MSK-ACCESS powered with SOPHiA DDM: Performance analysis of a decentralized MSK-ACCESS solution.

e15063 Background: Blood plasma cell-free DNA (cfDNA) testing of cancer patients allows for genomic profiling when tissue sampling is too invasive or of insufficient quality. This approach has demonstrated clinical utility to identify treatment targets, screen for resistance and monitor disease burden. MSK-ACCESS (Memorial Sloan Kettering – Analysis of Circulating cfDNA to Evaluate Somatic Status) is a validated, high sensitivity cfDNA assay employed at MSK to identify biologically actionable tumor alterations in clinical care. SOPHiA GENETICS is developing a decentralized solution of this assay that incorporates major features of MSK-ACCESS such as molecular barcoding, duplex-based variant calling, the inclusion of tumor-informed prior knowledge and identification of somatic alterations through the removal of both germline and clonal hematopoiesis (CH) variants. Here, we present the results of a performance study obtained during product development. Methods: A cohort of 24 plasma cfDNA and matched white blood cell (WBC) DNA samples were collected from patients with solid tumors and processed at MSK according to the validated standard operating procedures and bioinformatic analyses. In parallel, for the same set of materials libraries were constructed at SOPHiA GENETICS using the proprietary CUMIN molecular identifiers, captured with probes targeting the updated MSK-ACCESS v2 set of 146 genes and sequenced on an Illumina NovaSeq 6000. Variants were called using a duplex-aware approach by SOPHiA GENETICS and germline and CH variants were identified and filtered by paired cfDNA – WBC DNA analysis. Accuracy of the decentralized “MSK-ACCESS powered with SOPHiA DDM” solution was assessed through concordance to the orthogonal results obtained by MSK. Results: A total of 104 somatic SNV/Indels were detected with variant allele fractions (VAF) ranging from 0.1% to 97.0% with a median of 5 variants per sample. The positive percent agreement (PPA) for all variants above the limit of detection of the original MSK-ACCESS assay was 98.9% with 99.98% specificity. For the subset of prior-knowledge variants PPA was 95.6% down to 0.1% VAF. A total of additional 59 variants with ≤ 15% VAF were found both in cfDNA and WBC and assigned as CH-derived. Importantly, these occurred not only in CH-associated genes but also in common oncogenes such as BRCA1/2 and ERBB2, highlighting the utility of matched WBC analysis to correctly identify somatic variants. Conclusions: The results demonstrate that the decentralized “MSK-ACCESS powered with SOPHiA DDM” solution exhibits a high degree of agreement to the results achieved by MSK. Further studies will be conducted to expand the analytical performance evaluation.