Early cortical changes are detectable before amyloid positivity

Early cortical changes may be detectable long before amyloid-beta reaches the level classified as positive on a PET scan. The unexpected part is the direction of the effect: people who later became amyloid-positive did not show earlier cortical loss. Instead, their cortex was relatively thicker and thinned more slowly with age.

Amyloid-beta is now treated as a central biological marker of Alzheimer’s disease. In the standard sequence of disease biomarkers, amyloid accumulation comes first, followed by other pathological processes including tau, while structural brain atrophy appears later. Amyloid PET has therefore been considered one of the earliest imaging signals of the disease process.

A study by James Roe and colleagues in Nature Neuroscience challenges that sequence, at least for some cortical changes.

Looking backward from amyloid positivity

The researchers combined data from three longitudinal aging cohorts: LCBC in Norway, the Berkeley Aging Cohort Study and ADNI. Altogether, the analysis included 4,570 MRI scans from 1,051 cognitively healthy adults aged 30 and older. Among them, 691 participants also had amyloid PET data, providing 1,684 PET scans.

A particularly informative group consisted of 77 people whose PET scans were initially below the threshold for amyloid positivity but later crossed it. Their MRI trajectories were compared with those of participants who remained amyloid-negative across their available PET scans.

Rather than comparing a single scan before and after conversion, the authors modeled each participant’s age-related cortical thickness trajectory. They then repeatedly removed MRI scans obtained too close to the first positive PET scan. This allowed them to ask whether structural differences could still be detected using only MRI data collected many years before amyloid positivity.

They could. Differences remained detectable at least seven years before the estimated time at which participants crossed the PET threshold, with some of the clearest effects in frontal cortical regions.

Less thinning, not earlier atrophy

The cerebral cortex normally becomes thinner with age. In people who later became amyloid-positive, that thinning occurred more slowly. Their cortex also appeared thicker than expected for age.

That finding needs careful interpretation. The study does not show that early Alzheimer’s pathology makes the brain healthier or produces extra neurons. MRI-derived cortical thickness is an imaging measurement based on the estimated boundary between gray and white matter. It can be influenced by inflammation, glial changes, intracortical myelination and other tissue properties.

One possibility raised by the authors is a biphasic process. At an early stage, low and still subthreshold amyloid accumulation could be accompanied by local inflammatory or other tissue responses that make the cortex appear relatively thicker, or reduce the normal age-related rate of thinning. Later, as tau pathology and neurodegeneration become more prominent, the direction could reverse and cortical thinning could accelerate. But this remains a proposed explanation rather than a demonstrated mechanism: tau biomarkers were not measured in this study.

The researchers also found that regions with higher amyloid accumulation followed different thickness trajectories, and the temporal ordering of cortical changes showed moderate correspondence with the progression of amyloid deposition. That strengthens the case that the two processes are related, but it does not establish that amyloid directly causes the early cortical changes.

Why “seven years” is not a precise disease clock

The most important limitation concerns what “amyloid-negative” actually means. A negative PET scan does not prove that no amyloid is present. It means that the amyloid signal remains below the threshold used for detection or classification. The appropriate conclusion is therefore that cortical changes can precede PET-defined amyloid positivity, not that they necessarily occur before amyloid begins to accumulate.

PET scans were also performed at discrete intervals, so the exact moment when each person crossed the positivity threshold was unknown. The authors therefore estimated that timing from individual amyloid trajectories. In the stricter analysis based on predicted positivity, the cortical effect was still detectable seven years earlier, but only 16 converters remained at the seven-year cutoff. The result is therefore informative without making seven years a universal biological boundary.

The cohorts were also composed of cognitively unimpaired volunteers in long-running research studies, a group that is generally healthier and higher-performing than typical clinical populations. The findings should not be assumed to apply directly to people with symptomatic dementia. In addition, MRI-derived “thickness” can partly reflect tissue properties such as myelination rather than literal changes in the physical thickness of the cortex.

The study therefore does not establish a new diagnostic test for Alzheimer’s disease. Its more important implication is conceptual: the familiar sequence in which amyloid PET becomes positive first and structural MRI changes only later may be too simple. Some cortical alterations may begin substantially earlier, while amyloid burden is still below the conventional PET threshold.