Article Text
PDF
Abstract
Introduction Patients with COVID-19-related acute respiratory distress syndrome (ARDS) show limited systemic hyperinflammation, but immunomodulatory treatments are effective. Little is known about the inflammatory response in the lungs and if this could be targeted using high-dose steroids (HDS). We aimed to characterise the alveolar immune response in patients with COVID-19-related ARDS, to determine its association with mortality, and to explore the association between HDS treatment and the alveolar immune response.
Methods In this observational cohort study, a comprehensive panel of 63 biomarkers was measured in repeated bronchoalveolar lavage (BAL) fluid and plasma samples of patients with COVID-19 ARDS. Differences in alveolar–plasma concentrations were determined to characterise the alveolar inflammatory response. Joint modelling was performed to assess the longitudinal changes in alveolar biomarker concentrations, and the association between changes in alveolar biomarker concentrations and mortality. Changes in alveolar biomarker concentrations were compared between HDS-treated and matched untreated patients.
Results 284 BAL fluid and paired plasma samples of 154 patients with COVID-19 were analysed. 13 biomarkers indicative of innate immune activation showed alveolar rather than systemic inflammation. A longitudinal increase in the alveolar concentration of several innate immune markers, including CC motif ligand (CCL)20 and CXC motif ligand (CXCL)1, was associated with increased mortality. Treatment with HDS was associated with a subsequent decrease in alveolar CCL20 and CXCL1 levels.
Conclusions Patients with COVID-19-related ARDS showed an alveolar inflammatory state related to the innate host response, which was associated with a higher mortality. HDS treatment was associated with decreasing alveolar concentrations of CCL20 and CXCL1.
- ARDS
- COVID-19
- respiratory infection
- bronchoscopy
- critical care
- innate immunity
Data availability statement
Data are available upon reasonable request. De-identified participant data with data dictionary can be shared after approval of a proposal with a signed data access agreement and always in collaboration with the study group.
This article is made freely available for personal use in accordance with BMJ’s website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.
https://bmj.com/coronavirus/usageStatistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Patients with COVID-19-related acute respiratory distress syndrome (ARDS) show limited systemic hyperinflammation, but immunomodulatory treatments are effective. Little is known about the alveolar inflammatory response and if this could be targeted using high-dose steroids (HDS).
WHAT THIS STUDY ADDS
Patients with COVID-19-related ARDS showed an alveolar inflammatory state related to the innate host response, which was associated with a higher mortality. HDS treatment was associated with decreasing alveolar concentrations of innate immune system-related biomarkers.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
To engage in precision medicine for COVID-19-related ARDS, our data suggest focusing on immunomodulation in patients with increased neutrophil-related biomarker concentrations rather than an unselected population to allow for predictive enrichment.
Introduction
Patients with COVID-19 frequently require admission to the intensive care unit (ICU)1 2 because of acute respiratory failure due to acute respiratory distress syndrome (ARDS).3 4 ARDS is presumed to result from a hyperinflammatory state, initiated by proinflammatory neutrophils and macrophages.4–8 Indeed, concentrations of proinflammatory cytokines in the systemic compartment have shown to associate with the degree of respiratory failure in patients with COVID-19,9 10 but are lower than typically observed in ARDS not caused by COVID-19.11
Despite a limited systemic inflammatory response compared with other critically ill patients, early administration of corticosteroids and specific immunomodulators reduces mortality in critically ill patients with COVID-19.12–15 A possible hypothesis for such benefit is that a compartmentalised hyperinflammatory response is more pronounced to the alveolar space in COVID-19,16–18 as previous data suggest that systemic inflammation is a poor marker of the alveolar response.19–21 Indeed, expression profiling in tracheal aspirates of patients with severe COVID-19 has predicted that immunomodulation via corticosteroids would benefit the modulated host response in the lung.22 Yet, the alveolar host response in COVID-19 is understudied because alveolar samples are difficult to acquire.
In this study, we aimed to evaluate alveolar and systemic inflammatory responses in mechanically ventilated patients with COVID-19-related ARDS by measuring a comprehensive set of protein biomarkers in both bronchoalveolar lavage (BAL) fluid and blood plasma. We hypothesised that patients with COVID-19-related ARDS have a persistent alveolar inflammatory response and that such a response is associated with higher mortality. We also hypothesised that treatment with high-dose steroids (HDS), which is sometimes used in patients with ARDS who have persistent respiratory failure,23 is associated with decreasing concentrations of innate immune biomarkers that correlate with mortality. As we performed repeated sampling in patients who remained intubated, we did not only evaluate differences at baseline (as other studies did up to now), but also incorporated the dynamic nature of the inflammatory response in the analyses.
Methods
Study cohort
This observational cohort study was performed at Amsterdam University Medical Centers, locations AMC and VUmc, and has been described before.24 All patients with PCR-confirmed COVID-19 admitted to the ICU, who underwent a diagnostic bronchoscopy with BAL because of COVID-19-related ARDS between 12 June 2020 and 6 June 2021, were included. ARDS was defined according to the Berlin criteria.25 Details on the study cohort are described in the online supplemental methods. In brief, a diagnostic bronchoscopy with BAL was performed routinely once weekly if no respiratory improvement was observed, as determined by the treating clinicians. Per protocol, all patients with hypoxaemia received dexamethasone 6 mg once daily up to 10 days. When patients did not show respiratory improvement after completing their dexamethasone course, HDS treatment (ie, more or equal to 1 mg/kg prednisone) was considered. The decision to do a bronchoscopy or initiate HDS treatment was discussed during multidisciplinary team meetings and made by joint expert opinion.
Supplemental material
Within the study population, we defined two different cohorts to perform different kinds of analyses: the biomarker cohort (ie, all patients with reliable biomarker data) and the baseline cohort (ie, all patients with reliable biomarker data obtained in the first week of intubation).
Sampling and assays
During bronchoscopy, 4×20 mL 0.9% sodium chloride was inserted into single segments of the lungs. The second fraction was used for our analysis, as well as EDTA plasma obtained within 24 hours of the bronchoscopy. At the end of the study, a comprehensive set of 63 biomarkers representative of the main pathophysiological pathways in ARDS (online supplemental table 1) was analysed all at once in both BAL fluid and, if available (online supplemental table 2), paired plasma samples. The biomarkers were measured by Luminex multiplex assay (R&D Systems, Abingdon, UK), using the Bio-Plex 200 System (Bio-Rad Laboratories, Hercules, California, USA). In addition, a protein quantitation assay (Bio-Rad Laboratories, Hercules, California, USA) was performed to measure total protein concentrations in BAL samples. The sampling and assay procedures (including quality assessment (online supplemental tables 3 and 4)) and clinical data collection, collected until 90 days after intubation, are described in the online supplemental methods.
We performed all primary analyses with the log10-transformed absolute alveolar biomarker concentrations, that is, not corrected for total protein concentration. Systemic differences in the BAL fluid dilution factor were minimised by using a standardised bronchoscopy. As an additional method to correct for dilution effects, all analyses were repeated using total protein-corrected (ie, relative) alveolar biomarker concentrations.
Statistical analysis
To interrogate the alveolar inflammatory response, alveolar and plasma concentrations of the measured biomarkers were compared using a paired Wilcoxon signed-rank test using data from the first week of intubation (ie, baseline cohort). Only biomarkers with a higher concentration in BAL than in plasma with a p value of <0.05 after adjustment for multiple testing using the Benjamini-Hochberg false discovery rate were carried forward.
The persistency of the alveolar inflammatory response was assessed by modelling the dynamic changes in the absolute alveolar concentrations using joint model analysis (JM package),26 combining a linear mixed model and Cox proportional hazards model. This method is selected because it provides accurate estimates of time-related associations and because it handles informative censoring, relevant to this specific situation as patients who died are unable to get additional bronchoscopies and will not provide more data points.
The relationship between the host response and mortality was estimated at baseline and in dynamic analyses. In cross-sectional analyses, patient characteristics and protein biomarker differences between survivors and non-survivors were evaluated using a Student’s t-test, Mann-Whitney U test or Χ2 test, as appropriate. The aforementioned joint models were used to evaluate the association between changes in the absolute biomarker concentration and mortality, accounting for dynamic changes in biomarker concentrations during invasive ventilation. All joint models were corrected for age, body mass index (BMI) and the use of HDS at sampling time. To assess whether these associations were specific to the alveolar compartment, or also present in the systemic compartment, we repeated these joint models with plasma biomarker concentrations.
To explore the second hypothesis, changes in concentrations of the biomarkers related to mortality, as identified using joint model analyses, were compared between patients who did and did not receive HDS. Patients with a BAL measurement within 7 days both before and after HDS treatment initiation were selected as eligible cases, and patients without HDS therapy with two BAL measurements within a comparable time window were selected as eligible controls. Nearest-neighbour 1-to-1 propensity score matching was used to compose comparable study groups based on two matching criteria: number of days between ICU admission and BAL procedure, and sequential organ failure assessment (SOFA) score at the time of the BAL procedure (MatchIt package).27 Matching effectiveness was assessed by evaluating love plots and propensity score distribution plots. The association between HDS treatment and the longitudinal absolute biomarker levels was estimated using linear mixed-effects models (lme4 package).28 As secondary analysis, the linear mixed-effects models were repeated using the plasma biomarker concentrations.
Calculations were performed in R V.4.2.1 using the RStudio interface. P values of <0.05 were considered statistically significant. Details on patient selection and statistical analyses are provided in the online supplemental methods.
Results
Patient population
Between 12 June 2020 and 6 June 2021, 177 patients underwent a bronchoscopy with BAL and were screened for the study (figure 1). Four ineligible patients were excluded during screening. In 154 out of 173 patients with COVID-19 (89%), biomarker data could be obtained (ie, biomarker cohort) resulting in a total of 284 individual BAL samples and, when available, their paired plasma sample (online supplemental figures 1–3 and online supplemental table 2). Ninety-three patients had a BAL sample within the first week of invasive ventilation (ie, baseline cohort). Table 1 shows the demographics and clinical characteristics of both the baseline and biomarker cohort. Median age was 65 years (IQR 58–72) and 133 (73%) were male. The median time from ICU admission until intubation was 0 days (IQR 0–2). Sixty-two out of 154 patients survived to day 90 (40%). Median ventilator-free days was 0 (IQR 0–0) on day 28 and 0 (IQR 0–55) on day 90. Patients who died were older and had a lower BMI than survivors, but SOFA scores and routine clinical chemistry on the day of ICU admission were not different between survivors and non-survivors. Demographics and clinical characteristics are comparable between patients in the baseline cohort and patients in the biomarker cohort (online supplemental table 5).
Flow chart of screening and inclusion process; starting with all patients with COVID-19 admitted to the intensive care unit (ICU) of the Amsterdam University Medical Centers (AUMC) during the study period (12 June 2020 until 6 June 2021). Numbers of patients with numbers of BAL samples (per week after intubation) are displayed. The biomarker cohort consists of all patients with a good quality sample for analysis, and the baseline cohort consists of all patients with a BAL sample in the first week of intubation. BAL, bronchoalveolar lavage.
Demographics and clinical characteristics
Alveolar inflammatory response
We first determined the ratio between the alveolar and plasma concentrations at baseline (ie, all samples obtained within the first week of intubation). Biomarkers involved in the innate host response, including CXC motif ligand (CXCL)1, CXCL8, interleukin (IL)-10, tumour necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL), CC motif ligand (CCL)20, CCL2, CCL3, IL-1β, CCL4 and TNFα were found in higher absolute concentrations in the alveolar compartment compared with the systemic compartment (figure 2A)—indicative of a predominant alveolar inflammatory response. Matrix metalloproteinase (MMP)-8, angiopoietin (Ang-)2/Ang-1 and epidermal growth factor (EGF) also showed an alveolar-predominant response. In a sensitivity analysis focusing solely on samples obtained in the first 2 days of intubation, comparable results were found (online supplemental figure 4).
Several biomarkers that are higher in the alveolar compartment of patients with COVID-19 are associated with increased mortality. (A) Log2 fold change of BAL versus plasma absolute biomarker concentrations of all patients in the baseline cohort (n=93). Hypothesis testing was performed by a paired Wilcoxon signed-rank test. P values were adjusted for multiple testing using the Benjamini-Hochberg method. (B) The dynamic change in absolute alveolar biomarker concentration over time, as calculated by joint model analysis, corrected for age, BMI and the use of high-dose steroids at the day of sampling. (C) Association between increased absolute alveolar biomarker concentrations over time and 90-day mortality, of all patients in the biomarker cohort (n=154). Hypothesis testing was performed using the same joint model that was used in (B). All biomarker names and their abbreviations are shown in online supplemental table 1. BAL, bronchoalveolar lavage; BMI, body mass index.
We compared the baseline alveolar concentrations of the aforementioned biomarkers between patients with COVID-19 presented here and a historical non-COVID-19 ARDS cohort. Most biomarkers were comparable between the groups, with the exception of higher IL-10 and TRAIL concentrations in the COVID-19 cohort (online supplemental figure 5). Patient characteristics were unevenly distributed between the two groups; the non-COVID-19 ARDS group had more females, higher admission SOFA scores and higher prevalence of indirect ARDS causes (online supplemental table 6).
Repeated measurements were used to assess the longitudinal inflammatory response by evaluating the trajectories of the aforementioned biomarkers. Joint model analysis showed increasing absolute alveolar concentrations of several innate host response biomarkers (ie, IL-1β, CCL3, CXCL8, TRAIL, CCL20, CXCL1, CCL2 and CCL4) over time, suggesting a persistent alveolar inflammatory response (figure 2B). Fibroproliferative biomarker MMP-8 and growth factor EGF were also found to have increasing concentrations over time. The alveolar IL-10 concentration showed a decreasing trend, whereas Ang-2/Ang-1 and TNFα remained stable. These findings were replicated by modelling relative (protein-corrected) instead of absolute biomarker concentrations (online supplemental figure 6). When modelling the plasma biomarker concentrations, the observed trajectories showed complete distinct patterns compared with the trajectories observed in the alveolar compartment (online supplemental figure 7A,B).
Association between alveolar inflammation and mortality
We compared the baseline concentrations of the aforementioned biomarkers between survivors and non-survivors (online supplemental tables 7–10). After multiple testing correction and irrespective of the use of absolute and relative protein alveolar biomarker concentrations, or alveolar/plasma ratios, none of the biomarkers were significantly associated with mortality at baseline (online supplemental figure 8).
Within the biomarker cohort (including consecutive samples), we evaluated the association between persistent alveolar inflammation and clinical outcomes. Joint model analysis showed that persistently high absolute concentrations of innate immune biomarkers CCL20, CXCL1, TRAIL, CCL2, CXCL8, CCL3 and IL-10, and of EGF and MMP-8 were associated with increased mortality (figure 2C). TNFα, IL-1β, CCL4 and Ang-2/Ang-1 did not show a significant relationship between the longitudinal alveolar concentration and mortality. These results remained largely the same in different sensitivity analyses (online supplemental figure 9), including using relative protein concentrations and using different subsets of patients: only patients in the baseline cohort, only patients who received HDS and only samples obtained in the first 14 or 28 days. In the subset of patients who received HDS, the association between mortality and increasing biomarker levels was only significant for CCL20, CXCL1, CCL2 and CXCL8 (online supplemental figure 9C). To evaluate whether these associations were specific to the alveolar compartment, these analyses were repeated using plasma biomarker concentrations (online supplemental figure 7C,D). In plasma, CCL20, CCL2, CXCL8, MMP-8 and IL-10 showed similar positive associations with mortality, whereas TRAIL and EGF showed opposing effects.
Association between HDS treatment and alveolar biomarker levels
HDS for the treatment of ARDS was given to 72 patients (47%) during their ICU stay, whereas 82 patients (53%) did not receive HDS (online supplemental figure 10). The median time between intubation and initiation of HDS treatment was 13 days (IQR 9–18). The median duration of HDS treatment was 9 days (IQR 6–10). The propensity score-matched cohort consisted of 18 patients with, and 18 patients without HDS treatment (online supplemental figure 11). The matching covariates (ie, days to BAL procedure and SOFA score at the time of the BAL procedure) were well balanced after matching (online supplemental figure 12). Patient characteristics, sampling times, microbiological outcomes (ie, the development of secondary infections) and clinical outcomes were comparable between the two treatment groups (table 2). The patients used in this analysis were a representative sample of the biomarker cohort, with the exception of longer total duration of mechanical ventilation, but this was matched between cases and controls (online supplemental table 11). In addition, more patients were treated with antifungals in this analysis, although the amount of proven fungal infections did not differ between the groups (online supplemental table 11). Online supplemental table 12 shows that the HDS subgroup selected for this analysis was a representative sample of all patients who received HDS.
Demographics and clinical characteristics of patients in the exploratory analysis on the association between HDS and biomarker levels (no HDS vs HDS), focusing on two consecutive samples within each patient in a matched cohort
HDS-treated patients showed a significant decrease in the absolute alveolar concentration of CXCL1 (log10 decrease of 0.07 per day; 95% CI 0.01 to 0.12) and CCL20 (log10 decrease of 0.10 per day; 95% CI 0.01 to 0.19) after the start of HDS treatment (figure 3), while untreated patients showed a gradual increase. No significant differences between the two treatment groups were found in the longitudinal concentrations of the other biomarkers (online supplemental figure 13). In the systemic compartment, a significant decrease after HDS initiation was only observed in the plasma concentrations of CCL20, CXCL8 and CCL2 (online supplemental figure 14).
CXCL1 and CCL20 show a decrease in alveolar concentration after high-dose steroids (HDS) therapy in patients with COVID-19-related acute respiratory distress syndrome. Blue and green lines indicate the trajectories of absolute alveolar biomarker concentrations for patients treated with HDS (n=18), before and after initiation of HDS therapy, and matched controls without HDS therapy (n=18). Patient selection is shown in online supplemental figure 11. Linear mixed-effects models were used to provide p values. Dotted line indicates the 95% CI. CCL, CC motif ligand; CXCL, CXC motif ligand.
Discussion
In patients with COVID-19-related ARDS, we here provide evidence for an alveolar hyperinflammatory state. Alveolar concentrations of the measured biomarkers were not associated with mortality at baseline, but a persistently increased alveolar concentration of CCL20, CXCL1 and several other biomarkers related to neutrophil and macrophage activation was associated with a higher mortality risk. Patients who received HDS treatment showed a significant decrease in the alveolar CCL20 and CXCL1 concentrations.
Unique to our study, we evaluated the time-dependent changes in the alveolar inflammatory response and demonstrated a neutrophil-related persistent alveolar inflammatory response to be associated with a higher mortality risk, which appeared to be mostly specific to the alveolar compartment. These findings are in line with the established association between persistent alveolar hyperinflammation and higher risk of mortality in sepsis-induced ARDS.29 Moreover, our results support previous findings that a neutrophil-predominant bronchoalveolar phenotype is associated with a poor 28-day outcome in critically ill patients with COVID-19.30 Taken together, the data suggest that a subset of patients with COVID-19-related ARDS have persistent alveolar inflammation, which is related to an increased mortality rate.
The difference in alveolar to systemic inflammatory response observed in this study may even be underestimated in our results since the BAL samples are—inherent to the procedure—diluted epithelial lining fluid. Consistent with our findings, previous studies have shown that the alveolar host immune response in COVID-19 has unique local profiles that strongly differ from those in the peripheral blood.31 32 For example, COVID-19-infected alveolar macrophages have been proven to be part of a positive feedback loop with T cells by producing inflammatory cytokines that promote T cell activation.33 T cell-related cytokines were also shown to be elevated and to display increased cellular expression in the alveolar space in COVID-19, thereby further contributing to a strong inflammatory response in the alveolar space.34 35 Considering these findings and the presented data here, we highlight the importance of the bronchoalveolar inflammatory response in the pathophysiology of COVID-19.18 21 32 36 37
Because HDS could theoretically help to diminish alveolar inflammation and might facilitate recovery from persistent ARDS,23 we evaluated the association between HDS treatment and alveolar biomarker concentrations. In this exploratory analysis, we found that HDS treatment was associated with a decrease in alveolar CCL20 and CXCL1 concentrations, which are both proinflammatory cytokines of the innate immune system and play a role in neutrophil recruitment.7 38 Data on the role of CXCL1 and CCL20 in the pathophysiology of (COVID-19) ARDS are limited.39 CXCL1 is a CXC chemokine, structurally related to the chemokine CXCL8 (also called IL-8).40 41 CXC chemokines, including CXCL1 and CXCL8, are responsible for neutrophil recruitment to the lungs during lung injury7 and are elevated in patients with ARDS.42 Our data are therefore consistent with those of a previous randomised controlled trial (RCT) looking into the effect of HDS treatment in patients with ARDS not related to COVID-19, which showed decreasing neutrophil counts in the alveolar environment.23 Based on the presented data, HDS could possibly be a promising therapy in a subset of patients with COVID-19 with persistent alveolar inflammation.
The most important strengths of this study are the size of the cohort, the longitudinal data analysis using joint modelling and the inclusive and quantitative protein analysis. Inherent to any study relying on repeated alveolar sampling, there are limitations with regard to selection of patients, dilution of the sample and unavailability of follow-up samples. We tried to limit these issues by using protocolised indications for bronchoscopy to reduce selection bias, but our data cannot be generalised to patients with COVID-19 without indication for bronchoscopy (ie, rapid ARDS resolvers). This is reflected by the high mortality rates and low number of ventilator-free days, compared with study cohorts of patients with severe ARDS43 and severe COVID-19 ARDS.3 We recognise that there are many patients with resolving lung injury and good clinical outcomes that are not included in this analysis. Even though the alveolar profiles of non-COVID-19 ARDS controls showed large overlap with the COVID-19 cohort, we acknowledge that our findings may not be fully applicable to the entire population of patients with ARDS. Around one in four patients also received immunomodulators, such as IL-6R inhibitor tocilizumab, but we could not account for this exposure in the analyses.
Strict standardisation of the procedures was only possible because the study was performed in two academic hospitals in Amsterdam that shared the same protocol, which may limit external validity. We aimed to limit BAL dilution effects as much as possible by standardised volume instillation, and by replicating all analyses using protein-corrected concentrations as sensitivity analysis. However, we do recognise that there is no gold standard for correction of dilution effects.44 45 Joint model analysis was used to reduce the influence of informed censoring on the association between biomarker concentrations and outcome. Of note, for a high percentage of samples, CXCL8 was above the detection limit, creating challenges in modelling. Future studies using serial dilutions are required to further investigate the role of CXCL8. Even though we selected statistical methods to account for sample timing variability, we cannot exclude that this influenced the study results. We only analysed alveolar proteins in the supernatant and we do not report the cellular composition of the samples. Although we realised that such data would be valuable, we were unable to perform these analyses due to logistical constraints related to the pandemic.
The key implication of this study is that we ought to focus more on the inflammatory response in the bronchoalveolar space in patients with COVID-19-related ARDS, rather than focus on the systemic host response alone due to the simplicity of sampling this compartment. Our results suggest that a persistent alveolar neutrophil signature could potentially be targeted by HDS treatment. Although persistently high alveolar CCL20 and CXCL1 concentrations were associated with higher mortality and HDS treatment was associated with lower levels of these markers, mortality benefit could not be studied in this observational study. This implies the need for RCTs examining the use of HDS in ARDS with persistent alveolar inflammation. To engage in precision medicine, our data suggest focusing on patients with increased neutrophil-related biomarker concentrations rather than an unselected population to allow for predictive enrichment. Future studies are required to investigate predictive enrichment in the setting of COVID-19-related ARDS and HDS.
To conclude, we show here that COVID-19-related ARDS is associated with an alveolar inflammatory response, and that the dynamics of several biomarkers related to neutrophil recruitment are associated with increased mortality. Treatment with HDS was associated with a decrease in alveolar CCL20 and CXCL1 concentrations.
Data availability statement
Data are available upon reasonable request. De-identified participant data with data dictionary can be shared after approval of a proposal with a signed data access agreement and always in collaboration with the study group.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and the ethical boards of the participating hospitals approved the collection of data for the study purposes (ID AUMC 2020_065). Informed consent was deferred until ICU discharge. Patients were prospectively included if they provided written informed consent or if the patient or relatives did not use the opt-out form. The study procedure was approved by the Review Committee of the Amsterdam UMC Biobank and was performed in accordance with the Declaration of Helsinki and adheres to Dutch regulations.
Acknowledgments
The authors would like to thank Barbara Smids-Dierdorp and Tamara Dekker for their expert technical assistance in the performance of the Luminex multiplex assays, Stijn Klarenbeek, Fayola de Lange and Jacky de Leeuw for their help in the collection of the BAL samples, and Laura van den Heuvel and Tom Zwaan for the collection of clinical data.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
JdB and LSB are joint first authors.
JD and LDJB are joint senior authors.
JdB and LSB contributed equally.
Collaborators The ArtDECO consortium (Amsterdam UMC location University of Amsterdam and location Vrije Universiteit Amsterdam, Amsterdam, the Netherlands): E J Nossent, J W Duitman, A Saris, H de Vries, L J Meijboom, L D J Bos, S G Blok, A R Schuurman, T D Y Reijnders, F Hugenholtz, J J Garcia Vallejo, H Bontkes, A P J Vlaar, W J Wiersinga, R Lutter, T van der Poll, H J Bogaard and L Heunks. The Amsterdam UMC COVID Study group (Amsterdam UMC location University of Amsterdam and location Vrije Universiteit Amsterdam, Amsterdam, the Netherlands): M van Agtmael, A G Algera, B Appelman, F Baarle, M Beudel, H J Bogaard, M Bomers, P Bonta, L D Bos, M Botta, J de Brabander, G de Bree, S de Bruin, M Bugiani, E Bulle, D T P Buis, O Chouchane, A Cloherty, M Dijkstra, D A Dongelmans, R W G Dujardin, P Elbers, L Fleuren, S Geerlings, T Geijtenbeek, A Girbes, B Goorhuis, M P Grobusch, L Hagens, J Hamann, V Harris, R Hemke, S M Hermans, L Heunks, M Hollmann, J Horn, J W Hovius, M D de Jong, R Koning, E H T Lim, N van Mourik, J Nellen, E J Nossent, S Olie, F Paulus, E Peters, D A I Pina-Fuentes, T van der Poll, B Preckel, J M Prins, J Raasveld, T Reijnders, M C F J de Rotte, M Schinkel, M J Schultz, F A P Schrauwen, A Schuurman, J Schuurmans, K Sigaloff, M A Slim, P Smeele, M Smit, C S Stijnis, W Stilma, C Teunissen, P Thoral, A M Tsonas, P R Tuinman, M van der Valk, D Veelo, C Volleman, H de Vries, L A Vught, M van Vugt, D Wouters, A H Zwinderman, M C Brouwer, W J Wiersinga, A P J Vlaar and D van de Beek. The BASIC consortium (Amsterdam UMC location University of Amsterdam, Amsterdam, the Netherlands): F M de Beer, L D J Bos, T A Claushuis, G J Glas, J Horn, A J Hoogendijk, R T van Hooijdonk, M A Huson, M D de Jong, N P Juffermans, W A Lagrand, T van der Poll, B Scicluna, L R Schouten, M J Schultz, K F van der Sluijs, M Straat, L A van Vught, L Wieske, M A Wiewel and E Witteveen.
Contributors JdB, LSB, EJN, LMAH, APJV, PIB, MJS, TvdP, JD and LDJB designed the study. JdB, LSB and RFJK were involved in collecting the data. JdB, LSB, JD and LDJB had access to the raw data, did the analyses and drafted the manuscript. RFJK, SZ, EJN, LMAH, APJV, PIB, MJS and TvdP provided intellectual input and revised the initial draft. All authors and the collaborators approved the final version of the manuscript. LDJB was responsible for the overall content as the guarantor. JdB and LSB contributed equally and shared first authorship. JD and LDJB contributed equally and shared last authorship.
Funding This research was funded by the Netherlands Organization for Scientific Research (NWO) under VENI grant 016.1860.046 to JD and by an Amsterdam UMC fellowship to LDJB in 2020 (no award/grant number).
Competing interests LDB is supported by a research fund from the Amsterdam UMC and JD by a research fund from the Netherlands Organization for Scientific Research (NWO).
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.