Nomogram for predicting long-term survival in renal cell carcinoma patients undergoing thermal ablation.

Abstract

Objective: Thermal Ablation (TA) represents a valid option for management of renal cortical neoplasms. Recognizing paucity of tools to predict overall survival (OS) for patients undergoing TA, we developed a nomogram to offer personalized OS predictions utilizing the National Cancer Database.

Methods: We included patients diagnosed with primary renal tumors who underwent TA between 2004 and 2020. Cox proportional hazards (CPH) model included age, Charlson-Deyo Comorbidity Index (CCI), tumor size, insurance status, ethnicity, histology, and tumor grade. A nomogram was developed to predict OS at 1, 5, and 10 years using a multivariable CPH model. Model robustness was confirmed through bootstrap validation with 1,000 iterations. Model performance was evaluated using Harrell's C-index, calibration plots at 1, 5, and 10 years, and time-dependent area under the curve (AUC) from ROC curves for 1-, 5-, and 10-year OS predictions RESULTS: We identified 10,121 patients (median age: 69 years; median follow-up: 55 months). Significant predictors of worse OS included advanced age (Hazard Ratio [HR] = 1.04, P < 0.001), higher CCI (HR = 2.20, P < 0.001), larger tumor size (HR = 1.03, P < 0.001), non-private insurance (HR = 2.16, P < 0.001), high-grade (HR = 1.31, P < 0.001), and clear cell (HR = 1.14, P = 0.015). Bootstrap validation confirmed the stability of the model, which achieved a C-index of 0.68. Calibration plots showed agreement between predicted and observed survival probabilities at 1, 5, and 10 years, with AUC values of 0.70, 0.71, and 0.74, respectively.

Conclusion: We constructed a nomogram incorporating clinical, pathological, and socioeconomic factors to offer personalized OS prediction for TA. Future research should focus on external validation and clinical implementation.