Skip to main content
Download PDF

Novel subtypes of severe COVID-19 respiratory failure based on biological heterogeneity: a secondary analysis of a randomized controlled trial

Critical Care volume 28, Article number: 56 (2024) Cite this article

Abstract

Background

Despite evidence associating inflammatory biomarkers with worse outcomes in hospitalized adults with COVID-19, trials of immunomodulatory therapies have met with mixed results, likely due in part to biological heterogeneity of participants. Latent class analysis (LCA) of clinical and protein biomarker data has identified two subtypes of non-COVID acute respiratory distress syndrome (ARDS) with different clinical outcomes and treatment responses. We studied biological heterogeneity and clinical outcomes in a multi-institutional platform randomized controlled trial of adults with severe COVID-19 hypoxemic respiratory failure (I-SPY COVID).

Methods

Clinical and plasma protein biomarker data were analyzed from 400 trial participants enrolled from September 2020 until October 2021 with severe COVID-19 requiring ≥ 6 L/min supplemental oxygen. Seventeen hypothesis-directed protein biomarkers were measured at enrollment using multiplex Luminex panels or single analyte enzyme linked immunoassay methods (ELISA). Biomarkers and clinical variables were used to test for latent subtypes and longitudinal biomarker changes by subtype were explored. A validated parsimonious model using interleukin-8, bicarbonate, and protein C was used for comparison with non-COVID hyper- and hypo-inflammatory ARDS subtypes.

Results

Average participant age was 60 ± 14 years; 67% were male, and 28-day mortality was 25%. At trial enrollment, 85% of participants required high flow oxygen or non-invasive ventilation, and 97% were receiving dexamethasone. Several biomarkers of inflammation (IL-6, IL-8, IL-10, sTNFR-1, TREM-1), epithelial injury (sRAGE), and endothelial injury (Ang-1, thrombomodulin) were associated with 28- and 60-day mortality. Two latent subtypes were identified. Subtype 2 (27% of participants) was characterized by persistent derangements in biomarkers of inflammation, endothelial and epithelial injury, and disordered coagulation and had twice the mortality rate compared with Subtype 1. Only one person was classified as hyper-inflammatory using the previously validated non-COVID ARDS model.

Conclusions

We discovered evidence of two novel biological subtypes of severe COVID-19 with significantly different clinical outcomes. These subtypes differed from previously established hyper- and hypo-inflammatory non-COVID subtypes of ARDS. Biological heterogeneity may explain inconsistent findings from trials of hospitalized patients with COVID-19 and guide treatment approaches.

Background

Coronavirus disease 2019 (COVID-19) is clinically heterogeneous, ranging from asymptomatic disease to protracted critical illness with a high fatality rate [1]. Variability in the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the environment, and the host all contribute to this heterogeneity. Host-related clinical factors associated with poor outcomes include male sex, obesity, older age, and the presence of comorbidities such as diabetes mellitus and hypertension [2, 3]. Studies on the host immunologic response to SARS-CoV-2 have found associations between circulating levels of interleukin- (IL-) 6, tumor necrosis factor (TNF), interferon- (IFN-) α, IL-10 and other cytokines with outcomes in those with moderate to severe disease [4]. Trials of immunomodulators in this population, however, have met with mixed results, likely in part due to complex host–pathogen interactions. For instance, the RECOVERY trial found that dexamethasone led to a reduction in 28-day mortality in severe COVID-19 [5]. However, four other trials reached different conclusions [6,7,8,9]. Similar findings have been reported with IL-6 inhibitors [10,11,12,13,14,15] and the antiviral therapy, remdesivir [16,17,18,19]. These disparate findings suggest that trial participants may be biologically heterogeneous, leading to studies that are underpowered to detect meaningful treatment effects.

Latent class analysis (LCA) of clinical and protein biomarker data has identified two subtypes of non-COVID acute respiratory distress syndrome (ARDS), termed “hyper-inflammatory” and “hypo-inflammatory,” the former characterized by elevated plasma levels of inflammatory biomarkers such as IL-6 and soluble TNF receptor 1 (sTNFR-1) and lower levels of bicarbonate and protein C [20]. These subtypes have significantly different mortality in secondary analyses of numerous trials and observational studies [20,21,22,23,24] Moreover, the subtypes demonstrate different responses to therapies such as PEEP, fluid strategy, and simvastatin in secondary analyses of randomized trials [20, 21, 24]. A similar approach to subtyping patients with COVID-19 ARDS early in the pandemic revealed significant overlap with non-COVID ARDS subtypes, but with the exception of IL-6, data on biomarkers of lung injury and inflammation were unavailable [25]. Additionally, corticosteroids were associated with reduced 90-day mortality in one subtype, offering a plausible explanation for the mixed results from corticosteroid trials in severe COVID-19 [26]. An observational study similarly found two distinct subtypes of COVID-19 ARDS using clinical data from a single center, early in the pandemic. No data on biomarkers of lung injury or inflammation were collected [27]. A recent retrospective cohort study by Verhoef et al. used latent profile analysis of clinical and extensive protein biomarker data from a cohort of patients with COVID-19 who presented to acute care facilities and identified two novel subtypes with heterogeneous treatment response to corticosteroids [28]. Due to the observational nature of the recruited cohort, the study was limited by heterogeneous practices around steroid or other immunomodulator drug administration, timing of SARS-CoV-2 infection, and a mix of patients from the outpatient and inpatient settings, limiting the conclusions that can be drawn with regards to the biology of patients with severe COVID-19.

In this study, the primary objective was to apply LCA using clinical and protein biomarker data to test for evidence of latent subtypes in the I-SPY COVID randomized controlled trial of investigational agents for the treatment of severe COVID-19. [29]

Methods

Study design

Data were obtained from the I-SPY COVID Trial, a multi-institutional phase 2 platform randomized controlled open-label trial to evaluate pharmacotherapies for severe COVID-19 (NCT04488081). The trial protocol and results for the first seven agents, none of which met predefined criteria for benefit, have been reported [29, 30]. In brief, newly hospitalized adults with SARS-CoV-2 requiring ≥ 6 L/min supplemental oxygen were randomized to either a control arm receiving backbone therapy (Remdesivir and dexamethasone) or an investigational arm receiving backbone therapy plus a study drug.

Assay procedures

Plasma was collected on days 1 (baseline, prior to administration of any active study drug), 3, and 7 following trial enrollment. Seventeen candidate protein biomarkers of inflammation, disordered coagulation, endothelial, and epithelial lung injury were selected a priori based on evidence from non-COVID ARDS and early evidence from COVID-19 studies. Biomarkers were measured using customized and validated multiplex Luminex panels or in duplicate using single analyte enzyme linked immunoassay methods (ELISA) (Additional file 1: methods). SARS-CoV-2 nucleocapsid antigen levels were measured using a single-molecule immune bead assay (Quanterix, Billerica, MA, USA).

Statistical analysis

The primary outcome for these analyses was 28-day mortality. Secondary outcomes included 60-day mortality, time to death, and time to recovery. The association of each biomarker with mortality was tested using unadjusted logistic regression models and models adjusting for age, BMI, and level of respiratory support required at study enrollment. P-values were adjusted for multiple comparisons using False Discovery Rate (FDR) cutoff of 0.05. The Fine-Gray model for cumulative incidence function was used to estimate a subdistribution hazard ratio (SHR) for the association of each biomarker with time to death and time to recovery, as recovery and death were competing events in the trial (Additional file 1: methods).

A set of clinical and protein biomarker data were selected a priori to serve as subtype defining variables for LCA. Modeling was performed agnostic of clinical outcomes using standard methods [31]. A validated parsimonious model using IL-8, bicarbonate, and protein C was used for comparison with established hyper- and hypo-inflammatory LCA subtypes of non-COVID ARDS [32]. Using longitudinal biomarker measurements in the control arm, we fit linear mixed effects models to determine whether subtype assignment was associated with differences in biomarker trajectory over time.

Analyses were conducted in R 4.2.2, STATA 17.0, and MPlus 8.8.

Results

Cohort characteristics

Of 868 patients randomized in the I-SPY COVID trial from September 18th, 2020, until October 6th, 2021, 597 had plasma collected at baseline (Additional file 2: Figure S1). The first 400 were included in the analyses to provide adequate power for LCA analyses [33, 34]. Average age was 60 ± 14 years; 67% were male, and 28-day mortality was 25% (Table 1). At trial enrollment, 26% required positive pressure ventilation, and 97% were receiving dexamethasone. Patients excluded from these analyses had similar characteristics to those included (Additional file 2: Table S1).

Table 1 Characteristics of the study population and outcomes by assigned I-SPY COVID trial arm
Full size table

Baseline protein biomarkers and associations with outcomes

Baseline concentrations of several biomarkers were significantly associated with 28-day mortality. Specifically, increased concentrations of SARS-CoV-2 antigen, biomarkers of inflammation (IL-6, IL-8, IL-10, triggering receptor expressed on myeloid cells [TREM-1], sTNFR-1, interferon-gamma induced protein 10 [IP-10]), alveolar epithelial injury (soluble receptor for advanced glycation end products [sRAGE], surfactant protein D [SP-D]) and endothelial injury (thrombomodulin, intercellular adhesion molecule 1 [ICAM-1], and the ratio of Ang-2 to Ang-1 [Ang-2/Ang-1]) were significantly associated with increased odds of 28-day mortality (Additional file 2: Figure S2, Table S2).

After adjusting for confounders, SARS-CoV-2 antigen, IL-6, IL-8, IL-10, TREM-1, sTNFR-1, sRAGE, and thrombomodulin were still associated with 28-day mortality. In unadjusted analyses, all the biomarkers associated with 28-day mortality were also associated with increased odds of 60-day mortality (Additional file 2: Figure S2, Table S3). Additionally, increased concentrations of Ang-1, vascular endothelial growth factor (VEGF), and protein C were associated with a reduced odds of 60-day mortality. After adjusting for confounders, IL-6, IL-8, IL-10, IP-10, TREM-1, sTNFR-1, thrombomodulin, Ang-1, sRAGE, and SARS-CoV-2 antigen were still associated with 60-day mortality.

Given the impact of tocilizumab on inflammatory biomarkers, we performed a sensitivity analysis by removing the 68 patients (17%) who received tocilizumab during their hospitalization. We found similar associations of biomarkers with 28- and 60-day mortality, with some exceptions (Additional file 2: Figure S3). In the unadjusted analyses, ICAM-1 was no longer associated with 28-day mortality, and Ang-2 became significantly associated with 28- and 60-day mortality. In the adjusted analyses, IL-10 was no longer associated with 28-day mortality.

The same biomarkers associated with 60-day mortality were also associated with time to death and recovery (Additional file 2: Tables S4, S5). Additionally, protein C and Ang-1 were significantly associated with recovery. Results of extended Cox model analyses are provided in Additional file 2: Tables S6 and S7.

Latent class analysis

We fit four models with increasing number of classes and found that a two-class model was the best fit for the data based on fit characteristics (Table 2). Using this model, 292 participants (73%) were categorized as Subtype 1 and 108 (27%) as Subtype 2. The average probability for most likely latent class membership was 0.96 for Subtype 1 and 0.93 for Subtype 2. Subtype 2 patients were older and more likely to be on invasive mechanical ventilation (IMV) at trial enrollment (33% vs 9%) (Table 3). Time from symptom onset, not included in the LCA, was similar between the two groups (Table 3). Subtype 2 patients were more likely to have comorbidities such as cerebrovascular disease, congestive heart failure, hypertension, diabetes, and chronic kidney disease (Additional file 2: Table S8).

Table 2 Fit statistics of latent class analysis results using clinical and protein biomarker data at study enrollment
Full size table
Table 3 Summary of patient characteristics by latent class analysis subtype assignment
Full size table

Compared to Subtype 1, Subtype 2 had higher baseline levels of sTNFR-1, creatinine, B-type natriuretic peptide (BNP), Ang-2, sRAGE, partial thromboplastin time (PTT), neutrophil to lymphocyte ratio, IP-10, IL-18, MMP-8, IL-8, white blood cell count (WBC), IL-6, C reactive protein (CRP), total bilirubin, SP-D, and SARS-CoV-2 antigen (Fig. 1, Table 3). ICAM-1, plasminogen activator inhibitor 1 (PAI-1), and BMI were not significantly different between the two groups. Subtype 2 had lower levels of protein C, bicarbonate, hematocrit, platelets, Ang-1, and VEGF. Creatinine, sTNFR-1, protein C, and bicarbonate were the most subtype separating variables. Given the strong association between SARS-CoV-2 antigen levels and outcomes in the literature, we conducted a sensitivity analysis by repeating LCA after excluding this analyte. There were no notable differences between the models though five patients switched subtypes (three became Subtype 1, and two became Subtype 2), suggesting that antigen levels were not a significant driver of subtype differences (Additional file 2: Figure S4).

Fig. 1
figure 1

Differences in the standardized values of each continuous variable by subtype assignment. The variables are sorted based on the degree of separation between the two subtypes. A standardized value of + 1 signifies that the mean value for a given subtype was one standard deviation higher than the mean value in the cohort as a whole. Ang = angiopoietin; BMI = body mass index; BNP = brain natriuretic peptide; CRP = C reactive protein; ICAM1 = intercellular adhesion molecule-1; IL = interleukin; IP10 = interferon-gamma inducible protein of 10 kDa; MMP8 = matrix metalloproteinase-8; NLR = neutrophil to lymphocyte ratio; PAI1 = plasminogen activator inhibitor-1; PTT = partial thromboplastin time; RAGE = soluble receptor for advanced glycation end products; SPD = surfactant protein D; TNFR1 = tumor necrosis factor receptor-1; VEGF = vascular endothelial growth factors; WBC = white blood cell count

Full size image

Subtype 2 designated participants had higher 28-day (41% vs 20%, p < 0.0001) and 60-day mortality (45% vs 24%, p < 0.0001) compared to Subtype 1 (Fig. 2, Table 3). Subtype 2 was associated with a subdistribution hazard ratio (SHR) of death of 2.5 (95%CI 1.7 to 3.5, p-value < 0.001) compared to Subtype 1 (Fig. 2, Additional file 2: Table S9). Similarly, Subtype 2 was associated with longer time to recovery (SHR 0.6, 95%CI 0.4 to 0.8, p-value < 0.001) (Fig. 2, Additional file 2: Table S9). When the study cohort was re-categorized to those not on IMV (WHO 5) and those on IMV or additional life support (WHO ≥ 6), the association of subtypes with all outcomes was only significant amongst those not on IMV (Fig. 2, Additional file 2: Table S9).

Fig. 2
figure 2

Clinical outcomes by latent class analysis (LCA) subtype assignment. A Mortality rates by subtype assignment stratified by WHO ordinal scale for COVID-19 severity upon trial enrollment; Within strata comparisons done using Chi-squared test or Fisher’s exact test where appropriate and p-value < 0.05 is depicted via Asterix. B Survival and recovery for Subtype 2 compared with Subtype 1 stratified by WHO ordinal scale for COVID-19 severity upon trial enrollment; For time to death, estimates are derived from Fine-Gray subdistribution hazard model with recovery as the competing event; For time to recovery, estimates are derived from Fine-Gray subdistribution hazard model with death as the competing event. ***P value < 0.001. WHO 5: hospitalized, noninvasive mechanical ventilation or high-flow nasal cannula (HFNC); WHO 6: hospitalized, intubation and invasive mechanical ventilation (IMV); WHO 7: hospitalized, IMV + additional support such as pressors or extracardiac membranous oxygenation

Full size image

Additionally, we performed sensitivity analyses to ensure that associations between latent subtypes and clinical outcomes were not mediated by patients’ assigned treatment arms. We removed the 57 participants who received one of the four active study drugs for which outcome data from the I-SPY COVID Trial has not yet been published. The strong association of Subtype 2 designation with mortality at 28 days (42% vs 18%, p < 0.0001) and 60 days (46% vs 22%, p < 0.0001) persisted. Next, we adjusted Fine-Gray models for assigned treatment arm in the full cohort of 400 patients. Subtype 2 was still associated with a subdistribution hazard ratio (SHR) of death of 2.7 (95%CI 1.8 to 3.9, p-value < 0.001) and longer time to recovery (SHR 0.5, 95%CI 0.4 to 0.7, p-value < 0.001) compared to Subtype 1.

We used the parsimonious model described by Sinha et al [32] using IL-8, bicarbonate, and protein C levels to determine which patients in this cohort would have been classified as having hyper-inflammatory ARDS. Only one participant was classified as hyper-inflammatory using this approach, and this participant had a Subtype 2 designation.

Longitudinal biomarker assessment

Longitudinal analyses were restricted to the 142 patients in the control arm. Pro-inflammatory biomarkers were higher across all time points in Subtype 2 compared to Subtype 1, and either remained elevated or slightly decreased over time (Fig. 3A). TREM-1 decreased over time in Subtype 2 but increased in Subtype 1, whereas sTNFR-1 decreased at a faster rate in Subtype 2 and remained relatively stable in Subtype 1. Of endothelial biomarkers, Ang-2 and thrombomodulin were higher and Ang-1 and VEGF were lower in Subtype 2 over time (Fig. 3B). Both epithelial biomarkers, SP-D and sRAGE, were higher across all time points in Subtype 2, with SP-D increasing and sRAGE decreasing in all patients (Fig. 3C). Of the coagulation biomarkers, PAI-1 and protein C, the latter was lower across all time points in Subtype 2 (Fig. 3D). PAI-1 decreased over time in Subtype 2 but remained stable in Subtype 1. A higher proportion of Subtype 2 designated patients died in the first week of enrollment compared with Subtype 1 (Additional file 2: Table S10).

Fig. 3
figure 3

Change in log10 biomarker concentration over time by subtype assignment in the control arm (N = 142). A Biomarkers of inflammation, B endothelial injury, C alveolar epithelial injury, and D disordered coagulation. Y-axis is the log10 concentration of biomarker. Estimates with 95% confidence intervals derived from linear mixed effects models. Asterisks denote significant difference in the slope of biomarker change over time between Subtype 1 and Subtype 2 based on Chi-squared test of interaction. *p value < 0.05. **p value < 0.005. ***p value < 0.001

Full size image

Discussion

In this secondary analysis of data from a randomized controlled trial, we found that participants with severe COVID-19 were characterized by significant biological heterogeneity. Using a combination of clinical data and protein biomarkers, we detected two latent subtypes of severe COVID-19 respiratory failure with significantly different compositions of biomarkers of inflammation, endothelial injury, alveolar epithelial injury, and disordered coagulation. Subtype 2 had significantly worse outcomes and was defined by higher baseline values of creatinine, sTNFR-1, BNP, Ang-2, and sRAGE and lower levels of protein C and bicarbonate among other differences. These subtypes differed from previously established COVID and non-COVID ARDS hyper- and hypo-inflammatory subtypes.

Though the subtypes, on average, demonstrate some differences in clinical parameters, no single clinical or biological parameter alone is sufficient for subtype designation. For instance, an older patient on IMV may appear clinically more worrisome but can still be designated as Subtype 1 based on the biological parameters incorporated into the model in our study. Indeed, a higher proportion of Subtype 2 designated patients were receiving IMV at trial enrollment. However, in stratified analyses, the subtypes’ prognostic value was most notable among those not on IMV. The lack of association with mortality among those on IMV or additional life support may be related to the small number of patients in these strata at study enrollment; nonetheless, the robust association with mortality among those not on IMV suggests that the subtypes capture pathobiology beyond the severity of acute respiratory failure. These findings have several implications. First, trials enrolling patients with severe COVID-19 as defined by respiratory failure status likely included a biologically heterogeneous patient population, which may explain inconsistent results in trials of corticosteroids and immunomodulators [6, 8, 12, 35,36,37,38] Our findings suggest that current clinical approaches to recruiting trial participants may be generating biologically heterogenous samples. Second, the treatment paradigm of patients with COVID-19, which primarily relies on the degree of respiratory failure, may be insufficient to identify those most likely to respond to therapies and at highest risk of poor outcomes. As steroids and immunomodulators have side effects, fine-tuning the target population most likely to respond to these medications would be desirable. Third, though most patients in our cohort did not meet the Berlin criteria for ARDS, our findings suggest that biological derangements leading to poor outcomes start earlier than the onset of IMV. In line with the new Global Definition of ARDS that includes those on high flow nasal oxygen [39], studying subtypes of non-COVID ARDS earlier in the course of disease and prior to the initiation of mechanical ventilation may both inform the natural trajectory of ARDS subtypes and offer an optimal window to deliver targeted therapies. Understanding early signs of biological pathways of injury is the first step toward the identification of treatable traits [40]. Future studies comparing our findings to those of patients with other types of severe viral pneumonia and/or ARDS from non-COVID related triggers will further aid in identifying shared pathways of injury that can be targeted for therapies.

The latent subtypes in this cohort of patients with hypoxemic respiratory failure had remarkable differences from the hyper- and hypo-inflammatory subtypes of COVID and non-COVID ARDS [20, 25]. Early in the pandemic, Sinha and colleagues discovered two latent subtypes of COVID-19 ARDS, as defined by the Berlin criteria, using LCA of readily available clinical data and IL-6. The authors then applied a validated clinical classifier model for non-COVID ARDS subtypes to phenotype their study cohort into hyper- and hypo-inflammatory ARDS. Over 80% of the participants in each of the two COVID ARDS subtypes overlapped with the corresponding non-COVID hypo- and hyper-inflammatory ARDS subtypes. In contrast, nearly all patients in our study cohort were categorized as hypo-inflammatory using the same parsimonious model. The lack of overlap of our subtypes with hyper- and hypo-inflammatory ARDS as compared with those of Sinha et al. is likely multi-factorial. First, most of our study cohort is patients with less severe respiratory failure than ARDS as defined by the Berlin Criteria. Our ability to still identify a latent subtype with over 40% mortality further supports the notion that biological signals of adverse outcomes are detectable earlier than the onset of IMV. Indeed, similar to hyper-inflammatory ARDS, Subtype 2 designated patients demonstrated higher levels of sTNFR-1 and lower levels of bicarbonate, protein C, and platelets compared to Subtype 1, suggesting similar pathways of injury. Biomarkers such as IL-6, IL-8, PAI-1, and ICAM-1, key to class designation in hyper- and hypo-inflammatory ARDS [20], were notably less class separating in severe COVID-19 subtypes, potentially rendering them less useful in phenotyping this patient population. Altogether, these differences may be related to differences in host–pathogen interactions over time in COVID and non-COVID respiratory failure as well as the impact of dexamethasone on biomarker concentrations in our study cohort. A recent study by Verhoef et al. performed latent profile analysis in a cohort of COVID-19 patients who presented to four different medical centers using residual blood samples collected during routine care [28]. The authors similarly found two latent profiles associated with significantly different clinical outcomes. The profiles demonstrated similar trends in the association between biomarkers of inflammation and endothelial injury as the subtypes in our study, although in contrast to our findings, IL-6 levels were the most important variable in profile designation. We are unable to directly compare the subtypes we observed to these latent profiles, given important differences between the cohorts (e.g., inclusion of outpatients in that cohort, variable corticosteroid usage) and the lack of a classifier model for the latent profiles; however, taken together, these two studies indicate that important biological heterogeneity is clearly present even in relatively milder and earlier forms of acute respiratory illness. More studies on patients presenting earlier in respiratory disease onset are warranted to validate these findings.

Patients with a Subtype 2 designation in our study were characterized by a higher degree of inflammation, alveolar epithelial injury, endothelial injury, and coagulation abnormalities. Interestingly, SARS-CoV-2 antigen level, though higher in Subtype 2, was less important to the model than the aforementioned biomarkers. The association between viral RNA in plasma with biomarkers of inflammation, injury, and poor outcomes in COVID-19 has been established [41,42,43], and studies have found a similar association between viral antigen levels and clinical outcomes [44, 45]. Though our analyses do not allow conclusive causal inference, one potential mechanism for the pattern of biomarker derangements in this cohort may be that a higher viral burden at the onset of disease leads to more severe initial lung injury, triggering both local and systemic inflammation that mediate poor outcomes. Further investigation of pathogenic differences in host response in those with acute hypoxemic respiratory failure are needed.

Not only were the two subtypes in this study biologically different at baseline, but this difference was maintained over the first 7 days of study enrollment. Subtype 2 was characterized by persistent inflammation, endothelial dysfunction, epithelial injury, and abnormal coagulation throughout the first week of trial enrollment. Some biomarker trajectories demonstrated a temporal trend toward convergence. Given similar rates of immunomodulator and dexamethasone use between the subtypes, the improving trend over time in Subtype 2 may have been in part due to an imbalance between those who recovered and those who died between the two subtypes during the first week of the study (Additional file 2: Table S9). Nonetheless, the pattern of persistent biomarker derangements in Subtype 2 raises important considerations. Baseline biomarker differences between the two subtypes persist beyond the immediate period in which the subtypes are defined, rendering them more likely to be clinically meaningful. For instance, high levels of TREM-1, a marker of innate immune response, have been associated with disease severity, duration of mechanical ventilation, and clinical outcomes in COVID-19 [46, 47]. Indeed, TREM-1 inhibition in those with severe COVID-19 demonstrated a trend toward reduction in 28-day mortality in a phase 2 randomized controlled trial [48]. Surprisingly, higher baseline TREM-1 levels did not predict beneficial response from therapy, suggesting that cross-sectional single biomarker measurements may not adequately capture treatment responsive subsets of patients. Similarly, a randomized, double-blind, placebo-controlled trial of oral imatinib, a tyrosine-kinase inhibitor, found that imatinib significantly reduced 90-day mortality in hospitalized adults with severe COVID-19 [49]. In secondary analyses, the effect of imatinib on mortality was fully mediated by a reduction in IL-6 levels, yet baseline plasma IL-6 concentration did not moderate the effect of imatinib on outcomes [50]. Indeed, using a panel of biomarkers, the authors found that a cluster of patients marked by increased levels of biomarkers of alveolar epithelial injury was the only one that experienced a meaningful mortality reduction with imatinib, suggesting that a complex interplay of alveolar epithelial injury, endothelial barrier integrity, and systemic inflammation is likely needed to experience mortality reduction from imatinib. Subtype 2 designated patients in our study appear to capture such a population of patients. The persistent dysregulation in these pathways and the differential mortality rates early in our study further suggest that interventions targeting these pathways of injury may have the highest value at the earliest timepoint of delivery. As we transition out of the pandemic, applying these insights to patients with non-COVID acute hypoxemic respiratory failure is worth consideration. Tools for real time subtyping in non-COVID ARDS are emerging [51] and capacity building studies are underway, laying the foundation for future trials to randomize patients based in part on biological subtypes. Understanding similarities and differences in biological heterogeneity between COVID and non-COVID respiratory failure is important to aid in targeting therapies to those with shared biological pathways of injury.

Our study has notable strengths. Patients in this cohort were recruited from a variety of academic and community hospitals across the United States, yielding a diverse cohort which enhances generalizability. Furthermore, participants were enrolled within a clinical trial with largely uniform clinical practices. The large panel of biomarkers in our study were carefully chosen a priori based on data from non-COVID ARDS as well as early findings in COVID-19. Our study also has limitations. Only participants with biospecimens available at day 1 were included, which may have introduced selection bias, although the clinical characteristics and outcomes of the patients without plasma were similar to those included. In the absence of a separate validation cohort, we did not attempt to develop a classifier model for pragmatic identification of these subtypes for research or clinical care. Lastly, given the high mortality rate in this cohort, informative missingness due to different survival rates between the subtypes limits any conclusions that can be drawn from longitudinal biomarker analyses.

Conclusions

In this secondary analysis of data from a randomized controlled trial of hospitalized adults with severe COVID-19, we found evidence of two latent subtypes with divergent clinical outcomes. Subtype 2, comprising 27% of the cohort, was characterized by persistent inflammation, epithelial and endothelial injury, as well as disordered coagulation and had more than twice the mortality of Subtype 1. Latent biological heterogeneity may be contributing to negative findings in trials of pharmacotherapies. Early subtyping of patients with acute hypoxemic respiratory failure has the potential to identify biological pathways underlying clinical outcomes, reveal treatment-responsive subgroups of patients, and offer an earlier window for study of targeted therapies.

Availability of data and materials

The data reported in this manuscript is maintained by the study sponsor, Quantum Leap Healthcare Collaborative (QLHC). Source data supporting this manuscript is available to qualified investigators upon approval by QLHC and the ISPY COVID Data Access and Publications Committee. Requests can be initiated by contacting the corresponding author of this manuscript.

Abbreviations

Ang:

Angiopoietin

ARDS:

Acute respiratory distress syndrome

BNP:

B-type natriuretic peptide

COVID-19:

Coronavirus disease 2019

CRP:

C reactive protein

ELISA:

Enzyme linked immunoassay methods

FDR:

False discovery rate

ICAM-1:

Intercellular adhesion molecule 1

IFN:

Interferon

IL:

Interleukin

IMV:

Invasive mechanical ventilation

IP:

Interferon-gamma induced protein

LCA:

Latent class analysis

MMP:

Matrix metalloproteinase

PAI-1:

Plasminogen activator inhibitor 1

PEEP:

Positive end expiratory pressure

PTT:

Partial thromboplastin time

SARS-CoV-2:

Severe acute respiratory syndrome coronavirus-2

SHR:

Subdistribution hazard ratio

SP-D:

Surfactant protein D

sRAGE:

Soluble receptor for advanced glycation end products

sTNFR-1:

Soluble tumor necrosis factor receptor 1

TNF:

Tumor necrosis factor

TREM:

Triggering receptor expressed on myeloid cells

VEGF:

Vascular endothelial growth factor

WBC:

White blood cell count

WHO:

World Health Organization

References

  1. Morens DM, Fauci AS. Emerging Pandemic Diseases: How We Got to COVID-19. Cell. 2020;182(5):1077–92. https://0-doi-org.brum.beds.ac.uk/10.1016/j.cell.2020.08.021.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Docherty AB, Harrison EM, Green CA, et al. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ. 2020;369: m1985. https://0-doi-org.brum.beds.ac.uk/10.1136/bmj.m1985.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Cummings MJ, Baldwin MR, Abrams D, et al. Epidemiology, clinical course, and outcomes of critically ill adults with COVID-19 in New York City: a prospective cohort study. medRxiv. 2020. https://0-doi-org.brum.beds.ac.uk/10.1101/2020.04.15.20067157.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Schultze JL, Aschenbrenner AC. COVID-19 and the human innate immune system. Cell. 2021;184(7):1671–92. https://0-doi-org.brum.beds.ac.uk/10.1016/j.cell.2021.02.029.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Group RC, Horby P, Lim WS, et al. Dexamethasone in Hospitalized Patients with Covid-19. N Engl J Med. 2021;384(8):693–704. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2021436.

    Article  Google Scholar 

  6. Tomazini B, Maia I, Cavalcanti A, et al. Effect of dexamethasone on days alive and ventilator-free in patients with moderate or severe acute respiratory distress syndrome and COVID-19: the CoDEX randomized clinical trial. JAMA. 2020. https://0-doi-org.brum.beds.ac.uk/10.1001/jama.2020.17021.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Dequin P, Heming N, Meziani F, et al. Effect of hydrocortisone on 21-day mortality or respiratory support among critically ill patients with COVID-19: a randomized clinical trial. JAMA. 2020. https://0-doi-org.brum.beds.ac.uk/10.1001/jama.2020.16761.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Crothers K, DeFaccio R, Tate J, et al. Dexamethasone in hospitalised COVID-19 patients not on intensive respiratory support. Eur Respir J. 2022. https://0-doi-org.brum.beds.ac.uk/10.1183/13993003.02532-2021.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Angus DC, Derde L, Al-Beidh F, et al. Effect of hydrocortisone on mortality and organ support in patients with severe COVID-19: the REMAP-CAP COVID-19 corticosteroid domain randomized clinical trial. JAMA. 2020;324(13):1317–29. https://0-doi-org.brum.beds.ac.uk/10.1001/jama.2020.17022.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Salvarani C, Dolci G, Massari M, et al. Effect of tocilizumab vs standard care on clinical worsening in patients hospitalized with COVID-19 pneumonia: a randomized clinical trial. JAMA Intern Med. 2021. https://0-doi-org.brum.beds.ac.uk/10.1001/jamainternmed.2020.6615.

    Article  PubMed  Google Scholar 

  11. Hermine O, Mariette X, Tharaux P, Resche-Rigon M, Porcher R, Ravaud P. Effect of tocilizumab vs usual care in adults hospitalized With COVID-19 and moderate or severe pneumonia: a randomized clinical trial. JAMA Intern Med. 2021. https://0-doi-org.brum.beds.ac.uk/10.1001/jamainternmed.2020.6820.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Gordon A, Mouncey P, Al-Beidh F, et al. Interleukin-6 receptor antagonists in critically ill patients with Covid-19. N Engl J Med. 2021;384(16):25. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2100433.

    Article  Google Scholar 

  13. Rosas IO, Bräu N, Waters M, et al. Tocilizumab in hospitalized patients with severe Covid-19 pneumonia. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2028700 2021 (research-article) (in English).

  14. Veiga VC, Prats JAGG, Farias DLC, et al. Effect of tocilizumab on clinical outcomes at 15 days in patients with severe or critical coronavirus disease 2019: randomised controlled trial. 2021 (in English). https://0-doi-org.brum.beds.ac.uk/10.1136/bmj.n84.

  15. Soin AS, Kumar K, Choudhary NS, et al. Tocilizumab plus standard care versus standard care in patients in India with moderate to severe COVID-19-associated cytokine release syndrome (COVINTOC): an open-label, multicentre, randomised, controlled, phase 3 trial. Lancet Respir Med. 2021;9(5):511–21. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(21)00081-3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Ader F, Bouscambert-Duchamp M, Hites M, et al. Remdesivir plus standard of care versus standard of care alone for the treatment of patients admitted to hospital with COVID-19 (DisCoVeRy): a phase 3, randomised, controlled, open-label trial. Lancet Infect Dis. 2021. https://0-doi-org.brum.beds.ac.uk/10.1016/S1473-3099(21)00485-0.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Wang Y, Zhang D, Du G, et al. Remdesivir in adults with severe COVID-19: a randomised, double-blind, placebo-controlled, multicentre trial. Lancet (London, England). 2020;395:25. https://0-doi-org.brum.beds.ac.uk/10.1016/S0140-6736(20)31022-9.

    Article  Google Scholar 

  18. Beigel J, Tomashek K, Dodd L, et al. Remdesivir for the treatment of Covid-19—final report. N Engl J Med. 2020;383(19):25. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2007764.

    Article  Google Scholar 

  19. Consortium WST. Repurposed antiviral drugs for Covid-19—interim WHO solidarity trial results. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2023184 2020 (research-article) (in English).

  20. Calfee C, Delucchi K, Parsons P, Thompson B, Ware L, Matthay M. Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials. Lancet Respir Med. 2014;2(8):25. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(14)70097-9.

    Article  Google Scholar 

  21. Calfee C, Delucchi K, Sinha P, et al. Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial. Lancet Respir Med. 2018. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(18)30177-2.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Sinha P, Delucchi K, Thompson B, McAuley D, Matthay M, Calfee C. Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study. Intensive Care Med. 2018. https://0-doi-org.brum.beds.ac.uk/10.1007/s00134-018-5378-3.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Sinha P, Delucchi KL, Chen Y, et al. Latent class analysis-derived subphenotypes are generalisable to observational cohorts of acute respiratory distress syndrome: a prospective study. Thorax. 2021. https://0-doi-org.brum.beds.ac.uk/10.1136/thoraxjnl-2021-217158. (in English).

    Article  PubMed  Google Scholar 

  24. Famous K, Delucchi K, Ware L, et al. Acute respiratory distress syndrome subphenotypes respond differently to randomized fluid management strategy. Am J Respir Crit Care Med. 2017. https://0-doi-org.brum.beds.ac.uk/10.1164/rccm.201603-0645OC.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Sinha P, Furfaro D, Cummings MJ, et al. Latent class analysis reveals COVID-19-related ARDS subgroups with differential responses to corticosteroids. Am J Respir Crit Care Med. 2021. https://0-doi-org.brum.beds.ac.uk/10.1164/rccm.202105-1302OC.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Sinha P, Calfee CS, Cherian S, et al. Prevalence of phenotypes of acute respiratory distress syndrome in critically ill patients with COVID-19: a prospective observational study. Lancet Respir Med. 2020;8(12):1209–18. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(20)30366-0.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Ranjeva S, Pinciroli R, Hodell E, et al. Identifying clinical and biochemical phenotypes in acute respiratory distress syndrome secondary to coronavirus disease-2019. EClinicalMedicine. 2021;34: 100829. https://0-doi-org.brum.beds.ac.uk/10.1016/j.eclinm.2021.100829.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Verhoef PA, Spicer AB, Lopez-Espina C, et al. Analysis of protein biomarkers from hospitalized COVID-19 patients reveals severity-specific signatures and two distinct latent profiles with differential responses to corticosteroids. Crit Care Med. 2023;51(12):1697–705. https://0-doi-org.brum.beds.ac.uk/10.1097/CCM.0000000000005983.

    Article  CAS  PubMed  Google Scholar 

  29. Files DC, Matthay MA, Calfee CS, et al. I-SPY COVID adaptive platform trial for COVID-19 acute respiratory failure: rationale, design and operations. BMJ Open. 2022;12(6): e060664. https://0-doi-org.brum.beds.ac.uk/10.1136/bmjopen-2021-060664.

    Article  PubMed  Google Scholar 

  30. Consortium ISC. Report of the first seven agents in the I-SPY COVID trial: a phase 2, open label, adaptive platform randomised controlled trial. EClinicalMedicine. 2023;58: 101889. https://0-doi-org.brum.beds.ac.uk/10.1016/j.eclinm.2023.101889.

    Article  Google Scholar 

  31. Sinha P, Calfee CS, Delucchi KL. Practitioner’s guide to latent class analysis: Methodological considerations and common pitfalls. Crit Care Med. 2021;49(1):e63–79. https://0-doi-org.brum.beds.ac.uk/10.1097/CCM.0000000000004710.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Sinha P, Delucchi KL, McAuley DF, O’Kane CM, Matthay MA, Calfee CS. Development and validation of parsimonious algorithms to classify acute respiratory distress syndrome phenotypes: a secondary analysis of randomised controlled trials. Lancet Respir Med. 2020;8(3):247–57. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(19)30369-8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Lo Y, Mendell NR, Rubin DB. Testing the number of components in a normal mixture. Biometrika. 2001;88(3):767–78.

    Article  MathSciNet  Google Scholar 

  34. Henson JM, Reise SP, Kim KH. Detecting mixtures from structural model differences using latent variable mixture modeling: a comparison of relative model fit statistics. Struct Eq Model Multidiscip J. 2007;14(2):202–26.

    Article  MathSciNet  Google Scholar 

  35. Horby P, Lim W, Emberson J, et al. Dexamethasone in hospitalized patients with Covid-19. N Engl J Med. 2021. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2021436.

    Article  PubMed  Google Scholar 

  36. Group RC. Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet. 2021;397(10285):1637–45. https://0-doi-org.brum.beds.ac.uk/10.1016/S0140-6736(21)00676-0.

    Article  Google Scholar 

  37. Rosas IO, Brau N, Waters M, et al. Tocilizumab in hospitalized patients with severe covid-19 pneumonia. N Engl J Med. 2021;384(16):1503–16. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2028700.

    Article  CAS  PubMed  Google Scholar 

  38. Salama C, Han J, Yau L, et al. Tocilizumab in patients hospitalized with Covid-19 pneumonia. N Engl J Med. 2021;384(1):20–30. https://0-doi-org.brum.beds.ac.uk/10.1056/NEJMoa2030340.

    Article  CAS  PubMed  Google Scholar 

  39. Matthay MA, Arabi Y, Arroliga AC, et al. A new global definition of acute respiratory distress syndrome. Am J Respir Crit Care Med. 2023. https://0-doi-org.brum.beds.ac.uk/10.1164/rccm.202303-0558WS.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Maslove DM, Tang B, Shankar-Hari M, et al. Redefining critical illness. Nat Med. 2022;28(6):1141–8. https://0-doi-org.brum.beds.ac.uk/10.1038/s41591-022-01843-x.

    Article  CAS  PubMed  Google Scholar 

  41. Chen X, Zhao B, Qu Y, et al. Detectable serum severe acute respiratory syndrome coronavirus 2 viral load (RNAemia) Is closely correlated with drastically elevated interleukin 6 level in critically Ill patients with coronavirus disease 2019. Clin Infect Dis. 2020;71(8):1937–42. https://0-doi-org.brum.beds.ac.uk/10.1093/cid/ciaa449.

    Article  CAS  PubMed  Google Scholar 

  42. Prebensen C, Myhre PL, Jonassen C, et al. Severe acute respiratory syndrome coronavirus 2 RNA in plasma is associated with intensive care unit admission and mortality in patients hospitalized with coronavirus disease 2019. Clin Infect Dis. 2021;73(3):e799–802. https://0-doi-org.brum.beds.ac.uk/10.1093/cid/ciaa1338.

    Article  CAS  PubMed  Google Scholar 

  43. Bermejo-Martin JF, Gonzalez-Rivera M, Almansa R, et al. Viral RNA load in plasma is associated with critical illness and a dysregulated host response in COVID-19. Crit Care. 2020;24(1):691. https://0-doi-org.brum.beds.ac.uk/10.1186/s13054-020-03398-0.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Wick KD, Leligdowicz A, Willmore A, et al. Plasma SARS-CoV-2 nucleocapsid antigen levels are associated with progression to severe disease in hospitalized COVID-19. Crit Care. 2022;26(1):278. https://0-doi-org.brum.beds.ac.uk/10.1186/s13054-022-04153-3.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Group A-TS, Rogers AJ, Wentworth D, et al. The Association of Baseline Plasma SARS-CoV-2 Nucleocapsid Antigen Level and Outcomes in Patients Hospitalized With COVID-19. Ann Intern Med. 2022;175(10):1401–10. https://0-doi-org.brum.beds.ac.uk/10.7326/M22-0924.

    Article  Google Scholar 

  46. Van Singer M, Brahier T, Ngai M, et al. COVID-19 risk stratification algorithms based on sTREM-1 and IL-6 in emergency department. J Allergy Clin Immunol. 2021;147(1):99-106 e4. https://0-doi-org.brum.beds.ac.uk/10.1016/j.jaci.2020.10.001.

    Article  CAS  PubMed  Google Scholar 

  47. de Nooijer AH, Grondman I, Lambden S, et al. Increased sTREM-1 plasma concentrations are associated with poor clinical outcomes in patients with COVID-19. Biosci Rep. 2021;41(7):25.

    Article  Google Scholar 

  48. Francois B, Lambden S, Garaud JJ, et al. Evaluation of the efficacy and safety of TREM-1 inhibition with nangibotide in patients with COVID-19 receiving respiratory support: the ESSENTIAL randomised, double-blind trial. EClinicalMedicine. 2023;60: 102013. https://0-doi-org.brum.beds.ac.uk/10.1016/j.eclinm.2023.102013.

    Article  PubMed  PubMed Central  Google Scholar 

  49. Aman J, Duijvelaar E, Botros L, et al. Imatinib in patients with severe COVID-19: a randomised, double-blind, placebo-controlled, clinical trial. Lancet Respir Med. 2021;9(9):957–68. https://0-doi-org.brum.beds.ac.uk/10.1016/S2213-2600(21)00237-X.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. de Brabander J, Duijvelaar E, Schippers JR, et al. Immunomodulation and endothelial barrier protection mediate the association between oral imatinib and mortality in hospitalised COVID-19 patients. Eur Respir J. 2022. https://0-doi-org.brum.beds.ac.uk/10.1183/13993003.00780-2022.

    Article  PubMed  PubMed Central  Google Scholar 

  51. McAuley D. Clinical evaluation of a point of care (POC) assay to identify phenotypes in the acute respiratory distress syndrome—Full Text View—Clinical.Trials.gov. https://clinicaltrials.gov/study/NCT04009330.

Download references

Acknowledgements

Quantum Leap Healthcare Collaborative is the trial sponsor. Funding for this trial has come from: the COVID R&D Consortium, Allergan, Amgen Inc., Takeda Pharmaceutical Company, Implicit Bioscience, Johnson & Johnson, Pfizer Inc., Roche/Genentech, Apotex Inc., FAST Grant from Emergent Venture George Mason University, The DoD Defense Threat Reduction Agency (DTRA), The Department of Health and Human Services Biomedical Advanced Research and Development Authority (BARDA), and The Grove Foundation. Effort sponsored by the U.S. Government under Other Transaction number W15QKN-16-9-1002 between the MCDC, and the Government. The US Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the U.S. Government. The funders had no role in data collection or analysis or preparation of this manuscript; some funders reviewed this manuscript for accuracy.

MEMBERS OF THE I-SPY COVID COnSORTIUM (AUTHORS):

Writing Committee: Neil R. Aggarwal, MD, MHSc8, Timothy Albertson, MD27, Sara Auld, MD21, Jeremy R. Beitler, MD, MPH1, Paul Berger, DO6, Ellen L. Burnham, MD8, Carolyn S. Calfee, MD, MAS9, Nathan Cobb, MD19, Alessio Crippa, PhD29, Andrea Discacciati, PhD29, Martin Eklund, PhD29, Laura Esserman, MD, MBA12, D. Clark Files, MD2, Eliot Friedman, MD25, Sheetal Gandotra, MD4, Kashif Khan, MD32, Jonathan Koff, MD7, Santhi Kumar MD18, Kathleen D. Liu, MD, PhD15, Thomas R. Martin, MD33, Michael A. Matthay, MD16, Nuala J. Meyer MD3, Timothy Obermiller, MD26, MS3, Philip Robinson, MD20, Derek Russell, MD4,5, Karl Thomas, MD2, Se Fum Wong, MD32, Richard G. Wunderink, MD28, Mark M. Wurfel, PhD, MD33, Albert Yen, MD32, and Fady A. Youssef, MD17

Site Investigators and Staff: Anita Darmanian, MD1, Amy L. Dzierba, PharmD1, Ivan Garcia, RRT1, Katarzyna Gosek, PharmD1, Purnema Madahar, MD, MS1, Aaron M. Mittel, MD1, Justin Muir, PharmD1, Amanda Rosen, MD1, John Schicchi, MD1, Alexis L. Serra, MD, MPH1, Romina Wahab, MD1, Kevin W. Gibbs, MD2, Leigha Landreth, RN2, Mary LaRose, RN2, Lisa Parks, RN2, Adina Wynn, MPH, CCRP2, Caroline A.G. Ittner, PhD3, Nilam S. Mangalmurti, MD3, John P. Reilly, MD MS3, Donna Harris, BSN4, Abhishek Methukupally, MBBS4, Siddharth Patel, MBBS, MPH4, Lindsie Boerger, BA7, John Kazianis, MD7, Carrie Higgins8, Jeff McKeehan, MS8, Brian Daniel, RCP, RRT11, Scott Fields, PharmD13, James Hurst-Hopf, MS14, Alejandra Jauregui, BA9, Lamorna Brown Swigart, PhD14, Daniel Belvins, CCRP17, Catherine Nguyen, MD17, Alexis Suarez, MS17, Maged A. Tanios, MD17, Farjad Sarafian, MD20, Usman Shah, MD20, Max Adelman MD, MSc21, Christina Creel-Bulos, MD22, Joshua Detelich, MD21, Gavin Harris MD21, Katherine Nugent MD23, Christina Spainhour, RN, CCRC23, Philip Yang, MD21, Angela Haczku, MD, PhD27, Erin Hardy27, Richart Harper, MD27, Brian Morrissey, MD27, Christian Sandrock, MD, MPH27, G. R. Scott Budinger, MD28, Helen K. Donnelly, RN, BSN28, Benjamin D. Singer, MD28, Ari Moskowitz, MD30

Safety Working Group Chairs: Melissa Coleman, MD9; Joseph Levitt, MD34

QLHC Staff: Ruixiao Lu, PhD31, Paul Henderson, PhD31, Adam Asare, PhD31, Imogene Dunn, PhD, Alejandro Botello Barragan31

1Centre for Acute Respiratory Failure, Columbia University, 2Department of Internal Medicine, Wake Forest University , 3Department of Medicine, University of Pennsylvania Perelman School of Medicine, 4Pulmonary, Allergy, and Critical Care Medicine, University of Alabama, Birmingham, 5Pulmonary Section, Birmingham Veteran’s Affairs Medical Centre, 6Department of Critical Care, Sanford Health, 7Section of Pulmonary, Critical Care, and Sleep Medicine, Yale University, 8Department of Medicine, University of Colorado , 9Division of Pulmonary, Critical Care, Allergy and Sleep Medicine, University of California San Francisco , 10Division of Cardiothoracic Surgery, Department of Surgery, Critical Care Medicine, University of California San Francisco , 11Division of Critical Care Medicine, University of California San Francisco , 12Department of Surgery and Radiology, University of California San Francisco, 13Investigational Drug Service, University of California San Francisco, 14Department of Laboratory Medicine, University of California San Francisco, 15Divisions of Nephrology and Critical Care Medicine, University of California San Francisco, 16Cardiovacular Research Institute, University of California San Francisco, 17Department of Internal Medicine, Pulmonary Division, Long Beach Memorial Medical Centre, 18Division of Pulmonary, Critical Care and Sleep Medicine, Keck School of Medicine, USC, 19Georgetown University Medical Centre, 20Hoag Memorial Hospital Presbyterian Centre for Research and Education, 21Department of Medicine, Emory University, 22Department of Anesthesiology, Emory University, 23Department of Emergency Medicine, Emory University, 24Department of Surgery, Emory University, 25Department of Medicine, Main Line Health, 26Critical Care Medicine, Logan Health Research Institute , 27Division of Pulmonary, Critical Care and Sleep Medicine, University of California Davis, 28Department of Medicine, Pulmonary and Critical Care Division, Northwestern University Feinberg School of Medicine, 29Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, 30Division of Critical Care Medicine, Montefiore Medical Centre, 31Quantum Leap Healthcare, 32Critical Care Medicine, Keck School of Medicine, 33Pulmonary, Critical Care and Sleep Medicine, University of Washington, 34Pulmonary, Allergy and Critical Care Medicine, Stanford University.

Funding

NA is funded by NHLBI F32HL165828. Quantum Leap Healthcare Collaborative is the trial sponsor. Funding for this trial has come from: the COVID R&D Consortium, Allergan, Amgen Inc., Takeda Pharmaceutical Company, Implicit Bioscience, Johnson & Johnson, Pfizer Inc., Roche/Genentech, Apotex Inc., FAST Grant from Emergent Venture George Mason University, The DoD Defense Threat Reduction Agency (DTRA), The Department of Health and Human Services Biomedical Advanced Research and Development Authority (BARDA), and the Grove Foundation. Effort sponsored by the US Government under Other Transaction number W15QKN-16-9-1002 between the MCDC and the Government.

Author information

Authors and Affiliations

  1. Division of Pulmonary, Critical Care, Allergy, and Sleep Medicine, University of California, Room M-1083, 505 Parnassus Ave., San Francisco, CA, 94143, USA

    Narges Alipanah-Lechner, James Hurst-Hopf, Andrew Willmore, Benjamin LaCombe, Michael A. Matthay, Carolyn S. Calfee, Neil R. Aggarwal, Timothy Albertson, Sara Auld, Jeremy R. Beitler, Paul Berger, Ellen L. Burnham, Nathan Cobb, Alessio Crippa, Andrea Discacciati, Martin Eklund, D. Clark Files, Eliot Friedman, Sheetal Gandotra, Kashif Khan, Jonathan Koff, Santhi Kumar, Thomas R. Martin, Nuala J. Meyer, Timothy Obermiller, Philip Robinson, Derek Russell, Se Fum Wong, Richard G. Wunderink, Mark M. Wurfel, Albert Yen, Fady A. Youssef, Anita Darmanian, Amy L. Dzierba, Ivan Garcia, Katarzyna Gosek, Purnema Madahar, Aaron M. Mittel, Justin Muir, Amanda Rosen, John Schicchi, Alexis L. Serra, Romina Wahab, Kevin W. Gibbs, Leigha Landreth, Mary LaRose, Lisa Parks, Adina Wynn, Caroline A. G. Ittner, Nilam S. Mangalmurti, John P. Reilly, Donna Harris, Abhishek Methukupally, Siddharth Patel, Lindsie Boerger, John Kazianis, Carrie Higgins, Jeff McKeehan, Brian Daniel, Scott Fields, Alejandra Jauregui, Daniel Belvins, Catherine Nguyen, Alexis Suarez, Maged A. Tanios, Farjad Sarafian, Usman Shah, Max Adelman, Christina Creel-Bulos, Joshua Detelich, Gavin Harris, Katherine Nugent, Christina Spainhour, Philip Yang, Angela Haczku, Erin Hardy, Richart Harper, Brian Morrissey, Christian Sandrock, G. R. Scott Budinger, Helen K. Donnelly, Benjamin D. Singer, Ari Moskowitz, Melissa Coleman, Joseph Levitt, Ruixiao Lu, Paul Henderson, Adam Asare, Imogene Dunn & Alejandro Botello Barragan

  2. Department of Psychiatry and Behavioral Sciences, University of California, San Francisco, CA, USA

    Kevin Delucchi & Karl Thomas

  3. Department of Laboratory Medicine, University of California, San Francisco, CA, USA

    Lamorna Swigart

  4. Cardiovascular Research Institute, University of California, San Francisco, CA, USA

    Kathleen D. Liu

  5. Virus Isolation and Serology Laboratory, Applied and Developmental Directorate, Frederick National Laboratory, Frederick, MD, USA

    Robin Dewar & Perrine Lallemand

  6. Division of Clinical Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, USA

    H. Clifford Lane

  7. Division of Nephrology, University of California, San Francisco, CA, USA

    Kathleen D. Liu

  8. Department of Anesthesia, University of California, San Francisco, CA, USA

    Michael A. Matthay & Carolyn S. Calfee

  9. Department of Surgery, University of California, San Francisco, CA, USA

    Laura Esserman

Authors
  1. Narges Alipanah-Lechner
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. James Hurst-Hopf
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Kevin Delucchi
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Lamorna Swigart
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Andrew Willmore
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Benjamin LaCombe
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Robin Dewar
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. H. Clifford Lane
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Perrine Lallemand
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Kathleen D. Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Laura Esserman
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Michael A. Matthay
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Carolyn S. Calfee
    View author publications

    You can also search for this author in PubMed Google Scholar

Consortia

the I-SPY COVID Consortium

  • Neil R. Aggarwal
  • , Timothy Albertson
  • , Sara Auld
  • , Jeremy R. Beitler
  • , Paul Berger
  • , Ellen L. Burnham
  • , Carolyn S. Calfee
  • , Nathan Cobb
  • , Alessio Crippa
  • , Andrea Discacciati
  • , Martin Eklund
  • , Laura Esserman
  • , D. Clark Files
  • , Eliot Friedman
  • , Sheetal Gandotra
  • , Kashif Khan
  • , Jonathan Koff
  • , Santhi Kumar
  • , Kathleen D. Liu
  • , Thomas R. Martin
  • , Michael A. Matthay
  • , Nuala J. Meyer
  • , Timothy Obermiller
  • , Philip Robinson
  • , Derek Russell
  • , Karl Thomas
  • , Se Fum Wong
  • , Richard G. Wunderink
  • , Mark M. Wurfel
  • , Albert Yen
  • , Fady A. Youssef
  • , Anita Darmanian
  • , Amy L. Dzierba
  • , Ivan Garcia
  • , Katarzyna Gosek
  • , Purnema Madahar
  • , Aaron M. Mittel
  • , Justin Muir
  • , Amanda Rosen
  • , John Schicchi
  • , Alexis L. Serra
  • , Romina Wahab
  • , Kevin W. Gibbs
  • , Leigha Landreth
  • , Mary LaRose
  • , Lisa Parks
  • , Adina Wynn
  • , Caroline A. G. Ittner
  • , Nilam S. Mangalmurti
  • , John P. Reilly
  • , Donna Harris
  • , Abhishek Methukupally
  • , Siddharth Patel
  • , Lindsie Boerger
  • , John Kazianis
  • , Carrie Higgins
  • , Jeff McKeehan
  • , Brian Daniel
  • , Scott Fields
  • , James Hurst-Hopf
  • , Alejandra Jauregui
  • , Lamorna Swigart
  • , Daniel Belvins
  • , Catherine Nguyen
  • , Alexis Suarez
  • , Maged A. Tanios
  • , Farjad Sarafian
  • , Usman Shah
  • , Max Adelman
  • , Christina Creel-Bulos
  • , Joshua Detelich
  • , Gavin Harris
  • , Katherine Nugent
  • , Christina Spainhour
  • , Philip Yang
  • , Angela Haczku
  • , Erin Hardy
  • , Richart Harper
  • , Brian Morrissey
  • , Christian Sandrock
  • , G. R. Scott Budinger
  • , Helen K. Donnelly
  • , Benjamin D. Singer
  • , Ari Moskowitz
  • , Melissa Coleman
  • , Joseph Levitt
  • , Ruixiao Lu
  • , Paul Henderson
  • , Adam Asare
  • , Imogene Dunn
  •  & Alejandro Botello Barragan

Contributions

NA and CSC conceived of the study question, study design, and planned analyses. NA performed all data analyses, generated all figures, and wrote the manuscript summarizing the results. JH, LS, and BL performed protein biomarker measurements. KD assisted with planned analyses and performed accuracy checks on all analyses. AW performed protein biomarker quality control checks and maintained database. RD, HCL, and PL measured SARS-CoV-2 viral antigen levels. KDL, LE, and MAM were leaders of the I-SPY COVID trial and contributed to the study design and reviewed all analyses of this manuscript. All authors reviewed the manuscript.

Corresponding author

Correspondence to Narges Alipanah-Lechner.

Ethics declarations

Ethics approval and consent to participate

The I-SPY COVID trial is conducted under a central IRB mechanism (Wake Forest University Health Sciences IRB00066805). The trial was conducted in accordance with Good Clinical Practice guidelines and the Helsinki Declaration, and all participants, or their surrogates, provided informed consent prior to enrollment in the interventional arm of the study. The University of California, San Francisco internal review board approved the study.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

. Supplemental methods.

Additional file 2

. Supplemental tables and figures.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Alipanah-Lechner, N., Hurst-Hopf, J., Delucchi, K. et al. Novel subtypes of severe COVID-19 respiratory failure based on biological heterogeneity: a secondary analysis of a randomized controlled trial. Crit Care 28, 56 (2024). https://0-doi-org.brum.beds.ac.uk/10.1186/s13054-024-04819-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/s13054-024-04819-0

Keywords

Critical Care

ISSN: 1364-8535