The MS Severity Variant Shows Up in Brains

The variant rs10191329*A is the first common genetic variant reaching genome-wide significance for disease progression in multiple sclerosis (MS). It sits in the DYSF–ZNF638 locus, and beyond higher lesion load in brainstem and cortex it has been tied to increased neuroaxonal damage, cortical neuronal loss and chronic inflammation in postmortem MS brains, with a recent imaging study reporting 28% more brain atrophy per risk allele in people with MS. Two readings of that atrophy are possible, and they lead in different directions. It may be disease-specific, reflecting genetic vulnerability acting together with MS mechanisms such as chronic inflammation and neurodegeneration. Or the allele may confer a generic susceptibility to neuroaxonal damage, in which case what it does in MS is one instance of something more general. Corsten and colleagues tested the second reading where it can be tested cleanly: in 4,815 people aged 45 and over who do not have MS, across 10,308 MRI scans from the Rotterdam Study.

A Population Cohort Built to Answer This Kind of Question
The Rotterdam Study is a prospective population-based cohort on determinants of age-related disease, with participants examined every three to four years. Anyone carrying an MS diagnosis at baseline or diagnosed during follow-up was excluded. All 4,815 participants were of European ancestry, 55.1% women, with a median baseline MRI age of 62.86 years, and 3,365 of them had at least two scans with a median of 5.5 years between the first and the last. Genotypes came out at 129 AA, 1,343 AC and 3,343 CC, a minor allele frequency of 16.7% with no sign of a survival effect or age-related decline in carriage. The genotype groups did not differ on sex, age, smoking, hypertension, hypercholesterolemia, type II diabetes, body mass index, schooling or dementia during follow-up. Every scan came from the same 1.5 T scanner with identical hardware and software, segmented with FreeSurfer 6.0 into total brain volume plus thalamus, putamen, cortex, total white matter, subcortical gray matter and total gray matter.

Nothing at Baseline Until Age Enters the Model
Across the whole cohort, rs10191329*A showed no association with baseline volumes of total brain or any region of interest. Splitting by age changed the picture. Below 55 years, the allele was associated with a 0.9% lower total brain volume (0.991, 95% CI 0.983–0.998, p = 0.01), while at 55 and above there was nothing to see (1.001, 95% CI 0.998–1.005, p = 0.41). The same pattern appeared in cortical and total gray matter volumes, and not in the other regions. Post hoc analysis using 10-year age blocks confirmed the association with lower total brain and total gray matter volume below age 55, and with lower thalamic and white matter volumes. No consistent interactions between age and these associations turned up within the higher age ranges, though thalamic volume did associate with the allele across multiple age blocks.

The Longitudinal Model Puts Numbers on the Age Dependence
The longitudinal analysis used linear mixed-effects models with spline terms, the knot set at 64.7 years, corrected for age, sex, total intracranial volume and the first four genetic principal components, with log-transformed volumes and p values corrected for false discovery. rs10191329*A was associated with 0.7% lower total brain volume at intercept (0.993, 95% CI 0.989–0.997, adjusted p = 0.01). Its interaction with the first age spline was significant and showed a less steep decline over time (1.017, 95% CI 1.008–1.026, adjusted p = 0.003), while the second and older spline produced nothing (1.003, 95% CI 0.994–1.013, adjusted p = 0.63). Cortex followed the same shape, with an intercept effect of 0.992 (adjusted p = 0.01) and a first-spline interaction of 1.019 (adjusted p = 0.01), as did total gray matter at 0.993 and 1.017 (adjusted p = 0.01 and 0.004). Thalamus, putamen, white matter and subcortical gray matter showed no significant associations. Sensitivity analyses restricted to participants with at least two MRIs, and to the population without dementia, gave similar results.

Why the Effect Is Far Smaller Here Than in MS
The contrast with the 28% more atrophy per allele reported in people with MS is large, and the authors set out three reasons it might be. The rate of atrophy is higher in MS cohorts before midlife. The mechanisms differ, with chronic inflammation, demyelination and neuroaxonal stress producing more severe atrophy in MS than in a population cohort. And the metrics differ, since the earlier study used yearly percentage brain volume change while this one used linear mixed-effects models to capture age-dependent trajectories. They add a fourth consideration about who was measured: median age was 35 to 37 years in the MS population against 62 years here, so genetic influences could be stronger earlier in life and taper off with age. Separately, MS susceptibility genetics taken as a whole produced nothing here. Weighted genetic risk scores built with PRSice-2 from SNPs reaching genome-wide significance were regressed on total brain volume and every region of interest, with no significant associations in either the cross-sectional or the longitudinal analysis, which the authors report as supporting the specificity of rs10191329*A in these processes.

What the Locus Might Be Doing
The authors set their result in a line of work on the same allele. Their group has previously associated rs10191329*A with more neuroaxonal damage, more inflammation and mitochondrial gene upregulation, which points to an impaired cellular stress response in MS. Longitudinal serum neurofilament light chain elevations in homozygous patients fit reduced resilience to neuroaxonal damage, as does imaging work showing increased MS-associated brain atrophy and accelerated retinal layer atrophy in risk allele carriers. They address the studies that found no association with MS severity directly rather than around: a random-effects meta-analysis incorporating those negative data together with the discovery data strengthened the association. The same allele has been inversely associated with general intelligence in a population-based genome-wide association study, which they read as a broader link to cortical integrity and cognitive reserve. The precise functional role of the locus remains unclear, but modulating expression of the proteins encoded by DYSF and ZNF638 could reduce neuronal and glial resilience to normal stressors and impair cellular maintenance.

What the Age Pattern Supports, in the Authors' Terms
Their reading is that the genetic effect emerges during or before midlife and then diminishes, which would mean an earlier onset of atrophy in carriers rather than a faster one, with the similar age-related decline across individuals pointing toward diminished genetic influence after midlife. They are careful that the attenuation may be partly technical: in older populations, additional genetic effects on brain volume may be neutralized by increasing variability of age-related neurodegeneration, vascular damage and possible epigenetic changes, and the flattening may reflect a convergence of trajectories driven by dominant age-related atrophy, a floor effect, or reduced variability between individuals. What the midlife signal may point to, they suggest, is a function in preserving brain homeostasis or resilience during the early phases of adult brain aging. The strengths they claim are the large representative population, the longitudinal design, and all MRIs performed on one scanner with identical protocols. The limits they name are a possible dropout effect, since single-scan participants may differ from those with repeated imaging, though findings held in sensitivity analyses excluding them; white matter lesions, which may contribute to brain volume loss and were not analyzed specifically here; and the age floor of 45 years, which leaves them unable to say whether genetic effects emerge earlier in life, something longitudinal studies in younger cohorts would need to settle. They note that a population-based sample of children aged 9 to 13 showed no such associations, which makes a major role in early neurodevelopment less likely. Their conclusion is that rs10191329*A modulates general neuroaxonal resilience to neurodegeneration and inflammation, with the next work aimed at the cellular and molecular mechanisms and at modifiable pathways for the benefit of people with MS.

Disclaimer: This blog post is based on the cited study 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.

Reference:
Corsten, C. E. A., Marques, A. M., Vinke, E. J., de Mol, C. L., Neuteboom, R. F., Ikram, M. K., Ghanbari, M., Wolters, F. J., Wokke, B., Vernooij, M. W., & Smolders, J. (2026). Brain atrophy associated with risk variant rs10191329 extends beyond multiple sclerosis. Annals of Neurology, 99(4), 1083–1089. https://doi.org/10.1002/ana.78148