Measuring Tissue in MS Genetics
If central nervous system resilience in multiple sclerosis (MS) is partly inherited, the evidence has to come from studies that do three things in the same sample: genotype people, measure CNS tissue rather than a disability score, and have something to do with MS. That is a narrower set than it sounds. Three papers manage it. One reviews what genome-wide work has delivered for MS prognosis, brain atrophy included, and is blunt about how little is usable. One takes an allele borrowed from Alzheimer's research and reads it against four separate tissue measurements in people with MS. And one maps the genetic architecture of the cervical spinal cord in more than 80,000 people, which is the structure MS damages and the one imaging genetics has mostly ignored.
What Genome-Wide Work Has Delivered on Tissue
Ciampana and colleagues review the genetics of MS prognosis across clinical outcomes, radiological measures like lesion burden, gadolinium-enhancing lesions and atrophy, and laboratory measures like oligoclonal bands and IgG index. Their summary is that genetic influences on prognosis are modest and highly heterogeneous, with only a limited number of associations showing consistent confirmation while many findings come from small candidate-gene work and remain unconfirmed. On the tissue side specifically they name what has held up: the strongest genome-wide evidence concerns relapse rate, MRI measures including thalamic atrophy, and intrathecal IgG synthesis, with the largest genome-wide study of the age-related MS Severity Score finding one locus at genome-wide significance. The obstacles they list are methodological rather than biological, from small samples and limited confirmation to variable outcome definitions, treatment exposure and population ancestry.
An Allele Borrowed From Another Disease
Zeldich and colleagues took the APOE4 allele, the major genetic risk factor for late-onset Alzheimer's disease, and asked what it does inside an MS population. Using UK Biobank data they grouped patients with MS by carrier status into MS-E4 and MS-nonE4, and included age- and sex-matched people without MS as control-E4 and control-nonE4, which is the design that separates an MS-specific effect from a general one. The readouts were deliberately spread across compartments rather than pooled into a clinical score: retinal nerve fibre layer thickness from optical coherence tomography, blood glial fibrillary acidic protein and neurofilament light chain, cognitive performance, and white matter hyperintensities on MRI. Four different ways of asking how much tissue a person has kept.
Four Measurements Agreeing
The carriers with MS came out worse on all of them: thinner retinal nerve fibre layer, higher blood glial fibrillary acidic protein and neurofilament light chain, slower cognitive reaction times, and more white matter hyperintensities. Agreement across a retinal layer, two blood markers of neuroaxonal and astroglial injury, a behavioural measure and a structural MRI measure is worth more than any one of them, because the four do not share a measurement artefact. The statistical detail that matters most is that glial fibrillary acidic protein showed significant interactions between MS and APOE4 status, so the astroglial signal was not simply additive. The authors also report similar effects in healthy carriers, which keeps the finding from being read as something MS creates rather than something MS amplifies.
Mapping the Structure MS Damages
Sun and colleagues addressed a gap rather than a hypothesis. The spinal cord carries motor, sensory and autonomic function and is increasingly tied to neurological disease, yet its genetic architecture had been largely unexplored, partly because imaging genetics grew up around the brain. They phenotyped the upper cervical cord from C1 to C3 using brain MRI already acquired in more than 80,000 UK Biobank participants, extracting shape metrics including cross-sectional area, diameters and eccentricity. That is the whole trick: the cord was in the field of view of scans collected for other reasons, and nobody had measured it at this scale. The resulting phenotypes turned out to be substantially heritable, with SNP-based heritability from 0.16 to 0.42.
What the Cord Map Returned
They identified 179 independent genome-wide significant variants, along with sex-specific genetic signals that they read as pointing to biological sex differences in spinal cord development. And cord structure was associated with a range of neurological, metabolic and systemic conditions, multiple sclerosis among them, alongside neuropathies, diabetes and attention-deficit hyperactivity disorder. For MS the relevance is direct rather than analogical. Cord atrophy is one of the better predictors of disability in the disease, and until now there was no map of which common variants set how much cord a person starts with. The authors frame the cervical cord as a genetically informative and health-relevant structure, which is a modest way of saying the field had been measuring the wrong end of the CNS.
Three Levels, One Measurement Problem
The three sit at different levels and converge on the same practical point. The review has hundreds of susceptibility variants against modest and largely unconfirmed prognostic associations, with thalamic atrophy among the few MRI measures where genome-wide evidence has held. The APOE4 work shows what happens when a genotype is read against several independent tissue measurements at once rather than a disability score: four readouts agree, and one of them interacts with disease status. And the cord GWAS supplies 179 variants for a structure whose loss tracks disability closely, in people who do not have MS, which is precisely the sort of normative map needed before anyone can say whether a patient's cord is small because of the disease or because of their genome. Resilience genetics in MS may have been limited less by biology than by what was being measured.
Disclaimer: This blog post is based on the cited studies and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.
References:
Ciampana, V., Virgilio, E., Paciolla, L., Asaro, S., Franceschini, A., Thavamani, M., Mazzini, L., Comi, C., Barizzone, N., D'Alfonso, S., & Vecchio, D. (2026). Genetic Prognostic Factors in Multiple Sclerosis: Key Discoveries and Unmet Needs. International Journal of Molecular Sciences, 27(8), 3583. https://doi.org/10.3390/ijms27083583
Zeldich, D., Cheng, C. H., Guan, Y., Narain, S., & Koo, B.-B. (2026). Apolipoprotein E4 and its later-life health effects on the multiple sclerosis population. Multiple Sclerosis Journal – Experimental, Translational and Clinical, 12(1), 20552173251387833. https://doi.org/10.1177/20552173251387833
Sun, Z., Han, J., Gerring, Z. F., Jackson, V. E., Bhalala, O. G., Jarola, G. M., Harding, I. H., Rezende, T. J. R., & Bahlo, M. (2026). Genetic architecture of MRI-derived cervical spinal cord morphology reveals sensory-motor axis and systemic disease associations. Nature Communications, 17(1). https://doi.org/10.1038/s41467-026-77419-x
