Broncho-alveolar inflammation in COVID-19 patients: a correlation with clinical outcome

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), detected in Wuhan (China) in December 2019 [1], spread rapidly around the world, reaching pandemic proportions in March 2020. In Italy, from February up to June 17th, 237,500 SARS-CoV-2-positive adults have been documented with 34,405 deceased patients (http://​www.​salute.​gov.​it/​portale/​nuovocoronavirus​/​dettaglioNotizie​NuovoCoronavirus​). Since December 2019, researchers and clinicians have been trying to find out the pathogenic features of the new disease caused by SARS-CoV-2, named COVID-19, in order to develop an effective therapy. The typical clinical manifestation is severe pneumonia with hyperactivation of the inflammatory cascade and progression to acute respiratory distress syndrome (ARDS) in about 30% of patients [2‐4]. As of yet, most research papers have focused on the inflammatory status at the plasma level of COVID-19 patients [3‐7]. However, because the main target organ is the lung, it is crucial to understand the inflammatory status at the deep lung level during different stages of the infection. Currently, limited data are available about alveolar inflammatory status in COVID-19 patients because of concerns in relation to using bronchoscopy to avoid aerosol generation.

Herein, we aim to define innate inflammatory environment by the analysis of BAL cell differentials, ultrastructural cell analysis and cytokine profile in patients with severe COVID-19, comparing them with patients admitted to the intermediate medicine ward (IMW) enrolled by two major COVID-19 centers of Italy: Luigi Sacco Hospital (Milan, Italy) and the IRCCS Policlinico San Matteo Foundation Hospital (Pavia, Italy). Our hypothesis is that alveolar inflammation is mainly driven by innate immune effectors and correlate with clinical outcome of patients.

From March 7 to April 30, a total of 45 BAL samples were collected from 5 IMW and 28 ICU patients with proven COVID-19 infection. Patients characteristics are shown in Table 1, and data concerning administered treatment in Table 2. All patients from the IMW ward survived. Significant differences regarding laboratory abnormalities among ICU and IMW patients, when present, are shown in Table 1.

Firstly, we assessed BAL cell differentials in ICU and in IMW patients that presented a median of total cells of 0.25 × 10⁶ ml^− 1 (0.15–0.52) and 0.3 × 10⁶ ml^− 1 (0.15–0.52), respectively (p > 0.999). A different leukocyte profile between ICU and IMW patients was observed (Fig. 1a and b). In particular, ICU patients showed a marked prevalence of neutrophils (1.24 × 10⁵ ml^− 1, 0.85–2.07) with a lesser extent of macrophages (0.43 × 10⁵ ml^− 1, 0.34–1.62) and lymphocytes (0.97 × 10⁵ ml^− 1, 0.024–0.34) as compared to IMW patients, who had median macrophages at 2.14 × 10⁵ ml^− 1 (1.17–3.01), lymphocytes at 0.47 × 10⁵ ml^− 1 (0.28–1.01) and neutrophils at 0.095 × 10⁵ ml^− 1 (0.05–0.73).

IMW patients presented a significant decrease of macrophages (p = 0.029) and a slight, but not significant, increase of lymphocyte (p = 0.09) and neutrophils (p = 0.11) with respect to normal reference values from our lab (data not shown) [10].

Dividing all patient samples into survivors and non-survivors, only for lymphocytes we observed a significant difference (p = 0.03) between two groups, where non-survivors showed less percentage of lymphocytes compared to survivors (Fig. 1e). Regarding macrophages and neutrophils (Fig. 1c and d), even if there was a trend towards a lower macrophage fraction among non-survivors, they did not reach a significant difference between two groups of analysis.

Ultrastructural study of ICU BAL samples showed many cytopathic cells, with loss of integrity of cell membranes and cytoplasm vacuolization (Fig. S1a-d). Mononuclear cells and neutrophils were detected with morphological features of activation (Fig. 2a and b). Viral particles, both single and in small clusters, were identified in numerous cells, particularly in cytopathic epithelial cells and mononuclear cells, with morphology (spikes) and size (80–120 nm) consisting with coronavirus (Figs. S1a-d and S2), confirmed by immunohistochemistry (Fig. S1e, f and Fig. 1c, d). Infected VERO E6 cells used as positive control showed large amounts of clustered and single viral particles clustered (Fig. S3).

The following cytokines: IL6, IL8 and IL10 were quantified on cell-free BAL by ELISA. Figure 3a shows the different distribution of three cytokine levels between ICU and IMW patients. Pro-inflammatory cytokines, IL6 and IL8, were significantly higher in ICU than in IMW patients (IL6 p = 0.006 and IL8 p < 0.0001), while IL10 did not show a significant difference between groups (p = 0.249). When analyzing cytokines according to outcome, higher IL6 (Fig. 3b, p = 0.027) and IL8 (Fig. 3c, p = 0.05) BAL levels were detectable in non-survivors with respect to survivors, with no difference for IL10 (Fig. 3d).

Variations of cytokine BAL levels according to treatment strategies were also evaluated. We decided to consider only BAL samples derived from patients who started specific treatment within 7 days prior to BAL sampling in order to assess the potential influence of treatment on cytokine levels. Moreover, considered patients were not submitted to other treatment regiments before BAL sampling. Due to the limited sample size, we restricted the analysis to tocilizumab, steroids and antiviral treatments (remdesivir, lopinavir/ritonavir) also including in this last group, hydroxychloroquine. Figure 4 shows that we observed a significant difference between different groups only in the case of IL6, in particular in patients treated with steroids that showed lower BAL IL6 compared to those who received or antiviral drugs (p = 0.030) (Fig. 4a).

Finally, we investigated whether BAL cytokines correlated with differential cell counts quantified previously (Table S1) or between each other (Table S2), considering all collected BAL samples. We found that IL6 and IL8 correlated directly with neutrophils (Table S1 and Fig. 5b) and inversely with macrophages percentage (Table S1 and Fig. 5a). We observed that the level of IL8 also correlated inversely with lymphocytes percentage (Table S1). Analyzing the correlation between cytokines, levels of IL6 and IL8 were correlated between each other, while IL10 correlated inversely with IL8 (Table S1). These correlations between cytokines are clear in two representative cases of patients subjected to repeated bronchoscopies (Fig. S4).

To have a preliminary observation about pulmonary inflammation and its relationship to serum inflammation status, we analyzed correlation between CRP/IL6 BAL and IL6 serum/IL6 BAL. Regarding CRP/IL6 BAL we did not observe any direct correlation (spearman coefficient = − 0.038; p = 0.83). In contrast, for patients from whom we reported IL6 serum and BAL quantification sampling on the same day (20 from ICU and 4 from IMW), we analyzed the correlation and observed a direct correlation (Spearman coefficient = 0.53 with p < 0.01) (Fig. S5).

Up to now, there has been little information about the alveolar lung inflammatory status in SARS-CoV-2 patients, since most of the published studies assessed the inflammatory status of COVID-19 patients at peripheral level (plasma or serum) level [2‐7]. In the present study we analyzed the broncho-alveolar inflammatory environment of patients admitted to the ICU and IMW of two hospitals situated in the epicenter of the Italian epidemic outbreak. In particular, we evaluated cell differentials, cell activation by morphological features, and cytokines profile on retrieved BAL fluids. Among cytokines, we decided to focus our attention on IL6, a well-known inflammatory cytokine that has been used as a marker of macrophages activation in the peripheral blood of COVID-19 affected patients [11]; IL8, a pro-inflammatory CXC chemokine that exerts specific chemotactic and activating functions on neutrophils; and IL10, an anti-inflammatory and immunoregulatory cytokine, produced by several immune effectors, whose role in SARS-CoV-2 infected patients is still in debate [12].

Thanks to our study, we observed that in severe ICU COVID-19 patients the alveolitis was associated to an hyperactivation of innate effectors, macrophages and neutrophils, that showed pseudopodia and cell-to-cell contacts (Fig. 2a and b). Moreover, a large number of cytopathic cells with vacuolization, osmiophilic bodies, loss of integrity of both nuclear and plasma membranes and cell fragmentation were present (Figs. S1 and S2 and Fig. 2). Cell differential analysis confirmed what was expected after peripheral blood analysis. In ICU patients there is an excessive infiltration of neutrophils at alveolar level, in contrast to IMW patients (Fig. 1a and b). This result is confirmed by autopsy studies that revealed high neutrophilic infiltration inside the alveolar space [13]. However, even if there is a significant difference between IMW and ICU of neutrophil fraction (Fig. 1a and b), we did not observe any correlation between neutrophils and clinical outcome (Fig. 1d).

Regarding lymphocytes, ICU patients revealed a strongly and significantly reduced lymphocytes percentage compared to IMW (Fig. 1), corroborating findings recently reported by Liao and colleagues who showed a decreased lymphocyte counts in severe patients with respect to the mild ones [14].

Concerning the three analyzed cytokines, we demonstrated that ICU patients are characterized by significant high levels of the two pro-inflammatory cytokines, IL6 and IL8, with respect to IMW patients (Fig. 3a). This result is sustained by previous reports obtained for Middle East Respiratory Syndrome (MERS) pneumonia patients, demonstrating in respiratory specimens high expression levels of inflammatory cytokines IL-1α and IL-1β and IL-8 (CXCL8) [15]. We not only demonstrated that the two pro-inflammatory cytokines levels are related to the severity of illness, but also with the outcome of patients. In fact, we demonstrated that upon analyzing survivors in comparison to patients who did not survived, significant lower level of IL6 (Fig. 3b) and IL8 (Fig. 3c) were observed in the survivor group with respect to non-survivors.

In order to understand if the treatment regimen could have an impact on inflammatory status at alveolar level, we evaluated cytokines in a small group of ICU patients who received one type of treatment only. The significant result was obtained only for BAL-IL6 when we compared patients submitted to a steroid course against patients who received anti-viral drugs (Fig. 4a). The same trend towards a decrease was also observed for IL8, without reaching a statistical significance (Fig. 4b).

Thus, taking together all our findings, we can suggest that steroids seem to be able to limit the secretion of IL6 in severe COVID-19. This anti-inflammatory effect could be correlated to a better outcome, in fact the clinical trial called “RECOVERY”, recently reported that dexamethasone reduced deaths by one-third in ventilated patients [16].

Regarding the role of IL10 in COVID-19 infection is a matter of ongoing debate. Increased serum levels of IL10 have been described in severe SARS-CoV-2 patients with respect to those affected by mild disease. Moreover, it has been demonstrated a direct correlation between IL10 and IL6 at serum levels, suggesting that both cytokines could be used as predictors of higher risk of disease deterioration [17, 18]. However, in another paper, McElvaney et al. 2020 demonstrated that IL10 response in plasma of SARS-CoV-2 patients is inadequate to the high levels of pro-inflammatory cytokines (IL6 and IL1-beta) [19]. Unfortunately, our present study does not contribute in clarifying this issue. In fact, even if we detected a high median BAL levels for IL10 (334 pg mL^− 1, 133.2–372.1) in all patients, this cytokine was not associated to disease severity or to survival (Fig. 3). In addition, no significant correlation was found between IL10 and IL6 in BAL, while only a negative and significant correlation with IL8 levels was present (Table S2). Thus, on the basis of the present study, a role of BAL IL10 as a specific disease marker cannot be clearly inferred.

Lastly, we tried to understand if pulmonary inflammation correlates with peripheral one, but observing only a positive correlation between BAL and serum IL6, while BAL IL6 and serum CRP did not significantly correlate.

We believe that a significant role in driving lung inflammation in COVID-19 patients has to be ascribed also to IL8, that showed higher BAL values among ICU compared to IMW patients (Fig. 3a) and was associated to a negative outcome (Fig. 3c). IL8 acts as a specific chemoattractant and activating factor of neutrophils, in fact a direct correlation between BAL IL8 and neutrophils count is detectable in our patients (Table S1 and Fig. 5b). It has been already demonstrated that excessive neutrophil activation in the lung is associated to NETs formation and release of oxygen radicals and proteolytic enzymes, which can drive severe epithelial and endothelial injury [20]. Thus, modulation of IL8 pathway might represent a possible target of future intervention.

We are aware that this study has some limitations. We performed these preliminary analyses on limited sample size given the high risk of infection of the health care personnel. This limited our possibility to make further analyses, such as cell surface activation markers assessment by flow cytometry on macrophage or lymphocytes. Another limitation is the lack of paired assessment of cytokines in the peripheral blood due to the absence of stored serum samples from acute COVID19 patients in our institutions. Finally, we know that the infectious complications registered (Table 1) might have influenced cell differentials and cytokine levels however, the percentage of bacterial and fungal co-infection was very high in all patients in ICU without a significant difference between survivors and non-survivors.

In conclusion, in present study, we contributed to define the broncho-alveolar inflammatory environment in acute COVID-19 patients represented by high innate effectors infiltration, in particular neutrophils, that correlate with two pro-inflammatory cytokines, IL6 and IL8, whose BAL level reflect the outcome of COVID-19 patients.