Associations of Dietary Intakes and Serum Levels of Folate and Vitamin B-12 with Methylation of Inorganic Arsenic in Uruguayan Children: Comparison of Findings and Implications for Future Research

Abstract

Background:

In the human body, inorganic arsenic (iAs) is methylated via the one-carbon cycle to form monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). Lower proportions of iAs and MMA, and higher proportions of DMA in urine indicate efficient methylation; formation of DMA is thought to detoxify iAs and MMA. Studies on folate, vitamin B-12 and iAs methylation yield mixed findings, depending on whether folate and vitamin B-12 were assessed from diet, supplements, or using a blood biomarker.

Objective:

First, to compare the associations of serum concentrations and estimated intake of folate and vitamin B-12 with indicators of iAs methylation. Second, to highlight the implications of these different B-vitamin assessment techniques on the emerging evidence of the impact of dietary modifications on iAs methylation.

Methods:

The study was conducted among ~7-year-old children from Montevideo, Uruguay. Serum folate and vitamin B-12 levels were measured on the Horiba ABX Pentra 400 analyzer; urinary arsenic was measured using High-Performance Liquid Chromatography on-line with Inductively Coupled Plasma Mass Spectrometry. Dietary intakes were assessed using the average of two 24-hour dietary recalls. Linear regressions assessed the associations of serum levels, and dietary intakes of folate (n=237) and vitamin B-12 (n=217) with indicators of iAs methylation. Models were adjusted for age, sex, body mass index, total urinary arsenic, and rice intake.

Results:

Serum folate and vitamin B-12 levels were above the adequacy threshold for 99% of the participants. No associations were observed between serum folate, serum vitamin B-12, or vitamin B-12 intake and iAs methylation. Folate intake was inversely associated with urinary %MMA [β (95% confidence interval): −1.04 (−1.89, −0.18)].

Conclusion:

Additional studies on the role of B-vitamins in iAs methylation are needed to develop a deeper understanding of the implications of assessing folate and vitamin B-12 intake compared to the use of biomarkers. Where possible, both methods should be employed because they reflect different exposure windows and inherent measurement error, and if used individually, will likely continue to contribute to lack of consensus.

1. Introduction

Methylation of inorganic arsenic (iAs) occurs through the one-carbon metabolism cycle, resulting in the excretion of monomethylarsonic acid (MMA), and dimethylarsinic acid (DMA) in urine, besides some unmethylated iAs (Vahter, 2001). The proportions of urinary iAs, MMA, and DMA reflect a person’s efficiency in methylating iAs to its metabolites, MMA and DMA, with higher %DMA indicating more efficient methylation (Vahter, 2001). The majority of B-vitamins play an important role in the one-carbon cycle, which also methylates iAs. Specifically, folate acts as a methyl group donor, and vitamin B-12 acts as a cofactor in the one-carbon cycle (Hall et al., 2009b; Selhub, 2001). Importantly, folate supplementation has been associated with more efficient iAs methylation in randomized trials (Bozack et al., 2019; Gamble et al., 2006; Gamble et al., 2007; Peters et al., 2015).

Some findings from cross-sectional studies on the associations between folate and iAs methylation efficiency, assessing folate intake with questionnaires (Kordas et al., 2016; López-Carrillo et al., 2016; Steinmaus et al., 2005) or status with biomarkers (Gamble et al., 2005; Hall et al., 2009b; Kurzius-Spencer et al., 2017), have been consistent with the trial results on folate supplementation (Bozack et al., 2019; Gamble et al., 2006; Gamble et al., 2007; Peters et al., 2015). Other cross-sectional study findings have been inconsistent with trial findings, regardless of whether dietary folate intake was assessed (Argos et al., 2010; Heck et al., 2007; Kurzius-Spencer et al., 2017; Spratlen et al., 2017) or a folate biomarker was used (Kurzius-Spencer et al., 2017; Li et al., 2008). Mixed findings for the relationship between folate intake/status and iAs methylation efficiency have been obtained regardless of the arsenic exposure level in a given geographical area. Similarly, cross-sectional studies have yielded mixed evidence for associations between vitamin B-12 intake and iAs methylation efficiency (López-Carrillo et al., 2016; Spratlen et al., 2017; Steinmaus et al., 2005) and for associations between vitamin B-12 status assessed in plasma or serum samples and iAs methylation (Gamble et al., 2005; Hall et al., 2009a; Hall et al., 2009b; Kurzius-Spencer et al., 2017).

Previously, we assessed the relationships between dietary folate (Kordas et al., 2016) and vitamin B-12 intake (Desai et al., 2020b) and iAs methylation among school-aged children participating in the Salud Ambiental Montevideo study. We observed an inverse association with folate intake (Kordas et al., 2016) but not vitamin B-12 intake (Desai et al., 2020b). Serum folate and vitamin B-12 concentrations were measured thereafter in a subset of participants with dietary data. This short communication has two aims, first, to compare the associations of serum concentrations and estimated intake of folate and vitamin B-12 with indicators of iAs methylation, and second, to highlight the implications of these different B-vitamin assessment techniques on the emerging evidence of the impact of dietary modifications on iAs methylation.

2. Methods

2.1. Study Setting and Participant Recruitment

Salud Ambiental Montevideo is an ongoing cohort study conducted among schoolchildren in Montevideo, Uruguay; the present cross-sectional analysis used the baseline data collected on 357 children between 2009 and 2013. Children aged ~7 years and their mothers were enrolled, as described previously (Desai et al., 2018). Of 307 available samples, 275 had sufficient volume of serum to measure serum folate and 252 had sufficient volume to measure B-12 concentrations. Further exclusion of participants with missing data for the covariates of interest resulted in a complete case sample of 237 children for analyses involving serum folate and 217 for analyses involving serum vitamin B-12. The research protocol was approved by the Institutional Review Boards at the Catholic University of Uruguay, University of the Republic of Uruguay, Pennsylvania State University, and the State University of New York at Buffalo.

2.2. Measurements

2.2.1. Urinary arsenic measurement

We measured urinary concentrations of iAs, MMA, and DMA and calculated the sum concentration, hereafter called total urinary arsenic. The proportions of urinary %iAs, %MMA, and %DMA in urine were used as indicators of iAs methylation; sample collection and detailed arsenic assessment assay methods are described elsewhere (Desai et al., 2018). The limit of detection was 0.1 μg/L for inorganic arsenic (III) and MMA, 0.2 μg/L for DMA, and 0.3–0.5 μg/L for inorganic arsenic (V). The intra- and inter-assay coefficients of variation were ~4%. Seven urine samples (2.1%) were below limit of detection for inorganic arsenic (III) and 26 (7.9%) were below the limit of detection for inorganic arsenic (V). The measured values were used in the statistical analyses. The urinary arsenic metabolite concentrations were adjusted for urinary specific gravity to account for variations in the hydration status of participants.

2.2.2. Serum folate and serum vitamin B-12 assays

Fasting venous blood samples were collected from participants as described previously (Desai et al., 2020a). Folate and vitamin B-12 concentrations were measured in archived serum samples that had sufficient volume. The assays were carried out together in the Clinical Biochemistry and Molecular Diagnostics Research Laboratory at the University at Buffalo. If the available sample volume exceeded 300 μl, both folate and vitamin B-12 assays were carried out, with the vitamin B-12 assay being done first. If the sample volume was between 150 μl and 300 μl, only the folate assay was completed.

Serum folate levels were measured on the Horiba ABX Pentra 400 analyzer using the U.S. Food and Drug Administration (FDA) approved Folate Assay diagnostic reagent kits, calibrators and quality control materials from Diazyme Laboratories (Poway, CA). Sample preparation and treatment was done according to manufacturer instructions (Laboratories, 2019a). The limit of detection was 0.91 ng/mL. Quality control materials were analyzed at the outset of testing and repeatedly throughout the analytical run. The linear range (lower limit of quantification to upper limit of quantification) for serum folate assays were 2.0 to 20.0 ng/mL. Folate controls were at 2 and 4 ng/mL with coefficients of variation (CVs) of 15% and 9%, respectively, across all study batches. Of the 275 samples with sufficient volume, 5 were above the limit of detection. Values above the detection limit were replaced with the highest observed values.

Serum vitamin B-12 levels were also analyzed on the Horiba ABX Pentra 400 analyzer using FDA approved Vitamin B-12 Assay diagnostic reagent kits, calibrators and quality control materials (Diazyme Laboratories, Poway, CA). Sample preparation and treatment was done according to manufacturer instructions (Laboratories, 2019b). The limit of detection was 63.3 pg/mL. Quality control materials were analyzed at the outset of testing as well as repeatedly throughout the analytical run. Vitamin B-12 controls were at 580 and 1100 pg/mL with CVs of 8% and 2% respectively across all study batches. The linear range (lower limit of quantification to upper limit of quantification) for vitamin B-12 assays were 96.7 to 2000 pg/mL. Of the 252 with sufficient volume to carry out the assay, 23 were below the limit of detection. Values below the detection limit were entered as the detection limit divided by the square root of two.

2.2.3. Dietary Assessment

Two 24-hour dietary recalls were conducted with the child’s mother or another caregiver who was aware of the child’s diet. Details of dietary assessment in this study have been published previously (Desai et al., 2020b; Kordas et al., 2018; Kordas et al., 2016). B-vitamin intake was calculated using either the Uruguayan nutrient database or the United States Department of Agriculture (USDA) National Nutrient Database for Standard Reference, Release 28 (Version Current: September 2015) for the foods that were not listed in the Uruguayan database (Instituto de Nutrición de Centro América y Panamá (INCAP) and Organización Panamericana de la salud (OPS), 2012; Instituto Nacional de Alimentación (INDA), 2010; USDA, 2018). Intake of each B-vitamin was adjusted for total energy intake and expressed as intake per 1000 kcal/day.

2.2.4. Anthropometry

Anthropometric measures, including children’s height and weight were measured by trained nurses using a portable stadiometer (Seca 214, Shorr Productions, Colombia, MD) and a digital scale (Seca 872, Shorr Productions, Colombia, MD), respectively. Body Mass Index (BMI) was calculated as the measured weight in kilograms divided by the square of height measured in meters.

2.3. Statistical Analyses

Analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). Demographic, anthropometric, and biochemical characteristics were compared across participants in the original study (n=290), participants with complete data on serum folate (n=237; serum folate group), and those with complete data on both serum folate and serum vitamin B-12 (n=217; complete serum group). Linear regression analyses were conducted to assess the associations of dietary intakes, and serum levels of folate (n=237) and vitamin B-12 (n=217) with the proportions of urinary metabolites of iAs (%iAs, %MMA, %DMA) in separate models. Both the dietary intake variables were treated as continuous, as well as dichotomized at the respective medians (220 μg/1000 kcal for folate, and 1.6 μg/1000 kcal for vitamin B-12) for consistency with previous publications from our group (Desai et al., 2020b; Kordas et al., 2016). Covariates were chosen using a definition-based approach and were based on existing literature (Heck et al., 2007; Kurzius-Spencer et al., 2017; Steinmaus et al., 2005); models were adjusted for age, sex, BMI, total urinary arsenic, and rice intake (to account for the fact that rice is a source of both iAs and DMA, which would contribute to urinary iAs and DMA levels independently of methylation efficiency). While certain seafood is also a source of DMA (Navas-Acien et al., 2011), models were not adjusted for seafood intake because only 18 participants reported consuming any seafood (median intake: 25.0 g, range: 10.0, 100.0). Because rice may contain folate and vitamin B-12, correlations between these variables were assessed with the aim of further understanding the relationship of rice consumption with the concentrations of folate and vitamin B-12.

3. Results

Medians (range) and frequencies (n, %) of demographic, anthropometric, and biochemical characteristics of participants in the serum folate group (n=237), the complete serum group (n=217), and the 290 participants from the original study are presented in Supplemental Table 1. The medians (range) of serum folate and vitamin B-12 concentrations were 11.3 (1.90, 19.1) ng/ml and 455 (44.8, 1144) pg/ml, respectively among participants in the complete serum group. The medians (range) of dietary folate and vitamin B-12 intake in the complete serum group were 219 (75.0, 379) μg/1000 kcal and 1.61 (0.34, 8.30) μg/1000 kcal, respectively.

Table 1 shows the associations of serum folate and vitamin B-12 concentrations, and those of dietary folate and vitamin B-12 intakes with the proportions of urinary metabolites of iAs. No associations were observed for serum variables. The dichotomized folate intake variable was inversely associated with urinary %MMA [β (95% confidence interval): −1.04 (−1.89, −0.18)], the continuous variable was not. Vitamin B-12 intake was not associated with any of the metabolites of urinary iAs. Results remained unchanged even after the rice intake variable was removed from the models (Supplemental Table 2). Furthermore, the correlations between rice intake and the concentrations of both folate and vitamin B-12 were very low and not statistically significant (results not shown).

Table 1:

Associations of serum folate (n=237) and serum vitamin B-12 (n=217) concentrations, and those of dietary folate intake^* (n=237) and vitamin B-12 intake^* (n=217) with the proportions of urinary metabolites of inorganic arsenic among study participants.

	%iAs		%MMA		%DMA
	β (95% CI)	β1 (95% CI)	β (95% CI)	β1 (95% CI)	β (95% CI)	β1 (95% CI)
Serum Variables
Folate, ng/ml	−0.13 (−0.35, 0.09)	−0.14 (−0.34, 0.07)	0.05 (−0.09, 0.19)	0.05 (−0.08, 0.19)	0.08 (−0.21, 0.36)	0.08 (−0.18, 0.34)
Vitamin B-12, pg/ml	−0.0003 (−0.004, 0.003)	−0.001 (−0.004, 0.002)	−0.0004 (−0.003, 0.002)	−0.001 (−0.003, 0.001)	0.001 (−0.004, 0.005)	0.002 (−0.002, 0.01)
Dietary Intake Variables
Folate, continuous	0.01 (−0.01, 0.02)	0.002 (−0.01, 0.01)	−0.003 (−0.01, 0.004)	−0.01 (−0.01, 0.001)	−0.003 (−0.02, 0.01)	0.004 (−0.01, 0.02)
Folate, categorical	0.76 (−0.69, 2.22)	0.41 (−0.93, 1.76)	−0.83 (−1.74, 0.08)	−1.04 (−1.89, −0.18)	0.06 (−1.81, 1.94)	0.62 (−1.05, 2.28)
Vitamin B-12, continuous	0.71 (−0.25, 1.67)	0.73 (−0.17, 1.64)	0.00 (−0.60, 0.60)	−0.05 (−0.63, 0.52)	−0.70 (−1.93, 0.53)	−0.67 (−1.78, 0.44)
Vitamin B-12, categorical	0.99 (−0.56, 2.55)	0.53 (−0.94, 2.00)	0.31 (−0.67, 1.28)	0.17 (−0.76, 1.10)	−1.29 (−3.27, 0.69)	−0.69 (−2.49, 1.11)

^*Adjusted for energy and expressed as μg/1000 kcal;

¹Models adjusted for age, sex, BMI, specific gravity adjusted sum of iAs metabolites, rice intake. Abbreviations: iAs-inorganic arsenic, MMA-monomethylarsonic acid, DMA-dimethylarsinic acid.

4. Discussion

We found no evidence for an association between serum concentrations of folate and vitamin B-12, or dietary intake of vitamin B-12, and the percentages of the metabolites of urinary iAs in Uruguayan children. Dietary intake of folate, dichotomized, was inversely associated with urinary %MMA. Previously, we found no association between dietary intake of vitamin B-12 and urinary iAs metabolites (Desai et al., 2020b), but did observe an inverse association of urinary %MMA and a positive association of urinary %DMA with dietary folate intake dichotomized at the median (217 μg/1000 kcal/day) in a larger group of Uruguayan children (Kordas et al., 2016) that also included the children in the present study.

It is unclear why we observed an association between urinary iAs metabolites and dietary folate, but not serum folate. One potential explanation is that the folate-iAs methylation association itself reflects acute rather than habitual folate exposure. We observed a weak correlation between dietary folate and serum folate (r=0.07), which is lower than reported by previous studies (ranging from 0.49 to 0.69 when habitual folate intake was assessed with a food frequency questionnaire (Green et al., 1998; Hickling et al., 2005; Pufulete et al., 2002) and 0.17 to 0.31 with 24-hour dietary recalls (Chew et al., 2011; Scholl et al., 1996)). Although serum folate is a marker of short-term intake and is sensitive to recent dietary changes (Green, 2011), these previous correlations suggest that serum folate may reflect habitual diet more so than very recent diet assessed via 24-hour recalls. Another consideration is the variation in the two measures being compared; the greater the variation, the likelihood of a stronger correlation coefficient increases. The variation in both serum and dietary measures in our study was low, as presented in Supplemental Table 1, which could have contributed to the low correlation.

Furthermore, we measured arsenic exposure and iAs methylation efficiency in a single spot urine sample. Urinary arsenic is a marker of short term exposure, although existing literature does suggest that the relative distribution of urinary iAs metabolites remains approximately constant (NRC, 1999; Vahter, 2001). However, recent dietary sources of arsenic may influence the proportions of urinary iAs metabolites to some extent. Therefore, it is possible that our observed association between iAs and dietary folate reflects two acute exposures (dietary folate and iAs from diet), whereas these acute exposures and their relationships are not captured with the use of serum folate.

It is also possible that the findings of dietary folate and iAs methylation were spurious; we only saw a statistically significant finding when we dichotomized folate intake at the median. Either no association exists, or we were unable to detect the association in this sample. Perhaps we observed no associations because 90% of participants had intakes above the Recommended Dietary Allowance (RDA) for folate, and 99% exceeded the RDA for vitamin B-12. Serum levels ≥ 3 ng/ml for folate and >200 pg/ml for vitamin B-12 are considered adequate; 99.5% of our participants had values above these for folate and 89.9% for vitamin B-12 (Institute of Medicine Standing Committee on the Scientific Evaluation of Dietary Reference Intakes and its Panel on Folate, 1998). The fact that our participants had very high intake as well as serum levels of folate increases our confidence that B-vitamin nutrition is likely adequate in this population. The varying results in our study sample based on the parameterization of the folate intake variable highlight the importance of taking into account the way variables are parameterized in statistical models, the hypothesized relationship between the exposure and outcome variables (linear or otherwise), available statistical power, and statistical methods when considering and comparing literature.

It is also worth mentioning that longer-term biomarkers of B-vitamin status exist and may be appropriate for studies investigating the relationship between B-vitamins and arsenic exposure. For example, red blood cell (RBC) folate reflects tissue folate stores (Dietrich et al., 2005; Green, 2011). Among women of child bearing age who participated in the 1988–1994 National Health and Nutrition Examination Survey (NHANES), serum folate levels were considered adequate but their RBC folate was low, at levels associated with risk of neural tube defects (Dietrich et al., 2005). In a study among NHANES 2003–2004 participants that included assessment of serum, RBC, as well as intake of folate in relation to the metabolites of urinary iAs (Kurzius-Spencer et al., 2017), dietary folate intake and serum folate levels were not associated with urinary iAs metabolites (Kurzius-Spencer et al., 2017). However, RBC folate levels were inversely associated with urinary %iAs, which reflects unmethylated iAs, and were positively associated with urinary %DMA, albeit not in a statistically significant manner (Kurzius-Spencer et al., 2017). The study also showed an inverse association with vitamin B-12 intake but a null association with serum B-12 levels (Kurzius-Spencer et al., 2017). Serum vitamin B-12 is a marker of longer-term vitamin B-12 status, although a single measure does not always capture functional deficiencies (Green, 2011; Hannibal et al., 2016). To our knowledge, no other study has assessed long term folate status, as indicated by RBC folate measures, in relation to iAs methylation (Kurzius-Spencer et al., 2017). Together, these studies indicate the need to further understand whether long term versus short term folate and vitamin B-12 status measures are more pertinent to understanding iAs methylation.

Other factors may also play a role in iAs metabolism. We previously found that meat intake was associated inversely with urinary %iAs, and positively with urinary %DMA among Uruguayan children (Kordas et al., 2016); animal proteins, primarily red meats and poultry are good sources of selenium. Selenium and arsenic act as metabolic antipodes; selenium has been shown to interact with arsenic and offer protection against arsenic toxicity (Christian et al., 2006; Hsueh et al., 2003; Schrauzer, 1992). Only two studies have assessed the relationship between selenium intake/status and iAs methylation (Li et al., 2008; López-Carrillo et al., 2016). Among Mexican women, selenium intake was inversely associated with urinary %iAs (López-Carrillo et al., 2016), whereas no association was observed between pregnant Bangladeshi women’s plasma selenium levels and iAs methylation (Li et al., 2008). A majority of the studies on B-vitamin intake/status and iAs methylation have not included measures of selenium intake/status (Argos et al., 2010; Bozack et al., 2019; Gamble et al., 2006; Gamble et al., 2005; Gamble et al., 2007; Hall et al., 2007; Hall et al., 2009a; Hall et al., 2009b; Heck et al., 2007; Howe et al., 2017; Kordas et al., 2016; Kurzius-Spencer et al., 2017; Peters et al., 2015; Spratlen et al., 2017; Steinmaus et al., 2005), perhaps because its role in the one carbon cycle is not clear. Our study also did not include measures of selenium intake/status. While exclusion of selenium from this study is an important limitation, it does not preclude comparisons of our findings with those of other studies on B-vitamins and iAs methylation.

We found no associations of dietary intake of vitamin B-12, or serum levels of both folate and vitamin B-12 with iAs methylation among Uruguayan schoolchildren exposed to low levels of arsenic, but found an inverse association between folate intake, categorized at the median, and iAs methylation. Current studies relating B-vitamins and iAs methylation have included B-vitamins assessed in a variety of ways. Additional studies on the role of B-vitamins in iAs methylation are needed to develop a deeper understanding of the implications of assessing folate and vitamin B-12 intake compared to the use of biomarkers. Where possible, both methods should be employed because they reflect different exposure windows and inherent measurement error, and if used individually, will likely continue to contribute to lack of consensus. Additionally, we need a better understanding of whether important differences in the relationship between B-vitamins and iAs methylation exist between areas where dietary sources make proportionally small vs. large contributions to arsenic exposure. Finally, we need to also understand if our assumption that a single urinary arsenic measure is sufficient to assess associations of nutrients with iAs methylation is reasonable (Kurzius-Spencer et al., 2017).