Introduction

Prostate cancer is common but knowledge of modifiable risk factors is limited. Vitamin A (retinol), vitamin E, and vitamin D (total 25-hydroxyvitamin D, 25(OH)D; 1,25-dihydroxyvitamin, 1,25(OH)2D) may protect against prostate cancer, although harmful effects cannot be ruled out [1].

Retinol promotes cell differentiation and apoptosis, increases levels of other antioxidants, and regulates DNA transcription [2]. However, retinol may stimulate growth and de-differentiation of prostate cells, thus negating any anticancer activity [3]. Recent epidemiological studies of circulating retinol have reported positive [4], inverse [5], and null [6] associations, and there is insufficient evidence to draw conclusions regarding the effect of retinol on prostate cancer [7].

Vitamin E decreases DNA damage, enhances DNA repair, influences cellular response to oxidative stress, inhibits cell proliferation, enhances immune responses, and decreases cellular concentrations of testosterone [7]. Overall, there is limited, but largely suggestive, observational evidence that foods containing vitamin E and alpha-tocopherol supplements (a vitamin E metabolite) decrease the risk of prostate cancer [7]. This association, however, was not confirmed in two recent randomized controlled trials [1, 8].

Metabolites of vitamin D control cellular growth and differentiation [9], and the administration of vitamin D analogues inhibits the progression of prostate cancer in animal models [2] and in phase II trials [10]. Our recent meta-analyses found little evidence that circulating concentrations of 1,25(OH)2D (the active form of the hormone) were associated with prostate cancer but a potentially important protective effect of 1,25(OH)2D on advanced prostate cancer could not be excluded [7, 11]. Recent results from the cohort reported in this paper found a twofold increased risk of more aggressive (advanced stage and/or high grade) prostate cancers in men deficient in circulating 25(OH)D [12]. Some studies have shown positive associations of both low and high 25(OH)D with prostate cancer, but a review by Yin et al. [13] concluded that the overall literature is inconsistent. 1,25(OH)2D concentrations are tightly regulated [9] and only decrease significantly during severe deficiency of total 25(OH)D [14]. Therefore, 1,25(OH)2D may only become important for prostate cancer if there is a deficiency of 25(OH)D [15].

Retinol may mediate the effect of vitamin D on prostate cancer since retinol competitively binds to part of the same receptors as vitamin D [9, 16]. High retinol levels may also inhibit intestinal absorption, transport, or metabolism of vitamin D to its active form, or may stimulate degradation of vitamin D [16]. We hypothesize that high retinol potentially reduces any anticancer properties of vitamin D, thus increasing prostate cancer risk.

The current analysis investigates the associations of circulating retinol, vitamin E, and 1,25(OH)2D with PSA-detected prostate cancer, overall, and stratified by stage and grade, in a large UK-wide population-based case–control study [17]. We also investigated the possibility of an interaction between 25(OH)D and 1,25(OH)2D (i.e., whether 25(OH)D is associated with a reduced risk of prostate cancer in men with low 1,25(OH)2D and vice versa) and whether the association between 25(OH)D and prostate cancer differs in men with high compared with lower retinol levels.

Materials and methods

This case–control study is nested within a multi-center randomized controlled trial of treatments for localized prostate cancer: the Prostate Testing for cancer and Treatment (ProtecT) study, which has been described in detail previously [12, 17]. Briefly, men aged 50–69 years were offered a PSA test at a community-based “prostate check clinic,” and those with raised levels (≥3 ng/mL) were offered diagnostic biopsy. Detected tumors were histology confirmed, clinically staged (“localized”: T1–T2; “advanced”: T3–T4), and graded (“high grade”: Gleason score ≥7, combining intermediate (7) and high (8–10); low grade: Gleason score <7).

We randomly selected 1,500 cases and one control for each case from those men who had provided a non-fasted plasma heparin sample at the prostate check clinic and consented to the use of their samples for further prostate cancer research. Cases were men with a histology confirmed prostate cancer. All participants in the ProtecT prostate check clinics who had no evidence of prostate cancer were eligible for selection as controls, that is, men with a PSA test <3 ng/mL or a raised PSA (≥3 ng/mL) combined with at least one negative biopsy and no subsequent prostate cancer diagnosis during the follow-up protocol for negative biopsies. Controls were randomly selected from the same stratum—i.e., 5-year age band (age at PSA test) and GP/family practice—as the cases.

Plasma samples were drawn into heparinized tubes at the prostate check clinic and stored until required for use. Vitamins A and E are precipitated and extracted from plasma following the addition of an internal standard, separated by isocratic reverse phase HPLC using Chromsystems reagents and column (Manchester UK, product number 34,000) with UV detection of vitamin A at 325 nm and internal standard and vitamin E at 295 nm, using a programmable detector. 1,25(OH)2D samples were quantified by immunoassay [18] over a 2-month period using a single batch of reagents. 25(OH)D2 and 25(OH)D3 were measured using tandem mass spectrometry, in 31 batches over a period of approximately 3 months (50 % of 25(OH)D samples from matched cases–controls were assayed in the same batches; therefore, analyses including 25(OH)D were additionally adjusted for assay batch) [12].

Retinol and total vitamin E were measured in micromoles per liter (μmol/L). Circulating concentrations of 25(OH)D2 and 25(OH)D3 were measured in nanomoles per liter (nmol/L) where 1 ng/mL = 2.5 nmol/L and 1,25(OH)2D were measured in picomoles per liter (pmol/L) where 1 pg/mL = 2.6 pmol/L. Total 25(OH)D was calculated as the summation of 25(OH)D2 and 25(OH)D3.

Measures of baseline covariates (presented in Table 1) were collected at the time of the initial PSA test, either by questionnaire or by nurse interview. These measures were obtained prior to the knowledge of the PSA level or diagnosis in 85 % of men. We calculated body mass index (BMI; kg/m2), which represents general adiposity, and mean arterial pressure (MAP = ((2*diastolic) + systolic)/3; mmHg), which represents average arterial blood pressure during a single cardiac cycle.

Table 1 Baseline characteristics of cases and controls included in the study
Full size table

Statistical analysis

To allow for the matched sets of cases and controls, conditional logistic regression was used to estimate odds ratios (OR) and 95 % confidence intervals (CI) quantifying the association between exposure and all prostate cancers. The model included all case–control matching variables, as well as exact age. Matching on GP surgery accounted for study center (i.e., geographical location) and season of blood draw, as clinics were held over a number of consecutive weeks at each GP surgery. A case-only analysis used unconditional logistic regression, controlling for age, GP surgery, study center, and season of blood draw, to quantify the associations of circulating vitamin concentrations with prostate cancer stage (advanced vs localized) and grade (high vs low). A case-only analysis was used as all cases have undergone biopsy, therefore removing potential detection bias which could otherwise occur through misclassification of control status because of imperfect sensitivity of the PSA test [19].

Even-sized quintiles of 1,25(OH)2D, vitamin E, and retinol were calculated. Predetermined categories of 25(OH)D were defined as high: ≥75 nmol/L; adequate: 50–<75 nmol/L; insufficient: 30–<50 nmol/L; and deficient: <30 nmol/L as in our previous publication [12]. We computed ORs and 95 % CIs for the associations of prostate cancer per standard deviation (SD) increase in vitamin exposure. The association of 25(OH)D and 1,25(OH)2D with prostate cancer was repeated, stratified by retinol level (dichotomized around the median). Concentrations of 25(OH)D and 1,25(OH)2D were grouped into tertiles, and the association of each with prostate cancer was repeated stratified by tertile of the other, to assess whether there was an interaction between 25(OH)D and 1,25(OH)2D (tertiles of 25(OH)D: low 8.6–46.8 nmol/L, Q2 46.9–64.5 nmol/L, high 64.6–163.3 nmol/L; tertiles of 1,25(OH)2D: low 0–83.2 pmol/L, Q2 83.3–122.1 pmol/L, high 122.2–408 pmol/L). Sensitivity analyses were carried out additionally adjusting for family history of prostate cancer, BMI, weekly exercise, diabetes, and smoking status. Sensitivity analyses confirmed that these measures did not confound observed associations.

To avoid bias caused by complete-case analysis [20], we multiply imputed missing covariate values (i = 10) using chained equations [21], assuming those values could be predicted without bias from the observed relationships between covariates and the outcome measure, and substituting imputed values for missing values. Analyses were carried out in Stata 11 (StataCorp, 2009. College Station, TX, USA) using the-ice-for multiple imputation with chained equations [21] for imputing missing data. All tests of statistical significance were two-sided.

Results

Baseline characteristics of study participants

The current analysis includes 1,433 cases (1,277 (89.1 %) localized, 151 (10.5 %) advanced, 5 (0.3 %) missing stage; 963 (67.2 %) low grade, 466 (32.5 %) high grade, 4 (0.3 %) missing grade) and 1,433 controls that had a retinol and vitamin E measurement available. There were 1,282 cases (1,148 (89.5 %) localized, 129 (10.1 %) advanced, 5 (0.4 %) missing stage; 854 (66.6 %) low grade, 424 (33.1 %) high grade, 5 (0.3 %) missing grade) and 1,290 controls that also had an available 1,25(OH)2D measurement available. There were no substantial differences in baseline characteristics between cases and controls, except that more cases had a family history of prostate cancer versus controls and fewer cases had diabetes (Table 1). Of the 99.3 % of subjects who had recorded ethnicity, 98.9 % self-identified as white.

Characteristics of cases and controls by vitamin concentration

The overall unadjusted mean retinol was 1.8 μmol/L (standard deviation (SD) = 0.46; inter-quartile range (IQR): 1.5, 2.1), mean vitamin E was 15.8 μmol/L (SD = 10.7; IQR: 7.0, 24.1) and mean 1,25(OH)2D was 107.4 pmol/L (SD = 47.0; IQR: 73.7, 133.8). There was no correlation between PSA and any vitamin level (ρ between −0.03 and 0.02). Unexpectedly, men whose blood was drawn in winter had lower 1,25(OH)2D (p = 0.001 compared with summer), and men with prostate cancer whose blood was drawn in winter had higher vitamin E (p = 0.001 compared with summer), which may be diet-related.

Concentrations of vitamins and prostate cancer risk

There was no evidence that either increasing (a linear effect) or high and low (a nonlinear effect) circulating retinol, vitamin E, or 1,25(OH)2D concentrations were associated with overall prostate cancer risk (Table 2) or stage/grade of prostate cancer (Supplementary table 1). There was no evidence of either a linear or nonlinear association between 1,25(OH)2D and advanced versus localized or high-grade versus low-grade prostate cancer (Table 3). There was some evidence of an association between both high and low 1,25(OH)2D and advanced prostate cancer when compared with controls. Additionally, adjusting for family history, BMI, diabetes, smoking status, weekly exercise, or storage time of the sample did not alter the results.

Table 2 Associations of retinol (μmol/L), vitamin E (μmol/L), and 1,25(OH)2D (pmol/L) with risk of prostate cancer
Full size table
Table 3 Association between 1,25(OH)2D (pmol/L) and prostate cancer stage and grade in cases only, and between advanced/high-grade cases versus controls
Full size table

There was no evidence of an interaction of 25(OH)D and 1,25(OH)2D with prostate cancer risk (p interaction = 0.56). Similarly, there was no evidence of an interaction of 25(OH)D and 1,25(OH)2D with stage (p interaction = 0.24) or grade (p interaction = 0.93).

There was no evidence of an interaction between 25(OH)D and retinol with prostate cancer risk (p interaction = 0.34), stage (p interaction = 0.08), or grade (OR p interaction = 0.18).

Discussion

Summary of findings

We found no associations of circulating plasma retinol, vitamin E, or 1,25(OH)2D with overall prostate cancer risk or with more aggressive prostate cancer phenotypes (higher stage or grade). There was no evidence of an interaction between 1,25(OH)2D and total 25(OH)D on prostate cancer risk, and no evidence that the previously reported association of 25(OH)D with prostate cancer stage and grade [12] was modified by retinol level.

Previous evidence as to whether retinol, vitamin E, or 1,25(OH)2D increase or decrease prostate cancer risk has been conflicting. Results from the alpha-tocopherol, beta-carotene (ATBC) trial of male smokers (n = 29,104) found that higher serum retinol was associated with elevated risk of both total (n = 2,041) and advanced stage/high-grade (n = 461) prostate cancer, with a 20 % greater overall risk for men in the highest retinol quintile (HR = 1.19, CI: 1.03, 1.36, p trend = 0.009) [4]. Conversely, higher serum retinol was associated with decreased risk of advanced stage/high-grade prostate cancer in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial (692 cases, 844 controls; OR comparing highest vs lowest = 0.58, CI: 0.36, 0.92, p trend = 0.02) [5]. The current null findings are in line with results investigating circulating retinol from the European Prospective Investigation into Cancer and Nutrition (EPIC) study [6].

Two randomized controlled trials, PHS-II [8] (n = 14,641; median follow-up 8 years) and SELECT [1] (n = 35,533; median follow-up 5.5) found no evidence that vitamin E supplementation protected against prostate cancer. However, additional follow-up within SELECT (median follow-up 7 years) led to the observation that the risk of prostate cancer was increased by 17 % in men randomized to supplementation with vitamin E alone (HR = 1.17, CI: 1.00, 1.36, p = 0.008) [22]. Within ATBC, higher serum alpha-tocopherol was associated with lower risk of developing prostate cancer (RR for highest vs lowest levels = 0.80, CI: 0.66, 0.96, p trend = 0.03), particularly advanced stage cancer (RR = 0.56, CI: 0.36, 0.85, p trend = 0.002) [23]. There was no evidence of an association between circulating vitamin E concentrations and prostate cancer risk, stage, or grade in the current study, in line with results from EPIC [6].

Our previous meta-analysis found that a potentially important protective effect of 1,25(OH)2D on advanced stage prostate cancer could not be excluded (696 cases; pooled OR per 10 pg/mL increase in 1,25(OH)2D = 0.86, 95 % CI: 0.72, 1.02; p = 0.09) [11], although this result was based on only two studies. A protective effect was not observed in the current analysis. There was no association of 1,25(OH)2D with prostate cancer among men who were in the lowest tertile of 25(OH)D (<46.8 nmol/L; p interactions ≥0.24).

Our previous meta-analysis also found no association between 25(OH)D concentrations and prostate cancer or aggressive (advanced stage and/or high-grade) prostate cancer [11]. However, our recent results from the current cohort found a twofold increased risk of advanced stage versus localized prostate cancer (1,447 cases, 1,449 controls; OR for deficient vs adequate 25(OH)D = 2.33, CI: 1.26, 4.28) and high-grade versus low-grade cancer in men deficient in circulating 25(OH)D (OR for deficient vs adequate 25(OH)D = 1.78, CI: 1.15, 2.77) [12]. One previous study found that men who had both 25(OH)D and 1,25(OH)2D levels below the median had an increased risk of aggressive (defined as advanced stage, high grade, metastatic, or fatal) prostate cancer (OR = 2.1, CI: 1.2, 3.4), although the interaction between the two metabolites was nonsignificant (p for interaction = 0.23) [15]. The current study found no evidence that the association between 25(OH)D and prostate cancer was modified by a man’s retinol level, nor by his 1,25(OH)2D level, as suggested by others [15].

Strength and limitations of our study

It is possible that we are studying a relatively healthy population, within which there is insufficient variation in vitamin levels to be able to ascertain correlations with very high or very low levels. There is potential for residual confounding, as vitamin status is likely to be a marker of overall health and healthy diet but we have adjusted for the main risk factors for prostate cancer. It is difficult to separate the effects of particular vitamins from the effects of an overall healthy diet and from associated foods. Since the decision to biopsy was based on PSA level, some of the controls with PSA < 3 ng/mL will have unidentified prostate cancer [19] (misclassification bias) but this would not affect our analysis of advanced stage versus localized cancers (as all cancers were biopsy confirmed). Any misclassification of cancer status is likely to be non-differential with respect to vitamin status, at most moderately attenuating any effect estimates. We cannot rule out a role for retinol, vitamin E, or 1,25(OH)2D in prostate cancer progression [24, 25]. Prognostic significance will be tested once results are available. Most of the “advanced” stage cases are actually “locally advanced” as few have metastasized [9 of 151 (6 %) have distal metastasis (T4 or M1)]. Since the men had prostate cancer at the time of vitamin measurement, it is theoretically possible that a protective effect could be masked due to reverse causality if prostate cancers caused circulating levels of the vitamins to rise. However, there is no evidence to support this phenomenon. As the cancers were detected by screening and hence early on in their natural history, it is likely that our inference is restricted to associations of these vitamins with prostate cancer initiation rather than progression to clinically identifiable disease.

The strengths of our study are the large sample, about which we have extensive information recorded. Circulating plasma concentrations of vitamins were measured at one laboratory, in as few batches and in as short a time frame as possible (thus, attenuating any potential technical errors of measurement). Recall bias is unlikely since the questionnaire data were predominantly collected prior to the results of the PSA test being known (in 85 % of men). The study is population-based and thus subject to little selection bias. The case-only design used to investigate the associations of advanced/high-grade versus localized/low-grade cancers ensures that results are not affected by misclassification bias, as all cancers are biopsy confirmed.

Conclusion

Our study found no evidence that lower or higher retinol, vitamin E, or 1,25(OH)2D concentrations were associated with overall prostate cancer risk or more aggressive prostate cancer phenotypes (higher stage and/or grade). There was no evidence of an association of 1,25(OH)2D with prostate cancer when limited to men deficient in 25(OH)D. The association between 25(OH)D and prostate cancer was not modified by retinol levels. Future studies are needed to clarify why some previous studies have found harmful effects of these vitamins.