This is the first study investigating presynaptic dopaminergic function along with extensive clinical and neuropsychological characterization in a large cohort of patients fulfilling clinical criteria for ‘probable CBD’. The primary outcome of this study was to describe the pattern of striatal FP-CIT uptake reduction in CBS and correlate FP-CIT binding with clinical features. Compared to PD control subjects, individuals with CBS showed a large FP-CIT binding variability in terms of overall striatal binding as well as hemispheric asymmetry and caudate-to-putamen ratio ( Figures 1 and 2 ). In consistence with typical CBD pathological features [17], we found higher hemispheric asymmetry in caudate nucleus and putaminal FP-CIT uptake together with a more uniform reduction within striatal regions in CBS, thus reflecting the highly asymmetrical involvement and SNc neuronal loss involving both dorsal and ventral tiers in CBD [16], [17] in contrast to PD, where cell loss is typically confined to SNc ventral tiers [39]. These features well explain the difference in the caudate-to-putamen ratio of FP-CIT uptake between the two disorders. When correlation analysis was performed between nigrostriatal function and clinical features in CBS, FP-CIT binding showed did not show any correlation with disease duration and just a trend with disease severity, while this association was highly significant in PD control subjects ( Figure 3A and B ). Previous imaging studies already highlighted a less pronounced SNc neuronal loss in patients with CBS compared not only to severe atypical parkinsonian syndromes (such as progressive supranuclear palsy and multiple system atrophy) but also to idiopathic PD [40]–[44]. Interestingly, some CBS patients had symmetrical FP-CIT reduction despite strongly asymmetrical symptoms, a finding already described in a previous fluorodopa-PET study [41]. A recent description of autopsy-proven CBD with ante-mortem evidence of preserved presynaptic nigrostriatal terminals suggested a possible delayed timing of SNc cell degeneration, unrelated to the onset of motor symptoms [18], [19]. The lack of relationship between FP-CIT binding and disease duration and the very weak correlation with disease severity we found in our CBS cohort is in line with this hypothesis. Moreover, the presence of obvious bilateral symptoms but evidence of reduced DAT density only unilaterally in some CBS cases would further suggest that SNc pathology may require a variable time to occur bilaterally. These findings highlight a mismatch between SNc neuronal loss and clinical features (such as asymmetry of motor symptoms, disease duration and severity) and support the hypothesis of a prominent role played by cortical and/or striatal pathology in the clinical picture of this syndrome. The lack of any correlation between FP-CIT binding and cognitive performance is an expected finding that is consistent with ‘cortical’ rather than ‘subcortical’ pathology underlying cognitive deficits in CBS compared to PD.

As secondary objective, we aimed to assess the prevalence and characterize those CBS subjects with normal FP-CIT scans aiming to find any distinctive clinical or neuropsychological feature. To achieve this objective, we used very restrictive a priori criteria to minimize false positive cases and found preserved SNc neuronal density in approximately 10% of our cases. So far, normal dopamine transporter density has been reported only sporadically in CBS [19], [45], while none of the few systematic imaging studies in consecutive cohorts ever documented any normal scan [40]–. To our knowledge, there are only few reports of patients with clinical diagnosis of CBD with preserved SNc neuronal density at post-mortem, and most of them displayed either Alzheimer's or Pick's disease pathology [2], [3], [5], [7], [9], [10], [14], [15], [46]–[48]. The first possible explanation would come from a possible pathological involvement of the downstream postsynaptic neurons, even though the pathological involvement of postsynaptic striatal neurons in CBD is variable and unpredictable [17], [49]. A second hypothesis would posit that extrapyramidal symptoms in CBS with normal SNc density might be associated with frontal and/or parietal cortical rather than striatal pathology [4], [7], [19], [50], [51]. In the attempt to rule this out, we extensively investigated clinical, neuropsychological and MR imaging features of our CBS cohort and did not find any major clue suggesting an alternative diagnosis than CBD. Although early memory impairment was more frequently recorded in the history of CBS_N cases (Table S1), full neuropsychological assessment revealed similar features in both subgroups, including preserved memory functions, consistent with the pattern shown in definite CBD (Table S2) [52]. Notably, our cohort is representative of specialized movement disorders centres so that the number of cases presenting with prominent cognitive impairment and mild-to-absent extrapyramidal symptoms is relatively low, as these subjects would rather refer to third-level dementia clinics [20], [52]. SNc depigmentation is less evident in CBD cases presenting with severe cognitive symptoms [53] and it is thus conceivable that we underestimated the real prevalence of normal FP-CIT scans in CBS population [6], [20]. Moreover, CBS subjects with normal scans showed a mean longer disease duration than CBS patients with presynaptic nigrostriatal degeneration despite similar degrees of motor disability, suggesting a more indolent course.

We acknowledge that the interpretation of our findings is limited by the lack of pathological confirmation of the diagnosis. Indeed, some CBS patients with normal FP-CIT binding might still turn out to be focal cortical variants of other dementia than CBD [3], [7], [10], [46], [47], [54]–[59], because clinical features and the distribution of pathological lesions are closely associated whatever the histologic nature [3], [4], [9], [14]. Post-mortem assessment in CBS cases with preserved SNc neuronal density could also unravel the pathological underpinnings of extrapyramidal symptoms. A second limitation in the interpretation of the comparison between CBS_N and CBS_P comes from the marked unbalance in the number of cases between the two groups.

Taken as a whole, our findings support the hypothesis of a mismatch between SNc neuronal loss and clinical features with a variable contribution played by supranigral pathology in extrapyramidal phenotype. We speculate that the well-known heterogeneity of CBD pathology [60] might additionally include variants with predominant involvement of the cerebral cortex and disease progression in the rostro-caudal direction to the brainstem, in a kind of ‘cortico-to-basal’ progression of neuronal degeneration. This hypothesis is supported by pathological studies showing a less evident SNc depigmentation in CBD cases presenting with cognitive rather than motor symptoms [53] and already demonstrated in Lewy bodies dementia [61], [62]. Clinicopathological studies including in vivo DAT imaging are needed to assess the predictive value of normal presynaptic nigrostriatal function in the diagnostic work up of patients with CBD-like phenotype.