Comparison of genomic instability test scores used for predicting PARP activity in ovarian cancer.

Abstract

1586 Background: Clinical trials have explored the utility of various genomic instability (GI) scores or gene panels to assess deficiencies in the homologous recombination (HR) DNA repair pathway and support PARP inhibitor use in ovarian cancer; however, these methods of assessing homologous recombination deficiency (HRD) may not be equivalent. The myChoice HRD test is the only analytically and clinically validated, FDA-approved HRD test that includes BRCA1/2 mutation status and three measures of GI. We compared the proportion of patients identified as candidates for PARP inhibitor use by two measures of HRD [percent loss of heterozygosity (%LOH), 11-gene panel] to myChoice HRD. Methods: Whole-genome SNP analysis was used to reconstruct ovarian tumor genomic profiles to calculate the myChoice HRD score and %LOH in 2 cohorts (clinical laboratory cohort, N = 3,278; SCOTROC4 trial, N = 248). Mutation screening for a set of 11 genes in the HR pathway ( ATM, BARD1, BRCA1, BRCA2, BRIP1, CHEK2, MRE11A, NBN, PALB2, RAD51C, RAD51D) was performed for a subset of tumors from the SCOTROC trial (n = 187). Samples were considered positive if the myChoice HRD score was above the threshold (threshold scores of 42 and 33 were assessed), %LOH above the threshold (16%), or a pathogenic variant in one of the 11 HR genes. The correlation between positive results from %LOH and the 11-gene panel were compared to myChoice HRD. Percent positive agreement (PPA) was the proportion of positive test results from myChoice HRD that were also positive by %LOH or the 11-gene panel. Results: The table shows the correlation and PPA between myChoice HRD, %LOH, and the 11-gene panel. Overall, 19%-61% of patients identified as positive by myChoice HRD would have been missed by %LOH or the 11-gene panel in these two cohorts. Conclusions: These data show that HRD tests used in published and ongoing clinical trials are not equivalent, and they should not be considered interchangeable in predicting PARP inhibitor response in clinical practice. [Table: see text]