Personalized Surgical Decision-Making Model for Clinical Stage IA Pure-Solid Non-small Cell Lung Cancer.

Abstract

Background: Making an optimal surgical procedure decision (lobar or sublobar resections) remains challenging for some early-stage non-small cell lung cancer (NSCLC). This study aimed to evaluate the feasibility of developing a personalized model exclusively for surgical decision-making in early-stage NSCLC.

Methods: Propensity score matching was performed to diminish the influence of the surgeon's surgical procedure decision. Clinical and radiomic covariates were modified by the surgical procedure to reflect the interaction between the surgical procedure and covariates. The least absolute shrinkage and selection operator, Cox regression, was used for model development. Patients were divided into positive-score and negative-score groups based on a predicted score threshold of 0.

Results: After matching, 369 patients with clinical stage IA pure-solid NSCLC were included. Of the 248 modified covariates, 14 were selected, including four clinical and ten radiomic covariates. The surgical decision-making model generated a score for each patient (mean -0.43; standard deviation 0.97). In the positive-score group, sublobar resection was associated with significantly worse recurrence-free survival (RFS) than lobar resection (hazard ratio [HR], 4.15; 95% confidence interval [CI], 1.47, 11.7; P = 0.004). In contrast, sublobar resection was superior to lobar resection in terms of RFS (HR, 0.4; 95% CI, 0.22, 0.73; P = 0.002) for the negative-score group patients. For overall survival, lobar resection was favored in the positive-score group (P < 0.001), while sublobar resection was favored in the negative-score group (P < 0.04).

Conclusions: The personalized surgical decision-making model potentially helps to decide the optimal surgical procedure for early-stage NSCLC.