Metabolomic and proteomic dataset of trauma patients in PAMPer Trial

Trauma is a leading cause of death and morbidity worldwide. The lack of a comprehensive characterization of the human response to severe injury limits progress in this field. Here, we present the analysis of a longitudinal, multi-omic dataset comprised of clinical, cytokine, endotheliopathy biomarker, lipidome, metabolome and proteome data from severely injured humans. This includes the validation of seven distinct modules within the metabolome in an external dataset. A “Systemic Storm” pattern with release of 1061 markers together with a pattern suggestive of the “Massive Consumption” of 892 constitutive circulating markers are identified in the hyper-acute phase post-trauma. Data integration reveals two human injury response endotypes, which align with clinical trajectory. Prehospital allogeneic thawed plasma rescues only endotype 2 patients with traumatic brain injury (30d mortality: 30.3% vs. 75.0%, p=0.0015). A machine learning based approach identifies UCHL1 as the most predictive circulating biomarker for patients most likely to benefit from early plasma treatment. These response patterns re-define the paradigm and the datasets provide a resource for the study of critical illness, trauma and human stress responses.