In the past decade, in vivo cerebral amyloid load in the spectrum of cognitive impairment has been extensively explored using positron emission tomography (PET), and is now a key part of screening visits for most therapeutic studies. However, to date, little if no distinction is made between patients with low to medium amyloid load versus patients with high amyloid load, in terms of both the risk of developing AD and the progression of cognitive change. Here, we investigated the cognitive changes in frail participants to a multidomain interventional study (Andrieu et al., Lancet 2017), according to the degree of cerebral amyloid load: absent/low, medium, high.

Two hundred and forty five participants to the MAPT study were selected. All of them underwent a full cognitive assessment, and were follow-up for up to five years. They all underwent a cerebral amyloid PET scan using [18F]florbetapir. All of them had been randomized at the beginning of the trial into one of four intervention groups: multidomain intervention and daily omega-3 intake; multidomain intervention and placebo; omega-3 uptake alone; the placebo alone. Cortical amyloid load was quantified from PET imaging as the mean standard uptake value ratio in all regions of the brain (parcelated according to the AAL atlas; cerebellar grey matter as reference). Patients were categorized into “low”, “medium”, or “high” cerebral amyloid using k-clustering. Cognitive functions were measured as a composite score combining four tests. Using a linear regression model, we investigated how cognition progressed over time according to amyloid status and the intervention group.

Our findings were two-fold: patients in the “high amyloid” group showed cognitive decline over time, while patients in the “medium amyloid” group exhibited a cognitive profile stable and similar to the patients with “low amyloid” load (P<0.001). In addition, in the “medium amyloid” group only, we observed an improvement of cognitive performance over time for patients of all intervention group, except for patients in the placebo group (i.e., without intervention at all), who showed stable cognitive performance (P<0.05).

Our findings suggest that having a moderate or high amyloid load is not equivalent in terms of impact on cognitive progression in the short-term (5years), and that patients with moderate rather than high amyloid load may be a better suited target for drug studies.