Format

Send to

Choose Destination
Brain Behav Immun. 2016 Mar;53:183-193. doi: 10.1016/j.bbi.2015.12.008. Epub 2015 Dec 17.

Principal components derived from CSF inflammatory profiles predict outcome in survivors after severe traumatic brain injury.

Author information

1
Department of Physical Medicine and Rehabilitation, University of Pittsburgh, Pittsburgh, PA, USA.
2
Department of Mathematics, University of Pittsburgh, Pittsburgh, PA, USA. Electronic address: jonrubin@pitt.edu.
3
Safar Center for Resuscitation Research, University of Pittsburgh, Pittsburgh, PA, USA; Department of Pediatrics, University of Pittsburgh, Pittsburgh, PA, USA.
4
Safar Center for Resuscitation Research, University of Pittsburgh, Pittsburgh, PA, USA; Department of Critical Care Medicine, University of Pittsburgh, Pittsburgh, PA, USA.
5
Department of Physical Medicine and Rehabilitation, University of Pittsburgh, Pittsburgh, PA, USA; Safar Center for Resuscitation Research, University of Pittsburgh, Pittsburgh, PA, USA; Department of Neuroscience, University of Pittsburgh, Pittsburgh, PA, USA; Center for Neuroscience, University of Pittsburgh, Pittsburgh, PA, USA. Electronic address: wagnerak@upmc.edu.

Abstract

Studies have characterized absolute levels of multiple inflammatory markers as significant risk factors for poor outcomes after traumatic brain injury (TBI). However, inflammatory marker concentrations are highly inter-related, and production of one may result in the production or regulation of another. Therefore, a more comprehensive characterization of the inflammatory response post-TBI should consider relative levels of markers in the inflammatory pathway. We used principal component analysis (PCA) as a dimension-reduction technique to characterize the sets of markers that contribute independently to variability in cerebrospinal (CSF) inflammatory profiles after TBI. Using PCA results, we defined groups (or clusters) of individuals (n=111) with similar patterns of acute CSF inflammation that were then evaluated in the context of outcome and other relevant CSF and serum biomarkers collected days 0-3 and 4-5 post-injury. We identified four significant principal components (PC1-PC4) for CSF inflammation from days 0-3, and PC1 accounted for the greatest (31%) percentage of variance. PC1 was characterized by relatively higher CSF sICAM-1, sFAS, IL-10, IL-6, sVCAM-1, IL-5, and IL-8 levels. Cluster analysis then defined two distinct clusters, such that individuals in cluster 1 had highly positive PC1 scores and relatively higher levels of CSF cortisol, progesterone, estradiol, testosterone, brain derived neurotrophic factor (BDNF), and S100b; this group also had higher serum cortisol and lower serum BDNF. Multinomial logistic regression analyses showed that individuals in cluster 1 had a 10.9 times increased likelihood of GOS scores of 2/3 vs. 4/5 at 6 months compared to cluster 2, after controlling for covariates. Cluster group did not discriminate between mortality compared to GOS scores of 4/5 after controlling for age and other covariates. Cluster groupings also did not discriminate mortality or 12 month outcomes in multivariate models. PCA and cluster analysis establish that a subset of CSF inflammatory markers measured in days 0-3 post-TBI may distinguish individuals with poor 6-month outcome, and future studies should prospectively validate these findings. PCA of inflammatory mediators after TBI could aid in prognostication and in identifying patient subgroups for therapeutic interventions.

KEYWORDS:

Cluster analysis; Cytokines; Inflammation; Interleukins; Outcome; Principal component analysis; Prognosis; Rehabilomics; Soluble cell surface markers; TBI

PMID:
26705843
PMCID:
PMC4783208
[Available on 2017-03-01]
DOI:
10.1016/j.bbi.2015.12.008
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Elsevier Science Icon for PubMed Central
Loading ...
Support Center