Development and validation of a comprehensive clinical risk-scoring model for prediction of overall survival in patients with endometrioid endometrial carcinoma

Abstract

ObjectiveTo develop and validate a comprehensive overall survival (OS) risk-scoring model in women with endometrioid endometrial cancer (EC). MethodsPatients with EC diagnosed from 2004 to 2013 were identified through the National Cancer Database (NCDB). Patients with known lymphovascular space invasion (LVSI) status who were treated surgically (with or without adjuvant therapy) were included. Cox proportional hazards analysis was used to identify prognostic factors for OS. This model was used to assign points based on hazard ratios for risk factors and a risk score was obtained. Recursive partitioning analysis (RPA) was used to categorize patients into risk groups. Results were internally validated in a cohort of patients from our institution (CCF cohort). Risk scores were calculated and assessed in a Cox regression model, and Harrell's c-index was calculated to assess model fit. ResultsAmong 349,404 women with EEC during the study period, 42,107 fulfilled inclusion criteria. Factors associated with worse OS were age ≥ 60, African American race, Charlson-Deyo score 1 or 2+, higher grade, LVSI, tumor size ≥2 cm, and no lymphadenectomy performed. Six risk groups were identified (scores 0–30) and OS estimated for each risk group. Risk score per 1-point increase in HR were comparable between NCDB and CCF cohorts (HR 1.21 (1.20–1.22 p < 0.001 vs 1.18 (1.12–1.25), p < 0.001), and c-index 0.80 (0.79–0.81) vs. 0.77 (0.68–0.86). Similar analysis was done in stage IA and IB. Adjuvant therapy had a beneficial effect on survival in the majority of stage IB patients, but only one of the six risk groups in stage IA EC. ConclusionsWe report a comprehensive validated OS risk-scoring model for patients with.