The hyperacute plasma proteome reports injury severity and defines distinct trauma endotypes

Abstract The host response to traumatic injury is varied and unpredictable. Patients with subjectively similar injuries progress along divergent clinical paths, from uncomplicated recovery to extended hospitalisation, multiple organ dysfunction, and life-long ill health. The temporal certainty of trauma nonetheless provides immediate opportunity to predict clinical trajectory and intervene therapeutically. We demonstrate using machine-learning that the hyperacute plasma proteome can function as a physiology-driven indicator of injury severity. Moreover, it identifies two serious-injury endotypes that differentially predict unfavourable clinical trajectories. Notably, a prominent neuronal guidance protein signature, that reports vascular dysfunction and immune activation, together with an anti-coagulation/pro-fibrinolytic state, identify those patients who progress to adverse clinical outcomes. Together, these mechanistic insights into immediate host responses to serious trauma reveal tractable targets for future therapeutic interventions.