This study applied the Reporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement (16). We retrospectively collected data from patients who underwent surgery for OGC in Fondazione Policlinico Universitario “Agostino Gemelli”—IRCCS, Rome, between April 2012 and April 2022. All patients were aged 18 years or older, with a new diagnosis of advanced {stage IV according to American Joint Committee on Cancer (AJCC) 8th edition (17)} gastric or gastroesophageal junction adenocarcinoma, an Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≤ 2, and adequate baseline bone marrow, liver, and renal function. Tumors located proximally in the gastroesophageal junction were classified according to Siewert and Stein (18). Only Siewert type III tumors were included in the analysis. No one had received any previous chemotherapy for advanced disease and was considered eligible to start a doublet or triplet chemotherapy combination as a first-line treatment. All the radiological images and reports were revised at least twice, at diagnosis and preoperatively after induction chemotherapy by the multidisciplinary tumor board (MTB) to assess the TNM stage and disease response to medical treatment. The surgical indication for each OGC patient was discussed by our GC dedicated MTB and defined by the following criteria: (1) confirmed primary GC, (2) resectable hepatic metastasis, (3) minimal peritoneal dissemination [i.e., peritoneal cancer index (PCI) <6, uni- or bilateral Krukenberg tumors], (4) up to three lymph nodes involved in stations categorized as M1 according to AJCC manual 8th edition (17), (5) eligible for radical surgical treatment, and (6) in good general health conditions. Diagnostic laparoscopy and peritoneal cytology were performed to assess the presence of peritoneal dissemination in newly diagnosed advanced GC patients with suspected peritoneal dissemination before starting any kind of treatment as per routine clinical practice and PCI index was used to stratify patients according to living guidelines (4, 19). The decision about the chemotherapy administration was made by medical oncologists according to baseline disease and patient characteristics. The heterogeneity in treatment protocols and number of cycles performed were not considered for statistical analysis. Primitive tumor excision was combined with metastasectomy in one-time surgery. All patients with potentially curable lesions were treated by metastasectomy, gastrectomy, and extended lymphadenectomy (D2) on purpose, in agreement with current guidelines (4, 20). For tumors located in the middle and lower thirds of the stomach, a subtotal gastrectomy was generally preferred, provided that an adequate resection margin was maintained. After total gastrectomy with lymph node dissection, esophagojejunostomy (using a circular stapler, diameter 25 mm) was used routinely for Roux-en-Y reconstruction. In case of subtotal gastrectomy, intestinal continuity was restored by means of Billroth II or Roux-en-Y gastrojejunostomy, at the discretion of the surgeon. Resection was stated as potentially curative (R0 according to the UICC/AJCC staging system (17), if macro- and microscopically no tumor was left following surgery). At the end of the operation, the surgeon resected all lymph nodes from the surgical specimen and identified their distribution according to the Japanese Gastric Cancer Association classification (20). Pathological TNM classification, the status of margins and histological tumor regression grade (TRG) assessed in accordance with the Mandard criteria (21) were described in the pathologists’ report. The patients were monitored for up to 30 days by a surgeon to assess postoperative complications and mortality. All clinical and pathological data were stored in a GC database and retrospectively evaluated for this study. Patient status was investigated by follow-up examination or by telephone contact.

The primary endpoint was overall survival (OS) defined as the time from diagnosis to death or latest follow-up; the secondary endpoint was recurrence-free survival (RFS) defined as the time from surgery to radiological evidence of relapse. An exploratory analysis was performed to evaluate the survival impact of demographic and clinicopathological factors such as age, sex, ECOG PS, histotype (according to Lauren’s classification) (22, 23), primitive tumor site (esophagogastric junction, fundus/body, and antrum/pylorus) and extension (T), metastatic sites (liver, lymph nodes, and peritoneum) and burden (single metastatic site versus multiple sites), resection margins (R0 versus R1–2), and TRG according to the Mandard criteria (1–3 versus 4–5) (21, 24). Furthermore, we tested a prognostic score based on five relevant prognostic factors: (1) pathological tumor growth (T+) on surgical specimen; (2) presence of liver metastases (L+); (3) peritoneum (P+) or (4) non-locoregional lymph nodes (N+); and (5) presence of macroscopical residual disease on resection margin (R2 resection).

Statistical analyses were performed using SPSS^® software, version 29.0 Chicago, IL. The survival curves were generated by the Kaplan–Meier method and compared using the Log-rank test. Demographic and clinicopathological factors were collected and evaluated by univariate analysis among patients who received macroscopically complete resection to evaluate their impact on OS and RFS. Discrete variables were compared using the Chi-square test and continuous variables were compared using independent-samples t-test. p-value <0.05 was considered statistically significant. Only variables that were statistically significant at univariate analysis were tested in multivariate analysis with Cox proportional hazards model to identify independent predictors of specific survival and recurrence.

In the past 10 years, 39 patients with metastatic GC, 16 men and 23 women, matching the aforementioned criteria had undergone surgery with radical intent at our institution. Patients’ age at diagnosis ranged from 26 to 75 years, with a median age of 58 years. PS ECOG was 0–1 for 33 patients and 2 for 6 patients. Primary tumor site was located at the gastroesophageal junction in 8 cases, at the gastric body in 17 cases, and at the antrum/pylorus in 14 cases. All subjects had a histopathologically confirmed diagnosis of adenocarcinoma according to Lauren’s classification (22). The specimens of 14 patients were diagnosed as intestinal subtype, 15 were diffuse, 6 were mixed type, and 4 were indeterminate type. Signet ring cells have been detected in 13, all in patients with a diffuse subtype. Among the population, 18% of patients were found positive for HER2 amplification score ≥ 2+. All patients had a baseline clinical T4 stage except one T3, and most of them had a locoregional nodal involvement (N1–N2). A total of 34 study patients had only one metastatic site: 6 of them presented with only liver metastases, 22 presented with peritoneal metastases, and 6 presented with distant lymph nodes ( Table 1 ). The others had ≥2 synchronous metastatic sites ( Table 2 ). All of them were considered eligible to first-line chemotherapy and received 2–6 months of induction systemic platinum and fluoropyrimidine combination chemotherapy, with or without trastuzumab, according to histology and HER2 status (4, 5) ( Table 3 ). Any case was referred to our MTB and considered amenable to radical surgery after the induction chemotherapy.

Surgery was conducted in both primitive tumor and metastases at the same time. Regarding surgical approach to primitive tumor, for 59% of patients, a total gastrectomy was performed, while in 41%, a subtotal gastrectomy was carried out. Four (10.2%) patients with GEJ adenocarcinoma underwent a thoraco-abdominal gastrectomy, while in all other cases, a laparoscopic abdominal surgical approach was used. In 28 (71.8%) cases, a laparoscopic technique was preferred, whereas in 11 (28.2%) cases, patients underwent a laparotomy. Up to 87% of cases underwent a lymphadenectomy ≥ D2; in five cases, a D1 lymphadenectomy was performed. A total of 19 patients (49%) presenting peritoneal-only metastatic spread underwent hyperthermic intraperitoneal chemotherapy (HIPEC). HIPEC was carried out according to the Coliseum technique (25) using mitomycin C (MMC) at a dose of 15 mg/m² and cisplatin at a dose of 75 mg/m² administered for 90 min with an inflow temperature of 41–42°C and an outflow temperature of 39–40°C (26). An R0 resection was achieved in 27 patients (69%), while microscopic (R1) or macroscopic (R2) residual disease was found in 7 (18%) and 5 (13%) cases, respectively ( Table 4 ). TRG evaluated according to the Mandard criteria was available for 26 patients: 58% of them were pathologic responders (TRG 1–3) and 42% were non-responders (TRG 4–5). Combined surgery on both gastric primitive tumor and metastasis clearly affected operation duration but did not negatively affect patients’ outcome in terms of mortality and performance status. No major intraoperative complications or surgery-related deaths occurred in our series. During the hospitalization, up to 12 (30.7%) patients developed at least one postoperative complication. According to the Clavien–Dindo classification (27), the most frequent were grade I/II complications, while two patients (5.1%) reported a grade III complication. One patient reported an abdominal collection treated by percutaneous drainage plus antibiotic therapy; another experienced postoperative bleeding treated by angioembolization. Only one subject (2.5%) reported a grade IV complication: an acute renal failure requiring hemodialysis. A summary of postoperative complications is available in the Supplementary Materials ( Supplementary Table S1 ). Actually, the greater surgical stress related to an extensive resection did not deteriorate prognosis in patients without postoperative complications (69.2%). After surgery, 26 patients (66.7%) were able to receive an adjuvant treatment ( Table 3 ).

At the data cutoff analysis of November 2022, 17 patients were alive and, among them, 7 were free from any disease recurrence. Recurrence rate (RR) after surgery was 72%: in 78% of cases, disease relapse occurred in the peritoneum (22 cases); 11% (3) of patients progressed with liver metastases; one patient had a recurrence on both sites; one had only distant lymph node relapse; one patient progressed with multiple metastasis also involving brain and lungs. After a median follow-up of 33.6 months, median OS was 26.6 months (95% CI 23.8–29.4) ( Figure 1A ) and median RFS was 10.6 months (95% CI 6.3–14.8) ( Figure 2A ). Estimated 3-year OS was 13%. At univariate analysis, factors that significantly influenced OS were a good pathologic response according to the Mandard criteria [TRG 1–3, not reached (NR) versus 20.5 months for TRG 4–5, HR 0.23, p = 0.019] ( Figure 1B ), PS ECOG ≤ 1 (26.7 versus 11.2 months, HR 0.3, p = 0.022) ( Figure 1C ), and the number of metastatic sites: 26.7 months for patients with a single site versus 12.9 months for those with ≥2 sites (HR 0.34, p = 0.039) ( Figure 1D ). Having a single metastatic site was also the only variable that statistically significantly influenced RFS (12.8 vs. 4.4 months, HR 0.10, p < 0.001) ( Figure 2B ). The multivariate analysis confirmed having a single metastatic site as the only factor that statistically significantly influenced OS (HR 0.08, p = 0.048) but not RFS ( Supplementary Tables S2 , S3 ). Other clinicopathological features such as age, sex, histological type, HER-2 status, T-stage, N-stage, and primitive tumor site did not influence the benefits seen in these patients. In a study by Samarasam et al., patients were stratified according to four prognostic factors, demonstrating how the survival advantage of patients who underwent surgical resection disappeared when ≥3 negative factors were present (28). Thus, we tried to test an analogue score adding a relevant fifth prognostic factor: (1) pathological tumor growth (T+) on surgical specimen; (2) presence of liver metastases (L+); (3) peritoneum (P+) or (4) non-locoregional lymph nodes (N+); and (5) R2 resection. Moreover, in our population, there was a statistically significant difference between patients with a score of 1–2 and patients with a score of ≥3 in terms of both OS (26.7 vs. 12.9 months, HR 0.31, p = 0.023) ( Figure 1E ) and RFS (12.8 vs. 4.4 months, HR 0.10, p < 0.001) ( Figure 2C ).