Introduction:
Intervertebral disc degeneration (IVDD) is a leading cause of chronic low back pain, posing significant challenges to healthcare systems globally. IVDD is a complex, heritable condition influenced by both genetic and environmental factors, presenting with radiological features such as disc height reduction, herniation, and bulging, which can be observed through MRI. Traditional phenotyping for IVDD relies on manual MRI assessments, introducing variability and limiting the effectiveness of genome-wide association studies (GWAS) in identifying genetic risk factors. This study uses SpineNet, an automated MRI-based phenotyping tool, across three large cohorts (TwinsUK, Northern Finland Birth Cohort, and Michigan Genomics Initiative) to achieve consistent phenotypic data and enhance the statistical power required to uncover genetic contributors to IVDD.
Methods:
SpineNet was applied to MRI data from each cohort to extract eight IVDD-related phenotypes, including disc narrowing, herniation, and Pfirrmann score for disc degeneration severity. MRI sequences were filtered to match SpineNet training standards, ensuring consistent phenotype extraction across cohorts. GWAS analyses were conducted on each SpineNet-derived phenotype within each cohort and then meta-analyzed to identify significant and suggestive genetic associations. Genetic correlations across IVDD phenotypes were also examined, followed by a multi-trait meta-analysis leveraging shared genetic signals. Fine-mapping and expression quantitative trait loci (eQTL) analyses were performed to identify probable causal genes, with pathway enrichment analysis highlighting biological pathways relevant to IVDD.
Results:
Validation of SpineNet phenotyping accuracy against manual MRI grading showed high agreement, supporting the reliability of SpineNet-derived phenotypes for GWAS. Meta-analysis identified several suggestive loci, including CHST3 and SOX5, both of which have been previously implicated in disc degeneration. These and other suggestive genes identified in our analysis have been extensively linked to disc degeneration, reinforcing the relevance of known genetic pathways in IVDD. The study also identified significant genetic correlations across IVDD phenotypes, suggesting shared genetic influences across multiple disc traits. Pathway analysis of significant and suggestive loci highlighted pathways related to extracellular matrix organization and cellular processes involved in IVDD’s pathogenesis.
Discussion:
This study confirms the effectiveness of SpineNet in replicating and identifying genetic risk factors associated with IVDD, demonstrating the value of automated MRI phenotyping in large-scale genetic studies. The identification of suggestive loci, including CHST3 and SOX5, supports the importance of previously recognized pathways in IVDD’s complex etiology, particularly in pathways affecting structural integrity. The findings underline SpineNet potential to facilitate future genetic studies of disc degeneration, paving the way for targeted therapeutic approaches. By improving the accuracy and scalability of phenotyping across diverse cohorts, this research establishes a strong foundation for future studies to advance precision medicine in IVDD, supporting the development of targeted therapies and preventive strategies by offering a more comprehensive understanding of the genetic architecture underlying disc degeneration.