DOI: 10.1002/alz.070428 ISSN: 1552-5260

Blood‐based biomarkers of Alzheimer disease as measured by a mass spectrometry assay are as accurate as CSF tests

Suzanne E. Schindler, Nicolas R. Barthélemy, Benjamin A. Saef, Rachel L. Henson, Yan Li, Tammie L.S. Benzinger, John C Morris, Randall J. Bateman
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology



Especially because disease‐modifying treatments (DMTs) for Alzheimer disease (AD) are likely to become more available soon, highly accurate AD blood tests for use in clinical diagnosis are critically needed. Although amyloid PET scans and cerebrospinal fluid (CSF) tests are highly accurate in the identification of individuals with brain amyloidosis, blood tests are likely the only test modality that will enable an adequate scale of AD biomarker testing. Mass spectrometry‐based AD blood tests have consistently demonstrated very high performance.


Research participants enrolled in studies at the Knight Alzheimer Disease Research Center who underwent amyloid PET and measurement of established CSF biomarkers of AD with Lumipulse automated assays were included. The presence and severity of dementia was evaluated with the Clinical Dementia Rating® (CDR®) and Clinical Dementia Rating Sum of Boxes (CDR‐SB). Plasma samples collected at the same session as CSF underwent analysis for specific phosphorylated and non‐phosphorylated tau species with an immunoprecipitation‐mass spectrometry assay. Plasma T217 occupancy was calculated as the percent of phosphorylated to non‐phosphorylated tau at threonine 217.


The study cohort included 337 individuals with an average age of 69.8 ± 8.3 years (mean ± standard deviation); 175 (52%) were female, 128 (38%) were APOE ε4 carriers, 112 (33%) were amyloid PET positive, and 50 (15%) were cognitively impaired. Classification accuracy of amyloid PET status was equivalent for plasma T217 occupancy (AUC 0.973) and CSF Aβ42/Aβ40 (AUC 0.970), (p = 0.72 for the difference). Interestingly, plasma T217 occupancy more accurately classified individuals as cognitively impaired (CDR>0) or cognitively unimpaired (CDR 0), with an AUC of 0.876 versus 0.814 for CSF Aβ42/Aβ40 (p = 0.006 for the difference). Further, plasma T217 occupancy was better correlated with dementia severity as measured by the CDR‐SB (Spearman ρ = 0.455 versus ‐0.338 for CSF Aβ42/Aβ40, p = 0.02).


Plasma T217 occupancy as measured by mass spectrometry had equivalent classification accuracy of amyloid PET status compared to Lumipulse CSF Aβ42/Aβ40. Notably, plasma T217 occupancy was more strongly associated with cognitive impairment and dementia severity than CSF Aβ42/Aβ40. Overall, these data suggest that some AD blood tests have equivalent or better performance characteristics than clinically used CSF tests.

More from our Archive