Identified optimal candidates for pulmonary resection in octogenarians with non-small cell lung cancer: a web-based predictive model.

ABSTRACT

Background: A survival benefit from pulmonary resection was observed in octogenarians with non-small cell lung cancer (NSCLC). Meanwhile, the identification of patients who can indeed benefit can be difficult. Therefore, we aimed to establish a web-based predictive model to identify optimal candidates for pulmonary resection. Methods: Octogenarians with NSCLC in Surveillance, Epidemiology and End Results (SEER) database were enrolled and split into the surgery and non-surgery groups based on whether they received pulmonary resection. Propensity-score matching (PSM) was utilized to eliminate the imbalance. Independent prognostic factors were identified. Patients in the surgery group who lived longer than the median cancer-specific survival (CSS) time of the non-surgery group were assumed to benefit from the surgery. The surgery group was further divided into the beneficial group and the non-beneficial group based on the median CSS time of the non-surgery group. Among the surgery group, a nomogram was established through a logistic regression model. Results: A total of 14,264 eligible patients were extracted, with 4,475 (31.37%) patients receiving pulmonary resection. Surgery was an independent favorable factor of prognosis after PSM (median CSS time 58 vs. 14 months, P<0.001). A total of 750 (70.4%) patients lived longer than 14 months (beneficial group) in the surgery group. Factors including age, gender, race, histologic type, differentiation grade, and tumor-node-metastasis (TNM) stage were used to formulate the web-based nomogram. The precise discrimination and predictive capability of the model were validated through receiver operating characteristic curves, calibration plots, and decision curve analyses. Conclusions: A web-based predicted model was constructed to distinguish specific patients who can indeed benefit from pulmonary resection among octogenarians with NSCLC.