A machine learning approach identifies unresolving secondary pneumonia as a contributor to mortality in patients with severe pneumonia, including COVID-19 (preprint)

Background: Patients with severe SARS-CoV-2 pneumonia experience longer durations of critical illness yet similar mortality rates compared to patients with severe pneumonia secondary to other etiologies. As secondary bacterial infection is common in SARS-CoV-2 pneumonia, we hypothesized that unresolving ventilator-associated pneumonia (VAP) drives the apparent disconnect between length-of-stay and mortality rate among these patients. Methods: We analyzed VAP in a prospective single-center observational study of 585 mechanically ventilated patients with suspected pneumonia, including 190 patients with severe SARS-CoV-2 pneumonia. We developed CarpeDiem, a novel machine learning approach based on the practice of daily ICU team rounds to identify clinical states for each of the 12,495 ICU patient-days in the cohort. We used the CarpeDiem approach to evaluate the effect of VAP and its resolution on clinical trajectories. Findings: Patients underwent a median [IQR] of 4 [2,7] transitions between 14 clinical states during their ICU stays. Clinical states were associated with differential hospital mortality. The long length-of-stay among patients with severe SARS-CoV-2 pneumonia was associated with prolonged stays in clinical states defined by severe respiratory failure and with a lower frequency of transitions between clinical states. In all patients, including those with COVID-19, unresolving VAP episodes were associated with transitions to unfavorable states and hospital mortality. Interpretation: CarpeDiem offers a machine learning approach to examine the effect of VAP on clinical outcomes. Our findings suggest an underappreciated contribution of unresolving secondary bacterial pneumonia to outcomes in mechanically ventilated patients with pneumonia, including due to SARS-CoV-2.

The local and systemic response to SARS-CoV-2 infection in children and adults (preprint)

While a substantial proportion of adults infected with SARS-CoV-2 progress to develop severe disease, children rarely manifest respiratory complications. Therefore, understanding differences in the local and systemic response to SARS-CoV-2 infection between children and adults may provide important clues about the pathogenesis of SARS-CoV-2 infection. To address this, we first generated a healthy reference multi-omics single cell data set from children (n=30) in whom we have profiled triple matched samples: nasal and tracheal brushings and PBMCs, where we track the developmental changes for 42 airway and 31 blood cell populations from infancy, through childhood to adolescence. This has revealed the presence of naive B and T lymphocytes in neonates and infants with a unique gene expression signature bearing hallmarks of innate immunity. We then contrast the healthy reference with equivalent data from severe paediatric and adult COVID-19 patients (total n=27), from the same three types of samples: upper and lower airways and blood. We found striking differences: children with COVID-19 as opposed to adults had a higher proportion of innate lymphoid and non-clonally expanded naive T cells in peripheral blood, and a limited interferon-response signature. In the airway epithelium, we found the highest viral load in goblet and ciliated cells and describe a novel inflammatory epithelial cell population. These cells represent a transitional regenerative state between secretory and ciliated cells; they were found in healthy children and were enriched in pediatric and adult COVID-19 patients. Epithelial cells display an antiviral and neutrophil-recruiting gene signature that is weaker in severe paediatric versus adult COVID-19. Our matched blood and airway samples allowed us to study the spatial dynamics of infection. Lastly, we provide a user-friendly interface for this data as a highly granular reference for the study of immune responses in airways and blood in children.