Quantitative Assessment of Pathological Tau Burden in Essential Tremor: A Postmortem Study

Abstract

Essential tremor (ET) patients develop more cognitive impairment and dementia than controls, although there are surprisingly few data on the neuropathological basis for cognitive changes in ET. In this postmortem study, we assessed tau and other pathologies in 26 ET cases and 73 controls (non-ET) (1:3 matching). The mean age = 88.6 years; 55% were cognitively normal, 24% had mild cognitive impairment (MCI), and 20% had dementia. We found similar burdens of pathology using Braak, β-amyloid and Lewy body assessments in ET and controls. In contrast, among cognitively normal subjects, ET cases had a higher number of NFT-positive neurons in the neocortex than controls (p < 0.001); the number of NFT-positive neurons in the medial temporal lobe was similar in these 2 groups (p = 0.22). Among subjects with MCI, ET cases also had higher numbers of NFT-positive neurons in the neocortex than controls (p < 0.001) but again, not in the medial temporal lobe (p = 0.55). Among subjects with dementia, the number of NFT-positive neurons was similar in ET cases and controls. Cognitive function correlated with quantitative neurofibrillary tangle counts in ET cases and controls. In the context of ET, pre-dementia tau burden is higher than in the absence of ET, suggesting a predisposition to tau pathology.

INTRODUCTION

Essential tremor (ET) is a chronic, progressive and often disabling neurological disease whose hallmark clinical feature is an 8–12 Hz kinetic tremor of the arms (1). ET is among the most prevalent neurological diseases, is often familial, and affects an estimated 2.2% of the US population (i.e. approximately 7 million individuals) (2). The pathophysiology of ET is not fully understood, although in controlled postmortem studies, a range of pathological changes of a neurodegenerative nature have been documented in the cerebellar cortex; these are centered on the Purkinje cell and neighboring neuronal populations (3–6).

Given that its central clinical feature is tremor, ET has historically been viewed exclusively as a motor disorder. However, there is emerging evidence that non-motor features also occur (7–10). An increasing number of independent studies from around the world have demonstrated that ET cases have poorer cognitive performance than age-matched controls (11). There is also an emerging understanding that the cognitive problems in ET patients may be progressive (12), can become severe, and that the rate at which this occurs is beyond what occurs in age-matched controls without ET. One epidemiological study found an association between ET and mild cognitive impairment (MCI) (odds ratio = 1.57) (13). Two prospective, population-based, epidemiological studies, one in Madrid and the other in New York, demonstrated an association between ET and a clinical diagnosis of probable Alzheimer disease (AD) (14, 15). In these studies, 11.4%–25.0% of ET cases (mean age 79.1–80.9 years) had prevalent dementia versus only 6.0%–9.2% controls (14, 15). Furthermore, in both studies, the risk of incident dementia at follow-up was higher in ET cases who were non-demented at baseline than controls (relative risks = 1.64–1.89) (14, 15), indicating that for reasons that currently remain unclear, ET is associated with an increased risk of developing dementia. While cognitive impairment in ET may be mediated by the processes or pathways specifically involved in the disease (i.e. cerebellar pathology producing a cerebellar cognitive syndrome), an alternate and likely hypothesis is that cognitive impairment and dementia may occur as a result of concomitant neurodegenerative processes. In the general population, AD neuropathological changes including β-amyloid peptide accumulation in β-amyloid plaques and abnormal intracellular accumulation of hyperphosphorylated forms of the microtubule associated protein tau in neurofibrillary tangles (NFTs) are the most common drivers of MCI and dementia, followed by Lewy body pathology (16). However, there are surprisingly few data on the neuropathological basis for cognitive changes in ET. In one study, 40 ET patients had a higher Braak NFT stage than 32 controls (means: 2.2 ± 1.2 vs 1.2 ± 1.1; medians: 2.0 vs 1.0, p < 0.001) (17). In contrast, Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) scores for neuritic plaques were similar in ET patients and controls (means: 0.6 ± 0.9 vs 0.5 ± 0.6; medians: 0.0 vs 0.0, p = 0.83) (17). While the study raised the possibility that ET may predispose individuals to accumulate neuronal tau aggregates, and thus that tau may play a central role in the cognitive impairment that can accompany ET, the study was limited by (i) the use of a semi-quantitative rather than quantitative measure of NFTs and (ii) limited data on cognitive functioning. There is no further data of which we are aware.

The objective of this study was to determine whether or not ET is accompanied by greater neuropathological tau burden than expected for age, advancing beyond the prior study (17) by using a more quantitative and precise measure to assess NFTs in a sample of ET cases and non-ET controls previously assigned cognitive diagnoses based on comprehensive neuropsychological testing. Thus, we deployed semi-quantitative neuropathological assessments and disease staging systems, including the CERAD neuritic plaque burden and the Braak NFT stage (18), the widely used gold standard for assessing tau pathology given its convenience and degree of inter-rater reliability (16). We also obtained regional individual NFT-positive neuron counts, which can increase power to reveal clinically and biologically relevant correlations and elucidate how ET may interact with other neurodegenerative diseases (19, 20). Finally, we examined the association between NFT counts and cognitive performance across 3 domains including attention and processing speed, episodic memory, and working memory.

MATERIALS AND METHODS

Subjects

ET cases were enrolled in an ongoing longitudinal, prospective study of cognitive function in ET (Clinical Pathological Study of Cognitive Impairment in Essential Tremor, NINDS R01NS086736), which commenced enrollment in July 2014. The purpose of the study is to characterize the clinical features of a cohort of ET cases through the use of motor, neuropsychologic, and neuropsychiatric measures. The design of the study, including a description of the study assessments, the detailed cognitive test battery, the diagnosis of ET, and the assignment of diagnoses of cognitively normal, MCI and dementia, has been detailed elsewhere (21, 22). All ET cases were assessed longitudinally with periodic assessments until death. Specifically, neuropsychological evaluations were performed at baseline and every 18 months in the participants’ homes for the duration of the participants’ lifespan (21, 22). After the case had expired, brain tissue was evaluated at the Essential Tremor Centralized Brain Repository (Columbia and Yale Universities). The next of kin provided written consent for participation and brain donation. Institutional Review Board approval for collection of clinical data and brain donation was approved at Yale University and Columbia University Medical Center (CUMC).

The non-ET comparison group was derived from the Memory and Aging Project (MAP) at Rush University Medical Center (RUMC). All participants agreed to an annual detailed clinical evaluation and donation of brain, spinal cord, nerve, and muscle at death. Participants provided written informed consent and signed an anatomical gift act. The study was approved by the institutional review board of RUMC. Cases from this well-characterized cohort, with prospectively collected neuropsychological data, a modified Unified Parkinson’s Disease Rating Scale score (23), and extensive neuropathological data including quantitative tau-positive cell counts in 5 brain regions, were selected as controls. Details of the study have been previously reported (24).

For the current analyses, inclusion criteria for controls included group matching within each cognitive category (normal cognition, MCI, dementia) by age, gender, education, ethnicity (all Caucasian), duration from cognitive evaluation to death, and performance on 6 cognitive tests, which are described below. Exclusion criteria for controls included a diagnosis of ET or other movement disorder (e.g. Parkinson disease), or any signs of parkinsonism. Of these tissues, 73 normal elderly control brains from the cohort were selected and 3:1 group matched to the ET cases.

Cognitive Evaluation

The cognitive batteries administered to ET and controls shared 6 tests including measures of global cognition (Mini Mental Status Exam [MMSE]) (25), attention and processing speed (i.e. Digit Span Forward [26] and Oral Digit Symbol Modalities Test [27]), episodic memory (i.e. Immediate and Delayed Story Memory [26]), and working memory (Digit Span Backwards [26]).

Neuropathological Evaluation

For the ET brains evaluated at CUMC, at least 18 brain regions were sampled and examined following formalin-fixation and paraffin embedding as per our published protocol (28). Briefly, 7-μm-thick sections were stained with Luxol fast blue and counterstained with hematoxylin and eosin (H&E) or modified Bielschowsky silver stain. Immunohistochemical studies were performed using antibodies directed against α-synuclein, hyperphosphorylated tau (AT8), and β-amyloid (4G8, Invitrogen, Carlsbad, CA) as described elsewhere (28). A comprehensive neuropathological assessment was performed by a senior neuropathologist (Supplementary DataTable S3). A detailed description of the criteria used to assign neuropathological diagnoses can be found in “Twenty-first century brain banking. Processing brains for research: the Columbia University methods” (28). Sections were imaged using Aperio Digital Pathology Slide Scanners (Leica Biosystems, Buffalo Grove, IL) and counts performed. For the controls, a similar neuropathological evaluation was performed (Supplementary DataTable S2), with slight differences including systematic sampling protocols, and minor differences in sectioning and staining techniques, which can be reviewed elsewhere (29). While variability in staining quality and intensity was observed, neurofibrillary tangle staining was discernable. All reagents and antibodies were purchased from Thermo Fisher Scientific (Waltham, MA) unless otherwise noted.

NFT counts were performed by manually counting each pathological feature on modified Bielschowsky silver stained sections (both the controls and ET cases). Counts were performed in neocortical (frontal cortex, precuneus) and medial temporal regions (parahippocampal gyrus, CA1 of the hippocampus, entorhinal cortex) (Fig. 1). Given that NFT-positive cell counts had previously been performed on the control group as described by Bennett et al (30), a subset of controls (n = 20) from the Rush MAP cohort were blindly analyzed in 2 brain regions by the investigator at Mount Sinai who performed the counts on the ET brains, to assess inter-rater reliability. These sections were prepared and stained at Rush, and the digital images were transferred to the investigator at Mount Sinai and analyzed. The correlation between NFT counts performed by Rush investigators and ratings in this study in CA1 (Spearman’s r = 0.92, p < 0.001) and entorhinal cortex (Spearman’s r = 0.80, p < 0.001) were excellent.

FIGURE 1.

Representative Bielschowsky silver-stained brain sections. Neurofibrillary tangles were counted in the frontal cortex (A), medial temporal lobe (B), and parietal cortex (C). Counts were performed on positively stained neurons (inset). PHG, Parahippocampal gyrus; CA1, cornu ammonis 1; EC, entorhinal cortex; FC, frontal cortex; PC, parietal cortex (precuneous). Scale bar = 5 mm.

In ET cases, Purkinje cells and torpedoes (i.e. swellings of the Purkinje cell axon) were quantified from a standard 3- × 20- × 25-mm formalin-fixed tissue block obtained from each fixed hemi-brain. Next, a parasagittal slice was taken approximately 1–1.5 cm from the cerebellar midline containing anterior and posterior quadrangulate lobules and the underlying dentate nucleus. Purkinje cells and torpedoes were quantified in a single 7-μm thick, Luxol fast blue/H&E-stained section from that block. Purkinje cells were quantified by counting and averaging the number of Purkinje cells across 15 100× non-overlapping microscopic fields (31). Similarly, torpedoes were counted in one entire Luxol fast blue/H&E section, as described in (32). Both the Purkinje cell and torpedo counts were then normalized to Purkinje cell layer length (i.e. divided by Purkinje cell layer length in mm) to account for any potential variations in amount of cerebellar cortex in the tissue block.

Statistical Analysis

We first used a correlation matrix to determine whether the NFT counts in each of the 5 regions were correlated; these analyses used data from the entire sample. As expected, counts from the medial temporal lobe (parahippocampal gyrus, CA1 of the hippocampus, entorhinal cortex) were significantly correlated (Spearman’s r = 0.51–0.72, p < 0.001), and counts from the neocortex (frontal cortex and precuneus) were similarly significantly correlated (Spearman’s r = 0.47, p < 0.001), whereas counts across these 2 regions were less correlated. A factor analysis (principal component method with varimax rotation) also indicated that the NFT counts from the 5 regions loaded on 2 factors (factor 1 = parahippocampal gyrus, CA1 of the hippocampus, entorhinal cortex; factor 2 = frontal cortex and precuneus). Hence, as similarly described by Bennett et al, we created a summary measure of NFT counts for each of these two regions: medial temporal lobe and neocortex (30). As described previously, for each summary measure we converted raw NFTs counts to a standard distribution by dividing each person’s count by the standard deviation for that particular count (30). The summary measures summed the standardized counts and divided by the number of regions counted. We compared the number of NFT-positive neurons in ET and controls Wilcoxon rank sum test. Torpedo and Purkinje cell counts were compared to NFT summary measures and MMSE using Spearman’s correlation coefficients. Similarly, Spearman’s correlation coefficients were used to assess whether NFT counts were correlated with global cognition, attention and processing speed, episodic memory and working memory. Lastly a sensitivity (i.e. secondary) analysis was performed, excluding the 2 ET cases with Lewy bodies, to see whether this influenced our primary results. All computations were performed in SPSS (IBM, Armonk, NY) and R software packages.

RESULTS

There were 26 ET cases that were included in the study (Table 1). Among the ET group, the average age of death was 90.1 years (±4.97, range: 74–99). There were 9 males (34.6%) and 17 females (65.4%). They were categorized as cognitively normal (n = 11), (MCI; n = 8) and dementia (n = 7). The median age of onset of ET was 55 years and was also 55 years in those ET cases with dementia; hence, in those ET cases with dementia, the ET diagnosis preceded the dementia by many years.

TABLE 1.

Subject Data

	ET	Non-ET	Total
Sample size, n	26	73	99
Age of death, yr (%)
70–79	1 (3.8)	4 (5.5)	5
80–89	10 (38.5)	43 (58.9)	53
90–100	15 (57.7)	23 (31.5)	38
101–110	0 (0.0)	3 (4.1)	3
Sex
Male (%)	9 (34.6)	44 (60.3)	53
Female (%)	17 (65.4)	29 (39.7)	36

ET, essential tremor.

These ET cases exhibited a range of neurodegenerative pathology (Table 2). Mild or no neuritic β-amyloid plaque pathology was observed in 9/26 (34.6%) ET cases. Cerebral β-amyloid angiopathy was present in 10 ET cases (38.5%). Tau pathology was nearly universally present in ET cases (25 of 26, 96.2%). Hippocampal sclerosis was present in 3 ET cases (11.5%) and Lewy body pathology was detected in 2 ET cases (7.7%). Normalized Purkinje cell counts and torpedo counts are shown in Table 2. These values (means of 4.07, 4.31, and 4.52 in our 3 ET groups) are slightly below the mean value (4.9 ± 0.7) reported in 25 controls in a prior analysis (33).

TABLE 2.

Neurodegenerative Disease Pathology in Cognitively Assessed ET Patients and Non-ET (Controls)

	ET, n (%)			Non-ET, n (%)			p*
	Normal	MCI	Dementia	Normal	MCI	Dementia	Normal	MCI	Dementia
Sample size, n (%)	11 (42.3)	8 (30.8)	7 (26.9)	44 (60.3)	16 (21.9)	13 (17.8)
Aβ plaques
Sparse/absent	5 (45.1)	1 (12.5)	3 (42.8)	26 (59.0)	2 (12.5)	1 (7.6)	0.50	0.72	0.10
Present	6 (54.5)	7 (87.5)	4 (57.1)	18 (40.9)	14 (87.5)	12 (92.3)
Braak NFT stage
None (0)	1 (9.0)	0 (0.0)	0 (0.0)	1 (2.2)	0 (0.0)	0 (0.0)	0.73	0.56	0.53
Entorhinal (I–II)	2 (18.1)	1 (12.5)	0 (0.0)	14 (31.8)	1 (6.2)	1 (7.6)
Limbic (III–IV)	7 (63.6)	5 (62.5)	3 (42.8)	25 (56.8)	9 (56.2)	2 (15.3)
Neocortical (V–IV)	1 (9.0)	2 (25.0)	4 (57.1)	4 (9.0)	6 (37.5)	10 (76.9)
Lewy body
Absent	11 (100.0)	7 (87.5)	6 (85.7)	37 (84.0)	13 (81.2)	12 (92.3)	0.32	1.0	1.0
Present	0 (0.0)	1 (12.5)	1 (14.3)	7 (15.9)	3 (18.7)	1 (7.6)
Hippocampal sclerosis
Absent	11 (100.0)	6 (75.0)	6 (85.7)	43 (97.7)	15 (93.7)	12 (92.3)	1.0	0.52	1.0
Present	0 (0.0)	2 (25.0)	1 (14.3)	1 (2.2)	1 (6.2)	1 (7.6)
Cerebral amyloid angiopathy
None	8 (72.7)	3 (37.5)	5 (71.4)	19 (43.1)	2 (12.5)	0 (0.0)	0.66	0.44	0.32
Sparse	0 (0.0)	2 (25.0)	1 (14.3)	18 (40.9)	6 (37.5)	7 (53.8)
Moderate	1 (9.0)	0 (0.0)	0 (0.0)	7 (15.9)	5 (31.2)	5 (38.4)
Frequent	2 (19.1)	3 (37.5)	1 (14.3)	0 (0.0)	3 (18.7)	1 (7.6)
Purkinje cells (mm)†	4.55 ± 0.64, 4.52	4.26 ± 0.41, 4.31	4.41 ± 0.89, 4.07
Torpedoes (mm)†	0.09 ± 0.08, 0.08	0.05 ± 0.02, 0.05	0.07 ± 0.04, 0.08

^*Chi-squared test (χ²) or Fisher exact probability test (2-tailed) comparing ET to non-ET within each cognition category.

^†Mean ± standard deviation, median.

ET, essential tremor; MCI, mild cognitive impairment; NFT, neurofibrillary tangle.

Within each cognitive stratum (normal, MCI, dementia), no differences between ET and controls were observed for β-amyloid plaques, cerebral amyloid angiopathy, Braak NFT stage, presence of Lewy bodies, or hippocampal sclerosis (Table 2). Among cognitively normal subjects, ET cases had a higher number of NFT-positive neurons in the neocortex when compared to controls (p < 0.001) but the 2 groups had similar numbers of NFT-positive neurons in the medial temporal lobe (p = 0.22) (Fig. 2; Table 3). Among subjects with MCI, the ET cases similarly had a higher number of NFT-positive neurons in the neocortex when compared to controls (p < 0.001) but not in the medial temporal lobe (p = 0.55). Among subjects with dementia, ET cases and controls had similar NFT-positive neurons in both regions (p = 0.82 and 0.64). In a sensitivity analysis, we excluded the 2 ET cases with Lewy bodies, and this did not change our primary findings (Supplementary DataTable S1; see neocortical NFT-positive neurons in cases vs control, both for normal and MCI, where p = 0.001).

FIGURE 2.

Regional differences in NFT counts in ET versus non-ET subjects. Box plot showing the median and distribution of standardized tangle counts on the specified region broken up by cognitive status. *Mann-Whitney test. MTL, medial temporal lobe; NC, neocortex; MCI, mild cognitive impairment. *p < 0.001.

TABLE 3.

Standardized NFT Counts in Cognitively Assessed ET Patients and Non-ET

Cognitive Status	Region	ET, Median (Mean)	Non-ET, Median (Mean)	p*
Normal	MTL	0.26 (0.52)	0.28 (0.37)	0.22
	NC	0.59 (0.68)	0.00 (0.06)	<0.001
MCI	MTL	0.82 (0.94)	1.08 (0.96)	0.55
	NC	1.01 (0.99)	0.09 (0.18)	<0.001
Demented	MTL	0.96 (1.96)	1.78 (1.82)	0.64
	NC	0.68 (0.90)	0.66 (1.56)	0.82

^*Wilcoxon rank sum test with continuity correction.

ET, essential tremor; NFT, neurofibrillary tangle; MTL, medial temporal lobe; NC, neocortex.

Bold values indicate statistically significant differences.

We examined whether Purkinje cell or torpedo counts were correlated with MMSE and they were not (both Spearman’s correlation coefficient p values > 0.05). Neither Purkinje cell nor torpedo counts were correlated with neocortical tau or medial temporal lobe tau (all Spearman’s correlation coefficient p values >0.05).

In both groups, global cognition correlated inversely with number of NFT-positive neurons in both the neocortex and medial temporal lobe (ET NC r = −0.038, p = 0.07, ET MTL r = −0.039, p = 0.05, non-ET NC r = −0.019, p = 0.09, non-ET MTL r = −0.030, p = 0.007, Table 4). With regard to specific cognitive domains, among controls, NFT counts in both the neocortex and the medial temporal lobe were inversely associated with measures in each of the 3 domains. Within the ET cases, only medial temporal lobe tau counts were associated with cognitive functioning, and only in the domains of attention/processing speed and episodic memory.

TABLE 4.

NFT-Positive Neurons in the NC and the MTL in ET and NON-ET Cases

	ET Spearman’s r (p*)		Non-ET Spearman’s (p*)
	NFT-NC	NFT-MTL	NFT-NC	NFT-MTL
Global cognitive score
MMSE	−0.38 (0.07)	−0.39 (0.05)	−0.19 (0.09)	−0.30 (0.007)
Attention/processing speed
Digit span forward	0.14 (0.57)	−0.36 (0.12)	0.002 (0.99)	−0.11 (0.36)
Oral symbol digit modality	−0.17 (0.48)	−0.60 (0.007)	−0.65 (<0.001)	−0.33 (0.005)
Episodic memory
Logical memory immediate	−0.08 (0.73)	−.33 (0.16)	−0.53 (<0.001)	−0.47 (<0.001)
Logical memory delayed	−0.17 (0.49)	−0.48 (0.03)	−0.52 (<0.001)	−0.48 (<0.001)
Working memory
Digits backwards	0.14 (0.56)	−0.11 (.64)	−0.28 (0.02)	−0.19 (0.11)

Bolded values are significant.

^*Wilcoxon rank sum test with continuity correction.

MTL, medial temporal lobe; NC, neocortex.

DISCUSSION

An increasing number of studies have demonstrated that ET cases have poorer cognitive performance than age-matched controls (7, 9, 10, 34). There is also an emerging understanding that the cognitive problems in ET patients may be progressive (12), can become severe, and that the rate at which this occurs is above and beyond that expected in age-matched controls. Indeed, cognitive impairment progresses to dementia in ET more than in age-matched controls (14, 15). While cognitive impairment in ET may be mediated by the cerebellar changes seen in ET, an alternate hypothesis is that the more severe forms of cognitive impairment may reflect other cortical or subcortical pathological changes that are either concomitant with, or related to, the ET disease process.

Unfortunately, there are virtually no data on the neuropathological basis for cognitive changes in ET. One study used a semi-quantitative assessment of NFTs in ET cases, reporting a higher Braak NFT stage than in controls (17). However, the study had methodological limitations. The objective of this study was to use a more rigorous approach to determine whether or not ET is characterized by a higher tau burden than expected for age by assessing NFTs using a quantitative and precise measure in a sample of ET cases and controls with detailed cognitive characterization. Indeed, cognitively normal ET cases had more NFTs in the neocortex than cognitively normal controls; the same was found when comparing ET cases and controls with MCI. Among individuals with dementia, however, regional tau levels were similar across groups, suggesting that regardless of the presence of ET, dementia is often characterized by high levels of tau, as would be expected in the context of AD, for example.

The higher neocortical tau burden in pre-dementia ET indicates some baseline predisposition to accumulating tau pathology outside of the medial temporal lobe. This pattern of tau progression diverges from classical AD that preferentially affects the medial temporal lobe. This pattern may reflect an incipient and more aggressive tauopathy than that usually observed in aging, in which tauopathic changes tend to be restricted to the medial temporal lobe (35). Indeed, the pattern of tau across brain regions appeared to differ across ET and non-ET cases. Specifically, while both ET and non-ET cases had similar medial temporal lobe tau, neocortex counts were higher than medial temporal lobe in ET, but lower than medial temporal lobe counts in non-ET.

The pre-dementia vulnerability to tauopathic changes seen in ET may underlie, or at least contribute to, impairments in cognition observed in ET. Cognitive functioning was correlated with tau counts in ET cases as well as non-ET cases. Interestingly, however, the nature of the correlations may have differed across groups. First, in the ET group, cognition was associated only with medial temporal tau whereas controls showed associations with both medial temporal and neocortical regions. The lack of an association between neocortical tau and cognition in ET was somewhat surprising given the fact that the vulnerability to tau seen in pre-dementia ET was restricted to the neocortex. However, the number of ET cases was small, and additional work with a larger sample of ET cases is needed to more fully explore this issue. Second, NFT counts correlated with more aspects of cognition in controls than in the ET group, with no correlation between tau and working memory in ET. Again, additional work with a larger sample is needed; however, one could speculate that the root of working memory impairment in the ET group, a form of executive dysfunction, may be more highly attributable to a cortico-cerebellar syndrome than to tau (36).

These findings provide further evidence that patients with ET harbor a diffuse neural vulnerability that manifests, in part, by an elevation in pathological tau species. This spatial distribution is distinct from the neurofibrillary tau pathology in the medial temporal lobe that is observed nearly ubiquitously in aging (i.e. PART) (35). Further analysis is required to place ET on the spectrum of tauopathies. Previous studies have highlighted the potential importance of glial tau, which can be observed in ET in a pattern reminiscent of progressive supranuclear palsy (37), another movement disorder that affects not only the basal ganglia and brainstem, but also the neocortex resulting in variable degrees of cognitive impairment. Further focused studies addressing this relationship are required.

Given conflicting reports about the relationship between ET and Parkinson disease and Lewy bodies (38), we reported here our results with and without the ET cases with Lewy bodies and the results are essentially the same. It should be noted that Lewy body pathology in the ET group was very mild and restricted to the brainstem (brainstem predominant) in one case (Braak LB stage IV). The second case had significant numbers of Lewy bodies in the amygdala but not elsewhere despite extensive sampling. Such individuals that do not conform to the Braak Lewy body staging system have been widely reported and their clinical significance remains to be determined (39, 40).

In addition, we found no significant correlations between Purkinje cell counts and torpedo counts and MMSE or tau burden, which suggests that the ET-related cerebellar pathology does not seem to be driving the extent of neocortical and MTL tau deposition or the cognitive state. This result was not unexpected, as Purkinje cells are inhibitory neurons that when perturbed can result in decreased inhibitory modulation from the cerebellum, which primarily effects motor control, not cognitive function, although this research is ongoing (41).

This study has several limitations. The relatively small sample size likely limited our power to detect differences. As we continue to collect specimens from participants enrolled in the study, we expect new associations to emerge. Other neuropathologies, such as other manifestations of tau pathology (e.g. gliofibrillary pathology) and TDP-43 proteinopathy, will be addressed in future work. Importantly, the work presented here overcame several limitations of our previous analyses, as it included detailed premortem cognitive assessments and a more granular assessment of tau burden with regional NFT counts. Also, differences in staining quality and intensity across sites could have influenced our results. Despite this, the microscopic morphology of NFT staining in neurons is distinct and recognizable with sufficient experience, and hence, differences in staining intensity are not likely to have affected our counts, the primary analysis of this study. The fact that differences in NFT counts between ET and controls were regional rather than global (i.e. affecting only some brain regions), further supports our conclusion that our case-control differences were not secondary to differences in staining intensity.

In conclusion, we present the preliminary autopsy findings from a new cohort of ET patients and controls who underwent detailed cognitive characterization and whose brains were available for postmortem analyses including quantitative measurement of NFTs. Cognitively normal ET cases, as well as those with MCI, had more NFTs in the neocortex than their non-ET counterparts. These data indicate some predisposition to tau pathology among ET cases. Moreover, NFT counts correlated with cognition in both the ET and non-ET groups, suggesting that tau burden likely contributes to the cognitive impairment seen in ET. Finally, the ET-related differences in regional NFTs were not observed among individuals with dementia. Future work will continue to investigate the neuropathological basis of increased dementia risk in individuals with ET, including examination of other tau pathologies.

This work was supported by the National Institutes of Health [R01AG054008 and R01NS095252 to J.F.C., NINDS R01NS086736 to S.C. and E.L., and F32AG056098 to K.F], the Alzheimer’s Association [NIRG-15-363188 to J.F.C.], the Tau Consortium, and an Alexander Saint-Amand Scholarship [to J.F.C.].

The authors have no duality or conflicts of interest to declare.

Supplementary Data can be found at academic.oup.com/jnen.