Assessment of ctDNA somatic homologous recombination deficiency (HRD) in triple-negative breast cancer (TNBC) from SWOG S1416 trial.

1012 Background: HRD is observed in up to two-thirds of g BRCA- wildtype TNBC. S1416 (NCT02595905) showed that addition of a PARP inhibitor (veliparib) to cisplatin improved progression-free survival (PFS) in gBRCA -wildtype metastatic TNBC (mTNBC) with HRD phenotype (“ BRCA -like”). In this study, we sought to evaluate concordance between circulating tumor DNA (ctDNA)-based detection of somatic homologous recombination repair (sHRR) alterations and tumor-based HRD and to assess if sHRR deficiency (sHRR+) was associated with benefit from veliparib in g BRCA -wild type mTNBC in SWOG S1416. Methods: S1416 enrolled patients with mTNBC who had received ≤ 1 line of prior therapy and randomized them to cisplatin plus veliparib or placebo. Central gBRCA1/2 testing classified patients as gBRCA -mutated or -wildtype. An a priori defined biomarker panel classified gBRCA -wildtype patients into BRCA -like (HRD+) and non- BRCA -like (HRD-) groups. A third group with gBRCA -wildtype, but without tissue BRCA classification was also included. Pre-treatment and progression plasma samples were utilized for assessment of sHRR status. ctDNA was analyzed using the Guardant OMNI next-generation sequencing platform. sHRR+ was defined by detectable somatic alterations (SNVs, INDELs, fusions with a functional impact notation of deleterious, and/or CNVs with a functional characterization of homozygous deletion) in a 24 gene panel. Results: Among N =213 g BRCA -wildtype patients with evaluable pre-treatment blood samples, 25% were sHRR+. Among sHRR+ patients, alterations in CHEK2 (18%), BRCA1 (17%), BARD1 (8%), ATM (7%), BAP1 (7%), CDK12 (7%), NBN (7%), BRCA2 (5%), and FANCA (5%) accounted for 80% of sHRR alterations. Most sHRR+ patients (91%) had alterations in only one of 24 genes, suggesting mutual exclusivity of homologous recombination pathway alterations in ctDNA. sHRR+ status was numerically higher in BRCA -like compared to non- BRCA -like tumors or unclassified tumors (32% vs. 20% vs. 20%, respectively; P =0.12). Among n =98 patients with availability of evaluable pre-treatment and progression samples, 31% were sHRR+ at baseline and 28% were sHRR+ at progression. Numerically, conversion from sHRR+ to sHRR- was more common than conversion from sHRR- to sHRR+ (30% vs. 9%, respectively). sHRR was not prognostic for PFS (median 4.3 (sHRR+) vs. 4.1 (sHRR-) months, respectively; P =0.30) nor predictive of benefit from veliparib ( P =0.40). Conclusions: One-fourth of gBRCA- wildtype mTNBC patients have ctDNA sHRR alterations, and there is incomplete overlap between tumor- and ctDNA-assessed HRD. ctDNA sHRR alterations were mostly mutually exclusive. Approximately one-third of patients with baseline sHRR+ converted to sHRR- at time of progression while receiving DNA damaging chemotherapy. sHRR was not prognostic and did not predict benefit from veliparib in S1416.