One Brain Map, Two Diseases: Alzheimer’s and Multiple Sclerosis Share a Genetic Signature of Shrinkage

Alzheimer’s disease and multiple sclerosis could hardly look more different on the surface. One is a slow neurodegenerative illness of old age, defined by amyloid plaques and tau tangles; the other is an autoimmune attack on myelin that typically strikes people in their twenties and thirties. Yet a sweeping new study argues that when it comes to the structural erosion of the brain, the two conditions converge on a shared anatomical and genetic signature — one that may point toward therapies designed not for a single disease, but for the neurodegenerative process itself.

The research, published in Frontiers in Neuroimaging, is a trans-disease meta-analysis led by a team at Tianjin Medical University General Hospital in China. The investigators pooled voxel-based morphometry studies published between 2001 and 2024, assembling neuroimaging data from 2,070 patients with multiple sclerosis against 1,625 healthy controls, and 1,624 patients with Alzheimer’s disease against 1,607 healthy controls, drawn from 84 articles. Voxel-based morphometry is a technique that compares grey matter volume across the whole brain voxel by voxel, allowing researchers to detect regional shrinkage that often precedes symptoms. By statistically combining the peak coordinates of significant grey matter clusters from dozens of independent studies using the Seed-based Mapping software, the team produced disease-level atrophy maps of unusual statistical power.

The maps revealed a striking split. Multiple sclerosis-specific atrophy localized to the bilateral dorsomedial thalamic nucleus and the precentral gyrus — the deep grey matter and sensorimotor cortex that reflect the disease’s demyelinating character. Alzheimer’s-specific atrophy, in the age-unadjusted comparisons, centered on the bilateral hippocampus-amygdala complex, the medial temporal structures long known as the disease’s earliest casualties. Cytoarchitectonic mapping using the probabilistic Julich Brain Atlas added anatomical precision: the left hippocampal peak in Alzheimer’s disease was assigned to the CA1 subfield with a 52.5 percent probability, while the right parahippocampal peak fell within the laterobasal amygdala. The team also found that some of these between-disease differences, particularly in the thalamus and hippocampal-amygdala regions, were attenuated after age adjustment — a reminder that the twenty-year gap in typical onset age between the two conditions shapes part of the anatomical picture.

But the more consequential finding was the overlap. When the investigators computed the conjunction of the two atrophy maps, common grey matter loss emerged in the bilateral superior temporal gyrus, the insulae, the posterior cingulate cortex, the amygdalae, the right rectus gyrus, and striatal components including the caudate nucleus and putamen. These regions form the backbone of two major brain networks: the auditory-language network, anchored by the superior temporal gyrus, and the emotional processing network, a circuit spanning the amygdala, insula, orbitofrontal cortex, and striatum. Neurosynth decoding — a cross-sample spatial correlation between the atrophy map and thousands of brain activation maps for cognitive terms — confirmed the functional relevance, with the top terms encompassing speech production, auditory processing, face recognition, pain perception, and emotional regulation.

The convergence is not merely anatomical. Postmortem studies cited in the paper have documented pronounced cortical microglial activation in both multiple sclerosis and Alzheimer’s disease with closely similar patterns, and Alzheimer’s-type pathology, including beta-amyloid plaques and neurofibrillary tangles, has been identified in a subset of multiple sclerosis cases. Case reports have described patients with multiple sclerosis and progressive cognitive impairment who proved amyloid-positive and were subsequently diagnosed with coexisting Alzheimer’s disease. Conversely, recent work suggests that certain aspects of multiple sclerosis biology, or specific genetic factors in those patients, may offer unexpected protection against amyloid pathology — hinting at undiscovered neuroprotective crosstalk between the two conditions.

To probe the molecular underpinnings of the shared atrophy pattern, the team turned to spatial transcriptomics. They extracted gene expression profiles from the Allen Human Brain Atlas, a postmortem dataset mapping genome-wide expression across the adult human brain, and linked each tissue sample to the grey matter Z-value at its corresponding coordinate in the meta-analytic atrophy map. A partial least squares regression model then identified gene-expression patterns spatially associated with the atrophy profile. Of the five components estimated, PLS2 explained 13.10 percent of the variance and was the only one to survive a spatial autocorrelation-preserving permutation test — a critical safeguard, because brain maps are spatially smooth and naive statistics can inflate gene-imaging correlations. The PLS2 score correlated positively with the Z-scores of the overlapping atrophy regions, and 834 genes emerged as significantly associated with the common atrophy pattern, split into a positive set of 290 genes and a negative set of 544.

Functional enrichment analysis of these gene sets converged on four core biological themes: neurodegenerative injury, immune-inflammatory regulation, energy metabolic homeostasis, and synaptic-axonal maintenance. The positive-weighted genes predominantly encode how cells respond to internal and external stress — genomic damage, metabolic and energy challenges — and determine cellular fate through DNA stability maintenance, cell cycle arrest, metabolic remodeling, and quality control mechanisms such as mitophagy and apoptosis. This theme resonates deeply with established pathology. Mitochondrial dysfunction has been identified as a cause of axonal degeneration in multiple sclerosis, where energy deficits in demyelinated axons disrupt transport and promote cell death, while reduced mitochondrial density in Alzheimer’s disease signals an oxidative stress burden. The positive gene set also showed significant enrichment in adaptive immune cell differentiation and activation pathways, including T-cell receptor signaling — the very processes that drive inflammatory lesion formation in multiple sclerosis, where T-cell migration across the blood-brain barrier is a precondition for demyelination.

The negative-weighted genes tell the complementary story. They are functionally characterized by active suppression of excessive inflammation — inhibiting the production of interleukin-1 beta, interferon-beta, and interleukin-6 — along with specification of immune cell subsets such as marginal zone B cells and follicular dendritic cells, and maintenance of fundamental processes like DNA replication and synaptic vesicle transport. Cell-type-specific expression analysis showed the negative set was specifically expressed in immune cells, Pnoc-positive neurons of cortex, the habenula, and the hypothalamus, while the positive set was significantly expressed in the striatum. In other words, the shared atrophy pattern is spatially aligned with genes that both execute stress responses and actively restrain them — an immunological homeostasis imbalance that may represent the common molecular soil in which grey matter loss grows under two very different etiologies.

The authors are careful about what these findings do and do not mean. The spatial correlations link normative postmortem gene expression to disease-related atrophy; they are not evidence of differential gene expression in patients or direct causal effects of the identified genes on grey matter loss. Multiple sclerosis clinical subtypes carry distinct neurobiological features, and genetic heterogeneity across them could influence the results. Most importantly, because the two diseases have different primary neuropathological mechanisms, the shared atrophy regions and transcriptomic patterns may represent phenotypic outcomes induced by different disease processes rather than proof of a shared pathogenic mechanism — and the results cannot replace existing clinical diagnostic criteria for either condition.

Even with those caveats, the implications are considerable. The identification of a shared cortical-limbic-striatal atrophy nexus, validated across more than 6,900 brain scans and anchored to 834 genes spanning stress response, immune regulation, and energy metabolism, offers a template for transdiagnostic biomarkers of cognitive impairment and for therapeutic targets with relevance across disorders. The authors suggest that neuromodulation targeting the amygdala-insula-striatum circuit, such as transcranial magnetic stimulation, might simultaneously improve emotional symptoms in both conditions. As people with multiple sclerosis live longer and age into the risk window for neurodegenerative disease, the boundary between the two illnesses may matter less than the molecular machinery they share — and that machinery, this study argues, is now mapped.

Subject of Research: Neuroscience

Article Title: Grey matter atrophy signatures and their genetic associations: a trans-disease study of Alzheimer’s disease and multiple sclerosis

Article References: Zhao, Y., Sun, J., Reheman, B., Bai, X., Zhang, C., Ding, W., Zou, B., Zhang, H., Zhang, Z., & Zhang, N. (2026). Grey matter atrophy signatures and their genetic associations: a trans-disease study of Alzheimer’s disease and multiple sclerosis. Frontiers in Neuroimaging, 5, Article 1917609. https://doi.org/10.3389/fnimg.2026.1917609

Image Credits: AI Generated

DOI: 10.3389/fnimg.2026.1917609

Keywords: Alzheimer's disease, multiple sclerosis, grey matter atrophy, meta-analysis, spatial transcriptomics, Allen Human Brain Atlas, neuroimaging, neurodegeneration, gene expression, Frontiers in Neuroimaging, Grey, matter