ABSTRACT
RATIONALE: Some biomarkers of host response to viral infection are associated with COVID-19 outcomes, but these biomarkers do not directly measure viral burden. The association between plasma viral antigen levels and clinical outcomes has not been previously studied. Our aim was to investigate the relationship between plasma SARS-CoV-2 viral antigen concentration and proximal clinical deterioration in hospitalized patients. METHODS: SARS-CoV-2 nucleocapsid antigen concentrations were measured using a validated microbead immunoassay (Quanterix, NIH/NIAID laboratory) in plasma collected at enrollment from 256 subjects in a prospective observational cohort of hospitalized patients with COVID-19 from 3 hospitals, admitted between March 2020 and August 2021. Relationships between viral antigen concentration and clinical status at 1 week as measured by the World Health Organization (WHO) ordinal scale as well as ICU admission were assessed. Models were adjusted for age and sex, baseline comorbidities including immunosuppression, endogenous neutralizing antibodies, baseline COVID-19 severity, smoking status, remdesivir therapy, steroid therapy, and vaccine status. Missing covariate data were imputed using multiple imputation by chained equations. RESULTS: The median viral antigen concentration for the 35 subjects who deteriorated by 1 week was 4507 (IQR 1225-9665) pg/mL compared to 494 (IQR 18-3882) pg/mL in the 212 subjects who did not (p = 0.0004 Figure a). Using ordinal regression, each doubling in viral antigen concentration was significantly associated with a worse WHO ordinal scale at 1 week (unadjusted OR 1.07, 95% CI 1.02-1.13;adjusted OR 1.10, 95% CI 1.02-1.18). Among 168 patients not in the ICU at baseline, the median viral antigen concentration for the 40 patients who progressed to the ICU was 4697 (IQR 482- 10410) pg/mL vs. 459 (IQR 15-3062) pg/mL in the 128 patients who did not progress to require ICU care (p = 0.0001 Figure b). Using logistic regression, each doubling in viral antigen concentration was significantly associated with ICU admission (unadjusted OR 1.18, 95% CI 1.06-1.32, adjusted OR 1.40, 95% CI 1.11-1.76). CONCLUSIONS: Higher plasma viral antigen concentration at hospital admission is independently associated with a significantly worse clinical status at 1 week and a higher odds of ICU admission among hospitalized patients with COVID-19. This novel finding indicates that plasma viral antigen concentration may identify hospitalized COVID-19 patients at highest risk of short-term clinical deterioration in both clinical practice and research. Results of plasma antigen tests are available within 2-3 hours and could be integrated for identifying hospitalized COVID-19 patients who might benefit from early intervention.
ABSTRACT
Background: Latent class analyses in patients with acute respiratory distress syndrome (ARDS) have identified “hyper-inflammatory” and “hypo-inflammatory” phenotypes with divergent clinical outcomes and treatment responses. ARDS phenotypes are defined using plasma biomarkers and clinical variables. It is currently unknown if these phenotypes have distinct pulmonary biology and if pre-clinical models of disease replicate the biology of either phenotype. Methods: 45 subjects with ARDS (Berlin Definition) and 5 mechanically ventilated controls were selected from cohorts of mechanically ventilated patients at UCSF and ZSFG. Patients with COVID-19 were excluded from this analysis. A 3-variable classifier model (plasma IL-8, protein C, and bicarbonate;Sinha 2020) was used to assign ARDS phenotypes. Tracheal aspirate (TA) RNA was analyzed using established bulk and single-cell sequencing pipelines (Langelier 2018, Sarma 2021). Differentially expressed (DE) genes were analyzed using Ingenuity Pathway Analysis (IPA). Microbial community composition was analyzed with vegan. Fgsea was used to test for enrichment of gene sets from experimental ARDS models in genes that were differentially expressed between each phenotype and mechanically ventilated controls. Results: Bulk RNA sequencing (RNAseq) was available from 29 subjects with hypoinflammatory ARDS and 10 subjects with hyperinflammatory ARDS. 2,777 genes were differentially expressed between ARDS phenotypes. IPA identified several candidate upstream regulators of gene expression in hyperinflammatory ARDS including IL6, TNF, IL17C, and interferons (Figure 1A). 2,953 genes were differentially expressed between hyperinflammatory ARDS and 5 ventilated controls;in contrast, only 243 genes were differentially expressed between hypoinflammatory ARDS and controls, suggesting gene expression in the hypoinflammatory phenotype was more heterogeneous. Gene sets from experimental models of acute lung injury were enriched in hyperinflammatory ARDS but not in hypoinflammatory ARDS (Figure 1B). Single cell RNA sequencing (scRNAseq) was available from 6 additional subjects with ARDS, of whom 3 had hyperinflammatory ARDS. 14,843 cells passed quality control filters. Hyperinflammatory ARDS subjects had a markedly higher burden of neutrophils (Figure 1C), including a cluster of stressed neutrophils expressing heat shock protein RNA that was not present in hypoinflammatory ARDS. Expression of a Th1 signature was higher in T cells from hyperinflammatory ARDS. Differential expression analysis in macrophages identified increased expression of genes associated with mortality in a previous study of ARDS patients (Morell 2019). Conclusions: The respiratory tract biology of ARDS phenotypes is distinct. Hyperinflammatory ARDS is characterized by neutrophilic inflammation with distinct immune cell polarization. Transcriptomic profiling identifies candidate preclinical disease models that replicate gene expression observed in hyperinflammatory ARDS.
ABSTRACT
Background: The coronavirus disease 2019 (COVID-19) pandemic has led to a rapid increase in the incidence of acute respiratory distress syndrome (ARDS). The distinct features of pulmonary biology in COVID-19 ARDS compared to other causes of ARDS, including other lower respiratory tract infections (LRTIs), are not well understood. Methods: Tracheal aspirates (TA) and plasma were collected within five days of intubation from mechanically ventilated adults admitted to one of two academic medical centers. ARDS and LRTI diagnoses and were verified by study physicians. Subjects were excluded if they received immunosuppression. TA from subjects with COVID-ARDS was compared to gene expression in TA from subjects with other causes of ARDS (OtherARDS) or mechanically ventilated control subjects without evidence of pulmonary pathology (NoARDS). Plasma concentrations of IL-6, IL-8, and protein C also were compared between these groups. Upstream regulator and pathway analysis was performed on significantly differentially expressed genes with Ingenuity Pathway Analysis (IPA). Subgroup analyses were performed to compare gene expression in COVID to ARDS associated with other viral LRTIs and bacterial LRTIs. The association of interferon-stimulated gene expression with SARS-CoV2 viral load was compared to the same association in nasopharyngeal swabs in a cohort of subjects with mild SARS-CoV2. Results: TA sequencing was available from 15 subjects with COVID, 32 subjects with other causes of ARDS (OtherARDS), and 5 mechanically ventilated subjects without evidence of pulmonary pathology (NoARDS). 696 genes were differentially expressed between COVID and OtherARDS (Figure 1A). IL-6, IL-8, B-cell receptor, and hypoxia inducible factor-1a signaling were attenuated in COVID compared to OtherARDS. Peroxisome proliferator-activated receptor (PPAR) and PTEN signaling were higher in COVID compared to OtherARDS (Figure 1B). Plasma levels of IL-6, IL-8, and protein C were not significantly different between COVID and OtherARDS. In subgroup analyses, IL-8 signaling was higher in COVID compared to viral LRTI, but lower than bacterial LRTI. Type I/III interferon was higher in COVID compared to bacterial ARDS, but lower compared to viral ARDS (Figure 1C). Compared to nasopharyngeal swabs from subjects with mild COVID-19, expression of several interferon stimulated genes was less strongly correlated with SARS-CoV2 viral load in TA (Figure 1D). IPA identified several candidate medications to treat COVID-19, including dexamethasone, G-CSF, and etanercept. Conclusions: TA sequencing identifies unique features of the host response in COVID-19. These differentially expressed pathways may represent potential therapeutic targets. An impaired interferon response in the lung may increase susceptibility to severe SARS-COV2.