Choline functions in several important structural and cell signalling roles, integral in the formation of the phospholipids phosphatidylcholine and sphingomyelin, VLDL and the neurotransmitter acetylcholine(Reference Ueland1). Consequently, the metabolic implications of choline deficiency are wide ranging and severe and include fatty liver disease, DNA damage, cell apoptosis, altered gene expression and cognitive impairments(Reference Ueland1, Reference Zeisel2). The importance of choline as an essential nutrient was accepted in 1998 with the establishment in the USA of a choline-adequate intake (AI) of 550 mg/d for men and 425 mg/d for women(3). The only other source of choline is via de novo synthesis from phosphatidylethanolamine by the enzyme phosphatidylethanolamine-N-methyltransferase, with S-adenosylmethionine as the methyl donor(Reference Dudman, Guo and Gordon4), in a pathway enhanced by oestrogen activation of phosphatidylethanolamine-N-methyltransferase(Reference Drouva, LaPlante and Leblanc5). As a result, a woman's capacity to synthesise choline is likely to diminish after menopause and, consequently, dietary intake of choline is reported to assume much greater significance with respect to homocysteine metabolism and liver function in postmenopausal women(Reference Fischer, daCosta and Kwock6, Reference Fischer, da Costa and Kwock7).
As a source of the methyl donor betaine, produced by oxidation of choline primarily in the liver, choline is important in the conversion of homocysteine to methionine. Remethylation of homocysteine occurs primarily via methionine synthase, using methylcobalamin (vitamin B12) as a cofactor and 5-methyltetrahydrafolate as a cosubstrate. However, remethylation also occurs via the betaine–homocysteine methyltransferase (BHMT) pathway, which requires betaine, with the importance of this pathway increased in the face of sub-optimal folate status(Reference Holm, Hustad and Ueland8). A high plasma concentration of total homocysteine (tHcy) is associated with an increased risk of a number of age-related diseases(Reference Wald, Law and Morris9–Reference van den Kommer, Dik and Comijs14), and although clinical trials have failed to confirm a causative link between elevated tHcy and CVD in patients with existing pathology(Reference Marti-Carvajal, Sola and Lathyris15, Reference Clarke, Halsey and Lewington16), interest remains on the potential for primary prevention of disease, particularly stroke(Reference Wang, Qin and Demirtas17, Reference Lee, Hong and Chang18). Folic acid supplementation has been consistently shown to decrease tHcy(19), with additional roles for vitamin B12(Reference Eussen, de Groot and Clarke20), vitamin B6(Reference McKinley, McNulty and McPartlin21) and riboflavin(Reference McNulty, Dowey and Strain22) identified. Betaine and choline intake and status are determinants of tHcy(Reference Holm, Hustad and Ueland8, Reference Cho, Zeisel and Jacques23, Reference Lee, Jacques and Dougherty24) and betaine supplementation can lower tHcy(Reference Schwab, Torronen and Toppinen25–Reference Price, Keaveney and Hamill28). Studies suggest that plasma tHcy can be lowered postprandially in response to acute and short-term increases in choline intake(Reference Olthof, Brink and Katan29, Reference Atkinson, Elmslie and Lever30); however, effects on fasting plasma tHcy are equivocal.
Low choline intakes among adults have been reported previously(Reference Lee, Jacques and Dougherty24, Reference Chiuve, Giovannucci and Hankinson31–Reference Xu, Gammon and Zeisel33) and indeed research reported from our own centre indicated that mean choline intakes among men and women aged 54–65 years were 172 (sd 72) mg/d, considerably lower than the AI(Reference Price, Keaveney and Hamill28). Given the importance of choline in a wide range of physiological functions, together with consistent evidence of sub-optimal intakes, it has been suggested that the intake of choline-rich foods should be increased(Reference Zeisel, Da Costa and Franklin34). However, recent research(Reference Wang, Klipfell and Bennett35) which reported an association between choline status and risk of vascular disease would caution against such advice and rather highlights the need for more research into the physiological functions of this nutrient.
We assessed the effect on plasma concentrations of free choline, betaine and tHcy and on B vitamin status of choline supplementation at 1 g/d, a dose approximating to twice the current AI, over 12 weeks in healthy postmenopausal women, a group that would be expected to have a higher choline requirement. Previous research(Reference Olthof, van Vliet and Verhoef36) had reported that betaine and choline supplementation increased TAG concentration; therefore, this effect was also investigated.
Women aged 49–71 years were recruited to the study from within the University and from the local community. All interested subjects were interviewed regarding general health and well being and were considered eligible for screening if they were independently mobile and postmenopausal (defined as last spontaneous menstrual bleeding >12 months previously). The exclusion criteria were smoking, BMI ≥ 35 kg/m2, gastrointestinal, hepatic or renal disease, endocrine or metabolic abnormalities, use of drugs or dietary supplements known to influence B vitamin or choline status or metabolism, haematological disorder, serum vitamin B12 concentration < 111 pmol/l or serum creatinine concentration ≥ 130 μmol/l. Initially, eighty-five postmenopausal women were recruited and following screening, forty-four women were deemed eligible to participate. This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by the Research Ethical Committee of the University of Ulster. Written informed consent was obtained from all participants.
The study was a randomised, double-blind, placebo-controlled intervention trial undertaken during 2005–6. The sample size was estimated by using data for tHcy response to folic acid supplementation(Reference Ward, McNulty and McPartlin37). On the basis of the means and standard deviations of homocysteine response to treatments compared with that to placebo in this previous study, it was estimated that a sample size of nineteen subjects per group was needed to detect a difference of 1·8 μmol homocysteine/l with a power of 80 % at α = 0·05. To allow for a 20 % dropout rate, we estimated that a sample size of twenty-three subjects in each treatment group would be required.
Subjects were stratified according to reported use of hormone replacement therapy (HRT) and fasting tHcy and subjects in each stratum were randomised to receive 2·4 g choline bitartrate per d (1 g choline) or placebo (2·4 g tartaric acid per d) (Akzo Nobel Functional Chemicals). The supplement provided choline at approximately twice the AI of 425 mg/d; research suggests that intakes higher than the current AI may be required to optimise choline status in postmenopausal women(Reference Fischer, da Costa and Kwock7). Each choline bitartrate capsule contained 250 mg of choline and subjects were instructed to take two capsules twice a day with food. All supplements were identical in appearance and taste. To maximise compliance, supplements were distributed every 2 weeks in 7 d pillboxes. The pillboxes were then collected, and the number of unused tablets was recorded to monitor compliance. Volunteers were instructed not to consume any B vitamin or choline-containing supplements other than those provided, and were asked to maintain their usual diet and lifestyle throughout the intervention period. Previous research has reported the short-term effects of choline supplementation(Reference Holm, Ueland and Kvalheim38). In contrast, we wished to investigate the longer-term effects of low-dose supplementation on tHcy and on lipid profile and, therefore, individuals were supplemented for 12 weeks. Weight (kg) and height (m) were measured at recruitment.
Blood sampling and biochemical measurements
Blood samples were collected at the Human Interventions Study Unit, University of Ulster at baseline, week 6 and week 12. Following an overnight fast (12 h), a blood sample was obtained from each participant. Blood samples were collected into EDTA-containing tubes for plasma and erythrocyte extraction and into a serum separation tube for serum extraction. The EDTA-containing tube was wrapped in foil (owing to the light sensitivity of a number of the analytes) and was placed on ice immediately after collection. Sample preparation and fractionation were performed within 0·5–2·5 h of the time of sampling as described in detail elsewhere(Reference McKinley, McNulty and McPartlin21), and fractions were stored at − 70°C for batch analysis at the end of the study and at − 20°C for extraction of DNA.
Plasma free choline, betaine, tHcy, dimethylglycine (DMG) and methionine were analysed by liquid chromatography–tandem MS as described previously(Reference Holm, Ueland and Kvalheim38). Serum folate(Reference Molloy and Scott39) and vitamin B12(Reference Kelleher and Broin40) were measured by microbiological assay. Plasma pyridoxal 5-phosphate (the main active derivative of vitamin B6) concentrations were determined by reverse-phase HPLC with fluorescence detection(Reference Bates, Pentieva and Matthews41, Reference Frosst, Blom and Milos42). Riboflavin status was determined by erythrocyte glutathione reductase activation coefficient (EGRac), a functional assay which measures the activity of glutathione reductase before and after in vitro reactivation with its prosthetic group FAD; EGRac is calculated as a ratio of FAD stimulated to unstimulated enzyme activity with values ≤ 1·3 generally considered to represent adequate riboflavin status(Reference Powers, Bates and Prentice43). For each participant, DNA was extracted using a QIAamp DNA Mini Kit (Qiagen Limited). Genotyping for the methylenetetrahydrofolate reductase 677 C → T polymorphism was performed by PCR and Hinf1 digestion as described previously(Reference Frosst, Blom and Milos42).
Serum concentrations of total cholesterol, HDL-cholesterol and TAG were measured using a Hitachi 911 analyser (Roche) and enzymatic assays (Roche) and LDL-cholesterol concentrations were calculated(Reference Friedewald, Levy and Fredrickson44).
Data are presented as medians with 25th and 75th percentiles or means and standard deviations, as appropriate. Analysis of serum folate indicated a skewed distribution; therefore, data were transformed logarithmically to approximate normal distribution before statistical analyses. Baseline values for the placebo and choline groups were compared using an independent samples t test or χ2 test, as appropriate. Bivariate relationships between choline and B vitamin-related measures were evaluated using Spearman's rank correlation coefficients. Data were analysed by repeated-measures ANOVA using the general linear model with choline/placebo as the between-subject factor within time points. If there was a significant effect of treatment, specific contrasts were made using post hoc comparisons with Bonferroni's correction. To investigate if the response of tHcy to supplementation was dependent on baseline tHcy concentration or folate status, the intervention group was divided via median split into high (>median) and low ( ≤ median) tHcy or serum folate and the analysis repeated. Statistical analyses were carried out using SPSS 17.0 for Windows (SPSS, Inc.). Results were considered statistically significant when P < 0·05.
Baseline characteristics and study compliance
Of the forty-six individuals originally deemed eligible, four subjects did not complete the 12-week intervention and complete data sets were available for forty-two individuals, who were included in the final analysis. There were no significant differences in age, BMI, HRT use, concentration of tHcy, choline, betaine and B-vitamins between the choline and placebo groups at baseline (Table 1). The prevalence of the methylenetetrahydrofolate reductase 677TT polymorphism did not differ significantly between the choline-supplemented (n 2) and placebo groups (n 1). On the basis of pill counts, the compliance of subjects with the study protocol was excellent (>95 % of supplements were consumed by participants), with no reports of side effects of supplementation.
HRT, hormone replacement therapy; MTHFR, methylenetetrahydrofolate reductase; tHcy, total homocysteine; PLP, pyridoxal 5-phosphate; EGRac, erythrocyte glutathione reductase activation coefficient; DMG, dimethylglycine.
* The choline and placebo groups were not significantly different for any of the variables at baseline when examined using independent samples t test or χ2 test as appropriate.
† EGRac is a functional indicator of riboflavin status.
Baseline serum tHcy concentrations ranged from 4·62 to 19·8 μmol/l; eighteen of the forty-two participants were classified as having an elevated tHcy, i.e. tHcy >10 μmol/l(3). A value of 7 nmol/l for serum folate has been reported as the cut-off point for negative folate balance(3), and on this basis, no individual was classified as folate deficient. At baseline, no individual was found to be vitamin B12 deficient, i.e. no subject had a serum vitamin B12 concentration < 120 pmol/l(3). Vitamin B6 status as assessed by plasma pyridoxal 5-phosphate concentration at baseline ranged from 10·1 to 55·3 nmol/l; and twenty-three (55 %) of the participants had low vitamin B6 status as defined by a plasma pyridoxal 5-phosphate concentration < 20 nmol/l(Reference Olthof, van Vliet and Verhoef36). Riboflavin deficiency, as defined by an EGRac value of ≥ 1·3(Reference Holm, Hustad and Ueland8, Reference Holm, Ueland and Vollset45, Reference Konstantinova, Tell and Vollset46), was evident in thirteen volunteers at baseline. Serum tHcy concentrations at baseline were significantly correlated with serum folate (r − 0·458; P = 0·002), serum vitamin B12 (r − 0·428; P = 0·005), plasma creatinine (r 0·392; P = 0·01) and plasma methionine (r − 0·615; P = 0·001), but not plasma vitamin B6 or riboflavin. At baseline, choline was positively associated with betaine (r 0·389; P = 0·011) but not DMG or tHcy, while betaine was positively associated with DMG (r 0·367; P = 0·017).
At baseline, concentrations of free choline were higher among the 29 % of women who reported HRT use compared to non-users; median (5th, 95th percentile) 7·4 (5·2, 12·9) v. 6·9 (4·8, 9·3) nmol/l, a difference which approached statistical significance (P = 0·057; independent samples t test, data not shown).
Table 2 shows the response of plasma concentrations of free choline, betaine, DMG, tHcy and methionine to 12-week choline (1 g/d) supplementation. Repeated-measures ANOVA indicated that compared to placebo treatment, plasma choline at week 6 (P = 0·001) was significantly higher in the choline-supplemented group, a difference that remained at week 12 (P = 0·001), therefore, indicating a significant treatment effect between groups over time F(1,39) = 8·65, P = 0·001. In addition, plasma betaine was significantly higher following 6 weeks of choline supplementation, an effect that remained at week 12 (P = 0·001). There was a significant treatment effect in response to choline supplementation for betaine F(1,39) = 21·55, P = 0·0001 and DMG F(1,39) = 6·04, P = 0·005 over 12 weeks. At week 6 and week 12, there were no significant differences in tHcy or methionine concentrations in the choline-supplemented group compared to placebo. Serum folate, vitamin B12 and riboflavin status were not altered by choline supplementation; however, vitamin B6 status (as determined by plasma pyridoxal 5-phosphate) was significantly higher at week 12 in the choline-supplemented group compared to placebo (P < 0·01); however, there was no treatment effect in response to choline for vitamin B6 over time F(1,39) = 1·99, P = 0·17 (Table 3).
Median values were significantly different from the placebo group at corresponding time point: * P < 0·001, ** P < 0·05.
Median values were significantly different from change in the placebo group at corresponding time point: † P < 0·001, ††P < 0·01 (repeated-measures ANOVA, and post hoc comparisons with Bonferroni's correction).
PLP, pyridoxal 5-phosphate; EGRac, erythrocyte glutathione reductase activation coefficient.
* Median values were significantly different from the placebo group at corresponding time point (P < 0·01, repeated-measures ANOVA, and post hoc comparisons with Bonferroni's correction).
† EGRac is a functional indicator of riboflavin status.
The median (25th, 75th percentile) change in plasma tHcy concentration at week 6 in the choline-supplemented group was − 0·9 ( − 1·6, 0·2) μmol, a change which when compared to the median change observed in the placebo group, 0·6 ( − 0·4, 1·9) μmol, approached statistical significance (P = 0·058; independent samples t test). Furthermore, the response to treatment was not significantly different in women with high (>median, 9·8 μmol/l) v. low ( ≤ median) tHcy and comparing those with high (>median, 10 nmol/l) v. low ( ≤ median) serum folate (data not shown).
There were no significant differences in plasma concentrations of total cholesterol, LDL-cholesterol, HDL-cholesterol, TAG or glucose at week 6 or week 12 in the choline-supplemented group compared to the placebo group (Table 4).
HDL-C, HDL-cholesterol; LDL-C, LDL-cholesterol.
* There were no significant differences between treatment and placebo groups at baseline. No significant effect of treatment (repeated-measures ANOVA).
To our knowledge, the present study is the first long-term, randomised, placebo-controlled intervention study to examine the effect of choline supplementation, not only on plasma tHcy, but also on concentrations of free choline, betaine and DMG and on B-vitamin status. Although these variables are important with respect to the methionine cycle, they were not comprehensively reported in the previous intervention studies(Reference Olthof, Brink and Katan29, Reference Atkinson, Elmslie and Lever30, Reference Veenema, Solis and Li47). The present results showed that long-term supplementation with 1 g/d choline, as choline bitartrate, increased plasma free choline concentration, as well as concentrations of the choline metabolites, betaine and DMG. Supplementation resulted in a decrease in plasma tHcy after 6 weeks, which approached, but did not reach, statistical significance, possibly owing to the relatively high folate status and low tHcy of the present study population. In contrast to previous research(Reference Olthof, van Vliet and Verhoef36), we did not observe any significant adverse effect of choline supplementation on plasma lipid profiles.
In the present study, plasma choline, betaine and DMG concentrations at baseline appeared marginally lower than those reported previously(Reference Holm, Hustad and Ueland8, Reference Holm, Ueland and Vollset45, Reference Konstantinova, Tell and Vollset46) and these may be explained by our measurement of fasting blood samples, in contrast to other studies. However, the lower values may also reflect the fact that the present study population comprised only postmenopausal women, who would be expected to have lower choline status owing to the loss of oestrogen, which naturally enhances choline synthesis(Reference Drouva, LaPlante and Leblanc5). In support of this contention, we did note that free choline concentration at baseline was higher in women who reported using HRT, a source of exogenous oestrogen. Recently, Fischer et al. (Reference Fischer, da Costa and Kwock7) reported that oestrogen treatment was associated with a decrease in the requirement for dietary choline among postmenopausal women. Folate has a betaine-sparing effect(Reference Melse-Boonstra, Holm and Ueland48) and folate status of the women in the present study was high. This observation is most probably explained by the wide availability of folic acid-fortified foods, which can contribute significantly to folate status(Reference Hoey, McNulty and Askin49). Indeed, serum folate concentrations reported here were similar to values reported by us in a healthy adult cohort who habitually consumed between 40 and 98 μg folic acid/d from fortified foods(Reference Hoey, McNulty and Askin49). Plasma tHcy concentration was marginally lower than might have been expected among postmenopausal women; with only eighteen of the women observed to have a basal tHcy >10 μmol/l; again probably reflecting the generally high folate status of the study group. The inclusion of individuals who were regular consumers of fortified foods could be considered a limitation of our study design, in that such individuals would be less likely to respond to the homocysteine-lowering effect of choline. However, such individuals are representative of the general population in Northern Ireland, where over 75 % consume foods fortified with folic acid and other B vitamins at least once per week(Reference Hoey, McNulty and Askin49).
We observed a significant increase in circulating choline concentrations following supplementation with choline at a dose of 1 g/d, in the form of choline bitartrate. This dose would have resulted in a marked increase in dietary choline which we have recently estimated in a similar population to be approximately 300 mg/d(Reference Price, Keaveney and Hamill28). Whether the significant effects on status observed in the present study would have been evident with lower doses of supplementation is not known; a previous dose–response intervention reported that while supplementation with 500 mg choline/d, as choline chloride, did increase plasma free choline concentration, no effect was evident with 300 mg/d(Reference Veenema, Solis and Li47). We also recorded significant increases in betaine and DMG concentrations post-supplementation. Although betaine can be obtained directly from the diet (with wholegrain cereals, spinach and beetroot being particularly good dietary sources(50)), it can also be formed from choline in a two-step enzymatic process, occurring in the liver and kidney, where choline is first oxidised to betaine aldehyde, and then further oxidised to betaine(Reference Ueland, Holm and Hustad51). Our findings are in keeping with observational research showing that choline and betaine were positively associated(Reference Holm, Ueland and Vollset45), and with an intervention study which also reported an increase in plasma betaine in men supplemented with choline(Reference Veenema, Solis and Li47). BHMT-mediated remethylation of homocysteine results in the conversion of betaine to DMG, and plasma betaine concentration was positively associated with DMG concentration in the present study. However, previous studies have reported that this association is only evident at lower DMG concentrations, a finding which the authors concluded was indicative of inhibition of BHMT by higher concentrations of DMG(Reference Holm, Ueland and Vollset45). Supplementation had no significant effect on the concentration of folate, vitamin B12 or riboflavin, perhaps reflecting the overall adequate status of these nutrients. In contrast, supplementation was associated with a significant increase in vitamin B6, the status of which was low at baseline in just over half of the present study population. Observational research has previously reported an association between betaine and plasma vitamin B6(Reference Holm, Hustad and Ueland8), and therefore the increase in vitamin B6 in the present intervention study may be owing to the higher betaine concentration observed following choline supplementation. Although our result is in keeping with the literature and scientifically plausible in respect of a B6-sparing effect with enhanced betaine status, we cannot however discount the possibility that this was a chance finding.
In the present study, choline supplementation for 6 weeks was accompanied by a modest effect in lowering plasma tHcy median (25th, 75th percentile) (%) change in plasma tHcy concentration ( − 0·9 ( − 1·6, 0·2) μmol/l ( − 9 %), v. 0·6 ( − 0·4, 1·9) μmol/l (6 %) in the choline and placebo groups, respectively, P = 0·058). Phosphatidylcholine supplementation, providing 2·6 g choline/d, for 2 weeks was previously shown to significantly decrease fasting plasma tHcy by 3 μmol/l (18 %)(Reference Atkinson, Elmslie and Lever30); however, the mean baseline tHcy concentration in that study was considerably higher than in the present intervention (14·7 v. 10·1 μmol/l). Indeed, subjects were recruited to the study of Olthof et al. (Reference Olthof, Brink and Katan29) on the basis of a high tHcy and were supplemented with a dose of choline that was two-and-a-half times higher than in our investigation, which may also explain the differing results. Previously, the consumption of a choline-rich meal, providing 760 mg of choline, was associated with a lower plasma tHcy at 4 h post-meal(Reference Atkinson, Elmslie and Lever30). In contrast to our study, the latter study showed no concomitant increase in plasma betaine, and, therefore, the authors concluded that the homocysteine-lowering effect was independent of the BHMT pathway and was possibly owing to a decrease in choline synthesis, which also generates homocysteine. Other researchers have reported that while choline intake did not alter fasting tHcy concentration, supplementation did attenuate the rise in plasma tHcy after a methionine load(Reference Veenema, Solis and Li47), suggesting that the BHMT pathway of homocysteine remethylation may be more important in the fed state. Consequently, a more marked tHcy lowering may have been observed in the present study if non-fasting samples had been collected. The modest homocysteine-lowering effect with choline in this study may also be explained by the high folate status of our study population. Folate status was higher than in the two previously reported intervention studies(Reference Olthof, Brink and Katan29, Reference Atkinson, Elmslie and Lever30), which is of note as it has been reported that the BHMT pathway of homocysteine remethylation comes into play only under conditions of impaired B-vitamin status(Reference Holm, Hustad and Ueland8, Reference Holm, Ueland and Vollset45). In the present sample population, we did not exclude women on the basis of HRT use but rather stratified across groups by oestrogen use. Clearly, however, HRT composition and duration of use are likely to have varied and, therefore, given the established relationships among oestrogen, choline and tHcy, this may have confounded the response of tHcy to supplementation.
The concomitant lowering in plasma tHcy, a reported risk factor for vascular disease(Reference Wang, Klipfell and Bennett35), with increased choline intake, might be predicted to lower disease risk; however, not all research(Reference Cho, Zeisel and Jacques23, Reference Dalmeijer, Olthof and Verhoef32) concurs with this contention. Indeed a very recent study by Wang et al. (Reference Wang, Klipfell and Bennett35), using a targeted metabolomics approach, concluded that plasma concentrations of choline, betaine and a subsequent metabolite, trimethylamine N-oxide, were positively associated with the risk of vascular disease. However, while the authors found, using confirmatory animal studies, that gut microbiota metabolise choline into trimethylamine, which is converted into trimethylamine N-oxide in the liver, they did not establish whether increased trimethylamine N-oxide contributes directly to CVD progression or is simply a marker of disease risk. The authors did not observe any significant effect of choline on plasma cholesterol, TAG or glucose concentrations, in keeping with our findings. Earlier research indicated that choline, supplemented as phosphatidylcholine, increased plasma TAG concentrations(Reference Olthof, van Vliet and Verhoef36), which was attributed to the role of phosphatidylcholine, as an obligate component of VLDL.
In conclusion, research has linked choline intake and status with a number of health conditions(Reference Xu, Gammon and Zeisel52, Reference Bjelland, Tell and Vollset53), highlighting the benefits of optimal choline status. The present study shows that choline supplementation in postmenopausal women, at a dose approximately equivalent to twice the current AI for women, can significantly increase circulating free choline concentration as well as the concentration of the methyl donor betaine, and may also lower plasma tHcy. Despite the recent reports that excess dietary choline might lead to CVD(Reference Wang, Klipfell and Bennett35), choline is an essential nutrient for several key metabolic pathways, and consequently, we caution, as have others(Reference Rak and Rader54) that strategies to lower choline for therapeutic purposes should proceed with care.
The authors wish to thank all the women who participated in the present study. The authors also acknowledge Dr Ian Bradbury for his advice on statistical analysis. This study was supported by the Department of Employment and Learning, Northern Ireland. Akzo Nobel Functional Chemicals supplied the choline supplements and placebo capsules. J. M. W. W. was involved in hypothesis generation, organisation of the study, statistical analysis, data interpretation and manuscript preparation. P. M. W. and M. E. D. were responsible for execution of the study, sample collection and biochemical analysis. J. J. S., J. M. M., H. M., M. W., M. P. B. and P. J. R. assisted in hypothesis generation, study design and data interpretation. A. M. M., J. M. S. and P. M. U. were involved in sample analysis and data interpretation. All authors read and approved the manuscript. No author had any conflict of interest.