When evaluating levetiracetam (LEV; Keppra) for Alzheimer's disease (AD), the key question is not just "Does it enhance cognition?" but rather: Is there a specific Alzheimer's phenotype where circuit stabilization significantly alters its trajectory?

Over the last 15 years, levetiracetam has transitioned from a standard antiseizure medication to a potential precision neurophysiology intervention aimed at network hyperexcitability in early AD and mild cognitive impairment (MCI). Numerous trials have investigated its role in:

• Seizure-comorbid AD
• Subclinical epileptiform AD
• Amnestic MCI with hippocampal hyperactivity
• Broad, unselected MCI cohorts
• Genotype-stratified subgroups (APOE ε4)

The results are varied but not inconsistent. A comparative analysis reveals a pattern: LEV shows promise when hyperexcitability is present and is administered within a narrow therapeutic window.

This article integrates the historical development, mechanistic framework, and clinical trial landscape, providing a comparative interpretation across studies.

1. Historical Arc: From Racetams to Circuit Precision

1.1 The Racetam Origin

The racetam family emerged from Corneliu Giurgea's work on piracetam in the 1970s. The initial "nootropic" hypothesis suggested cognitive enhancement without sedation or psychostimulation. Piracetam was tested in dementia in small, varied trials, which were often underpowered and inconsistent. Meta-analyses ultimately determined that the evidence was insufficient for recommending routine use in dementia.

This background is crucial as many patients and some clinicians often equate racetams with "memory enhancers." However, levetiracetam’s modern rationale for AD is fundamentally different. It is not intended to stimulate cognition but to modulate pathological network overactivity.

1.2 From Etiracetam to Levetiracetam

Levetiracetam, the S-enantiomer of etiracetam, was discovered during antiseizure compound screening. Its binding to SV2A (synaptic vesicle protein 2A) distinguishes it mechanistically from traditional sodium channel-blocking antiepileptics.

FDA approval in 1999 established LEV as a well-tolerated anti-seizure agent with:

• Minimal hepatic metabolism
• Limited drug-drug interactions
• Renal clearance
• Favorable geriatric tolerability

This pharmacological profile made it appealing for dementia research, especially once the prevalence of seizures in AD was better understood.

2. Biological Framework: Hyperexcitability as an AD Phenotype

2.1 Seizure Overrepresentation in AD

Seizures are more common in AD, particularly in:

• Early-onset AD
• Autosomal dominant AD
• APOE ε4 carriers
• Advanced disease

Seizure presence correlates with faster cognitive decline. However, overt seizures are only a small part of electrophysiologic instability.

2.2 Subclinical Epileptiform Activity

Extended EEG and MEG studies have shown that subclinical epileptiform discharges are common in AD. Vossel et al. (2016) revealed:

• Increased epileptiform activity in AD vs. controls
• Association with accelerated cognitive decline
• Often undetected on routine EEG

This redefined AD as partially a disorder of abnormal network synchrony.

2.3 Hippocampal Hyperactivity

Functional MRI studies in amnestic MCI show paradoxical hippocampal hyperactivation during memory tasks. This hyperactivity, rather than being compensatory, appears maladaptive and predictive of progression.

Bakker et al. (2012) provided the first human interventional signal that low-dose LEV could normalize this hyperactivation and improve memory performance, becoming the foundation for later trials.

3. Trial Landscape: Comparative Review

3.1 Bakker et al., 2012 - aMCI, fMRI-Guided

Design: Randomized crossover
Population: Amnestic MCI
Dose: 125 mg BID (low dose)
Duration: 2 weeks

Findings:

• Reduced hippocampal hyperactivity on fMRI
• Improved memory task performance
• Dose-dependent effect (low dose effective; higher dose not beneficial)

Interpretation:

This was a mechanistic proof-of-concept, not a disease-modifying trial, demonstrating state normalization, not trajectory change.

3.2 LEV in AD with Epilepsy (2007-2010 observational studies)

These trials assessed LEV in AD patients with overt seizures.

Findings:

• Effective seizure control
• Improved attention and verbal fluency compared to older agents
• Favorable tolerability vs. phenobarbital or valproate

Interpretation:

LEV is cognitively safer than older antiepileptics in AD, but these studies do not establish benefit in non-epileptic AD.

3.3 LEV-AD (Vossel et al., 2021)

Design: Phase 2a randomized crossover
Population: Mild AD
Dose: 125 mg BID
Stratification: Epileptiform activity subgroup

Primary Outcome:

No overall improvement in executive function (NIH-EXAMINER composite).

Prespecified Subgroup (Epileptiform+):

• Significant improvement in executive function
• Improved spatial memory
• Reduction in epileptiform discharges

Interpretation:

This trial strongly supports a phenotype-dependent response. LEV does not benefit unselected AD patients but may help those with demonstrable network hyperexcitability.

3.4 ILiAD (Sen et al., 2024)

Design: Double-blind crossover
Population: Mild-to-moderate AD without known seizures
Dose: Titrated up to 500 mg BID
Limitations: Small N (COVID-19 disruption)

Findings:

• Well tolerated
• No clear cognitive benefit
• No mood deterioration

Interpretation:

Supports safety at moderate doses but does not establish efficacy in unselected AD.

3.5 HOPE4MCI (AGB101; Mohs et al., 2024)

Design: Phase 2b randomized
Population: Amnestic MCI due to AD
Dose: 220 mg once daily (extended-release AGB101)
Duration: 18 months
Primary Endpoint: CDR-SB composite

Result:

• Did not meet primary endpoint in full cohort
• APOE ε4 noncarriers showed 40% slower decline (numerical trend)
• Reduced entorhinal cortex atrophy in noncarriers

Interpretation:

The most ambitious trial to date. Failure on the primary endpoint tempers enthusiasm, but genotype-linked signal suggests biological heterogeneity matters.

4. Cross-Trial Comparison

Patterns emerge:

1. Low dose appears superior to standard epilepsy doses
2. Hyperexcitability presence predicts response
3. Unselected populations dilute signal
4. Genetic modifiers may influence efficacy

5. Mechanistic Reinterpretation

5.1 SV2A and Synaptic Stabilization

LEV binds SV2A and reduces excessive neurotransmitter release during hyperactivation states without uniformly suppressing baseline firing.

5.2 Amyloid Processing Hypothesis

Recent preclinical data (Rao et al., 2026) suggest LEV may reduce Aβ42 production via SV2A-dependent APP trafficking modulation. This is mechanistically intriguing but clinically unproven.

5.3 Tau Propagation Hypothesis

Network hyperexcitability may accelerate tau spread. By dampening excessive synchrony, LEV could theoretically reduce propagation. Again, this is plausible but not proven in humans.

6. Why Mixed Results Occur

6.1 Alzheimer's Is Not Electrophysiologically Uniform

Some patients demonstrate:

• Hyperexcitability
• Network disinhibition
• Increased hippocampal activation

Others demonstrate:

• Synaptic failure
• Hypoactivity
• Reduced throughput

Suppressing firing in the latter group may worsen cognition.

6.2 Dose Window Is Narrow

Preclinical data suggest a U-shaped curve:

• Too low - ineffective
• Optimal low dose - selective stabilization
• High dose - global suppression

Many negative experiences clinically may reflect inappropriate dosing.

6.3 Endpoint Mismatch

Short-term mechanistic correction (fMRI, EEG) may not immediately translate to long-term global functional scales like CDR-SB. Future trials likely need biomarker-linked endpoints.

7. Safety Across Trials

Across studies:

• Behavioral side effects occur but are modest at low dose
• No consistent mood worsening in AD trials
• Falls, fatigue, anxiety reported in HOPE4MCI
• Higher epilepsy doses increase irritability risk

Compared to older antiepileptics, LEV remains relatively favorable.

8. Is It Disease-Modifying?

At present:

• No Phase 3 trial demonstrates confirmed disease modification
• HOPE4MCI did not meet primary endpoint
• Mechanistic plausibility exists
• Human biomarker coupling is incomplete

LEV remains a precision symptomatic stabilizer candidate, not a proven disease-modifying therapy.

9. Clinical Synthesis for MDs

1. LEV is not FDA-approved for Alzheimer’s disease.
2. Evidence supports benefit primarily in:
   • AD with epileptiform activity
   • aMCI with hippocampal hyperactivity
   • Possibly APOE ε4 noncarriers
3. Empiric prescribing in unselected dementia populations is not supported by trial data.
4. Low-dose strategies (125 mg BID or equivalent) align with trial design.
5. Behavioral monitoring is essential.

10. Future Directions

• EEG-based enrichment strategies
• Genotype-stratified trials
• Longitudinal biomarker coupling
• Alternative SV2A ligands (e.g., brivaracetam)
• Combination therapy with anti-amyloid agents

The story is still unfolding, moving towards a precision neuromodulation framework rather than a generalized cognitive enhancer narrative.

Closing Perspective

The literature on levetiracetam and dementia neither supports unreserved enthusiasm nor outright dismissal. It reflects the early stages of defining Alzheimer’s network phenotypes.

LEV's potential lies in selectively stabilizing hyperexcitable circuits in biologically defined subgroups, rather than broadly enhancing cognition.

Future advancements will rely on:

• Biomarker-driven enrichment
• Dose optimization
• Genotype interaction modeling
• Long-term trajectory endpoints

Until then, LEV remains a compelling but targeted hypothesis - not yet a standard-of-care therapy.

References

• AgeneBio, Inc. (2026, February 17). Newsroom blog, updates, & neurodegenerative research articles.
• Alzheimer's Drug Discovery Foundation. (2024, February 16). Lithium (pharmaceutical doses): Cognitive vitality report.
• Alzheimer's Drug Discovery Foundation. (2024, October 25). Levetiracetam: Cognitive vitality report.
• Alzforum. (2024, October 27). AGB101.
• Badarny, S., Badarny, Y., & Mihilia, F. (2021). Unusual side effects of levetiracetam. BMJ Case Reports, 14(4), Article e242496.
• Devanand, D. P., Crocco, E., Forester, B. P., Husain, M. M., Lee, S., Vahia, I. V., Andrews, H., Simon-Pearson, L., Imran, N., Luca, L., Huey, E. D., Deliyannides, D. A., & Pelton, G. H. (2022). Low dose lithium treatment of behavioral complications in Alzheimer's disease: Lit-AD randomized clinical trial. The American Journal of Geriatric Psychiatry, 30(1), 32-42.
• Dutchen, S. (2025, August 6). Could lithium explain - and treat - Alzheimer's disease? Harvard Medical School News.
• Lam, A. D., & Shafi, M. M. (2022). Towards a coherent view of network hyperexcitability in Alzheimer’s disease. Brain, 145(4), 1195-1197.
• Mohamadi, M. H., Bavafa, A., Salehi, S., Abedi, M., Shahabi, F., Jafarlou, S., Kolivand, P., & Sahab-Negah, S. (2025). Cognitive effect of levetiracetam in patients with Alzheimer's disease or mild cognitive impairment: A systematic review. Current Therapeutic Research, Clinical and Experimental</