Neurodegeneration biomarkers in Alzheimer’s disease: axonal density index expands the “N” in the AT(N) framework

Neurodegeneration, along with amyloid and tau, define the AT(N) framework of Alzheimer’s disease that has shaped the development of diagnostics and therapeutics. Yet, biomarker development for neurodegeneration has lagged behind that for amyloid and tau, with limited definition of its heterogeneous microstructural aspects that may each serve as critical measures. In this issue of the JCI, Gong et al. leveraged diffusion MRI to derive a unique measure of axonal injury or axonal density index (ADI). Through cross-sectional and longitudinal analyses, they demonstrated that the ADI has superior performance in detecting, tracking, and predicting clinical impairment compared with prior diffusion MRI methods to evaluate axonal health and standard biomarkers of amyloid and tau. As such, the ADI measure may serve as an important expansion of the neurodegeneration biomarker repertoire.

Within the AT(N) framework, neurodegeneration is a critical biological endpoint indicating the culmination of amyloid- and tau-mediated damage and the development of clinical symptoms. Typically, neurodegeneration is assessed using structural MRI, with macroscale measurements of medial temporal volume loss and cortical thinning (2). However, these macrostructural changes may be the result of many microstructural changes, including axon loss, neuronal or glial cell body loss, and dendritic pruning (3). Measures of volume loss and cortical thinning also fail to account for changes in white matter, which are increasingly implicated in early AD (4). Such microstructural white matter changes may have unique predictive capabilities and could not only reflect diverse pathophysiological ramifications of amyloid and tau, but also other processes increasingly recognized as contributing to AD, namely, inflammation and cerebrovascular dysfunction (2). Thus, it is critical to refine imaging methods that can evaluate these changes in early AD.

Diffusion MRI has the potential to sensitively measure axonal injury in the white matter across disorders. Most work on imaging-based markers of axonal injury has been conducted in traumatic brain injury (5) and, for more advanced diffusion MRI models, in multiple sclerosis (6). In these disorders, diffusion MRI is generally more sensitive than traditional structural MRI, including both quantitative measures using diffusion tensor imaging and qualitative expert reading of apparent diffusion coefficient maps (5, 6). Diffusion tensor imaging measures of axonal injury also correlate well with CSF measurement of neurofilament light chain, a biofluid marker of axonal damage (7, 8). Research in multiple sclerosis has shown that microstructure models of diffusion MRI, such as neurite orientation dispersion and density imaging (NODDI) and standard model of imaging (SMI), can outperform diffusion tensor imaging measures of axonal injury (6). Microstructure models work by parsing the diffusion MRI signal into different compartments, often including intra-axonal water, extra-axonal water including extracellular matrix and glia, and free water including CSF (9). They can thus be used to generate an axon density index (ADI), which is the fraction of intra-axonal water compared with the total water volume. However, these measures of ADI become less reliable in the presence of high free water fraction (10), which is particularly an issue in neurodegenerative disorders such as AD (11). Despite this, there has been little research investigating the impact of free water correction techniques on ADI measures in AD. Further, there are limited longitudinal studies on ADI in AD.

In this issue of the JCI, Gong et al. (12) leveraged the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset to conduct a study of whole white matter ADI derived from a free water–corrected implementation of NODDI called constrained NODDI (C-NODDI), which they propose to be a more physiologically relevant measure of axonal health. Instead of using the diffusion MRI signal to estimate the free water fraction as done in the original NODDI implementation, C-NODDI estimates the free water fraction by segmenting a structural T2-weighted image using the FMRIB Software Library Automated Segmentation Tool (8). They subsequently compared the performance of their C-NODDI ADI metric with metrics derived from previous measures of axonal density, namely, NODDI and SMI, in a cohort of participants that were cognitively normal and cognitively impaired (comprising mild cognitive impairment and AD dementia).

They first demonstrated that their measure of ADI differentiated between subjects that were cognitively impaired and cognitively normal at baseline, with those that were impaired showing significantly lower C-NODDI measures of axonal density. Moreover, subjects with cognitive impairment showed a significantly greater longitudinal decline in ADI compared with those that were cognitively normal. They subsequently found that baseline C-NODDI ADI successfully predicted longitudinal cognitive and functional outcomes in the cognitively impaired group. The longitudinal decline in C-NODDI ADI also tracked with declines in cognition and function in this group. Importantly, all of the above findings were valid even when the impaired cohort was restricted just to participants with mild cognitive impairment. Furthermore, C-NODDI outperformed NODDI and SMI across detection, prediction, and tracking of clinical impairment.

They next evaluated ADI against standard biomarkers of proteinopathy. In cognitively impaired participants, the capability of C-NODDI ADI to predict cognitive and functional decline in impaired participants was comparable to PET and CSF biomarkers of amyloid and tau. Lastly, with respect to tracking, only longitudinal changes in C-NODDI ADI and amyloid PET correlated with longitudinal changes in cognition and function.

Taken together, the above findings suggest that axonal loss is a key aspect of neurodegeneration in AD, as well as a major contributor to cognitive and functional outcomes. Notably, changes in ADI were evident in the prodromal mild cognitive impairment stage. Strengths of the study include its reliance on the robust ADNI dataset, longitudinal analyses, which fill a critical gap in current microstructure MRI literature, and comparison with established AD biomarkers. Some limitations of the study include its short follow-up period, which did not permit evaluation of preclinical outcomes, and that analyses were restricted to whole white matter, with the spatial pattern of ADI values not extensively evaluated.

This work has important implications for expanding the repertoire of clinically relevant markers of neurodegeneration that have prognostic value and for understanding the pathways to neurodegeneration in AD and related disorders. From a methodological perspective, it provides an easily implemented and clinically scalable solution for microstructure model instability in the presence of high free water infiltration, as regularly seen in AD. Future studies should explore other free water correction strategies that may further improve the sensitivity of ADI. The expense, equipment, and expertise required to collect diffusion MRI are a limitation to accessibility, particularly in resource-scarce environments. However, current developments in low-field MRI may alleviate these limitations in the future. From a pathophysiologic perspective, this work highlights that neurodegeneration is not a monolithic process in AD and that axonal injury is an important determinant and predictor of impairment that may serve as a critical metric for evaluating therapeutic interventions. As such, future work should focus on defining the value of measuring axonal injury during the preclinical stage, spatial variation in axonal injury across the brain, and the relationship of this variation to the heterogeneity of pathology that defines AD and related disorders. Overall, this work shines an important and innovative spotlight on neurodegeneration in AD of relevance to improved detection, monitoring, and therapeutic development.

This work is the result of NIH funding, in whole or in part, and is subject to the NIH Public Access Policy. Through acceptance of this federal funding, the NIH has been given a right to make the work publicly available in PubMed Central.

• NIH grant P30AG066512.
• Agence National de la Recherche grant ANR-22-CE17-0060.

Conflict of interest: AVM is an inventor on US provisional patent application no. 63/638,866.

Copyright: © 2026, Flaherty et al. This is an open access article published under the terms of the Creative Commons Attribution 4.0 International License.

Reference information: J Clin Invest. 2026;136(3):e202411. https://doi.org/10.1172/JCI202411.

See the related article at Early axonal degeneration linked to clinical decline in Alzheimer’s disease progression revealed with diffusion MRI.

1. Jack CR, JrA/T/N: an unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology. 2016;87(5):539–547.
   View this article via: CrossRef PubMed Google Scholar
2. Jack CR, JrRevised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup. Alzheimers Dement. 2024;20(8):5143–5169.
   View this article via: CrossRef PubMed Google Scholar
3. Schilling KG, et al. Aging and white matter microstructure and macrostructure: a longitudinal multi-site diffusion MRI study of 1218 participants. Brain Struct Funct. 2022;227(6):2111–2125.
   View this article via: CrossRef PubMed Google Scholar
4. Nasrabady SE, et al. White matter changes in Alzheimer’s disease: a focus on myelin and oligodendrocytes. Acta Neuropathol Commun. 2018;6(1):22.
   View this article via: CrossRef PubMed Google Scholar
5. Lampros M, et al. The role of novel imaging and biofluid biomarkers in traumatic axonal injury: an updated review. Biomedicines. 2023;11(8):2312.
   View this article via: CrossRef PubMed Google Scholar
6. Vanden Bulcke C, et al. Comparative overview of multi-shell diffusion MRI models to characterize the microstructure of multiple sclerosis lesions and periplaques. Neuroimage Clin. 2024;42:103593.
   View this article via: CrossRef PubMed Google Scholar
7. Alirezaei Z, et al. Neurofilament light chain as a biomarker, and correlation with magnetic resonance imaging in diagnosis of CNS-related disorders. Mol Neurobiol. 2020;57(1):469–491.
   View this article via: CrossRef PubMed Google Scholar
8. Alsameen MH, et al. C-NODDI: a constrained NODDI model for axonal density and orientation determinations in cerebral white matter. Front Neurol. 2023;14:1205426.
   View this article via: CrossRef PubMed Google Scholar
9. Novikov DS, et al. Quantifying brain microstructure with diffusion MRI: theory and parameter estimation. NMR Biomed. 2019;32(4):e3998.
   View this article via: CrossRef PubMed Google Scholar
10. Zhang H, et al. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage. 2012;61(4):1000–1016.
    View this article via: CrossRef PubMed Google Scholar
11. Henf J, et al. Mean diffusivity in cortical gray matter in Alzheimer’s disease: the importance of partial volume correction. Neuroimage Clin. 2018;17:579–586.
    View this article via: CrossRef PubMed Google Scholar
12. Gong Z, et al. Early axonal degeneration linked to clinical decline in Alzheimer’s disease progression revealed with diffusion MRI. J Clin Invest. 2026;136(3):e196638.
    View this article via: JCI PubMed CrossRef Google Scholar