MRI Brain Changes May Precede PET-Visible Amyloid by at Least 7 Years

Rather than treating a single scan as a diagnosis, the researchers examined trajectories over time. They compared cortical thickness and its change in people who later became high-amyloid on PET with the corresponding patterns in people who remained amyloid-negative. The MRI data used for the comparison came from years before the participants converted to high amyloid.

This longitudinal design is important. A cross-sectional scan can show that two groups differ, but repeated scans can reveal whether the rate or pattern of structural change differs before a later biological event.

Why this challenges the role of amyloid PET

Amyloid PET is often used to identify the presence of substantial fibrillar amyloid plaque. But a scan becomes positive only after amyloid has reached the level that the imaging method and its interpretation threshold can detect. A negative or below-threshold scan may therefore mean that plaque burden is not yet detectable—not necessarily that Alzheimer’s-related biology has not begun.

The Oslo findings do not establish that cortical change causes amyloid accumulation, or that it is specific to Alzheimer’s disease. They do challenge a simpler interpretation of amyloid PET: that the first positive scan marks the beginning of the disease process. Structural changes may already be under way by that point.

What might be happening before PET-visible plaque?

Several explanations remain possible, and the study alone cannot distinguish among them:

• Amyloid below PET sensitivity: soluble amyloid species or smaller amounts of deposited amyloid might affect cells before plaques become visible on PET.
• Tau-related injury: tau pathology may contribute to cortical change independently of, or alongside, detectable amyloid.
• Synaptic or neuronal dysfunction: early disruption of communication between neurons could affect cortical structure before major symptoms appear.
• Inflammation, vascular injury, or metabolic disease: these processes can influence brain structure and may overlap with Alzheimer’s risk.
• Normal or accelerated aging: some thinning may reflect ordinary aging rather than a disease-specific process.

Other imaging research has found distinct relationships between cortical thickness, amyloid, and tau in cognitively healthy older adults, reinforcing the need to interpret thinning as a biological signal rather than a stand-alone diagnosis. Likewise, the vascular-burden findings show why MRI abnormalities should not automatically be attributed to amyloid.

What earlier detection could change

If the pattern is replicated and becomes sufficiently specific for individual patients, serial MRI could help researchers monitor people before symptoms or high amyloid-PET burden appear. It could also be combined with blood biomarkers, cognitive testing, genetic-risk information, and tau or amyloid imaging to create a more complete risk profile.

Earlier biological monitoring could help researchers:

• Identify suitable participants for prevention trials
• Measure disease progression before substantial symptoms develop
• Test treatments at an earlier stage
• Separate Alzheimer’s-related changes from vascular or age-related changes
• Determine which early signals actually predict future decline

For now, the finding does not support using cortical thinning on routine MRI as a population-wide screening test or as a diagnosis of Alzheimer’s disease. The key unanswered question is whether detecting these changes earlier—and acting on them—improves outcomes.

What the contrasting GLP-1 results add to the picture

The timing problem also matters for treatment research. A therapy may be tested after symptoms and established pathology are already present, even if the disease process began years earlier.

A smaller study of injectable liraglutide reported a possible reduction in brain-volume loss in regions vulnerable to Alzheimer’s disease, along with a signal for slower cognitive decline. Those findings are encouraging but remain an imaging and early clinical signal, not proof that liraglutide is an effective Alzheimer’s treatment.

The larger EVOKE and EVOKE+ phase 3 trials produced a different result for oral semaglutide. Across 566 sites, the trials randomly assigned 3,808 participants and tested once-daily oral semaglutide at doses up to 14 mg. After 104 weeks, semaglutide did not slow cognitive or functional decline compared with placebo in early symptomatic Alzheimer’s disease.

These results do not prove that every GLP-1 drug is ineffective for Alzheimer’s disease. They show that oral semaglutide, at the tested dose and duration, in the studied population, did not produce a clinical benefit. A difference between liraglutide and semaglutide could involve the specific molecule, dose, exposure, oral versus injectable formulation, treatment timing, disease stage, or how much active drug reaches relevant brain targets. Those are research hypotheses, not conclusions established by the trials.

The larger research gap

The MRI result and the GLP-1 results point to the same problem from different directions: Alzheimer’s biology may begin quietly, while reliable detection and treatment trials often start later.

Future work needs to establish which early structural changes are specific and predictive, validate low-burden biomarkers, replicate the Oslo finding in peer-reviewed studies, and test whether interventions given before substantial plaque burden can preserve cognition. Until then, the seven-year MRI signal is best understood as evidence that the detectable timeline of Alzheimer’s disease may begin earlier than amyloid PET alone can show—not as a ready-to-use diagnostic or treatment pathway.