Introduction
Frontotemporal lobar degeneration with tauopathy (FTLD-tau) is a neurodegenerative disease found at autopsy and is the second most common cause of dementia under age 65 [
53]. Much of the complexity in studying FTLD-related dementias is due in part to the phenomenon that the same pathology can be associated with different clinical dementia syndromes. For example, primary progressive aphasia (PPA) is a syndrome characterized by isolated and progressive impairment of language and focal atrophy of regions in the language dominant hemisphere [
30,
33,
34,
51]. In contrast, behavioral variant frontotemporal dementia (bvFTD) is characterized by progressive dysfunction in personality and atrophy in bilateral frontal regions [
1,
28,
48,
54]. Yet, both distinct clinical phenotypes can be associated with the same underlying pathology. In this study, we focus on PPA and bvFTD caused by the FTLD-tau subtype known as Pick’s disease [
48].
Pick’s disease (PiD) is one of three major subtypes of FTLD-tau, first coined by Arnold Pick who observed patients with progressive behavior and/or language deficits and focal frontotemporal atrophy [
24]. Those affected with PiD commonly display “knife-edge” frontal lobe as well as anterior temporal lobe atrophy on structural MRI and subsequently at autopsy [
45,
62]. It is now understood that PiD is defined by the postmortem neuropathology characterized by distinct, round cytoplasmic neuronal tau inclusions called Pick bodies [
21,
57]. PiD is considered a 3R-tauopathy, as its pathologic inclusions consist almost exclusively of pathologic tau containing three microtubule-binding repeat domains [
11]. Conversely, corticobasal degeneration (CBD) and progressive supranuclear palsy (PSP) are 4R-tauopathies, while neurofibrillary tangles in Alzheimer’s disease (AD) consist of both 3R and 4R tau [
7,
12,
16].
The goal of the present study was to investigate the neocortical and hippocampal distributions of Pick bodies in PPA and bvFTD to establish clinicopathologic concordance between PiD and the salience of the aphasic versus behavioral phenotype. Utilizing immunohistochemical techniques and an unbiased stereologic approach, we analyzed up to seven regions, six of which were acquired bilaterally, to closely investigate the relationship between anatomy, clinical syndrome, and regional and hemispheric distributions of Pick bodies. Findings provide further evidence of the notion that dementia symptoms are related to localization of pathology, while also presenting evidence of 3R tau selective vulnerability.
Discussion
Until recently, Pick’s disease was considered a relatively uniform clinicopathologic entity that described a frontotemporal lobar degeneration with pathologic 3R-tau-positive intraneuronal inclusions known as Pick bodies. In actuality, the pathologic diagnosis of PiD can be found in two separate, distinct clinical phenotypes: the aphasic dementia of PPA and the comportmental dementia of bvFTD. The present study aimed to distinguish regional and hemispheric distributions of Pick bodies in a well-characterized cohort of PPA- and bvFTD-PiD individuals. Pick bodies showed a leftward predilection in the neocortex in PPA cases concordant with the aphasic phenotype, while bvFTD cases were on average bilaterally affected. Differential vulnerability of the hemispheres in PPA versus bvFTD was particularly pronounced in the STG, where pathology was significantly left-predominant in PPA and right-predominant in bvFTD. Lastly, the hippocampus—primarily the dentate subregion—was uniquely and universally affected with a high burden of Pick bodies in all cases regardless of clinical phenotype.
The left-lateralization of atrophy in PPA is a core biologic feature of the aphasic phenotype [
15,
34,
38,
41]. Asymmetric pathologic inclusion distribution has been shown in PPA associated with TDP-43 [
26,
27] and in PPA associated with ADNC [
14]. A seminal study found that distribution of AD pathology differed between PPA and dementia of the Alzheimer’s type, where PPA due to AD showed pronounced leftward asymmetry of AD neurofibrillary tangle pathology compared to symmetric distribution in the amnestic phenotype. This study was one of the first to demonstrate clinicopathologic concordance between distribution of the same pathologic entity in different clinical syndromes [
14]. The current study is also the first to demonstrate this concordance in PPA due an FTLD-tauopathy using stereologic analysis. Our findings further showed symmetric overall neocortical predominance of Pick bodies in bvFTD, a finding that has been well-documented in both imaging and histologic analyses [
22,
47,
60]. Overall left neocortical Pick body densities in PPA are comparable to the overall left and right hemisphere Pick body densities in bvFTD, consistent with previous evidence of PPA as a disease of the language dominant hemisphere and bvFTD as a bilateral disease. Surprisingly, findings reveal that the observed asymmetry in PPA-PiD is not characterized by increased left hemisphere pathology, but rather decreased right hemispheric pathology compared to bvFTD-PiD. Such results provide insight into the progression of PiD in these phenotypes, where the right hemisphere appears relatively spared in PPA-PiD until later disease stages, while PiD underlying bvFTD may engender and spread bilaterally.
The STG emerged as an interesting region of differential vulnerability, whereby PPA cases showed significant left-predominant pathology while bvFTD patients showed significant right-sided predominance. The left STG is a well-established component of the human language network, though its exact function is still debated. The left STG is classically associated with Wernicke’s area, which was first characterized as a site of language comprehension based on lesion anatomy in stroke patients [
50,
59]. However, neurodegenerative research reveals that comprehension deficits can arise from a constellation of cortical atrophy sites in addition to the classic Wernicke’s area/left STG [
2,
10,
40,
46]. Several PPA studies have instead found the anterior and posterior segments of the left STG to underly separate and distinct functions that correlate with different language deficits [
39,
61]. The anterior temporal lobe, which includes the anterior STG, is found to assist with functions such as word comprehension and object naming, the proposed functions of Wernicke’s area [
9,
19,
36,
42]. Conversely, the posterior STG supports the phonological loop, which aids in word repetition, sentence comprehension, and the storage of auditory speech input in working memory for future articulation [
17,
39,
55]. The agrammatic phenotype more closely aligns with posterior STG impairment, which correlates with our pathologic findings. It is important to note that while a minority of cases show moderate to frequent neuritic plaques, this pathologic entity likely does not contribute to the aphasic or behavioral phenotype [
14]. Though the functional modularity of the right temporal lobe is less understood, it is generally responsible for auditory and emotional processing [
5,
43,
63]. In FTD cohorts, right-predominant temporal lobe atrophy is most often associated with severe comportment and personality changes [
23,
56]. Right-asymmetric temporal PiD pathology in bvFTD is therefore concordant with observed behavioral symptoms. However, bvFTD-PiD shows peak atrophy patterns in bilateral prefrontal and anterior temporal cortices; therefore, PiD pathology in right STG may serve as a neighborhood marker for more severe burden in anterior regions.
Despite relative sparing of memory functioning in the behavioral and aphasic phenotypes, we found the hippocampus to be greatly affected in both bvFTD- and PPA-PiD. The selective vulnerability of the granule cells of the dentate gyrus in PiD has previously been reported [
11,
12,
21,
25]. A recent study by Mesulam et al. (2021) demonstrated that PPA patients with underlying AD neuropathology show preserved memory functioning despite hippocampo-entorhinal postmortem neuropathologic burden that is comparable to those with amnestic dementia due to AD [
37]. The finding that DG and CA1 pathology is significantly greater than overall neocortical pathology in both diseases is striking. In a recent study, our group found that, on average, about 60% of dentate granule cells contained a Pick body [
25]. Yet despite intense hippocampal PiD pathologic burden, PPA and bvFTD patients show relative sparing of memory functioning. One variable that was thought to account for increased Pick body density in the hippocampus is higher packing density of granule cells in the DG and, to a lesser extent, pyramidal cells in the CA1, compared with the packing density of neocortical neurons. However, when we compared the inclusion-to-neuron ratios of granule cells to neurons in the left MFG in three PPA cases, we found that DG granule cells
still showed significantly greater burden of PiD. Further, the average inclusion-to-neuron ratio from the three chosen cases (66.9%) aligns with previous published findings of 60% [
25]. Hippocampal cells thus appear to be resilient to potential deleterious effects of Pick-related tau accumulation. One possible explanation for this resilience is that while the Pick body may form early in disease course, their presence may not necessarily lead to neurodegeneration; indeed, our group has shown relative preservation of granule cells in the DG in both PiD and TDP-43 proteinopathies, despite pathologic accumulation [
25,
26]. Another explanation is that despite early arrival to the dentate gyrus, PiD may progress at a slower rate in the hippocampus compared to neocortex, leading to preserved memory until later disease stages. Nevertheless, we found the DG showed at least 3-times more Pick bodies than neocortical regions in both bvFTD and PPA. Results suggest the hippocampus is vulnerable to Pick body pathologic accumulation, regardless of phenotype, but this vulnerability is not necessarily associated with expected impairments in memory domains.
Despite a unique clinical presentation of semantic impairments, Case 11 followed similar patterns of pathologic distribution compared to the other PPA cases with agrammatic presentations. There was one notable exception in regional vulnerability; the semantic PPA case was more heavily affected by Pick body pathology in the MFG whereas the agrammatic cases showed greater STG involvement. While differences between PiD causing agrammatic versus semantic deficits cannot be extrapolated from one case, comprehensive analyses can offer clues regarding relative spread and associated cognitive dysfunction.
The present study contains one of the largest cohorts of autopsy confirmed PiD leading to two disparate clinical phenotypes. Stereological analysis provided rigorous quantitative measures of inclusion density that allow for reliable comparison of pathologic density between a large host of regions. Limitations include small sample size and lack of ATL data in the bvFTD group and neuronal density data from all cases. Given severe degeneration of tissue, neuronal density data would illuminate whether lower pathologic inclusion density correlates with lower neuronal number, implying inclusions are cleared as neurons are lost [
26]. Lastly, our modified stereologic methodology analyzes adjacent 5 µm sections; this methodological approach may result in overestimation of inclusion densities. Subsequent analyses will include stereological quantification of subcortical regions in PiD as well as collection of similar data in 4R-FTLD-tauopathies, corticobasal degeneration and progressive supranuclear palsy, to determine shared versus unique neuropathologic signatures of 3R and 4R tauopathies. Additional analyses are also needed to understand the contributions of neuroinflammatory, synaptic, and genomic hallmarks that may lead to distinct clinical phenotypes within the same pathology. Together, these studies may determine the individual features that create the complex clinicopathologic picture of frontotemporal dementias.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.