GLP 1 safety review finds firm evidence only for gallbladder harm among seven feared risks

A new GLP 1 safety review in the Journal of Internal Medicine sorts seven feared side effects of the drug class by how strong the evidence really is. Its author, Peter Ueda of the Karolinska Institutet in Stockholm, concludes that only gallbladder and bile duct disease has consistent support for a causal link. For thyroid cancer, a rare optic nerve stroke, suicidal behaviour, pancreatitis and bowel obstruction, the answer is still uncertain, and for diabetic eye disease the pattern points to rapid blood sugar change.

Why a safety reckoning now

GLP 1 receptor agonists began as diabetes drugs and are now prescribed widely for weight loss and for heart and kidney protection. According to the News Medical summary of the review, around 5% of people in the UK and Finland had used one by 2025, and surveys from 99 countries suggest more than a quarter of adults aged 25 to 64 may qualify. Common side effects such as nausea and diarrhoea are well known. Rare but serious harms are much harder to pin down, because they occur too seldom for most trials to catch, are recorded inconsistently and need long follow up. As use spreads to younger, healthier and more varied groups, those rare events matter more in absolute numbers.

This GLP 1 safety review is a narrative appraisal, not a new study or a pooled analysis. Ueda looked at four kinds of evidence for each concern. Laboratory and animal studies can show whether a harm is biologically plausible but cannot show it happens in patients. Spontaneous reports to regulators can raise early alarms but lack a denominator and are skewed by publicity. Randomised trials remove most confounding but are usually too small, too short or too selective for rare outcomes. Large cohort studies cover more people for longer but can mislead when the comparison groups differ. He argues that a pile of studies does not settle a question when they define and record events differently, so the evidence has to be weighed together.

What the GLP 1 safety review concluded

Gallbladder and biliary disease stands apart. A meta analysis of 76 randomised trials cited in the review found a 37% relative increase in gallbladder or bile duct problems. Cohort studies agree, especially with higher doses, longer use and treatment for weight loss. Fast weight loss and direct effects on how the gallbladder empties offer a plausible mechanism, so the review treats this link as causal. Diabetic retinopathy is a different case. The worry dates to the SUSTAIN 6 trial, in which semaglutide was linked to a 76% relative rise in retinopathy complications, mostly in people with existing eye disease whose blood sugar fell fast. Most later studies show no rise, and the review reads this as early worsening after a quick drop in glucose.

A meta analysis of 76 randomised trials cited in the review found a 37% relative increase in gallbladder or bile duct problems.

For the other five concerns the verdict is uncertain. Thyroid cancer fears came from rodent studies, but human thyroid tissue carries fewer GLP 1 receptors, and two 2025 meta analyses disagreed, one finding a significant increase and one not. In the United States the drugs remain contraindicated for people with a personal or family history of medullary thyroid cancer or multiple endocrine neoplasia type 2. For non arteritic anterior ischaemic optic neuropathy, a sudden loss of blood flow to the optic nerve, the data are mostly observational and mixed, yet the European Medicines Agency listed it as a very rare side effect of semaglutide. Suicidal behaviour, pancreatitis and bowel obstruction showed no convincing increase in trials, though risk in vulnerable people cannot be ruled out.

Relative risk against absolute risk

The central message of the review is the difference between relative and absolute risk. Because the serious events examined are rare, even a large relative increase may add only a small number of extra cases among people at low baseline risk. The picture changes for patients who already have a history of the relevant problem, such as earlier optic nerve damage, pancreatitis, suicidal behaviour or symptomatic gallstones, where the same relative increase can mean many more events. The review illustrates this with hypothetical scenarios rather than treating uncertain links as proven. It also warns that safety fears can cause harm of their own, if patients stop or doctors withhold a drug with clear benefits because of a signal that may not be causal.

Regulators and specialists have taken different lines on the same evidence. According to the summary, the FDA and the EMA have both concluded that available data do not support a causal link with suicidal behaviour, while North American eye specialists stressed the limits of the optic nerve data and cautioned against stopping treatment that protects the heart and kidneys. In trials, discontinuation because of adverse events ranged from 4% to 16% on GLP 1 drugs against 3% to 8% on placebo. The review also notes a modest rise in resting heart rate and says other blood pressure and glucose lowering medicines may need adjusting as weight falls. Related studies in people are collected in our clinic section.

Where the evidence gaps remain

The weakest evidence concerns the people at highest risk. Trials have usually excluded patients with past pancreatitis, previous optic nerve damage or serious psychiatric illness, and trials designed for them may be hard to run for practical and ethical reasons. Self harm and suicidal thoughts are also poorly captured in both trials and health records, so reassuring population results may not apply to people with significant psychiatric risk. Placebo controlled trials are becoming rarer as more active drugs reach the market, which means future safety evidence for newer GLP 1 drugs will come increasingly from careful observational studies and indirect comparisons. That puts a premium on clear outcome definitions and on comparing like with like, for example against drugs that lower blood sugar by a similar amount.

For a reader trying to make sense of headlines, the review offers a practical frame. Ask whether a reported link comes from case reports, a cohort or a randomised trial, whether the comparison group was sensible, and whether the result is a relative or an absolute number. Only one of the seven concerns, gallbladder disease, passes those tests convincingly. That does not make the others imaginary, and it does not make the drugs risk free. It means the decision to start or continue treatment depends on each person's history and expected benefit, which is a conversation for a prescriber. The review abstract on Europe PMC lists all seven conclusions. Key numbers we track are on our figures page.

Conclusion

This GLP 1 safety review is a useful corrective to alarm in both directions. Gallbladder and bile duct disease is the one serious harm with consistent support, while retinopathy worsening seems tied to fast glucose change. Thyroid cancer, optic nerve damage, suicidal behaviour, pancreatitis and obstruction remain open questions that matter most for people with a history of those conditions. It is a single author narrative review, not new data.

Source Ueda P. GLP-1 receptor agonists and serious adverse events: Evidence, uncertainty and clinical implications. Journal of Internal Medicine, October 2026. DOI 10.1111/joim.70170

Amino Weekly reports research. It is not medical advice and never gives doses.