Semaglutide Long-Term Safety: What 5+ Years Show
Semaglutide Safety
17,604 patients
The SELECT trial followed patients on semaglutide for a mean of 39.8 months, finding a 20% reduction in major cardiovascular events alongside a 16.6% permanent discontinuation rate.
Lincoff et al., NEJM, 2023
Lincoff et al., NEJM, 2023
View as imageSemaglutide is the most prescribed GLP-1 receptor agonist in the world, approved for type 2 diabetes (Ozempic), weight management (Wegovy), and cardiovascular risk reduction. With tens of millions of people now taking this peptide drug, the question of semaglutide long-term safety has become one of the most consequential in modern medicine. The evidence base spans more than 30,000 patients across randomized controlled trials lasting up to four years, plus growing real-world pharmacovigilance data. This article examines what that evidence actually shows across every major safety concern, from gastrointestinal effects to cancer risk to cardiovascular outcomes. For a broader view of GLP-1 agonist adverse events beyond semaglutide specifically, see the dedicated article on GLP-1 agonist adverse events.
Key Takeaways
- The SELECT trial (17,604 patients, mean 39.8 months) showed semaglutide reduced major cardiovascular events by 20% in people with obesity but no diabetes (Lincoff et al., 2023)
- The SOUL trial (9,650 patients, median 49.5 months) confirmed oral semaglutide reduced cardiovascular events by 14% in high-risk type 2 diabetes, with no increase in serious adverse events (McGuire et al., 2025)
- Gastrointestinal side effects (nausea, vomiting, diarrhea) are the primary safety concern, causing permanent discontinuation in 10-16.6% of trial participants vs. 2-8.2% on placebo (Lincoff et al., 2023; Aroda et al., 2023)
- A systematic review and meta-analysis found no increased cancer risk with semaglutide use across all tumor types (Nagendra et al., 2023)
- Gallbladder events occur at slightly higher rates (2.8% vs. 2.3% in SELECT), while acute pancreatitis remains rare at approximately 0.2% (Smits et al., 2021)
- The retinopathy signal from SUSTAIN 6 (HR 1.76) has not been replicated in subsequent obesity trials, suggesting it may relate to rapid glucose improvement rather than semaglutide itself
The Major Trials: How Long We Have Actually Watched
The semaglutide safety database comes from three categories of evidence: cardiovascular outcome trials (the longest and largest), weight management trials, and pooled safety analyses across clinical programs.
SUSTAIN 6 was the first large cardiovascular outcome trial. Marso et al. randomized 3,297 patients with type 2 diabetes to weekly subcutaneous semaglutide (0.5 or 1.0 mg) or placebo for 104 weeks (2 years). The primary endpoint, a composite of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke, occurred in 6.6% of the semaglutide group versus 8.9% on placebo (HR 0.74, 95% CI 0.58-0.95). Nonfatal stroke was reduced by 39% (HR 0.61). Fewer serious adverse events occurred overall, though more patients stopped due to gastrointestinal side effects.[1]
SELECT expanded the evidence to people without diabetes. Lincoff et al. enrolled 17,604 patients aged 45+ with obesity (BMI 27+) and preexisting cardiovascular disease. Over a mean follow-up of 39.8 months (approximately 3.3 years), the primary cardiovascular endpoint occurred in 6.5% on semaglutide versus 8.0% on placebo (HR 0.80, 95% CI 0.72-0.90). This was the first trial to demonstrate cardiovascular benefit of a GLP-1 agonist independent of diabetes. The tradeoff: 16.6% of the semaglutide group permanently discontinued treatment due to adverse events, compared to 8.2% on placebo.[2]
SOUL provides the longest randomized follow-up. McGuire et al. studied 9,650 patients with type 2 diabetes and atherosclerotic cardiovascular disease or chronic kidney disease, randomized to oral semaglutide or placebo. With a median follow-up of 49.5 months (just over 4 years), the primary composite endpoint occurred in 12.0% on semaglutide versus 13.8% on placebo (HR 0.86, 95% CI 0.77-0.96). Nonfatal myocardial infarction drove much of the benefit (26% reduction). Serious adverse events were actually lower in the semaglutide group (47.9% vs. 50.3%).[3]
STEP 1 and STEP 5 provide the weight management data. STEP 1 (1,961 patients, 68 weeks) established the 14.9% body weight loss with semaglutide 2.4 mg weekly.[4] STEP 5 (304 patients, 104 weeks) showed this weight loss was sustained over two full years, the longest dedicated obesity trial for semaglutide.[5]
Together, these trials provide approximately 4 years of randomized, placebo-controlled safety data. That is substantial by pharmaceutical standards, though short relative to the decades many patients will take this drug. The title of this article references "5+ years" because semaglutide received its first regulatory approval in 2017 (for type 2 diabetes in the EU and US), meaning real-world exposure data now spans over 8 years for some patients, even though the controlled trial data maxes out at approximately 4 years. Post-marketing surveillance through the FDA Adverse Event Reporting System (FAERS) and equivalent international databases supplements the trial evidence with broader population data, though these passive reporting systems cannot establish causation or calculate true incidence rates.
Gastrointestinal Effects: The Dominant Safety Signal
Nausea, vomiting, diarrhea, and constipation are the most frequently reported adverse events with semaglutide. They are also the primary reason patients discontinue treatment. For dedicated coverage of these effects and how they compare across the GLP-1 class, see GLP-1 agonist adverse events.
Aroda et al. conducted a pooled safety analysis across the entire SUSTAIN and PIONEER phase III programs, representing thousands of patient-years of exposure. Gastrointestinal events were the most common treatment-emergent adverse events, predominantly nausea (11-20% of patients depending on the trial), occurring most frequently during dose escalation and typically resolving within weeks to months.[6]
Smits et al. reviewed the full semaglutide safety landscape in 2021 and noted that while GI events are common, they are generally mild to moderate in severity and transient. The clinical relevance lies in the discontinuation rates: across major trials, 5-10% of patients stop semaglutide specifically because of gastrointestinal intolerance, compared to 1-2% on placebo.[7]
The mechanism is pharmacological, not toxic. Semaglutide slows gastric emptying through the same GLP-1 receptor pathway that produces its metabolic benefits. Dose titration over 16-20 weeks mitigates the initial nausea for most patients. Vomiting rates range from 4% to 11.5% depending on the dose and trial, and diarrhea from 4.5% to 11.3%. Constipation, which receives less attention than nausea, affects approximately 3-5% of patients.
The key distinction is between prevalence and severity. Most GI events are grade 1-2 (mild to moderate). Grade 3 or higher GI events that require medical intervention are uncommon. In SUSTAIN 6, serious gastrointestinal adverse events occurred at similar rates in the semaglutide and placebo groups.[1] The clinical picture is one of frequent but manageable symptoms, not frequent dangerous ones.
Cardiovascular Safety: Better Than Expected
The cardiovascular data for semaglutide is unambiguously positive. Across SUSTAIN 6, SELECT, and SOUL, semaglutide consistently reduced major adverse cardiovascular events. The effect size ranges from a 14% reduction (SOUL, oral formulation) to a 26% reduction (SUSTAIN 6, injectable).[1][2][3]
| Trial | N | Follow-up | MACE HR | 95% CI |
|---|---|---|---|---|
| SUSTAIN 6 | 3,297 | 104 weeks | 0.74 | 0.58-0.95 |
| SELECT | 17,604 | 39.8 months | 0.80 | 0.72-0.90 |
| SOUL | 9,650 | 49.5 months | 0.86 | 0.77-0.96 |
SELECT demonstrated cardiovascular benefit independent of diabetes or weight loss, suggesting a direct cardioprotective mechanism beyond metabolic improvement. The benefit was consistent across prespecified subgroups, including patients stratified by age, sex, baseline BMI, and history of heart failure.
The SOUL trial extended this to the oral formulation, with the reduction in nonfatal myocardial infarction (26%) driving most of the composite benefit. Cardiovascular death was numerically lower but did not reach statistical significance individually in most trials. The SOUL trial also showed that oral semaglutide's cardiovascular benefit was consistent regardless of baseline SGLT2 inhibitor use, important because many patients with type 2 diabetes take both drug classes simultaneously.
The progressive increase in trial size (3,297 to 9,650 to 17,604 patients) and follow-up duration (2 to 3.3 to 4.1 years) across these three trials builds confidence that the cardiovascular safety signal is durable and not a statistical artifact of any single study.
Peter et al. reviewed the early cardiovascular safety data in 2020 and noted that the cardiovascular profile of injectable semaglutide compared favorably to all other diabetes treatments studied in outcome trials.[8] For a comprehensive review of cardiovascular outcomes across all GLP-1 drugs, see GLP-1 drugs and heart disease.
Cancer Risk: What the Evidence Shows
Semaglutide carries an FDA boxed warning for thyroid C-cell tumors based on rodent data. In rats and mice, GLP-1 receptor agonists caused dose-dependent increases in thyroid C-cell tumors, including medullary thyroid carcinoma. Whether this translates to humans remains unresolved after more than a decade of clinical use. For a detailed analysis of the rodent-to-human evidence gap, see GLP-1 agonists and thyroid cancer.
Nagendra et al. conducted a systematic review and meta-analysis of semaglutide and cancer risk across randomized controlled trials and real-world studies. They found no increased risk of any cancer type associated with semaglutide use, a conclusion supported by high-grade evidence.[9]
Within the clinical trial programs, thyroid cancer cases have been rare and isolated. Across studies analyzing approximately 7,830 patients, only a few cases of papillary thyroid cancer and medullary thyroid cancer were reported, each constituting less than 1% of study participants. No dose-response relationship has been observed in human data.
The uncertainty will persist until long-term observational studies covering 10-20 years of exposure accumulate. The rodent signal cannot be dismissed, but the human data through 4 years of controlled follow-up and years of real-world surveillance has not confirmed it.
A biological factor working in favor of human safety: rodent thyroid C-cells express GLP-1 receptors at much higher density than human C-cells. Primate studies have not reproduced the thyroid tumor signal seen in rats and mice. This species difference in receptor expression may explain why the rodent findings have not translated clinically, though it does not eliminate the possibility of a smaller, harder-to-detect effect in humans.
Pancreatitis: Rare but Real
Acute pancreatitis is a recognized risk with GLP-1 receptor agonists as a class. Across the semaglutide clinical trial programs, pancreatitis occurred in approximately 0.2% of treated patients. Smits et al. documented 22 cases in approximately 7,830 semaglutide-treated patients across the SUSTAIN and PIONEER programs.[7] For the complete meta-analytic evidence on GLP-1 drugs and pancreatitis, see GLP-1 agonists and pancreatitis.
An intriguing counterpoint: Nassar et al. found that semaglutide and tirzepatide were associated with a decreased risk of recurrent acute pancreatitis in patients with type 2 diabetes or obesity who had a prior episode. The mechanism may involve reduced triglycerides and improved metabolic health rather than a direct pancreatic effect.[10]
The practical implication: pancreatitis is a known but uncommon risk. Current labeling advises against use in patients with a history of pancreatitis, though the Nassar data raises questions about whether this blanket restriction is appropriate for patients whose metabolic risk factors contribute to their pancreatitis risk.
Individual case reports of fatal pancreatitis after years of semaglutide use exist in the literature. These cases tend to involve patients with multiple risk factors for pancreatitis (gallstones, high triglycerides, alcohol use), making it difficult to attribute causation solely to semaglutide. The absolute risk of approximately 2 cases per 1,000 patient-years is lower than the pancreatitis risk associated with uncontrolled type 2 diabetes itself. Monitoring for symptoms (severe abdominal pain radiating to the back) remains important regardless of treatment duration.
Gallbladder Events: A Consistent Signal
Gallbladder-related disorders, primarily cholelithiasis (gallstones) and cholecystitis, occur at modestly higher rates with semaglutide. In the SELECT trial, gallbladder-related events affected 2.8% of the semaglutide group versus 2.3% on placebo. In the STEP program, cholelithiasis was reported in 1.6% of semaglutide-treated patients versus 0.7% on placebo.[2]
The mechanism is likely related to rapid weight loss rather than a direct drug effect. Gallstone formation during caloric restriction and weight loss is well documented independent of GLP-1 agonism. When the body metabolizes stored fat rapidly, cholesterol concentrations in bile increase, promoting stone formation. Patients losing 15-20% of body weight on semaglutide face the same gallstone physiology as patients achieving similar weight loss through bariatric surgery or severe caloric restriction.
The absolute risk increase is modest: an additional 5-9 gallbladder events per 1,000 patients treated, depending on the trial. For context, the cardiovascular event reduction in SELECT prevented approximately 15 major cardiovascular events per 1,000 patients treated. The risk-benefit arithmetic favors treatment for the indicated population, though individual patients with known gallbladder disease or prior cholecystectomy face a different calculus.
Retinopathy: The SUSTAIN 6 Signal
SUSTAIN 6 found a statistically significant increase in retinopathy complications (vitreous hemorrhage, blindness, or conditions requiring treatment) with semaglutide compared to placebo (HR 1.76, 95% CI 1.11-2.78).[1] This was the most concerning safety signal in the entire semaglutide database.
Subsequent analysis suggested this was driven by rapid HbA1c improvement in patients with pre-existing retinopathy and poor baseline glucose control. The "early worsening of retinopathy" phenomenon is documented with insulin and other diabetes treatments that produce fast glycemic improvement. The SELECT trial, conducted in patients without diabetes, did not replicate this signal. The STEP weight management trials, also in predominantly non-diabetic patients, showed no retinopathy signal.
The current interpretation: the retinopathy risk is likely specific to patients with type 2 diabetes who have existing retinopathy and experience rapid glucose lowering, not a generalizable safety concern with semaglutide itself.
Discontinuation and Tolerability Over Time
One of the most practically relevant safety data points is how many patients actually stay on treatment. In SELECT, 16.6% of semaglutide patients permanently discontinued due to adverse events (mainly GI) versus 8.2% on placebo. This means roughly 1 in 6 patients could not tolerate the drug over 3+ years even with gradual dose titration.[2]
Aroda et al. noted that across the SUSTAIN and PIONEER programs, gastrointestinal adverse events peaked during dose escalation and declined with continued treatment. Patients who tolerate the initial months tend to remain on therapy long-term.[6] This pattern suggests the tolerability question is largely resolved in the first 3-6 months of treatment. Real-world data from Japan showed that treatment adherence at 12 months was higher for subcutaneous semaglutide than for oral semaglutide, likely reflecting both the convenience of weekly injection and the stricter dosing requirements of the oral form (taken fasting with minimal water). The framing matters: a 16.6% discontinuation rate also means 83.4% of patients tolerated semaglutide over 3+ years, a high retention rate for any chronic medication.
For a comparison of how semaglutide's tolerability profile differs from tirzepatide and other dual agonists, see tirzepatide side effect profile.
What We Still Do Not Know
The semaglutide safety database is large by trial standards, but several questions remain open.
Duration beyond 4 years. The SOUL trial provides the longest controlled follow-up at just over 4 years. Most patients taking semaglutide for obesity will take it for decades. Whether risks emerge at 5, 10, or 20 years of continuous exposure is unknown.
Thyroid C-cell tumor latency. Medullary thyroid carcinoma can take decades to develop. The 4-year trial window may be too short to detect a real but slow-emerging signal.
Pregnancy and fertility. Semaglutide is contraindicated in pregnancy, and the FDA recommends stopping it at least 2 months before planned conception. Long-term reproductive effects in patients who have used semaglutide for years before pregnancy are not characterized.
Muscle and bone effects. Weight loss with semaglutide includes loss of lean mass, not just fat. Whether this accelerates sarcopenia in older adults taking the drug long-term is an active area of concern, addressed in our article on GLP-1 weight loss and sarcopenia.
Rebound weight regain. The STEP 1 extension showed substantial weight regain after stopping semaglutide. The safety implications of cycling on and off the drug over years have not been studied.
Drug interactions at scale. With tens of millions of users, rare drug interactions that would not appear in trials of thousands may emerge. Post-marketing surveillance systems are actively monitoring this.
Kidney effects. Post-marketing case reports have described acute kidney injury associated with semaglutide, primarily in the context of severe dehydration from vomiting or diarrhea. The SOUL trial, which included patients with chronic kidney disease, did not show worsening renal function as a class signal. Whether semaglutide provides renal protection (as suggested by preliminary SUSTAIN 6 data showing reduced new or worsening nephropathy) or poses renal risk depends heavily on hydration status and patient monitoring.
Mental health effects. European regulators investigated reports of suicidal ideation associated with GLP-1 agonists in 2023. The European Medicines Agency's review concluded that the available evidence did not support a causal link between GLP-1 agonists and suicidal or self-harm thoughts. However, monitoring continues, and the psychological effects of rapid body composition changes on long-term mental health are not fully characterized.
The Bottom Line
Across more than 30,000 patients in randomized trials lasting up to 4 years, semaglutide shows a consistent safety profile: cardiovascular benefit (14-26% reduction in MACE), common but generally manageable gastrointestinal side effects, no demonstrated cancer risk increase, and small but real increases in gallbladder events. The retinopathy signal from SUSTAIN 6 appears specific to rapid glucose lowering in diabetic patients with existing eye disease. The largest gap in the evidence is duration: most patients will take this drug far longer than the trials have watched.