Leveraging genomic instability scores (GIS) to investigate BRCA variants of unknown significance (VUS) in ovarian cancer: An analysis of 14,513 patients in the Myriad Collaborative Research Registry (MCRR).

Abstract

e17549 Background: In ovarian cancer, use of homologous recombination deficiency (HRD) and BRCA mutation ( BRCAm) status is essential in guiding PARP inhibitor (PARPi) therapy, with the greatest survival benefit in patients harboring BRCAm pathogenic variants (PVs). In other indications, PARPi are only approved for patients with BRCAm or homologous recombination repair mutations (HRRm). Genomic instability scores (GIS), an HRD proxy, are also predictive of PARPi benefit in patients without BRCAm or HRRm, suggesting that other DNA alterations without definitive classification, such as BRCA VUS or variants of Special Interpretation (SI), may be contributing to HRD. Here, we evaluated the association of BRCA VUS and SI variants that have complex clinical interpretation with reduced penetrance. Methods: Retrospective analysis was conducted using de-identified genetic testing data from the MCRR. Eligible patients with ovarian, fallopian tube or primary peritoneal cancer were tested with the MyChoice CDx HRD test between 2016 and 2023, with HRD+ defined as GIS≥42 or presence of BRCAm PV. Patient BRCAm and HRRm status were defined using the most clinically significant variant observed in BRCA1/2 and an HRR 13-gene set, respectively. GIS distributions were compared using Kolmogorov-Smirnov tests and decomposed using mixture models. Results: Overall, 31.9% (4,628/14,513) of patients were HRD+ with 9.3% (N=1,344) having BRCAm PVs. An additional 3.7% (N=538) had a BRCA VUS [N=83 uncharacterized large rearrangement (LR), N=455 non-LR]. The distributions of GIS in both non-LR and LR VUS were significantly different from the distributions in BRCAm PVs and BRCA wildtype tumors (ps < 0.001), likely representing a spectrum of pathogenic and benign variants. Mixture models estimated that 28.8% [95% confidence interval (CI) 21.9, 35.8] of non-LR VUS and 55.8% (95% CI 40.0, 70.4) of LR VUS exhibited GIS similar to BRCAm PVs. Of the 15 non-LR VUS observed more than once, two were observed only in tumors with low GIS and are therefore presumed benign. BRCA SI variants (N=6) were observed in 16 patients and had a distribution of GIS significantly different from BRCA wildtype tumors (p < 0.001), but not significantly different from BRCAm tumors (p = 0.1). Two SI variants had high GIS in multiple patients and are therefore presumed pathogenic. One SI variant had low GIS in two patients and is therefore presumed benign. Conclusions: The BRCA VUS GIS distribution spanned across the BRCAm PV and BRCA wildtype GIS distributions, which suggests that BRCA VUS represent a mix of benign variants and PVs. In these patients, GIS is able to correctly classify HRD status. Additional data are needed to further characterize these uncertain BRCA variants and the appropriate therapeutic and preventive strategies required for these patients.