Inflammatory Endotypes of Sepsis: Preclinical Models and Clinical Investigation

Introduction: Although cytokine-specific biologics are available for immune modulation of systemic inflammatory conditions, their clinical utility is limited by the inability to rapidly discriminate various inflammatory processes. Recently, preclinical models have demonstrated that the clinical phenotype of sepsis can be induced in mice by pair-wise administration of tumor necrosis factor (TNF), plus either interleukin-1 beta (IL-1), IL-18, or interferon-gamma (IFN-). Unfortunately, direct measurement of these inflammatory cytokines has not proven useful. Instead, we hypothesized that these cytokine-induced sepsis models could be used to inform other biomarkers for discriminating distinct cytokine storm endotypes, and that these endotypes are present in children with hyper-inflammatory syndromes.


Methods: Cytokines were injected retro-orbitally into C57BL/6J mice, and serial cytokine measurements were obtained from tail bleeds. Mouse cytokines were measured by enzyme-linked immunosorbent assays (ELISA) or automated microfluidic immunoassays (Ella, Bio-Techne). Children admitted to Penn State Health Children’s Hospital with presumed hyper-inflammatory syndromes were enrolled and cytokines were measured by Ella.


Results: The pairwise administration of IL-1/TNF or IFN-/TNF was near universally fatal in mice and biomarker evaluation demonstrated that TNFR1 was the best discriminator of lethal inflammation. CXCL9 was induced by IL-18/TNF and IFN-/TNF, whereas IL-6 was associated with IL-1 signaling. The measurement of four cytokines (CXCL9, IL-6, IL-18 and TNFR1) was used to assess the presence of cytokine storm in children with various inflammatory conditions. Hierarchical clustering was used to divide subjects into two groups, presence or absence of cytokine storm. Cytokine storm was present in 11 of 28 (39%), and these children demonstrated a 28-day mortality of 36% compared to 0% without cytokine storm.


Conclusions: The presence of hyper-inflammation is associated with poor outcomes in children. This four- biomarker panel may provide insight into the presence of cytokine storm, and ultimately provide prognostic and predictive enrichment for future studies.