Conditional Survival Nomogram After An R0 Resection For Gastric Cancer

11 Background: The risk of dying of cancer is highest in the first two years after a curative resection for gastric cancer. Therefore, the prognosis of patients who did not recur in the first two years is improved because they survived this critical period, a phenomenon called conditional survival. The US-derived gastric cancer nomogram predicts disease-specific survival (DSS) based on pathological variables. However, a disease-free interval after surgery, which improves the prognosis, is not captured by the nomogram. Therefore, it has only been used directly after surgery and not in the follow-up setting. The purposes of this study were to develop a conditional survival nomogram for 1, 2 and 3-year survivors (step 1) and to test if the introduction of follow-up variables would improve predictive accuracy of the nomogram in the follow-up setting (step 2). Methods: In a combined US-Dutch population of 1642 patients who underwent an R0 resection for gastric cancer and for whom the old nomogram variables were available, a conditional survival nomogram based on the original variables was developed for one (N=1147), two (N=879) and three (N=721) year survivors (step 1). To improve predictive accuracy in the follow-up setting, weight loss, performance status (PS), hemoglobin (HGB), and albumin (ALB) at one year after resection were retrospectively collected and added to the baseline variables in a new nomogram (step 2). Results: The conditional survival nomograms for 1, 2 and 3-year survivors (step 1) showed a high predictive accuracy in the calibration plots. Surviving one, two and three years shows a median improvement of 5-year DSS of 4%, 9% and 14% respectively. The introduction of weight loss, PS, HGB, and ALB at one year after surgery (step 2) did not improve this nomogram, but availability of these variables was limited. Conclusions: A strongly predictive conditional survival nomogram was developed, giving an improved prognosis for 1, 2 and 3-year survivors of gastric cancer. Introduction of variables available at one year after resection did not further improve this nomogram. This might be caused by the limited availability of follow-up data, as well as the strong predictive accuracy of the original variables. No significant financial relationships to disclose.