Metformin to reduce metabolic complications and inflammation in patients on systemic glucocorticoid therapy: a randomised, double-blind, placebo-controlled, proof-of-concept, phase 2 trial

Background An urgent need to reduce the metabolic side-effects of glucocorticoid overexposure has been recognised, as glucocorticoid excess can lead to Cushing's syndrome, which is associated with high morbidity. We aimed to evaluate the potential of metformin to reverse such effects while sparing the anti-inflammatory benefits of glucocorticoids. Methods We did a randomised, double-blind, placebo-controlled, proof-of-concept, phase 2 trial involving four hospitals in the UK. Patients without diabetes were eligible if they were between the ages of 18 and 75 years with an inflammatory disease treated with continuous prednisolone (≥20 mg/day for ≥4 weeks and remaining on ≥10 mg/day for the subsequent 12 weeks, or its cumulative dose-equivalent). Eligible patients were randomly allocated (1:1) to either the metformin or placebo groups, using a computer-generated randomisation table stratified according to age and BMI. Metformin and placebo were administered orally for 12 weeks in escalating doses: 850 mg/day for the first 5 days, 850 mg twice a day for the next 5 days, and 850 mg three times a day subsequently. The primary outcome was the between-group difference in visceral-to-subcutaneous fat area ratio over 12 weeks, assessed by CT. Secondary outcomes included changes in metabolic, bone, cardiovascular, and inflammatory parameters over 12 weeks. Our analysis followed a modified intention-to-treat principle for the primary outcome. This study is registered with ClinicalTrials.gov, NCT01319994. Findings Between July 17, 2012, and Jan 14, 2014, 849 patients were assessed for study eligibility, of which 53 were randomly assigned to receive either metformin (n=26) or placebo (n=27) for 12 weeks. 19 patients in the metformin group and 21 in the placebo group were eligible for the primary outcome analysis. Both groups received an equivalent cumulative dose of glucocorticoids (1860 mg prednisolone-equivalent [IQR 1060–2810] in the metformin group vs 1770 mg [1020–2356] in the placebo group); p=0·76). No change in the visceral-to-subcutaneous fat area ratio between the treatment groups was observed (0·11, 95% CI −0·02 to 0·24; p=0·09), but patients in the metformin group lost truncal subcutaneous fat compared with the placebo group (−3835 mm2, 95% CI −6781 to −888; p=0·01). Improvements in markers of carbohydrate, lipid, liver, and bone metabolism were observed in the metformin group compared with the placebo group. Additionally, those in the metformin group had improved fibrinolysis, carotid intima–media thickness, inflammatory parameters, and clinical markers of disease activity. The frequency of pneumonia (one event in the metformin group vs seven in the placebo group; p=0·01), overall rate of moderate-to-severe infections (two vs 11; p=0·001), and all-cause hospital admissions due to adverse events (one vs nine; p=0·001) were lower in the metformin group than in the placebo group. Patients in the metformin group had more events of diarrhoea than the placebo group (18 events vs eight; p=0·01). Interpretation No significant changes in the visceral-to-subcutaneous fat area ratio between the treatment groups were observed; however, metformin administration did improve some of the metabolic profile and clinical outcomes for glucocorticoid-treated patients with inflammatory disease, which warrants further investigation. Funding Barts Charity and Merck Serono.