250: INDIVIDUAL-SPECIFIC PRINCIPAL COMPONENT ANALYSIS PREDICTS EARLY ORGAN DYSFUNCTION IN TRAUMA PATIENTS

Abstract

Introduction: The human systemic inflammatory response following blunt trauma is a multi-dimensional, complex process that manifests in patient-to-patient variability. We sought to elucidate principal drivers of this response using data-driven computational approaches, with the goal of improved diagnosis. Methods: From a cohort of 472 blunt trauma survivors studied following IRB approval, two separate sub-cohorts were studied: Cohort 1 (33 patients [19 males, 14 females; age 44 ± 3; Injury Severity Score {ISS} 24 ± 3]); Cohort 2: (33 patients [19 males, 14 females; age 44.3 ± 1.3; ISS 23.3 ± 1]). Serial blood samples were obtained from all patients (3 samples within the first 24 h and then from days 1 - 5 post-injury). To define the generalizability of this approach, inflammatory mediators in Cohorts 1 and 2 were assayed using two different but overlapping Luminex™ beadsets (Invitrogen and Millipore, respectively). NO2-/NO3- was measured using the nitrate reductase/Griess assay in both groups. Group-time interaction of plasma inflammatory mediators’ levels was determined by 2-way ANOVA, with P < 0.05 being considered statistically significant. Principal Component Analysis (PCA) was used to suggest patient-specific, early drivers of system inflammation in the form of “inflammation barcodes”, followed by hierarchical clustering of PCA-transformed data to define patient sub-groups. Results: PCA/hierarchical clustering segregated the patients into groups that differed significantly in their Marshall Score (MS), a composite measure of organ dysfunction (Cohort 1 MS: Group A: 3.1 ± 0.8 vs. Group B: 1 ± 0.2; Cohort 2 MS: Group A: 4 ± 0.6 vs. Group B: 1.8 ± 0.4, with a higher MS being worse) within the first 24 h post-injury and independently of the specific set of inflammatory mediators analyzed. Patients in Groups A and B in both cohorts remained segregated (significantly higher MS in Group A vs. Group B) for up to 5 d post-injury. However, there was no statistical difference in ICU length of stay (LOS), Total LOS, or days on mechanical ventilator between Groups A and B in both cohorts. Multiple inflammatory mediators were significantly different in Groups A and B in both cohorts. Patient sub-group PCA across Cohorts 1 and 2 suggested that systemic inflammation in blunt trauma patients with a lower degree of organ dysfunction was associated with plasma IL-13 and IL-10, while inflammation in patients with worse organ dysfunction was associated primarily with IL-2. Conclusions: These findings suggest that early systemic inflammatory responses post-trauma, though diverse, are characterized through PCA by defined primary drivers that predispose to different degrees of organ dysfunction.