A total of 4,221 eligible GC patients with definite organ metastases (liver, lung, bone, brain) were identified and included in the present study based on the inclusion and exclusion criteria. Of these, 3312 patients had single-site metastasis while 909 patients presented multiple-site metastases. In the overall patient cohort, there were 3133, 1041, 956, and 156 patients with liver, lung, bone, and brain metastases, respectively. Notably, the most common metastasis of single-site was liver (70.5%), followed by bone (15.8%), lung (11.9%), and brain (1.8%). The median age at diagnosis was 65 years. The clinicopathological characteristics of patients with metastatic GC are summarized in Table 1. A total of 457 patients underwent surgical resection, of these, surgical resection for primary GC resection was performed in 303 patients and in 218 GC patients the resection of metastasis was performed. In the subgroup of patients with primary tumor resection, the percent of patients with bone metastasis was the lowest, approximately 4.0%. While in the subgroup of patients with metastatic tumor resection, the percent of patients with liver metastasis was found to be the lowest, approximately 3.7%. Furthermore, a total of 2796 patients received chemotherapy and 972 patients were administered with radiotherapy.
OS and CSS in patients with advanced GC were evaluated based on single-site or multiple-site metastases and surgeries performed on the primary or metastatic tumors. The results indicated that advanced GC patients with single-site bone metastasis (p= 0.03 for CSS), brain metastasis (p= 0.03 for CSS), or liver metastasis (p< 0.001 for CSS) were benefited the most from primary surgery (p< 0.05) (Fig. 2). Furthermore, the evaluation of metastatic GC patients undergoing resection for metastatic lesions revealed that patients with lung metastasis (p= 0.007 for CSS) and liver metastasis (p= 0.02 for CSS) were benefitted significantly from surgery than patients with bone metastasis and brain metastasis (Fig. 3). However, there was no significant difference between patients who underwent resection for primary and metastatic lesions in GC patients with multiple-site metastases (Supplementary Fig. 1).
In the cohort of patients who received chemotherapy, advanced GC patients with the brain (p=0.02 for CSS) or liver (p<0.001 for CSS) metastasis were benefitted significantly from surgery of the primary tumor (Supplementary Fig. 2). However, patients with lung (p=0.01 for CSS) or liver (p=0.02 for CSS) metastasis benefitted significantly from surgery of the metastases (Supplementary Fig. 3).
The median OS for advanced GC patients with single-site bone, brain, lung, and liver metastases were 5, 4, 6, and 6 months, respectively. Advanced GC Patients with isolated lung or liver metastasis exhibited better CSS and OS compared with GC patients with bone metastasis (p =0.001 for CSS and p=0.002 for OS in lung metastasis; p =0.002 for both CSS and OS in liver metastasis). However, no significant difference was observed between patients with single-site lung and liver metastases (p>0.05 for both CSS and OS). There was no significant difference between patients with single-site bone and brain metastases. (p>0.05 for both CSS and OS) (Fig. 4).
Furthermore, the prognostic value of the pattern of metastases was also evaluated for stage IV GC patients. The median OS for patients with one, two, three, and four metastatic sites was 6, 4, 4, and 3 months, respectively. Patients with single-site metastasis exhibited survival benefits over patients with multiple-site metastases (p < 0.001 for both CSS and OS). However, no significant difference was observed between patients with two, three or four metastatic sites (p >0.05 for both CCS and OS) (Supplementary Fig. 4).
Analysis of prognostic factors using Multivariable Cox regression
Multivariate Cox regression analysis was performed to evaluate the association of significant variables with the prognosis of metastatic GC. The results indicated that both CSS and OS were significantly improved in GC patients who underwent primary resection (HR, 0.48; 95% CI, 0.41-0.56; p< 0.001) and those who received chemotherapy (HR, 0.37; 95% CI, 0.34-0.40; p< 0.001). Lung metastasis (HR, 0.77; 95% CI, 0.67-0.89; p< 0.001) and liver metastasis (HR, 0.82; 95% CI, 0.74-0.92; p= 0.001) was significantly associated with improved CSS and OS of GC patients compared with bone metastasis (Table 2). In the overall cohort, patients aged > 65, undifferentiated cancer, signet ring cancer, who were single, or had multiple-site metastases were found to be statistically significant independent prognostic factors for poor survival, while primary surgical treatment and chemotherapy were protective factors. However, in the single-site metastasis cohort and the overall patient cohort, there was no significant association between the GC patients who underwent metastatic surgery and CSS/OS. (Supplementary Table 1)
Prognostic factors associated with surgery
In a multivariable logistic regression model, patients with gastric cardia cancer (OR, 3.36; 95% CI, 2.37-4.83; p< 0.001), who were unmarried (OR, 1.75; 95% CI, 1.27-2. 42; p< 0.001), had signet ring cell cancer (OR, 1.82; 95% CI, 1.12-3.04; p= 0.018), or who underwent chemotherapy (OR, 1.59; 95% CI, 1.17-2.16; p= 0.003) were less likely to undergo primary surgery. Considering bone metastasis as a reference, GC patients with brain (OR, 0.20; 95% CI, 0.07-0.58; p= 0.002), lung (OR, 0.48; 95% CI, 0.25-0.91; p= 0.025) or liver (OR, 0.52; 95% CI, 0.30-0.86; p= 0.015) metastasis were more likely to receive primary tumor resection (Table 3). In the metastatic tumor resection cohort, GC patients older than 65 years (OR, 1.44; 95% CI, 1.03-2.02; p= 0.036) or male patients (OR, 1.61; 95% CI, 1.35-2.27; p= 0.007) were less likely to receive metastatic tumor resection. In addition, GC patients with mucinous adenocarcinoma of the stomach (OR, 0.27; 95% CI, 0.13-0.61; p < 0.001) and GC patients who underwent primary resection (OR, 0.13; 95% CI, 0.08-0.19; p < 0.001) were more likely to receive metastasis resection. Compared to GC patients with liver metastasis, GC patients with brain metastasis (OR, 0.07; 95% CI, 0.04-0.14; p < 0.001), lung metastasis (OR, 0.37; 95% CI, 0.24-0.58; p < 0.001) or bone metastasis (OR, 0.44; 95% CI, 0.28-0.69; p < 0.001) were more likely to undergo metastatic surgery (Supplementary Table 2).