Cytological patterns of bronchoalveolar lavage fluid in mechanically ventilated COVID‐19 patients on extracorporeal membrane oxygenation

Abstract Introduction Coronavirus disease‐2019 (COVID‐19) may lead to acute respiratory distress syndrome requiring extracorporeal membrane oxygenation (ECMO). Patterns of inflammatory bronchoalveolar cells in COVID‐19 patients treated with ECMO are not well described. Objective We aimed to describe inflammatory cell subpopulations in blood and bronchoalveolar lavages (BALs) obtained in critically ill COVID‐19 patients shortly after ECMO implementation. Methods BAL was performed in the middle lobe in 12 consecutive ECMO‐treated COVID‐19 patients. Trained cytologists analyzed peripheral blood and BAL cells using flow cytometry and routine staining, respectively. Data were interpreted in relation to dexamethasone administration and weaning from ECMO and ventilator. Results High neutrophil proportions (66% to 88% of total cells) were observed in the absence of bacterial superinfection and more frequently in dexamethasone‐free patients (83% [82–85] vs. 29% [8–68], P = 0.006), suggesting that viral infection could be responsible of predominantly neutrophilic lung inflammation. Successful weaning from ECMO/ventilator could not be predicted by the peripheral white blood and BAL cell pattern. Conclusion High neutrophil proportions can be observed in critically ill COVID‐19 patients despite the lack of microbiological evidence on BAL of bacterial superinfection. Dexamethasone was associated with lower neutrophil proportions in BAL. Our study was probably underpowered to provide BAL cell pattern helpful to predict weaning from ECMO/ventilator.

K E Y W O R D S ARDS, bronchoalveolar lavage, COVID-19, ECMO, lymphocyte, neutrophil 1 | INTRODUCTION Acute respiratory distress syndrome (ARDS) is a lifethreatening complication of coronavirus disease-2019 (COVID-19) pneumonia, which may require veno-venous extracorporeal membrane oxygenation (ECMO) in addition to optimized mechanical ventilation. 1,2 Dexamethasone was shown to reduce 28-day mortality among mechanically ventilated COVID-19 patients. 3 Therefore, investigating inflammatory cells in blood and lungs appears paramount to understand the overall process involved in the disease progression and how it is altered by steroid treatment. 4,5 Bronchoalveolar lavage (BAL) cells in COVID-19-attributed pneumonia are relatively well described. [6][7][8][9] However, description of BAL cell patterns and alterations resulting from dexamethasone administration in patients on ECMO remain scarce. Therefore, we designed this observational exploratory study (i) to describe inflammatory cell subpopulations in blood and BALs obtained in COVID-19 patients shortly after ECMO implementation and as far as possible (ii) to seek any pattern associated with the successful weaning from ECMO and ventilator.

| METHODS
All consecutive ECMO-treated COVID-19 patients with BAL admitted to our intensive care unit from March to May 2020 were included. ECMO was implemented in patients with refractory hypoxemia and/or hypercapnia despite muscular paralysis, optimized ventilation, and prone position, as recommended. 2 Ventilator and ECMO settings were adjusted to maintain PaO 2 ≥ 60 mmHg and sweep adapted to keep PaCO 2 in the normal range if possible. Dexamethasone was administered as 10 mg twice daily for 5 days followed by 10 mg daily for 5 days, 10 a higher dose regimen used before the Recovery trial. 3 This study was part of the COVID-ICU and French COVID-19 cohort registries and approved by our institutional ethics committee (IDRCB, 2020-A00256-33; CPP, 11-2020.02.04.68737).
BAL was performed in the middle lobe using ≥120 ml saline by trained pulmonologists and intensivists. Smears were analyzed microscopically after Papanicolaou, May-Grünewald Giemsa, Ziehl, Grocott, and Perls staining by one experienced pathologist blinded to the patient treatments. Cell count in BAL fluid (BALF) was expressed as percentages of the total calculated on a 300-cell sample. Tracheal aspirates were sampled every 24 or 48 h to detect bacterial infection. Within 24 h of BAL, peripheral white blood cells counts and CD3+, CD4+, CD8+, CD19+, CD16+, and CD56+ lymphocyte subsets were quantified using FACS-Canto™ flow cytometry (Becton Dickinson, Erembodegem, Belgium). Clinical and routine biological data were collected from the patients' records.
Results are presented as median [percentiles 25th-75th] for the quantitative parameters and percentages for categorical variables. Data were compared using Mann-Whitney and Fischer's exact tests, as appropriate. P values <0.05 were considered significant.
Clinical characteristics and biological parameters did not differ between the patients who were weaned from ECMO/ventilator and those who were not (Table 1). No significant differences were found in peripheral white blood cells or lymphocyte subpopulations in relation to successful weaning (Table 2).

| DISCUSSION
One major finding in this exploratory study is that BAL with >50% neutrophil count was not necessarily associated with evidence on BAL of bacterial superinfection in severe COVID-19 patients despite the high prevalence of pulmonary infections in ICU COVID-19 patients including those on ECMO, [11][12][13] and may thus just be the effect of the viral infection itself. This is important, as in patients with high BAL neutrophil count clinicians may strongly suspect bacterial superinfection and delay steroid initiation. Previously, BAL specimens mainly showed marked lymphocytosis with activated plasma cells. 6,7 Another important finding was that patients who received dexamethasone before BAL showed significantly lower BAL neutrophil proportions than dexamethasonefree patients, suggesting that dexamethasone might limit neutrophil recruitment. Whether this effect may contribute to dexamethasone-induced improvement in survival 3 remains to be established. Altogether, our results may have implications regarding the immunomodulatory therapy effects if confirmed in future studies. Nevertheless, as expected, the cytological pattern of BAL in our T A B L E 2 Ventilation and ECMO parameters, bronchoalveolar lavage fluid, and blood leukocyte subpopulations in 12 critically ill COVID-19 patients treated with optimized mechanical ventilation and veno-venous extracorporeal membrane oxygenation study did not provide predictive information on the possible successful weaning from ECMO/ventilator, which is a multifactorial outcome not easy to predict on the first days of ECMO. Limitations of our study include the single-center setting and the relatively small number of patients, which may have precluded evidencing significant differences in BAL neutrophil percentage between weaned and nonweaned patients or identifying plasma IL-6 thresholds categorizing patient severity, as previously reported. 14 Therefore, larger cohort studies are necessary to better explore these aspects. However, to the best of our knowledge, this is the first study to investigate BAL cells in ECMO-treated COVID-19 patients; a single case has been previously published. 6 Logistical difficulties related to the pandemic precluded a more complex design.
In conclusion, in this exploratory study, high neutrophil proportion in BALF can be observed in severely SARS-Cov2-infected patients with no microbiological evidence on BAL of bacterial superinfection at the time of BAL. Dexamethasone administration is associated with a decrease in BAL neutrophil proportion, a finding that needs further investigation.