Adriaan Kooy MD PhD, Department of Internal Medicine, Bethesda General Hospital Hoogeveen, Dr G.H. Amshoffweg 1, 7909 AA Hoogeveen, The Netherlands (fax: +31-528-286299; e-mail: firstname.lastname@example.org).
Abstract. Wulffelé MG, Kooy A, Lehert P, Bets D, Ogterop JC, Borger van der Burg B, Donker AJM, Stehouwer CDA (Bethesda General Hospital, Hoogeveen, The Netherlands; University of Mons, Mons, Belgium; Merck Nederland B.V., Amsterdam; Deaconesses’ Hospital, Meppel; Aleida Kramer Hospital, Coevorden; and Vrije Universiteit Medical Centre, Amsterdam; The Netherlands). Effects of short-term treatment with metformin on serum concentrations of homocysteine, folate and vitamin B12 in type 2 diabetes mellitus: a randomized, placebo-controlled trial. J Intern Med 2003; 254: 455–463.
Objective. Metformin is a key treatment option in type 2 diabetes. However, metformin may decrease vitamin B12 levels and increase levels of homo-cysteine, a cardiovascular risk factor. We investigated whether 16 weeks of treatment with metformin affects serum concentrations of homo-cysteine, folate and vitamin B12 in subjects with type 2 diabetes treated with insulin.
Design. Placebo-controlled, randomized trial. Measurements: at baseline and 16 weeks later.
Setting. This trial was conducted in the outpatient clinics of three general hospitals in The Netherlands.
Subjects. A total of 745 patients with type 2 diabetes, treated with insulin and not known with a contraindication for the use of metformin, were approached; 390 gave informed consent and entered the study. Thirty-seven subjects dropped out (12 placebo and 25 metformin users).
Intervention. Addition of metformin or placebo to insulin therapy.
Primary outcome parameters. Serum homocysteine, folate, vitamin B12, indices of glycaemic control and body weight.
Results. Amongst those who completed 16 weeks of treatment, metformin use, as compared with placebo, was associated with an increase in homocysteine of 4% (0.2 to 8; P = 0.039) and with decreases in folate [−7% (−1.4 to −13); P = 0.024] and vitamin B12 [−14% (−4.2 to −24); P < 0.0001]. In addition, the increase in homocysteine could be explained by the decreases in folate and vitamin B12.
Conclusion. In patients with type 2 diabetes, 16 weeks of treatment with metformin reduces levels of folate and vitamin B12, which results in a modest increase in homocysteine. The clinical significance of these findings remains to be investigated.
Type 2 diabetes mellitus is associated with a two- to fourfold increase in the risk of cardiovascular disease , which cannot be fully explained by important risk factors such as hyperglycaemia , hypertension  and dyslipidaemia .
Homocysteine is increasingly recognized as an independent [5–7], and potentially modifiable , risk factor that may be strongly linked to cardiovascular prognosis in type 2 diabetes [9, 10]. In addition, homocysteine may be a determinant of microalbuminuria and diabetic retinopathy [11–13].
Homocysteine is a sulphur-containing amino acid formed during the demethylation of methionine. Homocysteine can be remethylated to methionine or transsulphurated to cystathionine (Fig. 1). In homocysteine remethylation, 5-methyltetrahydrofolate is the methyl donor and vitamin B12 is an essential cofactor. Low serum folate and vitamin B12 levels are thus strongly associated with increased serum homocysteine [14, 15].
Serum vitamin B12 level is known to decrease during metformin treatment , probably due to malabsorption [17, 18]. Hence, during metformin treatment, homocysteine level might increase, but data on this issue are sparse and conflicting [19–21].
In view of these considerations, we studied the effects of metformin treatment on serum levels of homocysteine, vitamin B12 and folate in patients with type 2 diabetes in the setting of a randomized, placebo-controlled trial.
Material and methods
Patients and procedures
The HOME trial was designed to compare the metabolic and cardiovascular effects of metformin with those of placebo in insulin-treated type 2 diabetic patients . We aimed to include 400 patients with type 2 diabetes mellitus between 30 and 80 years of age who had received a diagnosis of diabetes after 25 years of age, had never had an episode of keto-acidosis, and whose blood-glucose-lowering treatment had previously consisted of oral agents but now exclusively consisted of insulin or insulin and metformin. We excluded pregnant women and women trying to become pregnant, patients with a Cockroft and Gault-estimated creatinine clearance <50 mL min−1  or low plasma cholinesterase (reference value ≥3.5 U L−1), and patients with congestive heart failure New York Heart Association class III/IV or other serious medical or psychiatric disease.
All patients gave written informed consent. The medical Ethical Committees of the three participating hospitals approved the trial protocol. The trial has been and is conducted in accordance with the Note for Guidance on Good Clinical Practice (CPMP/ICH/135/95) dated 17 July 1996 and in accordance with the Declaration of Helsinki (Revised Version of Hong Kong 1989).
The HOME trial is conducted in the outpatient clinics of three nonacademic hospitals (Hoogeveen, Meppel and Coevorden). All subjects were treated with only insulin for 12 weeks during a pre-randomization phase. From the beginning of this prerandomization phase all subjects were given advice for a healthy, moderately calorie-restricted diet and an active lifestyle . After these 12 weeks, all subjects were randomly assigned in a double-blind fashion to receive placebo or metformin in addition to insulin therapy for 16 weeks. All patients were numbered in order of study entry and received trial medication with the same number. The boxes and tablets of metformin and placebo had a similar appearance. Each subject successively increased the dose from one to finally three tablets of 850 mg day−1, if tolerated. The first tablet was taken at bedtime, the second at breakfast and the third at dinner. The treatment goals were fasting plasma glucose levels between 4 and 7 mmol L−1 and postprandial glucose levels between 4 and 10 mmol L−1. At the beginning and at the end of this 16-week short-term active treatment phase, fasting blood samples were drawn, a physical examination was carried out, and a complete medical history was taken. Results on glycaemic control, weight gain and insulin dose at the end of the 16 weeks have been reported elsewhere .
Homocysteine, folate and vitamin B12 were measured in serum. Homocysteine was measured by high-performance liquid chromatography (HPLC)  on a Gynkotek M580 LPG (Gynkotek/Softron GmBH, Germering, Germany) with a Spark Midas autosampler (Spark, Emmen, The Netherlands) and a Jasco 821-FP fluorescence detector (Jasco Spectroscopic Co., Hachioji City, Japan). Data were processed by Chromeleon v4.30 software (Gynkotek/Softron GmBH). Tributylfosfine was used as the reducing agent and 7-fluoro-benzo-2-oxa-1, 3 diazol-4-sulphonate as the thiol-specific fluorochromophore followed by HPLC with reverse phase chromatography and fluorescence detection. The intra- and interassay coefficients of variation are 2.5 and 9%, respectively.
Both vitamin B12 and folate were measured by electrochemiluminiscence immunoassay (ECLIA) [25, 26] on a Roche Elecsys 2010 (Roche Diagnostics, Basel, Switzerland). After vitamin B12 is released from intrinsic factor (IF), it is incubated with ruthenium-labelled IF. The resulting free, labelled IF is successively bound with biotine-labelled vitamin B12 and streptavidine-labelled paramagnetic microspheres as a solid phase. By using trypropylamine, the electrochemical emission was measured with a photomultiplier. The intra- and interassay coefficients of variation are 2.2 and 3.8%, respectively. The method of measuring folate was similar to that of vitamin B12, with use of ruthenium-labelled folate-binding protein and biotin-labelled folate. The intra- and interassay coefficients of variation are 5.5 and 8.5%, respectively.
We included only subjects with complete data sets. We log-transformed data on homocysteine, folate and vitamin B12 before analysis, because of their skewed distribution. Data are given as median (interquartile range).
The end-point of interest was the percentage change of each variable from baseline. The significance of the difference between the metformin and the control group was tested by a central t-test on log-transformed values. As log values are not directly interpretable, the antilogs are reported instead. These values are the geometric mean percentages of change from baseline.
We used structural equation modelling [27, 28] to investigate whether metformin-associated changes in homocysteine, if any, could be explained by changes in folate and (or) vitamin B12, and, if so, whether this was independent of age, gender and (changes in) serum creatinine, body mass index, insulin dose or glycated haemoglobin.
A P-value <0.05 was statistically significant.
We screened the medical files of all three participating outpatient clinics and identified 745 eligible patients. All were asked to enrol into the trial; 390 subjects gave written informed consent (Fig. 2). A total of 196 subjects were randomized to receive metformin (M) and 194 to receive placebo (P). The numbers of the metformin stoppers at the start of the ‘wash-out’ period of 12 weeks were comparable in both groups (placebo: 22; metformin: 23). Thirty-seven patients dropped out, 25 in the metformin and 12 in the placebo group, respectively. The actual mean dose in the metformin-treated group was 2163 mg. Each patient in this group maintained his/her maximally tolerated daily dose (one, two or three tablets of 850 mg) during the trial. Two patients never took their medication (P: 1, M: 1), nine withdrew their consent (P: 5, M: 4), and 26 experienced adverse effects (P: 6, M: 20). Of these 26, 11 experienced diarrhoea, five flatulence, four fatigue, one pruritus, one headaches, one pyrosis, one nausea, one myocardial infarction and one patient died suddenly. These 37 subjects are not included in the current analyses.
Table 1 shows baseline characteristics at the start of the short-term active treatment phase. Patients randomized to metformin were slightly older than patients randomized to placebo [63.2 (9.8) vs. 58.9 (11.1) years], but other characteristics were comparable between the two groups.
Table 1. Baseline characteristics
Placebo (n = 182)
Metformin (n = 171)
Data are normally distributed and given as mean (SD) or number (%).
Currently smoking, n (%)
Duration of diabetes (years)
Insulin treatment (years)
Cardiovascular, n (%)
Retinal coagulation and (or) cataract extraction, n (%)
Amputation, n (%)
Paresthesias, n (%)
Lipid-lowering drugs, n (%)
Blood pressure-lowering drugs, n (%)
Body mass index (kg m−2)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Daily dose of insulin (IU day−1)
Fasting plasma glucose (mmol L−1)
Total cholesterol (mmol L−1)
HDL cholesterol (mmol L−1)
LDL cholesterol (mmol L−1)
Triglycerides (mmol L−1)
Glycated haemoglobin (%)
Homocysteine, folate and vitamin B12
During placebo treatment, homocysteine increased by 0.40 μmol L−1 [4% (95% CI: 1 to 7)], folate increased by 1.43 nmol L−1 [13% (9 to 18)], and vitamin B12 decreased by −4.40 pmol L−1 [−2% (−6 to +2)] (Table 2). During metformin treatment, homocysteine increased by 1.10 μmol L−1 [8%(6 to 11); P = 0.04 versus placebo], folate increased by 0.88 nmol L−1 [6% (2 to 11); P = 0.02 versus placebo], and vitamin B12 decreased by −47.4 pmol L−1 [−16% (−12 to −19); P = 0.001 versus placebo]. Compared with placebo, metformin treatment was associated with an increase in homocysteine of 4% [(0.2 to 8); P = 0.039] and with decreases in folate [−7% (−1.4 to −13); P = 0.024] and vitamin B12 [−14% (−4.2 to −24); P < 0.0001]. The effects of metformin on the variables of Table 2 were re-analysed with adjustment for age, previous metformin treatment, duration of diabetes, gender, insulin dose and actual metformin dose (separately for each variable as well as all together); all of which did not materially change the results (data not shown).
Table 2. Changes in serum total homocysteine, folate and vitamin B12 after 16 weeks of treatment with either metformin or placebo
Change (95% CI)
Geometric mean change (95% CI)
P (metformin versus placebo)
t = 0
t = 16
t = 0
t = 16
Metformin versus placebo
Data are median (IQR or 95% CI). The change in geometric mean can be read as a fraction. For example, 1.08 change in homocysteine during metformin treatment is an 8% increase from baseline.
Total homocysteine (μmol L−1)
13.6 (10.8 to 16.6)
13.9 (11.4 to 16.3)
13.6 (11.5 to 15.9)
14.4 (12.3 to 17.2)
0.4 (−1.0 to 1.6)
1.1 (−0.3 to 2.6)
1.04 (1.01 to 1.07)
1.08 (1.06 to 1.11)
4 (0.2 to 8)
Folate (nmol L−1)
13.02 (10.38 to 17.25)
14.71 (12.11 to 18.49)
13.52 (10.33 to 16.89)
14.07 (11.58 to 17.28)
1.43 (−0.73 to 4.56)
0.88 (−1.52 to 2.85)
1.13 (1.09 to 1.18)
1.06 (1.02 to 1.11)
−7 (−1.4 to −13)
Vitamin B12 (pmol L−1)
397.9 (318.9 to 475.5)
394.5 (306.3 to 451.2)
390.1 (319.8 to 480.2)
332.6 (271.3 to 404.6)
−4.4 (−45.3 to 38.3)
−47.4 (−104.4 to −9.8)
0.98 (0.94 to 1.02)
0.84 (0.81 to 0.88)
−14 (−4.2 to −24)
According to the reference values used for vitamin B12 (180–700 pmol L−1) and folate (4.5–21.0 nmol L−1), the numbers of patients with vitamin B12 values below the lower reference limit were 8 (P: 6 and M: 2) at the start and 8 (P: 5 and M: 3) at the end of the study period, and the numbers of patients with folate values below the lower reference limit were 2 (P: 1 and M: 1) at the start and 1 (P: 0 and M: 1) at the end of the study period. The numbers of patients with vitamin B12 values above the upper reference limit were 11 (P: 5 and M: 6) at the start and 8 (P: 4 and M: 4) at the end of the studied period, and the numbers of patients with folate values above the upper reference limit were 43 (P: 24 and M: 19) at the start and 47 (P: 29 and M: 18) at the end of the studied period.
Additional analyses of the patients with vitamin B12 and/or folate concentrations in the lowest quartile at study start did not show a different effect of metformin on homocysteine levels, as compared with the (small) effect of metformin on those levels in the total group.
Structural equation modelling
Figure 3 shows that metformin treatment affected homocysteine levels at 16 weeks through its effects on folate and vitamin B12, and had no direct effect on homocysteine levels at 16 weeks. These results were not affected by adjustment for age, gender or (changes in) serum creatinine, body mass index, insulin dose or glycated haemoglobin (data not shown).
This is the first randomized, placebo-controlled study that reports on the effects of treatment with metformin on serum concentrations of homocysteine, folate and vitamin B12. We found that 16 weeks of metformin treatment in patients with type 2 diabetes was associated with an increase in serum homocysteine of ∼4% (about 0.7 μmol L−1) and with decreases in serum folate and vitamin B12 of ∼7 and ∼14%, respectively. In addition, further analysis indicated that the increase in serum homocysteine was mediated by the decreases in serum folate and vitamin B12. To our knowledge, no previous study has demonstrated a folate-reducing effect of metformin in patients with type 2 diabetes intensively treated with insulin.
Our results on vitamin B12 are consistent with previous studies, which have reported that metformin treatment was associated with decreases in serum vitamin B12 of 16 to 22% [16, 18, 29]. Metformin is thought to induce malabsorption of vitamin B12 and IF in the ileum, an effect that can be reversed by increased calcium intake [17, 30, 31].
How metformin affects folate status is not known, but findings similar to ours have been reported previously in nondiabetic individuals . Serum folate increased in both groups, but significantly less so in the metformin-treated group. The overall increase in folate levels we observed may be related to the fact that all trial participants received regular dietary counselling, which may have increased their vegetable intake.
The metformin-associated decreases in folate and vitamin B12 led to an increase in serum homocysteine of ∼4% or about 0.7 μmol L−1, which is similar to the estimate obtained in an earlier case–control study . The clinical significance of such an increase is not yet clear but may not be negligible. A recent meta-analysis estimated, in nondiabetic individuals, that a persistent increase in serum homocysteine of ∼3 μmol L−1 may be associated with increases in the risk of coronary heart disease and stroke of about 10 and 20%, respectively . There is evidence that such risk increases may be greater amongst diabetic individuals [9, 10], i.e. homocysteine may be a stronger cardiovascular risk factor amongst diabetic than amongst nondiabetic individuals.
Note that, as compared with routine clinical practice, we may have somewhat underestimated these adverse effects of metformin. First, the participants in our trial were folate- and vitamin B12-replete, and received dietary counselling, which may have attenuated the impact of metformin treatment on vitamin status and thus homocysteine levels. In routine clinical practice, poor folate and vitamin B12 status, which are common amongst the elderly , may easily go undetected. The effects of metformin treatment may then be much greater, because the relationship between folate and vitamin B12 status on the one hand and homocysteine on the other is nonlinear, i.e. greater at low levels of these vitamins . Secondly, we cannot exclude that the effects on vitamin status and homocysteine concentrations of prolonged treatment with metformin are greater than those of short-term treatment; this issue merits further investigation. Finally, it should be emphasized that the beneficial effects of metformin treatment with regard to complications of type 2 diabetes have been clearly demonstrated . Our data raise the possibility that these favourable effects of metformin may be even more pronounced if decreases in folate and vitamin B12 are avoided.
We cannot be certain that our results can be generalized to patients with type 2 diabetes not treated with insulin. However, previous studies have shown that metformin reduced vitamin B12 levels in patients who received metformin alone or combined with sulphonyl ureas [16, 18, 29]. It therefore seems likely that a rise in homocysteine levels may also occur in patients not treated with insulin.
In conclusion, we found that 16 weeks of metformin treatment in patients with type 2 diabetes was associated with decreases in serum folate and vitamin B12 of ∼7 and ∼14%, respectively, which led to an increase in serum homocysteine of ∼4%. The clinical consequences of these changes are unclear and merit further study.
Conflict of interest statement
D. Bets is a clinical research assistant, and P. Lehert is an occasional consultative statistician for Merck Nederland BV.
This part of the HOME trial was supported by grants from Byk, Lifescan, E. Merck/Lipha, Merck Sharpe & Dohme and Novo Nordisk. We thank the study nurses Liesbeth Breedland and Els van Driesum for their dedication to the concerns of the patients and the quality of the treatments, Jan van der Kolk and Steven Berends for their technical assistance and validation of the laboratory techniques, Gerard de Groot, Rob Hoorn and Erk Pieterse for their hospitality at their laboratories, and all other members of the HOME study group for their contribution to the HOME trial.