For millions of people living with schizophrenia, bipolar disorder, and other severe psychiatric conditions, the medications that quiet their symptoms often carry a hidden price. Second-generation antipsychotics, the mainstay of modern psychiatric treatment, are notorious for driving dramatic weight gain, pushing blood sugar upward, and derailing cholesterol levels. These metabolic side effects, collectively known as antipsychotic-induced metabolic disturbances, are among the leading reasons patients abandon treatment altogether, and they contribute to the stark two-decade gap in life expectancy that separates people with serious mental illness from the general population. Now, a large synthesis of randomized clinical evidence published in BMC Medicine offers the most detailed comparison yet of two candidate antidotes: the humble diabetes drug metformin and the newer, wildly popular class of GLP-1 receptor agonists, which includes semaglutide, the active ingredient behind blockbuster therapies for obesity and type 2 diabetes.
The research team, led by investigators at Dongzhimen Hospital and Beijing University of Chinese Medicine together with colleagues at Ningbo Municipal Hospital of Traditional Chinese Medicine and Tsinghua University Yuquan Hospital, conducted a systematic review and network meta-analysis, a statistical technique that allows researchers to compare multiple treatments simultaneously, even when some treatments have never been tested head-to-head in the same trial. The team searched four major biomedical databases, PubMed, Embase, the Cochrane Library, and Web of Science, for randomized controlled trials published through December 5, 2025. To be included, studies had to evaluate metformin or a GLP-1 receptor agonist for at least twelve weeks in patients continuing antipsychotic treatment, ensuring that any measured metabolic changes reflected an intervention layered on top of ongoing psychiatric care rather than a change in the psychiatric medication itself.
The search ultimately captured twenty-nine randomized controlled trials encompassing 1,721 patients. The typical participant was young, with a median study-level age of 36.2 years, and slightly less than half of the pooled population, 47.7 percent, was male. The median treatment duration across the trials was sixteen weeks, long enough to capture meaningful changes in body weight and glucose metabolism but short enough to leave questions about durability unanswered. The researchers assessed the risk of bias in each trial using the Cochrane Risk of Bias 2.0 tool, the current gold standard for judging whether randomized studies were designed and conducted in ways that protect against inflated or misleading results.
At the heart of the analysis was a random-effects frequentist network meta-analysis performed in Stata 17.0 MP, a framework that pools direct comparisons between treatments with indirect ones to estimate how any two interventions would compare even without direct trial evidence. To rank the competing treatments, the team calculated SUCRA values, the surface under the cumulative ranking curve, a metric that expresses the probability that a given treatment ranks best across outcomes. They also deployed univariate network meta-regression to probe whether study-level characteristics, such as baseline weight or antipsychotic type, modified treatment effects, and they graded confidence in the findings using the CINeMA framework, short for Confidence in Network Meta-Analysis, which evaluates how much trust each estimate deserves based on the structure and consistency of the evidence network.
The headline result concerns semaglutide, the GLP-1 receptor agonist that has transformed obesity medicine in recent years. Compared with control groups, semaglutide was associated with a reduction in body mass index of 3.55 kilograms per square meter, with a 95 percent confidence interval spanning from 4.27 down to 2.84, a magnitude that in clinical terms represents a substantial improvement for a patient population in which weight gain of twenty or thirty kilograms is not uncommon. The drug also produced meaningful reductions in waist circumference, averaging 6.34 centimeters, in glycated hemoglobin A1c, the three-month average of blood sugar, which fell by 0.44 percentage points, and in fasting blood glucose, which dropped by 0.53 millimoles per liter. Together, these figures paint semaglutide as a broad-acting agent against the weight and glycemic dimensions of antipsychotic metabolic harm.
Metformin, by contrast, showed a different signature. The decades-old biguanide, which works largely by suppressing glucose production in the liver and improving insulin sensitivity, was associated with reductions in lipid metabolism markers, specifically total cholesterol and triglycerides, the blood fats most closely tied to cardiovascular risk. Intriguingly, the analysis also flagged an exploratory potential benefit for psychiatric symptom scores, with a standardized mean difference of 0.31 in favor of metformin, a small-to-moderate effect that, if confirmed, would suggest the drug might ease the burden of psychotic symptoms themselves rather than merely offsetting the metabolic cost of their treatment. The authors were careful to frame this finding as preliminary, but it adds to a growing body of speculation about links between insulin signaling, inflammation, and brain function in psychiatric illness.
The divergence in outcome profiles between the two drug classes is arguably the study’s most clinically useful message. GLP-1 receptor agonists mimic the gut hormone glucagon-like peptide-1, slowing gastric emptying, enhancing satiety, and prompting insulin release in a glucose-dependent manner, which explains their strength on weight and glycemic endpoints. Metformin’s mechanism, centered on AMPK activation and hepatic glucose suppression, aligns more naturally with lipid effects. For clinicians, this suggests the choice of adjunctive therapy could be tailored to a patient’s dominant metabolic problem: a patient ballooning in weight on olanzapine or clozapine might be a stronger candidate for a GLP-1 agonist, while one with worsening triglycerides might do better on metformin, at least until direct comparative trials settle the question.
That caveat matters, and the authors emphasize it forcefully. Because most trials in the network compared each drug against placebo rather than against each other, the estimates linking semaglutide and metformin are indirect, and the transitivity assumption, the premise that the trials being connected are similar enough in their populations and designs to be meaningfully compared, remains uncertain. The team explicitly describes their conclusions as exploratory evidence intended to inform clinical discussion and future research, not as definitive treatment guidance. The CINeMA-based confidence ratings reflect this caution, and the fact that the article was shared early as a peer-reviewed, accepted manuscript subject to further editorial processing adds another layer of provisional status to the findings.
Even with those limitations, the stakes of the research are difficult to overstate. People with schizophrenia die on average fifteen to twenty years earlier than the general population, and cardiovascular disease driven by antipsychotic-associated metabolic dysfunction is a leading contributor. Adherence is another casualty: patients who gain alarming amounts of weight frequently stop taking the very medications that keep their psychosis at bay, triggering relapse, hospitalization, and worsening long-term outcomes. An effective, well-tolerated metabolic countermeasure could therefore ripple across the entire trajectory of severe mental illness, improving not just waistlines and lab values but treatment persistence, relapse rates, and quality of life.
The study also arrives at a moment of intense public fascination with GLP-1 drugs, which have been hailed as near-miraculous for obesity and are being tested for everything from addiction to neurodegeneration. Applying them to antipsychotic-induced metabolic disturbances is a natural extension, but questions remain about cost, access, tolerability in psychiatric populations, and the theoretical concern that GLP-1 agents’ appetite suppression could interact with the weight dynamics of patients on antipsychotics. The authors’ funding, drawn from Chinese national and institutional sources including traditional Chinese medicine endocrinology programs, underscores the global nature of the effort to solve this problem. What the analysis delivers is a rigorous map of the current randomized evidence and a clear agenda: head-to-head trials directly comparing metformin and GLP-1 receptor agonists, longer follow-up to test durability, and studies powered to confirm whether metformin’s apparent psychiatric benefit is real. For the millions of patients whose lifesaving medications come with a metabolic tax, that research cannot come soon enough.
Subject of Research: Comparing metformin and GLP-1 receptor agonists for treating antipsychotic-induced metabolic disturbances
Article Title: Metformin and GLP-1 receptor agonists for antipsychotic-induced metabolic disturbances: a systematic review and network meta-analysis
Article References: Chen, Y.-X., Yang, Q.-W., Lin, M.-X., Sun, M.-H., Dong, Y.-Y., Yu, R.-D., Mao, D.-D., Zhao, Y., Zhang, L., Zhao, J.-X., Zhang, Y.-F., & Xu, J. (2026). Metformin and GLP-1 receptor agonists for antipsychotic-induced metabolic disturbances: a systematic review and network meta-analysis. BMC Medicine. https://doi.org/10.1186/s12916-026-05280-2
Image Credits: AI Generated
DOI: 10.1186/s12916-026-05280-2
Keywords: metformin, GLP-1 receptor agonists, semaglutide, antipsychotics, metabolic side effects, network meta-analysis, schizophrenia, obesity, BMI, HbA1c, psychiatry, BMC Medicine
Cite Scienmag News
APA MLA Chicago
Copy citation Download RIS
Tags: and metabolicantipsychoticsbipolar disorderBMC MedicineBMIcomparative effectiveness of diabetes medications for mental health patientsDiabetes drugs for antipsychotic-induced weight gainGLP-1 receptor agonistsHbA1cimpact of antipsychotic medications on blood sugar and cholesterolmanaging metabolic side effects of antipsychoticsmetabolic disturbances in psychiatric treatmentmetabolic side effectsMetforminMetformin vs GLP-1 receptor agonistsnetwork meta-analysisobesityobesity treatment with GLP-1 receptor agonistspsychiatryschizophreniasecond-generation antipsychotics and weight gainsemaglutidesystematic review and network meta-analysis of diabetes drugs
