Oral Presentation GENEMAPPERS 2026

Genetic architecture of MRI-derived cervical spinal cord morphology reveals sensory-motor axis and systemic disease biomarkers (#30)

Jiru Han 1 2 , Zhuopin Sun 1 2 , Zachary Gerring 1 2 , Vicki Jackson 1 2 , Oneil G Bhalala 1 2 3 , Gustavo M Jarola 4 , Ian H Harding 5 6 , Thiago J.R. Rezende 4 7 , Melanie Bahlo 1 2
  1. Department of Medical Biology, The University of Melbourne, Melbourne, Parkville, Australia
  2. The Walter and Eliza Hall Institute of Medical Research, Parkville, VICTORIA, Australia
  3. Department of Medicine, Royal Melbourne Hospital, Melbourne, VIC, Australia
  4. Department of Neurology, School of Medical Sciences, University of Campinas (UNICAMP), Campinas, São Paulo, Brazil
  5. Queensland Institute of Medical Research, Herston, QLD, Australia
  6. Department of Neuroscience, School of Translational Medicine, Monash University, Melbourne, VIC, Australia
  7. Brazilian Institute of Neuroscience and Neurotechnology (BRAINN), School of Medical Sciences, University of Campinas (UNICAMP), Campinas, São Paulo, Brazil

Abstract:

The spinal cord is critical to motor, sensory, autonomic function, and increasingly implicated in neurological disease and human health, yet its genetic architecture remains largely unexplored.  Here, we present the first large-scale, AI-enabled genome-wide association study (GWAS) of spinal cord morphology derived from brain MRI in over 40,000 UK Biobank participants.

Methods: We applied a novel deep learning pipeline adapted from the ENIGMA-ataxia consortium, to extract high-quality phenotypes from upper cervical spinal cord segments (C1–C3). Through rigorous validation and longitudinal quality control, we derived 12 spinal cord phenotypes, including cross-sectional area, diameters, and eccentricity across these segments. 

Results: We identified 179 independent genome-wide significant variants, with cervical spinal cord morphology showing moderate to high SNP-based heritability (0.16 to 0.42). We observed strong phenotypic and genetic correlations among adjacent spinal cord segments, confirming the robustness of these measures. The C1 segment, located closest to the brainstem, exhibited the strongest correlation with brainstem and subcortical structures. We also uncovered sex-specific genetic signals, highlighting potential biological sex differences in spinal cord development. In addition, spinal cord structure was associated with a wide range of neurological, metabolic, and systemic conditions, such as multiple sclerosis, neuropathies, diabetes, and attention-deficit/hyperactivity disorder. To validate these findings, we assessed effect sizes in independent, non-European ancestry cohorts, and generated spinal cord polygenic risk scores (PRS) using the genetic data of remaining non-imaged UKB participants. The validation analyses recapitulated the disease associations identified using the MRI-derived phenotypes in independent samples.

Conclusions: These findings establish the cervical spinal cord as a genetically informative and health relevant structure, offering new opportunities to study its role in disease mechanisms and human health.