Voxel-based morphometry (VBM) allows for a voxel-based automated and rater-independent analysis of brain volumes between groups, either in specified regions-of-interest (ROI) or at a whole-brain level without a priori hypotheses (117). Alternatively, automated segmentation methods can be applied to quantitatively measure brain volumes, e.g., the cortical thickness, using freely available software, such as FreeSurfer (118).

Cerebellar atrophy is commonly reported in ET patients (15, 22, 26, 28–32, 36, 37, 43, 46, 48). However, an equivocal number of studies found no morphological cerebellar changes (18, 21, 25, 33–35, 38–42, 49), and even increased cerebellar gray matter volume in young ET subjects has been reported (27, 45). Findings of cerebral cortical and subcortical structural changes are even more heterogeneous. There is no consistent pattern of atrophy. Moreover, alongside volume loss, gray matter volume gain has been observed in various cortical regions, and some studies did not identify any cortical differences between ET patients and HC (15, 22, 23, 26, 33–36, 38, 41, 44, 45, 47). Of note, the clinical phenotype of ET is associated with distinct morphological brain changes. For example, ET patients presenting with additional head tremor display more pronounced or distinct patterns of cerebellar atrophy, as well as various cortical structural changes compared with classic ET (25, 26, 28, 30, 31, 46). Indeed, some of the studies reporting cerebellar atrophy found significant volume loss only in ET individuals exhibiting additional head or voice tremor (28, 30, 31, 46). Moreover, cognitive dysfunction in ET has been linked to specific cortical atrophy patterns not apparent in cognitively intact ET individuals (24). The heterogeneity of structural brain alterations reported in ET has been highlighted in a recent meta-analysis including 16 VBM studies and more than 350 ET individuals (119). The latter study did not identify any brain regions, including the cerebellum, that exhibited consistent gray matter volume loss in ET patients compared with HC (119).

DTI is utilized for the assessment of the brain's microstructural integrity and is particularly sensitive to alterations in cerebral white matter. DTI measures the random movement of water molecules, which is mainly directed along white matter fiber tracts (120). Two important measures are the mean diffusivity (MD) and the fractional anisotropy (FA). The MD depicts the average movement of water molecules in organic tissue, whereas the FA refers to the directionality of movement. FA values close to 1 reflect anisotropy, whereas values nearing 0 are isotropic and are suggestive of tissue damage. Conversely, high MD values are a surrogate for a loss of cellular integrity and indicative of neuronal damage (121).

Compared with conventional MRI, DTI studies more consistently point toward microstructural alterations of the cerebellum, particularly of the cerebellar peduncles and dentate nuclei (38, 49, 54, 55, 57, 59–61). In contrast to these 11 studies, only two studies, both employing an ROI-based approach, did not find any differences in DTI between ET patients and HC (19, 21). Beyond cerebellar changes, widespread microstructural alterations have been reported in various cerebral white matter tracts related to both motor and non-motor function and in the red nuclei (37, 38, 52, 53, 56, 59). For example, in a recent study by Revuelta et al., the authors reported decreased MD of fiber tracts connecting the ventral intermediate nucleus of the thalamus (Vim), the typical target for DBS in ET, with motor and supplementary motor cortical regions (58). Even though no alterations of FA were observed in the same tracts, both MD and FA in the Vim supplementary motor area tract correlated with tremor severity, suggesting a pathological reinforcement of this tract in ET (58). Similar to VBM studies, phenotype-specific changes of FA and MD have been reported. Specifically, ET patients presenting with additional resting tremor, but unremarkable dopamine transporter imaging, showed reduced structural connectivity in a network comprised of the globus pallidus, caudate nucleus, and supplemental motor area that was not apparent in ET patients without resting tremor (51). Moreover, distinct cortical microstructural changes, including the hippocampi, have been linked to cognitive dysfunction (41, 49, 50).

fMRI measures the blood oxygen level-dependent (BOLD) contrast—generally called “the BOLD signal” (123). The BOLD signal is affected by hemodynamic, vascular, and metabolic factors, but is generally assumed to be closely related to neural activity (123, 124). The first task-based fMRI study in ET patients identified increased activity in the contralateral sensory and motor cortices, thalamus, and globus pallidus and bilateral overactivation of the cerebellar hemispheres and dentate nuclei during arm posturing. In contrast, the authors observed increased activity in the ION in only two out of 12 patients, supporting a pivotal role of the cerebellum in tremor genesis and refuting the single oscillator ION hypothesis (16). Subsequently, numerous task-based fMRI studies have confirmed that altered cerebellar and cerebello-thalamico-cortical activity is correlated to clinical tremor manifestation and task performance (36, 40, 62–64, 66, 67). One study did not report a significant difference in functional connectivity during an emotion processing and finger tapping task in ET patients compared with HC. However, in the latter study, HC were not age matched to the ET cohort (32). In line with the findings from structural imaging, cognitive function has been associated with specific activity changes outside the classical tremor network (65, 68).

More recently, neuronal activity has been assessed in the resting state. Resting-state fMRI is advantageous over task-based paradigms in that it is independent of individual variability in task performance and interference of tremor with motor function. The most consistent finding reported by these studies was altered intrinsic cerebellar and cerebello-thalamo-cortical activation/connectivity, particularly of cerebello-motor cortical projections (33–35, 37, 61, 71–75, 77). There is also evidence that complex functional alterations outside the classical tremor axis are present in ET, including visual networks (69, 70). In this vein, Archer et al. have demonstrated that activity within the tremor and visual networks during a grip motor task could be modulated by visual feedback (21).

Of note, surgical interventions to treat ET, such as Vim-DBS or thalamotomy have been shown to restore connectivity in the tremor network partially and to cause widespread remodeling of other brain networks outside the classical tremor axis [e.g., (125, 126)]. In line with observations from structural MRI, the clinical phenotype appears to be associated with distinct functional brain changes. For example, ET individuals exhibiting head tremor showed distinct cerebellar activity compared with those who did not (76), and ET patients with resting tremor showed different activation patterns of various cortical and subcortical brain regions compared with classical ET (73).

MRS is utilized to assess neurometabolic alterations in brain tissue in vivo. N-acetylaspartate (NAA) is an abundant amino acid derivative synthesized in neurons. A reduction of NAA, therefore, is indicative of neuronal damage. The choline (Cho) fraction comprises several soluble components mainly located in myelin and cell membranes. Conditions resulting in increased turnover or damage to cellular membranes and myelin, such as inflammation, tumor, or neurodegenerative processes, result in increased Cho concentrations. Creatine (Cr) is found in most neurons and astrocytes. The Cr peak is very robust, which is why Cr is commonly used as a denominator to offset changes of NAA and Cho (127). MRS can also be applied to measure GABA (please see below).

Few studies have exploited MRS to investigate neurometabolite changes in ET. Louis et al. were the first to report a reduced NAA/Cr ratio in the cerebellum of 16 ET patients compared with 11 HC that was inversely correlated to tremor severity (81). The same group later found that NAA/Cr changes were similar between cerebellar hemispheres, in accordance with the symmetric clinical manifestation of ET (82). Similarly, reduced NAA/Cr ratios have been reported by Pagan et al. in a small cohort of 10 ET patients (83). That said, others found normal NAA/Cr ratios in the thalami of ET patients (78–80), whereas there was an increase in the excitatory neurotransmitter glutamate/glutamine evident in one of these studies (79).

The role of GABA in ET has been a topic of ongoing discussion for many years (13), and different lines of research have vindicated the significance of the GABAergic system in ET pathophysiology.

ET patients show lower GABA levels in the cerebrospinal fluid (128), and GABA receptor density in the cerebellar dentate nucleus has been reported to be reduced (129). Alcohol has agonistic GABAergic properties and alleviates tremor in many patients with ET (130), and the majority of drugs used to treat ET act via GABAergic pathways (131). Moreover, GABA-A₁ receptor knockout mice develop an ET-like disease that responds to GABAergic drugs commonly used to treat ET (132). The impact of reduced cerebellar GABAergic tone on neuronal activity in cerebello-thalamico-cortical tremor network activity has also been highlighted in a recent study applying a complex computational simulation model of ET (133).

Very few in vivo imaging studies have explored the role of GABA in ET. Using PET and ¹¹C-flumazenil, a ligand of the benzodiazepine site of the GABA receptor, Boecker et al. observed increased cerebellar, thalamic, and premotor cortical tracer uptake in a small cohort of ET patients compared with HC, hinting at reduced GABAergic function (90). In contrast, a separate study employing MRS failed to demonstrate a significant difference in GABA concentration in the dentate nuclei between ET individuals and HC (91). Another MRS study reported a positive correlation of the cerebellar GABA/glutamate + glutamine ratio with clinical tremor scores in a small ET cohort. However, neither GABA nor glutamate/glutamine levels differed between ET patients and HC in the latter study (92). Given the limited number of available imaging studies focused on GABA, we would like to mention the work by Gironell et al. even though their study did not meet our inclusion criteria; they found a significant correlation of cerebellar ¹¹C-flumazenil uptake and tremor severity in a cohort of 10 ET patients (134).

Different MRI techniques, such as susceptibility-weighted imaging, T2^*-weighted, or its inverse R2^*-weighted gradient echo imaging, can be used to measure brain iron (135). More recently, novel methods, such as neuromelanin and nigrosome-1 imaging have been developed to visualize the neuronal integrity of the substantia nigra (136). These techniques have been extensively used to detect iron depositions, which are assumed to be a surrogate of cellular damage in neurodegenerative disorders, such as PD (137).

In one study comparing 25 ET patients with 25 matched HC, no significant difference was found in the R2^* relaxation times of the substantia nigra (85). Similarly, three other studies reported normal nigral neuromelanin concentrations in ET patients (86, 88, 89), and nigral nigrosome-1 integrity has been found to be comparable with that of HC (84, 86). The focus of all these studies was on the substantia nigra, and only one study applied a whole-brain voxel-based approach (87). The authors reported increased iron levels in the bilateral pallidum, substantia nigra, and the right dentate nucleus. That being said, only the pallidal iron increase survived correction for multiple comparisons and was correlated to tremor severity (87).

Epidemiological studies suggest that ET populations have an increased risk of developing PD, and there is an ongoing controversy about a potential pathophysiological overlap between the two diseases (14).

FP-CIT SPECT (commercially known as DaTScan) is commonly used to assess the presynaptic striatal dopaminergic integrity (138). Striatal tracer uptake is significantly reduced in typical and atypical parkinsonism (138).

Most studies using FP-CIT SPECT or comparable techniques found no alterations of the dopaminergic system in ET (51, 78, 93–96, 100, 102–105, 107, 139). These findings were extended by two longitudinal studies showing constant tracer uptake over time (98, 100). Of note, a third longitudinal study not meeting our inclusion criteria did not reveal a decline of striatal dopamine transporter availability over a mean follow-up period of 28 months (140). One study reported normal DaTScan in classical ET patients, whereas tracer uptake was reduced in ET patients with additional resting tremor (101). However, resting tremor ET individuals were about 7 years older than the corresponding HC, and several subjects presented with subtle parkinsonian features, raising the question of whether they may have subsequently developed PD. Conversely, others have found normal striatal dopaminergic integrity in ET patients with resting tremor (102). That being said, some authors reported signs of slight striatal dopaminergic degradation in classical ET (97–99, 106). Of note, ET patients may show reductions of tracer uptake primarily in the caudate nucleus, contrasting the typical pattern of pronounced posterior putamenal reductions observed in PD (99, 106).

A series of small exploratory PET studies conducted in the 1990s, mostly using ¹⁵O-labeled H₂O and PET, revealed increased regional cerebellar blood flow (rCBF) during both rest and posture in ET patients compared with HC (17, 109, 110). These studies showed overactivation of additional regions comprising the tremor network, whereas olivary overactivation was not present in any of these studies (17, 109, 110). Boecker et al. demonstrated that abnormally increased cerebellar rCBF decreased toward normal values after ingestion of ethanol, and this decrease was correlated to concurrent tremor alleviation (108). Furthermore, there was an increase in ION activation following ethanol ingestion, suggesting increased afferent olivary input as a consequence of normalizing synaptic cerebellar activity (108). More recently, SPECT and ^99mTc-hexamethylpropylenaminoxom (HMPAO) have been used to measure rCBF in ET cohorts. Sahin et al. did not observe any difference of rCBF between 16 ET patients and matched HC, but reported an inverse correlation of frontal cortical rCBF with tremor severity (111). Employing the same method, Song et al. found no significant differences in rCBF between ET patients with and without head tremor (113). Interestingly, rCBF was reduced in various brain regions, including the cerebellum, in the overall ET cohort compared with HC in the latter and in a subsequent study conducted by the same group (112, 113).

¹⁸F-fluorodeoxyglucose (FDG) and PET can be used to quantitatively assess brain glucose consumption, which is largely equivalent to neuronal activity (141). FDG PET has been extensively used to characterize metabolic brain abnormalities in neurodegenerative disorders, such as PD, and has revealed disease-specific abnormal brain networks that correlate with disease severity and can discriminate PD from atypical parkinsonism (142).

Hallett and Dubisnky were among the first to report increased brainstem and thalamic activity in ET patients, whereas they did not observe significant changes in cerebellar metabolism (114). Similarly, two recently published studies did not find changes in cerebellar metabolism, but widespread cortical hypometabolism was reported in one of these studies (105, 115). In contrast, Song et al. found cerebellar hypometabolism accompanied by reduced tracer uptake in various cortical regions (116). In yet another study using an ROI-based approach, no metabolic differences were identified in ET patients compared with controls in the basal ganglia (95).