Serum and Dietary Vitamin E in Relation to Prostate Cancer Risk

The α-Tocopherol, β-Carotene Cancer Prevention (ATBC) Study (1), a large controlled trial, showed a significant 32% reduction in prostate cancer incidence in response to 5 to 8 years of supplementation with α-tocopherol (50 mg daily; ref. 2). Some observational studies of serum, dietary, and supplemental vitamin E are supportive in reporting inverse associations with prostate cancer (3-9), although not all risk estimates are statistically significant, and some studies are null (10-22). Several studies show reduced risks for higher serum concentrations of α-tocopherol or supplemental vitamin E among current or recent smokers with respect to advanced or fatal prostate cancer (23-26). In a recent nested case-control analysis in ATBC, with 100 cases and 200 matched controls, we found significant inverse associations with serum α-tocopherol and γ-tocopherol (27). Among the eight natural forms of vitamin E (α, β, γ, and δ tocopherol and α, β, γ, and δ tocotrienol), the majority of studies have focused on total vitamin E or α-tocopherol (the major isoform found in the bloodstream), whereas only a few studies have examined γ-tocopherol (7-10, 25-27). In addition, in the most recent update to the dietary reference intakes, α-tocopherol is the only form used to establish recommended intakes (28), and studies with vitamin E measured as α-tocopherol equivalents may not be as relevant as where data are presented as α-tocopherol only.

We report here findings about serum and dietary vitamin E and prostate cancer risk based on nearly 20 years of ATBC Study cohort follow-up and 1,732 incident cases. Our analysis included sufficient power to investigate the associations in subgroups defined by disease stage, smoking dose and duration, length of follow-up, and other prostate cancer risk factors of interest. Our previous prospective examination of these associations included only 317 cases and was essentially confined to the trial intervention period and showed only modest relationships between serum and dietary vitamin E and prostate cancer risk in the men receiving the trial α-tocopherol supplement (29).

The original ATBC Study was conducted in Finland as a joint project between the National Public Health Institute of Finland and the U.S. National Cancer Institute. The overall design, rationale, and objectives of this intervention study have been published (30). Briefly, this was a randomized, double-blind, placebo-controlled, primary prevention trial to determine whether daily supplementation with α-tocopherol, β-carotene, or both would reduce the incidence of lung or other cancers among male smokers. Between 1985 and 1988, 29,133 men ages 50 to 69 years, who smoked at least five cigarettes daily, were recruited from southwestern Finland. Participants were randomly assigned to receive either α-tocopherol as dl-α-tocopheryl acetate (50 mg/d), β-carotene as all-trans-β-carotene (20 mg/d), both supplements, or placebo capsules for 5 to 8 years (median, 6.1 years) until death or trial closure (April 30, 1993). Men who had prior cancer (other than nonmelanoma skin cancer or carcinoma in situ) or serious illness or who reported current use of vitamins E (>20 mg/d), A (>20,000 IU/d), or β-carotene (>6 mg/d) were ineligible. Written informed consent was obtained from each participant before randomization and the study was approved by the institutional review boards of the U.S. National Cancer Institute and the National Public Health Institute of Finland. Postintervention follow-up continued through the Finnish Cancer Registry with data available through April 2004.

Incident prostate cancer cases (International Classification of Diseases 9, code 185), diagnosed by April 30, 2004 (up to 19 years of follow-up), were identified through the Finnish Cancer Registry, which provides ∼100% case ascertainment (31). For cases diagnosed through April 1999, the medical records were reviewed centrally by two study oncologists for diagnostic confirmation and staging, and cases with histopathologic and cytologic specimens available were reviewed and confirmed by pathologists. Prostate cancer cases diagnosed after April 1999 had only the Finnish Cancer Registry data for site, histology, and diagnosis date available. Advanced cases (n = 319) were defined as those cases diagnosed through April 1999 with stage III or IV of the tumor-node-metastasis staging system, as defined by the American Joint Committee on Cancer (32). Finland has not adopted population-based prostate-specific antigen screening programs, and only 1 of 246 cases that occurred during the trial period was detected by prostate-specific antigen screening (2). Based on data on the prostate cancers diagnosed by April 1999, we estimate that only ∼10% or fewer of the prostate cancer cases were initially detected through an elevated prostate-specific antigen screen.

At baseline, study subjects completed a general risk factor, smoking, and medical history questionnaire (33), along with a food frequency questionnaire, including both portion size and frequency of consumption for 203 food items and 73 mixed dishes (34). This instrument was intended to measure usual consumption over the previous 12 months. Nutrient intake was estimated using a national food composition database available from the National Public Health Institute of Finland. In addition, food chemistry analyses were conducted to determine the content of the eight component tocopherols and tocotrienols in Finnish foods (35). The energy-adjusted correlation coefficient for vitamin E (corrected for attenuation) was 0.76 in a validation study comparing the food frequency questionnaire with food records (34). Height and weight were measured at baseline.

Fasting serum samples were collected at the prerandomization baseline visit and stored at −70°C until assayed. Serum α-tocopherol and β-carotene were determined on all cohort subjects by high-performance liquid chromatography as described (36). The between-batch coefficient of variation for serum α-tocopherol was 2.2%. Serum cholesterol was determined enzymatically (cholesterol oxidase/4-amino-phenazone method, (CHOD-PAP) Boehringer Mannheim).

The follow-up time for each participant was calculated from the date of randomization until the date of prostate cancer diagnosis, death, or April 30, 2004; 390,722 person-years of observation were accumulated. A total of 29,097 men, including 1,732 cases, who had baseline serum α-tocopherol and serum cholesterol values, were included in the serologic analyses. Only those with complete dietary information were included in the dietary analyses (n = 27,111; 1,641 cases). Baseline descriptive statistics are presented as means (continuous variables) or proportions (categorical variables), with significance tests calculated using the general linear models procedure in SAS. Cox proportional hazard models were used to determine relative risks (RR) and 95% confidence intervals (95% CI) for the association between prostate cancer and serum α-tocopherol as well as the individual dietary tocopherols and tocotrienols. The risk proportionality assumption, as tested with cross-product terms of follow-up time and the vitamin E exposure of interest, was met for all nutrients. Dietary intakes were log transformed and adjusted for energy intake using the residual method (37). Total α-tocopherol intake was the sum of α-tocopherol intake plus the contribution of vitamin E from multivitamin supplements. All exposure variables were divided into quintiles, based on the distribution of the entire cohort, and entered into the models as indicator variables with the lowest quintile as the referent category. Tests for linear trend were obtained by assigning to each nutrient quintile the median level and treating this as a continuous variable. All models were adjusted for age at randomization (continuous) and trial intervention group (separate terms for α-tocopherol and β-carotene arms). The serum α-tocopherol model was additionally adjusted for serum cholesterol, whereas the dietary vitamin E models were adjusted for total energy intake. The following factors were considered as potential confounders and entered simultaneously into a multivariate model: height, weight, body mass index, smoking (daily cigarettes and smoke-years), vitamin supplement use (any/none), physical activity (light-to-moderate work activity and ≥moderate leisure activity), urban residence, education, marital status, history of benign prostatic hyperplasia, serum β-carotene, serum high-density lipoprotein, and, for the dietary models, intakes of protein, fat, polyunsaturated fatty acids, cholesterol, vitamin C, and lycopene. A backward stepwise procedure that removed one factor at a time, beginning with those having the least significant χ² P value, was then used to arrive at a parsimonious model. If removal of the factor did not change the β-coefficients (i.e., <10% change), it was not retained in the model. This procedure was repeated for all potential confounders with χ² P values ≥0.05 until the only factors remaining were ones that altered the β-coefficients or ones with χ² P values of <0.05. If any factor was retained in one model, it was retained in all models (with the exception of the dietary factors that were not confounders in the serum model and whose inclusion would have reduced the number of subjects to those 27,111 with dietary data). Family history of prostate cancer was considered separately, as the data were available only on 64% of the cohort, and was not a confounder. Effect modification was assessed in stratified analyses and was statistically evaluated by including the cross-product term of the nutrient variables (quintiles) by the effect modifier (split at the median value of the cohort). For analyses by disease stage, cases were divided into nonadvanced and advanced (defined previously), and the two stratified analyses were both conducted using all noncases, censoring all subjects at April 1999. For a time period analysis, follow-up was divided into three periods consisting of the trial supplementation period (randomization date to April 30, 1993) and two postintervention periods (early postintervention, May 1, 1993-April 30, 1999; late postintervention, May 1, 1999-April 30, 2004). Person-years were calculated from the beginning of each time period to date of prostate cancer diagnosis, death, or end of each time period. Only those cases diagnosed during a specified time period were considered cases for that time period. All statistical analyses were done using SAS software version 8.02 (SAS Institute, Inc.).

Baseline characteristics of the cohort by quintile of serum α-tocopherol are shown in Table 1. Men with higher serum α-tocopherol tended to be younger, heavier, and of higher body mass index and tended to smoke fewer cigarettes for fewer years than men with lower concentrations. They also were more likely to report a history of benign prostatic hyperplasia, of being physically active, and of taking a multivitamin containing vitamin E and to be more educated, married, and live in an urban area. Not surprisingly, higher serum α-tocopherol was also related to higher intakes of the tocopherols and tocotrienols, as well as higher intakes of total fat, polyunsaturated fat, lycopene, and vitamin C, and higher concentrations of serum β-carotene and cholesterol. These patterns were similar when adjusted for age and serum cholesterol. It should be noted that in contrast to patterns in the United States, daily α-tocopherol intake in this cohort of Finnish men (mean, 10.4 mg) exceeded γ-tocopherol intake (mean, 8.2 mg). Mean serum α-tocopherol was 11.9 mg/L and the Spearman correlation coefficient for serum and dietary α-tocopherol was 0.24 (P < 0.0001).

Prostate cancer risk was not related to α-tocopherol intake, with or without the contribution of vitamin E from multivitamin supplements (Table 2). Risk was also unrelated to intake of other dietary tocopherols and tocotrienols. RRs for the component tocopherol and tocotrienol intakes did not vary by trial supplementation group, disease stage, or follow-up period (data not shown), with the exception of a significantly lower risk of advanced prostate cancer for γ-tocotrienol (RR, 0.50; 95% CI, 0.30-0.84 for highest versus lowest quintile; P_trend = 0.03).

Higher baseline serum concentrations of α-tocopherol were related to significantly lower prostate cancer risk (Table 3). Based on a priori hypotheses and previous observations about effect modification of the vitamin E-prostate cancer association, we conducted analyses stratified by several factors (Table 3). The inverse serum α-tocopherol-prostate cancer association was somewhat stronger among older men, and those who were supplemented with either α-tocopherol or β-carotene during the trial, had lower vitamin C intake, or smoked fewer cigarettes daily, but the interaction tests were not statistically significant. A substantial and statistically significant inverse relationship between serum α-tocopherol and prostate cancer was observed for advanced disease (RR, 0.56; 95% CI, 0.36-0.85 for highest versus lowest quintile; P_trend = 0.002). This association seemed to be stronger both in the α-tocopherol trial supplementation arm (RR, 0.36; 95% CI, 0.19-0.71 for highest versus lowest quintile; P_trend = 0.001) compared with those not supplemented with α-tocopherol (RR, 0.76; 95% CI, 0.43-1.33; P_trend = 0.20) and in the β-carotene trial supplementation arm (RR, 0.34; 95% CI, 0.18-0.63 for highest versus lowest quintile; P_trend < 0.0001) compared with those not supplemented with β-carotene (RR, 0.93; 95% CI, 0.51-1.72; P_trend = 0.78).

Serum α-tocopherol concentrations seemed unrelated to prostate cancer incidence during the trial supplementation period, but there was a strong inverse association during the first 6 years postintervention (Table 3). The latter finding was accentuated among those who during the trial received either the α-tocopherol supplement (RR, 0.55; 95% CI, 0.35-0.87 for highest versus lowest quintile; P_trend = 0.004) or the β-carotene supplement (RR, 0.52; 95% CI, 0.34-0.80; P_trend = 0.003). Only a weak, nonsignificant inverse association was suggested for the later postintervention period, with no modification by α-tocopherol or β-carotene supplementation during this, or the trial period. The lower prostate cancer risk for high serum α-tocopherol in the early post-trial period was also primarily evident for advanced disease (RR, 0.43; 95% CI, 0.25-0.73 for highest quintile; P_trend = 0.001) and not early disease (RR, 0.87; 95% CI, 0.60-1.25 for highest versus lowest quintile; P_trend = 0.52). One other finding of note is evidence of a possible stage shift during the trial period from advanced to nonadvanced cancers in the α-tocopherol supplemented study arm for the two highest quintiles of serum α-tocopherol: RRs for serum quintiles 4 and 5 (versus quintile 1) were 0.44 and 0.39 for advanced disease compared with 1.79 and 1.51 for nonadvanced disease (95% CIs and trend tests not formally significant, however).

We observed a significant inverse relationship between serum α-tocopherol and prostate cancer risk, which was particularly strong for advanced disease and cancers diagnosed in the early post-trial period. These associations were also stronger among those who were supplemented with either α-tocopherol or β-carotene during the trial. Intakes of component vitamin E tocopherols and tocotrienols were unrelated to risk.

Investigations of serum or plasma α-tocopherol and prostate cancer incidence or mortality provide some support for a protective association (7-12, 24, 25). Three nested case-control studies showed inverse associations for higher serum α-tocopherol concentrations with odds ratios ranging from 0.58 to 0.64 (refs. 7-9; ref. 9 for statistically significant trend). Similar to our finding, the Physicians' Health Study found a significant inverse association for plasma α-tocopherol (median concentration, 11.2 mg/L) among smokers with advanced disease (odds ratio, 0.51; 95% CI, 0.26-0.98; ref. 25), and a study in Switzerland found significantly higher prostate cancer mortality among smokers with low serum α-tocopherol concentrations (odds ratio, 3.26; 95% CI, 1.27-8.35; ref. 24). Two other studies (10, 11) found no associations, and a third showed a reduced risk for higher serum α-tocopherol concentrations that disappeared after adjustment for serum cholesterol (12). In the ATBC Study cohort, we observed previously a nonsignificant inverse association for serum α-tocopherol in the α-tocopherol supplemented arm (29), and our recent nested case-control study showed inverse associations for both serum α-tocopherol and γ-tocopherol with prostate cancer, with odds ratios of 0.49 (P_trend = 0.05) and 0.57 (P_trend = 0.08) for highest versus lowest tertiles, respectively, that were also stronger in the α-tocopherol supplemented arm (27). In the present investigation, which includes cases from our two previous analyses, we found a significant inverse relationship between serum α-tocopherol and prostate cancer risk, which was stronger among those who were supplemented with either α-tocopherol or β-carotene during the trial.

By contrast, only 3 (4-6) of 12 studies examining vitamin E intake and prostate cancer risk (4-6, 13-19, 22, 26) showed significant inverse associations, and we found previously that total vitamin E and γ-tocopherol intakes were significantly inversely associated with prostate cancer risk only among men who received the trial α-tocopherol supplement (29). In the current study, intakes of component vitamin E tocopherols and tocotrienols were unrelated to risk overall, or within subgroups of intervention, follow-up period, or disease stage, with the exception of a significant inverse association between γ-tocotrienol and advanced disease.

Of the seven observational studies that examined supplemental vitamin E (3, 19-23, 26), two [the Health Professionals Follow-up Study (23) and the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (26)] found no relationship overall but reported a reduced risk for advanced or fatal prostate cancer among current smokers and recent quitters who consumed supplemental vitamin E intake >100 IU/d (23) or 400 IU/d (26) or for greater than 10 years (26). The Cancer Prevention Study II from the American Cancer Society reported lower risk among current smokers who regularly consumed supplemental vitamin E (21) but did not find further reduced risks for advanced prostate cancer (based on only 19 cases, however).

In addition to the controlled trial component of the ATBC Study that showed lower prostate cancer incidence in response to α-tocopherol supplementation (2), the SU.VI.MAX trial found reduced incidence in the antioxidant supplementation arm that included 30 mg α-tocopherol, a finding that was significant among men with normal prostate-specific antigen at baseline (38). By contrast, the HOPE and Heart Protection studies found no effects of α-tocopherol supplementation (400 IU alone or 600 mg in combination with other antioxidants, respectively) on prostate cancer incidence (39, 40). Two other controlled trials in the United States are currently testing this hypothesis (41, 42).

The mechanisms underlying the effect of vitamin E on prostate cancer have not been determined. Oxidative stress has been implicated in prostate carcinogenesis (43), and vitamin E functions as a chain-breaking antioxidant nutrient that prevents propagation of free radical damage in biological membranes and plasma lipoproteins (44). Vitamin E may also affect carcinogenesis via interactions with other antioxidant nutrients, including carotenoids, vitamin C, and flavonoids (45, 46). Other vitamin E and α-tocopherol functions include inhibition of cell proliferation, cell adhesion, and protein kinase C activity; enhancement of immunity; and modulation of gene expression (46, 47). That α-tocopheryl succinate inhibits experimental prostate cancer cell line growth by decreasing expression of cell cycle regulatory proteins, prostate-specific antigens, and the androgen receptor (48, 49) provides a biologically plausible explanation for the observed lower incidence of advanced prostate cancer in men with higher serum α-tocopherol concentrations that supports a tumor growth-inhibitory influence of vitamin E in humans.

Given the underlying intervention and long follow-up period, we evaluated whether the relationship between baseline serum α-tocopherol and subsequent prostate cancer changed over time and found significantly reduced risk in the years immediately following the supplementation period for men with higher baseline serum levels. This suggests a cumulative, synergistic, or residual effect of the α-tocopherol intervention (50 mg daily for 5-8 years, which had resulted in 32% lower prostate cancer incidence and 40% lower incidence of advanced disease compared with the no α-tocopherol arm; ref. 2) on vitamin E status. Although serum α-tocopherol was not associated with incidence during the trial period overall, possibly because of overshadowing by the α-tocopherol supplementation, our data indicate that men in the higher quintiles of serum status in the α-tocopherol supplemented arm experienced a shift from advanced to nonadvanced disease incidence during supplementation that might represent more effective inhibition of tumor progression by the vitamin E in the setting of higher status or bioavailability. The fact that both α-tocopherol and β-carotene supplementation seemed to enhance the serum α-tocopherol protective association postintervention is intriguing and might reflect modulation of tocopherol metabolism and effectiveness by both of these agents (e.g., through altered regulation of specific cytochrome P450 or other enzymes; ref. 50).

The prospective design of this study, with long follow-up, is an important strength that minimized the possibility of disease or treatment effects on serum tocopherol concentrations and eliminated any potential recall bias of dietary intake. The study included only older male smokers who participated in the original prevention trial, limiting its generalizability to some degree. This is an important population group to study, however, because it is believed that smoking increases vitamin E requirements (51), and as detailed above, several studies have found reduced risks of advanced or fatal prostate cancer with higher supplemental vitamin E or serum α-tocopherol only among current smokers or recent quitters (23-26). We also observed that higher serum concentrations of α-tocopherol were associated with significantly reduced risks of advanced prostate cancer in our cohort. Chemical analyses of foods in the Finnish diet were comprehensively conducted to measure all eight vitamin E components (35) and thereby improve measurement of dietary vitamin E to a certain degree, although underreporting of dietary fat intake and difficulty in recalling or quantifying specific types of fats or oils consumed is a limitation (28). This fact, as well as the modest correlations between serum and dietary α-tocopherol, may explain why we and others have found associations with serum, but not dietary, vitamin E. As with all dietary assessment instruments, the potential for measurement error in a food frequency questionnaire can attenuate true associations, although energy adjustment may improve this situation (52).

In conclusion, whereas we found no association between vitamin E intake and prostate cancer risk in this cohort of smokers, higher serum α-tocopherol concentrations, prospectively determined, were significantly associated with reduced risks of prostate cancer, particularly advanced prostate cancer. The biological basis for these relationships, as well as better delineation of the “dose risk” continuum, should remain the focus of further research.