HRD-One: CLINICAL VALIDATION AND PERFORMANCE ASSESSMENT. Comparison between Myriad's myChoice(R), SOPHiA GENETICS SOPHiA Homologous Recombination Solution(R) and AmoyDx HRD Focus Panel(R)

Abstract

Homologous Recombination Repair (HRR) testing has become increasingly important in clinical genomic labs due to the use of poly-ADP-ribose polymerase (PARP) inhibitor therapy for epithelial ovarian, fallopian tube, or peritoneum cancer. While sequencing and copy number variation analysis can identify patients with a pathogenic mutation in BRCA1 or BRCA2 who can benefit from PARPi therapy, there are also patients who may benefit but do not have these mutations. To address this, our lab has developed a test called HRD-One, in partnership with SOPHiA GENETICS, that can detect sequence variants in genes involved in HRR, as well as genomic scars that indicate Homologous Recombination Deficiency (HRD), which may be present even when a pathogenic variant is not detected. We tested 59 high-grade serous epithelial ovarian cancer samples using HRD-One and found that it had an overall categorical concordance of 94.74% with Myriads myChoice(R) score, which is a commercial HRD test. 12 out of 13 samples that carried a pathogenic or likely pathogenic variant in BRCA1/2 also had a positive HRD-One score, and 9 samples in which a pathogenic variant in BRCA1/2 was not identified had a positive score in both HRD-One and myChoice(R). Of the samples that passed quality control, we observed an average of 1.62 points variation between replicates on a scale from -25.0 to +25.0. We also found that low-confidence results were associated with a low DNA input and the age of FFPE blocks, while the estimated tumor percentage in the block, NGS library yield, and score of genomic instability did not have a significant association. We determined that blocks older than 3 years or with a DNA input of less than 25ng are not reliable for producing high-quality results. Finally, we validated the HRD-One test with SOPHiA Homologous Recombination Solution (Library Prep kit II) and correlated it to myChoice(R), and found that the AmoyDx(R) HRD Focus Panel had the same sensitivity but a higher number of false positive samples and therefore lower specificity. Overall, we have shown that HRD-One can provide a reproducible and concordant score for inferring HRD, and an HRD-One score of 2.0 or greater predicts HRD and correlates to Myriads myChoice(R) score of 42 in high-grade serous epithelial ovarian cancers samples that meet our minimum quality criteria.