Olaparib and talazoparib are approved for patients with HER2- mBC carrying BRCA1/2 germline mutations. As a consequence, the National Comprehensive Cancer Network (NCCN) and the ESMO guidelines recommend the assessment for germline BRCA1/2 mutations in patients with mBC as soon as possible at diagnosis [24,77]. However, no specific companion test has officially been validated in this setting. The National Institute for Health and Care Excellence (NICE) and NCCN acknowledged the need for highly sensitive assays to identify large genomic rearrangements [78,79].

The BRCAnalyse Myriads genetic test, composed of a quantitative PCR, CGH-microarray and bi-directional sequencing, was used in OlympiAD and EMBRACA trials with high sensitivity and specificity for determining BRCA1/2 PV, although it misses some defects such as RNA transcript processing or balanced rearrangements [80,81,82] (Table 3).

In addition to BRCA1/2 status, ESMO ABC5 2020 recommendations advocating for considering other HR-related genes, such as PALB2 [71], is the first official effort to enlarge the PARPi target population. However, gene panel strategies are limited by many uncertainties about the actual impact of many unknown HR-related gene alterations not included in these panels and does not solve the issue of variants of unknown significance (VUS) [83].

In that context, assessing the DNA genomic scars induced by defective DDR systems, instead of gene alterations, is an attractive approach. Several studies demonstrated that quantification of large-scale state transitions (LST) [84], LOH [85], and telomeric allelic imbalances (TAI) [86] were associated with higher probability of BRCA1/2 mutation, especially BRCA1 mutations [87]. Mutational signatures have also been associated with HRD and BRCA1/2 mutations [12,39].

Currently, two commercial tests have been developed including FoundationOne CDx test for BRCA PV and MyriadMychoice for BRCA PV and genomic instability.

FoundationOne CDx test (Foundation Medicine, Cambridge, MA, USA) combines BRCA1/2 PV and percentage of genomic LOH to provide a score (considered as high when ≥ 16), alongside assessment of other HR-related gene alterations.

MyriadMyChoice genomic test associates BRCA1/2 PV, and the three biomarkers LST, LOH, and TAI combined together in a genomic instability score (GIS), categorized as low when <42, or high when ≥42. LST, LOH, and TAI genomics scars detection using MyriadMyChoice genomic test were found in TNBC with BRCA1/2 mutations, and then in the other breast cancer subtypes including ER+/HER2-. Of note, the mean HRD score was found to be similar in both ER+/HER2- and TNBC with BRCA1/2 mutations (mean around 14.5). In addition, the combination of these three individual biomarkers exhibited better predictive value regarding BRCA1/2 deficiency than each of them considered alone [88]. This assay is now recognized by the Food and Drug Administration for assessing the HRD status and the utility of PARPi in patients with advanced ovarian cancers [9,10].

Academic tests use additional information from HR mutational signatures in combination with genomic scars and genomic instability.

HRDetect [89] computes five weighted parameters, including microhomology-mediated small insertions and deletions (indels), GIS, single-base substitution (SBS) signature 3, rearrangement signature 3 and 5 [39]. The sensitivity of HRDetect for predicting BRCA1/2 mutation in the validation cohort was 86% in ER+/HER2- breast cancer patients [89] and showed sensitivity to detect PALB2 biallelic inactivation or RAD51C hypermethylation, and to reclassify BRCA1/2 VUS as germline PV [90]. A prospective study showed that HRDetect was predictive of rucaparib efficacy in a neoadjuvant setting [91].

The random forest-based Classifier of Homologous Recombination Deficiency (CHORD) was developed with the data from patients with metastatic solid tumors, including 19% of mBC. The score integrating a single nucleotide variant, indels, and structural variants, identified HRD status in 24% of primary and 12% of metastatic lesions from breast cancers. Of note, it distinguishes “BRCA1-type HRD”, associated with BRCA1 PV along with deficiency in BRCA1 binding proteins such as BRIP1, FAM175A, FANCA, and BARD1, and “BRCA2 type HRD” associated with PALB2, RAD51C, and BRCA2 PV [13].

In parallel, Bertucci and colleagues combined SBS signature 3 and LST and found a larger proportion of HRD tumors among patients with ER+/HER2- breast cancers. HRD status was found to be more frequent in ER+/HER2- mBC compared to ER+/HER2- eBC (15% versus 8.0%, respectively, p = 0.005) [42].

Whilst genomic instability and HRD scores can be seen as potential predictive biomarkers of PARPi efficacy, they do not provide direct information about the origin of HRD. Furthermore, HRD score was initially developed for maximizing the likelihood of BRCA1 mutations [87] and additional work is needed to optimize this tool to ER+/HER2- mBC known to be associated with a higher rate of BRCA2 or PALB2 PV and to harbor specific genomic features such as larger deletion and microhomology [92].

Finally, the main drawback of genomic scar signatures is the lack of consideration of the HRD dynamics or reversion, explaining the reduced value in multi-treated patients [93] whose tumor may change the HR function, which may not be captured by genomic scars, that are indelible [94].

Comparative genomic hybridization arrays were used to characterized copy-number (CN) profile of BRCA1/2 mutated breast cancers [95,96] and predict the benefit from chemotherapy [97]. Thereafter, specifically designed MLPA determined the CN profile of up to 50 different genomic regions [98] and demonstrated a good sensitivity and specificity to detect BRCA1-like tumors and predict the response to chemotherapy [97,99]. Then, digitalMLPA allowed to identify the CN profile of up to 700 genomic locations and distinguish non-BRCA-like, BRCA1-like, and BRCA2-like breast cancers. The accuracy was 91% and 82% for the BRCA1-like and BRCA2-like classification, respectively. Moreover, this test may also identify patients with triple-negative or ER+/HER2- breast cancers who could benefit from adjuvant chemotherapy [100]. An on-going phase III trial is assessing a combination of cytotoxic chemotherapy with olaparib in eBC patients with BRCA1-like tumors identified with this digital MLPA assay (NCT02810743).

RAD51 recruitment DNA breaks are a hallmark of HR pathways, that immunofluorescence can detect on formalin-fixed paraffin-embedded tumor samples.

RAD51 foci deficiency was significantly associated with a higher Myriad HRD score or biallelic inactivation of HR-related genes including BRCA1, BRCA2, CHEK2 [101], and PALB2 [102]. Functional HRD deficiency was correlated with PARPi and platinum-based chemotherapy efficacy and the subsequent resistance to these drugs in patients carrying the BRCA1/2 mutation [93,102,103,104]. These data suggested that RAD51 foci detection is a dynamic test that can diagnose HRD, and then restored pathways. However, this test cannot detect alterations occurring downstream from RAD51 intervention, such as ATM alterations.