Prediction of programmed death-ligand 1 expression levels in non-small cell lung cancer patients based on 18F-fluorodeoxyglucose PET/computed tomography radiomics features

OBJECTIVE: This study explored the predictive value of 18 F-fluorodeoxyglucose (FDG) PET/computed tomography (CT) radiomics for assessing programmed death-ligand 1 (PD-L1) expression in non-small cell lung cancer (NSCLC), aiming to noninvasively evaluate PD-L1 status and assist in selecting patients for immunotherapy. METHODS: We retrospectively analyzed 163 NSCLC patients with pretreatment 18 F-FDG PET/CT scans, randomly assigning them into training ( n = 130) and validation ( n = 33) cohorts. Optimal radiomics features were selected via least absolute shrinkage and selection operator and combined with clinical factors to construct five predictive models: CT, PET, radiomics, clinical, and a combined model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). RESULTS: All models showed predictive ability for PD-L1 expression. The combined model demonstrated superior performance, with AUCs of 0.839 [95% confidence interval (CI): 0.771-0.908] in training and 0.782 (95% CI: 0.610-0.954) in validation. Calibration curves indicated good agreement between predicted and observed probabilities (Brier scores: 0.163 and 0.191, respectively). DCA confirmed the highest net clinical benefit for the combined model. CONCLUSION: The multimodal combined model, integrating PET/CT radiomics with clinical factors, shows significant potential for noninvasively predicting PD-L1 expression in NSCLC, offering a novel strategy for precise patient selection for anti-PD-L1 immunotherapy.