis diagnosed by image is not recognized as O-meta.

Tokunaga et al. reported the 3-year and 5-year survival rates for patients with advanced GC with PALN metastasis who underwent curative surgery, which were 20.9% and 13.0%, respectively [12]. Kinoshita et al. reported the 3-/5-year overall and recurrence-free survival rates for patients with advanced GC with synchronous and metachronous liver metastasis who underwent curative surgery, which were 41.9/31.1%, and 32.4/30.1%, respectively [10], associated with the median overall and recurrence-free survival times of 31.1/9.4 months. In the all subjects of the present study, the 5-year overall survival and relapse free survival rates were 31.4% and 21.3%, respectively. Consistently, these data have indicated that surgical resection of O-meta has a potential to cure selected GC patients [19,20,21]. Considering the chemotherapy alone in patients with stage IV GC is not curative but palliative, surgical resection can be recommended for GC with resectable O-meta.

However, it should be emphasized that the median PFS was 10.5 months in the distant lymph metastasis group and 12.2 months in the liver metastasis group. Half of the recurrences developed within 12 months despite careful patient selection for R0 resection. More appropriate patient selection might be necessary. As for liver metastatis, some previous reports suggested that patients with a single liver metastasis had significantly higher 3-year survival rates than those with multiple liver metastases [9, 20, 21]. Patients with multiple liver metastasis are generally considered as poor candidates for hepatectomy. In the present study, GC patients with a single liver metastases tended to have a better prognosis than those with multiple liver metastasis. However, considering that systemic chemothrapy alone hardly achieve cure, while there is a substantial cure rate even in patients with multiple liver metastasis, limiting the surgical indication to a single liver mstastasis can not be recommended.

As for lymph node metastasis, both PALN and ERLN are defined as distant metastasis in the 15th edition of the Japanese Classification of Gastric Carcinoma. In the present study, there were no differences either in OS or RFS between PALN and ERLN. While PALN is the most common metastatic site which is candidate for surgical resection of O-meta, O-meta at ERLN can be included in the indication of surgical resection.

This study showed that Post-Cx, age (< 70) and liver metastasis were independent favorable prognostic factors. Qiu et al. reported that the median OS after hepatectomy in patients with GC who received post-operative chemotherapy was 43.0 months, whereas it was 32.0 months in those who did not (p = 0.022). Tiberio et al. also reported the benefit of post-operative chemotherapy after surgical resection of liver metastases in patients with GC. The median OS was 24.4 months in the post-operative chemotherapy group and 8.2 months in the surgery alone group (p = 0.001) [22, 23]. The consistent results with the present study suggest that post-operative chemotherapy may have some survival benefits for patients with GC with O-meta.

However, treatment results of curative surgery for GC with O-meta are still not satisfactory. It is necessary to develop more efficacious post operative chemotherapy after curative resection of GC with O-meta. For stage III GC, intensive adjuvant chemotherapy with doublet chemotherapy such as S-1 plus docetaxel and S-1 plus oxaliplatin improved survival over S-1 monotherapy. In this study, however, the CS regimen, which improved survivl of unresectable/recurrent GC patients compared with S-1 monotherapy, did not show statistically significant clinical benefits for OS over S-1 using univariable and multivariable analyses. Similarly, CS did not show a survival benefit over S-1 alone for GC patients after R1 resection of all visible tumors except CY1 in a large retrospective study in Japan[16]. Because CS was reported to be not feasible enough as adjuvant chemotherapy due to impaired oral intake immediately after curative surgery of GC, it is expected that present standard adjuvant doublet chemothrapy regimen such as S-1 plus docetaxel or S-1 plus oxaliplatin not using cisplatin may show better survival after curative resection of GC with O-meta.

Because neoadjuvant chemotherapy is more feasible than post-operative chemotherapy for GC patients, it is considered that peri-operative adjuvant chemotherapy is another treatment strategy for GC with O-meta. In Japan, preoperative chemotherapy followed by surgery with extensive lymph node dissection is recognized as standard treatment, because extensive lymph node metastasis such as multiple bulky regional lymph nodes and a few para-aortic lymph nodes (No. 16a2 and No. 16b1) are considered to be a resectable locoregional disease [5]. The results of the AIO-FLOT3 trial suggested a survival benefit in patients undergoing surgical resection of O-meta after neoadjuvant chemotherapy [24]. The ongoing AIO-FLOT5 (NCT02578368), which is a phase III trial comparing between peri-operative chemotherapy with FLOT (5-fuluorouracil, oxaliplatin and docetaxel) followed by gastrectomy and surgical resection of metastatic site, will answer the question about the optimal treatment strategy for O-meta with GC [25]. The combination of these cytotoxic agents with new agents, such as immune-checkpoint inhibitors and molecular-targeted agents, might be also a promising approach for future progress.

This study has several limitations. First one is selection bias. The elderly patients, worse PS, and pathological N category, which were known as poor prognostic factors, were significantly more in the Non-Cx group. To avoid selection bias, we adjusted the patient characteristics with multiple-variate analysis. However, it is difficult to adjust all factors, because our study was a small sample retrospective study. Second, the post-operative chemotherapy regimen was selected by an individual physician, leading to imbalances in patient characteristics among treatment groups. Third, the procedure of surgical resection and diagnosis of O-meta depended on each physician, and the chemotherapy dose and duration were not unified. Moreover, cases of resection of lymph node metastases identified during surgery were also included in the analysis. However, this study had not collected detailed diagnostic methods of lymph node metastasis for each patient. Thus, there might have been some differences in diagnostic and treatment approaches among the participating institutions. Finally, data regarding toxicities and quality of life were not collected during our study.

In conclusion, R0 resection followed by post-operative chemotherapy in patients with GC with synchronous O-meta showed favorable survival, and CS seems to have no additional benefit over S-1. However, the efficacy of post-operative chemotherapy remains unclear. Further development of a novel perioperative chemotherapeutic regimen or innovative multi-disciplinary treatment strategy that targets metastatic site is required to improve survival in patients with GC with O-meta.