|Home | About | Journals | Submit | Contact Us | Français|
Although laboratory studies linked zinc and heme iron to colorectal cancer, epidemiologic evidence is limited. We prospectively examined these associations in the Nurses’ Health Study and Health Professionals Follow-up Study. We used Cox proportional hazards regression analyses to calculate cohort-specific relative risks (RRs) and pooled results using a fixed effects model. We documented 2114 incident colorectal cancer cases during up to 22 years of follow-up. Compared highest to lowest quintile of dietary zinc intake, the pooled multivariable RRs (95%CIs) were 0.86 (0.73, 1.02) for colorectal cancer, 0.92(0.76, 1.11) for colon cancer, and 0.68 (0.47, 0.99) for rectal cancer. The significant inverse association between dietary zinc intake and risk of rectal cancer was mainly driven by data in women, although the difference in the sex-specific results was not statistically significant. For the same comparison, the pooled multivariable RRs (95%CIs) for heme iron were 1.10 (0.93, 1.30) for colorectal cancer, 1.06 (0.88, 1.29) for colon cancer, and 1.20 (0.83, 1.75) for rectal cancer. These associations were not significantly modified by alcohol consumption, body mass index, physical activity, menopausal status or postmenopausal hormone use. Total zinc intake, total iron intake, dietary iron intake, zinc or iron supplement uses were largely not associated with colorectal cancer risk. Our study does not support strong roles of zinc and heme iron intake in colorectal cancer risk, however, a suggestive inverse association of dietary zinc intake with rectal cancer risk in women requires further study.
Colorectal cancer is the third most common type of cancer in both men and women worldwide1. The positive association reported in most studies between insulin/C-peptide2,3 levels and type 2 diabetes4 and colorectal cancer risk suggest a possible role of hyperinsulinemia, insulin resistance, and insulin-like growth factor 1 (IGF-1) in colorectal carcinogenesis3,5. Zinc, an essential trace element with antioxidant properties, is vital for various cellular functions and as early as 1930, zinc was shown as an integral element of the insulin crystalline structure6. Subsequently, rodent models showed that zinc supplementation attenuated hyperinsulinemia in db/db mice7. In addition, zinc supplementation for 4 weeks significantly improved insulin sensitivity in obese women8. More recently, zinc intake was inversely associated with risk of type 2 diabetes among women in the Nurses’ Health Study9. Despite the potential beneficial effect of zinc on risk of colorectal cancer, the hypothesis that dietary zinc intake is associated with a lower risk of colorectal cancer has been examined in only two prospective studies in women; one study reported a significant10 inverse association whereas the other found a null11 association between dietary zinc intake and colon cancer risk. In the former study, the inverse association was markedly stronger among alcohol drinkers10.
In contrast to zinc, redox active iron is a pro-oxidant that may influence colorectal carcinogenesis by forming reactive oxygen species12. However, epidemiological studies of colorectal cancer risk in relation to iron intake or markers of body iron stores have been mixed10,11,13–19. Emerging evidence suggests that heme iron may play a more important role in colorectal carcinogenesis than other forms of iron20,21. Heme iron is mainly found in animal foods, whereas non-heme iron is present mainly in plant foods22. In general, heme iron is more bioavailable than other forms of iron and is the form in which iron is stored in the body23. Further, heme iron, but not non-heme, has been shown to have cytotoxic and hyperproliferative effects in rats21 and to increase formation of endogenous intestinal N-nitroso compounds24, established human carcinogens. We thus focused on heme iron in addition to examining other iron related variables in our analyses. To date, five studies10,11,14,18,25 have examined the association between heme iron intakes and colon or colorectal cancer risk and the results have been mixed. Two of these studies reported a stronger positive association with heme iron intake among alcohol drinkers10,11.
We used the dietary data collected every 2–4 years from the US female Nurses’ Health Study (NHS) and the US male Health Professionals Follow-up Study (HPFS) to examine whether zinc intake is associated with a lower risk and heme iron intake is associated with an increased risk of colorectal cancer. We further tested whether these associations differ by alcohol consumption because associations with zinc and heme iron intake have been more pronounced among alcohol drinkers in some previous studies10,11,18.
The NHS26 is an ongoing cohort study established in 1976 including 121,700 married female registered nurses at baseline who were 30 to 55 years old and resided in 11 states in the US The HPFS26 is an ongoing cohort study of 51,529 US male professionals who were aged 40 to 75 years at baseline in 1986. Questionnaires have been mailed to participants in both studies every 2 years since baseline to collect updated information on demographics, lifestyle factors, medical history, and disease outcomes. The follow-up rate has been greater than 90% in both cohort studies. These studies have been approved by the institutional review board at the Brigham and Women’s Hospital, Boston, Massachusetts.
Cancer and other disease outcomes were reported by the participants in both cohorts in the biennial questionnaires. Researchers received permission from the study participants or next of kin to obtain their medical records and pathological reports on colorectal cancer and, while blinded to exposure information, abstracted the information on anatomic location, stage, and histological type of the cancer. Colon cancer was further classified into proximal colon cancers (neoplasms from the cecum to the splenic flexure) and distal colon cancers (neoplasms in the descending and sigmoid colon). Rectal cancer was defined as that occurring in the rectosigmoid or rectum28.
Information on usual dietary intake over the past year was obtained using a validated semi-quantitative food frequency questionnaire (FFQ) in NHS participants in 1980, and repeated in 1984, 1986, and every 4 years thereafter. A total of 61 food items were included in the 1980 FFQ; the number of food items expanded to about 130 in 1984 and in subsequent FFQs. Similar FFQs were administered for men in 1986 and repeated every 4 years thereafter. Nine possible frequency choices were available, ranging from “almost never” to “6 or more times per day”. Nutrient intakes were calculated by multiplying the frequency of each food consumed and the nutrient content of specified portion sizes. Zinc or iron supplement use refers to zinc or iron from multivitamins and supplements in this study. Total intake of zinc and iron was calculated by summing up the amounts from food and supplemental sources. Nutrient intakes, including total and dietary zinc and iron, were adjusted for total energy intake using the residual method26. Because zinc or iron supplement use was not included in the 1980 food frequency questionnaire in the NHS, we thus treated 1984 as the baseline for these analyses in NHS. The validity of the FFQs has been evaluated in 173 women from the NHS30 and in 127 men from the HPFS31. The energy-adjusted Pearson correlation coefficients for total zinc intake comparing the FFQ and the average of multiple 1-week diet records (four for women and two for men) were 0.63 for women30,32 and 0.71 for men31. The Pearson correlation coefficients between FFQ and multiple dietary records for total iron were 0.55 for women33 and 0.54 for men31. Although dietary zinc and heme iron intakes were not evaluated in the validation study, we have evaluated the major contributors of dietary zinc and heme iron intakes in our validation studies34,35. The top 5 foods contributing approximately 60–70% of dietary zinc and heme iron intakes included red meat (beef, pork, lamb), cold breakfast cereal, chicken without skin, skim milk, and hamburgers in both women and men. The Spearman correlation coefficients between the FFQs and the dietary records ranged from 0.38 for hamburger to 0.81 for skim milk in women35 and ranged from 0.56 for chicken without skin to 0.88 for skim milk in men34..
In the baseline and biennial questionnaires, we inquired about colorectal cancer risk factors such as body weight, physical activity, cigarette smoking, family history of colorectal cancer, and aspirin use. In women, information on menopausal status and postmenopausal hormone use was also obtained at baseline and in the subsequent follow-up questionnaires.
We excluded participants with a history of cancer (except nonmelanoma skin cancer) and ulcerative colitis at baseline. In addition, we excluded participants with unreasonable baseline total energy intake (< 600 or > 3500 kcal/d for women and < 800 or > 4200 kcal/d for men). We calculated person-time for each participant from the date of baseline questionnaire return to the date of death, loss to follow-up, colorectal cancer diagnosis, or the end of follow-up (June 1, 2006 for NHS and January 1, 2006 for HPFS), whichever came first. We used a Cox proportional hazards model36 to calculate relative risks (RRs) and 95% confidence intervals (CIs) and adjusted simultaneously for age (in months) and year of questionnaire return using SAS PROC PHREG37. We observed no violation of the proportional hazard assumption based on the likelihood ratio test that compared the model with and without the interaction terms between zinc or heme iron intakes and age or follow-up time. All statistical analyses were two-sided and a P-value less than 0.05 was considered statistically significant. We conducted all analyses using the SAS software (SAS Institute, Inc., Version 9, Cary, NC).
For each cohort, energy-adjusted dietary zinc and heme iron intakes were categorized into quintiles based on the distribution in that study population. We used median values of these categories and entered these values as continuous variables into the model to conduct trend tests. In addition to age and year of questionnaire return, multivariable analyses also adjusted for established or potential risk factors for colorectal cancer (see Table 2 for the categorization of confounding variables). Given that vitamin C and phytate intakes might influence zinc and iron absorption, we conducted sensitivity analyses where we further adjusted for vitamin C intake (quintiles), and phytate intake (quintiles). Results were essentially unchanged. We modeled all covariates as time-varying variables to take into account potential changes over follow-up. To represent better long-term dietary intake38 and to minimize the impact of random measurement errors in the dietary assessment, we analyzed cumulative average nutrient intakes. In addition to dietary zinc and heme iron intake, we further analyzed total zinc intake (quintiles), zinc from supplements (nonuser, user), total iron (quintiles), dietary iron (quintiles), and iron from supplements (nonuser, user). Given that there were very few cases (<20) in the category exceeding the Tolerable Upper Intake Levels (UL, 40 mg/day for zinc and 45 mg/d for iron), we used approximately half of the UL (i.e., ≥25 mg/day) to further evaluate associations with relatively high zinc or iron supplement use to detect any potential harmful effect.
We investigated whether associations between differed between women and men using the Q statistic39,40, which follows an approximate χ2 distribution with 1 degree of freedom. We pooled the sex-specific results using a fixed effects model40.
For colorectal cancer and each subsite, we further evaluated whether the associations with intakes of dietary zinc and heme iron differed by alcohol consumption (non-drinkers, >0–<10, ≥10 g/day), body mass index (<25, ≥25 kg/m2), physical activity (low, high), or zinc or iron supplement use (zero, low, high). For these analyses, we categorized dietary zinc and heme iron intakes into tertiles to have a reasonable number of cases in each stratum. In addition, we further examined the potential interactions between zinc and heme iron intakes because some experimental studies41,42 have suggested that these two micronutrients may compete with each other due to their similar chemical properties. We used a Wald test to examine whether the beta coefficients for the cross-product terms between each of these variables and dietary zinc or heme iron intake were statistically significant.
In the NHS, 1,079 incident colorectal cancer cases were documented among 69,345 women during 1,434,574 persons-years from 1984 to 2006. In HPFS, 1,035 colorectal cancer cases were identified among 45,716 men during 804,210 person-years from 1986 to 2006. Intake levels of dietary zinc and heme iron were slightly higher in men than in women. In both men and women, participants in the highest quintile of intake were generally comparable with the participants in the lowest quintile of intake of dietary zinc and heme iron with respect to age, body mass index, family history of colorectal cancer, regular aspirin use, and total vitamin D intake. Participants with low dietary zinc intake drank more alcohol but consumed less total calcium and total folate than participants with higher intakes. In contrast, men with low heme iron intake consumed more total calcium and total folate, smoked less, and were more likely to be physically active (Table 1). Use of supplements containing zinc (including both multivitamin and zinc-specific supplements) increased over study time in both men and women. For example, the proportion of individuals using of zinc-containing supplements increased from 17% in 1990 to 52% in 2004 in women and increased from 18% to 44% in men for the same time frame. Use of iron-containing supplements increased slightly from 1990 to 2004 in both women (from 20% to 26%) and men (from 14% to 20%).
Table 2 shows the associations of dietary zinc intake and total zinc intake with risk of colorectal cancer. Because the age-adjusted results were similar to the multivariable-adjusted results, we only present the multivariable-adjusted results. A suggestive inverse association with dietary zinc intake was observed in women. Comparing the highest to the lowest quintile of dietary zinc intake, multivariable RRs (95%CIs) were 0.75 (0.59, 0.95; P trend=0.02) for colorectal cancer, 0.84 (0.64, 1.09; P trend=0.18) for colon cancer, and 0.51 (0.29, 0.86; P trend=0.01) for rectal cancer. Results for proximal and distal colon cancer were non-significant (multivariable RR=0.80; 95%CI: 0.57, 1.12 for proximal colon cancer and RR=0.84; 95%CI: 0.54, 1.30 for distal colon cancer for the extreme quintile comparison). Associations with total zinc intake were non-significant and weaker than those observed for dietary zinc intake (Table 2). For men, results of dietary and total zinc intakes were largely null (Table 2). The pooled multivariable RRs (95%CIs) comparing the highest to lowest quintile of dietary zinc intake for women and men combined were 0.86 (0.73, 1.02) for colorectal cancer, 0.92(0.76, 1.11) for colon cancer, and 0.68 (0.47, 0.99) for rectal cancer (Table 2). Again, the significant inverse association between dietary zinc intake and risk of rectal cancer was mainly driven by data in women, although the difference in results between men and women was not statistically significant. Further, zinc supplement use (user vs non user; data not shown) or relatively high dose of zinc supplement use (≥25 mg/day) was not significantly associated with risk of colorectal cancer in either men or women (data not shown).
Table 3 shows the associations of heme iron intake or total iron intake with risk of colorectal cancer. A non-significant positive association between heme iron intake and risk of colorectal cancer was observed in women. Comparing the highest quintile to the lowest quintile of intake, the multivariable RRs (95%CIs) were 1.21 (0.96, 1.52; P trend=0.10) for colorectal cancer, 1.13 (0.87, 1.47; P trend=0.30) for colon cancer, and 1.50 (0.90, 2.49; P trend=0.17) for rectal cancer (Table 3). For the same comparison, weaker non-significant positive associations were observed for total iron intake. For men, both heme iron and total iron intakes were not associated with risk of colorectal cancer or any subsite. Non-significant between-studies heterogeneity was observed and the pooled RRs for women and men combined were largely around the null value 1 (Table 3). In addition, intakes of heme iron and total iron were not associated with risk of either proximal colon or distal colon cancer (data not shown). Moreover, dietary iron intake and iron supplement use were not significantly associated with risk of colorectal cancer overall or with any of the subsites (data not shown). We did not adjust for red meat consumption in our main analysis because red meat is a major contributor to dietary zinc and heme iron in our study. However, in secondary analysis, to investigate whether the associations with dietary zinc and heme iron are independent of the constitutes in red meat, we further adjusted for red meat and processed meat, both modeled as quintiles. The results were essentially unchanged (data not shown). In addition, results were essentially the same after including dietary zinc and heme iron intake in separate models. Consistent with the results for heme iron intake observed in women, suggestive positive associations were observed for women with relatively high iron supplemental use, especially for rectal cancer (i.e., for ≥25 mg/day vs non-user, RR=2.54,95%CI:1.43,4.50; n=15 cases). For the same comparison, the associations were not statistically significant in men (RR=0.41, 95%CI: 0.13, 1.32; n=3 cases). For both zinc and iron intakes, comparing the highest with the lowest decile, the associations for colorectal cancer and subsites were similar in magnitude as those presented in Table 2 and Table 3 for the quintile analyses.
Because a previous study reported that the inverse association between zinc intake and colorectal cancer risk was markedly stronger among alcohol drinkers10, we specifically evaluated the interaction between dietary zinc intake and alcohol consumption (Table 4). We also examined the joint associations of dietary zinc (or heme iron) intake with supplemental zinc (or iron). For zinc, we found no statistically significant interactions, though the inverse association of dietary zinc intake with risk of rectal cancer was observed mainly among non-drinkers (Table 4). Zinc supplement use did not significantly modify the association between dietary zinc intake and colorectal cancer risk. Associations of heme iron intake with colorectal cancer risk were not significantly modified by alcohol consumption or iron supplement use; however,, among alcohol drinkers (10 g/day), a small elevated risk of colorectal or colon cancer was observed for heme iron intake but the interaction test was not significant. Suggestive stronger positive associations of rectal cancer were observed for women (RR=2.37, 95%CI: 1.33, 4.21) who were in the highest tertile of heme iron intake with high iron supplement use but not in men (RR=0.28, 95%CI: 0.07, 1.21) but these results might be due to chance because of the smaller number of cases in these strata (Table 4). Further, associations were not significantly modified by postmenopausal hormone use (never vs ever; in women only), menopausal status (premenopausal vs postmenopausal in women only), body mass index (< 25 vs ≥ 25 kg/m2), or physical activity (low vs high) (data not shown). Further, no significant interaction was observed between zinc and heme iron intakes (data not shown).
In these large prospective cohort studies of men and women, our results did not support a strong role of zinc or iron, whether from food sources only or including supplements, in colorectal carcinogenesis. However, a modest inverse association of dietary zinc intake with risk of rectal cancer in women cannot be excluded. In addition, these associations were not significantly modified by alcohol consumption or several other colorectal cancer risk factors.
Several mechanisms have been proposed to support a role of zinc in colorectal carcinogenesis. In addition to the potential effect of zinc on insulin, animal models have suggested a role of zinc to protect against chemically induced colonic proneoplastic progression43. Further, zinc has been shown in experimental studies to play an important role in antioxidant defense system, DNA synthesis, and immune function44. Despite the experimental evidence, only two epidemiological studies to date have examined the association between zinc intake and colon cancer risk and both were in women. The Swedish study10 reported a weak non-significantly lower risk of colon cancer (highest vs lowest quintile, RR=0.90, 95%CI: 0.65, 1.25, P trend=0.71). However, they had a relatively limited range in intake to examine this association (≥ 11.1 vs < 9.0 mg/day). In contrast, the study from Iowa11 had a wider range across quintiles of dietary zinc intake (≥ 17.6 vs ≤ 8.5 mg/day). Although associations with dietary zinc intake for colon cancer were not reported, they did observe a significantly lower risk of proximal colon cancer (highest vs lowest quintile, RR=0.38, 95%CI: 0.17, 0.74, P trend=0.01) and a non-significantly lower risk of distal colon cancer (highest vs lowest quintile, RR=0.58, 95%CI: 0.26, 1.30, P trend=0.15). We had a reasonable contrast in zinc intake (≥ 13.4 vs ≤ 8.7 mg/day) but we observed a non-significant lower risk for colon cancer (highest vs lowest quintile, RR=0.84; 95%CI: 0.64, 1.09) with no difference by colon cancer subsite. Although results were not reported in these earlier studies for rectal cancer, we observed a significant lower risk for rectal cancer (highest vs lowest quintile, RR=0.51, 95%CI: 0.29, 0.86) in women. Reasons for the inconsistent results observed in these few observational studies are unclear and the difference in ranges across quintiles of dietary zinc intake might partly explain the inconsistencies in the results. In addition, food sources of zinc differ between the US and Sweden. In the US diet, the major food sources of dietary zinc are meat, dairy foods, and fortified cereals while the primary sources in Sweden are mainly grains and red meat45. Zinc from meat is more bioavailable than zinc from plant sources46 and plant sources of zinc such as whole grains and beans contain phytates, which can inhibit zinc absorption46. However, further adjustment for phytate intake did not change the results in our study. Nonetheless, future studies in different populations with a wide range of zinc intake should help elucidate the of zinc intake on colorectal carcinogenesis.
Zinc supplement use was not associated with colorectal cancer risk in our study, which was consistent with the study from Iowa10. The mean dietary zinc intake in our population was higher than the US recommended dietary allowance, which is 8 mg/day for women and 11mg/day for men, and relatively few participants had high zinc supplement intake (i.e., ≥25 mg/d) or intakes less than 8mg/d, which limited our ability to examine associations with very high or low intake. Taken together, no firm conclusion regarding zinc intake and colorectal cancer incidence can be drawn based on the current limited epidemiological evidence.
In contrast to zinc, heme iron has been hypothesized to be associated with an increased risk of colorectal cancer. To date, five cohort studies have examined this hypothesis10,11,14,18,25. In the Iowa Women’s Health Study, heme iron intake was positively associated with risk of proximal (for ≥ 2.05 vs ≤ 0.76 mg/day, RR=2.18, 95%CI: 1.24, 3.86, P trend=0.01) but not with risk of distal colon cancer (RR=0.90, 95%CI: 0.45, 1.81, P trend=0.77)10. The association of proximal colon cancer with heme iron intake was particularly strong among women who consumed at least 10 grams of alcohol per day (RR=7.20, 95%CI: 1.33, 38.91, P trend=0.03). Similarly, a study from Sweden reported a positive association with heme iron intake among women drinking at least 20 grams of alcohol per week (≥ 2.06 vs ≤ 0.67 mg/day, RR=2.29, 95%CI: 1.25, 4.21, P trend=0.007)11. In contrast, in a study of women and men from the Netherlands, no association between heme iron intake and colon cancer risk was observed, but there was a suggestion of a positive association among women who drank at least 5 grams of alcohol per day (tertile 3 vs tertile 1, RR=1.50, 95%CI: 0.95, 2.36)18. The NIH-AARP study of both men and women suggested a positive association between heme iron intake and colorectal cancer risk (highest vs lowest quintile, RR=1.13, 95%CI: 0.99–1.29, P trend = 0.02) although whether the association was modified by alcohol consumption was not reported25. A Canadian study of women showed null results for heme iron intake (>2.95 vs <1.58 mg/day; RR=0.99, 95%CI:0.70, 1.40) and the associations did not differ by the amount of alcohol consumed14. A recent meta-analysis of these five studies suggested a modest positive association between heme iron intake and colon cancer risk (highest vs lowest quintile, RR=1.18, 95%CI: 1.06, 1.32)20. Our findings of non-significant associations for heme iron intake are somewhat consistent with the early two studies14,18 but not the others10,11,25. The narrower comparison across quintiles may partly explain the non-significant positive associations we observed. The stronger associations observed among alcohol drinkers in two of the previous studies10,11 may also be due to chance because of the relatively small number of cases in the highest quintile of heme iron and highest alcohol consumption categories (< 50 cases). Future studies in populations with a wide range of heme iron intakes should help confirm these findings or identify any potential threshold effect of heme iron. Notably, the significant positive associations between relatively high iron supplemental use (i.e., ≥ 25 mg/day vs none) and rectal cancer risk in women were unexpected. Given that this subgroup finding is based on a relatively small number of cases, the results might be due to chance. On the other hand, we cannot rule out the possibly that iron intake may influence the rectal cancer.
Limitations of our study need consideration. Our study had relatively narrow quintile intake ranges of zinc and heme iron, which limited our ability to evaluate associations with these micronutrients for high or low intake ranges (i.e., zinc deficiency). We lacked information on cooking methods, which might impact the amount of heme iron because heme iron can be partially converted to non-heme iron depending the type and extend of the cooking method47. The suggestive inverse associations between dietary zinc intake and rectal cancer risk, particularly in women, might be due to chance because of the relatively small number of rectal cancer cases. We cannot exclude the possibility that residual confounding may explain the suggestive inverse associations observed with dietary zinc intakes although we adjusted for multiple lifestyle and dietary confounders. In addition, our study population consisted of mainly Caucasians and the results may not be generalizable to other ethnic groups with different dietary patterns. Our study has several strengths, including its large size, prospective design, long follow-up time, and the comprehensive updated measurements of zinc and iron intakes.
In summary, although a beneficial effect of zinc and a detrimental effect of heme iron on colorectal carcinogenesis have been suggested by some experimental evidence, our study did not support a strong role of zinc or heme iron intake in colorectal cancer risk. However, a potential modest inverse association with dietary zinc intake and rectal cancer risk in women requires further study.
We are grateful for all participants in the Nurses’ Health Study and Health Professionals Follow-up Study for their continuous contribution and support. This work was supported by the NIH grants CA87969 and CA55075.
Conflict of interest: None. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.