Semaglutide

Semaglutide is the most source-rich peptide in the corpus and the molecule that defines the modern GLP-1 era — culturally, commercially, and in the trial chronology behind FDA approval as Ozempic (type 2 diabetes), Wegovy (chronic weight management), and Rybelsus (oral type 2 diabetes). The pivotal weight-loss case unfolds across three STEP trials. Wilding 2021 NEJM — the STEP 1 trial behind Wegovy's obesity approval — randomized 1,961 adults with overweight or obesity and without diabetes 2:1 to once-weekly subcutaneous semaglutide 2.4 mg or placebo plus lifestyle intervention for 68 weeks. Mean body-weight change was −14.9% on semaglutide versus −2.4% on placebo (a 12.4-point treatment difference); 86.4% of treated participants reached ≥5% loss, 69.1% reached ≥10%, and 50.5% reached ≥15%. Rubino 2021 JAMA — STEP 4 — used a withdrawal design: 803 participants who completed a 20-week run-in producing 10.6% loss were re-randomized to continue semaglutide or switch to placebo for 48 more weeks. The continued-semaglutide arm lost an additional 7.9% while the switched-to-placebo arm regained 6.9% — a 14.8-point divergence that defined the maintenance question. Wilding 2022 Diabetes, Obesity and Metabolism — the STEP 1 extension — followed 327 STEP 1 completers for 52 weeks off study drug and found participants regained 11.6 percentage points of body weight, roughly two-thirds of their active-phase loss, with cardiometabolic improvements reverting toward baseline in parallel. This is the empirical anchor for every modern claim about discontinuation rebound on the GLP-1 class, and the reason ongoing-treatment framing rather than time-limited-course framing has dominated the chronic-weight-management indication.

The cardiovascular case runs on a parallel trial chronology. Marso 2016 NEJM — SUSTAIN-6 — randomized 3,297 patients with type 2 diabetes and high cardiovascular risk (83% with established CVD or CKD) to weekly semaglutide or placebo and reported a 26% reduction in the composite primary endpoint of cardiovascular death, nonfatal MI, and nonfatal stroke (HR 0.74; 95% CI 0.58–0.95) over a median 2.1 years, alongside a concerning retinopathy-complication signal (HR 1.76) that is still being parsed in subsequent trials. Seven years later, Lincoff 2023 NEJM — the SELECT trial — randomized 17,604 adults aged 45 or older with established cardiovascular disease and BMI ≥27 but no diabetes to weekly semaglutide 2.4 mg or placebo and reported a 20% reduction in the composite primary cardiovascular endpoint (HR 0.80; 95% CI 0.72–0.90) over a mean 40-month follow-up. Mean weight loss was ~9%, but SELECT's authors and subsequent commentary have argued the cardiovascular benefit exceeds what weight loss alone would predict — consistent with the Skacel pair-fed mouse and Lake SLIM LIVER lipidomic mechanistic data. Akbar 2026 Am J Cardiol extended the SELECT signal into the real world: a propensity-matched TriNetX cohort of 14,844 non-diabetic ASCVD patients with BMI ≥27 found GLP-1RA initiation associated with a 32% relative reduction in 5-year all-cause mortality (HR 0.68; 95% CI 0.53–0.88), with directional consistency for MI and heart failure hospitalization.

The competitive context is set by Frias 2021 NEJM — SURPASS-2 — the only major head-to-head trial of the modern GLP-1 class. The 40-week open-label Phase 3 study randomized 1,879 T2D patients on metformin to tirzepatide 5/10/15 mg or semaglutide 1 mg weekly. Tirzepatide met both noninferiority and superiority on HbA1c at all three doses, and the 15-mg arm produced roughly twice the weight loss of semaglutide 1 mg. The honest caveat is that the semaglutide comparator was the 1-mg T2D dose, not the 2.4-mg obesity dose used in STEP and SELECT; a head-to-head against semaglutide 2.4 mg has not been published as of early 2026.

The 2026 indication expansion runs across psychiatry, addiction medicine, and reproductive endocrinology, with the evidence quality varying sharply by indication. Klausen 2026 The Lancet is the strongest single addition: 108 treatment-seeking adults with DSM-5 alcohol use disorder and comorbid obesity randomized 1:1 to weekly semaglutide or placebo for 26 weeks at a Copenhagen center, with heavy drinking days dropping −41.1 percentage points on semaglutide versus −26.4 on placebo (a 13.7-point between-group difference; p=0.0015). It is the first Lancet-tier RCT confirming a GLP-1RA effect on alcohol consumption, with the durability question and obesity-comorbid-only design as the principal limits. Gill 2026 JAMA Psychiatry randomized 72 adults with major depressive disorder and BMI ≥25 to adjunctive oral semaglutide or placebo for 16 weeks and reported reduced effort cost on the Effort-Expenditure for Rewards Task — the first RCT showing GLP-1RA modulation of reward-related motivation in clinical depression. Mehrhof 2026 Neuropsychopharmacology — a cross-sectional Cambridge computational-psychiatry study, n=228 across four groups — found T2D participants showed a blunted effort-acceptance bias relative to matched controls and that the deficit was not normalized by semaglutide; the contrast with Gill is best read as indication-dependent rather than contradictory. Jensterle 2026 Drugs is a narrative evidence map across the incretin class in polycystic ovary syndrome that finds liraglutide has the densest PCOS-specific evidence (reproducible weight loss, visceral and hepatic fat reduction, preliminary androgen-axis and fertility signals) and that semaglutide PCOS data remain sparse but conceptually plausible — a "watch, do not yet recommend" frame paired with the pregnancy-contraindication that complicates the actual clinical workflow.

The 2026 pharmacovigilance landscape has also matured. Dhivagaran 2026 Neurology — the largest independent academic meta-analysis of the nonarteritic anterior ischemic optic neuropathy signal — pooled five observational studies (n=1,593,554) and found semaglutide associated with a 2.52-fold increased NAION risk (95% CrI 1.56–4.72), with the strongest signal in diabetic patients (RR 2.41) and an estimated 85-per-100,000 incidence across the broader GLP-1RA class. Pooled RCT data published by Novo-affiliated investigators do not show the signal; the honest synthesis is that the question is unsettled and the absolute baseline NAION rate is low enough that informed consent and ophthalmologic vigilance for at-risk patients is the appropriate clinical posture rather than avoidance. Engström 2026 Diabetes, Obesity and Metabolism — a Scandinavian active-comparator new-user cohort across Denmark, Norway, and Sweden — compared 158,961 GLP-1RA new users (72.9% liraglutide-weighted) to 188,065 sulfonylurea new users and reported a 19% lower incidence of Parkinson's disease on GLP-1RA (HR 0.81; 95% CI 0.68–0.96), strengthening to HR 0.74 after censoring DPP-4 inhibitor exposure. The cohort is liraglutide-dominant rather than semaglutide-dominant, and a 2-year lag-time sensitivity analysis loses statistical significance — the directional signal is real but the magnitude is fragile. Aboukaoud 2026 J Affect Disord and Li 2026 Diabetes Ther together provide the most comprehensive WHO VigiBase disproportionality analyses of the class — Aboukaoud reports elevated reporting odds ratios for depressed mood and suicidal thoughts across semaglutide, liraglutide, and tirzepatide but no signal for suicide attempts or completed suicide, with the most parsimonious interpretation being that affective vulnerability in the GLP-1RA-prescribed population drives the signal rather than a uniform drug effect; Li reports drug-distinct rather than class-uniform safety signatures across seven approved GLP-1RAs. Both papers carry the standard disproportionality-analysis caveats (selection bias, no incidence denominator) but together collapse the simple class-effect framing.

The body-composition story is the conversation that follows every conversation about magnitude. Batsis 2026 Annals of Internal Medicine is the most rigorous independent academic synthesis to date — a systematic review of 36 RCTs across liraglutide, semaglutide, tirzepatide, and dulaglutide against prespecified benchmarks (approximately 25% of total weight loss expected as fat-free-mass loss by DXA, 15% as skeletal muscle by CT/MRI). The headline editorial reading is that the lean-mass loss is real but must be evaluated against what proportional loss any equivalent weight reduction produces — the question is not whether incretin therapy causes non-zero lean-mass loss (it does, because all substantial weight loss does) but whether the proportional split exceeds the lifestyle benchmark. The clinical implication is that incretin therapy should be paired with resistance exercise and adequate protein intake; the muscle-preservation decision guide walks the applied framework, and the fat-loss decision guide addresses the broader weight-loss-architecture question.

The discontinuation and rebound conversation is the one no single trial resolves. STEP 1 extension's 11.6-point regain on a 327-participant subset, STEP 4's 6.9% regain on placebo over 48 weeks after a 20-week run-in, and the broader observation that cardiometabolic improvements revert in parallel with weight regain — these together make ongoing-therapy the default clinical posture and time-limited-course an underspecified question. Whether semaglutide is a chronic therapy or a tool with a defined exit strategy is an open clinical and economic question, not a settled answer; the GLP-1 discontinuation playbook covers the framework.