Elevated plasma levels of epithelial and endothelial cell markers in COVID-19 survivors with reduced lung diffusing capacity six months after hospital discharge

This prospective, observational study included 215 adults who were hospitalized in our institution between May and November 2020 because of PCR-confirmed COVID-19 pneumonia and studied at 6 months after hospital discharge. Figure 1 presents the consort diagram of the study. STROBE guidelines were used to ensure the reporting of this observational study [20]. The study protocol was approved by the Ethical Review Board of our hospital (HCB/2020/0422), and all patients signed their informed consent.

Demographic, clinical and biological characteristics were recorded on hospital admission and at 6 months after discharge. During admission, patients were treated according to international recommendations [21, 22]. The severity of the acute disease was determined according to the seven-category severity scale recommended by WHO [21]: score 3: admitted to the hospital but not requiring supplemental oxygen; score 4: requiring oxygen by mask or nasal prongs; score 5: requiring non-invasive ventilation or high-flow oxygen; score 6: requiring intubation or mechanical ventilation; and score 7: requiring ventilation plus additional organ support—pressors, renal replacement therapy (RRT), extracorporeal membrane oxygenation (ECMO). ARDS was defined by the Berlin criteria [23]. Organizing pneumonia was defined as the presence of different degrees of bilateral ground glass opacities with peripheral, linear and perilobular consolidations, some of them with reverse halo appearance, associated with bronchial dilatation and architectural distortion [24].

All patients followed the current Spanish Society of Pulmonology and Thoracic Surgery (SEPAR) consensus for post-COVID-19 clinical follow-up [25]. In all patients, spirometry and DLCO were measured (Medisoft, Sorinnes, Belgium) 6 months after discharge following international recommendations [26, 27] adapted to the current pandemic situation [28, 29]. Reference values were those of a Mediterranean population [30].

Blood was collected in EDTA tubes by peripheral venepuncture 6 months after discharge and was then centrifuged at 600xg 10 min 4ºC to obtain the plasma which was stored at -80ºC until analysis. Plasma CC16, SP-D (Cloud-Clone Corp, TX, USA), SP-A (Novus Biologicals, CO, USA), SLPI, Angiopoietin-2, sICAM-1 (R&D Systems, Minneapolis, MN, USA) and sE-selectin (RayBiotech, Peachtree Corners, GA) levels were measured by validated commercially available ELISA kits following manufacturer instructions. Plasma samples were diluted 1/10 for CC16, SP-D, SP-A and Angiopoietin-2, 1/80 for SLPI, 1/25 for sE-selectin and 1/20 for sICAM-1. The limits of detection were 0.156 ng/ml for CC16, 0.625 ng/ml for SP-D, 15.625 pg/ml for SP-A, 62.5 pg/ml for SLPI, 24.7 pg/ml for sE-selectin, 1.6 ng/ml for sICAM-1 and 46.9 pg/ml for Angiopoietin-2.

Participants were categorized according their DLCO values at 6 months after discharge as normal (≥ 80% ref.) or low (< 80% ref.); the latter were further subdivided as moderate (80–50% ref.) or severe (< 50% ref.). Results are presented as n, frequency, mean ± standard deviation (SD), median [interquartile range -IQR]. Groups were compared using ANOVA, Student t-test, χ². Fisher exact test, Kruskal–Wallis or Mann–Whitney tests, as appropriate according to the normal data distribution Multiple comparisons between groups were analyzed by the Dunn’s test. Bivariate correlations were analysed using the Spearman rank test. Multivariable adjusted logistic regression models were used to investigate risk factors of low DLCO (< 80% ref.), with age, hypertension, smoking status (never, current, former) and WHO acute disease severity score as independent variables. A p value < 0.05 was considered significant. Analyses were performed using R version 3.6.1, SPSS version 20 (IBM Inc, Armonk, NY) or GraphPad Prism 7 (GraphPad Software, Inc, La Jolla, Calif).

Mean age was 61.4 ± 11.8 years, and 66% of patients were male. The most frequent comorbid condition prior to hospitalization was arterial hypertension (40%). Most frequent symptoms on admission were fever (73%), cough (59%) and dyspnoea (46%). 44% of patients were admitted to the ICU, and 22% required mechanical ventilation. ARDS was observed in 35% of patients, while organizing pneumonia and pulmonary embolism occurred in 50% and 7% respectively during hospital admission. Length of hospital stay was 21.3 ± 18.6 days.

At 6 months follow-up, 52% of patients remained symptomatic, predominantly with dyspnoea (26%) and fatigue (18%). On average, spirometry was normal (Forced expiratory volume in 1^st second (FEV₁) 94.7 [83.5–105.8] % ref, Forced Vital Capacity (FVC) 91.9 [81–103] % ref., and FEV₁/FVC 79.1 [73.9–83]). Mean DLCO was 76.6 [64.7–91] % ref. Collectively, 136 (63.7%) patients had one or more abnormal lung function value. Specifically, 43 patients (20%) had abnormal FEV₁, 47 patients (21.9%) had abnormal FVC and 125 patients (58.1%) had abnormal DLCO values (% ref) (Fig. 2). Among patients with abnormal DLCO, 90 (42%) had a moderate reduction (50–80% ref.) and 35 (16%) a severe one (≤ 50% ref.).

Table 1 shows that patients with reduced DLCO 6 months after hospital discharge were older, more frequently smokers and suffered more arterial hypertension before admission for the acute COVID-19 episode. During hospital admission they had higher levels of C-reactive protein, lactate dehydrogenase (LDH) and circulating leukocytes, suffered more severe disease and required mechanical ventilation more often. Their length of ICU and hospital stay was longer. Six months after hospital discharge, they showed persistent dyspnoea and cough more often, and their mean levels of LDH and circulating leukocytes remained higher. Interestingly, 35 patients (58%) from score 3 and 4 severity disease had reduced DLCO 6 months after hospital discharge without suffering from respiratory diseases nor smoking.

Table 2 presents the results of univariate and multivariate analyses of risk of abnormal DLCO at 6 months follow-up. Multivariate analysis showed that age ≥ 60 years (OR 2.92 [95CI 1.64–5.21], p < 0.001) and WHO severity score 6 (OR 3.03 [95CI 1.14–8.10], p = 0.02) were independent predictors of reduced DLCO after hospital discharge.

Patients with reduced DLCO 6 months after discharge had significantly higher circulating levels of both epithelial (SP-A, SP-D, SLPI) and endothelial (sICAM-1 and Angiopoietin-2) markers, particularly in those with severely reduced DLCO (< 50% ref.) (Fig. 3). Significant associations also existed between DLCO values and these biomarkers (SP-A, rho = − 0.32, p < 0.001; SP-D, rho = − 0.25, p < 0.001; SLPI, rho = -0.15, p = 0.03; sICAM-1, rho = − 0.26, p < 0.001; and Angiopoietin-2, rho = − 0.15, p = 0.03) (see Additional file 1). Once patients that required mechanical ventilation were excluded from the analysis, SP-A (p = 0.004), SP-D (p = 0.002) and sICAM-1 (p = 0.025) still being significantly higher in patients with altered DLCO compared with patients with normal DLCO values. After this exclusion, significant associations still existing between DLCO values and SP-A (rho = − 0.32, p < 0.001), SP-D (rho = − 0.25, p = 0.001) and sICAM-1 (rho = − 0.24, p = 0.002). Also, when patients with comorbidities such as respiratory diseases, hypertension or diabetes were excluded, SP-A, SP-D and sICAM-1 remained significantly higher in this group of patients. When we analyzed the relationship of these markers with blood parameters, we observed weak but significant associations between sICAM-1 levels and LDH (rho = 0.3, p = 0.001) and D-dimer (rho = 0.23, p = 0.03) levels. No differences in CC16 [ng/ml, 24 (14.3–35.4) vs 22.5 (14.8–36.4), p = 0.7] and sE-selectin [ng/ml, 12.7 (6.5–25.6) vs 12.1 (6.99–25.5), p = 0.8] levels were found among patients with normal and reduced DLCO.

Finally, we explored if the levels of these circulating markers during the follow-up were related to the acute COVID-19 event. Figure 4 shows that SP-A levels were higher in patients with more severe acute disease (WHO scores 6 and 7) or in those who had suffered ARDS, organizing pneumonia and pulmonary embolism during hospitalization. Additionally, patients who suffered ARDS had higher levels of SLPI (ng/ml, 160.6 (138.8–204) vs 143.6 (121.5–175.6), p = 0.008) and patients who develop organizing pneumonia had higher levels of SP-D (ng/ml, 160.7 (116.6–200.8) vs 147.3 (102–185), p = 0.048). Of note, endothelial biomarkers levels during follow-up were not related with events during hospitalization.

Our clinical and functional findings here are in keeping with previous studies by our group and others evaluating DLCO after hospital discharge in COVID-19 survivors [3‐8]. They are also in line with the reduction in DLCO observed in non-COVID-19-related ARDS survivors at 6- and 12-months after hospital discharge [31, 32]. Although recent studies have suggested cardiopulmonary recovery after COVID-19 [9], our findings indicate that a subgroup of patients (≥ 60 years requiring mechanical ventilation and/or additional organ support during admission because of acute COVID-19) still present reduced DLCO 6 months after discharge. The pathophysiological causes for persistent alterations in gas-blood exchange are not fully elucidated yet, but reduced DLCO may suggest interstitial or pulmonary vascular alterations caused by COVID-19 [3]. These structural lung changes may damage diffusion membrane, especially in those patients with severe disease. Their future follow-up will help to elucidate if these alterations are permanent or resolve with time.

It has been recently recommended that clinical observational studies on post-COVID-19 conditions investigate biomarkers after, at least, 6 months post-hospital discharge [33]. To our knowledge, this is the first study to do so. We found that the epithelial markers SP-A, SP-D and SLPI are increased in patients with abnormal DLCO 6 months after hospital discharge, reflecting a persistent epithelial injury. SP-A and SP-D are surfactant-related soluble pattern recognition receptors mainly produced by alveolar-epithelial type II cells (AEC-II) and involved in the lung innate immune response against viral and bacterial pathogens [34]. Some previous studies have reported elevated serum SP-D levels during the acute COVID-19 episode and suggested that this may be protective [35‐37]. Although Club cells can both secrete CC16 and surfactant proteins, CC16 levels were not associated with reduced DLCO in COVID-19 patients. These results may suggest that AEC-II cells are one of the key cells involved in the pathophysiology of reduced DLCO in COVID-19 survivors. Interestingly, SP-A also participates in the protease and antiprotease activity through the regulation of SLPI [38], suggesting that the elevated SLPI levels observed in patients with reduced DLCO may be induced by the high SP-A concentrations.

Regarding the endothelial markers, although sICAM-1 and Angiopoietin-2 are not specific biomarkers for pulmonary vasculature, we found that their levels were increased in patients with abnormal DLCO 6-months after hospital discharge. Both markers are elevated during the acute phase of SARS-CoV-2 infection in relation to disease severity [17‐19]. sE-selectin was found to be increased during the acute phase, especially in COVID-19-related ARDS [17], but we did not observe any significant relationship with reduced DLCO at 6-months after hospital discharge. This fact suggests that sE-selectin may have a role only during the initial steps of the pulmonary vasculature damage. Therefore, our findings highlight a persistent endothelial injury in patients with lung sequelae. However, there is a temporal evolution of the endothelial dysfunction markers during the progression of the disease that it should be considered. Viecelli Dalla Sega F. et al.showed that the time course for some endothelial dysfunction markers was different between COVID-19 survivors and non-survivors [39]. Further studies evaluating the levels of these markers during an extended follow-up are needed to know their stability over time.

We also observed, as other works, that patients with mild and moderate COVID-19 can also suffer long-term sequelae [33]. In fact, several of the biomarkers quantified here, including SP-D, SP-A and sICAM-1 still being higher after excluding by mechanical ventilation requirement. Also, these markers still being significantly different when patients with hypertension and/or diabetes are excluded, suggesting that our results are not biased by the comorbidities of the population. In contrast, we observed that SLPI and Angiopoietin-2 did not remain associated with reduced DLCO after mechanical ventilation exclusion, suggesting that they are markers dependent of ventilator-induced lung damage. However, the associations shown in our study may not be causative, and further studies are needed to address this question.