Validation of a gastric cancer nomogram using a cancer registry

Background A Memorial Sloan Kettering (MSKCC) nomogram predicts disease specific survival (DSS) for gastric adenocarcinoma. The goal of this study is to use a cancer registry to compare nomogram predicted survival with actual survival in the general population. Methods All patients undergoing surgery for gastric adenocarcinoma from the Surveillance, Epidemiology, and End Results (SEER) database (1988-2012) were studied. Results 6954 patients were identified. Majority of cancers were in the antrum (30.2%), and had intestinal histology (73.7%). Median follow-up was 8.2 years. Five year DSS for nomogram risk groups (0-25%, 26-50%, 51-75%, and 76-100%) was 23%, 48%, 57%, and 81% respectively. Actual DSS was 7-15% lower than nomogram predicted DSS. Relative to patients in the 76-100% 5-year DSS risk group, patients in the 0-25%, 26-50%, and 51-75% groups had significantly higher risks of death with hazard ratios of 6.84 (95%CI 6.12-7.65), 3.30 (95%CI 2.83-3.86), and 2.64 (95%CI 2.30-3.03), respectively (all P < 0.001). The concordance index for 5-year nomogram predicted DSS was 0.68 (95%CI 0.67-0.69). Conclusions The MSKCC gastric cancer nomogram over-estimates DSS from gastric cancer in the general population and has a moderate concordance index. Predictive tools generated at specialized institutions may not perform as well in the general population. J. Surg. Oncol. 2015; 112:377-380.