The study developed a non-invasive biomarker to predict durable clinical benefit from anti-PD-1/PD-L1 immunotherapy in patients with advanced NSCLC by integrating longitudinal clinical data, blood test variables, and CT-derived radiomics/deep-radiomics features. Using a retrospective cohort of 264 stage IV NSCLC patients, the combined model significantly outperformed models based on a single data source, achieving AUCs of 0.824 and 0.753 for predicting durable benefit at 6 and 9 months, respectively. The resulting signatures also successfully stratified patients into high- and low-risk groups and were significantly associated with both progression-free and overall survival, highlighting the value of combining longitudinal clinical and imaging biomarkers for treatment response prediction.

Farina, B., Guerra, A. D. R., Bermejo-Peláez, D., Miras, C. P., Peral, A. A., Madueño, G. G., Jaime, J. C., Vilalta-Lacarra, A., Pérez, J. R., Muñoz-Barrueta, A., Peces-Barba, G. R., Maceiras, L. S., Gil-Bazo, I., Gómez, M. D., & Ledesma-Carbayo, M. J. (2023). Integration of longitudinal deep-radiomics and clinical data improves the prediction of durable benefits to anti-PD-1/PD-L1 immunotherapy in advanced NSCLC patients. Journal of translational medicine, 21(1), 174.