Enhancing Prediction of Pathological Complete Response to Neoadjuvant Immunotherapy in Non-small cell lung cancer by Whole Lung Radiomics

Objective To evaluate the utility of whole lung radiomics for predicting pathological complete response (pCR) to neoadjuvant immunotherapy in patients with non-small cell lung cancer (NSCLC). We further investigated whether integrating whole lung radiomics with tumor-specific radiomics could enhance the predictive performance for pCR. Methods Consecutive patients with stage II-IIIB NSCLC who underwent neoadjuvant immunotherapy were retrospectively enrolled between August 2018 and June 2025. The intra-tumoral and whole lung radiomics features were extracted from baseline CT images. Radiomics models were constructed using the LightGBM classifier. Model performance was assessed using receiver operating characteristic (ROC) curve, and recurrence-free survival (RFS) outcomes were evaluated via Kaplan-Meier survival analysis. Results A total of 284 patients (median age [IQR], 62 [57–67] years; 239 men) were enrolled and subsequently randomized into a training cohort (n = 199) and a test cohort (n = 85). Among them, 124 patients (43.7%) achieved pCR after neoadjuvant immunotherapy. The combined model, which incorporates 6 tumor-specific radiomics features, 3 whole lung derived radiomics features, and a clinical indicator (carcinoma embryonic antigen, CEA), achieved areas under the curve (AUC) of 0.713 (95% CI: 0.600–0.827), sensitivity of 83.3%, and specificity of 63.6% in the test cohort. Patients classified as low-risk group by the combined model exhibited significantly prolonged RFS (log-rank P  < 0.05). Conclusion The proposed whole lung radiomics approach demonstrates potential for predicting pCR to neoadjuvant immunotherapy in NSCLC. The combined model has capability to predict therapeutic response and prognosis prior to the initiation of neoadjuvant immunotherapy, enabling non-invasive personalized prediction.