ERS International Congress 2020: highlights from the Clinical Techniques, Imaging and Endoscopy assembly

COPD

Targeted lung denervation (TLD) is a novel endoscopic treatment for patients with symptomatic COPD, which aims to disrupt the parasympathetic nerves around the main bronchi by radiofrequency ablation [1].

At the ERS congress in 2018, Slebos et al. [2] presented data from a randomised sham-controlled double-blinded multicentre trial showing the feasibility and safety of this treatment in patients with moderate-to-severe COPD. In 2019, the AIRFLOW-2 study group published an article in which the safety of TLD was prospectively evaluated, with the effect on respiratory adverse events as the primary outcome. The risk of COPD exacerbation requiring hospitalisation in the 12.5-month follow-up was significantly lower in the TLD group than in the sham group [3]. At ERS 2020, Shah et al. [4] presented the AIRFLOW-2 crossover patient outcomes, whereby, following the 12-month follow-up visit, patients in the sham group were also offered TLD treatment (crossover group). Acute COPD exacerbations (AECOPD) and lung function were then analysed in these 20 crossover patients. 15 patients had completed 12-month follow-up. The percentage of patients with a moderate or severe AECOPD and severe AECOPD alone following TLD decreased from 73% to 47% and 25% to 13%, respectively. Time to first event for moderate or severe AECOPD and severe AECOPD was also attenuated by TLD.

Frequent COPD exacerbations lead to a more rapid decline in lung function. As TLD reduces the number and severity of COPD exacerbations, Hartman et al. [5] studied the annual decline in forced expiratory volume in 1 s (FEV₁) before and up to 3 years after TLD. In 14 patients treated with TLD, the annual decline in FEV₁ was significantly less compared to before treatment (mean FEV₁ decline 1 year before treatment −57 mL per year (n=11) versus mean FEV₁ decline 3 years after treatment −24 mL per year (n=14)). The large clinical trial including a control group with 5-year follow-up (AIRFLOW-3), currently underway, will be needed to confirm these results.

A different treatment modality for COPD is endoscopic lung volume reduction (ELVR) with endobronchial valves. Previous studies have shown good clinical outcomes in selected COPD patients so that valve treatment is recommended as additional therapy in patients with advanced emphysema in the Global Initiative for Chronic Obstructive Lung Disease recommendation [6]. Due to a high post-operative mortality rate reported in an earlier surgical trial [7] in the high-risk COPD group, patients with a transfer factor of the lung for carbon monoxide (T_LCO) of <20% have usually been excluded from ELVR trials. Lenga et al. [8] presented outcomes (lung function, quality of life and exercise capacity) in relation to the T_LCO of patients treated with ELVR. 34 patients with a T_LCO <20% and 87 patients with a T_LCO >20% were included. They found a significant improvement in clinical outcomes 3 months after ELVR in patients with a low T_LCO. These data indicate that the treatment response after ELVR might be also beneficial for patients with a low T_LCO.

Severity of hyperinflation is one of the key factors for success of ELVR with valves. Prior multicentre randomised controlled trials (RCT) excluded patients with a residual volume (RV) ≤175%. Klooster et al. [9] included 12 patients in a single-centre retrospective study with an RV ≤175% and a clear target lobe. Lung function (FEV₁, RV), 6-min walk test (6MWT) and St George's Respiratory Questionnaire (SGRQ) improved at both 1 and 6 months after ELVR.

A systematic comparison of surgical lung volume reduction (LVRS) and ELVR with valves is still missing. The Lung Emphysema Registry, a multicentre nonrandomised prospective clinical study, collected data on patients with emphysema treated with LVRS or ELVR [10]. Since 2017, 246 patients have been recruited; the surgical group (n=88) and the ELVR group (n=158) both showed significant improvements in lung function and quality of life at 3 months. Both techniques seem to be beneficial to patients with severe emphysema. A multicentre RCT (the CELEB trial) is underway, which compares the outcomes at 1 year between LVRS and ELVR. Hopefully, this RCT and future studies, including safety data, might help to define which patient subgroups should be treated with LVRS or ELVR.

Chronic hypercapnic failure (CHF) may develop in patients with severe emphysema and hyperinflation. Previous analysis showed improvement in lung function as well as hypercapnia in patients treated with ELVR by coil implantation [11]. In a retrospective trial presented at ERS 2020, 138 emphysema patients with CHF (carbon dioxide tension (P_CO2) ≥45 mmHg) and treated with ELVR were enrolled. A significant decrease in mean P_CO2 (pre-ELVR mean P_CO2 49.8±4.8 mmHg) could be found at 3 months (mean P_CO2 46.1±12.1 mmHg; p<0.001) and 6 months (P_CO2 46.4±18.9 mmHg; p<0.001) after intervention [12]. Prospective trials are warranted.

Lung cancer

One of the biggest bronchoscopic challenges is how to diagnose small peripheral pulmonary nodules (PPN). The number of patients with PPN that require tissue confirmation is rapidly increasing with the use of computerised chest tomography (CT) and the implementation of lung cancer screening programmes. At ERS 2020, different methods for diagnosing PPN were presented.

Transbronchial lung cryobiopsy (TBLC) is an endoscopic procedure that has proven diagnostic value for interstitial lung diseases [13]; however, there are limited data investigating its use in suspected lung cancer. An observational, retrospective cohort study between 2015 and 2019 enrolled 512 patients who underwent TBLC for suspected lung cancer. The overall diagnostic yield was 93%. 442 cases were malignant and 70 had a nonmalignant pathology with clinical and radiographic follow-up [14]. The complication rate for pneumothorax was 6.6% and for any grade of bleeding was 18%. In most procedures, fluoroscopy and radial endobronchial ultrasound (EBUS) was used for navigation.

These results were confirmed by another observational study evaluating diagnostic efficacy of radial EBUS-guided TBLC of PPN in obtaining histological diagnosis and molecular subtyping [15]. Histopathology was conclusive in 41 (93%) out of 44 patients. Molecular subtyping was performed in 60% (nine out of 15) of squamous cell carcinoma cases and in 88% (15 out of 23) of adenocarcinoma and not-otherwise-specified lung cancer cases.

A late-breaking abstract at ERS 2020 evaluated the efficacy and safety of TBLC and transbronchial forceps biopsy (TBFB) in the diagnosis of PPN [16]. 39 patients were enrolled, with TBFB performed in all patients (n=39) and TBLC performed in addition in 14 patients. Radial EBUS and fluoroscopy was used for guidance with the probe position within the PPN in 33.3% of cases, and adjacent to the PPN in 66.7% of cases. Diagnostic yield of TBFB, TBLC and a combined approach (TBFB+TBLC) were 64.1% (25 out of 39), 64.3% (nine out of 14) and 69.2% (27 out of 39), respectively (p<0.05). The diagnostic yield was 85% (11 out of 13) when placing the radial EBUS probe inside the PPN. No complications occurred.

These studies show that TBLC use in PPN cases results in high diagnostic yields with an acceptable complication profile, especially when used in combination with radial EBUS. However, the added value of TBLC should be evaluated in further randomised trials.

Virtual bronchoscopy navigation (VBN) is a diagnostic modality used to access and sample PPN under fused fluoroscopic guidance. A study from Herth et al. [17] included 116 patients with solitary pulmonary nodules ≥8 mm and compared the biopsy yield and performance of VBN with or without radial EBUS guidance. Overall, biopsy yield was 94% and there was no difference in yield when radial EBUS was used.

Another new innovative technique to diagnose PPN is needle-based confocal laser endomicroscopy (nCLE) which enables real-time microscopic analysis at the needle tip. A study presented at ERS 2020 included 15 patients in whom the feasibility and safety of nCLE imaging for diagnosis of PPN was assessed. In 87% of patients (13 out of 15), good- to high-quality videos were obtained. All malignant lesions scored at least two out of three malignant CLE criteria [18].

EBUS combined with strain elastography might be a helpful diagnostic modality to differentiate malignant lymph nodes [19] by measuring stiffness of a lymph node with ultrasound, where the stiffer node (colouring blue) is more likely to have malignant infiltration. Teng et al. [20] presented a prospective study which included 47 patients in whom all lymph nodes >5 mm were sampled and evaluated with elastography. 30 lymph nodes were scored as non-blue, of which final cytology showed only one malignant case. 51 lymph nodes scored mixed blue/green or blue on elastography and cytology showed malignancy in 26 cases. This resulted in a 96% sensitivity rate, but 54% specificity in predicting malignancy, with a negative predictive value of 97%.

Malignant central airway obstruction occurs in ∼20% of primary lung carcinomas or pulmonary metastasis [21, 22]. A novel modality using microwave ablation (MWA) to treat malignant airway obstruction was presented at ERS 2020. MWA heats tissue by creating an electromagnetic field around an ablation device. In five cases successful airway recanalisation was achieved with no immediate complications noted [23]. In addition, MWA offers the potential for local therapy in inoperable peripheral lung cancer and for patients who cannot undergo stereotactic radiation [24]. A pilot study in 13 patients showed that bronchoscopic MWA is a feasible, efficacious and safe treatment for peripheral lung cancer (ineligible for surgery) [25]. The local control rate was 85.71% (12 out of 14) and the complete ablation rate was 78.57% (11 out of 14) at 6 months post-ablation.

Another diagnostic challenge can be the assessment of tumour invasion (T-status) in the mediastinum and large vessels (T4). A retrospective study included 104 patients in whom T-status was reviewed based on EBUS, CT and thoracotomy findings [26]. Surgical–pathological staging was the gold standard, with 36 (27.6%) patients diagnosed as stage pT4. EBUS showed a high specificity (84–98%) and negative predictive value (76–88%) for T4 assessment and could be of additional value to CT alone.

Interstitial lung disease

In interstitial lung disease (ILD), TBLC was a hot topic at ERS 2020. TBLC is a safe alternative to surgical biopsy as shown in the COLDICE trial [27] published in 2020.

One study evaluated safety in 480 patients who underwent TBLC for ILD [28]. Overall, 5.6% developed pneumothorax and 15.8% required a haemostatic agent for bleeding. Patients with antiplatelet or anticoagulant therapy or patients with emphysema had a tendency to develop pneumothorax (p<0.001 and p=0.029, respectively). The presence of emphysema as a risk factor for pneumothorax following TBLC is comprehensible due to the increased vulnerability of emphysematous lung tissue. However, the antiplatelet drug treatment and its association with the advent of pneumothorax is questionable. It may be explained by the fact that TBLC in patients with anticoagulant therapy is performed intentionally closer to the pleura to avoid the risk of bleeding following central TBLC. The safety of TBLC was again confirmed by Daniels et al. [29], who showed an overall pneumothorax rate of 15.1% in 93 patients (mean number of biopsies 3.8±1.3) over a 5-year period. Notably, the annual rate of pneumothorax declined from 29% in 2015 to 8% in 2018 and 2019, emphasising the learning curve.

The impact of various comorbidities on the safety and complications of TBLC are uncertain. In a prospective study of 196 patients with comorbidities including hypertension, diabetes mellitus, ischaemic heart disease, chronic kidney disease and chronic liver disease, none of the comorbidities, alone or in combination, had a significant impact on rates of pneumothorax or bleeding [30].

Tomassetti et al. [31] investigated whether the histological information obtained by lung biopsy (surgical or TBLC) had an impact on clinician management in 426 patients without a definite usual interstitial pneumonia pattern. Within this trial, a multidisciplinary team made a management decision before and after lung biopsy. The results showed changes in clinical management in 147 (35%) out of 426 cases. The most notable differences observed were a decrease in both the prevalence of untreated patients (from 15% to 4%; p<0.001) and in patients that would have been treated with steroids (from 54% to 37%; p<0.001). Antifibrotic and immunosuppressant use increased significantly after lung biopsy revision (from 23% to 44% and from 7% to 14%; both p<0.001). In uncertain ILD cases, lung biopsy can be of great value.

Interstitial lung disease

It is crucial to diagnose ILD at an early stage so that treatment can be started in time. Lung function and chest radiography are the first steps if lung disease is suspected. However, early ILD can be missed on chest radiograph examination alone. At ERS 2020, one study presented data from artificial intelligence (AI) software used to help physicians detect ILDs [32]. The area under the receiver operating characteristics curve (AUROC) for the detection capability of AI was 0.979, with a sensitivity and specificity of 90% and 100%, respectively, exceeding that of human physicians alone. Thus, this software may help to identify patients with ILD at a very early stage, providing an early start of further diagnostics and adequate treatment. However, further research is warranted.

Furthermore, the impact of magnetic resonance imaging (MRI) as an imaging method to assess ILD was explored. One trial examined the usefulness of oxygen-enhanced MRI as a novel biomarker in patients with different ILD phenotypes [33]. As biomarkers, baseline T1, change in delta oxygen tension (ΔP_O2), oxygen wash-in time and ventilated volume fraction were obtained at three different time points and compared with lung function tests. In the subgroup of patients with idiopathic pulmonary fibrosis (IPF), T1 correlated with the diffusion capacity and the change in ΔP_O2 between the different visits correlated with the change in the transfer coefficient of the lung for carbon monoxide. These positive correlations could only be observed in the IPF subgroup, but not in the other ILD subtypes.

The utility of 129-xenon (¹²⁹Xe) diffusion imaging in ILDs is still not well understood. The apparent diffusion coefficient (ADC) and the mean diffusive length scale (Lm_D) is a diffusion-weighted MRI measurement that provides information about lung microstructure down to the alveolar level. In one prospective observational study, three-dimensional diffusion-weighted MRI with hyperpolarised ¹²⁹Xe was performed twice in 14 patients with IPF [34]. There was a significant longitudinal change in mean ¹²⁹Xe Lm_D in the lower and middle lung zones, but no significant change in ADC or lung function parameters. Thus, ¹²⁹Xe Lm_D may be a useful imaging biomarker to assess disease progression and can provide further insight into the lung microstructure in IPF.

Lung cancer

Various RCTs, including the Nederlands-Leuvens Longkanker Screenings Onderzoek (NELSON) trial [35] and the National Lung Screening trial (NLST) [36] have demonstrated CT-based lung cancer screening to be efficacious in reducing lung cancer mortality. However, there are still some uncertainties leading to slow initiation of formalised lung cancer screening programmes. So far, recruitment for screening is based on the inclusion criteria of the lung cancer screening trials. One trial presented at ERS 2020 calculated the proportion of lung cancer patients who would have been eligible for lung cancer screening under major trial inclusion criteria [37]. The authors found that only one out of three lung cancer patients would have been invited for screening according to NLST and NELSON criteria. This finding reinforces that additional tools to stratify risk and facilitate earlier diagnosis are needed.

Cui et al. [38] evaluated the diagnostic accuracy of a deep-learning based computer-aided detection (DL-CAD) system. In this trial, 360 CT scans were evaluated retrospectively by the DL-CAD system and then independently by radiologists in a double-reading fashion. An additional senior radiologist checked all the results and made the consensus as to the reference standard. The DL-CAD system achieved a sensitivity of 89.3% with one false positive per scan, while radiologists had a sensitivity of 76.0% for detection during double reading. Thus, the DL-CAD system may assist radiologists in pulmonary nodule detection.

COPD

In recent years, radiological assessment in patients with COPD has gained importance, particularly in patient selection for ELVR. CT reflects disease severity, emphysema type and distribution.

One trial presented at the congress shows that CT assessment of erector spinae muscle (ESM) may predict long-term mortality of COPD patients [39]. In this prospective observational study, reduction in the cross-sectional area (CSA) of ESM, total airway count, wall area percentage of the segmental airways and airway tree to lung volume ratio were assessed in 247 COPD patients. Reduced ESM_CSA was associated with poor prognosis more strongly than the other CT parameters. This finding emphasises the impact of muscle loss on survival of COPD patients and confirms the importance of preventive strategies.

Functional respiratory imaging (FRI) is another CT tool, which provides an assessment of ventilation/perfusion mismatch in COPD patients [40]. Here, quantitative CT measures of lobe expansion and blood vessel volumes are obtained to evaluate ventilation/perfusion values. A pooled cohort of 96 patients were studied retrospectively and the FRI results correlated with clinical outcomes. It was shown that ventilation/perfusion levels correlated significantly with SGRQ, 6MWT and oxygen desaturation during 6MWT. This finding emphasises the impact of ventilation/perfusion mismatch on clinical performance of COPD patients and underlines the importance of restoring the ventilation/perfusion level to improve patients’ clinical status.

COVID-19

Coronavirus disease 2019 (COVID-19), caused by SARS-CoV-2, dominated 2020. Infections with SARS-CoV-2 were widespread all over the world leading to a global COVID-19 public health emergency. Besides clinical features and laboratory tests, radiological assessment plays an important role in COVID-19 detection, assessment of disease severity and disease monitoring. In particular, CT reveals the typical COVID-19 features of ground-glass opacities, consolidation and a reticular and/or crazy-paving pattern [41]. Zerka et al. [42] evaluated the development of a new AI software to detect COVID-19 patients. This software was shown to have accuracy of 85%, sensitivity of 70%, specificity of 92%, negative predictive value of 95% and positive predictive value of 59%. Thus, this AI software seems to provide an accurate means of detecting COVID-19 patients allowing rapid, accurate diagnosis and facilitating the implementation of isolation procedures and early intervention.

Limitations of CT may include patient transport, machine preparation and scanning time, such that chest radiography, which is more widely available, still plays an important role. However, as chest radiographs are nonspecific and often difficult to interpret, a machine-learned early-detection system using conventional chest radiography to identify COVID-19 and differentiate its phenotypes was evaluated [43]. The sensitivity, positive predicted value and negative predictive value of the machine early-detection system for COVID-19 was 91%, 97% and 99%, respectively, resulting in a diagnostic accuracy of 95%. This encouraging result demonstrates that a machine-learned early-detection system using conventional chest radiography can identify unique features of COVID-19, facilitating fast diagnosis.

Another trial evaluated the impact of chest radiography as predictor of prognosis in COVID-19 [44]. Chest radiographs in 595 patients with symptoms suggestive of COVID-19 were retrospectively analysed and correlated with intensive care unit (ICU) admission, continuous positive airway pressure (CPAP) use, swab positivity and mortality. It showed that in those with COVID-19, a normal chest radiograph was a statistically significant negative predictor of ICU admission, CPAP use, mortality and swab positivity (p<0.05).

COVID-19

The SARS-CoV-2 pandemic has changed the need for rapid evaluation and bedside imaging assessment in order to narrow differential diagnoses without compromising patient and healthcare worker safety, and to limit potential spread of the virus. Due to the increased number of patients in need of supportive care and thereby limited resources, it is of great importance to be able to stratify patients who will experience more mild or severe disease.

Thoracic ultrasound was explored as a tool for predicting severity and clinical outcomes in COVID-19 patients [45]. The authors used a 0–36-point system (12 scanning zones); at a cut-off score of 24 points, they found a sensitivity of 100% and specificity of 69.2%, and AUROC was 0.846 for identifying patients with a worse prognosis (defined as ICU admission or death within 14 days).

The sonographic findings interpreted as COVID-19 were few B-lines (1 point), multiple B-lines (2 points) or consolidation (3 points). No studies have explored this to date, but the prediction by the experienced ultrasound operators in the ERS assembly is that a high correlation is likely. More studies are expected to be published in this area over the next year, leading to a more solid and comprehensive pool of data at ERS 2021.

Suspected malignant lung lesions and malignant pleural effusion

The evidence for lung cancer screening programmes is increasing [46]. Without doubt, more clinical studies and potential implementation of nation-based screening programmes will lead to a higher number of interventional procedures in the diagnostic work-up of pulmonary nodules. Transthoracic ultrasound-guided biopsies have gained attention because of fewer complications compared to, for example, CT-guided transthoracic biopsies of peripheral and subpleural lesions.

Sharshar et al. [47] presented a study exploring the diagnostic yield and complication rate of ultrasound-guided transthoracic needle biopsies. The results, showing a diagnostic yield of 80% and complication rate of 6.66% for pneumothorax and 3.33% for pulmonary haemorrhage, are supported by a systematic review and meta-analysis of the same topic [48].

Noninvasive methods for the investigation of lung nodules are warranted, especially in patients with comorbidities or reduced lung function. Transthoracic shear-wave elastography has been explored, but so far it is not able to distinguish between benign and malignant lung nodules nor between different histological types of lung cancer [49]. It may be influenced by conditions such as fibrosis, necrosis, air in the surrounding lung parenchyma and pleural effusion.

Ultrasound has a high diagnostic accuracy for the detection of pleural effusion [50]. The presence of pleural effusion most often leads to a cytological examination to rule in or rule out malignancy, but first cytological evaluation implies a sensitivity of only 60% [51]. A systematic review and meta-analysis investigated whether thoracic ultrasound can identify and rule-in malignant pleural effusion (MPE) [52]. The ultrasound findings were not useful in the diagnostic work-up of suspected MPE; however, pleural nodularity in conjunction with pleural effusion could be used as a rule-in test for proceeding with repeated thoracocentesis or additional invasive procedures when malignant disease is suspected.

A systematic review of the current clinical application of contrast-enhanced thoracic ultrasound revealed that its use to guide transthoracic biopsies could increase diagnostic accuracy by 14.6% on average [53]. However, the use of contrast-enhanced ultrasound to distinguish between benign or malignant lesions is not yet well supported by evidence, as overall the studies were heterogeneous, small and with high risk of bias. This reinforces the need for larger, controlled studies before the true application of contrast-enhanced ultrasound can be assessed.

Diaphragm

The amount of literature investigating the sonographic assessment of respiratory musculature has exploded within the past 2 years, with the most commonly evaluated muscle being the diaphragm [54]. The most frequent outcome measures for diaphragmatic function were diaphragm thickness, thickening fraction and excursion, with patients more frequently assessed in the supine position and with an intercostal view. However, the real clinical utility of ultrasound for diaphragm assessment is still debated.

In relation to MPE, fluid drainage normally reduces the respiratory symptoms of most patients. Evidence as to whether pre-drainage motility of the diaphragm can predict the symptomatic outcome has been explored; in a small study (n=20 patients), ultrasound examination of the diaphragmatic shape prior to the pleural procedure did not predict how well the patients felt post-procedure, but excursion may be able to predict the response [55]. Although many studies have been presented within this topic over the past few years, there still remains a gap in the evidence that needs to be explored further [56].

Ultrasound in children

In adult pulmonology, emergency medicine and critical care, ultrasound is widely implemented with a solid evidence base [57, 58]. In the paediatric setting the evidence is less prominent, but is increasing, since the examination does not, for example, expose children to radiation or require movement to another room. Compared to conventional chest radiography, ultrasound has a high sensitivity (97%) and specificity (96%) for the detection of pneumonia in children [59]. Additionally, the sonographic consolidation dimension (measured on the largest consolidation surface) correlates with the serum C-reactive protein level. Despite the limited number of patients (n=49), one study suggested ultrasound as a tool for monitoring hospitalised infants and children [60]. In accordance with published literature, the presented abstracts further indicate that ultrasound may replace chest radiography as the first-line imaging modality in children with suspected pneumonia [61].

Another major topic that many researchers have explored, but which is not yet well anchored, is the role of ultrasound in diagnosing or monitoring ILDs. A pilot study of five children was presented exploring the feasibility of ultrasound and comparability of thoracic ultrasound and CT images [62]. The preliminary results demonstrated good correlation between the images. This result supports the main findings in one adult ILD study comparing ultrasound and high-resolution CT [63], but differs from another, which also explored ultrasound elastography for diagnosis and follow-up of ILD [64].

Education and training

The increased interest, use and research in thoracic ultrasound has provoked an increased interest in educational approaches. Despite a steep learning curve, implementation of point-of-care ultrasound without simultaneous implementation of a structured, competency-based training programmes is a hazard to patient safety [65]. A regional survey revealed that 72% of residents had performed thoracic ultrasound or unsupervised pleural procedures before being a level 1 practitioner despite the Royal College of Radiologists’ recommendations [66]. Half of residents had access to organised education and training of pleural skills, but only half responded that they had access to supervision for difficult procedures [67]. Even though training programmes are provided at a European and, in several countries, at a national level, there is plenty of room for improvement; all the way down to an institutional level, focus on competence should be prioritised [68].

In order to improve safety and reduce complications following pleural procedures, the British Thoracic Society has published safety standards for invasive procedures [69]. A study among respiratory registrars found that these checklists are not used enough, but that they are more likely to be used if they are available in a dedicated procedure room or next to the procedural equipment [70].

During an oral session on skills training in the field of respiratory medicine, the limited access to education and skills training due to COVID-19 restrictions was highlighted, as this is an immediate and direct threat to patient safety. Development of innovative methods for ensuring high quality and evidence-based education and skills training in setting with COVID-19 restrictions should be a high priority.