SNP analysis
No DNMT1 or DNMT3B SNPs analyzed in our study were significantly associated with breast cancer. However, our findings of DNMT1 genetic polymorphisms associated with breast cancer may be limited by the SNPs analyzed, as previous studies exploring DNMT1 genetic polymorphisms found significant associations with breast cancer. Analysis of a central European Caucasian population by Kullmann et al. identified a significant reduction in breast cancer risk associated with the G allele of DNMT1 SNP rs2228612, a SNP not analyzed in our study [13].
Ye et al. conducted a two-stage case‒control study examining DNMT1 and DNMT3B genetic polymorphisms with breast cancer in Chinese women [21]. The study analyzed 1 SNP on DNMT3B that was included in our study, rs6058896, similarly finding no significant association with breast cancer risk. However, our study differed in that our population lacked homozygous recessive genotypes for rs6058896 [21].
The only SNP significantly associated with increased breast cancer odds in our study was rs7605753 on DNMT3A, with the AA genotype conferring a 30% increase in breast cancer odds compared to the GG or AG genotype (aOR: 1.30, 95% CI: 1.06, 1.59). Rs7605753 is an intron variant for DNMT3A, located at chromosome 2, position 25,270,318 (GRCh38.p14) [22]. This SNP has been associated with expression quantitative trait locus (eQTL) activity in two studies conducted by the Boyle Lab, [23, 24] suggesting a regulatory effect in gene expression. To our knowledge, no published studies on clinical significance have been conducted on this SNP, indicating a novel finding [22]. However, as this study did not adjust the significance level for multiple comparisons, these results should be interpreted as exploratory and hypothesis-generating, underscoring the need for further research into this SNP to understand its relation to cancer development and susceptibility.
Few studies have analyzed the association between DNMT3A genetic polymorphisms and breast cancer. Deroo et al. analyzed DNMT3A genetic polymorphisms with breast cancer using a case-cohort study design with participants in the Sister Study, analyzing 1 SNP that we also analyzed [25]. While the SNP was not significantly associated with breast cancer in either study, our studies differed in the direction of the effect. Deroo et al. found rs7575625 to have a hazard ratio of 1.05 (95% CI: 0.85, 1.29),25 whereas our study reported a nonsignificant inverse association with breast cancer odds (aOR: 0.87, 95% CI: 0.70, 1.07).
Haplotype analysis without population stratification
In the haplotype analysis without population stratification, haplotypes TACGA and TACAT were significantly associated with increased breast cancer odds of 42% and 40%, respectively, compared to haplotype CGCGA.
Compared with CGCGA, TACGA had 2 different alleles: the major allele of rs12991495 and the minor allele of rs7605753. In the SNP analysis, the minor allele of rs7605753 had a statistically significant 30% increase in breast cancer odds (aOR: 1.30, 95% CI: 1.06, 1.59). For rs12991495 SNP analysis, the minor allele displayed a protective association compared to the major allele but this association with breast cancer odds did not reach statistical significance (Table 3). While not statistically significant in the SNP analysis, haplotype analysis revealed that this SNP may contribute to increased odds of breast cancer when combined with other polymorphisms, as TACGA had an increased odds of breast cancer compared to the analysis with SNP rs7605753 alone.
The other haplotype identified, TACAT, had 4 differing alleles from CGCGA: the same 2 alleles as TACAT, the major allele of rs7575625, and the minor allele of rs10196635. In the SNP analysis, the minor allele of rs7575625 had a nonsignificant protective effect on breast cancer odds. Rs10196635, while not statistically associated with breast cancer odds, was particularly interesting, as the estimated increase in breast cancer odds was 17% for every minor allele present (aOR: 1.17, 95% CI: 0.96, 1.43). The minor allele was very infrequent in the study population at only 6.15% (Table 2), potentially underpowering the study to detect any true associations present. These results spark interest in future research of this SNP, particularly for individuals with a homozygous recessive genotype. Additionally, this haplotype was interesting in that only 5.68% of the population was estimated to have this haplotype, producing wider variation in the 95% confidence interval estimation, yet still conferring statistically significant results. Therefore, this combination of SNPs warrants further investigation in populations with higher percentages of these SNPs present.
Haplotype analysis with population stratification
Significant differences between haplotypes and breast cancer associations were observed by race upon inclusion of the DNMT3A SNP rs2304429, which indicated population stratification. As Whites constituted the majority of the study population, similar results occurred as those in the analysis without population stratification. For the White stratum, including SNP rs2304429, we found haplotype CTACGA associated with a 47% increase in breast cancer odds compared to TCGCGA (aOR: 1.47, 95% CI: 1.13, 1.90). According to the SNP analysis, the minor allele of rs2304429 was insignificantly associated with increased breast cancer odds (Table 3), potentially explaining the 5% increase in breast cancer odds from TACGA to CTACGA.
In the Black stratum, CTACGA was significantly associated with increased odds of breast cancer; however, estimates had considerably less precision compared to the White stratum, likely due to the low frequency of the referent haplotype in the Black stratum at 4.87% compared to 13.60% in the White stratum. In this subgroup analysis, CTACGA was associated with a 5.72-fold increase in breast cancer odds compared to TCGCGA after adjusting for covariates (95% CI: 1.34, 24.39). Interestingly, CTACGA was more common in the Black stratum, with 17.73% of the population estimated to have the haplotype, compared to 15.21% in the White stratum. Another haplotype of interest was identified in this stratum: CTGCAA. This haplotype was associated with an 8.62-fold increase in breast cancer odds compared to TCGCGA (95% CI: 1.93, 38.45), an association not present in the White stratum.
CTGCAA has the major allele of rs7605753, an unusual result as the minor allele rs7605753 was the only SNP significantly associated with breast cancer in the SNP analysis (Table 3). However, examining the SNP by race, rs7605753 was not significantly associated with breast cancer in the Black stratum (aOR: 1.34, 95% CI: 0.63, 2.87) (Table S2). This result could indicate that there truly is no association between rs7605753 and breast cancer in the Black population of our study, or we failed to detect the association due to a small sample size of Black participants (n = 316) compared to White participants (n = 2032). This potential limitation highlights the importance for future studies to prioritize diversity and sufficient sampling so these exploratory findings in SNP and haplotypes associations by population may be sufficiently determined.
We also conducted DNMT3A haplotype analysis in the Black stratum using the most common haplotype identified in this population, CTGTAA, to increase stability and narrow confidence intervals for our analysis (Table S1). Utilizing this referent haplotype, CTACGA and CTGCAA remained significantly associated with increased breast cancer odds, with greater certainty of the association. CTACGA had an estimated 1.92-fold increase in breast cancer odds compared to CTGTAA after adjusting for covariates (95% CI:1.01, 3.63), indicating this haplotype is significantly associated with breast cancer regardless of racial background. CTGCAA had an estimated 2.89-fold increase in breast cancer odds compared to CTGTAA after adjusting for covariates (95% CI: 1.36, 6.12). This haplotype differed from the referent haplotype only by 1 allele: the major allele of rs11892646. In the SNP analysis, the major allele of rs11892646 had a nonsignificant increased association with breast cancer odds (aOR: 1.07, 95% CI: 0.89, 1.30). The haplotype analysis suggests rs11892646 may confer increased breast cancer odds that were masked in the overall analysis due to lack of population diversity, warranting further investigation into this SNP.
Our results suggest that in the general population, DNMT3A haplotype TACGA could be associated with increased breast cancer odds, with further large-scale studies needed to explore these exploratory results due to increased risk of false positive findings. Results also revealed important racial/ethnic differences upon population stratification, highlighting how aggregate results may conceal important association differences, as some populations have differing allele frequencies. This study was a majority White population, limiting statistical assessment of haplotypes among races, particularly for Black study participants. However, haplotype analysis with population stratification suggests that DNMT3A haplotype CTACGA could be significantly associated with increased breast cancer odds compared to TCGCGA, as Black women had an increased association (aOR: 5.72, 95% CI: 1.34, 24.39) compared to White women (aOR: 1.47, 95% CI: 1.13, 1.90).
While greater research is needed to assess these differences in measures of association according to race, a leading hypothesis may be potential gene‒environment interactions or epigenetic changes in this population due to factors related to systemic racism, such as environmental exposures, housing, or psychosocial stressors which may disparately affect Black women in our population. A study conducted by Jasmine M Miller-Kleinhenz et al. identified 5 differentially methylated CpG sites that were associated with breast carcinogenesis and contemporary redlining in breast cancer patients from Emory University Hospitals in Atlanta, Georgia. Their study revealed significant associations with this systemic racism practice and epigenetic age acceleration, providing evidence that social determinants and environmental exposures may contribute to one’s epigenome, affecting DNA methylation patterns and contributing to breast cancer development [26]. Additionally, a cross-sectional study in Baltimore found epigenetic differences in breast tumors attributable to neighborhood deprivation level, which was higher for Black women compared to their White counterparts [27]. These studies provide evidence that epigenetic modifications related to social and environmental influences negatively affect breast cancer, highlighting the need to research how these DNMT genetic polymorphisms may contribute to differing breast cancer risks in conjunction with these exposures.
