Thymic health on routine chest CT predicts who benefits from cancer immunotherapy
Mar 18th 2026
A deep-learning CT tool that measures thymic health predicts progression-free and overall survival after immune checkpoint therapy across cancers, and correlates with blood and tumour measures of T cell output and diversity, but needs broader validation before clinical use.
- A deep-learning model scored thymic health from standard chest CTs and was applied to 3,476 patients treated with immune checkpoint inhibitors across multiple cancer types.
- Patients with average or high thymic health had significantly lower risk of progression and death after immunotherapy, with high versus low thymic health showing hazard ratios about 0.65 for progression-free survival and 0.56 for overall survival in NSCLC.
- Thymic health added prognostic information independent of established tumour biomarkers and performed similarly to PD-L1 and tumour mutational burden when adjusted for clinical factors.
- Biological validation in a 464-patient NSCLC cohort showed higher thymic health correlated with higher sjTREC levels, greater T cell receptor diversity, and higher blood and tumour T cell fractions.
- Associations were strongest in first-line immunotherapy and weaker after prior chemotherapy, suggesting treatment line affects the signal.
- Key limitations are predominantly white study populations, the need for external and scanner-level validation, and the absence of a non-immunotherapy control to prove predictive value