Prediction of amyloid and tau brain deposition and cognitive decline in people with Down syndrome using plasma biomarkers: a longitudinal cohort study

Abstract

Background: Plasma biomarkers associated with Alzheimer's disease could improve prognostic assessment for people with Down syndrome in both clinical practice and research settings. We aimed to identify the plasma biomarkers that most accurately predict longitudinal changes in Alzheimer's disease-related pathology and cognitive functioning in individuals with Down syndrome.

Methods: This longitudinal cohort study included data from 258 adults (aged ≥25 years) with Down syndrome who were followed up prospectively every 16 months as part of the longitudinal Alzheimer's Biomarker Consortium-Down Syndrome study (recruited from seven university sites in the USA and UK between July 13, 2016, and Jan 15, 2019). Participants had baseline and longitudinal assessments of plasma tau phosphorylated at threonine 217 (p-tau217), glial fibrillary acidic protein (GFAP), amyloid β (Aβ)_42/40, neurofilament light (NfL), or total tau (t-tau). Associations of baseline plasma biomarkers and longitudinal changes in plasma biomarkers with changes in global cognitive functioning (Down Syndrome Mental Status Examination [DS-MSE] scores), Aβ-PET, and tau-PET were examined using linear regression models. Plasma biomarker-associated risk of progression to dementia was assessed using Cox regression analysis.

Findings: Baseline p-tau217, as well as GFAP, NfL, or t-tau, were individually associated with longitudinal changes in DS-MSE, Aβ-PET, and tau-PET, and with progression to dementia. However, in combined models, only baseline p-tau217 remained associated with changes in DS-MSE (β -0·30 [95% CI -0·45 to -0·15], p=0·0001, n=220), tau-PET (0·42 [0·14 to 0·70], p=0·0039, n=88), and progression to dementia (hazard ratio 3·51 [95% CI 1·76-7·00], p=0·0004, n=194), whereas baseline p-tau217 (0·29 [0·14-0·45], p=0·0003) and GFAP (0·37 [0·18-0·56], p=0·0003) were associated with changes in Aβ-PET (n=106 for both). Similar associations were shown between longitudinal p-tau217 or GFAP and changes in DS-MSE (p-tau217: β -0·33 [95% CI-0·52 to -0·13], p=0·0015, n=133), tau-PET (p-tau217: 0·61 [0·40 to 0·83], p<0·0001, n=87), and Aβ-PET (p-tau217: 0·35 [0·19 to 0·50], p<0·0001; GFAP: 0·49 [0·27 to 0·70], p<0·0001, n=88).

Interpretation: Baseline and longitudinal plasma p-tau217 were associated with subsequent decline in global cognition, progression to dementia, and increased tau burden, whereas baseline p-tau217 and GFAP were associated with Aβ accumulation. These findings suggest that plasma p-tau217 and GFAP might be valuable for prognostic assessment of Alzheimer's disease in people with Down syndrome in both clinical and research contexts. The results further support evaluation of these biomarkers for monitoring disease progression in clinical trials of Down syndrome-related Alzheimer's disease.

Funding: The European Research Council and National Institute on Aging (National Institute of Health).

Figure 1

Association between baseline plasma biomarkers and longitudinal cognition Spaghetti plots visualise associations of plasma p-tau217 (A), GFAP (B), NfL (C), and t-tau (D) with cognitive decline. Each transparent line represents an independent individual connecting longitudinal DS-MSE observations. Solid lines represent the estimated cognitive trajectories for plasma biomarker quartiles (with shaded areas representing 95% CI) that were derived using linear mixed effects models and the “ggeffects” package of R software. Linear mixed effects models included longitudinal DS-MSE as the outcome and interaction between plasma biomarker quartiles and time as the predictor adjusting for age, sex, premorbid intellectual disability, APOEε4 status, and site. Plasma NfL and t-tau data were missing for 33 and 37 participants, respectively. Further results for the associations between plasma biomarkers and cognition are presented in table 2. APOEε4=apolipoprotein E ε4. DS-MSE=Down Syndrome Mental Status Examination. GFAP=glial fibrillary acidic protein. NfL=neurofilament light. p-tau217=tau phosphorylated at threonine 217. t-tau=total tau.

Figure 2

Association of baseline plasma biomarkers with longitudinal tau-PET and Aβ-PET Scatterplots illustrate the association of plasma p-tau217 (A, B), GFAP (C), and NfL (D) with tau-PET slopes (A) and Aβ-PET slopes (B–D) stratified by Aβ status (cutoff point Centiloid >20). Each dot represents an individual biomarker value (log-transformed) and corresponding PET slope value. Regression lines with 95% CI (shaded areas) representing associations between plasma biomarkers and PET slopes were derived using linear regression models and the “ggeffects” package of R software. p values are from linear regression models, including the interaction between each plasma biomarker (log-transformed) and Aβ status as the predictor and subject-specific slopes of tau-PET or Aβ-PET as the outcome adjusted for age, sex, premorbid intellectual disability, APOEε4 status, and site. Plasma NfL data were missing for three participants. Further results for the associations between plasma biomarkers and PET measures are presented in the appendix (pp 17–18). Aβ=amyloid β. APOEε4=apolipoprotein E ε4. GFAP=glial fibrillary acidic protein. NfL=neurofilament light. p-tau217=tau phosphorylated at threonine 217.