Background: Guidelines recommend multiple types of cytologic and tissue samplings in the diagnosis of lung cancer by bronchoscopy, but differences of opinion exist as to the relative value of bronchial brushings and endobronchial or transbronchial biopsies. Our objective was to determine concordance of these procedures by a test of symmetry in a historical cohort referred to the pulmonary diagnostic laboratory.
Methods: From 1988 to 2001, patients with pathologic confirmation of primary lung cancer were examined by standard bronchoscopic techniques of that period. An electronic medical record system was used, with statistical analysis of symmetry between brushings and biopsies establishing the diagnosis.
Results: Of 968 patients, 98% had bronchoscopy for 624 central and 322 peripheral suspect lesions. Bronchial brushings from 915 patients confirmed pulmonary malignancy in 811 (89%) patients. Endobronchial or transbronchial biopsies from 739 patients showed lung cancer in 603 (82%) cases. Bronchial washings in 16 patients and transthoracic needle biopsies in 30 patients established diagnosis. Transbronchial needle aspiration of mediastinal nodes identified metastases in 94 patients. Only 14 patients required a surgical procedure for diagnosis, but 188 received surgical excision as primary treatment. Statistical evaluation used only patients with both bronchial brushings and endobronchial or transbronchial biopsies. Analysis by a test of symmetry showed a significant difference (P<0.0001).
Conclusions: Positive, suspicious, and negative specimens were consistent, with bronchial brushings being more sensitive with a lower false-negative rate than endobronchial or transbronchial biopsies. Multiple techniques are recommended for bronchoscopic confirmation of lung cancer, but bronchial brushings should be collected initially, as technical or patient limitations might preclude diagnostic tissue biopsies.
*Pulmonary/Critical Care Section
†Hematology/Oncology Section, Medical Service
∥Surgery Service, Memphis Veterans Affairs Medical Center
‡Department of Preventive Medicine
§Section of Thoracic and Cardiovascular Surgery, Department of Surgery, University of Tennessee Health Science Center, Memphis, TN. IRB Committee: Memphis Veterans Affairs Medical Center, FWA00001678
Disclosure: There is no conflict of interest or other disclosures.
Reprints: John P. Griffin, MD, College of Medicine, University of Tennessee Health Science Center, 956 Court Avenue, Room G228, Memphis, TN 38163 (e-mail: email@example.com).
Received July 11, 2011
Accepted November 16, 2011
Bronchofiberscopy has revolutionized the diagnosis of lung cancer1,2 since the admonition of Edwards that rigid “bronchoscopy should never be omitted.”3 Under fluoroscopic control, it has become the most commonly used instrument for confirmation of the clinical impression4 and initial staging of bronchogenic carcinoma. Usual sampling techniques during inspection with this instrument,5,6 at the time of this study included bronchial washings,7 brushings,8–10 endobronchial and/or transbronchial biopsies,11,12 and transbronchial aspiration of accessible, enlarged mediastinal lymph nodes.13–16 Several investigators had found that bronchial washings and postbronchoscopic sputum specimens did not significantly contribute to the diagnostic yield and suggested that they should not be routinely examined.17 This retrospective observational study attempts to clarify the relative value of cytologic bronchial brushings versus histologic endobronchial or transbronchial biopsies in establishing the diagnosis of lung cancer by fiberoptic bronchoscopy by statistical testing of concordance of these sampling techniques.
PATIENTS AND METHODS
Between 1988 and 2001, 968 patients referred to the pulmonary diagnostic laboratory at the Memphis Veterans Affairs Medical Center, because of suspected lung cancer, received an established pathologic diagnosis of primary lung neoplasia (Fig. 1). This historical cohort of patients with consecutively diagnosed bronchogenic carcinoma was evaluated using the initially available computerized patient record system of Veterans Affairs medical charts,18,19 correlated with log data from the bronchoscopy laboratory, and supplemented by American College of Surgeons Tumor Registry records. Bronchoscopy, considered the most appropriate diagnostic procedure for most patients, was performed under conscious sedation by standard methodology for this time period at this institution, with support of monoplane fluoroscopy. When medical circumstances permitted, bronchial brushings and washings as well as biopsies of all directly visualized endobronchial lesions were obtained. Multiple transbronchial biopsies were attempted only when fluoroscopically evident lesions could be convincingly identified and entered in the absence of complications such as bleeding or oxygen desaturation.
Summary statistics were computed using SAS software, version 22.214.171.124 Agreement between results from bronchial brushings and endobronchial or transbronchial biopsies was tested using the FREQ procedure. The test of symmetry or concordance was based on a χ2 test with 3 degrees of freedom21; a true positive was defined as a positive test result, and a false negative was defined as a negative test result. Although a suspicious report may be regarded as being suggestive of lung cancer, this result was not classified as either a positive or negative result, thereby remaining a separate category.
There were 961 men and 7 women, with a median age of 68 years, range of 34 to 92 years, and mean age of 66.7 years. Radiographically suspect lesions for primary lung cancer were classified as “central” in 66% and “peripheral” in 34% of patients. Follow-up until demise was obtained in 955 or 99% of these lung cancer patients.
Bronchoscopy under fluoroscopic control was performed on 946 or 98% of patients (Table 1). In 624 patients, there were definitely visualized suspect endobronchial lesions. No potentially neoplastic abnormalities were found by endobronchial examination in 322 patients with only fluoroscopically evident peripheral lesions.
Brushings were obtained from suspect lesions in 915 patients (97%) (Table 1). Positive bronchial brushings for lung cancer were obtained in 811 patients (89%). In 47 patients (5%), bronchial brushings were suspicious for lung cancer. In 57 patients (6%), bronchial brushings showed negative cytology.
In 738 patients (78%), endobronchial or transbronchial biopsies were performed from visualized lesions (Table 1). Positive endobronchial or transbronchial biopsies for lung cancer were obtained for 603 patients (82%). In 18 patients (2%), endobronchial or transbronchial biopsies were only suspicious for lung cancer. In 117 patients (16%), endobronchial or transbronchial biopsies were histologically negative for lung cancer.
Transbronchial Needle Aspirates of Lymph Nodes
Multiple transbronchial aspirations of enlarged mediastinal lymph nodes were obtained from 173 patients (18%) (Table 1). Transbronchial needle aspirates from 94 patients (54%) with enlarged accessible mediastinal nodes were positive for metastatic lung cancer. Cytologic diagnoses from these patients were as follows: 29 small-cell carcinomas, 19 squamous cell carcinomas, 30 non–small-cell carcinomas, 12 adenocarcinomas, 2 large-cell undifferentiated carcinomas, and 1 neuroendocrine carcinoma. In 23 patients (13%), transbronchial nodal aspirates were only suspicious. In 57 patients (33%), transbronchial nodal aspirates showed negative cytology.
Bronchial washings obtained during bronchoscopy but initially unexamined provided additional diagnoses in 16 patients (2%). The specific cell types were as follows: 7 squamous cell carcinomas, 5 adenocarcinomas, 3 non–small-cell carcinomas, and 1 small-cell carcinoma.
Transthoracic Needle Biopsy of Parenchymal Lesions
When bronchoscopy was nondiagnostic, 30 patients had successful confirmation of lung cancer by fluoroscopic or computed tomography-directed transthoracic needle biopsy (3%). The specific cell types were as follows: 9 squamous cell carcinomas, 14 non–small-cell, 3 large-cell undifferentiated, 3 adenocarcinomas, and 1 neuroendocrine carcinoma.
In 14 patients (1%), thoracotomy was required for confirmation of lung cancer diagnosis. The specific cell types were as follows: 5 adenocarcinomas, 4 squamous cell carcinomas, 3 non–small-cell carcinomas, 1 small-cell carcinoma, and 1 combined adenocarcinoma/squamous cell tumor.
However, of this study cohort, 188 patients or 19% received primary surgical treatment for their lung cancer by lobectomy, pneumonectomy, or wedge resection.
Diagnostic Procedural Summary
Therefore, 96% of patients referred with a clinical impression of possible lung cancer, who ultimately received confirmation of that diagnosis, had their diagnosis confirmed by histologic or cytologic examination of specimens obtained at the pulmonary bronchoscopic laboratory. Additional procedures for the confirmation of lung cancer diagnosis, outside this location, were required in only 4% of study patients. The final pathologic diagnosis of the entire cohort of 968 patients with established primary lung cancer is summarized in Table 2.
Results of Test of Symmetry
Of the 946 patients receiving a bronchoscopy, 727 (76.9%) patients underwent both bronchial brushings and endobronchial or transbronchial biopsy. Table 3 depicts the diagnostic results for this subgroup of patients. The diagnostic results of the 2 tests were found to be significantly different (P<0.0001) using κ statistics. The 2 diagnostic methods yielded the same results for only 560 (77%) patients, with 540 being definitely positive for lung cancer (true positives) and 20 being negative for lung cancer (false negatives) in both tests. More than 13% of the tests that were positive on the basis of bronchial brushings (n=646) were negative for lung cancer on endobronchial or transbronchial biopsy, whereas only 3% of the tests that were positive on the basis of endobronchial or transbronchial biopsy (n=592) were negative on bronchial brushing. Similarly, >19% of the tests that were suspicious on bronchial brushing (n=42) were negative on endobronchial or transbronchial biopsy, whereas none of the tests that were suspicious for lung cancer on the basis of endobronchial or transbronchial biopsy (n=18) was negative on bronchial brushing. Finally, of the 38 tests that were negative for lung cancer on the basis of bronchial brushing, >52% were negative on endobronchial or transbronchial biopsy, whereas only 17% of the 116 tests that were negative on the basis of endobronchial or transbronchial biopsy were also negative for lung cancer on bronchial brushing. On the basis of this historical cohort of patients with primary lung cancer, bronchial brushing was more sensitive with a lower false-negative rate compared with endobronchial or transbronchial biopsy.
Since its development by Ikeda in 1966,22,23 the flexible fiberoptic bronchoscope has replaced the rigid bronchoscope as the principal diagnostic tool in the majority of patients in whom lung cancer is suspected. Using brushings and forceps biopsies, Zavala24 obtained an overall diagnostic yield of 94% in 193 patients with bronchogenic carcinomas that were visible endoscopically. Although Kvale et al17 had found that bronchial washings and postbronchoscopic sputum specimens did not contribute significantly to the diagnostic yield, it has been the experience of other investigators that occasionally both brushings and forceps biopsies are negative, whereas bronchial washings are positive for carcinoma.1,25
The role of fiberoptic bronchoscopy in the diagnosis of primary lung cancer can be grouped into various presentations reflecting the requirement of different bronchoscopic techniques26: distinct endobronchial mass lesions visible through the bronchoscope,27 submucosal or peribronchial spread of cancer,28 and peripheral lesions5,28; the latter include masses, nodules with feeding bronchi, and infiltrates.
Reviewing available data on central lesions in patients with bronchogenic carcinoma in 1993, Arroliga and Matthay6 found an average diagnostic yield of 75% by forceps biopsy and 78% by bronchial brushing cytology in a histologically confirmed squamous cell type. The diagnostic yield by forceps biopsy was only 55%, whereas cytologic yield by bronchial brushings was 69% for adenocarcinomas. With small-cell carcinomas the yield of 85% for forceps biopsy was superior to that of 66% for bronchial brushing cytology. Bronchial needle aspiration has been used in the diagnosis of endobronchial masses, but its role is not clear in comparison with forceps biopsy.5,29
Tumors located in the peribronchial area cause extrinsic compression and submucosal infiltration of the bronchial wall. The involved area shows erythema, loss of bronchial markings, and thickening or nodular appearance of the bronchial mucosal lining. Although these findings strongly suggest the presence of lung cancer, pathologic confirmation is essential to determine the tumor histology.30 Bronchoscopic needle aspiration aided diagnosis in 71% of such lesions compared with detection by forceps biopsy in 55%.28,31
Peripheral tumors that are not visible on endobronchial examination are diagnosed less readily by fiberoptic bronchoscopy,32 and these lesions are typically approached under fluoroscopic guidance. Conventional procedures used to obtain diagnostic material from peripheral lung cancers are bronchial washings, bronchial brushings, and transbronchial biopsies. Transbronchial needle aspiration increases the diagnostic yield of peripheral masses ≥2 cm in diameter; in 1 study, it increased the yield to 76% over that of transbronchial biopsy, which showed a yield of 52%.31 Overall diagnostic yield is higher when multiple procedures are performed sequentially during fiberoptic bronchoscopy as compared with any single sampling technique.30
Peripheral lesions evident on imaging studies with no bronchoscopic findings are increasingly evaluated by percutaneous transthoracic techniques.33,34 Under fluoroscopic or computed tomography-directed guidance, diagnoses are readily obtained, if intrathoracic location and boney structures permit.35 In our series, this technique provided definitive diagnosis in 30 patients.
Although bronchial washing requires no additional instrumentation and provides a diagnosis in about 28% of peripheral lung cancers, it increases the diagnostic yield by only 3% when performed together with bronchial brushing and transbronchial biopsy.11
Bronchial brushing under fluoroscopic guidance is easy to perform,9 and it provided a diagnosis in approximately 45% of cases.32 This procedure is recommended for all cases in which bronchoscopy is performed for peripheral lung cancers. Transbronchial biopsy under fluoroscopic guidance is a widely used technique for the diagnosis of peripheral lung cancer, with yield improving as the number of specimens increases. Collective data on >1500 patients from 25 studies showed 52% diagnostic sensitivity with this procedure for peripheral lung cancers.30
Mediastinal lymph node enlargement is often seen in patients initially presenting with non–small-cell lung cancer. Because surgical excision is the predominant opportunity for cure of the patient, accurate pathologic staging is essential before recommending or denying this primary treatment option. Transbronchial needle aspiration has emerged as an important technique for this purpose,14,16,36 being less invasive than mediastinoscopy and avoiding the need for general anesthesia. Unfortunately, despite its established role and safety, transbronchial needle aspiration is underused.37
The strengths of this study are its large sample size, consecutive performance of the bronchoscopic procedures by the senior authors, its high diagnostic yield in a pulmonary fellowship training environment, and inclusion of all patients referred to the same pulmonary bronchoscopic laboratory with a final pathologic diagnosis of primary lung cancer. Additional strengths include the testing of the concordance of bronchial brushing and endobronchial or transbronchial biopsy as diagnostic sampling techniques, conducted during the initial availability of the electronic medical record at this institution, with accuracy of data cross-checked with those from the tumor registry, and virtually complete follow-up of all patients.
Relative weaknesses of this study are its single institution origin, its predominance of a male population, its observational format without controls or blinding, its conduct during 1988 to 2001 before initiation of positron emission tomography imaging at our hospital, and the availability of newer diagnostic techniques38 such as endobronchial ultrasound for fine needle aspiration of mediastinal nodes.39
The performance of diagnostic sampling techniques during bronchoscopy varies widely in practice. The most inclusive collection of studies of the sensitivity of fiberoptic bronchoscopy diagnostic procedures from multiple databases has been published in the American College of Chest Physicians Evidence-Based Clinical Practice Guidelines.40 An exhaustive review of >10,000 reported cases classifies various sampling techniques according to central or peripheral location of lesions. However, because most studies used diagnostic brushings and biopsies to a highly variable extent, averaging all reports cannot validly compare the relative efficiency of these techniques. We consider that our study is the initial report on a true measure of concordance of these techniques by a test of symmetry.
In this investigation, bronchoscopic brushings statistically outperformed biopsies in diagnostic yield in cases of suspicious diagnoses and in false-negative cases. This should favor initially performing brush samplings in most patients undergoing diagnostic bronchoscopy for suspected lung cancer. An additional benefit in obtaining brushings for cytology is its relative ease and safety as compared with biopsies in an aging patient population with multiple comorbidities or possibly undergoing procedure-limiting drug therapy.
However, as established by current guidelines, the recommendation for the diagnostic confirmation of lung cancer by bronchoscopy is to obtain as many of the standard samplings as possible to increase diagnostic yield, including tissue samplings by biopsy as available, with additional appropriateness relative to specific cell type differentiation with increasing availability of genetically targeted drug treatment.
1. Loke J, Matthay R, Ikeda S. Techniques for diagnosing lung cancer: a critical review. Clin Chest Med. 1982;3:321–329
2. Postmus P. Bronchoscopy for lung cancer. Chest. 2005;128:16–18
3. Edwards A. Carcinoma of the bronchus. Thorax. 1946;1:1–25
4. Buccheri G, Barberis P, Delfino M. Diagnostic, morphologic, and histopathologic correlates in bronchogenic carcinoma. Chest. 1991;99:809–814
5. Shure D. Fiberoptic bronchoscopy: diagnostic applications. Clin Chest Med. 1987;8:1–13
6. Arroliga A, Matthay R. The role of bronchoscopy in lung cancer. Clin Chest Med. 1993;14:87–98
7. Mak V, Johnston I, Hetzel M, et al. Value of washings and brushings at fiberoptic bronchoscopy in the diagnosis of lung cancer. Thorax. 1990;45:373–376
8. Zavala D, Richardson R, Mukerjee P, et al. Use of the bronchofiberscope for bronchial brush biopsy: diagnostic results and comparison with other brushing techniques. Chest. 1973;63:889–892
9. Zaman M. Bronchoscopic brush techniques in the diagnosis of lung cancer. J Bronchol. 1994;1:263–264
10. Richardson R, Zavala D, Mukerjee P, et al. The use of fiberoptic bronchoscopy and brush biopsy in the diagnosis of suspected pulmonary malignancy. Am Rev Respir Dis. 1974;109:63–66
11. Popovich J Jr, Kvale P, Eichenhorn M, et al. Diagnostic accuracy of multiple biopsies from flexible fiberoptic bronchoscopy. Am Rev Respir Dis. 1982;125:521–523
12. Govert J, Dodd L, Kussin P, et al. A prospective comparison of fiberoptic transbronchial needle aspiration and bronchial biopsy for bronchoscopically visible lung carcinoma. Cancer. 1999;87:129–134
13. Wang K, Terry P. Transbronchial needle aspiration in the diagnosis and staging of bronchogenic carcinoma. Am Rev Respir Dis. 1983;127:344–347
14. Shure D, Fedullo P. The role of transcarinal needle aspiration in the staging of bronchogenic carcinoma. Chest. 1984;86:693–696
15. Dasgupta A, Jain P, Minai O, et al. Utility of transbronchial needle aspiration in the diagnosis of endobronchial lesions. Chest. 1999;115:1237–1241
16. Cetinkaya E, Yildiz P, Altin S, et al. Diagnostic value of transbronchial needle aspiration by Wang 22-gauge cytology needle in intrathoracic lymphadenopathy. Chest. 2004;15:527–531
17. Kvale P, Bode F, Kini S. Diagnostic accuracy in lung cancer: comparison of techniques used in association with flexible fiberoptic bronchoscopy. Chest. 1976;69:752–757
18. Jha A, Perlin J, Kizer K, et al. Effect of the transformation of the Veterans Affairs Health Care System on the quality of care. N Engl J Med. 2003;348:2218–2227
19. Asch S, McGlynn E, Hogan M, et al. Comparison of quality of care in the Veterans Health Administration and patients in a national sample. Ann Intern Med. 2004;141:938–945
20. SAS for Windows. 2003 Cary, NC SAS Institute Inc.
21. Bowker A. A test for symmetry in contingency tables. J Am Stat Assoc. 1948;43:572–574
22. Ikeda S. Flexible bronchofiberscope. Ann Otol Rhinol Laryngol. 1970;79:916–923
23. Ikeda S Atlas of Flexible Bronchofiberscopy. 1974 Tokyo Igaku Shoin Ltd
24. Zavala D. Diagnostic fiberoptic bronchoscopy: techniques and results of biopsy in 600 patients. Chest. 1975;68:12–19
25. Van der Drift M, van der Wilt G, Thunnissen F, et al. A prospective study of the timing and cost-effectiveness of bronchial washing during bronchoscopy for pulmonary malignant tumors. Chest. 2005;128:394–400
26. Detterbeck F, Tanoue L, Boffa D. Anatomy, biology and concepts, pertaining to lung cancer classification. J Thorac Oncol. 2009;4:437–443
27. Shure D, Astarita R. Bronchogenic carcinoma presenting as an endobronchial mass: optimal number of biopsy specimens for diagnosis. Chest. 1983;83:865–867
28. Shure D, Fedullo P. Transbronchial needle aspiration in the diagnosis of submucosal and peribronchial bronchogenic carcinoma. Chest. 1985;88:49–51
29. Buirski G, Calverley PMA, Douglas NJ, et al. Bronchial needle aspiration in the diagnosis of bronchial carcinoma. Thorax. 1981;36:508–511
30. Mazzone P, Jain P, Arroliga A, et al. Bronchoscopy and needle biopsy techniques for diagnosis and staging of lung cancer. Clin Chest Med. 2002;23:137–158
31. Shure D, Fedullo P. Transbronchial needle aspiration of peripheral masses. Am Rev Respir Dis. 1983;128:1090–1092
32. Radke J, Conway W, Eyler W, et al. Diagnostic accuracy in peripheral lung lesions: factors predicting success with flexible fiberoptic bronchoscopy. Chest. 1979;76:176–179
33. Nordenstrom B. Transthoracic needle biopsy. N Engl J Med. 1967;276:1081–1082
34. Tao LC, Pearson FG, Delarue NC, et al. Percutaneous fine-needle aspiration biopsy: its value to clinical practice. Cancer. 1980;45:1480–1485
35. Sider L, Davis TM Jr. Hilar masses: evaluation with CT-guided biopsy after negative bronchoscopic examination. Radiology. 1987;164:107–109
36. Holty J-EC, Kuschner W, Gould M. Accuracy of transbronchial needle aspiration for mediastinal staging of non-small cell lung cancer: a meta-analysis. Thorax. 2005;60:949–955
37. Haponik E, Shure D. Underutilization of transbronchial needle aspiration: experience of current pulmonary fellows. Chest. 1997;112:251–253
38. Ernst A, Silvestri G, Johnstone D. Interventional pulmonary procedures: guidelines from the American College of Chest Physicians. Chest. 2003;123:1693–1717
39. Silvestri G, Hoffman B, Bhutani M, et al. Endoscopic ultrasound with fine needle aspiration in the diagnosis and staging of lung cancer. Ann Thorac Surg. 1996;61:1441–1446
40. Rivera MP, Mehta Atul C. Initial diagnosis of lung cancer: ACCP evidence-based practice guidelines (2nd edition). Chest. 2007;132:131S–148S