| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2005;129:623-631
© 2005 The American Association for Thoracic Surgery
General Thoracic Surgery |
a Departments of Thoracic and Cardiovascular Surgery and Biostatistics and Epidemiology
b The Cleveland Clinic Foundation, Cleveland, Ohio
Read at the Eighty-fourth Annual Meeting of The American Association for Thoracic Surgery, Toronto, Ontario, Canada, April 25-28, 2004.
Received for publication April 23, 2004; accepted for publication August 30, 2004. * Address for reprints: Thomas W. Rice, MD, Department of Thoracic and Cardiovascular Surgery, The Cleveland Clinic Foundation, 9500 Euclid Ave/Desk F24, Cleveland, OH 44195 (E-mail: ricet{at}ccf.org).
 |
Abstract |
|---|
 Top Abstract Patients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
METHODS: From March of 1996 to October of 2002, 274 patients with esophageal cancer underwent esophagectomy with gastric replacement and cervical esophagogastric anastomosis. Beginning in March of 2001, a modified Collard technique (stapled) was used in most patients (n = 86) for cervical esophagogastric anastomosis; a standard hand-sewn technique (sewn) was used in all others (n = 188). Using a propensity score based on 8 variables (age, gender, race, surgeon, surgical approach, pathologic stage, histologic cell type, and induction chemoradiotherapy), 85 patient pairs were matched and followed for time-related events. Outcome comparisons included cervical wound infection, cervical anastomotic leak, other hospital complications, length of stay, anastomotic dilatation, reflux symptoms, and survival.
RESULTS: At 30 days, freedom from cervical wound infection was 92% for stapled versus 71% for sewn anastomoses (P = .001), and freedom from cervical anastomotic leak was 96% versus 89% (P = .09), respectively. Other hospital complications occurred in 58% and 49%, respectively (P = .17). Median length of stay was 10 days for both (P = .3). At 2 years, freedom from anastomotic dilatation was 34% for stapled versus 10% for sewn anastomoses (P < .0001), and the mean number of dilatations per patient was 2.4 versus 4.1 (P = .0001), respectively. Reflux was rare for both. Thirty-day, 6-month, and 24-month survivals were 98%, 91%, and 77% for stapled anastomoses and 98%, 88%, and 69% for sewn anastomoses (P = .3).
CONCLUSIONS: The modified Collard anastomotic technique dramatically reduces morbidity after esophagectomy. It should replace hand-sewn esophagogastric anastomoses.
Reconstruction is as important as resection in patients undergoing esophagectomy, and complications of reconstruction might adversely affect the outcome of surgical treatment of esophageal cancer. One of the critical aspects of reconstruction is the esophagogastric anastomosis. Collard and colleagues2 proposed a partially stapled, partially hand-sewn anastomotic technique (terminalized, semimechanical, side-to-side cervical esophagogastrostomy) to increase the cross-sectional area of the anastomosis. The purpose of our study was to assess the effect of a modified Collard technique, compared with that of hand-sewn anastomosis, on outcome after esophagectomy for esophageal cancer.
|
Patients and methods |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
|
Hand-sewn technique
Hand-sewn anastomoses were end-to-side and constructed with a single-layer interrupted (polyglactin 910) suture technique. At the surgeon's discretion, an interrupted inverting 3-0 (polyglactin 910) Lambert technique was used, particularly along the posterior suture line.
Comparison of anastomotic techniques
Groups
Matched groups were identified using propensity-score methods to allow a valid comparison of outcome between patients with stapled and hand-sewn anastomoses.3,4 The propensity score was based on 8 variables: age, gender, race, surgeon, surgical approach, pathologic stage, histologic cell type, and induction chemoradiotherapy (Electronic Appendix Table 1 available at www.mosby.com/jtcvs). This yielded 85 well-matched pairs of patients (Table 1).
|
|
Patient follow-up
Outcome was assessed from review of hospital and outpatient visit records, contact with referring physicians, and patient questionnaires approved by the institutional review board. Follow-up was 100% complete for survival and all outcomes. However, only living patients could be followed up for symptoms of reflux. In the 85 propensity-matched patient pairs, median follow-up was 1.3 years, with 10% of patients followed less than 3 months and 10% followed more than 4.3 years. There were 320 patient-years of follow-up.
Analysis of outcome
Nonparametric estimates of freedom from various time-related outcome events, including length of stay, were obtained using the Kaplan-Meier method, and a parametric method was used to resolve the number of phases of instantaneous risk.5 (For additional details, see http://www.clevelandclinic.org/heartcenter/hazard.) An indicator variable for hand-sewn versus stapled anastomosis was forced into each hazard phase to compare outcome, and the log-rank test was used to compare nonparametric estimates. For repeated anastomotic dilatations, nonparametric estimates were obtained with Nelson's repeated-events method,6 and parametric estimates were obtained with hazard function methods, as described above.
Other hospital complications were compared using the 
2 or Fisher exact test when the numerator was 5 or less. Reflux symptoms were analyzed by ordinary logistic regression, with time of assessment as a covariable, because so few patients had more than minimal symptoms.
|
Results |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
|
|
|
|
Anastomotic dilatation
By 3, 6, and 12 months, freedom from anastomotic dilatation for any reason was 59%, 43%, and 37% (CL 31%-42%) for stapled and 32%, 17%, and 12% (CL 9%-15%) for hand-sewn anastomoses, respectively (P < .0001, Figure 5 and Electronic Figure 2 available at www.mosby.com/jtcvs). Mean number of dilatations per patient by 6 months, 1 year, and 2 years was 1.3, 1.9, and 2.4 (CL 2.1-2.6) for stapled and 2.4, 3.3, and 4.1 (CL 3.9-4.5) for hand-sewn anastomoses (Figure 6 and Electronic Figure 3 available at www.mosby.com/jtcvs). Early instantaneous risk of dilatation was substantially higher for hand-sewn anastomoses (P = .0001), but constant-phase instantaneous risk was similar (P = .8), resulting in similar risk of further dilatations after about 1 year.
|
|
|
|
Survival
Survival at 30 days, 6 months, and 24 months was 98%, 91%, and 77% for stapled anastomoses and 98%, 88%, and 69% for hand-sewn anastomoses (P = .3, Electronic Figures 4 and 5 available at www.mosby.com/jtcvs).
|
|
Discussion |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
Principal findings
Readers might be unfamiliar with presentation of these outcomes in a time-related fashion. In the past, outcomes have been presented at points in time, but complications continue to occur. Thus, failure to analyze them in a time-related manner leads to underestimation of prevalence and rates because the denominator (number of patients at risk) diminishes across time.
Cervical wound infection
Cervical wound infection, which occurred less often with the modified Collard technique, is infrequently reported. Even in their comprehensive report of the technical complications of esophagectomy, Rizk and associates11 relegated cervical wound infection to the category of "other infections" and not "technical complications."
The hazard functions for cervical wound infection and anastomotic leak were nearly identical in shape but different in magnitude, with wound infection recognized more frequently. In fact, all but one patient in each group had a cervical wound infection diagnosed before the anastomotic leak. Thus, cervical wound infection is a possible harbinger of anastomotic leak. Indeed, early treatment of such infection might prevent leaks.
Cervical anastomotic leak
Cervical anastomotic leaks were uncommon, but probably (P = .09) occurred less frequently in patients with a modified Collard anastomosis. Occurrence of anastomotic leak in the hand-sewn group was consistent with recent reports of large-volume esophageal surgery centers.11,12
Diagnosing anastomotic leak may be problematic. Although its diagnostic accuracy is unknown, the barium esophagogram can be falsely negative because of streaming of barium across the anastomoses or delay in barium leakage; it can be falsely positive because of the geometry of the anastomosis. Clinical importance of a radiographic finding interpreted as a leak (radiographic leak) has not been established. About half of radiographic leaks were detected by water-soluble contrast, and the other half were detected by high-density barium in the study by Swanson and colleagues,13 with few (3/46) demonstrating free extravasation. These were detected only with water-soluble contrast, a hypertonic agent that, if aspirated, might cause pulmonary edema and death. Management of patients with radiographic leaks was changed in 86% of leaks detected by water-soluble contrast and 53% of those detected by high-density barium.13 Similarly, Gollub and Bains14 found no mediastinitis on follow-up of 29 radiographic leaks in 12 patients. The use of barium in assessing gastric emptying in the recovery period might be misleading and might not reflect long-term gastric function. In addition, the use of barium in patients after gastrointestinal surgery has deleterious effects on recovery. For all these reasons, we have abandoned postoperative radiographic studies and aggressively treat only clinically evident anastomotic leak.
To further reduce requirements for diagnostic studies and in an effort to improve long-term gastric function, we delay feeding for up to 6 weeks, relying on a J tube for nutrition. This avoids gastric distention and poor emptying, allowing the anastomosis to heal and gastric tone to return.
Other hospital complications
Other hospital complications occurred frequently and similarly in both groups. Their occurrence was consistent with Medstat data based on 13 national cancer institutions that experienced 55% complications and on 88 community hospitals that experienced 68% complications from 1994 through 1996.15
In the past, complications that occur frequently, such as atrial fibrillation, have been trivialized because they are inconsequential in most patients and are thought by surgeons not to be directly related to technical aspects of the operation. Thus, the most common complications in this study were not technical ones but atrial fibrillation, which is generally unrelated to anastomotic technique. Patients and surgeons should be aware that the majority of patients will have a complication after a procedure of this magnitude and extent, ranging from inconsequential to life threatening.
Anastomotic dilatation
Function of an anastomosis is difficult to quantify. Patients might report having dysphasia but not have a stricture; however, their dysphasia might respond to dilatation.16 Therefore, we used anastomotic dilatation to reflect dysfunction and not necessarily stricture, and it was studied in a time-related fashion. All dilatations, regardless of indication or institution where performed, were recorded. The modified Collard anastomosis was associated with a marked reduction in both the number of patients requiring dilatation and the number of dilatations per patient in the first year postoperatively. After this time, the need for dilatation was similar for both anastomotic techniques, suggesting that severe anastomotic dysfunction can occur regardless of technique and, once established, has a prolonged treatment course.
It is not possible to compare directly our information about dilatation with the findings of past reports. We have reported all dilatations and not a selected proportion of them deemed to be for stricture and reported at a single point in time. Nevertheless, early 3-month freedom from dilatation of 59% (CL 54%-63%) for the modified Collard anastomosis is consistent with Orringer and colleagues' report10 of non–time-related occurrences within 3 months of 65%.
Although attempts to increase the luminal area of the anastomosis might increase the chance of reflux, this was not found. Cervical anastomotic site might account for this.
Survival
Unlike the inference of Rizk and colleagues,11 reducing cervical wound infections and anastomotic leaks by changing anastomotic technique did not improve survival. The study sizes were similar, but the majority of their leaks occurred in the chest, which is more lethal than a cervical leak. Thus, use of a safer cervical anastomosis transforms a potentially lethal event into an annoying complication.17-20 The spurious implication that a circular stapled anastomosis might increase mortality is similarly a reflection of anastomotic site rather than technique, a shortcoming of meta-analysis.21
Technical details
The fundamental modification we made to the Collard technique is stapler orientation, placing its longer and bulkier fork into the stomach. This more likely ensures that the entire staple line is deployed in the anastomosis because (1) the stomach is not fixed, as is the esophagus, by the pharynx; (2) angulation of the instrument caudally allows the mobile stomach to be placed over the fixed esophagus; and (3) the stomach has a larger capacity to accept the cartridge. Other factors critical to the success of the anastomoses are (1) accurate alignment of the esophagus and stomach to ensure maximum separation of the lesser curve staple line and anastomosis and (2) adequate mobilization of the proximal esophagus.
We close the anterior wall in a transverse fashion with a running absorbable monofilament suture. Because this anastomosis can be constructed faster than an interrupted hand-sewn one and handling of the open bowel is decreased, the potential for wound contamination is reduced. Avoiding a completely running anastomosis decreases the potential for ischemia. Completing the anterior wall with sutures rather than staples also avoids stricture by eliminating stapler overbite and crossed staple lines. Transverse closure allows maximal distraction of the stapled margins and anastomotic cross-sectional area.
Discussion
Dr Mark B. Orringer (Ann Arbor, Mich). This is yet another report documenting how use of the Endo GIA stapler for construction of the cervical esophagogastric anastomosis has reduced the anastomotic leak rate from the 14% to 15% average reported number to the 3% to 5% range. The authors' technique is essentially that reported by Collard in 1998, differing only in the orientation of the stapler and the method of closure of the anterior walls of the esophagus. The side-to-side stapled anastomosis we described in 1999 is another variation on this theme. Our anastomosis creates an acute angle between the esophagus and the stomach and places several centimeters of the gastric fundus behind the cervical esophagus, whereas your Collard anastomosis is an end-to-end configuration. Regurgitation might be less with our valve-like anastomotic construction. In discussing outcomes after an esophagogastric anastomosis, the incidence and degree of postoperative regurgitation should be reported. Therefore, the first question, Dr Ercan, is as follows: Is clinically significant regurgitation a problem after your Collard anastomosis?
Dr Ercan. Thank you very much, Dr Orringer. It is a privilege to have you discuss our paper.
In 116 living patients, most recent follow-up was by telephone questionnaire. We asked about and graded reflux symptoms. Eighty-eight percent of responders with a stapled anastomosis and 82% of responders with a sewn anastomosis were free of reflux.
Dr Orringer. As critical to the success of a cervical esophagogastric anastomosis as the Endo GIA stapler is minimizing trauma to the stomach. Our rule is that if the stomach is pink and healthy in the abdomen after it is mobilized and pink and healthy in the neck once it is brought up, an anastomotic leak should not occur. I agree that it is very important to separate the lesser curvature staple line from the anastomotic staple line to avoid ischemic necrosis of the intervening tissue. What steps does your group take to minimize gastric trauma, and have you encountered any instances of gastric tip necrosis necessitating takedown of the stomach and a cervical esophagostomy?
Dr Ercan. It is important to minimize trauma to the fundus to the stomach. It has been shown that oxygen tension, especially in the fundus of the stomach, drops considerably—by as much as 50%—during reconstruction. Knowing this, we minimize trauma while bringing the stomach to the neck, regardless of surgical approach.
With respect to gastric necrosis, 4 patients developed it, 1 in the stapled group and 3 in the hand-sewn group.
Dr Orringer. As for my final question, the Collard anastomosis that you have described requires sufficient esophageal length for the right angle and then for the anastomosis that is going to be done. Has the need for that length been problematic in patients with long-segment Barrett esophagus or those with midthird adenocarcinomas, where you need a little bit more length?
Dr Ercan. The Endo GIA stapler requires about 3 cm for firing. We believe the anastomosis should be constructed about 2 cm distal to the cricopharyngeus. Thus, 5 cm of esophageal length distal to the cricopharyngeus muscle is needed. In our series, early-stage cancers dominated, and there were few upper esophageal tumors. Pathologic analysis of resection margins during surgery was negative in all patients. We did not have any problems with esophageal length during anastomosis.
Dr Orringer. I have one final comment. I do not agree with the statement in the manuscript in which you state that the use of barium in a patient recovering from gastrointestinal surgery has deleterious effects on recovery and that the postoperative barium swallow should be abandoned. The postoperative barium swallow, in my mind, provides quality control that is essential in esophageal surgery. It is as fundamental after an esophagogastric anastomosis as it is after a fundoplication, and that is another operation in which failure to diagnose an early disruption might lead to far greater problems later.
Dr Tomasz Grodzki (Szczecin, Poland). I have one technical question. Why did you decide to complete the anastomosis with a running suture instead of a second stapler? It is quite easy.
Dr Ercan. That's a good question. The main purpose for doing a hybrid anastomosis is to obtain a wider anastomotic area. Applying another staple to the anterior row compromises the anastomotic lumen. We were also concerned that crossing staple lines might form a focal point for possible leak.
Dr Patterson. Do you want to make a comment on that, Tomasz? You asked a question. Do you use a double-staple technique?
Dr Grodzki. Yes, of course. We used it for about 40 patients since I learned it in Boston in Dr Sugarbaker's clinic. We have different problems with the esophageal patients, but never, ever with anastomosis.
Dr Murthy. Just to clarify the point that Dr Orringer made about the fundus of the stomach (this is something that Neil Christie brought up in his talk), we don't routinely tubularize the stomach. We leave a fair bit of fundus, which at some level helps it come up to the neck, and I think venous drainage is better. Not using a second stapler saves $100, and that is something to think about.
Also, concerning barium swallow, we find that it does not reliably predict or demonstrate leak, and an assurance of a negative barium swallow is confirmed only by no fever 2 or 3 days after that. Also, there is a constipating effect of barium in these patients whom we're trying to feed.
|
Limitations and advantages |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
|
Conclusions |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
Cervical wound infection: Required opening of the cervical incision for local findings of wound erythema, wound drainage, cervical subcutaneous emphysema, excessive pain, or tenderness. Positive wound culture was not required.
Cervical anastomotic leak: Detection of saliva, ingested material, gastric secretions, or bile in the wound or anastomotic dehiscence. Radiographic confirmation of leak was not required. Hospital stay: Interval between esophagectomy and hospital discharge. Complications: Complications other than cervical wound infection and anastomotic leak occurring after esophagectomy and within the index hospitalization were categorized as cardiac (primarily new-onset atrial fibrillation), pulmonary, pleural, abdominal wound, infectious, vascular, vocal cord paralysis, surgical intervention for bleeding, and nonanastomotic gastrointestinal leak. When a complication led to a sequence of associated complications (eg, venous thrombosis led to bleeding from its anticoagulant management), only the initiating complication was counted. Anastomotic dilatation: Any anastomotic dilatation at any institution, regardless of symptoms or radiographic or endoscopic findings. Reflux symptoms: Reflux symptoms reported by patients at follow-up were graded as none, minimal, or important. Mortality: All-cause mortality from date of esophagectomy to last follow-up.
|
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractPatients and methodsResultsDiscussionLimitations and advantagesConclusionsBox 1. Measures of...References |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. T. Cooke, G. C. Lin, C. L. Lau, L. Zhang, M.-S. Si, J. Lee, A. C. Chang, A. Pickens, and M. B. Orringer Analysis of cervical esophagogastric anastomotic leaks after transhiatal esophagectomy: risk factors, presentation, and detection. Ann. Thorac. Surg., July 1, 2009; 88(1): 177 - 185. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Deng, R.-W. Wang, Y.-G. Jiang, Q.-Y. Tan, Y.-P. Zhao, J.-H. Zhou, X.-L. Liao, and Z. Ma Functional and menometric study of side-to-side stapled anastomosis and traditional hand-sewn anastomosis in cervical esophagogastrostomy Eur. J. Cardiothorac. Surg., January 1, 2009; 35(1): 8 - 12. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. C. Austin Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement. J. Thorac. Cardiovasc. Surg., November 1, 2007; 134(5): 1128 - 1135. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. H. Blackmon, A. M. Correa, B. Wynn, W. L. Hofstetter, L. W. Martin, R. J. Mehran, D. C. Rice, S. G. Swisher, G. L. Walsh, J. A. Roth, et al. Propensity-Matched Analysis of Three Techniques for Intrathoracic Esophagogastric Anastomosis Ann. Thorac. Surg., May 1, 2007; 83(5): 1805 - 1813. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. A. Jobe, C. Y. Kim, R. C. Minjarez, R. O'Rourke, E. Y. Chang, and J. G. Hunter Simplifying Minimally Invasive Transhiatal Esophagectomy With the Inversion Approach: Lessons Learned From the First 20 Cases Arch Surg, September 1, 2006; 141(9): 857 - 866. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-M. Lee, M.-T. Wu, Y.-C. Lee, S.-Y. Yang, J.-S. Chen, H.-H. Hsu, P.-M. Huang, S.-W. Kuo, C.-J. Lee, and C.-J. Chen Association of GSTP1 Polymorphism and Survival for Esophageal Cancer Clin. Cancer Res., July 1, 2005; 11(13): 4749 - 4753. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
