Germline breast cancer susceptibility genes, tumor characteristics, and survival

Background Mutations in certain genes are known to increase breast cancer risk. We study the relevance of rare protein-truncating variants (PTVs) that may result in loss-of-function in breast cancer susceptibility genes on tumor characteristics and survival in 8852 breast cancer patients of Asian descent. Methods Gene panel sequencing was performed for 34 known or suspected breast cancer predisposition genes, of which nine genes (ATM, BRCA1, BRCA2, CHEK2, PALB2, BARD1, RAD51C, RAD51D, and TP53) were associated with breast cancer risk. Associations between PTV carriership in one or more genes and tumor characteristics were examined using multinomial logistic regression. Ten-year overall survival was estimated using Cox regression models in 6477 breast cancer patients after excluding older patients (≥75years) and stage 0 and IV disease. Results PTV9genes carriership (n = 690) was significantly associated (p < 0.001) with more aggressive tumor characteristics including high grade (poorly vs well-differentiated, odds ratio [95% confidence interval] 3.48 [2.35–5.17], moderately vs well-differentiated 2.33 [1.56–3.49]), as well as luminal B [HER−] and triple-negative subtypes (vs luminal A 2.15 [1.58–2.92] and 2.85 [2.17–3.73], respectively), adjusted for age at diagnosis, study, and ethnicity. Associations with grade and luminal B [HER2−] subtype remained significant after excluding BRCA1/2 carriers. PTV25genes carriership (n = 289, excluding carriers of the nine genes associated with breast cancer) was not associated with tumor characteristics. However, PTV25genes carriership, but not PTV9genes carriership, was suggested to be associated with worse 10-year overall survival (hazard ratio [CI] 1.63 [1.16–2.28]). Conclusions PTV9genes carriership is associated with more aggressive tumors. Variants in other genes might be associated with the survival of breast cancer patients. The finding that PTV carriership is not just associated with higher breast cancer risk, but also more severe and fatal forms of the disease, suggests that genetic testing has the potential to provide additional health information and help healthy individuals make screening decisions. Supplementary Information The online version contains supplementary material available at 10.1186/s13073-021-00978-9.

Conclusions: PTV 9genes carriership is associated with more aggressive tumors. Variants in other genes might be associated with the survival of breast cancer patients. The finding that PTV carriership is not just associated with higher breast cancer risk, but also more severe and fatal forms of the disease, suggests that genetic testing has the potential to provide additional health information and help healthy individuals make screening decisions.
Keywords: Breast cancer, Protein-truncating variants, Overall survival Background Breast cancer is the most common cancer among women worldwide. Breast cancer manifestations are biologically and molecularly heterogeneous with a high degree of diversity observed between and within tumors. Such phenotypic differences in tumor characteristics are clinically informative because they are prognostic and can improve therapy selection [1,2]. In particular, profiling of breast cancer can be based on the expressions of three immunohistochemical markers: estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor (HER2). Patients with ER-and PR-positive breast cancer respond well to endocrine therapy and have favorable outcomes [1]. In contrast, tumors which were ER/PR and HER2 negative are associated with worse survival and are typically treated with chemotherapy and radiotherapy [1].
There is a strong genetic component to the risk of breast cancer [3]. A large majority of disease-associated variants in susceptibility genes are protein-truncating variants (PTVs), a class of variants that usually results in an absence of functional protein [4]. Pathogenic PTVs are typically rare (allele frequency <1%) and are usually associated with a twofold or higher risk of breast cancer [5].
Previously, Li et al. [6] observed, in a study of 5099 Swedish breast cancer patients, that tumors arising in PTV carriers with known or suspected predisposition genes were phenotypically more aggressive and had worse survival as compared to tumors in non-carriers. However, there is growing concern that genetic markers identified in populations of predominantly European ancestry may not be equally informative in non-European populations, due to the modifier effect of lifestyle and genetic factors which may be distributed differently in diverse populations [7]. Here, we attempt to study how rare variants in breast cancer predisposition genes are associated with tumor characteristics and survival in 8852 breast cancer patients of Asian descent.

Study populations
The study population was derived from the patients enrolled in the Singapore Breast Cancer Cohort (SGBCC), the Malaysian Breast Cancer Genetic Study (MyBrCa), and the Korean Hereditary Breast Cancer Study (KOHBRA).

Singapore Breast Cancer Cohort (SGBCC)
The Singapore Breast Cancer Cohort (SGBCC) study includes women aged 21 years and above diagnosed with either breast carcinoma in situ or invasive breast cancer at one of the six participating restructured hospitals in Singapore (National University Hospital, KK Women's and Children's Hospital, Tan Tock Seng Hospital, Singapore General Hospital, National Cancer Centre Singapore, Changi General Hospital). These six hospitals collectively diagnose and treat~76% of all breast cancer cases in Singapore [8]. Patients were a mixture of prevalent and incident cases from the three main ethnic groups, Chinese (81%), Malay (13%), and Indian (6%). Ethnicity was self-reported. According to the 2010 census of Singapore, Indian ethnicity refers "to persons of Indian, Pakistani, Bangladeshi or Sri Lankan origin such as Tamils, Malayalis, Punjabis, Bengalis, Singhalese, etc." The ethnic distribution of SGBCC is similar to that of the general population of Singapore (Chinese 75.9%, Malay 15%, Indian 7.5%, Department of Statistics Singapore). Between April 2010 and December 2016, 7768 breast cancer patients were recruited into SGBCC. Of which 4538 patients had genetic information from blood or saliva samples collected at recruitment. After excluding duplicated individuals (n = 9), 4529 patients were included to test for associations between PTV carriership and tumor characteristics. A subset of 3213 patients were eligible for survival analysis.

The Malaysian Breast Cancer Genetic Study (MyBrCa)
The Malaysian Breast Cancer Genetic Study (MyBrCa), a hospital-based case-control study, was initiated in 2002 (described by Tan et al. [9]). Briefly, all patients diagnosed with breast cancer from two participating hospitals in Selangor, Malaysia (University Malaya Medical Centre, a public hospital, and Subang Jaya Medical Centre, a private hospital), were invited to participate. Participants were mainly from urban areas. These two hospitals treat more than 10% of the breast cancer cases in Malaysia [9]. The core ethnic groups of MyBrCa are Chinese (69%), Malay (16%), and Indian (12%). In Malaysia, the predominant ethnic group is Malay (67.4%), followed by Chinese (24.6%) and Indian (7.3%) (Department of Statistics Malaysia Official Portal). At recruitment, all participants (n = 3822) provided a blood or saliva sample and completed a detailed questionnaire that included lifestyle and reproductive-related factors for breast cancer as well as personal and family history of cancer. We excluded 46 related samples and 15 duplicated samples, resulting in 3761 patients included in the study of associations between PTV carriership and tumor characteristics. A subset of 3264 patients were eligible for survival analysis.

The Korean Hereditary Breast Cancer Study (KOHBRA)
The KOHBRA study is a prospective hereditary breast cancer cohort in Korea [10]. Between May 2007 and May 2010, the KOHBRA study recruited 1967 subjects from 35 hospitals registered in the Korean Breast Cancer Society. All participants received genetic counseling and BRCA genetic testing; the clinical information and blood samples for blood banking were collected. Included in the study were patients with a family history of breast or ovarian cancers, patients with non-familial breast and ovarian cancer but with other risk factors of genetic disease, and family members of breast cancer patients with BRCA1/2 mutations [11]. Patients with information on PTVs (n = 562) were included in the study of associations between PTV carriership and tumor characteristics. The KOHBRA study was omitted from the survival analysis as it is a BRCA1/2 case-control study oversampling BRCA1/2 carriers, which will result in a survival bias in our PTV carriership analysis if included.

Demographic information
Self-reported ethnicity (Chinese, Malay, Indian, and others) and family history of breast cancer (yes, no) were obtained from structured questionnaires, in SGBCC and MyBrCa. All patients from KOHBRA were Koreans and family history was obtained from a structured questionnaire.

Clinical data for breast cancer patients
Clinical characteristics were extracted from hospital breast cancer registries or hospital medical records-age at diagnosis (years), tumor stage (0, I-IV), tumor size (< 2 cm, 2-5 cm, and >5 cm, similar to TNM size reported by AJCC version 7), nodal status (positive, negative), tumor grade (well-differentiated, moderately differentiated, and poorly differentiated), and immunohistochemical markers ER, PR, and HER2. For ER status and PR status, staining of ≥1% was considered positive. HER2 status was classified as positive or negative (includes equivocal). Intrinsic-like subtypes were defined using immunohistochemical markers for ER, PR, and HER2 in conjunction with histologic grade: luminal A [ER+/PR+,  [12].
Treatment data, for SGBCC and MyBrCa, were extracted from hospital breast cancer registries or hospital medical records-surgery (yes, no), neo-adjuvant and/or adjuvant chemotherapy (yes, no), radiotherapy (yes, no), endocrine therapy (yes, no), and trastuzumab therapy (yes/no). Target-enriched sequencing libraries of germline DNA for the breast cancer cases and controls were prepared at the Centre for Cancer Genetic Epidemiology (University of Cambridge) as part of a larger effort (Breast Cancer Risk after Diagnostic Gene Sequencing, BRIDGES, https://bridges-research.eu) [13]. The gene panel, which was developed as part of the BRIDGES initiative, included coding sequences and intron/exon boundaries for a total of 34 genes for which there was prior evidence of association with breast cancer risk, including genes offered on commercial panels for breast cancer in early 2016 (Additional file 1: Table S1). The targeted sequencing workgroup in BRIDGES first performed rare variant detection in preliminary gene panels combined with data from whole-exome sequencing datasets before arriving at the 34 genes selected for the BRIDGES panel. Details of the library preparation, sequencing, variant calling, and quality control methods are given in Dorling et al. [13].

Protein-truncating variant (PTV) carriership
In this study, PTVs were defined as (1) variants predicted to introduce a premature stop codon (frameshift or nonsense mutations), (2) small insertions or deletions (indels) predicted to disrupt a transcript's reading frame, or (3) splice site mutations. PTVs occurring in the last exon of each gene were excluded to avoid including variants that do not lead to nonsense-mediated decay. Variants with less than 1% frequency in our study population were included. A list of PTVs from the 34 genes sequenced is listed in Additional file 1: Table S2. The distribution of PTVs in unselected breast cancer patients from SGBCC and MyBrCa was visualized in oncoplot and bar chart (Additional file 1: Fig. S1). As the PTVs are rare [14], the number of carriers in most genes was too small to analyze individually. Hence, we aggregated PTVs for each individual, creating a single binary variable: carrier of at least one PTV in any gene versus non-carrier. We hypothesized that the effects of PTVs on subtype and outcome were likely to be in the same direction and therefore the power to detect an association with this burden variable would be much greater.

Statistical analysis
Fisher's exact test was used to determine whether the proportion of PTV carriers for individual genes was different between two of the largest ethnic subgroups in this study (i.e., Chinese and Malay). To test for associations between PTV carriership and tumor characteristics in the full analytical cohort of 8852 breast cancer patients, we carried out multinomial logistic regression models (multinom function in the R package "nnet") with tumor characteristics as the outcome, adjusting for age at diagnosis, study, and ethnicity. Odds ratios (ORs) and corresponding 95% confidence intervals (CIs) were estimated. While SGBCC and MyBrCa included three core ethnic groups, KOHBRA consisted of a homogeneous population of Koreans; hence, adjustment for study and ethnicity was done as a joint variable.
Overall survival was studied in a subset of patients (n = 6477, 73% of analytical cohort, KOHBRA study excluded). Additional file 1: Fig Overall survival was studied using Cox proportional hazard models (survival package in R, where the Surv(time at entry, follow-up time, event)) command was used to estimate hazard ratios (HRs) and corresponding 95% CI. Adjustment for age at diagnosis, study (SGBCC, MyBrCa), ethnicity (Chinese, Malay, Indian, others), and tumor characteristics (stage, grade, ER status, nodal status, and tumor size) was done. As age 50 years is a common recommendation to start mammography screening, the survival analysis was repeated for subgroups of patients diagnosed at different age groups (<50 years and 50-75 years). A test for interaction (likelihood ratio test) between PTV and age group was performed using Cox proportional hazard models which included the following variables: PTV carriership, age group (<50, 50 to 75), study, ethnicity, and tumor characteristics.

PTV gene subset analysis
In the recent study by Dorling et al. [13] involving 60,466 female breast cancer cases and 53,461 controls from 44 studies, PTVs in nine of the 34 genes (ATM, BRCA1, BRCA2, CHEK2, PALB2, BARD1, RAD51C, RAD51D, and TP53) were found to be strongly associated with breast cancer risk (Bayesian false-discovery probability, <0.05). Association and survival analyses were repeated for PTV carriership coded for these nine genes and separately for the remaining 25 genes. In the analysis of the remaining 25 genes, carriers of any of the nine genes associated with breast cancer risk were excluded.

PTV subset analysis
As BRCA1/2 carriers tend to be associated with more aggressive tumor characteristics [15], all analyses were repeated without BRCA1/2 carriers to assess the combined effect of other breast cancer predisposition genes.
Intrinsic breast tumor subtypes are known to be highly predictive with breast cancer survival [16]. In addition, heterogeneity has been observed for breast cancer susceptibility risk genes across clinical subtypes for breast cancer [13,17]. Hence, we performed further analysis to study the association of PTV carriership with tumor characteristics and survival was within each proxy subtype (luminal A, luminal B [HER2−], luminal B [HER2+ ], HER2-enriched [HER2+], and triple-negative).
As ethnic Chinese breast cancer patients comprise the majority of the study population (71%), all analyses were repeated on a subset of 6265 Chinese breast cancer patients.
All statistical analyses were performed using R version 4.0.2.   (Table 2).

Characteristics of the breast cancer cohorts
Compared to PTV 34genes carriership, the observed associations with tumor characteristics were generally larger in effect size when PTV 9genes carriership (i.e., nine genes found to be significant in Dorling et al. [18]) was evaluated (

PTV carriership and overall survival within 10 years postdiagnosis
In 6477 invasive non-metastatic breast cancer patients aged <75 years at diagnosis from SGBCC and MyBrCa, a total of 790 deaths due to any cause occurred within 10 years after diagnosis with a median follow-up time of 5.6 years (IQR 3.4 to 8.5). The 10-year overall survival rate was 75% (95% CI 74 to 77%). In these 6477 patients, associations between PTV carriership and tumor characteristics were found to be similar with those from the entire analytical cohort of 8852 patients (Additional file 1: Table S6). Population structure of the major ethnic groups in SGBCC and MyBrCa was largely similar; Additional file 1: Fig. S4 shows a principal component analysis plot of SGBCC and MyBrCa breast cancer patients, colored by study (SGBCC or MyBrCa) and denoted by ethnicity (Chinese, Malay, or Indian).
PTV 34genes was not found to be associated with 10year overall survival (Additional file 1: Table S7, Fig. S5). The results were not appreciably different after adjustment for age at diagnosis, study, ethnicity, stage, grade, ER status, nodal status, and tumor size or when BRCA1/ 2 carriers were excluded (Additional file 1: Table S7,  Table 4, and Additional file 1: Fig. S6).
PTV 9genes carriership was not significantly associated with 10-year overall survival (Additional file 1: Table S7 and Fig. S7 Fig. S8). The likelihood ratio tests (LRT) for interaction between PTV and age group (<50 years, 50 to 75 years) in the adjusted models (study, ethnicity, stage, grade, ER status, nodal status, and tumor size) were not statistically significant (p-values, PTV 34genes 0.507, PTV 9genes 0.084, and PTV 25genes 0.556). Further adjustments for treatment variables for all survival analyses did not appreciably change the results (Additional file 1: Table S9). Kaplan-Meier curves for 10-year overall survival within each proxy subtype are presented in Additional file 1: Fig. S9 to S11. Significant associations were observed between PTV 34genes and PTV 25genes carriership and worse 10-year overall survival in the HER2-overexpressed proxy subtype. However, it should be noted that the results have to be interpreted with caution due to the limited number of death events within each subgroup.

Subset analysis in Chinese breast cancer patients
The results observed for Chinese breast cancer patients were similar to that of all breast cancer patients, for associations between PTV carriership and tumor characteristics and also associations between PTV carriership and survival. While the p-values were no longer statistically significant due to the reduced sample size comprising only Chinese breast cancer patients, the observed odds ratios and hazard ratios were not appreciably different. These results are presented in Additional file 1: Tables S9 and S10.

Discussion
Evaluating disease associations for rare variants in standard single-variant association analysis is challenging, since even if the effect size is large, the statistical power may still be low [19]. However, grouping variants in

0.804
All models are adjusted for study, ethnicity, and age at diagnosis. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using multinomial logistic regression models with each tumor characteristic as the outcome and genetic factors as explanatory variables. PTV  [19]. For example, multiple common variants are frequently collapsed into a single polygenic risk score [20]. For rare variant analysis in a gene-based burden test, the number of individuals carrying variants in a given gene is compared between affected and unaffected groups [21]. While rare variant analyses are commonly carried out in a region or gene-based manner [13,17], the collective effect of pathogenic variants across multiple breast cancer genes may also be studied [6,22,23]. In this large study of Asian breast cancer patients, we showed that PTV carriership in certain known or suggested breast cancer predisposition genes was associated with disease severity and fatality.
Two smaller studies on Asian Chinese breast cancer patients showed that while BRCA1/2 PTV carriers were associated with more aggressive clinical features, PTV carriers of non-BRCA1/2 breast cancer susceptibility genes when treated as a single group were not significantly associated with tumor characteristics (Wang et [22]). However, in spite of differences in population ancestry, results from our large study of Asian breast cancer patients closely replicated the findings of a study comprising 5099 breast cancer patients of European descent [6]. The European study examined PTVs in 31 genes in an earlier version of the BRIDGES panel, of which 30 genes overlap with the 34 genes in our study. The overlap between the genes included in Li et al. [6] and the current study is shown in Additional file 1: Fig.  S12. Li et al. [6] reported that compared to noncarriers, PTV carriers were more likely to have more aggressive tumors (i.e., ER-negative, large size, high grade, highly proliferative, luminal B, and triplenegative subtype). We observed the same significant associations in our Asian study and additionally found significant associations between PTV carriership and disease stage, tumor behavior, nodal involvement, PR status, and HER2 status.
Taking into account the known associations between BRCA1/2-related tumors and worse tumor biology, all analyses were repeated excluding BRCA1/2 carriers. Li et al. [6] observed that PTV carriership remained associated with high grade and worse survival. In our study, a significant association with grade was no longer seen after BRCA1/2 carrier exclusion, but other tumor characteristics common of fast-growing tumors, such as stage, tumor size, and luminal B [HER2−] subtype, remained significant. A more recent study reported that not all of the 34 reported or known breast cancer susceptibility genes on the targeted sequencing panel were clinically relevant for the prediction of breast cancer risk [13]. The results of a subset analysis of the nine genes found to be associated with breast cancer risk suggest that the genes most relevant for breast cancer development were also associated with more aggressive tumor phenotypes (larger effect sizes observed than for 34 genes). Nonetheless, the worse tumor characteristics were mostly driven by BRCA1/2 carriers-only stage, size, and subtype remained significantly associated with PTV carriership of the seven genes after the exclusion of BRCA1/2 carriers. PTV carriership of the remaining 25 genes was associated with a larger tumor size. What we found to be different from the Li et al. [6] Swedish study was that in our study, among BRCA1/2 non-carriers, there was a null relationship between PTV carriership (34 genes) and tumor grade. However, the gene subset analyses showed that while PTV 9genes carriership predisposed patients to worse tumor grade, a larger proportion of PTV 25genes carriers developed lower grade tumors.
We did not find a similar association between PTV 34genes carriership and 10-year overall survival as reported in Li et al.'s work on a Swedish dataset of 5099 breast cancer patients [6]. Interestingly, the gene subset analyses showed that PTV carriership of the nine genes, found to be most relevant for breast cancer risk, did not appear to be associated with worse survival. In particular, BRCA1/2 conferred a survival benefit for patients diagnosed above 50 years of age. PTV carriership of the remaining 25 genes, however, increased the risk of dying from any cause in all age groups. This observation is unexpected as while PTV 25genes carriership was associated with larger tumor size, the tumors were of a lower grade. A possible explanation could be that these 25 genes are associated with other cancers, for which we do not have information on to perform a competing risk analysis. An alternative explanation could be that some genes are important drivers of certain tumor subtypes associated with worse survival, which we do not have the statistical power to study. Further work in a larger study population, perhaps involving sub-analyses by further division of PTV 25genes carriership, would be helpful in understanding this observation. Nonetheless, the results suggest that PTVs in breast cancer genes influence survival. Larger studies with higher statistical power will be needed to elucidate which of the genes are specifically associated with disease outcome.
The role of BRCA1/2 in survival among breast cancer patients has been studied widely [24,25]. Most studies found that germline BRCA1/2 carriers were at a higher risk of death than their BRCA-negative counterpart, but there are indications that triple-negative breast cancer patients who are BRCA1/2 carriers have a better prognosis [23,24,26]. As BRCA1/2 carriership has the potential to affect the efficacy of chemotherapy, the effect on survival will need to be interpreted in light of the standard of care and patient population studied [26].
Among South-East Asians, it has been observed that Malay breast cancer patients tend to develop more aggressive tumors and have poorer survival rates [27]. The survival difference has been attributed to ethnic differences in lifestyle, socio-economic status, cultural values, tumor biology, and response to treatment [27]. In a previous study comparing Chinese and Malay breast cancer patients, a higher prevalence of BRCA2 mutations was found among Malay breast cancer patients [28]. Our larger study supports this finding and found significant differences in the prevalence of some breast cancer genes (BRCA1, MSH6, and MUTYH) between the two ethnic groups, suggesting that germline genetics may impact ethnic differences in breast cancer tumor characteristics and survival. However, a larger study is needed to validate these ethnic differences.
There are limitations to this study. Although this study is the largest to date to examine the impact of rare variants on breast cancer tumor characteristics and survival among patients of Asian ancestry, the number of carriers is still too limited for individual gene evaluation. The numbers are also too small to make definitive conclusions in specific ethnic subgroups. The effect of large germline structural variants (e.g., deletions, duplications, insertions, inversions, and translocations), which are not limited to genetic changes in coding regions, was not evaluated in this study [29]. Such variants have been documented to contribute to 10-25% of pathogenic variants for hereditary disorders [29]. However, germline structural variants are less frequent in breast cancer genes such as BRCA1 (0.8 to 6.9%) and BRCA2 (5%) in Asian populations [30,31]. As only germline variants were studied, we were not able to evaluate somatic second hits in tumors leading to the biallelic inactivation of the breast cancer genes. In view of potential bias, differences in the study design of each included cohort and how representative they are of the breast cancer demographics in general must be considered. For example, oversampling of BRCA1/2 carriers in the KOHBRA study could have led to bias in the estimates, but limiting the analyses to two hospital-based cohorts (SGBCC and MyBrCa, unselected breast cancer patients) showed no evidence of such bias. Finally, we were not able to consider breast cancer-specific survival as the data was not consistently ascertained across all the studies. However, it is noteworthy that this large Asian breast cancer population studied is novel and timely in view of possible Eurocentric biases in genetic studies [7,32].

Conclusions
It is important to identify germline carriers of high-risk breast cancer PTVs to prioritize individuals for inclusion in cancer surveillance programs, which has the potential to save lives [33,34]. The finding that PTV carriership is not just associated with higher breast cancer risk, but also more severe and fatal forms of the disease, suggests that genetic testing has the potential to provide additional health information and help healthy individuals make screening decisions.