Semaglutide Research: The Trials, the Mechanism, and the Data
Before the details
Semaglutide research is unusually deep for a single molecule — tens of thousands of people across the STEP (weight), SUSTAIN (diabetes), PIONEER (oral diabetes), SELECT (heart), and FLOW (kidney) programs. This page lays the evidence out in plain language and groups it by question: how it works in the brain, how much weight it moves, what it does for the heart and kidneys, and how it stacks up against tirzepatide.
A quick translation of the jargon. A "randomized trial" splits people by chance into drug or placebo, the fairest test there is. "HR 0.80" means a 20% lower risk than placebo. "-14.9%" means body weight fell by about fifteen percent on average. Every figure below points to the study that produced it, listed in full on the references page.
The mechanism: appetite control, in the brain
The defining finding of semaglutide research is where it acts. It is a long-acting agonist of the GLP-1 receptor — a molecular "on" switch that copies the gut hormone GLP-1. In rodents, semaglutide reached the brain directly: the brainstem, the area postrema (a brain region with a leaky blood-brain barrier), the hypothalamic arcuate nucleus, and the parabrachial nucleus. There it cut food intake and shifted food preference — without lowering energy expenditure [4]. In other words, the weight comes off because the brain asks for less food, not because the body burns more.
Zoom in on the arcuate nucleus and you find two opposing neuron crews: POMC/CART cells that signal "full," and NPY/AgRP cells that signal "hungry." GLP-1 receptor agonists turn the satiety crew up and the hunger crew down, and that balance shifts with the body's metabolic state [8]. Foundational work first showed the arcuate nucleus is required for this class of drugs to drive weight loss, using the related agonist liraglutide [9]. A 2024 mechanistic review ties these threads together, while noting that much of the deepest mechanism data is still preclinical — measured in rodents, then matched to human outcomes [12].
Semaglutide weight loss: how much, and for how long
On semaglutide weight loss, the anchor is STEP 1: in adults with overweight or obesity, once-weekly 2.4 mg produced a mean body-weight change of -14.9% at 68 weeks versus -2.4% on placebo — a roughly 12.4-percentage-point gap [1]. Two facts give that number its shape. First, durability: in the STEP 1 extension, stopping the drug led to a mean regain of about 11.6 percentage points within a year [18], and the STEP 4 withdrawal trial showed continued loss on the drug versus regain on placebo [19]. Second, composition: a body-composition substudy found the weight lost was both fat and a meaningful share of lean mass [17].
The through-line is that this is appetite-driven, ongoing treatment. The brain mechanism above explains the eating less; the regain-on-stopping data explains why the field frames obesity as a chronic condition rather than a one-and-done fix.
Heart and kidney outcomes
Beyond weight, the outcome trials are where semaglutide earned its newer approvals. In SUSTAIN-6, once-weekly semaglutide (0.5 or 1.0 mg) reduced the composite of cardiovascular death, nonfatal heart attack, or nonfatal stroke in type 2 diabetes at high risk (HR 0.74; 95% CI 0.58-0.95) [2]. SELECT then extended that to people without diabetes: across 17,604 adults with established cardiovascular disease and obesity, 2.4 mg cut major adverse cardiovascular events by 20% (HR 0.80; 95% CI 0.72-0.90) [3].
The kidney signal is just as concrete. In FLOW, among 3,533 people with type 2 diabetes and chronic kidney disease, once-weekly 1.0 mg reduced major kidney-disease events — kidney failure, a large drop in filtration, or kidney/cardiovascular death — by 24% (HR 0.76; 95% CI 0.66-0.88) [6]. And in 2025, the ESSENCE trial brought the liver into the picture, showing resolution of steatohepatitis without worsening fibrosis in 62.9% of patients versus 34.3% on placebo [24].
Semaglutide vs tirzepatide
On the head-to-head question, semaglutide vs tirzepatide, the cleanest answer comes from SURMOUNT-5: in 751 adults with obesity, tirzepatide produced greater mean weight loss than semaglutide at 72 weeks — -20.2% versus -13.7%, a roughly 6.5-percentage-point advantage, statistically significant (P<0.001) [7]. Tirzepatide is a dual agonist (it hits both the GIP and GLP-1 receptors), whereas semaglutide targets GLP-1 alone — a mechanistic difference often invoked to explain the gap.
The honest framing: both are highly effective, and the comparison is about degree, not direction. Semaglutide also carries the larger cardiovascular- and kidney-outcomes evidence base to date, which is a different axis of value than peak weight loss [3][6].
Compounded semaglutide
A practical chapter of the story is compounded semaglutide. During the federally declared shortage from roughly 2022 into early 2025, compounding pharmacies were permitted to prepare semaglutide outside the approved manufactured product. The approved-product evidence summarized on this site — STEP, SUSTAIN, SELECT, FLOW [1][2][3][6] — was generated with the manufactured drug; compounded or non-pharmaceutical sources fall outside that evidence base. After the shortage was declared resolved in 2025, those compounding pathways were curtailed, leaving ongoing regulatory and telehealth uncertainty. This site does not endorse, evaluate, or source any compounded preparation; it simply notes the distinction so the literature is read against the right product.
Where the research goes next
The frontier is widening past metabolism. A 2024 mechanistic review catalogs preclinical effects — adipose-tissue browning, mitochondrial biogenesis, autophagy — and flags how much remains rodent-stage [12]. Pharmacovigilance work continues to map the real-world adverse-event profile [15], and a thyroid-risk assessment has worked to separate the rodent-derived warning from the (so far unconfirmed) human signal [16]. The pattern across all of it is consistent: a peptide whose central, appetite-quieting mechanism keeps generating new clinical questions worth testing.