In vivo staging of regional amyloid deposition predicts functional conversion in the preclinical and prodromal phases of Alzheimer's disease

Highlights

• A confirmatory study in preclinical and prodromal Alzheimer's disease is presented.

• Amyloid stages from amyloid PET identify a high-risk group for conversion

• Effects are robustly replicated across 3 large independent cohorts.

• Identification of high-risk converters through amyloid stages has implications for selecting cohorts for clinical trials.

Abstract

We tested the usefulness of a regional amyloid staging based on amyloid sensitive positron emission tomography to predict conversion to cognitive impairment and dementia in preclinical and prodromal Alzheimer's disease (AD). We analyzed 884 cases, including normal controls, and people with subjective cognitive decline or mild cognitive impairment (MCI), from the Alzheimer's Disease Neuroimaging Initiative with a maximum follow-up of 6 years and 318 cases with subjective memory complaints with a maximum follow-up time of three years from the INveStIGation of AlzHeimer's PredicTors cohort (INSIGHT-preAD study). Cox regression showed a significant association of regional amyloid stages with time to conversion from a cognitively normal to an MCI, and from an MCI to a dementia status. The most advanced amyloid stages identified very-high-risk groups of conversion. All results were robustly replicated across the independent samples. These findings indicate the usefulness of regional amyloid staging for identifying preclinical and prodromal AD cases at very high risk of conversion for future amyloid targeted trials.

Introduction

Cerebral amyloid deposition is considered an upstream event in the pathogenesis of Alzheimer's disease (AD) (Thal et al., 2006). In vivo imaging using amyloid sensitive positron emission tomography (PET) detected increased levels of amyloid in 15%–30% of cognitively normal people older than 70 years, and in at least 50% of people with a clinical phenotype of amnestic mild cognitive impairment (MCI) (Quigley et al., 2010). However, the positive predictive value of increased amyloid signal in PET for subsequent cognitive decline in preclinical or prodromal AD cases is limited. In cognitively normal people, the positive predictive value of a positive amyloid PET status for subsequent conversion to MCI or dementia is only about 25% over 3–5 years of follow-up (Baker et al., 2017, Morris et al., 2009, Villemagne et al., 2011). In people with MCI, the positive predictive value of positive amyloid status for subsequent conversion to AD dementia is about 65%–84% for a follow-up period of 3–5 years (Martinez et al., 2017, Zhang et al., 2014).

The current standard of amyloid PET imaging data analysis is a dichotomous classification in amyloid-positive or amyloid-negative cases (Klunk et al., 2015). Recently, we have developed a more fine-grained (Grothe et al., 2017) and replicable (Sakr et al., 2019) PET-based in vivo amyloid staging scheme that considers five regional stages of progressive cerebral amyloid deposition. The staging identified neurobiologically meaningful regional variation of amyloid deposition even in people with an amyloid-negative status, as shown by associations of amyloid stages with cerebrospinal fluid (CSF) Aβ1-42 concentrations and cognitive performance. An alternative tripartite staging approach has been based on differential involvement of cortical versus subcortical structures (amygdala, putamen, and caudate nucleus) (Cho et al., 2018). This previous study showed promising results for the predictive utility of amyloid staging but lacked a differentiation of cortical stages and a comparison with the standard binary classification.

Here, we evaluated the usefulness of regional amyloid staging to predict conversion of cognitively normal people with and without subjective cognitive decline (SCD) or subjective memory complaints (SMC) to MCI or AD dementia and of MCI cases to AD dementia, respectively. We compared our results with classical binary amyloid classification. We studied replicability of effects in three different longitudinal samples: a sample from the Alzheimer's Disease Neuroimaging Initiative (ADNI) that had previously been used for establishing the regional amyloid staging approach (Grothe et al., 2017), a second sample from ADNI that was not part of the development of the staging scheme, and an independent cohort of SMC cases from the monocentric INveStIGation of AlzHeimer's PredicTors in subjective memory complainers (INSIGHT-preAD) cohort (Dubois et al., 2018). As an endpoint, we assessed functional conversion as defined by transition in clinical dementia rating scale (CDR) scores (Berg, 1988).