Abstract
Epidemiologic evidence supporting a protective role of calcium and vitamin D in colorectal carcinogenesis has been accumulating in Western populations, but it is limited in Asian populations, whose intake of calcium is relatively low. We investigated the association of intakes of these nutrients with colorectal cancer risk in Japanese. Study subjects were participants of a large-scale case-control study in Fukuoka, Japan. Diet was assessed through interview regarding 148 dietary items by showing typical foods or dishes on the display of a personal computer. In a multivariate analysis adjusting for potential confounding variables, calcium intake was significantly, inversely associated with colorectal cancer risk (P for trend = 0.01); the odds ratio for the highest versus lowest quintile of calcium intake was 0.64 (95% confidence interval, 0.45-0.93). Higher levels of dietary vitamin D were significantly associated with decreased risk of colorectal cancer among those who had fewer chances of sunlight exposure at work or in leisure (P for trend = 0.02). A decreased risk of colorectal cancer associated with high calcium intake was observed among those who had higher levels of vitamin D intake or among those who had a greater chance of daily sunlight exposure, but not among those with medium or lower intake of vitamin D or among those with potentially decreased sunlight exposure. These results add to support for a joint action of calcium and vitamin D in the prevention of colorectal carcinogenesis. (Cancer Epidemiol Biomarkers Prev 2008;17(10):2800–7)
Introduction
Colorectal cancer is a major cause of cancer death worldwide (1). In Japan, the mortality of colorectal cancer has sharply increased after the Second World War (2) and it is among the highest levels in the world (3). Such a high incidence may not fully be explained by known risk factors of colorectal cancer, including physical inactivity, obesity, and high consumption of alcohol and processed meat (4, 5).
Experimental data (6, 7) clearly show that calcium has an anticarcinogenic effect in the colorectum. Although calcium has not been considered to have a significant effect on colorectal cancer risk in humans until the late 1990s (7), evidence has accumulated in recent years (8-17) that supports the protective role of calcium in colorectal carcinogenesis. Vitamin D may also decrease cancer risk through various mechanisms, including regulation of cellular proliferation and differentiation, induction of apoptosis, and inhibition of angiogenesis (6, 7). The preventive role of vitamin D in colorectal carcinogenesis in human was first proposed by Garland and Garland (18) on the basis of an observation of a higher incidence of colorectal cancer mortality in regions with low solar radiation levels. This finding has been replicated in several studies including recent articles from the United States (19) and Japan (20). Moreover, nested case-control studies that measured circulating vitamin D concentrations (21) and one cohort study that estimated overall vitamin D status (22) have consistently shown a protective association.
Although experimental evidence supports the protective roles of calcium and vitamin D in colorectal carcinogenesis and epidemiologic evidence regarding calcium has been fairly consistent, several issues remain to be addressed. First, epidemiologic evidence regarding the association between these nutrients and colorectal cancer are not entirely consistent even among recent studies; several cohort studies reported virtually no association with calcium (23, 24) or vitamin D (13, 16, 23, 24). Besides measurement errors in dietary assessment, the null finding for vitamin D intake may be ascribed to a much greater contribution of cutaneous synthesis of vitamin D induced by sunlight exposure or to low levels of vitamin D intakes that do not exert any anticarcinogenic effect among study subjects. Furthermore, as vitamin D controls the calcium metabolism (25), the effect of calcium in colorectal carcinogenesis may depend on vitamin D status. This hypothesis was supported by an analysis (26) showing that calcium supplementation reduced the risk of colorectal adenoma recurrence only among subjects with higher vitamin D levels. Finally, evidence is conflicting regarding these relations according to subsite in the colorectum (14, 15, 17, 27).
We therefore investigated whether these nutrient intakes are associated with the risk of colorectal cancer using data of a case-control study in Japanese men and women, a population consuming a diet high in vitamin D but low in calcium (28). Specific considerations were given to occupational and leisure-time physical activities as a surrogate of sunlight exposure, potential effect modification of calcium by vitamin D status, and difference in the association by subsite of the colorectum.
Materials and Methods
Study Subjects
Data were derived from the Fukuoka Colorectal Cancer Study, a large-scale case-control study in Fukuoka, Japan. The study protocol has been approved by the ethics committee of Kyushu University and collaborating hospitals. Details of the study procedures were described elsewhere (29-31).
Cases were patients with histologically confirmed colorectal adenocarcinomas who were admitted to either of the two university hospitals and their affiliates in Fukuoka City for the first surgical treatment. Eligibility included being residents in Fukuoka City and three adjacent areas, between 20 and 74 years of age at the time of diagnosis, mentally competent to complete the interview, no previous history of colorectal cancer, no surgical resection of large-bowel, and no known familial adenomatous polyposis. Research nurses visited each hospital weekly, and determined the eligibility of cases by referring to admission logs and medical records. They then contacted each eligible patient and, if written informed consent was obtained, interviewed him/her. The survey started at two university hospitals and three affiliated hospitals in September or October 2000, and at another three affiliated hospitals in May 2001, and ended in December 2003. In the consecutive series of potentially eligible 1,099 cases, 19 were found to be mentally incompetent, 23 had no histologic diagnosis of colorectal adenocarcinomas, three were found to have a history of large-bowel resection after the interview, and one had familial adenomatous polyposis. A total of 840 cases (80% of eligible patients) participated in the interview.
Controls were randomly selected from the community by frequency-matching with respect to gender and 10-year age groups. Eligible criteria were the same as that for cases except that controls must not have a history of colorectal cancer. A total of 1,500 persons were selected a priori as control candidates by two-stage random sampling, using residential registry. Recruitment was initiated by a letter, sent to each control candidate, and a telephone call was made if the candidate was listed in the telephone directory. At most, three further letters of invitation were mailed to nonrespondents. The mail invitation revealed that 113 persons were ineligible for the study. In addition, five persons were diagnosed as having colorectal cancer after the interview survey. After the exclusion of these 118 persons, 833 (60%) of the 1,382 eligible candidates participated in the study.
After further exclusion of six subjects whose estimated daily energy intakes were >5,000 kcal, 836 cases and 861 controls remained for analysis.
Assessment of Lifestyles
Information on diet and lifestyles were ascertained by in-person interview. Interviews for cases were conducted in-hospital during admission, whereas those for controls were conducted mostly at public community centers or collaborating clinics. The index date was defined as the date of onset of symptoms or the screening leading to the diagnosis of colorectal cancer for cases and the date of interview for controls.
Interviewers asked each participant about diet over a period of 1 year prior to the index date using a personal computer software, which was developed with reference to existing dietary questionnaires in Japan (32-34). Details of the development of the software as well as the validity and reproducibility of estimated intakes have been described elsewhere (35). In short, the software developed was designed to assess usual dietary intake by obtaining information on the consumption frequencies and portion sizes of 148 food items. Typical dishes for each food item were shown on the display window of the personal computer. The collected information was the same as obtained by a quantitative food frequency questionnaire. Intakes of nutrients were calculated based on the food composition tables in Japan (36), supplemented by original data derived from dietary records in a validation study. Dietary intakes estimated from the interview data were validated against those obtained from dietary records over a period of 7 consecutive days in four seasons (35). Pearson correlation coefficients of energy-adjusted intake of selected nutrients and food groups between the two methods were as follows: milk (0.80), other dairy foods (0.77), calcium (0.59), and vitamin D (0.14). Average calcium intake estimated from the interview was 20% greater than that obtained from dietary records, whereas average vitamin D intake did not differ between the two methods.
Type of job engaged was asked with the following response options: sedentary or standing work, work with walking, labor work, hard labor work, and no job. For up to three leisure-time physical activities engaged at least once per week, type, number of months, and of days per week that individuals participated in each, and minutes of participation per occasion were reported. Intensity of each physical activity was determined in terms of metabolic equivalents according to the literature (37), and total metabolic equivalent-hours per week was calculated for each person. As regards smoking habit, participants who had ever smoked cigarettes every day for 1 year or longer were asked about cigarette consumption for each decade of age. For participants who had ever consumed alcoholic beverages at least once per week over a period of 1 year or longer in their lifetime, the frequency of consumption per week and amount of alcohol consumed on the day of alcohol drinking over the year at the time of 5 years prior to the interview were asked. The amount of alcohol was expressed in go, the conventional unit in Japan (1 go contains 23 g of ethanol). Height, recent body weight, and body weight 10 years prior to the index date were asked. Body mass index was expressed as the body weight 10 years previously in kilograms divided by the square of body height in meters. When body weight 10 years previously was not answered, recent body weight was used instead. The use of vitamin or mineral supplements and analgesics was recorded if these had ever been used at least once per week for a period of 6 months or longer. Parental history of colorectal cancer was also determined.
Statistical Analysis
Dietary intakes of calcium and vitamin D were adjusted for total energy intake using a residual method (38), with adjustment to the median of total energy intake. Because few subjects took supplements of calcium (n = 40) or vitamin D (n = 15), intakes from supplement were not considered. Subjects were categorized into quintiles according to intakes of these nutrients among controls. These procedures were done in a sex-specific manner. Because dairy foods are a rich source of calcium, the associations with intakes of total dairy foods, milk, and dairy foods other than milk were also evaluated. Logistic regression analysis was used to estimate the odds ratio of colorectal cancer for each quintile group taking the lowest quintile group as the reference. Confounding variables considered were residential area (Fukuoka City or others), sex, age (5-year age group), parental history of colorectal cancer, smoking (lifetime nonsmoker, past or current smoker having consumed <400, 400-799, ≥800 cigarette-years), alcohol drinking (nondrinker, drinker consuming <1, 1-1.9, ≥2 go/d), occupational physical activities (no work or sedentary or standing work, or others), leisure-time physical activity (no, <10, 10-19.9, ≥20 metabolic equivalent-hours per week), body mass index (<21, 21-22.9, 23-24.9, ≥25 kg/m2), intakes of total energy, vegetable, fruit, and red meat (sex-specific quartiles among controls). No adjustment was made for the use of nonsteroidal anti-inflammatory drugs, which was not common among the study subjects (n = 69). Two types of odds ratio were calculated, one with adjustment of residential area, sex, and age only, and another with adjustment for all of the potential confounding variables mentioned above. Trend association was assessed by assigning to each subgroup the corresponding median intake, which was treated as a continuous variable. Analysis was also conducted for cancer subsite in the colorectum (proximal colon including cecum and ascending and transverse colon, distal colon including descending and sigmoid segments, and rectum).
Because cutaneous production of vitamin D greatly contributes to systemic vitamin D levels (39), we used occupational and leisure-time physical activities as a surrogate of sunlight exposure; those who engaged in work with walking, labor work, or hard labor work or those who engaged in leisure-time physical activities that are usually done outdoor at least 120 min per week were regarded as having high levels of sunlight exposure, whereas those on sedentary or standing work (including no work) who did not engage in any outdoor physical activity at leisure were considered to have low levels of daily sunlight exposure. Then an analysis was done according to sunlight exposure status, dietary vitamin D intake, or their combination to examine whether the association between calcium intake and colorectal cancer risk differed according to vitamin D status. Statistical significance for the interaction was assessed by including a cross-product term of calcium intake (continuous) and sunlight exposure or dietary vitamin D intake. All tests were two-sided and statistical significance was declared at P < 0.05. Analysis was done with SAS software (40).
Results
Cases and controls did not significantly differ in terms of residence, sex, age, total energy intake, and other potential confounding variables except for body mass index and intakes of calcium and milk (Table 1). Compared with controls, cases were more likely to be obese and consumed smaller amounts of calcium and milk.
Characteristics of study subjects
. | Cases . | Controls . | ||
---|---|---|---|---|
No. | 836 | 831 | ||
Residence, Fukuoka city (%) | 60 | 65 | ||
Sex, women (%) | 40 | 38 | ||
Age (y)* | 61 (54-68) | 61 (52-68) | ||
Job | ||||
Work with walking (%) | 12 | 14 | ||
Labor or hard labor work (%) | 14 | 16 | ||
Parental history of colorectal cancer (%) | 7 | 6 | ||
Smoking, ever (%) | 55 | 59 | ||
Alcohol use, ≥1 go/d (%) | 37 | 35 | ||
Leisure-time physical activity, ≥10 MET-h/w (%) | 32 | 30 | ||
Body mass index, (kg/m2) | 23.1 (21.1-25.3) | 22.7 (20.7-24.8) | ||
Dietary intake | ||||
Total energy (kcal/d)* | 2,159 (1,770-2,561) | 2,183 (1,784-2,606) | ||
Vegetable (g/d)* | 268 (192-388) | 280 (194-379) | ||
Fruit (g/d)* | 114 (60-181) | 118 (65-188) | ||
Red meat (g/d)* | 43 (27-66) | 45 (26-67) | ||
Calcium (mg/d)*,† | 615 (499-744) | 632 (515-777) | ||
Vitamin D (μg/d)*,† | 8.6 (6.4-11.2) | 8.6 (6.4-11.9) | ||
Total dairy food (g/d)* | 111 (44-200) | 114 (40-204) | ||
Milk (g/d)* | 56 (7-186) | 86 (10-186) |
. | Cases . | Controls . | ||
---|---|---|---|---|
No. | 836 | 831 | ||
Residence, Fukuoka city (%) | 60 | 65 | ||
Sex, women (%) | 40 | 38 | ||
Age (y)* | 61 (54-68) | 61 (52-68) | ||
Job | ||||
Work with walking (%) | 12 | 14 | ||
Labor or hard labor work (%) | 14 | 16 | ||
Parental history of colorectal cancer (%) | 7 | 6 | ||
Smoking, ever (%) | 55 | 59 | ||
Alcohol use, ≥1 go/d (%) | 37 | 35 | ||
Leisure-time physical activity, ≥10 MET-h/w (%) | 32 | 30 | ||
Body mass index, (kg/m2) | 23.1 (21.1-25.3) | 22.7 (20.7-24.8) | ||
Dietary intake | ||||
Total energy (kcal/d)* | 2,159 (1,770-2,561) | 2,183 (1,784-2,606) | ||
Vegetable (g/d)* | 268 (192-388) | 280 (194-379) | ||
Fruit (g/d)* | 114 (60-181) | 118 (65-188) | ||
Red meat (g/d)* | 43 (27-66) | 45 (26-67) | ||
Calcium (mg/d)*,† | 615 (499-744) | 632 (515-777) | ||
Vitamin D (μg/d)*,† | 8.6 (6.4-11.2) | 8.6 (6.4-11.9) | ||
Total dairy food (g/d)* | 111 (44-200) | 114 (40-204) | ||
Milk (g/d)* | 56 (7-186) | 86 (10-186) |
Abbreviation: MET, metabolic equivalents.
Median (interquartile range) for continuous variables.
Energy-adjusted.
As shown in Table 2, there was a significant inverse association between calcium intake and colorectal cancer risk when residence, sex, and age were adjusted for (P for trend = 0.002). The association did not materially change after further adjustment for other potential confounding variables (P for trend = 0.01). The fully adjusted odds ratio comparing the highest versus lowest quintile group of calcium intake was 0.64 [95% confidence interval (CI) 0.45-0.93]. Among the highest intake group, the odds ratio did not materially differ between lower-half and higher-half intake groups, who were divided at the median value (900 and 947 mg for men and women, respectively).
Associations between intakes of calcium and vitamin D and colorectal cancer
. | Quintiles of intake* . | . | . | . | . | . | P for trend . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | . | |||||||
Calcium | ||||||||||||||
No. of cases/controls | 186/165 | 167/167 | 192/165 | 159/167 | 132/167 | |||||||||
OR (95% CI)† | 1 (reference) | 0.88 (0.65-1.19) | 0.98 (0.72-1.32) | 0.76 (0.56-1.04) | 0.62 (0.45-0.85) | 0.002 | ||||||||
Multivariate OR (95% CI)‡ | 1 (reference) | 0.92 (0.67-1.27) | 1.05 (0.76-1.45) | 0.82 (0.59-1.15) | 0.64 (0.45-0.93) | 0.01 | ||||||||
Lower-half | Higher-half | |||||||||||||
0.66 (0.43-1.02) | 0.63 (0.41-0.97) | |||||||||||||
Vitamin D | ||||||||||||||
No. of cases/controls | 157/165 | 175/167 | 182/165 | 179/167 | 143/167 | |||||||||
OR (95% CI) † | 1 (reference) | 1.03 (0.75-1.40) | 1.03 (0.75-1.40) | 0.99 (0.72-1.35) | 0.78 (0.56-1.07) | 0.10 | ||||||||
Multivariate OR (95% CI)‡ | 1 (reference) | 1.05 (0.76-1.45) | 1.05 (0.76-1.45) | 1.00 (0.72-1.38) | 0.79 (0.56-1.11) | 0.12 | ||||||||
Lower-half | Higher-half | |||||||||||||
0.87 (0.58-1.30) | 0.71 (0.47-1.08) | |||||||||||||
Multivariate OR (95% CI)‡,§ | 1 (reference) | 1.18 (0.72-1.91) | 0.82 (0.50-1.36) | 0.68 (0.40-1.15) | 0.63 (0.36-1.08) | 0.02 | ||||||||
Multivariate OR (95% CI)‡,∥ | 1 (reference) | 0.93 (0.61-1.59) | 1.40 (0.87-2.23) | 1.25 (0.79-1.98) | 0.94 (0.58-1.52) | 0.86 |
. | Quintiles of intake* . | . | . | . | . | . | P for trend . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | . | |||||||
Calcium | ||||||||||||||
No. of cases/controls | 186/165 | 167/167 | 192/165 | 159/167 | 132/167 | |||||||||
OR (95% CI)† | 1 (reference) | 0.88 (0.65-1.19) | 0.98 (0.72-1.32) | 0.76 (0.56-1.04) | 0.62 (0.45-0.85) | 0.002 | ||||||||
Multivariate OR (95% CI)‡ | 1 (reference) | 0.92 (0.67-1.27) | 1.05 (0.76-1.45) | 0.82 (0.59-1.15) | 0.64 (0.45-0.93) | 0.01 | ||||||||
Lower-half | Higher-half | |||||||||||||
0.66 (0.43-1.02) | 0.63 (0.41-0.97) | |||||||||||||
Vitamin D | ||||||||||||||
No. of cases/controls | 157/165 | 175/167 | 182/165 | 179/167 | 143/167 | |||||||||
OR (95% CI) † | 1 (reference) | 1.03 (0.75-1.40) | 1.03 (0.75-1.40) | 0.99 (0.72-1.35) | 0.78 (0.56-1.07) | 0.10 | ||||||||
Multivariate OR (95% CI)‡ | 1 (reference) | 1.05 (0.76-1.45) | 1.05 (0.76-1.45) | 1.00 (0.72-1.38) | 0.79 (0.56-1.11) | 0.12 | ||||||||
Lower-half | Higher-half | |||||||||||||
0.87 (0.58-1.30) | 0.71 (0.47-1.08) | |||||||||||||
Multivariate OR (95% CI)‡,§ | 1 (reference) | 1.18 (0.72-1.91) | 0.82 (0.50-1.36) | 0.68 (0.40-1.15) | 0.63 (0.36-1.08) | 0.02 | ||||||||
Multivariate OR (95% CI)‡,∥ | 1 (reference) | 0.93 (0.61-1.59) | 1.40 (0.87-2.23) | 1.25 (0.79-1.98) | 0.94 (0.58-1.52) | 0.86 |
Abbreviation: OR, odds ratio.
Cutoff values for quintiles of calcium intake were 464, 563, 656, and 794 mg/d for men and 535, 616, 720, and 841 mg/d for women; those for vitamin D intake were 5.8, 8.0, 10.0, 13.3 μg/d (232, 320, 400, 532 IU) for men and 5.6, 7.6, 9.2, 12.2 μg/d (224, 304, 368, 488 IU) for women; those dichotomizing the highest calcium intake group were 900 and 947 mg/d for men and women, respectively; the corresponding values for vitamin D intake were 15.7 and 14.7 μg/d (628 and 588 IU).
Adjusted for residence, sex, and age.
Adjusted for residence, sex, age, job, parental history of colorectal cancer, smoking, alcohol drinking, body mass index, leisure-time physical activity, and intakes of energy, vegetable, fruit, and red meat.
Among 685 subjects who engaged in sedentary or standing work (including no job) and did not engage in outdoor physical activity at leisure.
Among 875 subjects who engaged in work with walking or labor work or engaged in outdoor physical activity at leisure at least 120 min/wk.
Multivariate analysis showed that vitamin D intake was nonsignificantly, inversely associated with colorectal cancer risk (P for trend = 0.12), and that those in the highest quintile group of vitamin D intake had a nonsignificant 21% lower risk of colorectal cancer compared with those in the lowest quintile group. When the highest category of vitamin D intake was dichotomized at the median value [15.7 μg (628 IU) and 14.7 μg (588 IU) for men and women, respectively], the higher-half intake group had a lower odds ratio than the lower-half intake group (0.71 versus 0.87). Moreover, a statistically significant inverse association with dietary vitamin D intake was observed among subjects who presumably had low levels of daily sunlight exposure (P for trend = 0.02) but not among those who presumably had higher levels of sunlight exposure (P for trend = 0.86).
There was no clear association between total dairy food intake and colorectal cancer risk (Table 3). However, the odds ratio of colorectal cancer associated with 200 g or more milk intake per day comparing less than 50 g milk intake per day was significantly decreased (fully adjusted odds ratio, 0.60; 95% CI, 0.40-0.91), and the overall trend association with milk intake was marginally significant (P for trend = 0.065). In contrast, intake of dairy foods other than milk was significantly, positively associated with colorectal cancer risk (P for trend = 0.01); odds ratio (95% CI) for the highest versus lowest quintile group was 1.39 (0.95-2.02).
Associations between intakes of dairy foods and colorectal cancer
. | Intake levels . | . | . | . | . | P for trend . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | ||||||
Total dairy foods | ||||||||||||
Intake (g/d) | <50 | 50-99 | 100-199 | 200-299 | ≥300 | |||||||
No. of cases/controls | 221/237 | 150/120 | 250/249 | 169/156 | 46/69 | |||||||
OR (95% CI)* | 1 (reference) | 1.45 (1.07-1.97) | 1.07 (0.83-1.39) | 1.16 (0.86-1.55) | 0.72 (0.47-1.10) | 0.24 | ||||||
OR (95% CI)† | 1 (reference) | 1.57 (1.14-2.17) | 1.19 (0.90-1.55) | 1.24 (0.91-1.70) | 0.76 (0.48-1.18) | 0.38 | ||||||
Milk | ||||||||||||
Intake (g/d) | <50 | 50-99 | 100-149 | 150-199 | ≥200 | |||||||
No. of cases/controls | 371/353 | 88/84 | 110/107 | 218/212 | 49/78 | |||||||
OR (95% CI)* | 1 (reference) | 1.04 (0.74-1.46) | 0.96 (0.70-1.31) | 0.94 (0.74-1.20) | 0.60 (0.41-0.89) | 0.046 | ||||||
OR (95% CI)† | 1 (reference) | 1.10 (0.77-1.56) | 1.00 (0.72-1.36) | 0.97 (0.75-1.25) | 0.60 (0.40-0.91) | 0.065 | ||||||
Dairy foods other than milk | ||||||||||||
Intake (g/d) | <10 | 10-24 | 25-49 | 50-99 | ≥100 | |||||||
No. of cases/controls | 287/299 | 169/182 | 122/137 | 172/135 | 86/78 | |||||||
OR (95% CI)* | 1 (reference) | 1.04 (0.80-1.37) | 0.99 (0.73-1.34) | 1.41 (1.05-1.88) | 1.20 (0.84-1.72) | 0.056 | ||||||
OR (95% CI)† | 1 (reference) | 1.10 (0.83-1.46) | 1.11 (0.80-1.52) | 1.60 (1.17-2.19) | 1.39 (0.95-2.02) | 0.01 |
. | Intake levels . | . | . | . | . | P for trend . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | ||||||
Total dairy foods | ||||||||||||
Intake (g/d) | <50 | 50-99 | 100-199 | 200-299 | ≥300 | |||||||
No. of cases/controls | 221/237 | 150/120 | 250/249 | 169/156 | 46/69 | |||||||
OR (95% CI)* | 1 (reference) | 1.45 (1.07-1.97) | 1.07 (0.83-1.39) | 1.16 (0.86-1.55) | 0.72 (0.47-1.10) | 0.24 | ||||||
OR (95% CI)† | 1 (reference) | 1.57 (1.14-2.17) | 1.19 (0.90-1.55) | 1.24 (0.91-1.70) | 0.76 (0.48-1.18) | 0.38 | ||||||
Milk | ||||||||||||
Intake (g/d) | <50 | 50-99 | 100-149 | 150-199 | ≥200 | |||||||
No. of cases/controls | 371/353 | 88/84 | 110/107 | 218/212 | 49/78 | |||||||
OR (95% CI)* | 1 (reference) | 1.04 (0.74-1.46) | 0.96 (0.70-1.31) | 0.94 (0.74-1.20) | 0.60 (0.41-0.89) | 0.046 | ||||||
OR (95% CI)† | 1 (reference) | 1.10 (0.77-1.56) | 1.00 (0.72-1.36) | 0.97 (0.75-1.25) | 0.60 (0.40-0.91) | 0.065 | ||||||
Dairy foods other than milk | ||||||||||||
Intake (g/d) | <10 | 10-24 | 25-49 | 50-99 | ≥100 | |||||||
No. of cases/controls | 287/299 | 169/182 | 122/137 | 172/135 | 86/78 | |||||||
OR (95% CI)* | 1 (reference) | 1.04 (0.80-1.37) | 0.99 (0.73-1.34) | 1.41 (1.05-1.88) | 1.20 (0.84-1.72) | 0.056 | ||||||
OR (95% CI)† | 1 (reference) | 1.10 (0.83-1.46) | 1.11 (0.80-1.52) | 1.60 (1.17-2.19) | 1.39 (0.95-2.02) | 0.01 |
Abbreviation: OR, odds ratio.
Adjusted for residence, sex, and age.
Adjusted for residence, sex, age, job, parental history of colorectal cancer, smoking, alcohol drinking, body mass index, leisure-time physical activity, and intakes of energy, vegetable, fruit, and red meat.
An inverse association between calcium intake and colorectal cancer risk was consistently observed for all subsites in the colorectum (Table 4); decreased odds ratios for the fourth and highest quintiles compared with the lowest quintile were comparable in magnitude across the subsites. For vitamin D intake, the inverse association seems to be stronger for the distal colon (P for trend = 0.08) than for other subsites in the colorectum (P for trend, 0.23 and 0.21 for proximal colon and rectum, respectively), although their CIs overlapped considerably.
Multivariate odds ratio and 95% CI of colorectal cancer by quintile of intakes of calcium and vitamin D according to subsite of the colorectum
. | Quintiles of intake* . | . | . | . | . | P for trend . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | ||||||
Calcium | ||||||||||||
Colon | 1 (reference) | 0.97 (0.66-1.42) | 1.23 (0.84-1.81) | 0.83 (0.55-1.24) | 0.69 (0.45-1.07) | 0.045 | ||||||
Proximal | 1 (reference) | 0.81 (0.47-1.40) | 1.11 (0.66-1.88) | 0.76 (0.43-1.33) | 0.60 (0.33-1.09) | 0.08 | ||||||
Distal | 1 (reference) | 1.06 (0.67-1.67) | 1.35 (0.86-2.15) | 0.84 (0.51-1.37) | 0.77 (0.46-1.31) | 0.16 | ||||||
Rectum | 1 (reference) | 0.85 (0.56-1.28) | 0.88 (0.58-1.34) | 0.79 (0.51-1.22) | 0.56 (0.35-0.91) | 0.02 | ||||||
Vitamin D | ||||||||||||
Colon | 1 (reference) | 0.99 (0.68-1.45) | 1.00 (0.69-1.47) | 0.93 (0.64-1.37) | 0.69 (0.46-1.04) | 0.07 | ||||||
Proximal | 1 (reference) | 0.97 (0.57-1.66) | 1.05 (0.61-1.79) | 0.92 (0.53-1.58) | 0.72 (0.41-1.27) | 0.23 | ||||||
Distal | 1 (reference) | 0.98 (0.62-1.55) | 0.92 (0.58-1.46) | 0.92 (0.58-1.46) | 0.64 (0.39-1.06) | 0.07 | ||||||
Rectum | 1 (reference) | 1.13 (0.74-1.72) | 1.02 (0.66-1.57) | 1.03 (0.67-1.59) | 0.80 (0.51-1.26) | 0.21 |
. | Quintiles of intake* . | . | . | . | . | P for trend . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | ||||||
Calcium | ||||||||||||
Colon | 1 (reference) | 0.97 (0.66-1.42) | 1.23 (0.84-1.81) | 0.83 (0.55-1.24) | 0.69 (0.45-1.07) | 0.045 | ||||||
Proximal | 1 (reference) | 0.81 (0.47-1.40) | 1.11 (0.66-1.88) | 0.76 (0.43-1.33) | 0.60 (0.33-1.09) | 0.08 | ||||||
Distal | 1 (reference) | 1.06 (0.67-1.67) | 1.35 (0.86-2.15) | 0.84 (0.51-1.37) | 0.77 (0.46-1.31) | 0.16 | ||||||
Rectum | 1 (reference) | 0.85 (0.56-1.28) | 0.88 (0.58-1.34) | 0.79 (0.51-1.22) | 0.56 (0.35-0.91) | 0.02 | ||||||
Vitamin D | ||||||||||||
Colon | 1 (reference) | 0.99 (0.68-1.45) | 1.00 (0.69-1.47) | 0.93 (0.64-1.37) | 0.69 (0.46-1.04) | 0.07 | ||||||
Proximal | 1 (reference) | 0.97 (0.57-1.66) | 1.05 (0.61-1.79) | 0.92 (0.53-1.58) | 0.72 (0.41-1.27) | 0.23 | ||||||
Distal | 1 (reference) | 0.98 (0.62-1.55) | 0.92 (0.58-1.46) | 0.92 (0.58-1.46) | 0.64 (0.39-1.06) | 0.07 | ||||||
Rectum | 1 (reference) | 1.13 (0.74-1.72) | 1.02 (0.66-1.57) | 1.03 (0.67-1.59) | 0.80 (0.51-1.26) | 0.21 |
NOTE: Adjustments were made for residence, sex, age, job, parental history of colorectal cancer, smoking, alcohol drinking, body mass index, leisure-time physical activity, and intakes of energy, vegetable, fruit, and red meat.
The number of cases were 476, 190, 277, and 354 for the colon, proximal colon, distal colon, and rectum, respectively; cases having tumors in multiple sites were excluded in respective analysis.
Cutoff values for quintiles of calcium intake were 464, 563, 656, 794 mg/d for men and 535, 616, 720, 841 mg/d for women; those for vitamin D intake were 5.8, 8.0, 10.0, 13.3 μg/d (232, 320, 400, 532 IU) for men and 5.6, 7.6, 9.2, 12.2 μg/d (224, 304, 368, 488 IU) for women.
The association between calcium and colorectal cancer was assessed according to dietary vitamin D intake and sunlight exposure status (Table 5). A statistically significant inverse association with calcium intake was observed among those with the highest one-third levels of vitamin D intake (P for trend = 0.02; odds ratio for the highest versus lowest quintile of calcium intake, 0.40; 95% CI, 0.19-0.84) or among those who presumably had higher levels of sunlight exposure (P for trend = 0.008; odds ratio for the highest versus lowest quintile of calcium intake, 0.55; 95% CI, 0.34-0.91). In contrast, no apparent association with calcium intake was observed among those with lower two-thirds levels of vitamin D intake or among those who had low levels of sunlight exposure. P values for the interaction with calcium intake were 0.29 and 0.04 for vitamin D intake and sunlight exposure status, respectively. In the analysis by overall vitamin D status, defined by the combination of dietary intake and sunlight exposure, risk reduction with higher levels of calcium intake was greatest among subjects high in both indices, followed by those high in either.
Multivariate odds ratio and 95% CI for colorectal cancer by quintile of calcium intakes according to vitamin D status
. | No. of case/control . | Quintiles of calcium intake* . | . | . | . | . | P for trend . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | . | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | |||||||
Dietary vitamin D† | ||||||||||||||
Lower third | 275/278 | 1 (reference) | 1.06 (0.63-1.77) | 0.94 (0.53-1.65) | 0.85 (0.46-1.59) | 0.69 (0.32-1.48) | 0.29 | |||||||
Medium third | 328/276 | 1 (reference) | 1.24 (0.71-2.16) | 1.25 (0.73-2.14) | 0.99 (0.55-1.77) | 1.05 (0.55-1.99) | 0.85 | |||||||
Higher third | 233/277 | 1 (reference) | 0.46 (0.21-0.98) | 1.03 (0.49-2.15) | 0.77 (0.37-1.57) | 0.40 (0.19-0.84) | 0.02 | |||||||
Sunlight exposure‡ | ||||||||||||||
Low | 352/333 | 1 (reference) | 0.91 (0.56-1.48) | 1.03 (0.63-1.69) | 0.86 (0.50-1.49) | 0.92 (0.50-1.67) | 0.68 | |||||||
High | 435/440 | 1 (reference) | 0.93 (0.58-1.48) | 1.06 (0.66-1.71) | 0.81 (0.50-1.29) | 0.55 (0.34-0.91) | 0.01 | |||||||
Vitamin D status§ | ||||||||||||||
Neither high | 307/271 | 1 (reference) | 1.19 (0.74-1.93) | 1.05 (0.65-1.69) | 0.81 (0.46-1.41) | 1.03 (0.53-2.03) | 0.85 | |||||||
Either high | 390/403 | 0.94 (0.59-1.49) | 0.71 (0.44-1.14) | 0.96 (0.60-1.54) | 0.81 (0.50-1.30) | 0.71 (0.44-1.16) | 0.68 | |||||||
Both high | 139/157 | 1.03 (0.47-2.24) | 0.81 (0.40-1.62) | 1.12 (0.58-2.16) | 0.72 (0.40-1.31) | 0.38 (0.21-0.68) | 0.001 |
. | No. of case/control . | Quintiles of calcium intake* . | . | . | . | . | P for trend . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | . | 1 (lowest) . | 2 . | 3 . | 4 . | 5 (highest) . | . | |||||||
Dietary vitamin D† | ||||||||||||||
Lower third | 275/278 | 1 (reference) | 1.06 (0.63-1.77) | 0.94 (0.53-1.65) | 0.85 (0.46-1.59) | 0.69 (0.32-1.48) | 0.29 | |||||||
Medium third | 328/276 | 1 (reference) | 1.24 (0.71-2.16) | 1.25 (0.73-2.14) | 0.99 (0.55-1.77) | 1.05 (0.55-1.99) | 0.85 | |||||||
Higher third | 233/277 | 1 (reference) | 0.46 (0.21-0.98) | 1.03 (0.49-2.15) | 0.77 (0.37-1.57) | 0.40 (0.19-0.84) | 0.02 | |||||||
Sunlight exposure‡ | ||||||||||||||
Low | 352/333 | 1 (reference) | 0.91 (0.56-1.48) | 1.03 (0.63-1.69) | 0.86 (0.50-1.49) | 0.92 (0.50-1.67) | 0.68 | |||||||
High | 435/440 | 1 (reference) | 0.93 (0.58-1.48) | 1.06 (0.66-1.71) | 0.81 (0.50-1.29) | 0.55 (0.34-0.91) | 0.01 | |||||||
Vitamin D status§ | ||||||||||||||
Neither high | 307/271 | 1 (reference) | 1.19 (0.74-1.93) | 1.05 (0.65-1.69) | 0.81 (0.46-1.41) | 1.03 (0.53-2.03) | 0.85 | |||||||
Either high | 390/403 | 0.94 (0.59-1.49) | 0.71 (0.44-1.14) | 0.96 (0.60-1.54) | 0.81 (0.50-1.30) | 0.71 (0.44-1.16) | 0.68 | |||||||
Both high | 139/157 | 1.03 (0.47-2.24) | 0.81 (0.40-1.62) | 1.12 (0.58-2.16) | 0.72 (0.40-1.31) | 0.38 (0.21-0.68) | 0.001 |
NOTE: Adjustments were made for residence, sex, age, job, parental history of colorectal cancer, smoking, alcohol drinking, body mass index, and intakes of energy, vegetable, fruit, and red meat; leisure-time physical activity was additionally adjusted in the analysis stratified by vitamin D intake.
P values for the interaction with calcium intake were 0.29, 0.04, 0.008 for dietary vitamin D intake, sunlight exposure, and vitamin D status, respectively.
Cutoff values for quintiles of calcium intake were 464, 563, 656, and 794 mg/d for men and 535, 616, 720, 841 mg/d for women.
Cutoff values for tertiles of vitamin D intake were 7.3 and 11.0 μg/d (292 and 440 IU) for men and 7.2 and 10.3 μg/d (288 and 412 IU) for women.
Low, sedentary, or standing work (including no job) and no engagement in outdoor leisure-time physical activity; high, work with walking or labor work or engagement in outdoor leisure-time physical activity at least 120 min/wk; those not classified were excluded from the analysis.
Classified by dietary vitamin D and sunlight exposure; “high” indicates higher third of dietary vitamin D and high sunlight exposure as defined above.
Discussion
In a large-scale case-control study among a Japanese population, we found inverse associations of dietary calcium and vitamin D with colorectal cancer risk. The reduction in risk was mostly confined to the group with the greatest intake of these nutrients. These associations did not vary substantially according to subsites in the colorectum. Furthermore, both dietary vitamin D and sunlight exposure modified the relation of calcium intake to colorectal cancer risk.
To our knowledge, this is the second report from Japan of the association of calcium and vitamin D intakes with colorectal cancer. Data of the first report by Wakai et al. (41) were derived from a hospital-based case-control study, in which diet was assessed using a 47-item self-administered food frequency questionnaire. In the present study, controls were selected randomly from a general population and diet was assessed through in-person interview about 148 food items.
A meta-analysis published in 1998 of 10 case-control and cohort studies (8) did not find a consistent association between calcium and the risk of colorectal cancer. However, the present finding of an inverse association between calcium intake and colorectal cancer agrees with the results of recent studies, including a pooled analysis of Western cohort studies (27), other large-scale studies not included in the pooled analysis (11, 14, 15, 17), and Asian studies (41, 42). A protective role of this nutrient in colorectal carcinogenesis in humans has been further supported by evidence from randomized controlled trials of calcium supplement and colorectal adenoma recurrence (9).
In the present study, odds ratio for colorectal cancer associated with calcium intake started to decrease in the fourth (second highest) quintile group, and a statistically significant, 36% decreased odds ratio was observed in the highest quintile group, with a median daily calcium intake of 900 and 947 mg for men and women, respectively. When the highest intake group was dichotomized, there was no evidence suggesting a further reduction in risk associated with greater intake. This dose-response pattern seems to agree with the result of a pooled analysis of 10 prospective studies (27) and other studies (14-16), showing a decreasing trend in risk up to 1,000 mg of daily calcium intake, over which little additional benefit was attained. However, some studies (11, 17) indicated that >1,000 mg of daily calcium intake is required to achieve a maximum preventive effect against colorectal cancer. Variation in these data may reflect a methodologic difference in dietary assessment across studies, but dietary and nondietary factors that influence absorption or activation of calcium might also explain the discrepancy, as discussed later.
Dairy foods are a rich source of calcium and contain potent anticarcinogenic substances including conjugated linoleic acid (43). In the present study, high milk intake (200 g or more per day) was associated with a significantly decreased risk of colorectal cancer, a finding consistent with the results of several studies including a pooled analysis (27). With regard to dairy foods other than milk, epidemiologic evidence is conflicting. In our study, higher intake of dairy foods other than milk was associated with an increased risk of colorectal cancer. However, the pooled analysis (27) reported no association with yogurt or cheese, and a Swedish study (44) showed a nonsignificant inverse association with most dairy foods other than milk. Thus, more research is warranted to clarify this point.
Studies of dietary vitamin D intake and colorectal cancer have provided mixed results (11, 13, 14, 16, 23, 24, 41). One possible explanation is that dietary vitamin D intake among most study populations may be too low to exert a cancer-preventive action. In this study of a Japanese population, which consumed a large amount of fish, we were able to assess the effects of high levels of dietary vitamin D intake on colorectal cancer risk. We found a nonsignificant 22% reduction in colorectal cancer risk associated with the highest intake of vitamin D and, when the highest consumption category was dichotomized, a greater risk reduction was observed in the higher intake group, with a median vitamin D intake of 18.6 μg (744 IU) and 17.0 μg (680 IU) for men and women, respectively. The result agrees with the findings from studies that have taken into account supplementary vitamin D (11, 14), which assessed the association of colorectal cancer risk with very high levels of vitamin D intake. Another notable finding is that an inverse association with dietary vitamin D was observed only among those who probably had low levels of daily sunlight exposure. This result suggests that sunlight exposure may conceal the association with dietary vitamin D and underscores the importance of considering cutaneous photoinitiated production of vitamin D in exposure assessment. The null finding in a randomized controlled study (45) may be partly ascribed to a lack of consideration for sunlight-induced vitamin D. In fact, serum 25-hydroxyvitamin D levels, which reflect both dietary and skin-produced vitamin D (39), have been fairly consistently associated with a lower risk of colorectal cancer (45, 46).
Vitamin D controls the absorption of calcium in the gut (25) and, as recent studies suggest (6, 7), is involved in several mechanisms whereby calcium exerts anticarcinogenic effects. In line with these experimental data suggesting the joint action of calcium and vitamin D in reducing colorectal cancer risk, a decreased risk of adenoma recurrence associated with calcium supplementation was detected only among subjects with high serum vitamin D levels (26). Similarly, in a case-control study of colorectal adenomas (47), a stronger inverse association with serum vitamin D levels was observed only among those who had higher intake of calcium. However, a prospective study (24) found that the risk of colorectal cancer was increased, rather than decreased, among those who had higher dietary intakes of vitamin D, whereas another study (15) reported no difference in the calcium association across strata of dietary vitamin D. Null results for the interacting effect between vitamin D and calcium may be due to the small number of cancer cases in stratified analyses or the lack of consideration for sunlight exposure. In a pooled analysis of 4,995 cases of colorectal cancer (27), a significant inverse association with calcium was observed only among individuals in the highest-third category of dietary vitamin D intake. In our study, a reduced risk of colorectal cancer associated with high calcium intake was observed among subjects who consumed high amounts of vitamin D or those who had a potentially greater exposure to sunlight, a finding compatible with the hypothesis that calcium and vitamin D jointly, not independently, exert a preventive effect on colorectal cancer.
Data regarding the association between calcium or vitamin D and colorectal cancer according to cancer subsite is conflicting. In the pooled analysis of 10 cohort studies (27), calcium was associated with distal colon and rectal cancers but not with proximal colon cancer. The Cancer Prevention Study II Nutrition Cohort (14) observed a significant association with calcium intake only in the proximal colon. Another study (15) found an inverse association in the proximal and distal colon but not in the rectum. However, similar to our study, a cohort study of Swedish men (17) did not find significant variation in risk by subsite in the colorectum. For vitamin D, our study detected a somewhat clearer association for distal colon cancer, although CIs overlapped considerably across subsites in the colorectum. Similarly, there was a suggestion of a stronger association for distal colon cancer in the Cancer Prevention Study II Nutrition Cohort (14).
Our study had several strengths. First, trained interviewers asked about the frequency and quantity of intakes for 148 food items by showing photos of several typical foods or dishes for each food category on the display of a personal computer. This method was shown to be fairly valid regarding most of the nutrients and food groups (35). Second, in this large-scale study, we had more than 150 control subjects for each quintile group. This would allow us to detect a moderate association with reasonable precision. Third, because fortification of milk with vitamin D is not common in Japan, our data is suitable in assessing an independent association of calcium or vitamin D with colorectal cancer risk.
There are also limitations in our study. As with other case-control studies, potential bias in the recall of past diet cannot be ruled out. However, because calcium and vitamin D are not accepted as nutrients that prevent colorectal cancer in our society, we believe that disease status might not significantly influence the consumption of foods rich in these nutrients. Even so, the protective effects of these nutrients would be greater than the present estimates. Second, because the participation rate for controls was not sufficiently high (60%), selection bias may have influenced the results. Third, the validity of our estimate of vitamin D intake was low. Fish and eggs are major sources of vitamin D intake in Japanese (28). We have no clear explanation for the low validity, but the observed association between vitamin D and colorectal cancer risk was probably underestimated. Fourth, our study lacks measurements of blood 25-hydroxyvitamin D, a marker of systemic vitamin D exposure. Finally, we created a variable of sun exposure based on occupational and leisure-time physical activities, and thus, these variables were highly correlated. Although job-related and leisure-time physical activities were associated with a reduced risk of colorectal cancer in the present study (30), it is unlikely that physical activity itself confounded the interaction between vitamin D exposure index and calcium intake.
In conclusion, the present findings from a large-scale case-control study in a Japanese population add to evidence that calcium and vitamin D are associated with a lower risk of colorectal cancer. Furthermore, the present data support a hypothesis that vitamin D modulates the effects of calcium on colorectal carcinogenesis. If the estimated intake of calcium was calibrated according to the validation study (35), a sizable reduction in colorectal cancer risk was observed only for subjects with 700 mg or greater calcium intake per day plus 10 μg (400 IU) or greater vitamin D intake per day or engagement in any outdoor activity in daily life. Dietary modification to increase calcium intake while maintaining adequate vitamin D status through diet and moderate sun exposure have a large potential in the prevention of colorectal cancer among Japanese adults, whose average calcium intake remains low (520 mg/d; ref. 28).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology, Japan (18014022) and the Third Term Comprehensive 10-year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare, Japan.
Acknowledgments
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.