The Protein Your Brain Leaks Before Dementia Starts

Illustration of a human figure with a highlighted brain

A simple blood test may be able to flag your dementia risk 15 years before you show a single symptom — and the science behind that claim is stronger than most headlines let on.

Story Snapshot

Researchers found that a blood protein called GFAP begins rising up to 15 years before a dementia diagnosis.
People with high GFAP levels were more than twice as likely to develop dementia and nearly three times as likely to develop Alzheimer’s disease specifically.
A second protein, LTBP2, also showed strong predictive power — though some media outlets incorrectly named it “MENT,” which is not a real protein in this research.
No routine blood test for these proteins is currently available to patients, and elevated levels signal risk — not a confirmed diagnosis.

The Early Dementia Signal Hiding In Blood

Your brain has a cleanup crew. Cells called astrocytes help protect and support neurons. When those cells get stressed or damaged — often years before dementia shows up — they release a protein called glial fibrillary acidic protein, or GFAP, into the bloodstream. Researchers have now confirmed that rising GFAP levels in the blood are one of the earliest warning signs that the brain is under attack. The signal starts long before memory fades.

A 2024 study published in Nature Aging tracked thousands of adults over many years and found that GFAP levels began climbing roughly 15 years before a dementia diagnosis. That is not a typo. Fifteen years. That kind of early warning window is almost unheard of in medicine. For context, most dementia symptoms appear only after significant brain damage has already occurred. By then, the window for meaningful intervention may already be closing.

What the Numbers Actually Say

The statistics here are worth sitting with. Adults whose GFAP levels crossed a specific threshold faced a hazard ratio of 2.25 for developing all-cause dementia. That means their risk was more than double compared to people with lower levels. For Alzheimer’s disease specifically, the risk multiplier jumped to 2.91. Researchers also found that combining GFAP with another protein called neurofilament light chain produced prediction models with accuracy scores ranging from 0.80 to 0.91 — strong numbers for any biological risk tool.

A second protein, LTBP2 — short for latent-transforming growth factor beta-binding protein 2 — also emerged as highly specific to dementia risk, meaning it does not just flag general poor health. It flags brain-specific trouble. Some popular health websites reported this protein as “MENT,” which is simply wrong. The primary research is clear: the protein is LTBP2. Getting the name right matters because public confusion about what was actually discovered can erode trust in legitimate science.

Why This Is Not a Diagnosis — and Why That Matters

Here is where the excitement needs a guardrail. GFAP elevation is a risk signal, not a verdict. The predictive power of these proteins is notably stronger in adults under 65 than in those 65 and older. That age gap is important because most people worried about dementia are already in their 60s and 70s. The proteins also work best when combined with basic personal data — age, sex, education level, and whether someone carries the APOE ε4 gene variant linked to Alzheimer’s risk. Alone, they tell part of the story. Together with that context, they tell much more.

No Food and Drug Administration-approved blood test for GFAP or LTBP2 exists for routine clinical use today. That gap between what researchers know and what doctors can actually order for patients is real, and it is frustrating. This pattern is not new in dementia research. Over the past two decades, other promising blood markers — including amyloid beta ratios and total tau proteins — went through the same cycle: strong early findings, media buzz, then years of validation before any clinical tool emerged. GFAP appears more specific and more promising than many of its predecessors, but the validation work still lies ahead.

What Comes Next in This Research

Scientists still need to answer several key questions. Does catching high GFAP early and intervening — say, with anti-inflammatory therapies — actually delay or prevent dementia? Nobody knows yet. The current studies show association, not cause and effect. Researchers also need to test these biomarkers in populations beyond the UK Biobank, which skews heavily toward people of European descent. Whether GFAP predicts dementia as reliably in people of Asian or African ancestry is an open and important question. The science is genuinely exciting. The practical tool is not here yet — but it is closer than it has ever been.