The short answer
Lithium deserves to stay in Alzheimer’s research. It does not yet deserve a place in your prevention routine.
It’s a familiar story: a mouse study in Nature, a small human trial that meets none of its primary endpoints, and a few weeks later, bottles of lithium orotate “to protect your brain.” In between, the step that matters is missing: evidence in humans.
The LATTICE trial published in 2026 offers an interesting signal in older people with mild cognitive impairment. But it demonstrates neither prevention of Alzheimer’s disease, nor an overall cognitive benefit, nor the effectiveness of lithium orotate sold as a supplement. None of its six primary outcomes reached the prespecified statistical threshold.
If you already take prescription lithium for a mood disorder, do not change your dose or treatment based on these data. If your memory is declining, a medical assessment can look for treatable causes and characterize the problem far better than a personal supplement experiment.
Why lithium is back in the news
The renewed interest does not come from a single discovery. Three different publications have been lumped together in headlines, even though they do not answer the same question.
In 2025, a study in Nature found less available lithium in certain regions of postmortem human brains with mild cognitive impairment or Alzheimer’s. The researchers then reduced dietary lithium in mice and tested lithium orotate in mouse models. Deficiency worsened several markers of pathology, while orotate improved markers and memory test performance in mice. This is an important mechanistic lead, not proof that humans can avoid Alzheimer’s by swallowing lithium.
In March 2026, the randomized LATTICE trial provided more directly relevant human data. Then a narrative review in JAMA Psychiatry and a perspective in Translational Psychiatry proposed new trials targeting early stages, physiological doses and patients with biomarker-confirmed disease. These two articles lay out a research agenda. They are not new efficacy trials.
The analysis published by Peter Attia on July 25, 2026 put LATTICE back at the center of the conversation, despite its negative primary outcome results.
Three types of exposure you should not confuse
The word “lithium” hides very different interventions:
| Exposure | Where it appears in the data | What we can conclude |
|---|---|---|
| Endogenous or environmental lithium | Measurements in the brain, blood, diet or water | Associations and a deficiency hypothesis, with no established preventive dose for humans |
| Lithium carbonate | Prescription medication and the form tested in LATTICE | A low dose under supervision can be studied, but its effectiveness against cognitive decline has not been demonstrated |
| Lithium orotate | Form used in mice and sold as a supplement | Animal results prove neither effectiveness nor long-term safety in humans |
The mass of the salt on a label is not the mass of elemental lithium. More importantly, a so-called “micro” dose does not eliminate interactions or variability in absorption and elimination.
A small imaging study detected lithium in the brains of nine healthy men after 5 mg per day of orotate for up to 28 days. It addressed a measurement question, not a question of effectiveness or long-term safety. A dedicated orotate trial in biomarker-confirmed early Alzheimer’s is registered for 40 participants, with an estimated start in October 2026. It therefore has no clinical results yet.
What the LATTICE trial really shows
LATTICE was a randomized, double-blind pilot trial conducted at a single center in Pittsburgh. It included adults aged 60 or older with mild cognitive impairment but no dementia. Of 83 people randomized, 80 started the intervention: 41 in the lithium group and 39 on placebo.
Participants received lithium carbonate, initially 150 or 300 mg depending on their medical condition and medications. The dose was adjusted according to tolerability. Among participants who completed the study, the average was 195 mg of carbonate per day and the average blood lithium level was 0.17 mEq/L. These figures describe the trial. They are not a dosing regimen to copy.
The level of supervision shows why: weekly visits during dose adjustment, then roughly quarterly visits, with checks of blood lithium levels, kidney function, thyroid, parathyroid glands, electrolytes and sometimes an electrocardiogram. In other words, “low dose” did not mean “just a supplement.”
The six primary outcomes covered three cognitive tests, hippocampal volume, cortical gray matter volume and blood BDNF. The central result fits into one sentence: none of the six reached the significance threshold set at p below 0.01 to account for multiple comparisons.
A verbal recall test nevertheless produced the signal fueling interest. Its score fell by an average of 1.42 points per year on placebo, compared with 0.73 on lithium. The difference of 0.69 points per year had a p-value of 0.05 and a 95% confidence interval from 0.01 to 1.37. It would have looked positive in an isolated analysis, but it did not cross the threshold specified in the protocol.
Exploratory analyses also suggested larger effects in participants with detectable amyloid. The problem: only 21 participants were amyloid-positive at baseline, and the analysis involved an even smaller number of study completers. This observation helps design the next trial, not select patients for treatment today.
Human data remain conflicting
In 2011, a Brazilian trial randomized 45 people with amnestic mild cognitive impairment over 12 months. A target blood lithium level of 0.25 to 0.5 mmol/L was associated with a reduction in phosphorylated tau and better performance on certain tests. However, the sample remained very small.
By contrast, a meta-analysis published in 2026 pooled six randomized trials involving 435 participants with mild cognitive impairment or Alzheimer’s. Lithium did not significantly slow cognitive decline compared with placebo in the main analysis: a standardized difference of -0.21, with a confidence interval from -0.49 to 0.07. Subgroups by diagnosis, duration, dose or salt form did not produce a robust result. The authors conclude that efficacy has not been demonstrated.
A second meta-analysis of six trials and 394 participants reached a similar assessment: no consistent benefit for global cognition, function, neuropsychiatric symptoms or biomarkers. The evidence remains insufficient to recommend lithium for Alzheimer’s.
The mechanisms are plausible, not validated as a treatment
Lithium acts on several pathways relevant to neurodegeneration. It can inhibit GSK-3β, an enzyme involved in tau phosphorylation, and modulate BDNF signaling, mitochondrial function, oxidative stress, inflammation and synaptic plasticity. Recent work adds hypotheses involving the cholinergic system, myelin and brain connectivity.
But a mechanism is clinically useful only if an intervention improves a meaningful human outcome. In LATTICE, BDNF did not differ between groups, brain volumes decreased in both groups without a significant interaction, and the planned GSK-3 analysis failed quality control. The chain of “appealing mechanism, altered biomarker, preserved memory” is therefore not established.
The study in Nature deserves the same discipline. Reduced lithium was observed in human brain tissue after death. This does not establish with certainty whether it precedes the disease, results from it or whether the two processes reinforce each other. The most striking interventional findings involved mice. Moving from this model to a preventive supplement for humans would be extrapolation, not clinical translation.
What remains unknown
Before we can talk about prevention or treatment, several answers are missing:
- a benefit for independence, progression to dementia or a global cognitive measure
- an adequately powered trial in patients with biomarker-confirmed Alzheimer’s pathology
- the lithium dose, duration and form offering the best balance between brain exposure and risk
- kidney, thyroid, cardiac and metabolic safety over several years at very low doses
- independent replication of the brain lithium deficiency hypothesis
- randomized data in cognitively healthy people, which currently do not exist
The term “prevention” therefore goes too far. LATTICE studied people who already had mild cognitive impairment, most without detectable amyloid, and did not demonstrate prevention of dementia.
Why self-medication is a bad experiment
In France, lithium carbonate is a prescription medication indicated for certain mood disorders, not for Alzheimer’s prevention. Its official summary of product characteristics calls for an individualized dose based on blood lithium levels and regular checks of kidney function, thyroid, calcium, heart and fluid and electrolyte balance.
Pregnancy requires specialist advice, and breastfeeding is contraindicated with Téralithe.
The kidneys eliminate lithium. Dehydration, fever, vomiting, a marked change in salt intake or kidney disease can therefore alter exposure. Common medications can raise blood lithium to toxic levels, including certain nonsteroidal anti-inflammatory drugs, diuretics, angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers. Neurological or serotonergic interactions also occur with other treatments.
Signs that require prompt medical advice include:
- unusual nausea or diarrhea
- new or worsening tremors
- intense thirst and very large amounts of urine
- weakness, drowsiness, confusion or balance problems
Marked confusion, an inability to walk normally, seizures or impaired alertness warrant urgent care. These risks do not prove that a very low dose of orotate causes side effects at the same rate as a psychiatric dose. They show why a lack of data cannot be turned into a guarantee of safety.
The rational decision today
If you are cognitively healthy, do not take lithium to prevent Alzheimer’s outside a clinical trial. Vascular risk factors, physical activity, sleep, hearing, smoking, alcohol and diet offer better-supported ways to act. The MIND diet · FR puts this into practice on your plate, and the guide to the side effects of lithium orotate · FR details the supplement’s risks.
If you notice memory decline, seek an assessment rather than covering up uncertainty with a supplement. A sleep disorder, depression, hypothyroidism, a deficiency, an anticholinergic medication or a neurological cause do not call for the same treatment.
If a doctor already prescribes you lithium, continue your scheduled monitoring. LATTICE does not change the medication’s psychiatric indications or provide a basis for adjusting its dose for cognition.
The Verdict
The scientific bet is defensible: test lithium at an early stage of Alzheimer’s, in better-selected patients, with a low dose and structured monitoring. The personal gamble is not—yet. This uncertainty justifies research, not preventive supplementation.
My confidence is high on three points: LATTICE did not demonstrate efficacy on its primary outcomes, its findings do not apply to prevention in healthy adults, and monitored carbonate does not validate orotate supplements. My confidence remains low on the most interesting question: can very low, carefully targeted exposure slow biologically defined early Alzheimer’s? Only a larger trial can answer that.
For now, lithium is a clinical hypothesis to follow, not a protocol to try.
