Objective: This study investigated the key neuropathological features that could separate corticobasal degeneration (CBD) from related parkinsonian disorders by using a decision tree algorithm.

Background: The CBD neuropathologic criteria were defined and validated, but it is unknown which histopathological features could easily distinguish CBD from related disorders.

Method: We used data from the study that validated the CBD pathologic criteria (Dickson et al. 2002) in which six neuropathologists individually rated the histopathological features of 34 cases being unaware of their diagnosis. The cases consisted of slides of 9 CBD, 10 progressive supranuclear palsy (PSP), 6 Pick’s disease, 5 frontotemporal dementia and Parkinsonism linked to chromosome 17 (FDTP-17), 5 frontotemporal lobar degeneration lacking distinctive histopathology (FTLD), and 5 argyrophilic grain disease (AGD). The data was rated using semi-quantitative methods (0: none, 1: mild, and 2: severe). 
The decision tree analysis used the Towing splitting criteria to split the data into subsets.  The results were cross-validated.

Results: Severe centrum semiovale threads was the first decision-tree split that separated 90% of the CBD cases from other diseases, whereas no or mild neuronal loss in the entorhinal cortex separated 87.6% of the CBD cases. The third split, severe pontine tegmentum threads separated 70% of the CBD cases.  The cross-validation accuracy was 83.3%.  On the other hand, allowing 9 splits led to a cross-validation accuracy of 88.7% (Figure 1).

Conclusion: Severe centrum semiovale threads accurately separated 90% of the CBD cases from other diseases, and none or mild neuronal loss in the entorhinal cortex separated 87.6% of the CBD cases.   A limitation of the study is that astrocytic plaques were not rated because it had not been defined at the time of the study. Nevertheless, three areas can lead to a high accuracy in identifying CBD.

« Back to MDS Virtual Congress 2020

MDS Abstracts - https://www.mdsabstracts.org/abstract/key-neuropathological-features-separating-corticobasal-degeneration-from-other-related-disorders/