Shock and Acute Conditions OutcOmes Platform

Circulatory Shock, Sepsis, Cardiogenic Shock, Major Trauma, Surgical Shock

shock, heterogeneity of treatment effect, biomarkers, machine learning

In-hospital mortality of patients admitted in the intensive care unit (ICU) for circulatory shock remains high (between 20 and 40%). Currently, there are no markers that allow us to classify patients with circulatory shock at higher risk of early and late bad outcomes, or who may better respond to a specific intervention. To understand the contribution of biological heterogeneity to circulatory shock independently from its etiology, the ShockCO-OP Research Program aims to use clustering approaches to re-analyze existing clinical and molecular data from several large European and North American prospective cohorts and clinical trials. This will enable an improvement in risk prediction and a better patient selection in future clinical trials to assess a personalized therapy (i.e., prospective enrollment based on a biological/molecular signature).

Traditionally, circulatory shock subgroups are defined according to hemodynamic profile (e.g., hypovolemic, distributive, cardiogenic) and etiology (e.g., trauma, infection, myocardial infarction among others) and are incorrectly considered as homogeneous clinical syndromes. Emerging translational evidence highlights the existing molecular heterogeneity in the circulatory shock syndrome. Such findings raise a major issue in assessing neutral clinical trial results in circulatory shock as a given intervention effect (e.g., fluid management, vasopressors/inotropes, mechanical circulatory support) may preferentially impact different subgroups (i.e., heterogeneity of treatment effect). Accordingly, identifying distinct biological subphenotypes with different mechanistic signatures may provide new insights regarding the pathophysiology of circulatory shock. This may allow predictive enrichment (i.e., identifying those patients most likely to benefit from a particular therapy) and biomarker-driven or phenotype-driven patient selection in future clinical trials to assess a personalized therapy (i.e., prospective enrollment based on a biological signature). The ShockCO-OP Research Program aims to use unsupervised model-based clustering (i.e., regardless of outcome) to reanalyze existing clinical and biological data in several European and North American prospective cohorts and clinical trials to identify distinct biomarker-driven subphenotypes in circulatory shock syndromes, their underlying molecular signatures (proteomics, transcriptomics), their association with outcome and their response to different interventions.

Canada