The loss of TDP-43 splicing repression, long treated as the single shared lesion across the amyotrophic lateral sclerosis and frontotemporal dementia spectrum, does not fall on the same cells in both diseases, according to an analysis published on February 4, 2026 in Advanced Science by Chen Du, Yinming Li, Rong Wu and colleagues at Wuhan University. Reanalysing 1,603 bulk RNA sequencing samples from the New York Genome Center ALS Consortium alongside published single-nucleus data from patient cortex, the group reports that in the frontal cortex of ALS cases the aberrant splicing junctions were significantly enriched in glial genes, particularly oligodendrocyte genes, while in FTD cases they were enriched in neuronal genes (odds ratio above 2 by Fisher's exact test). The same pattern appeared in temporal cortex, and it tracked the direction of cell loss counted in the single-nucleus data: fewer oligodendrocytes in ALS cortex, fewer neurons in FTD cortex.
From that split the authors drew 31 oligodendrocyte-specific and 507 neuron-specific aberrant junctions as candidate biomarkers. A random forest classifier trained on those signatures was applied to an independent frontal cortex dataset of 60 ALS samples and assigned 56 of them to ALS at a confidence score above 0.8. One oligodendrocyte target, an aberrant junction in FOLH1, was checked directly in patient frontal cortex tissue by RT-PCR and confirmed by Sanger sequencing. Searching published cerebrospinal fluid mass spectrometry data, the team found 12 genes whose aberrant junctions could yield detectable peptides; a peptide from an oligodendrocyte junction in ENPP2 was found in 56 percent of ALS spinal fluid samples and in none of the FTD or control samples.
The context matters because the field's best known cryptic splicing targets, STMN2 and UNC13A, and the HDGFL2 neoepitope reported in spinal fluid by Irwin and colleagues in Nature Medicine in 2024, are all neuronal. The authors note that oligodendrocyte splicing changes have remained largely unconfirmed by comparison.
The work is a retrospective computational reanalysis of postmortem tissue, with one gene validated experimentally. Correlation between aberrant splicing and cell loss is not evidence of which came first, and bulk sequencing signal is itself shaped by which cells have already died. The FTD arm covers TDP-43 pathology only, not the tau forms of frontotemporal dementia, and the independent test set contained ALS cases alone. A separate long-read single-cell study of FTD published in Cell Reports in September 2025 found splicing dysregulation across many cell types, with a substantial share of cell-type-specific change masked when cell types are pooled.