Dentate Nucleus Neuronal Density: A Postmortem Study of Essential Tremor Versus Control Brains

Abstract

Background:

Essential tremor involves the cerebellum, yet quantitative analysis of dentate nucleus neurons has not been conducted.

Objectives:

To quantitatively compare neuronal density or neuronal number in the dentate nucleus of essential tremor versus age-matched controls.

Methods:

Using a 7-μm thick Luxol fast blue hematoxylin and eosin-stained paraffin section, dentate nucleus neuronal density (neurons/mm²) was determined in 25 essential tremor cases and 25 controls. We also applied a stereological approach in a subset of four essential tremor cases and four controls to estimate total dentate nucleus neuronal number.

Results:

Dentate nucleus neuronal density did not differ between essential tremor cases and controls (P = 0.44). Total dentate nucleus neuronal number correlated with neuronal density (P = 0.007) and did not differ between essential tremor cases and controls (P = 0.95).

Conclusions:

Neuronal loss, observed in the Purkinje cell population in essential tremor, did not seem to similarly involve the dentate nucleus in essential tremor.

Introduction

Essential tremor (ET) is a chronic, progressive, and highly prevalent neurological disease^1,2 whose pathophysiology is not completely understood. Numerous studies indicate that the cerebellum is an important contributor to the pathogenesis of ET^3–6; however, the exact mechanisms of dysfunction are still poorly defined. The majority of ET experimentation has focused on cerebellar cortex, where unique pathological changes, such as Purkinje cell (PC) loss^7,8 and increased PC axonal changes,^7,9 have been observed in ET brains relative to age-matched control brains. Other postmortem features associated with ET include reduced PC dendritic arborization and spine density,¹⁰ increased number of heterotopic PCs,¹¹ basket cell axonal alterations around the PC body and initial axon segment,^12–14 abnormal arrangement of climbing fiber–PC synapses along the PC dendritic arbor,^15,16 changes in dentate γ-aminobutyric acid (GABA) receptors¹⁷ and accumulation of cerebellar β-amyloid.¹⁸

In contrast, less post-mortem research has been conducted on the dentate nucleus (DN) in ET. The DN is the largest of four cerebellar nuclei and is functionally responsible for crucial neuronal outputs from the cerebellar neocortex. The DN contains anatomically separate and functionally different motor and non-motor regions, which are dorsally and ventrally located, respectively.¹⁹ These distinct processing regions undergo task-dependent activation and are associated with separate thalamocortical projections to motor and prefrontal regions.²⁰ The DN receives dense inhibitory input from PCs and excitatory input from collateral branches of mossy fiber and climbing fiber collaterals.²¹ In relation to ET, a recent study found defective GABA receptors in the DN of ET cases in comparison to Parkinson’s disease (PD) cases and controls.¹⁷ A case report documented extensive degenerative changes (ie, neuronal loss and reduced number of efferent fibers) in the DN of one ET patient²² and similar DN changes were noted in a second ET patient.⁷

In this study, we evaluated whether quantitative neuronal loss occurs in the cerebellar DN in ET. Using a 7-μm paraffin section from the neocerebellum, we first determined the DN neuronal density in 25 ET cases and 25 age-matched controls. Then, we validated this non-stereological method by using an optical fractionator stereological approach in a subset of ET cases and controls to obtain an unbiased estimate of the total number of neurons in the DN.

Methods

We selected 25 ET and 25 non-diseased control brains from the Essential Tremor Centralized Brain Repository (ETCBR). We broadly sampled ET cases to represent cases harvested during each year (from inception of the ETCBR-2015) while also taking into consideration the need to age-match cases with available controls. All brains underwent a complete neuropathological assessment by a senior neuropathologist.⁷

A single 7-μm thick paraffin section from a standard cerebellar block from 25 ET cases and 25 controls was stained with LH&E and used for neuronal density quantification. For a subset of four ET and four control brains, we applied an optical fractionator stereological approach to obtain an unbiased estimate of the total neuron number in the DN.

Parametric tests were used to analyze the DN neuronal density and DN neuronal counts. Spearman’s correlation coefficient (r_Spearman) was used to assess the association between total DN neuronal density and total DN neuronal number. We also used multivariate linear regression analysis to assess potential confounding variables.

All procedures are described in detail (Supplementary Appendix S1).

Results

ET cases and controls had similar age at death, gender, brain weight, Braak Alzheimer’s disease (AD) stage, and Consortium to establish a Registry for AD (CERAD) plaque score, but ET cases had a longer post mortem interval (PMI). Consistent with our previous studies, ET cases had lower PC counts and higher torpedo counts than controls (Table 1). None had PD or progressive supranuclear palsy.

TABLE 1.

Clinical and pathological data on controls and ET cases

	Controls	ET cases	P value
n	25	25
Age at death (yr)	81.7 ± 9.2	84.2 ± 6.2	0.27a
Gender			0.56b
Men	14 (56.0)	17 (68.0)
Women	11 (44.0)	8 (32.0)
ET medications		β-blocker 21 (84.0)
		Primidone: 17 (68.0)
		Other ET medication: 12 (48.0)
Ethanol (no. drinks/day)		0.8 ± 0.4
		Median = 0.9
Total tremor score (range = 0–36)		26.4 ± 5.3
		Median = 25.0
		Minimum = 18
		Maximum = 36
Brain weight (g)	1203.4 ± 143.8	1220.3 ± 136.2	0.67a
Postmortem interval (h)	16.2 ± 10.9	26.4 ± 8.1	<0.001c
	Median = 13.9	Median =27.5
Braak AD staged			0.22e
0	4 (16.0)	0 (0.0)
I-II	15 (60.0)	12 (50.0)
III-IV	5 (20.0)	10 (41.7)
V-VI	1 (4.0)	2 (8.3)
CERAD plaque scored			0.32e
0	11 (45.8)	11 (45.8)
A	10 (41.7)	7 (29.2)
B	3 (12.5)	3 (12.5)
C	0 (0.0)	3 (12.5)
Purkinje cell countf,g	11.6 ± 1.8	8.6 ± 1.4	<0.001a
Torpedoesh	6.2 ± 7.6	14.3 ± 12.3	0.003c
	Median = 4.0	Median = 9
Total DN neuronal density (neurons/mm2)	27.1 ± 6.20	28.8 ± 8.47	0.44a
Ventral DN neuronal density (neurons/mm2)	27.1 ± 7.01	29.3 ± 8.75	0.35a
Dorsal DN neuronal density (neurons/mm2)	27.3 ± 5.78	28.3 ± 8.63	0.62a
Total DN neuron number (×106)	2.11 ± 0.35	2.09 ± 0.31	0.95a
	Median = 2.14	Median = 1.99

Values represent mean ± SD or number (%).

For variables that were not normally distributed, the median is reported as well.

^aStudent t test.

^bFisher’s exact test.

^cMann-Whitney test.

^dData available on <50 brains.

^eχ² test.

^fMean number of Purkinje cells per 100× microscopic field, among 15 sampled fields.

^gFor the eight controls versus five ET cases not reported in Louis et al.⁷ or Choe et al.,⁸ mean ± SD Purkinje cell counts were 12.7 ± 1.8 versus 9.2 ± 0.6 (given the small number, these were not subjected to statistical testing).

^hFor the eight controls versus five ET cases not reported in Louis et al.⁷ or Choe et al.,⁸ median torpedo counts were 8.5 versus 20 (given the small number, these were not subjected to statistical testing).

Abbreviations: AD, Alzheimer’s disease; CERAD, the Consortium to establish a Registry for Alzheimer’s disease; DN, dentate nucleus; SD, standard deviation.

To determine DN neuronal density, we divided the DN in a 7-μm LH&E-stained section into outlined dorsal and ventral contours, and counted neurons with a visible nucleolus (Supplementary Figure S1). Total DN neuronal density (dorsal + ventral) was similar in ET cases and controls (Fig. 1A, Table 1). After dividing the DN into dorsal and ventral areas, we did not find a difference between ET cases and controls. Additionally, DN neuronal density was highly similar in dorsal and ventral regions (Table 1).

FIG. 1.

Dentate nucleus neuronal density and total neuronal number. Scatterplots of the dentate neuronal density determined in paraffin sections (A) and total dentate neuronal number determined by unbiased stereology analysis (B). The neuronal density and total neuronal number do not differ between ET cases and controls. The central bar in each graph indicates the median of the data set.

Total DN neuronal density of the entire sample was not associated with gender (females, 29.3 ± 7.3 neurons/mm²; males, 26.3 ± 7.3 neurons/mm², P = 0.15), PMI (r_Spearman = −0.05, P = 0.74), brain weight (r_Pearson = 0.20, P = 0.16), Braak AD stage (r_Spearman = 0.08, P = 0.58), or CERAD plaque score (r_Spearman = 0.00, P = 0.99). Older age was associated with higher DN neuronal density (r_Pearson = 0.31, P = 0.03). DN neuronal density did not correlate with total tremor scores in ET cases (Table 1; r_Pearson = 0.26, P = 0.22). To assess potential effects of confounding variables, we performed a multivariate linear regression analysis. In this analysis, the dependent variable was total DN neuronal density, the independent variable was diagnosis, and additional variables were age and PMI. In this adjusted analysis, there was no association between diagnosis and total DN neuronal density (β = −2.31, P = 0.38). Therefore, we report a null finding even when potential confounding effects are taken into consideration.

For the stereological component of the study, the four ET cases and four controls (both with mean age = 85 years) did not differ with respect to the mean or median total neuron number in the DN (Fig. 1B and Supplementary Table S1). Individual estimates for all eight subjects had a coefficient of error in the range of 0.03–0.07, demonstrating an adequate precision for our optical fractionator sampling scheme (Supplementary Table S1).

We examined the correlation between the stereological and non-stereological aspects of our study. There was a strong correlation between total DN neuronal number (stereology) and total DN neuronal density (measured in single sections) (r_Spearman = 0. 88, P = 0.007).

Discussion

To further investigate the extent of structural abnormalities in ET brains, we examined the DN, a crucial area of the cerebellar system, in a quantitative and systematic way. This is the only study to do so. We found no difference in DN neuronal density in ET cases compared to age-matched controls, indicating that extensive cell loss in the DN does not seem to be a feature of ET pathogenesis. Likewise, in a supplemental stereology study, we found no ET case–control difference in DN neuronal number. The stereological data strongly correlated with the DN neuronal density data, which, in turn, validates our single section density analysis approach. Overall, our null finding aligns with the hypothesis that the cerebellar cortex may be the source of dysfunction in ET. There may also be subtle anatomic changes, other than DN cell death, in the cerebellar network that contribute to ET.

Previous post-mortem studies of cerebellar abnormalities in ET reported that only 2 of 33 (6.1%) ET patients exhibited extensive neurodegenerative changes in the DN, as visualized in LH&E stained sections.^7,22 However, clear evidence of neuronal loss in the cerebellar nuclei has not been quantitatively established in a larger sample of ET cases, or in comparison to controls. In one study, defective GABA receptors were reported in the DN of ET cases in comparison to PD cases and controls.¹⁷ Positron emission tomography (PET) scans of tremor-related networks in ET cases and controls revealed decreased GABA_A receptor binding patterns in the ET DN.²³ These observations introduce the possibility of localized GABAergic dysfunction in the DN or receptor upregulation because of abnormal function in the cerebellum. We also previously quantified protein levels of the excitatory amino acid transporter 2 (EAAT2), the major glial glutamate transporter, in ET cerebellum versus controls, and demonstrated enhanced EAAT2 expression in the ET DN.²⁴ These findings raise the possibility that lower levels of inhibition in DN, because of PC loss in ET, could trigger a compensatory increase in EAAT2 to reduce the effects of glutamatergic synaptic inputs to the DN and maintain excitation-inhibition balance in cerebellar nuclei. Despite the above-mentioned alterations in the DN, the changes do not seem to result in significant neuronal loss in the ET DN.

There are only two prior stereological studies of neuronal number in human DN^25,26 and none in ET. Using the optical fractionator approach, we obtained an unbiased estimate of neuron number without the need to investigate the exact volume or area of the DN, which is important given concerns of volume reduction following formalin fixation and tissue preparation. Our stereological approach also allowed us to validate the null findings of our DN neuronal density data set. One issue is that stereology studies required representation of the entire DN in a hemi-cerebellum, which is not a feature of most brain banking protocols. Hence, our stereological analyses were able to draw on only a limited number of brains. Regardless, the sample size we were able to obtain for these analyses (4 brains/group) is common in other human stereology studies (ie, ≤5/group),^25,27–30 probably for similar reasons. Neuronal density and neuronal number can provide information about the physical population of cells in the DN, but these metrics in postmortem tissue do not necessarily contribute to functional knowledge about DN activity. Nonetheless, our study is the first to address the question of whether cell loss in the DN is a feature of ET pathogenesis. We separately examined both dorsal and ventral regions of the DN, representing distinct compartments having motor and non-motor functions, respectively. We had a well-powered study, with 25 in each group, to determine whether DN density differed by disease, and we evaluated a large number of potential confounding factors to rule out the presence of other causal relationships. Additionally, our samples were age-matched to minimize case–control differences.

Several limitations should be considered. First, our study was restricted to the DN and similar data for other cerebellar nuclei would be of value. Second, data on the structural integrity of this nucleus aids in but not does fully encompass all that is needed to decode abnormalities in neuronal oscillatory networks in ET. Third, more extensive use of stereological analysis would have allowed for more definitive conclusions.

In sum, cerebellar changes are a feature of ET, but the current study found no evidence for significant neuronal loss in the DN of ET brains. We investigated a possible pathological aspect of ET by quantifying the DN neuronal density and DN neuronal number in a large, age-matched group of ET cases and controls. Our results contribute to a greater understanding of the DN in the context of ET and the disease’s associated structural abnormalities.