GLP-1 Drugs Might Cut Dementia Risk in Half

This transcript has been edited for clarity. 

Welcome to Impact Factor, your weekly dose of commentary on a new medical study. I’m Dr F. Perry Wilson from the Yale School of Medicine.

It’s quite a thing that we live in an era with the biggest blockbuster weight-loss drugs in history, and I can look at you straight-faced and say, yeah, but honestly that’s not the most interesting thing about the GLP-1 receptor agonists such as Ozempic.

Beyond their clear utility in weight loss, we have good data that demonstrate the drugs lead to improvements in rates of heart disease, liver disease, and even all-cause mortality, and intriguing reports of strange off-target effects: a reduction in gambling, drinking, drug use, smoking, and compulsive shopping. It’s honestly hard to find an organ system or self-destructive behavior that GLP-1 drugs don’t seem to improve somehow. 

It’s almost strange. It’s as if the statins came out and everyone was excited about the cholesterol going down, and then we discovered that they reversed hair loss or took three strokes off your golf game. How does one drug do so many things?

I think the behavioral changes are the clue. The fact that people may drink or smoke less on these drugs should tell you that something is happening in their brains. And if these drugs are acting on the brain, you might well wonder if they can reduce the risk for brain disease — dementia. And, at least according to a new study, the answer appears to be yes.

As the GLP-1 drugs started getting more and more popular, I decided to perform some due diligence and refresh what I had learned in med school about this thing called “glucagon-like peptide.” These drugs bind to the same receptor that GLP does, primarily to cause the pancreas to release more insulin when blood sugar is high.

But we have plenty of drugs that stimulate insulin release. And they don’t seem to have all these other weird effects. So, what’s going on?

When a drug acts on a specific receptor, it is worth seeing what cells in the body have that receptor, and this is a pretty interesting figure that shows that.

You see high expression in the pancreas — no surprise. But next is expression in the heart muscle — interesting, considering these drugs have been shown to reduce the risk for heart disease. Then the salivary glands, and then… the cerebral cortex. The brain. And not just any part of the brain, the thinking part of the brain. 

A lot happens when these drugs bind to a receptor on a cell. But the brain is a mysterious and important organ. In contrast, the heart is relatively simple, and there are multiple randomized trials that show that GLP-1 drugs, as well as some other drugs for diabetes, such as the SGLT2 inhibitors, pioglitazone, and metformin, reduce the risk for cardiovascular disease in people with diabetes. So, it makes sense to ask whether these drugs would do the same for brain disease.

Enter this study, "Cardioprotective Glucose-Lowering Agents and Dementia Risk," from Catriona Reddin and colleagues in JAMA Neurology. It’s an impressive bit of work combining data from 23 different randomized trials of these drugs including more than 150,000 individual people, to determine whether they reduce the risk for dementia.

This is the best evidence we have to date for a neuroprotective effect of GLP-1 inhibitors. And this is the only type of study that could give us this evidence.

The reason you need to combine so much data to detect a signal is that, to date, no randomized trials of GLP-1 drugs have been designed with dementia as a primary outcome.

When you design a randomized trial, you need to pick a primary outcome — a single thing that you will use as the definitive test of whether your drug works or not. It could be anything, ranging from weight loss to death, but you have to decide in advance, to make sure there will be no cherry-picking of data. And once you decide, it behooves you as a researcher to make sure you have really good capture of that outcome. For example, if your primary outcome is weight loss, you want to make sure you are measuring weight very carefully, at a specific time of day, with excellent equipment.

Of course, if you’re enrolling all these people in a clinical trial, there’s nothing stopping you from tracking other outcomes. In fact, it would be silly not to, at least to suggest what outcomes you might want to use as a primary outcome in the next trial. Some outcomes are easier to track than others. Death is, well, relatively easy to determine. But dementia isn’t.

For a patient to receive a new diagnosis of dementia, quite a few things have to happen, and people can fall through the cracks at multiple steps in the path. They have to recognize that they might have a problem, and they have to seek help for the problem and undergo some fairly complicated tests. If dementia was the primary outcome of your clinical trial, you’d have a whole protocol to help people go through these steps. But if not, you’re likely missing cases. 

That’s certainly the case with some of the trials being analyzed here. In fact, most of the trials report dementia just as an adverse event — for example, in the HARMONY trial, a randomized trial of albiglutide with just under 10,000 participants. There were zero dementia adverse events in the albiglutide arm, compared with three in the placebo arm.

And as you scan down the other trials, that pattern repeats.

Zero cases vs one case. Three vs five. The study with the most dementia events had 12 in the GLP-1 group and 25 in the intervention group, which is still not a lot to work with.

But that’s the power of meta-analysis: You can combine all these small numbers and start to see a trend.

And that’s just what the authors show here. 

Overall, in trials of GLP-1 drugs, there was a 45% lower odds of dementia among those randomized to receive the GLP-1 drugs vs those who got placebo. That’s an impressive effect. It’s an effect that suggests a new route for dementia therapy.

The SGLT2 inhibitors, other very successful diabetes drugs which are protective of the heart, did not show such a benefit on the brain. The handful of dementia cases in those trials were evenly split between the treatment and placebo arms.

In terms of metformin, unfortunately no trials are reported in this meta-analysis. I was able to find some other studies that suggested there might be a benefit, but they don’t meet the criteria for this particular analysis.

So, how do we put this together? We know that diabetes increases the risk for dementia, likely because elevated sugar levels increase vascular disease. But if that’s the mechanism here, why is the benefit limited to the GLP-1 drugs? Why not a similar benefit with the SGLT2 inhibitors? To figure that out, we need to look at the differences between these two classes. One is the mechanism of action, and I pointed out earlier that cerebral cortex has a nice expression of GLP-1 receptor.

The other is in their other effects. Both classes of drugs reduce blood sugar, but only the GLP-1s have a profound effect on body weight. Perhaps the reduction in dementia risk is driven through that mechanism instead of through one mediated by sugar levels.

If that’s the case, a clear question emerges: Will these drugs prevent dementia in people without diabetes? There is no randomized trial data yet, but some trials, like the evoke and OxSENSE studies, may give us better answers. When those results come out, you can bet we’ll talk about them here.

F. Perry Wilson, MD, MSCE, is an associate professor of medicine and public health and director of Yale’s Clinical and Translational Research Accelerator. His science communication work can be found in the Huffington Post, on NPR, and here on Medscape. He posts at @fperrywilsonand his book, How Medicine Works and When It Doesn’t, is available now.