Billroth and Czerny described the first esophageal resections in the 1870s, and these consisted of resections of the cervical esophagus without reconstruction. Later, resection of gastroesophageal (GE) junction tumors was performed by laparotomy with gastroesophageal anastomosis to reestablish intestinal continuity. Because there were concerns over respiratory compromise, surgeons were hesitant to enter the chest to perform esophageal resection. In 1915, Torek described the first transthoracic esophageal resection.¹ He used a left thoracotomy to resect the esophagus but did not attempt reconstruction. Instead, a cervical esophagostomy and abdominal gastrostomy were performed. A 3-ft-long external rubber tube was used to connect the ostomies, and it allowed the patient to eat for 17 more years (Fig. 27-1). Turner performed the first transhiatal esophagectomy in 1933.² Oshawa reported the first transthoracic resection of the esophagus with esophagogastric anastomosis in 1933.³ Knowledge of this procedure did not become widespread in the Western community until Adams and Phemister described the procedure in 1938.⁴

Ivor Lewis is credited with popularizing transthoracic resection of the esophagus. Initially, he performed the procedure in two stages: first, mobilizing the stomach via laparotomy, and several days later resecting the intrathoracic esophagus and reconstructing with the stomach. The Ivor Lewis approach (which is an upper midline laparotomy for mobilization of the gastric conduit followed by right thoracotomy for resection and reconstruction) and the transhiatal approach are currently the two most commonly used techniques of esophageal resection. In 1962, McKeown described a tri-incisional approach. He used a right thoracotomy to mobilize the esophagus. The patient was then repositioned in the supine position, the gastric conduit was mobilized by laparotomy, and the anastomosis was performed in the neck.⁵ Minimally invasive options for surgical resection have also become increasingly popular.^6,7 Thoracoscopic and laparoscopic techniques, robotic surgery, as well as combined minimally invasive with open approaches have created a wider variety of experiences and are discussed in other chapters.

Historically, surgery has been the primary mode of treatment for localized esophageal cancer. Nonetheless, the long-term results of surgery alone for esophageal cancer are disappointing, with a 5-year survival of approximately 20%.^8–10 Given the poor results with surgery alone, preoperative chemotherapy with or without radiation has been proposed as a means of improving long-term survival. Twelve randomized trials have been performed using preoperative chemoradiation, while nine additional randomized studies have evaluated the benefit of preoperative chemotherapy without radiation. Although two of the larger randomized trials comparing preoperative chemoradiation followed by surgery to surgery alone showed no difference in survival,^11,12 several randomized trials have been used to support the use of preoperative chemoradiation.

Urba and colleagues looked at 100 total patients randomized to preoperative chemoradiation or surgery alone.¹³ Median survival was approximately 18 months in both groups; however, there was a trend toward improved survival at 3 years (30% vs 16%), albeit statistically insignificant.

Walsh and associates randomized 113 patients, and at 3 years there was a 32% survival benefit for those receiving preoperative cisplatin plus 5-FU and radiation (40 Gy) versus 6% for those undergoing surgery alone (p = 0.01).¹⁴ This study, however, has been heavily criticized for its lack of adequate pretreatment staging and survival in the surgical arm that is far below other reported series.

Cancer and Leukemia Group B 9781 (CALGB 9781) by Tepper and collegues, evaluated patients with stage I to III esophageal cancer. Patients were randomized to surgery alone or to preoperative cisplatin and 5-FU with concurrent radiation (50.4 Gy) followed by surgery. Poor accrual resulted in premature closure of this underpowered study of 56 patients. Nonetheless, with median follow-up of 6 years, 5-year survival was 39% for the trimodality group versus 16% for the surgery-alone group (p = 0.002).¹⁵

The 2012 Chemoradiotherapy for Oesophageal Cancer Followed by Surgery Study (CROSS) by Van Hagen et al. is a large multicenter study of 366 patients, and has been the most significant contribution supporting neoadjuvant chemoradiotherapy. The chemoradiotherapy cohort was dosed carboplatin/paclitaxel with 41Gy of concurrent radiation prior to surgery. Findings were encouraging, with the achievement of an R0 resection in 92% of patients randomized to preoperative chemoradiation versus 69% in the group managed with surgery alone (p < .001). Five-year overall and disease-free survival were also significantly improved in the preoperative chemoradiotherapy group as compared to the surgery-alone cohort, with 47% and 34% overall survival. Median survival was 49 months in the neodadjuvant group versus 24 months in the surgery-only group. However, most of the benefit was seen in the squamous cell population. Survival in patients with squamous cell cancer had a hazard ratio of 0.422 (p < .007). Adenocarcinoma patient survival had a hazard ratio of 0.71 but with a p value of 0.7, demonstrating a trend but not statistically significant difference.¹⁶

In an effort to settle controversies arising from several conflicting study results, there are a number of meta-analyses comparing neoadjuvant chemotherapy or chemoradiotherapy to surgery alone. The most recent is by Sjoquist and associates, a 2011 update of the 2007 study by Gebski et al.¹⁷ The 2007 analysis reported a significant survival benefit for neoadjuvant chemoradiotherapy as well as neoadjuvant chemotherapy in patients with esophageal cancer. The update, with nearly 4000 patients, strengthened previous results, returning a hazard ratio of 0.78 (95% CI 0.70-0.88; p < 0.0001) corresponding to an absolute survival benefit at 2 years of 8.7% and a number-needed-to-treat of 11 with trimodality treatment.¹⁸

A 2004 meta-analysis by Fiorica and collegues included 734 patients, and returned similar findings with a significant improvement in 3-year survival with preoperative chemoradiotherapy versus surgery alone. However, this study also reported a significant increase in postoperative morbidity including respiratory complications, heart failure, and anastomotic leak in patients treated with neoadjuvant therapy.¹⁹

Urschel and Vasan in 2003 combined the results of over 1100 patients from nine randomized controlled studies comparing neoadjuvant chemoradiotherapy followed by surgery versus surgery alone. This study also favored neoadjuvant chemoradiotherapy with surgery over surgery alone.²⁰ These studies have led to the most recent (2015) National Comprehensive Cancer Network (NCCN) guidelines for the use of neoadjuvant chemoradiation in medically and surgically fit patients diagnosed with resectable, locally advanced esophageal cancer of the GE junction.²¹

Similar to the studies evaluating the potiential survival gain with preoperative chemoradiotherapy, there is substantial comparative evidence of the benefit of neoadjuvant chemotherapy without radiation for locally advanced esophageal cancer. The largest of such studies is the Medical Research Council (MRC) esophageal cancer (OEO2) trial of 2002, which was updated in 2009.^22,23 Both the original and long-term studies demonstrated a statistically significant survival benefit (23% vs 17%) and progression-free survival in those patients who received preoperative chemotherapy. A comparable well-sized US study by Kelsen et al. was published in 1998. The original results of this study, however, were contradictory, displaying no evidence of a difference in the disease-free or overall survival between the two treatments.²⁴ Nonetheless, a 2009 update of this trial did report the necessity for complete resection with negative microscopic margins, as only patients who underwent an R0 resection had a substantial chance of long-term disease-free survival.²⁵

These reports were followed by the Medical Research Council Adjuvant Gastric Infusional Chemotherapy (MAGIC) trial in 2006. This study looked at GE junction tumors including those that were considered gastric, an important group that had been excluded from many earlier studies. This trial demonstrated an improved survival at 5 years of 36% versus 23% in patients with GE junction adenocarcinoma treated with preoperative chemotherapy.²⁶

Athough many of these studies suggest that neoadjuvant treatment may prove beneficial, there still remains controversy about the use of chemotherapy versus chemoradiotherapy. Two head-to-head comparisons of neoadjuvant chemotherapy versus neoadjuvant chemoradiotherapy have been completed to date. Both studies, however, were underpowered and consequently unable to determine a statistically significant survival advantage. The first trial by the German Esophageal Cancer Study Group did display a trend toward improved survival with trimodality therapy.²⁷ The second study by Bermeister and collegues did not share this trend.²⁸ Unfortunately, no clear determination has been made regarding which method is better. The relatively low incidence of esophageal cancer, the variable response to treatment between squamous cell carcinoma and adenocarcinoma, and the regional practice patterns make a large, randomized study difficult to envision.

It is important to recognize those patients with stage IV disease, because the mean survival in these patients is 6 to 10 months. In the past, palliative esophagectomy was thought necessary to restore swallowing and oral nutrition. With advances in photodynamic therapy, expandable endoscopic stents, and other endoluminal therapies, it is unusual for anyone to require esophageal replacement to reestablish swallowing ability. Hence, stage IV patients should be spared the perioperative mortality, morbidity, and recovery time associated with esophagectomy. The appropriate use of neoadjuvant treatment requires accurate staging. Patients with nodal involvement, invasion through the esophagus, or possibly even invasion into the muscularis often undergo preoperative chemoradiation, while patients with simple mucosal involvement generally proceed directly to endoscopic treatment or surgical resection.

The main staging modalities available today are computed tomography (CT) scan, positron emission tomography (PET) scan, and endoscopic ultrasound (EUS). CT scans are used mainly for detecting distant metastases in the lungs, liver, or other remote sites, including the brain. CT scan may be useful for excluding T4 tumors if a fat plane can be demonstrated between the adjacent structure and the esophagus. Such staging is often not possible if the patient is severely cachectic or if there are no natural fat planes, such as that between the trachea and esophagus. In regard to nodal status, CT is not as sensitive or as accurate as EUS.

PET scan is superior to CT scan for detecting distant metastatic disease. In a series of 91 patients, CT scan had a sensitivity of 46%, a specificity of 74%, and an overall accuracy of 73%. In contrast, PET scan had a sensitivity of 69%, specificity of 93%, and overall accuracy of 84%. All metastases that were missed by PET were less than 1 cm in size.⁷ Other studies have shown similar results.^29,30 In addition, PET scan may aid in the diagnosis of primary tumor where it may be difficult to perform biopsy because of obstruction. Conversely, a certain percentage of nonbulky tumors of the esophagus may be PET-negative.

EUS gives detailed images of the esophageal wall and nearby structures (Fig. 27-2). Accurate identification of the layers of the esophageal wall is possible. Muscle layers tend to be hypoechoic with intervening hyperechoic mucosal layers. The first and second hypoechoic layers correspond to the mucosa and muscularis mucosa, respectively. The third hyperechoic layer is submucosa. The fourth hypoechoic layer is the muscularis propria, and the fifth hyperechoic layer is the outside of the esophagus. Tumor infiltration of the wall disrupts the normal-layered appearance, and extent of penetration is usually clearly visible. EUS has an overall accuracy of 80% to 90% in ascertaining T status. The differentiation between T1 and T2 is most difficult. In addition, biopsy of deeper layers of tumor not accessible by traditional grasping forceps is possible. It should be noted that EUS is not accurate in defining postneoadjuvant treatment T status because of fibrosis induced by the chemoradiation.

Nodal status is determined by examining four characteristics. Malignant nodes tend to be round and hypoechoic. They have discrete borders and are larger than 1 cm in size. Nodes that meet such criteria have a 90% chance of being malignant. Fine-needle aspiration (FNA) further increases the accuracy in determining nodal status. If the tumor is from a node, the cytopathologist should be able to identify lymphoid tissue in the specimen. False positives can result with FNA if the needle passes through the primary tumor. The accuracy of EUS in N-status staging is between 70% and 80%. EUS is 10% to 15% more accurate than CT scan.³¹

Developments in EUS and PET scanning have lessened the enthusiasm for preresection operative staging of esophageal cancer patients. Operative staging involving laparoscopy and thoracoscopy is more invasive but may be superior to EUS. Luketich and associates studied 26 patients and detected N1 disease in a considerable number of patients staged N0 by EUS.³² It should be noted, however, that the sensitivity of EUS in this series was only 60%, considerably lower than that described in other series. In addition, 15% of patients with no radiographic metastatic disease were found to have liver metastases by laparoscopic staging. The average cost of surgical staging was $20,000 to $25,000 versus $2000 for EUS.

A common algorithm used in staging patients includes endoscopy for primary diagnosis, CT scanning with PET to evaluate for metastatic disease, and EUS if the patient is an operative candidate and neoadjuvant therapy is considered. In cases of esophageal obstruction, where EUS scanning is known to be less accurate, the incidence of lymph node metastasis is very high (90%), and neoadjuvant therapy should be considered.

The treatment of a cancer of the cervical esophagus is challenging and requires a multidisciplinary approach involving an otorhinolaryngologist, a thoracic surgeon, and occasionally a plastic surgeon. Frequently, radiation will be required preoperatively to maximize margins and spare the larynx, if possible. The neck incision is made along the anterior border of the sternocleidomastoid muscle and can be extended across the midline if additional exposure is needed. If the tumor is fixed to the spine or neck vessels, the procedure is aborted and palliative radiotherapy is considered. If the larynx is involved, it is removed en bloc with the upper esophagus along with the upper paraesophageal nodes bilaterally. A radical neck dissection is not routinely performed. The dissection spares the jugular vein, sternocleidomastoid muscles, and spinal accessory nerves. The trachea is transected, leaving enough length to allow construction of a permanent end tracheostomy. The endotracheal tube is inserted into the distal trachea and the hypopharynx is divided sharply.

By this point, a separate midline abdominal incision will have been performed, and blunt dissection is begun on the esophagus from the abdomen. A two-team approach should be considered, with one team at the neck while the other prepares the gastric conduit. The gastric conduit is elevated to the neck with traction and the gastroesophageal junction is divided. The pharyngogastrostomy anastomosis is performed using a single-layer, interrupted hand-sewn anastomosis with a nonabsorbable suture. The cervical tracheostomy is performed above the sternal notch. If too much trachea has been resected to allow for this, manubrial resection will permit placement of the end tracheostomy lower in the midline.

Lesions below the thoracic inlet can be divided according to their location in the upper esophagus (below the thoracic inlet but above the carina), midesophagus (between the carina and inferior pulmonary vein), or lower esophagus (below the inferior pulmonary vein). While we favor the tri-incisional approach for all malignant lesions (for reasons discussed later), lesions in the upper thoracic esophagus generally must be approached with this technique to ensure adequate proximal margins. If the lesion is in the midthoracic esophagus, either the tri-incisional approach or the Ivor Lewis approach may be adequate. Lower esophageal tumors can be resected with either of these two approaches, or additionally with a transhiatal approach, or left thoracotomy and distal esophagectomy. With any resection, accommodation must be made for additional resection with reconstruction if frozen margins are involved with tumor.

Transhiatal versus Transthoracic Techniques

Five prospective, randomized trials have been performed comparing transhiatal to transthoracic resection. The first was published in 1993 by Goldmine and associates.³³ Sixty-seven patients younger than 70 years with squamous cell cancer were randomized to Ivor Lewis resection or transhiatal resection. Operative time was longer (6 vs 4 hours) in the Ivor Lewis group. There was no difference in the incidence of pneumonia (20%), anastomotic leak, recurrent nerve injury, bleeding, perioperative mortality, or length of hospital stay. For those patients with nodal disease, however, none of the transhiatal patients were alive at 18 months, while 30% of the transthoracic patients were alive at 18 months.

The most recent randomized trial is from 2006. This study from Pakistan by Areja and associates found no significant difference between the two approaches. However, this study, like many before it, was limited, with only 30 patients.³⁴

Chu and coworkers randomized 39 patients with lower-third esophageal cancers to either Ivor Lewis or transhiatal resection.³⁵ Limitations of the study were small sample size, short follow-up (mean 15 months), and patient exclusions. Patients undergoing neoadjuvant therapy or those with forced expiratory volume in 1 second (FEV₁) less than 70% of expected were excluded. There were no perioperative deaths in either group. Intraoperative hypotension occurred in 60% of transhiatal patients but only in 5% of transthoracic patients. There was no difference in blood loss, pneumonia, or recurrent nerve injury. The mean proximal margin was 3 cm longer in the transhiatal group. No significant difference was seen in tumor recurrence or survival during the brief follow-up period.

A study comparing transhiatal resection to transthoracic, tri-incisional en bloc resection for distal adenocarcinoma of the esophagus or cardia was performed in the Netherlands. One hundred and six patients were randomized to transhiatal resection and 114 patients to transthoracic resection. In-hospital mortality was 2% to 4% in each group. Chyle leak was higher in the transthoracic resection group (10% vs 2%). Respiratory complications including atelectasis and pneumonia were higher in the transthoracic group (57% vs 27%). Although statistical significance was not reached, 39% of the transthoracic group was alive at 5 years, while only 29% of the transhiatal group survived 5 years.³⁶ An update of this study following with a full 5-year follow-up continued to show no statistically significant overall survival in either approach.³⁷ However, in a subgroup of patients who had one to eight positive lymph nodes in the resection specimen, the transthoracic approach (TTE) demonstrated improved overall survival compared with the transhiatal approach (THE) (39% TTE vs 19% THE, p = 0.05). Disease-free survival was similarly improved with the transthoracic approach (64% TTE vs 23% THE, p = 0.02).

Thirty-two patients were included in a trial by Jacobi et al.³⁸ This trial sought to determine if there was a difference in perioperative pulmonary function between the two approaches by measuring several cardiopulmonary parameters. They found a significant increase in operative time, blood loss, and need for transfusion in the transthoracic approach. Although there was a slight increase in introperative pulmonary strain during single-lung ventilation in the transthoracic patients, this was well compensated and transient. Additionally, the pulmonary disturbance did not correlate to a significant difference in postoperative cardiopulmonary events, 30-day mortality, anastomotic leak, or 1-year survival (70% THE vs 77% TTE). Of note, the patients in this study were highly selected and included in the study only if they were aged less than 75 years and free of cardiac, pulmonary, or renal dyscrasia.

The randomized trials show no statistically significant difference in survival, but they are small, and trends toward improved survival are observed in patients undergoing transthoracic dissection. No difference in mortality have been detected in the completed trials thus far.

The analyses summarized above have been performed in an attempt to quell the debate of the superiority of either transhiatal or transthoracic approach. Three meta-analyses to date have attempted to determine the most appropriate approach to the surgical treatment of esophageal cancer. Boshier and collegues reviewed 52 English-language studies between 1981 and 2009.³⁹ Comparable to other meta-analyses, there was no significant difference in 5-year survival, postoperative cardiac complications, hemorrhage, acute respiratory distress syndrome/acute lung injury, chyle leak, and renal insufficiency. Significant differences were found with the extent of lymphadenopathy, with a mean resection yield of eight more lymph nodes found with the TTE. Additionally, the TTE was found to have an increased incidence of pneumonia (21% vs 17%), early (<30 days or in hospital), mortality (10% vs 7%), and a prolonged length of stay of 4 days as compared to the THE. The THE, however, was found to have a statistically significant increase in anastomotic leak (17% vs 10%), anastomotic stricture (25% vs 21%), and vocal cord paralysis (11% vs 5%). A subgroup analysis of 22 studies published after 1999 was also completed. Surprisingly, this analysis of more recent studies determined that the differences in anastomotic leak, vocal cord paralysis, and pneumonia were no longer significant.

Rindani and associates reviewed 44 trials involving either Ivor Lewis or transhiatal esophagectomy that were published in the English language between 1986 and 1996.⁴⁰ Overall, the incidence of bleeding, cardiac complications, or pneumonia was no different between the two groups. Differences were seen in the anastomotic leak rate (16% transhiatal vs 10% Ivor Lewis), stricture rate (28% transhiatal vs 16% Ivor Lewis), and incidence of recurrent nerve injury (11% transhiatal vs 5% Ivor Lewis). Mortality was higher after the Ivor Lewis approach (9.5%) than the THE (6.3%). Long-term survival was approximately 25% with either technique.

Hulscher and colleagues performed a meta-analysis of 50 studies published between 1990 and 1999 involving transthoracic and transhiatal resection.⁴¹ Cardiac complications (20% vs 7%), anastomotic leakage (24% vs 7%), and vocal cord paralysis (10% vs 4%) were higher in the transhiatal group as opposed to the transthoracic group. Pulmonary complications (19% vs 13%), in-hospital mortality (9% vs 6%), and operative time (5 vs 4.2 hours) were higher in the transthoracic group. Overall long-term survival was similar between the two groups (23% for transthoracic and 21.7% for transhiatal resections).

It is important to note that these meta-analyses are retrospective and nonrandomized. In addition, reported surgical quality was suboptimal in both approaches, and the transthoracic group was noted to have more advanced cancer. Caution should therefore be used in applying these findings to individual institutions and patients.

Several patient factors must be considered when chosing the operative approach to esophageal resection. Some argue that transhiatal dissection may be less taxing on an elderly or debilitated patient (due to shorter operative time or avoidance of a thoracotomy). Many surgeons experience hemodynamic changes with transhiatal resection presumably due to compression during blunt dissection. Wong reported intraoperative hypotension in 60% of transhiatal dissections, but in only 5% of transthoracic dissections,⁴² suggesting that the transhiatal operation may be more taxing to a patient with severe cardiac valvular or atherosclerotic disease who cannot tolerate fluctuations in blood pressure. This finding, however, was not cooberated by the Jacobi study.

Tri-Incisional Esophagectomy (McKeown Technique)

The tri-incisional technique of esophageal resection combines the most attractive aspects of the Ivor Lewis and transhiatal approaches. It allows for dissection of the intrathoracic esophagus under direct vision with complete nodal resection and brings the anastomosis to the neck, allowing for maximal proximal margins and minimizing the risk of an intrathoracic leak.

Under general anesthesia, bronchoscopy is performed to rule out tracheal or bronchial (most commonly left main bronchial) involvement with tumor. Esophagogastroduodenoscopy is performed to localize the tumor and rule out disease of the stomach or duodenum. The patient is then reintubated with a double-lumen endotracheal tube and placed in the left lateral decubitus position. A right posterolateral thoracotomy incision is made large enough, approximately 10 cm in length, to introduce the surgeon’s hand (Fig. 27-3). The serratus muscle is spared. Division of the intercostal muscles anteriorly and posteriorly often permits adequate rib spreading without the need to remove a small portion, or shingle, of rib. The chest is entered through the fifth or sixth interspace, depending on the location of the tumor. The inferior pulmonary ligament is divided using electrocautery, and the lung is retracted anteriorly.

Dissection of the esophagus begins at a point away from tumor and any associated scarring, and the esophagus is encircled with a Penrose drain. Traction on the Penrose drain allows for cautery dissection encompassing all adjacent nodes. Arterial branches directly off the aorta are clipped or ligated. The settings on the electrocautery should be low when cauterizing near the trachea. The azygos vein is typically divided, although this is not always necessary (Fig. 27-4). At this level, the vagus nerves are identified. Dissection cranial to this level involves the vagus nerves; the vagus nerves are peeled off and away from the esophagus to avoid injury to the recurrent vagus branches.

Dissection between the trachea and esophagus must be done with care and with low cautery dissection to avoid injury to the membranous trachea. Much of the dissection high in the chest can be done bluntly (Fig. 27-5). The cranial aspect of the dissection is complete when one’s fingers reach easily above the first rib. The Penrose drain is knotted and passed into the lower neck with the knot against the vertebral body for later retrieval during the neck phase of the dissection (Fig. 27-6).

Another Penrose drain is used to gain traction on the lower esophagus, and dissection continues caudally. All tissue between the pericardium, aorta, and azygos vein is dissected and incorporated into the specimen. No effort is made to resect the thoracic duct, although it is sometimes injured. For tumors near the gastroesophageal junction, a rim of diaphragm is incorporated into the specimen. The knotted Penrose drain is placed in the abdomen for later retrieval (Fig. 27-7). At this point, careful inspection is made for hemostasis and injury to the thoracic duct. Often, injury to the thoracic duct is evident when slightly cloudy or crystallized fluid is seen pooling in the region of the duct. If an injury to the duct is seen, it should be closed with a pledgeted fine suture such as 5-0 Prolene. Mass ligature of the duct, as it enters the chest, is then performed by encompassing all tissue between the spine, aorta, and azygos vein at the level of the hiatus with a 0-silk suture. A 28-Fr straight chest tube is inserted via a separate stab incision and directed to the apex of the chest. An additional hole in the tube can be made to facilitate dependent fluid drainage. The ribs are reapproximated with 2-0 Vicryl sutures. The latissimus layer is closed using a running 0 Vicryl suture. A subdermal layer is closed with 2-0 Vicryl and the skin is closed in subcuticular fashion.

The patient is placed in the supine position and is reintubated with a single-lumen tube. A roll is placed under the back to permit neck extension, and the head is turned to the right. A midline laparotomy is performed from the umbilicus to the xiphoid. Exploration of the abdomen should include a careful palpation of the liver and inspection of the serosal surfaces for tumor implants. Palpation of the GE junction and proximal stomach should be performed to rule out gastric spread of tumor. The left lobe of the liver is mobilized and retracted to the right. The Penrose drain left from the chest dissection is used for retraction of the GE junction (Fig. 27-8). The gastroepiploic artery is identified and palpated. The pulse should be easily palpable provided the patient has a physiologic blood pressure. Staying at least 2 cm away from the gastroepiploic artery, the lesser sac is entered. Dissection continues cranially on the stomach along the greater curvature. Dissection may be performed by dividing tissue and ligating with 2-0 silk ties or by using an ultrasonic scalpel. The stomach is retracted medially and the omentum laterally. The artery itself should not be grasped or used for retraction. The gastroepiploic arcade ends near the point where the short gastric arteries begin. A pack placed behind the spleen often aids in exposure of the short gastric vessels (Fig. 27-9). The short gastric vessels can be ligated, double-clipped, or divided with an ultrasonic scalpel. Large vessels should be tied. Care should be taken not to incorporate stomach wall in the ligature, as this may result in delayed necrosis of stomach wall and a postoperative intrathoracic leak. Dissection on the greater curvature proceeds to the hiatus and is complete when the Penrose drain is reached.