External Validation of a Nomogram for Predicting Survival of Women With Uterine Cancer in a Cohort of African American Patients

Objective This study aimed to externally validate a nomogram for predicting overall survival of women with uterine cancer in an African American population. Methods After the institutional review board approval, data from the uterine cancer database from 2 major teaching hospitals in Brooklyn, NY, were analyzed. The predicted survival for each patient was calculated with the use of the nonogram; the data were clustered in deciles and compared with the observed survival data. Results High incidence of aggressive histologic types (22% carcinosarcoma, 16% serous/clear cell), poorly differentiated (53% grade 3), and advanced stage (38% stage III or IV) tumors was found in our study population. The median follow-up for survivors was 52 months (range, 1-274 months). The observed and predicted 3-year overall survival probabilities were significantly different (62.5% vs 72.6%, P < 0.001). Similarly, the observed 5-year overall survival probability was significantly lower than the predicted by the nomogram (55.5% vs 63.4%, P < 0.001). The discrepancy between predicted and observed survival was more pronounced in the midrisk groups. Conclusions The nomogram is not an adequate tool to predict survival in the African American population with cancer of the uterine corpus. Race seems to be a significant, independent factor that affects survival and should be included in predictive models.