Previous research has identified homologous recombination DNA repair deficiency (HRD) as a biomarker for sensitivity to poly( ADP-ribose) polymerase inhibitors (PARPi) and platinum-based therapies in patients with ovarian and breast cancers, wrote Niklas Krumm, MD, of the University of Washington, Seattle, and colleagues.
Currently, direct genetic testing is the most widely used method to identify mutations in HRD-associated genes, but not all genes underlying HDD have been identified, therefore current HRD assays that don’t rely on gene-specific information have been considered more diagnostically useful, the researchers noted. Two genetic tests are approved by the Food and Drug Administration, which are the FoundationFocus CDX BRCA and myChoice CDx, the researchers wrote. The Foundation Focus CDX BRCA was approved in 2016, and myChoice CDx was approved in 2019.
“However, transparent, well-defined methods and criteria for diagnosing HRD by genomic scarring that are practical for smaller, academic, or private laboratories have not yet been established or widely implemented,” they said.
In the paper published in JCO Precision Oncology, the researchers said they developed a molecular testing strategy involving the use of common, polymorphic single-nucleotide polymorphisms (SNPs).
They used a panel of approximately 3,000 SNPs distributed across the genome to create a loss of heterozygosity (LOH) score that could identify HRD.
To determine the ability of LOH to diagnose HRD in ovarian cancers, the researchers examined 99 ovarian neoplasm–normal pairs using the LOH method, and compared results with patient mutational genotypes and HRD predictors. LOH scores of 11% or higher showed greater than 86% sensitivity for identifying tumors with HRD-causing mutations in an independent validation set, and a sensitivity of 90.9% across training and validation sets.
When LOH scores were compared to a validated genome-wide mutational signature assay (HRDetect) the sensitivity and specificity of an LOH score of 11% or higher were estimated at 96.7% and 50%, respectively, for determining HRD-positive tumors.
However, the researchers found poor concordance (statistically insignificant correlation) using their LOH capture design to diagnose HRD based on mutational signatures only from targeted regions. “We conclude that mutational signatures inferred from our diagnostic tumor panel are unable to accurately ascertain HRD status, likely because the absolute number of somatic variants that it is able to identify is insufficient,” they said.
LOH scores were not significantly correlated with treatment outcomes, which suggests that LOH score can be used to infer HRD status, rather than serving as a direct predictor of patient response to primary platinum therapy, the researchers said. The average LOH score was higher in patients whose cancers responded to platinum therapy than in those with no treatment response (17% vs. 15%) but this difference was not significant.
Study limitations
The research was limited by several factors, including the validation only for high-grade non–clear cell ovarian carcinomas, and LOH scores likely vary across cancer types, therefore more studies will be needed to optimize the strategy for different cancers, the researchers noted. Other potential limitations include the high level of tumor cellularity needed (30%), which will eliminate some specimens, they said.