Chapter 11
Navigational bronchoscopy in solitary pulmonary nodules
Ralf Eberhardt1 and Joris van der Horst2
1Pneumology and Critical Care Medicine, Thoraxklinik, University of Heidelberg, Translational Lung Research Center Heidelberg (TLRCH) and Member of the German Center for Lung Research (DZL), Heidelberg, Germany. 2Respiratory Dept, Glasgow Royal Infirmary, University of Glasgow, Glasgow, UK.
Correspondence: Ralf Eberhardt, Pneumologie und Beatmungsmedizin, Thoraxklinik am Universitätsklinikum Heidelberg, Röntgenstrasse 1, 69126 Heidelberg, Germany. E-mail: ralf.eberhardt@med.uni-heidelberg.de
Suspicious peripheral pulmonary nodules need to be clarified histologically. Depending on the probability of malignancy and the patient’s comorbidities, the lesion can be resected surgically or a nonsurgical biopsy is necessary. The challenge in diagnosing a peripheral parenchymal nodule by bronchoscopy is to detect the nodule endoscopically, especially if it is not visible on fluoroscopy. Apart from TBB under fluoroscopic guidance, various navigation techniques such as radial EBUS, virtual bronchoscopy and electromagnetic navigation bronchoscopy are available to increase the diagnostic yield. Further developments are necessary in order to make bronchoscopic treatment of small malignant peripheral lesions possible, ideally in a one-step diagnostic and therapeutic procedure.
Cite as: Eberhardt R, van der Horst J. Navigational bronchoscopy in solitary pulmonary nodules. In: Herth FJF, Shah PL, Gompelmann D, eds. Interventional Pulmonology (ERS Monograph). Sheffield, European Respiratory Society, 2017; pp. 162–175 [https://doi.org/10.1183/2312508X.10003317].
Lung cancer is one of the most common cancers in both men and women worldwide. Prognosis is strongly dependent on the stage of disease at presentation. Although the risk factors are well known, the majority of patients are diagnosed at an advanced stage of disease, making cure with currently available techniques unlikely. In contrast, patients with an early tumour stage can be treated curatively with surgical resection or radiotherapy and they will have improved survival [1].
The desire for early detection of lung cancer has led to the idea of lung cancer screening. Initial attempts to screen for lung cancer with sputum analysis and/or radiographs of the thorax did not show a survival benefit for participants; however, more recent lung cancer screening trials utilising low-dose CT have been more promising. The 2010 National Lung Screening Trial (NLST), completed after enrolling more than 53 000 subjects, was for the first time able to show a decrease in lung cancer mortality [2]. In screening trials using chest radiographs, pulmonary nodules were noted in only 0.2% of patients, whereas 27.3% of patients undergoing the NLST had at least one nodule with a diameter of >4 mm detected on their CT, with the majority of these considered benign [2, 3]. Several other screening trials performed to date also detected a high number of pulmonary nodules [4, 5]. Furthermore, the generally increased use of thorax CT for other medical indications has led to a substantial increase in the number of incidental findings, including nodules (figure 1).
Figure 1. a) CT scan and b) corresponding PET scan from a 56-year-old patient with a 12 mm solitary pulmonary nodule in the middle lobe, suspicious for lung cancer. An acinar adenocarcinoma was diagnosed after surgical resection.
A solitary pulmonary nodule (SPN) is defined radiologically as an intraparenchymal lung lesion that is <3 cm in diameter and is not associated with atelectasis or adenopathy [6]. After radiological detection of a SPN the significant challenges are the classification of the nodule as benign or malignant and, more particularly, the diagnosis of a specific disease. The probability of malignancy of an incidentally detected pulmonary nodule in a CT scan depends on its size or diameter. Even in a high-risk group of heavy smokers the estimated risk of attaining a diagnosis of lung cancer is <1% when a pulmonary nodule with a diameter of <6 mm is found [5]. However, the risk of malignancy increases in line with the size of the nodule. The likelihood of malignancy for peripheral SPNs between 0.8 and 2.0 cm in diameter is reported to be ∼18%, and for nodules >2.0 cm, ∼50% [7, 8]. Suspicious morphology or upper lobe location are further features that increase the probability of malignant disease in these patients [9].
Diagnosis of solitary pulmonary nodules
One of the challenges in the early diagnosis of lung cancer remains the difficulty in reaching these small lung lesions, detected by radiography or CT, and successfully obtaining adequate tissue samples for pathological diagnosis. Due to the low prevalence of malignancy among small lung nodules <6 mm, nodules of this size are usually followed up by a low-dose CT scan. With both increasing size and pre-test probability of malignancy, e.g. due to predisposing risk factors, a nonsurgical biopsy and/or a surgical resection are/is usually recommended [1, 10].
In cases with a high probability of cancer, a direct surgical resection should be considered in order to diagnose and treat the nodule in one session. Ideally, a primary surgical approach would lead to resection of malignant nodules while sparing benign nodules. There are, however, several good reasons to avoid primary surgery and to perform a nonsurgical biopsy: 1) some patients may wish to have malignancy confirmed before contemplating surgery; 2) the investigator may be concerned about a benign diagnosis, which would not require resection; and 3) perhaps most importantly, significant patient comorbidities or poor surgical fitness may counsel against a primary surgical approach [11].
In a retrospective analysis of patients referred for surgical resection, 73% were found to be less than ideal candidates for surgery. More than half of the patients had to be excluded due to contraindications against surgery or an unacceptably high peri- or postoperative risk of morbidity and mortality [12]. In cases like these the confirmation of malignancy prior to any treatment should be attempted with minimally invasive methods.
The two standard approaches for diagnosing peripheral pulmonary lesions by minimally invasive methods used currently are transthoracic needle biopsy and bronchoscopy with TBB. A nonsurgical biopsy should be considered whenever imaging results and pre-test probability are discordant and the probability of malignancy is low to moderate [10, 12] or when comorbidities and surgical risk counsel against a surgical approach. Although no studies directly comparing these two approaches are available, the methods appear to be complementary. In general, transthoracic needle biopsy is preferred when the nodule is more peripheral or in the subpleural space, whereas a bronchoscopic approach is favoured for more centrally located lesions, especially where a bronchus sign is present, i.e. a bronchus is seen to lead towards the SPN [13].
The advantage of transthoracic needle biopsy under CT guidance is the high sensitivity with a specificity of nearly 100%. The sensitivity for malignancy is between 74% and 96%, and depends on the size of the lesion, location and distance to the pleura as well as the biopsy technique employed [10, 14]. The most common complication is a post-interventional pneumothorax; the risk is related to the size of the lesion, number of biopsies, distance to the pleura and degree of accompanying lung emphysema. A meta-analysis of 15 865 patients showed a pneumothorax rate of 15% and an insertion of a chest tube was needed in >6% of all procedures. The frequency of haemoptysis after transthoracic needle biopsy was ∼1% [15].
The standard endoscopic approach for diagnosing SPN is bronchoscopy with TBB using forceps. Other techniques described are cryobiopsy, TBNA or catheter aspiration [16–18]. A meta-analysis has shown that TBNA alone or combined with forceps is superior to TBB by forceps alone in diagnosing SPNs [18].
The bronchoscopy is usually performed under fluoroscopic guidance to aid steering the biopsy tool to the peripheral lung lesion. The value of this method depends on the size of the lesion, relationship of the nodule to the airways and visibility under fluoroscopy. In a meta-analysis the diagnostic yield for peripheral lesions >2.0 cm was 63%, whereas for nodules <2.0 cm the diagnostic yield dropped to 33% [19].
The use of CT for steering the biopsy tools to the lesion improves both the image quality and visual control, but despite that does not appear to improve the diagnostic yield or reduce the complication rate [20, 21]. Furthermore, the radiation exposures for the patient and staff as well as the demand on CT scanner time limit its usefulness in routine practice.
The advantage of the bronchoscopic approach in diagnosing SPNs over CT-guided transthoracic needle biopsy is its lower complication rate, with a pneumothorax rate of <3% [22]. However, the diagnostic yield has historically been lower than with CT-guided transthoracic needle biopsy. Bronchoscopic biopsies using forceps can be performed under medication with acetylsalicylic acid 100 mg·day–1, but should be considered carefully under clopidogrel due to the higher risk of bleeding [23, 24].
Navigational bronchoscopy
Improving the diagnostic yield of bronchoscopic sampling while preserving its advantage of lower complication rates when compared with a transthoracic approach remains a challenge, and will depend on improved techniques for navigation towards the lesion and better manoeuvrability of the sampling probes.
Radial EBUS
One of these technologies is radial EBUS, which can be used to locate and assess peripheral pulmonary lesions. Although EBUS probes have been developed for assessing the central airways and the mediastinum, the smaller calibre radial ultrasound probes can be placed through the working channel of a flexible bronchoscope into the peripheral lung, and can also be used to detect and analyse peripheral pulmonary lesions.
Technical aspects
These so-called miniprobes are available in different sizes with an external diameter of 1.4 or 1.9 mm (Olympus, Tokyo, Japan) or 1.7 mm (Fujinon, Tokyo, Japan). The most commonly used ultrasound frequency is 20 MHz, which provides high resolution and allows detailed imaging of the internal structures of peripheral lung lesions. In normal, ventilated and air-filled lung parenchyma, all ultrasound waves will be reflected and the miniprobe produces a snowstorm-like “white-out” ultrasound picture, even when the lesion is close to the tip of the probe, but separated by a small intervening layer of air. If the lesion can be reached endobronchially and the tip can be placed within or adjacent to the lesion, the image will change (figure 2).
Figure 2. a) Ultrasound image of a peripheral aerated lung: the ultrasound waves are reflected completely and only a whitish snowstorm-like image is visible. b) After reaching the peripheral lesion, the image changes and the pulmonary nodule is visible adjacent to the probe.
Solid tumours are usually clearly distinguishable against the normal lung tissue by a bright border. The sono-morphological image of a tumour appears grey and mostly homogeneous, although necrotic areas and vessels can be seen as circumscribed black areas. Furthermore, the existence of a continuous hyperechoic margin and the absence of a linear-discrete air bronchogram should raise the suspicion of malignancy. In contrast, ultrasound images of inflammatory tissue or atelectasis have a heterogeneous appearance, caused by the various different structures of the lung. Small bronchi containing trapped air are visible as sharp, white echo spots; the fluid-filled areas appear dark and the borders are slightly blurred (figure 3) [25–27]. However, in patients with lepidic growth pattern carcinoma, the ground-glass opacities have an appearance similar to inflammatory tissue, and a distinction between benign and malignant appearances is not possible here.
Figure 3. a) Typical EBUS image of a malignant tumour (nonsmall cell lung cancer) with solid structure and clear borders to the surrounding lung. The miniprobe is placed within the lesion. b) EBUS image of a pneumonic infiltrate with white spots and blurred borders.
Although the use of radial miniprobes in the peripheral lung is technically straightforward, it can sometimes be difficult to interpret the findings. Fluids appear black in the EBUS images, but no Doppler mode is available in the radial ultrasound to differentiate between necrotic areas and vessels. Here, tracing the course of the structure or looking for arterial vascular pulsation can be instructive. Trapped air leads to sharp, white spots with a “comet tail” sign behind which should not be confused with calcifications. Any strong reflections will create repeating echoes that can be recognised by their consistent interval distance.
EBUS-guided bronchoscopy
The EBUS miniprobes can be used like a flexible forceps probe and can be advanced into the relevant subsegmental bronchi where the lesion is suspected. This can be difficult when the probe has to be flexed; in addition, excess friction has to be avoided to protect the transducer and the connecting driving wire from damage. A water-filled balloon over the transducer is not usually necessary in the peripheral lung. The pulmonary nodule has been reached when the ultrasound image shows a solid round or oval lesion. Unfortunately, the biopsies cannot be taken under real-time visual control. The miniprobe has to be removed before introducing a biopsy tool into the working channel. The difficulty of then having to navigate back to the lesion can be reduced by using a guide sheath.
Following EBUS guidance, TBB using flexible forceps is the most common sampling technique. The diagnostic yield depends on the size and the visibility of the pulmonary nodule in the ultrasound image. Smaller SPNs are more difficult to reach, but the yield will be higher if a bronchus leads directly to the lesion and the probe can be placed within the lesion [28]. However, alternative or additional biopsy tools can be used. The additional use of a TBB needle (TBNA) or a thin aspiration catheter increases the diagnostic yield. This applies in particular if the miniprobe can only be placed adjacent to, but not directly inside, the nodule [16, 29]. Cryoprobes are used widely for taking endobronchial biopsies or TBBs to diagnose endobronchial tumours or ILDs [30, 31]. These probes are also effective in sampling peripheral lung nodules. The advantage of cryobiopsies are the lager size of the samples, better preserved histological architecture with fewer crush artefacts and possibility to obtain tissue from beyond the bronchial wall of the leading bronchus [17]. However, the bronchoscope has to be removed altogether with the cryoprobe and attached sample, which then requires repeated navigation towards the lesion.
The first description of the use of radial miniprobes for detecting peripheral lung lesions was in 2002 [32]. Currently, three different techniques for EBUS guidance are available: The first is to take biopsies “blindly” from the same subsegment after detecting the lesion and removing the EBUS probe. The second and most common approach is to use fluoroscopy to direct the miniprobe to the lesion and to check that the biopsy tool is in the same position that the EBUS probe was in at the time of tumour imaging, before then taking biopsies. The third technique involves advancing the EBUS miniprobe inside a guide sheath catheter used as an extended working channel (EWC). The probe is inserted into a small catheter and both together are advanced through the working channel of the bronchoscope to the peripheral lung [33]. The miniprobe is removed after reaching the lesion, whereas the guide sheath is left in place close to or within the nodule. It is then possible to introduce various biopsy tools through this EWC directly to the region of interest. This technique can also be used for diagnosing peripheral pulmonary lesions that are not visible under fluoroscopy or that are <20 mm [34, 35].
STEINFORT et al. [36] undertook a systematic review and meta-analysis of 13 studies and 1090 patients undergoing EBUS-guided bronchoscopy for diagnosing peripheral lung lesions. Although significant interstudy variation in the EBUS method was noted, the authors found a pooled sensitivity of 73% and specificity of 100% for radial EBUS-guided biopsies. The pneumothorax rate was 1.0% and no severe bleeding was observed. Moreover, the complications should be attributed to the use of forceps biopsy per se and not the use of radial miniprobes, and if the assessment time is limited no increased complication rate should be expected due to a prolonged sedation time. Therefore, EBUS is a safe and relatively accurate tool in the investigation of SPNs, increasing the likelihood of achieving a diagnosis and decreasing the need for surgical biopsy. In a retrospective analysis of four randomised trials with 481 patients, YE et al. [37] found that fluoroscopic TBB plus EBUS was superior to TBB under fluoroscopy alone and the clinical benefit was higher in smaller lesions.
The diagnostic sensitivity of EBUS is also influenced by the prevalence of malignancy in the patient cohort being examined. The yield for EBUS-guided TBB was only 57% in a clinical registry [38]. However, this can in part be explained by selection bias, because the choice of technique for diagnosis was not prescribed and varied from CT-guided biopsy or conventional TBB to TBB with EBUS guidance. In addition, EBUS was used particularly in smaller lesions.
The use of radial EBUS as an adjunct imaging modality is recommended in patients suspected of having lung cancer in the current guidelines of the American College of Chest Physicians, if the technology is available [10].
Combining radial EBUS with other navigational techniques may further improve diagnostic accuracy in peripheral pulmonary lesions.
Virtual bronchoscopy
Ultrathin bronchoscopes with an outer diameter of 3–4 mm have now been available for a few years. These bronchoscopes can be advanced far into the peripheral lung, which may improve access to more peripheral SPNs. However, the anatomical variability of bronchi increases with each generation of subsegments and orientation towards the target becomes increasingly difficult with each bifurcation passed, even for very experienced operators.
Raw CT scan data can be used to generate a three-dimensional virtual bronchoscopy with visual representation down to the sixth generation of bronchi, which allows planning of a virtual pathway towards the target lesion [39]. Following this virtual pathway towards the lesion will significantly simplify navigation in the periphery for the operator and has the potential to enable good results even for less experienced operators, without much additional training required. At present, two different systems are commercially available: LungPoint (Broncus Medical, Mountain View, CA, USA) and Bf-NAVI (Cybernet Systems, Tokyo, Japan). The latter has recently been replaced by the DirectPath system (Cybernet Systems). Generally, two phases can be distinguished: 1) the planning of the pathway and 2) the actual bronchoscopy procedure with biopsy sampling.
Virtual bronchoscopy planning
Although a virtual bronchoscopy can be easily reconstructed with raw diagnostic CT data, the use of virtual bronchoscopy-guided bronchoscopy for diagnosis of peripheral pulmonary nodules is not widespread. For high-quality virtual bronchoscopy a slice thickness of the CT of <1 mm is necessary to achieve a virtual bronchial tree far enough into the peripheral lung [39]. Thicker CT slices, breathing artefacts and/or excessive endobronchial secretion all shorten the visual reconstructed bronchial tree. The CT data have to be uploaded in the DICOM format (Digital Imaging and Communications in Medicine; National Electrical Manufacturers Association, Rosslyn, VA, USA).
The LungPoint computer-based navigation system allows reconstruction of the thorax and the pulmonary nodule in three dimensions. The peripheral lesion can be marked and several possible pathways can be calculated prior to performing the bronchoscopy. The pathway will be laid over the virtual bronchoscopy images and the distance from the point of view to the target can be shown.
An endobronchial pathway to a defined target can be calculated in similar fashion with the Bf-NAVI virtual bronchoscopy system. The CT data requirements are also similar. Nonvisible bronchi can be added using a manual editing or extraction process. The bronchial bifurcations as well as the airways to be intubated will be displayed in the planning of the route.
Virtual bronchoscopy-guided bronchoscopy
Performing virtual bronchoscopy-guided bronchoscopy is similar to that of normal bronchoscopy, but with the use of a thin or ultrathin bronchoscope. Repeated rotation of the instrument is needed in order to advance into more and more peripheral bronchi. This makes it much more difficult for the physician to maintain orientation within the bronchial tree.
In the LungPoint system the virtual bronchoscopy as well as the real endoscopic image are visible simultaneously. After initial synchronisation of the virtual and real images both the position and the rotation of the virtual bronchoscopy picture will be adapted automatically. The calculated pathway towards the target lesion that has to be followed by the investigator is projected over the images as a blue line (figure 4) [35]. After steering the bronchoscope as close as possible to the lesion, a biopsy can be taken through the working channel at the defined point of entry.
