Genome-wide variation in cell-free DNA end-motif entropy predicts immunotherapy response in head and neck cancer.
Summary
A novel study developed a non-invasive biomarker, the regional motif diversity score (rMDS), based on whole-genome sequencing of plasma cell-free DNA (cfDNA). This biomarker predicts immunotherapy response in patients with head and neck squamous cell carcinoma (HNSCC). The rMDS demonstrated superior performance compared to existing metrics in distinguishing responders from non-responders to pembrolizumab. Longitudinal rMDS changes were associated with genomic regions linked to immunity and keratinization, suggesting a connection to telomere biology. An rMDS-based classifier achieved a high AUC (0.89-0.99) and was associated with improved disease-free survival.
Analysis
Clinique: This prospective, multi-institutional phase II trial identifies rMDS as a predictive biomarker for immunotherapy response (pembrolizumab) in HNSCC patients. The rMDS demonstrated the ability to predict improved disease-free survival (HR 2.67), suggesting clinical actionability and potential to refine patient selection. It outperformed PD-L1 expression and tumor fraction, which could justify its integration into future risk assessment frameworks, with potential clinical impact in the medium term (3-5 years) after confirmation by phase III studies. Biomol: The main discovery is rMDS, a novel fragmentomic metric quantifying the entropy of cfDNA 5'-end motifs. The technology used is whole-genome sequencing (WGS) of plasma cfDNA, a non-invasive approach. The study analyzed longitudinal plasma samples, which is crucial for tracking disease and treatment response dynamics. rMDS is biologically meaningful, localizing to regions enriched for immune- and keratinization-related genes, and suggests a link to telomere biology. Bioinfo: The study developed rMDS, a novel fragmentomic metric, and an rMDS-based machine learning classifier to predict response. This model was validated across multiple settings, achieving an AUC of 0.89-0.99, and was favorably benchmarked against established biomarkers like PD-L1 expression and tumor fraction. The classifier's robustness and performance suggest potential for clinical deployment to refine patient selection for immunotherapy, potentially replacing or complementing less performant methods.