Blood Test for Neurofilament Light Predicts Clinical Decline in Early Alzheimer's Disease

Predicting how fast an individual patient with Alzheimer's disease (AD) will decline is hard, yet it matters most early on, when anti-amyloid therapies are expected to help the most. Blood biomarkers are now good at detecting AD pathology, but whether they predict progression in everyday memory clinic practice is far less established. It is also unclear whether changes over a few months reflect real disease activity or just biological and assay variability.

This prospective, observational cohort study recruited patients at a tertiary memory clinic at Rigshospitalet in Copenhagen between January 2022 and June 2023. Participants had mild cognitive impairment or mild-to-moderate dementia due to AD (MMSE above 19) and a baseline FDG-PET scan. Plasma was collected at baseline and again at a median of 4.4 months. Ninety-four patients had baseline and one-year follow-up, and 85 completed two years. Plasma NfL, BD-tau, p-tau217 and GFAP were measured on a single molecule array platform. Outcomes were annual change in the Clinical Dementia Rating Sum of Boxes (CDR-SB) and changes on FDG-PET and volumetric MRI. Linear regression models used baseline levels and short-term changes as predictors, adjusted for age, sex and creatinine.

Baseline plasma NfL was significantly associated with a rise in CDR-SB (worsening) at both one and two years. Baseline BD-tau, p-tau217 and GFAP were not associated with clinical progression. For imaging, short-term changes in NfL were associated with reduced glucose metabolism in the hippocampus and temporal cortex. Short-term changes in p-tau217 were associated with reduced hippocampal metabolism. Changes in the p-tau217/BD-tau ratio were associated with reduced metabolism in the hippocampus and isthmus cingulate cortex. No plasma biomarker was significantly associated with MRI volumetric change after correction for multiple testing.

The findings suggest NfL is a useful prognostic marker for clinical progression in early symptomatic AD, supporting its role in risk stratification and as a trial endpoint. Dynamic changes in p-tau217 and related ratios appear to track metabolic dysfunction in AD-vulnerable regions, so they may reflect underlying disease processes. The authors stress that short-term changes should be interpreted in light of baseline levels and clinical characteristics.

Several caveats apply. The cohort was modest and drawn from a single tertiary centre. The study ended early for logistical reasons, so not everyone reached two years. Plasma changes over about four months can be influenced by biological and analytical variability, and by factors such as kidney function and comorbidities. Multiple imaging regions were tested, and the MRI results did not survive correction. The results are associations, not evidence that biomarker-guided decisions improve outcomes.