High Serum Selenium and Reduced Risk of Advanced Colorectal Adenoma in a Colorectal Cancer Early Detection Program

Colorectal cancer is the third most commonly diagnosed cancer in U.S. men and women, leading to >50,000 deaths annually (1, 2). Identifying modifiable risk factors is important for the prevention of this disease.

Interest in the essential trace element selenium as a colorectal cancer preventive agent was stimulated by the observation of a 61% reduction in colorectal cancer in the treatment arm of the Nutritional Prevention of Cancer Trial, a randomized placebo controlled trial for skin cancer prevention with selenium supplementation (200 μg selenium per day). This result is limited, however, in that colorectal cancer was not a primary end point of the trial, and the number of colorectal cancer cases was small (selenium arm = 8, placebo arm = 19).

Most observational studies investigating selenium exposure are small, because reliable data on selenium is largely restricted to selenium measures in biological tissues, due to large variations in the selenium content of foods resulting in imprecise dietary selenium intake estimates from questionnaires (3-5). The selenium levels of U.S. grains, dairy products, eggs, meat, poultry, and fish-all good sources of selenium-vary up to 10-fold, as a result of differences in the selenium concentration of the soil where the crops are grown (3-5), with high selenium soil content, for example, in Nebraska, the Dakotas, and Colorado, and lower content in the South and Northeast of the United States (6, 7). U.S. population intakes range between 50 and 250 μg/d (8-10), encompassing levels within the dosing level (200 μg/d) associated with colorectal cancer prevention in the Nutritional Prevention of Cancer Trial (11, 12).

A recent pooled analysis (13) of 1,012 cases of recurrent colorectal adenoma showed an inverse association between serum/plasma selenium and adenoma recurrence, with a preponderance of less advanced tumors. Pooled risk estimates indicate a stronger association for advanced adenoma; however, the power was limited (13). To examine associations between serum selenium and risk for advanced colorectal adenoma (i.e., tumors with a greater potential for malignant transformation), we studied more than 750 advanced colorectal adenoma cases and controls. Our interest was to explore interactions of selenium with tobacco use, because substantial evidence from animal and human studies, including randomized trials, suggests that smoking is a strong effect modifier for other antioxidative nutrients, including β-carotene and vitamin E (reviewed in ref. 14).

This case-control study was nested in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, in which colorectal adenoma were identified following a standard protocol in a large population characterized for colorectal cancer risk factors. Because the trial is conducted at 10 centers throughout the United States, it also captures the wide geographic differences in selenium intake in the U.S. population, which increase the power to detect an association (6, 7).

This case-control study was nested within the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, which was designed to evaluate selected methods for the early detection of these cancers and to investigate etiologic factors and early markers of cancer (15, 16). Participants in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, ages 55 to 74 years, were recruited at 10 centers in the United States (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA; Salt Lake City, UT; St Louis, MO; and Washington, DC). Participants in the screening arm of the trial received sigmoidoscopic exam at baseline. If the sigmoidoscopy identified polyps or other suspect lesions, participants were advised to get further follow-up examination through their own medical care providers, which usually resulted in a full colonoscopy with polypectomy or surgical procedures, if indicated. All medical and pathologic reports of the follow-up examinations were obtained and coded by trained medical record abstractors. Written informed consent was obtained from participants, and the trial received approval from the institutional review boards of the U.S. National Cancer Institute and the 10 study centers.

All cases and controls for this study were selected from 42,037 participants in the screening group who underwent a successful sigmoidoscopic examination at baseline (insertion to at least 50 cm with >90% of mucosa visible or a suspect lesion identified). Baseline sigmoidoscopy screening of selected cases and controls was conducted between September 1993 and September 1999. Participants provided information on risk factors and donated a blood sample for use in etiologic studies. After exclusion of 4,834 participants with a self-reported history of cancer (except basal cell skin cancer), ulcerative colitis, Crohn's disease, familial polyposis, colorectal polyps, or Gardner's syndrome, we randomly selected 772 cases for study from among 1,234 cases with advanced distal adenoma (adenoma ≥1 cm or containing villous elements or high-grade dysplasia). An equal number of sex- and race-matched participants (n = 777) with a negative screening sigmoidoscopy (i.e., no polyp or other suspect lesion; n = 26,651) were randomly selected as controls. Blood samples collected at study entry were available for selenium analysis from 759 cases and 767 controls.

Serum selenium concentrations were determined using inductively coupled plasma mass spectrometry (for details, see ref. 17). Equal numbers of cases and matched controls were analyzed in the same or consecutive batches. Quality control samples were monitored throughout the analysis. We excluded one case and one control due to extremely high serum selenium levels (>300 ng/mL). The within-batch repeatability coefficient of variation estimated from 182 duplicates was 5.5%, and the between-batch coefficient of variation estimated from 361 replicates was 7.4%.

At initial screening, all participants were asked to complete a questionnaire about sociodemographic factors; medical history, including current and former use of hormone replacement therapy; and risk factors for cancer. Usual dietary intake over the 12 months before enrollment was assessed with a 137-item food frequency questionnaire, including an additional 14 questions about intake of vitamin and mineral supplements (18). 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 (19) using the nutrient database from the U.S. Department of Agriculture (20).

Adjusted means (least-squares means) were calculated by linear models. To estimate the association between selenium and colorectal adenoma, we calculated odds ratios (OR) and 95% confidence intervals (95% CI) using unconditional logistic regression analysis, with selenium entered as a continuous variable or in quintiles, with cut points based on the selenium distribution in controls. We used the continuous variable to estimate for a linear trend. All Ps are two sided.

The ORs were adjusted for the matching factors, sex, and race, and in addition for age at randomization, study center, and year of blood draw. We evaluated confounding for known and potential risk factors of colorectal tumors, including physical activity, body mass index, smoking, alcohol intake, aspirin use, ibuprofen use, hormone replacement therapy (women only), family history of colorectal cancer, race, educational attainment, energy intake, red meat intake, folate intake, fiber intake, and calcium intake. None of the factors was included in the analyses because none of them changed the β coefficient of the risk estimates of selenium by >10%. Because the variation of the quality control samples in the last 16 batches was substantially larger than in the first 180 batches (SD = 24.5 and 8.4, respectively), we included an indicator variable for this batch effect [excluding participants analyzed in the last 16 batches (7.5% of all participants) did not change the association between selenium and adenoma risk].

Selenium levels are reported to be related to age, sex, smoking, and alcohol consumption (9, 10, 21, 22). Furthermore, other antioxidants, particularly vitamin E, may modify the effect of selenium (23, 24). To explore potential effect modification by age, sex, smoking, alcohol, and vitamin E, we did stratified analyses and evaluated multiplicative interaction by creating product terms, investigating the statistical significance of multiplicative interactions by comparison of the -2 log likelihood statistics of the main effect model with the joint effects model. For this analysis, we combined current smokers and those that quit smoking <10 years ago as recent smokers to avoid small cell numbers and to account for the relevant time period of advanced adenoma development. Former smokers were defined as those that quit smoking ≥10 years ago.

The study population was mainly non-Hispanic White (94%) and male (70%). Controls were slightly younger than the advanced colorectal adenoma cases (average, 61.8 and 63.1 years, respectively). Mean serum selenium levels of the first and fifth quintile ranged from 108 to 174 ng/mL (Table 1). Those with high serum selenium levels were less likely to be female and tended to exercise more.

Adjusted mean serum selenium levels were significantly lower in cases (134.2 ± 23.3 ng/mL, mean ± SD) than controls (137.3 ± 23.3 ng/mL, mean ± SD; P_difference = 0.007). Considering serum selenium as a continuous measure, risks for advanced adenoma were reduced by 26% for each 50 ng/mL increase in serum selenium (P_trend = 0.01), although the risk pattern was not clearly monotonic, when examined by quintile of selenium (Table 2). The inverse association was strongest in older participants (>67 years), showing a 52% reduction in advanced adenoma risk comparing those in the highest quintile with those in the lowest quintile (P_trend = 0.005; Table 2; P_interaction = 0.09). In addition, men in the highest quintile of selenium had 43% lower risk for advanced colorectal adenoma compared with men in the lowest quintile of selenium (P_trend 0.001; Table 2), whereas serum selenium was unrelated to advanced adenoma risk in women (P_interaction = 0.03). Within older participants (>67 years), we observed an inverse selenium-adenoma association for both men and women (data not shown), but numbers in these subgroups were small, resulting in wide confidence intervals.

Among recent smokers, defined as current smokers or those quit smoking <10 years ago, risks for colorectal adenoma were reduced 47% in the highest serum selenium tertile compared with the lowest tertile (P_trend = 0.008; Table 3; tertile grouping is used to avoid smaller stratum-specific numbers), with evidence for interaction of selenium and smoking in adenoma risk (P_interaction = 0.03). This inverse association was even stronger when restricted to recent smokers that smoked >10 cigarettes per day (3rd versus 1st tertile: OR, 0.34; 95% CI, 0.16-0.72; P_trend = 0.0009). Smoking itself was a risk factor for advanced adenoma in this study population with strongest positive associations for recent smokers (OR, 2.61; 95% CI, 1.97-3.53 for recent smokers versus nonsmokers). The observed inverse association between serum selenium and advanced adenoma risk in recent smokers was present in men (3rd versus 1st tertile: OR, 0.51; 95% CI, 0.24-1.08; P_trend 0.02), in women (3rd versus 1st tertile: OR, 0.46; 95% CI, 0.09-2.31; P_trend 0.31), and in younger participants (≤67 years old; 3rd versus 1st tertile: OR, 0.48; 95% CI, 0.23-1.00; P_trend 0.02), although cell numbers, particular in women and older participants, were small. Among older participants, selenium was inversely associated with advanced colorectal adenoma risk in those who were former smokers (3rd versus 1st tertile: OR, 0.22; 95% CI, 0.07-0.70; P_trend = 0.01). Alcohol and vitamin E intake did not modify the association between serum selenium and adenoma risk (data not shown).

We conducted a large study of 758 advanced colorectal adenoma cases and 767 sigmoidoscopy-negative controls, randomly selected from the same study population and screened for colorectal cancer with a standardized procedure. Overall, higher serum selenium levels were inversely associated with reduced risk of advanced colorectal adenoma. This inverse association between serum selenium and advanced adenoma risk was particularly strong in older persons and recent smokers.

In the Nutritional Prevention of Cancer Trial, 1,312 patients with a history of skin cancer were randomized to receive 200 μg selenium per day or placebo. Although the trial found no significant association between selenium supplementation and skin cancer recurrence (primary end point), significantly reduced risks were noted for overall cancer death and incidence of lung, prostate, and colorectal cancer (11). During the intervention period of up to 10 years (average, 4.5 years), 19 cases with colorectal cancer were diagnosed in the placebo group, whereas only eight cases were observed in the selenium group; a 61% risk reduction (95% CI, −10% to −83%; Table 4; ref. 11). Two additional years of follow-up with one more case in the placebo group add further support to this finding (25).

Three recent nested case-control studies of colorectal adenoma recurrence conducted within adenoma prevention trials also support a preventive role of selenium (13, 26). Studies, including 247 to 403 recurrent adenoma per study, showed 25% to 40% risk reductions for subjects with high serum selenium, and pooled analysis of all three studies resulted in a significant inverse association with a highly significant linear trend (Table 4; ref. 13). Most earlier studies of this topic have entailed relatively small numbers of cases, often with <100 cases (27-34). Several of these studies generally supported a role for selenium in colorectal cancer (27-30, 34-36) and adenoma prevention (27, 32, 33), although not all studies were statistically significant. Relatively few observational studies have reported either no association (29, 37), including the largest prospective study on colorectal cancer (37), or a nonsignificant positive association (31, 38). In agreement with our observations in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, several but mainly small observational studies and one randomized trial found inverse associations between selenium and colorectal adenoma and cancer risk. Findings from our study and the pooled analysis (13) particularly support inverse association with advanced adenoma.

The inverse selenium-adenoma association in our study was limited to men, as was overall cancer incidence reduction in the Nutritional Prevention of Cancer Trial (results for colorectal cancer were not provided separately; refs. 11, 25). One observational study (29) also reported an inverse association in men only and another including only women found a positive association (31), but two studies reported inverse associations in both men and women (13, 28). Effect modification by gender may be due in part to gender differentials in selenium excretion rates (25, 39, 40), or to interactions with smoking, as observed in our study. Given the small percentage of women in our study and the Nutritional Prevention of Cancer Trial (refs. 11, 25; 30 and 25%, respectively), these gender differentials could also be attributed to chance; larger studies with sufficient numbers of women are needed to address this question (The Selenium and Vitamin E Cancer Prevention Trial, the largest ongoing randomized trial on selenium supplementation, includes men only as prostate cancer is the primary end point; ref. 41).

A large body of experimental data, including animal models for colon cancer (42-45), supports a role for selenium in cancer prevention (see for review, ref. 46), potentially acting through multiple pathways (47-49). Selenium, particularly methylated forms, directly affects cell cycle control and apoptosis (43, 50-52). In cell lines, selenomethionine activates p53, with related increases in p53-dependent DNA repair (53). Selenium may also interact with the folate/homocysteine pathway, potentially altering DNA methylation patterns (54-56), as an early step in colorectal cancer development (57).

Indirect effects of selenium in colorectal cancer prevention may be mediated through selenium-containing enzymes (such as glutathione peroxidase and selenoprotein P), particularly due to their antioxidative and anti-inflammatory properties (47, 58-61) in relation to food-derived oxidative stress (62, 63) and colorectal inflammation (64, 65).

The gastrointestinal tract expresses all four common glutathione peroxidases (GPX1, GPX2, GPX3, and GPX4), suggesting their importance in colonic function (62, 66). Targeted disruption of both cytosolic and gastrointestinal GPX genes (GPX1 and GPX2, respectively) results in high susceptibility to inflammation and colon cancer in mice (49, 67). Oxidative stress, related to aging and smoking (both are risk factors for colorectal adenoma in this study population), could account for the strong inverse selenium-adenoma associations found in our study among older participants and recent smokers, possibly mediated through the antioxidative properties of selenoenzymes (68, 69). Our study is the first large study to explore interactions of selenium with smoking in colorectal tumors, and it will be interesting if future studies result in similar strong evidence for selenium-smoking interactions, as seen for other antioxidative nutrients (i.e., β-carotene and vitamin E; reviewed in ref. 14).

Mean serum selenium levels in the lowest and highest quintile in our study were 108 and 174 ng/mL, respectively, comparable with increase in serum selenium due to long-term supplementation in the Nutritional Prevention of Cancer Trial (with supplements mean serum selenium increased from 114 to 190 ng/mL; ref. 11). We observed the lowest serum selenium levels in Alabama and the District of Colombia and the highest levels in the Midwest (Michigan, Wisconsin, and Minnesota); however, this pattern was only in part consistent with local soil selenium levels reported for the United States (6, 7), probably due to nationwide food distribution of foods high in selenium.

The wide range of selenium levels in our study and the relatively large sample size increased the power to detect significant selenium-adenoma associations, overall, and to conduct stratified analysis. Findings were unlikely affected by prior screening, as results were similar when stratified by conduct of previous endoscopy or fecal occult blood test (data not shown). Advantages of conducting the study in a screening trial were that cases and sigmoidoscopy-negative controls were selected from the same source population of largely asymptomatic individuals and were screened with a standardized procedure, avoiding many of the biases of studies carried out in adenoma cases identified due to symptoms or signs of colorectal cancer.

In summary, higher serum selenium levels were associated with reduced risk of advanced colorectal adenoma, which seems more prominent in recent smokers. Because oxidative stress is related to smoking, the observed interactions with selenium may be mediated through the antioxidative properties of selenoenzymes.