Prevalence of tumor genomic alternations in homologous recombination genes among Taiwanese breast cancers

Purpose Deleterious BRCA1 / 2 mutations are among the most highly pathogenic variants in hereditary breast and ovarian cancer syndrome. PARP (poly ADP ribose polymerase) inhibitors selectively cause failure of single-strand but does not affect double-strand DNA break repair, and ant-tumor activity is observed in BRCA mutant breast cancers. Recently genes implicated into the homologous recombination repair (HRR) pathways are investigated extensively, as defective HRR genes may indicate potential clinical benets from PARP inhibitors beyond BRCA1 / 2 mutations. Materials and Methods We evaluated the prevalence of BRCA1 / 2 mutations as well as alternations in HRR genes for Taiwanese breast cancers with targeted sequencing. Consecutive 648 breast cancer samples were assayed, and HRR genes by Heeke et al. and those interrogated in Talazoparib Beyond BRCA (TBB) trial were evaluated for prevalence from breast cancer tissues. Results Among 648 breast cancer samples, there were 18 truncating and 2 missense mutations in BRCA1 and 48 truncating and 2 missense mutations in BRCA2 , impacting 3% and 5% of study population (collectively altered in 6%) with co-occurrence of BRCA1 / 2 in 7 breast cancers. On the other hand, HRR genes dened by Heeke et al. were altered in 122 (19%) breast cancers while TBB interrogated genes (excluding BRCA1 / 2 ) were mutated in 107 (17%) patients. Beyond BRCA1/2 , the most prevalent HRR mutant genes came from ARID1A (7%), PALB2 (7%) and PTEN (6%). Collectively, 164 (25%) of the 648 Taiwanese breast cancer samples were impacted by at least one HRR gene. Conclusion The prevalence of high-penetrant BRCA1 / 2 mutations was far below one tenth of assayed samples while the prevalence of tumor DNA mutations in HRR pathways was much higher and approached one fth among Taiwanese breast cancers. Further studies to evaluate the ecacy of PARP inhibitors in patients with defective HRR gene(s) are warranted to broaden the targeted population of synthetic lethality.


Introduction
Breast cancer has overtaken lung cancer as the most common cancer, accounting for nearly 12% of new cases each year worldwide, according to the World Health Organization [1]. Breast cancer is also one of the leading causes of cancer-related death for women in Taiwan, and the annual incidence is continuously increasing throughout the past two decades [2][3]. Outcomes of breast cancer treatment have improved enormously with long-term survival achieved due to awareness, wide spread of screening and e cient therapeutic options. With advocacy of personalized medicine, the combination of clinical and genetic information is useful for decision making and precise selection of targeted agents [4]. Among all breast cancer susceptible genes, BRCA1/2 deletions are among the most highly pathogenic genetic alternations in hereditary breast and ovarian cancer (HBOC) syndrome [5]. Pathogenic/likely pathogenic BRCA mutations not only trigger genetic counseling for the infected proband and associated family members, germline mutations of BRCA are predictive biomarkers for synthetic lethargy [6][7][8]. PARP (poly ADP ribose polymerase) inhibitors selectively cause failure of DNA single-strand break (SSB) repair but does not affect double-strand break (DSB) repair, for which intact BRCA1 and BRCA2 are indispensable component enzymes within cascading pathway [9]. Under the circumstance of defective BRCA1/2 deleterious mutations, anti-tumor activity is observed with the treatment of PARP inhibitors [10].
Our previous study has evaluated mutational pro les of BRCA1/2 among Taiwanese breast cancers, and the preliminary results of 100 consecutive patients assayed with targeted sequencing of tumor-only samples showed 74 single nucleotide variants (SNVs), 2 copy number variations (CNVs), and 15 insertions/deletions (indels) with a prevalence rate of 4% (4 breast cancers). Regarding the distribution of ltered SNVs, 38 were with BRCA2 original and 36 with BRCA1 gene, while 40 out of them were reported as being benign or likely benign, 33 as variant of uncertain signi cance (VUS) or never referred, and 1 pathogenic. The only pathogenic SNV, BRCA1 (c.3607C > T), resulting in p.Arg1203Ter amino acid change from one triple negative breast cancer (TNBC). Additional four pathogenic variants were identi ed from four indels among three breast cancers. The doublet BRCA2 pathogenic indels (c.1612_1613insA and c.1792_1793insA; amino acid changes: p.Ser538fs and p.Thr598fs) came from one premenopausal TNBC patient with a positive family history. The remaining pathogenic BRCA2 (c.5994delA; p.Val1999fs) and BRCA1 deletion (c.2110_2111delAA; p.Asn704fs) were revealed from two luminal/human epidermal growth factor receptor II (HER2)-negative breast cancers [11].
Taking our series of non-selective 100 consecutive Taiwanese cases together with others, the prevalence of somatic/germline mutations of BRCA1/2 is extremely low and never exceed one-tenth, even among TNBC, further limiting the potential bene ciaries of synthetic lethargy [12][13]. Recently genes implicated in the homologous recombination repair (HRR) pathways are investigated extensively, as defective HRR genes other than BRCA1 and BRCA2 may indicate potential applications from PARP inhibitors to alleviate the prerequisite of defective BRCA genes [14]. In current study we evaluated the prevalence of mutations within BRCA1/2 and other HRR genes among Taiwanese breast cancers.

Materials And Methods
The aim of the study was to evaluate the prevalence of BRCA1/2 mutations as well as mutations of HRR genes for Taiwanese breast cancers with targeted sequencing of tumor-only samples.

Study population
The full protocol of the VGH-TAYLOR: Comprehensive precision medicine research on the heterogeneity of Taiwanese breast cancer patients, and results of initial 380 targeted sequencing have been descripted elsewhere [15]. The study comprised a broad clinical spectrum of breast cancers, including Group 1: planned to receive rstline surgery and followed by adjuvant therapy, or early relapse within three years. Group 2: planned to receive rstline neoadjuvant therapy and followed by surgery, and Group 3: de novo stage IV, or stage IV with recurrence beyond three years. Three years of enrollment and four years follow-up after enrollment were planned.
For clinical parameters, estrogen receptor (ER) and progesterone receptor (PR) positive was de ned as ≥1% of tumor cells exhibiting nuclear staining by immunohistochemistry (IHC) assay. Patients with HER2 testing scored as IHC 3+ (positive) or 2+ (equivocal) and with uorescence in situ hybridization (FISH) con rmed ampli cation were regarded as being HER2 positive. For molecular subtyping, luminal A subtype was de ned as ER positive, HER2 negative, with PR >20%, Ki-67 <=20%, and nuclear differentiation as grade I or II. Luminal B1 breast cancers were ER positive, HER2 negative and at least one held true for PR < 20%, Ki-67 >20%, or being nuclear poorly differentiated as grade 3. Luminal B2 breast cancers were those with both ER and HER2 positive phenotypically. Assay v3 (Thermo Fisher Scienti c, Waltham, MA) was adopted for next-generation sequencing (NGS), which enabled detections of 161 cancer-related genes and identi ed SNVs, CNVs, gene fusions, and indels [15].
Formalin-xed para n-embedded (FFPE) specimens were assayed; sequencing data were analyzed, aligned and annotated through the Torrent Suit (Thermo Fisher Scienti c) and Ion Reporter (Thermo Fisher Scienti c) software with the default Oncomine BRCA (5.12) lter applied.

Variant caller and annotation
Mutational consequences of called variants were ascertained with ClinVAR [21]. To further correct spurious ndings due to trans-ethnic discrepancy, the on-line VariED tool was consulted to lter out Taiwan Biobank polymorphisms [22].

Enrolled Taiwanese breast cancers
In current study, we presented the updated results of 648 TMO assays from 621 breast cancer patients (27 assayed twice) from the VGH-TAYLOR study. The distributions of clinical scenarios were: Group 1A (surgery rst, n=387), Group 1B (recurrence within 3 years, n=25), Group 2 (neoadjuvant therapy rst, n=93), Group 3-1 (de novo stage IV, n=36), and Group 3-2 (recurrence beyond 3 years, n=36). In addition, there were 71 subjects from retrospective biobank cohort. Twenty-eight patients were assayed twice, including 19 in Group 2 (diagnostic/postneoadjuvant pair) and 8 in Group 1B. Fig 1 displays the distributions of IHC results and molecular subtypes.

Discussion
Breast cancer is one of the leading causes of cancer-related death in women, and the annual incidence of breast cancer has increased continuously throughout the past two decades in Taiwan. There is a demand to identify therapeutics based on genetic pro ling from breast cancer samples to improve treatment response and suppress recurrence rate. Our previous study has determined the molecular alternations underpinning breast oncogenesis and identi ed potential biomarkers from a large cohort of Taiwanese breast cancers using tumor-only targeted sequencing [15]. In current study, we focused on the prevalence of HRR genes to shed light on the potential application of PARP inhibitors and subsequent synthetic lethargy.
The prevalence of BRCA1/2 mutations (ESMO Scale for Clinical Actionability of molecular Targets (ESCAT) Tier 1A for germline and Tier IIIA for somatic mutation) in an unselected breast cancer cohort was collectively 6% in Taiwan [23]. BRCA1/2 mutation status does not confer adverse prognosis other than contralateral breast cancer occurrence, and no disease-speci c or overall survival difference is observed between BRCA1/2 mutant and wildtype breast cancers [24][25]. Synthetic lethality is evidenced when occurrence in either of the two genetic events individually has no effect but combining both events leads to cell death. PARP inhibitor selectively causes failure of SSB but does not affect DSB repair pathway, and durable ant-tumor response was observed in patients with BRCA mutant breast cancers [26]. Both triple negative (HR-/HER2-) and HR+/HER2-breast cancers may bene t from the defective DSB repair pathway inherited from a deleterious BRCA gene mutation, and the recently approved PARP inhibitor could be a reasonable therapeutic choice in metastatic or recurrent scenarios after adjuvant chemotherapy. FDA-approved PARP inhibitors such as olaparib and talazoparib were responsive for advanced/metastatic breast cancers who had undergone previous chemotherapy if germline mutation was also con rmed [27][28]. Mechanisms of PARP inhibitor comes from targeting DNA damage response, destabilizing replication forks through PARP-DNA entrapment and inducing cell death through replication stress-induced mitotic catastrophe [6]. Breast cancers with BRCA1/2 mutations are candidates for synthetic lethargy if their mutations are proved germline origin. Low mutation rate of BRCA1/2 in sporadic/hereditary breast cancers, however, further limit the clinical applicability of PARP inhibitors.
To overcome the low prevalence of germline BRCA1 and BRCA2 mutations, numerous efforts have been made to evaluate the role of other DNA damage repair (DDR) genes beyond BRCA1/2, with the hope to extend the clinical applicability of PARP inhibitors [29]. Indeed, there are multiple players involved in the HRR pathways, such as sensors of DSB (ATM and ATR) which subsequently lead to activation signal mediators (BRCA1, BRCA2 and PALB2), as well as nuclear protein RAD51 for strand invasion and replication fork stabilization [30]. Dozens of genes have been reported as being interrogated within the HRR process, and a critical question raised is to determine an operable and reproducible genes list as a predictive biomarker for targeted therapy. It's also of great clinical potentiality to treat BRCA-wild breast cancers that can bene t from PARP inhibitors based on HRR de ciency.
In current study, two de nitions of HRR genes were adopted. Heeke et al. evaluated 56,426 solid tumors molecularly pro led with NGS or Sanger sequencing. HRR de ciency by linage was reported, with an overall mutation frequency of 17.4% reported [19]. Overall, the most frequently mutated genes were ARID1A (7.2%), BRCA2 (3%), BRCA1 (2.8%), ATM (1.3%), ATRX (1.3%), CHEK2 (1.3%) and BAP1 (1.1%). Breast cancer ranked the ninth over 21 cancer types in terms of HRR de ciency while all metastatic solid tumors should be pro led according to their conclusion. We also evaluated HRR genes purposed by the TBB study, which was a phase II trial evaluating talazoparib in advanced HER2-negative breast cancer or other solid tumors with a germline or somatic alteration in HRR genes excluding BRCA1/2 [20]. Clinical bene ts were observed for breast cancers with germline PALB2, CHEK1, FANCA and somatic PTEN, ATR mutations. In our tumor-only series, BRCA1/2 mutations collectively impacted 6% of Taiwanese breast cancers, and the proportion with mutant HRR genes ranged between 17% and 19% based on different de nition, and a total of 25% Taiwanese breast cancers harbored at least one alternation among HRR gene from targeted sequencing. Another proof of concept study comes from the TCBRC048 trial, which reported response rate of 82% among 11 germline PALB2 mutant breast cancers and 50% for 16 somatic BRCA1/2 mutant cases with olaparib [31].
It's interesting to see if there exists a dose-dependent effect of PARP inhibition based on dual BRCA alternations or doublet/triplet mutations of BRCA1/2 as our study revealed co-occurrence of BRCA1 and BRCA2 in seven Taiwanese breast cancers as well as one doublet BRCA1 and seven cases with multiple BRCA2 mutations. Clinical phenotypes do matter here as PARP inhibitors are indicated only for HER2 negative advanced/metastatic breast cancers. Although up to one-fourth of Taiwanese breast cancers reported tumor mutations in HRR genes, this did not necessarily translate into clinical actionability under current evidence of targeted therapy. Nowadays only germline BRCA1/2 mutations have therapeutic implications. Ongoing clinical trials are evaluating PARP inhibitors in other DDR pathway genes (both germline and somatic); the clinical signi cance of somatic BRCA1/2 mutations in breast cancer remains inconclusive, and the ongoing NCT03344965 and VIOLETTE (NCT03330847) trials are designed to answer the question whether BRCA1/2 somatic mutation could act as a predictive marker for PARP inhibitor response as olaparib in ovarian cancer [32][33][34].
In addition to PARP inhibitor monotherapy, the combination with immune checkpoint blockade such as TOPACIO (niraparib plus pembrolizumab in triple negative breast cancer regardless of BRCA1/2 status), MEDIOLA (olaparib plus durvalumab in triple negative breast cancer with germline BRCA1/2 mutation), DORA (olaparib plus durvalumab as maintenance therapy after platinum chemotherapy), NCT03801369 (olaparib plus durvalumab in BRCA-wild type triple negative breast cancer) and JAVELIN BRCA/ATM (avelumab plus talazoparib with either BRCA or ATM mutation) have been extensively investigated [35]. The combination of cisplatin chemotherapy with a less potent PARP inhibitor (veliparib) was evaluated in the SWOG1416 trial, and for the rst time that an improvement in progression-free and overall survival with a PARP inhibitor in metastatic triple negative breast cancer without a germline BRCA mutation had been demonstrated [36]. It was the "BRCA-like" group which derived the most clinical bene t from the combination therapy. Four biomarkers, HRD genomic instability score with cutoff 42, somatic BRCA1/2 mutation, BRCA1 promoter methylation, and germline HRR gene mutations excluding BRCA1/2, constituted the BRCA-like phenotype [37].
There were some limitations of the study. First, we conducted tumor-only sequencing while re ex testing for germline mutations were not carried out once alternations in HRR genes were reported. The tumor-only study design abolished the chance to discover germline mutations from targeted sequencing [38]. Without matched germline sequencing, it was di cult to differentiate somatic from germline mutations for investigated genes. Second, although the commercialized OCP assay was used with wide coverage of tumor associated genes, there were still missing HRR genes not interrogated such as BARD1, BLM, BRIP1, CHEK2, FANCC, FANCE, FANCF, FANCG, FANCL, MRE11A and WRN. Most of these missing genes were of rare variants and rarely addressed by most targeted panels. Third, "genomic scar", i.e. scores capturing large genomic aberrations such as loss of HRR using SNP arrays, myChoice HRD (Myriad Genetics) and Foundation Focus CDx BRCA LOH (Foundation Medicine) could not be mimicked by NGS experiments [39][40]. Only point mutations presented as SNVs, MNVs, Indels, as well as structural aberrations including CNVs and fusions were evaluated in HRR genes using tumor DNA sequencing.

Conclusion
Our study con rmed the feasibility of targeted sequencing with a commercial tumor-only panel while the prevalence of high-penetrant BRCA1/2 mutations was collectively up to 6% of an unselected Taiwanese population, which narrowed the potentially targeted population of PARP inhibitors substantially. The prevalence of tumor DNA mutations in HRR pathway was much higher and approached one fth among Taiwanese breast cancers. PARP inhibitor is a successful paradigm for synthetic lethargy and the frontline will be extended to early stage disease in near future [41]. Further studies to evaluate the e cacy of PARP inhibitors in patients with a defective HRR pathway are warranted to broaden the targeted population of synthetic lethality. Figure 1 Distributions of clinical variables across study groups. The distributions of clinical scenarios were: Group 1A (surgery rst, n=387), Group 1B (recurrence within 3 years, n=25), Group 2 (neoadjuvant therapy rst, n=93), Group 3-1 (de novo stage IV, n=36), Group 3-2 (recurrence beyond 3 years, n=36), and retrospective biobank cohort (non-PCR, n=9, stage III, n=21 and stage IV, n=41). IHC: immunohistochemistry, m_subtype: molecular subtype.    OncoPrinter showed HRR genes altered in 107 (17%, Talazoparib Beyond BRCA trail de nition, top) and 122 (19%, Heeke et al. de nition, bottom) breast cancer samples based on different de nition of HRR genes. IHC: immunohistochemistry, m_subtype: molecular subtype.

Figure 5
Chord diagram of mutant HRR genes among Taiwanese breast cancers.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.