Back

Circulating Biomarkers for Therapeutic Response to Immune Checkpoint Inhibitor Therapy in Patients with Advanced Lung Cancer

Jain, K.; Mehra, D.; Yadav, M.; Aggarwal, S.; Chugh, P.; tripathi, G.; Ganguly, S.; Rana, R.; Awasthi, A.; Ganguly, N. K.; Maras, J. S.

2025-05-24 cancer biology
10.1101/2025.05.21.655239 bioRxiv
Show abstract

Immune checkpoint inhibitors (ICIs) have demonstrated remarkable efficacy in the treatment of advanced lung cancer (ALC), but not all patients benefit. To identify responders, patient selection, categorization, and the identification of specific predictive biomarkers are crucial. Our research aims to identify circulating metabolites that could be used to stratify patients into potential responders prior to the initiation of ICIs and to monitor clinical response in real-time during treatment. In the discovery phase, plasma samples (n = 25) from ALC patients receiving ICIs were analyzed using unbiased metabolomics and validated in a distinct patient cohort (n = 39). Five potential biomarkers were identified (FC > 1.5, p 0.05, AUC 0.81-1), including Thiourea and a combined metabolite panel (CMP) consisting of sphingosine-1-phosphate, gentisic acid, glutathione, and 4-hydroxybutanone. Early on-treatment high plasma levels of Thiourea were significantly associated with 5-year progression-free survival [Hazard ratio (HR) = 0.038, 95% confidence interval (CI) = 0.013-0.107, p<0.001] and overall survival [HR = 0.048, 95% CI = 0.019-0.120, p<0.001]. Low levels of CMP early on treatment were significantly associated with progression-free survival [HR = 8.119, 95% CI: 3.767-17.501, p<0.001] and overall survival [HR = 7.367, 95% CI: 3.517-15.433, 3.517-15.433, p<0.001]. This predictive plasma metabolite panel could serve as promising circulating biomarkers for predicting the therapeutic response of ICIs in ALC patients.

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.