Introduction

Stage IV gastric cancer carries a poor prognosis. As per current guidelines [1, 2] the only therapeutic option for these patients could be palliative chemotherapy which may offer survivals up to 24 months [3]. However, there is growing literature reporting long survivals in those cases (stage IV gastric cancer) who responded to palliative chemotherapy and were subsequently submitted to gastrectomy [4,5,6,7,8,9,10,11]. In particular, this multimodal treatment, which was defined as conversion surgery, could be associated with significant improved survivals of more than 40 months when R0 resection could have been achieved [5, 11].

Still, literature about conversion surgery is scarce and studies are usually characterized by limited sample sizes and different definitions [7, 12]. In addition, to date, all but one study was performed on a cohort from a western center [7].

With this paper, we aim to report the experience of six GIRCG centers with conversion surgery, focusing our analysis on factors affecting survival and the risk of recurrence.

Materials and methods

Study design

A retrospective, multicenter cohort study was performed in patients who had undergone conversion gastrectomy April 1, 2005 and January 1, 2017 in 6 centers belonging to Gruppo Italiano Ricerca Cancro Gastrico (GIRCG) centers. Data were extracted from a GIRCG database including all metastatic gastric cancer patients submitted to surgery. Only stage IV GC which became resectable, (showing partial or complete response) after chemotherapy were included in this analysis.

Definitions

Patients’ diagnosis, treatment and perioperative care were performed as recommended by GIRCG guidelines in all participating centers [2]. Conversion surgery was defined as a surgical treatment aiming at R0 resection for tumors that were deemed unresectable before chemotherapy. As per study protocol, the definition of an unresectable tumor/metastasis before chemotherapy was based on technical and oncological reasons and included: peritoneal cancer index (PCI) (as defined by Sugarbaker et al. [13]) > 6, bilobar hepatic metastases, nodal involvement outside D1-3 stations, technically unresectable metastases.

Staging laparoscopy was performed in all patients before chemotherapy. The re-assessment for resectability was based on the imaging performed during treatment at 3–6 months. Surgery was proposed only to those patients in whom gastrectomy and associated resections could have been potentially radical. In particular, the peritoneal involvement was considered for treatment only in case of PCI < 6 at exploration after palliative chemotherapy.

All specimens were histopathologically classified according to Lauren’s microscopic criteria [14]. Tumor stage was presented as indicated by the Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) [15]. All cases were also grouped as proposed by Yoshida et al. [16].

Variables

Baseline characteristics collected included sex, age, American Society of Anaesthesiologists (ASA) score, type f preoperative chemo- and/or radiotherapy, preoperative staging (pTNM). Operative variables collected included type of procedure and lymphadenectomy, additional resections, intraoperative multimodal treatments and number of lymphnode harvested. Tumor-related variables included pathology stage (pTNM), resection margin status, tumor regression grade, Lauren’s histotype and grading. Complications were all collected up to the day of discharge and graded according to Clavien et al. [17]. Progression-free survival (PFS) and overall survival(OS) were the primary outcome measures.

Statistical analysis

Continuous data were presented as median and interquartile range (IQR). The Kaplan–Meier curve was used to calculate survival rates. OS was calculated as the time between both diagnosis and surgery and death/last follow-up. PFS was defined as the time between surgery and the first evidence of disease recurrence/progression and was calculated for all patients excluding R2 resections. Cox regression was performed to find independent factors affecting survival after surgery. Hazard ratios (HR) and 95% confidence intervals (95% CI) were calculated when required. Only variables with p < 0.050 at univariate analysis were entered in multivariable analyses in a stepwise manner. Follow-up was calculated as recommended elsewhere [18]. Analyses were performed with MedCalc Statistical Software (MedCalc Software bvba, Ostend, Belgium).

Results

In total, 287 stage IV GC patients were screened, of whom 242 were excluded for reasons (193 had no preoperative chemotherapy, 38 had preoperative chemotherapy for technically resectable disease, 11 had no details about preoperative period), leaving 45 patients for analysis. Patients’ characteristics at diagnosis are presented in Table 1. Reasons for being deemed unresectable at diagnosis were peritoneal involvement (n = 38, 84.4%), distant metastatic nodes (n = 3, 6.6%) and extensive liver involvement (n = 4, 8.8%).

Table 1 Patients’ characteristics at diagnosis
Full size table

Operative variables are shown in Table 2. In 33 (73.3%) cases peritoneal resections were required while in 3 these were not performed due to the extent of the disease. Two patients with peritoneal involvement had complete response after palliative chemotherapy and peritonectomy procedures were not performed as no residual disease was found. Postoperative complications occurred in 18 (40%) patients; and they had a Clavien-Dindo grade above 2 in 14 cases (31.1%). Tumor-related variables are presented in Table 3.

Table 2 Operative variables
Full size table
Table 3 Pathology variables
Full size table

Median follow-up was 25 months (IQR 9-50). Median overall survival was 15 months and 1-, 3- and 5-year survivals were 57.2, 36.1 and 24%, respectively (Fig. 1). Median OS calculated from diagnosis of GC was 20.6 months with 1-, 3- and 5-year survival of 78.2, 38.6 and 26.5%. Median PFS was 12 months with 1- and 3-year survival of 46.4 and 33.9%, respectively (Fig. 2).

Fig. 1
figure 1

Kaplan-Meier curve of overall survival for the whole cohort

Full size image
Fig. 2
figure 2

Kaplan-Meier curve of progression free survival

Full size image

At cox regression analysis the only independent prognostic factor for OS was the presence of more than one type of metastasis (HR 4.41, 95% CI 1.72–11.3, p = 0.002) (Table 4) (Fig. 3). A positive microscopic resection margin was the only risk factor for recurrence (HR 5.72, 95% CI 1.04–31.4, p = 0.045) (Table 5).

Table 4 Cox regression analysis of overall survival
Full size table
Fig. 3
figure 3

Kaplan-Meier curves of overall survival according to the type of extra-gastric involvement

Full size image
Table 5 Cox regression analysis of disease-free survival
Full size table

Discussion

Stage IV unresectable gastric cancer patients could benefit from surgery, which may offer prolonged survival.

More than 20 years ago, first, Nakajima et al. [8] described similar findings. Authors showed that unresectable GC could become resectable after intensive chemotherapy and found that long survivors were only in the group of patients who could have radical surgery after chemotherapy.

Since then, an increasing number of researchers tried to investigate the role of the combination of preoperative/palliative/induction chemotherapy followed by surgery in the treatment of advanced unresectable gastric cancer. Currently, literature demonstrated that so-called conversion surgery for unresectable stage III or stage IV gastric cancer was associated with longer survival than chemotherapy alone. The most recent studies on stage III/IV unresectable patients undergoing conversion surgery reported survivals ranging from 37 to 56 months [4, 5, 7, 9,10,11, 19,20,21,22].

The survival reported in our study is lower than those reported by others and this could be ascribed to the inclusion criteria we selected for our analysis. As such, our study population included only part of stage IV patients who had undergone chemotherapy plus surgery. In particular, we excluded from this study all stage IV GC patients who had chemotherapy plus surgery with PCI ≤ 6, unilobar technically resectable hepatic metastasis, only positive cytology or positive nodal metastases in D3 stations. As a consequence, our inclusion criteria identified a selected population composed mainly by Yoshida type 3 and 4 patients for whom an extraordinary response to chemotherapy could have led to the complete resection of the primary tumor and the peritoneal, nodal and liver metastases. As it has been highlighted by Yoshida et al. [16] and Yamaguchi et al. [11], peritoneal metastases in these cases may be technically resected but they often recur. In our study, a sub-population of Yoshida type 3 (macroscopic peritoneal dissemination without other organs involvement) patients showed unexpected survivals, being more than 35% alive 3 years after surgery. We believe that the results presented in this study may recommend not to exclude surgery a priori in selected advanced cases of stage IV GC.

In our analysis we found that more than one type of metastasis significantly affect prognosis. Similar findings were found by other authors [5, 10, 22], who showed that more than one non-curative factors were associated with a poor prognosis in unresectable GC patients. In light of those findings, it might be suggested not to proceed with surgery in those patients who have more than one district involved by metastases. Still, further data are required to confirm this hypothesis.

To date, the highest level of evidence about stage IV gastric cancer and the potential benefit of surgery in these patients is available thanks to the REGATTA trial [23]. This study showed that stage IV patients can benefit from surgery in terms of survival only when this is radical. Recently, these conclusions were also reached by the analyses of retrospective series of stage IV/unresectable patients undergoing induction chemotherapy followed by surgery [5, 11]. Yamaguchi et al. [11] in their multi institutional study on 77 stage IV GC patients who had undergone conversion surgery found that median survival was 41.3 for R0 patients while it was 21.2 months for R1-2.

Our study, which is the largest in the western literature, confirmed that R0 conversion surgery was associated with a significantly longer PFS than R1 resections. In our opinion, this latter finding further highlights the importance of radical surgery in the attempt to improve survival in metastatic GC patients.

Current literature about conversion surgery lacks of standardized definitions as it was also highlighted by Terashima [12]: the criteria for initial determination of oncologically unresectable or determination of resectability post-chemotherapy are heterogeneous among studies and results should always be interpreted in light of these differences. An effort from the scientific community would be required to find common definitions, and thus, reliable results in future studies.

The main limitation of this study is that the long study period covers several advances in oncology and, thus, in the chemotherapy regimens adopted. This resulted in a heterogeneous population in terms of type of chemotherapy used which made it difficult to understand the real impact of a regimen in the achievement of complete/partial response.

In conclusion, conversion gastrectomy is a treatment option for selected patients with stage IV GC. The main prognostic factor for these patients was the presence of more than one type of extra-gastric metastatic involvement. A radical procedure was significantly associated with a reduced risk of recurrence.