Assessing homologous recombination deficiency (HRD) in ovarian cancer: Optimizing concordance of the regulatory-approved companion diagnostic and a next-generation sequencing (NGS) assay kit.

e17571 Background: Olaparib + bevacizumab is approved as first-line maintenance treatment of ovarian cancer that is HRD positive, defined as the presence of a deleterious/suspected deleterious BRCA mutation (BRCAm) and/or genomic instability. HRD genomic instability score (GIS) has been assessed in clinical trials using a regulatory-approved companion diagnostic, Myriad myChoice ® CDx. We evaluated the performance of an in-development NGS assay kit that identifies BRCA variants based on genomic content from Illumina’s TruSight™ Oncology 500 research assay, and, with additional genomic content, measures HRD GIS in tumor tissue (Illumina test) in parallel. Analytic concordance of the Illumina test versus the Myriad myChoice ® PLUS assay (Myriad test) under improved algorithms for calculating GIS scores is reported. Methods: Ovarian cancer tissue samples were analyzed with the Illumina (40 ng DNA; N = 227) and Myriad (200 ng DNA; N = 254) tests. Agreement rates for BRCA analysis, HRD GIS, and overall HRD status (includes BRCAm and HRD GIS [cutoff, 42]) were analyzed. For the overall and the non-BRCAm analysis cohorts, Pearson correlation between the continuous HRD GIS of the Illumina and Myriad tests was determined. Analytic sensitivity and specificity of the Illumina-derived HRD GIS to classify HRD GIS (cutoff, 42) were evaluated using AUROC. After initial concordance calculations, modifications were made to increase the quality of the Illumina preprocessing data (ie, reducing allele dosage noise and allele bias) upstream of the GIS calling algorithm. Reprocessing of the existing raw data led to further improvement in the point estimates of positive and negative percent agreement. Results: Prevalence using the Illumina and Myriad tests was 51.2% and 49.2% (overall HRD positive) and 27.6% and 25.5% (BRCAm), respectively. Agreement rates are reported in the Table. Correlation of the Illumina HRD GIS with Myriad HRD GIS was 0.980 (all samples) and 0.976 (non-BRCAm cohort); AUROCs were 0.995 and 0.992, respectively. Conclusions: Comparison between the Illumina and Myriad tests showed that overall HRD status, BRCA analysis, and HRD GIS detection results were > 90% concordant; HRD GIS was highly correlated (r > 0.98); and prevalence estimates were similar. These data suggest that a distributable kit solution, such as the Illumina test under development, will approach the performance of the Myriad myChoice HRD assay.[Table: see text]