Interferon-Inducible Mediators Are Associated with SARS-CoV-2 and Related Severity Amongst Critically Ill Patients Suspected of COVID-19

Abstract

RATIONALE: Several studies have identified host immune signatures that are associated with COVID-19. However, it is unclear whether these immune signatures are specific to COVID-19 or are merely reflective of illness severity. In vitro studies have demonstrated that human T-cell responses to SARS-CoV-2-specific antigens are mediated through interferon-gamma (IFN-γ). Methods: We prospectively enrolled a multi-site cohort of patients admitted to the ICU under suspicion for COVID-19 who were then determined to be SARS-CoV-2-positive (n = 82) or-negative (n = 97) by RT-PCR. We measured multiple molecular and cellular immune profiles from blood and endotracheal aspirates (ETAs) collected on ICU admission. Our primary analysis tested for associations between IFN-γ and interferon-inducible mediators (CXCL10 and soluble PD-L1 (sPD-L1)) in blood or ETAs and SARS-CoV-2 status. We then stratified our cohort into SARS-CoV-2-negative and-positive groups and tested for associations between interferon-inducible mediators and clinical outcomes and SARS-CoV-2-copy-number. We used cytometry time-of-flight (CyTOF) to simultaneously measure 39 cell surface and intracellular markers on peripheral blood mononuclear cells collected from a subset of patients with ARDS. We then compared immune cell signatures in subjects with vs. without SARS-CoV-2. Results: The mean APACHE III score was higher in SARS-CoV-2-negative vs.-positive subjects (80±30 vs. 69±29), but the groups were otherwise well-matched. SARS-CoV-2-positive subjects had higher plasma concentrations of IFN-γ, CXCL10, and sPD-L1 relative to SARSCoV-2-negative patients adjusting for age, sex, and severity of illness (all p ≤ 0.01). The levels of IL-6, TNF-α, IL-8, MCP-1, and IL-17A were not significantly different between the two groups. SARS-CoV-2-positive subjects also had higher CXCL10 concentrations in ETAs than SARS-CoV-2-negative subjects. Higher plasma concentrations of CXCL10 and sPD-L1 were associated with higher mortality (Table 1) and more severe respiratory disease (ventilator-free days (VFDs), ARDS) in SARS-CoV-2-positive, but not-negative, patients. In contrast, higher IL-6 was associated with a lower number of VFDs and ARDS in both groups. IFN-γ and CXCL10 (but not IL-6) were associated with SARS-CoV-2-copy-number. Using CyTOF, we found SARS-CoV-2-positive subjects had a lower proportion of CXCR3+ (CXCL10 receptor) T-cells, a higher proportion of PD-L1+ monocytes, and less T-cell and monocyte intracellular cytokine staining vs. SARS-CoV-2-negative patients. Conclusion: Our findings suggest interferon-inducible mediator responses and immune cell hypofunction are characteristic of critically ill subjects with SARS-CoV-2 compared with similar patients without SARS-CoV-2. Our identification of immune signatures that are associated with SARS-CoV-2 infection but are distinct from other forms of critical illness clarifies COVID-19 pathophysiology.