Predictive Model for the Occurrence of Cerebral Vasospasm Complicating Subarachnoid Haemorrhage by Combined Analysis of the Kinetics of a Panel of Biomarkers.

Overview

The objective is to create a dynamic clinical prediction model that includes routinely measured care and biological biomarkers to predict cerebral vasospasm within 14 days of bleeding in patients treated in the neurosurgical intensive care unit for subarachnoid hemorrhage. Patients admitted to intensive care will be followed for up to 14 days (D14 time horizon of interest), or until discharge from intensive care if earlier. Blood samples will be taken from D1 to D10 to isolate the blood biomarkers of interest for each patient. The measurement of biomarkers and cerebral vasospasm will be blinded to each other.

Subarachnoid hemorrhage (SAH) is the rupture of a cerebral aneurysm, resulting in bleeding into the subarachnoid space. This condition has significant morbidity and mortality. The patient's functional outcome is primarily determined by the severity of cerebral ischemic lesions that develop during the first few weeks after the acute phase. The main focus of resuscitation management in patients are the 'delayed ischemic lesions'. These lesions are caused by various phenomena, with vasospasm being the most common mechanism. The caliber of cerebral arteries will shrink, reducing the blood flow delivered to the parenchyma, leading to a deficit of energy metabolites in neurons and causing their death. This complication typically occurs within a well-defined time frame, ranging from 3 to 21 days after bleeding, and peaking around the seventh day. The objective of this study is to create a novel predictive method for symptomatic vasospasm. This method will incorporate routine clinical and radiological biomarkers, as well as innovative biological assays. The aim is to enable earlier diagnosis and even pre-emptive treatment of this pathology. Several studies have examined the predictive potential of various blood biomarkers for neurological prognosis and the incidence of delayed brain damage in patients. These studies have demonstrated strong associations between them. For instance, one study found that patients with the most severe vasospasm had a significantly higher peak in cerebrospinal fluid of several biomarkers associated with neurodegeneration, such as Neuron Specific Enolase (NSE). Additionally, the plasma concentration-time curves demonstrated simultaneous elevations during periods of vasospasm. However, no study has examined the practical clinical use of these biomarkers during hospitalization to predict the occurrence of vasospasm on a daily basis or at specific times of interest. This is particularly important as the pathophysiological time sequence appears to be common to all patients. A single study has attempted to establish a predictive algorithm for delayed cerebral lesions, achieving a certain degree of effectiveness (over 90% correct predictions and sensitivity of around 93%), by combining a single biomarker assay and clinical parameters. However, this study only focuses on ischemic lesions at 6 weeks and cannot be used to guide therapy during initial management. Within the framework of a predictive statistical model, the investigators wish to study the possibility of combining routine clinical and radiological parameters with iterative assays of a panel of biomarkers covering several pathophysiological pathways (which would be easily assayable in the plasma of all patients) to predict on a daily basis (or at certain times of clinical interest) the risk of occurrence of cerebral vasospasm, with a view to being able to trigger diagnostic (or even therapeutic) procedures during initial management to prevent the ischemic cascade. To this end, the investigators have chosen 3 assays as a priority, targeting three previously described pathophysiological pathways. For neuroinflammation : * Interleukin-6 (IL-6) For cerebral cellular damage: * Neuron Specific Enolase (NSE) * The β-subunit of the S100 protein (S100 β) To enable this prediction, the investigators propose to use an innovative statistical method in the health sciences. Repeated biomarker assays can be integrated into complex event prediction models: the joint modeling of a longitudinal data model, for estimating individual biomarker trajectories over time, and a survival model, for estimating event risk with the current value or slope of the biomarker, enables precise event prediction. These models enable either static prediction (at a given time horizon) or dynamic prediction (with re-estimation of risk during follow-up). They also allow the concomitant integration of several biomarkers. Lastly, a model estimated on the study population could be transposed to other populations, thus making it possible to obtain risk models for this event. The aim of this research is to develop tools for daily clinical prediction of the onset of symptomatic vasospasm, using routine clinical and radiological parameters as well as innovative biological assays, with a view to triggering earlier diagnostic and even therapeutic responses than with the usual screening methods, which are severely limited and not always usable.