Longitudinal Associations of Blood Phosphorylated Tau181 and Neurofilament Light Chain With Neurodegeneration in Alzheimer Disease

Abstract

Importance: Plasma phosphorylated tau at threonine 181 (p-tau181) has been proposed as an easily accessible biomarker for the detection of Alzheimer disease (AD) pathology, but its ability to monitor disease progression in AD remains unclear.

Objective: To study the potential of longitudinal plasma p-tau181 measures for assessing neurodegeneration progression and cognitive decline in AD in comparison to plasma neurofilament light chain (NfL), a disease-nonspecific marker of neuronal injury.

Design, setting, and participants: This longitudinal cohort study included data from the Alzheimer's Disease Neuroimaging Initiative from February 1, 2007, to June 6, 2016. Follow-up blood sampling was performed for up to 8 years. Plasma p-tau181 measurements were performed in 2020. This was a multicentric observational study of 1113 participants, including cognitively unimpaired participants as well as patients with cognitive impairment (mild cognitive impairment and AD dementia). Participants were eligible for inclusion if they had available plasma p-tau181 and NfL measurements and at least 1 fluorine-18-labeled fluorodeoxyglucose (FDG) positron emission tomography (PET) or structural magnetic resonance imaging scan performed at the same study visit. Exclusion criteria included any significant neurologic disorder other than suspected AD; presence of infection, infarction, or multiple lacunes as detected by magnetic resonance imaging; and any significant systemic condition that could lead to difficulty complying with the protocol.

Exposures: Plasma p-tau181 and NfL measured with single-molecule array technology.

Main outcomes and measures: Longitudinal imaging markers of neurodegeneration (FDG PET and structural magnetic resonance imaging) and cognitive test scores (Preclinical Alzheimer Cognitive Composite and Alzheimer Disease Assessment Scale-Cognitive Subscale with 13 tasks). Data were analyzed from June 20 to August 15, 2020.

Results: Of the 1113 participants (mean [SD] age, 74.0 [7.6] years; 600 men [53.9%]; 992 non-Hispanic White participants [89.1%]), a total of 378 individuals (34.0%) were cognitively unimpaired (CU) and 735 participants (66.0%) were cognitively impaired (CImp). Of the CImp group, 537 (73.1%) had mild cognitive impairment, and 198 (26.9%) had AD dementia. Longitudinal changes of plasma p-tau181 were associated with cognitive decline (CU: r = -0.24, P < .001; CImp: r = 0.34, P < .001) and a prospective decrease in glucose metabolism (CU: r = -0.05, P = .48; CImp: r = -0.27, P < .001) and gray matter volume (CU: r = -0.19, P < .001; CImp: r = -0.31, P < .001) in highly AD-characteristic brain regions. These associations were restricted to amyloid-β-positive individuals. Both plasma p-tau181 and NfL were independently associated with cognition and neurodegeneration in brain regions typically affected in AD. However, NfL was also associated with neurodegeneration in brain regions exceeding this AD-typical spatial pattern in amyloid-β-negative participants. Mediation analyses found that approximately 25% to 45% of plasma p-tau181 outcomes on cognition measures were mediated by the neuroimaging-derived markers of neurodegeneration, suggesting links between plasma p-tau181 and cognition independent of these measures.

Conclusions and relevance: Study findings suggest that plasma p-tau181 was an accessible and scalable marker for predicting and monitoring neurodegeneration and cognitive decline and was, unlike plasma NfL, AD specific. The study findings suggest implications for the use of plasma biomarkers as measures to monitor AD progression in clinical practice and treatment trials.

Figure 1.. Associations of Baseline Plasma Phosphorylated Tau at Threonine 181 (P-Tau181) and Neurofilament Light Chain (NfL) With Decreasing Glucose Metabolism and Increasing Atrophy

Regression lines displayed in graphs were computed by setting covariates in the linear model to average group levels (cognitively unimpaired [CU] or cognitively impaired [CImp]) and categorical variables to the reference (female sex and, for atrophy measures, 3-T field strength). Age- and sex-adjusted associations of baseline plasma p-tau181 and NfL with hypometabolism progression are shown at the voxel (upper row) and Alzheimer disease (AD) meta–region of interest (ROI) level (bottom row) in cognitively unimpaired (A) and cognitively impaired (B) individuals. To account for the difference in sample sizes, results of voxelwise analyses were thresholded on the voxel level at P < .01 (uncorrected) for the CU group and P < .001 (uncorrected) for CImp. All maps were further thresholded at the cluster level by restricting to clusters with a number of voxels higher than the expected number of voxels as predicted using random field theory. Age- and sex-adjusted associations of baseline plasma p-tau181 and NfL with atrophy progression are shown at the voxel (upper row) and AD-signature ROI level (bottom row) in cognitively unimpaired (C) and cognitively impaired (D) individuals. Results of voxelwise analyses were thresholded at P < .01 (uncorrected) for the CU group and at P < .001 (uncorrected) for CImp. All maps were further thresholded at P < .05 (familywise error corrected) at the cluster level. The eTable in Supplement 1 shows ROI analyses using hippocampus volume. FDG indicates fluorine 18–labeled fluorodeoxyglucose; PET, positron emission tomography; and SUVR, standardized uptake value ratio.

Figure 2.. Associations of Baseline Plasma Phosphorylated Tau at Threonine 181 (P-Tau181) With Decreasing Glucose Metabolism and Increasing Atrophy in Amyloid-β–Positive (Aβ+) Cognitively Unimpaired and Impaired Participants

Associations of baseline plasma p-tau181 with longitudinal hypometabolism in Aβ+ cognitively unimpaired (CU) (A) and Aβ+ cognitively impaired (CImp) (B) and with longitudinal atrophy in Aβ+ CU (C) and Aβ+ CImp (D) at the voxel and region-of-interest (ROI) level. Models were adjusted for age, sex, and, for atrophy measures, for total intracranial volume and MRI field strength. Statistical maps were thresholded using a lenient threshold (P < .05 [uncorrected] at the voxel level and further thresholded at the cluster level by restricting results to clusters with a number of voxels higher than the expected number of voxels as predicted using random field theory) to maximize detection power in the Aβ– group while keeping identical thresholds for the Aβ+ group. Reported partial correlation coefficients were adjusted for the same covariates. Regression lines were computed by setting covariates in the linear model to average group levels (CU or CImp) and categorical variables to the reference (female sex and, for atrophy measures, 3-T field strength). The eTable in Supplement 1 shows ROI analyses using hippocampus volume. FDG indicates fluorine 18–labeled fluorodeoxyglucose; SUVR, standardized uptake value ratio.

Figure 3.. Associations of Plasma Phosphorylated Tau at Threonine 181 (P-Tau181) and Neurofilament Light Chain (NfL) Changes With Decreasing Glucose Metabolism and Increasing Atrophy

Regression lines displayed in graphs were computed by setting covariates in the linear model to average group levels (cognitively unimpaired [CU] or cognitively impaired [CImp]) and categorical variables to the reference (female sex and, for atrophy measures, 3-T field strength). Age- and sex-adjusted associations of plasma p-tau181 and NfL change with hypometabolism progression are shown at the voxel (upper row) and Alzheimer disease (AD) meta–region of interest (ROI) level (bottom row) in CU (A) and CImp (B) individuals. To account for the difference in sample sizes, results of voxelwise analyses were thresholded at the voxel level at P < .01 (uncorrected) for the CU group and at P < .001 (uncorrected) for CImp. All maps were further thresholded at the cluster level by restricting results to clusters with a number of voxels higher than the expected number of voxels as predicted using random field theory. Age- and sex-adjusted associations of plasma p-tau181 and NfL changes with atrophy progression are shown at the voxel (upper row) and AD-signature ROI level (bottom row) in CU (C) and CImp (D) individuals. Results of voxelwise analyses were thresholded at P < .01 (uncorrected) for the CU group and at P < .001 (uncorrected) for CImp. All maps were further thresholded at P < .05 (familywise error corrected) at the cluster level. The eTable in Supplement 1 shows ROI analyses using hippocampus volume. FDG indicates fluorine 18–labeled fluorodeoxyglucose; SUVR, standardized uptake value ratio.

Figure 4.. Associations of Longitudinal Plasma Phosphorylated Tau at Threonine 181 (P-Tau181) Change With Decreasing Glucose Metabolism and Increasing Atrophy in Amyloid-β–Positive (Aβ+) Cognitively Unimpaired (CU) and Cognitively Impaired (CImp) Participants

Associations of longitudinal plasma p-tau181 change with longitudinal hypometabolism in Aβ+ CU (A) and Aβ+ CImp (B) and with longitudinal atrophy in Aβ+ CU (C) and Aβ+ CImp (D) at the voxel and region-of-interest (ROI) levels. Linear models were adjusted for age, sex, and, for atrophy measures, total intracranial volume and MRI field strength. Statistical maps were thresholded using a lenient threshold (P < .05 [uncorrected] at the voxel level and further thresholded at the cluster level by restricting to clusters with a number of voxels higher than the expected number of voxels as predicted using random field theory) to maximize detection power in the Aβ- group while keeping identical thresholds for the Aβ+ group. Reported partial correlation coefficients were adjusted for the same covariates. Regression lines were computed by setting covariates in the linear model to average group levels (CU or CImp) and categorical variables to the reference (female sex and, for atrophy measures, 3-T field strength). The eTable in Supplement 1 shows ROI analyses using hippocampus volume. FDG indicates fluorine 18–labeled fluorodeoxyglucose; SUVR, standardized uptake value ratio.