Multimodal Resonance Imaging for Outcome Prediction on Coma Patients

Predicting the awakening of patients in comas is one of the principal stakes of the current neurointensive care unit (neuroICU). Several studies and clinical practice suggest that the multimodal MRI, which associates the traditional morphological sequences (T1, T2*, FLAIR/T2), the spectroscopy-MRI (MRS) and the diffusion tensor imaging, is a tool allowing such a prediction. However, this strategy has not been yet validated. Additionally, currently there is no method of analysis including the 4 different sequences.

Objective: The goal of this study is to develop a composite score able to predict the awakening of coma patients following events such as a severe cranial trauma, ischemic or hemorrhagic cerebrovascular accident and cerebral anoxia. This composite score will be built from the results of the multimodal MRI (quantified indicator) in combination with clinical covariables (e.g., age of the patient, the mechanism of the accident (high versus low speed), etc.). The final score will aim to predict the outcome of patients at 1 year, evaluated by one of the following categories: favourable (Glasgow Outcome Scale (GOS 3+, 4, and 5) or unfavourable outcome (GOS 1, 2, and 3). GOS 3- score has been defined as minimally conscious state and GOS 3+ score as severe disability excluding cognitive sequelae.

MRI Analysis: The lesions present on the MRI will be quantified by a neuroradiologist and a dedicated clinical engineer from the coordination centre (Pitié-Salpêtrière Hospital) in a blinded way regarding patients' clinical data. Lesion load-indicators will be calculated on the sequences of FLAIR/T2, T2*, MRS and diffusion tensor imaging from a predefined analysis grid allowing the regional study of the lesions as well as the appreciation of their nature, their uni- or bilateral character and if bilateral, their symmetry.

Hypothesis and applicability: The multivariate analysis of morphological MRI, MRS and diffusion tensor imaging data, combined with the clinical covariables, will aim to develop a statistical algorithm, able to predict the clinical outcome of the patients. In the long term, it will be integrated into an expert system which will be the subject of a patent submission. The final objective is to provide the clinicians a diagnostic tool able to determine outcome of patients with severe cranial trauma and other neurological conditions such as stroke, subarachnoid hemorrhage and cerebral anoxia.

Inclusion criteria

• Adult affiliated to the social security system
• Hospitalized in neuroICU and requiring artificial ventilation following a severe cranial trauma, an ischemic and/or hemorrhagic cerebrovascular accident and/or a cerebral anoxia
• In coma defined as not answering simple orders at least 7 days after the event
• Receiving an amount of sedatives and not being able to explain the coma
• Having a standardized intracranial pressure (≤ 15 mm Hg) and in absence of severe hemodynamic or respiratory failure so that the MRI does not represent any additional danger

Exclusion criteria

• Coma of infectious or toxic origin
• Coma explained by sedation
• MRI contraindication (pace maker, medical material not MRI compatible)
• Persistent intracranial hypertension contraindicating the setting in prolonged dorsal decubitus (ICP > 20 mm Hg during more than 10 minutes after positioning on decubitus)
• Severe hemodynamic failure
• Severe respiratory failure
• Life threatening extra-cerebral lesions
• Cranial trauma of ballistic origin
• Patient presenting a severe compromise of a major function bringing into play the vital outcome, with invalidity before the accident/event
• Former neurological pathology altering the clinical outcome of the cranial trauma
• Refusal of the family
• Patient protected by the law (under supervision or trusteeship)