How does metformin work in the body?

The best-known and most well-established effect of metformin is inhibiting glucose production in the liver. Normally the liver continuously produces new glucose, especially in the fasted state. Metformin suppresses this process, causing blood sugar to fall. This is the primary blood-sugar-lowering effect in type 2 diabetes, demonstrated at the doses people actually take.

Exactly how metformin inhibits that hepatic glucose production is not yet fully understood. One of the mechanisms currently receiving the most attention: metformin disrupts the chemical balance inside liver cells, making the building blocks needed for new glucose less available. Metformin also blocks the action of glucagon, a hormone that normally signals the liver to produce glucose, by interrupting the signalling cascade within the liver cell.

The gut likely plays a larger role than was long assumed. Metformin accumulates in the intestinal wall at concentrations hundreds of times higher than those found in the blood. There it influences sugar processing in intestinal cells, increases the production of a hormone that promotes insulin release (GLP-1), and alters the composition of gut bacteria. This gut effect is increasingly regarded as an essential part of how metformin works.

Early research showed that metformin activates an energy sensor in the cell (AMPK), which leads, among other things, to less fat synthesis and more fat burning. However, this effect has only been demonstrated at doses far above clinically relevant blood concentrations. Whether it also occurs at ordinary patient doses is disputed. The improvement in fat metabolism seen in practice may partly occur through this mechanism, but the evidence for the gut effect and the inhibition of glucagon signalling is more solid.

Despite decades of use and research, the complete mechanism of action of metformin has still not been fully elucidated. Multiple independent review studies acknowledge this explicitly. An additional complicating factor is that studies use widely varying doses, meaning findings from the laboratory do not always correspond to what happens in a patient.