select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
Two studies published in JAMA Neurology look at diabetes drugs and the risk of dementia and Alzheimer’s.
Comments on both studies:
Prof Atticus Hainsworth, Professor of Cerebrovascular Disease, St George’s, University of London, said:
“These two studies have looked at drug prescribing databases, to assess whether diabetic medications impact on risk of Alzheimer’s, vascular and other dementia types. The findings are intriguing, even though they are somewhat contradictory. But nothing can substitute for a prospective, hypothesis-testing experiment, which in this context means a prospective clinical trial.”
Prof Mark Evans, University Professor of Diabetic Medicine & Honorary Consultant Physician, Institute of Metabolic Science & Department of Medicine, University of Cambridge, said:
Background:
“It is increasingly apparent that there is an association between diabetes and increased risk of dementia. Although it is not clear how this association is mediated (and whether this is to do with elevated blood glucose from diabetes or a broader effect of diabetes for example on the circulation or inflammation pathways), the obvious pragmatic question arising is whether or not treating diabetes with glucose lowering therapies can reduce risk of dementia and importantly whether certain types of glucose lowering therapy are particularly effective.
“Newer glucose lowering treatments used in (type 2) diabetes seem to carry additional health benefits beyond just glucose lowering. For example, treatments that act via signalling at the GLP-1 receptor or by blocking SGLT glucose transporter channels have been shown to carry broader protective effects for the heart and kidneys.
“Some data already exist to suggest that these agents may also have brain-protective effects against the development of dementia. The evidence to date has largely fallen into two different types. Firstly, examining clinical trials of glucose lowering therapies where usually dementia is not the primary focus but in the clinical trials, investigators will document all health changes including a new diagnosis or change in dementia or cognition. Secondly, examining large real-world datasets for the association between diabetes, different types of therapy and a clinical record of dementia.
These two papers:
“These two papers cover each of these areas respectively. Seminer2 et al have performed an analysis of clinical trials broadly similar to ones previously reported. They found that glucose-lowering therapies in general were not significantly associated with a reduction in dementia, although when comparing different types of therapy, GLP-1R targeted drugs but not those acting on SGLT channels were associated with a reduction in dementia. There are cautions arising from these data and the authors have acknowledge these appropriately. Overall, the absolute rates of dementia reported in the studies contributing to their analysis were relatively low which thus reduces the ability for this type of analysis to identify differences. There may be a number of reasons for this, for example the clinical trials were not designed to look in detail for possible changes in dementia.
“The other paper from Tang1 et al was a real-world analysis. Consistent with previous data, this did seem to show less dementia in those using either GLP-1R agonists or SGLT2 inhibitors. The authors were aware of the limitations of observational data. In other words, it may be that the type of person receiving these agents was different from the type of person prescribed other glucose lowering therapies. As an example, one obvious potential confounder from their paper was that the groups receiving GLP 1 agonist or SGLT inhibitor therapy were younger than the comparison groups.
General Reflection:
“Overall these papers whet the appetite for two large clinical trials (EVOKE and EVOKE+) that will complete towards the end of this year. These were specifically designed to look at whether semaglutide, a GLP 1 receptor agonist, can reduce progression of Alzheimer’s dementia and may give us a more definitive answer, at least for GLP 1 receptor agonist treatment.”
Prof David Strain, Associate Professor in Cardiometabolic Health, and a clinical geriatrician, University of Exeter, said:
“Living with diabetes is associated with twice the rate of decline towards both vascular and Alzheimer’s Type dementia. Although long-term high blood sugar is recognised to be warm contributor to this, there are many others including genetics, the body’s ability to produce insulin, episodes of low blood sugar (hypoglycaemia) and inflammation. For years, the best way of preventing this was to attempt to keep the sugar in the “goldilocks zone“ that is not too high but not too low.
“GLP-1 receptor antagonists (such as semaglutide and dulaglutide) and SGLT-2 inhibitors (such as dapagliflozin and empagliflozin) have been demonstrated to control the sugar, and also reduce the inflammation (a key driver of Alzheimer’s disease) and reduce vascular risk (a key driver of vascular dementia) more than would be expected by the sugar control alone. It is therefore no surprise that these data show a lower risk of dementia in people who receive them as part of her routine care.
“We must be cautious how we interpret these data though. Firstly, there were very small numbers of events in the randomised control trials and these were not fully validated so there is a possibility that there were many other cases of dementia missed, indeed that patients reported as having Alzheimer’s type dementia may have had other problems. In the database study, we can never be certain of other unmeasured factors that influenced the doctor to prescribe one medication over another. These may also have had an impact on whether a person would progress to dementia or not.
“It is also important to say this is talking about the risk of dementia in people with diabetes. We have recently seen benefit of the GLP-1 RAs for cardiovascular health in people who do not have diabetes, likewise the SGLT 2 inhibitors are regularly used in other conditions. Today’s studies will need to be replicated prospectively, in people with and without diabetes such as the work we are performing at the University of Exeter, to determine if we can help reduce the progression towards dementia in many more people in the country.”
Prof Naveed Sattar, Professor of Cardiometabolic Medicine/Honorary Consultant, University of Glasgow, said:
“These two papers on potential for GLP-1RA medicines to lower dementia risk are somewhat encouraging but they are FAR from definitive, given the design of one is observational and the numbers of people with dementia in the trial meta-analyses were incredibly small, leading to limited power. Hence, we need to await the results of ongoing randomised trials in this area before drawing sensible conclusions. There is optimism GLP-1RAs (and related medicines) may lower future dementia risk, however, given they favourably impact multiple diseases (cardiovascular, hypertension, kidney and diabetes) known to increase dementia risk, it might be that it is having a lower risk of stroke /heart disease and diabetes that actually then is linked to a lower risk of dementia. Hopefully, far more robust trial evidence – needed before any clinical implications are drawn – should be forthcoming in the near future.”
Dr Ivan Koychev, Clinical Reader in Neuropsychiatry and Consultant Neuropsychiatrist, Imperial College London and Central North West London NHS Foundation Trust, said:
“Both papers are of good quality. Tang et al is a large epidemiological study; the limitation comes from the fact that such designs limit the ability to draw conclusions about causality. Seminer et al look at clinical trial data where dementia and cognitive effects were not the primary targets of the studies. Therefore, it is possible that some of these effects were missed or over-reported due to the opportunistic nature of the data collection.
“The two studies add to a growing and remarkably consistent body of evidence that GLP1 receptor agonists associate with a reduction in dementia incidence. The Seminer et al paper is significant as it shows that GLP-1 RAs outperform SGLT-2 inhibitors in clinical trial settings. This suggests that the dementia protection effects are not due to glucose control mechanisms. Instead, inflammation and cerebrovascular effects are likely involved. Overall, these data support the urgent exploration of GLP1 RAs as a preventative treatment in people at risk for dementia.”
Prof Tara Spires-Jones, Director of the Centre for Discovery Brain Sciences at the University of Edinburgh, Group Leader in the UK Dementia Research Institute, and President of the British Neuroscience Association said:
“The study by Tang1 and colleagues examined data from over 90,000 people with diabetes to determine whether treatment with two different glucose lowering drugs were associated with risk of developing dementia. Both glucagon-like peptide-1 receptor agonist (GLP-1RAs) and
sodium-glucose cotransporter-2 inhibitor (SGLT2is) drugs were associated with reduced risk of developing dementia over 8 years from starting treatment compared to people taking other glucose lowering treatments. There was no difference in risk between people taking the GLP-1Ras and SGLT2is. This is a strong study that adds to the growing data suggesting that diabetes increases risk for developing dementia.
“The paper from Seminer2 and colleagues examined data from 26 clinical trials to determine whether glucose lowering treatments were associated with developing dementia. They observed that GLP-1Ras but not but not SGLT2is were associated with a reduction in dementia in these trials.
“Together, these data are encouraging for the potential of using GLP-1Ras to lower dementia risk in people with diabetes, but even within these 2 strong studies, there are slightly conflicting results over SGLT2is highlighting the need for further research. It is important to note that these drugs do have side effects and that they are not guaranteed to prevent dementia. The studies had important limitations including a relatively short follow up time. Future work will be important to understand how risk factors like diabetes and obesity increase risk of dementia to develop effective treatments and prevention strategies.”
Dr Emma Anderson, Principal Research Fellow and Associate Professor of Epidemiology in the Division of Psychiatry, UCL, said:
“Regarding the systematic review of glucose lowering therapies2, my comments are as follows:
“These results should be interpreted with caution for several reasons. Firstly, although an effect of GLP-1RAs was identified for all cause dementia, the heterogeneity in the studies included in this analysis was high (meaning that the studies included are not necessarily comparable, undermining the validity of the combined results). There was tentative evidence that this heterogeneity could have been explained, at least in part, by the proportion of women included in these studies.
“Secondly, there was no evidence of an effect of GLP-1RAs with either Alzheimer’s disease or vascular dementia; the two most common causes of dementia. Thus, overall, there is still a question around whether GLP-1RAs would actually reduce dementia risk.
“For the emulated target trial1, my comment is:
“This study should be interpreted with caution, as emulated target trials are as susceptible to confounding by indication bias as traditional observational epidemiology studies. This means that there is a possibility that the results they have observed are actually due to the underlying reason people are prescribed these glucose-lowering medications in the first place, rather than the medication itself. More robust study designs, which overcome this very important limitation, are needed before such conclusions can be made.”
Prof Masud Husain, Professor of Neurology & Cognitive Neuroscience, University of Oxford, said:
“For me, these new retrospective analyses suggest that GLP-1 receptor agonists, particularly semaglutide, might reduce the risk of developing dementia in people with type 2 diabetes. But we need data from prospective trials to provide stronger evidence.
“The wider question of whether such drugs might also be protective against dementia in people who don’t have diabetes is a really intriguing one, and the focus of several ongoing clinical trials.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:
“I’ll restrict myself to pointing out a few things about these two studies, which may well all be obvious. I’m writing as a statistician and did not spot any important statistical flaws – but I can’t comment on non-statistical aspects.
“Both studies are only in people who already had type 2 diabetes. I believe there’s been wider interest in whether GLP-1 receptor agonist drugs (GLP-1RAs), such as semaglutide (e.g. Ozempic or Wegovy) might reduce dementia risk, in people who are taking them for other reasons than having diabetes. These studies can’t directly tell us anything about that – though the wider context discussed in the linked editorial by Dr Diana Thiara does make some points about the wider context. That editorial is definitely worth reading to make sense of all this, in my view.
“At first sight it might look like a sort of contradiction that one1 of the studies (Tang et al.) found evidence of a decreased risk of certain dementias in people with type 2 diabetes taking either GLP-1RAs or drugs of another class (SGLT2is) used to lower glucose in people with type 2 diabetes, compared to people with type 2 diabetes taking other glucose-lowering drugs, while the other2 study (Seminer et al.) found no evidence that passed the usual statistical criteria that people with type 2 diabetes taking any of GLP-1RAs, SGLT2is, or another drug called pioglitazone did reduce all-cause dementia. It looks as if one is saying that, in people with type 2 diabetes, taking GLP-1RAs and SGLT2is is associated with reduced dementia risk, and the other is saying that it isn’t associated with reduced dementia risk. But there’s no contradiction, for the following reasons:
“Therefore the two studies differ in terms of the treatments involved (taking the comparator, control, treatment into account), the type of study design, and the risk of bias. So direct comparison of their findings doesn’t make as much sense as you might think at first. Again, the Thiara editorial makes this all clearer, I think.
“Despite the overall finding of the Seminer2 meta-analysis that the three drug classes that it considered weren’t associated with a reduction in overall risk of dementia or cognitive impairment, on average, compared to controls taking a placebo, they found that GLP-1RAs (considered on their own, leaving out the other drug classes) were associated with a reduction in dementia risk.
“Tang found overall no difference between the risk of Alzheimer disease and related dementias between type 2 diabetes patients taking GLP-1RAs or SGLT2is, but it did find a lower risk in patients taking semaglutide compared to patients taking SGLT2is.
“Though both studies involved large numbers of patients overall, the number of patients who actually had a dementia diagnosis was pretty low, particularly for newer drugs, and the follow-up time was short, given how long it can take for dementias to develop. This is quite a big limitation.”
Dr Richard Oakley, Director of Research and Innovation, Alzheimer’s Society, said:
“Last year, the Lancet Commission in Dementia Prevention, Intervention and Care – funded by Alzheimer’s Society – highlighted diabetes as one of 14 risk factors for dementia.
“Drugs that lower blood glucose are often used as part of diabetes treatment, and research has shown they might also be effective in reducing dementia risk.
“Whilst both of these studies found a link between GLP-1RAs and reduced dementia risk, only one found SGLT2is, another class of diabetes drug, were also associated with a reduced risk. More research is needed to properly understand how diabetes treatments may lower the risk of dementia, by tracking people for longer, especially as they get older.
“Alzheimer’s Society is interested to hear results from ongoing clinical trials of the GLP-1RA drug semaglutide for people with early Alzheimer’s disease, to learn whether these drugs can slow the course of the disease.”
Comments on just the Tang et al study:
Dr Leah Mursaleen, Head of Clinical Research at Alzheimer’s Research UK, said:
“This study1 used health records from nearly 400,000 people who had type 2 diabetes who were taking drugs called SGLT-2 inhibitors and GLP-1RAs to control their condition. These drugs are usually taken to lower blood sugar levels in diabetes. The researchers compared these drugs with other standard medications used to manage diabetes to evaluate the risk of developing dementia.
“People who were taking the SGLT-2 inhibitors or GLP-1RA medications appeared to have a decreased risk of developing dementia compared with those taking other diabetes medicines.
“The results from this study support the growing evidence that GLP-1RA and SGLT-2 inhibitors medicines may be linked with a lower risk of developing dementia. We don’t know yet why these medicines may be protective, and will need more research to understand how they are affecting the brain.
“Clinical trials are already looking at the use of these types of drugs, including the EVOKE study, which is testing semaglutide (Ozempic) as a potential treatment for people with early Alzheimer’s.
“While the findings of this study are interesting, it’s important to consider whether any other factors might be influencing the results such as the severity of type 2 diabetes, health, income, and education.”
1: ‘GLP-1RA and SGLT2i Medications for Type 2 Diabetes and Alzheimer Disease and Related Dementias’ by Huilin Tang et al. was published in JAMA Neurology at 16:00 UK time on Monday 7 April 2025.
DOI: 10.1001/jamaneurol.2025.0353
2: ‘Cardioprotective Glucose-Lowering Agents and Dementia Risk A Systematic Review and Meta-Analysis’ by Allie Seminer et al. was published in JAMA Neurology at 16:00 UK time on Monday 7 April 2025.
DOI: 10.1001/jamaneurol.2025.0360
Declared interests
Prof Atticus Hainsworth: “I lead the Vascular Experimental Medicine team within DementiasPlatformUK:
Vascular Health — DPUK: https://www.dementiasplatform.uk/research-hub/experimental-medicine-incubator/vascular-health?68d44564-1335-11ed-b137-0aa7be39d6a6
No other conflicts.”
Prof Mark Evans: “I have received personal fees from Medtronic, Ypsomed, Dexcom, Abbott, Novo Nordisk, Eli Lilly, Sanofi, Zucara, Pila Pharma and research support from NovoNordisk, this includes a current PhD student who is funded by Novo Nordisk to examine the mechanisms underpinning the relationship between diabetes/ metabolic disease and dementia but does not include examining the effects of therapies. Abbott, Eli Lilly, Sanofi. The University of Cambridge has received salary support for MLE from the National Health Service in the East of England through the Clinical Academic Reserve.”
Prof David Strain: “I have received speaker fees from AstraZeneca (dapagliflozin) and Novo Nordisk (semaglutide). I lead the UK Stakeholders consensus document for the management of diabetes in older adults which the current UK guidelines are based.”
Prof Naveed Sattar “Has consulted for several companies that make diabetes medicines but also contributed to several lifestyle trials.
“For Novo Nordisk: have consulted for company in advisory boards but not on any of their weight loss drug trial committees; am on steering committee for ZEUS trial but this is not a weight loss trial product but anti-inflammatory. Do not have any shares either for any product in health etc.
“N.S. declares consulting fees and/or speaker honoraria from Abbott Laboratories, Afimmune, Amgen, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Hanmi Pharmaceuticals, Janssen, Merck Sharp & Dohme, Novartis, Novo Nordisk, Pfizer, and Sanofi; and grant support paid to his university from AstraZeneca, Boehringer Ingelheim, Novartis, and Roche Diagnostics.”
Dr Ivan Koychev: “IK has received speaker fees as well as an investigator initiated grant to explore the effects of semaglutide in people at risk for dementia from Novo Nordisk.”
Prof Tara Spires-Jones: “I have no conflicts with this study but have received payments for consulting, scientific talks, or collaborative research over the past 10 years from AbbVie, Sanofi, Merck, Scottish Brain Sciences, Jay Therapeutics, Cognition Therapeutics, Ono, and Eisai. I am also Charity trustee for the British Neuroscience Association and the Guarantors of Brain and serve as scientific advisor to several charities and non-profit institutions.”
Dr Emma Anderson: “I have no conflicts of interest.”
Prof Masud Husain: “I don’t have any conflicts of interest.”
Prof Kevin McConway: “Previously a Trustee of the SMC and a member of its Advisory Committee.”
Dr Richard Oakley: “No conflicts of interest to declare from Richard or the Society.”
Dr Leah Mursaleen: “Leah has no conflicts of interest to declare.”