Modules of genotypic variance reflect heterogeneity across TDP-43 proteinopathies

Barbara E Spencer, PhD; David J. Irwin, MD; Vivianna M Van Deerlin, MD, PhD; Eddie B Lee, MD, PhD; Lauren Elman, MD; Colin Quinn, MD; Murray Grossman, MD, EdD; David A. Wolk, MD and Corey T McMillan, PhD University of Pennsylvania, Philadelphia, PA, USA

Background: TAR DNA-binding protein (TDP-43) proteinopathies yield a variety of neurodegenerative conditions including amyotrophic lateral sclerosis (ALS), frontotemporal lobar degeneration (FTLD), and hippocampal sclerosis of aging (HS-Aging). However, the contribution of both shared and disparate single nucleotide polymorphisms (SNPs) to phenotypic presentations is unclear. We hypothesize that shared genetic variation may drive susceptibility across diseases while disparate genetic variation may contribute to our understanding of disease-specific drivers. Method: We compared the relative associations of GWAS summary statistics for ALS, FTLD-TDP, and HS-Aging using weighted correlation analysis to identify data-driven modules of highly correlated SNPs. A gene ontology enrichment analysis for biological processes was performed separately for each identified module. Polygenic risk scores (PRS) were constructed using module SNPs and tested for association with diagnosis in cases that met either neuropathological or genetic criteria. Result: We identified five modules of correlated SNPs: three modules associated with a single GWAS phenotype, ALS (M1, containing 1,910 SNPs), FTLD-TDP (M2, containing 1,728 SNPs), and HS-Aging (M3, containing 1,042 SNPs), as well as two modules associated with the phenotypic pairs, ALS-FTLD (M4, 144 SNPs) and FTLD-HS (M5, containing 121 SNPs). Gene ontology enrichment analysis for biological processes revealed pathways relevant to nervous system development. Notably, M1-ALS uniquely included pathways related to spinal cord, while pathways related to neurogenesis were uniquely observed in M3 HS-Aging. Module-specific PRS were constructed, and the difference between M2-FTLD-TDP and M1-ALS PRS was evaluated to assess relative risk. Higher M2-FTLD-TDP than M1-ALS PRS was observed in both ALS-FTD (n=54) and FTLD-TDP (n=170) cases compared to ALS (n=172) cases (p<.01). Conclusion: Examining genotypes revealed modules of both shared and disparate correlated SNPs across phenotypes associated with TDP-43 proteinopathies which have biologically plausible pathway and phenotype associations. We suggest that genetic variation across TDP-43 proteinopathies may contribute to individual-level presentations of these syndromes.