Mediastinal Lymph Node Dissection in Resected Lung Cancer: Morbidity and Accuracy of Staging

Mediastinal Lymph Node Dissection in Resected Lung Cancer: Morbidity and Accuracy of Staging Ewald C. M. Bollen, MD, Cees J. van Duin, MD, Paul H. M. H. Theunissen, MD, PhD, Bep E. v. t Hof-Grootenboer, BSc, and Geert H. Blijham, MD, PhD Departments of Surgery and Pathology, De Wever Hospital, Heerlen; Department of Pathology, University of Nijmegen; and Department of Oncology, University Hospital Utrecht, the Netherlands A study was performed to investigate the morbidity of mediastinal lymph node dissection (MND) and to establish its contribution to the accuracy of staging in surgically treated non-small cell lung cancer. Between 1988 and the middle of 1991 a systematic sampling of mediastinal lymph nodes was done in 20 patients and a MND was carried out in 65 patients. Data from these patients were compared with those from a control group of 70 patients operated on in 1986 and 1987, who would have had MND if they had been treated in the years after 1988. The groups were comparable according to important clinical characteristics. There was a significantly greater fluid production via the drains in the groups with systematic sampling and MND, compared with the controls. Volume of blood lost during the operation and number of units blood transfused perioperatively were not significantly different between the groups. Three lesions of the recurrent laryngeal nerve and two episodes of chylothorax were observed, all probably caused by MND. The discovery ratio for N2 disease in the MND and systematic sampling groups together compared with the control group was 2.1, with a 9 confidence interval from 1.04 to 4.2. (~nn rhorac surg z993;55:96z-a peration still offers the best prospects for patients 0 with non-small cell lung cancer. This certainly applies to the early stages of the disease. There is still controversy regarding the role of operation when the mediastinal lymph nodes (N2) are affected. In retrospective studies, a prognosis of 19% to 30% actuarial 5-year survival in N2 disease [14] has been claimed. Other writers doubt the general usefulness of operation in N2 disease, restricting it to patients whose nodal metastases escape the preoperative evaluation [5-7]. Once operation is indicated in N2 disease, it is generally agreed that mediastinal node dissection (MND) has to be performed, first, because all tumor has to be excised, and, second, because MND has been claimed to lead to a more accurate staging [8]. To what extent MND improves the accuracy of staging has not fully been studied. It is remarkable, for a procedure so generally recommended [9], that so few data are available with regard to its morbidity. In this study with historical controls, a comparison is made between plain resection and resection with exploration of the mediastinal nodes to determine whether, and to what extent, MND improves accuracy of staging, and to determine the morbidity of MND. Material and Methods Study Period The study period ran from 1986 to the middle of 1991. During this period, 210 patients with non-small cell lung Accepted for publication July 30, 1992. Address reprint requests to Dr Bollen, Department of Surgery, De Wever Hospital, PO Box 4446, 6401 CX Heerlen, the Netherlands. cancer, 190 men and 20 women, were operated on in De Wever Hospital, Heerlen. The mean age was 63 years, ranging from 33 to 80 years. A squamous cell carcinoma was diagnosed 142 times (67.6%), an adenocarcinoma 44 times (21%), a large cell carcinoma 15 times, and another non-small cell lung cancer 9 times (4.3%). Pulmonary resection consisted of pneumonectomy in 64 patients (30.), lobectomy in 102 patients (48.6%), sleeve lobectomy in 17 patients (8.1%), segmentectomy or wedge resection in 14 patients (6.6%), and exploratory thoracotomy in 13 patients (6.2%). One hundred fifteen patients (54.8%) proved to be in stage I, 23 (11%) in stage 11, 62 (29.) in stage IIIa, and 10 (4.7%) in stage 11%. Patient Population The patient population was divided into three consecutive cohorts: in 1987 and early 1988 patients with non-small cell lung cancer underwent pulmonary resection without special attention to the mediastinal nodes; in 1988 an exploration of the mediastinum was added to the resection enabling systematic sampling (SS) of all the mediastinal node stations. Since 1989, instead of the sampling, MND became the routine procedure. The experimental group consisted of the patients with an SS (n = 20) or an MND (n = 65) operated on after 1988. In the same period, 26 other patients had no extended exploration of the mediastinum for various reasons: poor pulmonary or physical condition, severe silicosis (40% of the patients were former coal miners), much blood loss during the operation, age greater than 76 years, palliative resection, and exploratory thoracotomy (n = 7). The control group (n = 70) consisted of those patients operated on in 1986 and 1987 who would have been 0 1993 by The Society of Thoracic Surgeons 0003-4975/93/$6.00

962 BOLLEN ET AL Ann Thorac Surg 1993;55:961-6 selected for an exploration if they had been operated on after 1988. The same reasons for omitting mediastinal exploration in patients operated on after 1988 also applied to those operated on in 1986 and 1987. Thus, 29 patients had to be excluded from the control group. In summary, of the 210 patients operated on in the study period, 155 could finally be included in the case control study. Pretreatment In all patients a computed tomographic scan was part of the pretreatment staging. Until 1989 a mediastinoscopy was done only when nodes larger than 1.0 cm in diameter were seen on the computed tomographic scan. Since 1989 cervical mediastinoscopy has been performed routinely. Preoperative parasternal mediastinal exploration for left upper lobe tumors was performed in 6 patients of the control group and in 1 patient of the MND group. Patients with positive nodes found at preoperative mediastinal examination were excluded from resectional therapy. In the study period, 40 patients were excluded for operation because of positive findings at cervical mediastinoscopy. Positive findings at parasternal mediastinoscopy never excluded patients for operation. Mediastinal Node Dissection Technique Mediastinal node dissection was performed according to the method described by Naruke, Martini, and their associates [2, 101. In case of a pneumonectomy or upper lobe resection, mediastinal nodes were dissected en bloc. On the right side this applied to nodes in the superior or paratracheal compartments, and on the left side this applied to nodes in the aortopulmonary window. The nodes in the supraaortic compartment on the left side were merely sampled rather than dissected free. In the case of a lower lobe resection, the above-mentioned procedure was done separately. In all resections excision of the nodes in the subcarinal and inferior mediastinum was done after resection of the tumor. The nodes were numbered according to the chart of Naruke and coworkers [ll]. They were examined histologically and classified as NO, N1, or N2 according to the TNM system [=I. In the control group, only palpable lymph nodes suspected of being malignant were sampled before or after resection. One drain was left in situ for 24 hours after a pneumonectomy. After every other resection two drains were left in situ, one in the sinus pleura and the other one in the upper thorax. The fluid production via the drains and the time that the drains remained in situ were only measured in patients who had not undergone pneumonectomy. Accuracy of Staging To determine the accuracy of N staging, the number of N2 diseases discovered through exploration of the mediastinum was compared with the number discovered when no exploration was done. The relative discovery probability or discovery ratio was calculated by analogy with the calculation of the relative risk ratio [13]. The confidence interval was also determined. Statistics The data from the patients were processed by SFSS (Statistical Package for Social Studies). The three groups (MND, SS, and control) were compared using the 2 test for nominal data and one-way analysis of variance or the Kruskal-Wallis test for metric data, depending on the distribution (normal or outliers). The 9 confidence intervals for the discovery ratio were calculated using the large-sample approximation as is usually done for relative risks. Results Patient Characteristics Relevant characteristics of the 155 patients studied are summarized in Table 1. No significant differences were found between the three groups (MND, SS, and control) for age, sex, coughing, dyspnea, hemoptysis, number of asymptomatic patients, weight loss 6 months before operation, albumin level, lactate dehydrogenase level, lung function (mean forced expiratory volume in 1 second, vital capacity, and diffusing capacity), type of resection, histology, and T stage of the tumor. Morbidity PERIOPERATIVE BLOOD LOSS. Three parameters for blood loss during and after the operation were studied: the amount of blood lost during the operation itself, blood loss after operation making repeat thoracotomy necessary, and the need for blood transfusions during and after the operation within the period of hospital admission. Results are summarized in Table 2. Although some differences in the investigated parameters were observed, none of these were significant. In the 4 patients who needed a repeat thoracotomy, this procedure was not related to the mediastinal exploration. Of the total of 155 patients, 54% received no blood transfusion during the hospital admission. FLUID PRODUCTION VIA THE DRAINS. There was a significantly higher fluid production in the MND and SS groups compared with that in the control groups (Table 3). This applies to the production during the first 2 days as well as to the total production. The higher drain production in the MND and SS patients was not related to the number of days in which the drains remained in situ. BRONCHOPLEURAL FISTULA. Fistulas were seen three times (Table 4). Two fistulas developed within 10 days of pneumonectomy, one in the control group and one in the SS group. These fistulas were closed surgically with an uneventful postoperative course. A bronchopleural fistula with empyema developed in 1 patient in the SS group after pneumonectomy with rib resections (T3 Nl). The condition of the patient was poor, and he died 3 months after the pneumonectomy. No fistula developed in the MND group. HOARSENESS. Two patients in the control group and 6 patients in the MND group had hoarseness (see Table 4).

Ann Thorac Surg 1993;55:%1-6 BOLLEN ET AL 963 Table I. Characteristics of the Three Groups MND SS Control Variable (n = 65) (n = 20) (n = 70) Demographic features Age (Y) Male Clinical features Coughing preop Dyspnea preop Hemoptysis preop Asymptomatic Weight loss > Biochemistry Albumin (gl) LDH (UL) Lung function tests FEV, (% predicted) Vital capacity (% predicted) Diffusion capacity (%) Thoracotomy side Left Right Resection performed Pneumonectomy Lobectomy Sleeve lobectomy Lesser resection Histology Squamous cell caranoma Adenocarcinoma Giant cell carcinoma Other histology Tumor staging T1 T2 T3 64 (8) 8 72% 43% 32% 22% 21 % 33.4 (4.8) 382 (124) 82 (18) 94 (15) 72 (16) 32 33 42% 46% 9% 3% 64% 24% 3% 9% 34% 53% 13% 63 (10) 9 80% 53% 3 30% 2 34.2 (3.0) 325 (68) 73 (15) 93 (14) 78 (21) 13 7 40% 4 10% 6 2 32% 63% 64 (9) 94% 66% 41% 24% 34% 16% 34.2 (3.8) 348 (94) 76 (22) 92 (18) 74 (16) 33 37 30% 53% 10% 7% 67% 23% 7% 3% 44% 50% 6% a Where applicable, values are shown as mean (standard deviation). No significant differences were found using one-way analysis of variance or J tests. FEV, = forced expiratory volume in 1 second; LDH = lactate dehydrogenase; MND = mediastinal lymph node dissection; SS = systematic sampling. Both patients in the control group underwent a left pneumonectomy. The recurrent laryngeal nerve had to be cut to make an adequate dissection of the N2 nodes in the Table 2. Blood Loss Variable MND SS Control avalue Bleeding with repeat thoracotomy 1 1 2 NS (No.) Median operative blood loss (ml) 700 500 350 NS Mean blood transfusions within 574 394 584 NS admission period (ml) By Kruskal-Wallis test or one-way analysis of variance MND = mediastinal lymph node dissection; NS = not significant; SS = systematic sampling. Table 3. Fluid Production Via Drains Day MND ss Control p Valueb 0 450 625 350 0.001 1 400 440 237 0.006 2 150 262 125 NS 3 100 250 50 NS Total 1,050 1,460 688 <0.001 Values are shown as median in milliliters. MND = mediastinal lymph node dissection; SS = systematic sampling. By Kruskal-Wallis test. NS = not significant; aortopulmonary window possible. For the same reason, the nerve was cut in 3 patients in the MND group. In 3 other patients in this group, left pneumonectomy, left lobectomy, and right lower lobectomy were performed but the nerve was damaged unintentionally. In these patients the hoarseness must be regarded as a complication of the MND. CHYLOTHORAX. Chylothorax developed in 2 patients in the MND group who underwent left lower lobectomy and right upper lobectomy, respectively (see Table 4). Both patients were treated conservatively with an uneventful course. Other complications encountered OTHER COMPLICATIONS. were subcutaneous emphysema (2), air leakage longer than 2 weeks (5), pneumonia (5), atelectasis (4), septicemia (2), thrombosis (l), paresis of the peroneal nerve (l), and cardiac complication (4)(see Table 4). No wound infection or other disturbances in wound healing were found. All these complications were proportionally dis- Table 4. Morbidity and Mortality MND ss Control Complication (n = 65) (n = 20) (n = 70) Bronchopleural fistula Without empyema... 1 1 With empyema... 1... Recurrent nerve lesion Intentional 3... 2 Unintentional 3...... Chylothorax 2...... Subcutaneous 1... 1 emphysema Air leakage > 2 wk 2 1 2 Pneumonia 2 1 2 Atelectasis 1... 3 Septicemia 1 1... Thrombosis... 1... Paresis of peroneal nerve 1...... Wound infection......... Cardiac complication 2 1 1 Death 3 1 3 h4nd = mediastinal lymph node dissection; Ph3 SS = systematic sam-

964 BOLLEN ET AL Ann Thorac Surg 1993:55:%14 Table 5. N2 Positive Node Stations and Total Node Stations Sampled bv Station in Each Grouv Node 1 2R 2L 3A 38 4R 4L 5 6 7 8 9 Total Station Superior mediastinal Right upper paratracheal Left upper paratracheal Low pretracheal Low retrotracheal Right tracheobronchial Left tracheobronchial Subaortic Paraaortic Subcarinal Paraesophageal Pulmonary ligament MND (n = 65) ss (n = 20) Control (n = 70) a Data are shown as number of positive node stations, with number of node stations sampled or removed in parentheses. MND = mediastinal lymph node dissection; SS = systematic sampling. tributed over the MND, SS, and control groups. There was no significant difference among the three groups with respect to time to discharge from hospital. Mortality The mortality rate (within 30 days postoperatively) for all 210 patients was 5.2%. The mortality rate for pneumonectomy was 9.4%, for lobectomy 3.8%, and for thoracotomy without resection 7.6%. There were no deaths after sleeve resection, segrnentectomy, or wedge resection. Of the 155 patients studied, 7 patients died within 30 days postoperatively; 3 patients were in the MND group, 1 in the SS group, and 3 in the control group (see Table 4). No death could be related to MND. Accuracy of S tuging Three hundred fifty-one node stations were dissected in the MND group, 97 node stations were sampled in the SS group, and 73 stations were sampled in the control group. The distribution over the different node stations in each group is given in Table 5. In the whole population of 210 patients 40 were classi- fied as having stage N2 disease. Eight patient& operated on before 1988 and 4 after-were in the exclusion gr0up:j. Sixteen of the remaining 32 patients with N2 disease were found in the MND group, 7 in the SS group, and 9 in the control group (Table 6). The discovery ratio of N2 disease for MND against the controls was 1.9 (9 confidence interval, 0.9 to 4.0); for SS against controls, 2.7 (9 confidence interval, 1.2 to 6.3); and for MND plus SS against controls, 2.1 (9 confidence interval, 1.04 to 4.2). The last two ratios are significantly different from 1 (2 tests). Calculations on the total number of observed N2 nodes emphasize these findings (see Table 6). Comment Mediastinal lymph node dissection has been claimed to improve both the accuracy of staging [8] and the survival of patients with N2 non-small cell lung cancers [1-4, 141, although this view is not universally held [15]. In some studies an extended mediastinal lymph node dissection, through a median stemotomy [16, 171, has even been Table 6. N2 Disease and N2 Positive Node Stations in the Three Groups MND ss MND + SS Control Variable (n = 65) (n = 20) (n = 85) (n = 70) N2 disease Number of patients 16 7 23 9 Percent of patients 24.6 35.0 27.1 12.9 Discovery ratio 1.9 2.7 2.1 1 9 Confidence interval of discovery ratio 0.9-4.0 1.2-6.3 1.04-4.2... N2 positive node stations Total 27 12 39 12 Mean per N2 patient 1.7 1.7 1.7 1.3 Mean per patient 0.42 0.60 0.46 0.17 MND = mediastinal lymph node dissection; SS = systematic sampling.

Ann Thorac Surg 19!33:55:%1-6 BOLLEN ET AL 965 performed to remove lymph nodes from the contralateral side. Few data are available on the morbidity of the procedure or its contribution to the accuracy of staging. In this study, morbidity and accuracy of staging have been determined by systematically comparing three groups of patients who had plain resection, resection with sampling of the mediastinal lymph nodes, or resection with dissection of the nodes. Although the patients in the groups were not selected at random, we have four reasons for believing that comparing the groups is a methodologically valid procedure. First, the resectional strategy has not changed during the past 5 years. Second, all the operations were performed by two of us. Third, there were no changes during the study period except a more accurate preoperative staging by means of a routine mediastinoscopy and intraoperative exploration of the mediastinum. Finally, we selected those patients as controls who would have been eligible for SS or MND had they been seen in later years. The comparability of the three groups was substantiated by their matching in the most important clinical characteristics. The expected morbidity of MND includes extra blood loss [18], devascularization of the bronchus, and damage to some intrathoracic structures, especially the recurrent laryngeal nerve and the thoracic duct. In the present study, blood loss during the operation did not differ among the groups. The MND group had no higher rate of blood transfusion during or after the operation. However, the higher fluid production via the two drains indirectly indicates a somewhat higher perioperative blood loss. It is generally assumed that MND may lead to devascularization of the bronchus, increasing the risks of a fistula [19]. Because no fistulas were found in patients within the MND group this assumption is not confirmed. The lack of fistulas might be the result of the fact that the bronchial stump in the MND patients was covered with a vital pleural or pericardial flap. The most important complication associated with MND appears to be damage to the recurrent laryngeal nerve and the lymphatic vessels. In N2 disease, especially in left upper lobe cancer affecting the nodes in the aortopulmonary window, the recurrent laryngeal nerve may be cut intentionally to clear out this compartment adequately. However, if the N2 nodes in the aortopulmonary window have not been affected, recurrent nerve paralysis should be considered a serious side effect. Its occurrence, as in thyroid operations 1201, underlines the necessity of identifying the recurrent laryngeal nerve before continuing the dissection. Chylous fluid production is generally accepted as an adverse effect of lymph node dissection in cancer operations. In nonthoracic operations this complication is often treated conservatively [21], but in the case of chylothorax a more aggressive policy is followed [22]. Fortunately the 2 patients with chylothorax in this study, probably due to the MND, could be treated conservatively. To prevent this complication it might be important to ligate as much tissue as possible during the dissection. In inguinal lymphadenectomy fairly good results have been achieved by covering the operative field with a film of fibrin glue ~31. The overall mortality of 5.2% is within acceptable ranges [24] taking into account that in this region 40% of the male population has worked for more than 10 years in coal mines and that many patients consequently demonstrated extensive fibrotic changes with distortion of the lung and mediastinal structures. Most studies on the accuracy of staging have dealt with preoperative staging. In a recent study from the Bromptom Hospital [25], clinical TNM staging (determined by means of routine computed tomographic scans and selective mediastinoscopy) has been compared with pathological TNM or postsurgical staging. Mediastinal lymph node dissection has routinely been performed. The preoperative underestimation of 23. of the N stage in that study is in accordance with the 27.1% in our study (23 patients with N2 disease by SS or MND out of 85 with clinically less than N2 disease). The slightly higher percentage in the present study might be due to the fact that not all patients in the MND and SS groups underwent mediastinoscopy, and only a few patients underwent preoperative parasternal mediastinal exploration in case of a left upper lobe tumor. With respect to intraoperative accuracy of staging, the main issue in this context, it can be concluded that the discovery ratio suggests that twice as many patients with N2 disease are detected if a mediastinal lymph node exploration is performed. Perhaps the real difference in discovery ratio is even higher, as in our control group no routine mediastinoscopy was performed, whereas during the MND period a number of patients were excluded from resection because of positive nodes found during mediastinoscopy but not detected by the computed tomographic scan. Relatively more N2 diseases are identified by SS than by MND. Because the number of SS patients is very small, it is not possible to draw any conclusion in this respect. We even decided to change from SS to MND because MND seems to be more accurate than SS and as easy to perform. In addition, we expected from MND a therapeutic effect in N2 disease. In this study 351 node stations were totally removed in the MND group. This means an average of 5.85 stations per patient. In the SS group an average of 4.85 node stations per patient were sampled. Theoretically, on the right side with a complete mediastinal node dissection eight stations (1, ZR, 3A, 38, 4R, 7, 8, 9) can be removed and on the left side, 10 stations (1, 2L, 3A, 3B, 4L, 5, 6, 7, 8, 9). In our series, however, 3B has never been removed in left-sided tumors and only eight times in right-sided tumors. Besides, we only removed upper mediastinal node stations in left-sided tumors if there was any suspicion of abnormality. In some cases not all stations contained nodes or they were not adequately separated over the different stations. Even with these restrictions, the difference between number of stations actually sampled and those theoretically accessible is striking. This reflects the learning curve. Especially in the beginning of the study period sometimes we were not yet fully motivated

966 BOLLEN ET AL Ann Thorac Surg 1993;51-6 to remove node stations when they were already sampled by mediastinoscopy and revealed to be negative. It is our present practice to remove all stations in right-sided tumors except for stations 1 and 3B, which are seldom excised. For left-sided tumors, all stations are removed except for stations 1, 2L, 3A, and 3B. Mediastinal node dissection is now an obligatory part of resectional therapy. It is regarded as a safe procedure and provides a superior method for staging the mediastinal nodes. However, the fluid production from the drains suggests a higher blood loss in the MND group. Lesions of the recurrent laryngeal nerve are preventable. Supported in part by a grant from IKL, Comprehensive Cancer Centre Limburg, the Netherlands. We wish to express our appreciation to Dr Ben I. Davies for his valuable linguistic corrections and Dr Jan A. van Noord, Department of Respiratory Diseases, De Wever Hospital for critical reviewing of the manuscript. References 1. Martini N, Flehinger BJ, &man MB, Beattie EJ Jr. Results of resection in non-oat cell carcinoma of the lung with mediastinal lymph node metastases. Ann Surg 1983;198:3%-97. 2. Naruke T, Goya T, Tsuchiya R, Suemasu K. The importance of surgery to non-small cell carcinoma of lung with mediastinal lymph node metastasis. Ann Thorac Surg 1988;46: 603-10. 3. Watanabe Y, Shimizu J, Oda M, et al. Aggressive surgical intervention in N2 non-small cell cancer of the lung. Ann Thorac Surg 1991;51:25=1. 4. Patterson GA, Piazza D, Pearson FG, et al. Significance of metastatic disease in subaortic lymph nodes. Ann Thorac Surg 1987;43:155-9. 5. Pearson FG, De Larue NC, Ilves R, Todd TR, Cooper JD. Significance of positive superior mediastinal nodes identified at mediastinoscopy in patients with resectable cancer of the lung. J Thorac Cardiovasc Surg 1982;83:1-11. 6. Goldstraw P. The practice of cardiothoracic surgeons in the perioperative staging of non-small cell lung cancer. Thorax 1992;471-2. 7. Shields TW. The significance of ipsilateral lymph node metastasis (N2 disease) in non-small cell carcinoma of the lung. J Thorac Cardiovasc Surg 1990;99:&53. 8. Gaer JA, Goldstraw P. Intraoperative assessment of nodal staging at thoracotomy for carcinoma of the bronchus. Eur J Ci rdiothorac Surg 1990;4207-10. 9. Consensus report of the IASLC Working Party on Pretreatment Minimal Staging. Lung Cancer 1991;77-9. 10. Martini N, Flehinger BJ. The role of surgery in N2 lung cancer. Surg Clin North Am 1987;67103749. 11. Naruke T, Suemasu K, Ishikawa S. Lymph node mapping and curability at various levels of metastasis in resected lung cancer. J Thorac Cardiovasc Surg 1978;76832-9. 12. Mountain CF. A new international staging system for lung cancer. Chest 1986;892233S. 13. Moms JA, Gardner MJ. Calculating confidence intervals for relative risks (odds ratios) and standardised ratios and rates. Br Med J Clin Res 1988;2961313-6. 14. Cahan WG. Radical lobectomy. J Thoracic Cardiovasc Surg 10;39:555-72. 15. Yasumitsu T, In K, Nakagawa K, Kotake Y. Necessity of mediastinal lymph nodes dissection in operable lung cancer (a randomized study). Lung Cancer 1991;7(Suppl):288. 16. Naruke T, Goya T, Tsuchiya R, Suemasu K. Extended radical oxration for N2 left lung cancer through median stemotomy. Lung Cancer 1988;4(SuppI):A89. 17. hata E, Miyamoto H, Mitoma Y, Hayakawa K. Systematic bllateral mediastinal dissection and en bloc pulmonary resection through the median stemotomy. Lung Cancer 1991; 7:Suppl):269. 18. Vogt-Moykopf I, Branscheid D, Bulzebruck H, Probst G. Pktuelle Aspekte der neuen Stadieneinteilung beim Bronchialcarcinom und ihre kliruschen Konsequenzen. Der Chirurg 1989;60: 16-23, 19. Wilkens EW Jr. Bronchopleural fistula: prophylaxis. In: Grill0 HC, Eschapasse H, eds. International trends in general tnoracic surgery, vol2. Major challenges. Philadelphia: Sauncers, 19873967. 20. Mountain JC, Colcock BP. The recurrent laryngeal nerve in thyroid operations. Surg Gynecol Obstet 1971;133:978-80. 21. Femgni RG, Novicki DE. Chylous asates complicating genitourinary oncological surgery. J Urol 1985;134377&6. 22. Shirai T, Amano J, Takabe K. Thoracoscopic diagnosis and treatment of chylothorax after pneumonectomy. Ann Thorac Surg 1991;52:30&7. 23. llouchot 0, Bouchot-Hermouet FB, Karam G, Glemain P, Pannier M, Auvigne J. Prevention des complications de la lymphad6nectomie inguinale. J Urol (Paris) 1990;96:279-83. 24. Ginsberg RJ, Hill LD, Eagan RT, et al. Modem thirty-day operative mortality for surgical resections in lung cancer. J horac Cardiovasc Surg 1983;86:654-8. 25. emando HC, Goldstraw P. The accuracy of clinical evaluaive intrathoraac staging in lung cancer as assessed by postsurgical pathologic staging. Cancer 1990;65:2503-6.