Associations among plasma, MRI and amyloid PET biomarkers of dementia and the impact of health‐related comorbidities
Marc D. Rudolph, Courtney L. Sutphen, Thomas C. Register, Christopher T. Whitlow, Kiran K. Solingapuram Sai, Timothy M. Hughes, James R. Bateman, Jeff L. Dage, Kristen A. Russ, Michelle M. Mielke, Suzanne Craft, Samuel N. Lockhart- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Knowledge regarding associations between plasma and neuroimaging biomarkers indexing neurodegeneration and neuropathology observed in dementia is limited. Further, it is uncertain how comorbid health complications (e.g., kidney function) may alter plasma levels and impact associations with neuroimaging biomarkers.
Method
We examined associations between plasma and neuroimaging biomarkers in cognitively normal participants (NC; N = 300) and individuals with consensus diagnosis (Dx) of mild cognitive impairment (MCI; N = 192) or dementia (DEM; N = 64) enrolled in the Wake Forest ADRC (Table 1). We examined plasma biomarkers (Quanterix SIMOA HD‐X: Aß42/40, GFAP, NfL, p‐tau181) and neuroimaging measures of amyloid deposition (global PiB PET SUVr; Aß‐PET), total brain volume (BVOL), global white matter hyperintensity volume (WMH), diffusion‐weighted fractional anisotropy (FA) and NODDI freewater (FW; white matter). Linear models adjusted for APOE‐e4 carrier status, demographics (age, sex, race, education), and cardiometabolic factors (estimated glomerular filtration rate (eGFR); BMI).
Result
Plasma biomarkers were moderately correlated with each other (absolute r = .22‐.64; all p<.001) and significantly elevated (Aß42/40 lower) in DEM and MCI (Figure 1) versus NC, and Aß‐PET‐positive (SUVr >1.21) versus negative individuals (all p<.001). Plasma and neuroimaging markers were significantly associated in both unadjusted models and models including eGFR and BMI (all p<.05; attenuation effect <10% with and without Dx; Figure 2a & 2b). In fully adjusted models (Figure 2b: Dx and all covariates), age, sex, and race differentially impacted associations of Aß42/40, p‐tau181, and NfL with neuroimaging biomarkers (coefficients p <.05). APOE‐e4 status exclusively impacted associations with Aß‐PET SUVr/status. GFAP remained significantly associated with all neuroimaging biomarkers after covariate adjustment; no Aß42/40 associations survived adjustment. P‐tau181 remained significantly associated with Aß‐PET and BVOL, while associations with NfL were reduced in models stratified by Dx.
Conclusion
Among aging community‐dwelling participants, plasma biomarkers significantly differed between diagnostic groups (DEM>MCI>NC), were elevated in Aß‐PET positive individuals, and associated with poorer brain health. Except for GFAP, and to an extent p‐tau181, associations between plasma and neuroimaging biomarkers were differentially impacted by inclusion of comorbidities and covariates when stratified by diagnosis. Future work will examine high‐dimensional interactions among comorbidities, demographic information, and plasma and neuroimaging biomarkers in individuals with or at‐risk of dementia.