Poster #30: Apoorva Babu MS

Integrated Longitudinal Multi-omic profiling identifies latent sources of biological variation in pediatric MODS using MOFA2

Apoorva Babu, M.S., Department of Biomedical and Health Informatics, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Sydney A. Sheetz, B.S., Division of Allergy and Immunology, Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Teresa Arroyo, RN, BSN, Division of Critical Care Medicine, Department of Anesthesia and Critical Care, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Steven D. Ham, M.D., M.P.H., Division of Critical Care Medicine, Department of Anesthesia and Critical Care, Children's Hospital of Philadelphia and the Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; Jocelyn E. Chau, B.S., Division of Critical Care Medicine, Department of Anesthesia and Critical Care, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Peyton Conrey, B.S., Division of Allergy and Immunology, Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Samir Sayed, M.S., Division of Allergy and Immunology, Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA, USA; Sarah E. Henrickson, M.D., Ph.D., Division of Allergy and Immunology, Department of Pediatrics, Children's Hospital of Philadelphia and the Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; Robert B. Lindell, M.D., Division of Critical Care Medicine, Department of Anesthesia and Critical Care, Children’s Hospital of Philadelphia and the Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

Sepsis is the leading cause of death among hospitalized children worldwide and is the most common cause of multiple organ dysfunction syndrome (MODS). Pediatric MODS shows heterogenous recovery trajectories that are unexplained by a single-omic or clinical severity measures alone. We hypothesized that integrated longitudinal multi-omic profiling would identify latent immune factors associated with organ dysfunction and recovery which is translatable to a deployable plasma protein classifier. We profiled 149 pediatric MODS patients across up to six timepoints using spectral flow cytometry (Omiq-preprocessed) and plasma proteomics (817 proteins, Nomic). Cell-type composition, FlowSOM metaclusters, and proteomics were integrated via MOFA2 with a continuous MODS-day covariate, yielding 12 latent factors. Factors 1–4, capturing the majority of variance, were annotated by top-loading proteins/views and compared using landmark survival status, recovery trajectory (rapid vs. persistent MODS), and used to benchmark a LASSO model predicting organ-dysfunction-free days from proteomics. Factors were annotated as Immune-endothelial interaction (F1), immunoparalysis reversal (F2), Adaptive immune activation (F3), and Vascular remodeling vs Inflammatory activation (F4). At the day-3 landmark, high F3/F4 and low F1 were associated with shorter time to recovery and death, respectively, and F3 trajectory distinguished rapid recovery from persistent MODS. These patterns were broadly consistent with previously described pediatric MODS endotypes (A/B/C based on increasing severity), with high F3/F4 aligning with the least-severe endotype. A LASSO model predicting organ-dysfunction-free days from 817 proteins identified 13 candidate proteins, predominantly mapping to Factors 1 and 4. This Multi-omic profiling reveals stable immune endotypes at pediatric MODS onset that stratify survival and recovery trajectory, consistent with established severity endotypes. A small proteomic panel may offer a path toward an assayable classifier of recovery, pending validation in an independent cohort.