A26-10 Cytokine-Based Strategies to Improve Prognostic Enrichment of Pediatric Critical Illness Syndromes

Abstract Introduction Hyperinflammation is common to critical care syndromes, including sepsis and acute respiratory distress syndrome (ARDS), and is associated with worse outcomes. A parsimonious signature of hyperinflammatory ARDS has been described using IL-6, TNFR1, and bicarbonate, with prognostic utility in pediatric ARDS. We aimed to determine whether the addition of CXCL9 and IL-18, which are elevated in hyperinflammatory conditions such as hemophagocytic lymphohistiocytosis (HLH) and macrophage activation syndrome (MAS), could improve on the prognostic model. Methods We leveraged a prospective multicenter cohort study of 500 children with ARDS (Linking Endotypes and Outcomes in Pediatric Acute Respiratory Distress Syndrome LEOPARDS; NCT04113434) study. Cytokines were measured using microfluidic immunoassay (Ella™). In preclinical studies, mouse recombinant cytokines were injected retro-orbitally into C57BL/6J mice, and serial cytokine measurements obtained were obtained by tail bleeds. Mouse cytokines were measured by Ella or singleplex ELISA. Results The traditional hyper-/hypo-inflammatory phenotype model demonstrated modest utility for 90-day mortality (AUC 0.62, sensitivity 26%, positive predictive value PPV 24%) in LEOPARDS (14% hyperinflammatory; 9% mortality). In children with hypo-inflammatory ARDS, stratification by median IL-18 levels showed differential mortality (log-rank p = 0.05). A four-term model (IL-6, TNFR1, CXCL9, IL-18) had AUC 0.69 (sensitivity 56%. PPV 23%). Preclinical mouse models of cytokine-induced inflammation demonstrated that IL-6 levels were primarily a marker of synergistic IL-1 and TNF signaling activity. Conclusions Cytokine-based prognostic enrichment shows promise for children with critical illness syndromes, including ARDS and sepsis. Future studies should focus on identifying whether cytokine-based enrichment can identify specific pathways to inform immunomodulatory interventions. This abstract is funded by: NIH NHLBI (R01HL148054)