Nutrition, One-Carbon Metabolism and Arsenic Methylation in Bangladeshi Adolescents

Abstract

Background:

Over 57 million people in Bangladesh are chronically exposed to arsenic-contaminated drinking water. Ingested inorganic arsenic (InAs) undergoes hepatic methylation generating monomethyl- (MMAs) and dimethyl- (DMAs) arsenic species in a process that facilitates urinary As (uAs) elimination. One-carbon metabolism (OCM), a biochemical pathway that is influenced by folate and vitamin B12, facilitates the methylation of As. OCM also supports nucleotide and amino acid synthesis, particularly during periods of rapid growth such as adolescence. While folate supplementation increases As methylation and lowers blood As (bAs) in adults, little data is available for adolescents.

Objectives:

To examine the associations between OCM-related micronutrients and As methylation in Bangladeshi adolescents chronically exposed to As-contaminated drinking water.

Methods:

We conducted a cross-sectional study of 679 Bangladeshi adolescents, including 320 boys and 359 girls aged 14–16 years. Nutritional status was assessed by red blood cell (RBC) folate, plasma folate, plasma B12 and homocysteine (Hcys). Arsenic-related outcomes included blood arsenic (bAs), urinary arsenic (uAs), and urinary arsenic metabolites expressed as a percentage of total urinary As: %InAs, %MMAs, %DMAs.

Results:

Boys had significantly lower B12, higher Hcys, higher bAs, higher uAs, higher %MMAs, and a trend toward lower RBC folate compared to girls. Therefore, regression analyses controlling for water As and BMI were sex stratified. Among girls, RBC folate was inversely associated with bAs, plasma B12 was inversely associated with uAs, and plasma Hcys was inversely associated with %MMA. Among boys, plasma folate was inversely associated with %InAs and positively associated with %DMA, RBC folate was inversely associated with %InAs and positively associated with %MMA, while Hcys was positively associated with %InAs.

Conclusions:

These findings suggest that associations between OCM nutritional status, bAs, and distribution of As metabolites in adolescents are similar to previously reported observations in adults and in children. The As methylation findings are statistically significant among boys but not among girls; this may be related to estrogen which more strongly influences OCM in females. The inverse association between Hcys and %MMA in girls is somewhat unexpected given that Hcys is known to be an indicator of impaired OCM and low folate/B12 in adults. Overall, these results indicate that the associations between OCM-related micronutrients and arsenic methylation in adolescents are generally similar to prior findings in adults, though these associations may differ by sex. Additionally, these findings suggest that more investigation into the role of Hcys in adolescent physiology is needed, perhaps particularly for girls. Additional studies are needed to evaluate the impact of OCM and As methylation on As-related adverse health outcomes (such as cancer and cardiovascular disease) in people exposed to As during adolescence.

Introduction

Chronic exposure to arsenic (As) in drinking water is a global public health problem affecting more than 57 million people in Bangladesh, and over 200 million people worldwide.^1,2 Chronic As exposure has been associated with adverse health outcomes including cardiovascular disease,³ cancers of the skin, bladder, liver, and kidney,⁴ impaired intellectual function,⁵ and increased overall mortality.^1,6 The WHO guideline for drinking water As concentrations is <10μg/L.⁷ In our study site in Araihazar, Bangladesh well water As concentrations ranged from 0.1 to > 900 μg/L.^8,9

In Bangladesh, the primary species of As in drinking water is inorganic As (InAs) in the form of arsenite (As^III). InAs undergoes a series of biotransformation and methylation reactions that facilitate urinary As elimination.¹⁰ InAs^III is methylated by arsenic-3-methyltransferase (AS3MT) in a reaction that uses S-adenosylmethionine (SAM) as a methyl donor.^11,12 This yields monomethylarsonic acid (MMAs^V), which is reduced to monomethylarsonous acid (MMAs^III).¹³ MMAs^III undergoes a second methylation by AS3MT, yielding dimethylarsinic acid (DMAs^V), which is readily excreted in urine. Of these metabolites, MMAs^III is the most cyto- and genotoxic, while DMAs^V is the least toxic.^11,14–16 Synthesis of S-adenosylmethionine (SAM) is dependent on one-carbon metabolism (OCM), a biochemical pathway influenced by folate, vitamin B12, and other micronutrients. Methylation reactions yield the methylated products and S-adenosylhomocysteine (SAH), which is then hydrolyzed to form homocysteine (Hcys). Hcys can be remethylated to form methionine in a reaction that is dependent on both folate and B12, and is impaired by folate and/or B12 deficiency.^{17, 35} Consequently, high Hcys is a good overall indicator of impaired OCM, as well as folate and/or B12 deficiencies.

Studies in adults and animal models have shown better status of OCM-related nutrients to be associated with lower urinary %InAs, lower %MMAs and higher %DMAs (reviewed in Bozack et al 2018).^10,18–20 For reasons that are not entirely understood, folate nutritional status among children and adolescents is generally higher than adults.^21–23 During growth, a larger proportion of one-carbon units are needed for nucleotide and amino acid synthesis. Furthermore, OCM is influenced by estrogen.²⁴ Thus, the associations between OCM-related nutrients and As metabolism in adolescents may differ from those seen in adults.

While children appear to be more efficient at methylation As than adults,²⁵ relatively few studies have described the impact of nutritional status on As metabolism in adolescents. ^26,27 One observational study of 45 adolescents from Guanajuato, Mexico found micronutrient supplementation, including folic acid and vitamin B12, to be linked to improved nutritional status and higher levels of arsenic excretion in adolescents. ²⁸ Here, we tested the hypotheses that, as seen in adults, lower OCM micronutrient status in adolescents may be associated with reduced As methylation capacity and higher total blood As (bAs) concentrations, and that these associations may differ by sex.

Materials and Methods

Study overview:

The Health Effects of Arsenic Longitudinal Study (HEALS) is an ongoing prospective cohort study located in a 35km² region in Araihazar, Bangladesh.²⁹ The HEALS study recruited >30,000 participants between the ages of 18–75, beginning in 2000.²⁹ A subset of age-eligible children of HEALS participants were identified for inclusion in a study designed to evaluate the effects of As exposure on cognitive function in adolescents.³⁰ Those adolescents are the participants of the current study.

Field work:

Field staff approached 927 households in 51 villages, each thought to house a child between the ages of 14–16 from December 2012 to December 2016. Among the 927 households approached, 21 home visits found no age-appropriate child residing in the home, 12 home visits revealed that the child had passed away, and 29 had moved away from the study site. An additional 16 households were excluded for other reasons (14 had a child with a severe deficit (e.g., autism), 2 potential participants were twins). Of the 849 remaining households, 111 refused participation, resulting in a final sample size of 738 participants who consented, joined the study, donated biological samples, and completed structured interviews.

Procedures:

Adolescents were recruited via home visits. Within one week of the home visit, mother-child pairs visited the field clinic where the adolescents’ urine and venous blood samples were collected. Participants were given a small, non-monetary, age-appropriate gift as a token of appreciation for their participation.

Ethics:

Parental informed consent and adolescent assent were obtained by staff physicians in Bangladesh. The Columbia University Medical Center IRB and the Bangladesh Medical Research Council IRB approved this study protocol.

Sociodemographic characteristics:

During the home visit, household sociodemographic characteristics were assessed via a structured interview with the parent. Recorded data included: parental age, education and occupation, quality of home construction (concrete walls versus other as a proxy for socioeconomic status), and other relevant demographics. Anthropometric measures, including height, weight, and head circumference were also collected.³⁰

Blood-based analyses:

Blood samples were collected at our field clinic in Araihazar, where they were immediately processed, frozen, stored at −80°C, and shipped on dry ice to Columbia University for measurement of RBC folate, and plasma folate, B12, and Hcys. Folate and B12 were analyzed using radioproteinbinding assay (SimulTRAC-S, MP Biomedicals, Orangeburg, NY). Hcys was measured using high-performance liquid chromatography with fluorescence.³¹ Blood samples were analyzed for bAs concentrations using a Perkin-Elmer NexION 350S equipped Elemental Scientific autosampler 4DX. ICP-MS-DRC methods for metals in whole blood were developed according to published procedures;^32,33 as reported previously³⁰ the intraprecision coefficient of variation for bAs was 3.7%, and the interprecision coefficient for bAs was 7.3%. Additionally, the intraprecision coefficient of variation for uAs was 4.1%, and the interprecision coefficient for uAs was 4.2%.

Urinary As Measurements:

Urine samples were collected in our field clinic, frozen at 80°C, and shipped to Columbia University (CU) for a nalysis. UAs metabolite concentrations were measured in the CU Trace Metals Core Laboratory. Graphite furnace atomic absorption spectrophotometry (GFAA) via a Perkin-Elmer Analyst 800 system was used to assess total uAs concentrations.³⁴ UAs metabolites were measured via HPLC separation of arsenobetaine, InAs^III, InAs^V, MMAs, and DMAs, and via ICP-MS-DRC.³⁵ In storage and analysis InAs^III can oxidize to InAs^V, so these species are reported jointly as total InAs.¹³

Study Sample:

From the 738 participants initially recruited into the study, the data cleaning process systematically removed participants based on the following criteria: disclosed being part of a multiple birth after data collection (N=4); had biologically implausible specific gravity measurement of SG ≤1.001 (N=12); had biologically implausible %InAs measurement of %InAs > 55% (N=6); blood sample was hemolyzed (N=5); As metabolites data were missing (N=15), bAs data was missing (N=13); Hcys or SG data were missing (N=4); Current wAs data were missing (N=20). This yielded an overall sample size of 659 participants that was used for the majority of data analyses. For RBC folate analyses, we further excluded participants with missing RBC folate data (N=80), resulting in a subsample of 579 participants who had complete RBC folate data.

Statistical analysis:

Predictors of interest for this study are OCM-related micronutrients, including plasma folate, RBC folate, plasma B12 and Hcys. Outcomes of interest are bAs, uAs, and arsenic methylation as measured by the relative urine concentrations of %InAs, %MMAs, and %DMAs.

Summary statistics (mean, standard deviation, percentage) were used to describe the sample characteristics separately by sex. Wilcoxon rank sum and Chi-square tests detected sex differences for quantitative and categorical variables. Measures of uAs were adjusted for SG by multiplying by the ratio: [(mean SG per sex)-1]/(SG-1). Spearman correlation coefficients examined bivariate associations between nutritional predictors and As-related outcomes. Linear regression models further assessed the associations controlling for BMI and recent water As exposure, where the variables with right skewed distributions were transformed to meet model assumptions, improve model fitting and reduce impact of extreme values. The variables of bAs, uAs adjusting for SG, %InAs, BMI and water As were log-transformed and %MMA was square-root transformed. The models may have predictors for categorized nutrient variables or multiple predictors of nutrients.

To further examine the pattern of sex-specific associations between folate and As-related outcomes, we used previously published deficiency cut points for plasma^36,37 and RBC folate³⁸ (age-specific cut points are not widely used in the literature).^39–42 For plasma folate, sex-specific medians of the non-deficient group were used to categorize three levels (low, mid, high). The plasma folate categories for males were: low < 9.0, mid 9.0–15.38, and high > 15.38 nmol/L. For females they were: low < 9.0, mid 9.0– 15.08, and high > 15.08 nmol/L.^36,37 For RBC folate, a deficiency cut point and the overall median were used to categorize three levels (low, mid, high). The RBC folate categories for both sexes were: low < 317, mid 317– 386.9 and high > 386.9 nmol/L.³⁸

The standardized estimated coefficients (β) are reported, to facilitate comparison across predictors; this β is calculated by dividing a parameter estimate by the ratio of the sample standard deviation of the dependent variable to the sample standard deviation of the regressor.

All analyses were conducted using SAS 9.4 or R, and a significance level of 0.05.

Results

The sample characteristics by sex, including demographics, anthropometrics, measures of OCM nutritional status, and uAs metabolite percentages are presented in Table 1. Many sample characteristics relevant to the nutritional predictors and As-linked outcomes differed by sex: boys had lower BMI, RBC folate and plasma B12 levels, and higher Hcys, bAs, uAs, SG, and %MMAs levels, though plasma folate levels were similar between boys and girls. Given these differences, subsequent analyses were sex-stratified. The prevalence of folate deficiency was 17% among girls and 21% among boys, similar to our previous study of 6 years children in this region.²² The prevalence of B12 deficiency was 9.8% for girls and 18.8% for boys, a bit higher than the sex-specific prevalence observed in 6 year old children.²²

Table 1.

Study Sample Characteristics

	Girls		Boys
Characteristic	N	Mean (SD) or Percentage	N	Mean (SD) or Percentage	P
Age (yrs.)	349	14.6 (0.6)	310	14.6 (0.7)	0.89
Education (yrs.)	349	7.2 (1.9)	310	6.5 (2.3)	<0.01
BMI	349	18.8 (2.9)	310	17.8 (2.6)	<0.01
Water As (μg/L)	349	37.6 (73.2)	310	38.2 (57.6)	0.52
Plasma Folate (nmol/L)	349	14.7 (6.7)	310	15.3 (7.9)	0.80
% Folate Deficient		17.3%		20.7%
RBC Folate (nmol/L)	309	424.9 (129.3)	270	407.4 (119.4)	0.10
% Low RBC Folate (nmol/L)		18.0%		23.0%
Plasma B12 (pmol/L)	349	316.9 (163.1)	310	258.6 (137.0)	<0.01
% B12 Deficient		9.8%		18.8%
Plasma Homocysteine (μmol/L)	349	9.7 (4.2)	310	12.1 (7.1)	<0.01
% High Homocysteine		6.4%		18.2%
Blood Arsenic (μg/L)	349	4.6 (4.7)	310	5.1 (4.5)	<0.01
Urinary Arsenic (μg/L)	349	79.8 (129.6)	310	84.5 (116.4)	0.05
Urinary Creatinine (μg/L)	349	57.2 (50.1)	310	66.1 (51.6)	<0.01
Specific Gravity	349	1.008 (0.005)	310	1.010 (0.006)	<0.01
Urinary Arsenic Metabolites
%InAs	349	14.1 (5.6)	310	13.7 (5.4)	0.20
%MMAs	349	10.7 (3.4)	310	11.9 (3.4)	<0.01
%DMAs	349	75.2 (7.0)	310	74.4 (6.9)	0.17

P-values from Wilcoxon / Chi-square tests for quantitative/categorical variables, respectively. Low RBC Folate <317nmol/L; Plasma Folate Deficient <9nmol/L; Plasma B12 Deficient <150pmol/L; High Hcys >15μmol/L.⁵⁰

Plasma folate, RBC folate and arsenic by sex:

Single-predictor linear models were used to assess the associations between OCM micronutrients and As-related outcomes, controlling for BMI and water As (Table 2). In boys, plasma folate was inversely associated with %InAs (β=−0.15, p<0.05) and positively associated with %DMA (β=0.14, p<0.05). RBC folate was inversely associated with %InAs (β=−0.14, p<0.01), and positively associated with %MMAs (β=0.16, p<0.01). Plasma B12 was not significantly associated with blood or urine As nor with As methylation profile. Plasma Hcys was positively associated with %InAs (β=0.13, P<0.05), and was inversely associated with %DMA (β=−0.11, p<0.05).

Table 2.

Standardized parameter estimates from linear models for covariate-adjusted association between OCM Micronutrient and As outcomes

	Girls (N=349, RBC Folate N=309)				Boys (N=310, RBC Folate N=270)
	Plasma Folate	RBC Folate	Plasma B12	Plasma Hcys	Plasma Folate	RBC Folate	Plasma B12	Plasma Hcys
Outcome	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)	β (SE)
bAs‡	0.02 (0.04)	−0.09 (0.04)*	−0.03 (0.04)	0.04 (0.04)	0.03 (0.04)	0.05 (0.05)	0.05 (0.04)	−0.05 (0.04)
uAs§	0.01 (0.04)	−0.05 (0.04)	−0.09 (0.04)*	0.07 (0.04)	0.00 (0.04)	0.06 (0.05)	−0.03 (0.04)	−0.04 (0.04)
Arsenic Metabolites
%InAs‡	0.03 (0.06)	−0.03 (0.06)	−0.01 (0.05)	−0.01 (0.06)	−0.15 (0.06)*	−0.14 (0.05)**	−0.04 (0.05)	0.13 (0.06)*
%MMAs#	0.05 (0.06)	0.09 (0.06)	0.09 (0.05)††	−0.13 (0.05)**	−0.03 (0.06)	0.16 (0.06)**	0.08 (0.06)	0.05 (0.06)
%DMAs	−0.08 (0.06)	−0.04 (0.07)	−0.04 (0.05)	0.04 (0.06)	0.14 (0.06)*	0.05 (0.05)	0.01 (0.05)	−0.11 (0.06)†

^‡Log-transformed.

^§uAs adjusted for SG and log transformed.

^#Square root transformed.

^†p=0.07

^††p=0.06

^*p<0.05

^**p< 0.01.

Covariates: BMI, current wAs.

In girls, plasma folate was not associated with bAs, uAs, or As methylation patterns. RBC folate was inversely associated with bAs (β=−0.09, p<0.05) but had no significant associations to uAs or As methylation profile. Plasma B12 was associated with decreased uAs (β=−0.09, p<0.05), and appeared to be associated with increased %MMAs (β=0.09, p=0.07) at marginal significance levels. Finally, plasma Hcys was inversely associated with %MMAs (β=−0.13, p<0.01).

Plasma folate categories and arsenic by sex:

In both boys and girls, participants with low plasma folate had the highest blood and urine As levels, although differences between plasma folate categories were not statistically significant (Table 3). Among boys, the higher-level plasma folate category tended to associate with lower %InAs (p=0.05) and a higher %DMA (p=0.11). These patterns in boys persisted when plasma folate categories were compared using a regression model adjusting for BMI and current wAs exposure (Table 4). In boys, the low and mid plasma folates groups have significantly higher %InAs than the high plasma folate group (low vs. high: β=0.16, 95% CI=0.03, 0.28; mid vs. high: β=0.13, 95% CI=0.016, 0.25). Additionally, the low and mid plasma folate groups had significantly lower %DMA than the high plasma folate group (low vs. high: β=−0.15, 95% CI=−0.28, −0.03, mid vs. high: β=−0.12, 95% CI=−0.24, −0.003).

Table 3.

Characteristics by Plasma Folate Category

	Girls				Boys
	Low (N=62)	Mid (N=143)	High (N=144)		Low (N=64)	Mid (N=123)	High (N=123)
Characteristic	Mean (SD)	Mean (SD)	Mean (SD)	P	Mean (SD)	Mean (SD)	Mean (SD)	P
Plasma Folate	7.0 (1.4)	11.8 (1.6)	20.8 (5.4)	<0.01	7.1 (1.4)	11.9 (1.9)	22.8 (7.0)	<0.01
BMI	19.0 (2.7)	18.9 (3.2)	18.6 (2.8)	0.65	18.4 (2.6)	17.5 (2.2)	17.6 (2.9)	0.03
wAs	43.1 (102.7)	36.5 (67.3)	36.4 (63.4)	0.78	43.2 (65.7)	34.6 (54.2)	39.5 (56.8)	0.38
bAs	5.0 (6.9)	4.5 (4.1)	4.6 (4.1)	0.89	5.5 (4.3)	5.1 (5.2)	5.1 (3.9)	0.40
uAs	92.7 (171.9)	80.5 (95.0)	80.1 (90.8)	0.84	95.0 (77.3)	82.6 (106.2)	88.6 (85.3)	0.11
Arsenic Metabolites
%InAs	13.9 (4.2)	13.8 (5.0)	14.6 (6.7)	0.79	14.7 (5.7)	14.2 (6.2)	12.7 (4.0)	0.05
%MMAs	10.9 (3.5)	10.3 (3.1)	11.0 (3.5)	0.30	12.3 (3.5)	11.9 (3.7)	11.7 (3.1)	0.43
%DMAs	75.2 (6.1)	76.0 (6.3)	74.8 (8.0)	0.36	73.0 (7.8)	74.0 (7.3)	75.6 (5.6)	0.11

Female Cut Points: Low<9.0, Mid=9.0–14.72, High >15.08; Male Cut Points: Low<9.0, Mid=9.0–15.375, High>15.375. uAs is adjusted for SG. P-values from Kruskal-Wallis test for differences between categories of plasma folate.

Table 4.

Standardized parameter estimates from linear models for differences in As outcomes between plasma folate categories

	Girls (N=349)			Boys (N=310)
	Low vs. High	Mid vs. High	P-value	Low vs. High	Mid vs. High	P-value
Outcome	β (SE)	β (SE)		β (SE)	β (SE)
bAs‡	−0.03 (0.05)	−0.02 (0.04)	0.74	0.02 (0.05)	−0.02 (0.05)	0.79
uAs§	0.01 (0.04)	−0.02 (0.05)	0.84	0.05 (0.05)	−0.04 (0.05)	0.30
Arsenic Metabolites
%InAs‡	−0.02 (0.05)	−0.05 (0.06)	0.69	0.16 (0.06)*	0.13 (0.06)*	0.02
%MMAs#	−0.01 (0.06)	−0.10 (0.06)	0.21	0.06 (0.06)	0.02 (0.06)	0.60
%DMAs	0.04 (0.05)	0.10 (0.06)	0.20	−0.15 (0.06)*	−0.12 (0.06)*	0.03

Female Cut Points: Low<9.0, Mid=9.0–14.72, High >15.08; Male Cut Points: Low<9.0, Mid=9.0–15.38, High>15.38.

^‡Log-transformed.

^#Square root transformed.

^§uAs adjusted for SG and log transformed.

^*p≤0.05

^**p≤ 0.01 from test for difference between two categories.

Covariates: BMI, current wAs. P-value was from F-test for covariate-adjusted differences between three folate categories.

Among girls, there were no significant differences in As methylation profile by plasma folate groups (Table 3), nor significant associations between plasma folate group and As methylation profile controlling for BMI and water As (Table 4).

RBC folate categories and arsenic by sex:

In boys, there was a trend toward an inverse association between ordinal RBC folate group and %InAs (p=0.06), while %MMAs increased as RBC folate group rank increased (p=0.03) (Table 5). Covariate-adjusted associations between RBC folate group and both %InAs and %MMAs were also seen in linear regression models (Table 6), which showed that the low RBC folate group had significantly higher %InAs (β=0.14, 95%CI=0.007, 0.27) and lower %MMAs (β=−0.16, 95%CI=−0.30, −0.03) than the high RBC folate group. In girls, no clear patterns were observed between As methylation biomarkers and RBC folate groups (Table 5) nor after adjustment for BMI and water As (Table 6).

Table 5.

Characteristics by RBC Folate Category

	Girls				Boys
	Low (N=57)	Mid (N=88)	High (N=164)		Low (N=59)	Mid (N=86)	High (N=125)
Characteristic	Mean (SD)	Mean (SD)	Mean (SD)	P	Mean (SD)	Mean (SD)	Mean (SD)	P
RBC Folate	275.7 (39.0)	355.1 (21.5)	507.6 (113.5)	<0.01	269.1 (35.5)	353.3 (19.7)	495.0 (98.0)	<0.01
BMI	18.7 (2.9)	18.9 (3.1)	18.8 (3.0)	0.80	17.9 (2.6)	18.0 (2.9)	17.5 (2.4)	0.78
wAs	42.3 (88.9)	50.0 (87.5)	28.3 (50.1)	0.80	44.6 (68.1)	24.3 (42.1)	42.9 (56.6)	0.10
bAs	5.8 (7.3)	5.2 (5.0)	3.9 (3.2)	0.14	5.5 (4.7)	4.5 (4.4)	5.7 (4.9)	0.09
uAs	101.5 (181.5)	91.5 (114.0)	72.1 (80.0)	0.14	98.9 (127.2)	69.7 (59.6)	99.4 (97.3)	0.05
Arsenic Metabolites
%InAs	14.8 (6.1)	13.6 (4.3)	14.1 (6.4)	0.33	14.8 (5.8)	14.2 (7.1)	12.8 (3.6)	0.06
%MMAs	10.3 (3.3)	10.5 (3.3)	10.6 (3.4)	0.81	11.1 (3.9)	11.8 (3.2)	12.3 (3.4)	0.02
%DMAs	74.9 (7.7)	75.9 (6.4)	75.3 (7.5)	0.91	74.1 (7.9)	74.0 (8.0)	74.9 (5.7)	0.91

Cut Points: Low<317, Mid=317–386.90, High>386.90. uAs is adjusted for SG. P-values from Kruskal-Wallis test for differences between categories of RBC folate.

Table 6.

Standardized parameter estimates from linear models for differences in As outcomes between RBC folate categories

	Girls			Boys
	Low vs. High	Mid vs. High	P-value	Low vs. High	Mid vs. High	P-value
Outcome	β (SE)	β (SE)		β (SE)	β (SE)
bAs‡	0.06 (0.05)	0.03 (0.05)	0.50	−0.07 (0.05)	−0.03 (0.05)	0.42
uAs§	0.02 (0.05)	0.00 (0.05)	0.88	−0.07 (0.05)	−0.07 (0.05)	0.26
Arsenic Metabolites
%InAs‡	0.04 (0.06)	−0.03 (0.06)	0.57	0.14 (0.07)*	0.10 (0.07)	0.09
%MMAs#	−0.07 (0.06)	−0.05 (0.06)	0.42	−0.16 (0.07)*	−0.04 (0.06)	0.04
%DMAs	0.00 (0.06)	0.06 (0.06)	0.59	−0.04 (0.07)	−0.08 (0.07)	0.45

Cut Points: Low<317, Mid=317–386.90, High>386.90.

^‡Log-transformed.

^#Square root transformed.

^§uAs adjusted for SG and log transformed.

^*p≤0.05 from test for difference between two categories.

Covariates: BMI, current wAs. P-value was from F-test for covariate-adjusted differences between three folate categories.

Joint effect of plasma B12 and plasma or RBC folate:

To examine the joint effects of folate and B12, we used linear models with plasma B12 as a continuous predictor and plasma folate or RBC folate as continuous or categorical predictors, with bAs or uAs as the outcome of interest.

In boys, when plasma B12 was included in the model with categorical plasma folate, B12 was not associated with any As outcomes. The associations between plasma folate category and the As-linked outcomes remained similar to the previously described findings for plasma folate, with continuous plasma folate showing an inverse association to %InAs (β=−0.15, 95%CI= −0.28, −0.024) and a positive association to %DMA (β=0.15, 95%CI= 0.02, 0.27), and with the categorical plasma folate showing pattern persisting as reported in Table 4 (data not shown). Similarly, plasma B12 was unrelated to the As outcome variables in the models with continuous or categorical RBC folate in boys while the previously described findings for RBC folate group in Table 6 persisted (data not shown), which was also reflected in the inverse association between continuous RBC folate and %InAs (β=−0.14, 95%CI= −0.25, −0.03), and the positive association between continuous RBC folate and %MMA (β=0.15, 95%CI= 0.02, 0.29).

For girls, when plasma B12 was included in the model with either continuous or categorical plasma folate, neither measure of plasma folate was associated with any As outcomes, which was consistent with previously described findings for plasma folate groups in Table 4 (data not shown); however, B12 was inversely associated with uAs (β=−0.09, 95%CI= −0.19, −0.002) in the models including continuous or categorical folate variable. In models including B12 and either continuous RBC folate or categorical RBC folate with reduced sample size due to missing RBC folate measures, B12 was not significantly associated with any As outcomes, however, continuous RBC folate was inversely associated with bAs (β=−0.09, 95%CI= −0.17, −0.001), while the association with categorical RBC folate was not statistically significant, consistent with findings reported in Table 6 (data not shown).

Discussion

Previous studies in adults have demonstrated that folate nutritional status is positively associated with As methylation patterns and that FA supplementation increases As methylation which facilitates urinary As elimination⁴³ and lowers bAs concentrations.⁴⁴ Similar but weaker patterns of association were observed in an earlier study of 6 years children.²² The influence of B12 nutritional status on As methylation in adults has been inconsistent across studies.¹⁰ To our knowledge, no previous studies have evaluated the associations between folate and B12 nutritional status and As methylation patterns in adolescents with high prevalence of nutritional deficiency; though some NHANES studies have examined these associations in US adolescents, nutritional deficiency is not common in the US.

In this study, we found that among adolescent boys exposed to high wAs levels in Bangladesh, plasma and RBC folate levels were inversely associated with %InAs while RBC folate was positively associated with %DMAs. These findings are in agreement with prior studies in adults¹⁹ and consistent with the important role folate plays in the synthesis of the methyl donor SAM.¹¹ Plasma vitamin B12 was not associated with As methylation patterns in boys, though in girls it was inversely associated with uAs and showed a trend towards a positive association with %MMAs.

Among girls, plasma and RBC folate showed no significant associations with As methylation patterns. Plasma folate is known to fluctuate at throughout the menstrual cycle, and OCM is strongly influenced by estrogen.^24,45 For example, choline synthesis is stimulated by estrogen and choline can serve as an alternative methyl donor for As methylation through its conversion to betaine.⁴⁶ A limitation of this study is that we do not have information on choline status. Thus, fluctuations in estrogen and choline status may have introduced noise in the analyses and could explain why associations between plasma folate and As methylation patterns are not seen in girls.

Despite the fact that folate status was not found to be associated with As methylation patterns among these adolescent girls, we did find that RBC folate was associated with lower total bAs in a linear manner throughout the range of RBC folate values. These findings likely reflect the biological role that As methylation plays in facilitating As elimination in urine and lowering total blood As.

Some findings were contrary to our hypotheses and to findings in adults where higher folate status and lower Hcys are both associated with complete methylation of InAs to DMA.^43,47 For example, RBC folate was positively associated with %MMAs in boys, and, in girls, Hcys was inversely associated with %MMAs. It is interesting to note that in our earlier study of 6 years children, we similarly found Hcys to be inversely correlated to %MMAs, though that finding was among boys.^22,47 It is possible that upregulation of OCM, needed to meet the considerable demands for DNA and protein synthesis during periods of rapid growth, results in both increased Hcys synthesis and increased As methylation.

This study has several strengths, including its use of a well-characterized sample with extensive data on As exposure, As methylation and nutritional status, and its focus on adolescents who are a relatively under-studied and potentially vulnerable population. The main limitation is the relatively small sample size to assess effect modification by sex. The cross-sectional design limits our ability to make inferences regarding temporality. Finally, the study would have benefitted from data on choline nutritional status.

In conclusion, this study shows that the patterns of associations between folate nutritional status and As metabolites and/or total bAs in adolescents are generally similar to previously reported associations in adults in children, with a few exceptions. No significant influence of B12 nutritional status on As methylation or bAs concentrations were observed. Overall, these results are consistent with the literature that emphasizes the importance of nutritional factors in As metabolism and toxicity,^27,48,49 and also reflects somewhat different nutrition-dependent As methylation patterns for different for children and adults.²⁵ Some of these associations may differ by sex, and were possibly influenced by estrogen. Additional studies are needed to clarify the mechanism underlying these observations. The results of this study underscore the need for additional studies of the regulation of OCM and its influence on As methylation during periods of rapid growth, time periods that may be important with regard to long-term health outcomes.

Highlights.

• The effect of One-Carbon Metabolism micronutrients folate and B12 on arsenic methylation in adolescents is similar to previously seen effects in adults and children.

• Among adolescents, the effect of folate and B12 on arsenic methylation may differ by sex, but further investigation is needed.