The pathological sequestration of TAR
DNA-binding protein 43 (TDP-43, encoded by TARDBP) into cytoplasmic pathological inclusions characterizes the distinct clinical syndromes of
amyotrophic lateral sclerosis and behavioural variant
frontotemporal dementia, while also co-occurring in a proportion of patients with
Alzheimer's disease, suggesting that the regional concentration of TDP-43 pathology has most relevance to specific clinical phenotypes. This has been reflected in the three different pathological staging schemes for TDP-43 pathology in these different clinical syndromes, with none of these staging schemes including a preclinical phase similar to that which has proven beneficial in other
neurodegenerative diseases. To apply each of these three staging schemes for TDP-43 pathology, the clinical phenotype must be known undermining the potential predictive value of the pathological examination. The present study set out to test whether a more unified approach could accurately predict clinical phenotypes based solely on the regional presence and severity of TDP-43 pathology. The selection of brain regions of interest was based on key regions routinely sampled for neuropathological assessment under current consensus criteria that have also been used in the three TDP-43 staging schemes. The severity of TDP-43 pathology in these regions of interest was assessed in four clinicopathological phenotypes:
amyotrophic lateral sclerosis (n = 27, 47-78 years, 15 males), behavioural variant
frontotemporal dementia (n = 15, 49-82 years, seven males),
Alzheimer's disease (n = 26, 51-90 years, 11 males) and cognitively normal elderly individuals (n = 17, 80-103 years, nine males). Our results demonstrate that the presence of TDP-43 in the hypoglossal nucleus discriminates patients with
amyotrophic lateral sclerosis with an accuracy of 98%. The severity of TDP-43 deposited in the anterior cingulate cortex identifies patients with behavioural variant
frontotemporal dementia with an accuracy of 99%. This identification of regional pathology associated with distinct clinical phenotypes suggests key regions on which probabilistic pathological criteria, similar to those currently available for
Alzheimer's disease and
dementia with Lewy bodies, can be developed for
TDP-43 proteinopathies. We propose and validate a simplified probabilistic statement that involves grading the presence of TDP-43 in the hypoglossal nucleus and the severity of TDP-43 in the anterior cingulate for the pathological identification of
TDP-43 proteinopathy cases with clinical
amyotrophic lateral sclerosis and behavioural variant
frontotemporal dementia.