Association of BRCA1/2defects with genomic scores predictive of DNA damage repair deficiency among breast cancer subtypes

Kirsten M Timms, Victor Abkevich, Elisha Hughes, Chris Neff, Julia Reid, Brian Morris, Saritha Kalva, Jennifer Potter, Thanh V Tran, Jian Chen, Diana Iliev, Zaina Sangale, Eliso Tikishvili, Michael Perry, Andrey Zharkikh, Alexander Gutin, Jerry S Lanchbury

IntroductionHomologous recombination (HR) DNA repair is of clinical relevance in breast cancer. Three DNA-based homologous recombination deficiency (HRD) scores (HRD-loss of heterozygosity score (LOH), HRD-telomeric allelic imbalance score (TAI), and HRD-large-scale state transition score (LST)) have been developed that are highly correlated with defects in BRCA1/2, and are associated with response to platinum therapy in triple negative breast and ovarian cancer. This study examines the frequency of BRCA1/2 defects among different breast cancer subtypes, and the ability of the HRD scores to identify breast tumors with defects in the homologous recombination DNA repair pathway.Methods215 breast tumors representing all ER/HER2 subtypes were obtained from commercial vendors. Next-generation sequencing based assays were used to generate genome wide SNP profiles, BRCA1/2 mutation screening, and BRCA1 promoter methylation data.Results BRCA1/2 deleterious mutations were observed in all breast cancer subtypes. BRCA1 promoter methylation was observed almost exclusively in triple negative breast cancer. BRCA1/2 deficient tumors were identified with BRCA1/2 mutations, or BRCA1 promoter methylation, and loss of the second allele of the affected gene. All three HRD scores were highly associated with BRCA1/2 deficiency (HRD-LOH: P =1.3¿×¿10¿17; HRD-TAI: P =1.5¿×¿10¿19; HRD-LST: P =3.5¿×¿10¿18). A combined score (HRD-mean) was calculated using the arithmetic mean of the three scores. In multivariable analyses the HRD-mean score captured significant BRCA1/2 deficiency information not captured by the three individual scores, or by clinical variables (P values for HRD-Mean adjusted for HRD-LOH: P =1.4¿×¿10¿8; HRD-TAI: P =2.9¿×¿10¿7; HRD-LST: P =2.8¿×¿10¿8; clinical variables: P =1.2¿×¿10¿16).ConclusionsThe HRD scores showed strong correlation with BRCA1/2 deficiency regardless of breast cancer subtype. The frequency of elevated scores suggests that a significant proportion of all breast tumor subtypes may carry defects in the homologous recombination DNA repair pathway. The HRD scores can be combined to produce a more robust predictor of HRD. The combination of a robust score, and the FFPE compatible assay described in this study, may facilitate use of agents targeting homologous recombination DNA repair in the clinical setting.