br AJCC stage groupings br
3.2. AJCC stage groupings
3.3. Modified prognostic groupings
Based on the aforementioned findings, an optimized staging system was constructed. Herein, the pT4aN0M0 category was incorporated into the modified stage IIIA from stage IIB, because its survival data mat-ched with stage IIIA. As obvious from the Kaplan-Meier plot, there were no overlapping survival curves among all the eight stage groups in the AJCC modified 8th edition (m8th edition) staging system (Fig. 4). Significant differences in the 5-year survival rate were observed be-tween the modified stage IIB and IIIA (72.4% vs. 61.0%, P = 0.031).
(caption on next page)
Fig. 1. The data extraction diagram.
Clinic-pathological characteristics and the univariate and multivariate survival analysis in gastric cancer patients.
The details of NCT-501 distribution of the AJCC 8th edition staging system (except the patients of staging IV, n = 1622).
The AJCC 8th edition staging system N0
Table 4 showed the prognostic impacts of the TNM system on sur-vival in the multivariate Cox proportional hazards models by adjusting for the effects of the covariates. Compared with stage IA, the hazard ratio (HR) of the 8th edition stage IIB did not significantly differ from that of stage IIIA (8.07 vs. 8.27, P = 0.844). With respect to the mod-ified staging system, the HR of the patients in stage IIB was lower than that of the patients in stage IIIA in comparison with stage IA (5.55 vs. 8.65, P = 0.036).
3.4. Prognostic ability of the staging system
The performance of the 8th edition and the m8th edition staging system was assessed by the linear trend χ , the likelihood ratio χ , and the AIC and BIC tests (Table 5). Compared with the 8th edition system, the m8th edition staging system had a better homogeneity (higher likelihood ratio χ  score), discriminatory ability, and monotonicity of the gradients (higher linear trend χ2 score). Moreover,
the m8th edition staging system had smaller AIC and BIC values, sug-gesting the optimum prognostic stratification.
According to the AJCC 8th edition staging system, there were four categories staged IIB including pT1N3aM0, pT2N2M0, pT3N1M0, and pT4aN0M0. The last category had serosal invasion only. In our study, we found uniformitarianism there were similar survival curves between stages IIB and IIIA (P = 0.516), while the Kaplan-Meier plot showed no overlapping survival curves among the other stages. Upon further analysis, the distance between the curves of pT4aN0M0 and IIIA became narrower in the OS Kaplan–Meier curves (Fig. 2), reflecting similar mortality risks for the patients categorized as pT4aN0M0 and staged IIIA. Significant differences in the 5-year survival rate were observed between the ca-tegories of pT1N3aM0/pT2N2M0/pT3N1M0 versus pT4aN0M0 (P = 0.036) and the categories of pT1N3M0/pT2N2M0/pT3N1M0
Fig. 2. The 5-year survival analysis for all the stages according to the AJCC 8th edition staging system.
versus stage of IIIA (P = 0.041), with the exception of the category of pT4aN0M0 versus stage of IIIA (P = 0.693). These finding revealed that the prognosis of the patients categorized pT4aN0M0 was similar to IIIA, but not pT1N3M0/pT2N2M0/pT3N1M0. The prognosis of the patients categorized pT4aN0M0 was worse than those categorized as pT1N3aM0/pT2N2M0/pT3N1M0, due to its
Fig. 4. The 5-year survival analysis for all the stages according to the AJCC m8th edition staging system.
Multivariate survival analysis of the TNM staging system.