Endocrine Abstracts

Introduction: Despite a plethora of suggested targets and pathways, the mechanism of anti-hyperglycaemic metformin action is still unknown. The present study critically re-analysed protocols broadly applied in preclinical rodent studies.

Methods: Obese male C57BL/6J mice on high fat diet were treated with metformin in the form of a single dose, daily intraperitoneal injections, admixture to drinking water, or continuous infusion via intraperitoneal minipumps. Glucose tolerance tests (GTT) were performed to evaluate effects on blood glucose homeostasis.

Results: Thirty min after intraperitoneal injection of 50mg/kg metformin, the plasma concentration of metformin was 56±22μmol/l and glucose tolerance was improved (AUC, min*g/dl: 40.4±1.8 vs 34.0±1.2, P=0.008). The beneficial effect on glucose tolerance was gone 3 h after drug administration, when plasma metformin was down to 2.8±0.7µmol/l (AUC, min*g/dl: 39.0±1.4 vs 36.4±1.7, n.s.). Rapid clearance of metformin accompanied by fading of its action suggests that mice under typical long-term treatment regimens (daily dosing; admixture to food or water) are not permanently exposed to effective drug concentrations. In mice under long-term treatment, we therefore found effects on glucose tolerance to depend on the time span between the preceding metformin dose and measurement of blood glucose. Under regular treatment, however, metformin also affected glucose tolerance indirectly via reduced appetite and blunted weight gain (g gained in 7 weeks on metformin 50 mg/kg/d: +6.7±1.0 vs +3.1±0.5, P=0.005). When dosing shortly before the GTT was avoided and weight-mediated actions were eliminated by restricted feeding of the control mice, counterintuitive worsening of glucose tolerance by metformin was unmasked (after 6 weeks with 4.1g/kg metformin in drinking water: AUC, min*g/dl: 32.2±0.8 vs 38.5±1.2, P=0.0003). In mice continuously infused with the drug (13mg/kg/d), plasma metformin was intra-individually associated with less weight gain (r=0.81) and lower blood glucose (r=0.78).

Conclusions: Our results suggest three superimposed components of metformin action on blood glucose in mice: (i) Glucose lowering shortly after dosing, which fades rapidly with the clearance of plasma metformin - we hypothesise that this component accounts for antidiabetic action in the clinic. (ii) Indirect action via reduced appetite and weight gain - we suspect that this component is not sufficiently considered in many rodent studies. (iii) Deterioration of glucose homeostasis by prolonged treatment - this component is unmasked by exclusion of weight-mediated effects in combination with avoidance of dosing shortly before glucose measurements. Dependent on the specific experimental protocol, the net effect of these components can be decreased, increased, or unchanged blood glucose.