GLP-1s have been a breakthrough for obesity and diabetes. A 2022 study examined data from trials of GLP-1s in nearly 16,000 patients with diabetes, and showed that the drugs were associated with a 53% reduction in dementia risk.

That promising signal was the impetus for the EVOKE and EVOKE+ trials, a 3-year initiative that included two studies and 3800 patients with Alzheimer’s disease in 40 countries that explored whether the GLP-1 semaglutide could slow disease progression and cognitive decline.

Then came the disappointing topline results in November: Although patients treated with semaglutide showed a 30% decrease in C-reactive protein (CRP) and other markers of inflammation, these changes did not translate into cognitive or functional benefit.

To explore the findings and understand what happens when a promising therapy fails to deliver the hoped-for results, Medscape Medical News spoke with Jeffrey Cummings, MD, principal investigator of the EVOKE trials and director of the Chambers-Grundy Center for Transformative Neuroscience at the University of Nevada, Las Vegas.

What was the inspiration for the EVOKE trials?

The original observation was that people with diabetes treated with semaglutide were less likely to experience the onset of Alzheimer’s disease or [other types of dementia?] dementia. That was followed up with registry studies that showed the same thing. And eventually that was followed up by studies in animal models and organoids that reinforced the biology. It all came together to be enough to suggest that the drug might be effective in symptomatic Alzheimer’s disease.

What was the most surprising finding?

The negative outcome. It wasn’t that there was a small effect and it wasn’t statistically significant, there was really no drug-placebo difference. So it was a highly informative study because the hypothesis was that if we reduced peripheral inflammation with semaglutide, we would demonstrate benefit in Alzheimer’s disease progression because of the commerce between peripheral and central inflammation. We succeeded in reducing peripheral inflammation — the CRP was reduced by over 30% and was robustly statistically significant. However, that wasn’t enough to stop cognitive decline related to Alzheimer’s disease.

What was the biggest challenge in conducting the EVOKE research?

It was a very large study. There were over 200 sites and 40 countries involved. So the operational challenges were substantial. Novo Nordisk, which ran this study, had great experience with diabetes. Those were large studies to determine cardiovascular endpoints, among other outcomes.

This research taught the Alzheimer’s disease community about how to conduct this type of large trial. Organizationally, they had meetings of regional councils. They got the principal investigators across Europe together and kept people on track through good communication. They involved patients and caregivers in the review of the protocols to determine whether there were operational elements that could be optimized from the patient’s point of view. It was terrific.

After the findings were released, was there any reaction that surprised or concerned you?

People were critical of some aspects, and these criticisms might well be legitimate. For example, the EVOKE trials were based on the initial observation that there was a decreased incidence of Alzheimer’s disease and dementia in patients with diabetes. But the EVOKE trials were in people with proven Alzheimer’s disease.

So it was a somewhat different population — the diagnosis was biologically confirmed, and participants were symptomatic, so they were further along in the disease course. The idea was, could we slow progression — not slow the development of symptoms in people who were cognitively asymptomatic. The observational findings from the diabetes research may not have translated directly into the interventional trial.

Preclinical studies in animal models and brain organoids demonstrated benefit, which provided a basis for optimism going into the trial. However, key biological differences — including those related to the blood-brain barrier — separate these models from humans. Ultimately, the preclinical findings were not predictive of clinical outcomes.

Can you share any insights from the trial that haven’t yet been reported?

We worried a lot about weight loss because this is a weight-loss drug, right? That’s why people take it. It would have been surprising if patients had not had weight loss, and we wondered whether it would be too much. In general, we don’t encourage weight loss in older patients with Alzheimer’s disease. It was very interesting because the weight loss is proportional to their baseline weight. If they had slight underweight, they lost around 1 kg, and if they had overweight, they could lose up to 15 kg — a lot of weight.

One finding that did surprise me was that 50% of patients in the trial had either overweight or obesity, which was much higher than I would have expected for an Alzheimer’s disease population. It may be that we just haven’t paid much attention to baseline weight until studying this drug. Patients were heavier than I anticipated, and the weight loss was substantial. However, it did not appear to be harmful — we did not observe an increase in falls or cardiovascular events — at least based on the available data so far.

What are the broader implications of this study?

One of the key implications in the field now concerns what these findings mean for other anti-inflammatory programs in development. We knew semaglutide doesn’t enter into the brain to a substantial extent and that it primarily targets peripheral inflammation. As a result, the EVOKE results suggest that targeting inflammation outside the brain may not be sufficient to alter the course of Alzheimer’s disease. Whereas therapies that directly enter the brain and reduce neuroinflammation would not necessarily be undermined by these findings.

These findings shift the focus toward brain-penetrate compounds, even though we know there is commerce between peripheral inflammation and central inflammation. In the EVOKE trial we absolutely reduced peripheral inflammation, but we didn't see a corresponding clinical benefit. Because inflammation is such an important part of Alzheimer’s biology and because anti-inflammatory approaches represent such a large portion of the therapeutic pipeline, we really need to better understand what these results mean for the development of anti-inflammatories for Alzheimer’s disease.

With the benefit of hindsight, are there elements of the trial design or execution you would reconsider?

I’ve asked myself that question a lot. Of course, we’d like to capitalize on these observations. One way to do it would be to focus on patients with diabetes, where cognitive impairment is very common and the population is large. We could essentially repeat this type of study in a diabetes population — enrolling individuals with diabetes who are cognitively unimpaired and randomize them to semaglutide or placebo. The question would be does semaglutide reduce the incidence of new-onset dementia in patients with diabetes?

I think the answer to that would be yes. But now we have blood testing for Alzheimer’s disease, we would be able to determine whether the effect is specifically on Alzheimer’s disease or on all-cause dementia, which probably has a substantial vascular component. We know that patients with diabetes are at an increased risk for small vessel disease in the brain. Another missing piece of the puzzle is the underlying cause of cognitive impairment in people living with diabetes whose risk appeared to be reduced with semaglutide. It may not have been just Alzheimer’s disease alone; it could reflect mixed pathology with a significant vascular contribution.

If you could design the next study in this space, what would it look like?

It would be valuable to study a brain-penetrant GLP-1 receptor agonist. GLP-1 receptors are not located only in the brainstem, so a compound that enters the brain could potentially have effects in regions such as the cortex and the hippocampus. That said, I’m not certain that even a brain penetrant GLP-1 receptor agonist would be beneficial in Alzheimer’s disease. But that remains an important and legitimate unanswered question.

Is there anything about the trial results that you feel hasn’t been fully understood?

Yes, a negative trial is a grieving process. People may not fully appreciate that this represents 5 years of commitment. Not every day, all day, but for 5 years I’ve served as the co-chair of the steering committee, really hoping that this would become something meaningful for my patients and for patients around the world.

We really need an oral medication — wouldn’t that be a major advance compared with intravenous monoclonal antibodies? So yes, it’s a terrific disappointment. I’m not sure that the public always appreciates the emotional investment scientists have in these projects. We are scientists but many of us are physicians too, and we want treatments for our patients. When you see negative data like these, you can’t help but think, “Wow, this is not what I wanted.”

The way I do it — and this is probably the way most scientists work — is that we start looking for the bright side. You can’t be in drug development without being an optimist. So you start asking: What did we learn about inflammation? What did we learn about the blood-brain barrier? What did we learn about trial execution? There were many positive aspects of this trial that we can build on. It helps the field and ultimately will help patients. But it just didn’t help the patients who were in this trial.

Dr Cummings reported that he is a member of the steering committee for the EVOKE+ trials.