Molecular profiling of plasma from matched NCI-MATCH gynecological cancers.

5579 Background: During the screening phase of the National Cancer Institute Molecular Analysis for Therapy Choice (NCI-MATCH; NCT02465060) trial, 5954 patients with advanced cancer, approximately 60% of whom had rare or less common cancers, underwent next-generation sequencing of their tumor tissue to determine study eligibility. The majority of the patients could not be assigned to a treatment arm because they did not have a study mutation of interest (MOI). The plasma collected from these patients may provide insight into the genomic landscape of circulating tumor DNA (ctDNA) and inform the potential utility of ctDNA testing for the identification of clinically relevant mutations in these rare histologies. Here we report the molecular profiles of ctDNA from a subset of the NCI-MATCH screened patients with gynecologic cancers. Methods: Cell-free DNA (cfDNA) was extracted and quantitated from plasma collected in Streck blood tubes at study enrollment. Libraries were constructed using the NCI ctDNA v2 assay with a minimum input of 10 ng and were sequenced on the Illumina NovaSeq 6000 with S4 flow cells. Results: Among 200 samples sequenced, four different histologies were represented: uterine endometrioid adenocarcinoma (UEC, n=51), serous endometrial adenocarcinoma (USC, n=35), uterine leiomyosarcoma (ULMS, n=40), and ovarian epithelial (OVT, n=74). Two samples failed, one ULMS due to sequencing error and one OVT due to contamination. The positive percent agreement (PPA) values calculated from clinically relevant MOI derived from OncoKb as both PPA refₜumor and PPA refctDNA, were 88. 6% and 44. 1% respectively. The most commonly mutated genes were TP53, KRAS, CTNNB1, and PIK3CA. Additionally, the ctDNA detected mutations not identified in the tumor, including 8 patients (4%) with a PIK3CA mutation. By employing predefined cutoffs of tumor fractions, ctDNA identified 21 out of 25 expected copy number amplifications demonstrating an 84% concordance. Microsatellite instability (MSI) using ctDNA was highest in the UEC (17. 6% of patients) followed by USC at 8. 6%, OVT at 6. 8%, and ULMS at 2. 6%. Among the patients with mismatch repair deficiency (MMRd) identified by tumor immunohistochemistry, 50% were identified as MSI-High in plasma. Conclusions: In this cohort of patients with advanced gynecological cancers from the NCI-MATCH trial, clinically relevant mutations detected by ctDNA were 88. 6% concordant with data derived from tumor biopsy. In addition, some mutations were identified only in ctDNA, which may be due to tumor heterogenity. The data suggest that liquid biopsy is valuable as a complement to tumor biopsy data and when tissue is unavailable.