Multivariate projection method to investigate inflammation associated with secondary insults and outcome after human traumatic brain injury: A pilot study

Abstract

Background: Neuroinflammation has been proposed as a possible mechanism of brain damage after traumatic brain injury (TBI), but no consensus has been reached on the most relevant molecules. Furthermore, secondary insults occurring after TBI contribute to worsen neurological outcome in addition to the primary injury. We hypothesized that after TBI, a specific pattern of cytokines is related to secondary insults and outcome. Methods: A prospective observational clinical study was performed. Secondary insults by computerized multimodality monitoring system and systemic value of different cytokines were collected and analysed in the first week after intensive care unit admission. Neurological outcome was assessed at 6 months (GOSe). Multivariate projection technique was applied to analyse major sources of variation and collinearity within the cytokines dataset without a priori selecting potential relevant molecules. Results: Twenty-nine severe traumatic brain injury patients undergoing intracranial pressure monitoring were studied. In this pilot study, we demonstrated that after TBI, patients who suffered of prolonged and severe secondary brain damage are characterised by a specific pattern of cytokines. Patients evolving to brain death exhibited higher levels of inflammatory mediators compared to both patients with favorable and unfavorable neurological outcome at 6 months. Raised ICP and low cerebral perfusion pressure occurred in 21 % of good monitoring time. Furthermore, the principal components selected by multivariate projection technique were powerful predictors of neurological outcome. Conclusions: The multivariate projection method represents a valuable methodology to study neuroinflammation pattern occurring after secondary brain damage in severe TBI patients, overcoming multiple putative interactions between mediators and avoiding any subjective selection of relevant molecules.