Neuroimaging Advances in Parkinson's Disease and Atypical Parkinsonian Syndromes

Structural Magnetic Resonance Imaging in PD

Structural changes on conventional magnetic resonance imaging (MRI) are minimal and less apparent, especially in early stages of PD (10). A six-stage model depicting the sequential progression of α-synuclein pathology in PD has been proposed by Braak et al. (7) (Figure 1). As per this “gut to brain” transmission model, neuroimaging studies may predominantly observe brainstem and subcortical involvement in early PD stages with greater cortical involvement in late stage PD as well as in PD-MCI, PDD and DLB cases. Compared to normal controls, voxel-based morphometry (VBM) studies in PD may identify atrophy in the basal ganglia (22) (as consistent with nigrostriatal degeneration and consequential dysfunction in the basal ganglia thalamocortical circuit), frontal lobe (23), and non-specifically in the right hippocampus, left anterior cingulate and superior temporal gyri (24). Although inconsistently shown, cortical thinning in the orbitofrontal, ventrolateral prefrontal, and occipitoparietal cortical regions has been identified in PD, along with volumetric reductions in the caudate and putamen (25, 26). Some studies do not observe appreciable differences on conventional MRI compared to controls (27, 28). In PD patients with olfactory disturbances, reduced volume in the olfactory bulb and tract was evident vs. MSA and controls (29). In PD patients with respiratory dysfunction, gray matter (GM) atrophy was reported in the left parahippocampal formation, right fusiform gyrus, right cerebellum crus, and left postcentral gyri compared to PD with normal pulmonary functions (30). Freezing of gait symptomatology in PD was associated with posterior GM atrophy (specifically, left cuneus, precuneus, lingual gyrus, and posterior cingulate cortex) (27). In advanced PD, atrophy in the subcortical GM structures was found to be more pronounced vs. those in early PD stages (31) (Figure 2).

Changes within the SN may emerge as promising early diagnostic biomarkers of PD (10). Within the SNpc, calbindin-negative pockets termed “nigrosomes” are observed (34). The greatest loss in neuromelanin containing neurons takes place in the nigrosome-1, which is located in the caudal and mediolateral portion of SNpc (35). Nigrosome-1 shows a significant loss of hyperintensity on T2^* and neuromelanin-sensitive MRI in PD, probably caused by decreased neuromelanin, increased iron content, or loss of paramagnetic neuromelanin–iron complexes (36–38). On susceptibility-weighted imaging (modality of choice), healthy nigrosome-1 and the surrounding neuroanatomy of the dorsolateral SN may appear as the tail of a swallow bird (“swallow-tail” sign). Loss of this feature in PD vs. controls may assist in the differential diagnosis (sensitivity 80%, specificity 89%) (39, 40) (Figure 3). Finally, neuromelanin-sensitive MRI has shown promise in the differentiation of PD from essential tremor and normal controls (37, 41), and this technique may prove invaluable as a marker of disease progression in PD.

In PD patients with polysomnography-confirmed rapid eye movement sleep behavior disorder (RBD), cortical thinning was reported in the right perisylvian and inferior temporal cortices together with shape changes in the putamen compared to PD without RBD (42). Likewise, decreased volume in the right putamen correlating with RBD symptom severity was identified in PD with RBD vs. those without (43). On susceptibility-weighted imaging, loss of dorsal nigral hyperintensity (corresponding to nigrosome-1) was observed in ~25% of patients with idiopathic RBD, which associated with lower putaminal dopamine transporter (DAT) binding on single-photon emission computed tomography (SPECT) (44). This may suggest nigrosome-1 degeneration in some of the RBD cases, likely those at risk of progression to PD (44). These studies propose a link between PD, the presence of RBD, and greater neurodegeneration especially in subcortical structures.

Volumetric changes in the SN have been inconsistently reported in PD vs. controls (45–48). A high resolution examination of structural alterations in the SN is possible using ultra-high-field MRI (49). For example, susceptibility-weighted imaging at 7T permitted the visualization of the anatomical layers of the SN, allowing excellent discrimination between PD and controls (sensitivity 100%, specificity 96.2%) (50). Correlations between motor symptoms and increased R2^* (apparent transverse relaxation rate, R2^* = 1/T2^*) values in the SN have also been reported in PD (46, 51–53), which may reflect ferritin-induced magnetic field inhomogeneities. A recent study applied quantitative susceptibility mapping in PD to evaluate the magnetostatic alterations caused by changes in the iron distribution across the whole brain (54). The apparent magnetic susceptibility was found to be elevated in the dorsal and ventral SN, rostral pontine regions, and the cortex (primarily in the temporal paralimbic, prefrontal, and occipitoparietal regions) (54). Conversely, reduced magnetic susceptibility was detected in the normally iron-rich cerebellar region of the dentate nucleus suggesting decreased iron content (54). Interestingly, the striatum as well as the primary motor and somatosensory fields were spared (54). Future research will highlight whether iron accumulation is the consequence or cause of neurodegenerative changes associated with PD.

Diffusion-Weighted and Diffusion Tensor Imaging in PD

Reduced fractional anisotropy (FA) in the SN is commonly observed in PD (55, 56). Increased mean diffusivity (D) in the olfactory tracts and decreased FA in the anterior olfactory structures have been reported (57, 58), as consistent with neuronal loss observed in these regions (29). Significant differences in D or FA may not be evident in early PD due to milder neurodegeneration (59, 60). Combined analysis of diffusion parameters and apparent transverse relaxation rates was found to be superior in differentiating PD vs. MSA-P (sensitivity 97%, specificity 100%, positive predictive value [PPV] 100%, negative predictive value [NPV] 93%), and PD vs. MSA-P/PSP group (sensitivity 86%, specificity 87%, PPV 88%, NPV 84%) (61). In SNpc (delineated using a probabilistic atlas based on neuromelanin sensitive imaging), asymmetric alterations predominantly in the diffusion metrices rather than anisotropy were observed (62).

PD-MCI cases showed increased D as well as lower GM volume in the nucleus basalis of Meynert vs. cognitively normal PD (60). The degeneration of this cholinergic structure may identify patients at risk of more significant cognitive decline and dementia (60). Another study delineated the cortical projections of five corpus callosum segments and highlighted the role of callosal white matter (WM) abnormalities in cognitive dysfunction, which can occur via disruption of interhemispheric information transfer along callosal-cortical projections (63). In PD, increased axial diffusivity was identified in the three anterior callosal segments (projecting to prefrontal, premotor, motor, and supplementary motor cortices) vs. controls (63). Cognitive performance strongly related to diffusion tensor imaging (DTI) metrics in the most anterior (projecting to prefrontal cortex) and most posterior callosal sections (projecting to parietal, temporal, and occipital cortex), which may contribute to “fronto-striatal” and “posterior cortical” types of cognitive deficits seen in PD, respectively (63).

Free water was elevated in PD in posterior SN vs. controls (52, 64, 65). Increase in free water in posterior SN over time was observed in de novo PD patients, which was associated with motor severity and putaminal DAT SPECT binding (64). Given the posterior-to-anterior pattern of degenerative changes reported in SN, elevated free-water may also be observed in the anterior SN (66), especially in late-stage idiopathic PD (67). In contrast, free-water-corrected FA values were found to be unchanged in PD vs. controls, as assessed using the bi-tensor diffusion model (66, 68). Free-water and free-water-corrected FA values beyond SN and from multiple brain regions may help distinguish PD, MSA, and PSP cases from each other (66).

Proton Magnetic Resonance Spectroscopy in PD

Reduced N-acetyl aspartate/creatine (NAA/Cr) ratios in the SN have been observed in PD vs. controls, which were shown to correlate with disease severity (69, 70). Lowered NAA or NAA/Cr values have also been detected in other regions, including the lentiform nucleus (LN) (comprises of putamen and globus pallidus; basic anatomy presented in Figure 2), temporoparietal and posterior cingulate cortex, and pre-supplementary motor area vs. controls (71–74). However, the correlation between NAA/Cr ratios in these regions with disease severity or duration is inconsistently reported (72, 73). Another study compared the NAA/Cr ratios in the rostral and caudal SN, and found lower values in the rostral region in PD, whereas this pattern was inverted in the atypical PS group and controls (75). In tremor-dominant PD, reduced NAA/Cr and Choline/Cr values were detected in the thalamus vs. patients with essential tremor presenting with resting tremor (76). Cerebellar NAA/Cr and NAA/myo-inositol ratios were smaller in the atypical PS group compared to PD and controls (77). One study has also suggested the utility of proton magnetic resonance spectroscopy for the evaluation of treatment efficacy in PD (78). Specifically, the putaminal levels of myo-inositol, total Cr, and total NAA metabolites were reduced in the drug-off condition in PD vs. healthy controls. The administration of levodopa resulted in the restoration of total Cr and total NAA levels suggesting therapeutic responsiveness (78).

Structural Magnetic Resonance Imaging in LBSD

In DLB and PDD, conventional MRI typically shows variable changes. Compared to controls, VBM studies in PDD have identified a diffuse pattern of cortical atrophy involving the occipital, temporal, right frontal, and left parietal lobe (23), as well as atrophy involving the putamen, hippocampus, parahippocampal region, anterior cingulate gyrus, nucleus accumbens and the thalamic nuclei (24). Although inconsistently reported (23), a greater cortical loss in the temporal, occipital, and parietal lobes was noted in DLB vs. PDD (79). Indeed, pathological heterogeneity evident in these two closely related α-synucleinopathies is in part responsible for variable findings. Compared to PD, atrophy in the occipital lobe and entorhinal cortex may help differentiate PDD (23, 80).

The relative preservation of total hippocampus compared to AD is a supportive diagnostic feature of DLB (8). Within the hippocampus, the CA1 subfield shows preservation (81), which aligns with the histopathological evidence showing Lewy body aggregates and neuronal loss largely localized to the CA2/3 subfields in DLB cases (81, 82). Conversely, atrophy in the CA1 subfield is indicative of neurofibrillary tangle pathology as evident in AD. Hippocampal atrophy is in fact observed in LBSD, controls and AD in a characteristic pattern [controls < PD < PDD/DLB < AD (9, 83)] and the extent of atrophy aligns with the underlying concomitant AD-type pathology (9). Smaller caudate and putaminal volumes have also been reported in PD and DLB vs. AD and controls, albeit inconsistently (26, 84). WM hyperintensities are more prevalent in PDD and DLB vs. PD and controls (85).

PD-MCI patients may show greater cortical thinning in temporoparietal, occipital, and supplementary motor area vs. cognitively-normal PD (86). A recent meta-analysis evaluating VBM studies identified pronounced GM atrophy in the left anterior insula in PD-MCI vs. cognitively-normal PD cases (87). Longitudinal cognitive decline in PD was associated with an AD-like pattern of cerebral atrophy at baseline, underscoring the contribution of the hippocampus and temporoparietal cortex in the cognitive sequelae of PD (88).

The ε4-allele of the apolipoprotein E gene (APOE-ε4) is a shared risk factor for AD, PDD, and DLB disorders (89). Therefore, the identification of neuroimaging and cognitive endophenotypes of APOE-ε4 irrespective of the clinical diagnosis has been pursued (9). Recent investigations indicate that APOE-ε4 is related to hippocampal atrophy along with learning and memory performance in DLB as well as across the AD/DLB spectrum, implicating APOE-ε4-associated shared neurodegenerative mechanisms across these disorders (9). Similarly, MRI-derived WM hyperintensity burden was inversely related to learning/memory, attention/executive and language performances in APOE-ε4 carriers across the AD/DLB spectrum (90). In addition to the APOE-ε4's influence on amyloidopathy, these results are consistent with the emerging evidence indicating an independent role of APOE-ε4 in modulating α-synucleinopathy in the brain (91).

Diffusion-Weighted and Diffusion Tensor Imaging in LBSD

In DLB vs. controls, diffusion imaging studies reveal D and FA abnormalities in the corpus callosum, pericallosal regions, caudate nucleus (suggestive of nigrostriatal involvement), amygdala, inferior longitudinal fasciculus, precuneus, as well as in the frontal, parietal, and occipital WM with milder involvement of the temporal lobe early in the disease (92–94). In DLB, elevated D in the amygdala (a region preferentially affected by Lewy pathology) was observed in the absence of significant GM density changes, probably suggesting microvacuolation or spongiosis as one of the possible mechanisms (93). Increased D in the longitudinal fasciculus was identified in DLB patients with hallucinations vs. those without (93). In PDD, reduced FA in the bilateral posterior cingulate bundle was reported vs. PD (95). Compared to AD, decreased FA was noted in the pons and left thalamus in DLB (96). However, given the significant pathological overlap between AD and DLB, diffusion imaging studies may not reveal consistent findings, particularly in subjects at advanced disease stages (94).

Proton Magnetic Resonance Spectroscopy in LBSD

In DLB, lower NAA/Cr ratios in the bilateral hippocampus were evident relative to controls (97), albeit milder than typically seen in AD. Lower NAA/Cr ratios were observed in the posterior cingulate gyrus in PDD vs. cognitively normal PD patients (98).

Magnetic Resonance Imaging in MSA

Several visible features that may be identifiable on conventional MRI have been reported. On T2-weighted images, these include the presence of bilateral hyperintense rim lining the dorsolateral borders of the putamen (“putaminal rim” sign) and putaminal hypointensity in MSA-parkinsonian subtype (MSA-P) cases (11, 99). In addition, atrophy of the putamen, cerebellum, middle cerebellar peduncles (MCP), or pons may be noticed. In MSA-cerebellar subtype (MSA-C), visible features on T2-weighted and fluid-attenuated inversion recovery images may include cruciform pontine hyperintensity (“hot cross bun” sign; specificity 100%, sensitivity 58%; Figure 4a) and hyperintensity in the MCP (MCP sign; specificity 100%, sensitivity 50%; Figure 4b) (11, 99). Atrophy of the putamen, pons, cerebellum or MCP may be evident on T1-weighted images (Figure 4c). Notably, these signs have low sensitivity values and the appearance of these MRI markers can be influenced by image acquisition factors. For example, the “hot cross bun” sign was suggested to be more conspicuous on T2^*-weighted vs. T2-weighted images (100), probably due to the presence of increased iron (100, 101).

The volume of the putamen was found to be significantly reduced in MSA vs. PD cases (102). While putaminal atrophy has been shown to differentiate MSA/MSA-P from PD with high specificity (~92%), a rather low sensitivity (~44%) was noted (99, 103). When putaminal changes are present and are asymmetric, they reliably correlate with the asymmetry of clinical features in patients with MSA-P. Several studies suggest that the analysis of multiple biomarkers may improve differentiation. For example, combined analysis of putaminal hypointensity visualized on gradient echo sequence along with putaminal atrophy improved the overall diagnostic accuracy of MSA-P cases vs. PD and PSP (104). Likewise, analyzing DTI and R2^* relaxation rate together enabled the identification of abnormal patterns unique to PD, PSP, and MSA disorders (61). On the other hand, a multimodal study (incorporating R2^*, R1, and R2 mapping, magnetization transfer and DTI) found elevated putaminal R2^* values to be superior in the differentiation of MSA-P from PD, likely representing iron accumulation in the putamen (105).

In MSA vs. controls, volumetric and VBM studies commonly show striatonigral and olivopontocerebellar atrophy (11). Reduced volumes in the striatum and brainstem areas were observed in MSA vs. PD (28, 106), although with extensive overlap with PSP cases (28). In MSA-P, VBM studies also show atrophy in the primary motor and supplementary motor cortex indicating sensorimotor cortical degeneration (106, 107), as well as in the prefrontal and insular cortices compared to PD (106). Likewise, VBM analysis in MSA-P vs. PD revealed atrophy in the superior cerebellar peduncle (SCP), MCP, cerebellum, pons, midbrain, and putamen, but not in the globus pallidus (71). Cerebellar atrophy was evident in both MSA-P and MSA-C subtypes to varying degrees (28, 108). In an analysis of cerebellar neuroanatomical differences in MSA subtypes, a relatively greater GM atrophy was observed in MSA-C cases (vs. MSA-P) in the right Crus II—a cerebellar region involved in executive motor control (108). Both MSA-P and MSA-C patients may exhibit putaminal and infratentorial atrophy with considerable overlap, however, putaminal/supratentorial atrophy was more predominant in MSA-P subtype, whereas infratentorial atrophy was more prominent in MSA-C subtype (109, 110).

MSA patients with cognitive impairment showed volumetric reduction in the left dorsolateral prefrontal cortex vs. cognitively-normal MSA cases (111). The authors suggested that cortical pathology contributes minimally to cognitive deficits in MSA, whereas frontostriatal degeneration may be the primary driver of cognitive dysfunction (as per the concept of “subcortical cognitive impairment”) (111). Another study identified cortical thinning in the parahippocampal and lingual cortices in MSA with dementia vs. cognitively-normal MSA patients (112).

Diffusion-Weighted and Diffusion Tensor Imaging in MSA

In MSA-P, higher D was identified in the putamen compared to subjects with PD, MSA-C and controls (113). Similarly, reduced FA and elevated apparent diffusion coefficient (ADC) values were observed in MSA-P in the putamen, cerebellum and pons vs. PD and controls (114). Combined analysis of elevated T2^* relaxation rate and putaminal D allowed discrimination of PD from MSA-P with high accuracy (113). Likewise, a multiregional evaluation of diffusivity changes in the pons, putamen and cerebellum was found to be more useful than single-region analysis (114). Compared to PSP, increased D was observed in MCP and pons in MSA cases, which correlated with cerebellar ataxia in these regions (115). In a comparison between MSA-P and MSA-C subtypes, elevated ADC values were observed in the putamen and pons in MSA-P cases vs. MSA-C, and in the cerebellum and MCP in MSA-C cases vs. MSA-P, highlighting distinct microstructural damage in these subtypes (116). Microstructural changes in the WM may be more pronounced early in the disease in MSA-C than in MSA-P (110). In MSA patients with cognitive impairment, a greater involvement of the cerebrum (specifically, reduced FA in the anterior corpus callosum) was identified compared to cognitively-normal MSA patients (117).

Using the bi-tensor diffusion analysis model, free-water was found to be elevated in the posterior SN in PSP, MSA, and PD vs. controls, however, this increase was significantly greater in PSP than in PD/MSA patients (118). In addition, depending upon the disease severity, free-water may be elevated in both the anterior and posterior SN in PD, MSA and PSP vs. controls (66). Free-water-corrected FA was elevated in MSA in the putamen and caudate vs. controls; whereas, it was decreased in the thalamus and increased in the SCP in MSA vs. PSP (66). Using a machine learning algorithm, combined analysis of free-water and free-water-corrected FA derived from selective regions-of-interest achieved excellent separation among PD, MSA, and PSP cases (66). Unsupervised machine learning-based classification of PD, MSA-P and MSA-C patients using multimodal neuroimaging measures (GM density, T2^* relaxation rate, and DTI) have also been pursued with favorable results (119). Free-water imaging using advanced diffusion models may become an effective tool in the differential diagnosis of parkinsonian disorders in the future.

Proton Magnetic Resonance Spectroscopy in MSA

Compared to controls, NAA/Cr ratios were smaller in the putamen in MSA-P, and in the pontine base in both MSA-P and MSA-C cases (120). Lower NAA/Cr ratios in the putamen and pontine base best discriminated MSA-P cases from PD (120). In another study, no significant differences were observed in the metabolites examined between MSA-P and PD, suggesting similar metabolic alterations in the two disorders (71). Likewise, cerebellar NAA/Cr and NAA/myo-inositol levels in MSA-P subjects were similar to those seen in PD (77). In MSA-C patients, cerebellar NAA/Cr and NAA/myo-inositol ratios were significantly reduced compared to PD, MSA-P, PSP-Richardson's syndrome (PSP-RS), and controls, whereas cerebellar myo-inositol/Cr ratios were elevated in MSA-C compared to controls (77).

Magnetic Resonance Imaging in PSP

Structural MRI typically shows atrophy of the midbrain and SCP in PSP compared to PD, MSA-P, CBD/CBS and controls (121–124). Several morphological markers suggestive of PSP on MRI have been reported, including midbrain atrophy compared to pons (“hummingbird” sign; specificity ~99.5%, sensitivity ~51%; Figure 5), atrophy of the midbrain tegmentum (“morning glory” sign, specificity ~97%, sensitivity ~37%; visualized as concavity of the lateral margins of the midbrain tegmentum on axial images; Figure 5) (125–127), midbrain T2 hyperintensity, as well as atrophy of the midbrain tegmentum with relative preservation of the midbrain tectum and cerebral peduncles (“mickey mouse” sign, visualized as rounded rather than rectangular midbrain peduncles on axial images) (99, 125). Importantly, the “hummingbird” and “morning glory” signs had high specificity but low sensitivity (99). Furthermore, image acquisition parameters may influence the appearance of these morphological features (4).

In PSP vs. controls, reduced volumes were reported in the brainstem, midbrain, and frontal GM (129), however, only the midbrain volume showed minimal overlap with the normal range on a case by case basis (129). In PSP-RS, atrophy of the midbrain was the predominant feature (4, 130). In a pathology-proven sample, atrophy of the midbrain and SCP was associated with PSP, whereas frontoparietal and pallidum degeneration in the absence of significant brainstem atrophy was suggestive of CBD (123). The premotor cortices and supplementary motor area were involved in both PSP and CBD cases (123). Besides midbrain atrophy, VBM studies also show degeneration in the subcortical structures including pons, thalamus and striatum, as well as widespread cortical atrophy in the frontal, prefrontal, insular, premotor and supplementary motor areas vs. controls (121, 131, 132). Moreover, WM degeneration was reported in the pulvinar, dorsomedial and anterior nuclei of the thalamus, superior and inferior colliculi, as well as in the mesencephalic, and frontotemporal regions (131, 132). Pontine atrophy rates were significantly slower in PSP vs. MSA-P, and rates in the frontal lobe and midbrain in PSP were associated with executive and motor impairment, respectively (133). The midbrain atrophy rate may serve as an effective outcome measure in PSP clinical trials (134). In addition, the support vector machine classification method yielded accuracy rates >80% for predicting PSP diagnosis using disease-specific regions-of-interest (pallidum, putamen, caudate nucleus, thalamus, midbrain and insula) compared to the whole-brain approach (135).

Predominant midbrain atrophy has been incorporated as a supportive imaging feature in the Movement Disorders Society's PSP diagnostic criteria (3). The ratios of the pons to midbrain area (P/M) (as shown in Figure 5) and MCP to SCP widths (MCP/SCP) were found to be larger in PSP compared to PD, MSA-P and controls (specificity and sensitivity, 100%) (128, 136). Using these ratios, an index was calculated as [(P/M) x (MCP/SCP)], termed the “magnetic resonance parkinsonism index,” which proved to be highly sensitive and specific for distinguishing PSP from PD, MSA-P and controls (128). A revised version incorporating the width of the third ventricle has been proposed [“magnetic resonance parkinsonism index” × (third ventricle width)/(frontal horns width)], which showed superior sensitivity (100%) and specificity (94.3%) in differentiating PSP-parkinsonism patients with slowness of vertical saccades from PD (137). It is important to recognize that midbrain atrophy may not be evident in all PSP subtypes and thus, atrophy patterns from other brain regions (such as the globus pallidus, frontal lobe, and cerebral peduncle) may provide additional information (138). It is unknown whether the above quantitative measures can distinguish PSP patients at early disease stages, or whether PSP subtypes with less pronounced midbrain atrophy can be distinguished, which provides impetus for further research.

Diffusion-Weighted and Diffusion Tensor Imaging in PSP

Compared to controls, diffusion imaging studies in PSP show variable findings and may reveal the following: increased D or decreased FA in the decussation of SCP, orbitofrontal WM, thalamus, cingulum, motor and supplementary motor area, as well as in the inferior fronto-occipital fasciculus, superior longitudinal fasciculus, anterior corpus callosum, arcuate fasciculus, posterior thalamic radiations, and internal capsule (115, 132, 139, 140). Elevated ADC values were observed in the putamen in PSP vs. controls over a 2 year period (141). Increased D in the decussation of SCP may discriminate PSP from MSA and PD (115), whereas increased ADC values in the putamen, globus pallidus and caudate nucleus may help distinguish PSP cases from PD (142). Another study calculated the FA score per subject for regions hypothesized to be involved in PSP (i.e., the SCP and frontal WM region), and reported >85% sensitivity and specificity for differentiating PSP from PD/DLB cases (143).

A multimodal study integrated volumetric MRI, DTI and neuromelanin-sensitive imaging, and identified several predictors for separating PSP-RS from controls (144). The best predictor was the neuromelanin-based SN volume followed by FA in the midbrain (144). The separation of PSP-RS cases from PD was achieved using neuromelanin-based SN volume, pons FA values, midbrain and globus pallidus volumes, and basal forebrain FA values (144). Another study identified greater atrophy, decreased FA, and increased D in the SCPs bilaterally in PSP-RS patients as compared to PD and controls (145). More advanced neuroimaging techniques have identified changes in the free water in the posterior SN (118). Specifically, free-water values derived from the bi-tensor diffusion model were significantly elevated in PSP in the posterior SN vs. MSA, PD and controls, and were observed in a characteristic pattern: PSP > MSA/PD > HC (118). In addition, free-water-corrected FA was elevated in PSP in the caudate, putamen, thalamus and vermis with accompanying decreases seen in the SCP and corpus callosum compared to controls (66).

Other studies have compared diffusion imaging parameters in PSP subtypes. Lower FA values were exclusively detected in the SCP in PSP-RS vs. PSP-parkinsonism patients, implicating SCP's involvement in postural instability (146). PSP patients with vertical supranuclear gaze palsy exhibited lower FA values in the midbrain vs. those with slowness of vertical saccades, highlighting the role of midbrain atrophy in vertical ocular dysfunction (146). PSP-RS patients also showed more severe and widespread diffusion abnormalities vs. PD, which reflects greater microstructural damage as consistent with greater overall brain atrophy often found in PSP-RS than in PD cases (147). Moreover, damage to the SCPs may be detected in both PSP-RS and PSP-parkinsonism subtypes (130, 145). When the two subtypes were directly compared, PSP-RS patients showed decreased FA and increased D in the left SCP vs. PSP-parkinsonism patients (145). Computer-aided diagnosis of PSP and its subtypes may be possible using diffusion-weighted/DTI measures (145, 147).

Proton Magnetic Resonance Spectroscopy in PSP

In PSP, reductions in the NAA/Cr ratios in the LN, brainstem, centrum semiovale, frontal, and precentral cortex, as well as reductions in the NAA/choline values in the LN are observed relative to controls (148, 149). A more prominent decline in NAA/Cr ratio was noted in the putamen vs. PD and MSA (150). PSP-RS patients had reduced cerebellar NAA/Cr and NAA/myo-inositol ratios vs. controls, and reduced cerebellar NAA/Cr ratio vs. PD patients (77). Compared to controls, PSP patients also showed a decrease in scyllo-inositol concentration (a stereoisomer of inositol) and scyllo-inositol/Cr ratio in the supplementary motor area, and both of these metabolic measures were directly related to attention and working memory functions (151). The pathological significance of scyllo-inositol reduction in PSP is currently uncertain (151).

Magnetic Resonance Imaging in CBD/CBS

In CBS/CBD, asymmetrical cortical atrophy in the frontoparietal lobe is commonly observed (Figure 6), contralateral to the clinically more affected side of the body (however, laterality may not be present in all cases). In CBS vs. controls, an asymmetric pattern of atrophy in the bilateral premotor cortex, superior parietal lobules, and striatum was identified (121). Compared to PSP, greater atrophy was observed in the dorsofrontal and parietal cortices in CBS (121, 129), whereas midbrain atrophy was more pronounced in PSP vs. CBS (121). Likewise, greater asymmetric GM degeneration in the inferior frontal and premotor cortex, parietal operculum, superior temporal gyrus, and hippocampus was detected, along with decline in FA primarily in the frontoparietal region vs. controls (152). In a meta-analysis of VBM studies, although a significant overlap was detected among PSP, MSA-P, and CBS cases, more prominent atrophy in the superior parietal lobe was observed in CBS (153). Importantly, CBS can be associated with significant pathological heterogeneity difficult to predict based on clinical presentation in life (15, 154). Thus, neuroimaging patterns using pathology-proven samples can improve diagnostic accuracy.

On MRI, the pattern of atrophy in CBS aligns with the “true” underlying pathology. Although, GM atrophy in a clinically-diagnosed CBS group was observed in the premotor cortices, supplemental motor area and insula, the pattern of atrophy aligned more closely with neuropathological diagnosis (155). For example, frontotemporal atrophy in CBS was associated with frontotemporal lobar degeneration with transactive response DNA binding protein-43 kDa (TDP-43) pathology, whereas temporoparietal atrophy was related to AD pathology (155). In contrast, focal atrophy predominantly involving the premotor cortex and supplementary motor area was observed in those pathologically diagnosed with CBD and PSP, although more severe changes in these regions suggested CBD over PSP (155). The degree of gross global atrophy is typically more severe in CBD vs. PSP (156). Another pathology-confirmed study observed GM degeneration in the dorsal prefrontal and perirolandic cortex, striatum, and brainstem in CBD vs. controls (6). Furthermore, in CBS due to frontotemporal lobar degeneration (tau or TDP-43), atrophy pattern progressed into the prefrontal cortex, striatum and brainstem, while in CBS due to AD, atrophy stretched into the temporoparietal cortex and precuneus regions (6). The predominant clinical syndrome in CBS (extrapyramidal vs. cognitive) was associated closely to the regional atrophy patterns (123). On fluid-attenuated inversion recovery images, subcortical WM hyperintensities with ventricular dilation (greater in the more affected lobe) were reported (157), however, these changes are not specific to CBS/CBD.

Diffusion-Weighted and Diffusion Tensor Imaging in CBD/CBS

Elevated D and reduced FA were detected in the posterior truncus of the corpus callosum in CBS compared to PD and controls suggesting transcallosal fiber degeneration (59). In addition to the corpus callosum (59, 152, 158), decreased FA in CBS was observed in the long frontoparietal connecting tracts, intraparietal associative fibers, and sensorimotor cortical projections (152). Similarly, reduced FA and increased D were noted in the WM of premotor, prefrontal and motor cortices, as well as in the middle cingulate bundle vs. controls—changes that were more pronounced contralateral to the more affected side (158). Compared to PSP, a more asymmetric, supratentorial and posterior pattern of WM tract degeneration was reported in CBS with greater involvement of the splenium of the corpus callosum, WM of the motor and premotor cortices, as well as the parietal lobes (158).

Proton Magnetic Resonance Spectroscopy in CBD/CBS

NAA/Choline and NAA/Cr levels were reduced in CBS patients in the frontoparietal cortex, LN and centrum semiovale vs. controls (149, 159). Lower NAA/Choline levels in the frontoparietal lobe may help differentiate CBS cases from PSP (149). In the parietal cortex of CBS patients, the NAA/Choline values were reduced contralateral to the clinically more affected side (159). A more pronounced reduction in NAA/Cr values was observed in the frontal cortex and putamen vs. PD, MSA and vascular parkinsonism, with prominent asymmetry in NAA/Cr ratios in the putamen (150). Lower putaminal NAA levels have been reported across the PD, MSA, PSP and CBS patients to varying degrees. Thus, laterality of metabolite ratios observed in the putamen in CBS may be helpful in differentiation in some cases.