Common BRCA1 and BRCA2 Mutations among Latin American Breast Cancer Subjects: A Meta-Analysis

Methods: Pubmed, EBSCO, and OVID databases, and study bibliographies were systematically searched for observational studies that examined for mutations in BRCA1 and BRCA2 until March 2015. The pooled prevalence was obtained using the inverse double arcsine square root method. Publication bias was assessed by Begg and Mazumdar’s test and the Egger’s test. The sensitivity was determined by reevaluation of the pooled estimate after removal of one study.


Introduction
Breast cancer is the most common cancer among Latin American women and the leading cause of cancer-associated deaths [1,2]. It is estimated that 114,900 new cases and 37,000 deaths occur in Latin American populations annually [3]. Unfortunately, Latin American women have a poor 5 year survival rate than most other ethnic groups [4] and the incidence is increasing annually in these countries [5][6][7]. Genetic cancer risk assessment has become an integral part of disease prevention, especially in countries such as Spain and USA; however, the limited availability of clinical gene testing has prevented the implementation of prevention programs in Central and South American countries.
Five to ten percent of all breast cancers in Latin American women are attributed to germ-line mutations in the breast cancer susceptibility genes BRCA1 and BRCA2 [8,9]. Conversely, in low-income countries with restricted financial resources for genetic testing, this percentage has been suggested to be underestimated. The lifetime risk of developing breast cancer increases up to 80% with certain BRCA1 and BRCA2 mutations [10]. BRCA mutations prevalence varies between country as well as ethnic groups [11]. With more than 300 documented BRCA1 and BRCA2 mutations found in Hispanic countries [12][13][14][15][16][17][18][19][20][21][22] and with limited reports describing the prevalence of BRCA1 and BRCA2 mutations, which can range from 0% to over 50%, we therefore conducted a meta-analysis to determine the prevalence of certain pathogenic mutations in Central and South American countries. studies had to meet the following criteria: studies focused on examining the prevalence of BRCA1 or BRCA2 mutations in human subjects, with breast cancer, from Latin American countries or their descendants. Non-human studies, in vitro or in vivo studies, reviews, studies that failed to indicate the prevalence of mutations, or focused on other than breast cancer were excluded.

Data extraction
Two of the authors extracted all data independently. If there was a disagreement, another author assessed the publication in question. If a single sample was believed to be use in multiple publications, the publications were assessed to determine which one was the most representative and that data was used for that mutation, or the corresponding author was contacted to resolve the issue. The data collected were geographical location, criteria use to select sample, sample size, the mutation and the number of positive individuals, method used to detect the mutations, and exons and introns examined.

Statistical analysis
The term "Prevalence" corresponds to relative prevalence rate, which was defined as the total number of positive individuals with breast cancer for a specific mutation divided by the total number of breast cancer cases. The prevalence and the 95% confidence interval (95% CI) were calculated for each mutation. Next, the pooled prevalence estimate was calculated using the inverse double arcsine square root method (Stuart-Ord) [23]. It is worthy to note that only studies that examined the exon or the intron of the BRCA1 or BRCA2 gene or specifically indicated they examined for the mutation of interest were included in the meta-analysis, independent if the authors found the mutation or not. Heterogeneity was determined using the ψ 2 -based Q test and its degree was assessed by the I 2 value (inconsistency index). The Fixed Effects Model was used when the sample was considered homogeneous (Mantel-Haenszel method) [24] and the random effects model was used when the sample was considered heterogeneous (DerSimonian and Laird Method) [25]. The stability and sensitivity of the results were assessed by removing one study and re-calculating the pooled prevalence. Publication bias was evaluated by Begg and Mazumdar adjusted rank correlation asymmetry test (Kendall's tau) and the Egger regression asymmetry test [26,27]. The Fisher's exact test was used to determined difference of frequencies between groups. Statistical analyses were performed using StatDirect Statistical Software version 3.0.147 (Cheshire, UK). P-values <0.05 (two-sided) were considered statistically significant.

Common BRCA1 mutations among Latin Americans
Using the 32 studies, we identified 175 BRCA1 mutations found in Latin American breast cancer patients (Supplemental Table 1). A majority of mutations were single nucleotide variants (n=105, 60.0%, Table 2). However, when only pathogenic mutations were identified (n=80, 45.7%), by using the CLINVAR database or the studies themselves, 36 were deletions (45.0%), 33 were single nucleotide variants (41.25%), and 11 were insertions (13.75%). Fifty pathogenic mutations (62.5%) were reported only once, whereas fifteen mutations (18.8%) were reported four times or more. The most identified mutation for BRCA1 were 185delAG and A1708E, found in eleven and ten studies, respectively.
Next, to determine if these mutations were consistent between groups, we examined the prevalence these mutations in a selected subgroup of subjects-Mexicans and Mexican descendants were selected because they had largest number of studies-and compared them to the rest of the Latin American population. Anton Culver 2000, Hall 2009, John 2007 (except for 185delAG analysis) and Vogel 2007 were excluded because Latin American origins could not be determined. The prevalence for the Mexican subjects for del exon 9-12 was 3.35% (95% CI: 1.18-6.57%, Figure 3A), for 185delAG was 0.94% (95% CI: 0.18-2.29%, Figure 3B), for R71G was 1.24% (95% CI: 0.67-1.98%, Figure 3C), and A1708E was 0.67% (95% CI: 0.35-1.08%, Figure 3D). The remaining Forest Plots can be found in supplemental material (Supplemental Figure 2). Comparison of these mutations indicated that the frequency between Mexicans and other Latin American countries was not similar. The frequency of the R71G mutation was higher in Mexicans than other Latin Americans (Fisher's exact test, p=0.0034), but there was no difference between the two groups for 185delAG (p=0.702) and A1708E (p=0.802). It is important to note that the deletion of BRCA1 exons 9-12 was only found in Mexican subjects.

Test for sensitivity and publication bias
We assessed the publication bias for pathogenic mutation prevalence for BRCA1 and BRCA2 ( Figure 6) American subjects, the pooled prevalence was resistance. These results suggest limited bias, which would minimally affect the results of the meta-analysis. The publication bias for each individual mutation was assessed for each meta-analysis and listed in Table 3.

Discussion
Breast cancer is the most common cancer among females in Latin American countries. While other risk factors, such as high estrogen exposure and age of menarche increases the risk of breast cancer development, mutations in the BRCA1 and BRCA2 genes have a more profound effect in certain population [53]. Genetic testing is an expensive procedure that can aid the development of specific treatment options. Unfortunately, with the large amount of possible mutations to test for, there is a need for a consensus on specific mutations.
In our study, we determined the prevalence of all BRCA1 and BRCA2 mutations among Latin American breast cancer subjects. This study is similar to Wang et al. and Forat-Yazdi et al., who determine the prevalence in breast cancer families and Iranians, respectively [54,55]; however, neither study examined the prevalence in nor contained Latin American subjects, specifically. Furthermore, Wang et al. and Forat-Yazdi et al. excluded studies from their meta-analysis that determine the absence of a mutation from their sample, which may have led them to overestimated the mutation's prevalence. Here, we did incorporate any report that examined the region for these 80 BRCA1 and 69 BRCA2 pathogenic mutations. For example, Lara 2012 and Solano 2012, the two studies that found BRCA2 H372N, determined the frequency to be 55.2% and 25.5%, respectively, which would give a pooled prevalence of about 38.8%; however, including reports that determine the absence of this mutation, gives the pooled prevalence of 0.88%. Since many studies failed to determine this mutation in larger samples, this result would suggest that including the negative data would yield a more accurate value. The most common BRCA1 mutation for Latin American breast cancer subjects was deletion of exons 9-12. This mutation leads to an inactive form of BRCA1. However, this mutation was only found in Mexican studies and no other country. It is believed to have originated in the state of Puebla [31]; however, Weitzel et al., found this mutation present in subjects that originated from other distant regions of Mexico [21]. This would suggest emigration has led to the dispersion of this mutation. Interestingly, this mutation has not been reported in Southern Mexico or Guatemala, maybe due to a lack of studies. More research is required to determine the regions in Mexico where this mutation is prevalent.
The second most prevalent BRCA1 mutation was 185delAG. Unlike the deletion exon 9-12, the 185delAG mutation has been found in many different regions of Latin America (Argentina, Brazil, Chile, Mexico and Peru). Numerous reports have demonstrated that the BRCA1 185delAG increases the risk of developing breast cancer. This has led to the additional screening of subjects from certain ethnicities for the presence of this mutation, such as the Ashkenazi Jewish descendants. Nonetheless, within the Latin American populations, there is inconsistent evidence about the prevalence of this mutation in breast cancer subjects, which ranged from 0.0% (19 studies) to 5.22%. Here, we provide evidence that 185delAG frequency was significantly prevalent and was not different between Mexico and all other Latin American countries (0.90% vs. 0.94%, p=0.70); then again, not all countries in Central and South America were properly represented. A similar result was determined for BRCA1 A1708E and BRCA2 E49X. On the other hand, BRCA1 R71G was found in Mexicans as well as Argentinians but was more prevalent in Mexico. Moreover, BRCA2 6174delT and H372N were found in Argentina, Brazil, Chile, Costa Rica, and Venezuela and not Mexico. As well, the BRCA2 3492insT, G273R, and W2586X were found only in Mexico and not any other Latin American country. These examples beg the question that should there be a select set of mutations examined for certain regions of Latin America. Furthermore, it demonstrates that specific regions of Latin American are associated with certain BRCAs mutations.
Central and South America were settled by many different groups from the Iberian Peninsula. We posit that due to emigration from certain regions of Europe has led to the presence of certain mutations in specific regions of Latin America. For example, the BRCA2 3492insT  polymorphism has been identified in many regions of Spain: Austurias [56], Valencia [57][58][59][60], Aragon [60], Castilla-Leon [61], and Madrid [62], where the mutation frequency ranged between 0.22-2.08%. Interestingly, this mutation has not been reported in Barcelona or Galicia [63], suggesting that this mutation originated from a specific region of Spain. In Mexico, a majority of the population ancestry is from Spain, implying that the BRCA2 3492insT could be found in Mexico. Indeed, this meta-analysis does provide evidence that BRCA2 3492insT was significantly present among Mexicans. A majority of the reports focused on Mexican subjects used in this meta-analysis failed to determine the presence of BRCA2 3492insT. These reports focused on highly-populated regions of Mexico City, the State of Mexico, Nuevo Leon and Veracruz and surrounding regions-most of the states located in the center of Mexico. However, Weitzel et al. did observe the mutation in subjects from Durango, Guerrero, Jalisco, Sinaloa, Sonora, and Zacatecas [21]. With the exception of Guerrero, these states are located in Western Mexico, which would posit the notion that these states present this mutation due to a more pronounced Spanish influence. This is supported by the work of Moreno-Estrada et al., which indicates that subjects from Western Mexico do have a greater Spanish genetic composition than states from Eastern or Southern Mexico [64]. Interestingly, the BRCA2 3492insT mutation has not been observed in other Central or South American countries [15,17,33,39,43,[47][48][49][50]. As seen with the San Luis Valley studies, due to Spaniard expeditions, the BRCA1 185delAG mutation was introduced and spread among the region, leading to its extraordinary prevalence [65,66]. This would also support the notion that the BRCA2 3492insT mutation is specific for Mexico and Spain. Overall, the region-specific mutational patterns are likely caused by original differences of the native population and induced differences by emigration.
In this meta-analysis there are a few limitations. First, the included studies collected genomic DNA from either blood or a buccal sample, or both. Recent reports have suggest that a blood sample is the superior method, suggesting that studies that used a buccal sample could underestimate or fail to determine the prevalence of BRCA mutations. Second, studies with smaller sample sizes does increase the prevalence of a single case found but also decrease the chance of discovering a positive case among the sample. Third, due to the emigration pattern that has led to the high diversity of Latin America, many regions of Central and South America are under-represented. For example, as mentioned above, the 3492insT mutation was mainly found on the Western part of Mexico and not in the Northern or the Eastern regions or any other Latin American country. This result demonstrates the need for more region-specific analyses.
In conclusion, this study identifies the most prevalent BRCA1 and BRCA2 mutations found in Latin American breast cancer subjects. Furthermore, we demonstrate that certain mutations are only specific for certain regions, whereas others are constant throughout Latin America. This information will aid in developing a more narrow genetic screening strategy based on the subject's background and lead to cheaper testing. However, with most Latin American countries have not been assessed for BRCA1 and BRCA2 mutations, further studies are required.