Antioxidant intake and pancreatic cancer risk
The Vitamins and Lifestyle (VITAL) Study
Abstract
BACKGROUND:
Oxidative stress causes damage to many components of human cells (ie, proteins, lipids, and DNA) and is involved in carcinogenesis. Nutrients with antioxidant properties may protect against oxidative stress. In this study, the authors examined the intake of antioxidants from diet and supplements in relation to pancreatic cancer risk among participants of the Vitamins and Lifestyle (VITAL) Study.
METHODS:
The participants included 77,446 men and women ages 50 to 76 years who were residents of western Washington State and who completed a baseline questionnaire between 2000 and 2002. Participants reported usual diet over the past year and use of supplements over the past 10 years in addition to demographic and lifestyle factors. During a median follow-up of 7.1 years, 184 participants developed pancreatic adenocarcinoma. Cox proportional hazards models were used to estimate multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for 7 antioxidants: β-carotene, lutein plus zeaxanthin, lycopene, vitamin C, vitamin E, selenium, and zinc.
RESULTS:
An inverse association was observed between dietary selenium and the risk of pancreatic cancer (medium vs low intake: HR, 0.58; 95% CI, 0.35-0.94; high vs low intake: HR, 0.44; 95% CI, 0.23-0.85; Ptrend = .01); however, when supplemental and dietary exposures were combined, the association was no longer statistically significant.
CONCLUSIONS:
Dietary selenium intake was inversely associated with the risk of pancreatic cancer, and the observed association was attenuated by selenium supplementation. Cancer 2013. © 2012 American Cancer Society.
INTRODUCTION
Despite discouraging results from randomized trials,1-3 there remains an interest in antioxidant nutrients and cancer risk. It has been observed that antioxidants reduce oxidative DNA damage and genetic mutations,4-6 and they may protect against pancreatic carcinogenesis. Epidemiologic data have produced inconsistent results regarding the relation between antioxidant intake and pancreatic cancer risk. Most case-control studies7-15 have observed an inverse association between dietary intake of vitamin C, vitamin E, or β-carotene; whereas other case-control studies16-19 and all prospective studies20-22 have reported null results. One cohort study20 and 2 meta-analyses of clinical trials23, 24 have examined antioxidant supplementation and pancreatic cancer, and all have reported null associations.
We examined the risk of pancreatic cancer associated with intake of 7 antioxidant nutrients, including β-carotene, lutein plus zeaxanthin, lycopene, vitamin C, vitamin E, selenium, and zinc, from dietary and supplemental sources in a large, population-based, prospective study, the Vitamins and Lifestyle (VITAL) cohort. We hypothesized that intake of antioxidants would be inversely related to incidence of pancreatic cancer.
MATERIALS AND METHODS
Study Population
Participants were members of the VITAL cohort, a population-based, prospective study that was designed specifically to investigate the association between vitamin, mineral, and other dietary supplements and cancer risk. Details of the study design, recruitment and study implementation have been reported previously.25 Briefly, men and women ages 50 to 76 years at baseline who lived in the 13-county region in western Washington State covered by the Surveillance, Epidemiology, and End Results (SEER) cancer registry were eligible to participate. Between October 2000 and December 2002, baseline questionnaires, followed by postcard reminders 2 weeks later, were mailed to 364,418 individuals based on a commercial mailing list. Among these, 77,719 reminders (21.3%) were returned and deemed eligible. The study was approved by the Institutional Review Boards of the University of North Carolina at Chapel Hill and the Fred Hutchinson Cancer Research Center (Seattle, Wash).
Exposure Assessment
Diet
Diet was assessed by using a semiquantitative food frequency questionnaire that captured the consumption frequency and portion size of 120 foods and beverage items over the last year and included adjustment questions on types of foods and preparation techniques.25 Average dietary nutrient intake per day was estimated by the food frequency questionnaire analytic program based on nutrient values from the Minnesota Nutrient Data System.26
Supplement use
Respondents were queried about multivitamin use and their intakes of 16 vitamins and minerals from all other mixtures and single supplements during the 10-year period before baseline, including duration in years, frequency in days/weeks, and dose. The 10-year average supplemental nutrient intake per day was computed as (dose per day) × (days per week/7) × (years/10), summed over individual supplements and micronutrient dose in participant-reported multivitamins. The VITAL supplement questionnaire exhibited excellent reliability in a 3-month test-retest reliability substudy of 220 randomly selected participants with intraclass correlation coefficients ranging from 0.69 to 0.87.27
Total nutrition intake
Total nutrient intake per day was calculated by combining the data on supplement use and dietary intake. Conversion factors were used for β-carotene28 and vitamin E29 to account for different chemical forms or differences in absorption. The total nutrient intake was not calculated for lutein plus zeaxanthin or lycopene, because supplement use information was available only in pills daily, whereas the dietary intake was available in mcg daily.
Covariates
Participants also reported on personal characteristics as part of the baseline questionnaire, including age, sex, ethnicity, education, height, weight, recreational physical activity, cigarette smoking, alcohol consumption, family history of cancer, and medical history. From data on height and weight, we calculated body mass index (BMI) (in kg/m2). Average total metabolic equivalent (MET) hours per week over the past 10 years were calculated using the years, frequency, and published energy expenditure for different activities. Cigarette smoking status was categorized as never, former (quit >10 years), recent (quit ≥10 years), and current.
Case Ascertainment and Follow-Up
Cohort members were followed for incidence of pancreatic cancer from enrollment to December 31, 2008; the median follow-up was 7.1 years. Incident pancreatic cancer was ascertained by linking the study cohort to the western Washington SEER cancer registry. All incident cancer cases except nonmelanoma skin cancer diagnosed within the 13-county area of western Washington State are reported to SEER. We ascertained 195 incident pancreatic cancer cases, including 184 adenocarcinoma (International Classification of Diseases for Oncology, 3rd edition [ICD-O-3] codes C25.0-C25.3 or C25.7-C25.9) and 11 neuroendocrine tumors (ICD-O-3 code C25.4). The remaining participants were right-censored from the analysis at the earliest date of the following events: withdrawal (n = 22), emigration from SEER catchment area (n = 4216), death (n = 5234), or December 31, 2008 (n = 67,790).
Exclusions
For the current study, participants were excluded if they reported a positive (n = 49) or missing (n = 213) history of pancreatic cancer at baseline. Eleven neuroendocrine tumors were also excluded, leaving 77,446 participants in the study. In addition, participants were excluded from the dietary analysis if they did not complete all pages of the food frequency section (at least 5 items per page), if their energy intake was <800 kcal for men or <600 kcal for women, or if their energy intake was >5000 kcal for men and >4000 kcal for women (n = 7132), leaving 70,332 participants in the dietary analysis. Participants were excluded from the supplement analysis if they did not provide use information of that supplement. Participants were excluded from the analysis of combining dietary intake and supplement use if either was missing.
Statistical Analysis
Cox proportional hazards models were used to estimate crude and adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) associated with antioxidant intake. For dietary intake and total intake, the exposure was categorized into tertiles. For supplement use, the intake levels were categorized as none, low (below median among users), and high (above median among users). P values for trend (Ptrend) were calculated by using the continuous variables with excluding values above the 98th percentile for each exposure.
We selected a priori potential confounders for the adjusted models: age (continuous), sex, ethnicity (white, nonwhite), education (less than high school graduate, some college, college or advanced degree), total energy intake (tertiles), BMI (<25 kg/m2, 25-30 kg/m2, ≥30 kg/m2), recreational physical activity (tertiles of MET for all recreational activities), cigarette smoking status (never, former, recent, current), total alcohol consumption (tertiles of average daily alcohol intake), family history of pancreatic cancer (yes/no), and use of medication for diabetes (yes/no). Adding history of pancreatitis and pack-years of smoking to the models did not materially change the results; thus, these 2 variables were not included in the final models. Total energy intake was dropped from the supplement-use models because it did not change the estimates. Missing values for covariates (9% with 1 or more covariates missing) were imputed by chained equations in IVEware software (version 0.2; University of Michigan, Ann Arbor, Mich). All statistical analyses were performed using the SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC).
RESULTS
The cohort of participants included in this analysis included 52% women, and 93% of participants were white. The average age of the participants was 62 years at baseline (Table 1).
Characteristic | Frequency: No. of Participants (%)a |
---|---|
Age: Mean ± SD, y | 61.99 ± 7.45 |
BMI: Mean ± SD, kg/m2 | 27.41 ± 5.17 |
Physical activity: Mean ± SD MET, h/wk | 10.88 ± 13.89 |
Alcohol: Mean ± SD, g/d | 8.17 ± 15.45 |
Total energy intake: Mean ± SD, kcal/d | 1855.12 ± 773.95 |
Sex | |
Women | 40,225 (51.94) |
Men | 37,221 (48.06) |
Race | |
White | 72,220 (93.25) |
Others | 5226 (6.75) |
Education | |
≤High school graduate | 15,377 (20.2) |
Some college | 29,161 (38.31) |
College or advanced degree | 31,587 (41.49) |
Smoking status | |
No | 36,459 (47.46) |
Long-term quitters, >10 y | 28,148 (36.64) |
Current and short-term quitters, ≤10 y | 12,216 (15.9) |
Family history of pancreatic cancer | |
Yes | 2834 (3.71) |
No | 73,640 (96.29) |
History of pancreatitis | |
Yes | 650 (0.84) |
No | 76,777 (99.16) |
History of diabetes | |
Yes | 5411 (6.99) |
No | 72,034 (93.01) |
- Abbreviations: BMI, body mass index; MET, metabolic equivalent for all activities; SD, standard deviation.
- a Not all totals are all equal to 77,446 because of missing values.
The data listed in Table 2 indicate the associations between average dietary antioxidant consumption and pancreatic cancer risk. After adjustment for all a priori confounders, we observed a statistically significant, inverse association between dietary selenium intake and the risk of pancreatic cancer (medium vs low intake: HR, 0.58; 95% CI, 0.35-0.94; high vs low intake: HR, 0.44; 95% CI, 0.23-0.85; Ptrend = 01). The crude HR and adjusted HR were very close except for lycopene, vitamin C, and selenium. Stepwise examination indicated that the difference for lycopene was mainly because of an adjustment for age (age was negatively correlated with dietary lycopene intake; r = −0.05; P < .0001); the difference for vitamin C was mainly because of adjustment for smoking (current smokers and recent quitters had a significantly lower intake of vitamin C [mean, 104.3 mg daily] than never-smokers [mean, 125.5 mg daily] and long-term quitters [mean, 125.5 mg daily]; P < .0001); and the difference for the trace element selenium was mainly because of adjustment for total energy intake (total energy intake was highly correlated with dietary selenium intake; r = 0.88; P < .0001).
Antioxidant Intake Level | No. of Participants | No. of Incident Pancreatic Cancers | Crude HR | PTrend | Adjusted HRa | PTrend |
---|---|---|---|---|---|---|
β-Carotene, mcg | ||||||
14.26-2564.48 | 23,440 | 66 | 1.00 | 1.00 | ||
2564.66-4753.24 | 23,441 | 54 | 0.81 (0.57-1.17) | 0.81 (0.56-1.18) | ||
4753.66-59,237.96 | 23,441 | 42 | 0.63 (0.43-0.93)b | .13 | 0.65 (0.42-0.99)b | .29 |
Lutein plus zeaxanthin, mcg | ||||||
13.48-1717.84 | 23,440 | 62 | 1.00 | 1.00 | ||
1717.84-3125.33 | 23,441 | 56 | 0.90 (0.63-1.29) | 0.92 (0.63-1.35) | ||
3125.43-100,884.19 | 23,441 | 44 | 0.71 (0.48-1.04) | .54 | 0.74 (0.48-1.14) | .84 |
Lycopene, mcg | ||||||
0-3768.61 | 23,440 | 67 | 1.00 | 1.00 | ||
3768.66-7222.54 | 23,441 | 47 | 0.70 (0.48-1.02) | 0.79 (0.54-1.17) | ||
7222.74-218,764.82 | 23,441 | 48 | 0.72 (0.50-1.04) | .07 | 0.82 (0.53-1.26) | .25 |
Vitamin C, mg | ||||||
0.57-76.64 | 23,440 | 62 | 1.00 | 1.00 | ||
76.64-135.01 | 23,441 | 51 | 0.82 (0.56-1.18) | 0.87 (0.59-1.28) | ||
135.01-1854.97 | 23,441 | 49 | 0.79 (0.54-1.15) | .17 | 0.89 (0.58-1.35) | .51 |
Vitamin E, IU | ||||||
0.97-9.24 | 23,440 | 69 | 1.00 | 1.00 | ||
9.24-14.93 | 23,441 | 44 | 0.64 (0.44-0.93)b | 0.65 (0.42-1.01) | ||
14.93-303.22 | 23,441 | 49 | 0.72 (0.50-1.03) | .05 | 0.67 (0.40-1.12) | .13 |
Selenium, mcg | ||||||
6.38-85.49 | 23,440 | 69 | 1.00 | 1.00 | ||
85.49-127.50 | 23,441 | 46 | 0.67 (0.46-0.97)b | 0.58 (0.35-0.94)b | ||
127.50-641.60 | 23,441 | 47 | 0.68 (0.47-0.99)b | .04b | 0.44 (0.23-0.85)b | .01b |
Zinc, mg | ||||||
1.45-8.76 | 23,440 | 61 | 1.00 | 1.00 | ||
8.76-13.14 | 23,441 | 48 | 0.79 (0.54-1.15) | 0.90 (0.56-1.45) | ||
13.14-67.94 | 23,441 | 53 | 0.87 (0.60-1.26) | .57 | 0.94 (0.52-1.71) | .98 |
- Abbreviations: HR, hazard ratio.
- a Models were adjusted for age, gender, ethnicity, education, body mass index, physical activity, cigarette smoking status, total alcohol consumption, family history of pancreatic cancer, history of diabetes and total energy intake.
- b These values are statistically significant.
Data in Table 3 indicate the associations between 10-year antioxidant exposure from supplement sources and pancreatic cancer risk. In the current study, most supplemental nutrient intake was from multivitamins except for vitamin C and vitamin E; therefore, supplement use results for vitamin C, vitamin E, and multivitamins are presented in the table. We observed no associations between the use of any individual supplement or multivitamin supplements and pancreatic cancer. For supplemental selenium (mainly from multivitamins), the adjusted HRs were 0.99 (95% CI, 0.70-1.41) for low intake (<20 mcg daily) and 0.73 (95% CI, 0.51-1.06) for high intake (≥20 mcg daily) compared with nonusers (Ptrend = .60).
Antioxidant Intake Level | No. of Participants | No. of Incident Pancreatic Cancers | Crude HR | PTrend | Adjusted HRa | PTrend |
---|---|---|---|---|---|---|
Vitamin C, mg | ||||||
Nonuser | 20,651 | 53 | 1.00 | 1.00 | ||
2.57-148.57 | 28,080 | 67 | 0.93 (0.65-1.34) | 0.93 (0.65-1.34) | ||
149.00-1750.00 | 28,083 | 59 | 0.82 (0.56-1.18) | .53 | 0.82 (0.56-1.19) | .44 |
Vitamin E, mg | ||||||
Nonuser | 20,194 | 53 | 1.00 | 1.00 | ||
1.29-100.86 | 28,329 | 67 | 0.90 (0.63-1.29) | 0.92 (0.64-1.32) | ||
101.00-1000.00 | 28,328 | 64 | 0.86 (0.60-1.24) | .80 | 0.80 (0.55-1.17) | .37 |
Multivitamin, pills | ||||||
Nonuser | 26,732 | 70 | 1.00 | 1.00 | ||
0.04-0.70 | 23,934 | 54 | 0.86 (0.60-1.23) | 0.95 (0.66-1.36) | ||
0.79-1.00 | 26,770 | 60 | 0.86 (0.61-1.21) | .34 | 0.81 (0.57-1.15) | .14 |
- Abbreviations: HR, hazard ratio.
- a Models were adjusted for age, sex, ethnicity, education, body mass index, physical activity, cigarette smoking status, total alcohol consumption, family history of pancreatic cancer and history, of diabetes.
When dietary and supplement use of antioxidants was combined (Table 4), a significant, inverse association was observed among those who consumed a medium level of selenium (HR, 0.59; 95% CI, 0.37-0.93) but not among those in the highest tertile of selenium intake (HR, 0.69; 95% CI, 0.39-1.20) (Table 4). Trend tests were suggestive of a possible inverse association between total β-carotene intake and pancreatic cancer risk (Ptrend = .03), but the HR did not reach statistical significance for either medium or high level of intake.
Antioxidant Intake Level | No. of Participants | No. of Incident Pancreatic Cancers | Crude HR | PTrend | Adjusted HRa | PTrend |
---|---|---|---|---|---|---|
β-Carotene, mcg | ||||||
37.05-4451.73 | 23,172 | 60 | 1.00 | 1.00 | ||
4451.99-8513.77 | 23,173 | 53 | 0.88 (0.61-1.27) | 0.86 (0.59-1.26) | ||
8513.82-109,006.02 | 23,172 | 42 | 0.70 (0.47-1.04) | .02b | 0.69 (0.46-1.04) | .03b |
Vitamin C, mg | ||||||
0.57-137.20 | 23,261 | 61 | 1.00 | 1.00 | ||
137.20-347.25 | 23,262 | 47 | 0.77 (0.53-1.12) | 0.81 (0.55-1.20) | ||
347.29-2629.64 | 23,262 | 49 | 0.80 (0.55-1.16) | .44 | 0.82 (0.56-1.21) | .46 |
Vitamin E, mg | ||||||
0.85-18.35 | 23,281 | 55 | 1.00 | 1.00 | ||
18.35-70.45 | 23,281 | 56 | 1.01 (0.70-1.47) | 1.01 (0.69-1.47) | ||
70.46-524.95 | 23,281 | 51 | 0.93 (0.63-1.36) | .74 | 0.84 (0.57-1.24) | .34 |
Selenium, mcg | ||||||
9.81-98.76 | 23,314 | 66 | 1.00 | 1.00 | ||
98.76-145.65 | 23,314 | 40 | 0.61 (0.41-0.90)b | 0.59 (0.37-0.93)b | ||
145.66-646.60 | 23,314 | 52 | 0.79 (0.55-1.14) | .09 | 0.69 (0.39-1.20) | .06 |
Zinc, mg | ||||||
1.73-13.75 | 23,273 | 55 | 1.00 | 1.00 | ||
13.75-23.10 | 23,273 | 56 | 1.02 (0.71-1.48) | 1.04 (0.71-1.53) | ||
23.10-158.28 | 23,273 | 46 | 0.84 (0.57-1.24) | .31 | 0.81 (0.53-1.24) | .24 |
- Abbreviations: HR, hazard ratio.
- a Models were adjusted for age, sex, ethnicity, education, body mass index, physical activity, cigarette smoking status, total alcohol consumption, family history of pancreatic cancer, history of diabetes, and total energy intake.
- b These values are statistically significant.
DISCUSSION
In this cohort study, we investigated the intake of 7 antioxidants from dietary and supplement sources in relation to pancreatic cancer risk. Our results provided evidence that dietary selenium intake is inversely associated with risk of pancreatic cancer. We did not observe strong evidence of an association with the intake of other antioxidants.
Epidemiological studies have indicated an inverse relation between selenium intake and the incidence of certain cancers, such as colorectal cancer,30 bladder cancer,31 lung cancer,32 and prostate cancer,33 although not all results were consistent.34 A cancer prevention trial in Finland22 assessed baseline dietary selenium intake with a dietary history questionnaire and reported no association with pancreatic cancer risk. A recent study of the European Prospective Investigation of Cancer (EPIC) cohort in the United Kingdom, in which dietary selenium was assessed using 7-day food diaries at baseline, indicated that high intake of selenium was associated with a reduced risk of pancreatic cancer.35 Moreover, 2 studies have identified an inverse association between biomarkers of selenium and pancreatic cancer risk, including a small, nested, case-control study with 22 cases and 44 controls using serum36 and a recent case-control study with 118 cases and 399 controls using toenails.37 In line with these biomarker studies, our results indicated that selenium intake from diet was inversely associated with pancreatic cancer, even after adjustment for several potential confounding variables. We also observed that selenium supplementation did not appear to provide an additional benefit beyond the effect observed for dietary intake of selenium alone, because the analysis of selenium supplement use (intake mostly from multivitamin supplements) produced null results, and the association between selenium intake and pancreatic cancer was substantially attenuated and became statistically nonsignificant when supplemental selenium was added to dietary selenium as the exposure.
Selenium is a trace element essential to human health. It plays an important role in thyroid hormone metabolism, antioxidant defense systems, and immune function.38 Selenium has several anticarcinogenic mechanisms, including inactivating oxygen-free radicals, initiating DNA repair, and inducing apoptosis.39 However, animal and in vitro studies have also demonstrated that selenium promotes malignant cell transformation and protects tumor cells from stress-induced apoptosis.40 Furthermore, a randomized trial of supplemental selenium indicated that selenium supplementation was linked to an increased risk of type 2 diabetes mellitus,41 which is a risk factor for pancreatic cancer. Our results suggest that there may be an optimal range of selenium level that maximizes its anticarcinogenicity or that reduces carcinogenicity. Further investigation will be necessary to confirm this finding and to determine the optimal dose. A study in British adults indicated that higher selenium status was associated with an adverse blood lipid profile,42 and another study among US adults indicated that increasing serum selenium levels were associated with decreased mortality up to 130 ng/mL and raised the concern that higher serum selenium levels beyond that may be associated with increased mortality.43 Along with the previous studies, our results suggest that selenium supplementation may not be beneficial, especially for individuals who already have high selenium status.
Four other prospective studies have investigated dietary intake of other antioxidants (vitamin C, vitamin E, β-carotene, lycopene) in relation to pancreatic cancer, and all of them reported null or weak findings.20-22, 35 In our current study, we also observed null findings for these antioxidants, with 1 exception: we observed a decreased risk with the highest intake of β-carotene. The discrepancy with previous studies may be because of different populations, or our finding could be due to chance.
This study has several strengths. The VITAL cohort is a prospective study that was designed specifically to investigate supplement use with cancer risk. Supplement users were targeted at recruitment to increase power, because the recruitment letter described this as a study on supplement use and cancer risk, but the study was not restricted to supplement users. Information on supplementation was collected for the 10 years before baseline, providing long-term intake. We collected extensive information on cancer risk factors, and we were able to carefully control for potential confounding effects.
This study also has several limitations. Although our detailed supplement assessment yielded very good validity and reliability results,27 recall error from the self-reported food frequency questionnaire and supplement use was inevitable. Moreover, dietary selenium in food varies, depending on where the food is grown (eg the selenium level in soil varies from 0.10 parts per million to 1.31 parts per million among counties in Washington44), and this information was not included in determining the selenium content of foods; therefore, these nondifferential measurement errors may have led to some attenuation of the results. Another limitation is that we examined a relatively large number of dietary exposures in this work, increasing the probability of a spurious result. Finally, although we adjusted for education level and race in our final model, there still may be residual confounding from socioeconomic status. However, we do not believe this is a big concern, because the current literature reveals only a weak or null association between socioeconomic status and pancreatic cancer risk.45-48 In our analysis, further adjustment for household income did not change the conclusions.
In summary, in the current study, we did not observe an association between the use of antioxidant supplements and the incidence of pancreatic cancer. Our data suggest that dietary selenium is associated with a reduced risk of pancreatic cancer; however, the findings need replication in other populations.
FUNDING SOURCES
This work was supported by a grant from the National Cancer Institute (R03CA139261). Dr. Han was supported by the Marilyn Gentry Fellowship from American Institute for Cancer Research; Dr. White and Dr. Brasky were partially supported by grants K05CA154337 and R25CA94880; and Dr. Gammon was partially supported by grant P30ES010126.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.