Traditionally, the diagnosis of lung cancer has been made with flexible bronchoscopy (FB) and its attendant procedures: bronchial washings , endobronchial or transbronchial brushes , bronchoalveolar lavage (BAL) , transbronchial biopsies , and conventional transbronchial fine-needle aspiration (TBNA) of mediastinal lymph nodes.

The sensitivity of bronchoscopic biopsy for central, endobronchial lesions has been reported to be very high; however, the yield for peripheral lesions is not so promising. In a review of 30 studies that reported diagnostic yield [15], the diagnosis of central, endobronchial tumors by bronchoscopy showed the highest sensitivity for endobronchial biopsies (74%) followed by bronchial brushing (59%) and washing (48%). The sensitivity for central tumors for all modalities combined was 88%. For peripheral lesions, cytobrushing demonstrated the highest sensitivity (52%), followed by transbronchial biopsy (TBB) (46%), and BAL/washing (43%). The overall sensitivity for all modalities for peripheral lesions was 69% [15]. The diagnostic yield of bronchoscopic sampling procedures is very much dependent on tumor visibility during bronchoscopy, the location of the bronchoscopically visible tumors, and, in the case of peripheral lesions, the size of the lesion (diagnostic yield higher for lesions greater than 3 cm in size) [1]. Other critical factors in the diagnostic yield of bronchoscopic biopsies are the forceps size and the number of biopsies obtained [16]. Forceps of 2 mm open diameter are felt to be the most useful in order to decrease artifacts that can impede the correct diagnosis. The more biopsies obtained, the better; however, increasing the number of biopsies taken results in increased risk of bleeding [16]. It is reported that between one third and one half of the bronchial biopsies taken from visible endobronchial tumors contain no viable tumor [17]. Cryobiopsies may be a more effective way to obtain larger biopsies, but the technique is not yet widely used in clinical practice [16].

Conventional TBNA (without endobronchial ultrasound) can be performed during flexible or rigid bronchoscopy in order to sample endoscopically visible bronchial abnormalities especially when there is evidence of extrinsic compression, submucosal infiltration, or an exophytic mass [18] as well as sampling hilar and mediastinal lymph nodes for staging of NSCLC [19]. It is particularly well suited for sampling paratracheal (stations 4R, 4L), subcarinal (station 7), and hilar (stations 10R, 10L) lymph nodes. Conventional TBNA is however a procedure that is performed without direct visualization of the lymph node being aspirated, and because of this limitation, the reported yield for TBNA for hilar and mediastinal lymph nodes varies widely (from 14% to 1%) [20]. In a meta-analysis of 12 studies in 910 patients, the sensitivity rate of TBNA was 76%, while the specificity rate was 96% [20]. In an updated systematic review performed by Silvestri and colleagues for the American College of Chest Physicians (ACCP) lung cancer guidelines update that included 2408 patients, the overall median sensitivity of TBNA was 78% (ranged from 14% to 100%), and the median negative predictive value (NPV) excluding studies with a prevalence of >80% was 77% [21]. The high false-negative rate of conventional TBNA makes it a less attractive modality for staging of the mediastinum. Therefore, TBNA would probably be the preferred minimally invasive method for patients with radiographic evidence of enlarged mediastinal lymph nodes adjacent to the airways, as bronchoscopy is usually performed in lung cancer patients and assessment for endobronchial lesions can be performed during the same procedure [22]. The optimal diameter of the needle is between 19G and 22G although the 19G needle is preferred as more clumps of tumor cells are sampled with the larger needle [16]. Rapid on-site evaluation (ROSE) of the aspirates by a cytopathologist/technologist improves the yield, is cost-effective, and eliminates unnecessary passes during the procedure [23].

EBUS-TBNA has revolutionized the approach to the diagnosis and staging of NSCLC. The technique is minimally invasive; provides access to nearly all lymph node stations (upper and lower paratracheal, subcarinal, hilar, and interlobar); has the ability to combine diagnosis and staging in a single procedure; has resulted in higher diagnostic yields than typically associated with conventional TBNA that are equivalent to, if not better than, diagnostic yield when compared with mediastinoscopy; and has the ability of providing adequate tissue for molecular analysis [24–30]. In a prospective cohort study of 108 patients, real-time EBUS-guided TBNA detected malignant lymph node involvement with a sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy of 95, 100, 100, 90, and 96%, respectively [29]. In a meta-analysis, EBUS-TBNA was reported to have a high pooled sensitivity of 93% and specificity of 100% for the confirmation of malignancy [31]. Even in patients with lymph nodes under 1 cm (cN0 by CT criteria), with the use of EBUS-TBNA, a significant percentage could still be shown to have pN2/pN3 disease (some despite also being negative on CT and PET-CT) [32, 33]. A randomized study evaluated a staging strategy combining endosonography and surgical staging compared with surgical staging alone [34]. Two hundred forty-one patients with potentially resectable NSCLC were randomized to surgical staging alone and to endosonography (EBUS and EUS) followed by surgical staging if negative. Nodal metastases were found in 41 patients (35%) by surgical staging vs. 56 patients (46%) by endosonography (P = 0.11) and in 62 patients (50%) by endosonography followed by surgical staging (P = 0.02). This corresponded to sensitivities of 79% vs. 85% (P = 0.47) and 94%, respectively (P = 0.02). Thoracotomy was unnecessary in 21 patients (18%) in the mediastinoscopy group vs. 9 patients (7%) in the endosonography group (P = 0.02). The complication rate was similar in both groups [34]. A systematic review performed for the ACCP lung cancer guidelines update of 2756 patients who met inclusion criteria for mediastinal staging with EBUS-TBNA revealed an overall sensitivity rate of 89% with a range from 46% to 97% and a median NPV of 91% [21].

In addition to its role in the diagnosis and staging of lung cancer, EBUS-TBNA has been shown to be a useful diagnostic modality in patients suspected of having lymphoma, metastatic disease to the mediastinal nodes from an extrathoracic cancer, and benign diseases such as sarcoid. Steinfort [35] et al. evaluated 98 patients who underwent EBUS-TBNA for evaluation of isolated mediastinal lymph nodes. Clinico-radiologic features suggested sarcoidosis as the likely diagnosis in 43 patients. In the remaining 55 patients, EBUS-TBNA achieved definitive diagnosis in 42 patients (76%; 95% confidence interval [CI] 55–90). Lymphoma was ultimately diagnosed in 21 of 55 patients (38%). EBUS-TBNA demonstrated lymphoma in 16 (76%) patients; however, in four patients, surgical biopsy was required to completely characterize lymphoma subtypes that were not readily amenable to diagnosis with low-volume specimens. Sensitivity and specificity for definitive diagnosis of lymphoma were 57% (95% CI 37–76) and 100% (95% CI 91–100), respectively [35]. Kennedy [36] et al. demonstrated EBUS-TBNA sensitivity of 90.9%, specificity of 100%, PPV of 100%, and NPV of 92.9% for the diagnosis of lymphoma. In a study by Tournoy et al., 92 patients with extrathoracic malignancies with suspicion of mediastinal or hilar metastases were evaluated with EBUS-TBNA. Mediastinal and hilar metastatic spreads were detected in 52 patients (57%) with a sensitivity and negative predictive value of 85% and 76%, respectively [37]. Garwood et al. [38] demonstrated noncaseating granulomas on EBUS-TBNA in 41 of 48 patients (85%) suspected of having pulmonary sarcoid. Factors affecting the diagnostic yield of EBUS-TBNA include decreased lymph node size (<5 mm), paratracheal location, airway distortion, and nodal calcification [38].

The mediastinal lymph nodes that are accessible through EUS include the aortopulmonary (station 5), subcarinal (station 7), paraesophageal (station 8), and inferior pulmonary ligament (station 9) [22, 39]. In a prospective cohort study of 104 patients, EUS-NA detected malignant lymph node involvement with a sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 92, 100, 100, 94, and 97%, respectively [40]. In addition to the mediastinal nodal stations, EUS-NA is particularly well suited to NA of the left adrenal gland and has been shown to frequently alter the staging and management of patients with NSCLC [41]. A major drawback of EUS-NA is the high false-negative rate; therefore, EUS-NA should be performed primarily on patients with radiological evidence of mediastinal lymphadenopathy [22]. Two studies report the combined use of EBUS and EUS to evaluate the mediastinum [42, 43]. For mediastinal staging, EUS provided additional information to EBUS-TBNA in 20 lung cancer patients with enlarged mediastinal lymph nodes or mediastinal lesions [42]. In a larger study of 33 patients for the staging of lung cancer, a total of 119 lesions were sampled by EUS-NA (n = 50) and EBUS-TBNA (n = 60) [43]. When EBUS-TBNA samples were compared with EUS-NA samples, 11 additional cancer diagnoses and three samples with suspicious cells were obtained by EBUS-TBNA that had not been obtained by EUS-NA. Conversely, EUS-NA diagnosed 12 additional cancer diagnoses, one suspicious and one specific benign diagnosis in addition to EBUS-TBNA. With a combined EBUS-EUS approach using a single bronchoscope, the sensitivity for cancer detection can be as high as 96% (EUS 89%, EBUS 91%), specificity 100%, and the negative predictive value 96% (EUS 82%, EBUS 92%) [44]. In an analysis of seven studies with 811 patients, the pooled sensitivity and specificity for combined EBUS-TBNA and EUS-NA were 91% and 100%, respectively, with a median NPV of 96% [21].

Ultrasound-guided needle techniques are minimally invasive and safe techniques with excellent performance characteristics (sensitivity rates of 89%, 89%, and 91% for EBUS-TBNA, EUS-NA, and combined EBUS-TBNA and EUS-NA, respectively) and are currently recommended as the best first diagnostic tools to obtaining tissue in the work-up of lung cancer [21].

Electromagnetic navigation bronchoscopy (ENB) is a localization device that guides endoscopic tools (forceps, brush, and needle) to preselected locations within the periphery of the bronchial tree, allowing the clinician to biopsy lesions with increased accuracy in areas that are traditionally either inaccessible or associated with low diagnostic yields when compared with traditional unguided or fluoroscopically guided bronchoscopy [15]. ENB has also been used to guide TBNA of peribronchial lymph nodes and placement of fiducial markers for stereotactic radiosurgery. Three companies currently make ENB systems, superDimension (Minneapolis, MN, USA), Veran Medical Technologies (St. Louis, MO, USA), and Broncus (Mountain View, CA, USA). The superDimension system utilizes a locatable guide inserted through a working channel catheter. Once navigation to the lesion(s) in question has been achieved, the locatable guide is removed, and instruments are deployed down the catheter for biopsy or fiducial placement. In a study by Gildea et al. [45], 54 patients with peripheral lesions underwent ENB. The mean lesion size was 23 mm (range, 8–78 mm), and 57% were less than 2 cm in diameter. A definitive diagnosis was made in 40/54 (74%) peripheral lesions and in 31/31 (100%) of the lymph nodes sampled. For all malignant lesions (total 43), 32 (74.4%) were successfully diagnosed by ENB. Pneumothorax occurred in two patients (3.5%). Eberhardt et al. [46], reported their experience performing ENB biopsy of 92 peripheral lesions in 89 subjects. No fluoroscopy was used. The mean lesion size was 24 mm (range, 10–58 mm). The overall diagnostic yield was 67% and appeared to be independent of size. The sensitivity for malignant disease was only 60%, and the NPV for malignant disease was 44%. The incidence of pneumothorax was 2.3%. Lamprecht [47] et al. studied ENB sampling using rapid on-site evaluation during the procedure, which showed a sensitivity and specificity of 84.6% and 100%, respectively. In a randomized trial using ENB, radial-probe EBUS, and EBUS combined with ENB [48], the authors hypothesized that the use of electromagnetic navigation along with radial-probe EBUS visualization of the peripheral lesion would increase the diagnostic yield. One hundred eighteen patients with peripheral nodules were randomized to EBUS, ENB, or EBUS combined with ENB. The diagnostic yield of 88% obtained by combined ENB and EBUS was superior to the diagnostic yield of either technique alone, 59% and 69%, respectively. More importantly, the NPV for malignant lesions increased from 44% to 75% with the combined use of ENB and a radial-probe EBUS [48]. The Veran system uses tip-tracked instruments via an ultrathin bronchoscope or scope catheter to navigate to peripheral lung lesions for biopsy. In addition, the system allows for electromagnetic navigation-guided transthoracic needle biopsy (EMTTNA) of anterior and lateral peripheral lung lesions. Yarmus et al. recently reported their experience with the system in which they sequentially enrolled 24 patients with peripheral lung nodules without radiographic evidence of lymphadenopathy (N0) for biopsy [49]. All patients underwent EBUS for lung cancer staging followed by ENB and EMTTNA; ROSE was not utilized during the procedures. The combined diagnostic yield of EMTTNA was 83% alone and 87% when ENB was combined with EMTTNA. The addition of EBUS to complete the staging paradigm further increased the diagnostic yield to 92%. Pneumothorax occurred in five patients (21%), of which only two (8%) required chest tube placement [49].

The success of peripheral lung biopsies has long been determined largely by lesion size and location. Lesions typically need to be greater or equal to 8 mm in size. Above this size, the diagnostic yield has depended most upon accessibility from the bronchial tree. Lesions in direct line with a bronchus that is visible on CT are more likely to be successfully biopsied. Lesions in the apical segments of the upper lobe and the superior segments of the lower lobes tend to be more challenging [45–48].

It must be emphasized that the false-negative rate of ENB (closely related to the NPV) is significant. The false-negative rate of transthoracic needle aspiration is in the range of 20–30% [1], and it is probable that this is a similar finding with ENB done without radial EBUS [50]. Thus, in a patient with a suspicious nodule, a negative or nondiagnostic biopsy result on ENB cannot be used to rule out malignancy. While the studies by Eberhardt and Yarmus et al. [48, 49] are encouraging, they have yet to be confirmed by other institutions or in large prospective multicenter trials.

ENB has also been used to place fiducial markers for stereotactic radiosurgery. Anantham et al. [51] placed 39 fiducials via navigation bronchoscopy into nine patients. A 10% migration rate after placement was reported, one patient suffered a chronic obstructive pulmonary disease exacerbation, and there were no instances of pneumothorax. In another study, a combination of ENB and radial EBUS was used to place fiducials in 43 patients. Although 13 of the patients suffered displacement of fiducials (30%), all were able to undergo stereotactic radiosurgery. Only one pneumothorax was seen [52].

Radial-probe EBUS (R-EBUS) consists of a flexible wire attached to a miniature ultrasound probe containing a rotating crystal tip that provides a 360° image of the surrounding structure. The ultrasound probe can be passed down the working channel of the bronchoscope and deployed into the lung parenchyma either alone or housed within a guide sheath catheter. In a recent meta-analysis [53] evaluating R-EBUS, the sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio were reported to be 73%, 100%, 27%, and 28% for the diagnosis of lung cancer, respectively. Significant interstudy variation was noted with the EBUS method used. In addition, significant interstudy heterogeneity for sensitivity of malignancy was noted, with prevalence of malignancy, lesion size, and reference standard reported as possible explanations. The rate of pneumothorax was only 1%. The authors concluded that R-EBUS is a safe and relatively accurate tool in the investigation of peripheral pulmonary nodules [53]. Recently, Chen et al. reported their 5-year experience utilizing R-EBUS in 467 cases. Nodules for biopsy were categorized by 1 cm size increments. R-EBUS views were classified as concentric or eccentric depending on lesion location in relation to the tip of the ultrasound probe. Successful identification of the lung nodules occurred in 96% of cases with an overall diagnostic yield of 69%. Diagnostic yield when comparing concentric vs. eccentric views was 84% vs. 48% (p < 0.0001) and 72% vs. 70% (p > 0.05) when comparing the use of a guide sheath to an ultrathin bronchoscope. The pneumothorax rate was 2.8% with a chest tube required in 1.5% [54].

Memoli et al. conducted a meta-analysis [55] to determine the overall diagnostic yield of several guided bronchoscopic techniques (electromagnetic navigation bronchoscopy (ENB), virtual bronchoscopy (VB), radial-probe endobronchial ultrasound (R-EBUS), ultrathin bronchoscope, and guide sheath) developed to improve the yield of transbronchial biopsy (TBBx) for diagnosing pulmonary nodules (PN). A total of 3052 lesions from 39 studies were included. The pooled diagnostic yield of guided bronchoscopic techniques was 70%, higher than the yield for traditional TBBx. The yield increased as the lesion size increased. The pneumothorax rate was only 1.6%, which is significantly smaller than the 15% reported for TTNA. The meta-analysis showed that the diagnostic yield of guided bronchoscopic techniques is better than traditional TBBx, and although the yield remains lower than TTNA (reported diagnostic yield of TTNA is 90%), the procedural risk is lower. Guided bronchoscopy may be an alternative or be complementary to TTNA for tissue sampling of PN [55].