Abstract
Metabolism and excretion of the phytoestrogen enterolactone (ENL), which has been associated with breast cancer risk, may be affected by variation in steroid hormone and xenobiotic-metabolizing genes.
We conducted a randomized, crossover flaxseed intervention study in 252 healthy, postmenopausal women [137 European ancestry (EA) and 115 African ancestry (AA)] from western New York. Participants were randomly assigned to maintain usual diet or consume 10 g/day ground flaxseed for 6 weeks. After a 2-month washout period, participants crossed over to the other diet condition for an additional 6 weeks. Urinary ENL excretion was measured by gas chromatography-mass spectrometry and 70 polymorphisms in 29 genes related to steroid hormone and xenobiotic metabolism were genotyped. Mixed additive genetic models were constructed to examine association of genetic variation with urinary ENL excretion at baseline and after the flaxseed intervention.
SNPs in several genes were nominally (P < 0.05) associated with ENL excretion at baseline and/or after intervention: ESR1, CYP1B1, COMT, CYP3A5, ARPC1A, BCL2L11, SHBG, SLCO1B1, and ZKSCAN5. A greater number of SNPs were associated among AA women than among EA women, and no SNPs were associated in both races. No SNP–ENL associations were statistically significant after correction for multiple comparisons.
Variation in several genes related to steroid hormone metabolism was associated with lignan excretion at baseline and/or after flaxseed intervention among postmenopausal women.
These findings may contribute to our understanding of the differences observed in urinary ENL excretion among AA and EA women and thus hormone-related breast cancer risk.
This article is featured in Highlights of This Issue, p. 237
Introduction
Higher circulating estrogen levels are associated with increased risk of breast cancer in postmenopausal women (1–4). Modification of estrogen metabolism through dietary and supplemental intervention thus provides a potential approach for breast cancer prevention. Lignans, phytoestrogens found in plant foods, are structurally similar to endogenous steroid hormones (5) and have received considerable research attention for breast cancer prevention. Flaxseed is the richest source of lignans known to date, and the use of flaxseed to increase lignan intake has been widely studied (6). In flaxseed, the predominant lignan is secoisolariciresinol diglucoside (SDG), which has low bioavailability (7, 8). As a result, SDG reaches the lower part of the gastrointestinal tract where it is deglycosylated to secoisolariciresinol (SECO) and converted by the intestinal microbiota to the enterolignans enterodiol (END) and enterolactone (ENL). END is rapidly further converted to ENL, the predominant circulating lignan (8, 9).
Enterolignans have both estrogenic and antiestrogenic effects, depending on the presence of stronger estrogens such as estradiol (10). When the concentrations of endogenous estrogens are high during reproductive ages, enterolignans can modify steroid hormone availability through competitive inhibition of estrogen receptors (ER) and by increasing production of serum hormone-binding globulin (SHBG), thus blocking the actions of endogenous estrogens (11). Enterolignans produced from flaxseed consumption have been shown to increase urinary excretion of 2-hydroxyestrone (2OHE) to 16α-hydroxyestrone (16αOHE) ratio in postmenopausal women (12, 13), which reflects the removal of estrogen from endogenous circulation. After menopause when the levels of endogenous estrogens decrease significantly, lignans can also act as weak estrogens (14). Experimental and epidemiologic studies have demonstrated that dietary phytoestrogens may be protective against menopausal symptoms and certain chronic diseases (15–18). There is also a growing body of evidence to support the health benefits of lignans in reducing breast cancer risk and improving survival among postmenopausal women (19–21). It has also been shown that lignan-based dietary intervention can alleviate treatment complications and improve the quality of life of patients with cancer (22–27).
Nevertheless, not all studies support a positive association between lignan exposure and optimal health effects, which may be a result of individual variation in metabolism (28). Inconsistent associations have been reported between dietary lignan intakes and breast cancer risk, which may be attributable, in part, to genetic variation (24, 29–32). Lignans are structurally similar to estrogen and undergo metabolism by sulfation and/or glucuronidation, similar to estrogens. Therefore, variation in steroid hormone–metabolizing genes may also impact the effects of dietary lignan exposure. Limited research describing associations between lignans and other phytoestrogens and genetic variation support that phytoestrogen biomarker levels may be dependent upon variation in genes related to steroid hormone metabolism, for example, CYP19, COMT, ESR1, SHBG, and CYP1B1 (13, 33–36). These findings highlight the importance of considering genetic variation in the assessment of the effects of flaxseed intervention or lignan exposure from diet. Moreover, most of the previous genetic studies have been conducted in populations of European ancestry (EA), and those from populations of African ancestry (AA) are lacking. Compared with EA women, AA women may have distinct genetic backgrounds (37), tend to have higher circulating levels of estrogens (38), and are at higher risk of aggressive breast cancer (39).
Therefore, we conducted a dietary flaxseed intervention study in healthy postmenopausal women of both African and European descent, which aimed to examine associations of single nucleotide polymorphisms (SNPs) in genes important in steroid hormone metabolism with enterolignan excretion at baseline and in response to the flaxseed intervention.
Materials and Methods
Study population
Data and biospecimens for this study were obtained from a randomized, crossover flaxseed intervention study conducted between March 2012 and June 2017. All study procedures were conducted in accordance with ethical standards on human experimentation and the Helsinki Declaration of 1975 (revised 1983). The study protocol was approved by Roswell Park Comprehensive Cancer Center Institutional Review Board and all participants provided written, informed consent. The trial is registered at clinicaltrials.gov as NCT01698294.
Healthy, postmenopausal women ages 45 to 75 years, without a menstrual cycle in the past 12 months and who were from the western New York region, were eligible to participate in the study. Women were excluded from the study if they had used any of the following in the 2 months prior to baseline: oral antibiotics, hormone replacement therapy, nonprescription hormones or herbal supplements for menopausal symptoms, black-cohosh, or flaxseed supplements. Participants were block randomized equally on the basis of self-reported race to either 10 g/day ground flaxseed for 6 weeks, or maintenance of usual diet. After a 2-month washout period, each participant crossed over to the other diet condition for an additional 6 weeks (see study design schema in Supplementary Fig. S1). The study statistician generated the random allocation schedule and study coordinators assigned subjects to groups as indicated by the randomization schedule. Blinding was not possible due to the nature of the intervention and each woman served as her own control (pre–post design). Each participant was provided with one 500-g bag of ground brown flaxseed (Heartland Flax; 32.62 mg SDG/g) for the 6-week intervention period and a standardized scoop for measuring daily flaxseed doses. Participants were instructed to consume one level scoop (approximately 10 g) of ground flaxseed per day mixed into water or juice and store the flaxseed in the refrigerator for the duration of the study. Any uneaten flaxseed was returned to monitor compliance. During the 5-month study period, participants were instructed to avoid other high dietary sources of flaxseed such as seeded bakery products and cereals. Fasting blood draws, overnight urine collections, and fecal samples were collected in the morning at baseline and after 6 weeks of each crossover period, for a total of 4 collections per subject over the study period. By the end of the funding period, a total of 115 AA women and 137 EA women were recruited and randomized (Fig. 1, Consort diagram). Adverse events were generally mild and primarily related to increased gas and bloating.
CONSORT Diagram: recruitment, enrollment, and study completion flow diagram for a flaxseed intervention study in AA and EA women.
CONSORT Diagram: recruitment, enrollment, and study completion flow diagram for a flaxseed intervention study in AA and EA women.
Data collection
During the first study visit, participants completed interviewer-administered questionnaires including demographic information, medical and reproductive histories, smoking history, medication and dietary supplement use, chronic disease history, physical activity, and other epidemiologic data relevant to diet and cancer. Anthropometric measures were obtained by study personnel using a standardized protocol. Height was obtained at visit 1 only and weight at visits 1–4. Body composition (lean and fat mass) was assessed by a bioimpedance analysis (BIA) system that has 8-contact electrodes (two on each hand and foot) allowing for rapid evaluation of whole body and regional body composition (50 kHz BC-418, Tanita Corporation). Dietary data were collected via 12 telephone-administered 24-hour dietary recalls assigned randomly throughout the 5-month active intervention period using the University of Minnesota Nutrient Data System for Research (Cary, NC) dietary analysis program.
SNP selection and genotyping
Genomic DNA was extracted from peripheral blood using QIAamp DNA Blood Mini-kit and stored at −80°C until analysis. Putatively functional SNPs in genes with known roles in steroid hormone metabolism pathways that have been reported in the literature were selected by surveying the HuGE Navigator Database, supplemented with top hits from several published genome-wide association studies (GWAS) of circulating steroid hormone levels, as well as SNPs in hormone-metabolizing genes that were identified in GWAS of any phenotypes (40–68). A total of 70 SNPs in 29 genes (Supplementary Table S1; ARPC1A, BAIAP2L1, BMF, COMT, CYP17A1, CYP19A1, CYP1A1, CYP1B1, CYP3A4, CYP3A5, ESR1, GCKR, HHEX, HSD17B1, JMJD1C, LHCGR, NR2F2, PRMT6, SHBG, SLCO1B1, SULT1A1, TDGF3, TSPYL5, UGT1A1, ZBTB10, ZKSCAN5, ZNF652) with a minor allele frequency (MAF) >10% in populations of either African or European ancestry were identified and multiplexed with another 60 SNPs selected as ancestry informative markers (AIM) to assess genetic admixture. SNP genotyping was performed using the MassARRAY technology and iPLEX Gold assay (Agena BioSciences) performed by the Roswell Park Comprehensive Cancer Center Genomics Shared Resource. Blind duplicates (5%) and HapMap trio samples were included for quality control purposes. The overall genotyping success call rate was 98.0%. SNPs with low call rate (<95%), poor clustering, Mendelian error, or Hardy–Weinberg error were excluded. As a result, data of 116 SNPs (including 60 putatively functional SNPs and 56 as AIMs) from 252 women were included in the final analysis.
Lignan excretion
Overnight urine collections were obtained for determination of SECO, END, and ENL excretion. SECO, END, and ENL were assayed in the laboratory of J.W. Lampe (Fred Hutchinson Cancer Research Center, Seattle, WA) by isotope dilution gas chromatography-mass spectrometry in the SIM mode (HP 6890 GC, HP 5973 MSD). The methods are based on Adlercreutz and colleagues (69) and adapted in J.W. Lampe's laboratory (22). To adjust for variation in urine output, all urinary lignan variable concentrations were normalized to creatinine concentrations. All samples from each participant were analyzed together in a batch, and participants were randomly assigned to batches by race and treatment group. As ENL is the primary circulating enterolignan, our analyses here focus on this metabolite.
Statistical analysis
Baseline ENL levels were log transformed for analysis and reported after back transformation from the logarithmic scale to the original scale of measurement. Paired t tests were utilized to examine the intervention effect, which measures the differences in urinary ENL levels between before (visit 1 for group 1 and visit 3 for group 2) and after (visit 2 for group 1 and visit 4 for group 2) flaxseed intervention. Carry-over effect was also assessed using a paired t test for treatment group 1 (flaxseed intervention during visit 1–2 and usual diet during visit 3–4). The difference in pre- to postintervention change in ENL values between the two treatment groups was examined using t test. A repeated measures ANOVA was performed to test the stability of nonintervention measurements of urinary ENL (visit 1, 3, and 4 for group 1 and visit 1, 2, and 3 for group 2).
Proportion of genetic ancestry was estimated using STRUCTURE v2.3.4 (70–72), with HapMap data from ASW (Southwest United States), YRI (Yoruba in Ibadan, Nigeria), and CEU (Utah residents with Northern and Western European ancestry from the CEPH collection) included in the analysis as reference populations. The proportion of European ancestry among the AA women in the study was relatively low, with an average of 2.8%. There were a few AA cases (<10%, n = 8) with high European admixture, as high as 54%, but, in general, the subjects were genetically homogenous within self-reported race with little evidence of admixture. The estimated proportion of European ancestry was related to baseline ENL levels and changes upon intervention, and was used a covariate in genetic association tests.
To investigate the impact of genetic variation on baseline ENL excretion, we used linear regression models. Change in ENL levels across the four visits were assessed by repeated measures mixed models. By design, only one of the four visits was postintervention, while the other three were visits with usual diet. ANOVA revealed similar levels of ENL excretion across nonintervention visits; hence, these visits were all coded as 0 in the mixed models, whereas the postintervention visits (visit 2 in group 1 and visit 4 in group 2) were coded as 1. Separate models were fit for each SNP, assuming an additive genetic model. Covariates including baseline age and body mass index (BMI) were adjusted in all models and adjustment for multiple testing was accomplished by using a Bonferroni-adjusted P value threshold of 8.3e−4 after correcting for a total of 60 SNPs. Before fitting the final mixed models, the interaction of treatment visit and group was examined and showed no significant interaction effects. All analyses were performed in R3.4.4 and two-sided P value of 0.05 was considered significant, if not otherwise stated.
Results
The baseline demographic characteristics of the participants are shown in Table 1. Compared with EA women, AA women were less likely to achieve higher education, be married, or have a healthier BMI, and more likely to be current smokers (P < 0.0001). Because of the differences between these two populations, we stratified subsequent analyses by race. Of the 258 enrolled (120 AA and 138 EA), 190 (74%; 63% AA and 83% EA) completed all 4 visits and 252 had complete data for the present analyses. The most prevalent reason for not completing the study was that the participant became ineligible (antibiotics or other medical reason) and were removed from the study (40.2%) followed by a failure to enroll, that is, the participant did not return after the consent visit (23.5%), or the participant did not return after enrollment (17.7%). With the exception of race, no significant differences were found in baseline characteristics between those who completed the study and those who did not (Table 1).
Baseline characteristics of participants in the flaxseed intervention study
Overall (n = 252) | AA (n = 115) | EA (n = 137) | Completed study (n = 188) | AA (n = 76) | EA (n = 112) | |
Mean ± SD | ||||||
Age, y | 59.4 ± 6.2 | 59.2 ± 6.0 | 50.7 ± 6.3 | 59.7 ± 6.2 | 59.3 ± 6.2 | 60.0 ± 6.2 |
BMIa, kg/m2b | 30.5 ± 7.6 | 33.7 ± 7.8 | 27.8 ± 6.3 | 30.1 ± 7.6 | 33.2 ± 8.0 | 27.9 ± 6.6 |
n (%) | ||||||
Educationb | ||||||
High schoolc | 64 (25.4) | 40 (34.8) | 24 (17.5) | 45 (23.9) | 26 (34.2) | 19 (17.0) |
Some college | 85 (33.7) | 51 (44.4) | 34 (24.8) | 60 (31.9) | 33 (43.4) | 27 (24.1) |
Bachelor degree | 56 (22.2) | 14 (12.2) | 42 (30.7) | 41 (21.8) | 10 (13.2) | 31 (27.7) |
Graduate degree | 47 (18.7) | 10 (8.7) | 37 (27.0) | 42 (22.3) | 7 (9.2) | 35 (31.3) |
Marital statusb | ||||||
Never married | 45 (17.9) | 33 (28.7) | 12 (8.8) | 29 (15.4) | 19 (25.0) | 10 (8.9) |
Married | 98 (38.9) | 20 (17.4) | 78 (56.9) | 80 (42.6) | 16 (21.1) | 64 (57.1) |
Widowed | 28 (11.1) | 18 (15.7) | 10 (7.3) | 21 (11.2) | 13 (17.1) | 8 (7.1) |
Divorced/separated | 81 (32.1) | 44 (38.3) | 37 (27.0) | 58 (30.9) | 28 (36.8) | 30 (26.8) |
Smoking statusb | ||||||
Current | 50 (19.8) | 41 (35.7) | 9 (6.6) | 34 (18.1) | 26 (34.2) | 8 (7.1) |
Former | 85 (33.7) | 29 (25.2) | 56 (40.9) | 65 (34.6) | 20 (26.3) | 45 (40.2) |
Never | 117 (46.4) | 45 (39.1) | 72 (52.6) | 89 (47.3) | 30 (39.5) | 59 (52.7) |
BMI categoryb | ||||||
Normal (<25) | 60 (23.8) | 12 (10.4) | 48 (35.0) | 48 (25.5) | 8 (10.5) | 40 (35.7) |
Overweight (25–30) | 71 (28.2) | 23 (20.0) | 48 (35.0) | 51 (27.1) | 16 (21.0) | 35 (31.3) |
Obese (≥30) | 121 (48.0) | 80 (69.6) | 41 (30.0) | 89 (47.3) | 52 (68.4) | 37 (33.0) |
Overall (n = 252) | AA (n = 115) | EA (n = 137) | Completed study (n = 188) | AA (n = 76) | EA (n = 112) | |
Mean ± SD | ||||||
Age, y | 59.4 ± 6.2 | 59.2 ± 6.0 | 50.7 ± 6.3 | 59.7 ± 6.2 | 59.3 ± 6.2 | 60.0 ± 6.2 |
BMIa, kg/m2b | 30.5 ± 7.6 | 33.7 ± 7.8 | 27.8 ± 6.3 | 30.1 ± 7.6 | 33.2 ± 8.0 | 27.9 ± 6.6 |
n (%) | ||||||
Educationb | ||||||
High schoolc | 64 (25.4) | 40 (34.8) | 24 (17.5) | 45 (23.9) | 26 (34.2) | 19 (17.0) |
Some college | 85 (33.7) | 51 (44.4) | 34 (24.8) | 60 (31.9) | 33 (43.4) | 27 (24.1) |
Bachelor degree | 56 (22.2) | 14 (12.2) | 42 (30.7) | 41 (21.8) | 10 (13.2) | 31 (27.7) |
Graduate degree | 47 (18.7) | 10 (8.7) | 37 (27.0) | 42 (22.3) | 7 (9.2) | 35 (31.3) |
Marital statusb | ||||||
Never married | 45 (17.9) | 33 (28.7) | 12 (8.8) | 29 (15.4) | 19 (25.0) | 10 (8.9) |
Married | 98 (38.9) | 20 (17.4) | 78 (56.9) | 80 (42.6) | 16 (21.1) | 64 (57.1) |
Widowed | 28 (11.1) | 18 (15.7) | 10 (7.3) | 21 (11.2) | 13 (17.1) | 8 (7.1) |
Divorced/separated | 81 (32.1) | 44 (38.3) | 37 (27.0) | 58 (30.9) | 28 (36.8) | 30 (26.8) |
Smoking statusb | ||||||
Current | 50 (19.8) | 41 (35.7) | 9 (6.6) | 34 (18.1) | 26 (34.2) | 8 (7.1) |
Former | 85 (33.7) | 29 (25.2) | 56 (40.9) | 65 (34.6) | 20 (26.3) | 45 (40.2) |
Never | 117 (46.4) | 45 (39.1) | 72 (52.6) | 89 (47.3) | 30 (39.5) | 59 (52.7) |
BMI categoryb | ||||||
Normal (<25) | 60 (23.8) | 12 (10.4) | 48 (35.0) | 48 (25.5) | 8 (10.5) | 40 (35.7) |
Overweight (25–30) | 71 (28.2) | 23 (20.0) | 48 (35.0) | 51 (27.1) | 16 (21.0) | 35 (31.3) |
Obese (≥30) | 121 (48.0) | 80 (69.6) | 41 (30.0) | 89 (47.3) | 52 (68.4) | 37 (33.0) |
NOTE: Differences by race and study completion assessed with t-tests for continuous variables and χ2 for categorical variables.
aBody mass index.
bP < 0.0001 for differences by race; no significant differences were observed by study completion status.
cIncludes 13 subjects (10 complete study) in AA who did not finish high school; no EA subjects had education <high school.
Mean ENL excretion levels across the four study visits are shown for treatment group 1 and group 2 in Figs. 2A and B, respectively. Compared with AA women, EA women had higher mean ENL excretion levels at each visit. After the flaxseed intervention, ENL excretion levels were significantly elevated, and the magnitude of the increase was similar between the two treatment groups and the two races. ENL levels remained stable among the three habitual time-points, with no significant carry-over effect observed.
The mean (95% confidence intervals) log-transformed creatinine-adjusted urinary ENL levels across visits in AA and EA for 93 participants (41 AA and 52 EA) in treatment group 1 (A) and for 95 participants (35 AA and 60 EA) in treatment group 2 (B).
The mean (95% confidence intervals) log-transformed creatinine-adjusted urinary ENL levels across visits in AA and EA for 93 participants (41 AA and 52 EA) in treatment group 1 (A) and for 95 participants (35 AA and 60 EA) in treatment group 2 (B).
Supplementary Table S1 provides the information of the selected SNPs in the analysis and Supplementary Tables S2 and S3 detail the associations of the selected SNPs with ENL excretion levels at baseline and with change in excretion after flaxseed intervention, stratified by race and adjusted for age and BMI. The P values for these associations are summarized in Figs. 3A and B (baseline ENL excretion) and Figs. 4A and B (change in excretion after flaxseed intervention). Among EA women, baseline ENL excretion was associated with rs6738028 in BCL2L11, rs4869739 in ESR1, and rs6259 in SHBG. Collectively, these three SNPs explained 12% of the variation in ENL excretion. Different SNPs were associated with baseline ENL excretion in AA women: rs3020418 in ESR1, rs740160 in ARPC1A, rs11761528 in ZSCAN5, rs11974702, and rs7809615 in CYP3A5, and rs4633 in COMT, which collectively explained 16% of the variation in ENL excretion. There was no overlap in the SNPs associated with ENL excretion between EA and AA women. It should be noted that none of the above associations remained statistically significant after correction for multiple testing, and none were significant in both AA and EA women.
Manhattan plots summarizing P values from associations between SNPs and baseline urinary ENL excretion, adjusted for baseline age and BMI in AA (A) and EA (B). In both figures, the light gray horizontal line represents P = 0.05; the darker gray horizontal line represents Bonferroni-corrected statistical significance.
Manhattan plots summarizing P values from associations between SNPs and baseline urinary ENL excretion, adjusted for baseline age and BMI in AA (A) and EA (B). In both figures, the light gray horizontal line represents P = 0.05; the darker gray horizontal line represents Bonferroni-corrected statistical significance.
Manhattan plots summarizing P values from associations between SNPs and postintervention urinary ENL excretion, adjusted for baseline age and BMI in AA (A) and EA (B). In both figures, the light gray horizontal line represents P = 0.05; the darker gray horizontal line represents Bonferroni-corrected statistical significance.
Manhattan plots summarizing P values from associations between SNPs and postintervention urinary ENL excretion, adjusted for baseline age and BMI in AA (A) and EA (B). In both figures, the light gray horizontal line represents P = 0.05; the darker gray horizontal line represents Bonferroni-corrected statistical significance.
Several SNPs were also nominally associated with ENL excretion in response to the flaxseed intervention. Among EA women, ENL excretion after flaxseed intervention was associated with rs4869739 in ESR1, rs6259 in SHBG, and rs4149056 in SLCO1B1. These three SNPs explained 17% of the variation in ENL excretion in response to the flaxseed intervention. Among AA women, several different SNPs were associated with ENL excretion after the intervention: rs1056827 in CYP1B1, rs3020418 in ESR1, rs740160 in ARPC1A, rs11761528 in ZSCAN5, and rs11974702 and rs7809615 in CYP3A5. These six SNPs explained 13% of the variation in ENL excretion after flaxseed consumption. However, none of the SNPs were shared across the two groups.
Among AA women, with the exception of rs4633 in COMT and rs1056827 in CYP1B1, variation in SNPs associated with ENL excretion at baseline were also associated with excretion after the intervention. Similarly, among EA women, variation in rs4869739 in ESR1 and rs6259 in SHBG was associated with both baseline and postintervention ENL excretion.
Discussion
Few previous studies have investigated the impact of genetic variation on urinary ENL excretion and individual differential response to lignan exposure (13, 28, 73, 74), and, to our knowledge, none have included substantial numbers of AA participants. In this study, we were able to examine a large number of SNPs in genes related to steroid hormone metabolism, thus expanding on previous investigations. We observed several SNPs associated with ENL excretion, and different SNPs were identified in the two populations. Furthermore, the impact of genotype on excretion was comparable regardless of whether we examined excretion at baseline or in response to the intervention. Given this consistency and that we also observed lower ENL excretion at all time-points for AA compared with EA women, our findings suggest that genetic variation may, in part, be involved in the racial differences in ENL excretion.
We observed considerable disparities by race for the effect of genetic variations on urinary ENL excretion. Consistent with some previous studies (33, 75), polymorphisms in ESR1 were associated with ENL excretion (P < 0.05) among both AAs and EAs, although different SNPs were identified in each population. Notably, the allele frequencies for both ESR1 rs4869739 and rs3020418 differ drastically by race, with the minor allele in one population being the major allele in the other population. Relatively little is known about the functional significance of these two ESR1 SNPs. In a GWAS meta-analysis in the GEFOS/GENOMOS Consortium, lower bone density was observed in participants with greater prevalence of the variant alleles in rs4869739. The other SNP rs3020418 was identified in another GWAS of human height, with the A allele associated with greater height, which is also much more common in the AA population. ESR1 codes for the estrogen receptor type 1, which is a DNA-binding transcription factor and also has ligand-binding properties (53). Lignans are weak phytoestrogens that can function as competitive inhibitors of endogenous estrogen binding. If the variant alleles are resulting in reduced availability of receptors for binding of estrogens and/or lignans, increased availability of circulating lignans could be channeled into enterohepatic recycling for eventual excretion.
In addition to polymorphisms in ESR1, among AA women, we also found urinary ENL excretion associated with SNPs in several hormone-metabolizing genes including CYP3A5, COMT, and CYP1B1. There is limited research on steroid hormone metabolism–related genes and lignan metabolism, especially for AAs. A previous flaxseed intervention study reported that the changes in the 2OHE1:16OHE1 ratio increased with increasing numbers of variant alleles in CYP1B1 and COMT genes in predominantly white postmenopausal women, which supported the putatively protective functions of these SNPs (13). CYP1B1 contributes to hydroxylation of E1 at the 4- and 2-positions, and variations in this gene have been shown to have higher activities (76). COMT, in particular, contributes to the excretion of hydroxylated estrogen metabolites through methylation and subsequent excretion (77). Given the structural similarity of ENL to estradiol, similar mechanisms may affect lignan excretion. In this study, associations between ENL excretion and variation in CYP1B1 and COMT were observed only among AA women. This may help explain the lower urinary ENL levels in AA compared with EA women in the study, and may also attribute to the higher circulating steroid hormone levels in AA women than EAs found in previous literature (78). Thus, further research on the SNPs in these two genes is of great importance to investigate the race disparity in hormone-related cancers.
Regarding postmenopausal EA women, three polymorphisms in SHBG, BCL2L11, and SLCO1B1 were associated with lignan metabolism in the study (P < 0.05). We found that the variant allele A of rs6259 in SHBG was inversely associated with urinary ENL excretion both at baseline and in response to flaxseed intervention. Similar inverse association was also shown between rs4149056 polymorphism in SLCO1B1 with ENL excretion after exposure to a lignan-rich diet. SLCO1B1 gene encodes the protein solute carrier organic anion transporter family member 1B1 (SLCO1B1), which assists the transport of compounds from the blood into the liver. A GWAS found significant relationships between rs4149056 variant and estrone conjugate (E1C) plasma concentrations (79). The study suggested that this variant resulted in reduced transporter activity and higher E1C plasma concentrations. Given that lignans have properties similar to estrogen, the same mechanism may exist for ENL. On the other hand, previous studies investigating lignan exposure and breast cancer risk have varied in assessment methods (dietary intake vs. enterolignan excretion), which may also contribute to inconsistencies in associations. Given our experimental design, this is unlikely to be a concern in this study. However, future investigations of lignan exposure and breast cancer risk will need to give appropriate consideration to exposure assessment.
A potential limitation of this study is the small sample size and the lack of a replication population. Also, because of the relatively small sample size, a GWAS approach was not feasible, and we relied on a candidate SNP approach to focus on major contributors to steroid hormone metabolism and xenobiotic metabolism. Furthermore, SNPs were chosen to have MAF (>10%) sufficient for statistical analysis, and thus the influence of less common SNPs cannot be studied.
Our study has several notable strengths. We utilized a cross-over design where each participant serves as her own control, and thus the influence of confounding covariates is reduced. Notably, this study is reasonably comprehensive research to investigate genetic effects on lignan metabolism. We included a relatively large number of SNPs in related genes, and involved both AA and EA populations. In addition, we performed ancestral analysis to detect hidden population substructure, which makes our findings of race disparities more reliable. Although none of the selected SNPs were significantly associated with ENL excretion after correction for multiple comparisons, the results of this study nevertheless may shed light on the potential impact of genetic variation on lignan metabolism of postmenopausal women and stimulate further research.
In summary, we observed that variation in a number of genes related to steroid hormone metabolism were associated with lignan excretion at baseline and after flaxseed intervention among postmenopausal women, which may help explain the difference in urinary ENL levels among AAs and EAs. To our knowledge, this is the first report suggesting potential explanations related to genetic variation for racial disparities in lignan metabolism among AA and EA postmenopausal women. Future chemoprevention and diet intervention strategies to reduce breast cancer incidence and mortality may benefit by incorporation of genetic variation and examination by race.
Disclosure of Potential Conflicts of Interest
J.W. Lampe is a consultant/advisory board member for U.S. Highbush Blueberry Council. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: S. Yao, J.W. Lampe, L.U. Thompson, S.E. McCann
Development of methodology: H. Chang, S. Yao, S.E. McCann
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.W. Lampe, S.E. McCann
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Chang, S. Yao, D. Tritchler, S.E. McCann
Writing, review, and/or revision of the manuscript: H. Chang, S. Yao, M.A. Hullar, J.W. Lampe, L.U. Thompson, S.E. McCann
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Chang, S.E. McCann
Study supervision: J.W. Lampe, S.E. McCann
Acknowledgments
This work was supported by National Cancer Institute (NCI) grants U01CA161809 (to S. McCann, J.W. Lampe, M.A. Hullar, S. Yao, and D. Tritchler) and P30CA016056 (to S. McCann and S. Yao) involving the use of Roswell Park Comprehensive Cancer Center's Databank and Biorepository and the Genomics Shared Resource.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.