New research finds that certain antidepressants and sleep medications significantly reduce Alzheimer’s progression in people with early memory loss. The finding was unexpected because these drugs were never designed to fight dementia.
The history of Alzheimer’s drug development is one of medicine’s most sobering stories. More than 99% of drugs specifically targeting the disease have failed in clinical trials. The first disease-modifying treatments to receive FDA approval, lecanemab and donanemab, remove amyloid plaques from the brain and produce modest slowing of cognitive decline, but they also cause brain swelling and bleeding in a significant proportion of patients and require intensive monitoring and infusion appointments. After decades of research and billions in investment, the treatments available for Alzheimer’s disease remain modest, expensive, and inaccessible to most of the people who need them.
A study published in Alzheimer’s & Dementia: Translational Research and Clinical Interventions, the official journal of the Alzheimer’s Association, has approached this problem from a completely different direction. Instead of testing new compounds, the research team asked what happens to Alzheimer’s risk in people who are already taking common medications originally designed for depression, sleep, and metabolic disease. The answer, for a specific subset of those drugs, was surprising.
Researchers led by Reina Tonegawa-Kuji at Stanford University used the MarketScan commercial claims database, a real-world dataset covering tens of millions of insured Americans, and identified two study populations. The first was people who had been diagnosed with mild cognitive impairment, the transitional stage between normal aging and dementia that represents the earliest clinically detectable sign that Alzheimer’s may be developing. The second was people over 70 who had no cognitive diagnosis but were at age-related risk.
From both populations, they identified individuals who were taking one of 54 drugs that had appeared in Alzheimer’s clinical trials, and asked whether their outcomes differed from people taking alternative drugs within the same medication class. This active comparator design is methodologically important: rather than comparing drug users to non-users, which can introduce bias from the reasons people take medications, it compares people who needed a medication and got different versions of it.
Among 54 drugs tested, five showed consistent, statistically significant reductions in Alzheimer’s progression or incidence.
The specific drugs and what the numbers showed
The most striking finding was for bupropion, sold under the brand name Wellbutrin, which is commonly prescribed for depression and as an aid to quitting smoking. In the mild cognitive impairment cohort, people taking bupropion had a hazard ratio of 0.57 for Alzheimer’s progression compared to people taking escitalopram, a widely used SSRI antidepressant. In practical terms, this represents a 43% lower risk of progressing to Alzheimer’s over a three-year follow-up period.
Trazodone, an older antidepressant frequently prescribed at low doses as a sleep aid, showed an 18% reduction in progression risk compared to sertraline, another SSRI. Venlafaxine, sold as Effexor and prescribed for depression and anxiety, showed a 28% reduction compared to escitalopram.
Zolpidem, sold as Ambien and one of the most commonly prescribed sleep medications in the United States, showed a 31% lower rate of Alzheimer’s progression compared to lorazepam, a benzodiazepine sedative. This finding requires careful interpretation. Zolpidem has been associated with elevated dementia risk in some long-term observational studies. The current finding does not suggest that zolpidem prevents Alzheimer’s in absolute terms; it suggests that, among people who need sleep medication and might receive either zolpidem or lorazepam, those who receive zolpidem show lower rates of Alzheimer’s progression than those who receive lorazepam. The comparison is within a class of sleep drugs, not against no treatment.
In the over-70 cohort with no cognitive diagnosis, liraglutide, the original GLP-1 receptor agonist in the same drug class as semaglutide (Ozempic, Wegovy), showed a 25% lower rate of Alzheimer’s incidence compared to metformin, the most common diabetes medication. This finding adds to a growing body of evidence suggesting that GLP-1 receptor agonists may have brain-protective effects beyond their metabolic actions.
Why these specific drugs might work
The four antidepressants and the sleep drug that showed the strongest signals share something that the SSRIs like escitalopram and sertraline do not: they all act on multiple neurotransmitter systems rather than primarily on serotonin alone.
Bupropion inhibits the reuptake of dopamine and norepinephrine, not serotonin. Venlafaxine inhibits both serotonin and norepinephrine reuptake. Trazodone blocks serotonin receptors in addition to inhibiting reuptake, producing a very different pharmacological profile from SSRIs. Mirtazapine, another drug that appeared in the consistent benefit group across both analyses, also works through multiple receptor systems.
The fact that SSRIs, which act primarily through serotonin, showed little benefit in this dataset, while drugs that affect dopamine, norepinephrine, and multiple receptor systems showed significant reductions in Alzheimer’s progression, suggests that the relevant mechanism may involve systems beyond serotonin signaling.
Dopamine and norepinephrine systems are known to be impaired early in Alzheimer’s disease. The locus coeruleus, the brain’s primary source of norepinephrine, shows significant neuronal loss in Alzheimer’s, often before amyloid plaques become detectable. Dopamine pathways involved in reward and motivation also deteriorate. Drugs that support these neurotransmitter systems may be providing compensatory protection that delays the functional consequences of early Alzheimer’s pathology.
For liraglutide and the GLP-1 class, the proposed mechanisms are different: GLP-1 receptors in the brain appear to promote neuronal survival, reduce inflammation, improve insulin signaling in neurons, and may directly influence amyloid and tau processing. The Alzheimer’s research community has been closely watching this drug class since observational studies began suggesting benefits, and the current dataset adds to the real-world evidence supporting further investigation.
What makes this method different from prior observational studies
The target trial emulation approach used in this study is a significant methodological advance over the simple observational analyses that have dominated repurposing research. Standard observational studies compare drug users to non-users, but people who take a medication differ from people who do not in many ways that are difficult to control for statistically. Someone prescribed an antidepressant may be cognitively different from someone who is not depressed and not taking antidepressants, regardless of what the drug itself does.
The active comparator design addresses this by comparing people who all needed treatment within the same drug category and received different drugs from that category. Everyone in the bupropion vs. escitalopram comparison needed an antidepressant. The factors that determined which drug they received are different from and more limited than the factors that determine whether someone takes an antidepressant at all. This produces a more reliable estimate of the drug’s specific effect rather than a composite of the drug effect and the selection effects that determine who takes it.
Inverse probability weighting was applied to further balance the comparison groups on demographic and clinical characteristics, making the analyses approximate the balance that would be achieved in a randomized trial without actually randomizing anyone.
The researchers describe their approach as trial emulation precisely because it is designed to produce an approximation of what a randomized trial would show, using existing claims data. It is stronger than naive observational analysis but not as definitive as an actual randomized controlled trial.
What the finding about SSRIs means
An important negative result in the study received less attention than the positive findings but may be equally significant. When the researchers looked at SSRIs as a class, including escitalopram, sertraline, and related drugs, they found no consistent reduction in Alzheimer’s progression in the MCI cohort.
This finding is significant because it is SSRIs that have most commonly been proposed as potential Alzheimer’s treatments in prior research, based on the role of serotonin in brain function and some early-phase trial data. The SELECT-SSRI trial, which tested escitalopram for Alzheimer’s prevention in amyloid-positive cognitively normal adults, is currently ongoing and its results are awaited.
The current data suggest that if SSRIs have any protective effect against Alzheimer’s progression, it may be smaller than the effects seen with the multi-system drugs that outperformed them in this analysis. This has implications for how the field prioritizes follow-up trials. Rather than assuming that any antidepressant might help, the data point toward specific mechanistic profiles, particularly those involving dopamine and norepinephrine rather than serotonin alone, as the most promising directions.
The path from a database finding to clinical change
The researchers are careful to position their findings as hypothesis-generating rather than practice-changing. This is a real-world evidence study, not a randomized controlled trial. Despite the methodological strengths of the target trial emulation design, residual confounding may explain some of the associations. People who receive bupropion instead of escitalopram may differ from their comparison group in ways that the available data cannot fully capture.
The appropriate next step from a finding like this is a randomized trial that tests the most promising drugs head-to-head in people with mild cognitive impairment, measuring Alzheimer’s progression as the primary outcome. Such trials take years and require substantial funding. The value of the current study is that it provides a prioritized list of candidates with supporting real-world evidence, which is exactly the kind of signal that funding agencies and trial designers need to decide where to invest.
Prazosin, an older blood pressure drug that appeared in the consistent benefit group alongside bupropion, trazodone, venlafaxine, zolpidem, and mirtazapine, has already attracted attention from Alzheimer’s researchers for different reasons: it blocks a specific receptor implicated in stress-related memory impairment. Its appearance in this dataset alongside drugs with plausible neurotransmitter mechanisms adds another data point to what is becoming a credible hypothesis about non-serotonergic pathways in Alzheimer’s protection.
For the millions of people currently taking these drugs for depression, anxiety, and sleep problems, the finding is not a recommendation to start or continue any medication for Alzheimer’s prevention. It is a signal that the drugs they are already taking may be doing something for their brain health that nobody anticipated, and that understanding why could open new directions in a field that has needed them.
The study, “Critical evaluation of real-world evidence of repurposable medicines in the Alzheimer’s disease drug development pipeline using a target trial emulation”, was authored by Reina Tonegawa-Kuji and colleagues at Stanford University and collaborating institutions, and published in Alzheimer’s & Dementia: Translational Research & Clinical Interventions in January 2026.
Source: Stanford University. DOI: 10.1002/trc2.70193
