Researchers at Erasmus MC in Rotterdam have published the first dedicated transcriptomic profile of the temporal cortex in frontotemporal lobar degeneration with TDP-43 type C pathology, reporting that genes serving the postsynaptic side of the synapse were consistently underexpressed in diseased tissue. The study, by Ana Rajicic, Merel Mol, Harro Seelaar, John van Swieten, Jeroen van Rooij and colleagues, appeared as advance access in Brain on 6 March 2026. It used bulk RNA sequencing on post-mortem temporal cortices from 18 people with FTLD-TDP type C and 23 sex-matched and age-matched controls without dementia, and identified 6,322 differentially expressed genes at a false discovery rate below 0.05, alongside five disease-related co-expression modules.
FTLD-TDP type C is the pathology found at autopsy in nearly all people diagnosed with semantic variant primary progressive aphasia, also called semantic dementia, in which word meaning erodes while speech remains fluent. Unlike much of frontotemporal dementia, type C is sporadic, with no causal gene identified, yet its clinical and pathological picture is unusually uniform from case to case. That uniformity is the study's premise: if the presentation is shared, a shared molecular pathway may sit beneath it. The broad pattern the authors report, upregulated damage response, cell structure and RNA splicing genes and downregulated synaptic genes, includes changes common to neurodegeneration generally, and genes related to TARDBP, which encodes TDP-43, were enriched among the differentially expressed set.
The more specific claim rests on two filters. The authors compared their results with published RNA sequencing from tauopathies (Alzheimer's dementia, progressive supranuclear palsy and FTLD with MAPT mutations) and from FTLD-TDP subtypes A and B, and removed genes dysregulated in those other datasets. They also applied cell deconvolution to adjust for differences in cell type composition between cases and controls, the statistical correction that addresses the standard objection to end-stage brain tissue, namely that losing neurons looks like switching neuronal genes off. Postsynaptic signal remained after both steps, centred on assembly of receptors at the postsynaptic membrane and on synaptic signal transmission. Eleven synaptic genes were altered at both RNA and protein level in the same cohort, among them CADM1 and NCAN in synaptic adhesion, SLC1A2 in signal transmission, and RAB3B and SYNPO in synaptic plasticity.
The counterpoint is built into the design. This is a cross-sectional autopsy series of 41 brains, sampled once, at the end of a disease that may have run for years; it can show which pathways differ in surviving tissue, not which pathways failed first or why. Cell deconvolution reduces the composition problem without eliminating it, and the protein-level corroboration comes from the same donors rather than an independent cohort. Type C has recently drawn attention for a separate reason: annexin A11 has been reported to co-aggregate with TDP-43 in this subtype, a strand of work that the Erasmus MC group's earlier proteomics contributed to. The authors present their pathways as candidates for further study, not as targets ready for testing.