Alzheimers Disease Project: Quantitative imaging of astrogliosis and neuroinflammation

National Institute On Aging

Investigators

Linked publications & trials

Abstract

As indicated in the Goals and Objectives section, the main goals of this research initiative are to discover diagnostic imaging biomarkers for astrogliosis and neuroinflammation by investigating the relationships between advanced MD-MRI signal signatures and increased glial fibrillary acidic protein (GFAP) deposition. Below, milestones from this project are summarized: 1. Imaging cortical astrogliosis in Alzheimer’s disease (AD) Development of a multidimensional MRI (MD-MRI) biomarker for cortical astrogliosis in Alzheimer’s disease (AD): We conducted ex vivo 7T MRI on postmortem human cortical tissue from two independent cohorts, comparing non-demented individuals with varying levels of AD pathology to individuals with both AD pathology and clinical dementia. MD-MRI data—including combined diffusion and relaxation contrasts—were co-registered with histopathology (Aβ, pTau, GFAP). The resulting biomarker showed strong and specific association with astroglial burden (GFAP: β = 0.658/0.709, pFDR < 0.0001) but not with Aβ or pTau, suggesting high specificity for neuroinflammation over classical AD lesions. Demonstrated predictive value of MD-MRI for cognitive status: Among all tested MRI parameters, the MD-MRI–derived astrogliosis signature was the only metric capable of differentiating cognitively unimpaired individuals from those with dementia despite equivalent levels of AD pathology. Conventional T1, T2, and diffusion tensor imaging (DTI) metrics failed to show similar associations. This work highlights the potential of MD-MRI as a sensitive, noninvasive imaging tool for detecting cortical neuroinflammation and predicting cognitive decline in the context of Alzheimer’s disease. 2. Characterizing the human hippocampus in Alzheimer’s disease using virtual histology derived from ex vivo MRI and histology Development of a multidimensional MRI (MD-MRI) approach for predicting Alzheimer’s pathology: We investigated whether T2-diffusion correlation MD-MRI could noninvasively predict neuropathology associated with AD. Brain tissue samples from 12 deceased human donors, ranging in age from 50 to 89 and with varying levels of cognitive impairment and AD pathology, were collected. The temporal lobe, including the hippocampus and surrounding white and gray matter, was scanned using a 9.4T MRI system to capture detailed diffusion-relaxation signal profiles at each voxel. These MRI data were matched with histological measurements of hallmark AD features—amyloid-β (Aβ), phosphorylated tau (p-tau), microglial activation, and myelin—using digital microscopy and image registration techniques. Demonstrated potential of MRI-based “virtual histology”: Using machine learning–based predictive modeling (elastic net regression), we showed that the MRI-derived signal patterns could successfully estimate the presence of each histological marker. The strongest predictions were observed for myelin (R² = 0.60, PCC = 0.78) and microglial activation (R² = 0.49, PCC = 0.70), suggesting that this imaging method is particularly sensitive to inflammation and structural tissue changes. These findings support the potential of T2-D MD-MRI as a “virtual histology” tool—capable of estimating microscopic brain pathology without the need for invasive procedures or postmortem analysis. Future efforts will focus on adapting this technique for clinical use and refining prediction models across brain regions. 3. Time-Dependent Diffusion-Relaxation Multidimensional MRI Detects Alzheimer’s Pathology in a Mouse Model Developing and applying time-dependent diffusion-relaxation MD-MRI in mice: This project advances the use of time-dependent diffusion-relaxation MD-MRI to detect microstructural changes associated with AD pathology. By combining advanced diffusion techniques with quantitative relaxometry in a multidimensional framework, MD-MRI offers increased sensitivity and reproducibility. Critically, the method integrates diffusion time (or frequency) information using varied diffusion encoding waveforms—an innovation that enables detection of microstructural alterations that evolve over time. In this study, MD-MRI was applied to ex vivo brain samples from 5xFAD transgenic mice, a widely used AD pathology model characterized by rapid amyloid plaque development, and compared with wild-type (WT) controls. Region-of-interest (ROI) analysis revealed statistically significant differences in diffusion-time–sensitive metrics between AD and WT mice, highlighting their potential as novel, noninvasive biomarkers for early AD pathology. Diffusion time dependency is sensitive to AD pathology: Building on our prior work that established the reproducibility of MD-MRI metrics in fixed mouse brains, this study demonstrated that several time-dependent diffusivity metrics can robustly differentiate AD from WT brain tissue. The consistency of these findings across individuals and the clear group-level contrasts visible in common-space metric maps emphasize the biological relevance and reliability of these parameters. While metrics such as isotropic diffusivity and diffusion anisotropy align with previous literature, the study’s novel contribution lies in identifying new diffusion-time/frequency–dependent metrics that show significant between-group differences despite a small sample size (n=4 per group). Future directions include increasing sample size for validation, developing smooth voxel-wise distribution maps, exploring customized metric binning, and studying temporal progression in younger mice. Together, these efforts aim to refine MD-MRI as a sensitive and practical imaging tool for preclinical AD diagnosis.