Introduction

The number of patients with type 2 diabetes (T2DM) is progressively increasing all over the world and the health-care costs of them are increasing [1]. One of the earliest changes in the development of T2DM is postprandial hyperglycemia. Postprandial hyperglycemia plays a major role in cardiovascular complications in patients with T2DM [2, 3]. Therefore, targeting the postprandial hyperglycemia is an important glucose-lowering strategy in patients with T2DM. One of the potential approaches is to target the incretin hormone, glucagon-like peptide-1 (GLP-1). Indeed, it has been reported that incretin-based therapies, such as GLP-1 receptor agonists [4] and dipeptidyl peptidase 4 (DPP-4) inhibitor [5], represses postprandial hyperglycemia in patients with T2DM.

Metformin, which has been considered as an insulin sensitizer [6], is a common first-line agent in managing hyperglycemia in patients with T2DM [7] and is taken with a meal (Sumitomo Dainippon Pharma Co., Ltd. METGULCO interview form [online]. [Accessed 25 May 2015]). Previous studies have revealed that metformin increased serum GLP-1 levels [8, 9]. However, it has been reported that compared with taking metformin without a meal, taking metformin with a meal decreased bioavailability of metformin [10]. Therefore, we hypothesized that taking metformin before a meal would have a favorable effect on postprandial hyperglycemia through increased serum GLP-1. Thus, we researched the effect of taking metformin before a meal on postprandial hyperglycemia both in an animal model and in patients with T2DM.

Materials and methods

Study design for animals

All experimental procedures were approved by the Committee for Animal Research, Kyoto Prefectural University of Medicine. Eighteen 9-week-old male Zucker fatty rats with jugular vein cannulation were purchased from Japan SLC (Kyoto, Japan).

All rats were fasted overnight and were divided into three groups as follows, (1) No medication (Control), 3 ml of saline was orally administrated 10 min before meal and a polymeric diet (Ensure® H, 375 kcal/250 mL, 1.5 kcal/mL; proteins: 15 % of total kcal, carbohydrates: 57 % of total kcal, fat: 28 % of total kcal; Abbott Japan, Tokyo, Japan) of 10 mL/kg was orally administrated with 3 mL of saline; (2) Metformin with meal (Met), 3 mL of saline was orally administered 10 min before meal and 3 mL of saline with metformin (metformin concentration was 600 mg/kg) and a polymeric diet of 10 mL/kg were orally administrated; and (3) Metformin 10 min before meal (pre-Met), 3 mL of saline with metformin (metformin concentration was 600 mg/kg) was orally administrated 10 min before meal and a polymeric diet of 10 mL/kg was orally administrated with 3 mL of saline.

Blood samples (t = 0, 20, 60, and 100 min) were obtained from the jugular vein. Blood glucose levels were measured using a glucometer (GLUTEST ACE, Sanwa Kagaku Kenkyusho, Nagoya, Japan). Serum insulin levels were measured using an insulin enzyme immunoassay system, the Morinaga Rat Insulin Assay Kit (Morinaga Institute of Biological Science, Kanagawa, Japan). Plasma total GLP-1 levels were measured using a GLP-1 enzyme immunoassay system, YK160 GLP1 EIA kit (Yanaihara Institute, Sizuoka, Japan).

Study design for patients with diabetes

We recruited five patients with T2DM who were being treated with metformin monotherapy and whose HbA1c were under 6.0 % from Hospital of Kyoto Prefectural University of Medicine for this randomized, open-label crossover pilot study. The research protocol was approved by the institutional review board of the Hospital of Kyoto Prefectural University of Medicine. The study was registered with UMIN (UMIN000015055) and all participants gave their written informed consent.

After a 1-week metformin washout period, subjects were randomized using a coin flip. Each subject consumed two polymeric diets separated by a washout period of at least 1 week. At the first visit, subjects were randomized to take metformin (1000 mg) either (1) 30 min before their meal (pre-Met protocol) or (2) with their meal (Met protocol). (1) For pre-Met protocol, a fasting blood sample (t = −30 min) was taken and then metformin was administrated. A blood samples were again taken at t = 0 min and immediately after subjects were administrated a polymeric diet (Ensure® H, 375 kcal/250 mL, 1.5 kcal/mL; proteins: 15 % of total kcal, carbohydrates: 57 % of total kcal, fat: 28 % of total kcal; Abbott Japan, Tokyo, Japan). (2) For Met protocol, fasting blood samples (t = −30, 0 min) were taken and then subjects were administrated metformin and were immediately administrated a polymeric diet. For both groups, blood samples were drawn thereafter, at t = 15, 30, 45, 60, 90, and 120 min, and plasma glucose, plasma total GLP-1 and serum insulin were measured. In addition, 13C-acetate breath tests were also performed for evaluation of gastric emptying [11]. Breath sample was collected from each subject before taking a polymeric diet and then 100 mg of 13C-acetate sodium (Otuka Phar, Tokyo, Japan) was administrated with a polymeric diet. Then breath samples were taken at t = 15, 30, 40, 50, 60, 70, 80, 90, and 120 min. 13C was measured as the 13CO2/12CO2 isotope ratio and was expressed as Δ over baseline per milliliter (‰). At the second visit, each subject was crossed over to the opposite protocol. Plasma total GLP-1 levels were measured using a GLP-1 enzyme immunoassay system, YK160 GLP1 EIA kit (Yanaihara Institute, Sizuoka, Japan).

Statistical analysis

The primary endpoint of the study was the peak plasma concentration of glucose. Secondary endpoint was concentration of serum insulin or plasma total GLP-1, and the 60-min AUC of concentration of glucose, insulin and total GLP-1 in the animal model and the 60 or 120-min AUCs of concentration of glucose, insulin and total GLP-1 in patients with T2DM. Exploratory analyses also included AUCs of other time intervals and peak time of 13CO2/12CO2 isotope ratio (t max).

Continuous variables were presented as the means ± standard deviations (SDs). In the animal model, Kruskal–Wallis analysis was performed to compare the three groups and P values <0.05 were considered statistically significant. Then, to correct for a family wise error, we performed a Bonferroni correction and P values <0.017 were considered statistically significant. In patients with T2DM, paired t test was performed to evaluate the two protocols and P values <0.05 were considered statistically significant. The statistical analyses were performed using the JMP version 10.0 software (SAS Institute Inc., Cary, North Carolina).

Results

Animal model

The results of glucose, insulin and total GLP-1 levels are shown in Fig. 1 and Table 1. Glucose levels, especially the glucose level at t = 20 min, area under curve (AUC)0–60 min and AUC0–100 min, of pre-Met group were significantly lower than those of the others. Insulin level at t = 20 min of pre-Met group was higher than that of Control group. Although there was no statistical difference in GLP-1 among the three groups, the total GLP-1 level at t = 0 min of pre-Met group (7.4 ± 2.7 pmol/L) tended to be higher than Control group (3.7 ± 2.0 pmol/L, P = 0.030) or Met group (3.9 ± 1.2 pmol/L, P = 0.020).

Fig. 1
figure 1

Result of polymeric diet test of Control, Met or pre-Met in an animal model. Control no medication, Met taking metformin (600 mg/kg) with a meal, pre-Met taking metformin (600 mg/kg) 10 min before a meal, GLP-1 glucagon-like peptide-1. a glucose, b insulin, c total GLP. The short-dashed line represented Control. The long-dashed line represents Met. The black solid line represents pre-Met. *P < 0.017 pre-Met versus Control. † P < 0.017 pre-Met versus Control

Full size image
Table 1 AUC of the variables in animal experiments among three groups
Full size table

Patients with type 2 diabetes

The results of glucose, insulin and total GLP-1 levels are shown in Fig. 2 and Table 2. Because the timings of peak glucose levels differed among the subjects, we compared the peak glucose levels. The peak glucose level of pre-Met protocol (7.0 ± 0.4 pmol/L) was lower than that of Met protocol (8.5 ± 0.9 pmol/L, P = 0.021). Glucose levels, including glucose level at t = 90 min, the AUC0–60 min and AUC0–120 min were lower in pre-Met protocol (30 min prior to meal) than those in Met protocol. Total GLP-1 levels, especially the total GLP-1 level at t = 30 min, AUC0–60 min and AUC0–120 min, were higher in pre-Met protocol than those in Met protocol. In 13C-acetate breath tests, the t max of pre-Met protocol (102.5 ± 17.9 min) was significantly longer than that of Met protocol (77.5 ± 9.0 min, P = 0.043) (Fig. 3).

Fig. 2
figure 2

Result of polymeric diet test of Met protocol or pre-Met protocol in patients with diabetes. Met protocol taking metformin (1000 mg) with a meal, pre-Met protocol taking metformin (1000 mg) 30 min before a meal, GLP-1 glucagon-like peptide-1. a glucose, b insulin, c total GLP. The dashed line represents Met protocol. The black solid line represents pre-Met protocol. *P < 0.05 pre-Met versus Met

Full size image
Table 2 AUC of the variables for the different time intervals in diabetic subjects
Full size table
Fig. 3
figure 3

Result of 13C-acetate breath test of Met protocol or pre-Met protocol in patients with diabetes. Met protocol taking metformin with a meal, pre-Met protocol taking metformin 30 min before a meal; t max, peak time of 13CO2/12CO2 isotope ratio. The t max of pre-Met protocol (102.5 ± 17.9 min) was significantly longer than that of Met protocol (77.5 ± 9.0 min, P = 0.043)

Full size image

Discussion

In the present study, we showed that taking metformin before a meal ameliorated the peak and incremental area for post-meal glucose both in the animal model and in the patients with T2DM. This glucose-lowering effect of taking metformin before a meal was associated with enhanced early prandial GLP-1 both in the animal model and in the patients with T2DM.

Previous researches provide important clues for an explanation of these results. Metformin increases GLP-1 levels [8, 9]. On the other hand, taking metformin with a meal decreased bioavailability of metformin compared with taking it without a meal [10]. In fact, we showed that taking metformin before a meal enhanced early prandial GLP-1. Interestingly, there was no difference in the insulin secretion between pre-Met and Met groups and between the two protocols. One of the reasons might be that taking metformin before a meal reduces the quantity of insulin needed. Earlier total GLP-1 secretion might enable the subjects to secrete insulin earlier. Another possible reason might be the effects on gastric emptying. A previous study showed that the slowing of gastric emptying by GLP-1 receptor agonist decreased the quantity of insulin needed [12]. Moreover, metformin inhibits secretion of ghrelin [13], which stimulates gastric motility [14]. In addition, it has been reported that slowing of gastric emptying ameliorated the postprandial hyperglycemia [15]. Thus, the required insulin might be lower when metformin is taken before a meal. Taking these finding together, the glucose levels of taking metformin before a meal were significantly lower than those of the others in both animal and patents with T2DM.

It has been reported that taking metformin with a meal is useful to avoid gastrointestinal side-effects [16]. We showed that taking metformin before a meal increased the early secretion of GLP-1. Thus, there is a possibility that entero-hormones’ release could be the mechanism underlying gastrointestinal side-effect [17, 18]. In fact, gastrointestinal side-effect is one of the common side-effects of GLP-1 agonist [19, 20].

Several limitations should be noticed. First, this was a single administration study. Thus, a further study using repetitive administration should be needed. Second, HbA1c of patients who participated in this study was under 6.0 %. Thus, the effect of taking metformin before the meal for patients with insufficient glucose control remains unclear. Third, we did not measure the levels of metformin. Thus, it is unclear whether the bioavailability of pre-Met and that of Met was different. Fourth, subjects took metformin 30 min before a meal in the pre-Met protocol in this study; however, the best timing of taking metformin is still unclear.

In conclusion, this is the first study to show that taking metformin 30 min before a meal ameliorated the postprandial hyperglycemia. From the viewpoint of both cost and therapeutic effectiveness, taking metformin 30 min before a meal would represent a novel approach for enhancing glucose-lowering strategies in patients with T2DM.