GLP-1 in Parkinson's: LixiPark Positive, Exenatide Null

Lixisenatide and exenatide gave different answers in Parkinson's disease because the two trials asked different questions of different molecules in different patients: LixiPark measured motor scores on medication at 12 months in people diagnosed within three years, and found a 3.08-point difference favouring lixisenatide; Exenatide-PD3 measured motor scores off medication at 96 weeks in a broader population, and found a 0.92-point difference favouring placebo that was consistent with chance 12. What the pair jointly establishes is narrow: one GLP-1 receptor agonist produced a measurable short-term motor-score difference in early disease, and a related one did not alter progression over two years. Neither result establishes disease modification.

This site covers the Alzheimer's trials of a different molecule separately. This article is about Parkinson's disease and about the inference problem that two contradictory trials in one drug class create. It is a comparative analysis of design, not a verdict on the class, and every section states the tier of evidence it rests on.

Why an incretin was tested in Parkinson's disease

Evidence tier for this section: animal models and in vitro work, plus one earlier human phase 2. The hypothesis predates both trials. GLP-1 receptor agonists had shown neurotrophic and neuroprotective properties in cell culture and in rodent models of Parkinson's disease, in which they reduced dopaminergic neuron loss after a toxic lesion 12. Impaired insulin signalling in the brain was proposed as a pathway shared between metabolic disease and neurodegeneration. Epidemiological studies and small randomised trials had suggested possible effects on risk and progression, with the usual limits of such comparisons 2.

The human precedent was a 2017 single-centre trial of 62 people with moderate Parkinson's disease, randomised to once-weekly exenatide or placebo for 48 weeks with a 12-week washout. Off-medication MDS-UPDRS part III scores improved by 1.0 point on exenatide and worsened by 2.1 on placebo at 60 weeks, an adjusted difference of −3.5 points (95% CI −6.7 to −0.3; p = 0.0318). The authors wrote that whether exenatide altered the underlying disease or produced a long-lasting symptomatic effect was uncertain 3. That sentence is the hinge of the whole story. Both later trials were attempts to answer it.

LixiPark: design, population and the 12-month result

Evidence tier: human randomised trial, phase 2. LixiPark was a multicentre, double-blind, placebo-controlled proof-of-concept trial run through the French Parkinson's clinical research network, registered in February 2018 with enrolment starting that June 4. It enrolled 156 people in whom Parkinson's disease had been diagnosed less than three years earlier, who were on a stable dose of symptomatic medication and had no motor complications. They were randomised 1:1 to daily subcutaneous lixisenatide, at the dose licensed for type 2 diabetes with a lower step permitted for intolerance, or matching placebo, for 12 months, followed by a 2-month washout 14.

The primary endpoint was change from baseline in MDS-UPDRS part III, assessed in the on-medication state at 12 months. Baseline scores were about 15 in both groups. At 12 months the lixisenatide group had changed by −0.04 points and the placebo group by 3.04 points; the difference of 3.08 points had a 95% confidence interval of 0.86 to 5.30 and P = 0.007. At 14 months, after washout, off-medication motor scores were 17.7 (95% CI 15.7 to 19.7) with lixisenatide and 20.6 (95% CI 18.5 to 22.8) with placebo. Other secondary endpoints did not differ substantially. Nausea occurred in 46% of the lixisenatide group and vomiting in 13% 1.

Two features of the design cut in opposite directions. Assessing on medication measures the combined effect of the trial drug and background levodopa, which is the clinically realistic state but makes a symptomatic contribution hard to exclude. The washout assessment partly addresses that: a difference that persists two months after the last injection is less easily explained by a direct symptomatic effect, though the two-month off-medication figures were a secondary endpoint and were reported as group means rather than as a tested difference. The investigators themselves concluded that longer and larger trials were needed 1.

Exenatide-PD3: longer, larger, and null

Evidence tier: human randomised trial, phase 3. Exenatide-PD3 was a multicentre, double-blind, parallel-group, placebo-controlled trial at six UK research hospitals, sponsored by University College London and registered in January 2020 5. Participants were aged 25 to 80, had a diagnosis of Parkinson's disease, were at Hoehn and Yahr stage 2.5 or less on treatment and had been on dopaminergic treatment for at least four weeks. They were randomised 1:1, with minimisation by stage and site, to once-weekly extended-release exenatide by pen injection or visually identical placebo for 96 weeks 2.

Between January 2020 and April 2022, 215 people were screened and 194 randomised, 97 to each arm; 29% were female. The primary outcome was MDS-UPDRS part III off dopaminergic medication at 96 weeks, analysed by linear mixed modelling in the 188 participants with at least one follow-up visit. Scores worsened by a mean of 5.7 points (SD 11.2) on exenatide and 4.5 points (SD 11.4) on placebo. The adjusted coefficient for exenatide was 0.92 points (95% CI −1.56 to 3.39; p = 0.47), numerically favouring placebo and statistically indistinguishable from zero. Nine participants (9%) on exenatide and 11 (11%) on placebo had at least one serious adverse event 2.

The authors wrote that they found no evidence to support exenatide as a disease-modifying treatment, and that studies with agents showing better target engagement, or in specific subgroups, would be needed before the class could be judged 2. The upper bound of the confidence interval, 3.39 points, is informative on its own. It excludes an effect of the size LixiPark reported at 12 months, for this molecule, at this duration, in this population.

What differs between the two trials besides the molecule

Discordant results invite a single explanation. Here there are at least six candidate explanations, and the trials cannot separate them.

Disease stage is the explanation most often offered. A drug acting on a degenerative process might show more in people diagnosed within three years than in a population with longer disease, in which fewer dopaminergic neurons remain to protect. The medication state is the second. On-medication assessment can capture a symptomatic contribution; off-medication assessment is the harder test. Duration is the third: a short-lived difference that fades by two years would look positive at 12 months and null at 96 weeks. The molecule is the fourth, and the one most relevant to this site's readers: exenatide and lixisenatide are both exendin-4 derivatives, but they differ in formulation, dosing interval and exposure profile, and neither trial measured either drug in cerebrospinal fluid 12.

MDS-UPDRS part III as an endpoint

The Movement Disorder Society revision of the Unified Parkinson's Disease Rating Scale, part III, is a clinician-rated motor examination scored from 0 to 132, higher meaning greater motor disability 1. It is the standard primary endpoint in disease-modification trials because it is validated, widely used and sensitive to dopaminergic state. Those same properties create the interpretive problem. A scale sensitive to dopaminergic state will register a symptomatic effect as readily as a disease-modifying one, and the two are indistinguishable at a single time point.

The 3.08-point difference in LixiPark should be read against that. It is roughly 20% of the baseline score of about 15, and it was obtained in 156 people. A confidence interval running from 0.86 to 5.30 points is consistent with an effect too small for a patient to notice and with one that would matter. The placebo group's 3.04-point progression over a year is itself a reminder of how slowly the scale moves in early disease, and of why 96 weeks in Exenatide-PD3 produced a mean change of only 4.5 points on placebo 12. A trial powered to detect a two-year slowing of that progression needs either a very large effect or a very large sample.

Target engagement: the unresolved question of brain exposure

Evidence tier: this section describes an absence of evidence. Neither trial reported drug concentrations in cerebrospinal fluid or any direct marker of receptor engagement in the brain. Both molecules are peptides of around 40 residues administered subcutaneously, and the fraction reaching brain tissue in humans is not established. The Exenatide-PD3 authors made this the centrepiece of their interpretation: future studies should use agents with better target engagement, or should select participants in whom engagement can be shown 2.

This matters for how the null result is read. A trial that fails because the drug never reached its target has tested delivery, not the hypothesis. A trial that fails despite adequate exposure has tested the hypothesis. Without an exposure measurement, Exenatide-PD3 cannot be assigned to either category, and the same limitation applies in reverse to LixiPark: a positive result without exposure data cannot be attributed to central receptor activation rather than to a peripheral effect that happens to register on a motor scale.

Adverse events reported in the three trials

Gastrointestinal effects dominated, as expected for the class. In LixiPark, nausea occurred in 46% of the lixisenatide group and vomiting in 13% 1. In Exenatide-PD3, serious adverse events occurred in 9% of the exenatide group and 11% of the placebo group, and the authors described the drug as safe and well tolerated in this population 2. In the 2017 exenatide trial, injection-site reactions and gastrointestinal symptoms were common in both groups; six serious adverse events occurred on exenatide and two on placebo, none judged related to the intervention 3. None of the three trials reported a safety signal specific to Parkinson's disease, and none was large enough to detect an uncommon one.

What the two results jointly establish, and what remains open

Established: in people with early Parkinson's disease on stable medication, 12 months of lixisenatide produced a motor-score difference of about three points on medication, with lower off-medication group means at a two-month washout assessment, in a phase 2 trial of 156 people 1. Also established: in a broader UK population, 96 weeks of weekly exenatide produced no detectable difference in off-medication motor progression in a phase 3 trial of 194 people, with a confidence interval excluding an effect of LixiPark's size 2. Neither trial measured brain exposure. One of the two reports carries an expression of concern whose outcome is pending 6.

Open: whether the LixiPark difference is symptomatic or disease-modifying; whether it survives a longer trial; whether it is specific to lixisenatide, to early disease, or to on-medication assessment; and whether any peripherally administered GLP-1 receptor agonist reaches human brain at a relevant concentration. The two trials together do not support a class claim in either direction. They support a specific, testable proposition: that a confirmatory trial of lixisenatide, in early disease, with exposure measured, over at least two years, would resolve most of what they left unresolved. No GLP-1 receptor agonist is approved for Parkinson's disease, and these trials provide no basis for use outside a trial.