Development of a predictive algorithm for patient survival after traumatic injury using a five analyte blood panel

Abstract

ABSTRACTSevere trauma can induce systemic inflammation but also immunosuppression, which makes understanding the immune response of trauma patients critical for therapeutic development and treatment approaches. By evaluating the levels of 59 proteins in the plasma of 50 healthy volunteers and 1000 trauma patients across five trauma centers in the United States, we identified 6 novel changes in immune proteins after traumatic injury and further new variations by sex, age, trauma type, comorbidities, and developed a new equation for prediction of patient survival. Blood was collected at the time of arrival at Level 1 trauma centers and patients were stratified based on trauma level, tissues injured, and injury types. Trauma patients had significantly upregulated proteins associated with immune activation (IL-23, MIP-5), immunosuppression (IL-10) and pleiotropic cytokines (IL-29, IL-6). A high ratio of IL-29 to IL-10 was identified as a new predictor of survival in less severe patients with ROC area of 0.933. Combining machine learning with statistical modeling we developed an equation (“VIPER”) that could predict survival with ROC 0.966 in less severe patients and 0.8873 for all patients from a five analyte panel (IL-6, VEGF-A, IL-21, IL-29, and IL-10). Furthermore, we also identified three increased proteins (MIF, TRAIL, IL-29) and three decreased proteins (IL-7, TPO, IL-8) that were the most important in distinguishing a trauma blood profile. Biologic sex altered phenotype with IL-8 and MIF being lower in healthy women, but higher in female trauma patients when compared to male counterparts. This work identifies new responses to injury that may influence systemic immune dysfunction, serving as targets for therapeutics and immediate clinical benefit in identifying at-risk patients.