Brain-based Evaluation of Autism Phenotypes Using Longitudinal, Multimodal Imaging Data

Informations projet

BEAUTIPHUL MIND

N° de convention de subvention: 101025785

DOI

10.3030/101025785

Projet clôturé

Date de signature de la CE 29 Mars 2021

Date de début 1 Septembre 2021

Date de fin 1 Octobre 2023

Financé au titre de

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Coût total

€ 191 149,44

Contribution de l’UE

€ 191 149,44

Coordonné par

UNIVERSITAT ZURICH
Switzerland

Periodic Reporting for period 1 - BEAUTIPHUL MIND (Brain-based Evaluation of Autism Phenotypes Using Longitudinal, Multimodal Imaging Data)

Période du rapport: 2021-09-01 au 2023-08-31

Résumé du contexte et des objectifs généraux du projet

Autism is a common neurodevelopmental condition with social-commuication challenges and restricted repetitive behaviours. It is exclusively diagnosed based on symptoms which vary considerably across individuals. Little progress has been made in identifying objective, biological markers informative of core features to support targeted clinical management. This is due to a lack of understanding the complex etiology involving different neurobiological systems underlying heterogeneous conditions like autism. In recent years, novel neuroimaging techniques have provided noninvasively a window into the structural and functional properties of the brain and are well suited to explore biomarkers. However, the prevailing approach in research is to identify clinical commonalities across patients (i.e. average group-differences) confined to single biological measures (i.e. unimodal) which has fallen short of acknowledging inter-individual heterogeneity and system-level biology. Here, I aimed to study different, complementary aspects of the brain simultaneously at the individual level in a large, deeply phenotyped sample, to identify a more fine-grained picture of associated neural features of core symptoms, such as face processing, in autism. Different neuroimaging markers within key regions of the face processing network, such as the fusiform gyrus (FFG), have previously been studied extensively across single modalities. However, how different neurobiological markers are simultaneously implicated in face processing and social functioning in autism remains elusive. Integrating data from multiple modalities is crucial to uncover the intricate links between brain structure and function. The overarching objective of this project was to establish an individual-level, multimodal characterization of the neural substrates of face processing in autism informative of different social features associated with autism. The development of objective, biological markers is pivotal for overcoming the urgent, unmet medical need for tailored interventions. Results suggest that the FFG is a central region differentially implicated in autistic and non-autistic individuals (NAI) across different biological levels simultaneously informing mechanisms associated with social functioning in autism. Integrating data from different modalities has the advantage of being biologically more plausible and comprehensive in characterizing complex, heterogenous conditions and is the most promising and powerful method to achieve significant advances in our understanding of system-level atypicalities in autism.

Travail effectué depuis le début du projet jusqu’à la fin de la période considérée dans le rapport et principaux résultats atteints jusqu’à présent

I leveraged the large EU-AIMS Longitudinal European Autism Project dataset and aggregated structural and functional neuroimaging data (i.e. volume, electroencephalography (EEG), resting-state functional magnetic resonance imaging (fMRI) and task-fMRI) within the FFG. After unimodal feature extraction, I trained normative models for each brain imaging modality independently to obtain a more sensitive individual-level measure capturing heterogeneity. Next, I merged the different individual-level deviations using Linked Independent Component Analysis which provides a decomposition of the brain features into independent components (ICs). To establish functional relevance, I compared ICs between autistic and NAI. To test for the added value of multimodal features in characterizing autism, I compared unimodal features individually between autistic and NAI and tested whether multimodal ICs outperformed unimodal ICs in differentiating autistic from NAI. Finally, I ran canonical correlation analyses to establish the association between different ICs and social measures (and non-social measures to establish specificity). Longitudinal analyses are still ongoing.

I identified different, multimodal ICs among which one showed a difference between autistic and NAI. This IC overlapped with both face-selective and retinotopic regions of the FFG implying the involvement of both low-level and higher-level processing atypicalities. When comparing unimodal deviations and ICs between autistic and NAI, there were no significant group differences, despite employing a more sensitive individual-level measures derived from normative modeling. Also, using multimodal features significantly outperformed unimodal features in differentiating autistic from NAI. This emphasizes the added value of employing a cross-modal approach and highlights that shared variance across different modalities when modelled properly increases sensitivity. Furthermore, a set of multimodal ICs showed a significant multivariate association with social features in autism. This was not the case for non-social features, pointing to specificity. Taken together, these findings highlight the value of cross-modal analyses in characterizing a key structure in the multilevel neurobiology of autism and its implication in core social functioning.

Furthermore, I (co-)published over 15 papers, two preprints and two papers are currently under peer-review. I extended previous and established new collaborations which resulted in joint publications. I was invited to give more than 10 talks at (inter-)national conferences, research institutions and companies. To secure funding of my future research goals, I received a competitive research grant as the principal investigator for another two years.

Progrès au-delà de l’état des connaissances et impact potentiel prévu (y compris l’impact socio-économique et les conséquences sociétales plus larges du projet jusqu’à présent)

There is still little knowledge about how structural and functional neurobiological markers are simultaneously implicated in face processing and social functioning in autism. Extracting the joint information across modalities is essential to better elucidate complex relationships between brain structure and function leading to a more comprehensive understanding of associated mechanisms of autism which will pave the way for more personalised support. My work employed an innovative research line promoting new ways of investigating brain structure and function simultaneously and their association with clinical presentations via biologically informed, integrative, and more sensitive, individual-level characterisations of neurodevelopmental conditions in combination with multivariate methods. I further contributed substantially to the state of the art moving beyond established analytical approaches. Multimodal methods have been available for several years; however, their use is still limited and most neuroimaging research does not exploit the rich information inherent to different modalities, which when combined yields a biologically more valid and comprehensive characterization of complex, multifaceted conditions such as autism. Furthermore, I benchmarked against current best practices in the field by comparing unimodal with multimodal results and showing the added value when incorporating multiple imaging modalities. Through dissemination activities, upcoming publications, and planned follow-up work, I am hoping to inspire the direction of future research to adopt a similar more integrative approach. In the final part of the project, I will implement functional outcome prediction by establishing which individual-level, multimodal features are most predictive of social outcome in autism. Eventually, results from this research line will inform translation of techniques and of biological signatures across different scales into practical tools for clinical phenotyping and decision-making in the context of diagnostics, prediction of course, and outcome events.

figure1.png