Individualised prognostic models using plasma biomarkers for clinical progression to Alzheimer’s disease dementia in non‐demented elderly

Abstract

AbstractBackgroundBlood‐based biomarkers can provide a low‐invasive and accessible way to identify neurodegenerative diseases before the clinical onset of dementia. We aimed to construct prognostic models for personalized risk of developing Alzheimer’s disease (AD) dementia in a memory clinic population of individuals with either subjective cognitive decline (SCD) or mild cognitive impairment (MCI), using plasma phosphorylated‐tau‐181 (pTau181), amyloid beta1‐42/1‐40 (Aβ42/40), glial fibrillary acidic protein (GFAP) and/or neurofilament light (NfL).MethodFrom the Amsterdam Dementia Cohort and SCIENCe project we included 303 individuals with SCD (age 61±9 years, n = 126(42%) female, MMSE 28±1) and 250 with MCI (age 65±7 years, n = 89(36%) female, MMSE 27±2), who had annual follow‐up visits for re‐evaluation of diagnosis (average follow‐up duration: 2.7±1.7 years.) Our outcome measure was progression to AD dementia. Plasma biomarkers were measured at baseline using SiMoa. Concentrations were then Z‐transformed for the statistical analysis and split into tertiles for risk chart visualisation. We evaluated adding the biomarkers individually, or as a panel, to a basic prognostic model including age, sex and baseline diagnosis. We selected a final model by its discrimination (Harrell’s C‐index, a value above 0.7 indicates a good model) and accuracy (Brier score, a value below 0.2 indicates a good model) and used it to calculate 5‐year risk scores for progression to AD dementia.ResultDuring follow‐up, 86 individuals developed AD dementia (8 with SCD, 78 with MCI at baseline, average time to progression: 2.8±1.7 years). Adding any of the biomarkers to the basic model improved the model discrimination (Table 1). The prognostic model with all four plasma biomarkers had the best discrimination and accuracy, followed by the model with GFAP only, and the model with pTau181 only (Table 1). Using the model with all four biomarkers, 5‐year individual progression risks to AD dementia varied from 1.9% (lowest tertiles for GFAP, pTau181 and NfL, highest tertile for Aβ42/40 ratio) to 40.6% (highest tertiles for GFAP, pTau181 and NfL, lowest tertile for Aβ42/40 ratio) (Figure 1).ConclusionPlasma biomarkers, particularly plasma GFAP, can be utilized to provide prognostic risk information about progression to AD dementia for individuals presenting at a memory clinic.