Tau368 improves p-tau diagnostic accuracy for FTLD-tau from FTLD-TDP.

Researchers struggle to distinguish between different types of dementia that look nearly identical in a clinical setting. While Alzheimer’s disease (AD) has robust fluid biomarkers to guide diagnosis, other neurodegenerative conditions—specifically frontotemporal lobar degeneration (FTLD)—remain a diagnostic "black box." FTLD-tau (a type of dementia caused by tau protein abnormalities) and FTLD-TDP (caused by the TDP-43 protein) often manifest with the same behavioral or language symptoms. This makes it difficult for clinicians to identify the underlying biological cause.

A new study published in Acta Neuropathologica proposes a way to resolve this ambiguity. By measuring a specific fragment of the tau protein called tau368 and using it as a "controller" to adjust other measurements, the authors report they can significantly improve the accuracy of distinguishing FTLD-tau from FTLD-TDP. This approach uses one protein fragment to normalize for the natural biological noise that typically obscures these signals in cerebrospinal fluid (CSF).

The diagnostic gap in frontotemporal dementia

Current diagnostic workflows for dementia rely heavily on identifying specific protein signatures in the CSF. For Alzheimer’s disease, this is highly effective. Biomarkers like phosphorylated-tau (p-tau181 and p-tau212) are significantly elevated. They serve as reliable proxies for amyloid and tau pathology.

However, the authors note that there are currently no robust or reliable fluid biomarkers specific to FTLD-tau. This lack of specificity creates a massive hurdle for clinical trials. FTLD is exceptionally heterogeneous, meaning it presents with vastly different clinical and biological profiles. This makes it difficult to recruit the right patients for targeted therapies. Many patients with FTLD-tau also harbor Alzheimer's neuropathologic changes (ADNC), which can contaminate biomarker readings. As shown in, while AD patients exhibit massive spikes in p-tau and total tau (t-tau), the differences between various FTLD subtypes are much subtler and harder to catch with standard assays alone.

Using tau368 as an inter-individual controller

The researchers' core insight lies in the dual nature of tau fragments. Most current assays target the N-terminal or mid-domain of the tau protein to measure phosphorylation (a chemical modification that often signals disease). However, the authors focus on tau368, a fragment from the microtubule-binding region (MTBR) that represents the "core" of aggregated tau.

The authors hypothesize that tau368 can act as an inter-individual "controller." Think of it like a reference voltage that stabilizes a circuit despite fluctuations in power supply. Their approach follows these logic steps:

1. Identify the noise: Individual variability in CSF production, aging, and protein handling creates a "baseline" level of tau that varies between people.
2. Measure the signal and the noise: Instead of looking at p-tau levels in isolation, the authors measure p-tau (the disease signal) alongside tau368 (the stability reference).
3. Calculate the ratio: By creating ratios like p-tau181/tau368 or p-tau212/tau368, they effectively normalize the phosphorylation signal against the person's specific level of tau aggregation.

The authors demonstrate that tau368 immunoreactivity (the visibility of the protein under a microscope) increases alongside the severity of neurofibrillary tangles in the brain, as seen in .

This confirms that tau368 is a meaningful proxy for the actual physical burden of tau pathology.

Improved accuracy in the ADNC-negative subcohort

To prove this worked, the authors focused on an "ADNC-negative" subcohort. These were patients who had little to no Alzheimer's-related pathology. This allowed them to isolate the signals belonging strictly to FTLD. The results were striking.

The paper reports that the p-tau212/tau368 ratio achieved an Area Under the Curve (AUC) of 0.95 when discriminating FTLD-tau from FTLD-TDP. An AUC of 0.95 indicates near-perfect diagnostic accuracy. For comparison, the p-tau181/tau368 ratio reached an AUC of 0.93. These are high-performance metrics for a diagnostic tool. The authors also found that these ratios were significantly higher in patients presenting with behavioral or language variants (PPA) of tauopathy compared to those with TDP-43 pathology, as illustrated in .

Crucially, the authors found that these ratios correlate more strongly with the actual physical tau burden in the brain than p-tau or t-tau measured alone. In, the researchers show that while tau368 itself doesn't correlate well with regional brain tau burden, the ratio of p-tau to tau368 shows significant Spearman correlations across various brain regions.

This suggests the ratio is a much cleaner window into the pathology.

Limitations and data constraints

While the results are promising, several caveats remain. First, the study relies on a "rare autopsy cohort." Because many of the definitive diagnoses were confirmed post-mortem, the findings are grounded in high-quality truth. However, they are inherently limited by the small sample sizes of specific clinical and pathological subgroups.

Second, the authors acknowledge that the study's clinical and racial diversity was limited. A biomarker must be validated across diverse populations to ensure that "inter-individual variability" does not include systemic biases. Finally, the study is cross-sectional. It captures a snapshot of the disease at the time of CSF collection. It does not track how these ratios evolve as a patient moves from early to late-stage dementia.

The verdict: A specialized tool for clinical trials

Is this ready for the clinic? Probably not yet, but it is a vital step for research. The authors present this as a "proof-of-concept" study. For a practitioner in a general neurology clinic, the immediate utility is low. These specific ratios are not yet part of standard diagnostic panels.

However, for the pharmaceutical industry, the verdict is much stronger. The ability to precisely select FTLD-tau patients for clinical trials is a high-value capability. Researchers can now filter out those with confounding Alzheimer's pathology. If the p-tau/tau368 ratio can reliably separate these proteinopathies, it becomes a powerful tool. It ensures that the right drugs reach the right patients. The transition from "standard p-tau" to "normalized p-tau ratios" represents a move toward much higher resolution in neurodegenerative diagnostics.

Figures from the paper