Functional magnetic resonance imaging (fMRI) is a very versatile tool to investigate brain activity, and in resting state conditions it allows the identification of several subnetworks subserving sensorimotor and cognitive functions. These subnetworks are composed of brain regions spatially distinct but functionally connected, and changes in their functional connectivity turned out to be a potential biomarker in the study of neurodegenerative pathologies. However, a comprehensive assessment of all brain subnetworks connectivity changes in neurodegeneration is still lacking. Even less is known about pathological changes affecting other network features, such as the excitatory/inhibitory (E/I) balance. A pivotal role of the imbalance towards excitation or inhibition has been recognized in the cascade of neurodegenerative pathophysiology. However, how subnetworks E/I balance acts in concert with connectivity changes to determine cognitive decline still needs a further assessment. The development of brain models recently enabled the overcome of previous limitations and the non-invasive investigation of intrinsic brain features. In this context, The Virtual Brain (TVB) is a novel neuroinformatic platform which makes possible to simulate the cerebral activity starting from MRI data and extract information about connectivity and E/I pathways. The main aim of this work was to exploit TVB power in multiple subnetworks to assess whether model-derived parameters could be used as new neurodegenerative biomarkers. A heterogeneous cohort of subjects was considered, including healthy controls and patients affected by typical forms of Alzheimer’s disease (AD) and atypical phenotypes. Participants underwent resting-state fMRI and diffusion weighted imaging (DWI) to reconstruct their functional and structural connectivity at whole-brain and in six functional subnetworks: the visual, the somatomotor, the attention, the limbic, the frontoparietal and the default mode network. Thanks to TVB simulations, four specific parameters (G, JNMDA, Ji, w+) linked to biological characteristics were derived for each subnetwork at single-subject level. These parameters gained a description of long-range connectivity (G), excitatory (NMDA) synapses strength (JNMDA), inhibitory (GABA) synapses strength (Ji) and recurrent excitation (w+), endowing a subject-specific representation of subnetwork connectivity and E/I balance. In this way it was possible to explore whether subnetworks showed different connectivity and E/I balance in subjects belonging to the same clinical group and whether inter-networks relationship changed passing from health to pathology. Further, TVB-derived parameters were used to assess neurophysiology and neuropsychology relationship and perform patients’ stratification. Our results showed differences in the E/I balance and connectivity between functional subnetworks, and these differences changed passing from healthy to pathological conditions. This evidence suggested the presence of a specific subnetwork-dependent E/I balance which was influenced by neurodegeneration. Moreover, the relationship between TVB-derived parameters and subjects’ neuropsychological performance in multiple cognitive domains was demonstrated, supporting TVB parameters clinical relevance. Finally, clustering analysis identified cognitive subnetworks properties as the most informative features to perform patients’ stratification. Subjects resulted distributed into seven clusters, outlining a personalized fingerprint based on cognitive subnetworks profile and sensitive to pathological severity. In conclusion, this study exploited TVB power to shed a new light on subnetworks functioning and their disruption in neurodegeneration, defining new potential biomarkers. The results achieved so far provide new means to answer pathology-specific questions and pave the way to a novel approach in differential diagnosis and tailored interventional workflows.
La risonanza magnetica è uno strumento versatile per investigare l’attività cerebrale e permette di identificare diverse network adibite a funzioni sensomotorie e cognitive. Queste network sono costituite da regioni cerebrali spazialmente distinte ma funzionalmente connesse, e cambiamenti nella loro connettività funzionale si sono rivelati un potenziale marcatore per la neurodegenerazione. Tuttavia, una valutazione completa delle alterazioni di connettività di tutte le network cerebrali non è ancora stata eseguita. Ancor meno si sa dei cambiamenti patologici a carico di altre caratteristiche intrinseche delle network, come l’“excitatory/inhibitory (E/I) balance”. Un ruolo fondamentale nella neurodegenerazione è stato riconosciuto allo squilibrio a favore dell’eccitazione o dell’inibizione. Tuttavia, è ancora da valutare come l’E/I balance delle network contribuisca a determinare il declino cognitivo. Lo sviluppo di simulatori avanzati ha recentemente permesso di superare i limiti e di esplorare in modo non invasivo nuove proprietà cerebrali. In questo contesto, The Virtual Brain (TVB) è un modello con cui simulare l’attività cerebrale a partire da dati di risonanza per estrarre informazioni riguardo la connettività e i fenomeni eccitatori/inibitori. Lo scopo di questo lavoro è stato quello di sfruttare la potenza del TVB per esplorare più network e stabilire se i parametri derivati dal modello possano essere usati come nuovi biomarcatori. Per questo studio è stata considerata una coorte di soggetti sani e pazienti affetti da forme tipiche di Alzheimer e da fenotipi atipici. Per ogni soggetto sono state acquisite immagini di risonanza magnetica funzionale e immagini pesate in diffusione per poterne ricostruire la connettività funzionale e strutturale di sei network cerebrali: visiva, somatomotoria, attentiva, limbica, frontoparietale e default mode network. Grazie alle simulazioni col TVB, è stato possibile ottenere per ogni soggetto quattro parametri (G, JNMDA, Ji, w+) con valenza biologica specifici di ogni network. Questi parametri descrivono la connettività (G), la forza (JNMDA) delle sinapsi eccitatorie (NMDA), la forza (Ji) delle sinapsi inibitorie (GABA) e la recurrent excitation (w+), fornendo una rappresentazione soggetto-specifica delle proprietà delle network. In questo modo è stato possibile investigare i valori network-specifici di connettività e di E/I balance in soggetti appartenenti allo stesso gruppo clinico e l’impatto patologico sui rapporti inter-network. Inoltre, i parametri derivati dal TVB sono stati utilizzati per esplorare la relazione tra neurofisiologia e neuropsicologia ed eseguire la stratificazione dei pazienti. L’analisi ha mostrato delle differenze nelle proprietà delle network, che cambiano da condizioni fisiologiche a patologiche. Ciò suggerisce che per ogni network è presente una specifica E/I balance e che il rapporto inter-network viene influenzato dalla neurodegenerazione. In più, è stata dimostrata la relazione tra i valori estratti dal TVB e la performance neuropsicologica dei soggetti, supportando la rilevanza clinica dei parametri del TVB. Infine, un’analisi di clustering ha indicato i parametri delle network cognitive come i più informativi per eseguire la stratificazione dei pazienti. I soggetti sono stati distribuiti in sette cluster, delineando un fenotipo personalizzato basato sul profilo delle network cognitive e sensibile alla gravità della patologia. In conclusione, questo studio ha sfruttato la potenza del TVB per far luce sulla fisiologia delle network cerebrali e sulla loro disfunzione nella neurodegenerazione, definendo nuovi biomarcatori. Questi risultati forniscono nuovi mezzi per rispondere a domande patologiche ancora in sospeso e aprono la strada ad un nuovo approccio nella diagnosi differenziale e nella medicina personalizzata.
Simulazione delle dinamiche cerebrali in reti funzionali per identificare nuovi biomarcatori della malattia di Alzheimer
ARGENTINO, FRANCESCA
2021/2022
Abstract
Functional magnetic resonance imaging (fMRI) is a very versatile tool to investigate brain activity, and in resting state conditions it allows the identification of several subnetworks subserving sensorimotor and cognitive functions. These subnetworks are composed of brain regions spatially distinct but functionally connected, and changes in their functional connectivity turned out to be a potential biomarker in the study of neurodegenerative pathologies. However, a comprehensive assessment of all brain subnetworks connectivity changes in neurodegeneration is still lacking. Even less is known about pathological changes affecting other network features, such as the excitatory/inhibitory (E/I) balance. A pivotal role of the imbalance towards excitation or inhibition has been recognized in the cascade of neurodegenerative pathophysiology. However, how subnetworks E/I balance acts in concert with connectivity changes to determine cognitive decline still needs a further assessment. The development of brain models recently enabled the overcome of previous limitations and the non-invasive investigation of intrinsic brain features. In this context, The Virtual Brain (TVB) is a novel neuroinformatic platform which makes possible to simulate the cerebral activity starting from MRI data and extract information about connectivity and E/I pathways. The main aim of this work was to exploit TVB power in multiple subnetworks to assess whether model-derived parameters could be used as new neurodegenerative biomarkers. A heterogeneous cohort of subjects was considered, including healthy controls and patients affected by typical forms of Alzheimer’s disease (AD) and atypical phenotypes. Participants underwent resting-state fMRI and diffusion weighted imaging (DWI) to reconstruct their functional and structural connectivity at whole-brain and in six functional subnetworks: the visual, the somatomotor, the attention, the limbic, the frontoparietal and the default mode network. Thanks to TVB simulations, four specific parameters (G, JNMDA, Ji, w+) linked to biological characteristics were derived for each subnetwork at single-subject level. These parameters gained a description of long-range connectivity (G), excitatory (NMDA) synapses strength (JNMDA), inhibitory (GABA) synapses strength (Ji) and recurrent excitation (w+), endowing a subject-specific representation of subnetwork connectivity and E/I balance. In this way it was possible to explore whether subnetworks showed different connectivity and E/I balance in subjects belonging to the same clinical group and whether inter-networks relationship changed passing from health to pathology. Further, TVB-derived parameters were used to assess neurophysiology and neuropsychology relationship and perform patients’ stratification. Our results showed differences in the E/I balance and connectivity between functional subnetworks, and these differences changed passing from healthy to pathological conditions. This evidence suggested the presence of a specific subnetwork-dependent E/I balance which was influenced by neurodegeneration. Moreover, the relationship between TVB-derived parameters and subjects’ neuropsychological performance in multiple cognitive domains was demonstrated, supporting TVB parameters clinical relevance. Finally, clustering analysis identified cognitive subnetworks properties as the most informative features to perform patients’ stratification. Subjects resulted distributed into seven clusters, outlining a personalized fingerprint based on cognitive subnetworks profile and sensitive to pathological severity. In conclusion, this study exploited TVB power to shed a new light on subnetworks functioning and their disruption in neurodegeneration, defining new potential biomarkers. The results achieved so far provide new means to answer pathology-specific questions and pave the way to a novel approach in differential diagnosis and tailored interventional workflows.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per contatti: unitesi@unipv.it
https://hdl.handle.net/20.500.14239/15182