A prognostic PET radiomic model for risk stratification in non-small cell lung cancer: integrating radiogenomics and clinical features to predict survival and uncover tumor biology insights.

purposeTo develop a survival risk score using

methodsWe utilized several NSCLC cohort datasets from the Cancer Imaging Archive (TCIA) for radiomic analysis, where transcriptomics data were available through the Cancer Genome Atlas (TCGA). A total of 945 radiomic features were extracted from the segmented tumors. A survival-based radiomic model was developed, from which a radiomic risk score was calculated. Radiogenomic analyses were then performed to explore correlations between radiomic risk cohorts and tumor transcriptomics, oncogenic pathways, and genetic mutations. We also constructed a nomogram by combining clinical and radiomic risk factors.

resultsThe PET-radiomic model significantly predicted the 5-year survival rate of patients, with AUCs of 0.78, 0.71, and 0.73 in the training, validation, and testing cohorts, respectively. Integration of clinical features and the radiomic risk score in a nomogram demonstrated enhanced efficacy, achieving AUCs greater than 0.85. Radiogenomic analysis revealed that while the low-risk group indicated anti-tumor immunity, the high-risk group exhibited transcriptomic characteristics associated with enhanced tumor aggressiveness, with consistent correlations between risk group membership, oncogenic pathways, immune cell types, and critical gene alterations.

conclusionPET-radiomic features successfully delineated high- and low-risk NSCLC patient groups. Supporting radiogenomic analysis identified tumor-promoting characteristics and immune-suppressing activity in the high-risk group, which is consistent with these patients' prognoses.