1. Introduction
Breast cancer is the most common cancer among women in populations of European descent, representing 25% of all cancers [
1]. As in any complex disease, there are multiple genetic and environmental factors involved in its development. Risk factors, such as hormonal levels and lifestyle, have been reported as some of the most important [
2]. However, genetic alterations underlying breast cancer have not been fully described, and only 15 to 30% of cases are heritable or due to genetic transmission. Susceptibility genes with high or moderate penetrance (i.e.,
BRCA1, BRCA2, PALB2, ATM, and
CHEK2) have been identified in 30 to 40% of patients with a family history of breast cancer, but only in 5% of sporadic breast cancer cases. A large proportion of heritability still needs to be explained by low-penetrance variants and, therefore, a major effort has been underway to identify single-nucleotide polymorphism (SNPs) associated with breast cancer risk. These SNPs are not very informative by themselves, but together may account for almost half of the familial relative risk (FRR) [
3]. A set of 170 SNPs strongly associated with breast cancer was described for familial cases, explaining up to 18% of the FRR in European populations [
3]. Breast cancer heterogeneous incidences and mortality rates indicate a complex interaction between risk factors, one of the most relevant being a population’s geographic location [
1]. The higher rates of incidence and mortality observed in populations of European descent suggest a population specific association [
1,
4]. These marked differences between European and non-European populations are expected to have an impact on the clinical presentation of breast cancer in admixed populations [
5,
6].
Latin American countries show great disparities in incidence and mortality rates across the continent. Uruguay, Argentina, and southern Brazil have incidence rates as high as southern Europeans (72, 72, and 70 cases per 100,000 women per year, respectively), while Bolivia has an incidence rate of 13 cases per 100,000 women per year [
4]. This translates into almost 2000 annual cases of breast cancer for Uruguay. Uruguay also has the highest breast cancer mortality rate in Latin America [
7]. The reasons for these high rates are not well understood. Environmental factors, such as increased consumption of red meat and fat, as well as a reduced intake of vegetables, have been identified as risk factors for breast cancer in Uruguay [
8]. However, genetic factors associated with the European ancestral genetic contribution could also be playing an important role. In fact, countries with the highest breast cancer incidence in Latin America exhibit considerable European genetic contributions to their populations. For instance, urban admixed populations from Argentina, Brazil, or Uruguay have European genetic contribution ranging from 70 to 90%, whereas European ancestry in Bolivia amounts to less than 20% [
9,
10].
In Latin America, genetic analyses have been focused on
BRCA1 and
BRCA2. Founder mutations that have been described in Chile, Brazil, Colombia, Uruguay, and Argentina can be explained by the complex demographic history of Latin American populations. The admixture and demographic dynamics promoted the drift to higher frequencies of rare variants [
11]. Almost all Latin American countries show more mutations in
BRCA1 than in
BRCA2, as observed in European countries, except from Uruguay, Costa Rica, Puerto Rico, and Cuba, with higher numbers of BRCA2 mutations [
12].
The attempt to understand the genetic contribution to breast cancer in Latin America was centered in familial, hereditary, and early onset breast cancer, sometimes without making a distinction between patients regarding familial history or age at diagnosis [
13]. Just a few studies of sporadic breast cancer have been conducted in Latin America, and they have focused on different candidate genes than our study [
14,
15,
16,
17].
In Uruguay, European ancestral proportions from autosomal markers show no substantial association with breast cancer, but a weak association of European mitochondrial haplogroup H with breast cancer risk was found in a previous study [
18]. This represents a controversial issue; genetic ancestry seems to be a risk factor for breast cancer in Mexican and Hispanic women [
19]. In other populations, from the Asia-Pacific region, and Europe, breast cancer risk is correlated with other factors, such as body mass index (BMI) [
20,
21]. The heterogeneous composition of the Latin American populations represents a challenge to understand the development of breast cancer. A higher number of specific-population mutations have been described for Uruguay compared to neighboring countries Argentina and Brazil [
22,
23,
24].
However, given the strong European genetic contribution, the small population size, and quasi-null immigration since the 1950s, it is likely that the same candidate gene variants found amongst breast cancer cases in Europe could be increasing disease risk in Uruguay as well.
In this study, we analyzed a set of 144 candidate gene SNPs, previously associated with sporadic breast cancer in European patients, in the Uruguayan population. Common susceptibility loci are reported in genes whose rare variants explain a major fraction of hereditary breast cancer risk, which may, together, explain an appreciable fraction of the genetic variation in sporadic breast cancer.
3. Results
After quality control, data was available for 118 SNPs in 172 cases and 160 controls. All of the variants genotyped were found to be under HWE in controls after FDR correction (
Supplementary Table S1).
The logistic regression results for the eight SNPs associated with breast cancer are listed in
Table 1. Although none of them remained significant after FDR correction, the association of SNPs in
CNTNAP2 and
VDR genes with the disease is supported by the randomized analysis.
All, but rs1799950 of BRCA1, are risk variants with moderate effects ranging from 38–88% enhanced risk. BRCA1 rs1799950 minor allele (G) showed a protective effect of ~50% decreased risk for the carriers of this variant.
Since ESR1 rs2941740 had a genotyping rate lower than 90%, it was excluded from the traditional analysis. Nevertheless, it stands out with the highest percentage of significant sub-samples in the randomized analysis (86% as a single SNP and up to 91% within haplotypes) and is, therefore, worthy of mention.
Despite
BRCA2 SNPs not showing an association with breast cancer using the traditional approach, this gene appears associated in the haplotype analysis (
Table 2). The haplotype frequencies for
ESR1, FGFR2, VDR, and
BRCA2 genes are presented in
Table 2.
We found a statistically significant association between haplotypes in these genes and breast cancer risk by both methods of analysis. Even though, after FDR correction, none of them remained significant with the traditional approach, they are strongly supported by the RDM analysis.
BRCA2, VDR, and ESR1 genes showed two significant haplotypes, one that acts like a protective factor, and the other that acts like a risk factor. Both BRCA2 haplotypes and the ESR1 protective haplotype revealed a strong support from RDM. The FGFR2 gene exhibited a strong association (also with a strong support from RDM) with breast cancer only for the protective haplotype, which includes the major allele from rs2981582. This SNP was previously found as significantly associated with the disease, its minor allele increasing risk.
To understand the genetic structure of
BRCA1, BRCA2, ESR1, and
VDR genes in the Uruguayan population, a set of SNPs was chosen with a worldwide representation. The frequency of the Uruguayan haplotypes was compared with that of parental populations, i.e., African, European, and Asian populations (as a surrogate for Native Americans), and other admixed Latin American populations (
Supplementary Table S2). These haplotypes from the selected set of SNPs show no statistically significant differences between patients and controls. In general, haplotype frequencies in the Uruguayan population are very similar to those in Puerto Rico (already similar in
BRCA2 frequencies) [
12].
Figure 1 illustrates the geographic distribution of
VDR haplotypes (rs7975232|rs1544410|rs2239182|rs2239179). Some differences are remarkable. For example, the
VDR ACCC haplotype frequencies in Uruguay and Colombia are close to each other, but are higher than those in European populations.
The PCA helps to understand the general genetic structure of the samples. When the Uruguayan samples are compared against the European, African, or Asian populations, they show a clear European pattern. However, the controls are more homogeneous, while the patients comprise a more heterogeneous group (
Figure 2).
4. Discussion
Breast cancer is a common cancer in Latin American countries. Previous studies in Latin American populations focused mainly on hereditary breast cancer and
BRCA1/
2 mutations [
31]. This study was performed in a sample of women with sporadic breast cancer genotyped for 118 SNPs in 98 genes. Most of the SNPs (110) are present in the iCOGs microarray [
32]. The results indicate that there is evidence for an association between polymorphisms in well-known candidate genes for breast cancer and the disease, however, with multiple testing stringent criteria the associations faded.
To the best of our knowledge, this is the first study to explore these candidate genes in sporadic breast cancer patients in South America. Studies conducted in other Latin American countries either examined hereditary breast cancer or did not take into account family history at the time of selecting the patients. The few sporadic breast cancer studies available analyzed a small number of SNPs that are different from those included in the present study [
13].
The admixture structure in Latin American populations represents an important factor to be considered in breast cancer studies. In the past decades, admixture mapping was developed as a powerful method of gene mapping [
33]. This method takes advantage of the linkage disequilibrium generated during the admixture process in recently admixed populations. It relies on the idea that genes are involved in population-specific diseases (such as breast cancer in Europeans) and could be identified by ancestry informative markers. Even though this method has strong theoretical support, few studies were performed using Latin American samples, and even fewer involving breast cancer [
34]. Although our study was conceived as a candidate gene study, at a population level, the candidate gene SNP frequencies showed differences between geographic populations, with the Uruguayan population closely resembling European populations. This confirms what was observed in a previous study where there was no increased risk of breast cancer associated with European ancestry [
18], contrary to findings in other Latin American and Hispanic populations [
35,
36]. This is controversial under the hypothesis that breast cancer is a population-specific disease. When we consider other diseases, a clear excess of European genetic ancestry was detected in melanoma patients on chromosome 16 [
37], which is expected for this type of cancer [
38]. Moreover, other diseases, such as hemoglobinopathies or diabetes, showed an increased representation of non-European mutations, indicating the genetic contribution of Native American or African populations as expected [
39,
40,
41].
In familial breast cancer,
BRCA1 and
BRCA2 have been the primary targets in Latin American studies, and one of their main achievements is the detection of new variants only present in those populations. Our experimental design did not allow us to identify new variants in the Uruguayan population. The
BRCA1 or
BRCA2 haplotype architecture is not different from that present in Europeans or other Latin Americans. However, the
BRCA1 rs1799950 G variant described as benign in hereditary breast cancer, was protective in our study, in agreement with other sporadic breast cancer studies in European populations [
42]. In fact, the rs1799950 G allele is present mainly in Europe (6%) and in Latin American populations (5%). However, it is important to consider that most of the variants known today were discovered in populations of European descent. The large proportion of European ancestry along with the small sample size of this study are probably contributing to us failing to discover additional haplotypes that could explain risk in the Uruguayan population.
The sub-sampling approach provided a different approximation to the analysis of the variants. Using this method, a random set of sub-samples was created by a variable selection procedure. The method reveals that
ESR1,
FGFR2,
VDR,
CNTNAP2, and
BRCA2 have a weak association with sporadic breast cancer in the Uruguayan population. The reduced sample size and the bias detected for SES decreased statistical power and confounded the traditional analysis. Nevertheless, the randomized approach suggests that genetic risk factors for sporadic breast cancer in Uruguay are similar to those in other populations of European ancestry [
43]. Moreover,
BRCA2 seems to be a good candidate gene to explore in the future given the information available from previous studies in hereditary breast cancer in Uruguay. In these previous studies more mutations were described in
BRCA2 than in
BRCA1 with some of them being population-specific mutations [
22,
44]. A plausible explanation is that the Uruguayan population is the most aged in South America [
45], and
BRCA2 mutations are slightly more common between older women with sporadic breast cancer [
46]. Moreover, an association of common variants in genes, such as
FGFR2 (rs2981582) and
MAP3K1 (rs16886165), with
BRCA2 mutant carriers, has been described [
47], but it is not present in our study (data not shown). In addition, rs2981582 in
FGFR2, rs16886165 in
MAP3K1 and rs3803662 in
TOX3 are associated with breast cancer in populations of European ancestry [
48], and the first two were also associated with breast cancer in Uruguay. In U.S. Hispanics [
49], the association is weaker, a difference likely explained by the high proportion of Native American genetic contribution to southwest Hispanics [
50,
51]. Even if the relationship is not causative, these SNPs may be proxies for major candidate genes.
The subsampling method allowed us to reintroduce
ESR1 rs2941740 in the haplotype analysis and the haplotype rs2941740_G|rs1999805_T|rs827423_C showed a strong support. In a recent study, the authors identified the 6q25 region, where
ESR1 is located, as associated with breast cancer in U.S. Latinas, Mexican, and Colombian patients. The rs2941740 is one of the 48 SNPs showing robust evidence of association in GWAS. The 6q25 region was also previously described as associated with breast cancer and Native American ancestry in U.S. Latinas. In this region, breast cancer risk decreases with higher Native American genetic contribution, which suggests that a particular genomic structure related with breast cancer risk may exist [
34]. Most of the variants described in the Hoffman et al. study are common in non-European populations and no direct relationship with breast cancer could be established. The
ESR1 gene, which encodes the estrogen receptor α, is an important candidate gene in the 6q25 region and risk allele carriers show abnormal gene expression [
52]. The mechanism behind gene expression regulation by the risk alleles seems to be unclear. There are examples of the influence that the variability in the genomic context can cause in the chromatin status of a chromosomal region in cancer, or cardiovascular diseases [
53]. However, perhaps the most paradigmatic example is the adult persistent expression of the lactase gene due to a single nucleotide change in Europeans [
54].
Recently, we described the influence of epigenetic ancestry in sporadic breast cancer. DNA genome demethylation was positively correlated with African ancestry in breast cancer patients, but not in control individuals. The data suggests that the usual DNA hypomethylation in breast cancer patients is reinforced by the genome structure due to population-specific contributions [
55]. Complex interaction between population-specific variants and DNA methylation status, some of them synergistic, could be influencing overall risk. How each variant modifies this risk in specific populations still needs to be elucidated.
Even though we found a weak association between European ancestry and breast cancer in Uruguay [
18], there are risk genes detected in Europeans that appear to be risk genes in Uruguay too. The data suggest that European-specific variants are important, but also that the ethnic genetic contribution might be affecting through mechanisms that need further exploration in the admixed population of Uruguay.