Abstract 3697: Cell-free DNA fragmentome dynamics as a biomarker for immune-checkpoint inhibition in advanced carcinoma

Background: Immune-checkpoint inhibitors (ICIs) have revolutionized cancer treatment, but a large proportion of patients is resistant to the treatment. Genome-wide cfDNA fragmentation profiles (fragmentome) represent a promising non-invasive approach able to monitor and predict response. Methods: The SChISM (Size cfDNA Immunotherapies Signature Monitoring) enrolled 128 advanced pan-cancer ICI-treated patients. Using Adelis technologies (France), based on the patented BIABooster system, pre- and on-treatment cfDNA size profiles were generated from the plasma, without prior DNA extraction. First, associations of pre-treatment fragmentome-derived metrics (e.g., concentration within multiple size ranges, position of peaks in the distribution) with early progression (EP, defined as progression at the first imaging evaluation) and progression-free survival (PFS) were assessed using logistic/Cox proportional hazard regression models. To assess predictive performances, data were split into training (n=90) and test (n=38) sets. Then, a mechanistic model for the joint dynamics of cfDNA and tumor lesions at the systemic level was developed. It relies on the following biological hypotheses: 1), wild-type cfDNA is primarily released through the apoptosis of hematopoietic cells, 2) cancer cells can be either resistant or sensitive to treatment, 3) cancer cells release a higher amount of DNA fragments due to the chaotic apoptosis-proliferation cycle characteristics of tumor cells, 4) cfDNA fragments are continuously cleared by liver and kidneys, with half-life ranging from 15min to 2h, 5) immunotherapy induces apoptosis in cancer cells, leading to the release of short cfDNA fragments, and 6) in the tumor microenvironment, cells undergo necrosis as the tumor grows, resulting in the release of longer fragments. Results; The ratio of long fragments (≥ 1650 base pairs (bp), LF) - normalized by total concentration - was significantly associated with non-EP (odds ratio = 0.3 (95%: 0.16-0.55), p <0.001) and longer PFS (p<0.001, hazard ratio 0.44 (95%: 0.31 - 0.64)). It also showed the best predictive performances of EP, on the test set: AUC = 0.74 (95% (CI): 0.65-0.82)), accuracy = 0.71 (95% CI: 0.62-0.77), positive predictive value 0.58 (0.46-0.69). The mechanistic model was able to accurately describe cfDNA-cancer dynamics under ICI, including steady decrease/increase of cfDNA concentration in either the long- or short-size ranges, but also non-trivial patterns, such as initial cfDNA concentration bumps. Conclusions: High-molecular-weight fragments are associated with early progression in ICI-treated patients. Mechanistic modeling offers biological insights for the interplay between cfDNA and tumor kinetics. Citation Format: Linh Nguyen Phuong,Frédéric Fina,Romain Zakrajsek,Laurent Greillier,Pascale Tomasini,Jean-Laurent Deville,Audrey Boutonnet,Frédéric Ginot,Jean-Charles Garcia,Sébastien Salas,Sébastien Benzekry. Cell-free DNA fragmentome dynamics as a biomarker for immune-checkpoint inhibition in advanced carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3697.