Epidemiological Evidence for Vitamin D and Colorectal Cancer


  • Edward Giovannucci MD, ScD

    Corresponding author
    1. Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women's Hospital, Boston, Massachusetts, USA; Department of Nutrition and Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, USA
    • Address reprint requests to: Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA
    Search for more papers by this author

  • The author states that he has no conflicts of interest.


Since Garland and Garland formulated the hypothesis that vitamin D may protect against colorectal cancer in 1980, various epidemiological approaches have been undertaken to evaluate this hypothesis. These approaches include studies based on regional solar UVB radiation, plasma- or serum-based studies, dietary studies, and those examining multiple factors that influence vitamin D status. Studies over the past several decades have tended to support that higher levels of vitamin D may decrease risk of colorectal cancer. An important implication is that current recommended dietary intakes such as 200–400 IU/d may be too low to exert appreciable benefits. To substantially reduce risk, higher levels of vitamin D associated with sunshine exposure or considerably higher intakes may be required. Recent studies also suggest a potential benefit of vitamin D on other digestive system cancers. One study suggested that a better vitamin D status at the time of diagnosis and treatment, as indicated by season of diagnosis, may improve survival from colorectal cancer. Darker-skinned individuals who tend to make less vitamin D may be at particularly high risk for digestive system cancer. The strong biological evidence for a protective role of vitamin D supports the epidemiological data. More study is needed to determine the optimal levels and intakes of this vitamin to optimally reduce colorectal cancer risk.


Based on observations of higher rates of skin cancer but lower rates of other malignancies in United States Navy personnel in the 1930s, Peller and Stephensonmath image hypothesized that acquiring skin cancer conferred some immunity against other cancers. Several years later, Apperlymath image found an inverse association between latitude and cancer mortality rate and stated that “The presence of skin cancer is really an occasional accompaniment of a relative cancer immunity in some way related to the exposure to solar radiation.” These suggestions received little scrutiny until 1980, about four decades later, when Garland and Garlandmath image hypothesized that vitamin D status accounted for the inverse association between solar UVB radiation exposure and risk of colon cancer. In later years, Garland and colleagues proposed similar benefits of sunshine and vitamin D for breast cancermath image and ovarian cancer.math image When Garland and Garlandmath image formulated their hypothesis in 1980, little was known about the biological functions of vitamin D beyond the classical functions on mineral homeostasis and bone physiology. Subsequently, a large number of studies have found that many cell types contain vitamin D receptors; furthermore, when these receptors are activated by 1,25(OH)2D, the most active metabolite of vitamin D, they induce potential cancer protective properties such as increasing differentiation and inhibiting proliferation, invasiveness, angiogenesis, and metastatic potential.math image Circulating 25(OH)D was shown to be potentially beneficial, because many cell types including cancer cells express 1-α-hydroxylase, which is required to convert 25(OH)D into 1,25(OH)2D. This brief review considers the current epidemiological evidence of the role of vitamin D for the potential prevention of colorectal cancer.


First considered here is the evidence from the ecological studies based on surrogates of sun exposure, which is the main source of vitamin D in most populations. The initial observation of an inverse correlation between solar UVB and risk of colon cancermath image has been confirmed subsequently in the United States using more detailed data and analyses. For example, Grantmath image showed that regional UVB radiation correlated inversely with mortality rates of numerous cancers. The strongest associations were for cancers of the colon and rectum; of all the preventable cancers estimated attributable to living in a low sun area, 60% were caused by colorectal cancer in men, and 35% were caused by colorectal cancer in women (the smaller proportion in women is because of the relative large proportion [42%] also attributable to breast cancer in women). Thus, although other cancers may be affected, the large bowel and breast appear to be quantitatively most important in the United States. A limitation of these studies is that, whereas solar UVB would correlate with vitamin D production, it is possible that other biological effects of sun exposure could lower risk, although apparently no plausible alternative hypotheses have been offered.

Another limitation of these studies is the questionable ability of adequately controlling for potentially confounding factors; in essence, other uncontrolled factors correlated with solar UVB could be the actual causal factors. One way to address this limitation is to examine the consistency of this finding across diverse populations, because it would be unlikely, although not impossible, that the same causal factors would correlate with regional solar UVB across different populations. In this regard, a study by Mizouemath image is of interest because it was conducted in Japan, an entirely different population. Mizoue calculated Pearson correlation coefficients between averaged annual solar radiation levels for the period from 1961 through 1990 and cancer mortality in 2000 in 47 prefectures in Japan, adjusting for regional per capita income and dietary factors. An inverse correlation between averaged annual solar radiation levels and mortality from cancers of the large bowel were found; for colon cancer, the correlations were –0.53 in men and –0.46 in women, and for rectal cancer, the correlations were –0.53 in men and –0.47 in women.math image Similar to the United States,math image other digestive system cancers (esophagus, stomach, pancreas, and gallbladder and bile ducts) had significant inverse correlations with regional solar UVB radiation level. These cancers have diverse risk factors, and it seems unlikely that an unknown confounding factor would be operative for all these cancers in a similar fashion both in the United States and Japan. Although Japan has a relatively high intake of vitamin D from fatty fish, the prevalence of hypovitaminosis D is still fairly high, and sun exposure is relevant.math image


The generally accepted best indicator of vitamin D nutritional status is plasma or serum 25(OH)D level. This metabolite accounts not only for skin exposure to UVB radiation but also for total vitamin D intake and the hydroxylation of all sources of cholecalciferol in the liver.math image In addition, it has a relatively long half-life (t1/2) in the circulation of ∼2–3 wk.math image The initial studies that examined 25(OH)D levels prospectively in relation to risk of colorectal cancer generally supported an inverse association.math image Recently, the relationship between plasma 25(OH)D and colorectal cancer risk based on 193 incident cases was examined in the Nurses' Health Study.math image The multivariable relative risk (RR) decreased monotonically across quintiles of plasma 25(OH)D concentration. In an another study, an inverse association was observed between baseline 25(OH)D level and colorectal cancer risk in the Women's Health Initiative (WHI), a randomized placebo-controlled trial of 400 IU vitamin D plus 1000 mg/d of calcium in 36,282 postmenopausal women; however, as discussed below, the interventional component of this study did not support a protective role of vitamin D intake.math image In general, across the studies of circulating 25(OH)D, individuals in the high quartile or quintile of 25(OH)D had about one half the risk of colorectal cancer as did those in the lowest group. The dose–response appears fairly linear up to a 25(OH)D level of at least 35–40 ng/ml.

There are several strengths of the serum- or plasma-based studies. The best single indicator of vitamin D status is assessed at the individual level, and subsequent risk of cancer is examined. Potentially confounding factors such as body mass index and physical activity could be accounted for in the analyses. Importantly, these studies were either conducted in a single region or controlled for region so the variation in 25(OH)D levels was completely independent of region. Thus, these results can be considered as completely independent corroborating evidence from the studies based on regional solar UVB level. A limitation is that, although the t1/2 in the circulation of ∼2–3 wk is fairly reasonable, studies have been based on a single measurement throughout the year, and the correlation with long-term (e.g., over decades) vitamin D status is unclear. In addition, uncontrolled confounding, while less likely than in the ecological studies, is still a possibility.


Studies that examine regional solar UVB are in essence examining one of the multiple determinants of individual variation in circulating 25(OH)D. A more comprehensive look at multiple determinants of 25(OH)D was taken recently using data from the Health Professionals Followup Study.math image In a sample of 1095 men in this cohort of health professionals who had circulating 25(OH)D levels measured, the authors used geographical region, skin pigmentation, dietary intake, supplement intake, body mass index, and leisure time physical activity (a surrogate of potential exposure to sunlight UVB) to develop a predicted 25(OH)D score. The score, which can be interpreted as an estimate of 25(OH)D level, was calculated for each of the >47,000 cohort members. This score was examined prospectively in relation to subsequent risk of cancer incidence and mortality. Based on multivariable models, a 25-nM increment in predicted 25(OH)D score was associated with a 37% statistically significant lower risk of colorectal cancer. Other cancers of the digestive system were also reduced in incidence.


Numerous case-control and cohort studies have examined vitamin D intake in relation to risk of colorectal cancer or adenoma. These studies have been reviewed in detail and have generally shown an inverse association between vitamin D intake and risk of colorectal cancer or adenoma.math image Many of the studies controlled for known or suspected risk factors for colorectal neoplasia, although the independent effects of vitamin D and calcium intakes may be difficult to separate entirely. The magnitudes of the risk reductions have been relatively modest, ∼20–30% reductions, so uncontrolled confounding cannot be ruled out. Studies in the United States have generally yielded stronger associations than studies in European populations, where vitamin D from supplementation and fortification is generally much less and thus the range of intake is low. Even with added vitamin D from supplementation and fortification, vitamin D intake at typical levels currently do not raise 25(OH)D levels substantially, and most variability in populations comes generally from sun exposure. Even in a country such as Japan with relatively high intake of vitamin D from fatty fish, solar UVB is an important source of vitamin D. For example, in the Tokai area, a relatively sunny region, the prevalence of hypovitaminosis D (defined as <20 ng/ml) was 86.7%, 33.4%, 1.0%, and 26.0% in March, June, September, and December, respectively.math image

To date, there has been one adequately sized randomized-controlled trial of vitamin D intake and colorectal cancer risk, the WHI. This WHI was a recent randomized placebo-controlled trial of 400 IU vitamin D plus 1000 mg/d of calcium in 36,282 postmenopausal women. This study did not support a protective role of these nutrients over a follow-up of 7 yr.math image However, the vitamin D supplement was of relatively low dosage, and it probably was inadequate to produce a substantial contrast between the treated and the control groups. The expected increase of 25(OH)D for an increment of 400 IU/d would be ∼3 ng/ml. Moreover, the adherence was suboptimal, and some women took nonstudy supplements, so the true contrast of 25(OH)D in the intent-to-treat analysis is unclear. In contrast, in the observational studies of 25(OH)D, the difference between high and low quintiles was generally 20 ng/ml (50 nM) or higher.math image Finally, the epidemiological data on duration, although limited, indicated the effects of calcium and vitamin D intakes may require at least 10 yr to emerge.math image Thus, this randomized trial was probably inadequate in testing the vitamin D colorectal cancer hypothesis.


In Norway, all cancer diagnoses since 1953 have been registered in the Cancer Registry. Norwegian investigators examined the influence of latitude and season of diagnosis on survival from colon cancer.math image The analysis included 12,823 men and 14,922 women with colon cancers diagnosed between 1964 and 1992. Latitudinal and season variation and survival rate were assessed. No significant annual variation in the incidence rates of colon cancer was found, suggesting that there was no seasonal bias in the diagnosis of cancers. The death rates at 18, 36, and 45 mo were 20–30% lower in the cancers diagnosed in autumn months compared with those diagnosed in the winter months. No significant north–south gradient was found for the death rate for colon cancer. However, there may be minimal variation of 25(OH)D by latitude in these countries. These findings indicate that high sun exposure during the recent time period at the time of diagnosis may improve ultimate survival from colon cancer. What explains this association is unclear. In the late summer in Norway, the serum concentration of 25(OH)D is ∼50% higher than that in late winter. Effects of vitamin D in late carcinogenesis stages such as reduction in metastases are observed in numerous animal models. In some animal studies, vitamin D may improve tumor control by radiation treatment, possibly by promoting apoptosis.math image These results are provocative, but alternative explanations to an effect of vitamin D are possible, such as seasonal intake of other micronutrients related to fruits and vegetables. This finding should be considered hypothesis-generating but suggests the possibility that if vitamin D status is improved at the time of treatment, an important effect on survival could ensue.


A significantly higher mortality rate from cancer has been documented in United States blacks compared with whites.math image Whereas differences in incidence may contribute to the higher mortality rate, blacks experience a more advanced stage at presentation and poorer survival across all stages for many cancer sites.math image A number of explanations have been offered, including biological or genetic differences, unequal access or use of medical care, and differences in lifestyle and diet. Vitamin D deficiency may also be important. Blacks have markedly lower vitamin D levels caused primarily by higher melanin in their skin.math image Poor vitamin D status has been hypothesized to account for higher prostate cancer rates,math image more aggressive prostate and breast cancer,math image and higher total cancer incidence and mortalitymath image in blacks. Furthermore, an inverse association between regional solar UVB radiation and mortality rate of breast, colon, esophageal, and gastric cancers was shown for blacks in the United States.math image

The Health Professionals Follow-Up Study consists of highly educated, generally health conscious male health professionals. In this cohort, even after adjusting for multiple dietary, lifestyle, and medical risk factors, black men were at 32% higher risk of total cancer incidence and 89% higher risk of total cancer mortality; the relative risk (RR = 2.24) was especially high for digestive system cancer mortality (including colon, rectum, oral cavity, esophagus, stomach, and pancreas).math image In addition, compared with whites with relatively few risk factors for hypovitaminosis D, blacks also with few risk factors for hypovitaminosis D were not at appreciably higher risk of total cancer incidence (RR = 0.95) or mortality (RR = 1.55), but black men with additional risk factors for poorer vitamin D status had a higher cancer incidence (RR = 1.57) and cancer mortality risk (RR = 2.27). Furthermore, blacks with few risk factors for hypovitaminosis D were not at appreciably higher risk of colorectal cancer incidence (RR = 0.71), but black men with additional risk factors for poorer vitamin D status had an increased risk (RR = 2.43). Our results suggest that the high frequency of hypovitaminosis D in blacks may be an important, and easily modifiable, contributor to their higher risk of cancer incidence and mortality, particularly for cancer of the large bowel and other digestive system organs.


Since Garland and Garland formulated the hypothesis that vitamin D may protect against colorectal cancer in 1980, various approaches have been undertaken to evaluate this hypothesis. Geographical residence contributes to variation in vitamin D levels, but skin pigmentation, sunlight exposure behaviors, and adiposity also contribute, in addition to diet and supplements. Intake generally contributes only to a modest proportion of total vitamin D but may be important in some populations. Regions with higher solar UVB exposure generally have lower rates of colorectal cancer. Higher intake of vitamin D has been associated with a moderately lower risk of colorectal cancer, consistent with the magnitude of the influence that typical vitamin D intakes would have on circulating 25(OH)D levels. Higher circulating 25(OH)D has generally been associated with a markedly lower risk of colorectal cancer. In one study, a predicted 25(OH)D score computed from various factors that influence 25(OH)D levels was associated with a lower risk of colorectal cancer. Finally, one study in Norway indicate that patients who are diagnosed and treated in the sunnier months, where vitamin D levels are much higher, may have a better prognosis from colorectal cancer.

All these data do not prove a benefit of vitamin D on colorectal cancer, but they are strongly suggestive of such an effect. Confounding cannot be ruled out using observational (nonrandomized data), but it would seem unlikely that confounding could entirely account for all these associations. For example, living in a sunny region, having a higher vitamin D intake, higher 25(OH)D level controlling for season and diet, lighter pigmented skin, and season of diagnosis in the sunnier months are all associated with a lower risk or better prognosis. These factors are not inherently correlated, and thus multiple uncontrolled confounding factors would have to be posited. The hypothesis is also buttressed by a large body of in vitro animal and clinical studies that indicate that vitamin D may have anticancer benefits against colorectal cancer and possibly other cancers.

Confirming the hypothesis that improving vitamin D status could be potentially beneficial against either incidence or mortality, or both, is critical and timely. Among those at high risk for vitamin D deficiency are individuals with low intakes, those who live in regions with low sunlight intensity, those who avoid sunlight or thoroughly use sunscreen, darker-skinned individuals, the elderly, those who live in nursing homes, and those who are overweight or obese. In essence, many people could be at risk for low vitamin D, even in relatively sunny climates.math image The current recommended intakes of vitamin D (200–600 IU/d depending on age) may be suboptimal for many if not most people. Given the potential benefits of vitamin D against colorectal cancer, and possibly other conditions, higher intakes may be warranted. Further research should be a top priority. Randomized interventional studies may be required, but obstacles abound as shown in the recent WHI trial. A randomized intervention of adenomas is ongoing (J Baron, personal communication). A randomized trial of providing high doses of vitamin D to boost 25(OH)D levels at the time of treatment in diagnosed cases should be quite feasible and safe and could provide an answer regarding the potential benefit of vitamin D on prognosis.