Recherche translationnelle en santé,
technologie pour la santé et recherche clinique
Soutenance Benjamine Sarton – 11/10/2024
Benjamine Sarton soutiendra sa thèse intitulée « Exploration mécanistique des états de coma : Structure, fonction, activation neuroimmune« , le Vendredi 11 Octobre à 9 heures en salle de conférence du 1er étage du pavillon Baudot.
Composition du jury:
Rapporteurs
Dr. Sophie ACHARD (Grenoble)
Dr. Nicolas LEJEUNE (Liege)
Membres du Jury:
Pr. Tarek SHARSHAR (Paris)
Pr. Pierre PAYOUX (Toulouse)
Co-Direction:
Pr. Stein SILVA (Toulouse)
Dr. Patrice PERAN (Toulouse)
Membres invités:
Dr. Clovis TAUBER (Tours)
Pr. Jean Marc OLIVOT (Toulouse)
Mechanisms of coma : Structure, function and neuroimmune activation
Coma resulting from acute cerebral injury (ACI), such as traumatic brain injury, anoxic-ischemic events due to cardiac arrest, or encephalitis , is a major cause of death and disability worldwide. Early assessment of neurological prognosis is challenging for intensivists. Prognostic evaluations suggesting a lack of consciousness recovery can lead to decisions of withdrawal of life sustaining therapies, and death. Despite advances in neuroscience identifying key brain structures and networks involved in coma (e.g., the medial fronto-parietal network and mesocircuit), prognostic assessments and therapeutic proposals remain inadequate. No treatment currently exists to modulate recovery from altered states of consciousness. Future progress in these areas depends on a better understanding of the brain mechanisms involved in coma and their recovery. This thesis focuses on three approaches to evaluating coma states across different ACI models using multimodal imaging techniques: structural, functional, and molecular imaging. The goal is to identify coma signatures, propose new mechanistic pathways, and assess their link with patient prognosis.
-Approach 1: We investigated structural prognostic signatures of poor functional outcomes in patients with severe herpes simplex encephalitis (HSE) within a multicentric French retrospective cohort. Early MRI scans were systematically reviewed by two radiologists blinded to prognosis. Extensive FLAIR hyperintensities affecting more than three lobes, bilateral diffusion abnormalities, and left thalamic involvement were significantly associated with a lack of functional autonomy at three months (modified Rankin Scale 3-6).
-Approach 2: We conducted a proof-of-concept study using a deep learning model based on 3D convolutional neural networks (3D-CNN) to analyze and merge structural and functional MRI data from post-anoxic coma patients and healthy subjects. The model performed excellently in differentiating between these two groups. Visualization tools revealed that functional variables provided the best performance, although both structural and functional data were used for final discrimination. Regions identified by the model as crucial for discrimination were key to consciousness networks. This work confirms the effectiveness of 3D-CNNs for discriminating complex multimodal MRI datasets and highlights the interpretability provided by visualization tools, enhancing confidence in the model.
-Approach 3: We explored in vivo neuro-immune activation during the acute phase in post-anoxic or post-traumatic coma patients. Patients were compared to healthy subjects using molecular imaging with the tracer 18F-DPA-714, targeting TSPO receptors expressed by activated immune cells. Results showed significantly higher tracer uptake in comatose patients, indicating neuro-immune mechanisms involvement. The affected regions were part of consciousness networks but differed between etiological contexts. Higher whole-brain tracer uptake was associated with a lack of consciousness recovery at three months. A partial least squares (PLS) regression model demonstrated good performance in predicting prognosis based on voxel-based tracer uptake.
To conclude, the research presented in this thesis confirms and identifies structural, functional, and neuro-immune activation signatures specific to coma states. It proposes new evaluation methods, particularly through integrating AI in multimodal imaging analysis, and explores innovative mechanistic pathways, marking a paradigm shift in the field. These advances pave the way for new perspectives in prognostic evaluation and therapeutic development. Future research should focus on developments in structural and functional imaging, AI applications and model explainability, exploring central and peripheral immune interactions, as well as new modifiable mechanistic pathways based on consciousness theories.