Development and External Validation of a Novel Nomogram Predicting Cancer-specific Mortality–free Survival in Surgically Treated Papillary Renal Cell Carcinoma Patients

Abstract

BackgroundAccurate prediction of cancer control outcomes in renal cell carcinoma (RCC) patients is important for counselling, follow-up planning, and selection of appropriate adjuvant trial designs. ObjectiveTo develop and externally validate a novel contemporary population-based model for predicting cancer-specific mortality–free survival (CSM-FS) in surgically treated papillary RCC (papRCC) patients and to compare it with established risk categories (Leibovich 2018). Design, setting, and participantsWithin the Surveillance, Epidemiology, and End Results database (2004–2019), we identified surgically treated papRCC patients (n = 3978). The population was randomly divided into development (50%, n = 1989) and external validation (50%, n = 1989) cohorts. Of the external validation cohort, 97% (n = 1930) of patients were included in a head-to-head comparison of the Leibovich 2018 risk categories addressing nonmetastatic patients. Outcome measurements and statistical analysisUnivariable Cox regression models tested the statistical significance in the prediction of CSM-FS. The most parsimonious model with the best validation metrics was selected as the multivariable nomogram. Accuracy, calibration, and decision curve analyses (DCAs) tested the Cox regression–based nomogram, as well as the Leibovich 2018 risk categories in the external validation cohort. Results and limitationsAge at diagnosis, grade, T stage, N stage, and M stage qualified for inclusion in the novel nomogram. In external validation, the accuracy of the novel nomogram was 0.83 at 5 yr and 0.80 at 10 yr. In nonmetastatic patients, 5- and 10-yr accuracy of the novel nomogram was 0.77 and 0.76, respectively. Conversely, 5- and 10-yr accuracy of the Leibovich 2018 risk categories was 0.70 and 0.66, respectively. The novel nomogram exhibited smaller departures from ideal predictions in calibration plots and higher net benefit in DCAs, when it was compared with the Leibovich 2018 risk categories. Limitations include the retrospective nature of the study, absence of a central pathological review, and inclusion of only North American patients. ConclusionsThe novel nomogram may represent a valuable clinical aid, when papRCC CSM-FS predictions are required. Patient summaryWe developed an accurate tool to predict death due to papillary kidney cancer in a North American population.