Abstract 1948: Real-world insights into the accuracy and precision of the decentralized MSK-ACCESS powered with SOPHiA DDM solution

Abstract Objectives: Liquid biopsy (LBx) is a powerful, non-invasive approach to identify treatment selection or resistance markers, assess treatment efficacy, monitor disease, and predict relapse. To facilitate genomic profiling of cell-free DNA (cfDNA) in cancer SOPHiA GENETICS has developed a decentralized version of MSK-ACCESS (Memorial Sloan Kettering Cancer Center - Analysis of Circulating cfDNA to Evaluate Somatic Status). The MSK-ACCESS powered with SOPHiA DDM solution interrogates 146 cancer-associated genes for somatic variants and exhibits a high concordance to the original single-site test. It incorporates white blood cell (WBC) sequencing to reduce the risk of including germline and clonal hematopoiesis (CH) alterations in the tumor variant set. To evaluate the real-world performance of MSK-ACCESS powered with SOPHiA DDM an observational, multi-center study was initiated to evaluate accuracy (concordance to orthogonal tests and tumor tissue analyses) and precision (repeatability & reproducibility). Methods: In participating laboratories sequencing libraries were generated from peripheral blood plasma cfDNA and WBC DNA samples and a common set of reference materials using MSK-ACCESS powered with SOPHiA DDM and sequenced on Illumina NovaSeq 6000 or NextSeq 2000. Variant detection and annotation of germline and CH variants were performed with the SOPHiA DDM platform. Accuracy was determined in the clinical samples through the concordance to orthogonally reported variants, tumor tissue genomic profiling results and to the reference cfDNA variants. Repeatability and reproducibility were measured through reference material variant allele frequency (VAF) variability. Results: In clinical samples 94.1% of reported SNVs/Indels with an expected VAF ≥ 0.5% were detected (159/169, positive percent agreement per site: 71.4% - 100%). Across all externally reported variants (n = 246), 58 were not enriched in cfDNA compared to WBC DNA and annotated as germline or CH. The average percent agreement in the 0.5% VAF reference samples was 99.1%. The coefficient of variation of the 0.5% VAF samples had the same magnitude within intra-lab repeatability (0.29, mean VAF 0.48, SD 0.14) as for inter-lab reproducibility comparisons (0.29, mean VAF 0.48, SD 0.14). Conclusion: Interim real-world results of the MSK-ACCESS powered with SOPHiA DDM solution illustrate its accuracy in clinical samples. By distinguishing cfDNA from CH/germline origin variants, potential biological false positives were removed. Further analyses of patient-matched tissue will be used to establish concordance between cfDNA and tumor tissue genomic profiles and investigate causes of discordance, including biological, technical and protocol factors. The assay’s observed precision demonstrates that site-to-site discordance, a major barrier to the adoption of LBx testing, has been addressed by this solution. Citation Format: Florian Klemm, 1 Alex Tuck, 1 Fuad Mohammad, 1 Xiaobin Xing, 1 Anne Pradines, 2 Laura Keller, 2 Daniel Kazdal, 3 Markus Ball, 3 Morten Lapin, 4 Oddmund Nordgård, 4 Persephone Du-Parcq, 5 Alasdair Heasman, 5 Sasha Hansel, 5 Julia Stevenson, 6 Gareth Gerrard, 7 Justin Wheelan, 8 Eric Antoniou, 8 Rehan Faridi, 9 Faisal Khan, 9 Leah Thompson, 10 Thomas Stricker, 10 Cristovam Scapulatempo-Neto, 11 Marie Piecyk, 12 Lea Payen, 12 Paul Hofmann, 13 Zhenyu Xu1. Real-world insights into the accuracy and precision of the decentralized MSK-ACCESS powered with SOPHiA DDM solution [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 1948.