Thoracoscopy for the Diagnosis of Pleural Disease Richard Menzies, MD, MSc; and Marc Charbonneau, MD

Transcription

1 Thoracoscopy for the Diagnosis of Pleural Disease Richard Menzies, MD, MSc; and Marc Charbonneau, MD Objective: To assess the accuracy and safety of thoracoscopy for the evaluation of pleural disease. Design: Prospective evaluation of patients referred for thoracoscopy. Setting: University hospital specializing in chest diseases. Patients: We studied 102 patients with pleural disease, the cause of which had not been determined after initial investigation, including thoracentesis and needle biopsy. Eighty-six patients had pleural effusion, 11 had pleural mass, and 5 had pleural effusion in association with a known primary lung carcinoma. Intervention: All patients had thoracoscopy under local anesthesia with mild sedation. Visually directed biopsies were done of parietal pleura. Measurements: We recorded clinical characteristics, laboratory data, findings and duration ofthoracoscopy, and any complications associated with the procedure. Hospital and clinic follow-up records were reviewed, and patients were contacted by telephone 12 and 24 months after thoracoscopy to assess their health status. Main Results: One hundred and four thoracoscopies were done in 102 patients. A definitive diagnosis was established in 95 patients: 42 had malignant pleural disease and 53 had benign pleural disease. A diagnosis of benign pleural disease using thoracoscopy could not be confirmed in the remaining 7 patients because of insufficient follow-up information. Overall, thoracoscopy was 96% accurate with a sensitivity of 91%, a specificity of 100%, and a negative predictive value of 93% for the diagnosis of pleural malignancy. Thoracoscopy was well tolerated under local anesthesia and entailed hospitalization for less than 24 hours in most cases. No deaths occurred, although 1.9% of patients had major complications, and 5.5% had minor complications. Conclusions: Among patients with pleural disease remaining undiagnosed after usual initial investigation, thoracoscopy done under local anesthesia is a rapid, safe, and well-tolerated procedure with an excellent diagnostic yield that is equivalent to that of thoracotomy. 1 he diagnosis of pleural disease is often problematic. Even after thoracentesis, pleural biopsy, and other studies relevant to the clinical circumstances have been done, 21% to 27% of patients remain undiagnosed (1-4). Of these, approximately 50% will ultimately be diagnosed with malignancy and a few may prove to have tuberculosis, fungal disease, or pulmonary emboli, but 20% to 45% of patients will continue to be classified as having idiopathic disease (1-4). The clinician faced with a patient who remains undiagnosed after initial studies can choose several further diagnostic procedures. Both bronchoscopy and computed tomography have been found to have low yield (5, 6). Immunologic studies (7) and alpha-fetoprotein and carcinoembryonic antigen assays (8) have proved to be too nonspecific. Thoracotomy entails major surgery with attendant risks and discomfort, as well as several days of hospitalization. Observation alone is often not acceptable to the patient and family because of the high likelihood of malignancy. The diagnostic option that remains is thoracoscopy. Thoracoscopy was widely used by phthisiologists in North America until the late 1940s in collapse therapy for the lysis of adhesions in patients with tuberculosis. With the advent of antibiotic therapy, collapse therapy was abandoned and with it thoracoscopy. In the last 10 to 15 years, thoracoscopy has been reintroduced by chest physicians in many European centers for the diagnosis of pleural diseases (2, 9-11). For 1500 thoracoscopies done in eight major centers (2, 9-15), the diagnostic accuracy of thoracoscopy averaged over 90%. Major complications, such as hemothorax, empyema, and arrhythmia, occurred in less than 3% of cases, and there was one death (14). Despite this success, thoracoscopy is done in only a few North American centers. This may reflect lack of awareness of the advantages of this procedure. Over the last 4 years, thoracoscopy has been done at our center in over 100 patients who were referred after conventional evaluation of pleural disease failed to yield a diagnosis. We have prospectively gathered data on the accuracy, tolerance, and safety ofthoracoscopy, as well as on clinical and laboratory characteristics of these patients. Methods Patients Annals of Internal Medicine. 1991;114: From Montreal Chest Hospital, Montreal, Quebec. For current author addresses, see end of text. Thoracoscopy was considered for patients with pleural effusion only after routine evaluation, including at least one thoracentesis and percutaneous needle biopsy (Abrams needle, Downs Surgical, London, England), proved to be nondiagnostic. This evaluation was not required for patients with pleural mass. Transthoracic needle biopsies, fiberoptic bronchoscopies, and chest computed tomographic scans were done in many patients at the discretion of the referring physician. Thoracoscopy was not done if patients were judged to be unable to 1991 American College of Physicians 271

2 tolerate a 20% to 30% pneumothorax, were confused or otherwise unable to cooperate, or had a Paco 2 of more than 50 mm Hg. Data Collection Before thoracoscopy, we examined all patients in order to complete a questionnaire concerning demographic data, symptoms, and medical, smoking, and occupational histories. Results of all previous investigations were also recorded. The duration, complications, and findings of thoracoscopy were recorded, as was the duration of tube thoracostomy and hospital stay. Thoracoscopy Before the procedure, patients with pleural effusion had thoracentesis to remove 300 to 500 ml of fluid. In all patients, approximately 300 ml of air was insufflated through a millipore filter. Pleural pressure was monitored using a water manometer in order to maintain pressure below + 5 cm H 2 0. Thoracoscopy was done with the patient lying in the lateral decubitus position with the affected side upward. Before thoracoscopy, all patients were administered 50 to 75 mg of meperidine intramuscularly. During the procedure, local anesthesia was used, although intravenous diazepam or fentanyl was also administered if necessary. Rigid endoscopes (Karl Storz Endoscopes, Tuttlingen, Federal Republic of Germany) with viewing angles of 0, 45, and 90 degrees were used. Biopsy specimens of parietal pleura were obtained under direct vision and were sent for mycobacterial culture, routine histologic examination, and immunochemistry and electron microscopy studies. At the end of thoracoscopy, an intercostal drain was inserted to evacuate any remaining air and fluid and to ensure that there was no bleeding or air leak. Patients were discharged that evening or the following morning. Follow-up The diagnosis made using thoracoscopy was considered to be established if confirmed by thoracotomy or autopsy or if another definite (benign) diagnosis was made (for example, pulmonary embolus), the pleural disease and all associated symptoms and signs resolved, or the pleural disease remained stable after follow-up of more than 1 year without appearance of other disease manifestations. Follow-up was done by reviewing records from the hospital or clinic where patients were followed after thoracoscopy. Patients were also contacted by telephone annually, most recently in the late fall of If a patient's status remained unclear, either the attending physician was called or the patient was seen by one of us. Data Analysis Bivariate associations were tested for significance using the chi-square test for dichotomous variables and the Student r-test for continuous variables. A P value of less than 0.05 was considered statistically significant. The 95% confidence intervals (CI) for proportions were calculated for sensitivity and specificity, as suggested by Colton (16). whom an ipsilateral primary lung malignancy had already been diagnosed. Among patients with pleural effusion, 30% had had one thoracentesis, 43% had had two, and the remaining 27% had had three. Among the patients, 5% had not had an adequate pleural needle biopsy (insufficient tissue or other technical problems), 65 (74%) had had one biopsy; 19%, two biopsies; and 2%, three biopsies. In addition, 62% of the patients had had fiberoptic bronchoscopy, and 51% had had a chest computed tomographic scan. Final Diagnoses The final diagnoses of the 102 patients are shown in Table 1. Forty-two patients were diagnosed with malignant pleural disease and 53 with benign disease. A definitive diagnosis of the pleural disease could not be made for 7 patients. In 1 patient, the procedure could not be completed because of an arrhythmia. Two patients with known primary lung cancer had no thoracoscope evidence of pleural metastases but had mediastinal involvement confirmed by mediastinoscopy, precluding further evaluation. Another had an inadequate thoracoscopic examination because of extensive pleural adhesions but declined further evaluation. The remaining three patients died within 9 months, one each from cirrhosis, stroke, and esophageal cancer. Of the 53 patients with benign pleural disease, only 7 had this diagnosis confirmed by thoracotomy. As Table 2 shows, in 31 patients, another diagnosis of benign disease was made: Three patients had tuberculosis based on the finding of caseating granulomas on histologic examination; three had pulmonary emboli; two patients had clinical rheumatoid arthritis with a positive serologic test, and another had pleuropericarditis with a strongly positive antinuclear antibody assay; three patients with chylothoraces proved to have mediastinal involvement with lymphoma without direct pleural involvement; one patient was judged after more than a year of follow-up to have rounded atelectasis on the basis of typical radiographic and clinical findings. Asbestos effusion was diagnosed in 8 patients on the basis of a history of asbestos exposure, asbestos plaques visualized at thoracoscopy, and follow-up of at least 1 year; and all 5 patients with cardiac-related pleural effusions had resolution of disease. The remainder are listed in Table 2. Of the 22 with idiopathic effusions, 19 had resolution of disease and the other 3 patients have remained stable after follow-up ranging from 6 to 11 months. Results Indications and Patient Selection Between November 1985 and December 1989, 108 patients were referred for thoracoscopy for the diagnosis of pleural disease. Of these, 4 were judged to be too ill to undergo the procedure, one was unable to cooperate because of dementia, and one refused. Thoracoscopy was done in 102 patients and repeated in 2 patients. Thoracoscopy was done in 86 patients with undiagnosed effusion, in 11 patients with undiagnosed mass, and for the evaluation of effusion in 5 patients in Table 1. Final Diagnosis in 102 Patients Indication Cancer Benign No Total Disease Diagnosis < n» Effusion Mass Effusion and known primary lung carcinoma Total February 1991 Annals of Internal Medicine Volume 114 Number 4

3 Table 2. Final Diagnosis in 95 Patients with a Definite Diagnosis Diagnosis Patients, n Benign cause 53 Idiopathic 22 Other 31 Asbestos 8 Chylothorax 3 Droessler syndrome 3 Rheumatoid arthritis or systemic lupus erythematosus 3 Pulmonary embolus 3 Trauma or hemothorax 3 Tuberculosis 3 Congestive heart failure 2 Post-obstructive pneumonia 2 Rounded atelectasis 1 Malignant cause 42 Mesothelioma 20 Metastatic 22 Adenocarcinoma 12 Squamous cell 2 Other or undifferentiated 8 Clinical Characteristics The patients included 65 men and 37 women. Their mean age was 61.7 years. The characteristics of the 95 patients with a definite diagnosis (malignant mesothelioma, metastatic malignancy, or benign disease) are shown in Table 3. The results for 3 patients who had chylothorax due to mediastinal lymphoma and for 2 patients who had primary lung cancer with benign pleural disease are not included in the analysis of clinical characteristics and symptoms. Patients with mesothelioma were more likely to be male, to have had asbestos exposure, and to have a higher pleural fluid lactate dehydrogenase level and a lower ph and glucose level. However, patients who had idiopathic effusions were more likely to be female and to have fewer symptoms and a lower pleural fluid lactate dehydrogenase level. Results of Thoracoscopy Of 42 patients with malignant pleural disease, 38 were diagnosed using thoracoscopy. Fifty-seven patients were diagnosed with benign disease on the basis of thoracoscopic findings and this diagnosis was ultimately confirmed in 53 patients. Therefore, the overall accuracy of this procedure was 96%. For the diagnosis of malignancy, thoracoscopy was 91% sensitive and 100% specific with a negative predictive value of 93%. Five patients who were diagnosed with benign disease using thoracoscopy were lost to follow-up. Table 4 shows the effect of including these patients on the calculated sensitivity, specificity, and negative predictive value of thoracoscopy. Sensitivity would have been only 81% and negative predictive value 85% if all 5 had actually had malignancy. On the other hand, if it is assumed that the rate of malignancy among these 5 patients was the same as that seen among the 95 patients with a definite diagnosis (45%), then sensitivity would have been 86%; specificity, 100%; and negative predictive value, 90%. Of the four patients in whom thoracoscopy gave false-negative results, the final diagnosis in all cases was malignant mesothelioma. In three of these patients, the examination was limited by extensive adhesions; in two of these patients, the pathologic appearance suggested malignancy. All three were referred immediately for thoracotomy for a definitive diagnosis. In only one instance was the examination adequate and the pathologic appearance unequivocally benign, so that the patient was falsely reassured and discharged. Thoracotomy and pleurectomy were done 1 month later to control recurrent massive effusion, at which time epithelial mesothelioma with extensive desmoplastic reaction was found. Acceptability and Safety The mean duration of the 104 thoracoscopies was 46 minutes. Intravenous diazepam was required in 38% and fentanyl in 33% of procedures. The duration of tube thoracostomy averaged 11.9 hours, and the patients were judged ready for discharge after a mean of 26.5 hours. As shown in Table 5, the duration of tube thoracostomy and hospitalization was significantly shorter among those who had benign disease compared with those who had malignant pleural disease. These latter patients had a greater incidence of air-leak complications. No deaths occurred, but two patients had major complications (rate, 1.9%). A 35-year-old woman who had received combined local irradiation and chemotherapy including adriamycin for breast cancer 1 year earlier experienced ventricular tachycardia and vascular collapse during insufflation of air at the beginning of the Table 3. Clinical Characteristics of Patients with Benign and Malignant Effusions* Characteristic Malignant Effusion Benign Mesothe Meta Effusion! lioma static Demographic factors Male, % 90$ 57$ 59 Age, y Smoking Yes, % Pack-years Asbestos exposure, % 63* 24$ 25 Symptoms, % Pleuritic chest pain Anorexia and weight loss Fatigue and lethargy None of these symptoms Pleural fluid assessments Lactate dehydrogenase, ixkatlh (u/l) 9.2 (553) 7.0 (420) 6.6 (394) Protein, gil Glucose, mmolll 3.8* 6.3$ 5.4 ph 7.31$ 7.43$ 7.40 Outcome Alive at follow-up, % * For 95 patients with a definite diagnosis. t Five patients were excluded from the "benign effusion" group: three with lymphoma and two with primary lung cancer. X Difference between mesothelioma compared with other: P < Difference between benign compared with other: P < Mean values are given. 15 February 1991 Annals of Internal Medicine Volume 114 Number 4 273

4 Table 4. Sensitivity and Specificity of Thoracoscopy Situation Patients Sensitivity Speci- Negative (95% CI) ficity Predictive Value (95% CI) n < % > Known outcome (85 to 97) (88 to 98) Assumption 1* (79 to 93) (84 to 96) Assumption 2t (73 to 89) (78 to 92) * Assumption 1: The proportion of patients with malignancy was the same (45%) in the five patients lost to follow-up as in the known population, so that two of the five had had a false-negative diagnosis. t Assumption 2: All five patients lost to follow-up actually had pleural malignancy and thus had had a false-negative (worst case) diagnosis. procedure. She was resuscitated successfully and remains alive and well 3 years later, although the effusion remains unexplained. In the second patient, thoracoscopy was limited by extensive adhesions and biopsy could not be done. Massive subcutaneous emphysema developed within 6 hours requiring tube thoracostomy. Significant air leak persisted until thoracotomy was done 1 week later to close this leak, at which time mesothelioma was diagnosed. Minor complications developed in 8 patients (rate, 7.5%). Air leak developed only among those with malignant effusions. Fever developed in two patients, one of whom had rheumatoid arthritis and one of whom proved to have pulmonary emboli; thus, fever may have been related to their underlying disease rather than to the procedure. One patient had excessive bleeding at the biopsy site, but bleeding stopped without sequelae after 10 minutes. No patient with idiopathic effusion developed complications. When the first 52 procedures were compared with the latter 52, there was no difference in duration of thoracoscopy nor in the amount of sedation used. However, the median duration of tube thoracostomy decreased from 6 to 3 hours, and the mean duration of hospitalization decreased from 32 to 23 hours (P < 0.01). Complications occurred in 13.5% of the first 52 cases but in only 5.5% of the latter 52. Discussion Thoracoscopy was done for the diagnosis of pleural disease in 102 patients. A definite diagnosis was made in 95 patients. Even though this procedure was done only in patients in whom extensive investigations had proved to be nondiagnostic, thoracoscopy was 90% sensitive with a negative predictive value of 93% for the diagnosis of malignant pleural disease. Accuracy of Thoracoscopy The diagnostic accuracy of pleural disease with thoracoscopy in other series has been influenced by the methods used to confirm the diagnoses and the length of follow-up. In four series reporting a diagnostic accuracy of 90% to 100%, follow-up was either not stated or lasted less than 6 months (9, 12, 14, 15). By contrast, in three series in which a total of 822 patients were followed for 1 to 5 years, accuracy was 62% to 85% (2, 11, 13). Boutin and colleagues (2) reported false-negative results in 15% of patients after 1 year of follow-up, whereas Enk and Viskum (11) reported an accuracy of 69% among patients followed for up to 5 years. Even after thoracotomy established a diagnosis of benign pleural disease, 13 of 51 (25%) patients were later diagnosed to have malignancy after a median interval of 6 months (3). The accuracy of diagnosis of malignancy and tuberculosis in our series compares well to that reported in series with longer follow-up (2, 9-15). There are several reasons why thoracoscopy should be diagnostic when cytologic examination of pleural fluid or percutaneous blind needle biopsy proves to be nondiagnostic. Cytologic examination, even with optimal facilities and repeated examinations of large amounts of pleural fluid, will be diagnostic in only 60% to 80% of patients with metastatic disease (8) and in fewer than 20% of patients with mesothelioma (17). Canto and colleagues (18) reported that among 79 patients with proven metastatic pleural disease, only 53% had involvement of the costal pleura. In an autopsy study of 53 patients with metastatic pleural malignancy, Meyer (19) found that 17 (32%) had involvement of visceral pleura only. Therefore, among patients with metastatic pleural disease, 32% to 47% will have localized disease inaccessible to percutaneous needle biopsy, whereas specimens from needle biopsy will rarely be of sufficient size or depth to allow a definitive diagnosis of mesothelioma. Thoracoscopy allows inspection of approximately 75% of the visceral pleural surface as well as of the parietal pleural surface. Biopsies can be visually directed in instances where tumor deposits appear to be localized. In addition, biopsy specimens can be obtained from multiple sites and are of greater size and depth, factors which improve the diagnostic yield, particularly for malignant mesothelioma (8, 17, 20). All four patients in whom thoracoscopy provided a false-negative result were diagnosed to have mesothelioma. Contributory factors were inadequate visualization due to extensive adhesions and the difficulties involved in the pathologic diagnosis of mesothelioma, difficulties that have been previously described (8, 17, 20, 21). In other series of patients undergoing thoracoscopy, mesothelioma was correctly diagnosed in 21 of 35 patients (60%) (14) and in 15 of 20 patients (75%) (22). Even among patients undergoing thoracotomy, me- Table 5. Acceptability and Safety of Thoracoscopy Factor Malignant Benign Overall Procedures, n 43* 54* 97 Duration of thoracoscopy, min Sedation Valium, mg Fentanyl, fig Tube thoracostomy, h Hospitalization, h Complications, % Major Minor * Two patients (one with malignancy and one with benign disease) had two thoracoscopies February 1991 Annals of Internal Medicine Volume 114 Number 4

5 sothelioma was correctly diagnosed in only 31 of 35 patients (88.5%) (3), which emphasizes the difficulty of diagnosing this condition. Acceptability and Safety of Thoracoscopy No thoracoscopy had to be terminated because of patient intolerance caused by pain or dyspnea. Fewer than 50% of the patients required additional sedation or analgesia other than the meperidine given before the procedure. Six other centers have also reported that thoracoscopy has been well tolerated under local anesthesia (9, 11-13, 15, 22). We did not routinely monitor arterial oxygen saturation because our early experience showed that subjective dyspnea preceded arterial desaturation and because of the findings in three other series in which arterial saturation was monitored (13, 22, 23). In these series (total of 26 patients), arterial oxygen saturation decreased by an average of 1% and maximal decrease was 4% (13, 22, 23). We used a rigid thoracoscope exclusively. Chest physicians are more familiar with the flexible fiberoptic scopes. However, in three series in which the two techniques were compared, diagnostic accuracy was significantly greater with the rigid scope (13, 15, 22). In most centers, it is standard practice to insert an intercostal drain at the end of the procedure (2, 9, 14). However, Enk and Viskum (11) reported that only 9 of 556 patients (4%) required this drainage, and Davidson and colleagues (22) reported that intercostal drainage was needed for only a few minutes, during which time all the air could be expelled. We inserted an intercostal drain in all patients, but as our experience grew, we steadily decreased the duration of intercostal drainage to reduce patient discomfort, risk for intrapleural infection (9, 22), and duration of hospitalization. Among our patients, 1.9% had major complications and 5.6% had minor complications. The complication rate in our study was similar to that found in 1000 patients undergoing percutaneous pleural needle biopsy (2) and to those reported by other centers doing thoracoscopy under local anesthesia (9, 13, 22, 24). Our rate is significantly lower than that reported by Page and colleagues (14). These investigators did thoracoscopy under general anesthesia in 121 patients, 1 of whom died from myocardial infarction and 2 of whom developed respiratory failure requiring mechanical ventilation within 24 hours (14). Diagnostic yield reported by centers doing thoracoscopy under local anesthesia (9, 11, 13, 22) is the same as that reported by centers using general anesthesia (2, 10, 14). Given the excellent tolerance, low rate of complications, and high diagnostic yield found in our study as well as in others in which thoracoscopy was done under local anesthesia, the routine use of general anesthesia for thoracoscopy does not appear to be justified. Significant air leak occurred exclusively among patients with malignant pleural disease. Of the six patients who had air leak, none had biopsies of lung or visceral pleura, but all had adhesions between visceral and parietal pleura. A possible explanation is that during initial induction of the pneumothorax, these adhesions may have been stretched and torn with resultant trauma to the underlying lung. Prolonged air leak, or pneumothorax, is a common complication reported by other centers (22), but the association with malignancy has not been previously noted. In our study of patients undergoing thoracoscopy after the usual initial investigation had proved to be nondiagnostic, clinical and laboratory characteristics were found to be too nonspecific to be of diagnostic utility. Thoracoscopy was well tolerated under local anesthesia, requiring minimal additional sedation. The duration of tube thoracostomy steadily diminished as our experience grew, so that the procedure could be done on an outpatient basis. Complications were uncommon, occurring mostly in the first 52 patients to have thoracoscopy and in those with malignant disease. Thoracoscopy provided a definitive diagnosis of malignancy or tuberculosis with a high degree of accuracy, and for patients in whom there was adequate visualization and an unequivocal pathologic diagnosis of benign disease the prognosis was excellent. Thoracoscopy is rapid, safe, and highly accurate. This procedure is also cost effective because it can be done in a bronchoscopy suite (2, 11, 22), it is well tolerated under local anesthesia with minimal additional sedation, and patients do not require hospitalization (after 6 hours of post-procedure observation, patients can be discharged). Consideration should be given to the re-introduction of thoracoscopy in centers where it is not available. When cytologic examination of pleural fluid and percutaneous needle biopsy are nondiagnostic, thoracoscopy can provide a definitive diagnosis with a high degree of accuracy and minimal risk or discomfort to the patient. Acknowledgments: The authors thank Drs. J. Martin and L. Knight for their critical review of the manuscript. Grant Support: In part by the Canadian Lung Association and by the Research Institute of the Royal Victoria Hospital. Requests for Reprints: Richard Menzies, MD, MSc, Montreal Chest Hospital, 3650 St. Urbain Street, Montreal, Quebec, Canada H2X 2P4. Current Author Addresses: Drs. Menzies and Charbonneau: Montreal Chest Hospital, 3650 St. Urbain Street, Montreal, Quebec, Canada H2X 2P4. References 1. Gunnels JJ. Perplexing pleural effusion. Chest. 1978;74: Boutin C, Viallat JR, Cargnino P, Farisse P. Thoracoscopy in malignant pleural effusions. Am Rev Resp Dis. 1981;124: Ryan CJ, Rodgers RF, Unni KK, Hepper NG. The outcome of patients with pleural effusion of indeterminate cause at thoracotomy. Mayo Clin Proc. 1981;56: Poe RH, Israel RH, Utell MJ, Hall WJ, Greenblatt DW, Kallay M. Sensitivity, specificity, and predictive values of closed pleural biopsy. Arch Intern Med. 1984;144: Feinsilver SH, Barrows AA, Braman SS. Fiberoptic bronchoscopy and pleural effusion of unknown origin. Chest. 1986;90: Pugatch RD, Fating LJ, Robbins AH, Sider GL. Differentiation of pleural lesions using computed tomography. J Comput Assist Tomog. 1978;2: Halla JT, Schrohenloher RE, Volanakis JE. Immune complexes and other laboratory features of pleural effusions: a comparison of rheumatoid arthritis, systemic lupus erythematosis, and other diseases. Ann Intern Med. 1980;92: Light RW. Pleural Diseases. Philadelphia: Lea & Febiger; Canto A, Blasco E, Casillas M, et al. Thoracoscopy in the diagnosis of pleural effusion. Thorax. 1977;33: Weissberg D, Kaufman M. Diagnostic and therapeutic pleuroscopy: experience with 127 patients. Chest. 1980;78: February 1991 Annals of Internal Medicine Volume 114 Number 4 275

6 11. Enk B, Viskum K. Diagnostic thoracoscopy. Eur J Respir Dis. 1981;62: DeCamp PT, Moseley PW, Scott ML, Hatch HB. Diagnostic thoracoscopy. Ann Thorac Surg. 1973;16: Oldenberg FA, Newhouse MT. Thoracoscopy, a safe, accurate diagnostic procedure using rigid thoracoscope and local anesthesia. Chest. 1979;75: Page RD, Jeffrey RR, Donnelly RJ. Thoracoscopy: a review of 121 consecutive surgical procedures. Ann Thorac Surg. 1989;48: Wu MH, Hsiue RH, Tseng KH. Thoracoscopy in the diagnosis of pleural effusions. Jpn J Clin Oncol. 1989;19: Colton T. Statistics in Medicine. Boston, Massachusetts: Little, Brown and Company; Legha SS, Muggia FM. Pleural mesothelioma: Clinical features and therapeutic implications. Ann Intern Med. 1977;87: Canto A, Ferrer G, Ramagosa V, Moya J, Bernat R. Lung cancer and pleural effusion: clinical significance and study of pleural metastatic locations. Chest. 1985;87: Meyer PC. Metastatic carcinoma of the pleura. Thorax. 1966;21: McCaughey WT, Al-jabi M. Differentiation of serosal hyperplasia and neoplasia in biopsies. Pathol Annu. 1986;1: Cantin R, Al-jabi M, McCaughey WT. Desmoplastic diffuse mesothelioma. Am J Surg Pathol. 1982;6: Davidson AC, George RJ, Sheldon CD, Sinha G, Corrin B, Geddes DM. Thoracoscopy; assessment of a physician service and comparison of a flexible bronchoscope used as a thoracoscope with a rigid thoracoscope. Thorax. 1988;43: Faurschou P, Madsen F, Viskum K. Thoracoscopy: influence of the procedure on some respiratory and cardiac values. Thorax. 1983; 38: Viskum A, Enk B. Complications of thoracoscopy. Poumon-Coeur. 1981;37: February 1991 Annals of Internal Medicine Volume 114 Number 4