D tion therapy, complete resection of a tumor offers

Determinants of Perioperative Morbidity and Mortality After Pneumonectomy Rakesh Wahi, MBBS, Marion J. McMurtrey, MD, Louis F. DeCaro, MD, Clifton F. Mountain, MD, Mohamed K. Ali, MD, Terry L. Smith, MS, and Jack A. Roth, MD Departments of Thoracic Surgery, Medical Specialties (Cardiopulmonary Section), and Biomathematics, The University of Texas, M.D. Anderson Cancer Center, Houston, Texas A total of 197 consecutive patients undergoing pneumonectomy at the M.D. Anderson Cancer Center from 1982 to 1987 were reviewed. Sixty-five variables were analyzed for the predictive value for perioperative risk. The operative mortality rate was 7% (14/197). Patients having a right pneumonectomy (n = 95) had a higher operative mortality rate (12%) than patients having a left pneumonectomy (l%, p < 0.05). The extent of resection correlated with the operative mortality rate (chest wall resection or extrapleural pneumonectomy, n = 39, 15%; versus simple or intrapericardial pneumonectomy, n = 158, 5%; p < 0.05). Patients whose predicted postopera- tive pulmonary function, by spirometry and xenon 133 regional pulmonary function studies, was a forced expiratory volume in 1 second > 1.5 L, forced expiratory volume in 1 second > 58% of the preoperative value, forced vital capacity > 2.5 L, or forced vital capacity > 0% of the preoperative value had a lower operative mortality rate (p < 0.05). Atrial arrhythmia was the most common postoperative complication (23%). Xenon 133 regional pulmonary function studies are useful in predicting the risks of pneumonectomy. ( ) espite recent progress in chemotherapy and radia- D tion therapy, complete resection of a tumor offers the best possibility for long-term survival in patients with lung cancer [l]. In some patients, a pneumonectomy with or without chest wall or pleural resection may be required to remove the entire tumor. Long-term survival after resection of pulmonary metastases has been well documented, and surgical resection may be a useful adjunct in the multimodality treatment of metastatic cancer [2]. In select cases, a pneumonectomy may be required for resection of the metastases. Extrapleural pneumonectomy may offer palliation and local control in some patients with pleural mesotheliomas [3]. It is well recognized that recent advances in anesthesia, intensive care, and invasive and noninvasive perioperative evaluation and monitoring have reduced the mortality rates for major pulmonary resections. However, pneumonectomy continues to be associated with the highest mortality rate for all types of pulmonary resection. In 1987, Krowka and associates [4] reported an 11% 30-day mortality rate for pneumonectomy. Weiss [l] reported a series of patients undergoing pulmonary resection for lung carcinoma with a mortality rate of 17% for pneumonectomy. An accurate current estimation of surgical risk is useful in weighing various therapeutic options. We, therefore, undertook a review of all patients undergoing pneu- Presented at the Thirty-fifth Annual Meeting of the Southern Thoracic Surgical Association, Marco Island, FL, Nov 10-12, 1988. Address reprint requests to Dr Roth, The University of Texas, M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Box 109, Houston, TX 77030. monectomy at M.D. Anderson Cancer Center from July 1982 to July 1987. The purpose of the study was to define the risk factors for patients undergoing pneumonectomy and to identify subgroups of patients at higher risk with preoperative criteria. Material and Methods Records of 197 patients undergoing pneumonectomy at M.D. Anderson Cancer Center were available for review. One hundred sixty-one patients underwent operation for primary carcinoma of the lung, 15 for mesothelioma, 15 for metastatic disease, 3 for carcinoid, 2 for lung sarcomas, and 1 for plasma cell granuloma. The operations were performed under the direct supervision of a faculty thoracic surgeon. Patients had a thorough preoperative evaluation to determine the tumor stage, as well as to estimate the risk of operation. All patients were evaluated by a senior staff surgeon for resectability. The patients were evaluated by the Section of Cardiopulmonary Medicine, where they underwent pulmonary function tests including regional pulmonary ventilation and perfusion using xenon 133 gas. Patients were not systematically excluded from operation based on the test results. For the xenon 133 test, the patients were studied in an upright position at rest during normal breathing. Eight scintillation detectors were positioned against the patient s back, with four over each lung. Regional pulmonary blood flow was determined by injection of 1 to 2 mci of xenon 133 dissolved in saline solution. Regional ventilation was determined from the distribution of a single tidal breath of xenon 133-air mixture and from the wash 0 1989 by The Society of Thoracic Surgeons 0003-4975/89/$3.50

34 WAHI ET AL in and wash out of the equilibration volume curve using the half time in seconds [5]. Using previously described formulas [, 71, we were able to predict lung functions after resection for these patients. Data regarding the patients cardiac status were retrieved, with special attention to any history of arrhythmias, coronary artery disease, hypertension, or cardiac failure. Using the criteria described by Goldman [8], a numerical value was assigned to each patient. Other preoperative variables analyzed in this study are as follows: General medical condition Diabetes Creatinine level Hemoglobin level Serum albumin level Tumor characteristics Type Preoperative stage Final stage Surgical factors Extent of resection Additional procedures Completion pneumonectomy Side of resection Time taken Stapler use Staff surgeon Transfusion requirements Postoperative variables Length of mechanical ventilation Length of intensive care unit stay Length of hospitalization Pulmonary edema Bronchopleural fistula Atelectasis Wound infection Other complications Patients dying within 30 days of operation or during the same hospitalization were considered operative deaths. Based on the clinical course of the patients, we were able to classify five deaths as resulting from cardiac causes, two deaths from pulmonary causes, six deaths from cardiopulmonary causes, and one death from technical causes. All postoperative and intraoperative complications, regardless of their impact on the hospital stay, were noted. These included arrhythmias, premature ventricular contractions, pulmonary edema, pneumonia, bronchopleural fistula, wound infections, intubation longer than 48 hours, atelectasis, and secretions requiring therapeutic bronchoscopy. The necessity for pneumonectomy was determined by the staff surgeon at the time of operation. All patients with primary lung cancer underwent complete lymph node mapping. The extent of resection was primarily determined by the tumor size and local extension. In this series, 92 patients (4%) had a simple pneumonectomy with lymph node dissection, whereas patients (34%) had an intrapericardial pneumonectomy. The other 39 patients (20%) required resection of the chest wall, pleura, or diaphragm. Eighteen patients (10%) had prior pulmonary resection and had a completion pneumonectomy. In addition, 39 patients had mediastinoscopy, 5 had an anterior mediastinotomy, 2 had a concomitant tracheostomy, 1 had a staging adrenalectomy, 1 had a concomitant thymectomy, and 2 were operated on via a midline sternotomy incision. The incidence of surgical complications including deaths was compared for patients according to the demographic and disease characteristics, pulmonary functions, and surgical variables. Differences were compared by,$ tests. All p values are two-tailed. Results The mean (t standard error of the mean) age of patients in this series was 55 t 0.8 yr. There were 137 men and 0 women. The mean postoperative hospital stay was 10 2 0.4 days; the mean intensive care unit stay was 3 2 0.2 days. Patients required a mean of 3 2 5 hours of mechanical ventilation. The mortality rate was 7%, and this was not significantly influenced by age, sex, or Goldman criteria (Table 1). Patients who had a right pneumonectomy had a higher operative mortality rate (12%) compared with patients who had a left pneumonectomy (1%)( p < 0.05). This difference was independent of other factors. Patients who had chest wall resection, extrapleural pneumonectomy, or resection of the diaphragm had a mortality rate of 15%, compared with 5% for patients who had a simple or intrapericardial pneumonectomy (p < 0.05). Transfusion requirements greater than 3 units increased the operative mortality rate from 4% to 17% ( p < 0.05). However, there was a statistically significant correlation between the extent of resection and transfusion requirements, indicating that transfusion was not an independent determinant of operative mortality. The mean total volume of fluids given during the perioperative period (24-hour period after induction of anesthesia) was 5.4 k 0.8 L. The total volume of fluid was not significantly increased in those patients who died after operation. The volume of fluids given perioperatively did not differ significantly for patients receiving a right compared with left pneumonectomy. Patients undergoing a completion pneumonectomy had a mortality rate of 11% compared with.7% for other patients. This difference was not statistically significant. Patients with a preoperative forced vital capacity (FVC) less than 85% of the preoperative predicted normal value had a mortality rate of ll%, which was significantly ( p < 0.05) higher than that of patients with an FVC greater than or equal to 85% of predicted (3%). We were unable to show any statistically significant differences in mortality based on preoperative forced expiratory volume in 1 second (FEV,), ratio of FEV,/FVC, absolute FVC, or forced expiratory flow (FEF). Using the regional ventilation perfusion studies and standard pulmonary function tests, we calculated the

~~ WAHIETAL 35 MORTALITY AFTER FNEUMONECTOMY Table 1. Differences in Operative Mortality for Various Criteria Criterion Age <70 years 270 years Sex Male Female Goldman s criteria 59 >9 Side Right Left Extent of resection Simple In trapericardial Extended Transfusions None 1-3 >3 Completion pneumonectomy a Determined by 2 analysis = not significant. No. of Deaths/ No. of Mortality P Patients (%) Value 121181 2/1 121137 2/0 11117 1/21 13/95 Ill02 4/92 4/ /39 3/93 4/4 7/40 18. 13 9 3 14 14 1 C0.05 4 15 C0.05 3 18 11 predicted postresection FEV, and FVC (Table 2). Patients with a predicted postoperative FEV, of 1.5 L or more had no perioperative deaths, compared with a perioperative mortality rate of 9% in patients with a lower predicted FEV, (p < 0.05). Patients with a predicted FEV, less than 41% of normal had a mortality rate of 1% compared with a mortality rate of 3% for patients with a predicted FEV, greater than or equal to 41% of normal. Patients with a predicted postoperative FVC greater than or equal to 2.13 L had a mortality rate of approximately 2.5% compared with 11% in patients with a lower predicted FVC (p < 0.05). Patients with a predicted postoperative FVC greater than or equal to 51% of normal had a mortality rate of approximately 1% compared with 12% in patients with a lower predicted FVC (p < 0.05). Additional procedures such as mediastinoscopy and mediastinotomy did not adversely affect mortality rate. Atrial arrhythmias were the most common complication, occurring in 4 patients (23%). Atrial arrhythmias were seen in 23% of patients undergoing simple pneumonectomy compared with 20% of patients requiring intrapericardial dissection (not significant). Patients who developed atrial arrhythmias had a longer stay in the intensive care unit (4.8 versus 2.5 days) and a longer postoperative hospital stay (13.7 versus 9.3 days) compared with patients who did not develop atrial arrhythmias (p < 0.05). Patients undergoing simple pneumonectomy had a 3% incidence of complications compared with 53% for intrapericardial resection. Other procedures such as chest wall resection, extrapleural pneumonectomy, and diaphragm resection increased the complication rate to 7%. The incidence of all nonfatal complications increased with the number of transfusions required. Patients who required no transfusion had a 39% incidence of complications, whereas patients requiring three or more transfusions had a 1% incidence of complications. Because of the correlation between extent of resection and transfusion, the transfusion requirement was not an independent predictor of operative complications. We were unable to show any significant effect of age, sex, preoperative lung function tests, or predicted lung function on the incidence of all postoperative complications. Cardiac failure developed in 18 patients (9%), showing no correlation with age or Goldman s criteria. Eighteen patients required ventilation longer than 48 hours. These patients had a mean preoperative FEV, of 9% of predicted normal compared with a mean preoperative FEV, of 79% of predicted normal for patients requiring ventilation for 48 hours or less (p < 0.05). No significant differences in other measures of preoperative spirometry could be demonstrated. However, when analyzing predicted postoperative lung function, significant differences were noted (Table 3). Patients requiring prolonged ventilation had a mean predicted FEV, of 43% of normal compared with 48% ( p < 0.05) and mean predicted FVC of 45% compared with 52% ( p < 0.05) for patients requiring mechanical ventilation for 48 hours or less. Other pulmonary complications, such as major atelectasis, pneumonia, and retained secretions, occurred in 13 patients (.%). Technical complications such as excessive blood loss (> units), loss of arterial or venous control, failure of the stapler to fire correctly, and inadvertent injury to unrelated structures occurred in 13 patients (.%). Eight patients (4%) developed a bronchopleural fistula. All bronchopleural fistulas developed on the right side (p < 0.05). Table 2. Predicted Postoperative Lung Function Variable No. of P Patients Deathsa Valueb FEV, (L) 21.5 47 0 (0) <1.5 12 12 (9) FVC (L) 22.13 78 (2.5) <2.13 95 10 (11) Percent FEV, Percent FVC 241% 251% 122 83 4 (3) 1 (1) <41% (51% 51 90 8 (1) 11 (12) c0.05 (0.05 a Numbers in parentheses are percentages. ysis. FEV, = forced expiratory volume in 1 second. capacity. Determined by,$ anal- FVC = forced vital

3 WAHIETAL 1989;48:3%7 Table 3. Lung Function in Patients Requiring Prolonged Ventilation Ventilation" >48Hours 548Hours p Variable (n = 18) (n = 179) Valueb Percent predicted 9% f 0.04% 79% -+ 0.01% normal FEV, (preoperative) Predicted 1.2? 0.04 1.49 f 0.07 FEVI (L) Percent predicted 43% f 0.02% 48%? 0.01% FEV, Predicted 1.93 f 0.12 2.08 2 0.05 FVC (L) Percent predicted 45% * 0.02% 52% 5 0.01% FVC a Data are reported as mean 2 standard deviation. Student's t test. FEV, = forced expiratory volume in 1 second; capacity; = not significant. Comment Determined by FVC = forced vital Two large series [4, 91 describing the risks of pneumonectomy have recently been published. One multicenter study [9] reported a mortality rate of approximately 7% for patients undergoing simple pneumonectomy for carcinoma of the lung. The other series [4] described patients with a variety of diseases that required a pneumonectomy. The mortality rate in that series was 13%. Our mortality rate of 7% is consistent with these and earlier reports [lo]. In a separate study, Reichel [lo] showed no significant increase in mortality related to age. Kohman and associates [ll] reported an increase in the operative mortality rate from 2.4% for patients younger than 0 years of age to 7.4% in patients older than 0 years. This study considered all types of lung resections. Ginsberg and co-workers [9] reported an increased mortality rate in patients older than 70 years. However, further analysis of their data for pneumonectomies alone shows a mortality rate of 5.9% for patients older than 70 years compared with 8% in patients younger than 70 years. Our results show no significant increase in mortality related to age. Reviewing patients older than 70 years of age, Breyer and co-workers [12] reported a mortality rate of 5% in patients with either lobectomy or pneumonectomy. The increased risk associated with right pneumonectomy is documented in various publications [lo, 13, 141. Gerson [15] reported a higher mortality rate in patients with a preoperative FEV, of less than 70% of the predicted normal value. Similar results were reported independently by Boushy and co-workers [1]. However, Keagy and associates [17] reported no significant effects of preoperative FEV, and FVC on mortality. In this study, ventilation perfusion scans were not performed and no attempt was made to estimate postresection lung function. In an experimental clinical study, Adams and colleagues [18] showed the importance of cardiopulmonary reserve as a determinant of risk. Since then, many investigators have reported the effects of tests of predicted postresection lung function on mortality. Earlier studies were done using a double-lumen Carlen tube placed under local anesthesia. Neuhaus and Cheriack [19], in such a study, showed good correlation between predicted lung function and complications. The introduction of ventilation perfusion studies using xenon 133 gas obviated the need for double-lumen tubes for such tests. Various investigators [, 71 have reported the technique and the reliability of these studies in calculating predicted lung function. These predictors are very useful clinically because regional and overall pulmonary functions remain stable after pneumonectomy [20]. In our series, we were able to show significant differences in mortality based on predicted lung function. The incidence of atrial arrhythmias after pneumonectomy varies from % to 29% in various series [21, 221. We report an incidence of 23%, but were unable to demonstrate any differences based on the extent of resection. This study shows that the operative mortality rate after pneumonectomy has not changed greatly over the last two decades. Most of the deaths in this series were from cardiac and cardiopulmonary causes. Using the preoperative xenon 133 regional ventilation and perfusion studies in conjunction with overall pulmonary function tests and demographic criteria, we are able to identify patients at higher risk. These patients should receive more intensive perioperative evaluation and hemodynamic monitoring to reduce their operative risk. Such efforts should primarily be directed at the cardiopulmonary system. Furthermore, the risk of operation must be carefully balanced with a better chance for survival in patients requiring extended resection. References 1. Weiss W. Operative mortality and five year survival rates in patients with bronchogenic carcinoma. Chest 1974;:483-7. 2. Mubn JF, Holmes EC, Vernon SE, et al. Pulmonary resection for metastatic sarcoma. Am J Surg 1980;140:9-1. 3. Davalle ML, Faber LP, Kittle CF, et al. Extrapleural pneumonectomy for diffuse malignant mesothelioma. Ann Thorac Surg 198;42:12-8. 4. Krowka MJ, Pairolero PC, Trastek VF, et al. Cardiac dysrythmias following pneumonectomy. Chest 1987;91:490-5. 5. Miller JM, Ali MK, Howe CD. Clinical determination of regional pulmonary function during normal breathing using xenon 133. Am Rev Respir Dis 1970;101:21%29.. Ali MK, Mountain CF, Ewer MS, et al. Predicting loss of pulmonary function after pulmonary resection for bronchogenic carcinoma. Chest 1980;77337-42. 7. Bria WF, Kanarek DJ, Kazemi M. Prediction of postoperative pulmonary function following thoracic operations. J Thorac Cardiovasc Surg 1983;8:18-92.

WAHIETAL 37 8. Goldman L. Complications of noncardiac surgery. Ann Surg 1983;198:78&91. 9. Ginsberg RJ, Hill LD, Eagan RT, et al. Modern thirty day mortality for surgical resection in lung cancer. J Thorac Cardiovasc Surg 1983;8:54-8. 10. Reichel J. Assessment of operative risk of pneumonectomy. Chest 1972;2:570-. 11. Kohman LJ, Meyer JA, Ikino PM, et al. Random versus predictable risk of mortality after thoracotomy for lung cancer. J Thorac Cardiovasc Surg 198;91:5514. 12. Breyer RH, Zippe C, Pharr W, et al. Thoracotomy in patients over age seventy years. J Thorac Cardiovasc Surg 1981; 81:187-93. 13. Evans EWT. Resection for bronchial carcinoma in the elderly. Thorax 1973;28:8-8. 14. Higgins GA, Beebe GW. Bronchogenic carcinoma. Factors in survival. Arch Surg 197;9453949. 15. Gerson G. Respiratory function tests and postoperative mortality. Br J Anaesth 199;41:97-71. 1. Boushy SF, Billings DM, North LB, et al. Clinical course related to preoperative and postoperative lung functions in patients with bronchogenic carcinoma. Chest 1971;59:383-91. 17. Keagy BA, Schorlemmer GR, Murray GF, et al. Correlation of preoperative pulmonary function testing with clinical course in patients after pneumonectomy. 1983; 3~253-7. 18. Adams WE, Perkins JF, Harrison RW, et al. The significance of cardiopulmonary reserve in the late results of pneumonectomy for carcinoma of the lung. Dis Chest 1957;32:280-8. 19. Neuhaus H, Cheriack. A bronchospirometric method of estimating the effect of pneumonectomy on the maximal breathing capacity. J Thorac Cardiovasc Surg 198;55:144-8. 20. Ali MK, Ewer MS, Atallah MR, et al. Regional and overall pulmonary function changes in lung cancer. J Thorac Cardiovasc Surg 1983;8:1-8. 21. Cerny CI. The prophylaxis of cardiac arrhythmias complicating pulmonary surgery. J Thorac Surg 1957;34:10510. 22. Shields TW, Ujiki GT. Digitalization for prevention of arrhythmias following pulmonary surgery. Surg Gynecol Obstet 198;12:743-.