Serum selenium and risk of prostate cancer—a nested case-control study 1,2,3

, Abstract Background— Selenium is a potential chemopreventive agent against prostate cancer, whose chemoprotective effects are possibly mediated through the antioxidative properties of selenoenzymes. Interrelations with other antioxidative agents and oxidative stressors, such as smoking, are poorly understood. Objectives— The aims were to investigate the association between serum selenium and prostate cancer risk and to examine interactions with other antioxidants and tobacco use. Design— A nested case-control study was performed within the screening arm of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Serum selenium in prospectively collected samples was compared between 724 incident prostate cancer case subjects and 879 control subjects, frequency-matched for age, time since initial screen, and year of blood draw. The men were followed for up to 8 y. Results— Overall, serum selenium was not associated with prostate cancer risk ( P for trend = 0.70); however, higher serum selenium was associated with lower risks in men reporting a high (more than the median: 28.0 IU/d) vitamin E intake [odds ratio (OR) for the highest compared with the lowest quartile of selenium: 0.58; 95% CI: 0.37, 0.91; P for trend = 0.05; P for interaction = 0.01] and in multivitamin users (OR for highest compared with the lowest quartile of selenium: 0.61; 95% CI: 0.36, 1.04; P for trend = 0.06; P for interaction = 0.05). Furthermore, among smokers, high serum selenium concentrations were related to reduced prostate cancer risk (OR for the highest compared with the lowest quartile of selenium: 0.65; 95% CI: 0.44, 0.97; P for trend = 0.09; P for interaction Conclusion— Greater prediagnostic serum selenium concentrations were not associated with prostate cancer risk in this large cohort, although greater concentrations were associated with reduced prostate cancer risks in men who reported a high intake of vitamin E, in multivitamin users, and in smokers.


INTRODUCTION
Interest in selenium as a nutrient with potential preventive effects against prostate cancer was heightened in the mid-1990s, after reports from the Nutritional Prevention of Cancer Trial showed that men who received 200 μg selenium/d had a significantly reduced risk of this disease (1)(2)(3). This trial was conducted in areas of the southeast United States notable for low soil content of selenium. Lower risks were found only among participants with low baseline concentrations of serum selenium (1,3,4). Further epidemiologic evidence for the preventive role of selenium in selenium-poor populations comes from studies conducted in malnourished populations in Linxian, China (5), where combined intervention with selenium, vitamin E, and β -carotene was related to reduced incidence and mortality of gastric cancer and total cancer.
Men who were randomly assigned to the screening arm of the trial were offered prostate cancer screening by serum prostate-specific antigen (PSA) at entry and annually for 5 y and digital rectal examination (DRE) at entry and annually for 3 y. If the PSA test result was ≥ 4 ng/mL or the DRE was suspicious for prostate cancer, the men were referred to their medical care providers for prostate cancer diagnostic evaluation. In addition, follow-up for recent diagnosis of cancer was carried out by annual mailed questionnaires and through periodic searches of the National Death Index. All medical and pathologic records related to the diagnosis were obtained for the participants suspected of having prostate cancer by either screening examination or annual questionnaire. Furthermore, death certificates and supporting medical or pathologic records were collected. Data related to the diagnosis of prostate cancer were abstracted by trained medical record specialists. All trial participants are followed for incidence of cancer and all causes of mortality for ≥ 13 y from the randomization date. The screening arm participants were asked to provide a blood sample at each screening visit. The institutional review boards of the US National Cancer Institute and the 10 study centers approved the trial and the participants provided written informed consent.

Study population
Of the 38 352 men randomly assigned to the screening arm of the trial, we excluded men reporting a history of prostate cancer (other than nonmelanoma skin cancer), men whose first valid screen (PSA test or DRE) was after 1 October 2001 (the censor date for the present analysis), men who received a screening exam but for whom there was no subsequent contact, men who did not complete a baseline risk factor questionnaire, men with an ethnic or racial background other than non-Hispanic white or non-Hispanic black, men without a signed informed consent for etiologic studies on cancer, and men without any blood collections for etiological studies at any of the screening visits. After exclusions, the analytic cohort comprised 26 975 men. All men were followed from their initial valid prostate cancer screen (PSA, DRE, or both), to first occurrence of prostate cancer, loss-to-follow-up, death, or censor date (1 October 2001), whichever came first. Case subjects are men diagnosed with adenocarcinoma of the prostate. Staging procedures corresponded to the Tumor, Nodes, and Metastases stage of disease classification (40). Cases were defined as advanced prostate cancer if they were stages III or IV (regional or distant) or Gleason Score ≥ 7.
The eligible 26 975 men included 1320 prostate cancer cases. For the present study, we included non-Hispanic white prostate cancer cases diagnosed ≥ 1 y after baseline blood draw (n = 803). For comparison, we selected control subjects by incidence-density sampling (41) with a casecontrol ratio of 1:1.2, frequency-matched by age (5-y intervals), time since initial screening (1-y time windows), race, and year of blood draw (n = 949).

Serum selenium analysis
Serum selenium concentrations were determined by using an inductively coupled plasma mass spectrometry method [for details, see Stürup et al (42)]. Serum for selenium analysis was available for 724 (90.2%) cases and 879 (92.6%) controls. Cases and their matched controls were analyzed in the same batch to minimize interassay variability. Blinded quality control samples (15%) were randomly inserted within each batch and monitored throughout the analysis. The CV, estimated from 181 blinded duplicates, was 9.4%.

Assessment of questionnaire-based covariates
At enrollment, all participants were asked to complete a questionnaire to obtain information on age, ethnicity, education, occupation, current and past smoking behavior, history of cancer and other diseases, use of selected drugs, recent history of screening exams, and prostate related health factors. Usual dietary intake over the 12 mo before enrollment was assessed with a 137item food-frequency questionnaire, which included 14 additional questions about intake of vitamin and mineral supplements and 10 additional questions on meat cooking practice (43). Daily dietary nutrient intake was calculated by multiplying the daily frequency of each consumed food item by the nutrient value of the sex-specific portion size (44) with the use of the nutrient database from the US Department of Agriculture (45). Total vitamin and mineral intake was calculated by adding dietary and supplemental intake. Multivitamin (and mineral) users were defined as men taking a one-a-day type vitamin, therapeutic type vitamin, highdose type vitamin, stresstabs, or B-complex in the last 2 y before enrollment (yes or no). Within a subset of controls, the partial Spearman correlation between intake of β-carotene, lycopene, and α -tocopherol and serum concentrations was 0.44, 0.31, and 0.58, respectively (coefficients were adjusted for months of blood draw, serum cholesterol concentrations, smoking, body mass index (BMI), age, and energy intake).

Statistical analysis
Adjusted means (least-squares means) were calculated by linear models. We used conditional logistic regression models to estimate odds ratios (ORs) of prostate cancer. Serum selenium was modeled as quartiles based on the distribution among the controls. We used the continuous variable to estimate for linear trend. All P values are two-sided. The analyses were conditioned on the matching factors (age, time since initial screening, and year of blood draw) and adjusted for study center. We evaluated confounding due to potential risk factors for prostate cancer, including average numbers of prostate cancer screening, family history of prostate cancer, educational attainment, physical activity, BMI, aspirin and ibuprofen use, diabetes, alcohol, smoking, energy, fat, tomatoes, fruit and vegetable intake, dairy products, red meat, heterocyclic amines from meat, vitamin E, β -carotene, lycopene, and calcium. None of the factors changed the β coefficient of the risk estimates of selenium by > 10%, and, therefore, none of these factors were included in the analyses. To explore potential effect modification by smoking, reported intake of antioxidants, and multivitamin use, we performed stratified analyses and evaluated the statistical significance of multiplicative interactions by comparison of the − 2 log likelihood statistics of the main effect model for selenium with that of the joint effects model, including cross-product terms. All analyses were carried out with SAS version 9.1 (SAS Institute, Cary, NC).

RESULTS
The average age of controls was 65 y and did not vary significantly by quartile of serum selenium. Reported intake of β -carotene, lycopene, and vitamin E tended to be higher in men with high selenium concentrations than in men with low selenium concentrations, whereas BMI, energy, and red meat and alcohol intake was lower in men with high serum selenium concentrations (Table 1). Other baseline study characteristics were not significantly different across quartiles of serum selenium. Compliance with the PLCO screening protocol also did not vary significantly by selenium concentrations and was very high because the average number of screens per year was close to 1-the goal for the screening intervention. The average serum selenium concentrations in the study population (controls) was 141.3 ng/mL, with mean serum selenium concentrations significantly higher in areas with high soil selenium content (146.8 ng/mL) than in those with intermediate and low soil selenium content areas (136.8 ng/ mL; P for mean difference < 0.0001; Table 2). Median serum selenium concentrations of the 4th quartile (170.4 ng/mL) was 50% higher than the 1st quartile (113.7 ng/mL; Table 3).
Serum selenium was not associated significantly with prostate cancer incidence overall: men in the highest quartile had a non-significant 16% reduction in prostate cancer risk compared with men in the lowest quartile of serum selenium, and there was no suggestion of a linear trend (P for trend = 0.70; Table 3). Similarly, no significant association with serum selenium was observed for advanced prostate cancer (stage III and IV OR in a comparison of the highest with the lowest quartile: 0.62; P for trend = 0.57). When stratified by study areas with high and intermediate or low soil selenium content, serum selenium was not significantly associated with prostate cancer in either group (study regions with high soil selenium content: OR for the highest compared with the lowest quartile of selenium: 0.68; 95% CI: 0.42, 1.09; P for trend = 0.42; study regions with intermediate or low soil selenium content: OR for the highest compared with the lowest quartile of selenium: 0.96; 95% CI: 0.63, 1.47; P for trend = 0.82).
The association between serum selenium and incident prostate cancer did not differ significantly by total reported intakes of vitamin C, β-carotene, or lycopene (Table 4); however, an inverse association between serum selenium and prostate cancer (OR for the highest compared with the lowest quartile: 0.58; 95% CI: 0.37, 0.91; P for trend = 0.05) was observed in men who reported a high intake of total vitamin E (equal to or more than the median, which was 28.0 IU/d-a dose similar to the one used in the Alpha-tocopherol, Beta-carotene Trial), showing a significant interaction between vitamin E and selenium (P for interaction = 0.01). High serum selenium was also nonsignificantly associated with a lower risk of prostate cancer in men taking multivitamins (OR for the highest compared with the lowest quartile of selenium: 0.61; 95% CI: 0.36, 1.04; P for trend = 0.06), but not in nonusers (P for interaction = 0.05). Because most vitamin E supplementation was in the form of multivitamins, we could not separate the effects of these 2 vitamin sources.
An analysis stratified by smoking status is shown in Table 5. We observed an inverse association between selenium and prostate cancer in smokers (OR for the highest compared with the lowest quartile of selenium: 0.65; 95% CI: 0.44, 0.97; P for trend = 0.09), and selenium-related risks tended to increase nonsignificantly in men who never smoked (P for interaction = 0.007).

DISCUSSION
In this nested case-control study, which included 724 incidence prostate cancer cases and 879 controls, we observed no overall association between prediagnostic selenium concentrations and prostate cancer. However, greater serum selenium concentrations were associated with lower risks of this disease in men who reported a high vitamin E intake, in multivitamin users, and in smokers.
The strongest support for a chemopreventive effect of selenium in human prostate carcinogenesis comes from the Nutritional Prevention of Cancer Trial, a randomized study to evaluate selenium supplementation (200 μg/d) and skin cancer prevention, which found, as secondary endpoints, reduced risks of total cancer mortality (50%) and prostate cancer incidence (52% reduced risk; average intervention period 6.4 y, with 64 prostate cancer cases) (1,3). In a second trial (SU.VI.MAX), no overall association was found with selenium supplementation; however, among men with a normal baseline PSA (< 3 ng/mL), the risk of prostate cancer was 48% lower in the group of selenium-treated men than in the group of placebo-treated men (47). The result of this study cannot be attributed directly to selenium (dose: 100 μg/d), because 5 other antioxidative vitamins and minerals were given simultaneously as a multivitamin supplement.
Because the enzyme activity of some selenoenzymes, such as the glutathione peroxidases, tend to plateau at high serum selenium concentrations (58,59), selenium supplementation may be most effective in populations with low selenium exposure. The Nutritional Prevention of Cancer Trial, which was conducted specifically in areas with low selenium intake, supported this hypothesis showing the strongest inverse associations with prostate cancer in men with low baseline serum selenium concentrations (1st tertile: < 106 ng/mL, and 2nd tertile: 106-121 ng/mL) and no association in men with high baseline concentrations (3rd tertile: > 121 ng/ mL) (4). However, the inverse selenium-prostate cancer associations observed in epidemiologic studies do not appear to be limited to settings with low mean serum selenium concentrations (Figure 1 and Figure 2), and, from our study, the strongest inverse associations were noted in areas with high soil selenium content. Furthermore, it is unknown how these circulating concentrations translate to the prostate, which also expresses selenoenzymes not found in the circulating system, eg, selenoprotein 15 (60)(61)(62). In addition, selenium may also prevent prostate cancer directly through active selenium metabolites, in particular methylated forms; however, such effects, as shown in experimental studies, are achieved only at supranutritive doses (16,20,21).
Our results suggest a synergistic relation between selenium and vitamin E, showing little evidence that one antioxidative nutrient can replace the other in prostate cancer prevention. Our finding is consistent with 2 observational studies (12,13) and a nutritional intervention trial conducted in Linxian China (5), although not all observational studies (49,51,55,56) found such interactions. However, our study lacks specificity on this point because most vitamin E supplementation was in the form of multivitamins, making it difficult to separate the effects of vitamin E from those of other multivitamin constituents. Because the Selenium and Vitamin E Cancer Prevention Trial, one of the largest ongoing intervention trials, makes a multivitamin without vitamin E and selenium, available to trial participants who prefer to continue using multivitamin while participating in the trial, this trial will be able to further explore interaction between selenium, vitamin E, and multivitamins. The trial is expected to be completed in 2013 (7).
We also observed a strong inverse association between serum selenium and prostate cancer risk in smokers. Although smoking itself was not associated with prostate cancer risk in the present study (data not shown) and several other studies (63), it is noteworthy that another antioxidant, vitamin E, is associated with reduced risk of this disease primarily in smokers, as seen in this cohort (64) and most other studies (65)(66)(67)(68)(69)(70). Additional exploration of a 3-way interaction between vitamin E, smoking, and selenium was beyond the scope of the present study, because the numbers of cases and controls in these subgroups were small. Effect modification of the selenium-prostate cancer association by smoking was found in 3 observational studies (49,50,56), but not in another (51) and not in an investigation of prostate cancer risk in smokers and asbestos workers (53). Smoking results in increased exposure to radical oxidative species (35)(36)(37)(38), and selenium inhibits the damaging effect of oxidative species on DNA and other biomolecules. The protective role of selenium in smokers could also be enhanced by the presence of oxidative-response elements in the promoter regions of genes encoding for selenoenzymes, such as GPX1 (71), and their increased transcription related to exposure to oxidative stressors (72)(73)(74).
Given an increasing nationwide distribution of foods, we were somewhat surprised to observe statistically significant differences in serum selenium concentrations by regional soil selenium content. However, this difference was not accounted for by regional differences in dietary pattern [eg, differences in the level of consumption of foods high in selenium, such as grains, eggs, meat, and fish (75)], because adjustment for these and other foods did not significantly change the results (data not shown).
A limitation of our study was the relatively short follow-up, to a maximum of 8 y. To avoid potential effects of disease on selenium concentrations, we only included cases diagnosed ≥ 1 y after blood draw; excluding cases diagnosed within the first 2 y showed similar results (data not shown). We only measured selenium at a single point in time, and multiple measurements ideally over the entire period of cancer development would have reduced the possibility of attenuated risk estimates due to random error. Stratified analysis by antioxidative nutrients was based on questionnaire data, which may introduce measurement error. Correlations of serum selenium with BMI and intakes of alcohol, red meat, vitamin E, β-carotene, lycopene, and energy suggest that combined lifestyle factors may contribute to prostate cancer prevention and that observational studies such as ours only incompletely control for unmeasured confounding. Clinical trials with selenium as an intervention could address this.
The present study was large (Figures 1 and 2), and the men studied had a broad range of serum selenium concentrations [almost as wide as the intervention effect in the Nutritional Prevention of Cancer Trial, in which mean serum selenium concentrations rose from 114 ng/mL at baseline to 190 ng/mL at the end of the intervention (1)]. By restricting our analysis to men randomly assigned to the screening arm of the trial, disease detection bias was limited. Compliance with the PLCO protocol for prostate cancer screening was very high and similar across quartiles of selenium. Our large sample size ensured sufficient power to observe ORs of ≤ 0.68 in comparisons of the 4th with the 1st quartile, similar to the summary OR of a recent metaanalysis (OR: 0.72) (76) and within the range of expected associations (Figure 1).
In conclusion, overall we observed no inverse association between prediagnostic serum selenium concentrations and the risk of prostate cancer in this large cohort, which was followed up by standardized screening procedures. However, higher serum selenium may be associated with lower prostate cancer risk in men who report a high intake of vitamin E, in multivitamin users, and in smokers. Clinical trial and observational studies on selenium and the risk of prostate cancer listed by mean or median selenium concentrations. ORs and 95% CIs were obtained in a comparison of the highest with the lowest quantile of selenium. ◆, randomized clinical trials with selenium concentrations obtained from serum or plasma; ▲, nested case-control study (CCS) and case cohort study with selenium concentrations obtained from serum or plasma; •, population-based CCS with selenium concentrations obtained from serum or plasma; △, nested CCS and case cohort study with selenium obtained from toenails; ○, population-based CCS with selenium obtained from toenails. Observational studies were excluded if they used a questionnaire-based assessment of selenium intake, included patients with benign prostatic hyperplasia as controls, or included < 15 prostate cancer cases. Observational studies on selenium and risk of advanced prostate cancer (1) listed by mean or median selenium concentrations. ORs and 95% CIs were obtained in a comparison of the highest with the lowest quantile of seleniuim. ▲, nested case-control study (CCS) and case cohort study with selenium concentrations obtained from serum or plasma; •, population-based CCS with selenium concentrations obtained from serum or plasma; △, nested CCS and case cohort study with selenium concentrations obtained from toenails; ○, population-based CCS with selenium concentrations obtained from toenails. Observational studies were excluded if they used a questionnaire-based assessment of selenium intake, included patients with benign prostatic hyperplasia as controls, or included < 15 prostate cancer cases. Advanced cancer was defined as a tumor that extends through the prostatic capsule or as a metastatic disease (stage C or D, stage III or IV, or regional or distant stage); the exception was Vogt et al (55), who defined advanced cancer as regional or distant stage, Gleason Score ≥ 7, or both.   2 Conducted by using serum selenium as a continuous variable.
3 Low and high are defined as below or above the median, respectively. The median for vitamin C was 241.7 mg/d, that for vitamin E was 28.0 IU/d, that for β-carotene was 4817.8 μg/d, and that for lycopene was 9693.6 μg/d. Vitamin C, vitamin E, and β-carotene are based on dietary and supplemental intakes, and lycopene is based on dietary intake from a 137-item food-frequency questionnaire that included 14 additional questions on intakes of vitamin and mineral supplements (43).
4 Obtained from a multivariate conditional logistic regression analysis adjusted for age, time since initial screening, year of blood draw, and study center.
5 For stratified analyses with no significant interaction, the P for trend for the combined analysis per