Cell-free DNA fragmentome profiles predict immunotherapy response

Mutation-based liquid biopsy approaches face challenges in accurately identifying the cellular origin of variants, which complicates the determination of circulating tumor DNA (ctDNA) molecular response. We hypothesized that these limitations could be addressed by utilizing genome-wide cell-free DNA (cfDNA) fragmentome methods. To this end, we performed whole-genome sequencing of 528 serial cfDNA samples and quantified the cfDNA tumor fraction (TF) using DELFI-TF, a tumor- and mutation-naive machine learning cfDNA fragmentome classifier, across two prospective clinical protocols for immunotherapy-treated non-small cell lung cancer (NSCLC). DELFI-TF values were validated independently by hybrid capture fixed gene panel next-generation sequencing, and the landmark cfDNA fragmentome molecular response (fMR) was determined within 3-9 weeks from the initiation of immunotherapy. Individuals with fMR attained significantly longer progression-free and overall survival than those without fMR (logrank p=2.7e-14 and p=3e-9), including in patients with mutation-based molecular progression (logrank p=1.2e-4 and p=2e-3). Our findings suggest that analyses of the cfDNA fragmentome can provide a clinically meaningful measure of circulating tumor fraction and validate the clinical sensitivity of plasma-only cfDNA fragmentome-based molecular response as an early indicator of immunotherapy response.

• Type: Cancer Genomics
• Archive: European Genome-phenome Archive (EGA)