What it is
Cagrilintide is a synthetic analogue of human amylin, the hormone pancreatic beta cells release alongside insulin. It is a 37 amino acid peptide with six substitutions against the native sequence and a C20 dicarboxylic acid on the first residue, joined by a gamma-glutamate linker, injected subcutaneously once weekly. Novo Nordisk developed it as AM833 and NNC0174-0833. It is not approved anywhere.
Cagrilintide alone is not CagriSema, and this entry hangs on the distinction. CagriSema is a fixed-dose combination of cagrilintide and semaglutide in one dual-chamber pen. In REDEFINE 1, the same trial over the same 68 weeks at the same 2.4 mg dose, cagrilintide monotherapy gave 11.8 percent weight loss against 2.3 percent on placebo while the combination gave 22.7 percent (Garvey 2025 in the New England Journal of Medicine, with the four-arm breakdown from the sponsor announcement of 20 December 2024). Pages that hand the combination figure to cagrilintide are off by roughly a factor of two.
How it works
Amylin has no receptor gene of its own. The amylin receptors are heterodimers of the calcitonin receptor, a class B G-protein-coupled receptor encoded by CALCR, paired with one of three receptor activity-modifying proteins to give AMY1, AMY2 and AMY3, which swings preference from calcitonin towards amylin. Cagrilintide is deliberately non-selective across both. Fletcher and colleagues, profiling it in 2021 across 25 endpoints, recorded the observation behind that choice: amylin agonists that also hit the calcitonin receptor appear more efficacious in obesity, even though, in their words, selective activation of calcitonin receptors is not efficacious. Cryo-electron microscopy in 2025 traced the dual activation to the Glu14 to Arg17 salt bridge and the Pro37 contact with the receptor extracellular domain.
The site of action is the hindbrain, not primarily the hypothalamus: the area postrema and the adjacent nucleus of the solitary tract, outside the blood-brain barrier. A 2026 cross-species atlas in Nature Metabolism found that the acute response, in area postrema Calcr and Ramp3 neurons, does not sustain weight loss, while long-term treatment raises prolactin-releasing hormone expression in nucleus of the solitary tract Calcr and Prlh cells. Knocking down dorsal vagal complex Prlh abrogated the effects of cagrilintide but not semaglutide. That is a rat experiment, and it is the cleanest evidence that the two mechanisms are genuinely separate.
Native amylin could not be made into a drug at all. It aggregates into amyloid fibrils, the process that deposits islet amyloid in type 2 diabetes, and it clears in minutes, the obstacle Kruse and colleagues open their 2021 design paper on. Three of cagrilintide’s substitutions are anti-aggregation prolines borrowed from rat amylin, the same three that define pramlintide. The C20 diacid gives reversible albumin binding and a half-life of 159 to 195 hours, 6.6 to 8.1 days (Enebo 2021).
What the human trials found
The dose-finding trial is Lau 2021 in the Lancet, NCT03856047: 906 adults with overweight or obesity, no diabetes, 26 weeks including up to six weeks of escalation, randomised to cagrilintide 0.3, 0.6, 1.2, 2.4 or 4.5 mg weekly, liraglutide 3.0 mg daily or placebo. On the trial product estimand weight fell 6.0 to 10.8 percent across the cagrilintide doses against 3.0 percent on placebo, and 4.5 mg beat liraglutide at 10.8 against 9.0 percent. The per-arm estimates for 0.3, 0.6, 1.2 and 2.4 mg are not in the public abstract, so only the range and those anchors appear here.
The 15.4 to 17.1 percent figures often quoted come from the phase 1b trial, Enebo 2021, where all 95 exposed participants received semaglutide 2.4 mg, placebo groups included; they are exploratory combination results. In the phase 2 diabetes trial, Frias 2023, 92 patients, cagrilintide alone lowered HbA1c 0.9 percentage points against 1.8 for semaglutide, and the combination was not superior to semaglutide at p equals 0.075.
REDEFINE 1 randomised 3417 adults for 68 weeks and met both coprimary endpoints: minus 20.4 against minus 3.0 percent on the treatment policy estimand, difference minus 17.3 percentage points, 95 percent confidence interval minus 18.1 to minus 16.6, P less than 0.001. Its four arms on the trial product estimand were 22.7 percent combination, 16.1 semaglutide, 11.8 cagrilintide and 2.3 placebo, and 31.6 percent of the monotherapy group reached 15 percent or more weight loss against 4.7 on placebo. REDEFINE 2, the other filing trial, gave minus 13.7 against minus 3.4 percent in 1206 patients with type 2 diabetes.
The REIMAGINE programme published on 7 June 2026. REIMAGINE 2, 2713 patients, holds the largest cagrilintide monotherapy arm in type 2 diabetes, 152 people. Its primary comparison against semaglutide 2.4 mg on HbA1c gave minus 1.91 against minus 1.75 percentage points, a difference of minus 0.16 percentage points, confidence interval minus 0.27 to minus 0.05, p equals 0.0035: significant in 1208 patients, clinically marginal, and reported as a win without being shown at that size. REIMAGINE 3 added the combination to basal insulin in 274 patients and reported, verbatim, that no severe hypoglycaemia was reported.
REDEFINE 4, NCT06131437, is the only head-to-head the molecule has been in: 809 people with obesity, 84 weeks, against tirzepatide 15 mg. On 23 February 2026 the sponsor announced 23.0 against 25.5 percent on the efficacy estimand and 20.2 against 23.6 on the treatment regimen estimand, stating verbatim that CagriSema 2.4/2.4 mg did not meet the primary endpoint of showing non-inferiority on weight loss compared to tirzepatide 15 mg at 84 weeks. There is no publication and no disclosed confidence intervals. Failing non-inferiority is not being ineffective, but the claim that this is the most effective obesity treatment available is not supported by the only head-to-head that exists.
The REDEFINE 1 expectation gap, and the arithmetic
Novo Nordisk had guided to roughly 25 percent. REDEFINE 1 read out at 22.7 on the trial product estimand and 20.4 on treatment policy, the share price fell about a fifth in a session and the trial has been called a failure ever since. It met every prespecified endpoint. The sponsor’s own explanation is the strongest: dosing was flexible, and at week 68 only 57.3 percent of combination recipients were still on the highest dose, against 82.5 percent on cagrilintide monotherapy and 70.2 on semaglutide.
The same arithmetic settles the word synergy. On that estimand semaglutide gave 13.8 percentage points of effect over placebo and cagrilintide gave 9.5, summing to 23.3. The combination gave 20.4. Additive at best, arguably sub-additive, and nothing in it is synergistic.
A standalone drug again, which is new
Until late 2025 cagrilintide appeared alone only as a comparator arm. On 16 September 2025 the sponsor presented the REDEFINE 1 monotherapy data and announced a dedicated phase 3 programme, RENEW. RENEW 1, NCT07220642, randomised 300 adults without diabetes against placebo to a week 64 endpoint, and a companion diabetes trial, NCT07220759, enrolled 330, both from November 2025. The thesis is tolerability, not potency, and the 82.5 against 57.3 percent dose retention is the argument. NCT07607587 states it out loud, testing tolerability in 114 people who stopped GLP-1 receptor agonists because of gastrointestinal side effects.
Safety, with the denominators
Gastrointestinal events are the class effect, mostly mild to moderate and transient. In the dose-finding trial they affected 41 to 63 percent across cagrilintide doses against 32 on placebo. In REDEFINE 1 the combination produced them in 79.6 percent against 39.9, with nausea 55 and vomiting 26.1. In REIMAGINE 2 the monotherapy arm had any adverse event in 82.2 percent, 125 of 152, between placebo at 70.5 and semaglutide at 81.2. Rates for the same regimen vary widely between trials.
Three findings rarely seen elsewhere. A 2026 meta-analysis of three randomised trials put administration-site conditions at a relative risk of 3.27 against semaglutide, 95 percent confidence interval 1.27 to 8.46, and serious adverse events on cagrilintide monotherapy at 1.83, interval 1.03 to 3.24, weak because it barely excludes 1.0 but the only such comparison there is. The cagrilintide plus semaglutide arm of the MASH trial NCT05016882 was not substantially different from placebo on fibrosis while semaglutide alone was nominally significant, a negative result. And the 2026 BMJ network meta-analysis of 262 trials and 99,791 participants placed the combination among the agents with the highest discontinuation for adverse events, with a fatigue relative risk of 3.2, about 92 more cases per 1000 in a year.
Hypoglycaemia is not an intrinsic risk: amylin receptor agonism is not insulinotropic, and pramlintide’s hypoglycaemia came from the mealtime insulin beside it. No amylin receptor agonist has completed a cardiovascular outcomes trial.
Regulatory status
A new drug application for the combination went to the US FDA on 18 December 2025, on REDEFINE 1 and 2. It is not approved. No action date has been made public, no approval in any jurisdiction, and the sponsor says only that the FDA is expected to review it in 2026. No European application has been publicly confirmed, so European status is unknown rather than filed. REDEFINE 4 was not part of the filing basis. Cagrilintide monotherapy is filed nowhere.
What circulates, and why no numbers appear here
Cagrilintide circulates as lyophilised powder under research-use framing, alongside written protocols that read like prescriptions. This site does not reproduce those figures, because the compound is unapproved for every use, because the escalation is where the harm sits, and because nothing in that supply chain establishes that a vial holds what its label claims.
What the record does show is this. One widely copied page carries a dose given three days a week and an eight weeks on, eight weeks off cycle. Every human cagrilintide trial since first-in-human in 2014 has dosed once weekly, which is what a half-life of 159 to 195 hours dictates, and three injections a week of a peptide with a one week half-life would accumulate towards roughly triple the studied exposure. No trial has tested that frequency or any cycling schedule. A second item is a five-step titration ladder attributed to Lau 2021: it is the semaglutide ladder with cagrilintide’s name on it, and its lowest step has never been given as a cagrilintide dose in any published trial.
The rest of the amylin field
Calling cagrilintide the amylin was defensible in 2022. A 2026 review lists cagrilintide, eloralintide, petrelintide, MET-233i, ABBV-295 and AZD6234 in human development, plus zenagamtide, the compound formerly called amycretin, a rename that breaks older cross-links. Eloralintide, a selective agonist, reached roughly 20 percent weight loss at 48 weeks as a single drug in Billings 2025, against cagrilintide’s 11.8 at 68 weeks, and petrelintide had 98 percent of its phase 2 participants reach the maintenance dose.
What the record gets wrong
That the lipid is a C18 fatty acid. It is eicosanedioic acid, a C20 alpha,omega-dicarboxylic acid, on Lys1 through a gamma-glutamate linker. That is not cosmetic: a diacid presents a free terminal carboxylate that changes the albumin-binding geometry. Semaglutide uses a C18 diacid, so the claim reads as semaglutide chemistry stapled onto cagrilintide.
That CagriSema failed REDEFINE 1, when it met every prespecified endpoint and missed a commercial guidance figure, which are different events. That it is the amylin, when six others are in development and two already beat it on the measures above.
The accurate sentence about cagrilintide in September 2026 is a narrow one. It is an investigational once-weekly amylin analogue that produced 11.8 percent weight loss over 68 weeks as a single drug, it is the most clinically advanced amylin programme and neither the most potent nor the best tolerated, it is approved nowhere for anything, and almost every number larger than 12 percent attached to its name belongs to two drugs rather than to this one.