Objective: We aim to identify the clinical phenotypes of immunocompromised patients with pneumonia-related ARDS, to investigate the lung microbiota signatures and the outcomes of different phenotypes, and finally, to develop a machine learning classifier for a specified phenotype.
Methods: This prospective study included immunocompromised patients with pneumonia-related ARDS. We identified phenotypes using hierarchical clustering to analyze clinical variables and serum cytokine levels. We then compared outcomes and lung microbiota signatures between phenotypes. Based on lung microbiota markers, we developed a random forest classifier for a specified phenotype with worse outcomes.
Results: This study included 92 patients, who were divided into three phenotypes, namely “type α” (N = 33), “type β” (N = 12), and “type γ” (N = 47). Compared to type α or type β, patients with type γ had no obvious inflammatory presentation and had significantly lower IL-6 levels and more severe oxygenation failure. Type γ was also related to higher 30-day mortality and lower ventilator free days. The microbiota signatures of type γ were characterized by lower alpha diversity and distinct compositions than those of other patients. We developed a lung microbiota-derived random forest model to differentiate patients with type γ from other phenotypes.
Conclusion: Immunocompromised patients with pneumonia-related ARDS can be clustered into three clinical phenotypes, namely type α, type β, and type γ. Phenotypes were distinguished from each other with different outcomes and lung microbiota signatures. Type γ, which was characterized by insufficient inflammation response and worse outcomes, can be detected with a random forest model based on lung microbiota markers.

Keywords: immunocompromised host, respiratory failure, microbiota