Endobronchial ultrasonography (EBUS) is one of the most important technological advances in the diagnosis and staging of lung cancer in recent history. EBUS-guided transbronchial needle aspiration (EBUS-TBNA) has excellent diagnostic accuracy for mediastinal staging of non–small cell lung cancer.1–6 EBUS-TBNA has also been shown to be cost-beneficial compared with mediastinoscopy when used as the initial diagnostic procedure.7,8 As such, EBUS-TBNA is becoming increasingly accepted as the procedure of choice for initial evaluation of patients with mediastinal and hilar lymphadenopathy. Perceived barriers may prevent the widespread use of EBUS. Among these are the misunderstanding that this procedure requires use of general anesthesia and that it can only be adequately performed by dedicated interventional pulmonologists with extensive experience and a large annual caseload.9 Although there are multiple reports on the diagnostic accuracy of EBUS-TBNA, relatively little has been published on the determinants of EBUS-TBNA sampling adequacy (defined as acquisition of tissue sufficient for pathologic analysis and/or diagnosis).
Our EBUS practice may differ from other large institutions in that multiple pulmonologists, who are not exclusively proceduralists, routinely perform EBUS bronchoscopy. The time spent by each of the 10 bronchoscopists performing EBUS-TBNA does not exceed 20% of his or her overall activity. As part of a quality improvement project, we sought to determine our overall and individual sampling adequacies. We analyzed whether sampling adequacy is affected by the pulmonologist performing the procedure, lymph node (LN) size, procedural environment (hospital procedural suite vs. clinic-based procedural suite), and LN station targeted.
This study was approved by the Institutional Review Board (IRB) of the Mayo Clinic College of Medicine (IRB#06-005951). We reviewed the EBUS-bronchoscopies performed from January 2009 through August 2011 at our institution. Procedures lacking TBNA of at least 1 LN station and/or prior patient research authorization (95 procedures) were excluded from the analysis. The final cohort included 1275 patients and 1304 procedures. A retrospective analysis of the electronic medical records was completed with abstraction of the following data: primary proceduralist, procedure location, sampled LN stations, short axis diameter of LN stations by chest computed tomography (CT) preceding EBUS-TBNA, and pathologic (histologic and/or cytologic) diagnoses from initial EBUS-TBNA and all subsequent confirmatory procedures (surgery and/or repeat EBUS-TBNA). Pathologic diagnoses were recorded as 1 of 5 possibilities: malignant, atypical/suspicious, granulomatous, lymphoid, or inadequate. Inadequate tissue sampling was defined as absence of sufficient material for analysis as determined by the pathologist. Both sampling adequacy and diagnostic yield (per procedure) were defined in keeping with a recent report of the AQuIRE Registry results by Ost et al.10 Sampling adequacy was defined as acquisition of tissue sufficient for pathologic analysis and/or diagnosis and was calculated as follows: sampling adequacy=(samples resulting in a malignant, atypical/suspicious, or granulomatous pathologic diagnosis+samples demonstrating lymphoid tissue consistent with lymph node sampling)/total samples acquired. Diagnostic yield was defined as determination of a specific diagnosis made by EBUS-TBNA per procedure and was calculated as follows: EBUS-TBNA procedures resulting in a malignant and/or granulomatous pathologic diagnosis/total EBUS-TBNA procedures. Nodal stations were initially defined according to the 1997 International Staging System until the updated International Association for the Study of Lung Cancer LN map was published in 2009, which was utilized thereafter.11,12 Rapid on-site evaluation for cytology was not utilized.
EBUS with TBNA
All EBUS-TBNA procedures were performed by 10 pulmonologists with or without the assistance of fellows. All 10 pulmonologists had similar EBUS-TBNA training and between 3 and 5 years of experience; however, total years of general bronchoscopy experience varied ranging from 5 to 40 years. Each proceduralist was assigned in weekly rotations to 1 of 2 procedural services and was responsible for performing EBUS-TBNA on all patients referred to their service in sequential manner. All procedures were performed in 1 of 2 locations: either the hospital procedural suite or our clinic-based procedural suite. Procedural location was determined based on scheduling availability and/or need for concurrent transbronchoscopic lung biopsy requiring fluoroscopic guidance, which was only available in the hospital procedural suite. All procedures were completed using only conscious sedation. In the hospital procedural suite, sedation was independently provided by a nurse anesthetist using midazolam and fentanyl with or without propofol. In the clinic-based procedural suite, the proceduralist administered sedation with midazolam and fentanyl with the assistance of a nurse. The majority of patients (>99%) were intubated with an uncuffed wire spiral endotracheal tube over a conventional flexible video-bronchoscope (Olympus BF-P160/BF-Q180), which was then used for routine airway inspection before EBUS-TBNA. A few patients (<1%) were not intubated. The conventional bronchoscope was then exchanged for the linear array ultrasonic bronchoscope (Olympus BF-UC160F-0L8). A comprehensive and systematic ultrasonic examination of the mediastinum and hila was performed followed by TBNA of LNs larger than 10 mm in short axis diameter and additional nodes at the discretion of the proceduralist. For sampling of supraclavicular and/or high paratracheal stations, either no endotracheal tube was used or the endotracheal tube was temporarily withdrawn to the level of the subglottis and manually stabilized until completion of the TBNA after which it was replaced in the midtrachea. Absolute number of needle passes per biopsy was not routinely recorded; however, per procedural protocol for all biopsies were obtained with a minimum of 3 needle passes unless 2 needle passes resulted in visible core obtained with each pass.13
Generalized estimating equations were used to assess variables associated with sampling adequacy. The generalized estimating equations models account for the repeated assessments (multiple stations) within each patient. In all cases, P-values <0.05 were considered statistically significant.
A total of 1275 patients and 1304 EBUS-TBNA procedures resulting in 2414 LN biopsies were included in the analysis. No serious adverse events were reported. Overall, 986 procedures (75.6%) were performed in the hospital procedural suite whereas 318 (24.4%) were performed in the clinic-based procedural suite. Malignant cells or granulomatous inflammation were found in at least 1 LN in 410 (31.4%) and 154 (11.8%) procedures, respectively, resulting in a per procedure diagnostic yield of 43.2%. In total, 118 (9.1%) procedures resulted in at least 1 LN with insufficient sampling accounting for a total of 139 (5.8%) LN biopsies. As most procedures adequately sampled at least 1 node, there were only 41 (3.1%) procedures that resulted in complete absence of adequate diagnostic material.
In a subset of 727 procedures, specific measurement of the LN size (short axis diameter) was available. Analysis of this subgroup revealed a significant association between sampling adequacy and LN size overall (parameter estimate=0.08; P=0.002). More specifically, sampling adequacy was significantly lower when sampling LN <10 mm versus LNs ≥10 mm (parameter estimate=−0.7; P=0.002).
Sampling adequacy was analyzed according to each LN station and results are listed in Table 1. There was no association between LN station and sampling adequacy overall (P=0.69). This was true for both procedural suites (hospital P=0.56, clinic-based P=0.66).
Procedures were performed by a total of 10 pulmonologists. Performance data for each of the 10 proceduralists are shown in Table 2. Sampling adequacy was excellent overall at 94.2% and ranged from 87% to 99.2% by proceduralist. There was a significant difference in sampling adequacy per proceduralist overall (P<0.001). There was no difference in sampling adequacy depending on general bronchoscopy experience divided by decades: <10 years (n=4), 10 to 20 years (n=2), and >20 years (n=4) (P=0.09). There was an association between sampling adequacy and average number of LN stations biopsied per procedure when analyzed by proceduralist favoring the proceduralists who sampled more LN stations per procedure (parameter estimate=1.32; P=0.007). The average number of EBUS-TBNA procedures performed annually by each proceduralist was not associated with a difference in sampling adequacy (P=0.21).
There was no difference in overall sampling adequacy between our 2 procedural suites (P=0.08). On average, 1.8±0.9 and 1.9±1 LN stations were biopsied per procedure in the hospital and clinic-based procedural suites, respectively (P=0.50). Amongst procedures performed in the hospital procedural suite, there was no difference in sampling adequacy between patients that received propofol [715/986 (73%)] versus those managed with midazolam and fentanyl alone (P=0.90).
Subsequent histologic specimens were available for 216 patients who underwent repeat EBUS-TBNA, mediastinoscopy, video-assisted thoracoscopic surgery, or thoracotomy with mediastinal LN dissection. Only 6 patients (2.8%) had inadequate sampling at the initial EBUS-TBNA that potentially resulted in a missed diagnosis of cancer. One patient had inadequate sampling of station 1L and was subsequently found to have metastatic lung adenocarcinoma in stations 4R and 4L by mediastinoscopy. The remaining 5 patients had confirmation of malignancy present within the same LN station inadequately sampled during EBUS-TBNA. In all 5 cases the inadequate sampling was located in the lower paratracheal stations (four 4R/one 4L).
The majority of the available literature on EBUS has been published by dedicated interventional pulmonologists from well-established procedural practices at large academic medical centers. Although the reported data has been impressive with respect to the diagnostic utility of EBUS, there have been concerns that these data may be difficult to generalize to other less subspecialized practices. In addition, relatively little data has been published on the determinants of EBUS-TBNA sampling adequacy. Rather than readdressing the diagnostic accuracy of EBUS-TBNA, we chose to focus on determinants of sampling adequacy, which have direct implications for widespread dissemination of the procedure. In our study, we found that mean sampling adequacy (94.2%) was excellent overall and well within performance standards reported in the literature.2,5,10,14 Differences in sampling adequacy according to proceduralist may have been related to the absence of a standardized procedure protocol allowing for proceduralist-dependent differences in technique. These differences in sampling adequacy among proceduralists are being addressed in an ongoing quality improvement project.
Per procedure diagnostic yield was 43.2% in our study, which is in keeping with previously reported values including the AQuIRE Registry results.10 Diagnostic yield was calculated per procedure due to the fact that 29 patients (2.2%) underwent repeat EBUS-TBNA during the inclusion timeframe allowing for repeated analysis. Although this type of analysis differs slightly from the more commonly performed per patient analysis, the results are unlikely to be significantly different given the very small number of patients with repeat procedures. Diagnostic accuracy was unable to be determined in this study due to the fact that the majority of patients did not have confirmatory surgery after EBUS-TBNA; therefore, a reference standard was not available for the majority of patients with a negative EBUS-TBNA. All reasonable efforts to determine the reasons for lack of documented follow-up and/or surgical confirmation were completed and all available data was extracted. The reasons for a relatively low number of confirmatory surgical procedures are varied and include (but are not limited to) a sufficiently low posttest probability of cancer to warrant continued radiographic surveillance, completion of the evaluation at an institution more geographically accessible to the patient, patient preferences, and/or interim mortality. Ultimately we were limited in this respect by the retrospective nature of the study. Although we recognize that a significant number of patients may have had radiographic follow-up and/or surgical confirmation at other institutions, this data was not readily available and collection of this data was outside the scope of our IRB approval.
Interestingly, we found that the average number of procedures performed annually by each proceduralist was not associated with sampling adequacy. Four of the proceduralists in our group performed between 28 and 40 EBUS-TBNA procedures annually on average and yet demonstrated excellent sampling adequacy. Thirty to 40 annual EBUS procedures is a target that should be within reach for most busy community pulmonology practices, and is in keeping with previously published guidelines as a number needed for maintenance of proficiency.15,16
We found that that the proceduralists with a higher average number of LN stations biopsied per procedure had better sampling adequacy. This is somewhat counterintuitive as one may expect that sampling adequacy will suffer as proceduralists attempt more biopsies per procedure, potentially targeting smaller or more technically challenging LNs. One possible explanation for this difference may be that the proceduralists with more confidence in their procedural skills attempt to sample more challenging LNs, which may be reflected in their improved sampling adequacy.
The procedural location (hospital procedural suite vs. clinic-based procedural suite) did not seem to influence sampling adequacy, nor did it affect the average number of LN stations sampled per procedure. The primary difference between the locations involved the administration and depth of sedation, and for this reason we felt it was an appropriate variable to evaluate. All procedures were completed using only conscious sedation and all patients received midazolam and fentanyl regardless of venue. The hospital procedural suite did, however, allow for administration of deep conscious sedation with propofol (used in 73% of cases) and all sedation was independently provided by a nurse anesthetist allowing the proceduralist to focus entirely on the procedure itself. The practice of the clinic-based procedural suite was similar to the majority of other practices wherein the proceduralist administered only moderate sedation with the assistance of a nurse while performing EBUS-TBNA. All but one of the proceduralists performed procedures at both locations. This is reassuring and suggests that an outpatient or clinic-based venue using only moderate conscious sedation is adequate for EBUS-TBNA. Furthermore, there was no difference in sampling adequacy between patients that received propofol in addition to the routine regimen of midazolam and fentanyl confirming that moderate sedation is sufficient for achieving excellent sampling adequacy.
Both LN size and station have been reported to be associated with EBUS-TBNA sampling adequacy. Kennedy et al5 reported a reduction in sampling adequacy associated with LN size ≤5 mm and paratracheal station in univariate analysis. Ost et al10 reported data from the AQuIRE Registry showing a reduction in sampling adequacy associated with the 4L station in multivariate analysis; LN size was associated with sampling adequacy in univariate analysis but did not retain significance in the multivariate model. Contrary to the AQuIRE data, we found no significant association between sampling adequacy and LN station. We did, however, find a reduction in sampling adequacy when sampling LNs <10 mm in short axis diameter on CT imaging.
We sought to determine how often inadequate sampling during EBUS-TBNA would have potentially led to a missed diagnosis of cancer had no other procedures been pursued after a negative EBUS-TBNA. We use the term “potentially” to account for the fact that the sensitivity of EBUS-TBNA is not 100% and hence an adequate specimen would not have assured a diagnosis of cancer. A total of 216 patients had subsequent LN sampling by at least 1 modality after EBUS-TBNA. Six patients (2.8%) with inadequate sampling of at least 1 LN station were subsequently found to have cancer involving the mediastinum. Five of these patients were found to have cancer present in the same lower paratracheal LN station that was inadequately sampled by EBUS-TBNA and 1 was found to have lower paratracheal involvement after inadequate sampling of an upper paratracheal LN. Sampling inadequacy potentially contributed to additional diagnostic testing in a very small number of patients in our study. Nonetheless, 6 individuals had to undergo additional diagnostic evaluations that may have been avoided had their EBUS-TBNA specimens been adequate. Any potential for a missed diagnosis of malignancy warrants investigation for quality improvement.
The strengths of our study include the large sample size (largest reported to our knowledge), high inclusion rate, focus on sampling adequacy, different procedural environments allowing for internal comparison, and a procedural practice comparable with community-based practices, which strengthens the external validity of our results. We also acknowledge that our study has several limitations. First, the sample size for some of the LN stations is relatively small when compared with the lower paratracheal and subcarinal stations, which may have obscured an association between sampling adequacy and LN station, specifically regarding stations 1, 5, and 8, which are rarely accessible by EBUS-TBNA.12 However, the numbers of samples taken from nearly every station are still substantially larger than those reported in the majority of previously published literature. Second, due to inconsistent recording we did not have LN sizes for all patients, though our subgroup analysis was arguably still large enough to yield sufficient power. Furthermore, the sizes that were available were based on preprocedural CT imaging rather than ultrasonic imaging, which would have been preferable but was not routinely recorded. Third, EBUS-TBNA procedural techniques (eg, exact number of needle passes per LN station) may have varied and, due to inexact recording of this information at the time of the procedure, did not allow us to account for this variable. Fourth, although it is true that an individual proceduralist in our practice performs an average annual number of EBUS procedures that may be moderate compared with other interventional pulmonology practices, the procedures are performed at an institution that averages over 500 EBUS procedures annually. This high annual volume is further advantaged by experienced procedural support staff and superb pathologist support. This expertise may limit the generalization of our results to small community practices. Finally, many of the patients that underwent EBUS-TBNA at our institution did not have subsequent surgical pathology or radiographic follow-up available within our electronic medical record for reasons previously discussed. This limited our ability to determine the diagnostic accuracy of EBUS-TBNA in our study and to fully determine the frequency of potential adverse outcomes related to inadequate sampling.
In summary, EBUS-TBNA sampling adequacy was excellent overall and was associated with LN size and proceduralist, but not procedure location (hospital procedural suite vs. clinic-based procedural suite), sedation (moderate vs. deep), or LN station. Our data suggest that pulmonologists not exclusively performing procedures can still achieve excellent sampling adequacy. Smaller LNs may require additional passes or confirmation of adequate sampling by rapid on-site evaluation. Although multiple perceived barriers to widespread implementation of EBUS-TBNA may exist, our data support EBUS-TBNA sampling can be successfully implemented using moderate conscious sedation by appropriately trained individuals who are not dedicated proceduralists.
Members of the Mayo Pulmonary Procedural Group include: Fabien Maldonado, MD, Eric Edell, MD, David Midthun, MD, John Mullon, MD, Karen Swanson, DO, Udaya Prakash, MD, Craig Daniels, MD, Rodrigo Cartin-Ceba, MD, James Utz, MD, Otis Rickman, DO (former), Becky Scherbring, RN, Patty Schmidt, Jolisa Sok, CST, Aliesha Schlaak, CST, Gordon Griffith, RN, Hope St Jean, RN, Theresa Coleman, RN, and Amanda Ellison, RN.
1. Yasufuku K, Chiyo M, Sekine Y, et al. Real-time endobronchial ultrasound-guided transbronchial needle aspiration of mediastinal and hilar lymph nodes. Chest. 2004;126:122–128
2. Herth FJ, Eberhardt R, Vilmann P, et al. Real-time endobronchial ultrasound guided transbronchial needle aspiration for sampling mediastinal lymph nodes. Thorax. 2006;61:795–798
3. Herth FJ, Ernst A, Eberhardt R, et al. Endobronchial ultrasound-guided transbronchial needle aspiration of lymph nodes in the radiologically normal mediastinum. Eur Respir J. 2006;28:910–914
4. Vincent BD, El-Bayoumi E, Hoffman B, et al. Real-time endobronchial ultrasound-guided transbronchial lymph node aspiration. Ann Thorac Surg. 2008;85:224–230
5. Kennedy MP, Jimenez CA, Morice RC, et al. Factors influencing the diagnostic yield of endobronchial ultrasound-guided transbronchial needle aspiration. J Bronchol. 2010;17:202–208
6. Adams K, Shah PL, Edmonds L, et al. Test performance of endobronchial ultrasound and transbronchial needle aspiration biopsy for mediastinal staging in patients with lung cancer: systematic review and meta-analysis. Thorax. 2009;64:757–762
7. Harewood GC, Pascual J, Raimondo M, et al. Economic analysis of combined endoscopic and endobronchial ultrasound in the evaluation of patients with suspected non-small cell lung cancer. Lung Cancer. 2010;67:366–371
8. Steinfort DP, Liew D, Conron M, et al. Cost-benefit of minimally invasive staging of non-small cell lung cancer: a decision tree sensitivity analysis. J Thorac Oncol. 2010;5:1564–1570
9. Shrager JB. Mediastinoscopy: still the gold standard. Ann Thorac Surg. 2010;89:S2084–S2089
10. Ost DE, Ernst A, Lei X, et al. Diagnostic yield of endobronchial ultrasound-guided transbronchial needle aspiration: results of the AQuIRE Bronchoscopy Registry. Chest. 2011;140:1557–1566
11. Mountain C, Dresler C. Regional lymph node classification for lung cancer staging. Chest. 1997;111:1718–1723
12. Tournoy K, Annema J, Krasnik M, et al. Endoscopic and endobronchial ultrasonography according to the proposed lymph node map definition in the seventh edition of the tumor, node, metastasis classification for lung cancer. J Thorac Oncol. 2009;4:1576–1584
13. Lee HS, Lee GK, Kim MS, et al. Real-time endobronchial ultrasound-guided transbronchial needle aspiration in mediastinal staging of non-small cell lung cancer: how many aspirations per target lymph node station? Chest. 2008;134:368–374
14. Stoll LM, Yung RC, Clark DP, et al. Cytology of endobronchial ultrasound-guided transbronchial needle aspiration versus conventional transbronchial needle aspiration. Cancer Cytopathol. 2010;118:278–286
15. Bolliger CT, Mathur PN, Beamis JF, et al. ERS/ATS statement on interventional pulmonology. European Respiratory Society/American Thoracic Society. Eur Respir J. 2002;19:356–373
16. Ernst A, Silvestri GA, Johnstone D, et al. Interventional pulmonary procedures: Guidelines from the American College of Chest Physicians. Chest. 2003;123:1693–1717