A comprehensive landscape of BRCA1 versus BRCA2 associated molecular alterations and survival outcome across 35 cancer types.

Abstract

3120 Background: Poly (ADP-ribose) polymerase inhibitors (PARPi) are effective therapies for some patients with both germline and somatic BRCA1/2 mutations (MTs) or with homologous recombination repair deficiency (HRD). We aimed to characterize molecular differences between BRCA1 and BRCA2 MTs and their prognostic and/or predictive impact on PARPi outcomes in various cancer subtypes using real world data (RWD). Methods: Tumor samples obtained from patients with 35 types of cancer were analyzed by whole exome sequencing (WES, Novaseq) at Caris Life Sciences (Phoenix, AZ). High genomic loss of heterozygosity (gLOH-H) was defined as LOH-H in ≥16% of tested loci. MSI/MMR was tested by fragment analysis, IHC, and WES. Overall survival (OS) extracted from insurance claims was calculated from start of treatment or tissue collection until last contact or death using Kaplan-Meier curves. P-values adjusted for multiple comparisons (q-value of < 0.05 was considered to be significant). Results: In total, 17,640 tumors were included, of which 776 (4.3%) had tumor-based BRCA1/2 MTs. BRCA1/2 MTs were most commonly seen in ovarian (N = 221/2187, 10.1%), breast (138/2506, 5.5%), prostate (61/1131, 5.4%), pancreatic (48/1430, 3.4%), and non-small cell lung (100/4046, 2.5%) cancers. BRCA2 MTs were more frequent than BRCA1 except in ovarian cancers. BRCA1 MTs were more common in younger pts (median age, 61 vs 65 years, p <.001). When compared to BRCA2 MTs, BRCA1 MTs were more often associated with gLOH-H (64% vs 51%, p <.001) and TP53 MT (80% vs 53%, p <.001) in all tumor types. In NSCLC, EGFR mutations were exclusively seen in BRCA2 compared to BRCA1 (10.3 vs. 0%, P = 0.038). The EGFR mutations that co-occurred with BRCA2 mutations were L858R (N = 1), Exon19del (N = 4), and L861Q (N = 1). KRAS was more frequently mutated in BRCA1-mutated NSCLC ( BRCA1: 32% vs. BRCA2: 16%, p =.056). In univariate analyses, overall BRCA1/2 MTs were associated with improved OS compared to wild type (HR 1.38, 95% CI [1.31-1.45], P <.0001). This effect was seen in ovarian (1.42 [1.29-1.57], p < 0.0001) and triple-negative breast cancers (TNBC) (1.18, [1.09-1.28], p <.001); but was not observed in prostate, pancreatic, or non-TNBC breast cancer subtypes. In all breast cancers, BRCA2 MTs had a superior OS (0.68, [0.51-0.89], p =.005) compared to BRCA1, while no differences were seen in other cancers. Using RWD, PARPi treated-patients with BRCA2 MTs had worse OS than BRCA1 MTs (HR 1.4, [1.09-1.80], p = 0.009); but this was not significant when individual cancers were considered. Conclusions: BRCA1 and BRCA2 MTs had variable power to be prognostic and predictive for PARPi efficacy among different cancer types using RWD. About 2.5% of NSCLCs harbor BRCA1/2 MT. Additional genomic exploration may refine biomarkers predictive of response to PARPi and may highlight features within the tumor microenvironment of importance in the setting of HRD.