Abstract 5714: Cell-free DNA fragmentation profiling for monitoring therapeutic response in metastatic colorectal cancer

Abstract BACKGROUND: Cell-free circulating tumor DNA (ctDNA) assays have been adopted to monitor therapeutic response in both early- and late-stage cancer. However, tests currently available require deep-targeted sequencing to detect cancer-specific mutations at low mutant allele frequency (MAF) levels in the circulation. Recently, we developed a tumor-agnostic approach called DELFI Tumor Fraction (DELFI-TF), a Bayesian probabilistic model designed to predict plasma tumor fractions based on genome-wide fragmentation-related features. METHODS: Overall, 692 longitudinal plasma samples collected from 153 patients with initially unresectable colorectal cancer (CRC) liver-only metastases participating in the phase III study CAIRO5 (NCT02162563) were split across a training (RAS/BRAF mutant) and a validation cohort (RAS/BRAF wildtype). For all cell-free DNA (cfDNA) samples of patients within the training cohort (n=312), the tumor burden was quantified as the MAF of the tumor-tissue-proven RAS/BRAF variant measured by digital droplet PCR (ddPCR). Using cfDNA fragment-sequencing statistics from low-coverage WGS data, a Bayesian regression model was trained against the MAF of the tumor-specific driver RAS/BRAF variant in all longitudinal timepoints in the training cohort. Changes in DELFI-TF scores during first-line FOLFOX/FOLFIRI and bevacizumab therapy were assessed to predict disease response. A tissue-informed approach was applied to detect minimal residual disease in the subgroup of patients treated with curative intent (complete resection of the primary tumor and liver metastases). RESULTS: In the training cohort, DELFI-TF scores strongly correlated with RAS/BRAF MAF measured by ddPCR (r=0.85, p<0.001). DELFI-TF captured plasma tumor fractions associated with copy number changes when ddPCR failed to detect RAS/BRAF mutations. Before treatment initiation, DELFI-TF correlated with the sum of the largest diameters on CT scans according to RECIST1.1 (r=0.49, p<0.001). Patients with negative and positive changes in DELFI-TF more often exhibited response and progressive disease based on consecutive RECIST assessments, respectively. Median progression-free survival was 9.9 and 12.8 months for patients with DELFI-TF low versus DELFI-TF high, respectively (p=0.007). Overall survival was 28.8 and 58.9 months for patients with DELFI-TF low versus DELFI-TF high, respectively (p=0.001). In patients treated with curative intent, tissue-informed focal and arm-level copy number changes were detected 4-12 weeks after metastasectomy. Most patients presenting molecular relapse were later diagnosed with clinical recurrence by conventional CT imaging. CONCLUSIONS: DELFI-TF demonstrates the ability to use cfDNA fragmentomes to estimate cfDNA tumor burden with a performance comparable to standard approaches for treatment response monitoring and clinical outcome prediction. Citation Format: Bahar Alipanahi, Iris van 't Erve, Keith Lumbard, Laurel K. Millberg, Zach Skidmore, Lorenzo Rinaldi, Jacob Carey, Jennifer Tom, Cornelis J. A. Punt, Nicholas C. Dracopoli, Gerrit A. Meijer, Robert B. Scharpf, Victor E. Velculescu, Remond Fijneman, Alessandro Leal. Cell-free DNA fragmentation profiling for monitoring therapeutic response in metastatic colorectal cancer. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5714.