Low-Dose Lithium Shows Early Promise for Slowing Memory Decline in MCI
A pilot RCT found low-dose lithium slowed verbal memory decline by ~50% vs placebo in older adults with mild cognitive impairment, though results didn't reach statistical significance.
Summary
The LATTICE pilot trial randomized 80 adults aged 60+ with mild cognitive impairment (MCI) to two years of low-dose lithium carbonate or placebo. While none of the six coprimary outcomes met the prespecified significance threshold, verbal memory (CVLT-II delayed recall) declined at 1.42 points/year in the placebo group versus 0.73 points/year in the lithium group—roughly a 50% slower rate. The trial established feasibility and safety, with no serious adverse events definitively linked to lithium. Exploratory analyses hinted at potentially larger benefits in amyloid-positive participants. The authors conclude this pilot provides crucial effect size estimates to power a definitive confirmatory trial, particularly targeting amyloid-positive MCI.
Detailed Summary
Alzheimer disease (AD) remains a leading cause of dementia globally, and disease-modifying treatments at the MCI stage represent a critical prevention opportunity. One underexplored hypothesis holds that lithium deficiency—potentially caused by sequestration of lithium by amyloid plaques—may contribute to AD pathogenesis. Lithium inhibits GSK-3α/β, a kinase involved in tau phosphorylation and amyloid processing, and upregulates brain-derived neurotrophic factor (BDNF), both plausible neuroprotective mechanisms. Prior meta-analyses of small RCTs suggested lithium reduces cognitive decline in AD and MCI, but no trial had combined cognitive, neuroimaging, and plasma biomarker endpoints in a rigorous prospective design.
The LATTICE trial was a single-site, randomized, double-blind, placebo-controlled pilot feasibility study conducted at the University of Pittsburgh from 2018 to 2024 with two-year follow-up. Of 170 individuals screened, 83 were randomized (41 lithium, 42 placebo) and 80 initiated treatment. Participants were adults aged 60 or older meeting Petersen criteria for MCI, free of major psychiatric or neurological illness, and without contraindications to lithium. Six coprimary outcomes were prespecified: CVLT-II delayed recall, Brief Visuospatial Memory Test-Revised (BVMT-R), preclinical Alzheimer cognitive composite (PACC), hippocampal volume, cortical gray matter volume, and BDNF. A Bonferroni-corrected significance threshold of P ≤ .008 was applied. Linear mixed-effects models were used in an intention-to-treat framework.
None of the six coprimary outcomes reached the prespecified significance threshold. However, the verbal memory finding was directionally compelling: CVLT-II delayed recall declined 1.42 points per year in the placebo group versus 0.73 points per year in the lithium group (difference: 0.69 points/year; 95% CI, 0.01–1.37; P = .05). Hippocampal and cortical gray matter volumes declined in both groups with no significant treatment-by-time interaction. BDNF levels also did not differ significantly between groups. Exploratory subgroup analyses suggested potentially larger cognitive benefits among amyloid-positive participants, an important signal for future trial design.
The safety profile was reassuring. Serious adverse events occurred in 29% of lithium-treated participants versus 23% of placebo participants; none were definitively attributed to lithium. Common adverse events included increased creatinine (29% lithium vs 31% placebo), diarrhea (29% vs 15%), tiredness (29% vs 15%), and tremor (24% vs 15%). One death occurred in the placebo arm. Notably, serum lithium levels were maintained in a low, sub-therapeutic range (target 0.25–0.5 mEq/L), well below the standard therapeutic range for bipolar disorder, which contributed to the favorable tolerability.
The LATTICE trial is the first prospective RCT to integrate cognitive, neuroimaging (including 7T MRI), and plasma AD biomarker endpoints in a lithium MCI trial. Its primary value lies in establishing feasibility, confirming tolerability at low doses, and generating the effect size estimates needed to power a definitive Phase 3 trial. The authors specifically recommend that future trials enrich enrollment for amyloid-positive participants and consider CVLT-II delayed recall as the primary endpoint, given the observed signal. The pilot nature and small sample size preclude definitive conclusions about efficacy.
Key Findings
- Verbal memory declined ~50% more slowly with lithium (0.73 pts/yr) vs placebo (1.42 pts/yr), though P=.05 did not meet the corrected threshold.
- None of 6 coprimary outcomes (memory, visuospatial memory, PACC, hippocampal volume, cortical volume, BDNF) reached Bonferroni-corrected significance.
- Exploratory analyses suggested potentially larger cognitive benefits in amyloid-positive participants.
- No serious adverse events were definitively attributed to lithium; tolerability was acceptable at sub-therapeutic serum levels (0.25–0.5 mEq/L).
- The trial established feasibility and generated effect size estimates to power future confirmatory trials in amyloid-positive MCI.
Methodology
Single-site, randomized, double-blind, placebo-controlled pilot RCT (n=80) over 2 years at University of Pittsburgh. Six coprimary endpoints analyzed via linear mixed-effects models (ITT); Bonferroni-corrected significance threshold P≤.008 applied for multiplicity. Target serum lithium level was sub-therapeutic at 0.25–0.5 mEq/L.
Study Limitations
The pilot was underpowered (n=80) and conducted at a single site, limiting generalizability and statistical confidence. Six coprimary outcomes required strict Bonferroni correction, increasing the risk of false negatives in a small sample. Amyloid status was not used as an enrollment criterion, potentially diluting treatment effects in an enriched high-risk subgroup.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.