Abstract
Chest imaging in patients with acute respiratory failure plays an important role in diagnosing, monitoring and assessing the underlying disease. The available modalities range from plain chest X-ray to computed tomography, lung ultrasound, electrical impedance tomography and positron emission tomography. Surprisingly, there are presently no clear-cut recommendations for critical care physicians regarding indications for and limitations of these different techniques.
The purpose of the present European Respiratory Society (ERS) statement is to provide physicians with a comprehensive clinical review of chest imaging techniques for the assessment of patients with acute respiratory failure, based on the scientific evidence as identified by systematic searches. For each of these imaging techniques, the panel evaluated the following items: possible indications, technical aspects, qualitative and quantitative analysis of lung morphology and the potential interplay with mechanical ventilation. A systematic search of the literature was performed from inception to September 2018. A first search provided 1833 references. After evaluating the full text and discussion among the committee, 135 references were used to prepare the current statement.
These chest imaging techniques allow a better assessment and understanding of the pathogenesis and pathophysiology of patients with acute respiratory failure, but have different indications and can provide additional information to each other.
Abstract
A variety of chest imaging techniques are now available for assessing patients with acute respiratory failure. This statement highlights characteristics, clinical indications and limitations of each technique as a guide for patient management. http://bit.ly/2XxYOd7
Introduction
Patients with acute respiratory failure require one or several imaging studies of the chest to diagnose underlying diseases, assess progression and evaluate treatment efficacy. Until a few decades ago, chest imaging of the critically ill consisted solely of chest X-ray (CXR). Additional imaging techniques have become widely available for the critically ill, including chest computed tomography (CT) and, more recently, positron emission tomography (PET) and bedside techniques such as lung ultrasound (LU) and electrical impedance tomography (EIT). Surprisingly, there are presently no clear recommendations for critical care physicians regarding indications for and limitations of these five imaging techniques.
To date, the limited use of chest magnetic resonance imaging (MRI) in pulmonary diseases is due to the physical properties of the pulmonary parenchyma and the long scan time. Although recent technical advances (i.e. parallel imaging, multi-array phase coils and ultra-short echo-time techniques enabling higher image quality and shorter scan time) have provided more detailed information on lung ventilation, inflammation, perfusion and structure, currently MRI is not used in the management of patients with acute respiratory failure. For these reasons, MRI was not included by the current task force.
The purpose of this European Respiratory Society (ERS) statement is to provide physicians with a comprehensive clinical review of chest imaging techniques for the assessment of patients with acute respiratory failure, based on the scientific evidence, as identified by systematic searches. Of the five imaging techniques selected by the Task Force Chairs, three are applicable at the bedside (CXR, LU, EIT) and two (CT and PET) require transfer to the radiological department.
For each of the included imaging techniques, the panel evaluated the following items: indications, technical aspects, qualitative and quantitative analysis of lung morphology and interaction with mechanical ventilation.
Methods
The ERS Scientific Committee approved the development of a document on imaging techniques in acute respiratory failure by a task force (TF-2016-01) on May 2016 aimed at summarising the relevant literature. The task force was composed of several experts and chaired by D. Chiumello and P. Navalesi. All members disclosed potential conflicts of interest according to ERS policies.
Search
A systematic search of the literature on the five imaging techniques (CXR, LU, CT, PET and EIT) was performed from inception to September 2018 on Medline/PubMed (National Library of Medicine, USA). The search was limited to articles in English and on humans aged >18 years. The keywords included in supplementary tables S1–S4 were used as literature search terms and limited to original studies.
Manuscript preparation
Task force members were divided into five groups (LP and AA for CXR; DC and GSP for PET; LH, MS, MZ and KM for CT; GP and BB for LU; PS for EIT). Each group focused on a single technique and its use in a disease entity selected a priori by the Task Force Chairs (pneumonia, chronic obstructive pulmonary diseases (COPD), acute heart failure, pneumothorax, pleural effusion, acute lung injury and acute respiratory distress syndrome (ARDS)). For each technique, a narrative summary was provided for contextualisation, which summarised certainty of evidence and relevant features. Feedback was provided by electronic communication. This is a statement document, thus no formal assessment of the evidence quality was performed and it does not include recommendations for clinical practice. The final approved version was peer reviewed.
Results of the literature search
Figure 1 provides an overview of the literature search. The initial search identified 1833 papers. After manual screening of titles and abstracts 1220 papers were excluded and 613 were considered as potentially relevant. After evaluation of the full texts and discussion among the committee, 493 were excluded for the following reasons: 350 were case reports, 61 were reviews, 25 were deemed too small a series (the expert and the methodologist agreed on two cut-offs: 10 subjects for PET and EIT, and 20 for CXR, LUS and CT), 16 were considered to be inappropriate for subject characteristics (e.g. paediatric studies) and 41 references were discarded for other reasons.
Flow chart of the study protocol process.
An additional search of bibliographies and authors' personal files provided nine papers. Six additional papers were added after updating the search in September 2018. Thus, a total of 135 references were used to prepare the current statement.
Chest X-ray
The studies on CXR used to prepare this statement are listed in table 1. Bedside CXR remains one of the most frequently prescribed imaging techniques, providing helpful information for the monitoring of critically ill patients. Although lots of effort has been made to improve technical aspects, numerous limitations of bedside CXR persist, including bedside attenuation factor, X-ray intensity and distance to the thorax, and synchronisation to mechanical ventilation. In addition, patients are often uncooperative and difficult to position, thus increasing procedure time and cost. All these factors may hinder the correct interpretation of CXR.
Studies included on CXR
ARDS
ARDS is defined by the acute onset of bilateral opacities on chest radiography not fully explained by effusions, lobar/lung collapse, nodules or oedema [1].
Although the radiographic criterion of “bilateral pulmonary infiltrates or opacities” on CXR is essential for the diagnosis of ARDS, significant inter-observer variability is reported [2]. Compared to chest CT, the accuracy of CXR in identifying pulmonary abnormalities in patients meeting criteria for ARDS is limited and unrelated to disease severity, as defined by the extent of oxygenation derangement [3]. A retrospective single-centre study showed that only 31% of patients with ARDS according to the current definition presented with histopathological signs of diffuse alveolar damage (a finding pathognomonic for ARDS) [4]. A recent study evaluated whether diagnostic accuracy could be improved by an educational intervention based on a set of chest radiographs. Though some improvement in diagnostic accuracy after the educational intervention was observed, the overall accuracy remained poor [5]. A multicentre randomised trial using an online educational module of the Berlin definition on ARDS failed to demonstrate any improvement in the accuracy or in the identification of ARDS patients among critical care clinicians and research staff [6].
Acute heart failure
CXRs performed at the bedside with portable equipment are often used to identify and quantify the presence of pulmonary oedema in the intensive care unit (ICU). However, a supine position and the potential presence of other multiple radiographic abnormalities often hinder CXR interpretation. In addition, the distinction between hydrostatic and permeability pulmonary oedema, as well as quantitating the amount of lung water, is extremely problematic [7]. Although a relationship has been found between a score based on CXR findings and the amount of extravascular lung water, the evolution of extravascular lung water was not associated with score changes [7]. To improve diagnostic accuracy in patients with acute heart failure, some radiological signs, such as the cardiothoracic ratio, vascular pedicle width and dimensions of the mediastinal silhouette of the great vessels, have been proposed [8, 9]. A vascular pedicle width exceeding 70 mm has acceptable sensitivity and specificity in discriminating cardiogenic from non-cardiogenic pulmonary oedema [10].
COPD
Although CXR is usually performed in patients with COPD exacerbation, it has low sensitivity for the detection of airflow obstruction. During acute exacerbation the predominant pathological changes are found within the airways, with abnormal images only in very few cases. In this setting, CXR is useful to rule out pneumonia or to exclude alternative diagnoses and complications such as decompensated heart failure, massive pleural effusion, atelectasis or pneumothorax.
Pneumonia
Conventional CXR is the preferred imaging modality in critical care settings for both the detection of new infiltrates and for monitoring response to therapy, with the possibility of detecting early complications including cavitations, abscesses, pneumothoraces and pleural effusions. However, owing to the well-known difficulties involved in radiographic interpretation of portable CXR in critical care, a diagnosis of pneumonia may be difficult. Using protected brush catheter specimens as the gold standard for the diagnosis of nosocomial pneumonia, Lefcoe et al. [11] demonstrated a CXR sensitivity in detecting pneumonia associated with positive cultures of 0.60 and 0.64, as interpreted by two radiologists, with low reproducibility between the two examiners. In a small group of patients with clinical suspicion of ventilator-associated pneumonia, the sensitivity of CXR in determining the presence of ventilator-associated pneumonia was 25%, with a specificity of 75% and an accuracy of 0.45 using protected brush catheter and microbiology as reference [12]. A multicentre French study found that the occurrence of alveolar consolidations with lobar distribution was more frequently associated with severe pneumococcal pneumonia, although the value of this prediction was rather limited [13].
Pneumothorax
Controversy exists regarding the optimal time interval to identify a pneumothorax on CXR after both chest tube positioning and removal. Schulman et al. [14] evaluated both an immediate and a delayed approach (following morning) with CXR after chest tube positioning. Of the 31 patients with a pneumothorax on follow-up, 22 were “early” and nine “late”, but none of the patients in the late group presented with a clinically significant pneumothorax. Concerning the timing of CXRs after tube removal, Pizano et al. [15] found that serial CXRs performed at approximately 1, 10 and 36 h did not identify different rates of pneumothoraces in mechanically ventilated patients.
Pleural effusion
Owing to technical limitations, pleural effusions, particularly when small, can be difficult to diagnose on CXR performed in a supine position. A previous report on the detection of pleural effusions on supine CXR showed an overall accuracy of 82%, with better results for larger pleural effusions (>300 mL) [16]. The accuracy for parapneumonic effusions is similar for anteroposterior and lateral CXR [17].
Ultrasound
The studies on LU used to prepare this statement are listed in table 2. Because the ultrasound beam does not penetrate the lung, LU is able to explore only the pleural line and the related artefacts that are generated at the pleural line. The pleural line appears as a hyperechoic sliding line, moving forwards and backwards in the course of inspiration and expiration. The key artefacts are A- and B-lines. B-lines correspond to various degrees of lung aeration and the quantity is related to the amount of extravascular lung water. Multiple and well-separated vertical B-lines correspond to a moderate decrease in lung aeration resulting from interstitial syndrome. Coalescent B-lines correspond to a more severe decrease in lung aeration resulting from partial filling of alveolar spaces by pulmonary oedema or confluent bronchopneumonia. Lung consolidation is a tissue-like echotexture pattern due to a loss of aeration of lung parenchyma [18].
Studies included on LU
ARDS
Lung infiltrates of the pulmonary alveolar interstitial space caused by ARDS may show consolidations and B-line patterns. However, these findings are nonspecific of ARDS because the syndrome encompasses consolidated regions, ground glass areas and normally aerated regions. As a consequence, LU patterns of ARDS may encompass B-lines, pleural line abnormalities (absent or reduced lung sliding, thickening or irregularities), consolidations and spared areas [19, 20]. These LU findings, together with impaired oxygenation, are simple tools to diagnose ARDS [21, 21], especially when combined with echocardiography [22]. LU has helped assess the incidence and outcomes of patients with ARDS in settings with limited resources [23].
The increase of lung water in ARDS can be detected by LU, and the LU score of B-lines is closely related to several prognostic indices [24].
Given that LU patterns are determined by the level of aeration, lung re-aeration can be assessed by tracking LU changes. This was first demonstrated by trans-oesophageal echography [25–27]. It has been proven that during positive end-expiratory pressure (PEEP) recruitment manoeuvres, the gas increase can be monitored by measuring the area of consolidated, dependent right lower lobes [28]. However, it may be more accurate to use scores based on a whole lung examination protocol [29]. Despite the fact that LU scores have proven to be a valid tool for assessing regional and global lung aeration, global LU score variations should not be used for bedside assessment of PEEP-induced recruitment. In addition, LU cannot be used to estimate lung hyperinflation [30].
Acute heart failure
Although nonspecific, B-lines can detect extravascular lung water. Simple scores based on numbers of B-lines correlate with surrogate markers of pulmonary oedema [31–33] because B-lines decrease during haemodialysis [34, 35]. B-lines could also be correlated with pulmonary wedge pressure [24, 36].
Cardiogenic pulmonary oedema results from increased intravascular hydrostatic pressure. As a consequence, the distribution of alveolar flooding is homogeneous. The LU pattern of cardiogenic pulmonary oedema is characterised by the presence of multiple B-lines in all thoracic regions [19]. Pleural effusion, moderately or severely decreased left ventricular function, and a small inferior vena cava diameter point towards cardiogenic pulmonary oedema instead of ARDS in acutely hypoxaemic patients [20]. LU has been used in the pre-hospital setting [37] and in the emergency department to manage patients with cardiogenic pulmonary oedema, thus proving to be a reliable bedside tool to guide therapy [38, 39].
COPD
In COPD patients, acute exacerbation usually shows a normal LU pattern despite the presence of acute respiratory failure. On the contrary, the presence of B-lines suggests the presence of an associated alveolar interstitial syndrome with an acceptable accuracy [40]. Chest ultrasonography (heart and lung) in patients admitted with acute respiratory failure allows a more accurate diagnosis of decompensated COPD compared to a standard diagnostic approach, based on physical examination, CXR and biological data [41].
Pneumonia
LU is a valid alternative for bedside diagnosis of lung consolidations in community-acquired pneumonia in adults [42], and can provide early detection of interstitial lung involvement in viral pneumonia [43]. In addition, LU can also provide guidance for transthoracic needle aspiration for aetiological diagnosis of patients with complicated pneumonia. This has been confirmed in the ICU and the emergency department [44–52].
Performing an accurate diagnosis of ventilator-associated pneumonia in mechanically ventilated patients frequently represents a challenge. By identifying ventilator-associated pneumonia-specific signs (focal areas of interstitial syndrome, small subpleural consolidations, large consolidations and fluid bronchogram), LU can discriminate pneumonia from resorptive atelectasis [53]. Furthermore, LU can accurately estimate the changes in lung aeration in patients with ventilator-associated pneumonia treated with antibiotics [54].
In a multicentre study, the diagnostic performance of a score based on the presence of subpleural consolidations, lobar consolidations and dynamic arborescent/linear air bronchograms was investigated in patients with suspected ventilator-associated pneumonia [55]. The LU-based score had a higher sensitivity and specificity in predicting ventilator-associated pneumonia than the clinical-based score (i.e. the Clinical Pulmonary Infection Score) [55].
Pneumothorax
Pneumothorax is the interposition of gas between visceral and parietal pleura. LU findings in pneumothorax are A-lines and the absence of lung sliding, B-lines and visualisation of a lung point [56]. The lung point corresponds to the area on the chest wall adjacent to the pneumothorax where the respiratory movement of the lung reappears. This transition between sliding and non-sliding pattern represents the limit of the pneumothorax and is a measure of its extension and volume [57].
In trauma patients in the emergency department, LU performs better than CXR in diagnosing pneumothoraces and can also detect the presence of occult pneumothoraces [58, 59]. A CXR is routinely requested after central venous catheter placement to exclude the presence of an iatrogenic pneumothorax. LU combined with contrast-enhanced ultrasonography has high accuracy in excluding both a pneumothorax and catheter malposition [60, 61], with a significant reduction in the mean time required for the examination [62]. In addition, LU can be used to exclude a pneumothorax, as an alternative to CXR, after chest tube removal [63].
Pleural effusion
LU can reliably identify free pleural effusion that appears as a dependent echo-free space between the parietal and visceral pleura [64]. LU may also allow semi-quantitative, clinically useful estimations of effusion volume. The expiratory interpleural distance measured at the thoracic base with ultrasonography has been proven to correlate with the fluid volume [65]; LU is able to detect pleural effusion in different clinical conditions, irrespective of the underlying disorder [66–69].
Ultrasound-guided thoracentesis in patients receiving mechanical ventilation reduces the risk of pneumothorax to <1% [70].
Miniaturised ultrasound systems such as hand-carried ultrasound imagers are now available. These systems allow a more prompt bedside diagnosis and immediate therapeutic measures, and could provide a helpful technique for the primary assessment of pleural effusions [71].
Computed tomography
The studies on CT used to prepare this statement are listed in table 3. CT scanning is frequently performed in critically ill patients, either at admission or later in cases of worsening respiratory failure. A retrospective analysis conducted in medical critically ill patients reported that a CT scan was performed in 11.5% of all patients admitted to the ICU [72]. Consolidations, pleural effusions and parenchymal abnormalities were each present in more than one-fifth of the patients. The most common CT findings included consolidations (46%), other parenchymal abnormalities (29%) and pleural effusions (35%). Clinical changes clearly linked to chest CT were observed in 24% of the patients [72].
Studies included on CT
ARDS
Description of the findings
Since the first description of chest CT in ARDS, the described disease patterns have included ground glass opacifications, parenchymal distortion, areas of consolidation, and reticular and linear opacities [73]. These alterations detected by CT scan are significantly related to the impairment of gas exchange and to the lung injury score [74]. ARDS patients are characterised by a lower end-expiratory lung gas volume and an increase in lung oedema with a typical diffuse or patchy distribution of attenuation in the lung [75–77]. However, owing to the inhomogeneous distribution of the disease, a single-slice CT, compared to an overall lung study, cannot accurately describe the amount of reopening of collapsed lung regions due to PEEP changes [78]. An important technical issue with CT is the spatial resolution. Vieira et al. [79] demonstrated that low spatial resolution CT can underestimate the degree of hyperinflation due to PEEP compared to high spatial resolution CT, particularly when the lung morphology has a focal loss of aeration.
ARDS is characterised by different levels of hypoxaemia due to different amounts of non-aerated lung regions (i.e. alveolar shunt), which can be precisely quantified by CT. The logarithmically transformed arterial oxygen tension (PaO2)/inspiratory oxygen fraction (FiO2) due to pure oxygen ventilation allows CT shunt to be estimated [80]. According to the recent Berlin definition of ARDS that proposed three exclusive categories according to the degree of hypoxaemia, a PEEP of 5 cmH2O should be applied to stratify patients at intensive care admission. This relatively low PEEP level is accurate in predicting the severity of hypoxaemia and the recruitability of the lung compared to higher PEEP levels [81]. At 5 cmH2O of PEEP the potential for lung recruitment is significantly different according to each ARDS category of the Berlin definition, being two and three times higher in patients with moderate and severe ARDS compared to mild ARDS [81], suggesting that low PEEP levels should be applied upon intensive care admission to stratify patients according to the severity of disease.
Patients with diffuse attenuations have a higher mortality rate compared to lobar attenuations [82]. In one study, up to 50% of patients with sepsis and ARDS had a CT scan score higher than that of survivors and fewer ground glass opacities [83]. Pulmonary findings on CT did not allow discrimination between a pulmonary and extrapulmonary focus of infection [83].
In a study of patients with ARDS caused by H1N1 influenza, the amounts of total lung consolidation and ground glass opacity were not different. However, the total lung consolidation significantly increased, whereas total lung ground glass opacity decreased from the anterior towards the posterior. The total lung disease was significantly higher in patients who required extracorporeal membrane oxygenation (ECMO) compared to those who did not require ECMO [84]. Chest CT has substantially changed the understanding and management of patients with ARDS. Simon et al. [85] reported that chest CT affected treatment in 27% of cases, in particular resulting in alteration in antibiotic therapy (8%), drainage of pleural fluid (8%) and modification in antimycotic therapy (4%). Major disadvantages of CT are that the patient has to be transported to the radiological department and the radiation exposure. Chiumello et al. [86] demonstrated that low-dose CT has a high agreement with conventional CT for quantitative analysis in ARDS patients. The current standard technique for quantitative CT scan analysis is based on a manual lung segmentation, which is time-consuming and depends on the skill of the operator. Klapsing et al. [87] reported very good precision for an automatic lung segmentation software program compared to manual segmentation. This automatic lung CT segmentation was able to reduce the processing time by >99%.
Trauma patients are at risk for developing ARDS, and CT could be used to detect possible lung and heart disease [85]. In a study of chest trauma patients with pulmonary contusion, the volume of the contusion was related to higher risk of ARDS [88]. Quantitative CT scan analysis offers the possibility of computing the total lung weight and could be used to discriminate lung atelectasis from consolidation. In a group of trauma patients with ARDS, 60% of the patients had a lung weight volume similar to that of trauma patients without ARDS, suggesting a higher amount of atelectasis compared to consolidation [89].
In the early phase of ARDS, the amount of pleural effusion is quite modest (an average of 340 mL) and does not affect the respiratory system elastance, amount of lung collapse or degree of oxygenation [90].
Treatment effect
Chest CT has been used to determine complications of mechanical ventilation or in the follow-up of ARDS patients. Late ARDS has a significantly higher incidence of pneumothoraces and number of bullae compared to early ARDS [91]. Treggiari et al. [92] performed chest CT in ARDS patients on prolonged ventilation (interval between ARDS and CT scan 22±19 days). They found that development of air cysts and bronchiectasis in ventilated patients with ARDS mainly occurred in non-dependent lung regions and severity correlated with peak pressures.
Although the mortality rate of ARDS patients has significantly decreased through the years, it still ranges between 40% and 50%, with surviving patients having a significant reduction in their quality of life. In a study of acute lung injury survivors, decrements in quality of life attributable to pulmonary dysfunction were strongly associated with higher radiological scores [93]. In a small study, 87% of patients with ARDS (13 out of 15) exhibited fibrotic changes in the lung, in particular in the ventral parts, as assessed by high-resolution CT [94]. A significant correlation was found between the severity of ARDS and the severity of CT findings. In a subsequent study [95], lung abnormalities were found on high-resolution CT in 75% of patients 6 months after recovery. A reticular pattern was the most frequent finding. A predominance for the ventral parts was noted in 37% of the patients. Kim et al. [96] found that patients with pulmonary ARDS had more severe lung sequelae on chest CT after 20±12 months compared to patients with extrapulmonary ARDS.
CT has previously been used to evaluate or predict the response to PEEP changes and to recruitment manoeuvres in patients with ARDS [97]. Analysing lung recruitability as the decrease in non-aerated tissue from 5 to 45 cmH2O of PEEP, an average of 15% of the total lung tissue was found to be unrelated to the amount of compressive forces (lung oedema and characteristics of the chest wall). This suggests that the amount of PEEP required to keep the lung open is independent of the amount of tissue which should be kept open, and that factors such as the distribution of oedema within the lung being mainly intra- or extra-alveolar and the nature of the disease play an important role [97]. In patients with diffuse attenuation, PEEP induced significant alveolar recruitment without over-distension; in patients with lobar CT attenuation, PEEP induced mild alveolar recruitment with over-distension of the already inflated lung regions [98]. Patients with focal ARDS at zero end-expiratory pressure are at increased risk of hyperinflation during recruitment manoeuvres and are less likely to show recruitment compared to patients with a non-focal lung morphology [99]. The effect of body mass index in ARDS is not associated with significant differences in lung recruitability and respiratory mechanics [100]. In addition, according to protective ventilation, between 10% and 30% of the potentially recruitable lung always remains closed. Furthermore, increasing PEEP up to 15 cmH2O does not prevent the cyclic lung tissue opening and closing [101].
In an ARDS lung the distribution of the lesions (consolidations and atelectases) is inhomogeneous, promoting a regional increase in transpulmonary pressure, acting as stress raiser. Thus, a safe transpulmonary pressure could become harmful (i.e. reaching high levels) in the presence of stress raisers. It has been found that the extent of lung inhomogeneity increases with the severity of ARDS; increasing PEEP significantly decreases the amount of lung inhomogeneity [102].
Randomised clinical trials comparing high and low PEEP values in ARDS have not find any difference in the outcome, probably owing to several factors not being taken into account, such as the potential for lung recruitability, the amount of oedema or disease severity. When different PEEP selection methods (based on lung mechanics, oesophageal pressure and oxygenation) were compared according to lung recruitability, the oxygenation method provided higher PEEP levels (i.e. higher PEEP in patients with higher recruitability) [103].
Constantin et al. [104] compared two recruitment manoeuvres, CPAP with 40 cmH2O for 40 s versus PEEP maintained at 10 cmH2O above the lower inflection point of the pressure volume curve for 15 min. Although the increase in oxygenation was different, lung recruitment estimated by CT was significantly lower with the CPAP manoeuvre. Galiatsou et al. [105] demonstrated that pronation in patients with ARDS recruited lung tissue in dependent lung areas and reversed overinflation of the ventral areas. Surfactant deficiency in ARDS also contributes to alveolar derecruitment. The administration of surfactant in mechanically ventilated patients is associated with a significant increase in the volume of gas in poorly/non-aerated lung areas and a significant increase in tissue volume in normally aerated lung areas [106]. Concerning the estimation of lung recruitment with CT or the pressure volume curve, they are well related despite having very large limits of agreement [107].
Two studies have used CT to compare different ventilator modes on lung aeration in patients with ARDS [108, 109]. The first found that airway pressure release ventilation significantly decreases the amount of atelectasis and increases the normally aerated lung volume compared to pressure support ventilation [108]. By contrast, a latter study failed to identify any differences between airway pressure release ventilation and pressure support ventilation from admission to day 7 [109].
Acute heart failure
Practical issues limit the application of CT in the acute phase for the diagnosis of acute cardiogenic pulmonary oedema. In a retrospective study, CT markers for acute pulmonary oedema (i.e. engorged peripheral pulmonary vessels, thickening of inter- and intra-lobular septa, ground glass opacities and consolidations) were compared with transpulmonary thermodilution technique variables [110]. The authors concluded that haemodynamic parameters obtained with transpulmonary thermodilution cannot be accurately estimated by CT. In a small study in ARDS patients, a good correlation was found between transpulmonary thermodilution and CT markers for pulmonary oedema [111]. Patients with acute pulmonary oedema presented with a similar amount of ground glass attenuation and a lower amount of airspace consolidation [112].
Pneumonia
CT is more sensitive than CXR in detecting pulmonary infiltrates in patients with clinical suspicion of pneumonia [113, 114]. Similarly, in patients admitted to the emergency department with clinically suspected community-acquired pneumonia, CT modified the likelihood of diagnosing community-acquired pneumonia in 58% of cases [115].
In addition, CT can also provide a detailed morphological description of patients with ventilator-associated pneumonia. CT scans of a group of patients with ventilator-associated pneumonia at diagnosis and at day 7 of antimicrobial therapy were characterised by the presence of intraparenchymal and subpleural rounded CT attenuations disseminated within the upper and lower lobes, with consolidations of the lower and upper lobes [54]. Patients who responded successfully to antimicrobial therapy showed predominant disappearance of the rounded opacities, whereas antibiotic failures correlated with new onset of rounded opacities within the lungs. A significant correlation was found between chest CT diagnosis of pneumonia and electronic nose sensor of the expired gases, a new promising adjunct tool for the diagnosis of pneumonia [116].
COPD
Patients with severe acute exacerbation of COPD, due to infection or cardiac failure, frequently require mechanical ventilation with PEEP to improve oxygenation and to reduce the work of breathing. However, the increase in PEEP is associated with an increase in lung volume with possible risks of overinflation [117].
In addition, COPD patients are at increased risk of developing pulmonary thromboemboli. During a 5-year follow-up from an acute exacerbation of COPD, 17% of patients developed pulmonary embolism; the ICU length of stay and mortality were significantly higher in patients with pulmonary embolism [118].
Pneumothorax
CT is commonly used for the diagnosis of pneumothorax. It is known that supine CXR is not sensitive for the diagnosis of pneumothorax in non-ICU patients [119, 120].
Very few studies have compared conventional CXR with CT in ICU patients. However, in a prospective study of 42 ICU patients, none of the eight pneumothoraces diagnosed by CT were seen with CXR [121].
Pleural effusion
The quantitative computation of pleural effusion with whole chest CT has been demonstrated to significantly relate to the amount of pleural effusion computed with LU using a multiplanar ultrasound approach considering the cephalocaudal extension and the area measured at mid length [122].
PET
The studies on PET used to prepare this statement are listed in table 4. PET provides a functional examination, detecting the presence of a radioactive tracer that is usually administered to patients linked to a biological molecule. One of the most common tracers is [18F]-2-fluoro-2-deoxy-d-glucose (18FDG). In the presence of an inflammatory status there is an increase in cellular metabolism and glucose consumption, mainly linked to neutrophilic activity.
Studies included on PET
ARDS
ARDS lung is characterised by an increase in pulmonary vascular permeability in addition to abnormalities in gas exchange [1]. Pulmonary vascular permeability can be assessed by the pulmonary transcapillary escape rate for transferrin with PET (PTCER), which evaluates the protein flux between the pulmonary intravascular and extravascular compartments. ARDS and pneumonia patients have been found to present with a significantly higher PTCER than heart failure patients and healthy subjects [123]. In a group of pneumonia patients, PTCER was also higher in the regions contralateral to focal pneumonia [123]. ARDS in the early phases has been found to have a higher PTCER compared to the late phases [124], which is still higher than that of healthy subjects [124].
The current lung ARDS model, extensively explained by CT, indicates that regions of normal aeration coexist with poorly and non-aerated lung regions, and that lung densities are mainly located in the dependent lung regions [125]. Similarly, PET shows a significant increase in lung density in dorsal compared to ventral lung regions, with a higher amount of the same lung regions compared to healthy subjects [126]. Surprisingly, no difference has been found in the PTCER distribution between the dorsal and ventral lung regions. Despite a lack of difference between the ventral and dorsal regions, PTCER was not uniformly distributed in ARDS patients, thereby suggesting a possible blood-borne delivery of injurious agents to the lung [126].
Combining PET and CT with 18FDG it is possible to assess the distribution and magnitude of inflammation within the lung. Different approaches have been proposed, such as the simple static model that measures the standardised uptake volume, dynamic models that analyse the spectral analysis filter, and the Patlak analysis. In a comparison of the static and dynamic models in ARDS patients, the dynamic model provided a better description of lung inflammation [127]. In ARDS patients, the metabolic activity of the lungs was significantly higher than in healthy subjects, and did not correlate with the mean lung density or with the relative weight of either non-aerated or normally aerated tissue [128]. The inflammation activity negatively correlated with oxygenation levels. In the normally aerated tissue the metabolic activity was significantly higher, up to seven times, than that of healthy subjects. Additionally, lung inflammation was very differently distributed considering the distribution of inflation (from non-inflated to well-inflated regions) [128].
The same authors analysed the relationship between gas volume changes induced by tidal ventilation, from end expiration to end inspiration, and pulmonary inflammation [129]. The lung regions undergoing intra-tidal recruitment and de-recruitment during tidal breathing had similar lung inflammation to the collapsed ones. Airway pressure positively correlated with lung inflammation [129].
Owing to the greater distribution of lung oedema in the dorsal regions, greater lung inhomogeneity of lung parenchyma may be present along the sternum–vertebral axis. Concerning the distribution of lung inhomogeneity and inflammation within the lung, the amount of lung inflammation and inhomogeneity has been shown to increase from mild to severe ARDS [130]. In that study, the homogeneous lung compartment with normal PET signal was mainly composed of well-inflated tissue and was located in the ventral regions. By contrast, the inhomogeneous compartment with high PET signal was composed of non- or poorly inflated tissue and located in the dorsal regions. The homogeneous lung compartment with high PET signal comprised mixed lung aeration from non-inflated to well-inflated regions and was similarly distributed within the lung [130].
Acute heart failure
Pulmonary hypertension, which is frequent in ARDS patients, has been associated with pulmonary vasoconstriction in response to hypoxia, which could redistribute pulmonary blood flow within the sick lung. The ventral to dorsal regional distribution of pulmonary blood flow has been analysed in a group of patients with pulmonary lung oedema and in healthy subjects [131]. Although the amount of lung water concentration was significantly higher in ARDS and cardiogenic pulmonary oedema, the regional distribution of pulmonary blood flow was similar among ARDS and healthy subjects [131].
Pneumonia
By analysing the uptake of tracer by activated inflammatory cells, PET can provide a quantitative assessment of lung infection and assess the response to therapy. The pulmonary transcapillary escape rate has proved to be significantly higher in areas of radiographic infiltrates in patients with pneumonia compared to normal subjects [123].
Cystic fibrosis patients are characterised by persistent lung inflammation with high levels of neutrophil activation, translating clinically into frequent episodes of lung infections. In patients with cystic fibrosis, PET showed a higher uptake of 18FDG compared to healthy subjects, and this feature positively correlated with the number of neutrophils in the bronchoalveolar lavage fluid [132].
PET has also been proposed in the diagnosis of interstitial lung diseases. In a small group of patients with diffuse interstitial lung disease, the tracer uptake was not different in patients with and without idiopathic pulmonary fibrosis (IPF) [133]. By contrast, in cryptogenic organising pneumonia the tracer uptake was significantly higher compared to IPF and nonspecific interstitial pneumonia (NSIP), while similar levels were detected between IPF and NSIP [134].
COPD
COPD patients are characterised by significant alterations in the distribution of ventilation and perfusion. By applying an innovative PET analysis, Vidal et al. [135] analysed ventilation and perfusion within the lung imaging resolution unit (voxel). There was greater perfusion heterogeneity in COPD, compared to in healthy subjects, with no dorsoventral ventilation gradient.
Pleural effusion
Owing to the intrinsic differences in glucose metabolism between normal and tumour cells, PET can distinguish benign from malignant pleural effusions. Malignant pleural effusions have shown a significantly higher glucose uptake compared to benign effusions, with the technique having good sensitivity and a relatively low specificity in detecting malignant pleural effusions (93% and 68%, respectively) [136, 137].
EIT
EIT uses multiple electrodes applied to the external chest surface and the application of a low voltage current to measure both absolute and relative variations of body impedance. A two-dimensional image, of approximately 10 cm, is created with good correlation to intrapulmonary lung gas volume and intrathoracic blood volume. EIT applications range from monitoring mechanical ventilation (PEEP selection, lung recruitability, distribution of ventilation) to estimating lung perfusion and pulmonary function [138–140].
Based on our criteria selection, no articles were found on this topic.
Conclusions and need for future research
Patients with acute respiratory failure due to different lung causes and high mortality risk require numerous lung studies. As a first level examination, CXR, despite its intrinsic limitations and low accuracy, may still play a relevant role. CT remains the gold standard, but it requires patient transportation and use of radiation, which preclude extensive use especially within the same patient. LU, after proper physician training, is able to provide greater accuracy than CXR, similar to that of CT. EIT is gaining a more clinical role after many years as a research tool, while PET has a minimal role in the acute phase of respiratory failure.
As bedside lung imaging techniques, LU and EIT will become more frequently used in patients with acute respiratory failure. Future studies will assess if the information provided will improve clinical management and outcome. Regarding CT, the possibility of dose reduction protocols and safer patient transport to the radiology department will extend its application. There is a clinical need for studies combining different methods for diagnosis and patient monitoring.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary table S1. Keywords used for search on Medline for chest-x-ray. ERJ-00435-2019.Table_S1
Supplementary table S2. Keywords used for search on Medline for lung ultrasound. ERJ-00435-2019.Table_S2
Supplementary table S3. Keywords used for search on Medline for computed tomography. ERJ-00435-2019.Table_S3
Supplementary table S4. Keywords used for search on Medline for positron emission tomography. ERJ-00435-2019.Table_S4
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00435-2019.Shareable
Footnotes
This statement was endorsed by the ERS Executive Committee on June 7, 2019.
This article has supplementary material available from erj.ersjournals.com
Conflict of interest: D. Chiumello has nothing to disclose.
Conflict of interest: G.F. Sferrazza Papa has nothing to disclose.
Conflict of interest: A. Artigas reports grants from Grifols, Fisher & Paykel, Fundacion Areces and Instituto Carlos III, outside the submitted work.
Conflict of interest: B. Bouhemad has nothing to disclose.
Conflict of interest: A. Grgic reports personal fees from MSD, Boehringer Ingelheim, Roche and Bayer Vital, outside the submitted work.
Conflict of interest: L. Heunks reports personal fees for travel and lecturing from Maquet critical care, and grants from Ventfree and Orionpharma, outside the submitted work.
Conflict of interest: K. Markstaller has nothing to disclose.
Conflict of interest: G.M. Pellegrino has nothing to disclose.
Conflict of interest: L. Pisani has nothing to disclose.
Conflict of interest: D. Rigau works as a methodologist for the ERS.
Conflict of interest: M.J. Schultz has nothing to disclose.
Conflict of interest: G. Sotgiu has nothing to disclose.
Conflict of interest: P. Spieth has nothing to disclose.
Conflict of interest: M. Zompatori has nothing to disclose.
Conflict of interest: P. Navalesi has nothing to disclose.
- Received March 1, 2019.
- Accepted May 16, 2019.
- Copyright ©ERS 2019
References
