Lithium carbonate has been the gold standard for the treatment of bipolar disorder since the 1950s. For nearly two decades, researchers have also wondered if this old, cheap drug might slow neurodegeneration as well.
From observational studies in people treated with lithium for psychiatric illness to epidemiological data of the drug in drinking water, research has long tied the substance to lower dementia rates. A 2025 Nature paper even hinted at a potential role for lithium in Alzheimer’s disease (AD) by suggesting the disorder may begin with a fundamental lithium deficiency in the brain. But the data still weren’t strong enough to justify a larger clinical trial to measure the substance’s neuroprotective qualities.
Researchers had hoped that results from a pilot project — the first prospective randomized controlled trial to study the effects of lithium on memory and brain aging — would finally settle the question. However, findings published earlier this year were mixed.
The trial missed every one of its six primary endpoints, which spanned cognition, brain volume, and a plasma biomarker. Still, there was one glimmer: Verbal memory declined at roughly half the rate in older adults who received low-dose lithium compared to those who received a placebo. While those results were promising, the data fell short of statistical significance.
“Our results were encouraging but not definitive,” lead investigator Ariel Gildengers, MD, professor of psychiatry at the University of Pittsburgh School of Medicine, told Medscape Medical News. “What lithium appears to do — if the signal holds up — is slow deterioration.”
Researchers say the findings can inform future trials, while outside experts have questioned whether the signal they observed holds up at all. The field had hoped for answers. Instead, it seems that there are just more questions.
A Neuroprotective Effect?
More than a decade ago, Gildengers observed signals of better brain health among older bipolar patients on long-term lithium. Those findings raised a question: Might lithium’s apparent neuroprotective effects extend beyond mood disorders, and could they be tested in people with mild cognitive impairment (MCI) who had no psychiatric diagnosis?
To find out, Gildengers designed the Lithium as a Treatment to Prevent Impairment of Cognition in Elders (LATTICE) trial in adults aged 60 and older with MCI. They randomized 41 patients to lithium (mean age, 72.9 years; 56% female) and 39 to placebo (mean age, 71.2 years; 56% female). Participants received treatment for 2 years, with a mean daily dose of approximately 195 mg of lithium carbonate — roughly one-fifth the standard psychiatric dose — producing serum levels around 0.17 mEq/L.
When researchers analyzed the data, they found no meaningful group differences in cortical gray matter, visuospatial memory, a composite cognitive score, and plasma levels of brain-derived neurotrophic factor (BDNF).
Of the six coprimary outcomes, only verbal memory showed a nominally significant treatment-by-time interaction. Scores on the California Verbal Learning Test-II declined 1.42 points per year in the placebo group vs 0.73 in the lithium group (P = .05), though this did not meet the study’s prespecified significance threshold of P < .01. Hippocampal volume trended favorably (P = .09) but also did not reach significance.
The trial did confirm that low-dose lithium was safe and well tolerated, with serious adverse events in 29% of the lithium group vs 23% on placebo.
“This does not mean lithium improves memory or reverses cognitive impairment,” Gildengers said. “At best, it may slow decline under certain conditions.”
Questioning Signal Strength
But Lon Schneider, MD, who wasn’t part of LATTICE, has a hard time looking past the six missed primary outcomes.
“The positive evidence is still lacking. This kind of profile could come out if you used a wholly inactive ingredient,” Schneider, a professor of psychiatry, neurology, and gerontology at the University of Southern California Keck School of Medicine, Los Angeles, told Medscape Medical News.
And the verbal memory finding, he added, is statistically fragile. “If you did the study again with a different group of 80 people, there’s roughly a 50% chance you just wouldn’t see this effect.”
The study also had a number of methodological problems, said Orestes Forlenza, MD, PhD, a professor of psychiatry at the University of São Paulo, Brazil. Among them, the inclusion of nonamnestic MCI patients, a low proportion of amyloid-positive participants, and a low lithium dose.
“I think the results were not as good as expected,” Forlenza told Medscape Medical News. The null BDNF finding was particularly surprising, he added, given that lithium has increased the factor in virtually every prior experimental model tested, including in a study he led in patients with amnestic MCI.
For that trial, Forlenza and colleagues followed 61 patients over 4 years, using roughly double the serum lithium levels that LATTICE achieved. The findings showed what LATTICE did not: that lithium was reaching the brain and was associated with an approximately 50% inhibition of glycogen synthase kinase-3 beta — the enzyme whose inhibition by lithium is considered central to its proposed neuroprotective mechanism. The study also tracked cerebrospinal fluid (CSF) biomarkers of AD pathology over 36 months.
In addition, Forlenza found a significant increase in CSF amyloid-beta 42 that suggested improved amyloid clearance, though an earlier reduction in phosphorylated tau at 12 months did not persist at 36 months. Those changes indicated the drug was acting on both molecular hallmarks of AD, providing evidence of disease modification, not just symptom management.
Designing the Next Study
Gildengers said he is planning a larger trial informed by what they learned from LATTICE, especially the need to enroll participants based on blood biomarkers of AD pathology. “If we were designing this study today, we would enroll participants based on amyloid status from the start,” he said.
When LATTICE was conceived nearly a decade ago, blood-based AD biomarkers like plasma p-tau217 had not yet been identified. Participants were enrolled based on clinical symptoms alone. Only about a quarter turned out to be amyloid-positive. About one-fifth had nonamnestic MCI, a subtype less commonly linked to AD progression.
If lithium’s mechanism of action is specific to AD biology, testing it in a mostly non-AD population dilutes the signal, Forlenza said.
“You actually need Alzheimer’s disease pathology to find the effect from lithium. It doesn’t work if you don’t have these pathophysiological abnormalities in your patient population,” he noted.
In LATTICE’s exploratory analyses of amyloid-positive completers, lithium effect sizes were larger: Hedges g was 0.74 for verbal memory and 0.82 for hippocampal volume, compared with 0.32 and 0.09, respectively, in amyloid-negative completers. Effect sizes in this range are considered medium to large, meaning lithium’s apparent benefit was concentrated almost entirely in participants who had AD pathology, and largely absent in those who did not.
The study wasn’t powered to compare results based on amyloid positivity, but the results are informative for the next trial design, Gildengers noted.
Other Challenges to Address
The next trial may also need to consider a different lithium formulation. Findings from a 2025 Nature study revealed that amyloid plaques carry a negative charge that attracts free lithium ions, trapping them before they reach neurons. The lithium carbonate used in LATTICE dissociates rapidly in the bloodstream, making its lithium ions particularly vulnerable to this sequestration.
But lithium orotate, the formulation used in the Nature study, remains bound to its carrier longer, allowing the delivery of lithium directly to neurons, said study investigator Bruce Yankner, MD, PhD, a professor of genetics and neurology at Harvard Medical School, Boston.
“Lithium carbonate was a reasonable choice when LATTICE was initiated in 2018,” Yankner told Medscape Medical News. “In our 2025 study, however, lithium carbonate was the least active of all the lithium salts tested.”
Another consideration might be to do a dose-ranging study before launching a larger trial, Forlenza suggested. This step would measure central nervous system penetration to establish if enough lithium is reaching the brain to engage its molecular targets.
But there’s an even greater challenge in lithium studies: funding. Because lithium is generic, no company stands to profit from proving it works. The entire burden falls on public funding.
“We really need a multicentric study, as we would do with any new pharmaceutical compound,” Forlenza said. “But who will pay for that?”
Clinical Takeaways
One thing Gildengers, Schneider, and Forlenza agree on is that the research to date does not support the drug’s clinical use beyond its approved indication, which does not include MCI or dementia.
“Patients should not self-medicate with over-the-counter lithium supplements. Lithium is a potent medication that requires careful dosing and close monitoring of kidney and thyroid function,” Gildengers said.
That warning comes as the lithium orotate supplement market continues to expand — valued at an estimated $411 million in 2024 and projected to nearly double by 2032. The supplements, sold online without prescription, are marketed for mood support, cognitive clarity, and brain health.
A 2025 survey published in the Canadian Journal of Psychiatry found that among 211 adults self-reporting use of over-the-counter lithium supplements, cognitive improvement was the most commonly experienced benefit, and side effects and withdrawal symptoms were more prevalent than anticipated.
Forlenza cautioned that the evidence does not yet support self-medication. “People take some experimental evidence and epidemiological data, and they decide it’s good to take lithium,” Forlenza said. “But it’s too early. We don’t have enough consistent evidence.”
The study was supported primarily by the National Institute on Aging (R01 AG055389). Gildengers reported receiving honoraria for invited lectures, according to the study’s published disclosures. Schneider, Forlenza, and Yankner reported no financial relationships relevant to this article.
https://www.medscape.com/viewarticle/verdict-low-dose-lithium-cognitive-preservation-2026a1000esx
