|
Calcium+Vitamin D Reduces Risk for Colorectal Cancer
|
|
|
Calcium and Vitamin D Intake and Risk of Colorectal Cancer: The Multiethnic Cohort Study
American Journal of Epidemiology April 1, 2007
Song-Yi Park1, Suzanne P. Murphy1, Lynne R. Wilkens1, Abraham M. Y. Nomura1, Brian E. Henderson2 and Laurence N. Kolonel1
1 Cancer Epidemiology Program, Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI
2 Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA
The associations of intakes of calcium and vitamin D with colorectal cancer risk were examined in the Multiethnic Cohort Study (Hawaii and Los Angeles, California). In 1993-1996, 85,903 men and 105,108 women aged >/=45 years completed a quantitative food frequency questionnaire. A total of 2,110 incident cases of colorectal cancer (1,138 in men and 972 in women) were identified through December 31, 2001. Cox proportional hazards models were used to calculate multivariate-adjusted relative risks and 95% confidence intervals. Total calcium intake (from foods and supplements) was inversely associated with colorectal cancer risk in both men (highest quintile vs. lowest: relative risk (RR) = 0.70, 95% confidence interval (CI): 0.52, 0.93; p for trend = 0.006) and women (RR = 0.64, 95% CI: 0.50, 0.83; p for trend = 0.003). The inverse association was also seen for total vitamin D intake in men (RR = 0.72, 95% CI: 0.51, 1.00; p for trend = 0.03) but not in women. Intake of dairy products was inversely associated with colorectal cancer risk, especially among nonusers of supplemental calcium (men: RR = 0.77, 95% CI: 0.59, 1.01; women: RR = 0.66, 95% CI: 0.49, 0.89). The findings support the hypothesis of protective roles for calcium, vitamin D, and dairy products in the risk of colorectal cancer.
INTRODUCTION
Colorectal cancer is one of the most commonly diagnosed cancers worldwide, and dietary factors are considered to be important in its risk (1, 2). A possible protective effect of calcium and vitamin D against colorectal carcinogenesis has been suggested from results of in vivo and in vitro studies (3-7). A number of mechanisms have been proposed to explain the roles of calcium and vitamin D in reducing colorectal cancer risk, including binding of long-chain fatty acids and bile acid in the small intestine, thereby protecting colonic epithelial cells from mutagens, as well as effects on cell proliferation and differentiation, apoptosis, angiogenesis, and cell-cycle regulation (4, 8-14).
Analytic epidemiologic studies have extensively examined the relation of these nutrients and their main food sources, dairy products, to colorectal cancer risk (15-17). Randomized clinical trials have suggested that use of calcium supplements may reduce the risk of recurrent colorectal adenomas (18-22), although the dose-response relation could not be examined in these studies because of the fixed doses of calcium used in trials. In a recent pooled analysis of 10 cohort studies, Cho et al. (23) found an inverse relation between calcium and milk consumption and colorectal cancer risk, while an older meta-analysis did not support an inverse association (15). The results from studies finding a reduction in risk associated with intakes of calcium, vitamin D, and/or dairy products have been mostly weak or statistically nonsignificant (15-17, 24).
Methodological limitations may be one reason for the lack of association of calcium and vitamin D with colorectal cancer risk in some studies (25). Dietary supplements, including multivitamin products, are important contributors to intake of these nutrients (26). However, only a few prospective studies have examined intake from both foods and dietary supplements in relation to colorectal cancer risk (25, 27-29). In this investigation, we examined the associations between calcium and vitamin D intakes from foods and dietary supplements and colorectal cancer risk in a large multiethnic cohort.
RESULTS
Among the Multiethnic Cohort Study participants, mean daily intakes of total calcium, adjusted for age and ethnicity, were 960 mg (411 mg/1,000 kcal) in men and 1,066 mg (567 mg/1,000 kcal) in women. For total vitamin D, mean values were 335 IU (156 IU/1,000 kcal) for men and 340 IU (192 IU/1,000 kcal) for women. Thirty-seven percent of men and 41 percent of women reported taking multivitamin supplements, and 9 percent of men and 27 percent of women reported using calcium supplements.
Table 1 presents baseline characteristics of the cohort according to total calcium and vitamin D intakes. Participants with higher intakes of calcium and vitamin D tended to be older and, in general, more health-conscious, being more likely to be nonsmokers, to be physically active (women), to use nonsteroidal antiinflammatory drugs, multivitamins, and hormone replacement therapy (women only), and to consume more dairy products and less alcohol, than those with lower intakes of both nutrients. There was no apparent association of calcium and vitamin D intakes with history of intestinal polyps or with family history of colorectal cancer. Intakes of the nutrients were not associated with body mass index in men but were inversely associated in women. Intakes of total and saturated fat were not closely associated with calcium and vitamin D intakes, while energy intake was inversely associated.
There were inverse associations between calcium intake from foods, supplements, or both and colorectal cancer risk in both men and women, except for calcium from foods adjusted for supplement use in women (table 2). However, calcium intake from foods was inversely associated with risk in men and women who did not take multivitamins or calcium supplements. The multivariate-adjusted relative risk for the highest quintile of total calcium intake was 0.70 (95 percent confidence interval (CI): 0.52, 0.93; p for trend = 0.006) in men and 0.64 (95 percent CI: 0.50, 0.83; p for trend = 0.003) in women.
Total vitamin D intake was inversely associated with colorectal cancer risk in men but not in women. The multivariate relative risk for the highest quintile was 0.72 (95 percent CI: 0.51, 1.00; p for trend = 0.03) in men and 0.89 (95 percent CI: 0.63, 1.27; p for trend = 0.80) in women (table 3). Women in the highest quintile of vitamin D from foods showed a significantly lowered risk (relative risk (RR) = 0.78, 95 percent CI: 0.63, 0.96) after adjustment for supplement intake, but the trend was not statistically significant (p = 0.12). The inverse association for vitamin D from foods was more obvious in women who did not take supplements (RR = 0.69, 95 percent CI: 0.52, 0.93; p for trend = 0.03).
We examined the association of colorectal cancer risk with intake of dairy products, the main food sources of calcium and vitamin D. There was an inverse association between dairy product and milk intakes and the risks, which were somewhat stronger in men (table 4). After exclusion of multivitamin or calcium supplement users who obtained calcium or vitamin D from nonfood sources, the inverse associations with dairy products became stronger than those for overall participants, particularly in women.
We examined risk according to calcium supplement use and levels of calcium intake from foods. Calcium supplement use appeared to have further benefit among male and female participants with lower calcium intake (table 5). However, among participants with a high intake of calcium from foods, calcium supplement use did not lower colorectal cancer risk for either men or women.
In ethnicity-specific analyses, inverse associations between total calcium and vitamin D intakes and colorectal cancer risk were seen consistently across groups (men and women combined) (table 6). The relative risks for the highest quartiles ranged from 0.44 to 0.79 for calcium and from 0.56 to 0.98 for vitamin D, although the trends for calcium were statistically significant in the two ethnic groups (Japanese Americans and Whites) with the largest number of cases, because of limited statistical power.
Because calcium absorption is affected by vitamin D, we attempted to separate the effects of the two nutrients. However, they were highly correlated, with Spearman correlation coefficients of 0.71 in men and 0.70 in women (p < 0.001). When both nutrients were included in multivariate models together, the association with calcium remained, while that for vitamin D was diminished. Among men, the relative risk for the highest quintile of total calcium intake was attenuated from 0.70 to 0.75, and the corresponding relative risk for total vitamin D intake was also attenuated from 0.72 to 0.86. For women, the relative risk for the highest quintile of calcium intake decreased slightly from 0.64 to 0.62, although the corresponding relative risk for vitamin D increased from 0.89 to 1.20.
TABLE 6. Multivariate-adjusted relative risk of colorectal cancer according to quartile of total intakes of calcium and vitamin D, by ethnic group, Multiethnic Cohort Study, 1993-2001
![Quartile-1.gif](../images/040207/Quartile-1.gif)
We investigated whether use of nonsteroidal antiinflammatory drugs affected the associations of calcium and vitamin D intake with colorectal cancer risk. The inverse associations were somewhat stronger in nonusers or short-term (<2 years) users than in long-term (2 years) users, but interactions between nonsteroidal antiinflammatory drug use and intakes of calcium/vitamin D were not statistically significant in either men or women (data not shown).
We repeated the analyses of total calcium and vitamin D intake and colorectal cancer risk after excluding all of the cases (n = 524) diagnosed within 2 years of cohort entry, because of the possibility that preclinical disease symptoms might have altered participants' dietary intake. Relative risk estimates for calcium and vitamin D intake were similar to those shown in tables 2 and 3 (data not shown).
DISCUSSION
In this large cohort study, we found an inverse association of total calcium intake with colorectal cancer risk in both men and women. The observed reduction in risk among participants with the highest intake range was 30 percent in men and 36 percent in women. Supplemental calcium intake was inversely associated with risk in both men and women. Calcium intake from foods was also associated with a lower risk in both men and women, although the relation in women was masked unless those who did not use supplemental calcium were excluded. Intakes of total vitamin D and dairy products were also related to a lower risk in men and in women who were supplemental calcium nonusers.
The relation between calcium intake and colorectal cancer risk was reported in recent cohort studies investigating intake from foods and supplements (25, 27-29). The results of these studies were inconsistent, however. The Nurses' Health Study and the Health Professionals Follow-up Study found an inverse association of total calcium intake with colon cancer in men (>1,250 mg/day vs. =500 mg/day: RR = 0.64, 95 percent CI: 0.43, 0.95) but not in women (RR = 0.94, 95 percent CI: 0.66, 1.33) (28). In Cancer Prevention Study II, a modest inverse association of total calcium intake with risk was found in men (highest quintile vs. lowest: RR = 0.82, 95 percent CI: 0.58, 1.16; p for trend = 0.04) but not in women (27). An inverse association between total calcium intake and colorectal cancer risk was reported in the Cohort of Swedish Men (highest quartile vs. lowest: RR = 0.68, 95 percent CI: 0.51, 0.91; p for trend = 0.01) (29), whereas no association was found in the Women's Health Study (25). Investigators from the former two studies reporting an inverse association of calcium intake with colon/colorectal cancer risk (27, 28) suggested that moderate doses of calcium may reduce risk but higher doses may be of no further benefit. However, the findings from our study population, which had a range of calcium intakes similar to that of those two studies, suggested a continued dose-response effect on colorectal cancer risk at higher doses of calcium. Findings from other prospective studies considering calcium intake from food sources only were inconsistent (36-40), whereas the results from case-control studies (41-45) more consistently supported an inverse association. A recent meta-analysis (46) based on three randomized controlled trials (18, 21, 47) of supplemental calcium intake and recurrence of colorectal adenomas, which are precursors for colorectal cancer, suggested a preventive role of calcium supplementation. However, a randomized clinical trial from the Women's Health Initiative found no effect of calcium plus vitamin D supplementation on colorectal cancer incidence among postmenopausal women (48).
Our finding that total vitamin D intake was significantly associated with a lower risk of colorectal cancer in men but not in women was similar to that from the Cancer Prevention Study II Nutrition Cohort (27). However, in our study, vitamin D intake from foods was significantly associated with risk among women who did not take multivitamins or calcium supplements. Two other cohort studies in women also found no association or only weak inverse associations of total vitamin D intake with risk (25, 49). It is not clear why risk reduction in women was not as apparent as in men. In part, it is probably due to higher misclassification of vitamin D intake in women than in men. Calcium supplementation is much more common in women, and these mineral tablets often include vitamin D; unfortunately, we did not have information on the addition of vitamin D to calcium supplements. Sex hormone metabolism may modify or mask associations between vitamin D status and colorectal cancer risk (27), since the vitamin D receptor has been linked with the estrogen receptor system (50, 51). Several cohort studies failed to find an association of vitamin D intake from only food sources with colorectal cancer risk (36, 38-40), because foods are only one source of vitamin D; the balance is provided by supplements and the synthesis of vitamin D in the skin in response to solar ultraviolet B radiation (17).
A pooled analysis of 10 cohort studies (23) and a review (24) found a protective effect of dairy product or milk consumption against colorectal cancer, although the results from recent cohort studies are not consistent (25, 27, 36, 38, 39). In our study, dairy product consumption and milk consumption were inversely related to colorectal cancer risk in men and in men and women who did not use multivitamins or calcium supplements. A possible protective effect of dairy products could be related to various constituents of these products in addition to calcium and vitamin D (through fortification), including conjugated linoleic acid, sphingolipids, butyric acid, and fermentation products (24). Although dairy product consumption was significantly correlated with both total calcium intake and total vitamin D intake (Spearman correlation coefficients were 0.61 and 0.40, respectively; p < 0.001), the inverse association of dairy products or milk with colorectal cancer risk appears to be less strong than that of total calcium or vitamin D intake. A possible reason is that milk fats, particularly saturated fats, might increase risk (24). However, in our study, adjustment for total fat or saturated fat intake did not affect the associations between dairy product and milk intake and colorectal cancer risk.
This study had several limitations. As we noted above, it is difficult to accurately measure vitamin D exposure in humans (17). We did not have information on sunshine exposure at baseline. In addition, we only considered multivitamins as a supplemental source of vitamin D, because use of single vitamin D supplements was not common in our population. However, calcium supplement products often contain vitamin D. A higher proportion of women than of men reported using calcium supplements, and misclassification may be one reason why we did not observe the protective effect of total vitamin D intake in women. Another limitation of our study was the single assessment of intake at baseline. Dietary intake from both foods and supplements may have changed during the follow-up period. However, there did not appear to be a strong difference in results for cases diagnosed within the first 2 years of follow-up and those diagnosed later. Lack of information on colonoscopy or sigmoidoscopy was also a limitation, although we adjusted for history of intestinal polyps in all multivariate analyses.
In addition to the prospective design and the large number of participants of several different ethnic groups, a strength of our study was the availability of detailed information on food consumption as well as use of common vitamin and mineral supplements. Although measurement error in dietary intake could not be eliminated, we had shown in a previous calibration study that adjustment for energy intake using nutrient densities improved the quality of the assessments (32). When we analyzed the data using calibration-adjusted values, the results were similar. Our brief questionnaire for quantifying intake of vitamin and mineral supplements was demonstrated to accurately capture data on supplement use, in comparison with three 24-hour recalls (33). A further strength was our ability to control for a wide range of potentially confounding factors for colorectal cancer in the analyses. The consistency of findings among ethnic groups added to the strength of the study.
In conclusion, our findings support the hypothesis of a protective role for calcium in colorectal cancer risk in both men and women, as well as a protective role for intakes of vitamin D, milk, and dairy products in men and in women who did not use supplemental calcium. Although previous cohort studies suggested that moderate calcium intake may reduce colorectal cancer risk, with no appreciable further effect from higher intakes (27, 28), our results suggest a continued dose-response relation.
MATERIALS AND METHODS
Study cohort
The Multiethnic Cohort Study was established to study the relation between diet and cancer in Hawaii and Los Angeles, California (30). It was designed to include adults aged 45-75 years from five targeted ethnic/racial groups: African Americans, Native Hawaiians, Japanese Americans, Latinos, and Whites. In 1993-1996, more than 215,000 participants completed a 26-page mailed questionnaire on diet, medical history, and lifestyle. The study was approved by the review boards of the University of Hawaii and the University of Southern California. For the analyses, participants who were not in one of the targeted ethnic groups (n = 13,994) or who reported implausible dietary intakes (n = 8,265) were excluded (31). We also excluded participants with previous colorectal cancer reported on the baseline questionnaire (n = 2,154) or identified from the cancer registries of Hawaii and California (n = 407). Therefore, the analyses included 191,011 participants.
Dietary assessment
Nutrient intakes from foods consumed at baseline were assessed by means of a quantitative food frequency questionnaire containing questions on more than 180 items. The questionnaire was developed from 3-day measured food records for approximately 60 men and women of each ethnic group (30). A calibration study was conducted and showed satisfactory correlations between the quantitative food frequency questionnaire and three repeated 24-hour recalls for all ethnicity-sex groups being studied (32). Correlations for nutrient densities ranged from 0.57 to 0.74 and were 0.67 for calcium and 0.61 for vitamin D from foods. Daily nutrient intakes from the quantitative food frequency questionnaire were calculated using the food composition table developed and maintained at the Cancer Research Center of Hawaii for use in the Multiethnic Cohort Study (30).
In addition to the quantitative food frequency questionnaire, the baseline questionnaire included questions about the use of eight vitamin and mineral supplements. Subjects were asked to indicate whether they had used any of these supplements at least weekly during the past year. Subjects were also asked about duration and frequency of use of each supplement. For this analysis, only regular supplement use, defined as use for more than 1 year, was considered. Subjects were asked to indicate the approximate dosage per tablet, with the following choices for calcium: =250 mg, 300-600 mg, 625-1,000 mg, >/=1,250 mg, and not known. Calcium supplement users who did not provide a dosage were assigned to the lowest category (=250 mg). Nutrient levels in multivitamins were estimated as composites of levels in the two brands of multivitamins most frequently reported in the dietary recalls of the calibration study (Centrum Silver and Centrum Hi Potency; Wyeth Consumer Healthcare, Madison, New Jersey) (33). Therefore, 181 mg of calcium and 400 IU of vitamin D were used as dosages obtained from one multivitamin tablet.
Total calcium and vitamin D intakes included contributions from foods and regularly used supplements. Supplemental calcium intake was based on the use of individual calcium supplements and multivitamins. Supplemental vitamin D intake was based on the use of multivitamin supplements.
Identification of colorectal cancer cases
Incident cases of colorectal cancer were identified by means of linkage to the Surveillance, Epidemiology, and End Results cancer registries covering Hawaii and California through December 31, 2001. Deaths were identified by linkage to death-certificate files in Hawaii and California and the National Death Index through December 31, 2001. The closure date was the earliest of the following: the date of diagnosis of colorectal cancer, the date of death, or December 31, 2001. Colorectal cancer cases were limited to participants diagnosed with invasive adenocarcinoma of the large bowel. A total of 2,110 incident cases (1,138 in men and 972 in women) were identified during an average follow-up period of 7.3 years.
Statistical analysis
Intakes of calcium, vitamin D, and dairy products were divided into quintiles based on the distribution of intakes among all participants. Selected baseline characteristics were adjusted for age and ethnicity via poststratification (34), weighted to the 10-year age and ethnic-group distribution of the entire cohort, and compared across quintiles of nutrient intakes.
We estimated the relative risks using Cox regression with age as the metric to model time to colorectal cancer incidence. All models were adjusted for two strata variables: ethnicity and time since cohort entry, categorized as =2 years, 2-5 years, and >5 years. Initially, analyses were performed with further adjustment for age at cohort entry (years). The multivariate models additionally included adjustment for the following factors assessed at baseline: pack-years of cigarette smoking, family history of colorectal cancer, physical activity (hours spent in vigorous work or sports per day), history of intestinal polyps, use of nonsteroidal antiinflammatory drugs, body mass index (weight (kg)/height (m)2), total energy intake, dietary fiber intake, and use of hormone replacement therapy (in women only). We included separate variables for intake from foods and intake from supplements in models together to examine the independence of their associations with colorectal cancer. For the models of total calcium, total vitamin D, and dairy product intakes, we additionally adjusted for regular multivitamin use as an indicator variable, to determine the role of the nutrient or related food beyond any influence of general multivitamin use. Trend tests were conducted by inclusion of a continuous variable in the model assigned the sex- and ethnicity-specific median values within the appropriate overall quantile.
We investigated whether the association of calcium intake from foods (< median, >/=median) with colorectal cancer risk differed between calcium supplement users and nonusers. Statistical interactions between calcium intake from foods and calcium supplement use were assessed by means of the likelihood ratio test, comparing a main-effect, no-interaction model with a fully parameterized model including all possible interaction terms for the variables of interest. In the ethnicity-specific analyses, we combined men and women and adjusted the results for sex as a stratum variable. This was justified because the associations were similar for men and women within each ethnic group.
We generally used daily nutrient and food intakes in terms of density (intake per 1,000 kcal) in the analyses. Stram et al. (32) and Kipnis et al. (35) previously found that energy-adjusted nutrient intakes are better recalled than absolute values. Measurement-error-corrected nutrient densities were computed as predicted values from models regressing the average intakes of three 24-hour recalls on the quantitative food frequency questionnaire intakes (32). Supplemental intakes of calcium and vitamin D were analyzed as absolute amounts, since it is unlikely that the energy coverage of the questionnaire would affect reporting of supplement use. However, the relative risks were similar for supplemental intake, regardless of whether it was parameterized as absolute amounts or as densities. Total intake was computed as intake from foods plus supplements, divided by calories per day, and multiplied by 1,000. All p values were two-sided.
|
|
|
|
|
|
|