Deep-Learning Serial CT Prediction of Survival in Immunotherapy-Treated Non–Small Cell Lung Cancer

This prognostic study developed and validated Serial CT Response Score (Serial CTRS), a fully automated deep learning biomarker that uses pretherapy and 12-week follow-up CT scans to predict overall survival in patients with advanced non-small cell lung cancer receiving immune checkpoint inhibitor therapy. Across 1,830 patients drawn from real-world clinical practice datasets spanning 10 US and European institutions and an independent multinational clinical trial of dostarlimab, Serial CTRS consistently outperformed standard imaging measures including RECIST and tumor volume change in discriminating survival outcomes. The biomarker demonstrated strong prognostic value across programmed death-ligand 1 and RECIST subgroups, including patients with stable disease who are particularly difficult to risk-stratify with conventional tools.

These findings address a significant unmet need in oncology, where reliable early response biomarkers for immunotherapy remain elusive despite the growing importance of immune checkpoint inhibitors across cancer care. By achieving superior predictive accuracy using the same CT scans already collected in routine practice, Serial CTRS offers a scalable and resource-efficient alternative to existing measures, with meaningful implications for clinical decision-making, treatment monitoring, and the design of future clinical trials evaluating immunotherapy response in lung cancer.