Altered brain metabolic connectivity at multiscale level in early Parkinson’s disease

www.nature.com/scientificreports OPEN Received: 24 January 2017 Accepted: 11 May 2017 Published: xx xx xxxx Altered brain metabolic connectivity at multiscale level in early Parkinson’s disease Arianna Sala1,2, Silvia Paola Caminiti1,2, Luca Presotto2, Enrico Premi3, Andrea Pilotto3,4, Rosanna Turrone3, Maura Cosseddu3, Antonella Alberici3, Barbara Paghera5, Barbara Borroni3, Alessandro Padovani3 & Daniela Perani1,2,6 To explore the effects of PD pathology on brain connectivity, we characterized with an emergent computational approach the brain metabolic connectome using [18F]FDG-PET in early idiopathic PD patients. We applied whole-brain and pathology-based connectivity analyses, using sparse-inverse covariance estimation in thirty-four cognitively normal PD cases and thirty-four age-matched healthy subjects for comparisons. Further, we assessed high-order resting state networks by interregional correlation analysis. Whole-brain analysis revealed altered metabolic connectivity in PD, with local decreases in frontolateral cortex and cerebellum and increases in the basal ganglia. Widespread longdistance decreases were present within the frontolateral cortex as opposed to connectivity increases in posterior cortical regions, all suggestive of a global-scale connectivity reconfiguration. The pathologybased analyses revealed significant connectivity impairment in the nigrostriatal dopaminergic pathway and in the regions early affected by α-synuclein pathology. Notably, significant connectivity changes were present in several resting state networks especially in frontal regions. These findings expand previous imaging evidence of altered connectivity in cognitively stable PD patients by showing pathologybased connectivity changes and disease-specific metabolic architecture reconfiguration at multiple scale levels, from the earliest PD phases. These alterations go well beyond the known striato-cortical connectivity derangement supporting in vivo an extended neural vulnerability in the PD synucleinopathy. Parkinson’s disease (PD) is a neurodegenerative disease predominantly characterized by abnormal intracellular accumulations of insoluble α-synuclein into fibrils1. It has been postulated that the synaptic dysfunction caused by the small aggregates of α-synuclein is the initial event leading to neurodegeneration in PD and in other synucleinopathies2. Since physiological α-synuclein plays a key role in the regulation of the dopaminergic normal synaptic function, dopaminergic neurons are particularly vulnerable to α-synuclein pathology2. Synaptic dysfunction might impair both neurotransmitter release and regulation of synaptic plasticity mechanisms, also producing widespread effects on functional connectivity among distant brain regions which may result in neural networks alterations3. All above evidence suggests that PD can be regarded as a “network-opathy”4, and that adopting a network perspective is essential to understand its pathophysiology5. During the last years, an increasing number of neuroimaging studies reported networks alterations in the PD brain. The most consistently reported connectivity change is the alteration of the striato-cortical loop, both in the form of microstructural damage, as shown by diffusion tensor imaging, or as functional connectivity changes by resting state fMRI (rs-fMRI (see refs 5 and 6). Connectivity changes between striatal and cortical regions have been associated with resting tremor7, freezing of gait8 and overall motor symptoms severity, as measured by UPDRS-III score9, thus suggesting that impairment of the cortico-striatal loops is a key phenomenon for the occurrence of motor symptoms in PD6. A limited amount of rs-fMRI studies assessed connectivity in non-motor resting state networks, reporting reduced connectivity in (i) default mode network 1 Vita-Salute San Raffaele University, Via Olgettina, 58, 20132, Milan, Italy. 2Division of Neuroscience, San Raffaele Scientific Institute, Via Olgettina, 58, 20132, Milan, Italy. 3Neurology Unit, Department of Clinical and Experimental Sciences, University of Brescia, piazzale Spedali Civili, 25123, Brescia, Italy. 4Parkinson’s disease Rehabilitation Centre, FERB ONLUS S. Isidoro Hospital, Trescore, Balneario, Italy. 5Nuclear Medicine Unit, Azienda Ospedaliera “Spedali Civili”, Spedali Civili Hospital, 25123, Brescia, Italy. 6Nuclear Medicine Unit, San Raffaele Hospital, Via Olgettina, 60, 20132, Milan, Italy. Correspondence and requests for materials should be addressed to D.P. (email: ) Scientific Reports | 7: 4256 | DOI:10.1038/s41598-017-04102-z 1 www.nature.com/scientificreports/ (DMN), correlating with working memory and visuo-spatial scores10, (ii) attentional network, also associated with presence of visual hallucinations11 and (iii) fronto-parietal network, correlating with executive performances12. Altogether, these studies show that several changes in connectivity underlie the symptomatic aspects of PD5, and that the dopaminergic degeneration inducing loss of striato–cortical functional connectivity is only part of a larger picture9. Notably, as recently suggested, not the dopaminergic cell death, but the abnormal accumulation of α-synuclein could be the initial event leading to neurodegeneration in PD and in other synucleinopathies2. α-synuclein aggregation affects key synaptic proteins, impairing neuronal function and axonal transport1, 2. These effects on both neurotransmitter release and regulation of synaptic mechanisms can affect activity-dependent signalling and gene expression3 leading to synaptic disconnection and network disintegration3, 5. A unique tool to capture in vivo the pathological events that contribute to synaptic dysfunction is [18F]fluorodeoxyglucose with positron emission tomography ([18F]FDG-PET). [18F]FDG-PET is considered as an effective measure of energy consumption in neurons (specifically in synapses)13, and its signal has also been associated with synaptic density and function14, altered intracellular signalling cascades, impaired neurotransmitter release, spreading of proteinopathies, and long distance disconnection (see ref. 15). Crucially, brain energy consumption as measured by [18F]FDG-PET reflects neuronal communication signalling16, and it is thus strictly interrelated with functional connectivity17, 18. Thus, a metabolic network perspective might allow for important insights into local neural vulnerabilities, long-range disconnection, and the effects of neuropathology. In the last years, several approaches have been developed in order to assess metabolic connectivity in the human brain19, 20. Based on the assumption that regions whose metabolism is correlated are functionally interconnected21, seed-based voxel wise analysis20 and Sparse Inverse Covariance Estimation (SICE) method19 represent suitable approaches to measure functional interconnections between brain regions. These methods provide results comparable to those derived by rs-fMR (...truncated)