Creatine 20g for 8 Weeks Shifts Brain Energy in Alzheimer's Patients, CABA Trial 2026
Creatine is strongly associated with athletic performance. That association is accurate, but incomplete. Before creatine reaches muscle, the brain already has a claim on it. The brain accounts for just 2 percent of body weight yet consumes 20 percent of the body’s total energy. In Alzheimer’s disease, the first and most pervasive system to fail is precisely this: the brain’s ability to produce and use energy.
A paper published in February 2026 in Alzheimer’s & Dementia: Translational Research & Clinical Interventions reports bioenergetic data from the CABA (Creatine to Augment Bioenergetics in Alzheimer’s) pilot trial, the first study to examine whether oral creatine supplementation can shift cellular energy metabolism in people with Alzheimer’s dementia.
When the Brain Runs Low on Energy
Alzheimer’s is often introduced through the lens of amyloid plaques and tau tangles. But before either becomes visible, something quieter happens: mitochondrial function in neurons declines, reducing ATP, the cell’s primary energy currency. Creatine operates directly within this cycle. It acts as a phosphate shuttle, rapidly regenerating ATP when demand spikes. This mechanism is why Alzheimer’s researchers have been watching creatine with interest.
The CABA Protocol: 8 Weeks, 20 g/Day, 20 Patients
The team at the University of Kansas Medical Center enrolled 20 patients with clinical Alzheimer’s dementia (mean age 73.1, 90 percent amyloid-positive, 65 percent APOE4 carriers) and administered 20 g/day of creatine monohydrate for 8 weeks, split into two 10-gram doses. This was a single-arm pilot with no placebo group.
The trial’s first paper, published in May 2025, reported brain creatine and cognitive outcomes. Measured by MR spectroscopy, brain creatine increased in 85 percent of participants, with an average elevation of 11 percent. Scores on working memory and oral reading recognition tests improved significantly. Fluid cognition and global cognitive composite scores also moved upward.
The February 2026 paper goes one level deeper, reporting what happened at the cellular level.
ATP Rose. Mitochondria Responded. Sex Made a Difference.
Researchers collected blood samples to measure mitochondrial function directly in lymphocytes and platelets. The findings were clear.
Lymphocyte ATP increased significantly (p < 0.001). ADP also rose (p < 0.02), and the ATP/ADP ratio held steady, meaning energy production and utilization both increased without creating imbalance. This pattern indicates activated energy metabolism, not just an artifact of more substrate.
The sex-specific finding was striking. Female participants showed significant increases in mitochondrial respiration across multiple measurement states. Male participants did not show the same effect. This suggests creatine’s impact on brain energy metabolism may differ by sex, a finding with practical implications for future trial design.
Two brain MR spectroscopy markers, N-acetylaspartate (NAA) and glutathione (GSH), did not change significantly over 8 weeks. Both reflect longer-term neuronal integrity and oxidative stress, and 8 weeks is likely too short a window to expect movement in either.
Why 20 g/Day Is Not a Standard Recommendation
The CABA dose is several times higher than typical creatine maintenance doses of 3 to 5 g/day. It was selected to rapidly saturate brain creatine stores, functioning similarly to a loading phase. All 20 participants completed the full 8 weeks, and adverse events were mild: cramping, gastrointestinal discomfort, nausea, facial flushing, and sleep disturbance, totaling 13 reports across the group.
For people with compromised kidney function, high-dose creatine warrants discussion with a physician before starting. The trial did not report kidney-related adverse events, but its small size and short duration limit how much that absence can tell us.
What the CABA Data Does and Does Not Say
CABA has real limitations: no control group, 20 participants, 8 weeks. These are features of a well-designed feasibility study, not a definitive efficacy trial. The researchers are explicit about this, and a larger randomized controlled trial is now in planning.
What the data does establish, for the first time, is a sequence of biological events: oral creatine crosses the blood-brain barrier and raises cerebral creatine stores in people with Alzheimer’s dementia. Those elevated creatine stores correspond with measurable increases in systemic cellular ATP. And cognitive assessments in domains tied to energy-intensive prefrontal processes showed improvement.
Three things made this study worth watching. First, the blood-brain barrier penetration is confirmed in an Alzheimer’s population, where transport mechanisms are often compromised. Second, cellular ATP elevation provides a mechanistic footprint, not just a symptom score. Third, the female-specific mitochondrial response opens a question about sex-differentiated supplementation protocols that has not been studied.
What This Means for Everyday Supplementation
This trial enrolled Alzheimer’s patients and used a high-loading protocol. Healthy adults interested in cognitive support are in a different situation. The relevant context comes from broader meta-analyses: a 2024 systematic review in Frontiers in Nutrition covering 16 RCTs and 492 participants found creatine supplementation significantly improved memory (SMD = 0.31), with stronger effects in older adults aged 66 to 76. Those trials used maintenance doses of 3 to 5 g/day.
Creatine monohydrate is among the most studied supplements available and is relatively inexpensive. A maintenance dose of 3 to 5 g/day dissolves easily in water or a beverage. If you already take a multivitamin or pre-workout blend, check whether creatine is already included before adding a separate dose.
Creatine is not about to replace approved Alzheimer’s therapies. But the CABA data positions creatine as a bioenergetic candidate worth a serious randomized trial, not because of what researchers expect it to do, but because the first cellular data suggests it may actually do something.