Transcriptomic mortality signature defines high-risk neonatal sepsis endotype.

Introduction: Neonatal sepsis remains a leading cause of global childhood mortality, yet treatment options are limited. Clinical and biological heterogeneity hinders the development of targeted therapies. Gene-expression profiling offers a potential strategy to identify neonatal sepsis subtypes and guide targeted intervention. Methods: We performed secondary analyses of publicly available gene-expression datasets. Differential gene expression analysis and T-distributed Stochastic Neighbor Embedding (t-SNE) identified biologically relevant patient clusters. Mortality and organ dysfunction were compared across clusters to determine clinical relevance. Results: We identified three endotypes of neonatal sepsis based on the 100 gene expression mortality signature, distinguishing five non-survivors from 72 survivors across datasets. Compared with other endotypes, Endotype A was associated with high mortality (22% vs. 0%, p=0.003) and cardiac dysfunction (61% vs. 31%, p=0.025). Pathobiology among endotype A patients was primarily driven by neutrophil progenitors. Conclusions: Gene-expression profiling can be used to disentangle neonatal sepsis heterogeneity. Dysregulated hyperinflammatory response with emergency granulopoiesis was pathognomonic of high-risk endotype A. Pending further validation, gene-expression-based subclassification may be used to identify at-risk neonates and inform the selection of targeted sepsis therapies.