Accuracy of clinical diagnosis in neurodegenerative diseases - a study of 455 autopsy cases

TL;DR

This study evaluates how accurate clinical diagnoses of neurodegenerative diseases are compared to autopsy results, finding significant variation in accuracy across diseases.

Contribution

The study provides detailed accuracy metrics for clinical diagnoses of nine specific neurodegenerative diseases using neuropathological data from 455 cases.

Findings

Clinical sensitivity varied widely (0–100%) while specificity was consistently high (89.5–100%).

Accuracy was very good (AUC > 0.9) for Huntington’s disease, motor neuron disease, and multiple system atrophy.

Argyrophilic grain disease had no discriminatory capacity (AUC = 0.5).

Abstract

Precision of clinical diagnosis in neurodegenerative diseases is critically important for clinical care and study recruitment. This study aimed to investigate the clinical accuracy using gold-standard neuropathological reference.

Neuropathological diagnoses from the Neurobiobank München were correlated with real-world clinical diagnoses from hospitals in Germany. Accuracy metrics, including sensitivity, specificity, and area under the curve (AUC) of clinical diagnoses, were calculated.

Among nine neuropathologically diagnosed neurodegenerative diseases (Alzheimer’s disease, argyrophilic grain disease, corticobasal degeneration, frontotemporal lobar degeneration, Huntington’s disease, Lewy body disease, motor neuron disease, multiple system atrophy, and progressive supranuclear palsy) with a total of 455 cases, clinical sensitivity varied widely (0–100%) whereas specificity was consistently high (89.5–100%). Accuracy was very good (AUC > 0.9) for Huntington’s disease, motor neuron disease, and multiple system atrophy; good (AUC = 0.8–0.9) for Alzheimer’s disease, dementia with Lewy bodies/Parkinson’s disease, and progressive supranuclear palsy; moderate (AUC = 0.7–0.8) for frontotemporal dementia, limited (AUC = 0.51–0.7) in the n = 20 cases with corticobasal degeneration, and no discriminatory capacity (AUC = 0.5) in the n = 6 cases with argyrophilic grain disease.

Clinical diagnostic accuracy of neurodegenerative diseases varies, with sensitivity as the main limiting factor. Improving diagnostic sensitivity will be essential for early and accurate patient identification, especially as disease-modifying therapies targeting causal proteinopathies become available. Achieving this will depend on the development and clinical implementation of reliable molecular biomarkers that indicate the causal proteinopathies of neurodegenerative diseases.

Figures