Abstract Details

We conducted a systematic review and bivariate meta-analysis to determine the pooled diagnostic accuracy of combined plasma biomarkers (NfL and the Aβ₄₂/Aβ₄₀ ratio) for the critical differential diagnosis between Alzheimer’s disease (AD) and frontotemporal dementia (FTD).

Alzheimer’s disease (AD) and frontotemporal dementia (FTD) share early clinical overlap, creating diagnostic uncertainty. Current definitive diagnosis relies on costly, invasive CSF or PET imaging. While individual plasma markers (e.g., Aβ₄₂/Aβ₄₀ ratio or NfL) show promise, their reliability for AD/FTD differentiation is inconsistent. Validating a practical, non-invasive panel requires aggregating data on the combined evaluation of both amyloid pathology and neurodegeneration.

We searched for diagnostic accuracy studies that compared AD (biologically confirmed Aβ⁺status) against FTD (FTLD, bivariate FTD). Eligible studies measured plasma NfL and/or the Aβ₄₂/Aβ₄₀ ratio via high-sensitivity platforms (SIMOA/MS). From 6 reports covering 7 cohorts (over 680 participants), 2×2 contingency data were extracted. Analysis used a hierarchical summary ROC (HSROC) framework via a bivariate random-effects model, with sensitivity analysis probing heterogeneity by FTD subtype.

The HSROC model converged successfully despite sparse data and zero cell events, demonstrating strong overall discriminative power for the combined plasma biomarker panels. Heterogeneity across the studies was low (I² <50%). The pooled Summary Area Under the Curve (AUC) was 0.942 (95% CI: 0.852-0.981). This strong accuracy was confirmed by the pooled Summary Operating Point (SOP) showing, a sensitivity was 76.4% (95% CI: 59.3%-87.8%) and specificity of 86.1% (95% CI: 67.8%-95.7%). Individual study AUCs for the panels ranged up to 0.94, highlighting the reliability of these biomarker combinations.

These findings indicate that combined plasma biomarker panels may provide high diagnostic accuracy for the differential diagnosis between AD and FTD. Supported by an AUC exceeding 0.94, the test's reliability demonstrates its utility as a powerful, accessible screening tool in cognitive clinics to effectively triage patients.