Background
Lung cancer is a major health problem and the most common cause of cancer-related mortality worldwide [
1]. In patients with suspected malignant lesions a rapid and precise diagnosis is crucial to determine optimal treatment. Flexible bronchoscopy (FB) and computed tomography-guided transthoracic needle aspiration (CT-TTNA) are the main modalities employed to achieve this purpose. The appropriateness of each method depends on numerous factors such as tumor size and location, accessibility to the primary tumor, local availability of expertise with a particular technique and potential complications of that procedure. The preferable technique is the one that can be performed on an outpatient basis and gives the most information about diagnosis with minimum risk to the patient [
2]. CT-TTNA is a standard diagnostic procedure however it cannot be executed in all lesions and holds a significant risk for complications, namely pneumothorax [
3]. Flexible bronchoscopy and its ancillary procedures (bronchial washings, brushing and biopsies) are frequently used in lung cancer diagnosis and have a high diagnostic yield in endobronchial tumors but they have a limited ability in diagnosing peripheral, submucosal or peribronchial lesions [
4]. The addition of “blind” transbronchial needle aspiration (TBNA) is able to improve the diagnostic yield in some extraluminal tumors but it is influenced by lesion size and location, operator experience, and tumor type, among other factors [
5]. Some patients remain without a definitive diagnosis despite being submitted to several procedures and have to undergo a surgical biopsy that, although definitive, is invasive and not always suitable for those with advanced disease and significant comorbidities.
The accuracy and safety of TBNA has been increased by the development of endosonography. Linear real-time endobronchial (EBUS) and esophageal endoscopic ultrasound (EUS) are minimally invasive techniques that are able to image and sample, under direct vision, the intended structures in order to obtain specimens from pulmonary and mediastinal lesions. Both procedures have proven their value, individually or in combination, in mediastinal staging [
6,
7] but there is limited data on the importance of combining these two techniques in lung cancer diagnosis.
The aim of this study was to evaluate the value of linear EBUS and EUS guided fine needle aspiration (FNA), performed with one echoendoscope, in patients with a suspicious lung lesion who failed diagnosis after conventional procedures.
Discussion
The present study demonstrates that in selected patients with central lesions without an endobronchial component, EBUS and EUS-FNA, performed with one linear ultrasound bronchoscope, are accurate methods to diagnose lung cancer.
The diagnosis of indeterminate pulmonary nodules and masses can constitute a considerable challenge in some cases. A broad range of factors must be considered when selecting a specific diagnostic modality to assess a suspicious lesion. These factors include the clinical and radiological information, sensitivity and specificity of the test, invasiveness of the procedure, safety profile, institutional availability of technology, presence of qualified clinicians and cost-effectiveness of the procedure [
2].
Several options are available for the diagnosis of indeterminate pulmonary lesions. Less invasive techniques include CT-TTNA and FB however these approaches have both advantages and limitations. FB with its attendant diagnostic modalities is extremely valuable for the diagnosis of a patient with suspected lung cancer but its yield is quite variable. A systematic review [
3] reported a diagnostic yield of 88% for centrally located lesions with the combination of different sampling methods (bronchial biopsy, fine needle aspiration, bronchial brushing and washing) therefore in 12% of cases the diagnosis is not reachable by these methods. This percentage is even higher for extraluminal and peripheral lesions since the sensitivity of the FB with all modalities combined is much lower when compared to endobronchial lesions [
2,
3].
CT-TTNA is frequently used by pulmonologists and interventional radiologists because it provides a more accurate way of diagnosing peripheral pulmonary nodules and masses than FB although it has the potential to cause a non-negligible rate of iatrogenic complications, especially pneumothorax [
3]. Its sensitivity clearly depends on the size and location of the lesion, the size and type of the needle, the number of needle passes, and the presence of on-site cytopathology examination [
4,
8]. The pooled sensitivity of TTNA reported for 12363 patients in 61 studies was 90% [
4].
A small yet important number of patients with radiologically suspicious pulmonary lesions remain without a definitive diagnosis after extensive workup (7.9% in our study). Some patients are referred to other procedures such as mediastinoscopy, thoracoscopy or even thoracotomy [
9] however they are more invasive, more expensive and associated with increase morbidity and mortality when compared to ultrasonography guided-FNA. Others, not fitted for surgical interventions, are submitted to the repetition of the previous attempted conventional procedures. In the present study, all patients had at least one non-diagnostic FB or CT-TTNA and 39% were submitted to a second non-diagnostic technique which lead to a delay diagnosis and increased patient’s anxiety. Non-ultrasound guided TBNA is a well-established technique for lung cancer diagnosis and mediastinal staging however it is not routinely done in many institutions. In the present study, blind TBNA was attempted in only 14% of cases. The absence of formal training programs, the existence of technical problems with the procedure, concerns regarding its safety, needles cost, inadequate cytopathology support and variable diagnostic yield have been responsible for limited use of this procedure [
5,
10].
Over the last few years, EBUS and EUS-FNA have been introduced to medical practice and numerous studies have evaluated their impact in lung cancer mediastinal staging.
They have a high diagnostic yield and are simpler, less invasive, cost-effective and allow sampling of lymph node stations not accessible to mediastinoscopy [
6]. These techniques have gained their place in lung cancer staging algorithm substituting and complementing more invasive techniques such as mediastinoscopy. Furthermore, real-time linear EBUS and EUS-FNA can constitute an important option to diagnose lung cancer in a single procedure, principally in patients who present centrally located tumors not visible on routine bronchoscopy. Eckardt and coworkers have assessed the value of EBUS-TBNA for the diagnosis of radiologically suspicious chest lesions in a 36 months retrospective study [
11]. The technique was able to provide a diagnosis in 55% of cases and the diagnostic yield was higher in central parenchymal lesions compared to enlarged lymph nodes. These results are consistent with the work published by Tournoy et al. [
12] that reported a sensitivity of 84% in the diagnosis of central lung lesions not visible at routine bronchoscopy and by Nakajima et al. [
13] that obtained 94.1% sensitivity and 94.3% diagnostic accuracy rate in a small population of 35 patients.
EUS-FNA has also been used to diagnose centrally located lung tumors abutting the esophagus [
14,
15]. The choice between EBUS and EUS totally depends on the availability of equipment, expertise and the location of the suspicious lesion. Two studies [
7,
16] reported that the combination of EUS-FNA and EBUS-TBNA was better than either alone for mediastinal assessment and were able to achieve a very high diagnostic yield. Herth et al. have shown that the use of a single scope is feasible, effective and safe in primary mediastinal staging [
7]. In our population, a final diagnose could merely be achieved by EBUS-TBNA in 56.8% of cases and 6.8% of cases had a definitive diagnosis based of the performance of EUS-FNA. The combination of these procedures accomplished a wider sampling method and there was a clear advantage to perform them in the same setting since it maximized resources. There were also benefits regarding equipment handling and costs by using one ultrasound bronchoscope to perform EBUS and EUS-FNA for diagnostic purposes.
A further relevant point, demonstrated in our study, is that tissue may be acquired from highly suspicious metastatic central lymph nodes to diagnose lung cancer if the main peripheral lesion cannot be easily accessed. Since an important percentage of patients with lung cancer also have mediastinal lymph node enlargement at the time of clinical presentation, EBUS and EUS-FNA can play an important role in the diagnostic algorithm. However, it is important to distinguish patients with an established lung cancer diagnosis sent for mediastinal staging and those referred with enlarged lymph nodes without a definitive diagnosis since the described yield in this second group of patients is usually lower [
11,
17,
18]. In contrast, one of the first published studies reported 96% sensitivity of EUS-FNA in diagnosing mediastinal lymph node malignancy in a small sample of patients with suspected lung cancer in whom bronchoscopy failed to establish the diagnosis [
19]. In a more recent study, Lee et al. reviewed their results in a heterogeneous population of 126 patients who underwent EBUS-TBNA to diagnose highly suspicious lung cancer lesions (151 lymph nodes and 44 lung masses) [
20]. Eight patients had endobronchial tumor invasion, only 48.4% of patients had a previous attempt to obtain diagnosis and in these cases the most common cytopathological technique was sputum cytology which has a low sensitivity especially when the tumor is extraluminal. The overall sensitivity and diagnostic accuracy was 97.2% and 97.6% respectively. Our results also demonstrate that highly suspicious radiological and PET positive lymph nodes constitute good targets for EBUS and EUS-FNA lung cancer diagnosis. The sensitivity of these minimally invasive techniques combined (89.8%) was lower when compared to their use for mediastinal staging in patients with known lung cancer (96%) [
7], however they allowed a broader access to the target lesions compared to other diagnostic studies. In addition, the high lung cancer prevalence in this population – not entirely unexpected since all the included patients had a strong suspicious clinical picture complemented by suggestive accessory exams – and the correct selection of cases that could benefit from these techniques may explain the high sensibility. A further important factor, that may contribute to increase the diagnostic yield in such patients, is related to the number of aspirates per lesion. Lee et al. [
21] reported optimal results obtained with successive aspirates until the third EBUS-TBNA pass per lesion and these results were confirmed in subsequent studies [
22]. Since the number of needle passes may modify the sensitivity of the technique we have done at least four passes per lesion (mean 5.4/lesion) in order to get an adequate sample.
Another concern regarding the diagnostic procedures is safety. EBUS and EUS-FNA are usually associated with a very good safety profile with a low risk of adverse events. In this study they were also found to be safe since there were no complications associated with the procedures. The use of Doppler mode avoided vessels adjacent or within to the target lesion and in three cases dictated that the lesion could not be safely punctured.
EBUS plus EUS-FNA had a significant influence on the diagnostic management of our patients since they were able to spare more invasive procedures such as surgical exploration in 87.6% of cases. This approach may constitute an important strategy for patients in whom surgery is not an option because of comorbid conditions or advanced metastatic disease.
Regarding the economic impact, a study by Steinfort et al. [
23] confirmed that EBUS-TBNA staging of NSCLC was a cost-beneficial strategy in comparison with surgical techniques. For lung cancer diagnosis the use of conventional techniques is still less expensive compared to EBUS and EUS-FNA but our results reinforce the use of linear ultrasonography when there are no endobronchial tumor signs and the intended lesion is adjacent to the main airway or esophagus since it is a cost-effective approach.
Some limitations in our study justify further discussion. It should be noted that it is not a randomized trial and while this is a prospective study, it included a selected population with peripheral and central lung lesions without endobronchial visible signs. Patients were excluded whenever the tumor was not abutting the airways or the esophagus; if there was aerated lung parenchyma surrounding the lesion; or if the primary peripheral lesion was not associated to PET positive enlarged central lymph nodes. Even though the population represented in this study embraces just a part of the suspected lung cancer patients that are frequently assess in the clinical practice we are aware that this particular group of patients usually causes a more challenging diagnosis. Further selection bias may exist since the work was carried out in a referral interventional pulmonology unit although this may be minimized by the fact that in our center CT-TTNA is executed by the authors in association with interventional radiologists. Nevertheless, our data must be confirmed in larger multicenter randomized trials.
Finally, one can always argue that some of the included patients had a primary indication for EBUS or EUS-FNA instead of being submitted to the described conventional procedures. Previous work has shown that EBUS-TBNA can be performed as the first diagnostic procedure in patients with pulmonary masses and concomitant mediastinal lymph nodes [
24]. We cannot forget that although EBUS and EUS-FNA are frequently advocated for lung cancer lymph node staging, they are not widely available and, at least in some countries, are performed in selected and referral centers.
Competing interests
All authors declare that they do not have any competing interests related to this work.
Authors’ contributions
AB and DF conceived the study, participated in its design, collected the data and drafted the manuscript. RE was involved in the design, progression of the study and helped to draft the manuscript. SSD participated in the design of the study, data collection and statistical analysis. PAV, FJH and LC participated in the coordination of the study and revised the manuscript. All authors read and approved the final manuscript.