To our knowledge, this report is the first to suggest a possible association between variants in BRCA1/2 and CVD in human populations. However this observation requires confirmation, given that the effect size appears to be small, and our meta-analysis did not reach statistical significance. It is possible that the association we observed in the SHARE studies was due to chance alone, and is not a true association.
Very little has been published with respect to shared genetic predisposition to cancer and CVD. However, mechanistically, several common pathways are emerging. Most prominent in recent publications is the endothelin axis, with reports linking endothelin-1 to prostate cancer tumorigenesis [25
]. Additionally, animal models demonstrate upregulation of DNA repair mechanisms in heart failure induced by ischemic damage [26
]. As an example, topoisomerase-II-alpha, a protein involved in various aspects of DNA repair, is overexpressed in unstable plaque from human carotid atheromas [27
]. Given its integral role in genomic stability and DNA repair, a similar role in the pathogenesis of coronary atherosclerosis, plaque rupture, and response to ischemic damage may exist for BRCA1/2
. Finally, several SNPs associated with metabolic diseases, identified through genome wide association studies, have roles in cancer and chronic disease predisposition. For example, variants in TCF7L2
predispose to both colon cancer and type 2 diabetes while variants in HNF1B
predisposes to type 2 diabetes and prostate cancer [28
Our study suggested a possible association between rs11571836 and rs1799943 and CVD in a multi-ethnic population. Analysis by ethnicity confirmed a statistically significant association with Aboriginal and South Asian ethnicities for rs11571836, while for rs1799943 the association was statistically significant only in the Aboriginal group. It is worth noting however, that with the exception of the Chinese group, all odds ratios trended in the same direction (ie. below 1) for both SNPs. This supports the possible explanation that the current analysis was underpowered to detect an association by individual ethnicity. In fact, only 17% and 6.9% of individuals with CVD in the study were European or Chinese respectively. Alternatively, lack of association between these SNPs and CVD in specific ethnicities may represent differences in LD in the region encompassing BRCA2,
as there are significant differences in LD structure in this region among ethnicities (Figure ). In silico
analysis of these two SNPs suggests both may modulate BRCA2
transcript levels [29
], but there have been no published functional studies to date. Similarly, neither SNP has been definitively associated with breast cancer susceptibility. Thus, it is quite possible that rs11571836 and rs1799943 are not directly associated with CVD, but instead are in LD with other functional variants associated with CVD. Finally, there may be ethnic-specific differences in the pathogenesis of CVD such that BRCA
variation is only relevant to CVD in certain ethnicities, although this seems unlikely given the consistency of other risk factors for CVD across ethnicities [15
We were unable to demonstrate a statistically significant association between rs11571836 and incident MI in two large case controls studies of incident MI in South Asians, although a trend towards an association was present in INTERHEART. It is possible in INTERHEART that a larger sample size may have resulted in a statistically significant association, as the current sample size was underpowered (44% power) to detect a 15% risk reduction in MI risk for allele frequencies of 20%. A sample size of approximately 2600 cases and a similar number of controls is required to reach 80% power. Based on sample size, the PROMIS analysis was adequately powered, however, it is possible that utilizing imputation to infer the genotype of rs11571836 may have resulted in divergent results from what would have been observed had the SNP been directly genotyped. While the literature suggests good concordance between imputation and direct genotyping [30
], it is possible that the HapMap reference samples were not adequate for this region. Alternatively, subtle differences in LD patterns between South Asian populations could explain the lack of a reproducible association between rs11571836 and acute MI in the SHARE, INTERHEART and PROMIS studies. Conversely, the lack of replication would be anticipated if the association between BRCA2
variants and CVD demonstrated in the SHARE studies occurred by chance. One means of further clarifying this possibility is to examine the association between BRCA2
variants and CAD in published GWAS studies. There was no association between either rs11571836 or rs1799943 and CAD in the Wellcome Trust Case Control Consortium (WTCCC) GWAS study [31
]. However, the WTCCC study was composed exclusively of White Europeans, unlike our multi-ethnic or South Asian study populations. Unfortunately, lack of publically available data from addition multi-ethnic or South Asian GWAS studies limited our ability “look-up” these SNPs directly in additional studies.
Alternatively, variants in BRCA2 may be more relevant to the pathogenesis of other components of the composite CVD outcome utilized in SHARE (for example ischemic stroke), rather than acute MI. One significant limitation of utilizing the composite CVD outcome in genetic association studies is the heterogeneous pathogenesis of the outcomes studied. In the SHARE studies, the low number of events precluded an assessment of individual components of the composite CVD outcome. Testing the association between BRCA variants and measures of CVD other than MI, ideally from multiple ethnicities, could confirm or refute this hypothesis. In addition, experimental data has thus far only demonstrated a role for BRCA1 in the modulation of cardiomyocyte function following ischemic damage. Due to resource constraints, our analysis was limited to SNPs from BRCA2 that had been previously genotyped, however, investigating the role of SNPs in BRCA1 would be desirable. Finally, data from the murine knock-out models described earlier allude to the possibility that BRCA1/2 may play a greater role in modulating the severity of cardiomyocyte damage following an insult, rather than being directly implicated in initiation of cardiac injury. Therefore, testing the association between variants in the BRCA genes and the severity of a CVD outcome might prove more useful.