Introduction

Much of the knowledge about COronaVIrus Disease 2019 (COVID-19) comes from studies focusing on the acute phase of the disease. On the other hand, few data are available regarding the respiratory sequelae of COVID 19 infection. Referred dyspnoea/breathing difficulties is one of the most frequently reported symptoms in long-COVID syndrome [1,2,3] but the correlation of symptoms with functional and/or morphologic sequelae remains to be assessed. Considering the high proportion of patients experiencing long COVID [4] the impact of a clinical/diagnostic assessment in symptomatic patients could be very high, both at the single patient level and for the health care systems. Thus, evidence regarding long-term effect on lung function and structure is highly relevant.

An abnormal lung function with impaired diffusing capacity for carbon monoxide (DLCO) has been observed in a significant percentage of patients several months after hospital discharge [5,6,7]. A higher incidence of DLCO impairment was shown in those who experienced a severe course of the acute disease [8]. In addition, recent studies showed that many patients still have residual lung CT abnormalities 6 months after discharge, with peripherally-distributed ground-glass opacities (GGOs) being the most common findings [6, 7]. Residual radiological changes were significantly associated with the severity of acute disease [6,7,8].

In the last decades several investigations aimed to detect the long-term fibrotic burden after severe acute respiratory syndrome coronavirus (SARS-CoV) [9, 10]. Since the early phase of the pandemic, the similarities between COVID-19 and SARS have brought to hypothesize a similar risk of progression to lung fibrosis. In addition, the advancing age and multiple comorbidities in COVID-19 hospitalized patients may further increase the severity of long-term pulmonary sequelae. Moreover, the dramatic diffusion of SARS-CoV-2 worldwide makes the possible fibrotic evolution highly impacting in terms of morbidity and mortality [11]. To date several studies aimed to quantify the burden of long-term pulmonary fibrosis in hospitalized patients with COVID-19 pneumonia and identified risk factors predicting persisting lung abnormalities [12,13,14]. Han et al. showed lung ‘fibrotic-like’ changes in more than one-third of patients after 6 months from severe COVID-19 pneumonia [12]. These changes were associated with older age, acute respiratory distress syndrome, longer in-hospital stays, non-invasive mechanical ventilation and chest CT involvement during acute disease [12]. Lung ultrasonography (LUS) is a safe, versatile and cost-effective imaging modality for the diagnosis and monitoring of many pulmonary diseases in acute care setting [15,16,17]. Its use has spread up during SARS-CoV-2 pandemic, both for diagnostic and prognostic purposes [18, 19]. In the last decades it has emerged as a promising technique for the screening and burden assessment of fibrosis in interstitial lung disease (ILD), particularly in connective tissue disease (CTD) [20]. In the presence of lung fibrosis, the pulmonary sonographic pattern is characterized by an irregular pleural line with reduced sliding and the presence of vertical artifacts (B-lines) representing the pulmonary interstitial involvement [21, 22]. In particular, B-line burden has shown a high sensitivity for the detection of ILD when compared to CT [23]. Thus, lung ultrasound could theoretically be useful in detecting long-term pulmonary sequelae of COVID-19 pneumonia. To the best of our knowledge most studies focused on the role of LUS in acute COVID-19 infection while limited data are available regarding LUS findings after hospital discharge [24,25,26].

In our study, we aimed to evaluate the potential of LUS performed at six months in detecting fibrotic-like changes revealed by High-resolution chest tomography (HRCT). Furthermore, we analysed the possible role of clinical, biochemical and echographic parameters recorded during hospitalization for COVID-related pneumonia in predicting long-term evolution to fibrotic-like changes.

Methods

Study population

The study is part of a single centre, prospective, observational study including all adult patients affected by COVID-19 pneumonia and admitted to Luigi Sacco Hospital, Milan, Italy, since February 21st 2020. The study was approved by the local Ethics committee (Protocol number 16088/2020).

Patients discharged from two COVID-19 medium-intensity care units during the first pandemic wave (February–end of May 2020) have been scheduled for a comprehensive follow-up evaluation in our outpatient clinic at 6 months, including clinical assessment, LUS and pulmonary function tests. In the study we included all patients with a diagnosis of severe pneumonia [27] who needed a chest CT scan according to clinical judgement after 6 months from hospitalization. Pre-existing conditions that may mislead the evaluation of lung ultrasound (i.e. congestive heart failure, lung neoplasms, pre-existing lung interstitial diseases) were considered as exclusion criteria.

Clinical, biochemical and LUS data collected during the hospitalization were recorded; moreover, symptoms, oxygen saturation (SO2), heart rate, LUS, pulmonary function tests and CT scan data obtained at 6 months were collected for all the patients.

Pulmonary function tests were performed using a constant-volume plethysmograph (MasterScreen Body; Erich Jaeger GmbH, Würzburg, Germany) following current recommendations [28]. The following items were analysed: forced vital capacity (FVC), forced expiratory volume in one second (FEV1), diffusing capacity for carbon monoxide (DLCO). Diffusion lung capacity for carbon monoxide (DLCO) was measured with the single breath manoeuvre (MasterScreen Body; Erich Jaeger GmbH, Würzburg, Germany), assessing the VA with the inert gas dilution technique [29]. Measured DLCO < 80% of the predicted value indicated pulmonary diffusion impairment. Lung function tests were performed following the latest national and international safety standards for lung function testing during the COVID-19 pandemic [30, 31].

Lung ultrasound

LUS was conducted by trained ultrasonographers who were blinded to CT data and performed with Epiq 7 or CX50 (Philips) ultrasound devices, using a convex probe. US machines setting were optimized following the subsequent modalities: low mechanical index (0.7 or less); a single focus, positioned on the pleural line; no harmonic modality; no persistence.

The examinations were conducted through bilateral scans, dividing the thoracic surface into 12 zones, 6 for lung: two anterior (upper and lower), two lateral and two posterior regions [32]. The anterior axillary, the posterior axillary line and the internipple line were used to divide the regions.

Each lung area was scored as follows:

  • Score 0: presence of horizontal artifacts (A-line pattern); < 3 B lines can be present.

  • Score 1: presence of at least 3 B lines in at least one scan of the region; the B lines are well separated and do not merge one in the other. Small subpleural consolidations ≤ 1 cm diameter and irregular pleural line may be present.

  • Score 2: multiple, converging B-lines, usually determining a so-called “white lung” in at least one scan of the region. Small subpleural consolidations ≤ 1 cm diameter and irregular pleural line may be present.

  • Score 3: at least one consolidation with major axis > 1 cm in at least one scan of the region.

The presence of pleural effusion was recorded on the record form. Each zone was accurately scanned, with both longitudinal and transversal scanning sections, and scored according to the highest score obtained in that area. Total LUS score was obtained by summing up the scores obtained for each area. As previously reported, the left anterior inferior area was not considered due to likely issues about scoring reliability in that area related to the presence of the heart. Thus, total score was derived from 11 areas (range 0–33) [33].

For each LUS we evaluated total score, bilateral involvement, evidence of pleural line abnormalities, subpleural nodules, consolidations (i.e. with major axis > 1 cm).

Lung CT imaging-CT scanning protocol

High resolution chest CT (HRCT) was performed using a 64-MDCT Brilliance scanner (Philips Medical System, Netherlands). The parameters used for the scanning protocols were as follows: patients in the supine position; end-inspiratory acquisition; tube voltage, 120–140 kVp; automatic tube current modulation: 100–300 mAs; pitch, 0.5; and section thickness after reconstruction, 1.25 mm. Unenhanced CT scans were obtained for all patients.

CT scans were independently reviewed by two expert radiologists, blinded to clinical data. They firstly assessed the presence of lung residual abnormalities due to COVID-19 pneumonia (ground-glass opacities, consolidation, thickening of pleural septa, parenchymal bands) and in particular of fibrotic-like changes, defined as traction bronchiectasis, parenchymal bands and/or honeycombing, according to a previous study by Han et al. [12]. Secondly, to quantify the involvement of the lung, we used a semi-quantitative CT score system, currently used in COVID-19 literature [34], assigning a score from 0 to 5 to each pulmonary lobe, depending on the extension of pulmonary abnormalities (regardless the type of alteration): 0, no lesion; 1, < 5%; 2, 5–25%; 3, 26–49%; 4, 50–75%; 5, > 75%.

Statistical analysis

Categorical variables were reported as counts (percentages) whereas continuous variables were reported as mean (standard deviation) or median (interquartile range, IQR), as appropriate. Univariate and multivariate logistic regression analysis was performed to assess the association between LUS score, presence of symptoms and DLCO 6 months after hospital discharge and CT scan fibrotic-like changes. Only those variables statistically significant at univariate analyses were considered in multivariate models. A second univariate and multivariate logistic regression analysis was performed to assess the value of some predefined baseline factors (age, sex, worst PaO2/FiO2, ventilator support, worst C reactive protein (CRP) value, worst D-dimer value and worst LUS score during hospitalization) to predict fibrotic-like changes on 6-month CT scan. Results were expressed as odds ratio (OR) with their 95% confidence intervals (CI). P values lower than 0.05, two sided, were considered statistically significant. All the statistical analyses were conducted with the statistical software SAS (release 9.4, SAS Institute, Inc., Cary, North Carolina).

Results

Seventy-four patients were enrolled in the study. Median time point for follow-up evaluation was 195 days for clinical visit and LUS (IQR 184–210), 198 days for CT scan (IQR 183–211) and 202 days for pulmonary function tests (IQR 187–215).

Mean age was 65 (IQR 56–73). Male sex was more represented in the study population (54 patients, 73%), and the most common comorbidity was arterial hypertension (42%). Characteristics of study population are summarized in Table 1.

Table 1 Population characteristics
Full size table

LUS was performed in all patients, showing abnormalities in 50 of them (69.4%), with a median LUS score of 2 (IQR 0–5.25). When compared to LUS during hospitalization a decrease in total score greater than 50% was observed in 76% of patients after 6 months from COVID-19 pneumonia. Thirty-one patients presented bilateral involvement and the posterior inferior areas resulted the most frequently affected. In our population, all large consolidation had reabsorbed and only small subpleural nodules were found, often associated with fragmented and thickened pleural line. All LUS characteristics are summarized in Table 2. Examples of LUS abnormalities are presented in Fig. 1.

Table 2 LUS findings at six months
Full size table
Fig. 1
figure 1

Examples of LUS findings at 6 months after Covid-19 pneumonia. a Interstitial pattern with separated B-lines, LUS score 1. b Interstitial pattern with confluent B-lines, LUS score 2. c Image blow-up for better appreciation of the irregular/fragmented pleural line. d Sub-centimetric pleural nodule, irregular pleural line

Full size image

All the 74 patients underwent HRCT. Twenty-four out of 74 patients (32%) showed lung abnormalities suitable for fibrotic-like changes, while the remaining patients showed either complete resolution (33/74; 44%) or non-fibrotic-like abnormalities (17/74; 23%) (complete data are provided with Online resource 1, Table e1).

Almost two thirds of patients complained about long-lasting symptoms six months after discharge, with 49% reporting fatigue and/or dyspnoea. The analysis of lung function test, showed as the most significant finding a reduction in DLCO value < 80% of predicted in 27 patients (41.5%) (complete data are provided with Online resource 2, Table e2).

The results of univariate and multivariate analyses assessing the value of LUS to predict fibrotic-like changes on 6-month CT scan are summarized in Table 3. LUS score was significantly associated with fibrotic-like pattern on CT scan at univariate analysis. LUS score maintained the statistically significant association with fibrotic-like changes on CT scan in the multivariate analysis when performed with age as the confounding variable. Based on these findings, we assessed the accuracy of 6-month LUS score in identifying patients with fibrotic-like changes on CT scan by building a ROC curve model. With an AUC of 0.85 (95% CI 0.76–0.93), ROC curve showed that 6-month LUS score is moderately accurate when used as an indicator of 6-month fibrotic-like CT scan pattern. In particular, a LUS score lower than 2 can rule out fibrotic-like changes with a sensitivity of 0.92 (95% CI 0.73–0.99) and a specificity of 0.60 (95% CI 0.45–0.74). The ROC curve analysis for 6-months LUS score is shown in Fig. 2.

Table 3 Univariate and multivariate analysis
Full size table
Fig. 2
figure 2

ROC curve model: LUS score accuracy for detecting patients with 6-month fibrotic-like changes on CT scan

Full size image

The results of univariate and multivariate analyses assessing the value of intra-hospitalization factors (age, sex, worst PaO2/FiO2, ventilator support, worst CRP value, worst D-dimer value and worst LUS score during hospitalization) to predict fibrotic-like changes on 6-month CT scan are summarized in Table 4.

Table 4 Univariate and multivariate analysis
Full size table

Univariate analysis showed that only age and worst D-dimer were significantly associated with the presence of six-month fibrotic-like changes. After a multivariate analysis D-dimer value remained the only independent predictive variable of fibrotic-like changes on 6-month lung CT scan.

Discussion

The main finding of our study is that LUS performed after 6 months from a severe COVID-19 pneumonia is a reliable tool in detecting pulmonary fibrotic sequelae. Moreover, D-dimer level during the acute disease was a strong predictor of subsequent development of fibrotic abnormalities.

To our knowledge, this is the first study where LUS was performed after a follow-up of 6 months from COVID-19 pneumonia. When comparing baseline and 6-month LUS findings in our study, a significant improvement of total score was evident in most patients; however, a relevant percentage of patients still had a bilateral involvement after 6 months. These data suggest that COVID-19-associated sonographic abnormalities may take several months to resolve thus adding evidence to what recently reported at a 3-months control by Hernandez-Piriz et al. [35]. Our study observed a significant association between LUS score and the presence of fibrotic-like changes on chest HRCT after a follow-up of 6 months from the acute disease. Our results are consistent with two previous studies, in which patients were evaluated 3 [24] and 2–5 [26] months after COVID-19 hospitalization. In particular, the study by Clofent et al. reported a strong correlation between LUS and CT scores in a patient cohort characterized by a wide range of clinical severity in the acute phase, including a relevant amount of patients who did not require oxygen supplementation [26]. In contrast with our study, a high variability in the timing of follow-up evaluation (median 90 days, IQR, 64–114) was described.

In our study, all patients were evaluated at 6 months; in our population, the percentage of patients with fibrotic-like changes at follow-up chest CT was similar to previous investigations [12, 13]. To note, in our population all patients had a severe COVID-19 pneumonia in acute stage; mean age and need of ventilatory support were higher as compared to Han et al. [12]. Our findings differ from Caruso et al. who reported fibrotic-like changes in 72% of patients at 6-month follow-up [14] probably reflecting a population with increased number of critical patients during acute disease and a different definition of lung fibrosis. Han et al. and Caruso et al. observed that age and the severity of COVID-19 pneumonia were significantly associated to the presence of fibrotic-like changes at chest CT scan after 6 months from acute disease [12, 14]. Although age was associated with the development of pulmonary fibrotic sequelae, D-dimer level during acute disease was the only independent predictive factor in our study. This finding suggests the potential value of this test as a prognostic marker even in the long-term setting of COVID-19 disease.

Although fibrotic-like changes were present at chest CT scan after 6 months from COVID-19 pneumonia in a significant percentage of individuals, their clinical meaning remains debatable. In our study there was not association between symptoms and pulmonary function tests and fibrotic-like changes on chest CT, suggesting the need for further research in order to define the clinical relevance of these radiologic abnormalities. Furthermore, recent studies showed a partial reversibility of fibrotic lesions, thus suggesting that–at least in some cases–they may not reflect a permanent damage to lung parenchyma [36,37,38,39]. In this context, it does not seem appropriate to perform a follow-up chest CT for detecting the presence of fibrotic-like changes in all patients with COVID-19 pneumonia. On the contrary, LUS may be the first-line ideal tool in the long-term follow-up of COVID-19 patients due to low cost, absence of radiation and repeatability. Our data suggest that LUS can potentially be used to follow-up lung lesions after hospital discharge. In particular, LUS could be used for ruling-out the presence of fibrotic lesions in patients who have multiple risk factors for evolution to lung fibrosis (older age, severe pneumonia, ventilatory support, elevated D-dimer levels) thus avoiding a CT examination; moreover, the possibility to follow-up lung alterations could allow a watch and wait strategy, particularly in patients without functional abnormalities. Overall, LUS may reduce the need for chest CT scan thus sparing radiation related damage for patients and healthcare economical resources.

Limits

Our study has several limitations.

Firstly, sample size was small and follow-up was too short to determine whether fibrotic-like changes are permanent, progressive or reversible.

Secondly, only a small percentage of patients performed a chest CT scan during hospitalization so we could not assess dynamic CT changes.

Finally, inter-reader agreement was not performed for LUS.

Conclusions

Our data suggest that LUS performed after 6 months from severe COVID-19 pneumonia may be a promising tool for detection and follow-up of pulmonary fibrotic sequelae.

Further studies on greater populations and new evidences on the long-term evolution and clinical impact of fibrotic-like changes at chest-CT will be necessary to better define the role of LUS in long-term management of patients with COVID-19 pneumonia.