The global pandemic caused by SARS-CoV-2 has affected more than 200 million people, according to the Coronavirus Resource Center at Johns Hopkins University.1 There have been more than four million deaths and 180 million people have recovered. However, there are concerns about long-term effects2 of COVID-19 survivors, irrespective of their initial severity, and these may have a significant impact on quality of life and disability.

Lessons learned from the previous epidemic of severe acute respiratory syndrome (SARS) in 2003 indicate that a subset of patients who survived the disease eventually developed residual pulmonary fibrosis. So, studies that analyzed patients who had recovered from SARS-CoV infection at 3, 6 and 12 months, showed that about 30% had features of fibrosis such as reticular changes in the lung parenchyma.3,4 These findings occurred mostly in elderly and severe patients.

One report concluded that up to one third of admitted hospital patients with COVID-19 might develop acute respiratory distress syndrome (ARDS).5 Further, many patients who survived ARDS had a restrictive pattern on lung function that persisted one year after discharge from the intensive care unit.6

Now in SARS-CoV-2 infection, severe COVID-19 has already been associated with impairment in pulmonary function and radiological abnormalities.7 Several studies had data three months after hospital admission, showing a reduced lung diffusing capacity between 21%8 and 57%9 of patients and one-fourth of discharged COVID-19 patients had chest computed tomography (CT) scan abnormalitites.10

A dysregulation in the wound healing process of acute lung injury caused by SARS has been described and some pathways controlled by receptor tyrosine kinases have been implicated.11 The role of matrix metalloproteinases (MMPs) in the lung repair process has been studied in ARDS12 and as potential biomarkers in idiopathic pulmonary fibrosis.13 However, to date the links between SARS-CoV-2 and lung fibrosis remains unclear. We hypothesized that SARS-CoV-2 may cause overexpression of biomarkers, which can induce a fibrotic environment in lung tissues. Within an observational, prospective study design, we aimed to evaluate functional respiratory impairment of patients at two and six-months after SARS-CoV-2 pneumonia, residual changes in their chest CT, and any correlation with potential peripheral blood biomarkers of lung fibrosis.

Between May 1, and July 31, 2020, this prospective, multicenter, cohort study evaluated 932 eligible patients who were admitted with pneumonia for SARS-CoV-2 from Respiratory Departments in participating hospitals, of whom 481 were initially considered for follow-up. Reasons for exclusion were logistic reasons as living out of area, previous diagnosis of COPD or ILD, investigator's decision or lack of inform consent. Mainly due to the overload of the health system, as we were in the first peak of the pandemic, some centres were unable to collect all data, therefore the final number of patients available for the second follow-up was 313 (65.1%) (Both groups of patients were similar, appendix S-2) (Fig. 1).

Fig. 1.

Flow chart of patients discharged from participating hospitals included in the cohort COVID-FIBROTIC.

(0.2MB).

The mean time for outpatient visits after discharge was 63±12 days for V1 and 181±10 days for V2. The group consisted of 184 males (59%), aged 61.12±12.26 years with a body mass index (BMI) of 28.23±4.78. Other demographic and clinical characteristics of the cohort are given by COVID-19 severity (moderate vs. severe) in Table 1. With the exception of sex distribution, there were no differences between both severity groups in terms of age, ethnicity, comorbidities, smoking history and BMI. Severe patients had higher RALE scores and a longer hospital stay (p<0.001). Severe patients also showed more prominent laboratory abnormalities as lymphocytopenia, elevated D-dimer and higher levels of lactate dehydrogenase, C-reactive protein, ferritin and fibrinogen, than moderate patients (all p<0.05).

At 60 days, the proportion of patients with lung diffusion impairment (DLCO<80%, of predicted) was 54.6%. At 180 days, after hospital discharge, there was a slight improvement but there were still 147 patients (47%) with impaired pulmonary diffusion (p=0.011). Half of patients complained of dyspnoea at V1 with improvement at V2 when 111 patients (35.46%) had one or more points at mMRC scale (Table 2).

Severe patients (group 2) showed statistically worse levels of FVC and DLCO at V1. However, these levels improved at V2 and no significant differences were found between severe and non-critical or moderate patients (Fig. 2).

Fig. 2.

Interaction plot of functional parameters variation between V1 (two months) and V2 (six months) according to severity (group 1: non critical; group 2: critical). A: FVC%, of predicted. B: DLCO%, of predicted.

(0.25MB).

Factors associated with diffusion impairment are presented in Table 3. According to Akaike's information criterion, factors included in the final model were sex (reference category: female) (OR: 2.97, 95%CI 1.74–5.06, p=0.001), age (OR: 1.03, 95% CI: 1.01–1.05, p=0.005), peak RALE score (OR: 1.22, 95% CI 1.06–1.40, p=0.005), and D-dimer (OR: 1.00, 95% CI: 0.99–1.00, p=0.056) all significantly associated with DLCO<80% (of predicted) at 6 months. Interestingly, a critical clinical course (patients with respiratory support by high flow oxygen or mechanical ventilation) was not a predictor of altered diffusion at follow-up.

According to study protocol, 226 patients underwent CT scan, of whom 149 (66%) had radiological sequelae of COVID-19. Two thirds of those who required respiratory support by high flow oxygen or mechanical ventilation (invasive or non-invasive) had GGO on CT scan; and reticulations, bronchiectasis and parenchymal bands that may indicate early progression to fibrosis were seen in more than one third of patients (Table 4).

Circulating serum biomarkers were determined in a subset of sites and characteristics of this sub-sample were similar to main cohort, except that they were younger (59.51±13.40 vs 62.29±11.84 years, p=0.047) (appendix, S-3). Several biomarkers as MMP-7 (14.11±10.09 vs 7.36±5.17, p=0.001), MMP-1 (10.04±7.33 vs 6.93±9.19, p=0.001), and periostin (1.28±0.89 vs 0.87±0.4, p=0.001) were elevated in patients with severe disease compared to those with a moderate clinical course (Table 5). Moreover, patients with altered lung diffusion showed higher levels of MMP-7 (11.54±8.96 vs 6.71±4·25, p=0.001), periostin (1.11±0.07 vs 0.84±0.40, p=0.001), and VEGF (230.07±223.33 vs 166.74±198.15, p=0.035) than patients with normal DLCO% (Table 6a). Finally, early fibrotic changes in CT scan were associated with even higher levels of MMP7 (13.20±9.20 vs 7.92±6.32, p=0.001), MMP1 (10.40±8.21 vs 6.97±8.89, p=0.023), and periostin (1.36±0.93 vs 0.87±0.39, p=0.001) (Table 6b).

A logistic regression analysis showed that MMP7 (OR: 1.07, 95% CI 1.02–1.12, p=0.002), and periostin (OR: 3.49, 95% CI: 1.83–6.68, p=0.001) remained as independent factors related with the presence of early fibrotic changes in the CT scan (appendix, S-4, S-5, S-6).

This multicenter, prospective study of hospitalized patients with moderate and severe COVID-19 pneumonia showed chest CT abnormalities, elevated blood biomarkers related to fibrogenesis, and alterations in lung function up to 6 months after hospital discharge.

The results of lung function assessment showed that 47% of patients had decreased pulmonary diffusion after 6 months and one third of the cohort suffered dyspnoea. CT findings as a reticular pattern, traction bronchiectasis or parenchymal bands, that may indicate early progression to fibrosis were seen in 20% of patients.

Several studies have presented data on lung function and CT findings in survivors of COVID-19. First reports from discharged patients with SARS-CoV-2 pneumonia showed that 47% of them had impairment of diffusion capacity at one month.22 A cohort of 133 patients evaluated at 60 and 100 days showed a recovery in diffusion capacity, from being impaired in 31% of patients to 21% patients.8 Data from 103 patients at three months showed that 24% of them had reduced DLCO, and opacities on CT scan were seen in 25% of patients. They found no differences between patients admitted to ICU compared to those who were not admitted.10 In contrast, in a study with a longer follow-up assessment of lung diffusion, 6 months after symptom onset, alteration of DLCO was found in 56% of patients who required HFNC or IMV, compared to 29% of those who required only supplemental oxygen.23

We have seen, in our patients, a clear improvement in pulmonary function tests between two and six months after hospital discharge, and no differences are found among severe and moderate patients regarding FVC and DLCO at six months.

Our results showed that only 9% of patients with severe disease had normal CT findings, and GGO was present in almost 70%. A cohort of patients followed 153 days after hospital discharge showed that GGO was seen in 44%,22 however they included in their study mild patients (not requiring supplemental oxygen). One-fifth of our patients had bronchiectasis at 6 months, this finding requires further comments as they may be related to pulmonary fibrotic changes (traction bronchiectasis) but the potential of viral infections to produce bronchiectasis should not be underestimated.24

Our multivariate analysis revealed that peak RALE score, and age, as expected, were associated to diffusion impairment; yet surprisingly, female sex is a prognostic factor for DLCO<80% in our cohort. We do not have an explanation for that, but such finding was also reported in another study where women had two-fold risk for diffusion impairment compared to men.23 Also, a prospective study in COVID-19 patients at 12 months following hospitalization showed an odds ratio of 8.61 of impaired DLCO associated with female sex.25

Clinically, the most striking effect in COVID-19 survivors is diffusion abnormality. Previous reports from SARS outbreak survivors showed that after one year 20% of them had altered diffusion.4 Further, in a small cohort of health workers, followed for 15 years, 38% had a mild impairment in lung diffusion.26 All coronaviruses share the same pathogenic mechanism that binds to a cell surface receptor to enter the host (angiotensin-converting enzyme 2 [ACE-2]). Pneumocytes overexpress ACE-2 that results in viral expansion and diffuse alveolar damage. It has been hypothesized that endothelial cell dysfunction may play a role in complicated COVID-19, as initiating post-infection pulmonary fibrosis and vascular remodelling.27 As described in ARDS, after lung injury, a subset of patients seem to be unable to clear the provisional extracellular matrix (ECM) and exuberant fibroproliferation can lead to residual fibrosis with pulmonary dysfunction.28

Serum biomarkers involved in pulmonary fibrosis were evaluated in selected sites in our cohort and MMP-7, MMP-1 and periostin were elevated in patients with severe disease. Furthermore, increased levels of MMP-7, and periostin were obtained in patients with lower DLCO, and remarkably the highest values of MMP-7, MMP-1 and periostin were seen in patients with early fibrotic changes in CT scan. MMP-7 and MMP-1 are described as potential peripheral blood biomarkers in idiopathic pulmonary fibrosis.13 MMP-7 serum levels have been related to early fibrotic changes on CT scan in asymptomatic patients with familial interstitial pneumonia.29 Serum levels of periostin were higher in patients with IPF and inversely correlated with pulmonary function in these patients.30 Our results have not shown different levels according to severity of disease in some biomarkers related to SARS-induced pulmonary fibrosis as soluble epidermal growth factor receptor (sEGFR)11 or osteopontin which is related to activation of several precursors responsible of worse outcomes in COVID-19 patients.31

Our study has several limitations. First, we do not have baseline data of pulmonary function. Second, chronic interstitial lung abnormalities may be underestimated, as we do not have previous CT scan. Although patients with COPD or ILD were excluded, asymptomatic underlying lung disease, whatever unlikely, cannot be excluded. Third, not all patients underwent CT scan, as per protocol, only those with alterations in chest X-ray and/or diffusion impairment at first follow-up had a CT scan. This approach, in line with the guidance of the British Thoracic Society,32 was adopted to mitigate the overload of the health system. Fourth, we could not study serum biomarkers in the whole group, mainly due to logistic reasons. Patients included in the study were similar to the rest of the group with the exception that they were younger, which may underestimate the results as senescence cells increase the expression of MMPs.33

Patients in our study were those admitted and then discharged from pulmonary departments, and this may carry some bias. During the first pandemic wave, patients with more severe COVID-19 were controlled mainly by pulmonologists in the majority of hospitals. This scenario gives us an insight into the cases that will have the highest risk of developing complications.

To conclude, patients with moderate or severe COVID-19 pneumonia showed an improvement in lung function tests at 6 months. However 47% of them still had impaired pulmonary diffusion. Alterations in CT scan as parenchymal bands that may indicate early fibrotic lesions were seen in 38% of severe COVID-19 patients. Moreover, higher values of serum biomarkers related to the progression of pulmonary fibrosis were identified at hospital discharge in severe COVID-19 patients. Following recommendations for clinical observational studies on post COVID-19 condition,34 we think that severe COVID-19 patients should be offered protocolized longer follow-up, and specialist ILD services should be considered to manage these patients.

JS-C, BS, and JT contributed to the literature search, study design, data interpretation and critical revision of the work. ER-B, AL-M, and IM-T did all laboratory analysis. BS, JT, EF-F, JNS-C, VM, CL-R, LC, ALA, SH, CL, JAR, JLR-H, and AM did all study assessments, study visits, and completed data entry. JAC-A was the independent statistician and analyzed all data. JBS critically revised the manuscript and provided additional statistical support. All authors had full access to all the data and contributed to article writing and editing and approved its submission. JS-C was responsible for the decision to submit the article.

De-identified clinical data might be made available to other investigators after approval by the institutional review board, when all follow-up studies are finished as we have investigations for COVID-19 in progress.

The results presented have not been published previously in whole or part. The authors declare no competing interests.