Metformin for Glucocorticoid-Induced Toxicity: Why Not? Free

Since their discovery in the 1940s, glucocorticoids have become an essential tool for the treatment of acute and chronic diseases (1). In the acute setting, glucocorticoids are lifesaving, as in the cases of adrenal insufficiency and severe respiratory infections (2). Glucocorticoids also improve morbidity and quality of life in people living with chronic inflammatory diseases, and they are mainstay treatments for organ transplantation and cancer. As a result, approximately 1%–3% of the global population receives glucocorticoid prescriptions annually (3). Despite steroid-sparing treatments for chronic diseases, glucocorticoid prescriptions have increased in the last several decades (4) and are likely to further increase following recent updates for the management of critical illness and the prevalence of their use during the coronavirus disease 2019 pandemic (2,5).

Unfortunately, both short- and long-term glucocorticoid treatment can result in adverse effects, including glucocorticoid-induced hyperglycemia. Meta-analyses suggest that >1 month of glucocorticoid treatment associates with hyperglycemia in 32% of people without preexisting diabetes (6). Moreover, glucocorticoids account for as many as 2% of incident cases of diabetes in the general population (7). Yet, few guidelines exist to guide detection or prevention of glucocorticoid-induced hyperglycemia outside of the hospital setting (8), and data to inform glucocorticoid-induced hyperglycemia treatment in people living with diabetes who are not treated with insulin are limited (9). Given the scope of the problem, the paucity of evidence to prevent glucocorticoid-induced hyperglycemia is striking.

To that end, reported in this issue of Diabetes Care, Thierry et al. (10) conducted a placebo-controlled, crossover design study to examine the role of metformin for the prevention of glucocorticoid-induced hyperglycemia and explore its potential to modify risk of other glucocorticoid-induced adverse effects. The study comprised two 7-day study periods separated by a 28-day washout period. During each study period, 18 healthy lean male participants (mean age 27 years and BMI 23 kg/m²) received 30 mg prednisone daily with escalating doses of metformin (maximum dose 2,000 mg) or matched placebo. Each participant received both placebo and metformin, but the treatment sequence (metformin + prednisone → placebo + prednisone vs. placebo + prednisone → metformin + prednisone) was randomized. Before and after each study period, participants underwent a series of tests, including a mixed-meal tolerance test (MMTT) and blood metabolomics. A subset of participants also underwent adipose tissue biopsy for RNA sequencing.

Following prednisone treatment, glucose area under the curve (AUC) during the MMTT increased significantly with placebo but did not change with metformin. Close evaluation of glucose curves demonstrates that the maximum glucose excursion (30 min following mixed meal) was higher with prednisone treatment (∼7.0 mmol/L [126 mg/dL]) than prior to prednisone exposure (6.5 mmol/L [117 mg/dL]). Metformin treatment reduced the peak glucose level to 6.0 mmol/L (108 mg/dL), but glucose was not different from that with placebo at 120 min (∼5 mmol/L [90 mg/dL]). These data are consistent with data from prior studies in individuals without diabetes (11,12). Whether these changes are clinically meaningful is debatable, especially since glucose tolerance remains well within normal limits. It seems unlikely that additional treatment would be warranted in people who are at low risk for hyperglycemia.

The authors next investigated the mechanism by which metformin improves glycemia. In this study, insulin AUC during MMTT increased with prednisone + placebo in comparison to pre–glucocorticoid exposure. Conversely, insulin AUC decreased with metformin + prednisone, resulting in an overall increase in the Matsuda index, a commonly used measure that highly correlates with whole-body insulin sensitivity (13) (Fig. 1). The impact of metformin on insulin sensitivity has been suggested in prior studies (14,15). In a study of individuals with prediabetes or type 2 diabetes, metformin significantly increased Matsuda index when paired with exercise, but only a trend toward an increase was observed with metformin alone (14). While metformin may require potentiation to achieve a significant change in insulin sensitivity in people who already have metabolic disturbances, metformin may be enough to overcome a metabolic challenge in people at low risk for diabetes. Whether these data can be applied to individuals at high risk for glucocorticoid-induced metabolic dysfunction therefore remains a question. Interestingly, a sequencing effect for the Matsuda index was observed, such that metformin had a greater impact on insulin sensitivity in those who were treated with steroids + placebo first. Glucose curves were not stratified by treatment sequence, but this finding could indicate that metformin’s impact on insulin sensitivity may be enhanced in individuals without diabetes who require multiple courses of glucocorticoids.

Glucagon-like peptide 1 (GLP-1) AUC significantly increased with metformin + prednisone but not with placebo + prednisone. Previous studies demonstrated that in individuals with type 2 diabetes, metformin-induced increases in GLP-1 stimulate insulin secretion (16). Further, GLP-1 receptor blockade abrogates metformin-induced insulin secretion and reduces metformin-induced improvements in postprandial glucose excursions by 75% (16). As described above, no increases in insulin secretion were observed despite a significant increase in GLP-1, suggesting that the overall metabolic improvements induced by metformin may not be GLP-1 mediated. However, infusion of exenatide, a GLP-1 receptor agonist, during MMTT in healthy individuals treated with a 2-day course of prednisolone prevented glucocorticoid-induced insulin intolerance and reduced insulin secretion (17). Together, these data could be interpreted multiple ways: 1) they highlight differences in response between people with and without diabetes or 2) they suggest that GLP-1 may have non–insulin-mediated mechanisms for improving metabolic parameters in the setting of glucocorticoids. Parsing out the metformin-induced impact on glucocorticoid-induced hyperglycemia from the GLP-1–mediated effects is an important future direction for research, as it is conceivable that GLP-1 receptor agonists may be a more effective therapy for treating glucocorticoid-induced hyperglycemia. Studies in people who are at high risk of glucocorticoid-induced hyperglycemia should be prioritized, as response may differ in people who have metabolic dysfunction.

The remainder of the investigations explore the possible mechanisms by which metformin may improve other glucocorticoid-induced toxicity. Bone turnover markers, lipids, weight, and blood pressure were not different between the groups, which, given this short course of steroids in healthy males, is not surprising. Metabolomics and RNA sequencing identified changes in pathways associated with free fatty acids, bile acids, and the urea cycle in comparisons with placebo, and the authors suggest plausible mechanisms by which these changes could counteract glucocorticoid-induced myopathy, dyslipidemia, and obesity. However, given the absence of evidence to support that metformin impacts these outcomes, consideration of the mechanism is premature and remains speculative. Regardless, the authors must be commended on a thoughtful study that brings an area of continued and significant unmet need to the forefront.

The potential impact of glucocorticoids on glycemia is too often overlooked, particularly in individuals at high risk. Factors associated with high risk of glucocorticoid-induced hyperglycemia are established, including prior history of impaired glucose tolerance, family history of diabetes, and type 2 polygenic risk score (1,18). Clinical trials that evaluate methods for preventing glucocorticoid-induced hyperglycemia are required to inform evidence-based guidelines. In the meantime, should metformin be prescribed to people receiving glucocorticoids? Without data in people who are at high risk, recommendation for metformin use with steroids in the general population is a hard sell. Nevertheless, there is no doubt that evidence suggests the potential for benefit. In certain individuals, with the understanding that the data are limited, in the absence of clear harm, and with the potential for benefit for other glucocorticoid-mediated toxicities, the real question is, why not?