Meta-Analysis of ALS Post-Mortem Cerebral Cortex Transcriptome Studies Identifies Robust Shifts in Gene Expression Despite Significant Between-Study Heterogeneity

Abstract

Background: ALS is characterized by motor neuron failure with potential neurocognitive comorbidity. Cortical neurons, and their support cells, exhibit pathologic features, some of which have been identified in post-mortem transcriptome studies of the cortex from ALS patients and control subjects. These studies, however, have utilized varying technologies with few attempts to reconcile findings. Methods: This study used meta-analysis to evaluate differential gene expression (ALS vs. control) from 16 prior transcriptome studies utilizing multiple technologies (microarray, bulk RNA-seq or snRNA-seq). Normalization with GC content correction was used to mitigate downstream biases impacting differential expression. Results: Approximately one-third of protein-coding genes (5930 of 16,579) exhibited significant between-study effect size heterogeneity (FDR < 0.05), which was frequently associated with variation in average post-mortem interval. Nonetheless, we identified 269 and 273 differentially expressed genes (DEGs) with relatively consistent increased or decreased expression in the ALS cortex, respectively (FDR < 0.05 with SMD ≥ 0.20 or SMD ≤ −0.20). ALS-increased DEGs were associated with transcription (RNA polymerase), cell signaling (TGF-beta, PI3K), interneurons, microglia, endothelial cells, and deep laminar layers (L5, L6). ALS-decreased DEGs were associated with transcription (RNA polymerase), glial cells (astrocytes, oligodendrocytes) and the L1 layer. There was significant overlap between DEGs from the ALS cortex and those previously identified from ALS spinal cord and blood. Conclusions: These findings highlight signaling pathways, cell types and cortical layers associated with robust changes in ALS cortex gene expression. Our results illustrate the importance of meta-analysis as a tool for understanding between-study heterogeneity and identifying repeatable findings across technologies.