- Email: [email protected]
Contributions of bronchoscopic microsampling in the supplemental diagnosis of small peripheral lung carcinoma
GENERAL THORACIC
Contributions of Bronchoscopic Microsampling in the Supplemental Diagnosis of Small Peripheral Lung Carcinoma Masazumi Watanabe, MD, PhD, Akitoshi Ishizaka, MD, PhD, Eiji Ikeda, MD, PhD, Akira Ohashi, PhD, and Koichi Kobayashi, MD, PhD Departments of Surgery and Pathology, Keio University School of Medicine and Tokyo Electric Power Company Hospital, and Pharmacia-Keio Research Lab, Tokyo, Japan
Background. Making a preoperative pathologic diagnosis in patients with small lung nodules remains challenging. We have developed a new, noninvasive bronchoscopic microsampling probe to examine biochemical substances in epithelial lining fluid. We used this probe to measure tumor markers in fluid from tissues surrounding lung nodules less than 30 mm in diameter to test its adjunctive diagnostic utility in lung cancer. Methods. In 12 patients, epithelial lining fluid was collected in triplicate or duplicate from tissue within 2 cm of small peripheral lung nodules and from the contralateral lung. The diagnosis of adenocarcinoma was surgically confirmed in all patients. Fifteen patients without lung cancer served as controls. Concentrations of
carcinoembryonic antigen, cytokeratin fragment 19, and sialyl SSEA-1 were measured in the fluid. Results. Carcinoembryonic antigen and cytokeratin fragment 19 concentrations were significantly higher in fluid near the nodules (median, 8.7 and 87.2 ng/mg, respectively) than on the contralateral sides (median, 1.5 and 3.7 ng/mg, respectively) or in fluid collected from the controls (median, 2.0 and 2.8 ng/mg, respectively). Conclusions. Measurements of carcinoembryonic antigen and cytokeratin fragment 19 collected by our microsampling probe may be a useful diagnostic adjunct in patients with small peripheral lung nodules. (Ann Thorac Surg 2003;76:1668 –73) © 2003 by The Society of Thoracic Surgeons
A
biochemical substances, including tumor markers, in epithelial lining fluid (ELF) near small peripheral lung nodules. This simple procedure can be repeated without the need for saline administration into the lungs. Furthermore, under fluoroscopic guidance, ELF can be sampled from more-precisely defined pulmonary regions compared with standard BAL. The present study applied the new BMS probe to measure concentrations of tumor markers and test its clinical contributions in patients with small peripheral lung nodules. A pathologic diagnosis was eventually made by surgical resection in all patients.
large proportion of patients with small lung nodules are currently investigated with the assistance of new imaging methods such as high-resolution computed tomography scanning [1]. A preoperative pathologic confirmation of the diagnosis by standard bronchoscopy is difficult to obtain in such cases [2, 3]. Occasionally an invasive percutaneous biopsy guided by computed tomographic scan is performed to make a tissue diagnosis. Complications, including pneumothorax and seeding of the tumor to the chest wall, have been observed with that procedure [4 – 6]. In addition, lung biopsy by a more invasive thoracoscopic wedge resection has also been performed. On the other hand, local bronchial lavage has been attempted to detect malignant cells in patients with lung cancer, although their detection rate is low [7, 8]. Furthermore, measurements of tumor marker concentrations in bronchoalveolar lavage (BAL) fluid have been occasionally described [9, 10], although BAL fluid is not reliable for quantitative determinations in presence of localized lesions such as a small adenocarcinoma. We have developed a new bronchoscopic microsampling (BMS) probe [11] to measure the concentrations of Accepted for publication April 22, 2003. Address reprint requests to Dr Watanabe, Department of Surgery, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku, Tokyo 1608582, Japan; e-mail: [email protected].
© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
Patients and Methods This study prospectively enrolled 12 consecutive patients with peripheral pulmonary nodules less than 30 mm in diameter, who had been referred to the Keio University Hospital or Tokyo Electrical Power Company Hospital between April 2001 and April 2002. These nodules were ultimately resected surgically and confirmed to be adenocarcinoma on histologic examination. Fifteen patients without malignant disease were enrolled as controls during the same period. Bronchoscopy was performed in control subjects for investigation of hemoptysis or persistent cough. All patients and subjects had granted their informed consent to participate in the study. 0003-4975/03/$30.00 doi:10.1016/S0003-4975(03)01015-4
1669
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
Fig 1. Illustration of the bronchoscopic microsampling probe.
Fig 3. Cytokeratin fragment 19 in epithelial lining fluid. Epithelial lining fluid weight– uncorrected values are correlated with weight– corrected values (p ⬍ 0.001).
GENERAL THORACIC
Ann Thorac Surg 2003;76:1668 –73
Microsampling Probe and Procedure The BMS probe (Olympus Co, Tokyo, Japan; Fig 1) and sampling procedure have been described in detail previously [11]. In brief, after premedication of the patient with atropine and pentazocine, and administering local anesthesia with aerosolized lidocaine hydrochloride, a flexible BF-XT40 fiberoptic bronchoscope (Olympus) was inserted in the trachea and, after flushing with air to minimize contamination of the samples, advanced into the target bronchus. The 2.8-mm-diameter sheath, containing an inner 1.7-mm cotton probe attached to a stainless-steel guidewire, were advanced under fluoroscopy through the bronchoscope. After wedging of the sheath in a bronchus, the inner probe was slowly advanced toward the distal airway, as near as possible to the fluoroscopic image of the nodule, as is performed in standard curettage methods (Fig 2). The BMS probe was maintained in position for 10 seconds to absorb the ELF. The inner probe was then withdrawn into the outer tube, and both were withdrawn together to avoid contamination. The inner probe was cut at 3 cm distal from its tip and placed in a previously weighed tube. After weighing, the wet probe was frozen at ⫺80°C until use. Before the
collection of ELF near the small nodule, ELF was collected in each patient from the contralateral lung. In control patients, ELF was collected by the same method from one or two pulmonary sites. Each procedure was repeated in duplicate or triplicate and the measurements were averaged.
Measurement of Tumor Markers The biochemical markers were recovered from the frozen probe as described previously [11]. Concentrations of carcinoembryonic antigen (CEA), cytokeratin fragment 19 (CYFRA), and sialyl SSEA-1 (SLX) in ELF were measured by enzyme-linked immunosorbent assay. Carcinoembryonic antigen was measured with the CEA Kit Daiichi II (Daiichi Radioisotope Labs, Tokyo, Japan), CYFRA with the CYFRA 21-1 IRMA Kit (TFB Inc, Tokyo, Japan), and SLX with the SLX Otsuka Kit (Otsuka Assay Labs, Tokushima, Japan). In patients with lung carcinomas these markers were also measured in serum by the same methods.
Immunohistochemistry Immunohistochemical staining of surgically resected cancerous and noncancerous (control) lung tissues was performed using CEA (A0115, polyclonal antibody immunized against rabbit, DAKO Japan, Tokyo, Japan) and CYFRA (M0888, monoclonal antibody immunized against mouse, DAKO Japan) antibody.
Data Management and Statistical Analysis
Fig 2. Bronchoscopic microsampling procedure under fluoroscopy. The arrow indicates the radiopaque marker.
In the first 87 samples, the concentrations of tumor markers were corrected for ELF weight as described earlier. A simple regression test was performed to examine correlations between concentrations of tumor markers with and without ELF weight correction. Because the corrected and uncorrected values were statistically correlated (Fig 3), only uncorrected measurements were used to simplify the analysis. Because data for tumor markers were not normally distributed, the nonparametric Kruskal-Wallis analysis
1670
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
Ann Thorac Surg 2003;76:1668 –73
Table 1. Patient Characteristics
GENERAL THORACIC
Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 a
Age (y)
Sex
Tumor Size (mm)
56 63 85 67 50 72 62 65 68 67 56 66
M M M F M M F M M M M M
22 25 28 20 15 16 18 8 10 10 10 12
Serum Concentrations
ELF Concentrations
CEA (ng/mL)
CYFRA (ng/mL)
SLX (U/mL)
CEA (ng/mL)
CYFRA (ng/mL)
Cytologic Diagnosisa
3.5 20.2 2.5 4.1 2.5 5.9 2.6 2.4 4.5 1.1 3.9 56.6
0.7 NE 3.9 1.3 1.1 1 1 0.4 1.8 0.8 0.5 1.3
43 36.5 25 23.6 44 29 43 37.3 3.7 26.1 53.9 36.4
181.2 35.1 14.4 28.9 3.1 2.9 1.2 2.9 14.4 0.7 0.8 20.9
150.8 833.7 359.8 34.4 163 67.9 13.2 46.3 106.5 1.7 2.5 9.1
⫹ ⫹ ⫺ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹
Cytologic diagnosis indicates cytologic examination by conventional bronchoscopic curettage.
ELF ⫽ epithelial lining fluid.
was used to compare groups. Receiver operating characteristic curves [12, 13] were constructed for each tumor marker. A p value less than 0.05 was considered significant.
Immunohistochemistry
Results
The expression of CYFRA was prominent in tumor tissue, mild in tissue surrounding the tumor, and absent in normal lung tissue. Microscopic findings in patient 8 are shown in Figure 6. The expression of CEA in each tissue was equivocal.
Patient Characteristics
Procedure Duration and Outcomes
The characteristics of the 12 patients, including tumor size, tumor marker concentrations in serum and ELF, and results of cytologic examination by conventional bronchoscopic curettage, are presented in Table 1. The mean (⫾standard deviation) diameter of the adenocarcinomatous nodules was 16 ⫾ 7 mm.
The duration of individual BMS procedures was approximately 10 minutes, for a total of 20 or 30 minutes per sampling site. No serious hemorrhage, marked hypoxia, or other procedure-related complication was observed in this series of patients.
Tumor Marker Values in Epithelial Lining Fluid Obtained by Bronchoscopic Microsampling
Comment
Figure 4 shows the values of the three tumor markers sampled by BMS from near the tumor, from the contralateral lung, and from the lung of noncancerous patients. Cytokeratin fragment 19 concentration in ELF from tissue near the tumor (median, 87.2 ng/mL) was significantly higher than in ELF from contralateral lung tissue (median, 3.7 ng/mL; p ⫽ 0.0021) or from noncancerous lungs (median, 2.8 ng/mL; p ⫽ 0.0001). Similarly, CEA concentration in ELF collected near the tumor (median, 8.7 ng/mL) was significantly higher than in ELF from the contralateral lung (median, 1.5 ng/mL; p ⫽ 0.0263) or from noncancerous lung tissue (median, 2.0 ng/mL; p ⫽ 0.0116). Differences in SLX among the three groups were not significant.
Receiver Operating Characteristic Analysis Receiver operating characteristic curves constructed for each tumor marker suggest that CYFRA is the most reliable of the three markers (Fig 5).
A growing number of small lung adenocarcinomas are being detected by new imaging methods such as highresolution computed tomographic scan, and may have a favorable prognosis after surgical resection [14]. Therefore, an early diagnosis of small lung nodules is of great value and importance. The mean tumor diameter of 16 mm in our present series suggests that our patients were good operative candidates. Our BMS probe was uniquely developed to collect and measure biochemical substances in ELF with minimal invasion compared with BAL or transbronchial biopsy. In an earlier study, we showed that the procedure can detect increased concentrations of interleukin 6, basic fibroblast growth factor, and neutrophil elastase in diffuse pulmonary diseases such as acute respiratory distress syndrome or acute lung injury [11]. Furthermore, because the BMS procedure could be performed serially, even in patients with respiratory failure, we were able to show that changes in cytokine concentrations reflected the severity of disease. These results prompted us to
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
1671
GENERAL THORACIC
Ann Thorac Surg 2003;76:1668 –73
Fig 5. Receiver operating characteristic curves for cytokeratin fragment 19 (CYFRA), carcinoembryonic antigen (CEA), and sialyl SSEA-1 (SLX). The areas under the curve values were 0.933 for cytokeratin fragment 19, 0.753 for carcinoembryonic antigen, and 0.704 for sialyl SSEA-1. Cytokeratin fragment 19 versus carcinoembryonic antigen, p ⫽ 0.0343; cytokeratin fragment 19 versus sialyl SSEA-1, p ⫽ 0.0197; carcinoembryonic antigen versus sialyl SSEA-1, p ⫽ 0.6331. Cytokeratin fragment 19 was the most reliable marker. (FPR ⫽ false-positive rate; TPR ⫽ true-positive rate.)
apply this procedure to localized pulmonary diseases, including lung cancer. In this series, we simplified the measurements of tumor markers in ELF to facilitate its introduction into clinical practice by comparing the results with versus without correction for ELF weight. Because we found a close correlation between the two sets of results, only direct measurements were performed thereafter to simplify the investigations. The sensitivity of measurements of tumor markers in
Fig 4. (A) Cytokeratin fragment 19 values in epithelial lining fluid. (B) Carcinoembryonic antigen values. (C) Sialyl SSEA-1 values. The box-whisker plots show the 25th and 75th percentiles, the median (horizontal line within the box), and the 10th and 90th percentiles (whiskers). *p ⬍ 0.05, **p ⬍ 0.01 (cancerous tissue group versus other groups).
Fig 6. Microscopic immunohistochemical staining of cytokeratin fragment 19. The expression of cytokeratin fragment 19 was prominent in tumor tissue, weak in tissue near the tumor (A and B), and absent in normal lung tissue (C and D). Arrowheads show cancer cells and arrows show cells at marginal region. Bar in A and C, 100 m; B and D, 50 m.
1672
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
GENERAL THORACIC
the serum for the detection of nonsmall cell lung cancer is 42% for CEA, 58% for CYFRA in the literature [15], and 32% for SLX in another article [16]. We hypothesized that the concentrations of these tumor markers are much higher in the tumor itself, or in the tissue surrounding the tumor, than in serum, and thus applied our BMS procedure to collect local ELF and measure the tumor markers as a supplemental method to diagnose peripheral lung cancer. Both CYFRA and CEA concentrations in ELF collected near the tumor were significantly higher than in ELF from the contralateral lung or from lung of patients without cancer. In addition, whereas the serum concentrations of CYFRA were within normal limits, the concentration in ELF was high in these patients with lung cancer. Cytokeratin fragment 19 was the most reliable marker, followed by CEA, according to the results of our receiver operating characteristic analyses. In contrast, SLX in ELF was as unreliable as in serum in this series. Attempts to diagnose peripheral lung cancer by measuring CYFRA, CEA, and SCC antigens in BAL fluid have been reported by other investigators [9, 10], although these attempts were unsuccessful. Our BMS probe appears more successful as it can precisely select the target bronchus under fluoroscopy. There are, however, technical problems associated with our BMS method. Insertion of the probe into the Ba1⫹2 (R. apicodorsalis, Rm. apicalis) or Ba6 (R. (lobi inferioris) superior) segments may be difficult because of the stiffness of the catheter. In addition, if the probe is contaminated by some blood, as occasionally happens, the concentrations of tumor marker in ELF may be underestimated by the blood contamination because blood tumor marker concentration might be suggested to be lower than that in ELF. We have observed the immunohistochemical expression of tumor markers in several cases. The mode of expression of CYFRA in lung cancer tissues is reported equivocally in the literature [17, 18]. In our present study the expression of CYFRA, which was prominent in the tumor and mild in the neighboring tissue, suggests that the marker diffuses from the tumor where it is produced to the surrounding area, and that it can be sampled by the probe even without direct contact with the tumor. This BMS procedure was a useful adjunct in the diagnosis of peripheral lung carcinoma. Because DNA or RNA can be collected by this same procedure, the detection of cancer-related genes in ELF, for example telomerase, may be considerably simplified in the near future. In conclusion, these results suggest that tumor markers produced in cancerous lung nodule diffuse to surrounding areas and can be sampled by our BMS procedure even if the probe is not in direct contact with the tumor. Microsampling methods are capable of detecting tumor makers more discriminately than BAL. Cytokeratin fragment 19 and CEA concentrations in ELF play visible roles in the diagnosis of primary lung adenocarcinoma. It is suggested that our BMS procedure during bronchoscopy supplements the histologic diagnosis in patients with small peripheral lung cancer. These results will need to be confirmed in larger groups of patients and further validated in prospective studies.
Ann Thorac Surg 2003;76:1668 –73
This study was supported by Grant-in-Aid No. 12671329 (to MW) from the Ministry of Education, Science, Sports and Culture, Japan. We thank Satoru Fukinbara, PhD (Department of Biochemical Engineering and Science, Faculty of Computer Science and Systems Engineering, Kyushu Institute of Technology) for contributing to the statistical analysis.
References 1. Swensen SJ, Jett JR, Sloan JA, et al. Screening for lung cancer with low-dose spiral computed tomography. Am J Respir Crit Care Med 2002;165:508 –13. 2. Mori K, Yanase N, Kaneko M, Ono R, Ikeda S. Diagnosis of peripheral lung cancer in cases of tumors 2 cm or less in size. Chest 1989;95:304 –8. 3. Obara M, Satoh H, Ishikawa H, et al. Diagnostic procedures in obtaining pathological materials for pulmonary nodules due to primary lung cancer. Oncol Rep 1998;5:1237–9. 4. Seyfer AE, Walsh DS, Graeber GM, Nuno IN, Eliasson AH. Chest wall implantation of lung cancer after thin-needle aspiration biopsy. Ann Thorac Surg 1989;48:284 –6. 5. Voravud N, Shin DM, Dekmezian RH, Dimery I, Lee JS, Hong WK. Implantation metastasis of carcinoma after percutaneous fine-needle aspiration biopsy. Chest 1992;102:313–5. 6. Yoshikawa T, Yoshida J, Nishimura M, Yokose T, Nishiwaki Y, Nagai K. Lung cancer implantation in the chest wall following percutaneous fine needle aspiration biopsy. Jpn J Clin Oncol 2000;30:450 –2. 7. de Gracia J, Bravo C, Miravitlles M, et al. Diagnostic value of bronchoalveolar lavage in peripheral lung cancer. Am Rev Respir Dis 1993;147:649 –52. 8. Pirozynski M. Bronchoalveolar lavage in the diagnosis of peripheral, primary lung cancer. Chest 1992;102:372–4. 9. Zaleska J, Pirozynski M, Kwiek S, Sakowicz A, RowinskaZakrzewska E. CEA, NSE, and SCC Ag in bronchial lavage in patients with lung cancer. Rocz Akad Med Bialymst 1997; 42(Suppl 1):179 –89. 10. Cremades MJ, Menendez R, Pastor A, Llopis R, Aznar J. Diagnostic value of cytokeratin fragment 19 (CYFRA 21–1) in bronchoalveolar lavage fluid in lung cancer. Respir Med 1998;92:766 –71. 11. Ishizaka A, Watanabe M, Yamashita T, et al. New bronchoscopic microsample probe to measure the biochemical constituents in epithelial lining fluid of patients with acute respiratory distress syndrome. Crit Care Med 2001;29:896 –8. 12. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29 –36. 13. Song HH. Analysis of correlated ROC areas in diagnostic testing. Biometrics 1997;53:370 –82. 14. Noguchi M, Morikawa A, Kawasaki M, et al. Small adenocarcinoma of the lung. Histologic characteristics and prognosis. Cancer 1995;75:2844 –52. 15. Schneider J, Velcovsky HG, Morr H, Katz N, Neu K, Eigenbrodt E. Comparison of the tumor markers tumor M2-PK, CEA, CYFRA 21–1, NSE and SCC in the diagnosis of lung cancer. Anticancer Res 2000;20:5053–8. 16. Ueoka H, Ohnoshi T, Moritaka T, et al. A statistical analysis of serum sialyl Lewis X-1 (SLX), CEA, SCC and NSE levels in patients with lung cancer. Nihon Kyobu Shikkan Gakkai Zasshi 1991;29:1022–8. 17. Quillien V, Ramee MP, Bansard JY, et al. Serum and tissue distribution of a fragment of cytokeratin 19 (cyfra 21–1) in lung cancer patients. Anticancer Res 1995;15:2857–63. 18. Chyczewski L, Niklinski J, Chyczewska E, Laudanski J, Furman M. Immunohistochemical analysis of tissue localization of cytokeratin 19 in lung cancer. Rocz Akad Med Bialymst 1997;42(Suppl 1):162–72.
Contributions of Bronchoscopic Microsampling in the Supplemental Diagnosis of Small Peripheral Lung Carcinoma Masazumi Watanabe, MD, PhD, Akitoshi Ishizaka, MD, PhD, Eiji Ikeda, MD, PhD, Akira Ohashi, PhD, and Koichi Kobayashi, MD, PhD Departments of Surgery and Pathology, Keio University School of Medicine and Tokyo Electric Power Company Hospital, and Pharmacia-Keio Research Lab, Tokyo, Japan
Background. Making a preoperative pathologic diagnosis in patients with small lung nodules remains challenging. We have developed a new, noninvasive bronchoscopic microsampling probe to examine biochemical substances in epithelial lining fluid. We used this probe to measure tumor markers in fluid from tissues surrounding lung nodules less than 30 mm in diameter to test its adjunctive diagnostic utility in lung cancer. Methods. In 12 patients, epithelial lining fluid was collected in triplicate or duplicate from tissue within 2 cm of small peripheral lung nodules and from the contralateral lung. The diagnosis of adenocarcinoma was surgically confirmed in all patients. Fifteen patients without lung cancer served as controls. Concentrations of
carcinoembryonic antigen, cytokeratin fragment 19, and sialyl SSEA-1 were measured in the fluid. Results. Carcinoembryonic antigen and cytokeratin fragment 19 concentrations were significantly higher in fluid near the nodules (median, 8.7 and 87.2 ng/mg, respectively) than on the contralateral sides (median, 1.5 and 3.7 ng/mg, respectively) or in fluid collected from the controls (median, 2.0 and 2.8 ng/mg, respectively). Conclusions. Measurements of carcinoembryonic antigen and cytokeratin fragment 19 collected by our microsampling probe may be a useful diagnostic adjunct in patients with small peripheral lung nodules. (Ann Thorac Surg 2003;76:1668 –73) © 2003 by The Society of Thoracic Surgeons
A
biochemical substances, including tumor markers, in epithelial lining fluid (ELF) near small peripheral lung nodules. This simple procedure can be repeated without the need for saline administration into the lungs. Furthermore, under fluoroscopic guidance, ELF can be sampled from more-precisely defined pulmonary regions compared with standard BAL. The present study applied the new BMS probe to measure concentrations of tumor markers and test its clinical contributions in patients with small peripheral lung nodules. A pathologic diagnosis was eventually made by surgical resection in all patients.
large proportion of patients with small lung nodules are currently investigated with the assistance of new imaging methods such as high-resolution computed tomography scanning [1]. A preoperative pathologic confirmation of the diagnosis by standard bronchoscopy is difficult to obtain in such cases [2, 3]. Occasionally an invasive percutaneous biopsy guided by computed tomographic scan is performed to make a tissue diagnosis. Complications, including pneumothorax and seeding of the tumor to the chest wall, have been observed with that procedure [4 – 6]. In addition, lung biopsy by a more invasive thoracoscopic wedge resection has also been performed. On the other hand, local bronchial lavage has been attempted to detect malignant cells in patients with lung cancer, although their detection rate is low [7, 8]. Furthermore, measurements of tumor marker concentrations in bronchoalveolar lavage (BAL) fluid have been occasionally described [9, 10], although BAL fluid is not reliable for quantitative determinations in presence of localized lesions such as a small adenocarcinoma. We have developed a new bronchoscopic microsampling (BMS) probe [11] to measure the concentrations of Accepted for publication April 22, 2003. Address reprint requests to Dr Watanabe, Department of Surgery, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku, Tokyo 1608582, Japan; e-mail: [email protected].
© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
Patients and Methods This study prospectively enrolled 12 consecutive patients with peripheral pulmonary nodules less than 30 mm in diameter, who had been referred to the Keio University Hospital or Tokyo Electrical Power Company Hospital between April 2001 and April 2002. These nodules were ultimately resected surgically and confirmed to be adenocarcinoma on histologic examination. Fifteen patients without malignant disease were enrolled as controls during the same period. Bronchoscopy was performed in control subjects for investigation of hemoptysis or persistent cough. All patients and subjects had granted their informed consent to participate in the study. 0003-4975/03/$30.00 doi:10.1016/S0003-4975(03)01015-4
1669
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
Fig 1. Illustration of the bronchoscopic microsampling probe.
Fig 3. Cytokeratin fragment 19 in epithelial lining fluid. Epithelial lining fluid weight– uncorrected values are correlated with weight– corrected values (p ⬍ 0.001).
GENERAL THORACIC
Ann Thorac Surg 2003;76:1668 –73
Microsampling Probe and Procedure The BMS probe (Olympus Co, Tokyo, Japan; Fig 1) and sampling procedure have been described in detail previously [11]. In brief, after premedication of the patient with atropine and pentazocine, and administering local anesthesia with aerosolized lidocaine hydrochloride, a flexible BF-XT40 fiberoptic bronchoscope (Olympus) was inserted in the trachea and, after flushing with air to minimize contamination of the samples, advanced into the target bronchus. The 2.8-mm-diameter sheath, containing an inner 1.7-mm cotton probe attached to a stainless-steel guidewire, were advanced under fluoroscopy through the bronchoscope. After wedging of the sheath in a bronchus, the inner probe was slowly advanced toward the distal airway, as near as possible to the fluoroscopic image of the nodule, as is performed in standard curettage methods (Fig 2). The BMS probe was maintained in position for 10 seconds to absorb the ELF. The inner probe was then withdrawn into the outer tube, and both were withdrawn together to avoid contamination. The inner probe was cut at 3 cm distal from its tip and placed in a previously weighed tube. After weighing, the wet probe was frozen at ⫺80°C until use. Before the
collection of ELF near the small nodule, ELF was collected in each patient from the contralateral lung. In control patients, ELF was collected by the same method from one or two pulmonary sites. Each procedure was repeated in duplicate or triplicate and the measurements were averaged.
Measurement of Tumor Markers The biochemical markers were recovered from the frozen probe as described previously [11]. Concentrations of carcinoembryonic antigen (CEA), cytokeratin fragment 19 (CYFRA), and sialyl SSEA-1 (SLX) in ELF were measured by enzyme-linked immunosorbent assay. Carcinoembryonic antigen was measured with the CEA Kit Daiichi II (Daiichi Radioisotope Labs, Tokyo, Japan), CYFRA with the CYFRA 21-1 IRMA Kit (TFB Inc, Tokyo, Japan), and SLX with the SLX Otsuka Kit (Otsuka Assay Labs, Tokushima, Japan). In patients with lung carcinomas these markers were also measured in serum by the same methods.
Immunohistochemistry Immunohistochemical staining of surgically resected cancerous and noncancerous (control) lung tissues was performed using CEA (A0115, polyclonal antibody immunized against rabbit, DAKO Japan, Tokyo, Japan) and CYFRA (M0888, monoclonal antibody immunized against mouse, DAKO Japan) antibody.
Data Management and Statistical Analysis
Fig 2. Bronchoscopic microsampling procedure under fluoroscopy. The arrow indicates the radiopaque marker.
In the first 87 samples, the concentrations of tumor markers were corrected for ELF weight as described earlier. A simple regression test was performed to examine correlations between concentrations of tumor markers with and without ELF weight correction. Because the corrected and uncorrected values were statistically correlated (Fig 3), only uncorrected measurements were used to simplify the analysis. Because data for tumor markers were not normally distributed, the nonparametric Kruskal-Wallis analysis
1670
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
Ann Thorac Surg 2003;76:1668 –73
Table 1. Patient Characteristics
GENERAL THORACIC
Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 a
Age (y)
Sex
Tumor Size (mm)
56 63 85 67 50 72 62 65 68 67 56 66
M M M F M M F M M M M M
22 25 28 20 15 16 18 8 10 10 10 12
Serum Concentrations
ELF Concentrations
CEA (ng/mL)
CYFRA (ng/mL)
SLX (U/mL)
CEA (ng/mL)
CYFRA (ng/mL)
Cytologic Diagnosisa
3.5 20.2 2.5 4.1 2.5 5.9 2.6 2.4 4.5 1.1 3.9 56.6
0.7 NE 3.9 1.3 1.1 1 1 0.4 1.8 0.8 0.5 1.3
43 36.5 25 23.6 44 29 43 37.3 3.7 26.1 53.9 36.4
181.2 35.1 14.4 28.9 3.1 2.9 1.2 2.9 14.4 0.7 0.8 20.9
150.8 833.7 359.8 34.4 163 67.9 13.2 46.3 106.5 1.7 2.5 9.1
⫹ ⫹ ⫺ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫺ ⫹
Cytologic diagnosis indicates cytologic examination by conventional bronchoscopic curettage.
ELF ⫽ epithelial lining fluid.
was used to compare groups. Receiver operating characteristic curves [12, 13] were constructed for each tumor marker. A p value less than 0.05 was considered significant.
Immunohistochemistry
Results
The expression of CYFRA was prominent in tumor tissue, mild in tissue surrounding the tumor, and absent in normal lung tissue. Microscopic findings in patient 8 are shown in Figure 6. The expression of CEA in each tissue was equivocal.
Patient Characteristics
Procedure Duration and Outcomes
The characteristics of the 12 patients, including tumor size, tumor marker concentrations in serum and ELF, and results of cytologic examination by conventional bronchoscopic curettage, are presented in Table 1. The mean (⫾standard deviation) diameter of the adenocarcinomatous nodules was 16 ⫾ 7 mm.
The duration of individual BMS procedures was approximately 10 minutes, for a total of 20 or 30 minutes per sampling site. No serious hemorrhage, marked hypoxia, or other procedure-related complication was observed in this series of patients.
Tumor Marker Values in Epithelial Lining Fluid Obtained by Bronchoscopic Microsampling
Comment
Figure 4 shows the values of the three tumor markers sampled by BMS from near the tumor, from the contralateral lung, and from the lung of noncancerous patients. Cytokeratin fragment 19 concentration in ELF from tissue near the tumor (median, 87.2 ng/mL) was significantly higher than in ELF from contralateral lung tissue (median, 3.7 ng/mL; p ⫽ 0.0021) or from noncancerous lungs (median, 2.8 ng/mL; p ⫽ 0.0001). Similarly, CEA concentration in ELF collected near the tumor (median, 8.7 ng/mL) was significantly higher than in ELF from the contralateral lung (median, 1.5 ng/mL; p ⫽ 0.0263) or from noncancerous lung tissue (median, 2.0 ng/mL; p ⫽ 0.0116). Differences in SLX among the three groups were not significant.
Receiver Operating Characteristic Analysis Receiver operating characteristic curves constructed for each tumor marker suggest that CYFRA is the most reliable of the three markers (Fig 5).
A growing number of small lung adenocarcinomas are being detected by new imaging methods such as highresolution computed tomographic scan, and may have a favorable prognosis after surgical resection [14]. Therefore, an early diagnosis of small lung nodules is of great value and importance. The mean tumor diameter of 16 mm in our present series suggests that our patients were good operative candidates. Our BMS probe was uniquely developed to collect and measure biochemical substances in ELF with minimal invasion compared with BAL or transbronchial biopsy. In an earlier study, we showed that the procedure can detect increased concentrations of interleukin 6, basic fibroblast growth factor, and neutrophil elastase in diffuse pulmonary diseases such as acute respiratory distress syndrome or acute lung injury [11]. Furthermore, because the BMS procedure could be performed serially, even in patients with respiratory failure, we were able to show that changes in cytokine concentrations reflected the severity of disease. These results prompted us to
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
1671
GENERAL THORACIC
Ann Thorac Surg 2003;76:1668 –73
Fig 5. Receiver operating characteristic curves for cytokeratin fragment 19 (CYFRA), carcinoembryonic antigen (CEA), and sialyl SSEA-1 (SLX). The areas under the curve values were 0.933 for cytokeratin fragment 19, 0.753 for carcinoembryonic antigen, and 0.704 for sialyl SSEA-1. Cytokeratin fragment 19 versus carcinoembryonic antigen, p ⫽ 0.0343; cytokeratin fragment 19 versus sialyl SSEA-1, p ⫽ 0.0197; carcinoembryonic antigen versus sialyl SSEA-1, p ⫽ 0.6331. Cytokeratin fragment 19 was the most reliable marker. (FPR ⫽ false-positive rate; TPR ⫽ true-positive rate.)
apply this procedure to localized pulmonary diseases, including lung cancer. In this series, we simplified the measurements of tumor markers in ELF to facilitate its introduction into clinical practice by comparing the results with versus without correction for ELF weight. Because we found a close correlation between the two sets of results, only direct measurements were performed thereafter to simplify the investigations. The sensitivity of measurements of tumor markers in
Fig 4. (A) Cytokeratin fragment 19 values in epithelial lining fluid. (B) Carcinoembryonic antigen values. (C) Sialyl SSEA-1 values. The box-whisker plots show the 25th and 75th percentiles, the median (horizontal line within the box), and the 10th and 90th percentiles (whiskers). *p ⬍ 0.05, **p ⬍ 0.01 (cancerous tissue group versus other groups).
Fig 6. Microscopic immunohistochemical staining of cytokeratin fragment 19. The expression of cytokeratin fragment 19 was prominent in tumor tissue, weak in tissue near the tumor (A and B), and absent in normal lung tissue (C and D). Arrowheads show cancer cells and arrows show cells at marginal region. Bar in A and C, 100 m; B and D, 50 m.
1672
WATANABE ET AL BRONCHOSCOPIC MICROSAMPLING FOR LUNG CANCER
GENERAL THORACIC
the serum for the detection of nonsmall cell lung cancer is 42% for CEA, 58% for CYFRA in the literature [15], and 32% for SLX in another article [16]. We hypothesized that the concentrations of these tumor markers are much higher in the tumor itself, or in the tissue surrounding the tumor, than in serum, and thus applied our BMS procedure to collect local ELF and measure the tumor markers as a supplemental method to diagnose peripheral lung cancer. Both CYFRA and CEA concentrations in ELF collected near the tumor were significantly higher than in ELF from the contralateral lung or from lung of patients without cancer. In addition, whereas the serum concentrations of CYFRA were within normal limits, the concentration in ELF was high in these patients with lung cancer. Cytokeratin fragment 19 was the most reliable marker, followed by CEA, according to the results of our receiver operating characteristic analyses. In contrast, SLX in ELF was as unreliable as in serum in this series. Attempts to diagnose peripheral lung cancer by measuring CYFRA, CEA, and SCC antigens in BAL fluid have been reported by other investigators [9, 10], although these attempts were unsuccessful. Our BMS probe appears more successful as it can precisely select the target bronchus under fluoroscopy. There are, however, technical problems associated with our BMS method. Insertion of the probe into the Ba1⫹2 (R. apicodorsalis, Rm. apicalis) or Ba6 (R. (lobi inferioris) superior) segments may be difficult because of the stiffness of the catheter. In addition, if the probe is contaminated by some blood, as occasionally happens, the concentrations of tumor marker in ELF may be underestimated by the blood contamination because blood tumor marker concentration might be suggested to be lower than that in ELF. We have observed the immunohistochemical expression of tumor markers in several cases. The mode of expression of CYFRA in lung cancer tissues is reported equivocally in the literature [17, 18]. In our present study the expression of CYFRA, which was prominent in the tumor and mild in the neighboring tissue, suggests that the marker diffuses from the tumor where it is produced to the surrounding area, and that it can be sampled by the probe even without direct contact with the tumor. This BMS procedure was a useful adjunct in the diagnosis of peripheral lung carcinoma. Because DNA or RNA can be collected by this same procedure, the detection of cancer-related genes in ELF, for example telomerase, may be considerably simplified in the near future. In conclusion, these results suggest that tumor markers produced in cancerous lung nodule diffuse to surrounding areas and can be sampled by our BMS procedure even if the probe is not in direct contact with the tumor. Microsampling methods are capable of detecting tumor makers more discriminately than BAL. Cytokeratin fragment 19 and CEA concentrations in ELF play visible roles in the diagnosis of primary lung adenocarcinoma. It is suggested that our BMS procedure during bronchoscopy supplements the histologic diagnosis in patients with small peripheral lung cancer. These results will need to be confirmed in larger groups of patients and further validated in prospective studies.
Ann Thorac Surg 2003;76:1668 –73
This study was supported by Grant-in-Aid No. 12671329 (to MW) from the Ministry of Education, Science, Sports and Culture, Japan. We thank Satoru Fukinbara, PhD (Department of Biochemical Engineering and Science, Faculty of Computer Science and Systems Engineering, Kyushu Institute of Technology) for contributing to the statistical analysis.
References 1. Swensen SJ, Jett JR, Sloan JA, et al. Screening for lung cancer with low-dose spiral computed tomography. Am J Respir Crit Care Med 2002;165:508 –13. 2. Mori K, Yanase N, Kaneko M, Ono R, Ikeda S. Diagnosis of peripheral lung cancer in cases of tumors 2 cm or less in size. Chest 1989;95:304 –8. 3. Obara M, Satoh H, Ishikawa H, et al. Diagnostic procedures in obtaining pathological materials for pulmonary nodules due to primary lung cancer. Oncol Rep 1998;5:1237–9. 4. Seyfer AE, Walsh DS, Graeber GM, Nuno IN, Eliasson AH. Chest wall implantation of lung cancer after thin-needle aspiration biopsy. Ann Thorac Surg 1989;48:284 –6. 5. Voravud N, Shin DM, Dekmezian RH, Dimery I, Lee JS, Hong WK. Implantation metastasis of carcinoma after percutaneous fine-needle aspiration biopsy. Chest 1992;102:313–5. 6. Yoshikawa T, Yoshida J, Nishimura M, Yokose T, Nishiwaki Y, Nagai K. Lung cancer implantation in the chest wall following percutaneous fine needle aspiration biopsy. Jpn J Clin Oncol 2000;30:450 –2. 7. de Gracia J, Bravo C, Miravitlles M, et al. Diagnostic value of bronchoalveolar lavage in peripheral lung cancer. Am Rev Respir Dis 1993;147:649 –52. 8. Pirozynski M. Bronchoalveolar lavage in the diagnosis of peripheral, primary lung cancer. Chest 1992;102:372–4. 9. Zaleska J, Pirozynski M, Kwiek S, Sakowicz A, RowinskaZakrzewska E. CEA, NSE, and SCC Ag in bronchial lavage in patients with lung cancer. Rocz Akad Med Bialymst 1997; 42(Suppl 1):179 –89. 10. Cremades MJ, Menendez R, Pastor A, Llopis R, Aznar J. Diagnostic value of cytokeratin fragment 19 (CYFRA 21–1) in bronchoalveolar lavage fluid in lung cancer. Respir Med 1998;92:766 –71. 11. Ishizaka A, Watanabe M, Yamashita T, et al. New bronchoscopic microsample probe to measure the biochemical constituents in epithelial lining fluid of patients with acute respiratory distress syndrome. Crit Care Med 2001;29:896 –8. 12. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29 –36. 13. Song HH. Analysis of correlated ROC areas in diagnostic testing. Biometrics 1997;53:370 –82. 14. Noguchi M, Morikawa A, Kawasaki M, et al. Small adenocarcinoma of the lung. Histologic characteristics and prognosis. Cancer 1995;75:2844 –52. 15. Schneider J, Velcovsky HG, Morr H, Katz N, Neu K, Eigenbrodt E. Comparison of the tumor markers tumor M2-PK, CEA, CYFRA 21–1, NSE and SCC in the diagnosis of lung cancer. Anticancer Res 2000;20:5053–8. 16. Ueoka H, Ohnoshi T, Moritaka T, et al. A statistical analysis of serum sialyl Lewis X-1 (SLX), CEA, SCC and NSE levels in patients with lung cancer. Nihon Kyobu Shikkan Gakkai Zasshi 1991;29:1022–8. 17. Quillien V, Ramee MP, Bansard JY, et al. Serum and tissue distribution of a fragment of cytokeratin 19 (cyfra 21–1) in lung cancer patients. Anticancer Res 1995;15:2857–63. 18. Chyczewski L, Niklinski J, Chyczewska E, Laudanski J, Furman M. Immunohistochemical analysis of tissue localization of cytokeratin 19 in lung cancer. Rocz Akad Med Bialymst 1997;42(Suppl 1):162–72.