Background In a phase III study for advanced gastric cancer (AGC), the Japan Clinical Oncology Group (JCOG) 9912 trial, we previously identified the following four prognostic factors-performance status >= 1, number of metastatic sites >= 2, no prior gastrectomy, and abnormal serum alkaline phosphatase levels-and proposed a prognostic index (good risk with 0 or 1 factor, moderate risk with 2 or 3 factors, and poor risk with all 4 factors). To assess the generalizability of this index, we attempted an external validation study using an independent data set.
Methods Individual patient data from the SPIRITS and G-SOX trials were applied to the JCOG prognostic index. The accuracy of the index for predicting survival was assessed by the Cox proportional hazards model.
Results The available data were obtained from 936 (94.5%) of the 990 patients in these trials. The three risk groups categorized by the JCOG prognostic index demonstrated highly significant survival differences; the hazard ratios (95% confidence interval) were 1.71 (1.46-2.01) between the good (n = 338) and moderate (n = 537) risk groups and 3.32 (2.47-4.46) between good and poor (n = 61) risk groups. The median overall survival times of the good, moderate, and poor risk groups were 17.2, 12.0, and 7.8 months, respectively.
Conclusions The JCOG prognostic index was externally validated and can be widely utilized for clinical trials. Further studies are needed to apply this index to the Western population.