The assessment of folate status has a long tradition in the National Health and Nutrition Examination Survey (NHANES). It has been carried out since 1974 with the use of different measurement procedures and serum and erythrocyte (RBC) folate as biomarkers( Reference Yetley, Pfeiffer and Phinney 1 ). Early measurements were conducted by microbiologic assay (1974–8), followed by two variants of a radio protein-binding assay (1978–1991 and 1991–2006), and more recently by a much improved microbiologic assay (2007–2010). Folate assays have had continued issues with comparability across laboratories and methods, necessitating the adjustment of data to allow the assessment of trends over time( Reference Yetley, Pfeiffer and Phinney 1 , Reference Pfeiffer, Hughes and Durazo-Arvizu 2 ). The NHANES 2011–2 survey assessed folate status in the US population for the first time by a combination of two analytical methods: serum folate forms were measured by HPLC–MS/MS, while whole-blood folate was measured by microbiologic assay. RBC folate was then calculated using the data from both assays. This approach was the result of a 2010 expert roundtable that advised Centers for Disease Control and Prevention (CDC) on folate biomarkers and methods for future NHANES surveys( Reference Yetley, Pfeiffer and Phinney 1 ). It allows clinicians, public health practitioners and researchers to obtain information on the full profile of folate forms, including unmetabolised folic acid (UMFA). This is important for monitoring purposes in the post-fortification era where an extremely low prevalence of folate deficiency is paired with higher folic acid intake from dietary supplements( Reference Yang, Cogswell and Hamner 3 ). HPLC–MS/MS is currently the best tool to measure individual folate forms with high sensitivity, specificity and accuracy( Reference Yetley, Pfeiffer and Phinney 1 ).While this analytical method produced results fully comparable to those of the microbiologic assay for serum folate, this was not the case for the more complex analysis of whole-blood folate( Reference Yetley, Pfeiffer and Phinney 1 ). Because these assay differences need to be better understood, HPLC–MS/MS was not used for whole blood folate.
Despite the interest in serum folate forms, currently only information on total folate can be interpreted clinically in the context of folate status. Clearly, the sum of biologically active folate vitamers (5-methyltetrahydrofolate (5-methylTHF), UMFA, tetrahydrofolate (THF), 5-formyltetrahydrofolate (5-formylTHF) and 5,10-methenyltetrahydrofolate (5,10-methenylTHF)) constitutes total folate. However, it remains to be determined whether MeFox (pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF) should be included in the calculation of total folate. This compound is an oxidation product of 5-methylTHF that lacks vitamin biologic activity( Reference Gregory 4 ) and is therefore not captured as part of the total folate measurement with the microbiologic assay. While it has been shown that MeFox can be formed in vitro after blood collection as a result of suboptimal sample handling( Reference Hannisdal, Ueland and Svardal 5 , Reference Hannisdal, Ueland and Eussen 6 ), its possible existence in vivo is unclear( Reference Hannisdal, Ueland and Svardal 5 ). If the formation of MeFox occurs solely in vitro (i.e. after blood collection), then a small part of the formerly active folate pool is lost and MeFox should be included in the total folate calculation to avoid underestimating the biologically active amount of folate. However, if some or all of the MeFox may already be present in vivo for an extended period of time, including it in the total folate calculation may slightly overestimate the biologically active amount of folate (i.e. folate status will appear better than it is because a small part of the total folate is not biologically available).
Our main objective was to describe serum concentrations of several folate forms in the US population ≥ 1 year participating in the NHANES 2011–2 by selected demographic, physiological, and lifestyle variables. Our other objective was to update information on total folate status and to provide the first nationally representative data for non-Hispanic Asians. We report information on serum total folate with and without the inclusion of MeFox to provide much needed insight on this new topic.
Methods
Participants and study design
The NHANES is conducted by the CDC and collects cross-sectional data on the health and nutritional status of the civilian non-institutionalised US population by use of a stratified, multistage, probability sample design. In addition to obtaining information in a home interview setting, participants undergo a physical examination and blood draw in a mobile examination centre. In 2011–2, the NHANES oversampled Asian and Hispanic persons in addition to oversampling some other population groups( 7 – Reference Mirel, Mohadjer and Dohrmann 9 ). The unweighted response rates for participants ≥ 1 years of age were 72·2 % for the interview component and 69 % for the examination component( 10 ). All respondents gave their informed consent, and the NHANES protocol was reviewed and approved by the CDC Research Ethics Review Board.
Biomarker measurement
Serum and whole-blood haemolyate samples from participants ≥ 1 year were analysed by the CDC laboratory for serum folate forms (folate cofactors and MeFox) by use of HPLC–MS/MS( Reference Fazili and Pfeiffer 11 – 13 ) and for RBC total folate by use of microbiologic assay( Reference Pfeiffer, Zhang and Lacher 14 – 16 ), respectively. We did not obtain valid final results for a few samples ( < 30), resulting in different sample sizes among compounds: serum 5-methylTHF (n 7454), UMFA (n 7462), THF (n 7461), 5-formylTHF (n 7466), 5,10-methenylTHF (n 7466), MeFox (n 7469), and total folate including MeFox (n 7442), and RBC total folate (n 7867). Sample sizes for the folate biomarkers by covariate categories are presented in Table 1. Because concentrations of the three minor folate forms THF, 5-formylTHF, and 5,10-methenylTHF were often below the limit of detection (LOD) and can be a result of folate interconversions at slightly acidic pH during sample preparation( Reference Fazili and Pfeiffer 17 ), we calculated the sum of these three forms as non-methyl folate. Serum total folate was calculated as the sum of the six folate forms including MeFox. CDC released results < LOD as imputed values (LOD divided by the √2); we used the imputed values in our calculation when the folate form result was < LOD( Reference Fazili and Pfeiffer 11 ). The serum total folate including MeFox result was missing if one of the folate forms was missing. We calculated serum total folate excluding MeFox by subtracting MeFox from total folate including MeFox. RBC total folate was calculated from the measured whole-blood folate concentration after subtracting the serum total folate including MeFox concentration (as determined by HPLC–MS/MS) and adjusting for RBC volume( 15 ). Assay performance is summarised in the online Supplementary Table S1 and has been described previously with regard to international reference materials for the HPLC–MS/MS ( Reference Yetley, Pfeiffer and Phinney 1 , Reference Fazili, Whitehead and Paladugula 18 ) and microbiologic assay( Reference Pfeiffer, Zhang and Lacher 14 ).
Unweighted sample sizes for serum folate forms and serum and erythrocyte (RBC) total folate by selected demographic, physiological and lifestyle variable categories for the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2*
5-methylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid; MeFox, pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF; eGFR, estimated glomerular filtration rate; BSA, body surface area.
* Serum folate forms and serum total folate (sum of all forms including MeFox) were measured by HPLC–MS/MS; RBC total folate was measured by microbiologic assay; non-methyl folate represents sum of three minor forms: tetrahydrofolate, 5-formyltetrahydrofolate, and 5,10-methenyltetrahydrofolate.
† Hispanic sub-group represents sum of Mexican American and other Hispanic ethnicity; other sub-group includes persons with multi-ethnic background.
‡ Used to assess renal function; available for persons ≥ 12 years; impaired renal function was defined as eGFR < 60 ml/(min × 1·73 m2).
§ BMI definitions: < 18·5 kg/m2 (underweight); 18·5–>25 kg/m2 (normal weight); 25– < 30 kg/m2 (overweight); and ≥ 30 kg/m2 (obese).
∥ Calculated as √((height in cm × weight in kg)/3600) or √((height in inches × weight in pounds)/3131).
¶ Biomarker of tobacco smoke exposure; concentrations >10 μg/l are considered to represent smokers.
** Calculated for participants ≥ 18 years as average daily number of ‘standard’ drinks ((quantity × frequency)/365·25); 1 drink approximately 15 g ethanol.
†† Folic acid-containing dietary supplements used during the last 24 h prior to visiting the Mobile Examination Center.
Study variables
We categorised the demographic variables as follows: age (1–5, 6–11, 12–19, 20–39, 40–59, and ≥ 60 years), sex (males and females), and race–ethnicity (Hispanic (Mexican American+other Hispanic), non-Hispanic Asian, non-Hispanic Black, and non-Hispanic White; other racial/ethnic groups were included in overall estimates). We also reported separate estimates for Mexican Americans to allow comparison to previous reports. We examined physiological and lifestyle variables previously shown to be associated with folate concentrations( Reference Pfeiffer, Sternberg and Schleicher 19 , Reference Haynes, Pfeiffer and Sternberg 20 ): fasting time ( < 3, 3– < 8 and ≥ 8 h), kidney function as determined by estimated glomerular filtration rate (0– < 60, 60– < 90, and ≥ 90 ml/(min × 1·73 m2))( Reference Stevens, Coresh and Greene 21 ), BMI ( < 18·5 kg/m2 (underweight), 18·5– < 25 kg/m2 (normal), 25– < 30 kg/m2 (overweight) and ≥ 30 kg/m2 (obese))( 22 ), body surface area (BSA, calculated as √(height in cm × weight in kg/3600); < 1·5, 1·5– < 1·8, 1·8– < 2·0, and ≥ 2·0 m2)( Reference Mosteller 23 ), smoking (serum cotinine ≤ 10 μg/l (nonsmoker) and >10 μg/l (smoker))( Reference Pirkle, Flegal and Bernert 24 ), alcohol intake (average daily number of ‘standard’ drinks (one drink approximately 15 g alcohol): no drinks, < 1 (not 0), 1– < 2, and ≥ 2 drinks/d; only available for participants ≥ 18 years)( Reference Pfeiffer, Sternberg and Schleicher 19 ), and use of folic acid-containing dietary supplements (self-reported use during the 24 h prior to visiting the mobile examination centre (self-reported use during the last 30 d is no longer available in NHANES 2011–2); yes and no). We presented the association between folate concentrations and body size by two variables: BMI, used traditionally in nutrition studies, and BSA, used in exposure studies to provide insight on ‘body burden’. A BSA value < 1·5 m2 generally represents children, while a value ≥ 2·0 m2 generally represents adult men. This additional variable may shed light on the distinction between metabolism v. ‘dilution effects’ due to body size.
Statistical analysis
We applied no exclusion criteria to our data analysis and used pairwise deletion for missing values in a particular analysis. We used the mobile examination centre weights to account for differential non-response or non-coverage and to adjust for oversampling of some groups. We calculated the mean percent contribution of each folate form to serum total folate including MeFox. We also calculated the mean absolute and percent contribution of each folate form to serum total folate including MeFox by weighted decile of serum total folate including MeFox. Bivariate associations between geometric mean folate biomarker concentrations (to normalise for right-skewed distributions) and each study variable were described. Geometric means were compared across the categories without (Wald F P value) and with (Satterthwaite F P value) controlling for additional covariates (age, sex and race–ethnicity). We used Spearman's coefficients to assess pairwise correlations among folate biomarkers as well as between folate biomarkers and selected physiological and lifestyle variables. We assessed the distributions (geometric means and selected percentiles (95 % CI)) for each folate biomarker among all participants, fasted ( ≥ 8 h) participants and non-fasted ( < 8 h) participants ≥ 1 year by demographic variables. Significance was defined as a two-sided P value of < 0·05. Statistical analyses were performed using SAS (version 9; SAS Institute, Inc.) and SUDAAN (version 9.2; RTI) software.
Results
A summary of the characteristics of the study population for each variable of interest is given in the online Supplementary Table S2. Among US population ≥ 1 year in the unweighted NHANES 2011–2 sample, 27 % were children (1–11 years), 14 % were adolescents (12–19 years) and 59 % were adults ( ≥ 20 years). Half of the participants were female and almost one-third of the participants were non-Hispanic white. Approximately half (43 %) of the participants were fasted for ≥ 8 h, 8 % had an impaired estimated glomerular filtration rate ( < 60 ml/min/1·73 m2), 24 % were obese, 30 % had a small BSA ( < 1·5 m2), 17 % were considered smokers (serum cotinine concentrations >10 μg/l), one-third of participants ≥ 18 years reported not consuming any alcoholic beverage, and about 20 % of participants reported using folic acid-containing dietary supplements during the last 24 h.
Folate biomarker concentrations and contribution of folate forms to serum total folate
The concentration ranges of serum total folate excluding MeFox, serum total folate including MeFox and RBC total folate were 3·26–375, 3·50–377 and 149–5490 nmol/l, respectively. Concentration ranges of serum folate forms were: 1·88–295 nmol/l for 5-methylTHF, < LOD (0·14)–282 nmol/l for UMFA, < LOD (0·37)–11·5 nmol/l for THF, < LOD (0·30)–31·6 nmol/l for 5-formylTHF, < LOD (0·34)–4·38 nmol/l for 5,10-methenylTHF and < LOD (0·34)–20·4 nmol/l for MeFox. Concentrations of 5-methylTHF (100 %), UMFA (99·9 %), MeFox (98·8 %) and THF (85·2 %) were detectable in all or most samples, while concentrations of 5-formylTHF (3·6 %) and 5,10-methenylTHF (4·4 %) were detectable in only a few samples.
On average, 5-methylTHF (86·7 %) was the biggest contributor to serum total folate including MeFox, while UMFA (4·0 %), non-methyl folate (4·7 %), and MeFox (4·5 %) contributed smaller amounts. When we calculated the contribution of these folate forms by decile of serum total folate including MeFox (Fig. 1 and online Supplementary Table S3), we noted some fluctuation in the proportion of 5-methylTHF (79·9–90·2 %), a decreasing proportion of non-methyl folate (9·1 % in the first, 2·8 % in the last decile) and MeFox (6·6 % in the first, 3·0 % in the last decile), and a generally U-shaped proportion of UMFA (4·5 % in the first, approximately 3 % in the fifth, and 10·4 % in the last decile) with increasing decile of serum total folate including MeFox.
Mean absolute (a) and relative (b) contribution of folate forms to serum total folate by weighted decile of serum total folate in the US population ≥ 1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2. Serum folate forms and serum total folate (sum of folate forms including MeFox (pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF, □)) were measured by HPLC–MS/MS. Non-methyl folate (
) represents the sum of three minor forms: tetrahydrofolate, 5-formyltetrahydrofolate and 5,10-methenyltetrahydrofolate. 5-MethylTHF, 5-methyltetrahydrofolate (■); UMFA, unmetabolised folic acid (
).
Folate biomarker concentrations by demographic characteristics
We noted approximately U-shaped age patterns for all serum folate forms except MeFox, for which the concentration was significantly higher in persons ≥ 60 years compared to all other age groups and the proportion of MeFox relative to serum total folate including MeFox was significantly higher in persons ≥ 60 years compared to the three youngest age groups (Table 2 and online Supplementary Fig. S1). Age was a significant factor for all serum folate forms as well as for serum total folate including and excluding MeFox, and RBC total folate whether or not we controlled for other demographic covariates (sex and race–ethnicity). While females had significantly higher serum total folate including and excluding MeFox, RBC total folate, 5-methylTHF and UMFA concentrations with and without controlling for age and race–ethnicity, there were no sex differences for non-methyl folate and MeFox concentrations. All folate biomarker concentrations except non-methyl folate varied significantly by race–ethnicity, with non-Hispanic whites having the highest concentrations and non-Hispanic Asians having similar concentrations compared to Hispanics. These observations did not change after we controlled for age and sex. The concentration difference between geometric means of serum total folate excluding v. including MeFox was approximately 2 nmol/l and it was reasonably consistent across demographic groups.
Concentrations of serum folate forms and serum and erythrocyte (RBC) total folate by demographic variable categories for the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2* (Geometric mean values and 95 % confidence intervals)
5-methylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid; MeFox, pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF; MA, Mexican American; NH, non-Hispanic.
* Serum folate forms and serum total folate (sum of folate forms excluding or including MeFox) were measured by HPLC–MS/MS; RBC total folate was measured by microbiologic assay; non-methyl folate represents sum of three minor forms: tetrahydrofolate, 5-formyltetrahydrofolate, and 5,10-methenyltetrahydrofolate; for sample sizes, see Table 1; P value is the unadjusted Wald F P-value, while P value adjusted is the Satterthwaite F P value adjusted for age, sex and race–ethnicity.
† Hispanic sub-group represents the sum of MA and other Hispanic ethnicity; P values for race–ethnicity show comparison of Hispanic, NH White, NH Black, NH Asian and other (not shown).
Correlations among folate biomarkers
Spearman's correlations were significant for most pairwise comparisons of folate biomarkers in persons ≥ 1 year (see online Supplementary Table S4). We observed significant and strong (r≥ 0·7) correlations for 5-methylTHF with serum total folate including MeFox (r 0·99) and for THF with non-methyl folate (r 1·0). We observed significant and moderate (0·4 ≤ r< 0·7) correlations for 5-methylTHF with RBC total folate (r 0·59), for UMFA with 5-methylTHF (r 0·44) and serum total folate including MeFox (r 0·50), and for serum total folate including MeFox with RBC total folate (r 0·59). We observed significant but weak (r< 0·4) correlations for 5-methylTHF with MeFox (r 0·25). After stratifying by age group, we noted a strengthening of the correlations in persons ≥ 60 years (trace covered a larger area than for persons 1–19 or 20–59 years), but the same patterns overall (Fig. 2).
Spearman's correlation between various folate biomarkers by age group in the US population ≥ 1 years, National Health and Nutrition Examination Survey (NHANES) 2011–2. Only statistically significant correlations are shown (see online Supplementary Table S4 for complete information). Serum folate forms and serum total folate (sum of folate forms including MeFox (pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF)) were measured by HPLC–MS/MS. Non-methyl folate represents sum of three minor forms: tetrahydrofolate, 5-formyltetrahydrofolate and 5,10-methenyltetrahydrofolate. Erythrocyte (RBC) total folate was measured by microbiologic assay. 5-MethylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid. ···, 1–19 years; - - -, 20–59 years; —, ≥ 60 years.
Associations between folate biomarkers and selected physiological and lifestyle characteristics
We observed generally significant but weak Spearman's correlations between folate biomarkers and the continuous physiological and lifestyle variables (Table 3). Fasting, kidney function, smoking and alcohol intake were negatively associated with most folate biomarkers. BMI and BSA showed positive associations with MeFox and negative associations with other folates. We noted generally significant differences in concentrations between the levels of the categorical variables, including categorised versions of the physiological and lifestyle variables, whether or not we controlled for demographic covariates (age, sex and race–ethnicity) (Table 4). All folate biomarkers showed significantly higher concentrations with recent folic acid-containing dietary supplement use whether or not we controlled for demographic covariates. The concentration difference between serum total folate excluding v. including MeFox was approximately 2 nmol/l and it was again reasonably consistent across categories of variables.
Spearman's correlations between various folate biomarkers and selected physiological and lifestyle variables for the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2*
RBC, erythrocyte; 5-methylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid; MeFox, pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF; eGFR, estimated glomerular filtration rate; BSA, body surface area.
* Serum folate forms and serum total folate (sum of folate forms excluding or including MeFox) were measured by HPLC–MS/MS; RBC total folate was measured by microbiologic assay; non-methyl folate represents sum of three minor forms: tetrahydrofolate, 5-formyl-tetrahydrofolate, and 5,10-methenyltetrahydrofolate; for sample sizes, see Table 1.
† Used to assess renal function; available for persons ≥ 12 years; impaired renal function was defined as eGFR < 60 ml/(min × 1·73 m2).
‡ BMI definitions: < 18·5 kg/m2 (underweight); 18·5–>25 kg/m2 (normal weight); 25– < 30 kg/m2 (overweight); and ≥ 30 kg/m2 (obese).
§ Calculated as √((height in cm × weight in kg)/3600) or √((height in inches × weight in pounds)/3131).
∥ Biomarker of tobacco smoke exposure; concentrations >10 μg/l are considered to represent smokers.
¶ Calculated for participants ≥ 18 years as average daily number of ‘standard’ drinks ((quantity × frequency)/365·25); 1 drink approximately 15 g ethanol.
Concentrations of serum folate forms and serum and erythrocyte (RBC) total folate by selected physiological and lifestyle variables for the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2* (Geometric mean values and 95 % confidence intervals)
5-methylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid; MeFox, pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF; eGFR, estimated glomerular filtration rate; BSA, body surface area.
* Serum folate forms and serum total folate (sum of folate forms excluding or including MeFox) were measured by HPLC–MS/MS; RBC total folate was measured by microbiologic assay; non-methyl folate represents sum of three minor forms: tetrahydrofolate, 5-formyl-tetrahydrofolate and 5,10-methenyltetrahydrofolate; for sample sizes, see Table 1; P value is the unadjusted Wald F P-value, while P value adjusted is the Satterthwaite F P value adjusted for age, sex and race–ethnicity.
† Used to assess renal function; available for persons ≥ 12 years; impaired renal function was defined as eGFR < 60 ml/(min × 1·73 m2).
‡ BMI definitions: < 18·5 kg/m2 (underweight); 18·5–>25 kg/m2 (normal weight); 25– < 30 kg/m2 (overweight); and ≥ 30 kg/m2 (obese).
§ Calculated as √((height in cm × weight in kg)/3600) or √((height in inches × weight in pounds)/3131).
∥ Biomarker of tobacco smoke exposure; concentrations >10 μg/l are considered to represent smokers.
¶ Calculated for participants ≥ 18 years as average daily number of ‘standard’ drinks ((quantity × frequency)/365·25); 1 drink approximately 15 g ethanol.
** Folic acid-containing dietary supplements used in the last 24 h prior to visiting the Mobile Examination Center.
Reference intervals and distributions of folate biomarker concentrations
Because fasting was a significant factor for most folate biomarkers, we calculated the central 95 % reference intervals (2·5th–97·5th percentile) for all, fasted, and non-fasted ‘generally healthy’ persons ≥ 1 year (Table 5). Reference intervals were fairly comparable among these three groups for 5-methylTHF, non-methyl folate, serum total folate excluding and including MeFox, and RBC total folate. However, we noted a lower upper end of the reference interval in fasted persons for UMFA and MeFox.
Central 95 % reference intervals for serum folate forms and for serum and erythrocyte (RBC) total folate in the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2*
5-methylTHF, 5-methyltetrahydrofolate; UMFA, unmetabolised folic acid; THF, tetrahydrofolate; 5-formylTHF, 5-formyltetrahydrofolate; 5,10-methenylTHF, 5,10-methenyltetrahydrofolate; MeFox, pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF.
* Serum folate forms and serum total folate (sum of folate forms excluding or including MeFox) were measured by HPLC–MS/MS; RBC total folate was measured by microbiologic assay; non-methyl folate represents sum of three minor forms: THF, 5-formylTHF, and 5,10-methenylTHF; for sample sizes, see Table 1.
Selected percentiles (5th–95th) presented by age, sex and race–ethnicity for all, fasted, and non-fasted persons ≥ 1 year generally showed the greatest variation by age group (see online Supplementary Tables S5–S11). We observed distinct differences in the distributions of serum total folate excluding and including MeFox, RBC total folate and 5-methylTHF by age group. We also observed differences in the upper end of the distribution of UMFA by age group, higher non-methyl folate concentrations at the lower end of the distribution for children 1–5 years compared to any other age group, and a right-shift in the distribution of MeFox with increasing age group. The differences we observed in the central 95 % reference intervals between all and fasted persons were also notable in the entire distribution of folate concentrations.
Folate status time trend
Serum total folate excluding MeFox concentrations was similar in 2011–2 (geometric mean 41·4 (95 % CI 40·1, 42·9) nmol/l) compared to that in the previous two survey cycles, when the microbiologic assay was used, which does not respond to MeFox: 2007–8 (geometric mean 39·5 (95 % CI 37·7, 41·3) nmol/l) and 2009–2010 (geometric mean 38·2 (95 % CI 37·2, 39·3) nmol/l). RBC total folate concentrations measured in all three survey cycles by microbiologic assay also appeared to be similar: 2007–8 (geometric mean 1120 (95 % CI 1070, 1160) nmol/l), 2009–2010 (geometric mean 1040 (95 % CI 1010, 1070) nmol/l) and 2011–2 (geometric mean 1050 (95 % CI 1010, 1090) nmol/l). As in the previous two survey cycles, < 1 % of the US population in NHANES 2011–2 had serum ( < 10 nmol/l) or RBC total folate ( < 340 nmol/l) concentrations at risk for deficiency( Reference de Benoist 25 ).
Discussion
The present study provides the first national reference information for serum folate forms measured by HPLC–MS/MS in a population exposed to folic acid fortification. It also offers a better understanding of variables associated with concentrations of serum folate forms. Based on the newest serum and RBC total folate concentrations from NHANES 2011–2, the folate status of the US population was comparable to that in the previous years and non-Hispanic Asians had similar folate concentrations compared to Hispanics.
Previous studies that assessed the profile of serum folate forms used convenience samples and were small in size (mostly < 100 subjects). Most studies investigated special population subgroups such as pregnant women, older adults or haemodialysis patients( Reference Ghandour, Bagley and Shemin 26 – Reference Obeid, Kirsch and Kasoha 30 ), while a few studies measured serum folate forms in apparently healthy US, German or Norwegian adults, though generally as part of method validations( Reference Hannisdal, Ueland and Svardal 5 , Reference Fazili and Pfeiffer 17 , Reference Summers, Mitchell and Stanislawska-Sachadyn 31 – Reference Kirsch, Knapp and Herrmann 34 ). Given that the population in the present study was exposed to folic acid fortification and known to have a historical prevalence of folic acid supplement use of approximately 35 %( Reference Bailey, Dodd and Gahche 35 ), the higher 5-methylTHF (38·5 nmol/l) and UMFA (0·991 nmol/l) median concentrations compared to the small convenience sample reports for German (15·8 and 0·10 nmol/l, respectively, ( Reference Kirsch, Knapp and Herrmann 34 )) or Norwegian (16·4 and 0·0 nmol/l, respectively( Reference Hannisdal, Ueland and Svardal 5 ) adults from countries with no folic acid fortification were not surprising. However, caution should be used when comparing data from different populations, in part due to potential method differences that have historically plagued folate analyses( Reference Shane 36 ).
The 5-methylTHF concentration was the biggest and a constant contributor to serum total folate regardless of the population (86·7 % in US (this study) compared to 87·2 % in German( Reference Kirsch, Knapp and Herrmann 34 ) and 85·8 % in Norwegian persons( Reference Hannisdal, Ueland and Svardal 5 )). The higher mean UMFA concentration (13·5 nmol/l) and relative contribution (10·2 %) in the highest decile of serum total folate including MeFox compared to the lower deciles (0·78–2·87 nmol/l, 2·75–4·45 %) in the present study are likely due to the larger intake of folic acid from dietary supplements and/or fortified foods and the incomplete conversion of folic acid to 5-methylTHF upon absorption( Reference Kelly, McPartlin and Goggins 37 – Reference Patanwala, King and Barrett 39 ). A previous report from NHANES 2007–8 showed that UMFA concentrations >1 nmol/l were largely explained by total folic acid intake from diet and supplements apart from fasting status( Reference Pfeiffer, Sternberg and Fazili 40 ). Not surprisingly, we found significantly higher serum folate forms as well as serum and RBC total folate concentrations in persons who reported consuming folic acid-containing dietary supplements during the last 24 h.
Given that MeFox is an oxidation product of 5-methylTHF, the correlation between these two folate forms (r 0·25) was lower than expected. This may indicate that factors beyond the amount of circulating 5-methylTHF may influence the generation of MeFox. Thus, the relevance of MeFox in relation to folate status is likely of interest in any population, regardless of whether they have high folate status as a result of fortification or supplementation or not. The high correlations between 5-methylTHF and serum total folate (r 0·99) and between THF and non-methyl folate (r 1·00) were expected, as these two folate forms were the major contributors to serum total folate and non-methyl folate, respectively. We found a correlation between UMFA and 5-methylTHF (r 0·54 for persons ≥ 60 years) similar to that reported for older German adults (r 0·42 at baseline and r 0·56 after supplementation with folic acid)( Reference Obeid, Kirsch and Kasoha 30 ). We found lower correlations between UMFA and THF (r 0·22) or between 5-methylTHF and THF (r 0·30) in US older persons compared to the report in German older adults (at baseline: r 0·39 and r 0·51, respectively; after supplementation: r 0·45 and r 0·56, respectively)( Reference Obeid, Kirsch and Kasoha 30 ).
Among demographic variables studied, we found interesting patterns with age. While most folate forms displayed the typical U-shaped age pattern previously documented with serum and RBC total folate( Reference Pfeiffer, Hughes and Lacher 41 ), concentrations of MeFox showed a linear pattern and were highest in persons ≥ 60 years. The distribution of MeFox concentrations showed a right-shift with increasing age group, resulting in higher detection rates of MeFox in persons ≥ 60 years (99·7 % compared to 95·5 % in children 1–5 years). Conversely, detection rates of THF were highest in children 1–5 years (96·1 % compared to 82·9–89·1 % for other age groups). These observations may indicate altered folate metabolism, possibly as a result of ageing, and will have to be confirmed in other studies. It is interesting though to note that older age was associated with less bioactive folate (THF) and more biologically inactive folate (MeFox), possibly pointing to an increased catabolism.
The associations of serum folate forms with physiological and lifestyle variables were generally consistent with reports for serum total folate from NHANES 2003–6( Reference Pfeiffer, Sternberg and Schleicher 19 , Reference Haynes, Pfeiffer and Sternberg 20 ). While most serum folate forms showed higher concentrations in non-fasted persons, we found no difference in RBC total folate concentrations by fasting time, consistent with the prevailing knowledge that RBC total folate concentrations are not affected by fasting( Reference Pfeiffer, Schleicher, Caldwell and Caballero 42 ). The negative association of serum folate forms and of serum and RBC total folate with decreasing kidney function was no longer significant for several folate biomarkers (5-methylTHF, non-methyl folate, total folate excluding and including MeFox) after we controlled for demographic covariates. Most serum folate forms showed lower concentrations in smokers and in persons with higher alcohol intake.
Some observations with physiological and lifestyle variables, particularly for MeFox, were unexpected and may indicate altered folate metabolism, possibly challenging the notion that MeFox occurs solely in vitro. BMI and BSA were positively associated with MeFox, but negatively with 5-methylTHF concentrations. This is counter-intuitive if one assumes that higher 5-methylTHF concentrations will lead to higher MeFox concentrations. It may in fact indicate a greater ‘loss’ of active folate in obesity. Whether this has any bearing on epidemiologic observations linking obesity with a higher rate of neural tube defects( Reference Rasmussen, Chu and Kim 43 ) is unclear, particularly if one considers that in the present study obese v. normal weight persons had approximately 5·5 nmol/l lower 5-methylTHF, but only approximately 0·2 nmol/l higher MeFox concentrations. We need a better understanding of these relationships to appropriately interpret them. The lower serum total folate (and higher RBC total folate) concentrations reported previously with higher BMI led to the hypothesis that cellular uptake and tissue distribution of folate may be altered by BMI( Reference Tinker, Hamner and Berry 44 ). The existing data by BSA category may instead imply a ‘dilution effect’ whereby most folate forms and serum total folate concentrations were lower in larger persons. This interpretation was also consistent with the finding that children < 11 years had higher 5-methylTHF and serum total folate concentrations compared to all other age groups.
We found a median MeFox concentration (1·49 nmol/l) similar to that reported in Norwegian adults (2·3 nmol/l; the authors called this compound 4α-hydroxy-5-methylTHF but stated that it was the pyrazino-s-triazine derivative( Reference Hannisdal, Ueland and Svardal 5 )), suggesting that concentrations of this oxidation product are low as long as the blood is processed under controlled conditions. Yet, we found apparent inconsistencies regarding MeFox and 5-methylTHF concentrations (Table 6), which raised the possibility that MeFox may be formed in vivo rather than solely in vitro as part of 5-methylTHF oxidation. In rats, administration of radio-labelled folic acid resulted in the excretion of several labelled products: a precursor of MeFox (4α-hydroxy-5-methylTHF), 5-methylTHF, 10-formyl-folic acid, and p-aminobenzoyl-l-glutamate, a folate breakdown product( Reference Barford, Staff and Blair 45 ). It is therefore conceivable that a small portion of MeFox measured in serum may already be in circulation at the time of the blood draw. We know that MeFox can be formed during the pre-analytical phase (blood collection and processing)( Reference Hannisdal, Ueland and Svardal 5 , Reference Hannisdal, Ueland and Eussen 6 ), leading to a small loss of ‘active’ folate after blood is collected. We have shown that the analytical phase of the CDC LC–MS/MS method does not generate additional MeFox( Reference Fazili and Pfeiffer 11 ). Thus, it appears that failure to include MeFox as part of serum total folate may slightly underestimate folate status if MeFox is mostly formed in vitro, while including it may slightly overestimate folate status if MeFox is mostly formed in vivo. Mechanistic studies that further explore the origins of this oxidation product are needed.
Inconsistencies between serum MeFox (pyrazino-s-triazine derivative of 4α-hydroxy-5-methylTHF) and 5-methyltetrahydrofolate (5-methylTHF) concentrations in the US population ≥1 year, National Health and Nutrition Examination Survey (NHANES) 2011–2
eGFR, estimated glomerular filtration rate; BSA, body surface area.
The present study is subject to some limitations. The data are based on only one NHANES survey period limiting our ability to generalise findings for some stratifications. While we evaluated the association of folate biomarkers with recent use of folic acid-containing dietary supplements as part of selected lifestyle factors, evaluating dietary folate intake overall and according to intake sources (fortified cereal-grain foods, ready-to-eat cereals, supplements, and combinations thereof) was beyond the scope of the present study. Lastly, because these are the first national estimates of folate vitamers, we are limited in our ability to compare findings to other studies of similar magnitude.
In summary, these novel data on serum folate forms generally show associations between these compounds and selected demographic, physiological and lifestyle variables similar to those reported previously for serum total folate. However, particularly for MeFox, we observed distinct patterns with the variables studied that may suggest altered folate metabolism dependent on biological characteristics. Thus, measuring MeFox as part of the folate profile may provide relevant information in populations with high or low folate status. While we cannot as yet answer the question of whether it is more accurate to include or exclude MeFox from the total folate, the difference between the two approaches is rather small (approximately 5 %). Based on the findings of the present study and for practical reasons, we suggest that until an unequivocal answer is found, MeFox should not be included in the calculation of serum total folate, but should be separately reported to allow an assessment of the quality of sample handling as well as potential insight into folate metabolism. This approach has two advantages: the total folate without MeFox can be directly compared to the microbiologic assay and other assays that do not measure this biologically inactive form and one errs on the side of caution with the interpretation of folate status by slightly underestimating it. The new reference intervals for serum folate forms in a population that has been exposed to folic acid fortification for over 15 years provide a much-needed benchmark to researchers and public health officials in those nations in which folic acid fortification has already occurred (e.g. the USA and Canada) and where folic acid intakes are significant contributors to total folate intakes, but also to nations that consider folic acid fortification (e.g. the UK).
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114515001142
Acknowledgements
We acknowledge contribution from the following laboratory members: Neelima Paladugula and Daniel Rabinowitz (CDC National Center for Environmental Health).
The authors' contributions to the study were as follows: C. M. P. conceptualised and designed the overall research project. C. M. P., M. R. S., Z. F. and M. Z. conducted most of the research. M. R. S. and C. M. P. analysed most of the data. C. M. P. drafted the manuscript. C. M. P. has primary responsibility for all content. All authors have read and approved the final manuscript.
The findings and conclusions in this manuscript are those of the authors and do not necessarily represent the official views or positions of the CDC/Agency for Toxic Substances and Disease Registry, the National Institutes of Health, the Food and Drug Administration or the Department of Health and Human Services.
Data collection and laboratory analyses of folate supported by funding from the Office of Dietary Supplements, NIH. None of the authors has reported a conflict of interest.
