Abstract A016: cfDNA fragmentome predicts cancer treatment response similarly to mutant allele fraction

Abstract Background: Monitoring response and progression in late-stage cancer patients presents unique challenges; conventional imaging does not often reflect tumor or clinical response early in treatment. Liquid biopsy assays may predict treatment response earlier than radiographic imaging, with mutation-based sequencing having been adopted as the benchmark to measure changes in circulating tumor DNA (ctDNA). Maximum somatic variant allele frequency (MSAF) assays, however, are complicated, costly, may require tumor tissue sequencing, and may be uninformative where ctDNA yield is limiting. These hurdles have delayed the widespread adoption of ctDNA dynamics into clinical practice and could be mitigated by inexpensive, non-mutation-based whole-genome sequencing approaches. Methods: We analyzed longitudinal plasma samples (n=53) from patients with metastatic rare cancer (n=14) receiving anti-neoplastic agents on two NCI clinical protocols (NCT02211755 and NCT04449549). From all samples, we performed comprehensive genomic profiling (CGP) of cfDNA using a 523-gene targeted panel (∼20 ng) and, in parallel, utilized 1 ng of cfDNA to measure ctDNA dynamics through low-coverage whole-genome sequencing (LC-WGS, ∼4x). The Illumina DRAGEN TSO 500 ctDNA analysis software v2.1 was used to calculate MSAF from CGP data, with manual review to exclude potential clonal hematopoietic (CH) variants and considering the longitudinal feature of the data, while the DELFI Tumor Fraction (DELFI-TF) algorithm was implemented from the LC-WGS data. DELFI-TF is a tumor- and mutation-independent approach that integrates cfDNA fragmentation patterns and aneuploidy across the genome by machine learning, resulting in a fragmentome-based tumor fraction. DELFI-TF scores and the MSAF of tumor-specific variants were tracked over time and clinical outcomes were assessed by RECIST 1.1; clinical responders had a Partial Response (PR) or Stable Disease (SD); non-responders had Progressive Disease (PD). Results: Tumor fraction data measured by DELFI-TF were highly concordant with MSAF measured by CGP for all 53 samples (Pearson, R=0.92, p < 2.2e-16). 77.8% of patients experiencing treatment response showed a significant reduction in DELFI-TF and CGP (both assays detected ctDNA reduction of at least 50%) in 7 out of 9 responders at first tumor restaging C2D1 and following on-treatment timepoints. Conversely, 75% of patients who experienced progressive disease (3 out of 4 non-responders) exhibited stable tumor fraction measurements captured by DELFI-TF or CGP. Several cases showed an increase in DELFI-TF and CGP tumor fraction earlier than progressive disease as measured by RECIST, confirming the lead time of liquid biopsy assays compared to conventional imaging. Conclusions: Analysis of the cfDNA fragmentome offers an opportunity to measure tumor burden longitudinally in a non-invasive, clinically meaningful, and cost-effective manner. Citation Format: Lorenzo Rinaldi, Li Chen, Chris Karlovich, Jamie Medina, Zachary Skidmore, Michael Rongione, Bryan Chesnick, Nicholas C. Dracopoli, Alice Chen, James Doroshow, Brooke Augustine, Murielle Hogu. cfDNA fragmentome predicts cancer treatment response similarly to mutant allele fraction [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2025 Oct 22-26; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2025;24(10 Suppl):Abstract nr A016.