| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Heart, Lung and Esophageal Institute, University of Pittsburgh, Pittsburgh, Pennsylvania
b Division of Pulmonary, Allergy, and Critical Care Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania
c School of Dental Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania
Accepted for publication May 8, 2009.
* Address correspondence to Dr Gilbert, University of Pittsburgh Medical Center Presbyterian, Suite C-800, 200 Lothrop St, Pittsburgh, PA 15213 (Email: gilberts{at}upmc.edu).
Presented at the Forty-fifth Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Jan 26–28, 2009.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Methods: The medical records of patients who underwent EBUS and had abnormal mediastinal lymph nodes (short-axis >1 cm and [or] positron emission topography-positive) over a 25 month period at the University of Pittsburgh were reviewed. Demographic and clinical data, cytology, and pathology results were entered in a database and analyzed.
Results: A total of 172 patients [male to female = 1.8:1; median age, 67 years (range, 20 to 90]) were included. The diagnostic yield of EBUS cytology was 79.7% (137 of 172). Pathologic testing was available in 68% (117 of 172) and 82% (96 of 117) had a diagnostic EBUS. The diagnostic accuracy of EBUS was 91.7%. The sensitivity, specificity, positive (PPV) and negative (NPV) predictive values were 88.1% (95% confidence interval [CI], 77.3 to 94.3), 100% (95% CI, 85.9 to 100), 100% (95% CI, 92.4 to 100), and 80.6% (95% CI, 63.4 to 91.2), respectively. In 67 patients who had a suspected or biopsy-proven primary lung cancer, diagnostic yield was 86.6% and accuracy was 94.8%. In this subgroup the sensitivity, specificity, PPV, and NPV were 93% (95% CI, 76.5 to 98.9), 100% (95% CI, 69.9 to 100), 100% (95% CI, 85 to 100), and 83.3% (95% CI, 56.2 to 97.5).
Conclusions: Diagnostic performance data support the clinical usefulness of EBUS in the evaluation of patients with a radiographically abnormal mediastinum. It should be considered complementary to mediastinoscopy rather than substitutive.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Technical Aspects
The technique and equipment necessary to perform EBUS have been described extensively [5]. Briefly, procedures were performed using topical anesthesia and monitored sedation or general anesthesia with endotracheal intubation. The equipment included the Evis Exera ultrasonic bronchoscope model BF-UC160F-OL8, the Evis Exera II video processor model CV-180, the Evis Exera II xenon light source model CLV-80, the EUS Exera CLA processor model EU-C60, and 22-gauge aspiration needles with syringe model NA-201SX-4022-A (Olympus, Center Valley, PA). During ultrasound examination of a given lymph node station, needle aspiration of the largest, accessible lymph node was performed. A different needle was used at each lymph node station to avoid potential specimen cross-contamination. Whenever possible, a cytologist was immediately available for rapid, on-site evaluation (ROSE) of transbronchial needle aspirates. The ROSE specimens were air-dried and stained with either Dif-Quick (American Scientific Products, McGaw Park, IL) or quick hematoxylin and eosin. Another slide was prepared from the same ROSE specimen but was fixed in alcohol to be processed at a later time using Papanicolaou staining. Any additional material retrieved from the aspiration needle was put in a preservation solution (CytoLyt; Cytyc Corp, Marlborough, MA). After centrifugation, the solid component was embedded in paraffin and processed using standard histology and immunohistochemistry techniques.
Interpretation of Cytology Results
When sufficient cellular material was obtained during EBUS, our institution's pathology department classified the cytology results according to the following categories: negative for malignant cells; atypical cells present; suspicious for malignant cells; and positive for malignant cells. The ROSE results were communicated immediately to the EBUS operator but the final cytology report only became available a few days later after all biopsy materials had been processed. The results of EBUS were classified as positive, negative, or nondiagnostic. A result was considered positive only when the cytologic diagnosis was "positive for malignant cells." A negative result was assigned only when the report was "negative for malignant cells" and the cytologist confirmed that the concentration of lymphocytes within the specimen was representative of lymph node tissue. All other results (ie, "atypical cells present," "suspicious for malignant cells") were classified as nondiagnostic. The EBUS procedure was labeled "diagnostic" only if results were either positive or negative. Reference pathologic tests considered acceptable for validation of EBUS cytology results included mediastinoscopy, mediastinal lymph node sampling or dissection, and confirmatory immunohistochemistry panel on positive cytology specimens.
Statistics
Continuous variables are expressed as the mean value ± standard deviation. A p value less than 0.05 was considered significant and all tests of significance were two-sided. Discrete and continuous variables were analyzed for statistically significant differences using the Pearson 
2 method and the independent samples Student t test, respectively. The prevalence of cancer represents the proportion of patients who were diagnosed with cancer, by EBUS or any other form of tissue biopsy, by the end of the study period. The diagnostic yield represents the proportion of patients with a diagnostic cytology report (positive or negative) within the entire study group or within a given subgroup. Diagnostic accuracy, sensitivity, specificity, positive (PPV) and negative (NPV) predictive values were calculated using patients who had a diagnostic EBUS cytology (positive or negative) and available reference pathologic testing. The SPSS software (version 16; SPPS Inc, Chicago, IL) was used for statistical calculations.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
Rapid On-Site Evaluation
Rapid on-site evaluation was available in 166 cases (96.5%) and was diagnostic in 68.8% (114 of 166) (Table 1). The ROSE report was consistent with the final EBUS cytology results in 86% (143 of 166). In patients with a positive ROSE for cancer cells, the final cytology report was uniformly positive. In those with a negative ROSE (51 of 166), the EBUS cytology was negative in 94.1% and nondiagnostic in 5.9%. When ROSE was nondiagnostic, final EBUS cytology was nondiagnostic in 61.6% and diagnostic in the remainder (positive = 19.2%; negative = 19.2%).
|
The diagnostic performance of EBUS could only be determined using patients in whom pathologic reference testing was available (117 of 172; 68%) and in whom the EBUS procedure was diagnostic (ie, positive or negative) (96 of 117; 82%) (Table 2). The overall diagnostic accuracy of EBUS was 91.7%. The sensitivity and specificity of EBUS to detect cancer in lymph nodes were 88.1% (95% CI, 77.3 to 94.3) and 100% (95% CI, 85.9 to 100), respectively. The PPV was 100% (95% CI, 92.4 to 100) and the NPV was 80.6% (95% CI, 63.4 to 91.2). The overall prevalence of cancer in the study group was 55.2% (95 to 172).
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
In our opinion, the false-negative rate (ie, 1-sensitivity) is a key diagnostic attribute of EBUS as a test to detect cancer in mediastinal lymph nodes. In large case series, the false-negative rate for mediastinoscopy was 1.6% to 8.8%. (3, 4, 7) Beyond stage misclassification, the implications of a false-negative result may include suboptimal timing of surgical resection and delay in the delivery of definitive chemoradiation. In prospective studies where negative EBUS results were systematically evaluated by mediastinoscopy and (or) open lymph node dissection, the false-negative rate of EBUS was 6.2% to 13% [12–15]. In this review, the false-negative rate was 11.9%. Although there is some overlap in the range of false-negative rates, whether or not EBUS is equivalent to surgical lymph node biopsy remains to be proven scientifically.
Independent of the manner in which EBUS is performed, variability in the sample size, cancer prevalence, criteria for acceptable reference standards, and cytologic criteria for diagnostic samples can lead to differences in diagnostic performance between reported case series. Several characteristics of our study may have had a negative impact on the diagnostic yield and performance of EBUS. These include a larger sample size, a relatively low overall cancer prevalence (55.2% vs 25.5% to 100%), a relatively high proportion of patients with reference pathologic testing (68% vs 6% to 80%), and the use of strict cytologic interpretation criteria. Indeed, our overall diagnostic yield (79.7% vs 87.5% to 100%) and the sensitivity (88.1% vs 89% to 98.7%) were lower than what is found in the literature [8, 9, 12–17]. However, the specificity (100% vs 100%), the positive predictive value (100% vs 100%), and the negative predictive value (80.6% vs 11% to 97.4%) were all comparable [8, 9, 12–17]. Beyond technical proficiency, patient selection and the prevalence of cancer in a given study population have a direct impact of the sensitivity and diagnostic yield of EBUS. The sensitivity and diagnostic yield of EBUS may be overestimated in situations where the prevalence of cancer is very high [8]. We found that our results are comparable to published literature when the patient populations are comparable [2].
Diagnostic failures were examined for possible causative factors. No single factor had a statistically significant impact on diagnostic yield. Recent evidence suggests that diagnostic yield improves with a higher number of needle passes per lymph node biopsy [12]. In the present study, the mean number of needle passes in nondiagnostic EBUS cases was significantly higher than in the rest of the study group. Therefore, we think that the proportion of nondiagnostic EBUS was probably not the result of an insufficient number of needle passes per lymph node station sampled. Beyond the absolute number of needle passes, other factors (eg, needle size, lymph node consistency and vascularity, airway cartilage, etc) may influence the diagnostic yield of EBUS. In an attempt to mitigate the potential effects of the learning curve on outcomes, institutional guidelines were established for training and proctoring of physicians performing EBUS. Although this study includes several operators, the variability in operator case load may have limited our ability to identify a case number threshold beyond which diagnostic yield significantly improves. The ideal setting to identify such a threshold would be a prospective protocol including a larger number of operators having performed at least 50 cases. Major pulmonary medicine societies currently recommend 40 to 50 supervised cases to establish EBUS competency [18, 19]. To our knowledge, specific requirements have not been established for surgeons. Given that surgical training and practice involve routine operative experience with mediastinal structures, the number of cases needed to establish competency may be different.
We are hesitant to provide specific guidelines regarding the use of EBUS because of the retrospective study design, the heterogeneity of the study population, and the relatively small number of patients in each subgroup. Nevertheless, this type of analysis provides useful information on the interaction between different diagnostic tests in patients with radiographically abnormal mediastinal lymph nodes. For instance, our results suggest that PET-CT scanning and EBUS may have complementary diagnostic performance characteristics in the evaluation of mediastinal lymphadenopathy. The relatively higher specificity and PPV of EBUS could fill some of the gaps in the diagnostic performance of a positive mediastinal PET-CT [20, 21]. As a result the number of mediastinoscopies performed for tissue diagnosis may decrease using this approach. Given the diagnostic yield and false-negative rate obtained in this study, EBUS would be unlikely to provide additional, clinically useful information in the majority of patients with a negative mediastinal PET-CT. At this point in time, the relationship between EBUS and mediastinoscopy appears to be complementary rather than substitutive. Endobronchial ultrasound is one of many tools which have expanded the pool of physicians who can successfully sample mediastinal lymph nodes when it is clinically indicated. If surgeons are to remain at the forefront of diagnosis, staging, and management of lung cancer, as well as other diseases involving the mediastinum, they must consider investing the time and effort to assimilate new technology.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
DR GILBERT: Typically we start with N3 stations, followed by level 7, and then go to other N2 stations.
DR CERFOLIO: How many patients had a false-negative EBUS and how many patients got a thoracotomy? You really can't calculate a false-negative rate unless you take out every node that you said is negative. You didn't do that in this study, did you?
DR GILBERT: This study was retrospective and therefore we could not control what happened to the patient after the EBUS was performed. Since we integrated EBUS to our clinical practice, we have made efforts to use every means possible to confirm negative EBUS results by obtaining additional tissue for pathologic analysis. We used mediastinoscopy as well as transthoracic lymph node sampling to accomplish this goal. However, unless EBUS is performed only in patients who are potential candidates for surgical resection of a primary lung cancer, it is not always feasible or clinically indicated to pursue additional surgical lymph node biopsy. In the present study, examples of such clinical scenarios include previous thoracic procedure (eg, mediastinoscopy, lung resection), poor operative candidate, diagnostic tissue obtained via other procedures (eg, image-guided biopsy of another site), specific benign diagnosis obtained (eg, sarcoidosis), more than two lymph node stations positive for cancer, abnormal lymph node station not accessible by mediastinoscopy, et cetera.
DR CERFOLIO: Did they get a thoracotomy if the mediastinoscopy was negative?
DR GILBERT: For pathologic confirmation, we use either a mediastinoscopy, or in some cases, VATS [video-assisted thoracic surgery] or thoracotomy. Most patients who had a mediastinoscopy did not undergo subsequent lymph node dissection. Upon reviewing our experience with EBUS, we realize that there was a significant selection bias favoring patients who had a positive mediastinal PET-CT [positron emission tomography-computed tomography]. There were only ten EBUS patients who had a normal PET-CT scan and they were excluded from the review, as stated in the methods. Over two thirds of our patients had a diagnosis other than primary lung cancer, and would therefore have been unlikely to undergo VATS or thoracotomy. Moreover, among the EBUS patients with suspected or biopsy-proven primary lung cancer and available pathologic testing, less than a third had negative mediastinal lymph nodes. The latter patients were managed surgically.
DR JOHN R. ROBERTS (Nashville, Tennessee): I wanted to explore a little bit with you the patients you excluded in which no node was sampled. As I understand, you had a group of patients in which no node was sampled that you excluded on the front end and then a second group that was nondiagnostic in which you sampled nodes. Is that correct?
DR GILBERT: There were a total of 184 patients who underwent EBUS over the study period and 12 were excluded from this review. Two of these 12 patients did not have any nodes sampled and the other 10 had a normal PET-CT scan. Everybody else had an abnormal PET-CT and had at least one lymph node station sampled.
DR ROBERTS: I guess the point I was going to make is that a nondiagnostic EBUS ought to include those patients in which you don't get any nodal tissue, in my opinion.
DR GILBERT: I completely agree with you. However, these patients have to be excluded from the specificity and sensitivity analyses because we need diagnostic results to generate a contingency table and calculate these values. We decided to include nondiagnostic EBUS results in the overall analysis because important information can be gained on the diagnostic yield of EBUS in various clinical situations where the surgeon is faced with a positive mediastinal PET-CT scan. Our review also provides what we believe is useful clinical data on the diagnostic yield of intraoperative, rapid on-site evaluation [ROSE], which may be available to surgeons from different institutions. In addition, these results allow for comparison of the diagnostic yield of ROSE and final cytology.
DR CERFOLIO: But your point is, if the test is negative, it's negative.
DR ROBERTS: Right. If the test is negative, it's negative, whether it's because you're not getting diagnostic tissue or you're not able to.
DR GILBERT: I think that the definition of a negative EBUS must be clarified. A negative EBUS means that lymphocytes were identified in the needle aspirate, in excess of what would be expected in the peripheral blood, and that no cancer cells were visualized. As with most CT-guided biopsies of lung nodules, EBUS provides a fine-needle aspirate. When a FNA [fine needle aspiration] of a lung nodule is reported as containing bronchial epithelial cells and pneumocytes with no malignant cells visualized, it does not provide the clinician with a specific benign diagnosis, such as hamartoma. If the latter CT-guided FNA report was obtained from a male smoker with a spiculated lung nodule, one would be hard pressed to interpret it as negative for cancer. Sampling error would be suspected and the CT-guided FNA results would be labeled "nondiagnostic." The same can be said of EBUS cytology in clinical scenarios where the degree of suspicion for mediastinal lymph node involvement with malignancy is relatively high. A significant proportion of our study group falls into this clinical category.
DR RAFAEL S. ANDRADE (Minneapolis, Minnesota): I'm very happy to see that this topic is getting revisited now at every meeting.
I do have a couple of questions for you. First, you started with about 170 patients and you subtracted the nondiagnostic tests and so on, but how did you end up with only 96 on your final analysis of sensitivity, positive predictive value, negative predictive value, et cetera?
DR GILBERT: We ended up with the subset of patients you are referring to by using strict criteria to validate EBUS cytology results. In order to create a contingency table and perform a sensitivity and specificity analysis, we had to exclude patients who had a nondiagnostic EBUS. Of the remaining patients who had a diagnostic EBUS, meaning either a positive or a negative result, only those who also had pathologic reference testing could be used for the diagnostic performance analysis. In contrast to other investigators, we decided that clinical follow-up was not acceptable as a reference standard against which EBUS results could be validated. In our review, a true-negative EBUS implies that the mediastinal lymph nodes were confirmed to be negative on pathologic sampling. In the case of a positive EBUS for cancer, pathologic sampling or confirmatory immunohistochemistry panel were considered acceptable reference standard. The immunostains were performed on the remaining EBUS biopsy material which had not been smeared onto cytology slides. Our early results and the literature suggest that EBUS has a very high positive-predictive value and that the probability of specimen cross-contamination is extremely low. Based on these facts, we thought that additional surgical sampling was not necessary to confirm positive EBUS results. In our study population, the fraction of patients who had pathologic reference testing that we could use to verify the results of EBUS was 68% (117 of 172). Of these patients, 21 had a nondiagnostic EBUS, and therefore, we were left with 96 patients to include in our two by two table to calculate sensitivity and specificity, and predictive values.
DR ANDRADE: Second, a nondiagnostic yield of roughly 30% on ROSE and 20% eventually is somewhat high. I saw that you looked at a series of possible predictive factors, but I didn't see that you analyzed for lymph node station. For example, station 7 tends to be more often nondiagnostic because it's very vascular. Another factor is the cytologist's experience reading transbronchial needle aspiration biopsies. I think it is very important to know that you are only as successful as your cytologist is.
DR GILBERT: I think the experience of the team of cytologists may have an influence on diagnostic yield. Our cytopathologists collectively analyze over 6,000 nongynecologic specimens per year. None were coinvestigators in this study. They were most likely not influencing the diagnostic yield of EBUS one way or another. We agree that lymph node vascularity may play a role in contamination of the EBUS sample with blood and could affect diagnostic yield. However, our results show that level 7 had the second highest diagnostic yield. This disparity may be explained by the fact that level 7 is technically easier to sample than some other stations, and that it was sampled in a higher proportion of our patients.
DR CERFOLIO: We have two esteemed European questioners. Give us quick questions and, please, quick, concise answers.
DR ERINO A. RENDINA (Rome, Italy): I noticed that you performed most of your procedures under general anesthesia. May I ask you why?
DR GILBERT: Initially, we performed all procedures under sedation because we only had a scope available in the bronchoscopy suite. We then purchased scopes for the operating room. As stated before, we had decided that, whenever possible, negative or nondiagnostic EBUS procedures should be followed by mediastinoscopy. Therefore, in most instances, we thought that it would be more convenient and comfortable for the patient to have an EBUS under general anesthesia, and be prepared to have a mediastinoscopy performed as a combined procedure.
DR RENDINA: But then positive patients get an anesthesia for free.
DR GILBERT: Some of the positive EBUS patients did get general anesthesia. Personally, I think that both approaches are safe and can be used according to operator preference and institutional policy.
DR GAETANO ROCCO (Naples, Italy): I would like to take the discussion to the next level. We know that the current views on chemotherapy regimens administered in a neoadjuvant setting seem to favor a specific histotype diagnosis in order to administer effective histotype-specific platinum-based regimens. So my question is, how many of the EBUS-diagnosed non-small cell lung cancers in your series had a specific histological diagnosis?
DR GILBERT: By histotype, I assume you are referring to adenocarcinoma, large cell, squamous cell, et cetera. Although this would definitely be interesting to examine, we do not have the information available at this point in time. I recall that histologic subtype was sometimes identified at the time of rapid on-site evaluation, and also sometimes provided in the final cytology and immunohistochemistry report. Unfortunately, my knowledge of cytopathology is insufficient to estimate the probability that a positive EBUS needle aspirate can be characterized in that manner. Thank you very much for your comment.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionAcknowledgmentsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. W. Connell, M. Berry, G. Cooke, and O. M. Kon Update on tuberculosis: TB in the early 21st century Eur. Respir. Rev., June 1, 2011; 20(120): 71 - 84. [Full Text] [PDF]
|
|
|
|
 |
|
 S. S. Groth and R. S. Andrade Endobronchial and Endoscopic Ultrasound-Guided Fine-Needle Aspiration: A Must for Thoracic Surgeons Ann. Thorac. Surg., June 1, 2010; 89(6): S2079 - S2083. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
