Dietary supplement use is not associated with recurrence of colorectal adenomas: A prospective cohort study
Diet and lifestyle influence colorectal adenoma recurrence. The role of dietary supplement use in colorectal adenoma recurrence remains controversial. In this prospective cohort study, we examined the association between dietary supplement use, total colorectal adenoma recurrence and advanced adenoma recurrence. Colorectal adenoma cases (n = 565) from a former case–control study, recruited between 1995 and 2002, were prospectively followed until 2008. Adenomas with a diameter of ≥1 cm and/or (tubulo)villous histology and/or with high grade dysplasia and/or ≥3 adenomas detected at the same colonic examination were considered advanced adenomas. Hazard ratios (HRs) and 95% confidence intervals (CIs) for dietary supplement users (use of any supplement during the past year) compared to nonusers and colorectal adenoma recurrence were calculated using stratified Cox proportional hazard models for counting processes and were adjusted for age, sex, educational level and number of colonoscopies during follow-up. Robust sandwich covariance estimation was used to adjust for the within subject correlation. A number of 165 out of 565 adenoma patients had at least one colorectal adenoma recurrence during a median person-time of 5.4 years and of these, 37 patients had at least one advanced adenoma. One-third of the total study population (n = 203) used a dietary supplement. Compared to no use, dietary supplement use was neither statistically significantly associated with total colorectal adenoma recurrence (HR = 1.03; 95% CI 0.79–1.34) nor with recurrent advanced adenomas (HR = 1.59; 95% CI 0.88–2.87). This prospective cohort study did not suggest an association between dietary supplement use and colorectal adenoma recurrence.
Dietary supplement use is rising in the Western world.1 More than half of the American population uses dietary supplements,1 and in Europe a wide variation in use is present.2 Supplement use appears to be higher among women,1 more educated people1, 3 and people with a healthier lifestyle.3 Despite limited scientific support for the efficacy of dietary supplements, frequent supplement users take dietary supplements as they believe it may be beneficial in the treatment of acute or chronic illnesses4 or for the prevention or recurrence of a serious disease, such as cancer.1
Colorectal cancer is one of the most common types of cancer in the Western world.5 It is often proceeded by colorectal adenomas,6 a precancerous and generally asymptomatic condition.7 Individuals with multiple and recurrent colorectal adenomas are at higher risk for colorectal cancer.8 Approximately 15–40% of postpolypectomy patients will develop recurrent colorectal adenomas within 3 years.9, 10 A history of colorectal adenomas,9 a higher number of prior colorectal adenomas,9 a family history of colorectal cancer,11 multiple adenomas at the same colonic examination,10 an adenoma size of 10 mm or larger8, 12 and having adenomas with villous features or high-grade dysplasia12 are positively associated with colorectal adenoma recurrence.
As dietary supplement use increases in countries where colorectal cancer is prevalent,5 it is important to provide recommendations for use among those with a high risk of colorectal cancer, that is, those with recurrent colorectal adenomas. Dietary supplement use in relation to the occurrence of a first colorectal adenoma has been extensively studied,13–17 while the role of dietary supplements in the recurrence of colorectal adenomas has been studied less comprehensively.18–23 For specific supplements, there is substantial evidence to support or refute a role in colorectal adenoma recurrence: meta-analyses showed that calcium supplements significantly reduced the risk of recurrent adenomas,24 while no protection was observed for folic acid supplements against recurrence of colorectal adenomas.25 Despite the fact that multivitamins are abundantly used,26 no prospective studies on the association of multivitamin use and colorectal adenoma recurrence have been published. One case–control study showed an inverse association between the use of multivitamin supplements and colorectal adenoma recurrence,27 but no marked association was found in another case–control study.28
We examined the association between the most frequently used dietary supplements, including multivitamin use, and colorectal adenoma recurrence by conducting a prospective cohort study among those with a history of colorectal adenomas in the Netherlands. Moreover, we investigated whether the association differed for colorectal adenomas with advanced pathology.
Material and Methods
This prospective cohort study originates from an endoscopy-based case–control study on risk factors for colorectal adenomas. In the original case–control study, participants were recruited among those undergoing endoscopy of the large bowel in ten outpatient clinics in The Netherlands between June 1997 and June 2002. Further details about the original case–control study have been described elsewhere.29
Colorectal adenoma patients were prospectively followed in our study. All patients had at least one histologically confirmed colorectal adenoma ever in their life, and were Dutch speaking, of European origin, aged 18–75 years at time of the recruitment endoscopy, were not suspected to have hereditary colorectal cancer syndromes (i.e. Lynch Syndrome, familial adenomatous polyposis coli, Gardners syndrome), did not suffer from inflammatory bowel disease and did not have a history of colorectal cancer or (partial) bowel resection. Of the 768 colorectal adenoma patients in the original case–control study, 143 adenoma patients were not included into our study because they did not have endoscopies during follow-up. Furthermore, we were not able to retrieve medical information of 50 colorectal adenoma patients. Seven patients had to be excluded due to surgery for colorectal neoplasms detected at recruitment, and three patients became ineligible due to having a histological unconfirmed adenoma only, being diagnosed with proctitis ulcerosa or being entered twice into the study. Overall, 565 colorectal adenoma patients were included in our study and were prospectively followed. All participants gave informed consent. The Medical Ethics Committee of Radboud University Nijmegen Medical Centre in the Netherlands approved the study.
Assessment of diet, dietary supplement use and lifestyle factors
Upon recruitment, patients filled out a standardized and validated semiquantitative food frequency questionnaire to report their habitual dietary intake.30, 31 Dietary supplement use in the year previous to the last endoscopy or bowel complaints was assessed by a self-administered questionnaire. Participants were asked to indicate their use of dietary supplements in general and to report use and frequency of intake (per day, per week, per month or per year) of multivitamins, B-vitamins, vitamin A, vitamin C, vitamin E, vitamin D, calcium, iron and garlic supplements. Dosage was only asked for vitamin C and vitamin E supplements. Participants were classified as users of dietary supplements if they reported to take any dietary supplement during the past year. When patients took no dietary supplements at all during the past year, they were considered as nonusers. General lifestyle information was collected through a lifestyle questionnaire containing questions about age, sex, weight, height, smoking habits, medication use, physical activity, family history of colorectal cancer and medical history.
Colorectal adenoma recurrence
Information on medical history and recurrence of colorectal adenomas was gathered via medical records from hospital registries just after recruitment and between 2007 and 2009. From each colonic examination, data on adenoma type (villous, tubulovillous, tubular, serrated), adenoma size (mm), adenoma location (cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid, rectum or overlapping location in the bowel) and the number of excised polyps was collected. Any histological confirmed colorectal adenoma detected at least 1 year after the last positive endoscopy was counted as a recurrent adenoma. In general, endoscopies performed within a year after the last positive endoscopy are mainly done to ascertain if the removal of the adenoma was adequate, rather than to check for recurrences.32 We assumed that recurrent adenomas, detected within a year after the recruitment endoscopy, were missed during the previous endoscopy. Advanced adenomas were those with a diameter of ≥1 cm and/or villous or tubulovillous histology and/or with high grade dysplasia and/or ≥3 adenomas detected at the same colonic examination.8 The pathological features of the largest and/or most advanced adenoma were used to characterize the adenomas when there were multiple adenomas at the same colonic examination. If no information about the size, histology and dysplasia could be retrieved from the endoscopy and/or pathology reports, we assumed that an adenoma was not advanced.
Descriptive data (means ± standard deviation [SD], medians and interquartile range [IQR] or percentages) were calculated for the total study population, for patients with adenoma recurrence and for patients with at least one recurrent adenoma with advanced pathology. In addition, demographic and lifestyle characteristics for dietary supplement users and nonusers were computed.
Stratified Cox proportional hazard models for counting processes33, 34 were used to examine the association of dietary supplement use with total colorectal adenoma recurrence and for the subgroup of recurrent advanced adenomas. Both analyses were compared to the total study population. The conditional approach of the stratified Cox proportional hazard model focuses on survival time between recurrent events, which is in this case colorectal adenomas. In this approach, the survival time between two events always starts at 0 for the earlier event and stops at the later event.33 In this case, the time until the first event does not influence the composition of the risk set for a second or a later event.33 The robust sandwich covariance estimation technique was used to adjust for the within subject correlation, if one person had more than one recurrent adenoma.33 Hazard ratios (HRs) were reported with 95% confidence intervals (95% CIs). The Cox proportional hazard models were tested for and met the assumption of proportionality; this was done by visually inspecting whether the distance between the log(-log) survival curves was approximately constant. Follow-up time started at the recruitment endoscopy and ended at the date of the last endoscopy during follow-up. Patients were censored at the date of the last known colonic examination, at the date of diagnosis of colorectal cancer or at the date of death. Patients who deceased with unknown date of death and patients who reached the end of the follow-up period without a colorectal adenoma recurrence were censored at the date of their last endoscopy.
Covariates were considered as confounders if they correlated with both dietary supplement use and colorectal adenoma recurrence and if they changed the HR by 10% or more using backward elimination of variables. We evaluated the following factors as potential confounders: age (years), sex, educational level (low, middle, high), smoking status (current, former, never), alcohol intake (g/day), physical activity (low, medium, high), use of nonsteroidal anti-inflammatory drug (NSAIDs) (<1 time/month, ≥1 times/month), family history of colorectal cancer (yes/no), history of adenomas preceding the recruitment endoscopy (yes/no), number of colonoscopies during follow-up (1, 2, ≥3), total energy intake (kJ/day), total fibre intake (g/day), total vegetables intake (g/day), total fruit intake (g/day) and total red meat intake (g/day). In the basic model, we adjusted for age and sex. The fully adjusted model included age, sex, educational level and number of colonoscopies during follow-up.
Predefined stratified analyses were performed for sex. To assess possible effect measure modification of the association between dietary supplement use and colorectal adenoma recurrence by smoking status and by total fruit and vegetables intake, we stratified our analysis for smoking status (never, former, current) and quartiles of total fruit and vegetables intake (<185, 185–276, 276–385, ≥385 g/day). HRs stratified for smoking status were estimated with nonusers, who never smoked, as reference group. In addition, nonusers, who had a relatively low intake of fruit and vegetables of <185 g/day, were defined as reference group to evaluate HR stratified for total fruit and vegetables intake. The p value for interaction was calculated by χ2 test of the likelihood ratio test, comparing the models for nonusers and users by smoking status and by total fruit and vegetables intake.
In our cohort, all participants (n = 565) had an endoscopy at recruitment. However, part of the cohort did not have a colorectal adenoma at the recruitment endoscopy but were diagnosed with an adenoma before recruitment. Therefore, a sensitivity analysis was performed in which the cohort was restricted to those persons (n = 406) who had a colorectal adenoma at the recruitment endoscopy. Furthermore, three other sensitivity analyses were done, one for persons who had a history of adenomas preceding the recruitment endoscopy compared to persons who did not have a history of adenomas before recruitment, an analysis restricted to those persons without a family history of colorectal cancer (n = 424) and an analysis in which nonusers were defined as taking less than four dietary supplements per month instead of no dietary supplement at all. A level of significance of less than 0.05 was considered statistically significant. The statistical software program SAS 9.2 (SAS, Cary, NC, 2004) was used for statistical analysis.
Characteristics of the study population are shown in Table 1. The mean (± SD) age at recruitment of the total study population (n = 565) was 58.8 (± 9.7) years and the median person-time was 5.4 (P25–P75 = 3.7–7.0) years. About one-third of the total study population (n = 203) used a dietary supplement, and about two-third (n = 344) had advanced colorectal adenomas at recruitment endoscopy. A number of 165 out of 565 patients (29%) had one or multiple colorectal adenoma recurrences during follow-up. Compared to the total population, patients with a colorectal adenoma recurrence used slightly more dietary supplements, were more often men, were slightly older and had more often former adenomas. Recurrent adenomas showed advanced pathology in 37 (22%) out of the 165 patients with a colorectal adenoma recurrence, and of these 27 (73%) already had an advanced adenoma at recruitment endoscopy. Patients who had an advanced adenoma recurrence used more dietary supplements, were slightly older, were lower educated, used less NSAIDs and had more often a history of adenomas preceding the recruitment endoscopy compared to both the total study population and patients with total colorectal adenoma recurrence.
Table 1. General characteristics for the total population, for cases with total colorectal adenoma recurrence and for cases with at least one adenoma recurrence with advanced pathology in a prospective cohort study of 565 adenoma patients
Table 2 shows general characteristics of dietary supplement users and nonusers. Of the 203 dietary supplement users, 66 (32.5%) users experienced a recurrent colorectal adenoma, and 16 (7.9%) users had a least one recurrent adenoma with advanced pathology. Among the nonusers (n = 362), 99 (26.5%) had an adenoma recurrence, and 21 (5.8%) experienced at least one recurrent advanced adenoma. Compared to nonusers, dietary supplement users were more often women, were higher educated, were more physically active, had a slightly lower body mass index, had more often a family history of colorectal cancer and consumed more vegetables and fruit.
Table 2. General characteristics of dietary supplement users versus nonusers in a prospective cohort study of 565 adenoma patients
Dietary supplement use was neither associated with total colorectal adenoma recurrence (HR = 1.03; 95% CI 0.79–1.34) nor statistically significantly associated with recurrent adenomas that showed advanced pathology (HR = 1.59; 95% CI 0.88–2.87) after adjustments for age, sex, educational level and number of colonoscopies during follow-up (Table 3). Multivitamin supplements (HR = 0.88; 95% CI 0.62–1.24), vitamin C supplements (HR = 0.83; 95% CI 0.58–1.17) and calcium (including vitamin D) supplements (HR = 0.83; 95% CI 0.45–1.50) did not show an association with total and advanced colorectal adenoma recurrence either. In exploratory analyses, B-vitamin supplements were associated with a statistically significantly increased risk of total colorectal adenoma recurrence (HR = 1.58; 95% CI 1.05–2.36) but not with recurrent adenomas that showed advanced pathology (HR = 1.59, 95% CI 0.59–4.27).
Table 3. Association of dietary supplement use with total colorectal adenoma recurrence and advanced adenoma recurrence in a cohort of 565 colorectal adenoma cases
Associations did not markedly differ between men (HR = 1.07; 95% CI 0.72–1.60) and women (HR = 0.98; 95% CI 0.69–1.38). No association for colorectal adenoma recurrence and dietary supplement use was shown in the lower and higher quartiles of total fruit and vegetables intakes (p for interaction: 0.93). Effect measure modification of smoking status was not present for the association of dietary supplement use and adenoma recurrence: p for interaction was 0.63 (Table 4).
Table 4. Hazard ratios for dietary supplement use and total colorectal adenoma recurrence by smoking status and total fruit and vegetables intake in a prospective cohort study of 565 adenoma patients
Sensitivity analyses showed that restricting the cohort to persons who all had an adenoma recurrence at the recruitment endoscopy did not markedly change the association between dietary supplement use and total colorectal adenoma recurrence (HR = 1.11; 95% CI 0.84–1.47). In additional sensitivity analyses, no differences in associations for dietary supplement use and total colorectal adenoma recurrence were seen for persons with a history of adenomas preceding the recruitment endoscopy (HR = 1.24; 95% CI 0.76–2.02) and for persons without former adenomas at baseline recruitment (HR = 0.94; 95% CI 0.69–1.28). A borderline statistically significant increased risk was found for dietary supplement use and colorectal adenoma recurrence (HR = 1.39; 95% CI 1.02–1.89) in persons with no known family history of colorectal cancer. Finally, sensitivity analyses restricted to those who used no or less than four dietary supplements per month as the reference category did not change the association for dietary supplements and total colorectal adenoma recurrence in our study either (HR = 1.13; 95% CI 0.86–1.48).
In this prospective cohort study, dietary supplement use was neither associated with total colorectal adenoma recurrence nor with recurrence of adenomas with advanced pathology. The lack of association for colorectal adenoma recurrence also applied to multivitamin supplements, vitamin C supplements and calcium (including vitamin D) supplements. Exploratory analyses suggested that B-vitamin supplements were associated with a statistically significantly increased risk of total adenoma recurrence but not with recurrence of colorectal adenomas with advanced pathology.
To our knowledge, no other prospective study evaluated the association between multivitamin supplement use and colorectal adenoma recurrence. Our prospective study showed no statistically significant association for use of multivitamin supplements with colorectal adenoma recurrence, which was in line with a case–control study of 198 recurrent adenoma cases and 347 controls who had a history of one or more polyps.28 Our findings contrast with a second case–control study with 183 recurrent adenoma cases and 265 controls who had a past history of a colonic neoplasm; there they found a decreased risk of adenoma recurrence with multivitamin supplement use.27 Nevertheless, case–control studies may be subject to recall bias. Prospective cohort designs have been largely perceived as an effective strategy to avoid exposure recall bias, which make prospective studies more reliable to predict associations.
The association between colorectal adenoma occurrence and use of dietary supplements has been extensively evaluated in many epidemiological studies13–17; however, less comprehensive results have been reported for colorectal adenoma recurrence. According to a meta-analysis of randomized controlled trials, calcium supplementation statistically significantly reduced the risk of colorectal adenoma recurrence.24 A prospective observational study within the Wheat Bran Fiber Trial showed no association for supplemental vitamin D and colorectal adenoma recurrence,18 whereas in observational analyses in the Polyp Prevention Trial an inverse association was observed for any use of vitamin D supplements and recurrent adenomas compared to no use.23 This difference in association could be explained by the fact that participants in the Wheat Bran Fiber trial may have more homogeneous levels of vitamin D due to higher sunlight exposure.23 There is no other study that investigated the association of supplemental vitamin C, as a single supplement, and total colorectal adenoma recurrence. One small intervention study found a significant reduction in the recurrence of polyps with a combination of a vitamin A, C and E supplement in 70 randomly assigned patients compared to 78 patients who received no treatment.35 No effects were found in an intervention study among 143 patients randomly assigned to a combined supplement of vitamin C and E or placebo and recurrence of colorectal adenomas and in another intervention study of 864 randomly assigned patients to four treatment groups.19, 36 Another randomized controlled trial did not provide evidence for a potential role of supplemental vitamin E in colorectal adenoma recurrence.19 Due to the low number of users of these specific dietary supplements in our study, it was not possible to show an interpretable association for supplemental calcium and vitamin D, vitamin C and vitamin E and colorectal adenoma recurrence.
In our study, B-vitamin supplements, including folic acid, were associated with a statistically significantly increased risk of total adenoma recurrence but not with recurrence of colorectal adenomas that showed advanced pathology. According to randomized controlled trials, supplemental folic acid may not prevent recurrence of colorectal adenomas25, 37, 38 but may instead increase the recurrence of multiple and advanced colorectal adenomas.39 We were not able to evaluate the role of folic acid supplements separately because in our questionnaire supplemental B-vitamin use was measured together as one dietary supplement as those B-vitamin supplements are mostly used. No other study evaluated whether a combination of B-vitamins in one supplement was associated with colorectal adenoma recurrence. However, due to the small number of B-vitamin supplement users in our study, we should interpret this possible chance finding with caution.
Generally, diets of dietary supplement users are higher in fruit and vegetables compared to those of nonusers.40 Randomized controlled trials with diets high in fruit and vegetables did not find significant benefits for higher fruit and vegetables intakes in reducing the risk of adenoma recurrence.41, 42 In our study, we found no association for dietary supplement use and colorectal adenoma recurrence in the lower and higher quartiles of total fruit and vegetables intakes. To our knowledge, no other study investigated whether the association between dietary supplement use and colorectal adenoma recurrence is different for low versus high intakes of fruit and vegetables. As fruit and vegetables are rich sources of micronutrients and bioactive compounds43 and a small reduction in risk for colorectal cancer and fruit and vegetables has been observed in large prospective studies,44, 45 it is useful to evaluate whether fruit and vegetables can reduce the risk in patients with recurrent colorectal adenomas.
Several methodological issues concerning dietary supplement use need to be considered. First, comparison of dietary supplement use between different studies is challenging because of different definitions of supplements and different frequencies of use required for a subject to be defined as user. In our cohort of adenoma cases, supplement use was defined as an intake of any dietary supplement during the past year. In observational analyses within the Polyp Prevention Trial, participants were defined as user when they used any calcium or vitamin D supplement during any of the time periods used for analysis, and use was categorized in tertiles of intake of supplemental calcium and vitamin D.23 No other observational studies were done that investigated dietary supplement use and colorectal adenoma recurrence. Moreover, no or infrequent users, who used no or less than four dietary supplements per month, may also be representative for subjects who did not use any dietary supplement at all. Yet, sensitivity analyses with subjects who used less than four dietary supplements per month as the reference did not change the associations in our study. Thus, those no or infrequent users might not differ much from nonusers, who did not take any dietary supplement at all, in our study.
A second issue regarding supplement use we need to consider is the fact that we were not able to calculate the total nutrient intake by foods and dietary supplements together as information on dosage and duration was not assessed. Observational analyses within two randomized controlled trials showed results for colorectal adenoma recurrence and total intake of micronutrients from supplements and diet together.18, 21, 22 The Wheat Bran Fiber Trial and the Polyp Prevention Trial indicated that higher intakes compared to lower intakes of total vitamin B6,22 total folate,22 total calcium18 and total vitamin A21 were inversely associated with recurrence of colorectal adenomas, whereas nonsignificant associations were shown for total vitamin B12.22 In our study, it would have been preferable if a detailed assessment on dietary supplement use and changes in use was available.
A third issue is that use of dietary supplements can vary throughout the year. Supplement users may take more dietary supplements during the winter months compared to the summer months, whereby confounding by season might bias the results. However, as subjects were recruited for the study throughout the year and supplement use was assessed by asking over the preceding year, adjusting for seasonal influences did not change the associations between dietary supplement use and colorectal adenoma recurrence in our study.
Fourth, in our study we relied on self-reported dietary supplement use, which made misclassification of the exposure possible. However, prior studies have demonstrated such data to be reliable.46, 47 Moreover, dietary supplement use was recorded before any knowledge of colorectal adenoma recurrence, thus reducing the likelihood of reporting biases. Participants could also have changed their intake of dietary supplements after polypectomy based on their health status and disease risk,48 but according to Almendingen et al. there is no existing evidence for meaningful lifestyle change after the diagnosis of adenoma.49 We therefore assume that intake of dietary supplements by colorectal adenoma patients is relatively constant during the follow-up period in our study, which makes self-reported dietary supplement use in the year prior to polypectomy a reliable indicator for the actual intake.
The last issue we want to address is the fact that the lack of association between use of any dietary supplement and colorectal adenoma recurrence could be a mere product of a neutralization of individual effects of each nutrient supplement. Unfortunately, we were not able to further explore the association of individual dietary supplements and recurrent colorectal adenomas due to the small study size.
Besides the methodological issues regarding dietary supplement use, another limitation of our study is the small sample size, which decreased the power to detect associations. One of the strengths of our study is the long person-time of this prospective cohort, which enabled us to detect several recurrent adenoma events. Another strength is that residual confounding will be unlikely in our study, because we acquired extensive dietary, lifestyle and medical information from the adenoma cases. Consistent with literature, dietary supplement users in our study were more often women,1 were higher educated,1, 3 had a healthier lifestyle1, 3 and were less likely to smoke.50 We were able to adjust for these potential confounders. However, as dietary supplement users probably have a healthier diet and are more health conscious than nonusers, which involves many different factors, residual confounding can never be completely ruled out.
In conclusion, we did not find an indication that dietary supplements have a beneficial or harmful role on colorectal adenoma recurrence. To make public health policies for prevention of recurrent adenomas and colorectal cancer, future studies with extensive information on habitual dietary supplement use are needed.
The authors are grateful to all the participants in our study and to the endoscopy staff in the following hospitals in the Netherlands: Slingeland Hospital (Doetinchem), Hospital Gelderse Vallei (Ede), Radboud University Nijmegen Medical Centre (Nijmegen), St. Antionius Hospital (Nieuwegein), Meander Medical Centre (Amersfoort), Hospital Rijnstate (Arnhem), Hospital Rivierenland (Tiel), Slotervaart Hospital (Amsterdam), Jeroen Bosch Hospital ('s Hertogenbosch) and Canisius Wilhelmina Hospital (Nijmegen). They thank Dorien Voskuil, Edine Tiemersma, Maria van Vugt, Elly Monster, Brenda Diergaarde, Petra Wark and Mariken Tijhuis for their roles in the conduct of baseline data, Marga Ocké from the National Institiute for Public Health and the Environment (RIVM, Bilthoven) for micronutrient calculations and Maria van Vugt and Elly Monster (Division of Human Nutrition, Wageningen University, Wageningen) for collecting blood samples and assistance with the conduct of our study. They thank Audrey Jung, Marlieke Visser, Leontien Witjes and Ursula Oldenhof for their roles in the conduct of follow-up data.