Areas where scientific evidence is lacking or incomplete.
Pivotal trials consistently excluded patients with advanced CKD, recent malignancy, or multiple coexisting chronic conditions. The safety profile in these real-world populations, who carry higher baseline risks for pancreatitis and other adverse events, remains incompletely understood.
Implications: Most semaglutide prescriptions in practice will go to patients with comorbidity profiles that differ from trial populations. Real-world effectiveness and safety may diverge from controlled trial results.
Semaglutide is studied almost exclusively in overweight and obese populations. Its safety in lean individuals is uncertain, particularly for newer indications like alcohol use disorder. Earlier exenatide data showed a BMI-dependent split where lean participants saw increased heavy drinking.
Implications: Expanding semaglutide to addiction indications may encounter a population for whom the risk-benefit ratio is fundamentally different from the well-studied metabolic cohort.
Limited data exists on long-term effects on bone accrual, pubertal development, and neurocognitive endpoints in children and adolescents. The developmental implications of chronic GLP-1 receptor agonism during growth periods are unknown.
Implications: Paediatric obesity is a growing indication, but the safety evidence lags well behind the adult data. Long-term follow-up studies are needed before broad paediatric adoption.
While the rebound effect is well-documented (two-thirds of weight returns within a year), research is needed on strategies to prevent metabolic rebound after withdrawal. Are there stepped-down dosing protocols, combination approaches, or lifestyle interventions that can sustain benefits?
Implications: Without discontinuation strategies, semaglutide may function as a lifelong therapy by default rather than by design. The economic and practical implications of this are significant.
Current MASH data relies on surrogate histological markers (biopsy scores). Hard clinical outcomes, specifically progression to cirrhosis, need for liver transplant, and liver-related mortality, remain unmeasured and require multi-year follow-up.
Implications: FDA accelerated approval was based on biopsy improvement. Confirmatory trials with hard endpoints will determine whether semaglutide changes the natural history of MASH or improves intermediate markers only.
It is not well understood why some patients are non-responders or experience severe GI intolerance while others do not. Gaps exist in using genomics, metabolomics, and microbiomics to identify predictive biomarkers for treatment response and adverse effects.
Implications: Precision medicine approaches could optimise patient selection and dosing, but the biomarker research is still early-stage. Currently, treatment is empirical: start low, titrate, hope for tolerance.
Semaglutide appears to reduce the 'wanting' (motivational drive) for addictive substances, but whether it also reduces 'liking' (hedonic memory of the substance) is unclear. The distinction matters because persistent liking despite reduced wanting could drive relapse.
Implications: If semaglutide only suppresses motivation without addressing hedonic memory, relapse rates after discontinuation may be high. This parallels the weight-regain problem in a new domain.
Expert disagreements and competing evidence.
Large RCTs and meta-analyses show no statistically significant increase in acute pancreatitis risk with semaglutide. SELECT trial: 0.2% semaglutide vs 0.3% placebo.
SELECT trial (N=17,604), SUSTAIN programme, and class-wide GLP-1 meta-analyses consistently show neutral pancreatitis risk in controlled settings.
Source: Controlled clinical trial data
Post-marketing case reports and surveillance identify semaglutide as a plausible precipitating factor, particularly in multimorbid patients who are systematically excluded from the trials that show neutral risk.
Individual case reports in patients with CKD, recent malignancy, or multiple comorbidities document severe pancreatitis episodes attributed to semaglutide.
Source: Post-marketing surveillance and case series
Verdict Note
Both positions have merit. The RCT data is robust within its population, but the exclusion criteria that make trials clean also make them less representative of real-world prescribing. The truth likely depends on the individual patient's comorbidity profile.
Early Phase 2b trials reported that semaglutide effectively induced MASH resolution but showed no significant signal for fibrosis improvement.
Phase 2b MASH studies with smaller sample sizes and shorter follow-up.
Source: Phase 2b trial publications
The Phase 3 ESSENCE trial contradicted earlier findings, achieving statistically significant fibrosis improvement by at least one stage (37% vs 22.5% placebo).
ESSENCE Phase 3 with larger sample size, longer follow-up, and fibrosis improvement as a co-primary endpoint.
Source: ESSENCE Phase 3 trial (2025)
Verdict Note
The Phase 3 data supersedes the Phase 2b finding. The earlier 'no signal' was a power and duration issue, not a biological one. ESSENCE provided the evidence that led to FDA accelerated approval for MASH.
The SELECT trial reported 10.2% mean body weight reduction over 208 weeks at 2.4 mg weekly dose.
SELECT (N=17,604) enrolled patients with established cardiovascular disease, older population, higher comorbidity burden.
Source: SELECT trial (Lincoff, NEJM 2023)
The STEP trials reported 15-17% mean body weight reduction over 68 weeks at 2.4 mg weekly dose.
STEP trials (N=1,961) enrolled overweight/obese adults without established CVD, generally younger and metabolically less complex.
Source: STEP programme publications
Verdict Note
Both figures are accurate for their respective populations. The difference is explained by patient selection (age, comorbidity, metabolic resistance) and trial duration. SELECT's 10.2% over 4 years represents a sustained, real-world-relevant result. STEP's 15-17% represents peak efficacy in a selected population.
Earlier exenatide studies showed a sharp phenotype split: heavy drinking was only reduced in patients with BMI above 30, while lean participants (BMI <25) actually saw an increase in heavy drinking.
Exenatide AUD studies showing differential response by BMI category.
Source: Exenatide alcohol studies
Current semaglutide data from the JAMA Psychiatry trial suggests a broader therapeutic window, with consistent effects across the overweight-to-obese spectrum without the rigid BMI-dependence seen with exenatide.
JAMA Psychiatry semaglutide AUD trial showing broader efficacy.
Source: JAMA Psychiatry 2024
Verdict Note
Semaglutide data is encouraging but sample sizes for the lean subgroup remain small. Until a dedicated trial enrols lean AUD patients, the safety and efficacy question in this population remains open. Clinical caution is warranted.
Injectable semaglutide causes nausea that peaks 2-3 days after injection and improves as the week progresses.
SUSTAIN and STEP trial adverse event timing data.
Source: Injectable semaglutide trial data
Oral semaglutide causes more consistent day-to-day nausea due to daily dosing, though some sources suggest it is most pronounced in the hours immediately after taking the tablet.
Rybelsus real-world prescribing experience and patient-reported outcomes.
Source: Oral semaglutide clinical experience
Verdict Note
The nausea pattern difference is real and reflects the pharmacokinetic difference between formulations. Neither pattern is objectively worse; patient preference determines which is more tolerable. Both improve with time and titration.