Metformin, UDCA no better than placebo in treating long COVID symptoms

Treatment with either metformin or ursodeoxycholic acid (UDCA) falls short of improving recovery from postacute sequalae of SARS-CoV-2 infection (PASC), a study has shown.

“[T]his randomized clinical trial did not show a statistically significant improvement in recovery with metformin or UDCA among patients with long COVID,” the researchers said.

A total of 666 adults were screened for this double-blind, placebo-controlled trial conducted at two tertiary hospitals in South Korea from July 2024 to April 2025. Of these, 396 (median age 36 years, 72 percent women) with a PASC index score of ≥12 were randomized 1:1:1 to oral metformin (n=132; uptitrated to 1,500 mg/day), UDCA (n=132; 900 mg once daily), or double placebo (n=132) for 14 days.

Patients with long COVID had a mean interval from SARS-CoV-2 infection of 9.8 months and a mean baseline PASC score of 19.3. Among the participants, 84 (63.6 percent) in the metformin arm, 90 (68.2 percent) in the UDCA arm, and 90 (68.2 percent) in the placebo arm recovered from their symptoms. [Ann Intern Med 2026;doi:10.7326/ANNALS-25-04883]

The mean changes in PASC scores from baseline to week 8 were ‒10.05 (95 percent confidence interval [CI], ‒11.35 to ‒8.76) with metformin and ‒10.62 (95 percent CI, ‒11.79 to ‒9.45) with UDCA relative to ‒10.43 (95 percent CI, ‒11.69 to ‒9.18) with placebo.

“[B]oth the intervention and control groups showed improvement in PASC index scores over the 8-week period, and these improvements were significantly associated with decreased levels of various cytokines between baseline and 8 weeks,” the researchers said.

“Although this trial did not find evidence supporting the efficacy of repurposed metformin or UDCA for long COVID, we established the potential utility of PASC index scoring for long COVID interventional trials and highlighted the need for further studies to support the development of targeted therapies addressing immune dysregulation in long COVID pathophysiology,” they added.

Immune dysregulation

Furthermore, the associations between changes in cytokine levels and improvements in long COVID symptoms validate the role of immune dysregulation in the pathophysiology of long COVID. A potential reason for the lack of observed benefit in this study is that metformin or UDCA appears to elicit no significant influence on the dysregulated immune system, including monocytes.

“Therefore, more targeted immunomodulatory drugs, such as baricitinib or monoclonal antibodies targeting pathogenic monocytes, are warranted in this context,” the researchers said.

Another likely reason for the lack of benefit is that UDCA or metformin could not reverse a more prolonged PASC. In exploratory analyses, cytokine change patterns appeared to be more pronounced in patients enrolled earlier (within 2‒6 months) after COVID-19, according to the researchers.

“Further in-depth immune phenotyping is needed to identify biomarkers and therapeutic targets addressing the heterogeneity of long COVID,” they said.

Currently, validated tools or biomarkers to define long COVID or to evaluate clinical outcomes in interventional trials targeting this condition do not exist yet. The PASC index score only examines the severity of long COVID symptoms that occurred during the period of postacute COVID and does not include the concept of symptom duration, according to the researchers.