A precision approach to GLP-1 therapy in Parkinson's – Q&A with Dr Cynthia Sandor

Weight-loss drugs, specifically the GLP-1 class of drugs behind Ozempic, have shown promising but inconsistent results in Parkinson's disease.

They appear to work only in patients whose neurons are insulin-resistant. For Dr Cynthia Sandor, Edmond J Safra Assistant Professor in Parkinson's Disease in the Department of Brain Sciences at Imperial College London, that subtype can be identified in advance through genetic data and routine blood tests. 

Her new study, STRAT-GLP1, funded by a grant from The Michael J. Fox Foundation for Parkinson's Research, sets out to answer a more detailed question than previous trials. Instead of asking whether GLP-1 drugs work in Parkinson's, Dr Sandor wants to learn which patients are likely to benefit, and whether they can be identified before treatment begins.

In this Q&A, Dr Sandor discusses the science behind that question, and what it could ultimately mean for patients, families, and the future of Parkinson's research.

Could you explain how drugs like Ozempic, originally developed for diabetes and weight loss, could potentially slow Parkinson’s disease?  

Dr Cynthia Sandor: Parkinson’s disease involves the gradual loss of a very specific population of brain cells, the dopamine-producing neurons in a region called the substantia nigra. What my colleagues at Cardiff and I have found is that a particular subtype of these neurons, identifiable by a marker called AGTR1, appears to be selectively vulnerable in Parkinson’s. Crucially, these cells show the hallmarks of insulin resistance, the same biological state that GLP-1 drugs like Ozempic are designed to correct. When these neurons become insulin-resistant, they activate stress pathways, lose the ability to regulate inflammation, and become far more susceptible to the toxic protein clumps, alpha-synuclein, that define Parkinson’s pathology. 

What population-scale data gives us is the statistical power to find the signal within the noise: to compare genetic and metabolic profiles across hundreds of thousands of people with Parkinson’s. Dr Cynthia Sandor Edmond & Lily Safra Assistant Professor in Parkinson's Disease, Department of Brain Sciences | Imperial College London

GLP-1 receptor agonists appear to restore normal signalling in these cells: reducing inflammation, protecting mitochondrial function, and slowing alpha-synuclein aggregation. The key insight is that this isn’t a generic brain effect: it is tied to a specific, identifiable neuronal population. And that is precisely why we believe only a subset of Parkinson’s patients, those whose disease is driven by this metabolic vulnerability, are likely to respond. 

 
What is unique about your methodology, and what can large population-level data tell us that smaller clinical trials cannot? 

ELAD was an elegant mechanistic study in Alzheimer’s disease, asking whether liraglutide has biological effects in the brain, and it delivered important positive signals. Our question in STRAT-GLP1 is fundamentally different, and it is a Parkinson’s-specific question: not whether GLP-1 drugs work in Parkinson’s broadly, but *which patients* with Parkinson’s are likely to benefit, and can we identify them in advance? 

Two Phase II trials [1] in Parkinson’s showed promising results, but a Phase III [2] trial last year found no overall benefit. The most likely explanation, supported by post-hoc analyses, is that the trial simply enrolled too few patients from what we believe is the true responder population: the insulin-resistant biological subtype. A clinical trial, however well-designed, cannot easily answer that question on its own, because it tests an average effect across a heterogeneous group. What population-scale data gives us is the statistical power to find the signal within the noise: to compare genetic and metabolic profiles across hundreds of thousands of people with Parkinson’s, and to ask whether those who were already taking GLP-1 drugs in real life, over 6,000 patients in UK health records showed a different disease trajectory depending on their metabolic subtype. That is a question that requires scale, and it is one that no single clinical trial can answer. 

 If this research is successful, how does the proposed genetic and blood-based test compare to current diagnostic tools available to clinicians? 

 At the moment, there is no validated tool that allows a clinician to tell a Parkinson’s patient whether a GLP-1 drug is likely to help them. Diagnosis of Parkinson’s itself relies on clinical examination, and while there are emerging blood biomarkers for the disease, none currently stratify patients by the biological subtype driving their condition. What we are aiming to develop is a practical stratification tool based on information that is, in large part, already routinely collected: blood sugar levels, cholesterol, blood pressure, BMI, the kind of measurements any GP takes. We will complement these with a small number of blood protein markers to capture the underlying metabolic and inflammatory biology more precisely. If successful, this would be the first precision tool for GLP-1 therapy selection in Parkinson’s, and it is designed from the outset to be clinically usable rather than confined to specialist research centres. 
 

Would this test be affordable and accessible, or is there a risk that it remains out of reach for patients outside well-resourced healthcare systems? 

 This is something we have thought about carefully from the start. The core metabolic markers we are targeting HbA1c, triglycerides, blood pressure, and BMI are already measured routinely in most healthcare systems worldwide, including in lower-resource settings.

What we are aiming to develop is a practical stratification tool based on information that is, in large part, already routinely collected: blood sugar levels, cholesterol, blood pressure, BMI , the kind of measurements any GP takes. If successful, this would be the first precision tool for GLP-1 therapy selection in Parkinson’s. Dr Cynthia Sandor

The more specialist protein assays we are using in the research phase are currently expensive, but their costs are falling rapidly, and our explicit aim is to identify the simplest possible combination of markers that works clinically. We have also deliberately included the All of Us Research Program in our genetic analyses, a US cohort in which nearly half of the participants are from historically underrepresented groups, including African and Hispanic/Latino backgrounds, precisely because we want our findings to be generalisable beyond the predominantly European populations that have dominated this kind of research. Equity of access is not an afterthought here; it is built into the study design. 

 Are there any sex-related differences that you are aware of that are worth looking into? 

 First, at the level of disease prevalence, Parkinson's is consistently more common in men than women, roughly 1.5 times more frequent, and men tend to have a more aggressive disease course. Women, on the other hand, are more likely to be diagnosed later and may present with a somewhat different symptom profile. The biological reasons for this are not fully understood, but oestrogen is thought to play a neuroprotective role, which may partly explain why premenopausal women have lower rates of Parkinson's. 

Second, when it comes to metabolic risk factors specifically, there are real sex differences in how insulin resistance and obesity manifest. Women tend to accumulate fat differently from men – more subcutaneous and less visceral, and the relationship between metabolic syndrome and neurological risk may therefore differ by sex. Some observational data suggest that obesity in midlife is a stronger risk factor for dementia and neurodegeneration in women than in men, though the Parkinson's-specific evidence is less established. 

In our own study, we will explicitly test for sex-specific effects in both aims, looking at whether metabolic stratification of Parkinson's risk and GLP-1 treatment response differs between men and women. This is built into the analysis plan rather than treated as an afterthought. Given the sex differences in both Parkinson's biology and metabolic disease, it would be surprising if sex did not turn out to be a relevant modifier,  but the honest answer is that we do not yet have definitive data, and this is one of the questions STRAT-GLP1 is positioned to address. 

 What do you see as the biggest societal impact of this research, both for patients with Parkinson's and for the broader field of personalised medicine? 

For patients, the most immediate hope is that a drug already approved, widely prescribed, and well-tolerated could be shown to slow down their disease, if we can identify who it works for. That is an enormously powerful prospect, because it would potentially compress the timeline to an effective treatment by years compared to developing a new molecule from scratch.

The era of one-size-fits-all trials in complex, heterogeneous diseases may need to end, and this project is, in a small way, part of building what comes next.  Dr Cynthia Sandor

More broadly, the methodology we are building is not Parkinson’s-specific. The pattern of a promising drug failing a Phase III trial because the responder subgroup was diluted by non-responders is not unique to Parkinson’s; it is a recurring problem across neurodegeneration and beyond. If we can show that genetics and blood biomarkers can prospectively identify treatment-responsive subgroups, that framework could become a template for other conditions facing the same challenge. The era of one-size-fits-all trials in complex, heterogeneous diseases may need to end, and this project is, in a small way, part of building what comes next. 
 

Latest results

• Volpato et al. (2025). A deep cellular atlas of the human ventral substantia nigra in Parkinson’s identifies a genetic and molecular overlap with insulin resistance. “medRxiv” (preprint), link 

Related research 

• Athauda et al. (2017). Exenatide once weekly versus placebo in Parkinson’s disease: a randomised, double-blind, placebo-controlled trial. “The Lancet”, 390, 1664–1675; link 

• Meissner et al. (2024). Trial of Lixisenatide in Early Parkinson’s Disease. “New England Journal of Medicine”, 390, 1176–1185, link. 

• Vijiaratnam et al. (2025). Exenatide once a week versus placebo as a potential disease-modifying treatment for people with Parkinson’s disease in the UK: Phase 3 trial. “The Lancet”, 405, 627–636, link. 

[1] Athauda, D. et al. Exenatide once weekly versus placebo in Parkinson's disease: a randomised,  double-blind, placebo-controlled trial. Lancet 390, 1664-1675 (2017). https://doi.org/10.1016/S0140-6736(17)31585-4; Meissner, W. G. et al. Trial of Lixisenatide in Early Parkinson's Disease. N Engl J Med 390, 11761185 (2024). https://doi.org/10.1056/NEJMoa2312323 

[2] Vijiaratnam, N. et al. Exenatide once a week versus placebo as a potential disease-modifying  treatment for people with Parkinson's disease in the UK: a phase 3, multicentre, double-blind, parallel-group, randomised, placebo-controlled trial. Lancet 405, 627-636 (2025). https://doi.org/10.1016/S0140-6736(24)02808-3