Pulmonary microbiome and transcriptome signatures reveal distinct pathobiologic states associated with mortality in two cohorts of pediatric stem cell transplant patients
Matt S Zinter,Christopher C Dvorak,Madeline Y Mayday,Gustavo Reyes,Miriam R Simon,Emma M Pearce,Hanna Kim,Peter J Shaw,Courtney M Rowan,Jeffrey J Auletta,Paul L Martin,Kamar Godder,Christine N Duncan,Nahal R Lalefar,Erin M Kreml,Janet R Hume,Hisham Abdel-Azim,Caitlin Hurley,Geoffrey D E Cuvelier,Amy K Keating,Muna Qayed,James S Killinger,Julie C Fitzgerald,Rabi Hanna,Kris M Mahadeo,Troy C Quigg,Prakash Satwani,Paul Castillo,Shira J Gertz,Theodore B Moore,Benjamin Hanisch,Aly Abdel-Mageed,Rachel Phelan,Dereck B Davis,Michelle P Hudspeth,Greg A Yanik,Michael A Pulsipher,Imran Sulaiman,Leopoldo N Segal,Birgitta A Versluys,Caroline A Lindemans,Jaap J Boelens,Joseph L DeRisi,Pediatric Transplantation and Cell Therapy Consortium
DOI: https://doi.org/10.1101/2023.11.29.23299130
2023-11-29
MedRxiv
Abstract:Lung injury is a major determinant of survival after pediatric hematopoietic cell transplantation (HCT). A deeper understanding of the relationship between pulmonary microbes, immunity, and the lung epithelium is needed to improve outcomes. In this multicenter study, we collected 278 bronchoalveolar lavage (BAL) samples from 229 patients treated at 32 children's hospitals between 2014-2022. Using paired metatranscriptomes and human gene expression data, we identified 4 patient clusters with varying BAL composition. Among those requiring respiratory support prior to sampling, in-hospital mortality varied from 22-60% depending on the cluster (p=0.007). The most common patient subtype, Cluster 1, showed a moderate quantity and high diversity of commensal microbes with robust metabolic activity, low rates of infection, gene expression indicating alveolar macrophage predominance, and low mortality. The second most common cluster showed a very high burden of airway microbes, gene expression enriched for neutrophil signaling, frequent bacterial infections, and moderate mortality. Cluster 3 showed significant depletion of commensal microbes, a loss of biodiversity, gene expression indicative of fibroproliferative pathways, increased viral and fungal pathogens, and high mortality. Finally, Cluster 4 showed profound microbiome depletion with enrichment of Staphylococci and viruses, gene expression driven by lymphocyte activation and cellular injury, and the highest mortality. BAL clusters were modeled with a random forest classifier and reproduced in a geographically distinct validation cohort of 57 patients from The Netherlands, recapitulating similar cluster-based mortality differences (p=0.022). Degree of antibiotic exposure was strongly associated with depletion of BAL microbes and enrichment of fungi. Potential pathogens were parsed from all detected microbes by analyzing each BAL microbe relative to the overall microbiome composition, which yielded increased sensitivity for numerous previously occult pathogens. These findings support personalized interpretation of the pulmonary microenvironment in pediatric HCT, which may facilitate biology-targeted interventions to improve outcomes.