Metformin cuts rifampicin exposure — a real problem in India

Design

population pharmacokinetic modelling nested in a randomised controlled trial, intensive and semi-intensive sampling at week 5

Population

78 adults with HIV-associated tuberculosis and without diabetes (43 on metformin 500 mg twice daily), median weight 60.8 kg, 79.5% on antiretroviral therapy

Primary outcome

bioavailability and exposure of rifampicin, isoniazid and pyrazinamide with and without metformin

Effect

rifampicin bioavailability -24.0% (95% CI 7.83-36.3, p<0.007), AUC0-24 35.9 to 27.1 mg.h/L; isoniazid -15.6% (4.45-26.4, p<0.009); no effect on pyrazinamide

Metformin is being investigated as host-directed therapy for tuberculosis, and separately a very large number of people in India take metformin and anti-tuberculosis therapy at the same time for the ordinary reason that both diseases are common. This analysis nested in a randomised trial asked whether metformin changes the pharmacokinetics of first-line tuberculosis drugs, using intensive sampling at week 5 in 78 adults with HIV-associated tuberculosis and no diabetes, 43 of whom received metformin 500 mg twice daily.

Metformin reduced rifampicin bioavailability by 24.0% (95% CI 7.83-36.3, p<0.007) and isoniazid bioavailability by 15.6% (4.45-26.4, p<0.009). Mean rifampicin exposure over 24 hours fell from 35.9 to 27.1 mg.h/L and isoniazid from 18.2 to 15.6. Pyrazinamide was unaffected. Simulations suggested a single additional 150 mg of rifampicin would restore exposure.

Rifampicin exposure already sits at the lower end of what is effective at standard doses, and low exposure is associated with slower culture conversion and relapse. This is 78 participants, all with HIV co-infection and most on dolutegravir, so it is not a study of a person with type 2 diabetes and pulmonary tuberculosis. But the direction is the concerning one, the mechanism is plausible, and Indian practice puts these drugs together constantly. It is a reason to be alert to slow response rather than to stop metformin, and a reason to watch for a dose recommendation.

• Do not stop metformin because of this, but note the interaction when tuberculosis treatment responds slowly
• Consider therapeutic drug monitoring for rifampicin where available and response is poor
• Glycaemic control worsens on rifampicin anyway through enzyme induction — review agents at TB diagnosis
• Pyrazinamide was unaffected; the concern is rifampicin above all, and isoniazid to a lesser degree
• The cohort was HIV co-infected and non-diabetic — extrapolate to diabetic patients with caution

The statistics, in plain English

A 24.0% reduction with a confidence interval from 7.83% to 36.3% is clearly a real effect in direction but poorly pinned down in size — the true reduction could be modest or substantial. Bioavailability estimates come from a population pharmacokinetic model rather than direct measurement, so they depend on the model structure being right. With 43 metformin recipients the study cannot detect an interaction with acetylator status or with weight bands, and it measures drug concentrations rather than treatment outcomes: nobody was followed to culture conversion or relapse.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.