Breast cancer is the most common cancer among women in the United States, accounting for 29% of all newly diagnosed cancers [1]. Although breast cancer incidence at older ages is lower among African American (AA) women than European American (EA) women, the incidence rate is higher in AA women at younger ages (50 years) [2]. More importantly, AA women are more likely to be diagnosed with aggressive tumors that are high grade and negative for estrogen receptors (ER), which are often associated with poorer disease prognosis [3]. An overrepresentation of ER negative breast cancers has also been observed in African women [4,5]. Explanations for these racial differences in breast cancer characteristics, however, are still largely unknown.

Chronic inflammation has been implicated in tumor initiation, promotion, progression, invasion, and metastasis [6]. A state of chronic inflammation develops through a complex process that involves the host’s immune system and inflammatory mediators. The innate immune system, as a dominant system of host defense, provides the first line of defense against exogenous threats such as pathogens. Activation of innate immunity promotes various inflammatory reactions and triggers the release of inflammatory cytokines as well as other inflammatory mediators, such as tumor necrosis factor alpha (TNFα), interleukin-1 beta (IL1β) and interleukin-6 (IL-6) [7]. Elevated circulating levels of inflammatory biomarkers, i.e., IL6, TNFα, C-reactive protein (CRP), have been associated with a greater risk for several types of cancer and cancer prognosis, including breast [8,9]. Several studies have also identified inflammation and immune-related signatures as being important for disease prognosis for triple negative breast cancer, the aggressive breast cancer subtype often observed in young AA women [10]. Indeed, studies have shown that AAs have higher CRP and IL6 levels than EAs [11,12]. Because of evolution over millennia in Africa, and adaptation to endemic infectious diseases, it is possible that innate immune factors may differ by ancestry, with a more robust inflammatory response among AAs [13–15], which could contribute to the differential risk between AA and EA women of developing more aggressive breast cancer phenotypes.

Variants in genes involved in the innate immune response pathway may influence the production or action of inflammatory cytokines and subsequently modulate inflammatory response, influencing risk of breast cancer. Racial differences in allele frequencies of single nucleotide polymorphisms (SNPs) for certain cytokine genes have been reported in several studies, and certain genetic variants associated with increased levels of pro-inflammatory biomarkers are more frequent in AAs than that in EAs [16,17]. While a number of studies have examined inflammatory gene polymorphisms and breast cancer risk, none have focused on associations in AA women [18–21]. Furthermore, previous studies primarily focused on a small number of genes in the innate immunity pathway, most often TNF, IL1α, IL1β, IL1 receptor antagonist (IL1RN), and IL6, with mixed results.

In this case-control study, we used a two stage design to examine potential associations between breast cancer and variants in genes involved with chronic inflammation within innate immunity-related pathways in AA and EA women. We hypothesized that the differential distribution of ‘at-risk’ alleles could contribute to the higher incidence of aggressive breast cancer among AA women, particularly risk of ER-negative breast cancers.

Recently, a number of genome-wide association (GWA) studies focused on breast cancer have been completed, and have identified novel genetic variants as potentially being associated with breast cancer risk [28]. GWA studies are able to screen a large number of SNPs covering the whole genome, but it may not always be the most optimal approach to detect certain important variants and newly discovered genetic variants may only explain a small fraction of population risk [29,30]. Evaluating functional gene variants in candidate pathways is an important hypothesis driven complementary method for increasing our knowledge of potentially important biological pathways in breast cancer risk. In this case-control study, we comprehensively examined common genetic variants within innate immunity pathways with overall risk of breast cancer, as well as breast cancer risk by menopausal and ER status in AA and EA women. We found that genotype frequencies for 35 out of 42 SNPs were significantly different between AA and EA women, with only one SNP, TNFA-rs1799724, being significantly associated with overall breast cancer risk among EA women. SNP associations with breast cancer risk, however, were found to vary substantially between AA and EA populations when menopausal or ER status was considered. Our findings suggest that different gene networks may be associated with breast cancer in AA versus EA women, pre- versus post-menopausal women, and in ER positive versus ER negative breast cancers, and provide insights into the etiology of breast cancer within these subgroups, indicating areas for further research into reasons for early onset/aggressive breast cancer in AA women.

The potential impact of inflammation-related susceptibility loci that are unevenly distributed within populations as a contributor to observed heterogeneity in breast cancer phenotypes and risk between different racial groups has been understudied. As reviewed by Pennington and colleagues [31] and Chapman and Hill [13] the high burden of infectious diseases in tropical Africa and the pressure to survive such life-threatening illnesses likely led to selection for those with more robust innate immune responses. It is possible that an exuberant innate immune response, in the form of robust inflammation, while being beneficial for resisting and surviving infectious diseases, may play a negative role in malignant transformation and cancer risk in later life [32–34]. Differences in genotype frequencies and LD structures in genes involved in mounting an inflammatory innate immune response between AA and EA women may partly explain differential risk profiles for breast cancer between these two groups. Moreover, the same genetic variants may have different effects in the two populations due to interactions with host and environmental factors that are differentially distributed between the two races [35], although this was not a focus of our study. A number of differences by menopausal status were also observed in our study consistent with the possibility that immune-related etiologic pathways for pre- and postmenopausal breast cancer may be different in many respects [36], perhaps due to the role of sex hormones in modulating both the innate and subsequent adaptive immune response [37]. Overlaid on this are potential racial differences in exposure to sex hormones that can further modulate relationships with breast cancer risk and phenotype, with AA women having higher lifetime fertility rates [38], and higher circulating estrogens compared to EA women [39], potentially due in part to higher rates of obesity in this group [40].

The strongest association observed in the study was for TNFA- rs1799724 in EA women, with combined CT+TT genotypes associated with a 2.3-fold increased breast cancer risk in premenopausal women, and a 2-fold increased risk of ER positive cancers. This variant located within the promoter region of the gene has been associated with increased [41] and decreased TNFα [42] production, as well as increased risk of radiation-induced toxicity after treatment for lung cancer [43]. Associations with breast cancer risk have not been examined in either AA or EA women, although two studies found no association within Asian populations [44,45]. TNFα is implicated in chronic inflammation and can support tumor growth and breast cancer progression. Positive cross-talk between 17β-estradiol and TNFα in inflammatory and angiogenic pathways, and the ability of TNFα to modulate gene regulation by 17β-estradiol may explain, in part, why associations were strongest for premenopausal EA women and ER positive breast cancers [46]. Because this variant was much less common in AA women, with an allele frequency of 7% in AAs compared to 35.6% in EAs, examination of associations among AA women was hampered by sample size. Similarly, limited sample size prevented adequate examination of potential relationships between this variant and risk of ER negative breast cancer. We also considered 4 other SNPs (rs361525, rs1800629, rs1799964, rs1800630) located in the promoter region of TNFA, but no significant associations were found in EA women, similar to findings from other studies [18,20,47]. TNFA-rs361525 was associated with a modest increase in breast cancer risk among EAs in a study of ~5300 cases and 4900 controls [21], but this was not replicated within the Breast Cancer Association Consortium of 30,000 breast cancer cases and 30,000 controls [20]. A recent Meta-Analysis reported a small decreased risk of breast cancer (OR=0.91, 95% CI, 0.85-0.97) associated with the TNFA- rs1800629 variant in Caucasians, although this appeared to be largely driven by results from one study (OR=0.55); no associations were observed with TNFA-rs1799964, rs1800630, or rs361525 [48]. In our study, TNFA-rs1799964 and rs1800630 were associated with a similar 1.7-fold increased breast cancer risk in premenopausal AA women who carry the variant C allele of TNFA-rs1799964 or the variant A allele of TNFA-rs1800630, but not among EA women.

CRP is an inflammatory effector that has been linked to breast cancer risk and poorer prognosis, with higher circulating levels observed in AA women compared to EAs [49]. Among premenopausal women, the CT and TT genotypes of CRP-rs1205 were associated with decreased breast cancer risk in EA, but not AA women. Located in the 3’ untranslated region of the gene, this variant is consistently associated with decreased serum CRP levels [50,51], but was not associated with breast cancer risk in a prospective cohort of ~3,800 EA women [52]. The total number of breast cancers diagnosed in that study, however, was limited to 172 events, and all women were 55 years of age and therefore might not have shown associations with this genetic variant if the effect is confined to younger women, as suggested by our findings. CXCL5, which encodes for epithelial neutrophil-activating peptide (ENA-78), is up-regulated in breast tumors and plays a role in regulating neutrophil homeostasis, an essential component of innate immunity, and a major contributor to inflammation-associated tissue damage [53]. In our study, the CXCL5-rs425535 variant A allele was associated with a decreased risk of breast cancer in AA premenopausal women, but not EAs, and was much more common among AA (MAF= 39.7%) than among EA women (MAF=13.5%). This SNP is located in an exon splicing enhancer site that could be important for transcriptional control and is highly linked with the CXCL5-rs352046 variant (r²=0.94), which is associated with significantly higher ENA-78 plasma concentrations [54]. Why this might translate to lower breast cancer risk among AAs is unclear and warrants further study, especially given understanding that AAs have reduced absolute neutrophil counts compared with EAs [55].

CCL2 is an important inflammatory chemokine involved in macrophage recruitment and expression of angiogenic factors that are highly expressed within breast tumors and associated with the development and progression of breast cancer [56]. In our study, several CCL2 SNPs (rs1024611, rs13900, rs4586) in LD were suggested to be associated with increased risk of premenopausal breast cancer in AA women, although an elevated risk observed for rs4586 in EA women in the initial analysis was attenuated towards the null in our second stage analysis using a larger study population. Of these, rs1024611 located in the promoter region of the gene has been demonstrated to affect CCL2 protein levels by changing transcription factor binding sites, with the variant C allele being associated with higher CCL2 levels [57]. The other two SNPs may also similarly alter CCL2 levels. Studies have shown that CCL2 gene expression is inhibited by 17β-estradiol [58,59], our findings may indicate that this genetic polymorphism, associated with higher levels of CCL2, is most relevant with respect to breast cancer risk against a background of low CCL2 expression. Others, like us, did not observe associations between CCL2-rs1024611 and breast cancer risk among Caucasians [60,61].

Among postmenopausal women, two SNPs in high LD in the IL1B gene, rs1143627 and rs16944, were associated with an increased risk of breast cancer in EA, but not AA women. Pro-inflammatory IL-1 is implicated in cancer progression, and intratumoral levels of IL1β are higher compared with normal adjacent breast tissue [62]. These SNPs are located in the promoter region of the gene and the IL1-rs16944 variant CT/TT genotypes are associated with higher CRP levels than the CC genotype in healthy individuals [63]. The IL1B-rs1143627 variant has been associated with increased breast cancer risk in Asian populations [64,65], but has not been examined in either an EA or AA population. Several hospital-based case-control studies have examined the SNP rs16944 in Caucasians but no associations were found [66,67]. Because IL-1β levels in breast tissue might be controlled in vivo by estradiol and is correlated with abdominal subcutaneous fat [62], genetic polymorphisms in this gene might be most relevant among postmenopausal women, as suggested by our findings, since peripheral fat is a main source of estrogen production after menopause. The NOD2-rs2066842 variant was associated with decreased breast cancer risk among postmenopausal EA women, and reduced overall risk of ER negative cancers. Although this SNP has not been implicated in risk of breast cancer, it has been shown to be associated with Crohn’s disease in Caucasians [68].

Several other SNPs in CCL2, CCL5, NOD2, FGF2 were found to be associated with either risk of ER positive or negative breast cancer in either AA or EA women, although only the NOD2 SNP association with ER positive tumors differed between AA and EA women, and none have been examined in relation to breast cancer risk. Notably, variant alleles for two SNPs (rs2107538, rs3817655) in high LD in CCL5 (RANTES) were associated with approximately 35% decreased risk of ER negative breast cancer in AA women. RANTES is a chemokine that assists in the recruitment of inflammatory cells, and evolutionary pressures have been shown to have significant impact on genetic variation in this gene across various populations [69]. Tumor expression of CCL5 promotes breast cancer progression [70], and both SNPs have been associated with decreased risk of prostate cancer in men of African descent [71]. Overall, we did not identify SNPs associated with strong increased risk with ER negative tumors in AA women in our study, and this could be due to our limited statistical power in these analyses. Another potential reason could be due to our SNP selection strategy, based upon existing literature, which focused primarily on Caucasians.

One limitation of this study is that we focused on a select panel of SNPs in each gene thought to be important in cancer risk based on previous studies without including a comprehensive set of variants. This candidate gene and SNP selection approach based on limited literature for AA populations could have affected our ability to identify novel genetic variants, especially for AA women. In addition, although this is a study with a large number of AA and EA women to examine racial differences for these genetic variants with breast cancer risk, our sample size was limited when analyses were stratified by menopausal and ER status. We were able, however, to test and strengthen several promising relationships in a larger study sample after additional participant accrual. In the larger sample, controls were slightly younger than cases (stage II dataset) because eligibility criteria for cases and controls, originally up to age 65, was expanded to include older women up to age 75, fewer older controls were recruited towards the end of the study. This difference, however, was unlikely to confound or bias our findings given that all models included age as a covariate, and we performed a sensitivity analysis excluding older women (70 years old) and found results were very similar to those reported using all participants. Finally, it must be emphasized that our findings should be interpreted with caution because few associations remained significant after correction for multiple testing. Nevertheless, we expect the probability of making a type I error is attenuated by our choice to evaluate primarily functional SNPs shown to affect gene expression by changing transcriptional binding sites.

In summary, this is the most comprehensive study that was designed to specifically examine putatively functional genetic variants in the innate immunity related inflammatory pathway with breast cancer risk and risk of ER-positive and ER-negative disease in AA and EA women simultaneously. Our findings indicate that genetic variants in innate-immunity pathways are associated with breast cancer risk in both AA and EA women, although susceptible and protective loci differed by race, and played a role in the etiology of both ER-negative as well as ER-positive breast cancers. Nevertheless, as the first study to comprehensively assess these genetic variants in both AA and EA women by menopausal or ER status, our findings could provide valuable information for better understanding of the etiology of this disease in both AA and EA women. Future studies with comprehensive resequencing or functional analyses are needed to further explore these associations.