An investigation of metformin's etiological role via AMPK-dependent pathways in major chronic diseases using Mendelian randomization

Abstract

The rapidly increasing global diabetes epidemic has several important implications for cardiovascular disease and cancer occurrence and prevention. Metformin is the first-line nontoxic medication for the treatment for type 2 diabetes mellitus and on the World Health Organization essential medication list. Emerging evidence from basic science, observational studies and short-term randomized controlled trials suggests that metformin may be protective against cardiovascular disease and cancer, unlike other anti-hyperglycaemia agents (such as insulin and sulfonylureas). Repositioning metformin for use beyond traditional glycaemic control could be a viable option in promoting global health. However, the aetiology and mechanisms by which metformin operate are not fully understood. Furthermore, previous observational studies are often open to both confounding and selection bias. As such, whether metformin’s benefits are genuine is unclear.

Mendelian randomization utilizes genetic variants, randomly allocated during the conception, as instruments to infer causality and is less prone to confounding than conventional observational studies. I used Mendelian randomization to explore the etiological role of metformin in major chronic diseases using individual-level data in the UK Biobank and publicly available summary-level data from consortia, if applicable. I similarly evaluated possible non-glycaemic pathways by which metformin may confer its effect on these major chronic diseases, within the context of well-established scientific theories, specifically the evolutionary biology theory that growth and reproduction trade-off against longevity, implying a key role their drivers, sex hormones, sex hormone-binding globulin and insulin-like growth factor 1, in health.

The effect of metformin was instrumented by adenosine monophosphate-activated protein (AMPK) variants, one of the key targets of metformin, where AMPK variants were selected upon based on their associations with glycated haemoglobin. Compared with the reference group, participants in the metformin use group (AMPK score above the median) had a lower risk of coronary artery disease and possibly overall cancer. Those with lower glycated haemoglobin instrumented by AMPK variants also had increased overall sex hormone-binding globulin level but decreased insulin-like growth factor 1 level, decreased serum total testosterone level in women but increased serum total testosterone level in men, suggesting a possible steroidal pathway for metformin’s action. Genetically predicted higher serum total testosterone was associated with increased risks of ischemic stroke, thromboembolism and heart failure, and probably coronary artery disease and myocardial infarction, especially in men. Given testosterone stimulates erythropoiesis and alters massive red blood cell indices, I also assessed the role of blood cell traits using multivariable Mendelian randomization Bayesian mode averaging, where the analyses suggested higher haemoglobin concentration is the most relevant factor for the increased venous thromboembolism risk.

This study provides novel insight that AMPK-dependent effects of metformin may operate via glycaemic and non-glycaemic pathways. Specifically, hormone-related mechanisms may have an essential role in the health effects of metformin. These findings provide potentially more credible evidence in support of the use of metformin as an adjunction treatment.