Association of radiomic features with disease-free survival following neoadjuvant chemoimmunotherapy in resectable NSCLC.

Abstract

8077 Background: Neoadjuvant chemoimmunotherapy (chemo-IO) has emerged as a promising approach to improve disease-free survival (DFS) in patients with surgically resectable non-small cell lung cancer (NSCLC). While recent clinical trials have demonstrated the efficacy of combining chemotherapy with immune checkpoint inhibitors in this setting, DFS remains variable among patients. Currently, there is no reliable biomarker to predict the risk of recurrence in this population. Predictors such as PD-L1 expression have shown limited utility, underscoring an unmet clinical need to identify robust biomarkers for DFS. This study investigates whether radiomic texture features derived from pre-treatment CT scans are associated with DFS in patients with NSCLC undergoing neoadjuvant chemo-IO prior to surgery. Methods: The study included 101 patients with NSCLC (median age: 66 years, range: 34–85) treated at the Cleveland Clinic. All patients received neoadjuvant platinum-doublet chemotherapy combined with an anti-PD-1 inhibitor prior to surgery. Radiomic features characterizing tumor heterogeneity were extracted from pre-treatment CT images. Patients were divided into a training set (St=50) and a validation set (Sv=51). A least absolute shrinkage and selection operator (LASSO) Cox regression model was used to identify prognostic features for DFS in St. A radiomic risk score (RRS) was computed as a linear combination of the selected features and their corresponding coefficients. High- and low-risk groups were determined based on the median RRS in St. A Cox regression analysis was performed to assess the impact of each factor on DFS. Kaplan–Meier survival analysis, accompanied by log-rank tests, was conducted to evaluate the prognostic performance of the biomarkers. Results: In a univariable analysis, the RRS was significantly associated with DFS in both St (HR = 2.77, 95% CI: 1.84 – 4.1, P < 0.0001) and Sv (HR= 2.28, 95% CI: 1.48 – 3.5, P = 0.0002). Kaplan-Meier analyses revealed significantly shorter DFS in the high-risk group compared to the low-risk group in both St (P < 0.0001) and Sv (P < 0.011). In a multivariable analysis that included clinicopathologic factors (age, race, tumor stage, and PD-L1 expression) along with RRS, both RRS and PD-L1 expression were significantly associated with DFS in St (RRS: HR = 2.99, 95% CI: 1.89 – 4.7, P < 0.0001; PD-L1: HR = 1.57, 95% CI: 1.14 – 2.15, P = 0.005). However, in Sv, RRS was the only factor significantly associated with DFS (RRS: HR= 2.31, 95% CI: 1.49 – 3.6, P = 0.0001), while PD-L1 expression was not (HR = 1.27, 95% CI: 0.35 – 4.6, P = 0.71). Conclusions: Identifying patients with locoregional NSCLC at risk of recurrence after neoadjuvant chemo-IO is crucial for effective treatment planning. Preliminary findings suggest that radiomic features hold promise as a reliable, non-invasive biomarker for risk stratification and guiding treatment decisions.