This retrospective project was approved by the review board for the human ethics committee of Sao Paulo University (CAPPesq-FMUSP; CAAE: 67771417.0.0000.0068).

We have selected 47 patients with a clinical diagnosis of ARDS as defined by the Berlin definition [9], in addition to histological findings of DAD. We further selected 30 patients with clinical diagnosis of sepsis defined according to Singer et al. [10] and without clinical criteria for ARDS. For both groups, we excluded individuals with a previous history of smoking and/or chronic lung disease and individuals that the medical records were not available and that laboratory exams necessary for the diagnosis (e.g., serum lactate, cultures, arterial blood gas analysis, and lung image exams) were not available or not performed.

As controls, we selected 27 patients who died from non-pulmonary causes, without a previous history of smoking and/or chronic lung disease and/or mechanical ventilation, with preserved lung tissue at histological analysis.

All patients had their lung tissue collected during the routine autopsy performed at the Sao Paulo Autopsy Service – University of Sao Paulo (SVOC-USP) between 2002 and 2014. To better represent the lung tissue, we have selected two to three post-mortem lung samples of each case, avoiding areas of abscess and/or necrosis and/or additional relevant tissue destruction which would compromise all the immunostainings and analysis as blocks with limited amount of tissue.

Clinical data were retrospectively collected from the medical records during the hospital admission period and laboratory exams performed 24 h prior to death. Since the control individuals died mostly from sudden deaths, we did not had access to their laboratory exams.

We collected the following clinical information: duration of hospitalization, mechanical ventilation (MV), ICU admission length, Sequential Organ Failure Assessment (SOFA) score, and information regarding mechanical ventilation, such as the fraction of inspired oxygen (FiO2), positive end-expiratory pressure (PEEP), partial pressure of arterial oxygen (PaO2) to FiO2 ratio. We also collected information regarding the arterial blood gas analysis such as arterial blood pH, PaO2, partial pressure of carbon dioxide (PaCO2), bicarbonate (HCO3), base excess, oxygen saturation (SO2), the fraction of oxyhaemoglobin (FO2Hb), Fraction of carboxyhaemoglobin (FCO2Hb), Fraction of methaemoglobin (FMetHb), Fraction of deoxyhaemoglobin (FHHb), and the oxygen tension at which haemoglobin is 50% saturated (p50).

We collected information about the full blood count in addition to serum lactate and C-reactive protein (CRP).

Slides stained with haematoxylin and eosin (H&E) were blinded and evaluated by an experienced pathologist who quantified the proportion (%) of the following histological patterns: normal lung tissue, exudative DAD, fibroproliferative DAD, and acute bronchopneumonia. The histological criteria used were: (a) normal tissue: lung parenchyma with normal histology or minimal non-specific changes as mild oedema and congestion; (b) exudative DAD: interstitial and/or intra-alveolar oedema, interstitial inflammation, variable amounts of alveolar haemorrhage and fibrin deposition, intra-alveolar hyaline membranes and type II pneumocyte hyperplasia; (c) fibroproliferative DAD: any degree of fibroblastic proliferation within the interstitium and/or alveolar spaces, including loose aggregates of fibroblasts admixed with scattered inflammatory cells and collagen deposition, intermingled with areas with hyaline membranes or densely fibrotic areas [11]. To further explore the features of each DAD pattern, each slide was scored for septal thickening, oedema, inflammation, hyaline membrane, alveolar haemorrhage, and proliferation of type II pneumocytes, with the following graduation: 0- absent, 1-mild, 2-moderate, and 3- severe. The cases were also classified by the type of inflammation: 0- Absent, 1- Predominantly neutrophilic inflammation, 2- Predominantly lymphomononuclear inflammation, and 3- Mixed inflammation [12, 13].

Lung tissue was immunostained using anti-elafin (Abcam, UK; cat. #ab81681; 1:300), anti-RAGE (Santa Cruz Biotechnology, Dallas, TX, USA; cat. #sc-365,154; 1:300), and anti-SP-D (Santa Cruz Biotechnology, Dallas, TX, USA; cat. #sc-25,324; 1:750). We also stained the lung slides with Sirius Red for collagen quantification. The epithelial damage markers and collagen were quantified in the lung septa, excluding airways, large blood vessels (only capillaries were not excluded), pleura, and loose cells in the alveolar space [14]. Positive-stained area per septa length (µm²/µm) was measure measured in 20 high-power fields using the Image-Pro Plus 4.1 software (Media Cybernetics, Silver Spring, MD, USA).

SPSS 23 software (SPSS Inc/IBM Chicago, USA) was used for the statistical analyses. GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA) and RStudio, version 4.1.1 (RStudio, PBC, Boston, MA, USA) were used for data visualization.

Categorical variables were analysed using the chi-square test and simple correspondence analysis (ANACOR). We further analysed the dependency relationships between each pair of categories using the adjusted standardized residuals of the chi-square test, adopting an alpha value of 0.05.

Quantitative variables distribution was assessed by the Shapiro-Wilk normality test. Depending on the data distribution, T-student or Mann-Whitney tests were used to compare two groups and Kruskal–Wallis or one-way ANOVA tests, followed by the Bonferroni or Tukey posthoc test to compare four groups. Statistical difference was assumed at the 5% significance level. The coefficient of variation (CV) was calculated for every case by dividing the standard deviation by the mean. We performed the Spearman correlation test between variables; coefficients (r) were considered statistically significant at p < 0.05.

As a strategy of data analysis, we compared all histological variables among the 3 groups: Control, Sepsis, and ARDS. We also compared the sepsis group to the subgroups formed by the division of the ARDS group according to its severity: Mild: 200 mmHg < PaO2/FiO2 ratio ≤ 300 mmHg, Moderate: 100 mmHg < PaO2/FiO2 ratio ≤ 200 mmHg, and Severe: PaO2/FiO2 ratio ≤ 100 mmHg. Within the ARDS group, we compared the pulmonary ARDS and extrapulmonary ARDS. Data from the laboratory exams were compared between the ARDS and sepsis groups and correlated with histological variables.