Folate fortification, plasma folate, homocysteine and colorectal adenoma recurrence

Authors

  • María Elena Martínez,

    Corresponding author
    1. Arizona Cancer Center, University of Arizona, Tucson, AZ
    2. Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, AZ
    • University of Arizona, Arizona Cancer Center, PO Box 245024 Tucson, AZ, USA
    Search for more papers by this author
    • Fax: +520-626-9275.

  • Edward Giovannucci,

    1. Department of Nutrition, Harvard School of Public Health, Boston, MA
    2. Department of Epidemiology, Harvard School of Public Health, Boston, MA
    3. Channing Laboratories, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Ruiyun Jiang,

    1. Arizona Cancer Center, University of Arizona, Tucson, AZ
    Search for more papers by this author
  • Susanne M. Henning,

    1. Center for Human Nutrition, University of California, Los Angeles, Los Angeles, CA
    Search for more papers by this author
  • Elizabeth T. Jacobs,

    1. Arizona Cancer Center, University of Arizona, Tucson, AZ
    2. Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, AZ
    Search for more papers by this author
  • Patricia Thompson,

    1. Arizona Cancer Center, University of Arizona, Tucson, AZ
    2. Department of Pathology, University of Arizona, Tucson, AZ
    Search for more papers by this author
  • Stephanie A. Smith-Warner,

    1. Department of Nutrition, Harvard School of Public Health, Boston, MA
    2. Department of Epidemiology, Harvard School of Public Health, Boston, MA
    Search for more papers by this author
  • David S. Alberts

    1. Arizona Cancer Center, University of Arizona, Tucson, AZ
    2. Mel and Enid Zuckerman College of Public Health, University of Arizona, Tucson, AZ
    3. College of Medicine, University of Arizona, Tucson, AZ
    Search for more papers by this author

Abstract

In 1996, the US Food and Drug Administration mandated the fortification of grain products with folic acid, a nutrient that has been associated with lower risk of colorectal neoplasia. We assessed the relation of plasma folate and homocysteine and colorectal adenoma recurrence separately in 2 studies: the first involved an intervention of a cereal supplement that contained folic acid, wheat bran fiber (WBF), and the second was conducted primarily during postfortification of the food supply using ursodeoxycholic acid (UDCA). Analyses were stratified for multivitamin use. Results show that plasma folate and homocysteine concentrations were associated with adenoma recurrence among nonusers of multivitamins only. Among nonmultivitamin users, the odds ratio [OR] (95% confidence interval [CI]) for those in the highest versus the lowest folate quartile was 0.65 (0.40–1.06) for the WBF study and 0.56 (0.31–1.02) for the UDCA; likewise, individuals in the highest versus the lowest quartile of homocysteine had higher odds of adenoma recurrence, in both the WBF (OR = 2.25; 95% CI = 1.38–3.66) and UDCA (OR = 1.93; 95% CI = 1.07–3.49) populations. Analyses comparing multivitamin users to different plasma folate concentrations among nonusers show that odds of recurrence for supplement users was lower only when compared to nonusers who had lower concentrations. Our results show that higher plasma folate or lower homocysteine levels are associated with lower odds of recurrence among nonusers of multivitamins in both studies. Our finding, suggesting that multivitamins or supplemental folate only benefit individuals with lower plasma folate concentrations, should be taken into consideration when designing and interpreting results of intervention studies. © 2006 Wiley-Liss, Inc.

Colorectal cancer is thought to be the result of a multistep evolutionary process that involves a precursor lesion, the adenomatous polyp. Among individuals who undergo removal of their adenoma(s), recurrence of these lesions at surveillance colonoscopy is common, making adenoma recurrence an important endpoint to study in relation to lifestyle and other risk factors.

Folate is a water soluble B-vitamin that functions as a coenzyme in single-carbon transfer in nucleic and amino acid metabolism. Folic acid is the fully oxidized monoglutamyl form of folate that is used in supplements and fortified foods. A recent meta analysis showed a significant 20–25% lower risk of developing colorectal cancer among individuals in the highest compared to those in the lowest category of dietary folate1; inverse associations between blood folate and colorectal neoplasia (i.e., cancer and adenoma) have also been shown.2, 3, 4 Mechanisms responsible for the inverse association of folate on risk of colorectal cancer are not entirely understood and may involve disturbances in DNA synthesis, methylation and repair,5, 6, 7 albeit alternate mechanisms have been proposed.8

Homocysteine, a biochemical marker of folate status, is an amino acid that is metabolized via remethylation into methionine, facilitated by folic acid and vitamin B12 or trans-sulfuration to cysteine, facilitated by vitamin B6.9 Homocysteine levels can be increased by the presence of genetic factors, physiological defects and by use of certain drugs10 as well as by deficiencies in folate, vitamins B12 and B6, necessary cofactors for its conversion to methionine.9 Few data exist on the relation between homocysteine and colorectal neoplasia.3, 4

In 1996, the United States Food and Drug Administration (FDA) issued a regulation for the prevention of neural tube defects (NTDs), requiring that all enriched cereal grain products (i.e., flour, rice, pasta, breads, etc.) be fortified11 with 140 μg of folic acid/100 g, which would provide an estimated additional 80–100 μg/day to women of childbearing age and 70–120 μg to middle-aged and older adults. Full compliance with this mandate was required by January 1998. Mandatory fortification also took place in Canada in 1998.12 However, several European countries, New Zealand, and Australia have been reluctant to implement similar initiatives. Specifically, the United Kingdom's Food Standards Agency and the Dutch Health Council decided against the fortification on the grounds that not enough was known about the potential adverse effects. Thus, fortification of the food supply with folic acid remains controversial.

Using data from the 2 colorectal adenoma study populations, we assessed the relation between plasma folate and homocysteine and adenoma recurrence. In addition, we conducted analyses stratified for multivitamin supplement use.

Abbreviations:

CI, confidence interval; CV, coefficient of variation; FDA, Food and Drug Administration; NTD, neural tube defects; OR, odds ratio; UDCA, ursodeoxycholic acid; WBF, wheat bran fiber.

Material and methods

Participants

Analyses were conducted in participants from the wheat bran fiber (WBF) and the ursodeoxycholic acid (UDCA) trials; details of the studies have been reported elsewhere.13, 14 A brief description of each study is provided later.

The WBF trial was a double-blind controlled trial that compared the effect of a high-fiber versus a low-fiber cereal supplement on adenoma recurrence, among individuals who had 1 or more adenoma(s) removed. The high and low fiber supplements were manufactured and provided by the Kellogg's Company specifically for our trial. The cereal products were fortified with folic acid; we estimate that the participants received an additional 140 μg of folic acid/day as a result of the intervention, after taking into account adherence to the intervention. A total of 1,429 participants were randomized between 1990 and 1995 and were followed for an average of 36.8 months; 1,304 (91.3%) completed the study by undergoing 1 or more colonoscopies after randomization. Baseline plasma samples were available on 1,014 (71%) individuals for analyses of folate and homocysteine.

The UDCA trial was a double-blind controlled trial that compared the effect of UDCA versus placebo on adenoma recurrence, among patients who had undergone polypectomy. A total of 1,285 participants were randomized between 1996 and 2000 and were followed for an average of 31.0 months; 1,192 participants (92.8%) completed the study of which 1,111 (86.5%) had baseline plasma samples available for analysis. Although 431 (39%) of the 1,111 participants were randomized before July 1, 1997, a cut-point that is thought to distinguish pre- and postfolate fortification,15 and is supported by the distribution of plasma folate concentrations in our study (data not shown), only 11 individuals had more than two-thirds of their follow-up prior to fortification.

Eligibility criteria were generally similar in both studies. Adenoma recurrence was defined as recurrence of 1 or more colorectal adenomas(s) or carcinoma after randomization. Neither intervention had a significant effect on adenoma recurrence.13, 14 Informed consent was obtained from all participants, and the studies were approved by the University of Arizona Human Subjects Committee and the Institutional Review Board of the local hospital committees.

Procedures

Plasma folate and homocysteine

Blood was drawn from fasting participants and placed in tubes containing heparin. Analyses of folate and homocysteine in plasma were conducted at the University of California, Los Angeles under the direction of one of the authors (S.M.H.). For quality control purposes, pooled plasma samples and commercially available controls were analyzed with each batch of samples for all assay determinations.

Folate was analyzed using the BioRad Quantaphase II Radioassay kit (Bio-Rad, Hercules, CA) in WBF samples, as previously described.4 The intra-assay coefficient of variation (CV) ranged from 4.1 to 5.2% and the inter-assay CV ranged from 4.1 to 8.7%. In the UDCA study, the IMMULITE folic acid analysis kit, which uses a competitive immunoasay methodology, was used (Diagnostic Products Corporation, Los Angeles, CA). The intra-assay CV was 5.2% for higher concentration (13 ng/ml) and 9.0% for lower concentrations (2.1 ng/ml), and the inter-assay CV was 7.4% (13 ng/ml) to 8.9% (1.8 ng/ml).

For homocysteine, the reversed-phase HPLC method by Kuo et al.16 was used for the WBF samples, as previously described.4 The intra-assay CV was 5.7% and the inter-assay CV was 4.4%. In the UDCA study, homocysteine was determined using an IMMULITE 2000 solid-phase, competitive chemiluminescent enzyme immunoassay (Diagnostic Products Corporation). The intra-assay CV was 9.9% for concentrations between 11 and 36 μmol/l, and the inter-assay CV was 9.1% for concentrations lower than 11 μmol/l and 6.3% for those greater than 36 μmol/l.

Because 2 different methods were used to measure folate and homocysteine in our 2 study populations, we randomly selected 20 samples from each study and analyzed these using the IMMULITE method and compared this to RIA for folate and HPLC for homocysteine. Correlation coefficients were 0.94 for folate and 0.88 for homocysteine. Mean (± SD) for RIA and IMMULITE folate methodology were 11.65 (± 6.39) and 11.54 (± 5.37), respectively; corresponding values for HPLC and IMMULITE homocysteine methodology were 9.64 (± 3.13) and 9.03 (± 3.81).

Risk factor and covariate data

Self-administered questionnaires were used to obtain data on sociodemographic variables and other risk factors. Data on adenoma characteristics were obtained from medical record review and the pathology reports, as previously reported.17 Dietary intake was assessed using the Arizona Food Frequency Questionnaire, a computer-scannable questionnaire that contains a vitamin/mineral supplement section, which ascertains current, regular use of multiple vitamin preparations and individual supplements. Data were also collected on the brand of multiple vitamin/mineral supplements and the dose and frequency of single vitamin and mineral supplements. We did not ask about the use of folic acid supplements; however, use of these is low in the age group of our populations.18 The vitamin/mineral supplement database contains over 180 nutritional supplement preparations.

Statistical analysis

Summary data for baseline characteristics were compared using t-tests for continuous variables and χ2analysis for categorical variables. We assessed the main effects of plasma folate and homocysteine on adenoma recurrence in each study (WBF and UDCA), using study-specific quartile cut-points by unconditional logistic regression.19 We also conducted analyses of the combined studies and derived the quartiles from the combined population. A priori we considered potential confounding variables that are suspected or established risk factors for colorectal neoplasia (i.e., age, gender, cigarette smoking, family history of colorectal cancer, aspirin use, etc.). Variables that were independently associated with both adenoma recurrence and baseline plasma markers were initially considered. Year of randomization was included as a covariate in the analyses, given that it was related to folate fortification, and we found that it controls for undefined variables that may have changed in the 2 studies. We then tested additional covariates, including number of colonoscopies conducted during the trial and baseline adenoma characteristics (i.e., number, size, location and histology), since these have been shown to be significant predictors of adenoma recurrence in our study populations.17 Lastly, we considered the simultaneous adjustment of plasma folate and homocysteine as well as potential dietary confounders (vitamins B6, B12, etc.) in the statistical models. Our final results present multivariate models that comprise the final list of covariates that were significant in the model or that resulted in a difference of 10% or more in the main independent variable. We conducted trend analyses by modeling either the quartile value or the continuous variable, based on the best transformation of the data determined by results from fractional polynomials.19

Results

While eligibility criteria for the WBF and UDCA trials were similar, there were some differences in the 2 populations (Table I). Differences in baseline nutrient intakes between the 2 studies mostly reflect slight differences in the dietary questionnaires between the 2 studies and the higher prevalence of multivitamin use in UDCA study; dietary folate intake was also higher in UDCA because those for WBF reflect values prior to the cereal intervention whereas those for UDCA reflect values postfortification of the US food supply. As observed for total folate intake, plasma folate was significantly higher in the UDCA than in the WBF study population, given that the baseline levels in the latter were taken prior to the cereal intervention. Although no difference was observed for homocysteine levels between the 2 studies, when the cut-point of July 1, 1997 was used to more precisely categorize these baseline values, we observed a significantly lower concentration in post- versus prefortification groups (8.97 vs. 9.86 μmol/l; p < 0.05), which is primarily confined to nonusers of multivitamins (9.53 vs. 10.59 μmol/l; p < 0.05) as previously reported.15 The difference in plasma homocysteine concentrations between multivitamin users and nonusers following fortification suggests that users were already consuming enough folic acid to reduce homocysteine levels optimally, as it has been shown that at adequate levels of serum or plasma folate, there is a plateau in the homocysteine concentrations.20

Table I. Baseline Characteristics1 of Study Participants in the WBF and UDCA Trials
 WBF trialUDCA trial
  • 1

    Values are given in means ± SD.

  • 2

    p < 0.05 between studies.

  • 3

    Values in parentheses indicate percentages.

  • 4

    Data missing for 20 participants in UDCA.

  • 5

    Regular use of aspirin in the previous month.

  • 6

    History of previous polyps prior to qualifying colonoscopy; excludes 129 participants with missing data in WBF and 61 in UDCA.

  • 7

    History of colorectal cancer in parent or sibling.

  • 8

    Data missing for 32 participants in UDCA.

  • 9

    Number of colonoscopies after qualifying procedure.

  • 10

    Presence of any adenoma ≥1 cm in size. Data missing for 13 participants in WBF and 1 in UDCA.

  • 11

    Presence of any tubulovillous or villous adenomas. Data missing for 13 participants in WBF and 2 in UDCA.

  • 12

    Presence of any proximal adenoma. Data missing for 13 participants in WBF and 2 in UDCA.

Characteristic
 No. of participants1,0141,111
 Plasma folate (nmol/l)9.4 ± 5.713.5 ± 4.72
 Plasma homocysteine (μmol/l)10.1 ± 3.610.1 ± 3.7
 Age (years)65.3 ± 9.066.2 ± 8.52
 Male (n)715 (70.5)3749 (67.4)
 White4 (n)967 (95.4)1,033 (94.7)
 BMI (kg/m2)27.5 ± 4.527.3 ± 4.7
 Aspirin use5 (n)291 (28.7)302 (27.2)
 Previous polyps6 (n)337 (38.1)497 (47.3)2
 Family history CRC7 (n)157 (15.5)307 (27.6)2
 Current smoker8 (n)139 (13.7)131 (12.1)
 No. of colonoscopies92.1 ± 0.81.5 ± 0.72
Daily dietary intake
 Energy (kcal)1948.6 ± 696.81991.2 ± 819.9
 Dietary folate (μg)326.7 ± 135.2375.3 ± 174.02
 Total folate (μg)480.2 ± 259.0618.4 ± 321.92
 Total vitamin B6 (μg)5.5 ± 13.48.7 ± 21.62
 Total vitamin B12 (μg)14.6 ± 53.429.2 ± 126.42
 Alcohol (g)6.8 ± 13.08.0 ± 13.82
 Methionine (mg)1.5 ± 0.61.4 ± 0.6
 Fiber (g)22.0 ± 9.821.8 ± 10.7
 Calcium (mg)896.3 ± 380.0969.8 ± 458.92
 Multivitamin use (n)378 (37.3)607 (54.6)2
Adenoma characteristics
 Number1.8 ± 1.41.6 ± 0.92
 Large10 (n)424 (42.4)463 (41.7)
 Villous histology11 (n)207 (20.7)228 (20.6)
 Proximal location12 (n)274 (27.4)371 (33.5)2

We first assessed the relation between plasma folate and homocysteine and adenoma recurrence separately for each study. As noted in the Material and Methods, the WBF population received a cereal supplement fortified with folic acid, whereas the UDCA was conducted primarily after fortification of the food supply. Plasma folate was inversely associated with the odds of recurrence in the WBF population (Table II); however, stratification by multivitamin use indicated that the inverse association was confined to nonusers. The odds ratio (OR) for nonusers of multivitamins in the upper vs. the lower quartile of folate concentration was 0.65 (95% confidence interval (CI) = 0.40–1.06). As also shown in Table II, a substantial upward shift in plasma folate concentrations occurred in the UDCA vs. WBF study population because the baseline plasma levels reflect fortification of the food supply and the higher prevalence of multivitamin use in UDCA. For the UDCA group, a similar association was observed among nonusers of multivitamins (OR = 0.56; 95% CI = 0.31–1.02 comparing the upper vs. the lower category; p-trend = 0.04). Among UDCA participants who reported multivitamin use, a weak, nonsignificant positive association between plasma folate and recurrence was observed.

Table II. Plasma Folate and Adenoma Recurrence in the WBF and UDCA Studies among Vitamin Supplement and Nonsupplement Users
 Plasma folate quartile
1234
  • 1

    Values in parentheses indicate percentages.

  • 2

    Adjusted for age, gender, number of colonoscopies and year of randomization.

  • 3

    Values in square brackets indicate 95% confidence intervals.

  • 4

    Trend test conducted by modeling the quartile value or continuous variable as determined by fractional polynomials analysis.

WBF
 Total population (n = 1,014)
  Plasma folate (nmol/l)<5.025.02–8.198.20–12.75>12.75
  No. of recurrence/total133/254 (52.4)1115/254 (45.3)128/253 (50.6)119/253 (47.0)
  OR21.000.61 [0.42–0.89]30.80 [0.55–1.17]0.64 [0.44–0.93]
  p-trend4<0.01
 Nonmultivitamin users (n = 636)
  Plasma folate (nmol/l)<4.054.05–6.336.34–9.05>9.05
  No. of recurrence/total85/159 (53.5)75/159 (47.2)79/159 (49.7)81/159 (50.9)
  OR21.000.70 [0.44–1.13]0.63 [0.39–1.03]0.65 [0.40–1.06]
  p-trend4<0.01
 Multivitamin users (n = 378)
  Plasma folate (nmol/l)<9.619.61–12.7412.75–16.15>16.15
  No. of recurrence/total38/96 (39.6)48/93 (51.6)48/95 (50.5)41/94 (43.6)
  OR21.001.69 [0.91–3.12]1.52 [0.82–2.81]1.03 [0.55–1.92]
  p-trend40.24
UDCA
 Total population (n = 1,111)
  Plasma folate (nmol/l)<10.1010.10–13.5013.60–17.10>17.10
  No. of recurrence/total131/280 (46.8)109/279 (39.1)118/277 (42.6)112/275 (40.7)
  OR21.000.71 [0.49–1.03]0.79 [0.55–1.14]0.79 [0.54–1.15]
  p-trend40.35
 Nonmultivitamin users (n = 504)
  Plasma folate (nmol/l)<8.308.30–11.0011.10–14.00>14.00
  No. of recurrence/total63/126 (50.0)60/134 (44.8)46/120 (38.3)44/124 (35.5)
  OR21.000.88 [0.51–1.51]0.67 [0.38–1.20]0.56 [0.31–1.02]
  p-trend40.04
 Multivitamin users (n = 607)
  Plasma folate (nmol/l)<12.3012.30–15.6015.70–18.80>18.80
  No. of recurrence/total65/152 (42.8)67/154 (43.5)59/151 (39.1)66/150 (44.0)
  OR21.000.99 [0.61–1.61]0.84 [0.52–1.37]1.23 [0.75–2.02]
  p-trend40.33

No material shift in homocysteine concentration was shown between the 2 studies (Table III). As observed for plasma folate, the positive association between homocysteine and adenoma recurrence was confined to nonusers of multivitamins. Individuals who were in the upper quartile compared to the lower quartile of homocysteine concentration and who did not report using multivitamins had about a 2-fold odds of adenoma recurrence in each study. No association between homocysteine level and adenoma recurrence was observed for multivitamin users in either study.

Table III. Plasma Homocysteine and Adenoma Recurrence in the WBF and UDCA Studies among Vitamin Supplement and Nonsupplement Users
 Plasma homocysteine quartile
1234
  • 1

    Values in parentheses indicate percentages.

  • 2

    Adjusted for age, gender, number of colonoscopies and year of randomization.

  • 3

    Values in square brackets indicate 95% confidence intervals.

  • 4

    Trend test conducted by modeling the quartile value or continuous variable as determined by fractional polynomials analysis.

WBF
 Total population (n = 1,014)
  Plasma homocysteine (μmol/l)<7.847.84–9.399.40–11.58>11.58
  No. of recurrence/total110/254 (43.3)1107/253 (42.3)138/254 (54.3)140/253 (55.3)
  OR21.000.90 [0.61–1.31]31.31 [0.89–1.92]1.42 [0.96–2.09]
  p-trend4<0.01
 Nonmultivitamin users (n = 636)
  Plasma homocysteine (μmol/l)<8.458.45–10.0810.11–12.29>12.29
  No. of recurrence/total60/160 (37.5)80/158 (50.6)88/159 (55.4)92/159 (57.9)
  OR21.001.67 [1.03–2.70]1.81 [1.12–2.95]2.25 [1.38–3.66]
  p-trend4<0.01
 Multivitamin users (n = 378)
  Plasma homocysteine (μmol/l)<7.167.16–8.448.45–10.00>10.00
  No. of recurrence/total38/95 (40.0)48/94 (51.1)45/95 (47.4)44/94 (46.8)
  OR21.001.33 [0.71–2.50]1.11 [0.59–2.10]1.01 [0.52–1.97]
  p-trend40.72
UDCA
 Total population (n = 1,111)
  Plasma homocysteine (μmol/l)<7.907.90–9.409.50–11.50>11.50
  No. of recurrence/total101/281 (35.9)123/292 (42.1)128/270 (47.4)118/268 (44.0)
  OR21.001.25 [0.87–1.80]1.58 [1.09–2.30]1.30 [0.88–1.91]
  p-trend40.11
 Nonmultivitamin users (n = 504)
  Plasma homocysteine (μmol/l)<8.308.30–10.0010.10–12.70>12.70
  No. of recurrence/total37/126 (29.4)56/129 (43.4)64/126 (50.8)56/123 (45.5)
  OR21.001.97 [1.12–3.44]2.57 [1.47–4.49]1.93 [1.07–3.49]
  p-trend4<0.01
 Multivitamin users (n = 607)
  Plasma homocysteine (μmol/l)<7.607.60–8.909.00–10.60>10.60
  No. of recurrence/total64/160 (40.0)68/149 (45.6)64/148 (43.2)61/150 (40.7)
  OR21.001.24 [0.76–2.02]1.01 [0.61–1.68]0.88 [0.52–1.47]
  p-trend40.48

Given the large upward shift in baseline plasma folate concentration in UDCA compared to WBF, we were unable to assess the relation of this marker with adenoma recurrence using common cut-points in the 2 populations separately. We therefore conducted analyses of the combined groups and generated quartiles based on this combination (Table IV). These results support our overall findings of stronger associations between plasma folate and homocysteine and adenoma recurrence among nonusers of multivitamins. To examine whether the associations of the plasma biomarkers differed by the severity of the recurrent lesion(s), we conducted analyses of the combined groups using 3 endpoints: (1) multiple adenomas, (2) large adenomas (i.e., lesions with minimum diameter of 1 cm or more) and (3) advanced adenomas (i.e., villous architecture comprising more than 25% of the lesion or large adenomas). Overall, we observed no material differences than those reported in Table IV, with the exception of a suggestion of stronger associations between plasma folate and large adenomas (OR [95% CI] for upper vs. lower quartile: 0.55 [0.33–0.90]) and multiple adenomas (OR [95% CI] for upper vs. lower quartile: 0.58 [0.41–0.83]). No material differences were shown for associations with homocysteine.

Table IV. Plasma Folate and Homocysteine and Adenoma Recurrence in the Combined WBF and UDCA Study Populations among Vitamin Supplement and Nonsupplement Users
 1234
  • 1

    Values in parentheses indicate percentages.

  • 2

    Adjusted for age, gender, number of colonoscopies, year of randomization and study.

  • 3

    Values in square brackets indicate 95% confidence intervals.

  • 4

    Trend test conducted by modeling the quartile value or continuous variable as determined by fractional polynomials analysis.

Plasma folate quartile
 Total (n = 2,125)
  Plasma folate (nmol/l)<7.207.20–11.2111.22–15.38>15.38
  No. of recurrence/total261/530 (49.3)1242/532 (45.5)244/532 (45.9)218/531 (41.1)
  OR21.000.83 [0.64–1.09]30.85 [0.65–1.12]0.74 [0.56–0.98]
  p-trend4<0.01
 Nonmultivitamin users (n = 1,140)
  Plasma folate (nmol/l)<5.405.40–8.298.30–11.90>11.90
  No. of recurrence/total147/286 (51.4)141/284 (49.7)127/287 (44.3)118/283 (41.7)
  OR21.000.78 [0.55–1.12]0.73 [0.50–1.07]0.63 [0.43–0.94]
  p-trend4<0.01
 Multivitamin users (n = 985)
  Plasma folate (nmol/l)<11.0611.06–14.6014.68–18.10>18.10
  No. of recurrence/total107/247 (43.3)123/251 (49.0)99/244 (40.6)103/243 (42.4)
  OR21.001.19 [0.81–1.73]0.84 [0.57–1.23]1.00 [0.67–1.48]
  p-trend40.56
Plasma homocysteine quartile
 Total (n = 2,125)
  Plasma homocysteine (μmol/l)<7.817.81–9.399.40–11.50>11.50
  No. of recurrence/total210/531 (39.6)225/530 (42.5)266/533 (49.9)264/531 (49.7)
  OR21.001.08 [0.83–1.40]1.40 [1.07–1.82]1.37 [1.04–1.79]
  p-trend4<0.01
 Nonmultivitamin users (n = 1,140)
  Plasma homocysteine (μmol/l)<8.388.38–10.0610.07–12.41>12.41
  No. of recurrence/total94/285 (33.0)138/285 (48.4)153/286 (53.5)148/284 (52.1)
  OR21.001.95 [1.35–2.80]2.19 [1.51–3.15]2.24 [1.54–3.25]
  p-trend4<0.01
 Multivitamin users (n = 985)
  Plasma homocysteine (μmol/l)<7.357.35–8.808.81–10.40>10.40
  No. of recurrence/total100/247 (40.5)119/254 (46.9)111/240 (46.3)102/244 (41.8)
  OR21.001.22 [0.83–1.79]1.10 [0.74–1.65]0.89 [0.59–1.36]
  p-trend40.44

Although we conducted analyses using data for multivitamin use at baseline, we also assessed its use at year 1 and 3 in the studies. The prevalence of multivitamin use was only significantly increased at year 3 compared to baseline (46.5% vs. 37.3% in WBF and 61.3% vs. 54.6% in UDCA). Thus, it is quite unlikely that substantial bias occurred as a result of misclassification of multivitamin use during follow-up.

As noted earlier, only 11 individuals in the UDCA study had over two-thirds of their follow-up conducted prior to fortification. When we repeated analyses excluding these 11 individuals, the results for plasma folate and homocysteine in relation to recurrence were not materially altered (data not shown).

We next examined multivitamin use in relation to adenoma recurrence in comparison to quartiles of plasma folate among nonusers of multivitamins (Table V). This analysis allowed us to directly assess the effect of supplement use compared to participants with high and low folate concentrations. Compared to nonmultivitamin users in the lowest plasma folate quartile, the OR (95% CI) for adenoma recurrence for multivitamin users was 0.64 (0.43–0.95) in the WBF population and 0.74 (0.48–1.13) in the UDCA group. These ORs were similar to those for the highest quartile of plasma folate among nonusers of multivitamins, which suggests that supplement use may not confer additional protection among those with higher concentrations.

Table V. Odds Ratio (95% CI) for Adenoma Recurrence Comparing Multivitamin Use to Quartiles of Plasma Folate in Nonmultivitamin Users in the WBF and UDCA Studies
 Nonmultivitamin usersMultivitamin users
1234
  • 1

    Represents mean ± standard deviation.

  • 2

    Values in parentheses indicate percentages.

  • 3

    Adjusted for age, gender, numbers of colonoscopies and year of randomization.

  • 4

    Values in square brackets indicate 95% confidence intervals.

  • 5

    p-trend was calculated by including the median folate level for each category in the multivariate model.

WBF
 Plasma folate (nmol/l)<4.054.05–6.336.34–9.05>9.0513.26 ± 6.071
 No. of recurrence/total85/159 (53.5)275/159 (47.2)79/159 (49.7)81/159 (50.9)175/378 (46.3)
 OR31.000.69 [0.43–1.11]40.62 [0.39–1.00]0.65 [0.40–1.05]0.64 [0.43–0.95]
 p-trend50.07
UDCA
 Plasma folate (nmol/l)<8.308.30–11.0011.10–14.00>14.0015.36 ± 4.461
 No. of recurrence/total63/126 (50.0)60/134 (44.8)46/120 (38.3)44/124 (35.3)257/607 (42.3)
 OR31.000.89 [0.52–1.51]0.70 [0.40–1.21]0.59 [0.34–1.03]0.74 [0.48–1.13]
 p-trend50.09

Discussion

In our study populations, the US FDA-mandated requirement to fortify the food supply with folic acid resulted in a substantial upward shift in the distribution of baseline plasma folate levels and no material reduction in homocysteine, which is consistent with the results of some published studies.15, 21 Our results show that the inverse relation of plasma folate and the positive relation with homocysteine and adenoma recurrence were confined to nonusers of multivitamins in both studies.

Few data exist on the role of homocysteine in relation to colorectal neoplasia.3, 4 Data from the present analyses show that higher homocysteine is a significant risk factor for adenoma recurrence among nonusers of multivitamins. This suggests that despite folate fortification, variation remains in homocysteine that predicts adenoma risk. This might imply that other determinants of homocysteine (i.e., genetic or dietary) are critical. As noted earlier, B6 facilitates the breakdown of homocysteine to cysteine and glutathione.9 Recent data22 indicate that vitamin B6 is an important factor. Assuming that the one-carbon pathway is important in the etiology of colorectal neoplasia, even with additional dietary folate, it does not appear that this will optimally reduce risk through this pathway; therefore, further attention should be given to factors such as B6.

Results of our analyses suggest that nonusers of multivitamins with lower plasma folate concentration have a higher odds of recurrence, even though the distribution of plasma folate was different between the 2 populations. However, because we measured plasma levels prior to the cereal intervention in the WBF study, plasma folate levels most likely increased after the intervention, making the levels more similar in the 2 populations. It is also plausible that one or more factors related to multivitamin use other than folate are responsible for the observed results.

Reasons for the lack of association between plasma folate and homocysteine on adenoma recurrence among multivitamin users are unknown. A possible explanation is that once an individual takes a multivitamin supplement containing folic acid, he or she reaches the minimal risk associated with this factor. Our analyses directly comparing multivitamin users to plasma folate levels in nonusers (Table V) support the notion that only participants with lower plasma folate concentrations have higher odds of recurrence compared to users of multivitamins.

Strengths of the present work include the relatively large sample, availability of plasma biomarker data rather than solely dietary intake data and the prospective nature of the study. Limitations include our inability to study these associations in a nonfortified population, given that both studies involved a fortification strategy. However, it will also be important to evaluate these associations in populations that have been exposed to fortification for longer periods of time. It is also worth noting that due to the use of an existing biospecimen bank, we were unable to analyze red cell folate, which is a more stable biochemical marker of folate status. Limitations related to studies of adenoma recurrence also have to be considered,23 including a short length of follow-up and inability to assess late events in neoplastic progression.

Kim24 has suggested that, on the basis of data from animal studies, folate possesses contrasting properties of tumor prevention as well as tumor promotion, depending on the timing and dosage of this agent. Results of these animal models indicate that folate deficiency enhances colorectal carcinogenesis, and modest levels of folate supplementation above the required doses suppress tumor development.25 However, high levels of folate supplementation do not appear to provide additional protection, and in some cases may enhance carcinogenesis.26, 27, 28 Kim24 further noted that progression of established neoplasia is quite likely with supplemental folate. Although caution when extrapolating animal data to humans is warranted, our data on multivitamin use support the notion that additional folate or multivitamins in an environment of adequate folate confers no added protection. Clearly, the optimal chemopreventive dose and timing of folate supplementation in humans has not been established. Although rates of NTDs have declined in the US and Canada following folic acid fortification of food products,21, 29 whether this mandate will also reduce rates of colorectal cancer is unknown. Furthermore, any long-term consequences of folic acid fortification are yet to be determined.

Intervention and observational studies of folate and folic acid conducted in the postfortification era will face important challenges. If deficient or suboptimal folate status is the primary risk factor for colorectal neoplasia, any protective effect of folate or folic acid will be difficult to uncover, given that the prevalence of folate deficiency has been dramatically reduced after fortification of the food supply.15, 30, 31 Our results suggest that intervention studies assessing the effect of supplemental folate or multivitamins, which contain 400 μg of folic acid, are quite likely to result in null findings unless nonvitamin supplement users with low plasma folate levels are enrolled. In fact, soon-to-be published results of a double-blind controlled trial of 1,021 participants who were randomized to folic acid supplementation (1 mg/day) or placebo and were followed for the endpoint of adenoma recurrence showed no effect on overall recurrence rates. However, in secondary analyses, folic acid supplementation was associated with a significantly higher number of adenomas and a nonsignificantly higher rate of advanced adenomas when compared with placebo (Cole BF, personal communication), which may support Kim's hypothesis.24 Aside from that study, clinical trial evidence on the chemopreventive effects of folate on colorectal neoplasia are sparse, given that these studies have included a small number of participants, and have had short follow-up periods, and at times have included only biomarker endpoints.32, 33, 34 In addition, it is quite unlikely that these studies will shed light into the precise dosage, timing and nature of the association between folate and risk of colorectal neoplasia. Therefore, results from recently completed intervention studies of folic acid will need to be interpreted in the light of these limitations. Furthermore, it is possible that future intervention studies will need to be conducted in countries where fortification has not taken place.

Acknowledgements

We wish to thank Ms. Ellen Graver, Mr. José Guillén-Rodríguez, Mr. Vern Hartz and Ms. Robin Whitacre for their expert assistance. We are indebted to the staff at the Phoenix and Tucson clinic study sites as well as Dr. Janine Einspahr, Ms. Julie Buckmeier and the Analytical Core staff for their valuable contribution.

Ancillary