Pulmonary function and chest CT in patients recovering from COVID-19 pneumonia

Background COVID-19 is a new and highly contagious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, there is a paucity of data regarding long-term CT findings and pulmonary function in COVID-19 survivors. The aim of this study was to investigate the influence of COVID-19 pneumonia on pulmonary function and chest high-resolution computed tomography (CT) in convalescent patients. A retrospective study of COVID-19 pneumonia patients in the Beijing Youan Hospital, Capital Medical University, was conducted. Serial assessments, including pulmonary volumes (TLC), spirometry (VC, FVC, FEV1), pulmonary diffusing capacity for carbon monoxide (DLCO, DLCO/VA), and chest high-resolution CT were collected 3 months after discharge. received glucocorticoids had higher chest CT scores but a similar incidence of DLCO impairment than those in the nonglucocorticoid group. Conclusions Three months after discharge from the hospital, impaired diffusing capacity and CT abnormalities were detected in more than one third of COVID-19 patients. Compared with patients with non-severe disease, those with severe illness had a higher incidence of lung imaging abnormalities and similar lung function impairment.


Background
The coronavirus disease 2019  caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a worldwide pandemic, which poses a serious threat to human health [1,2]. Pneumonia caused by SARS-CoV-2 is one of the main clinical manifestations, and the severity of pneumonia directly affects the prognosis. Chest highresolution computed tomography (CT) can show abnormal conditions in the lung and depict the patterns and extent of the abnormalities. Previous scholars who maked scientific researches on imaging features of COVID-19 patients found that bilateral, multifocal groundglass opacity (GGO) and consolidation, predominantly in the subpleural and peribronchovascular regions, were the typical characteristics [3,4].
COVID-19 leads to impaired lung diffusing capacity in about 52.6% of survivors 1 month after discharge [5]. However, there is a paucity of data regarding long-term CT findings and pulmonary function in COVID-19 survivors. The aim of this study was to assess changes in lung CT and lung function in 46 patients 3 months after discharge from our hospital. We also compared the outcomes of patients with severe disease to those with non-severe illness.  We used the same method as Nöbauer et al to quantify pulmonary function [7].Lung function was considered abnormal if lung volumes were less than 80% of predicted values.

Patient selection
Obstruction was defined as FEV1/VC less than 80% of the predicted value, whereas restriction was defined as VC and TLC less than 80% of the predicted value. Reduced dispersion function was defined as less than 80% of the predicted value. An impulse oscillation system (IOS) was used to measure airway resistance. Total airway resistance was defined as an oscillation frequency of 5 Hz(R5), central airway resistance was an oscillation frequency of 20 Hz (R20), and peripheral elastic resistance was reactance of an oscillation frequency of 5 Hz(X5). Upper-airway resistance was an oscillation frequency of 35 Hz(R35).

Lung imaging acquisition and CT quantitative evaluation
All chest CT scans were performed with a 256-section scanner (Brilliance iCT, Philips Healthcare, Cleveland, OH, USA) without intravenous contrast. CT examination parameters were as follows: 120 kV; automatic tube current (100 mA-400 mA); iterative reconstruction technique; detector collimation, 128 × 0.625 mm; section thickness, 5 mm; rotation time, 7 0.4 s; pitch, 0.914; matrix, 512 × 512. Images were assessed by 2 radiologists, both of whom were blinded to clinical information and lung function [8].
We used the same method as Li et al to quantify pulmonary inflammation severity [9]. To quantify lung parenchymal changes, we divided the lung parenchyma into 12 compartments on each side. Horizontal lines along the eminence and lower pulmonary vein divided the lungs into 3 parts: upper, middle, and lower. Each part included 4 subparts: central, peripheral, ventral, and dorsal. If the lung parenchymal lesion in each region did not cover more than one-third of the total surface area of the region, it was defined as a focal lesion.
In this study, pulmonary parenchymal lesions were classified into 3 types [9]. Mild pulmonary parenchymal changes mainly included changes in the line shadow. Each compartment was assigned a score based on the following: 0 (no involvement), 1 (focal lesion), 2 (widespread lesions). Moderate changes included GGO-like changes, bronchiectasis, grid-like fibrosis, and pulmonary hyperinflation symptoms, and severe changes mainly included honeycombing lesions or solid-shadow changes accompanied by traction bronchiectasis. CT scores of lung involvement were also obtained after discharge. All CT scores were independently performed by 2 radiologists, and agreement was reached by consensus.

Statistical methods
Statistical analysis was performed using Statistical Package for Social Science (SPSS) Version 22.0. Normal variables were expressed as mean ± standard deviation (x±s) and 8 analyzed by the independent or paired sample t-test. Nonnormal variables were expressed as medians (Q1, Q3) and analyzed by the Mann-Whitney U test. Categorical variables were described as percentages and compared using χ 2 or Fisher's exact test between groups. All statistical tests were 2-tailed. Statistical significance was defined as P<0.05. GraphPad Prism 8.0 was used to produce charts.

Characteristics of the enrolled COVID-19 patients
This study evaluated a total of 108 patients. Seven patients were excluded because they were younger than 18 years. Twenty-five patients were excluded because it had been less than 3 months since discharge, 3 were excluded because they had a history of severe respiratory diseases or mental illness, and 7 were lost to follow-up. Forty-six patients completed the serial assessments in the study (Fig. 1 0.013). The use of methylprednisolone in patients with severe disease was more common in patients with severe disease than in those with non-severe disease (P=0.000), and those with severe illness had higher serum C-reactive protein (CRP) peaks and neutrophil counts than those with non-severe disease. However, there were no significant difference in levels of white blood cells, hemoglobin, platelets, alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase (CK), PaO2/FiO2 ratio, or length of hospitalization (Table 1).

Pulmonary function tests and high-resolution chest CT
Three months after hospital discharge, 37.8% of COVID-19 patients still had diffusion dysfunction, and 60.9% of their chest CTs had not completely returned to normal (Table2-2).
One (2.2%), 2 (4.3%), and 17 (37.8%) patients had FEV1/FVC ratios and TLC and DLCO values less than 80% of predicted values, respectively. There was no significant difference in recovery of pulmonary function between the severe and non-severe group ( Table 2-1, Table   2-2). On the contrary, there was a significant difference in recovery observed on chest CTs between the severe and non-severe group, and that of the non-severe group was better than that of the non-severe group [2.0 vs 5.5, P=0.004] ( Table 2-1). Chest CT was more sensitive than the lung function test in reflecting the difference in pulmonary lesions between the severe and non-severe groups (Table 2-1, Fig.2-1). In addition, CT scans showed significant improvement in both groups, compared with the worst CT in the early phase (P<0.001) (Fig.2-2, Fig.2-3). 10 Nine patients (39.1%), 2 men (4.3%) and 7 women (15.2%), with an average age of 47.74 ± 12.12 years (range, 39 to 69 years) were treated with glucocorticoids, and that of the nonglucocorticoid group was 44.59 ± 11.39 years (P=0.013). In the glucocorticoid group, 6 patients (13.0%) had severe illness, and 2 (4.3%) had non-severe disease (P=0.000). There were no significant differences in PaO2/FiO2 ratio or lymphocyte or hemoglobin levels.

3.Pulmonary function tests and CTs in glucocorticoid and nonglucocorticoid groups
Through analysis of pulmonary function in the recovery period, we found that the diffusion function in the glucocorticoid group was worse than that in the nonglucocorticoid  Table 3).

Discussion
Since COVID-19 spread worldwide over the last year, its mechanism, clinical characteristics, prognosis, and effective treatment have been gradually elucidated through much hard work. Consistent with previous studies, the clinical symptoms of COVID-19 patients in our research were similar to those of those with influenza or severe acute respiratory syndrome (SARS) [10][11]. We found that, compared with patients with nonsevere COVID-19, those with severe COVID-19 were older and more likely to have high BMIs and CRP levels, as well as hypertension and cardiovascular disease. Age was a strong 11 risk factor for severe illness, complications, and death [10,12,13]. The case-fatality rate was higher for patients with comorbidities, at 10.5% in those with cardiovascular disease. People with hypertension may be at increased risk for severe COVID-19 and should continue to take their medications as prescribed [14]. High CRP levels may be associated with more severe illness [15].
Preliminary studies on imaging features of patients with COVID-19 found that bilateral, multifocal GGO and consolidation, predominantly in the subpleural and peribronchovascular regions, were typical features. Recent research and our data showed that more than half of the discharged patients had residual abnormalities on chest CT [16][17].
Small GGO, fiber cord shadow, and subpleural lines were still more intense in patients with severe versus non-severe disease 3 months after discharge. Meo et al reported that SARS and COVID-19 have similar biological and clinical characteristics [11]. Lung fibrosis was found at 3 and 6 months in recovering SARS patients in previous studies [18][19] and could be a long-term sequela of SARS-CoV-2 infection as well [20].
More concerns have been focused on the assessment of the pulmonary injury in discharged persons of COVID-19 . This study showed that in convalescence, about one-third of COVID-19 patients had pulmonary function impairments, the most common of which was impaired diffusing capacity. Research results indicated COVID-19 pneumonia has a characteristic of impaired diffusion pathways in the intra-alveolar space. Various types of viral pneumonia usually self-limited have a feature of diffuse alveolar damage, and imaging abnormalities usually disappear within three weeks in immunocompetent patients [21].
Autopsies of COVID-19 decedents showed different degrees of destruction in the alveolar structure and pulmonary interstitial fibrosis [22][23]. Pathological changes in the lungs can explain the impaired DLCO to an extent. Surprisingly, many patients' DLCO normalized but with residual imaging abnormalities. We think that mild lung lesions in convalescence (eg, fibrous cord shadow, subpleural line, small ground glass shadow) may not cause severe pulmonary diffusion disorder and would not cause the significant difference in pulmonary function between patients with severe versus non-severe illness. We will continue to perform long-term follow-up on these patients to observe DLCO impairment trends.
In this study, glucocorticoids were used more often in patients with severe COVID-19 than in those with non-severe illness. The recovery seen on chest CT in patients who received glucocorticoids was not as advanced as that in patients not given the drugs. As for the small sample size, we did not compare CT expression in patients with severe disease who received or did not receive glucocorticoids. Use of glucocorticoids in COVID-19 patients has been controversial. Some experts have said that glucocorticoids may prolong the duration of illness. However, many studies have shown that proper use of glucocorticoids can reduce the mortality rate [24]. In a meta-analysis of 7 trials with 1703 critically ill COVID-19 patients, glucocorticoids reduced 28-day mortality compared with standard care or placebo (32% vs. 40%; odds ratio 0.66; 95% confidence interval, 0.53 to 0.82) and were not associated with increased risk of severe adverse events [25].
The US Centers for Disease Control and Prevention recommend dexamethasone for severely ill COVID-19 patients on supplemental oxygen or ventilatory support. In this study, 13 recovery observed on CT in the glucocorticoid group was poor, but glucocorticoids were clearly beneficial, which may be related to the poor status of basic CT in patients with severe disease [26]. Though the lung function parameters were not different between the nonsevere and severe groups, DLCO values in the glucocorticoid group were lower than those in the nonglucocorticoid group. Furthermore, the use of glucocorticoids, total methylprednisolone dosage, and length of treatment were negatively correlated with DLCO (data not shown).
There are some limitations in our study. First, it was a prospective study with a small sample size, which provides only a short follow-up of 3 months. The heterogeneity of our research conclusions was not comprehensive. Second, only 46 of 108 COVID-19 patients (42%) completed the serial assessments; therefore, the results might not be representative of the entire population. Last, because our research object was not drawn by random sampling, the calculation of seroprevalence was restricted to potential sampling bias.

Conclusions
Three months after hospital discharge, impaired diffusing capacity and lung imaging abnormalities were detected in more than one third of COVID-19 patients. Compared with patients with non-severe disease, those with severe illness had more lung imaging abnormalities. Longer follow-up studies in COVID-19 patients should be performed to investigate clinical outcomes after recovery.  Values are shown as mean ± standard deviation (x±s) , medians (Q1,Q3) or percentage, severe vs non-severe with p values; *Statistically significant.  The CT score was 30. B. Worst CT scan of the same patient five days later showed diffuse GGO and consolidation. The CT score was 42.C. CT scan of the same patient nine days later showed less GGO on discharge. The CT score was 32. D. Followup CT of the same patient at three months after discharge from hospital showed that patchy GGO had obvious absorption. The CT score was 2.