A cohort study using IL-6/Stat3 activity and PD-1/PD-L1 expression to predict five-year survival for patients after gastric cancer resection

OBJECTIVES: The expression/activation of IL-6, p-Stat3, PD-1 and PD-L1 in gastric cancer (GC) tissues were examined to evaluate their abilities in predicting the survival prognosis in postoperative patients with GC. METHODS: The clinicopathological data and paraffin-embedded tissues of 205 patients who underwent gastric cancer resection were collected at the First Affiliated Hospital of Shihezi University School of Medicine, and the patients were followed-up annually after surgery. Immunohistochemistry (IHC) was used to detect the expression of IL-6, p-Stat3, PD-1 and PD-L1 proteins using tissue microarrays derived from these patients. Statistical analyses were performed using non-parametric tests, Spearman’s correlation, ROC curves, Kaplan-Meier survival analysis, Cox single-factor and multifactor regression models. In comparison, the analyses were also performed for GC patients from public databases (407 patients from TCGA and 433 patients from GEO, respectively). RESULTS: (1) The expression levels of IL-6, p-Stat3, PD-1 and PD-L1 in GC tissues were significantly higher than adjacent normal tissues (ANT) (81.01% vs. 52.78%, P<0.001; 100% vs. 93.41%, P<0.001; 58.58% vs. 40.12%, P<0.001; 38.20% vs. 26.90%, P = 0.025, respectively). The mean optical density (MOD) values of IL-6, p-Stat3, PD-1 and PD-L1 were significantly higher in GC tissues. (2) The higher the levels of IL-6 (P<0.001), p-Stat3 (P<0.001), and PD-L1 (P = 0.003) were, the worse the survival prognoses were observed, respectively, among GC patients. The expression of PD-1 was not correlated with the prognosis of GC patients (P>0.05). The lower the degree of cell differentiation (P<0.001) was, the worse the survival prognoses were observed among GC patients. (3) Independent risk factors for postoperative prognosis in GC patients included age (≥60 years old), poor cell differentiation, invasion depth (T3/T4), lymph node metastasis (N1-3), distant metastasis (M1), and high levels of IL-6 (2+/3+). (4) A multi-factor combination (cell differentiation+IL-6+p-Stat3+PD-1+PD-L1) appeared to be the best survival predictor for GC patients as indicated by AUC (AUC 0.782, 95% CI = 0.709, 0.856, P<0.001). This combination may be the optimal predictor for postoperative survival of GC patients. (5) The levels of IL-6, p-Stat3, PD-1 and PD-L1 correlated with the infiltration levels of various tumor-infiltrating immune cells. (6) The analyses of ROC curves, calibration, DCA and Kaplan-Meier (KM) survival curves in TCGA dataset confirmed that the nomogram model could accurately predict the prognosis in GC patients. CONCLUSIONS: (1) The expressed levels of IL-6, p-Stat3, PD-1 and PD-L1 are higher in GC tissues than in adjacent normal tissues. (2) The high levels of IL-6, p-Stat3 and PD-L1 are correlated with poor survival in GC patients. (3) The high levels of IL-6, p-Stat3, PD-1 and PD-L1 have influences in GC tumor microenvironment. (4) The multi-predictor combination of "IL-6+p-Stat3+PD-1+cell differentiation" serves as an optimal survival predictor for postoperative GC patients and better than the TNM staging system. As these molecules can be examined in preoperative biopsies, these observations may provide a useful guide for clinicians to strategize individualized surgical plans for GC patients before surgery.