Patient cohort, sampling and data collection

22 COVID-19 patients and 13 non-COVID-19 pneumonia patients in this study were enrolled from the University Hospitals Leuven, between March 31st 2020 and May 4th 2020. Disease severity was defined as ‘mild’ or ‘critical’, based on the level of respiratory support at the time of sampling. Specifically, ‘mild’ patients required no respiratory support or supplemental oxygen through a nasal cannula, whereas ‘critically ill’ patients were mechanically ventilated or received extracorporeal membrane oxygenation.

The demographic and disease characteristics of the prospectively recruited patients studied by scRNA-seq are listed in Supplementary information, Table S1. Diagnosis of COVID-19 was based on clinical symptoms, chest imaging and SARS-CoV-2 RNA-positive testing (qRT-PCR) on a nasopharyngeal swab and/or BAL fluid sample. Non-COVID-19 pneumonia cases all tested negative for SARS-CoV-2 RNA using a qRT-PCR assay on BAL.

All 35 patients underwent bronchoscopy with BAL as part of the standard of medical care, because of (i) high clinical suspicion of COVID-19 yet negative SARS-CoV-2 qRT-PCR on nasopharyngeal swab (ii) established COVID-19 with clinical deterioration, to rule out opportunistic (co-)infection and/or to remove mucus plugs. Lavage was performed instilling 20 mL of sterile saline, with an approximate retrieval of 10 mL. 2–3 mL of the retrieved volume was used for clinical purposes. The remaining fraction was used for scRNA-seq.

The retrieved BAL volume was separated into two aliquots, as explained above, by the performing endoscopist. The aliquot used for scRNA-seq was immediately put on ice and transported to a Biosafety Level 3 Laboratory (REGA Institute, KU Leuven) for scRNA-seq.

Demographic, clinical, treatment and outcome data from patient electronic medical records were obtained through a standardized research form in Research Electronic Data Capture Software (REDCAP, Vanderbilt University). A CT score from each patient was calculated by converting the percentage of lung parenchyma opacity for each lobe into a 5-points Likert scale (a score of 0 for 0% lung opacity (LO), 1 for 1% to < 5% LO, 2 for 5%–25% LO, 3 for 26%–50% LO, 4 for 51%–75% LO, and 5 for 76%–100% LO). The total CT score was the sum of the individual lobular scores and ranged from 0 (no area with increase in lung opacity) to 25 (all five lobes show more than 75% increase in lung opacity).

This study was conducted according to the principles expressed in the Declaration of Helsinki. Ethical approval was obtained from the Research Ethics Committee of KU / UZ Leuven (S63881). All participants provided written informed consent for sample collection and subsequent analyses.