N,N-dimethylglycine

Methods

Patients with type 2 diabetes (n = 34) and control subjects (n = 17) were recruited. We measured baseline fasting plasma and 4-hour post-methionine load (L-methionine, 0.1 mg/kg body weight) concentrations of homocysteine, betaine, and the betaine metabolite N,N-dimethylglycine. Baseline urine excretions of betaine, dimethylglycine and glucose were measured on morning urine samples as the ratio to urine creatinine. Statistical determinants of the post-methionine load increase in homocysteine were identified in multiple linear regression models.

Results

Plasma betaine concentrations and urinary betaine excretions were significantly (p < 0.001) more variable in the subjects with diabetes compared with the controls. Dimethylglycine excretion (p = 0.00014) and plasma dimethylglycine concentrations (p = 0.039) were also more variable. In diabetes, plasma betaine was a significant negative determinant (p < 0.001) of the post-methionine load increase in homocysteine. However, it was not conclusive that this was different from the relationship in the controls. In the patients with diabetes, a strong relationship was found between urinary betaine excretion and urinary glucose excretion (but not with plasma glucose).

Methionine load test study

The study was approved by the Canterbury Ethics Committee (New Zealand) and all subjects gave written informed consent. Patients with Type 2 diabetes were identified through the Canterbury retinal screening programme database and invited to participate. Controls were recruited by posting notices around the hospital campus. Thirty four patients (aged 51–75 yrs, 20 males) and 17 controls (aged 54–69 yrs, 10 male) were recruited. They were asked to fast for 16 hours and to withhold morning medications. Their height, weight, age, and medical history were recorded. Venous blood was drawn into EDTA tubes that were placed on ice immediately for determining fasting plasma betaine and homocysteine concentrations and a urine sample was collected. They were then given a methionine load of 0.1 g/kg body weight of L-methionine dissolved in 100 ml fruit juice. They also ate a standard snack consisting of a jam (jelly) sandwich to avoid hypoglycaemia. They took no further food or fluid until their return 4 hours later, when a second venous blood sample was collected. The betaine, dimethylglycine and homocysteine from the pre- and post- methionine load blood samples were measured in the same analytical batches.

Differences between subjects with and without diabetes

The post-methionine load plasma betaine and homocysteine concentrations were also significantly more variable in the subjects with diabetes (on log-transformed data, p < 0.001 for betaine and p = 0.0016 for N,N-dimethylglycine), as was the post-methionine change in plasma N,N-dimethylglycine (p = 0.013), but not the post-methionine change in plasma betaine (p = 0.7).

Methods

Patients with type 2 diabetes (n = 34) and control subjects (n = 17) were recruited. We measured baseline fasting plasma and 4-hour post-methionine load (L-methionine, 0.1 mg/kg body weight) concentrations of homocysteine, betaine, and the betaine metabolite N,N-dimethylglycine. Baseline urine excretions of betaine, dimethylglycine and glucose were measured on morning urine samples as the ratio to urine creatinine. Statistical determinants of the post-methionine load increase in homocysteine were identified in multiple linear regression models.

Methionine load test study

The study was approved by the Canterbury Ethics Committee (New Zealand) and all subjects gave written informed consent. Patients with Type 2 diabetes were identified through the Canterbury retinal screening programme database and invited to participate. Controls were recruited by posting notices around the hospital campus. Thirty four patients (aged 51–75 yrs, 20 males) and 17 controls (aged 54–69 yrs, 10 male) were recruited. They were asked to fast for 16 hours and to withhold morning medications. Their height, weight, age, and medical history were recorded. Venous blood was drawn into EDTA tubes that were placed on ice immediately for determining fasting plasma betaine and homocysteine concentrations and a urine sample was collected. They were then given a methionine load of 0.1 g/kg body weight of L-methionine dissolved in 100 ml fruit juice. They also ate a standard snack consisting of a jam (jelly) sandwich to avoid hypoglycaemia. They took no further food or fluid until their return 4 hours later, when a second venous blood sample was collected. The betaine, dimethylglycine and homocysteine from the pre- and post- methionine load blood samples were measured in the same analytical batches.

Glucose, betaine and homocysteine

Betaine excretion positively correlated with glucose excretion (log transformed data; r = +0.47, p = 0.005, n = 34) in patients with diabetes. Dimethylglycine excretion also correlated with glucose excretion (log transformed data; r = +0.53, p = 0.001, n = 34). There was an inconclusive correlation between plasma glucose and log-transformed betaine excretion (r = +0.33, p = 0.056, n = 34); the Spearman’s correlation coefficient was significant (rs = +0.37, p = 0.03). Neither fasting homocysteine, nor the post-methionine load increase in homocysteine, correlated with either plasma glucose or glucose excretion. Hemoglobin A1c did not strongly correlate with betaine excretion, but there was a significant positive correlation (r = +0.52, p = 0.003, n = 30) between hemoglobin A1c and the post-methionine load change in plasma dimethylglycine (measured as the difference in log(plasma dimethylglycine) before and after the load). Hemoglobin A1c did not correlate with either fasting homocysteine or the post-methionine load increase in homocysteine.

Methods

Patients with type 2 diabetes (n = 34) and control subjects (n = 17) were recruited. We measured baseline fasting plasma and 4-hour post-methionine load (L-methionine, 0.1 mg/kg body weight) concentrations of homocysteine, betaine, and the betaine metabolite N,N-dimethylglycine. Baseline urine excretions of betaine, dimethylglycine and glucose were measured on morning urine samples as the ratio to urine creatinine. Statistical determinants of the post-methionine load increase in homocysteine were identified in multiple linear regression models.

Methionine load test study

The study was approved by the Canterbury Ethics Committee (New Zealand) and all subjects gave written informed consent. Patients with Type 2 diabetes were identified through the Canterbury retinal screening programme database and invited to participate. Controls were recruited by posting notices around the hospital campus. Thirty four patients (aged 51–75 yrs, 20 males) and 17 controls (aged 54–69 yrs, 10 male) were recruited. They were asked to fast for 16 hours and to withhold morning medications. Their height, weight, age, and medical history were recorded. Venous blood was drawn into EDTA tubes that were placed on ice immediately for determining fasting plasma betaine and homocysteine concentrations and a urine sample was collected. They were then given a methionine load of 0.1 g/kg body weight of L-methionine dissolved in 100 ml fruit juice. They also ate a standard snack consisting of a jam (jelly) sandwich to avoid hypoglycaemia. They took no further food or fluid until their return 4 hours later, when a second venous blood sample was collected. The betaine, dimethylglycine and homocysteine from the pre- and post- methionine load blood samples were measured in the same analytical batches.

Predictors of the post-methionine load increase in homocysteine

In the subjects with diabetes, the statistically significant (p < 0.05) predictors of the post-methionine load increase in plasma homocysteine, identified by best subset regression, were fasting plasma homocysteine, either fasting or post-methionine load betaine, and (in most models) serum vitamin B12. Variables that did not enter any of 36 models with p < 0.1 included age, body mass index, red blood cell folate, hemoglobin A1c, plasma glucose concentrations, plasma N,N-dimethylglycine concentration (either fasting or post-methionine load), and the urinary excretions of glucose, betaine and N,N-dimethylglycine. Log-transformed variables were used where the raw data was not normally distributed (Kolmogorov-Smirnov test). The three strongest candidate predictors (all log-transformed) were included in an explicit multiple linear regression model (Figure ) which confirmed that the fasting plasma homocysteine was the strongest (positive) predictor, and the post-methionine load plasma betaine concentration was also a highly significant (p < 0.001) negative predictor. The serum vitamin B12 concentration also a positive predictor (p = 0.021). The model explained most of the variance in the post-methionine load increase in homocysteine (r2 = 0.69). When the fasting plasma betaine concentration was included in the model instead of the post-methionine concentration, a similar result was obtained (r2 = 0.56) but the fasting betaine concentration appears to be a weaker negative predictor (p = 0.013).