Circulating Vitamin D Metabolites, Polymorphism in Vitamin D Receptor, and Colorectal Adenoma Risk

That U.S. colorectal cancer rates are greater in the northern United States, where sun exposure is lower, led to the hypothesis that sunlight-mediated dermal vitamin D synthesis provides protection from this disease. Several epidemiological studies support this hypothesis (1–6), showing decreased risk for colorectal tumors in relation to higher serum levels of vitamin D, an integrative measure of dermal vitamin D production and dietary and supplemental vitamin D intake; however, evidence from these studies is limited because of the small number of studies that have been done and the small sample sizes of half of these studies (34–146 cases; Refs. 4–6).

The cellular effects of vitamin D are primarily mediated through binding to the nuclear vitamin D receptor (VDR), which regulates the transcription of more than 60 genes, including genes involved in cellular differentiation and inhibition of proliferation (7, 8). A series of polymorphisms in the 3′ untranslated region (UTR) of VDR may affect mRNA stability (9, 10) possibly through linkage to other variants (11); however, functional studies have shown contradictory results (12, 13).

We assessed the association of serum vitamin D metabolite levels and a TaqI polymorphism in the 3′ UTR of VDR with risk of colorectal adenoma, a precursor lesion of colorectal cancer. We also evaluated the effects of calcium intake and hormone replacement therapy (HRT) in relation to vitamin D and adenoma risk because of their potential physiological correlates with vitamin D (14–21) and observed associations with colorectal tumor risk (22–24).

We conducted our analysis within a large colorectal cancer screening trial (25, 26) in which subjects received a standardized screening exam. The sample size allowed us to limit case to advanced adenomas, which have a higher potential for malignant transformation. Because of the standardized screening, we could randomly select for control subjects who had a negative sigmoidoscopy examination from the same study population.

The study was done as a nested case-control study within the Prostate, Lung, Colorectal and Ovarian Cancer Screening (PLCO) trial (25, 26). The trial recruited 154,952 participants, aged 55–74 years, at 10 U.S. study centers (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA; Salt Lake City, UT; St. Louis, MO; and Washington, DC). Participants randomly assigned to the screening arm of the PLCO trial were offered a sigmoidoscopy examination at study entry and referred for colonoscopy follow-up examination if polyps or other suspect lesions were identified. Medical and pathological reports on follow-up were abstracted and coded by trained medical record abstractors. Written informed consent was obtained from participants and the trial received approval from the institutional review boards of the U.S. National Cancer Institute and the 10 study centers.

Subjects for this study were selected from 42,037 participants in the screening group who underwent a successful sigmoidoscopic examination (insertion to at least 50 cm with >90% of mucosa visible or a suspect lesion identified) between September 1993 and September 1999. Subjects provided information on risk factors and donated a blood sample for use in etiological studies. After exclusion of 4834 subjects with a self-reported history of cancer (except basal cell skin cancer), ulcerative colitis, Crohn's disease, familial polyposis, colorectal polyps, or Gardner's syndrome, we randomly selected 772 cases for study from among 1234 cases with advanced distal adenoma (adenoma ≥1 cm or containing high-grade dysplasia or villous elements). An equal number of gender- and ethnicity-matched participants (n = 777) with a negative screening sigmoidoscopy (i.e., no polyp or other suspect lesion; n = 26,651) were selected as controls. Genotype analysis was attempted on samples from all selected cases and controls; genotypes were successfully obtained from 763 case samples and 774 control samples. Analyses of vitamin D metabolite levels were conducted in samples of 400 cases and 400 controls selected at random. Stored blood samples collected at study entry were unavailable for six cases and three controls. Therefore, the total number was 394 for cases and 397 for controls for the vitamin D metabolite analysis.

Levels of the two vitamin D metabolites, 25-hydroxyvitamin D [25(OH)D] and 1,25-dihydroxyvitamin D [1,25(OH)₂D], were determined by RIA using radioiodinated tracer in serum derived from blood collected at study entry (27). Laboratory personnel were blinded to case-control status. Multiple blinded quality control samples from two different individuals were included in all batches (total n = 96). The coefficient of variation was 12.5% for 1,25(OH)₂D and 16.3% for 25(OH)D.

The TaqI polymorphism, a C-to-T base substitution (dbSNP rs731236) in exon 9 of the VDR gene, was detected by PCR amplification followed by restriction enzyme digestion (for details, see Appendix A. Alleles were scored as TT, Tt, or tt; lowercase letters indicate the presence of restriction site. Laboratory personnel were blinded to disease status of the participants. The P value of the Hardy-Weinberg equilibrium (28) was 0.31 in non-Hispanic white controls and 0.33 in all controls.

At the initial screening, all participants were asked to complete a questionnaire about sociodemographic factors, medical history including current and former use of HRT, and risk factors for cancer. Usual dietary intake over the 12 months before enrollment was assessed with a 137-item food frequency questionnaire including an additional 14 questions about intake of vitamin and mineral supplements (29). Daily dietary nutrient intake was calculated by multiplying the daily frequency of each consumed food item by the nutrient value of the gender-specific portion size (30) using the nutrient database from the U.S. Department of Agriculture (31). Total calcium and vitamin D intake was calculated by adding dietary and supplemental intakes.

To account for seasonal differences and demographic factors, we adjusted all mean values of 25(OH)D and 1,25(OH)₂D for month of blood draw, gender, ethnic origin, study center, and age. Adjusted means (least squares means) and differences between adjusted means were calculated by general linear models. Using unconditional logistic regression analysis, we calculated prevalence odds ratios (ORs) for the relation between VDR and serum vitamin D metabolite levels and advanced distal colorectal adenoma with vitamin D metabolites entered continuously or in quintiles with cut points based on the distribution among controls. Additionally, we used season-specific cut points for quintiles (December to May and June to November). We also estimated the ORs comparing serum levels of 25(OH)D and 1,25(OH)₂D above and below clinically normal levels (cut points: 15 ng/ml and 26 pg/ml, respectively; Refs. 32, 33).

We frequency matched on ethnic origin to allow stratified genotype analysis. Because of a strong imbalance between male and female cases and controls, we further frequency matched on gender. To avoid small cell numbers, we did not match on other risk factors but adjusted for other potential risk factors in the statistical analysis. We adjusted all ORs for the matching factors, gender, and ethnic origin as well as for age at randomization, study center, and month of blood draw. In addition to this basic model, we included potential confounders based on a priori hypotheses for colorectal neoplasia risk factors if they changed the risk estimate (OR) by more than 10%. None of the factors (educational attainment, smoking, alcohol use, aspirin use, ibuprofen use, physical activity, body mass index, HRT use, dietary fiber, total calcium, total folate, and red meat) was included in the analyses because the factors, either separately or together, did not change the risk estimates of 25(OH)D, 1,25(OH)₂D, or the VDR TaqI polymorphism by more than 10%.

To explore effect modification, we performed stratified analyses and evaluated multiplicative interaction by creating product terms. We investigated the statistical significance of multiplicative interaction terms by comparing the log likelihood statistics of the main effect model with the joint effects model. Continuous variables were used to calculate the P value for trend. All P values are two-sided.

Adenoma cases were slightly older than controls (average 62.9 and 62.3 years, respectively; Table 1). The study population included about 31% female and 94% non-Hispanic white participants. Because we matched on ethnic origin and gender, the case and control distributions were very similar for these two variables. Controls attained higher education levels than did cases.

The average serum level of 25(OH)D in adenoma cases was lower than in controls (P = 0.06; Table 2). In contrast to 25(OH)D levels, mean serum levels of 1,25(OH)₂D were not different for cases and controls (P = 0.66). Within the control group, the average levels of 25(OH)D and 1,25(OH)₂D were higher in women than in men (P = 0.15 and 0.01, respectively). Non-Hispanic black controls had significantly lower 25(OH)D levels (16.6 ng/ml) than did non-Hispanic white controls (28.6 ng/ml; P = 0.0007), whereas 1,25(OH)₂D levels were similar (35.9 and 36.6 pg/ml; P = 0.85 respectively).

Overall, risk reduction [(OR − 1) × 100] of advanced distal adenoma was 13% for each 10 ng/ml increase in serum 25(OH)D level (Table 2). The overall inverse association between adenoma risk and serum 25(OH)D level was limited to women, showing a 41% reduction in risk with each 10 ng/ml increment in 25(OH)D and a highly significant trend (P = 0.0002; Table 2). The P value for the multiplicative interaction between gender and 25(OH)D was 0.02. Similar to the continuous analysis, the categorical analysis of serum 25(OH)D showed no association between 25(OH)D and adenoma risk in men but a strong inverse association in women (Table 3). Comparing women in the highest (5th) quintile with women in the lowest (1st) quintile of 25(OH)D, adenoma risk decreased by 73%. Further, a 70% decrease in risk was evident when women with normal serum 25(OH)D levels were compared with those with subnormal levels (<15 ng/ml). The association between 25(OH)D and advanced adenoma risk was stronger in older participants (Table 2). For non-Hispanic whites, adenoma risk decreased by 16% for each 10 ng/ml increment in 25(OH)D. Sample sizes of groups of other ethnic origins were too small to estimate ORs.

Serum levels of 1,25(OH)₂D (as continuous term or quintiles) were not associated with adenoma risk in either all participants or separately in men and women (Tables 2 and 3). Using season-specific cut points for the quintiles or stratifying on season of blood draw resulted in adenoma risk estimates for 25(OH)D and 1,25(OH)₂D (data not shown) very similar to those reported in Tables 2 and 3.

In women, mean serum 25(OH)D levels were significantly higher in current HRT users than in former or never HRT users (within cases, P = 0.002; within controls, P = 0.006; Table 4). However, the association between 25(OH)D and adenoma risk was similar in both groups and the P value for the multiplicative interaction between 25(OH)D and HRT use was 0.43. Current HRT users also had significantly higher mean levels of 1,25(OH)₂D than did former/never HRT users (within cases, P = 0.003; within controls, P = 0.03). 1,25(OH)₂D was not associated with adenoma risk in current HRT users or former/never HRT users (data not shown). Total intake of vitamin D and calcium was similar for current and former/never HRT users (mean vitamin D: 509 and 512 IU/day; mean calcium: 1319 and 1325 mg/day, respectively).

The VDR TaqI polymorphism was not associated with adenoma risk in all participants or in non-Hispanic whites (Table 5). The association between VDR TaqI genotype and adenoma risk was similar in non-Hispanic white women (OR for Tt versus TT and tt versus TT: 1.00, 1.10, and 1.03) and men (OR for Tt versus TT and tt versus TT: 1.00, 1.16, and 1.11).

The association between VDR TaqI genotype and adenoma risk did not vary when stratified on total calcium intake (median or tertiles; data not shown). In addition, the association of 25(OH)D and 1,25(OH)₂D and adenoma risk did not vary after dichotomization on the median total calcium intake either in the whole study population or in each gender separately (data not shown). Gender-specific adenoma risk estimates of serum vitamin D level varied little by VDR TaqI genotype; only for women with the VDR TaqI genotype tt, the risk reduction for each 10 ng/ml increment in 25(OH)D was slightly stronger than that for women with genotypes TT or Tt (data not shown). None of the P values for multiplicative interaction between serum vitamin D level and VDR TaqI genotype or total calcium was below 0.2.

Within our nested case-control study of participants in a colorectal cancer screening trial, we observed a 73% reduction in advanced adenoma risk among women with serum 25(OH)D levels in the highest quintile compared with those in the lowest quintile. No association between serum 25(OH)D levels and adenoma risk was observed among men. Serum 1,25(OH)₂D levels and VDR TaqI genotype were not associated with adenoma risk in either men or women. The associations of adenoma risk with 25(OH)D, 1,25(OH)₂D, and VDR TaqI genotype were not appreciatively modified by each other or by calcium intake or HRT use.

Two out of three case-control studies, with 236 cases (1) and 473 cases (2), reported inverse associations between 25(OH)D and adenoma risk. The third study (3) with 326 female cases found no clear association between 25(OH)D and risk of adenoma but did find an inverse association between 1,25(OH)₂D and adenoma. In addition to these adenoma studies, three nested case-control studies (4–6) of colorectal cancer with 34 (4), 57 (5), and 146 (6) cases were reported. All three studies showed an inverse association between serum 25(OH)D and colorectal cancer risk. Two of the studies (5, 6) also investigated 1,25(OH)₂D levels and found no association between 1,25(OH)₂D levels and colorectal cancer. Thus, although most observational studies, including the present study, suggest an inverse association between 25(OH)D and colorectal neoplasia, results concerning 1,25(OH)₂D have been less consistent.

A potential anticarcinogenic effect of 25(OH)D on colon tissue is supported by recent studies demonstrating that normal colon tissue expresses 1-α vitamin D hydroxylase and hence can locally produce the metabolically active form of vitamin D, 1,25(OH)₂D, from 25(OH)D (34–36). These findings suggest that 25(OH)D as well as 1,25(OH)₂D may have a localized effect in colon tissue to reduce cell proliferation and induce differentiation (37, 38). The potential significance of circulating 25(OH)D levels on colon neoplasia is further supported by the fact that serum levels of 1,25(OH)₂D are tightly regulated as part of their function to control calcium homeostasis, whereas circulating 25(OH)D levels fluctuate and reflect vitamin D status as determined by dietary intake and endogenous production (39–41). The half-life of 1,25(OH)₂D is only a few hours whereas the half-life of 25(OH)D is about 1 month (21, 42, 43). These data suggest that 25(OH)D levels, which are modifiable by behavioral changes (diet and sun exposure), may affect colorectal cancer development.

It is puzzling that we found an association between adenoma risk and levels of 25(OH)D in women but not in men. The mechanism by which vitamin D mediates an anticarcinogenic effect does not appear to be gender specific. Most case-control studies have not stratified their analysis of vitamin D and colorectal neoplasia on gender (1, 2, 4, 5). However, one study that only included men showed an inverse association between 25(OH)D and colorectal cancer (6), and one study that only included women found no consistent association between 25(OH)D and adenoma risk (3). We reanalyzed an earlier published study (1) and found that the inverse association between 25(OH)D levels and adenoma risk was not substantially stronger in women than in men (quintiles 1–5, women: 1.0, 0.5, 0.4, 0.2, and 0.3; men: 1.0, 0.4, 0.8, 0.6, and 0.5). Based on the results of these previous studies, our finding of a gender-specific inverse association between vitamin D and colorectal adenoma was unexpected and might be due to chance. Therefore, we believe that the potential anticarcinogenic effect of 25(OH)D is unlikely to be gender specific.

Current HRT users had significantly higher serum 25(OH)D and 1,25(OH)₂D levels than did former/never users and the difference was not accounted for by differences in vitamin D intake. This finding is consistent with previous reports of higher 25(OH)D and 1,25(OH)₂D levels in women with increased levels of endogenous estrogens and in women taking exogenous estrogen formulations (14–20). The estrogen-related increase in 25(OH)D and 1,25(OH)₂D levels is possibly due to stimulation of renal 1-α hydroxylase and inhibition of 24-vitamin D hydroxylase, resulting in a reduced proportion of 25(OH)D being metabolized to 24,25(OH)₂D (14, 15, 19). Alternatively, estrogen may alter the relative proportions of free and protein-bound vitamin D: some but not all studies have shown that estrogen increases levels of vitamin D binding protein or that estrogen increases levels of free vitamin D (44, 45). The decreased risk of colorectal cancer associated with HRT observed in other studies (22, 23) may be related in part to the HRT effect on vitamin D levels. However, HRT and estrogen likely affect colorectal cancer risk through multiple pathways, such as interaction with insulin and insulin-like growth factor axis (23) or reduction in CpG methylation in the VDR gene, which results in increased VDR expression (46).

We did not observe an association between VDR TaqI polymorphism and colorectal adenoma. Two case-control studies reported a decreased risk of colorectal neoplasia in participants with the TaqI tt genotype compared with those with Tt or TT genotype either overall (11) or in conjunction with low calcium or vitamin D intake (9). In contrast, in our study and two case-control studies (47, 48), TaqI genotype was not associated with colorectal adenoma risk and neither vitamin D nor calcium modified this association. Results concerning the association between TaqI polymorphism in the 3′ UTR of VDR and risk of colorectal neoplasia remain inconsistent.

A theoretical limitation of our study is that serum nutrient levels may be influenced by disease status at the time of blood draw. However, we studied screen-detected adenomas so it is not likely that disease-related behavioral modifications such as dietary changes or exposure to sun affected vitamin D levels.

Our design had the advantage of ensuring that cases and controls came from the same source population and were screened with a standardized procedure (i.e., cases were not screened based on symptoms). The large study population allowed us to confine the analysis to cases with advanced adenoma, which have a higher potential for malignant transformation and are a particularly meaningful intermediate outcome for studying factors related to colorectal cancer. However, because we restricted our study to advanced adenomas, we could not examine whether vitamin D metabolites and the TaqI polymorphism prevented any adenoma formation but only whether they prevented the formation and/or subsequent persistence of adenomas with increased malignant potential.

In summary, results from our study suggest an inverse association between serum 25(OH)D level and risk of developing colorectal neoplasia. Our study does not support an association between adenoma and TaqI polymorphism in the 3′ UTR of VDR or between adenoma and serum 1,25(OH)₂D level.

The TaqI polymorphism was detected by PCR amplification followed by restriction enzyme digestion. Genomic DNA was amplified by PCR using the following primers: 5′-CAG AGC ATG GAC AGG GAG CAA-3′ (sense) and 5′-GCA ACT CCT CAT GGC TGA GGT CTC-3′ (antisense). The amplified product is a 740-bp fragment containing the variant site. PCR was started with a denaturation step of 94°C for 3 min followed by 30 cycles at 94°C for 45 s, 58°C for 60 s, and 72°C for 90 s. The concentrations of the sense and antisense primers were 250 nm in an amplification reaction. PCR products were digested by the TaqI restriction enzyme (1 unit/20 μl reaction) and electrophoresed through a 3% agarose gel at 125 V for 30 min. The gels were stained with ethidium bromide and photographed under UV light. Alleles were scored as TT, Tt, or tt; lowercase letters indicate the presence of restriction site. In each batch, we included samples of the homozygote wild-type (TT) and heterozygote (Tt) genotypes as positive controls and blank restriction reactions as negative controls.