From symptoms to biology: the case for a unified framework

Fifteen years ago, the Alzheimer’s field shifted from defining disease by symptoms to defining it by underlying biology — amyloid and tau. A parallel move is now being proposed for TDP-43, sometimes framed as TDP-43-Associated Neurodegeneration (TAN): lumping these historically separate diagnoses under one biological definition anchored on the hallmark pathology. The rationale is practical as much as conceptual. A biological definition enables earlier detection, cleaner clinical-trial enrichment, and shared therapeutic targets across ALS, FTD, and TDP-43-driven dementias. But a biological definition is only as good as the tools available to measure the biology — which is where the field hits its central bottleneck.

The Alzheimer’s analogy is instructive. The amyloid/tau framework only became actionable once amyloid PET and, later, blood-based p-tau assays let researchers confirm pathology in living patients, enrich trials with biomarker-positive participants, and read out target engagement. TDP-43 has no equivalent validated tool yet. That gap explains why so much energy at recent scientific meetings has shifted from arguing about nomenclature to solving measurement — a validated TDP-43 biomarker is the rate-limiting reagent for the entire TAN program, from patient stratification to demonstrating that a candidate therapy actually lowers pathology.

The core problem: direct TDP-43 measurement has not delivered

The obvious approach — measure soluble TDP-43 directly in cerebrospinal fluid (CSF) or blood — has repeatedly disappointed. Across assay platforms, soluble TDP-43 in biofluids yields only a weak, inconsistent signal that cannot reliably support differential diagnosis or clinical trials. The pathological protein is largely aggregated and sequestered, the healthy protein is ubiquitous, and blood in particular is dominated by a high-abundance plasma proteome that swamps neuronal signal. As a result, the field has pivoted toward four indirect but more tractable strategies.

There is also a practical dimension to where a biomarker is measured. CSF requires a lumbar puncture, which limits screening, repeat sampling, and use in frail elderly patients — exactly the LATE population. A blood-based test would enable population-scale screening, longitudinal monitoring of progression, and decentralized clinical trials. The prize, therefore, is not merely any TDP-43 signal but one that survives the journey into peripheral blood while remaining specific to neuronal pathology.

Four biomarker strategies gaining ground

1. Cryptic peptides — reading TDP-43’s loss of function

When TDP-43 leaves the nucleus, its splicing-repression function fails and cryptic exons appear in transcripts such as HDGFL2, STMN2, and UNC13A. Some in-frame cryptic exons encode novel peptides that can be detected in biofluids. A landmark Nature Medicine study showed a cryptic HDGFL2 neoepitope is elevated in CSF — and even in presymptomatic C9orf72 carriers — with ultra-high-sensitivity assays reaching limits of detection near 0.016 pg/mL [5], [6]. Because expression varies between patients, groups are building multi-marker panels (HDGFL2, STMN2, UNC13A, and others), and a recent NfL:HDGFL2 ratio outperformed either marker alone for discriminating ALS/FTD from controls and tracked progression [7].

2. Seed amplification assays — catching misfolded conformers

Borrowed from the prion field and already transformative in Parkinson’s and Lewy body dementia, seed amplification assays (SAA / RT-QuIC) use patient-derived misfolded TDP-43 as a template to nucleate aggregation of recombinant monomer, generating an amplified fluorescent readout. New digital SAA formats isolate individual aggregates in nanoliter compartments and have quantified elevated TDP-43 seeds in FTLD-TDP CSF that correlate with severity [8], [9]. The studies remain small and somewhat inconsistent, and a persistent technical hurdle is the reproducible production of recombinant TDP-43 substrate.

3. PET imaging — seeing the pathology in vivo

A TDP-43-selective PET tracer would let clinicians visualize and stage pathology directly. The first-in-class candidate [18F]ACI-19626 was characterized in Nature Communications in 2025, showing selective binding to aggregated TDP-43 over amyloid-β, tau, and α-synuclein; early first-in-human imaging reported higher tracer uptake in expected regions in C9orf72-FTD than in controls [10]. Multiple academic and industry groups are racing to develop competing tracers, though sensitivity to intracellular, conformationally heterogeneous aggregates remains the key challenge.

4. Extracellular vesicles — a route to a blood test

Repeatedly highlighted as one of the most promising paths to a blood biomarker, extracellular vesicles (EVs)concentrate proteins and RNA of neuronal origin and can be immuno-isolated by cell of origin, bypassing the high-abundance plasma background that has defeated conventional assays [11]. Critically, neuron-derived EVs can carry both aggregated TDP-43 and the downstream cryptic-peptide and RNA targets described above — offering multiple orthogonal analytes from a single peripheral draw [14].

The remaining challenges for EV biomarkers are analytical rather than conceptual: reproducible, high-specificity capture of neuron-derived vesicles from plasma, standardized workflows, and demonstrating that the TDP-43 species recovered genuinely reflect CNS pathology rather than peripheral background. These are precisely the problems that determine whether a promising research signal becomes a validated clinical assay — and they are the problems on which the next wave of the field will be judged.