Exenatide

Identity, and a set of registries that agree

Exenatide is a 39-residue peptide amide. PubChem indexes it as CID 45588096 with the empirical formula C184H282N50O60S. The FDA Global Substance Registration System holds it under UNII 9P1872D4OL with CAS registry number 141758-74-9 as the primary code, 141732-76-5 marked superseded, INN 8219, USAN NN-05, ChEMBL CHEMBL414357 and ATC code A10BJ01, the record also carrying the earlier A10BX04 classification. The GSRS sequence reads HGEGTFTSDLSKQMEEEAVRLFIEWLKNGGPSSGAPPPS, with a single structural modification logged at residue 39 converting the terminal serine to serinamide. The approved labelling prints the same sequence and the same formula.

Three molecular weights circulate, and the reason is arithmetic rather than structural. The prescribing information gives 4186.6 daltons. PubChem returns 4187. GSRS displays 4188, typed as calculated from the sequence. All three attach to the identical empirical formula, so the spread reflects how each database rounds and which isotopic convention it uses, not a disagreement about what the molecule is. That position is unusual for the compounds catalogued on this site, where registry formulas more often diverge outright.

One derived figure resists tracing to a derivation. The statement that exendin-4 shares 53 per cent sequence homology with human GLP-1 appears across secondary literature. Aligning the first thirty residues of exenatide against human GLP-1(7-37) gives sixteen identical positions, or 53.3 per cent, so the number is reproducible from two published sequences. The earliest primary statement of it located is Young and colleagues in 1999, whose opening paragraph reads that exendin-4 shows 53 per cent sequence similarity to GLP-1 (PMID 10331407); that report gives no alignment, no method and no citation for the figure, and its subject is glucose lowering in mice, rats and monkeys rather than sequence characterisation. The literal PubMed queries run for this page, with the date and the count each returned, are printed in the unsourced record below so the search can be repeated. The labelling itself gives no percentage and says only that the sequences partially overlap.

What the lizard actually does with the peptide

Eng and colleagues isolated exendin-4 from Heloderma suspectum venom in 1992, using a sequencing assay for peptides carrying an amino-terminal histidine. The peptide differs from exendin-3, isolated earlier from a related species, at two positions, Gly2-Glu3 in place of Ser2-Asp3. In dispersed acini from guinea pig pancreas, natural and synthetic exendin-4 produced a monophasic rise in cyclic AMP beginning at 100 pM and plateauing at 10 nM, progressively inhibited by the antagonist exendin-(9-39) amide. Unlike exendin-3, it did not stimulate amylase release or displace radiolabelled vasoactive intestinal peptide.

Chen and Drucker established five years later that exendin-4 is not the lizard's version of GLP-1 but the product of a separate gene. Lizard proglucagon transcripts were detectable in pancreas and intestine and not in salivary gland; a single class of proexendin cDNA, encoding exendin-4 preceded by a 45-residue amino-terminal peptide, was expressed in salivary gland and not in pancreas or intestine. The two peptides are related in sequence and distinct in origin, which is a point that secondary summaries describing exendin-4 as a lizard incretin routinely collapse.

Christel and DeNardo tested the folk premise directly in 2007. Gila monsters received one of four treatments: live prey, force-fed dead prey under anaesthesia, force-fed prey with exendin-4 injected immediately afterwards, and force-fed prey with injection delayed by 24 hours. Plasma glucose and triglyceride rose over time after ingestion in every group, with no significant treatment effect. Plasma exendin-4 showed significant time and treatment effects that did not correspond to the glucose or triglyceride curves. The report gives no per-group counts. Christel and DeNardo concluded that exendin-4 did not affect circulating glucose in Gila monsters as it had been reported to in mammals (PMID 16972064).

The registration trials, and what they measured

Three placebo-controlled trials of 30 weeks each supported the original approval, all of them adding twice-daily subcutaneous exenatide to existing oral therapy. Buse and colleagues randomised 377 patients at 101 US sites on a sulfonylurea background: HbA1c changed by -0.86 per cent at 10 micrograms, -0.46 at 5 micrograms and +0.12 on placebo, with end-of-study weight down 1.6 kilograms in the 10-microgram arm. DeFronzo and colleagues randomised 336 patients at 82 sites on metformin, recording HbA1c changes of -0.78, -0.40 and +0.08 per cent and weight changes of -2.8 and -1.6 kilograms.

Kendall and colleagues ran the third, 733 subjects on metformin plus a sulfonylurea (PMID 15855571). HbA1c changed by -0.8 per cent at 10 micrograms, -0.6 at 5 micrograms and +0.2 on placebo; weight fell 1.6 kilograms in both exenatide arms against 0.9 on placebo. Mild or moderate hypoglycaemia occurred in 28 per cent, 19 per cent and 13 per cent of the three groups, and appeared lower where the sulfonylurea was given at the minimum recommended rather than the maximally effective amount. That interaction, rather than exenatide alone, is what the hypoglycaemia figures describe.

DURATION-1 compared the two presentations. Drucker and colleagues randomised 295 patients to once-weekly extended-release 2 milligrams or twice-daily 10 micrograms in an open-label non-inferiority design over 30 weeks. HbA1c fell 1.9 per cent on the weekly regimen against 1.5 per cent twice daily, a difference of -0.54 to -0.12 per cent by 95 per cent confidence interval, p equal to 0.0023; 77 per cent against 61 per cent of evaluable patients reached 7.0 per cent or below. Buse and colleagues extended 258 of them to 52 weeks (PMID 20215461), reporting a least-squares mean HbA1c change of -2.0 per cent and weight down by more than four kilograms in both arms. The trial was not masked.

The cardiovascular outcome trial, and one number that travels without its caveat

EXSCEL randomised 14,752 adults with type 2 diabetes, 10,782 of them with previous cardiovascular disease, to once-weekly extended-release exenatide 2 milligrams or matching placebo, and followed them for a median of 3.2 years. The primary composite of cardiovascular death, non-fatal myocardial infarction or non-fatal stroke occurred in 839 of 7,356 on exenatide (11.4 per cent) against 905 of 7,396 on placebo (12.2 per cent), hazard ratio 0.91 with a 95 per cent confidence interval of 0.83 to 1.00. That met the prespecified noninferiority margin for safety and missed superiority at p equal to 0.06.

Death from any cause was recorded in 507 patients (6.9 per cent) against 584 (7.9 per cent), hazard ratio 0.86, confidence interval 0.77 to 0.97. The interval excludes one, and the figure is quoted widely on that basis. The publication states that the difference was not considered statistically significant under the hierarchical testing plan: superiority on the primary composite was tested first and failed, and the plan specified that formal testing stop at the first non-significant result. Rates of cardiovascular death, myocardial infarction, stroke, heart failure hospitalisation, acute coronary syndrome hospitalisation, acute pancreatitis, pancreatic cancer, medullary thyroid carcinoma and serious adverse events did not differ significantly between groups.

Two Parkinson's programmes, and a formal integrity flag

Athauda and colleagues reported a single-centre phase 2 trial in 2017. Sixty-two patients with moderate Parkinson's disease were randomised, 32 to once-weekly exenatide 2 milligrams and 30 to placebo, for 48 weeks followed by a 12-week washout. The primary outcome, MDS-UPDRS part 3 in the practically defined off-medication state at 60 weeks, improved by 1.0 points in the exenatide group and worsened by 2.1 points on placebo, an adjusted mean difference of -3.5 points with a confidence interval of -6.7 to -0.3 and p equal to 0.0318. Six serious adverse events occurred on exenatide and two on placebo, none judged related to the interventions.

Vijiaratnam and colleagues reported the phase 3 trial in February 2025. Six UK research hospitals randomised 194 participants, 97 to each arm, to extended-release exenatide 2 milligrams once weekly or placebo over 96 weeks. MDS-UPDRS part III off medication worsened by a mean of 5.7 points on exenatide and 4.5 points on placebo, adjusted coefficient 0.92 with a confidence interval of -1.56 to 3.39 and p equal to 0.47. Serious adverse events occurred in nine participants on exenatide and eleven on placebo. The trial found no support for a disease-modifying effect.

That phase 3 report now carries an Expression of Concern. PubMed types the notice accordingly and links it to the paper: The Lancet 2026;407(10548):2588, PMID 42330995, epub 22 June 2026. The notice text could not be retrieved during preparation of this page. The publisher returned HTTP 403 on two attempts, PubMed carries no abstract for the record, and no PubMed Central copy exists. Secondary reporting attributes the notice to a regulatory inspection at one participating hospital trust, with findings classified as critical and major, but that account was not confirmed against a primary document and is recorded here as unverified. What the primary record establishes is the existence of the notice, its date, and the paper it attaches to.

A separate molecule complicates the neurology reading. NLY01 is a pegylated, longer-acting derivative of exenatide, not exenatide, and McGarry and colleagues reported it as negative in early untreated Parkinson's disease across 58 US clinics: 255 randomised, 36 weeks, MDS-UPDRS parts II and III difference against placebo of -0.39 at 2.5 milligrams and 0.36 at 5.0 milligrams, neither significant (PMID 38101901). PT320, a sustained-release exendin-4 formulation, is a third distinct entity with its own registration, NCT04269642. Findings reported under any of these headings are not interchangeable.

Pharmacokinetics, antibodies and the rodent carcinogenicity record

The approved labelling gives the pharmacokinetics for the twice-daily presentation in patients with type 2 diabetes: median peak plasma concentration at 2.1 hours, mean peak concentration 211 picograms per millilitre and mean AUC 1,036 picogram-hours per millilitre after a 10-microgram subcutaneous dose, mean apparent volume of distribution 28.3 litres, mean apparent clearance 9.1 litres per hour, and a mean terminal half-life of 2.4 hours. Concentrations remain measurable for roughly ten hours after a dose in most individuals. Nonclinical work indicates elimination predominantly by glomerular filtration with subsequent proteolytic degradation.

Linnebjerg and colleagues gave a single 5 or 10 microgram subcutaneous dose to 31 subjects stratified by creatinine clearance and recorded mean half-lives of 1.5 hours with normal function, 2.1 hours with mild impairment, 3.2 hours with moderate impairment and 6.0 hours in end-stage renal disease requiring haemodialysis. Tolerability was reported as acceptable in the mild and moderate impairment groups and not in the end-stage group, where nausea and vomiting were the stated reason. The report gives no event counts for the mild and moderate groups, so it does not establish that those groups were free of either symptom.

Fineman and colleagues pooled twelve controlled twice-daily trials (2,225 patients), five uncontrolled ones (1,538), and four controlled once-weekly trials (653) with one uncontrolled period (128). At 30 weeks 36.7 per cent of twice-daily patients were antibody-positive, 31.7 per cent at low titre and 5.0 per cent at higher titre, falling to 16.9 per cent by three years. Among once-weekly patients 56.8 per cent were positive at 24 to 30 weeks and 45.4 per cent at 52 weeks. Treatment-emergent antibodies tested did not cross-react with human GLP-1 or glucagon; in the 12 per cent of once-weekly patients with higher titres, mean HbA1c reduction was attenuated. That analysis covers the exenatide programmes only and contains no comparator agonist, so it supports no statement about how other GLP-1 receptor agonists behave.

A 104-week rat study at 18, 70 or 250 micrograms per kilogram per day by subcutaneous bolus produced benign thyroid C-cell adenomas in females at every dose, at incidences of 14, 11 and 23 per cent against 8 and 5 per cent in two control groups, at systemic exposures 5, 22 and 130 times the human maximum. The dose-response is not monotonic. Vetter and colleagues pooled the controlled arms of 35 trials covering 5,596 exenatide and 4,462 non-exenatide patients and found 14 pancreatitis diagnoses, 8 against 6, an incidence of 0.1195 against 0.1276 events per 100 patient-years (PMID 31077072).

What is registered, and what is still supplied

ClinicalTrials.gov holds 377 studies with exenatide as an intervention: 273 completed, 93 at phase 3, 87 at phase 4, and 11 currently recruiting. Seven name Parkinson's disease as a condition, one names Alzheimer's disease (NCT01255163, phase 2, terminated), fifty-six name obesity, and three name alcohol. Klausen and colleagues reported the largest of the alcohol studies, NCT03232112, in which patients received once-weekly exenatide or placebo alongside cognitive-behavioural therapy for 26 weeks; the primary endpoint, heavy drinking days, did not separate from placebo, with imaging differences reported as secondary observations (PMID 36066977). The two records of that trial disagree on its size: the publication states that 127 patients were enrolled, while the ClinicalTrials.gov record gives an actual enrolment of 152. Neither record explains the 25-participant gap, and it is left standing here rather than resolved to one figure.

Mitchell and colleagues recruited 16 women with active idiopathic intracranial hypertension, 15 of whom completed, and measured pressure through telemetric catheters: -5.7 centimetres of cerebrospinal fluid at 2.5 hours (P equal to 0.048), -6.4 at 24 hours (P equal to 0.030) and -5.6 at 12 weeks (P equal to 0.058), against an alpha set a priori at 0.1 (PMID 36907221). GSRS records an FDA orphan-drug designation, number 530516, and a European designation, EU/3/16/1629, both for that condition. The single registered phase 3 trial in idiopathic intracranial hypertension, NCT05347147, is recorded as terminated after 14 participants.

Drugs@FDA records the twice-daily solution as application NDA 021773, approved 28 April 2005, the once-weekly extended-release presentation as NDA 022200, approved 27 January 2012, and a later autoinjector form as NDA 209210, approved 20 October 2017. All three carry a marketing status of Discontinued. A generic twice-daily injection, ANDA 206697, was approved on 19 November 2024 and is recorded as marketed. The once-weekly extended-release presentation used in EXSCEL, in both Parkinson's trials, in the alcohol trial and in the intracranial pressure trial is therefore no longer supplied in the United States.

What is not known

No trial of exenatide outside type 2 diabetes has met a primary endpoint in a completed phase 3 programme. The only positive controlled findings outside diabetes are two small ones: a fifteen-completer telemetric trial in idiopathic intracranial hypertension, which lowered pressure by 5.7 centimetres of cerebrospinal fluid at 2.5 hours and 6.4 at 24 hours against an alpha set a priori at 0.1 (PMID 36907221), and secondary imaging observations in the alcohol trial, which reported attenuated fMRI alcohol cue reactivity in ventral striatum and septal area while its primary endpoint of heavy drinking days did not separate from placebo (PMID 36066977). Against those, the cardiovascular outcome trial was neutral on its primary composite; the phase 3 Parkinson's trial was null and now carries an Expression of Concern whose text has not been made retrievable, so the standing of its data is formally unresolved; the Alzheimer's trial was terminated; and the phase 3 trial in idiopathic intracranial hypertension was terminated after fourteen participants, leaving that sixteen-woman phase 2 as the only controlled evidence in the condition. Exposure data are bounded: the longest randomised follow-up is the 3.2-year median of EXSCEL, and the longest published continuous glycaemic dataset is 52 weeks. Human beta-cell mass has never been measured under exenatide and cannot presently be measured directly in life, so the rodent regeneration finding has no human counterpart in either direction. Brain penetration is characterised only in mice. Comparative immunogenicity against the newer analogues has not been tested head to head. The rat thyroid C-cell adenoma signal has no established human counterpart, and its dose-response is not monotonic. All three originator applications are recorded as discontinued, so the once-weekly extended-release presentation that generated the cardiovascular, Parkinson's, alcohol and intracranial pressure data is no longer supplied; material sold outside a regulated supply chain has not been subject to the identity, purity or sterility controls that applied to any of it.

Questions