Circulating folate levels and colorectal adenoma: a case-control study and a meta-analysis.

BACKGROUND/
OBJECTIVES: The relationship between folate and colorectal neoplasia remains controversial. We examined the association between serum folate concentrations and colorectal adenomas in a case-control study of Korean adults and conducted a meta-analysis. SUBJECTS/
METHODS: Our case-control study included 113 pairs of case and control who underwent colonoscopy and provided blood samples. We used multivariable conditional logistic regression models to obtain the odds ratios and 95% confidence interval (CIs). For meta-analysis, we identified the relevant studies by searching the PubMed database up to February 2017, included our case-control study and combined the study-specific relative risks (RRs) using a random-effects model.
RESULTS: In this case-control study, we included 58 men and 55 women with colorectal adenomas and sex and fasting status matched the controls. We did not find any significant association between the serum folate levels and colorectal adenomas in either men or women. For meta-analysis, a total of eleven studies were included in our analysis and classified into two groups; polyp clearance group (PC) for the studies that included participants who underwent endoscopies and had their polyps removed at baseline; and no polyp clearance group (NPC) for the studies that included participants whose histories of endoscopies were unknown or who underwent their first endoscopies. Four PC (1,311 cases and 1,672 non-cases) and eight NPC studies (3,501 cases and 11,347 non-cases) were included. The combined RRs (95% CIs) comparing the bottom with the top categories of circulating folate levels were 1.07 (0.97-1.18) for the NPC group but 1.45 (1.16-1.74) for the PC group.
CONCLUSIONS: Low circulating folate levels were associated with new adenoma formation.

INTRODUCTION

Colorectal cancer is the third most common cancer worldwide [1] as well as in Korea [2]. The incidence of colorectal cancer dramatically increased by 4.6% annually from 1999 to 2013 in Korea [2] partly due to the shift towards a Western lifestyle and diet such as high intake of meat, low intake of vegetables, or sedentary physical activity.

Fruits and vegetables are rich sources of micronutrients including folate and have been actively investigated as food groups that exert preventive effects against the development of colorectal cancer. Furthermore, folate has gained attention as a primary mediator of the anti-cancer effects of fruits and vegetables. Folate is one of the vitamins in the vitamin B series and is known for its role in deoxyribonucleic acid (DNA) methylation, synthesis and replication [3]. Folate may be a key epigenetic regulator because of its primary role as a methyl donor in one-carbon metabolism [4], which controls the transport of the one-carbon moiety (methyl group) to biological methylation reactions.

Given that colorectal adenomas are precursor lesions of colorectal cancer [5] and may share common etiological factors with colorectal cancer, it has been hypothesized that nutrients related to one-carbon metabolism may be involved in colorectal adenoma development. Several studies have examined the associations of folate intake [678910] and circulating folate levels [1112] with colorectal cancer and colorectal adenomas. In the pooled analyses of 13 prospective cohort studies, the dietary folate intake from food was not associated with colon cancer, while the dietary folate intake from food and supplements was associated with a 15% decreased risk of colon cancer [10]. A meta-analysis of intervention trial studies on folic acid supplementation and colorectal adenoma recurrence [13] and nested case-control studies on circulating folate levels and colorectal cancer [11] did not observe significant associations. However, the recent study on folate suggested that the timing of folate interventions relative to colorectal carcinogenesis may be important [14]. A pooled analysis of two observational studies suggested that folate has a role only in the early stage, not in the late stage of carcinogenesis [8].

Based on trials in which the participants had polyps removed and were subsequently followed, the development of recurrent adenomas may indicate the early occurrence of colorectal neoplasia within the sequence of colorectal carcinogenesis [1516]. We hypothesized that folate may be inversely associated with newly developed colorectal adenomas once they are removed, but the association become less clear for advanced adenomas. Therefore, we categorized the studies by the histories of polyp clearance, including our case-control study, into two groups and conducted a meta-analysis of the circulating folate levels and colorectal adenomas. The two groups consisted of polyp clearance group (PC) and no polyp clearance group (NPC); PC was composed of studies that included participants who had undergone endoscopy and had polyps removed prior to the baseline, and NPC was composed of studies that included participants whose histories of endoscopy were unknown or patients who underwent their first endoscopies.

The aim was to examine 1) the association between circulating folate levels and colorectal adenomas in a matched case-control study in Korean adults and 2) whether the association between circulating folate levels and colorectal adenomas varied according to the clearance of previous polyps in a meta-analysis.

SUBJECTS AND METHODS

Case-control study

Study population

The study participants included 382 men and women aged 45-71 years who underwent a colonoscopy from August 2011 to September 2012 and provided blood samples at a university hospital in Daegu, Korea. We excluded participants who had any type of cancer (n=14). We matched the controls (n=113) with cases (n=113) by sex and fasting status. Written informed consent was obtained from all the participants. This study was approved by the Institutional Review Board of Daegu Catholic University Medical Center (No. CR-11-069-RES-003-R).

Ascertainment of colorectal adenomas

The subtype, size, and number of colorectal adenomas were determined by colonoscopy and histological examination. The colorectal polyps were classified into adenomatous, hyperplastic, and other nonadenomatous polyps. The adenomatous polyps were included as cases. A total of 113 (58 men and 55 women) adenoma cases were identified.

Measurement of the serum folate levels

Blood samples were collected from each participant between January and February 2013. The samples were centrifuged and sent on ice to the Neodin Medical Institute (Seoul, South Korea). The serum folate concentrations of the participants were measured with an electrochemiluminescence immunoassay at the Neodin Medical Institute. All laboratory technicians were blinded to the case status. The intra-assay coefficient of variation was 3.7-5.4%.

Information about potential risk factors for colorectal adenomas

The participants were asked about their sociodemographic characteristics, medical conditions, family histories of colorectal cancer, and lifestyles. Dietary information was assessed with a validated food frequency questionnaire [17]. The heights and weights of the participants were measured with the X-scan Plus II Professional (Jawon Medical, Gyeongsan, South Korea), and body mass index (BMI; kg/m2) was calculated by dividing the weight in kilograms by the square of the height in meters.

Statistical analysis

The distributions of the characteristics were compared between the case group and control group with the Mantel-Haenszel test and paired t-test. The means and standard deviations were calculated for continuous variables, and the frequencies and percentages were calculated for categorical variables.

To examine the associations between the serum folate levels and the prevalence of colorectal adenomas, we calculated the odds ratios (ORs) and two-sided 95% confidence interval (CIs) using multivariable conditional logistic regression models. Cases were categorized according to the tertiles based on the distribution among the controls. In the multivariable models, we adjusted for age (continuous; years), BMI (continuous; kg/m2), family history of colorectal cancer (yes, no), history of colorectal polyps (yes, no), total energy intake (continuous; kcal/d), education level (less than or equal to elementary school graduate, middle school graduate, high school graduate, more than or equal to college graduate), marital status (spouse, spouseless), frequency of red meat intake (less than or equal to once per month, 2-4 times per month, more than or equal to 2 times per week), ethanol intake (0, 0-10, ≥10 g/d in men; 0, 0-5, ≥5 g/d in women), and smoking (0, 0-15, ≥15 cigarettes/d in men; never smoker, ever smoker in women). To test for trends, the participants were assigned the median value of their tertile, and this variable was used as a continuous term in the model. We also examined whether the associations between the serum folate concentrations and colorectal adenoma prevalence differed according to age (<60, ≥60 years), BMI (<25, ≥25 kg/m2), ethanol intake (0, 0-10, ≥10 g/d), smoking status (never smoker, ever smoker), and frequency of red meat intake (less than or equal to once per month, 2-4 times per month, more than or equal to 2 times per week). We used the likelihood ratio test to examine the null hypothesis that there were no interactions by potential risk factors. All statistical analyses were done with SAS, version 9.3 (SAS Institute, Inc., Cary, NC, USA). Two-sided P-values of <0.05 were considered statistically significant.

Meta-analysis

Identification and selection of studies for the meta-analysis of circulating folate and colorectal adenomas

We performed a PubMed search for epidemiological studies that examined the association between circulating folate levels and colorectal adenomas for the period until February 28, 2017. We used the following search term ”(((((plasma folate) OR serum folate) OR red blood cell folate) OR erythrocyte folate) OR whole blood folate) AND colorectal adenoma”. Among the total of fifty-nine extracted articles, we included the articles that met the following criteria: (1) the exposure of interest was circulating folate level, including plasma folate, serum folate, red blood cell folate, and whole blood folate; (2) the outcome of interest was either an initial or recurrent colorectal adenoma; (3) the association between circulating folate levels and colorectal adenomas was presented with either the risk ratio (RR) or OR estimates and the 95% CIs or we were able to calculate the OR or RR and 95% CI; and (4) the articles were human studies and were published in English. Two authors (Youn J and Lee JE) independently assessed the eligibility criteria. The circulating folate levels were measured at baseline or in cross-sectional or case-control design settings. Regarding the randomized double-bind clinical trial studies of supplements [12181920], if the supplement in the intervention group contained folic acid, we used the RRs and 95% CIs for colorectal adenoma recurrence only in the placebo group [18] or calculated the crude RRs and 95% CIs using the number of cases and non-cases in the placebo group [1920]. One study combined a wheat bran fiber trial and ursodeoxycholic acid trial but reported the levels of folate at baseline; therefore, we included the RRs and 95% CIs of all the participants according to the reported baseline folate levels [12]. We did not include one randomized double-blind clinical trial of a folic acid supplement that did not report the estimates or numbers of cases and participants among the placebo group [21]. When the same data were reported in more than one study [2223], we included the study with the greater number of participants [22]. When the results were presented alone [24] or in a pooled analysis of two datasets [12], we included the estimates from the pooled analysis [12]. For three studies, we calculated the crude ORs or RRs and 95% CIs using the numbers of cases and non-cases [192025]. The following data were extracted from the selected articles: the first author, published year, country, study design, sex, the number of cases and controls or total, age at blood draw, specimen type, endpoint, type of endoscopy, the ORs or RRs and 95% CIs according to circulating folate level category, and adjusted covariates. We performed this meta-analysis according to the Meta-analysis of Observational Studies in Epidemiology guidelines [26].

We combined the ORs or RRs and 95% CIs in a random effects model developed by DerSimonian and Laird [27] and tested for heterogeneity using Q and I2 statistics [28]. Weight was allotted to the individual studies based on the inverse proportion of their variances.

To construct dose-response models, we estimated the ORs or RRs per 10 nmol/L increase in circulating folate level using generalized least squares [29] or a variance-weighted least squares [30]. If units of ng/mL were used, we converted those values to nmol/L (1 ng/mL = 2.265 nmol/L). We used the median of each circulating folate level category and assumed that the level had the same amplitude as the neighboring categories if the category was open-ended. We did not include studies that did not report circulating folate levels in each category or that compared only two circulating folate level categories [29], which resulted in the inclusion of six studies for the dose-response models.

We conducted a meta-regression analysis to investigate whether the association between circulating folate levels and colorectal adenomas differed according to the PC or NPC groups, sex, geographic region (Europe and Americas, Asia), specimen type (plasma, serum), study design (observational study, clinical trial with observational data analysis), age at blood draw (mean or median <60, ≥60 years), and folic acid fortification at blood draw (yes, no). To assess the publication bias, we conducted an Egger's asymmetry test [31]. All statistical analyses were done with STATA11 (Stata Corp., College Station, TX, USA). Two-sided P-values of <0.05 were considered statistically significant.

RESULTS

Case-controlstudy

Table 1 presents a comparison of the general characteristics between the case group and control group. The cases were more likely to drink alcohol compared with the colorectal adenoma-free controls. However, there were no statistically significant differences in the other variables between the case group and control group. The mean levels of circulating folate were 8.2 ng/mL among the men and 9.9 ng/mL among the women in our study. The minimum level of circulating folate was 3.1 ng/mL, in other words, there were no participants who were folate deficient (<3 ng/mL) in this population.

Table 1

General characteristics of participants according to colorectal adenoma status

BMI, body mass index; SD, standard deviation.

1)P-values were obtained by using paired t-test for continuous variables or Mantel-Haenszel test for categorical variables.

2)Mean ± SD for continuous variables and numbers (percentage) for categorical variables

3)The total number of participants in each category were not equal to the total number of participants because some participants did not provide the relevant information.

Table 2 shows the associations with the serum folate levels and colorectal adenomas for men and women. We found no significant associations in men or women. In the age-adjusted model, the ORs (95% CIs) for the highest tertile compared with the lowest tertile were 0.47 (0.19-1.17; P for trend=0.13) in men and 0.66 (0.25-1.70; P for trend=0.36) in women. In the multivariable model with adjustments for additional potential confounding factors, the ORs (95% CIs) for the highest tertile compared with the lowest tertile were 0.89 (0.19-4.13; P for trend=0.98) for men and 1.20 (0.32-4.46; P for trend=0.78) for women. When we examined whether the associations were modified by age, BMI, daily ethanol intake, smoking status, or the frequency of red meat intake, there were no significant interactions according to these factors (data not shown; P for the interactions ≥0.22).

Table 2

Odds ratios and 95% confidence intervals for colorectal adenoma according to serum folate levels

OR, odds ratio; CI, confidence interval.

1)None of the study participants were deficient (< 3 ng/mL) in circulating folate levels.

2)Models were adjusted for age (continuous, years).

3)Multivariable models were adjusted for age (continuous, years), body mass index (continuous, kg/m2), family history of colorectal cancer (yes, no), history of colorectal polyps (yes, no), total energy intake (continuous, kcal per day), education level (less than or equal to elementary graduate, middle school graduate, high school graduate, more than or equal to college graduate), marital status (spouse, spouseless), frequency of red meat intake (less than or equal to once per month, 2-4 times per month, more than or equal to 2 times per week), ethanol intake (grams per day; 0, 0-10, ≥10 in men; 0, 0-5, ≥5 in women), and smoking (cigarettes per day; 0, 0-15, ≥15 in men; never smoker, ever smoker in women)

A total of eleven articles reporting 4,812 colorectal adenomas and 13,019 non-cases were included in the meta-analysis (Table 3). Of the fifty-nine identified articles, twenty-four articles did not examine the association between circulating folate levels and colorectal adenomas; two articles were reported in non-English languages, and thirteen articles were not human studies. Out of twenty articles that examined the association between circulating folate levels and colorectal adenomas, ten articles were excluded because of the absence of estimates (n=7) and data overlap (n=3), and our study was included (Fig. 1). Consequently, three cross-sectional [323334], three case-control [222535], and four clinical trial studies [12181920] plus our study were included in the present meta-analysis. Among these four trial studies, the major endpoint of three of the trials was recurrent colorectal adenomas [121820]. Folic acid was the treatment reagent for three trials [181920]. Three trials included participants who underwent colonoscopies or sigmoidoscopies and had their adenomatous polyps removed [121820], and the other study followed the participants without polyp removal [19].

Table 3

Included studies of circulating levels of folate and colorectal adenoma

NPC, no polyp clearance; PC, polyp clearance; OR, odds ratio; RR, risk ratio; C, men and women; M, men; W, women; Q, quantile; BMI, body mass index; UDCA, ursodeoxycholic acid trial; WBF, wheat bran fiber trial.

1)The number of cases/total participants

2)Crude odds ratio was calculated.

3)Crude risk ratio was calculated.

Fig. 1

Flow chart of publication selection for the meta-analysis of the association between circulating folate levels and colorectal adenoma

We observed a statistically significant inverse association between circulating folate levels and colorectal adenomas in the meta-analysis (Fig. 2). The combined RR (95% CI) of the comparison of the bottom and top categories of the circulating folate levels was 1.15 (95% CI: 1.03-1.27). When we separated the PC and NPC groups, the RRs (95% CIs) when comparing the bottom and top categories of the circulating folate levels were 1.07 (0.97-1.18) and 1.45 (1.16-1.74) for the NPC and PC groups, respectively. We found that the estimates were different between the NPC and PC groups (P for difference = 0.06). When we examined the publication bias, P values for the publication bias were 0.097 and 0.988 for the NPC and PC groups, respectively.

Fig. 2

Forest plot for bottom versus top categories of circulating folate levels in relation to colorectal adenoma by PC or NPC groups.

NPC indicated no polyp clearance group and PC indicated polyp clearance group. M, W, and C represented men, women, and combined sex, respectively. The black circles indicate the study specific relative risks; the horizontal lines indicate the 95% confidence intervals. The gray squares represent the study specific weights, which are inverse of the variance. The dash line indicates the overall combined RR and the diamonds indicate the 95% CIs for the combined RRs. P for difference between NPC and PC groups was 0.06.

When we examined the dose-response relationship in the meta-analysis of six studies [1218223235] including our data, we also observed a statistically significant inverse association between the circulating folate levels and colorectal adenomas. Based on a 10 nmol/L increase in the circulating folate levels, the combined RRs (95% CIs) for colorectal adenomas were 0.94 (0.91-0.98) for all, 0.95 (0.89-1.01) for the NPC group, and 0.94 (0.89-0.98) for the PC group (Supplementary Fig. S1).

We conducted a meta-regression to examine whether the association between the circulating folate levels and colorectal adenomas differed according to potential interactions (Table 4). Regarding the sex-specific association, we included four studies, including ours [192533], and we observed a more pronounced association among men than among women. The associations did not vary according to the other factors.

Table 4

Combined relative risk (RR)s and 95% confidence interval (CI)s for bottom versus top categories of circulating folate levels in relation to colorectal adenoma according to sex, geographic region of study, specimen type, study design, age at blood draw, and folic acid fortification at blood draw

RR, relative risk; CI, confidence interval.

1)A meta-regression analysis was used to estimate P value for difference.

2)Our study was included.

3)We included each of odds ratios of ursodeoxycholic acid trial and wheat bran fiber trial in Martinez study according to folic acid fortification at blood draw.

4)We categorized studies into two groups (yes, no) based on the presence of mandatory folic acid fortification legislation.

DISCUSSION

We found that low circulating folate levels were associated with new colorectal adenoma formation. A higher colorectal adenoma occurrence was observed in those whose polyps were removed (PC group), but the association was not significant in the NPC group. Overall, low circulating folate levels were associated with a higher prevalence or occurrence of adenomas in the meta-analysis of 11 epidemiologic studies. To our knowledge, only a few epidemiologic studies have investigated the association between folate and colorectal cancer in Korea [36], and no Korean studies have examined the association of colorectal adenomas with circulating folate levels. Although we found no association in a case-control study of Korean adults, our meta-analysis supports the potential protection of folate against the new development of colorectal adenomas.

The circulating folate levels increased with increasing intake of folate from food and supplements. A recent review of 17 articles suggested that the correlation was stronger when folate from supplement use was included than when only dietary folate from foods was considered [37]. A recent meta-analysis of eight nested case-control studies found no associations between circulating folate levels and colorectal cancer [11]. A combined study of three large intervention trials suggested that folic acid supplementation may reduce colorectal adenoma recurrence among those with lower folate circulating levels but did not find any overall effect of folic acid supplementation on colorectal adenoma recurrence [7].

Evidence from experimental and epidemiological studies conducted in the 1990s and early 2000s suggests that folate deficiency might lead to carcinogenesis possibly by the misincorporation of uracil into DNA [38], decreased DNA methylation [39], and impaired DNA repair processes [40]. However, recent studies have provided evidence indicating that the effect of folate on colorectal carcinogenesis might not be this simple. A large randomized clinical trial involving the intake of 1 mg of folic acid per day reported an increase in the number of advanced colorectal adenomas [41]. The results regarding the relation between circulating folate levels and colorectal neoplasia are not consistent [124243444546]. Some studies have found positive associations between circulating folate levels and colorectal cancer [46], whereas other studies have found inverse [47] or no associations [7424548]. These inconsistent findings across studies may reflect heterogeneities between the studies including differences in the baseline folate levels, which could partially be due to folic acid fortification and supplement use, study design, and influences from confounding factors.

The findings from our meta-analysis suggest the potential benefit of folic acid for colorectal adenomas. We observed a stronger association when we combined the studies that included participants who had undergone endoscopies and had their adenomatous polyps removed at baseline. After the clearance of polyps at baseline, we observed a greater recurrence with low folate levels, which may support the hypothesis that early intervention with folate is beneficial in the prevention of colorectal carcinogenesis. Although folic acid supplementation increased or did not affect the recurrence in the trials [204149], the low baseline folate levels in the non-supplementation group resulted in a higher rate of colorectal adenoma recurrence in our meta-analysis of three studies [121820]. The pooled analysis of the three trials of folic acid supplementation suggested a potential interaction according to the baseline folate status, i.e., a possible benefit for those with low folate levels and a lack of effect from folate supplementation in those with high folate levels [7]. A recent pooled analysis of the Nurses' Health Study and the Health Professionals Follow-Up Study found that folate intake 12-16 years before colorectal cancer diagnosis was inversely associated with colorectal cancer risk, but intake in the recent past was not associated with colorectal cancer risk, suggesting the importance of timing of folate intake in relation to colorectal carcinogenesis [8].

In our case-control study of Korean adults, we did not find a statistically significant association between the serum folate levels and the prevalence of colorectal adenomas. The possible explanations for our findings include the small sample size, the small variance in the folate levels, a potential reverse causation and the potentially minimal influence of the folate levels compared with other risk factors such as alcohol consumption and smoking. Among men, we observed a suggestive inverse association in the age-adjusted model, but this association was attenuated after adjusting for confounding factors. Due to the high prevalence of smoking and alcohol consumption among Korean men, it is possible that the effect of folate by itself might not have been sufficiently large to detect in our study.

In conclusion, we observed that low folate levels were associated with a higher prevalence or occurrence of colorectal adenomas in a meta-analysis of eleven epidemiologic studies, and this association was apparent in a meta-analysis of studies in which participants were followed after polyp removal at baseline. Our study suggests that the adverse effect of low folate may be relevant to the early stage of colorectal cancer.