Development of a new HRD algorithm which was highly associated with anthracycline-containing neoadjuvant chemotherapy.

Abstract

e13575 Background: Homologous recombination deficiency (HRD) can be resulted from dysfunction of BRCA and is associated with sensitivity to platinum, PARP inhibitor and other DNA-damaging drugs. The results from a neoadjuvant trial showed that pathological complete response (pCR) was not significantly higher with cisplatin than with doxorubicin-cyclophosphamide in BRCA1/2-mutated breast cancers (BC). It suggests that BC with HRD might benefit from anthracycline-containing regiment. There are many commercial HRD detection assays, including the FoundationFocus CDx BRCA LOH and myChoice CDx, but there is still no uniform standard in China. Methods: A total of 96 in-house BC samples and 6 HRD positive standard cells (Cat No. CBP90023) were analyzed by whole-genome sequencing (WGS). Besides, 122 BCs from the TCGA database were down-sampled to ̃1X WGS. We constructed a new algorithm for HRD score based on WGS at low coverage as input data to estimate large-scale copy number alteration (LCNA) events on the genome. The sensitivity and specificity were compared between our algorithm and the ShallowHRD. A clinical cohort of 50 BCs (15 cases carrying BRCA mutation) was used to assess the association between HRD status and anthracyclines-containing neoadjuvant treatment outcomes. Results: A 100kb-window was defined as the optimal size by using 41 in-house cases and the TCGA dataset. The threshold of HRD was determined as the number of 10 LCNAs by using 55 in-house BCs with BRCA mutation, with the goal of achieving 95% sensitivity. The sensitivity and specificity of our algorithm were both 100%, while those of the ShallowHRD were 40% and 100%, respectively, by testing standard samples with positive HRD. Meanwhile, similar results were also observed that the sensitivity of our algorithm was far superior to ShallowHRD (87% vs 13%) in the clinical cohort. The association between HRD status and BRCA mutations was compared between our algorithm and the ShallowHRD by 120 BC WGS samples (20 cases carrying BRCA mutation) from the TCGA database. The results showed that BRCA status was significantly associated with HRD status by our algorithm and ShallowHRD (P = 0.00838 and P = 0.00284, respectively). However, our algorithm had a higher positive concordance rate than the ShallowHRD algorithm (70% vs 60%). In the clinical cohort of neoadjuvant treatment, HRD group was more likely to respond to anthracycline-containing chemotherapy than non-HRD group, with outcomes of pCR (OR = 9.5, 95% CI: 1.11–81.5, p = 0.04) and residual cancer burden score of 0 or 1 (RCB0/1) (OR = 10.29, 95% CI: 2.02–52.36, p = 0.005). Among 35 patients lacking BRCA mutations, HRD group tended to have RCB0/1 responses compared to non-HRD group (OR = 6.0, 95% CI: 1.00–35.91, p = 0.05). Conclusions: Here, we developed a new stable algorithm for HRD score. It’s a promising assay for clinical application to predict sensitivity of DNA-damaging drugs.