Thoracoscopic and laparoscopic esophagectomy: initial experience and outcomes

Surg Endosc (2005) 19: 1597–1601 DOI: 10.1007/s00464-005-0185-7 Ó Springer Science+Business Media, Inc. 2005

Thoracoscopic and laparoscopic esophagectomy Initial experience and outcomes D. J. Martin, J. R. Bessell, A. Chew, D. I. Watson Flinders University Department of Surgery, Flinders Medical Centre, Bedford Park, South Australia 5042, Australia Received: 1 April 2005/Accepted: 7 July 2005/Online publication: 17 October 2005

Abstract Background: Although surgical resection currently is the preferred treatment for fit patients with resectable esophageal cancers, it is associated with a relatively high risk of morbidity and significant perioperative mortality. Currently, a range of open surgical approaches are used. More recently, minimally invasive approaches have become feasible, with the potential to reduce perioperative morbidity. This study investigated the outcomes from one such approach. Methods: Outcome data were collected prospectively for 36 consecutive patients who underwent a minimally invasive esophagectomy for esophageal cancer. A threestage approach was used, with all the patients undergoing a thoracoscopic esophageal mobilization, combined with either open or hand-assisted laparoscopic abdominal gastric mobilization, and open cervical anastomosis. An open abdominal approach was used for 15 of the patients and a hand-assisted laparoscopic approach for 21. A total of 34 patients had invasive malignancy, whereas 2 had preinvasive disease. A group of 23 patients (68%) who had invasive malignancies also received neoadjuvant chemotherapy and radiotherapy. Results: The mean operating time ranged from 190 to 360 min (mean, 263 min). The median postoperative hospital stay was 16 days. In-hospital mortality was 5.5% (2/36), and perioperative morbidity was 41%. The perioperative outcomes for patients undergoing an open abdominal approach and those who had hand-assisted laparoscopic surgery were similar. For the patients who underwent a hand-assisted laparoscopic abdominal procedure, the total operating time was shorter (248 vs 281 min), and the blood loss was less (223 vs 440 ml). The median follow-up period was 30 months. The 4-year survival predicted by Kaplan–Meir for the 34 patients with invasive malignancy was 44%. Conclusion: The outcome for esophagectomy using thoracoscopic esophageal mobilization, with or without

Correspondence to: D. I. Watson

hand-assisted laparoscopic abdominal surgery, was comparable with data from conventional open surgical approaches. These approaches can be performed with an acceptable level of perioperative morbidity. Further application of these techniques, with close scrutiny of outcome data, is appropriate. Key words: Esophageal cancer — Esophagectomy — Thoracoscopic esophaeal mobilization

Since Czerny first successfully resected a cancer of the cervical esophagus in 1877, esophagectomy has had a long history of high morbidity and mortality, followed by a relatively poor long-term survival. Published perioperative mortality rates are available since the 1940s, and the initial reported rate was 72% [6]. By the 1970s, a review of all published data showed a reduction in the rate to 29% [6]. In the 1980s it was 13%, and in the 1990s, it declined to 9% [6]. Selected centers now publish 30-day mortality rates of approximately 2% to 5% [4, 5, 7, 8]. However, the published rates do not always tell the full story. The 30-day mortality for esophagectomy recorded in U.S. Medicare data from 1994 to 1999 actually was 23% [2]. Nevertheless, incremental improvements have occurred over time, and these can be explained by a range of factors including improved perioperative care, better patient selection, and improved surgical technique and experience. Surgeons have used a myriad of approaches to perform esophagectomy. After Czerny, these were pioneered by the likes of Torek (1913), who performed the first successful transthoracic resection, Ohsawa (1933), who first did reconstruction with the stomach, and Tanner (1947), who introduced the evergreen two-stage Ivor–Lewis procedure. Different surgical units use a diverse selection of these procedures, and any of the cervical, transthoracic, or abdominal approaches may be selected, either in combination or alone.

1598

Minimally invasive techniques, including thoracoscopic esophageal mobilization, and laparoscopic techniques that replicate all the described open surgical approaches also have been described [8, 11, 13, 15, 18, 19]. These have the potential to reduce perioperative and short-term morbidity for patients undergoing esophageal resection. In our practice, we have selectively applied either a thoracoscopic approach or a thoracoscopic approach combined with hand-assisted laparoscopic abdominal surgery [19]. These techniques apply a hybrid of minimally invasive and open techniques in a three-stage procedure. In this study, we determined the outcome for patients who had undergone one of these approaches to assess the perioperative outcome and median-term survival rate for this group of patients.

Patients and methods The study recruited all patients between January 1998 and January 2004 with resectable esophageal malignancy who underwent a cervicothoracoabdominal (three-stage) surgical resection with thoracoscopic esophageal mobilization and anastomosis in the neck. Patients with tumors that extended onto the gastric side of the esophagogastric junction underwent resection using abdominal and thoracic access, with anastomosis in the upper chest, and were excluded from this study. All demographic, perioperative, and postoperative outcome data were collected prospectively and managed using a computerized data base (FileMaker Pro Version 5.5; FileMaker Inc., Santa Clara, CA, USA). All the patients underwent preoperative staging with computed tomography of the chest and abdomen, and operative fitness was determined by clinical assessment, pulmonary function testing, and echocardiography. Endoscopic ultrasound was used to determine Tand N-stage after it became available in 2001. Laparoscopic staging was used preoperatively for tumors located in the lower third of the esophagus. Patients with more advanced tumors often underwent preoperative neoadjuvant treatment, whereas patients with early-stage lesions progressed directly to esophagectomy. Neoadjuvant treatment entailed two courses of 5-flurouracil and cisplatin chemotherapy combined with 45 Gy of radiotherapy, followed by surgery 4 to 8 weeks after the completion of treatment. The technique for minimal access esophagectomy has been described in detail elsewhere [19]. All patients initially underwent thoracoscopic esophageal mobilization. The patients were positioned fully prone after placement of a double-lumen endotracheal tube. The esophagus was exposed and mobilized via the right chest cavity. Three ports usually were placed, respectively, in the fifth, seventh, and ninth intercostal spaces posterolaterally at approximately the level of the posterior axillary line. An 11-mm port was used for a 10-mm laparoscope, a 5-mm port for the surgeonÕs right-hand instruments, and a 12mm port for the left-hand instruments and an endoscopic stapler. A diathermy hook was used to dissect the esophagus from the thoracic inlet to the diaphragmatic hiatus, and an endoscopic stapler was used to divide the azygos vein. Patients then were repositioned in the supine position, with legs extended in stirrups to allow the surgeon to stand between the patientÕs legs for the abdominal component of the procedure. This was performed concurrently with open cervical esophageal dissection and anastomosis. In some patients, abdominal mobilization of the stomach was performed using a conventional open surgical approach via an upper midline abdominal incision. In others, hand-assisted laparoscopic abdominal gastric mobilization was used. The type of abdominal approach was determined by surgeon preference, with the one surgeon (D.W.) applying the laparoscopic approach routinely and the other (J.B.) using an open approach in the early part of the experience and later adopting the laparoscopic approach. The hand-assisted laparoscopic approach entailed an 8-cm right transverse subcostal incision for the hand port, an 11-mm umbilical port for the laparoscope, and a 12-mm port in the left upper quadrant for dissecting and stapling instruments. A Nathanson liver retractor

(Cook Medical Technology, Eight Mile Plains, Queensland, Australia) was placed through a 5-mm epigastric incision to elevate the left lobe of the liver. Varied types of hand ports were used in this series, and all were found to be satisfactory. During both the laparoscopic and open surgical approaches, the stomach was mobilized using ultrasonic shears (Johnson and Johnson Medical, North Ryde, Sydney, Australia). The right gastric and right gastroepiploic arteries and their arcades were preserved. A vascular stapler was used to divide the left gastric vessels close to their origins, and the esophagus was divided in the neck after mobilization through a separate cervical incision. If the laparoscopic approach was used, the thoracic esophagus and stomach were delivered through the incision used for the hand port, and a gastric tube was created under direct vision using a linear stapling device. A pyloromyotomy also was performed under direct vision. If an open abdominal approach was used, the same operative techniques were used to fashion the gastric tube. After the hiatus had been widened by division of the left hiatal pillar with the ultrasonic shears, the gastric tube was passed up to the neck and anastomosed to the proximal cervical esophagus in the neck using a conventional hand-sewn technique. A feeding jejunostomy was inserted into the proximal jejunum, and the wounds were closed. In both the chest and abdomen, we performed a resection that included the lymph nodes immediately adjacent to the esophagus and lymph nodes around the base of the left gastric artery in the abdomen. Nodes below the tracheal bifurcation also were removed as a separate package of tissue. This replicates our usual dissection during open surgery. We did not attempt to perform a radical en bloc procedure. Information about patient demography, tumor stage, operative details, postoperative complications, and follow-up outcomes were collected. Data were analyzed using SPSS (version 12.0, Chicago, IL, USA). The Mann–Whitney test was used to compare nonparametric data sets, and the unpaired t-test was used to compare sets of parametric data. FisherÕs exact test was used to compare proportions, and the Kaplan–Meier and log-rank tests were used for survival analysis.

Results Altogether, 36 patients underwent an esophagectomy with thoracoscopic esophageal dissection. Of these, 21 underwent a hand-assisted laparoscopic abdominal approach, and 15 underwent a conventional open abdominal component. The two subgroups were well matched for age, sex, tumor stage, and histologic stage. Over the same period, the same surgeons performed an esophagectomy on an additional 65 patients using a conventional open surgical approach. The median age for the 36 patients in this study was 64 years (range, 35–79 years). The majority (72%) were men. Eleven (31%) had significant preexisting comorbidities. A tumor was located in the lower third of the esophagus in 25 patients (69%), the middle third in 10 patients (28%), and the upper third in 1 patient. A total of 34 patients (94%) had an invasive malignancy (stage T2 or greater). These malignancies included 25 adenocarcinomas (74%) and 9 squamous cell carcinomas (26%). The treatment for 23 patients was neoadjuvant chemotherapy and radiotherapy. According to the American Joint Commission on Cancer staging classification system, 10 tumors were stage II, 22 were stage III, and 2 were stage IV. For the patients who underwent neoadjuvant treatment, the tumor stage was determined to be whichever stage was greater at pretreatment endoscopic ultrasound or histopathology of the resected tumor. One patient had in situ squamous cell carcinoma, and one had severe dysplasia in BarrettÕs esophagus.

1599

The operating time ranged from 190 to 360 min (median, 240 min), and blood loss ranged up to 1,500 ml (median, 200 ml). Two procedures were not completed by a minimal access approach, and the surgery was converted to an open operation. In the one patient, dense pleural adhesions were present, preventing thoracoscopy, and the procedure was converted to open thoracotomy. However, the dense adhesions between the right lung and the chest wall made mobilization of the lung very difficult, and safe transthoracic access could not be achieved. The procedure was converted eventually to an open transhiatal procedure with blunt mobilization of the midesophagus via the abdomen. In the second patient, the hand-assisted laparoscopic approach was converted to an open surgical approach by transverse extension of the hand port incision because the ultrasonic shears failed to achieve adequate hemostasis when dividing omental vessels on multiple occasions. At open surgery, it was apparent that the veins draining the stomach were up to 7 mm in diameter, possibly because of mild portal hypertension. A splenectomy also was required in this patient to control bleeding. There were no complications or difficulties associated with the use of the prone position for the thoracoscopic phase of the procedures. A pyloromyotomy was performed in 14 of the 21 procedures undertaken laparoscopically. None of these procedures were converted to a pyloroplasty. A drainage procedure was not performed in seven patients because of difficult direct access to the pylorus through the transverse incision in the right upper abdomen. There were no adverse outcomes associated with this. The postoperative hospital stay ranged from 8 to 61 days (median, 16 days), and the median duration of postoperative stay in the intensive care unit was 1 day. There were two postoperative hospital deaths (5.5%). One patient died on postoperative day 18 after a vomiting and aspiration event that had occurred 9 days earlier. The other patient died on postoperative day 30 of bilateral pneumonia and adult respiratory distress syndrome. Overall, 15 patients (41%), including the 2 patients who died, had a perioperative complication. There were seven anastomotic leaks (19%). Two of these were asymptomatic and observed only on a routine radiologic contrast x-ray. They did not require specific treatment. Of the five clinically significant leaks (14%), two drained spontaneously via the neck wound, one required operative drainage by reopening of the neck wound, and two required open transthoracic drainage, respectively, on postoperative days 4 and 10. In the patient who underwent transthoracic drainage on day 4, the anastomosis also was reconstructed. The patient who had drainage placed via the neck incision also underwent abdominal reexploration and washout for leakage around the jejunostomy insertion site. One patient had an intercostal drain inserted directly into the gastric tube in the intensive care unit, and this resulted in a controlled gastric ‘‘fistula,’’ which closed spontaneously 1 month after esophagectomy. A further two patients re-

quired perioperative reoperation: one for evacuation of a mediastinal hematoma and one involving debridement of two areas of ischemic skin from the abdominal wall, which complicated postoperative bacterial endocarditis. There was a further case of postoperative chylothorax, which resolved spontaneously on day 11 after conservative management. Late complications included an incarcerated diaphragmatic hernia with herniation of bowel into the left chest 2 years after the original thoracoscopic and laparoscopic esophagectomy. The details of this problem have been described elsewhere [1]. In another patient, a port-site metastasis developed at the site of the previous umbilical incision just before the patient died from extensive systemic metastatic disease. This patient had a positive distal margin and extensive celiac lymphadenopathy at the original resection, and the operation was palliative in intent. Pathologic examination of the resected specimens showed a positive distal margin in one patient who had a large stage IV tumor that involved the gastroesophageal junction. Another patient had a positive proximal margin with a proximal squamous cell carcinoma that involved the cervical esophagus. For this patient, a more proximal resection was not feasible. Three further patients, none of whom had undergone neoadjuvant treatment, had tumor extending to the radial margins. The perioperative outcomes for the 21 patients who underwent a handed-assisted laparoscopic abdominal approach were compared with the outcomes of 15 patients who underwent surgery via an open abdominal approach. The mean operative time was 248 ± 44.33 min for those who had a laparoscopic approach, as compared with 281 ± 61.28 min for those who had an open abdominal approach (p = 0.071). For the abdominal part of the procedure alone, the times were 92.8 min for the laparoscopic approach and 125.6 min for the open procedures (p = 0.084). Blood loss was less for the patients who underwent hand-assisted laparoscopic surgery (mean, 223 vs 440 ml; median, 150 vs 500 ml; p = 0.05). There was no difference in postoperative hospital stay. The median stay was 15 days for the laparoscopic approach and 16 days for the open abdominal approach. There also was no difference in perioperative morbidity (odds ratio, 1.65 for laparoscopic vs open; p = 0.46). The follow-up periods ranged from 6 months to 6 years (median, 30.3 months). The patients had not been followed long enough to allow determination of the median survival (i.e., fewer than 50% had died during the follow-up period). Figure 1 shows the Kaplan–Meir survival rate for the 34 study patients who underwent surgery for invasive malignancy. In this analysis, the predicted 1-year survival was 72%, and the predicted 4year survival was 44%. There was no difference in survival rates between the patients who underwent neoadjuvant treatment and those who did not, between the patients who underwent a laparoscopic procedure and those who had an open abdominal procedure, nor between those who had a postoperative complication and those who did not.

1600

Fig. 1. Kaplan–Meir survival rate for the patients undergoing surgery for invasive malignancy.

Discussion Laparoscopic approaches have now been applied to the full range of abdominal and thoracic surgery, and they have the potential to improve short-term outcomes by reducing wound-related morbidity. Their application to cancer surgery has been more controversial, with early reports of port-site metastases after laparoscopic colectomy highlighting the potential for adverse outcomes [10]. Hence, the potential advantages of laparoscopic approaches should be carefully considered against any disadvantages. Currently, the role of minimal access approaches for the treatment of esophageal cancer is undefined. There is, however, no doubt that laparoscopic and thoracoscopic approaches are feasible, and some surgeons have reported experience with laparoscopic and thoracoscopic approaches to esophagectomy [3, 9, 19]. Although no randomized studies have compared minimally invasive and open esophageal resection, several units have reported results at least comparable with those obtained by open surgery. A paper from the University of Pittsburgh reported the outcomes for 222 patients who underwent a three-stage thoracoscopic and laparoscopic approach with an open cervical anastomosis [8]. For these patients, the median postoperative hospital stay was 7 days, the 30 day mortality was 1.4%, and survival was comparable with open surgical data [8]. Smithers et al. [15] reported their experience with 162 patients who underwent thoracoscopic esophageal mobilization in the prone position combined with an open abdominal approach. In this series, the 30-day mortality rate was 3.3%, and the median survival time was 29 months. This outcome was similar to the same

groupÕs experience with an open Ivor–Lewis technique [15]. The prone position for thoracoscopic mobilization, although not favored in the United States, Hong Kong, Japan, or continental Europe, in which the left lateral decubitus position is more popular, has been widely used in Australia [15, 19]. With this position, the deflated lung lies forward out of the operating field and, in virtually all cases, requires no extra port for a lung retractor, as is often required when the lateral position is used. The use of a hand-assisted approach for the laparoscopic component of esophagectomy has been described in a previous paper by one of the authors [19]. The current report describes more extensive experience with this technique. The use of a hand-assisted approach offers several advantages over both the totally laparoscopic and the open approaches to the abdominal component of esophagectomy. In our experience, the use of a hand port actually sped up the operation, reducing the time required for the abdominal component of esophagectomy by approximately 30%. In addition, overall blood loss was reduced by 50%, as compared with that of the open approach. The handassisted laparoscopic approach also has the potential advantage of less pain associated with the smaller transverse subcostal incision than experienced with a midline laparotomy wound. Hence, return to full activity might be quicker, although this has not been proved by our data. The hand-assisted laparoscopic approach also is a technique that is easily applied by upper gastrointestinal surgeons familiar with laparoscopic surgery at the gastroesophageal junction. As compared with a totally laparoscopic technique, the hand-assisted approach also allows the more difficult and time-consuming parts of a complete laparoscopic approach to be performed on the abdominal surface, namely, the pyloromyotomy and fashioning of the gastric conduit. For this reason, the median operating time of 4 h for the hand-assisted laparoscopic approach in our series compares favorably with the mean total operative time of 7.5 h reported for the first 77 patients undergoing a totally laparoscopic approach at the University of Pittsburgh [9], and 6 h reported in another series of 46 patients [12]. Esophageal resection is not without substantial risks, and there is evidence that both hospital and operator experience correlate with outcome. There also is likely to be a learning curve associated with the application of minimally invasive esophageal resection [17]. For this reason, it is important to compare critically the outcome of patients undergoing the new procedure with established outcomes from open surgery in large-volume centers. A large study from the Memorial Sloane Kettering Cancer Centre reported the outcome of esophageal resection for 510 patients [14]. As reported, 27% of their procedures were associated with a technical complication, and anastomotic leak occurred after 23% of the procedures. Our outcomes compare favorably with this experience. Furthermore, the clinical leak rate of 14% in our series compares well with the published literature, in

1601

which leak rates vary between 6% and 50% [16]. A metaanalysis suggests that although cervical anastomoses (as in our series) have a higher rate of leakage than intrathoracic anastomoses, cervical leaks are less likely to be associated with mortality [16]. Nevertheless, it is important to remember that the morbidity of any esophageal leak is significant. The survival rate for the patients in our current study is promising, and the Kaplan–Meir survival rate is very acceptable. Nevertheless, it should be remembered that there could be selection bias in our series because there was a tendency to exclude patients with large bulky tumors extending below the gastroesophageal junction from the minimal access approach. A report of South Australian experience from 1985 to 2003 with predominantly open esophageal resection techniques described an actual 1-year survival rate of 80% and a 5-year survival rate of 20% for patients undergoing esophagectomy for adenocarcinoma [20]. The survival outcome for the current series compares favorably with these outcomes. Both the short- and long-term outcomes after the minimal access technique described in this article are at least comparable with those of open surgery for esophageal malignancy. The hand-assisted laparoscopic method is easily applied by upper gastrointestinal surgeons who are already familiar with open esophageal surgery and upper abdominal laparoscopy, and few intraoperative technical difficulties were encountered in our patients. Nevertheless, further studies are needed before techniques such as this should be recommended for routine use.

References 1. Aly A, Watson DI (2004) Diaphragmatic hernia after minimally invasive esophagectomy. Dis Esoph 17: 183–186 2. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, Welch HG, Wennberg DE (2002) Hospital volume and surgical mortality in the United States. N Engl J Med 346: 1128– 1137 3. Bonavina L, Incarbone R, Bona D, Peracchia A (2004) Esophagectomy via laparoscopy and transmediastinal endodissection. J Laparendosc Adv Surg Tech A 14: 13–16

4. Feith M, Stein HJ, Siewert JR (2003) Pattern of lymphatic spread of BarrettÕs cancer. World J Surg 27: 1052–1057 5. Hulscher JB, van Sandick JW, de Ber AG, Wijnhoven BP, Tijssen JG, Fockens P, Stalmeier PF, ten Kate FJ, van Dekken H, Obertop H, Tilanus HW, van Lanschot JJ (2002) Extended transthoracic resection compared with limited transhiatal resection for adenocarcinoma of the esophagus. N Engl J Med 347: 1662– 1669 6. Jamieson GG, Mathew G, Ludemann R, Wayman J, Myers JC, Devitt PG (2004) Postoperative mortality following esophagectomy and problems in reporting its rate. Br J Surg 91: 943–947 7. Law S, Wong J (2004) Therapeutic options for esophageal cancer. J Gastrenterol Hepatol 19: 4–12 8. Luketich JD, Alvelo-Rivera M, Buenaventura PO, Christie NA, McCaughan JS, Litle VR, Schauer PR, Close JM, Fernando HC (2003) Minimally invasive esophagectomy: outcomes in 222 patients. Ann Surg 238: 486–494 9. Luketich JD, Schauer PR, Christie NA, Weigel TL, Raja S, Fernando HC, Keenan RJ, Nguyen NT (2000) Minimally invasive esophagectomy. Ann Thorac Surg 70: 906–911 10. Neuhaus SJ, Texler M, Hewett PJ, Watson DI (1998) Port-site metastases following laparoscopic surgery. Br J Surg 85: 735–741 11. Nguyen NT, Follette-Wolfe BM (2000) Comparison of minimally invasive esophagectomy with transthoracic and transhiatal esophagectomy. Arch Surg 135: 920–925 12. Nguyen NT, Roberts P, Follette DM, Rivers R, Wolfe BM (2003) Thoracoscopic and laparoscopic esophagectomy for benign and malignant disease: lessons learned from 46 consecutive procedures. J Am Coll Surg 197: 902–913 13. Osugi H, Takemura M, Higashino M, Takada N, Lee S, Kinoshita H (2003) A comparison of video-assisted thoracoscopic esophagectomy and radical lymph node dissection for squamous cell cancer of the esophagus with open operation. Br J Surg 90: 108– 113 14. Rizk NP, Bach PB, Schrag D, Bains MS, Turnbull AD, Karpeh M, Brennan MF, Rusch VW (2004) The impact of complications on outcomes after resection for esophageal and gastroesophageal junction carcinoma. J Am Coll Surg 198: 42–50 15. Smithers BM, Gotley DC, McEwan D, Martin I, Bessell J, Doyle L (2001) Thoracoscopic mobilization of the esophagus: a 6-year experience. Surg Endosc 15: 176–182 16. Urschel JD (1995) Esophagogastrostomy anastomotic leaks complicating esophagectomy: a review. Am J Surg 169: 634–640 17. Watson DI, Baigrie RJ, Jamieson GG (1996) A learning curve for laparoscopic fundoplication: definable, avoidable, or just a waste of time? Ann Surg 224: 198–203 18. Watson DI, Davies N, Jamieson GG (1999) Totally endoscopic Ivor–Lewis esophagectomy. Surg Endosc 13: 293–297 19. Watson DI, Jamieson GG, Devitt PG (2000) Endoscopic cervicothoracoabdominal esophagectomy. J Am Coll Surg 190: 372– 378 20. Zhang X, Watson DI, Jamieson GG, Lally C, Bessell JR, Devitt PG (2005) Outcome of esophagectomy for adenocarcinoma of the esophagus and esophagogastric junction. ANZ J Surg 75: 513–519