Development of a ¹⁸F-FDG PET/CT-based radiomic risk score as an independent prognostic factor in small cell lung...

Prognosis in small cell lung cancer (SCLC) remains poorly predicted by conventional staging and metabolic parameters. This study aimed to develop a positron emission tomography/computed tomography (PET/CT) based Radiomic Risk Score (Rad-Score), evaluate its independent prognostic value for overall survival (OS), and assess the clinical utility of an integrated model using Decision Curve Analysis (DCA).

143 SCLC patients (120 male, 23 female) who underwent pre-treatment staging ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET/CT between January 2012 and June 2024 were included. Using LIFEx (v25.06.1), 129 Image Biomarker Standardisation Initiative (IBSI) compliant radiomic features were extracted alongside standard metabolic parameters (SUVmax, SUVmean, MTV, TLG). After eliminating highly correlated features (|r|> 0.90), five key features were selected via tenfold cross-validation LASSO-Cox regression to construct the Rad-Score. Patients were stratified into Low and High Radiomic Risk groups. Independent prognostic factors were identified through Cox regression analyses, and the incremental value of the Rad-Score was assessed by hierarchical C-index comparisons and DCA.

Median OS was 10.8 months (95% CI: 8.2-13.3), with a mortality rate of 83.9%. Multivariable analysis identified the Radiomic Risk Group (HR: 2.12; 95% CI: 1.33-3.40; p = 0.002) and Curative CRT (HR: 0.17; 95% CI: 0.06-0.48; p < 0.001) as independent prognostic factors. The C-index improved from 0.657 to 0.688 (p = 0.018) and to 0.701 with Rad-Score integration (p = 0.0017). DCA demonstrated superior net benefit at high threshold probabilities (≥ 65%).

The ¹⁸F-FDG PET/CT-based Rad-Score independently predicts survival in SCLC beyond conventional staging and metabolic parameters. The integrated model provides meaningful incremental prognostic value; however, multicenter external validation is warranted prior to clinical implementation.