(750) Early Changes in Gene Expression Predict Severity of Critical Illness in Children

Introduction: Children admitted to pediatric critical care have a spectrum of diseases, including sepsis, acute respiratory distress syndrome (ARDS), trauma and severe respiratory and neurological disease. Gene expression analysis in children with sepsis have characterised clinical phenotypes associated with trends in gene expression (subgroups/endotypes), We hypothesised that early gene expression profiles may be shared across the spectrum of children requiring critical care and may represent treatable immune traits regardless of etiology of disease.

Methods: We conducted global gene expression analysis using microarrays (Affymetrix, UK) on whole blood samples from 380 children during emergency retrieval to 4 UK pediatric intensive care units (PICU) within a prospective cohort. Differential gene expression analysis was performed and data were subject to K-means clustering to derive patient subgroups. The clinical relevance of the subgroups was evaluated against the length of PICU stay (PICU-free days at day 30). Duration of ventilation and vasoactive support were secondary outcomes.

Results: In this cohort, 96% of patients were ventilated and 48% of patients received vasoactive support. Three subgroups were initially identified by cluster analysis. Due to limited clinical differences between subgroups 2 and 3, these groups were combined. Differential gene expression analysis identified 461 genes (log-fold change>1 and adjusted P< 0.05). Differences in the expression of immune response gene included B-cell and T-cell function in addition to innate response. Subgroup 1 membership was associated with poorer outcomes with fewer median PICU-free days (20 vs 24, P=0.045), greater median duration of ventilation (168 vs 83 hours, P=0.002) and greater median duration of vasoactive support (52 vs 0 hours, P=0.013). These differences were independent of age, gender and infection status and reason for admission.

Conclusions: This study investigated gene expression profiles of a large cohort of critically ill children. The results suggest that agnostic of disease type, infectious or non-infectious status, early gene expression profiles may indicate disease severity. Further, focus on the trajectory of gene expression profiles could enhance biological understanding of disease processes and provide prognostic insight.