Low levels of plasma vitamin D have been implicated as a possible risk factor for both prostate cancer incidence and advanced disease, and recent phase II trials suggest that vitamin D supplementation might delay progression of prostate cancer. Common polymorphisms in the vitamin D receptor (VDR) are associated with VDR activity and are therefore potentially useful proxies for assessing whether vitamin D is causally related to advanced prostate cancer. We genotyped five well-known VDR polymorphisms in 1,604 men with prostate cancer from the Prostate Testing for Cancer and Treatment study. Our aim was to examine the association between VDR polymorphisms and cancer stage (localized versus advanced) as well as cancer grade (Gleason score <7 versus ≥7). Moreover, we also carried out a systematic review and meta-analysis of 13 similar studies. As a result of our meta-analysis, we revealed three polymorphisms, BsmI, ApaI, and TaqI, associated with high Gleason score with an overall summary odds ratios (95% confidence intervals) of 1.12 (1.00-1.25; bb versus BB + Bb), 1.25 (1.02-1.53; aa versus AA + Aa), and 0.82 (0.69-0.98; Tt + tt versus TT), respectively. The haplotype analysis revealed that the BsmI (B)-ApaI (A)-TaqI (t) participants compared with BsmI (b)-ApaI (a)-TaqI (T) individuals were less likely to have high Gleason scores (odds ratio, 0.84; 95% confidence interval, 0.71-1.00; Punadjusted = 0.050; Padjusted = 0.014). Our finding provides some support for the hypothesis that low levels of vitamin D may increase the risk of prostate cancer progression. (Cancer Epidemiol Biomarkers Prev 2009;18(11):2874–81)

Low levels of vitamin D were first implicated as a possible risk factor for prostate cancer in a landmark study by Apperly (1) who reported a positive correlation between prostate cancer mortality and reduced UV radiation exposure. Since that study, evidence from several sources has implicated low levels of vitamin D as a risk factor for prostate cancer development and disease severity at diagnosis, the latter being a marker of cancer aggressiveness and progression. In vitro experiments have shown that the presence of vitamin D can inhibit growth of prostate cancer cells (2-4). However, prospective cohort studies of the association of prediagnosis vitamin D status with prostate cancer risk have found inconsistent associations, with inverse (5, 6), null (7-11), “U”-shaped (12), and even positive associations with aggressive cancer being reported (13, 14). Most of these studies have small sample sizes and it has been suggested that many that reported null associations may have been underpowered to detect an inverse association (13). Furthermore, the hypothesized effect of vitamin D on prostate cancer (primarily thought to be via inhibition of cancer cell growth) may mean that it is more likely that low levels of vitamin D are associated with how rapidly the disease progresses rather than as a primary etiologic risk factor for disease occurrence. With the pharmaceutical industry attempting to develop high-dose vitamin D drugs with limited side-effects for the treatment of men with prostate cancer (15, 16), it is important to examine the extent to which variation in vitamin D level or activity is likely to be related to disease severity and progression.

Vitamin D is initially metabolized to the intermediate compound 25-hydroxyvitamin D in the liver and then subsequently binds to intracellular receptors, which regulate gene expression (17). The vitamin D receptor (VDR) is an important regulator of the vitamin D pathway, which involves the conversion of serum 25-hydroxyvitamin D into the active hormone, 1,25-dihydroxyvitamin D (18). There are five important common polymorphisms within the VDR gene region that are likely to exert functional effects on VDR expression. These polymorphisms are located in three linkage disequilibrium (LD) blocks: Cdx2, located in the promoter region upstream of exon 1, affects the binding ability of VDR and subsequent VDR transcriptional activity (19); FokI, located in the translation start of exon 2, produces a shorter protein with greater transcriptional activity (20); the three other variants, BsmI (intron 8), ApaI (intron 8), and TaqI (exon 9), are all located in one LD block, which is thought to contain other variants at 3′-untranslated region of the VDR gene that may influence VDR expression by altering the mRNA stability. Thus, these three LD blocks within the VDR are likely to have independent effects on vitamin D metabolism and hence levels or activity (17, 21).

Because genetic variants are unrelated to the many environmental characteristics that tend to confound traditional risk factor-disease association studies (22, 23), it has been suggested that genetic variants with robust associations with modifiable risk factors (in this case, circulating vitamin D) could be used as proxies to determine the unconfounded and unbiased effect of such risk factors on disease outcome (here, prostate cancer stage/grade at diagnosis). To date, several studies have examined the association between VDR polymorphisms and advanced prostate cancer. However, the results from those studies are inconsistent and inconclusive, with most studies being of small sample size (the majority including <200 men with prostate cancer; refs. 13, 24-34).

In the present study, we undertook an analysis of VDR genetic variants in 1,604 prostate cancer patients who were identified prospectively through population-based PSA testing and had the clinical status of their cancer determined at the time of diagnosis. We then completed a systematic review of all published studies of the association of VDR genetic variants with advanced prostate cancer, and in a meta-analysis, we combined our results with those from the systematic review to estimate the overall association between VDR variants and prostate cancer stage/grade at diagnosis.

Study Participants

Participants in this study were selected from the Prostate Testing for Cancer and Treatment (ProtecT) study (35), which is a randomized controlled trial taking place in nine regions of the United Kingdom with the aim of evaluating the efficacy, cost-effectiveness, and acceptability of treatments for localized prostate cancer. Men, without known prostate cancer, aged 50 to 69 years were invited to attend a nurse-led prostate-check clinic and have a prostate-specific antigen (PSA) test. Participants with a single raised PSA level over 3.0 ng/mL were invited to attend the center's urology department for digital rectal examination, repeat PSA test, and transrectal ultrasound-guided biopsy (10 cores) to determine prostate cancer status. Participants' age when they attended the primary care clinic, PSA level, height, weight, smoking status, physical activity level, and self-reported ethnicity were collected either by questionnaire or by nurse interview at the prostate check clinic. Multicenter research ethics committee approval was obtained from Trent MREC, and consent for performing anonymized genetic tests on stored blood was obtained from individual participants.

Stage and Grade of Prostate Cancer

Prostate cancer stage was defined according to the 2002 tumor-node-metastasis (TNM) staging system (36). Patients with cancer stage T1 or T2 were defined as having localized cancer, while individuals with a cancer stage of ≥T3 were defined as advanced cases (Table 1). Cancer grade was defined as lower grade with Gleason score of 1 to 6 and higher grade with a Gleason score ≥7. Of the total number of cancer cases we genotyped, 99.8% with cancer stage and/or grade information were included in this study.

Table 1.

Correlation between TNM staging system and Gleason score in ProtecT participants

2002 TNM stagingGleason score <7Gleason score ≥7MissingTotal
Clinically localized 1,054 308 1,364 
Clinically advanced 53 142 197 
Missing   
Total 1,109 452   
2002 TNM stagingGleason score <7Gleason score ≥7MissingTotal
Clinically localized 1,054 308 1,364 
Clinically advanced 53 142 197 
Missing   
Total 1,109 452   

DNA Extraction and Genotyping

Five VDR variants were genotyped in 1,604 prostate cancer cases as part of a genetic association study examining the effect of 70 diet/nutrition relevant single nucleotide polymorphisms on prostate cancer risk. DNA was extracted by Tepnel7

at the University of Sheffield, and genotyping was undertaken by KBioscience,8 who use their own form of competitive allele-specific PCR (KASPar) and Taqman, for single nucleotide polymorphism analysis. Samples with >10% genotype failure (7 single nucleotide polymorphisms) were defined as having poor DNA quality (2.6%) and dropped from further analysis. Genotyping was repeated in 10% of the study samples (with independent assessment), and for 99.98% of those samples, there was exact agreement between the two. Based on standard nomenclature, genotypes for the four VDR variants, FokI (rs10735810), BsmI (rs1544410), ApaI (rs7975232), and TaqI (rs731236), were reported using lowercase and uppercase letters to reflect the presence or absence of a restriction site, which is consistent with the literature (FokI: T-f and C-F; BsmI: G-b and A-B; ApaI: T-A and G-a; and TaqI: T-T and C-t). The genotypes for Cxd2 (rs11568820) were labeled by their nucleotide bases.

Statistical Analysis

Hardy-Weinberg equilibrium of genotype frequencies among participants was tested using the Pearson χ2 test. Genotypic effect was assessed by using information from the systematic review to inform the genetic models in our primary study, but we also tested alternative models. One-way ANOVA was used for testing mean differences among genotypes for quantitative variables, such as age, body mass index, and PSA level, whereas χ2 tests were used for ordered categorical variables such as smoking status (current smoker, ever smoker, and nonsmoker), social class status (professional, intermediate, and manual), and weekly physical activity sessions (none, 1-2 times, 3-4 times, and ≥5 times). Logistic regression models were used for assessing genotype associations with prostate cancer stage/grade. Participants' characteristics associated with prostate cancer stage/grade were assessed using t and χ2 tests. Statistical tests were two-sided and carried out in Stata statistical software (version 10). Pairwise LD (D′) between single nucleotide polymorphisms was calculated using Haploview (37) and haplotype analysis was done using the haplo.stats package (38) in R. Odds ratios (OR) were calculated by using the most common haplotype as baseline reference. Logistic regression was used to determine the association between VDR haplotypes and prostate cancer stage/grade.

Systematic Review and Meta-analysis

Articles published or available online before August 14, 2008 were retrieved from Medline,9

ISI Web of Knowledge,10 and Embase11 using the search terms: “prostate cancer” or “prostatic neoplasms (MeSH keywords)” in combination with “vitamin D (MeSH keywords)” and “vitamin D receptor” or “VDR.” Relevant articles cited in the bibliography of retrieved articles were also included. Selection of articles was done by two independent researchers (L.C. and W.Y.) to avoid bias. Our inclusion criteria were that the article had to report an association between any genetic variant in VDR and advanced prostate cancer (advanced versus localized). Studies that only used PSA level or bone scan to assess prostate cancer stage were excluded, as these are not valid or reliable methods for cancer staging (n = 2 studies; refs. 39, 40). No articles were excluded based on language. Where studies appeared to fulfill our inclusion criteria but did not include all the information necessary for inclusion in our meta-analyses, the corresponding authors were contacted for relevant data. For studies that published more than once, only the most recent publication was included.

Meta-analysis was carried out using data from all articles that determined the association between genetic variants in the VDR gene and prostate cancer progression by comparing advanced cancer patients to individuals with localized cancer. Genotype distributions in advanced and localized cases were retrieved from publications. Unadjusted allelic ORs were calculated for all VDR polymorphisms. Different systems for determining whether an individual had localized or advanced prostate cancer were used in the retrieved articles. Articles that reported results based on the Jewett-Whitmore system were converted into TNM system using the following conversion: T1a-A1; T1b-A2; T1c-B0; T2a-B1; T2b-B2; T3a-b-C1; T3c-C2-3; N+-D1 based on “Clinical Staging of Prostate Cancer.”12

Gleason score and relative tumor behavior was converted to cancer grade according to Huang et al. (34). Where possible, we divided studies into two groups depending on whether they used Gleason score or TNM staging system and performed two fixed-effects meta-analyses to calculate summary OR estimates for these two methods of tumor evaluation.

We carried out a test for heterogeneity using I2 and tested for small study bias using the Egger (41) and Begg (42) tests. Our results from ProtecT were also included in the meta-analysis.

Genetic Association Analysis

Five VDR genetic variants, Cdx2, FokI, BsmI, ApaI, and TaqI, were genotyped and 43 participants who are not of European origin were excluded. Finally, 1,561 European origin men with prostate cancer that had been clinically staged or graded were included in the analysis. All participants had at least one successfully genotyped VDR polymorphism. Participants' baseline characteristics are presented by cancer stage/grade in Table 2. The mean age of participants in the study was 62.5 years and participants with advanced cancer were older than patients with localized cancer. Mean PSA level and the percentage of current smokers were also higher in advanced cancer patients (Table 2).

Table 2.

Basic characterization of ProtecT participants

Prostate cancer grade (Gleason score)PProstate cancer stage (clinical stage)P
<7≥7LocalizedAdvanced
No. participants 1,109 452 — 1,364 197 — 
Age (mean ± SD), y 62.3 ± 5.2 63.3 ± 4.9 <0.001 62.3 ± 5.2 64.0 ± 4.5 <0.001 
PSA (mean ± SD), ng/mL 6.4 ± 5.6 16.3 ± 46.8 <0.001 6.3 ± 4.8 30.0 ± 68.8 <0.001 
Body mass index (mean ± SD), kg/m2 26.6 ± 3.5 26.8 ± 3.8 0.514 26.7 ± 3.5 26.7 ± 3.7 0.965 
Social class, n (%) 
    Professional 516 (47.1) 204 (45.8) 0.340 635 (47.1) 84 (43.5) 0.610 
    Intermediate 164 (15.0) 80 (17.9) 214 (15.9) 31 (16.1) 
    Working 415 (37.9) 161 (36.2) 499 (37.0) 78 (40.4) 
Smoking, n (%) 
    Nonsmoker 323 (40.2) 105 (30.4) 0.007 375 (37.2) 52 (36.4) 0.795 
    Ever smoker 373 (46.5) 185 (53.5) 488 (48.5) 73 (51.0) 
    Current smoker 107 (13.3) 56 (16.2) 144 (14.3) 18 (12.6) 
Weekly physical activity sessions, n (%) 
    None 369 (47.2) 157 (46.3) 0.425 456 (46.5) 69 (46.3) 0.940 
    1-2 times 256 (32.8) 114 (33.6) 327 (33.3) 44 (31.4) 
    3-4 times 110 (14.0) 40 (11.8) 133 (13.5) 18 (12.9) 
    ≥5 times 47 (6.0) 28 (8.3) 66 (6.7) 9 (6.4) 
Prostate cancer grade (Gleason score)PProstate cancer stage (clinical stage)P
<7≥7LocalizedAdvanced
No. participants 1,109 452 — 1,364 197 — 
Age (mean ± SD), y 62.3 ± 5.2 63.3 ± 4.9 <0.001 62.3 ± 5.2 64.0 ± 4.5 <0.001 
PSA (mean ± SD), ng/mL 6.4 ± 5.6 16.3 ± 46.8 <0.001 6.3 ± 4.8 30.0 ± 68.8 <0.001 
Body mass index (mean ± SD), kg/m2 26.6 ± 3.5 26.8 ± 3.8 0.514 26.7 ± 3.5 26.7 ± 3.7 0.965 
Social class, n (%) 
    Professional 516 (47.1) 204 (45.8) 0.340 635 (47.1) 84 (43.5) 0.610 
    Intermediate 164 (15.0) 80 (17.9) 214 (15.9) 31 (16.1) 
    Working 415 (37.9) 161 (36.2) 499 (37.0) 78 (40.4) 
Smoking, n (%) 
    Nonsmoker 323 (40.2) 105 (30.4) 0.007 375 (37.2) 52 (36.4) 0.795 
    Ever smoker 373 (46.5) 185 (53.5) 488 (48.5) 73 (51.0) 
    Current smoker 107 (13.3) 56 (16.2) 144 (14.3) 18 (12.6) 
Weekly physical activity sessions, n (%) 
    None 369 (47.2) 157 (46.3) 0.425 456 (46.5) 69 (46.3) 0.940 
    1-2 times 256 (32.8) 114 (33.6) 327 (33.3) 44 (31.4) 
    3-4 times 110 (14.0) 40 (11.8) 133 (13.5) 18 (12.9) 
    ≥5 times 47 (6.0) 28 (8.3) 66 (6.7) 9 (6.4) 

Because genetic variants are randomly allocated at gamete formation, we expected no association between VDR genotypes and participants baseline characteristics. Consistent with our expectations, there was no association between VDR genetic variants and any of the baseline characteristics in Table 2 (all P values > 0.05; full results available from authors on request). Because none of these characteristics are associated with genotype (our exposure), they could not confound its association with cancer stage/grade. However, in all our association analyses, we present unadjusted and adjusted P values, with the latter adjusted for characteristics (age, PSA, and smoking), which are strongly associated with cancer status. This improves statistical efficiency in analyses and is equivalent to adjustment for baseline characteristics in randomized controlled trials.

Pearson χ2 tests found that the FokI genotype was slightly out of Hardy-Weinberg equilibrium in our controls (P = 0.02), but no disequilibrium was found for other variants (BsmI: P = 0.79; ApaI: P = 0.33; TaqI: P = 0.92; Cdx2: P = 0.09). Table 3 displays the genotype counts for each VDR variant stratified according to disease status. Based on previous research collected by systematic review, we fitted an additive model to Cdx2, a dominant model to FokI and TaqI, and a recessive model to ApaI and BsmI. No VDR variant was associated with tumor stage. By contrast, all variants, except Cdx2, were associated with cancer grade accessed by Gleason score (Table 3).

Table 3.

Genotype distribution of VDR variants according to prostate cancer stage and grade in the ProtecT participants

Prostate cancer grade (Gleason score)OR (95% CI)PAdjusted PProstate cancer stage (clinical stage)OR (95% CI)PAdjusted P
<7≥7LocalizedAdvanced
Cdx2 
    G:G 696 (63.4) 279 (62.1) — — — 860 (63.6) 116 (59.5) — — — 
    G:A 352 (32.1) 140 (31.2) 0.99 (0.78-1.26) 0.949 0.623 426 (31.5) 66 (33.8) 1.15 (0.83-1.59) 0.402 0.928 
    A:A 50 (4.5) 30 (6.7) 1.22 (0.97-1.55) 0.095 0.143 66 (4.9) 13 (6.7) 1.21 (0.88-1.65) 0.235 0.032 
Fok
    G:G (FF) 416 (37.9) 144 (32.3) — — — 489 (36.2) 70 (36.1) — — — 
    G:A (Ff) 534 (48.6) 241 (54.0) 1.30 (1.02-1.66) 0.032 0.042 681 (50.5) 96 (49.5) 0.98 (0.71-1.37) 0.927 0.378 
    A:A (ff) 148 (13.5) 61 (13.7) 1.09 (0.91-1.30) 0.333 0.434 180 (13.3) 28 (14.4) 1.04 (0.82-1.32) 0.729 0.822 
Bsm
    G:G (bb) 379 (34.3) 177 (39.4) — — — 491 (36.2) 65 (33.0) — — — 
    A:G (Bb) 547 (49.6) 205 (45.7) 0.80 (0.63-1.02) 0.072 0.04 656 (48.4) 96 (48.7) 1.11 (0.79-1.55) 0.559 0.848 
    A:A (BB) 178 (16.1) 67 (14.9) 0.90 (0.76-1.06) 0.204 0.126 209 (15.4) 36 (18.3) 1.14 (0.92-1.42) 0.239 0.361 
Apa
    C:C (aa) 217 (19.9) 107 (23.8) — — — 282 (21.0) 43 (22.0) — — — 
    A:C (Aa) 558 (51.1) 229 (51.0) 0.83 (0.63-1.10) 0.196 0.120 696 (51.7) 90 (46.2) 0.85 (0.57-1.25) 0.406 0.126 
    A:A (AA) 316 (29.0) 113 (25.2) 0.85 (0.73-1.00) 0.046 0.012 367 (27.3) 62 (31.8) 1.05 (0.85-1.30) 0.632 0.954 
Taq
    T:T (TT) 377 (34.4) 176 (39.2) — — — 492 (36.5) 61 (31.5) — — — 
    T:C (Tt) 539 (49.2) 206 (45.9) 0.82 (0.64-1.04) 0.103 0.034 647 (47.9) 98 (50.5) 1.22 (0.87-1.72) 0.249 0.799 
    C:C (tt) 179 (16.4) 67 (14.9) 0.90 (0.76-1.06) 0.193 0.089 211 (15.6) 35 (18.0) 1.16 (0.93-1.45) 0.201 0.355 
Prostate cancer grade (Gleason score)OR (95% CI)PAdjusted PProstate cancer stage (clinical stage)OR (95% CI)PAdjusted P
<7≥7LocalizedAdvanced
Cdx2 
    G:G 696 (63.4) 279 (62.1) — — — 860 (63.6) 116 (59.5) — — — 
    G:A 352 (32.1) 140 (31.2) 0.99 (0.78-1.26) 0.949 0.623 426 (31.5) 66 (33.8) 1.15 (0.83-1.59) 0.402 0.928 
    A:A 50 (4.5) 30 (6.7) 1.22 (0.97-1.55) 0.095 0.143 66 (4.9) 13 (6.7) 1.21 (0.88-1.65) 0.235 0.032 
Fok
    G:G (FF) 416 (37.9) 144 (32.3) — — — 489 (36.2) 70 (36.1) — — — 
    G:A (Ff) 534 (48.6) 241 (54.0) 1.30 (1.02-1.66) 0.032 0.042 681 (50.5) 96 (49.5) 0.98 (0.71-1.37) 0.927 0.378 
    A:A (ff) 148 (13.5) 61 (13.7) 1.09 (0.91-1.30) 0.333 0.434 180 (13.3) 28 (14.4) 1.04 (0.82-1.32) 0.729 0.822 
Bsm
    G:G (bb) 379 (34.3) 177 (39.4) — — — 491 (36.2) 65 (33.0) — — — 
    A:G (Bb) 547 (49.6) 205 (45.7) 0.80 (0.63-1.02) 0.072 0.04 656 (48.4) 96 (48.7) 1.11 (0.79-1.55) 0.559 0.848 
    A:A (BB) 178 (16.1) 67 (14.9) 0.90 (0.76-1.06) 0.204 0.126 209 (15.4) 36 (18.3) 1.14 (0.92-1.42) 0.239 0.361 
Apa
    C:C (aa) 217 (19.9) 107 (23.8) — — — 282 (21.0) 43 (22.0) — — — 
    A:C (Aa) 558 (51.1) 229 (51.0) 0.83 (0.63-1.10) 0.196 0.120 696 (51.7) 90 (46.2) 0.85 (0.57-1.25) 0.406 0.126 
    A:A (AA) 316 (29.0) 113 (25.2) 0.85 (0.73-1.00) 0.046 0.012 367 (27.3) 62 (31.8) 1.05 (0.85-1.30) 0.632 0.954 
Taq
    T:T (TT) 377 (34.4) 176 (39.2) — — — 492 (36.5) 61 (31.5) — — — 
    T:C (Tt) 539 (49.2) 206 (45.9) 0.82 (0.64-1.04) 0.103 0.034 647 (47.9) 98 (50.5) 1.22 (0.87-1.72) 0.249 0.799 
    C:C (tt) 179 (16.4) 67 (14.9) 0.90 (0.76-1.06) 0.193 0.089 211 (15.6) 35 (18.0) 1.16 (0.93-1.45) 0.201 0.355 

NOTE: Adjusted P, adjusted P value by age, PSA level (logarithm transformed), and smoking status.

Strong LD exists between the genetic loci, BsmI, ApaI, and TaqI (BsmI and ApaI D′ = 1, R2 = 0.58; ApaI and TaqI D′ = 0.99, R2 = 0.58; BsmI and TaqI D′ = 0.98, R2 = 0.96) but not with the loci Cdx2 or FokI (Fig. 1). Table 4 presents the result of a haplotype analysis using the BsmI, ApaI, and TaqI loci. There was a weak association between the B-A-t haplotype and Gleason score [OR, 0.84; 95% confidence interval (95% CI), 0.71-1.00; Punadjusted = 0.050; Padjusted = 0.014] relative to the common haplotype b-a-T, suggesting the protective effect of the B-A-t haplotype in prostate cancer grade.

Figure 1.

LD map of five VDR variants.

Figure 1.

LD map of five VDR variants.

Close modal
Table 4.

Haplotype distribution of BsmI-ApaI-TaqI variants by prostate cancer stage and grade in the ProtecT participants

Prostate cancer grade (Gleason score)OR (95% CI)PAdjusted PProstate cancer stage (clinical stage)OR (95% CI)PAdjusted P
<7≥7LocalizedAdvanced
b-a-T 0.451 0.491 — — — 0.467 0.439 — — — 
B-A-t 0.406 0.373 0.84 (0.71-1.00) 0.050 0.011 0.393 0.424 1.15 (0.91-1.45) 0.246 0.594 
b-A-T 0.132 0.123 0.86 (0.67-1.10) 0.219 0.119 0.131 0.121 0.99 (0.70-1.41) 0.971 0.976 
Prostate cancer grade (Gleason score)OR (95% CI)PAdjusted PProstate cancer stage (clinical stage)OR (95% CI)PAdjusted P
<7≥7LocalizedAdvanced
b-a-T 0.451 0.491 — — — 0.467 0.439 — — — 
B-A-t 0.406 0.373 0.84 (0.71-1.00) 0.050 0.011 0.393 0.424 1.15 (0.91-1.45) 0.246 0.594 
b-A-T 0.132 0.123 0.86 (0.67-1.10) 0.219 0.119 0.131 0.121 0.99 (0.70-1.41) 0.971 0.976 

NOTE: Adjusted P, adjusted P value by age, PSA level (logarithm transformed), and smoking status.

Systematic Review and Meta-analysis

Figure 2 shows the flow of articles identified and finally included in our meta-analysis. Our systematic search identified 747 articles that were potentially relevant. However, only 17 of these examined the relationship between VDR polymorphisms and prostate cancer stage/grade. Of these 17 articles, 2 were excluded because they used PSA or bone scan only to stage prostate cancer. Of the remaining 15, after contact with authors, we were able to obtain relevant data to include 12 in our meta-analyses (together with our own results). Table 5 summarizes the studies. Most of the studies are small, with <200 participants. The one exception is a recent study with similar numbers to our primary analyses: 1,034 men assayed for BsmI and 1,010 men genotyped for FokI (13). No association was detected between VDR polymorphisms and prostate cancer disease status (assessed by either Gleason grade or TNM stage) in that study.

Figure 2.

Flow chart of articles identification in systematic review and meta-analysis.

Figure 2.

Flow chart of articles identification in systematic review and meta-analysis.

Close modal
Table 5.

Association study between VDR polymorphisms and prostate cancer progression

StudyCity or countryNo. cases*No. controls*Cancer evaluationPolymorphism(s) reported
Blazer, 2000 (24) North Carolina 59 12 TNM stage TaqI [0.42] 
CGEMS, 2008 (42) United States 689 448 Gleason score and TNM stage BsmI [0.41] 
Cicek, 2006 (25) Detroit, MI and Cleveland, OH 225 188 Gleason score or TNM stage TaqI [0.37], BsmI [0.38], ApaI [0.47], FokI [0.38], Cdx2 [0.25] 
Furuya, 1999 (26) Japan 53 13 TNM stage TaqI [0.19] 
Gsur, 2002 (27) Vienna, Austria 111 79 Gleason score TaqI [0.40] 
Hamasaki, 2002 (28) Japan 43 (G) 77 (G) Gleason score TaqI [0.11] 
76 (T) 44 (T) TNM stage  
Holick, 2007 (46) Washington 80 (G) 545 (G) Gleason score TaqI, Bsm
20 (T) 602 (T) TNM stage  
Huang, 2004 (29) Taiwan 71 (G) 89 (G) Gleason score TaqI [0.04], BsmI [0.05], ApaI [0.39] 
73 (T) 84 (T) TNM stage  
Huang, 2006 (30) Taiwan 102 204 TNM stage FokI [0.46] 
Keitheri Cheteri, 2004 (31) Washington 76 (G) [B] 459 (G) [B]  BsmI [0.42], FokI [0.40] 
78 (G) [F] 466 (G) [F] Gleason score  
141 (T) [B] 396 (T) [B] TNM stage  
143 (T) [F] 403 (T) [F]  
Li, 2007 (48) United States 466 [F] 544 [F] Gleason score and TNM stage BsmI [0.40], FokI [0.39] 
487 [B] 547 [B]  
Liu, 2004 (32) Beijing, China 57 (G) [T,B]   TaqI [0.06], BsmI [0.05], ApaI [0.32] 
55 (G) [A] 24 (G) Gleason score  
54 (T) 27 (T) TNM stage  
Maistro, 2004 (49) Brazil 81 (G) 80 (G,T) Gleason score TaqI, Apa
43 (T) TNM stage  
Mikhak, 2007 (14) HPFS 35.1% (G) 64.9% (G) Gleason score BsmI, FokI, Cdx2 
8.7% (T) 92.3% (T) TNM stage  
Suzuki, 2003 (33)§ Japan 58 (G) 23 (G) Gleason score TaqI [0.15], BsmI [0.18], ApaI [0.40] 
21 (T) 58 (T) TNM stage  
Williams, 2004 (34) Detroit, MI 267 159 Gleason score BsmI [0.39] 
Williams, 2004 (34) Detroit, MI 208 102 Gleason score BsmI [0.27] 
StudyCity or countryNo. cases*No. controls*Cancer evaluationPolymorphism(s) reported
Blazer, 2000 (24) North Carolina 59 12 TNM stage TaqI [0.42] 
CGEMS, 2008 (42) United States 689 448 Gleason score and TNM stage BsmI [0.41] 
Cicek, 2006 (25) Detroit, MI and Cleveland, OH 225 188 Gleason score or TNM stage TaqI [0.37], BsmI [0.38], ApaI [0.47], FokI [0.38], Cdx2 [0.25] 
Furuya, 1999 (26) Japan 53 13 TNM stage TaqI [0.19] 
Gsur, 2002 (27) Vienna, Austria 111 79 Gleason score TaqI [0.40] 
Hamasaki, 2002 (28) Japan 43 (G) 77 (G) Gleason score TaqI [0.11] 
76 (T) 44 (T) TNM stage  
Holick, 2007 (46) Washington 80 (G) 545 (G) Gleason score TaqI, Bsm
20 (T) 602 (T) TNM stage  
Huang, 2004 (29) Taiwan 71 (G) 89 (G) Gleason score TaqI [0.04], BsmI [0.05], ApaI [0.39] 
73 (T) 84 (T) TNM stage  
Huang, 2006 (30) Taiwan 102 204 TNM stage FokI [0.46] 
Keitheri Cheteri, 2004 (31) Washington 76 (G) [B] 459 (G) [B]  BsmI [0.42], FokI [0.40] 
78 (G) [F] 466 (G) [F] Gleason score  
141 (T) [B] 396 (T) [B] TNM stage  
143 (T) [F] 403 (T) [F]  
Li, 2007 (48) United States 466 [F] 544 [F] Gleason score and TNM stage BsmI [0.40], FokI [0.39] 
487 [B] 547 [B]  
Liu, 2004 (32) Beijing, China 57 (G) [T,B]   TaqI [0.06], BsmI [0.05], ApaI [0.32] 
55 (G) [A] 24 (G) Gleason score  
54 (T) 27 (T) TNM stage  
Maistro, 2004 (49) Brazil 81 (G) 80 (G,T) Gleason score TaqI, Apa
43 (T) TNM stage  
Mikhak, 2007 (14) HPFS 35.1% (G) 64.9% (G) Gleason score BsmI, FokI, Cdx2 
8.7% (T) 92.3% (T) TNM stage  
Suzuki, 2003 (33)§ Japan 58 (G) 23 (G) Gleason score TaqI [0.15], BsmI [0.18], ApaI [0.40] 
21 (T) 58 (T) TNM stage  
Williams, 2004 (34) Detroit, MI 267 159 Gleason score BsmI [0.39] 
Williams, 2004 (34) Detroit, MI 208 102 Gleason score BsmI [0.27] 

*For the articles using different cancer progression measurements and/or reporting data on multiple polymorphisms, if the numbers of cases and/or control are different by progression measurement, G for Gleason scores and T for TNM stage, will be labeled in parentheses; if the numbers of cases and/or control are different by polymorphisms, the polymorphisms will be labeled in square brackets as A for ApaI, B for BsmI, C for Cdx2, F for FokI, and T for TaqI.

The number in the squared brackets is minor allele frequency in study populations.

Not included in the meta-analysis due to lack of data.

§TNM staging data are T1-3 versus N+.

White population only.

African American only.

In addition to our systematic search for published VDR advanced prostate cancer association studies, we also searched the currently available whole-genome association databases of prostate cancer (as of August 2008). That search showed that BsmI is the only polymorphism included in the current genome-wide chips. Data from a genome-wide association study (43)13

has also been included in our meta-analysis.

A fixed-effects meta-analysis was done on four genetic variants, FokI (n = 5 studies contributing), ApaI (n = 5 studies contributing), BsmI (n = 9 studies contributing), and TaqI (n = 9 studies contributing), using Gleason score or TNM stage. Only one other study (in addition to ours) had examined the association of Cdx-2 variant with advanced prostate cancer; therefore, we did not complete a meta-analysis for that variant. Figure 3 shows the results of the meta-analyses for these four variants with Gleason score and prostate cancer stage. ApaI aa homozygotes were more likely to have a high cancer grade compared with ApaI-A allele carriers (AA and Aa; OR, 1.25; 95% CI, 1.02-1.53; P = 0.034; Fig. 3C). The same effect has been found at the BsmI locus. Patients with two copies of the BsmI-b allele are at high risk of advanced cancer compared with BsmI-B carriers (OR, 1.11; 95% CI, 1.00-1.25; P = 0.079; Fig. 3B). In contrast, patients with TaqI-t allele are at low risk of high Gleason score compared with TT carriers (OR, 0.82; 95% CI, 0.69-0.98; P = 0.03; Fig. 3D). There was no strong evidence that the FokI polymorphism affected prostate cancer grade (Fig. 3A). None of the variants were associated with cancer stage in these meta-analyses (Fig. 3). Heterogeneity between studies in these different meta-analyses varied, with I2 between 0% and 50.7%. Given these relatively low levels of between study heterogeneity, we did not conduct further meta-regression analyses. An Egger test (41) and a Begg test (42) found no strong evidence of small study bias (all P values > 0.05).

Figure 3.

A. Funnel plots of meta-analysis of four VDR variants: FokI. B. Funnel plots of meta-analysis of four VDR variants: BsmI. C. Funnel plots of meta-analysis of four VDR variants: ApaI. D. Funnel plots of meta-analysis of four VDR variants: TaqI.

Figure 3.

A. Funnel plots of meta-analysis of four VDR variants: FokI. B. Funnel plots of meta-analysis of four VDR variants: BsmI. C. Funnel plots of meta-analysis of four VDR variants: ApaI. D. Funnel plots of meta-analysis of four VDR variants: TaqI.

Close modal

We carried out a genetic association study and meta-analysis of genetic variants in VDR and prostate cancer stage and/or grade at diagnosis. We found no consistent association of genetic variation in the VDR with cancer stage. However, there was an association between BsmI, ApaI, and TaqI genotypes and prostate cancer grade in both the ProtecT study and a meta-analysis of 13 studies published to date. The overall summary ORs showed that ApaI-a and BsmI-b increased the risk of a high cancer grade, whereas the Taq-t allele was protective. Haplotype analysis of BsmI, ApaI, and TaqI variants in the VDR gene confirmed the associations of individual genotypes with cancer grade, showing that patients with the B-A-t haplotype are at lower risk of having progressed at diagnosis compared with those with the b-a-T haplotype.

A study of 599 healthy men reported that those with the bb genotype at the BsmI locus had, on average, 2.3 pg/mL lower levels of 1,25-dihydroxyvitamin D compared with BB carriers (44). If we assume that this association is causal (as our biological understanding of VDR would suggest), the summary OR of 1.12 from the meta-analysis of BsmI locus would be equivalent to an OR for advanced prostate cancer at diagnosis of 1.63 for a 10 pg/mL reduction of plasma 1,25-dihydroxyvitamin D. However, this is an over simplistic calculation given that VDR is not only involved in the metabolism of vitamin D but also is a nuclear receptor and a transcription factor, which mediates the biological functions of vitamin D (45). Changes in VDR may therefore also have direct effects on vitamin D signaling and downstream effects on gene expression, which will in turn control cell proliferation. Our results provide some support for the hypothesis that plasma vitamin D level is an important factor in prostate cancer progression. This study represents a potentially important public health finding, as it suggests that oral vitamin supplementation (1,25-dihydroxyvitamin D) may help to prevent prostate cancer progression (15, 16). However, randomized controlled trial evidence is ultimately required to confirm the effectiveness of vitamin D supplementation in men with prostate cancer. Our results support the importance of funding for such trials.

An important finding from our study and others identified in our systematic review was that associations of VDR with prostate cancer status were only consistently found for cancer Gleason grade and were not found for TNM stage. Both Gleason score and TNM are accepted as valid and reliable measures of assessing the aggressiveness and extent of prostate cancer and for predicting pathologic characteristics (46). However, Gleason score was found to have higher predictive accuracy for biochemical recurrence compared with TNM staging system in a study of 971 advanced prostate cancer patients (47). In our study, we found little overlap between advanced prostate cancer cases defined by these two methods (see Table 1). Only 19.4% of patients with Gleason score ≥7 have cancer stage ≥T2c, whereas 72.2% of patients at stage ≥T2c have Gleason score ≥7. A similar situation also occurs in other studies in our systematic review, suggesting that Gleason score may be a more sensitive measure of aggressive prostate cancer than TNM stage (14, 32, 48). However, it is important to recognize that complete pathologic staging was not available in the majority of our cases and that histologic evaluation of Gleason scores is well known to be subject to variability, with both these factors potentially confounding the lack of correlation shown in our study.

Study Limitations

As with all genetic association studies, population admixture may have biased our findings. This may be particularly important here because prostate cancer prevalence varies with ancestry (49) as does VDR genotype (Table 5). However, only British individuals of European origin were included in our analysis and there was no difference in genotype distribution across the centers that recruited men for the study. Despite being from different populations, there was little evidence of between-study heterogeneity in the associations assessed in our meta-analysis. All of the men with prostate cancer in this study were identified through a population-based PSA testing protocol (no one had been diagnosed with prostate cancer at recruitment); therefore, survivor bias is unlikely to have affected our results. This is supported by the fact that mean age did not differ by genotype for any of the VDR variants.

The meta-analysis combined studies from diverse populations, although the recent larger studies, such as Li et al. (13) and CGEMS (43), are based on American Caucasians. Considering the small effect of VDR variants, population diversity and small sample size could reduce the power of the meta-analysis to reveal causal associations and our results would benefit from further replication in other large studies.

To conclude, our finding of associations between genetic variants in VDR and advanced prostate cancer at diagnosis provides some support for the hypothesis that low levels of vitamin D may increase the risk of prostate cancer progression. Further large studies that replicate these findings are needed to confirm this hypothesis.

No potential conflicts of interest were disclosed.

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.

We thank the tremendous contribution of all members of the ProtecT study research group and especially the following people who were involved in this research: Prasad Bollina, Sue Bonnington, Debbie Cooper, Andrew Doble, Alan Doherty, Emma Elliott, David Gillatt, Pippa Herbert, Peter Holding, Joanne Howson, Gemma Marsden, Mandy Jones, Roger Kockelbergh, Howard Kynaston, Teresa Lennon, Norma Lyons, Hilary Moody, Philip Powell, Stephen Prescott, Liz Salter, and Pauline Thompson.

1
Apperly
FL
. 
The relation of solar radiation to cancer mortality in North America
.
Cancer Res
1941
;
1
:
191
5
.
2
Schwartz
GG
,
Hill
CC
,
Oeler
TA
,
Becich
MJ
,
Bahnson
RR
. 
1,25-Dihydroxy-16-ene-23-yne-vitamin D3 and prostate cancer cell proliferation in vivo
.
Urology
1995
;
46
:
365
9
.
3
Moreno
J
,
Krishnan
AV
,
Peehl
DM
,
Feldman
D
. 
Mechanisms of vitamin D-mediated growth inhibition in prostate cancer cells: inhibition of the prostaglandin pathway
.
Anticancer Res
2006
;
26
:
2525
30
.
4
Krishnan
AV
,
Swami
S
,
Moreno
J
, et al
. 
Potentiation of the growth-inhibitory effects of vitamin D in prostate cancer by genistein
.
Nutr Rev
2007
;
65
:
S121
3
.
5
Corder
EH
,
Guess
HA
,
Hulka
BS
, et al
. 
Vitamin D and prostate cancer: a prediagnostic study with stored sera
.
Cancer Epidemiol Biomarkers Prev
1993
;
2
:
467
72
.
6
Ahonen
MH
,
Zhuang
YH
,
Aine
R
,
Ylikomi
T
,
Tuohimaa
P
. 
Androgen receptor and vitamin D receptor in human ovarian cancer: growth stimulation and inhibition by ligands
.
Int J Cancer
2000
;
86
:
40
6
.
7
Braun
MM
,
Helzlsouer
KJ
,
Hollis
BW
,
Comstock
GW
. 
Prostate cancer and prediagnostic levels of serum vitamin D metabolites
.
Cancer Causes Control
1995
;
6
:
235
9
.
8
Gann
PH
,
Ma
J
,
Hennekens
CH
, et al
. 
Circulating vitamin D metabolites in relation to subsequent development of prostate cancer
.
Cancer Epidemiol Biomarkers Prev
1996
;
5
:
121
6
.
9
Nomura
AM
,
Stemmermann
GN
,
Lee
J
, et al
. 
Serum vitamin D metabolite levels and the subsequent development of prostate cancer (Hawaii, United States)
.
Cancer Causes Control
1998
;
9
:
425
32
.
10
Jacobs
ET
,
Giuliano
AR
,
Martinez
ME
, et al
. 
Plasma levels of 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D and the risk of prostate cancer
.
J Steroid Biochem Mol Biol
2004
;
89–90
:
533
7
.
11
Platz
EA
,
Leitzmann
MF
,
Hollis
BW
,
Willett
WC
,
Giovannucci
E
. 
Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and subsequent risk of prostate cancer
.
Cancer Causes Control
2004
;
15
:
255
65
.
12
Tuohimaa
P
,
Tenkanen
L
,
Ahonen
M
, et al
. 
Both high and low levels of blood vitamin D are associated with a higher prostate cancer risk: a longitudinal, nested case-control study in the Nordic countries
.
Int J Cancer
2004
;
108
:
104
8
.
13
Li
H
,
Stampfer
MJ
,
Hollis
JB
, et al
. 
A prospective study of plasma vitamin D metabolites, vitamin D receptor polymorphisms, and prostate cancer
.
PLoS Med
2007
;
4
:
e103
.
14
Mikhak
B
,
Hunter
DJ
,
Spiegelman
D
, et al
. 
Vitamin D receptor (VDR) gene polymorphisms and haplotypes, interactions with plasma 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D, and prostate cancer risk
.
Prostate
2007
;
67
:
911
23
.
15
Brawer
MK
. 
Recent progress in the treatment of advanced prostate cancer with intermittent dose-intense calcitriol (DN-101)
.
Rev Urol
2007
;
9
:
1
8
.
16
Beer
TM
,
Ryan
CW
,
Venner
PM
, et al
. 
Double-blinded randomized study of high-dose calcitriol plus docetaxel compared with placebo plus docetaxel in androgen-independent prostate cancer: a report from the ASCENT Investigators
.
J Clin Oncol
2007
;
25
:
669
74
.
17
Ogunkolade
WB
,
Boucher
BJ
,
Bustin
SA
, et al
. 
Vitamin D metabolism in peripheral blood mononuclear cells is influenced by chewing “betel nut” (Areca catechu) and vitamin D status
.
J Clin Endocrinol Metab
2006
;
91
:
2612
7
.
18
Lou
YR
,
Qiao
S
,
Talonpoika
R
,
Syvala
H
,
Tuohimaa
P
. 
The role of vitamin D3 metabolism in prostate cancer
.
J Steroid Biochem Mol Biol
2004
;
92
:
317
25
.
19
Arai
H
,
Miyamoto
KI
,
Yoshida
M
, et al
. 
The polymorphism in the caudal-related homeodomain protein Cdx-2 binding element in the human vitamin D receptor gene
.
J Bone Miner Res
2001
;
16
:
1256
64
.
20
Jurutka
PW
,
Remus
LS
,
Whitfield
GK
, et al
. 
The polymorphic N terminus in human vitamin D receptor isoforms influences transcriptional activity by modulating interaction with transcription factor IIB
.
Mol Endocrinol
2000
;
14
:
401
20
.
21
Morrison
NA
,
Qi
JC
,
Tokita
A
, et al
. 
Prediction of bone density from vitamin D receptor alleles
.
Nature
1994
;
367
:
284
7
.
22
Davey Smith
G
,
Ebrahim
S
. 
'Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease?
Int J Epidemiol
2003
;
32
:
1
22
.
23
Davey Smith
G
,
Lawlor
DA
,
Harbord
R
, et al
. 
Clustered environments and randomized genes: a fundamental distinction between conventional and genetic epidemiology
.
PLoS Med
2007
;
4
:
e352
.
24
Blazer
DG
 III
,
Umbach
DM
,
Bostick
RM
,
Taylor
JA
. 
Vitamin D receptor polymorphisms and prostate cancer
.
Mol Carcinog
2000
;
27
:
18
23
.
25
Cicek
MS
,
Liu
X
,
Schumacher
FR
,
Casey
G
,
Witte
JS
. 
Vitamin D receptor genotypes/haplotypes and prostate cancer risk
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
2549
52
.
26
Furuya
Y
,
Akakura
K
,
Masai
M
,
Ito
H
. 
Vitamin D receptor gene polymorphism in Japanese patients with prostate cancer
.
Endocr J
1999
;
46
:
467
70
.
27
Gsur
A
,
Madersbacher
S
,
Haidinger
G
, et al
. 
Vitamin D receptor gene polymorphism and prostate cancer risk
.
Prostate
2002
;
51
:
30
4
.
28
Hamasaki
T
,
Inatomi
H
,
Katoh
T
,
Ikuyama
T
,
Matsumoto
T
. 
Significance of vitamin D receptor gene polymorphism for risk and disease severity of prostate cancer and benign prostatic hyperplasia in Japanese
.
Urol Int
2002
;
68
:
226
31
.
29
Huang
SP
,
Huang
CY
,
Wu
WJ
, et al
. 
Association of vitamin D receptor FokI polymorphism with prostate cancer risk, clinicopathological features and recurrence of prostate specific antigen after radical prostatectomy
.
Int J Cancer
2006
;
119
:
1902
7
.
30
Keitheri Cheteri
MB
,
Stanford
JL
,
Friedrichsen
DM
, et al
. 
Vitamin D receptor gene polymorphisms and prostate cancer risk
.
Prostate
2004
;
59
:
409
18
.
31
Liu
JH
,
Li
HW
,
Tong
M
,
Li
M
,
Na
YQ
. 
Genetic risk factors of prostate cancer in Han nationality population in northern China and a preliminary study of the reason of racial difference in prevalence of prostate cancer
.
Zhonghua Yi Xue Za Zhi
2004
;
84
:
364
8
.
32
Suzuki
K
,
Matsui
H
,
Ohtake
N
, et al
. 
Vitamin D receptor gene polymorphism in familial prostate cancer in a Japanese population
.
Int J Urol
2003
;
10
:
261
6
.
33
Williams
H
,
Powell
IJ
,
Land
SJ
, et al
. 
Vitamin D receptor gene polymorphisms and disease free survival after radical prostatectomy
.
Prostate
2004
;
61
:
267
75
.
34
Huang
SP
,
Chou
YH
,
Wayne Chang
WS
, et al
. 
Association between vitamin D receptor polymorphisms and prostate cancer risk in a Taiwanese population
.
Cancer Lett
2004
;
207
:
69
77
.
35
Donovan
J
,
Hamdy
F
,
Neal
D
, et al
. 
Prostate Testing for Cancer and Treatment (ProtecT) feasibility study
.
Health Technol Assess
2003
;
7
:
1
42
.
36
Wittekind
C
,
Compton
CC
,
Greene
FL
,
Sobin
LH
. 
TNM residual tumor classification revisited
.
Cancer
2002
;
94
:
2511
6
.
37
Barrett
JC
,
Fry
B
,
Maller
J
,
Daly
MJ
. 
Haploview: analysis and visualization of LD and haplotype maps
.
Bioinformatics
2005
;
21
:
263
5
.
38
Sinnwell JP, Schaid DJ. Haplo.stats: statistical analysis of haplotypes with traits and covariates when linkage phase is ambiguous: R package version 1.2.2; 2005.
39
Luscombe
CJ
,
French
ME
,
Liu
S
, et al
. 
Prostate cancer risk: associations with ultraviolet radiation, tyrosinase and melanocortin-1 receptor genotypes
.
Br J Cancer
2001
;
85
:
1504
9
.
40
Chokkalingam
AP
,
McGlynn
KA
,
Gao
YT
, et al
. 
Vitamin D receptor gene polymorphisms, insulin-like growth factors, and prostate cancer risk: a population-based case-control study in China
.
Cancer Res
2001
;
61
:
4333
6
.
41
Egger
M
,
Davey Smith
G
,
Schneider
M
,
Minder
C
. 
Bias in meta-analysis detected by a simple, graphical test
.
BMJ
1997
;
315
:
629
34
.
42
Begg
CB
,
Mazumdar
M
. 
Operating characteristics of a rank correlation test for publication bias
.
Biometrics
1994
;
50
:
1088
101
.
43
Yu
K
,
Wang
Z
,
Li
Q
, et al
. 
Population substructure and control selection in genome-wide association studies
.
PLoS ONE
2008
;
3
:
e2551
.
44
Ma
J
,
Stampfer
MJ
,
Gann
PH
, et al
. 
Vitamin D receptor polymorphisms, circulating vitamin D metabolites, and risk of prostate cancer in United States physicians
.
Cancer Epidemiol Biomarkers Prev
1998
;
7
:
385
90
.
45
Makishima
M
,
Lu
TT
,
Xie
W
, et al
. 
Vitamin D receptor as an intestinal bile acid sensor
.
Science
2002
;
296
:
1313
6
.
46
Partin
AW
,
Kattan
MW
,
Subong
EN
, et al
. 
Combination of prostate-specific antigen, clinical stage, and Gleason score to predict pathological stage of localized prostate cancer. A multi-institutional update
.
JAMA
1997
;
277
:
1445
51
.
47
Steuber
T
,
Erbersdobler
A
,
Graefen
M
, et al
. 
Comparative assessment of the 1992 and 2002 pathologic T3 substages for the prediction of biochemical recurrence after radical prostatectomy
.
Cancer
2006
;
106
:
775
82
.
48
Holick
CN
,
Stanford
JL
,
Kwon
EM
, et al
. 
Comprehensive association analysis of the vitamin D pathway genes, VDR, CYP27B1, and CYP24A1, in prostate cancer
.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
1990
9
.
49
Gronberg
H
. 
Prostate cancer epidemiology
.
Lancet
2003
;
361
:
859
64
.