Use of Folic Acid–Containing Supplements after a Diagnosis of Colorectal Cancer in the Colon Cancer Family Registry

Background: Supplement use among cancer patients is high, and folic acid intake in particular may adversely affect the progression of colorectal cancer. Few studies have evaluated the use of folic acid–containing supplements (FAS) and its predictors in colorectal cancer patients.

Objective: To assess the use of FAS, change in use, and its predictors after colorectal cancer diagnosis.

Design: We used logistic regression models to investigate predictors of FAS use and its initiation after colorectal cancer diagnosis in 1,092 patients recruited through the Colon Cancer Family Registry.

Results: The prevalence of FAS use was 35.4% before and 55.1% after colorectal cancer diagnosis (P = 0.004). Women were more likely than men to use FAS after diagnosis [odds ratio (OR), 1.47; 95% confidence interval (95% CI), 1.14-1.89], as were those consuming more fruit (P_trend < 0.0001) or vegetables (P_trend = 0.001), and U.S. residents (P < 0.0001). Less likely to use FAS after diagnosis were nonwhite patients (OR, 0.66; 95% CI, 0.45-0.97), current smokers (OR, 0.67; 95% CI, 0.46-0.96), and those with higher meat intake (P_trend = 0.03). Predictors of FAS initiation after diagnosis were generally similar to those of FAS use after diagnosis, although associations with race and vegetable intake were weaker and those with exercise stronger.

Conclusions: Our analysis showed substantial increases in the use of FAS after diagnosis with colorectal cancer, with use or initiation more likely among women, Caucasians, U.S. residents, and those with a health-promoting life-style.

Impact: Studies of cancer prognosis that rely on prediagnostic exposure information may result in substantial misclassification. Cancer Epidemiol Biomarkers Prev; 19(8); 2023–34. ©2010 AACR.

The B vitamin folate may affect both the development and progression of colorectal cancer, yet information about folate intake among colorectal cancer patients is lacking. Folate mediates the transfer of one-carbon moieties both in the synthesis of nucleotides necessary for DNA synthesis, replication, and repair and in DNA methylation reactions; these functions may play a role in cancer prevention (1, 2). Folic acid is the oxidized synthetic form of folate used in supplements and fortified foods, with greater stability and bioavailability than naturally occurring folate (3). Overall, epidemiologic evidence supports an inverse association between high folate intake (from diet and supplements) and risk of colorectal adenomas and colorectal cancer (3-6). Rodent models support the antineoplastic effect of folate on colorectal tissue before lesions develop, with modest doses of supplemental folic acid suppressing the development of colorectal cancer (3-7).

However, recent evidence suggests that folic acid may in fact accelerate the progression of existing colorectal cancer precursors. First, in rodents, folic acid supplementation promotes the progression of colorectal cancer after early lesions, aberrant crypt foci, are present (3). Second, a clinical trial in patients with a recent history of colorectal adenoma, the Aspirin/Folate Polyp Prevention Study, found no evidence that folic acid prevented the development of new colorectal polyps. Instead, its use was associated with multiple adenomas, and there was a suggestive positive association with advanced adenomas after longer term treatment and follow-up (6-8 y; ref. 8). In this high-risk group of patients, early lesions may have been present but missed or undetectable at baseline colonoscopy, and it has been hypothesized that folic acid could have accelerated their growth (3, 5). Two other trials have recently been completed. A randomized trial of folic acid supplementation among members of the Health Professionals' Follow-Up Study and the Nurses' Health Study with history of colorectal adenoma showed no significant associations between folic acid and recurrent adenomas (any, advanced, or multiple; ref. 9). Similarly, the UK Colorectal Adenoma Prevention trial found no effect of folic acid supplementation on the occurrence of any, advanced, or multiple new adenomas (10). However, the duration of follow-up and folic acid supplementation for both trials was shorter than for the Aspirin/Folate Polyp Prevention Study, and for the UK Colorectal Adenoma Prevention trial, the folic acid dose was lower and no population-wide fortification program was in place. Because the elevated risk in the Aspirin/Folate Polyp Prevention Study did not occur until 6 to 8 years postdiagnosis, additional follow-up in the recent trials is needed to evaluate whether the results are inconsistent.

Both normal and neoplastic colorectal tissues are rapidly dividing with high rates of DNA replication; the provision of nucleotides for DNA synthesis is the most likely mechanism through which folate may both protect normal tissue from DNA damage and carcinogenesis, and accelerate progression of established neoplasia (3, 4, 6, 7). Of concern is that a cancer-promoting effect may also be present for other tissues, as suggested by an increased risk for prostate cancer in the folic acid arm of the Aspirin/Folate Polyp Prevention Study trial (11). A recent analysis of data from a trial in Norway showed an increase in overall cancer incidence and mortality with folic acid supplementation (12). The statistical power of the study was inadequate to evaluate individual cancer sites, although hazard ratios for colorectal cancer were close to 1.0.

Consistent with high folate intake promoting growth of established colorectal tumors, antifolate agents inhibit their growth, an effect exploited in the conventional treatment of colorectal cancer. The chemotherapeutic agent 5-fluorouracil (5-FU), which is widely used in colorectal cancer treatment, inhibits the folate-metabolizing enzyme thymidylate synthase (TS), blocking pyrimidine synthesis, inducing DNA damage, and slowing tumor growth (3, 6, 13, 14). The effect of chronic use of folic acid supplements on 5-FU treatment efficacy is currently unknown.

Nutritional supplement use is high in the U.S. population (3, 15, 16) and higher still among cancer patients (3, 5, 17-19). In the limited number of studies reporting change in supplement use with diagnosis, all showed increases in use. Cancer patients report that they use supplements and other alternative therapies to feel better, gain a sense of control, reduce stress, support their immune systems, and improve chances of a cure (17, 20-22). Interest in research on health behaviors and cancer prognosis is growing (23-26), and thus, it is critical to evaluate and quantify life-style after a cancer diagnosis to design appropriate prognostic studies. Many epidemiologic studies evaluate associations between prediagnostic health behaviors and disease outcomes, an approach that may lead to substantial exposure misclassification. Limited information is available on the use of supplements, folic acid–containing supplement (FAS) intake in particular, and determinants of FAS intake among colorectal cancer patients (18, 19, 27). To date, only one study has addressed changes in FAS and other supplement use after colorectal cancer diagnosis (18). Thus, we here evaluate FAS initiation and use after colorectal cancer diagnosis in a large international cohort (n > 1,000) of colorectal cancer patients, and evaluate a comprehensive set of factors associated with FAS use.

The Cancer Family Registry (C-CFR) is an international study of colorectal cancer cases, their family members, and controls recruited by investigators at six sites: Seattle, Washington; the Mayo Clinic in Rochester, Minnesota; a consortium led by the University of Southern California (USC) in Los Angeles; the University of Hawaii; the University of Melbourne in Australia; and Cancer Care Ontario in Toronto. Study protocols were reviewed and approved by Institutional Review Boards at each participating site. Eligibility criteria and sampling schemes differed somewhat by site and have been previously described (28). Briefly, phase I of recruitment was conducted between 1998 and 2002, and included cases identified both from cancer registries (population-based recruitment) and from clinics serving families with early-onset or multiple colorectal cancer cases (clinic-based recruitment). Response rates varied by study site due to differences in approach protocols. Of those eligible at baseline, between 35% and 78% agreed to participate across the six C-CFR sites. Active follow-up of phase I participants occurred ∼5 years after initial enrollment; 22% of cases died before this follow-up, and of the survivors, 70% to 95% completed the follow-up questionnaire.

Life-style and risk factor questionnaires were administered at enrollment, asking about the period before diagnosis, and at follow-up ∼5 years later. Data collected included participants' use of supplements and medications, race, ethnicity, and other demographic information, physical activity, height, weight, smoking history, alcohol consumption, and some information on diet. Questions on use of folic acid and multivitamins at enrollment asked if participants had ever taken the supplements regularly (at least twice a week for more than a month), and if they had taken them regularly ∼2 years ago (use before diagnosis). At follow-up, participants were asked if they had taken folic acid or multivitamins regularly in the interval between baseline and follow-up interviews (use after diagnosis). Three of the six C-CFR sites, the USC Consortium, Ontario, and Hawaii, also administered a detailed food frequency questionnaire (FFQ) to participants at baseline (29), allowing quantitative assessment of prediagnostic folic acid intake for a subset of participants.

The present analysis uses an initial data set available in 2008 and includes 1,092 colorectal cancer cases recruited during phase I (28). The subset of 1,092 cases represents all those with data available from questionnaires administered both at enrollment and follow-up, and with data available on covariates such as education, exercise, and diet. Compared with phase I C-CFR cases not included in this subset, cases included had generally similar demographic characteristics and health behaviors. Supplement use was similar between the two groups. Thirty-four percent of those included used multivitamins before diagnosis, compared with 30% of those not included; ∼3% of each group used single-supplement folic acid. Those included in our analysis were somewhat younger and more educated, differences likely related to the required survival to follow-up and completion of the follow-up questionnaire.

Nearly all commonly used multivitamins include folic acid, and therefore, we included both multivitamins and single-supplement folic acid in our definition of FAS. We first studied the outcome of “FAS use after diagnosis with colorectal cancer,” by comparing all patients who used FAS after diagnosis (n = 602) to those who did not (n = 490). Second, we analyzed “FAS initiation” after colorectal cancer diagnosis by comparing cases who did not use FAS regularly in the 2 years before diagnosis and used FAS after diagnosis (“new use,” n = 307) to those who used FAS neither immediately before nor after diagnosis (“no use,” n = 399).

To investigate predictors of FAS use or initiation in colorectal cancer cases, we assessed characteristics that had been associated with supplement use in previous studies (15, 17, 30-32), including age; sex; race (white versus nonwhite); family history (one or more versus no first-degree relatives with colorectal cancer); study site; education; income; smoking; alcohol intake, diet including intake of fruit, vegetables, and red meat; lifetime exercise, and body mass index (BMI). We chose lifetime rather than recent exercise as better reflecting attitudes toward health rather than current physical capabilities.

We used multivariate logistic regression to evaluate associations between patient characteristics and FAS use or initiation, adjusting each for age, sex, and site as appropriate. For patient characteristics described by multiple ordinal categories, we used Wald trend tests to evaluate any trends apparent in odds ratios (OR) across categories. To identify the most important predictors of FAS use or initiation overall, we used forward stepwise regression analyses, which concurrently evaluated the covariates listed above, using a significance level of 0.1 for both entering and remaining in the model. To evaluate robustness, we also derived models using the entry and stay criteria of 0.05 and 0.1, respectively. Income and family history were excluded from the forward stepwise analyses because of incomplete data.

Table 1 shows baseline characteristics of the 1,092 colorectal cancer cases. Most were younger than age 60 years at diagnosis, and more than a third had a family history of colorectal cancer, reflecting the focus of the C-CFR cohort on recruiting cases more likely to have familial colorectal cancer, including those diagnosed at a young age (28, 33). Nearly a third of cases were from outside the United States. More than a third of cases took a multivitamin at baseline (before diagnosis), and 30 (3%) used single-supplement folic acid. Dietary intake data were available for 424 participants included in this analysis (39%) and showed mean folic acid intake from diet (postfortification) and supplements of 448 μg for non-FAS users and 883 μg for FAS users. Thus, FAS use is a major source of interindividual variability in folate intakes. For 11%, intake exceeded 1 mg folic acid, the tolerable upper intake level established by the Institute of Medicine to avoid masking vitamin B12 deficiency (34).

The proportion of cases using FAS, including multivitamins, increased substantially after colorectal cancer diagnosis (Table 2). A total of 35.4% of cases reported FAS use before colorectal cancer diagnosis, whereas after diagnosis, 55.1% reported FAS use (P = 0.004). Of those who did not use FAS before diagnosis, 43.5% initiated use after diagnosis (n = 307) and 56.5% did not (n = 399; these are the two groups of cases compared in our analysis of predictors of FAS initiation). Former users were more likely to initiate use after a cancer diagnosis than were those who had never used FAS (58.4% versus 36.7%, P < 0.0001). Of those cases who were already using FAS before diagnosis, 76.4% maintained use after diagnosis. FAS use was substantially higher, both before and after diagnosis, at U.S. compared with non-U.S. sites (P < 0.0001 at both time points), although use increased after diagnosis in all countries studied.

Table 3 shows that women were more likely than men to use FAS after colorectal cancer diagnosis [OR, 1.47; 95% confidence interval (94% CI), 1.14-1.89], and nonwhite cases were less likely than Caucasians to use FAS (OR, 0.66; 95% CI, 0.45-0.97). Subjects at the Ontario and Australian sites were much less likely to use FAS than were those from U.S sites [ORs of 0.40 (95% CI, 0.26-0.61) and 0.16 (95% CI 0.10-0.26], respectively, compared with Seattle). Current smokers were less likely to use FAS than were those who had never smoked (OR, 0.67; 95% CI, 0.46-0.96). There were positive associations between both higher fruit (P_trend < 0.0001) and vegetable intake (P_trend = 0.001) and FAS use, whereas those who consumed more red meat were less likely to use FAS (P_trend = 0.03). No strong associations between FAS use and age, family history, education, income, alcohol use, exercise, or BMI were observed. In forward stepwise analysis, sex, race, study site, and meat and fruit intake emerged as significant predictors of FAS use after diagnosis (Table 4). Using the more stringent criteria of P_enter < 0.05 and P_stay < 0.1 resulted in a model in which study site and intake of meat and fruit, but not sex and race, predicted FAS use (data not shown).

The analysis of predictors of FAS initiation after diagnosis (Table 5) gave similar results, with sex, site, smoking, and consumption of meat and fruit showing statistically significant associations. The associations of race and vegetable intake with FAS initiation were weaker than those with FAS use. There was an association between lifetime exercise and FAS initiation that had not been observed for FAS use (OR of 1.48 and 95% CI of 1.01-2.16 for active versus inactive cases). Forward stepwise analysis showed sex, study site, meat intake, and exercise to be predictors of FAS initiation (Table 6), although exercise did not seem as a predictor in a model using P_enter < 0.05 (data not shown).

We observed a substantial increase in the proportion of patients using FAS after colorectal cancer diagnosis compared with use before diagnosis. Those more likely to use FAS included women, Caucasians, U.S. residents, those consuming less meat and more fruit, and possibly nonsmokers and those consuming more vegetables. Predictors of FAS initiation after diagnosis were similar to those of any FAS use, although associations with race and consumption of fruits and vegetables were weaker and that with exercise somewhat stronger.

Few studies have addressed supplement use among colorectal cancer patients (19, 27); only one has addressed postdiagnostic initiation of FAS use and none has assessed predictors of FAS initiation in these patients. A relatively small study of colon cancer patients (n = 278), diagnosed between 1996 and 2000 in North Carolina, assessed change in supplement use after diagnosis (18). It showed that 31% of patients used multivitamins before diagnosis and 1.5% used single-supplement folic acid; 2 years after diagnosis, the proportions increased to 43% for multivitamins and 7% for folic acid. Our results suggest higher use both before and after diagnosis than did the North Carolina study, particularly at U.S. sites, possibly reflecting geographic differences and/or a time trend; nutritional surveys such as the National Health and Nutrition Examination Survey have generally shown increases in the use of dietary supplements over time (15). In the North Carolina study, factors associated with use of any new supplement after diagnosis included age, female sex, and white race, although predictors of change in FAS use were not reported.

The results of our analyses are in part consistent with earlier analyses of predictors of supplement use in general populations and in cancer patients, which have shown positive associations with female sex, white race, and fruit and vegetable consumption, and inverse associations with smoking and meat consumption (15, 17, 18, 30-32). The results of our multivariate analysis suggest that meat and fruit intake (and geographic location) are more important predictors of FAS use than sex and race. Our results for education contrast with earlier studies of supplement use in cancer patients and general populations, which showed consistent positive associations; our analysis did suggest that those with more education were more likely to use FAS, but differences were not statistically significant. In a separate analysis of supplement use before diagnosis in our colorectal cancer cases (data not shown), we found generally similar, slightly stronger, associations of FAS use with education and other indicators of socioeconomic status or health-promoting behaviors, consistent with reports from studies of individuals without cancer (15, 30-32).

Our analysis suggests that the use of FAS among colorectal cancer patients is widespread, with postdiagnostic use substantially higher than prediagnostic use. The effect of supplement use and nutrient status on prognosis in cancer patients is an important new area of epidemiologic research (23-26); however, many studies have measured supplement use or nutrient status before cancer diagnosis, rather than during cancer treatment and follow-up. Our results suggest that studies using prediagnostic supplement use as a proxy for postdiagnostic use may suffer from substantial exposure misclassification.

FAS use may adversely affect disease course. Results from animal studies and preliminary clinical data indicate that high folic acid intake may accelerate progression of established colorectal adenomas and cancer (3, 8). Effects on disease recurrence and metastasis are unknown. Experimental evidence suggests that folic acid may also alter the efficacy of 5-FU treatment, with effects depending on the timing of its use. The drug leucovorin is a reduced form of folic acid given immediately before 5-FU administration that stabilizes binding of 5-FU to thymidylate synthase and improves response rates (14, 35-37). However, chronic use of folic acid may have the opposite effect. Experimental results in human cancer cell lines and in mouse tumor models suggest that folate deficiency, in culture medium or diet, improves response to 5-FU especially when given with leucovorin (38-40). Mechanisms identified include the downregulation of thymidylate synthase expression and activity with low folate status, enhancing inhibition by 5-FU and drug efficacy (39). These results suggest that continuing use of folic acid supplements among colorectal cancer patients receiving 5-FU could reduce treatment efficacy; however, this question has not been studied in a clinical setting (4, 5).

Strengths of our analysis include data from a large, prospective, international study of colorectal cancer cases, with detailed information available on use of specific supplements and on covariates likely associated with this use. The diverse populations from which participants were recruited are a strength of the C-CFR and of our study; however, those included in our analysis still may not fully represent the larger population of colorectal cancer patients. In addition, FAS use among colorectal cancer patients in 2010 may differ from that measured in C-CFR participants between 1998 and 2002 because of media coverage of concerns about the use of folic acid in cancer patients.

Another limitation of the analysis is that data on baseline diet and supplement use were collected at enrollment, asking about intake 2 years previously, before participants' cancer diagnoses. However, validation studies comparing intake of food and supplements elicited by questionnaire to those obtained by diet record or 24-hour diet recall show comparable correlations for studies of current diet or supplements (41-43), and diet 3 to 4 years (44) or 10 years (45) in the past. A nested case-control study of long-term dietary recall (46) showed generally similar results for cancer cases and controls, although recall was somewhat poorer for cases with colorectal cancer compared with noncases.

A further limitation of our data are that FFQs were administered at only three C-CFR sites and only for the prediagnostic time period. Our analysis therefore focuses on qualitative rather than quantitative measures of change in folic acid intake after diagnosis and on intake from supplements but not from food. However, supplement use is a major source of variation in folate intakes, especially among cancer patients. A pilot study of 176 Ontario CFR participants did collect FFQ data both before and after colorectal cancer diagnosis, and found little change in dietary folate intake after diagnosis, although the subset of participants was not randomly selected and results may not apply to C-CFR participants overall.¹⁰

Our analysis of 424 CFR participants completing baseline FFQs at three sites showed that prediagnostic folic acid intake among FAS users was twice that of nonusers, and that intake exceeded 1 mg folic acid for 1 in 10 participants. The data for these subsets of C-CFR participants thus suggest that folic acid from supplements contributes substantially to overall intake and especially to change in intake after colorectal cancer diagnosis. Intakes were substantial before diagnosis, and although quantitative data on postdiagnostic intake are not available, the results of our qualitative analysis of FAS use suggest that these would be higher still.

The cases in our analysis can only include those surviving 5 years to follow-up. If FAS use accelerated progression of and death from disease, then postdiagnostic use in all colorectal cancer patients would be higher than our estimate of 55.1%. Some participants were also lost to follow-up or refused to complete the follow-up questionnaire; however, the response rate among survivors was high (82% across all sites). Analyzing the effects of FAS use on survival would address these questions, and such an analysis will be feasible once data on stage at diagnosis, clinical treatment, and mortality are available from all C-CFR sites. A further limitation associated with follow-up 5 years after enrollment is that our analysis could not assess changes in FAS use directly after colorectal cancer diagnosis, perhaps the most critical time period because of potential interactions with treatment. Another is that we cannot exclude secular trends in supplement use as one cause of the increase in FAS use observed.

We here report that FAS use is prevalent among colorectal cancer cases and identify characteristics of those likely to be using FAS. These findings are important, given evidence that folic acid may accelerate progression of colorectal cancer, and the unknown, but plausibly detrimental, effects that FAS intake may have on the efficacy of cancer treatment and on survival. The factors associated with FAS use in our analysis suggest that those who choose to use supplements may be receptive to information on health-promoting behaviors. However, many cancer patients do not discuss supplement use with their physicians (17). Our research may help clinicians identify patients likely to use supplements after colorectal cancer diagnosis, and allow them to discuss both the benefit and harm that may result. Finally, our study highlights the need to measure supplement use and other critical exposures at time points after diagnosis, to avoid substantial misclassification in studies of cancer prognosis.

Wyeth provided study materials to J. Baron's folic acid trial. P. Limburg, an ongoing consulting agreement with Genomic Health, Inc. J. Baron declared a conflict of interest in that Wyeth, manufacturers of Centrum multivitamins and other dietary supplements, provided study agents for a trial on folic acid that he conducted. No other authors reported a conflict of interest.

The content of this article does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the CFRs, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government or the CFR. CFR centers providing data for the analysis include the following: Australasian Colorectal C-CFR (U01 CA097735), Familial Colorectal Neoplasia Collaborative Group (U01 CA074799), Mayo Clinic Cooperative Family Registry for Colon Cancer Studies (U01 CA074800), Ontario Registry for Studies of Familial Colorectal Cancer (U01 CA074783), Seattle Colorectal C-CFR (U01 CA074794), University of Hawaii Colorectal C-CFR (U01 CA074806), and University of California, Irvine Informatics Center (U01 CA078296). R. Holmes and C. Ulrich were responsible for designing the study and drafting the manuscript. Y. Zheng, J. Baron, L. Li, G. McKeown-Eyssen, P. Newcomb, M. Stern, R. Haile, W. Grady, J. Potter, L. Le Marchand, P. Campbell, J. Figueiredo, P. Limburg, M. Jenkins, and J. Hopper all critically revised the manuscript. Y. Zheng, J. Baron, G. McKeown-Eyssen, P. Newcomb, R. Haile, J. Potter, L. Le Marchand, P. Campbell, J. Figueiredo, P. Limburg, M. Jenkins, and J. Hopper provided the data upon which this work was based. R. Holmes, Y. Zheng, L. Li, and P. Campbell conducted analyses. We thank Allyson Templeton for her generous support of this study, Liren Xiao for data set assistance, and Shannon Rush for manuscript support.

Grant Support: NIH grants: R01C105437, R25CA94880 and 5T32DK007742. This work was supported by the National Cancer Institute,NIH under RFA # CA-95-011 and through cooperative agreements with members of the C-CFR and Principal Investigators.

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.