Carcinoma of the esophagus continues to carry a

Pulmonary Complications After Esophagectomy Christopher E. Avendano, MD, Patrick A. Flume, MD, Gerard A. Silvestri, MD, Lydia B. King, MPH, and Carolyn E. Reed, MD Departments of Medicine, Biometry and Epidemiology, and Surgery, Medical University of South Carolina, Charleston, South Carolina Background. Pulmonary complications are common in patients who have undergone esophagectomy. There are no good predictive variables for these complications. In addition, the role that preoperative treatment with chemotherapy and radiation may play in postoperative complications remains unclear. Methods. We performed a retrospective review of all patients who underwent esophagectomy by a single surgeon at our institution over a 6-year period. Data were analyzed for a correlation between patient risk factors and pulmonary complications, including mortality, prolonged mechanical ventilation, and hospital length of stay. Results. Complete data were available on 61 patients. Nearly all patients had some pulmonary abnormality (eg, pleural effusion), although most of these were clinically insignificant. Pneumonia was the most common clinically important complication, and 19.7% of patients required prolonged ventilatory support. Significant risk factors identified included impaired pulmonary function, especially for patients with forced expiratory volume in 1 second (FEV 1 ) less than 65% of predicted, preoperative chemoradiotherapy, and age. Conclusions. Impaired lung function is a significant risk factor for pulmonary complications after esophagectomy. Patients with FEV 1 less than 65% of predicted appear to be at greatest risk. There also seems to be an associated risk of preoperative chemoradiotherapy for pulmonary complications after esophagectomy. (Ann Thorac Surg 2002;73:922 6) 2002 by The Society of Thoracic Surgeons Accepted for publication Nov 20, 2001. Address reprint requests to Dr Flume, Medical University of South Carolina, 812-CSB, 96 Jonathan Lucas St, Charleston, SC 29425; e-mail: flumepa@musc.edu. Carcinoma of the esophagus continues to carry a poor long-term prognosis. Worldwide, esophageal cancer is the sixth most common cause of cancer-related death. Curative resection remains the cornerstone of therapy [1]. Despite advances in anesthesia, operative techniques, and postoperative management, perioperative mortality rates continue to range from 3% to 10% [2, 3]. Postoperative pulmonary complications occur frequently, in approximately 30% of cases, and include pleural effusions, atelectasis, chylothorax, pneumonia, respiratory failure, and pulmonary embolism [4 17]. Furthermore, pulmonary complications have been associated with postoperative mortality [3, 8, 13, 14, 18]. Several authors have attempted to predict postoperative complications and mortality based on preoperative and operative factors [3, 4, 6 8, 10 13, 16, 19 23]. Despite many retrospective reviews, conclusions regarding which patients may be at undue risk for pulmonary complications are difficult to infer. This may be because of variations in surgical techniques and surgical skills among those performing resection in each of these studies. In addition, many patients are now receiving preoperative adjuvant therapy such as combined chemotherapy and radiotherapy (chemoradiotherapy) as a means of improving survival. The effect of preoperative chemoradiotherapy on postoperative pulmonary complications remains unclear. The purpose of this study was to quantify postoperative pulmonary complications and to evaluate the influences of various preoperative and perioperative factors on clinical outcomes in esophagectomy patients. Of particular interest was the effect of preoperative adjuvant therapy on clinical outcomes. Material and Methods We performed a retrospective review of patients who had an esophagectomy for esophageal cancer performed at one center by a single cardiothoracic surgeon (C.E.R.) from January 1994 to October 2000. Data were extracted from hospital and clinic records as well as from the clinical outcomes department and cancer registry of our institution. A total of 17 preoperative variables (see Appendix) were recorded and classified as patient, disease, or treatment factors. Seven postoperative pulmonary complications (see Appendix) including pneumonia, chylothorax, and adult respiratory distress syndrome were recorded for the duration of the postoperative hospital stay. Pneumonia was defined as febrile illness with the presence of new pulmonary infiltrates. Chylothorax was defined as a pleural effusion with documentation of chylomicrons on lipoprotein electrophoresis or a triglyceride level greater than 110 mg/dl. Patients were considered to have acute respiratory distress syndrome (ARDS) if they had a Po 2 :Fio 2 ratio less than 250 for more than 24 hours with pulmonary infiltrates, without clinical 2002 by The Society of Thoracic Surgeons 0003-4975/02/$22.00 Published by Elsevier Science Inc PII S0003-4975(01)03584-6 7

Ann Thorac Surg AVENDANO ET AL 2002;73:922 6 PULMONARY COMPLICATIONS AFTER ESOPHAGECTOMY 923 suspicion of volume overload. Four clinical outcomes (ie, ventilator hours, intensive care unit [ICU] and hospital lengths of stay, and mortality) were recorded. Mortality was defined as death occurring during hospitalization after esophagectomy, regardless of length of stay (LOS). Values are reported as the mean standard deviation. Analysis of variance was used to determine the effect of preoperative treatment and surgical techniques on outcome variables. Correlation coefficients for each treatment variable were calculated for the two main outcome variables, namely, ventilation hours and hospital length of stay. From this information we were able to determine which variables were most significantly correlated with outcome variables. Odds ratios and confidence intervals were calculated with logistic regression. Additional comparisons of outcomes based on categorical distribution of pulmonary function were made using nonparametric methods (Wilcoxon rank sum test). A p value of less than 0.05 was chosen as indicating a difference unlikely to be due to chance. All statistical analyses were performed with the Statistical Analysis Systems package, version 8 (SAS Inc, Cary, NC). Results Of the 96 patients identified for study, hospital charts documenting perioperative hospitalization were available for 81. Preoperative data were nearly complete for 61 patients and the analysis was confined to this dataset. Data regarding patient demographics as well as cancer type and stage are shown in Table 1. Three surgical approaches were used in this population; however, transhiatal esophagectomy was considered the preferred route for resection of esophageal cancer for patients who were believed to be at increased risk for pulmonary complications. Preoperative adjuvant therapy was given to patients with T3 or N1 disease with the decision of chemotherapy versus chemoradiotherapy made on an individual case basis. Although almost all patients developed some radiographic pulmonary abnormalities, significant pulmonary complications occurred in 22 (36.1%) of patients reviewed. In all, 53 patients (86.9%) had pleural effusion/ atelectasis, with 10 (16.4%) requiring placement of chest tubes, in addition to those placed at the time of surgery. Pneumonia was the most common clinically important pulmonary complication occurring in 20 (32.8%) cases. A total of 12 patients (19.7%) required mechanical ventilation for greater than 48 hours; all had been extubated within 48 hours of surgery but subsequently required reintubation. Six patients (9.8%) developed ARDS. Five patients (8.2%) developed a chylothorax. Seven patients (8.6% of total number of procedures; 11.5% of patients in analysis) died during their postoperative hospital stay. All of the patients who died postoperatively had developed pneumonia, and 5 of the 7 had progressed to ARDS; that is, 2 patients with ARDS survived to leave the hospital. Clinical outcomes included ventilator hours (mean 91.8 226.2 hours, median 12 hours), ICU days (mean Table 1. Characteristics of Study Population Number of subjects 61 Age, y 62.1 10.5 Age, y (range) 27 79 Male:female ratio 46:15 FVC (% predicted, n 49) 92.4 21.1 FEV 1 (% predicted, n 49) 90.5 23.7 FEV 1 80% predicted, n (%) 34 (69.4) FEV 1 65% 79% predicted, n (%) 7 (14.3) FEV 1 65% predicted, n (%) 8 (16.3) Pco 2 (mm Hg, n 39) 38.5 3.1 Po 2 (mm Hg, n 39) 86.2 10.1 Alveolar-arterial oxygen gradient (n 38) 16.2 9.3 Body mass index (n 61) 24.9 6.8 Serum albumin (mg/dl, n 45) 3.7 0.5 Tumor histology, n (%) Adenocarcinoma 37 (60.7%) Squamous cell 21 (35.0%) Other a 3 (4.3%) Pathologic stage (n 54) Stage 0, n (%) 12 (19.7%) Stage I 8 (13.1%) Stage IIA 9 (14.8%) Stage IIB 9 (14.8%) Stage III 16 (26.2%) Stage IV 7 (11.5%) Data are given as mean SD unless otherwise noted. a Two cases of high grade dysplasia and one unknown. FEV 1 forced expiratory volume in 1 second; FVC forced vital capacity; P CO2 partial pressure (tension) of carbon dioxide; P O2 partial pressure (tension) of oxygen. 6.1 11.1 days, median 2 days), and hospital days (mean 17.7 12.5 days, median 13 days). Those patients requiring prolonged mechanical ventilation ( 48 hours) were also the only patients with prolonged ICU length of stays ( 6 days). Thus, statistical analyses are reported to evaluate factors influencing duration of mechanical ventilation and hospital length of stay only. Patients who had been treated preoperatively with chemoradiotherapy (n 20; mean 187.3 314.2 hours, median 13.5 hours) had a longer duration (p 0.05) of mechanical ventilation than did patients who received preoperative chemotherapy (n 15; mean 10.7 hours, median 9 hours). There was no difference in hospital LOS among patients who received no preoperative treatment, those who underwent chemotherapy alone, and those had chemoradiotherapy. Preoperative factors that significantly correlated with duration of mechanical ventilation include preoperative chemoradiotherapy (r 0.30; p 0.02), forced vital capacity (FVC) (r 0.43; p 0.01), and forced expiratory volume in 1 second (FEV 1 )(r 0.34; p 0.02). Age (r 0.31; p 0.02), FVC (r 0.48; p 0.01), FEV 1 (r 0.38; p 0.01), and preoperative chemoradiotherapy (r 0.26; p 0.04) correlated with a prolonged hospital LOS. We compared outcomes (duration of mechanical ventilation and hospital LOS) between groups based on

924 AVENDANO ET AL Ann Thorac Surg PULMONARY COMPLICATIONS AFTER ESOPHAGECTOMY 2002;73:922 6 Table 2. Pulmonary Impairment and Outcomes FEV 1 (% pred) n Ventilator Hours LOS (Days) Mean SD Median Mean SD Median 80 34 66.1 155.6 15.5 16.3 7.6 13 65 79 7 48.7 93.8 8 13.9 4.4 12 65 8 356.9 469.7 a 31 33.8 25.9 b 22 a p 0.01. b p 0.001. FEV 1 forced expiratory volume in 1 second; LOS length of stay; SD standard deviation. severity of lung disease. We separated our patients into categories of normal (FEV 1 80% predicted), mildly impaired (FEV 1 65% to 79% predicted), and more severely impaired (FEV 1 65% predicted). Median and mean values of ventilator hours and LOS for each group are shown in Table 2. The patients with more severe impairment had longer duration of mechanical ventilation (p 0.01) and a longer hospital LOS (p 0.001). Those patients with an FEV 1 less than 65% predicted were at a significantly increased risk for requiring mechanical ventilation more than 48 hours (odds ratio 7.5; 95% confidence interval 1.24 to 45.4), even after controlling for age and type of preoperative treatment. Patients who were treated with chemoradiotherapy or who had an FEV 1 65% to 80% predicted were also at increased risk for prolonged mechanical ventilation and an increased risk for death, but these risks did not meet statistical significance. There were additional important findings. The patients who died were on the ventilator an average of 11 days, as compared with 3 days for the survivors (p 0.001), and were more likely to have had ARDS (p 0.0001). Those who died had an average age of 66.9 years, as compared with 60.9 years for survivors (p 0.01). Those who died also had lower albumin levels than did those who survived (3.2 mg/dl vs 3.8 mg/dl, p 0.01). Comment We found a high rate of pulmonary complications associated with esophageal resection for esophageal cancer. Atelectasis and pleural effusion were the most common pulmonary complication, occurring in 87% of patients, but were of little clinical importance. Pneumonia was the most common clinically important pulmonary complication, occurring in nearly one third of our study population, which is slightly higher than what has been previously reported (15% to 25%) [2, 9, 13, 16]. The overall mortality rate in our series was 8.6%, consistent with other series [3, 9, 16, 18]. Mortality rates as low as 1.2% have been reported [10]; however, factors such as patient selection and the willingness of the surgeon to undertake high-risk cases may influence these outcomes. We reviewed a number of preoperative clinical variables to determine whether they contributed to postoperative pulmonary complications as well as other clinical outcomes (notably, mortality, duration of mechanical ventilation, and length of stays in the ICU and the hospital). In general, measures of pulmonary function (FVC and FEV 1 as a percentage of predicted), preoperative chemoradiotherapy, and age were the only factors associated with prolonged mechanical ventilation or a prolonged hospital LOS. Pulmonary function has long been known to be associated with pulmonary complications after esophagectomy. Fan and colleagues [6] suggested that a preoperative peak expiratory flow rate (PEFR) of less than 65% predicted correlated well with the incidence of pulmonary complications. However, PEFR had less predictive value than other variables including age, albumin level, and arterial oxygen tension. Nagawa and coworkers [12] reported a significant difference in vital capacity between patients who developed pulmonary complications and those who did not. However, the mean vital capacity of patients who developed pulmonary complications was still 91.8% of predicted, which is normal. Our analysis demonstrates a negative correlation between pulmonary function (both FVC and FEV 1 ) and clinical outcomes; that is, worse lung function predicts longer duration of mechanical ventilation and hospital LOS. A specific degree of pulmonary impairment that would predict undue risk is not derived from our analysis; however, those patients with an FEV 1 less than 65% predicted had considerably and significantly longer durations of mechanical ventilation and a prolonged LOS. It should be noted that a bias is present in such an analysis, in that patients with very severe pulmonary impairment are unlikely to have been offered an operation because of presumed greater risk of complications, meaning that the population is generally selected for those with better lung function. The finding of an association between preoperative chemoradiotherapy and pulmonary complications (ie, duration of mechanical ventilation) could have implications with regard to adjuvant therapy for resectable esophageal cancer. The purpose of adjuvant chemoradiotherapy is to improve survival, although there are conflicting results as to whether it offers a survival advantage in patients with adenocarcinoma [2, 24 30]. Our study was not designed to determine a survival benefit of adjuvant chemoradiotherapy but to see whether such treatment causes increased morbidity; if it does, we will need to take this into consideration in determining its risk benefit ratio. Some authors have reported complications that may have been attributable to preoperative chemoradiotherapy [31], whereas others report that it was well tolerated [2, 24, 25]. Only Swisher and colleagues [25] used multivariate analysis to look at any correlation of preoperative adjuvant therapy and associated morbidity; however, no association was found. In contrast to previous studies, we found no significant

Ann Thorac Surg AVENDANO ET AL 2002;73:922 6 PULMONARY COMPLICATIONS AFTER ESOPHAGECTOMY 925 correlation between nutritional status and duration of mechanical ventilation or LOS. There was an association with albumin and mortality, although the numbers are small. The minimal influence of serum albumin on outcomes may be due to the relatively well-preserved nutritional status of our patients. Only 2 of our patients had serum albumin levels less than 3.0 mg/dl. Although our study may be limited by the small numbers of patients, there are certain features that separate it from other analyses of pulmonary complications after esophagectomy. Our series occurred over a relatively short period (6 years) so that there would be few effects due to changes in anesthesia or surgical approach. Although more than one surgical approach was used in our series, there was no significant difference in outcomes. Our series was performed by 1 surgeon, which should reduce the variability due to the operator. Only one other study may have had only 1 attending surgeon perform all esophagectomies and manage all patients postoperatively; these data were collected for more than 40 years [5]. Matthews and colleagues [32] found that there is a highly significant difference in operative mortality rate for surgeons based on the number of resections performed annually. By having a single surgeon perform all of the procedures in the study, we hoped to control for poorly quantifiable variables, which may influence outcome such as surgical experience and preferences of the individual surgeon. Our surgeon had performed 15 to 20 esophagectomies per year consistently for 10 years before the start of this series, so it is doubtful that a learning curve could be implied. In conclusion, we report a high frequency of pulmonary complications after esophagectomy. Although most were not clinically important, there was an associated mortality and morbidity. Impaired pulmonary function was associated with prolonged mechanical ventilation and prolonged hospital LOS, particularly in those patients with FEV 1 less than 65% predicted. Furthermore, there may be an association with preoperative chemoradiotherapy and increased morbidity, a finding that should be investigated further. References 1. Lee R, Miller J. Esophagectomy for cancer. Surg Clin North Am 1997;77:1169 96. 2. Naunheim K, Petruska P, Roy T, et al. Preoperative chemotherapy and radiotherapy for esophageal carcinoma. J Thorac Cardiovasc Surg 1992;103:887 995. 3. Ferguson M, Martin T, Reeder L, Olak J. Mortality after esophagectomy: risk factor analysis. World J Surg 1997;21: 599 604. 4. Ong G, Lam K, Wong J, Lim T. Factors influencing morbidity and mortality in esophageal carcinoma. J Thorac Cardiovasc Surg 1978;76:745 54. 5. Postlethwait R. Complications and deaths after operations for esophageal carcinoma. J Thorac Cardiovasc Surg 1983;85: 827 31. 6. Fan S, Lau W, Yip W, et al. Prediction of postoperative pulmonary complications in oesophagogastric cancer surgery. Br J Surg 1987;74:408 10. 7. Nishi M, Hiramatsu Y, Hioki K, Hatano T, Yamamoto M. Pulmonary complications after subtotal oesophagectomy. Br J Surg 1988;75:527 30. 8. Sugimachi K, Matsuoka H, Ohno S, Mori M, Kuwano H. Multivariate approach for assessing the prognosis of clinical oesophageal carcinoma. Br J Surg 1988;75:1115 8. 9. Lund O, Kimose H, Aagaard M, Hasenkam J, Erlandsen M. Risk stratification and long-term results after surgical treatment of carcinomas of the thoracic esophagus and cardia. J Thorac Cardiovasc Surg 1990;99:200 9. 10. Collard J, Otte J, Reynaert M, Michel L, Carlier M, Kestens P. Esophageal resection and by-pass: a 6 year experience with a low postoperative mortality. World J Surg 1991;15:635 41. 11. Tsutsui S, Moriguchi S, Morita M, et al. Multivariate analysis of postoperative complications after esophageal resection. Ann Thorac Surg 1992;53:1052 6. 12. Nagawa H, Kobori O, Muto T. Prediction of pulmonary complications after transthoracic oesophagectomy. Br J Surg 1994;81:860 2. 13. Law S, Fok M, Wong J. Risk analysis in resection of squamous cell carcinoma of the esophagus. World J Surg 1994;18: 339 46. 14. Zhang D, Cheng GY, Huang GJ, et al. Operable squamous esophageal cancer: current results from the east. World J Surg 1994;18:347 54. 15. Watson A. Operable esophageal cancer: current results from the west. World J Surg 1994;18:361 6. 16. Millikan K, Silverstein J, Hart V, et al. A 15-year review of esophagectomy for carcinoma of the esophagus, and cardia. Arch Surg 1995;130:617 24. 17. Kuwano H, Sumiyoshi K, Sonoda K, et al. Relationship between preoperative assessment of organ function and postoperative morbidity in patients with oesophageal cancer. Eur J Surg 1998;164:581 6. 18. Fok M, Law S, Wong J. Operable esophageal carcinoma: current results from Hong Kong. World J Surg 1994;18: 355 60. 19. Saito T, Kinoshita T, Shigemitsu Y, et al. Risk factors associated with postoperative mortality in patients with esophageal cancer. Int Surg 1993;78:93 8. 20. Nishi M, Hiramatsu Y, Hioki K, et al. Risk factors in relation to postoperative complication in patients undergoing esophagectomy or gastrectomy for cancer. Ann Surg 1988;207: 148 54. 21. Putnam J, Svell D, McMurtrey M. Comparison of three techniques of esophagectomy within a residency training program. Ann Thorac Surg 1994;57:319 25. 22. Moon M, Schulte W, Haasler G, Condon R. Transhiatal and transthoracic esophagectomy for adenocarcinoma of the esophagus. Arch Surg 1992;127:951 5. 23. Giuli R, Gignoux M. Treatment of carcinoma of the esophagus. Ann Surg 1980;192:44 52. 24. Terz J, Leong L, Lipsett J, Wagman L. Preoperative chemotherapy and radiotherapy for cancer of the esophagus. Surgery 1993;114:71 5. 25. Swisher S, Holmes E, Hunt K, Doty J, Zinner M, McFadden D. The role of neoadjuvant therapy in surgically resectable esophageal cancer. Arch Surg 1996;131:819 25. 26. Ilson D, Kelsen D. Combined modality in the treatment of esophageal cancer. Semin Oncol 1994;21:493 507. 27. Walsh T, Noonan N, Hollywood D. A randomized trial of multimodality therapy versus surgery for esophageal adenocarcinoma. N Engl J Med 1996;335:462 7. 28. Stewart J, Hoff S, Johnson D, et al. Improved survival with neoadjuvant therapy and resection for adenocarcinoma of the esophagus. Ann Surg 1993;218:571 6. 29. Urba S, Orringer M, Perez-Tamayo C, Bromberg J, Forastiere A. Concurrent preoperative chemotherapy and radiation therapy in localized esophageal carcinoma. Cancer 1992;69: 285 91. 30. Hoff S, Stewart J, Sawyers J, et al. Preliminary results with neoadjuvant therapy and resection for esophageal carcinoma. Ann Thorac Surg 1993;56:282 6. 31. Eguchi R, Ide H, Nakamura T, et al. Morbidity and mortality after esophagectomy with neoadjuvant therapy. Jpn J Thorac Cardiovasc Surg 1999;47:552 8. 32. Matthews H, Powell D, McConkey C. Effect of surgical experience on the results of resection for oesophageal carcinoma. Br J Surg 1986;73:621 3.

926 AVENDANO ET AL Ann Thorac Surg PULMONARY COMPLICATIONS AFTER ESOPHAGECTOMY 2002;73:922 6 Appendix Variables Collected and Used in Analysis Preoperative/Intraoperative Variables Age Comorbidities Forced vital capacity (FVC) as a percentage of predicted Forced expiratory volume in 1 second (FEV 1 ) as a percent of predicted Arterial carbon dioxide tension (Pco 2 ) Arterial oxygen tension (Po 2 ) Alveolar-arterial oxygen gradient Body mass index Preoperative albumin level Tumor histology Clinical tumor stage Preoperative (adjuvant) treatment (chemotherapy, chemoradiotherapy, or no treatment) Surgical technique (transhiatal, three-field cervical esophagostomy, or abdominal laparotomy/right thoracotomy [Ivor-Lewis approach]) Total anesthesia time Perioperative fluid repletion including both crystalloid and colloid replacement Estimated blood loss Blood transfusion Postoperative Pulmonary Complications Pneumonia Atelectasis/effusions Pleural effusion requiring placement of additional chest tubes Chylothorax Adult respiratory distress syndrome Respiratory failure requiring mechanical ventilation Pulmonary embolism Clinical Outcomes Ventilator hours Intensive care unit length of stay Hospital length of stay Mortality