| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Click here to read this article as a CME activity
Ann Thorac Surg 2007;83:1805-1813
© 2007 The Society of Thoracic Surgeons
a The University of Texas MD Anderson Cancer Center, Houston, Texas
b The Methodist Hospital, Houston, Texas
Accepted for publication January 23, 2007.
* Address correspondence to Dr Blackmon, 3741 Robinhood St, Houston, TX 77005 (Email: shblackmon{at}tmh.tmc.edu).
Presented at the Fifty-third Annual Meeting of the Southern Thoracic Surgical Association, Tucson, AZ, Nov 8–11, 2006.
| General thoracic surgery:
The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
|
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Methods: All patients undergoing transthoracic esophagectomy from 1999 to 2005 for esophageal cancer with gastric replacement were reviewed. A prospective quality improvement database, telephone interview, and chart review were used to collect data. A side-to-side stapled anastomosis was done in 44 patients, circular-stapled anastomosis in 147, and hand-sewn anastomosis in 23. Propensity scores were generated from 14 variables, which were then used to generate 23 patient triplets. End points included leak, dysphagia, stricture, other major complications, and overall survival. Follow-up was available on all patients.
Results: For matched triplet comparison, no significant difference was noted in anastomotic leaks (8.7% with side-to-side stapled, 4.3% with circular-stapled, and 4.3% with hand-sewn; p = 0.78). Postoperative dysphagia was significantly higher in hand-sewn anastomoses at 56.5% versus 26.1% with side-to-side stapled and 21.7% with circular-stapled (p = 0.04). Stricture requiring esophageal dilation was also increased in hand-sewn at 34.8% versus 8.7% with side-to-side stapled and 8.7% with circular-stapled (p = 0.04). No difference was noted in perioperative mortality, long-term survival, or locoregional recurrences between techniques.
Conclusions: In this carefully matched group of patients, intrathoracic use of the side-to-side stapled esophagogastric anastomosis in esophageal cancer patients is safe and effective. Postoperative dysphagia and need for stricture dilation may be decreased using a stapled compared with a traditional hand-sewn anastomosis.
An estimated 14,500 patients are diagnosed with esophageal cancer in the United States every year [1]. The standard of care for patients with localized disease is resection. Although mortality rates in esophageal cancer patients undergoing esophagectomy appear to be decreasing [2], morbidity is still significant [3, 4]. Anastomotic complications have a negative affect on patient survival [4]. Postoperative leak, dysphagia, and stricture may be indicators of the quality of an anastomosis [5–8].
Recently published retrospective data by Ercan and colleagues [8] suggest a cervical side-to-side stapled (SSS) esophagogastric anastomosis appears to decrease morbidity compared with traditional hand-sewn (HS) techniques. The three current basic anastomotic techniques used to create an esophagogastric anastomosis for patients undergoing reconstruction after transthoracic esophagectomy are the HS technique, the SSS technique, and the circular-stapled (CS) technique.
Of the five randomized prospective trials comparing a HS with a stapled anastomosis (predominantly CS), the leak rate appeared to be similar, regardless of technique [9–12]. The stricture rate was markedly higher in only one study [10], and favored a HS anastomosis over CS. A retrospective study by Casson and colleagues [13] found the leak rate decreased with a SSS versus HS, but there was no significant difference in stricture rates [13]. When these were pooled into meta-analyses [14, 15], there appeared to be no difference between the two techniques with regard to leak. One of these meta-analyses found stricture increased with a stapled anastomosis [15]. Stapled anastomoses appeared to have a lower leak rate and a higher stricture rate when analyzed as a pooled patient analysis of seven nonrandomized studies [15]. A recent retrospective unmatched review of 280 patients having both cervical and intrathoracic anastomoses found HS anastomoses were more likely to leak, potentially stricture, and require dilatation [16].
Disparity in the literature exists because of variations in technique, location of anastomosis, use of induction therapy, or other variables that have yet to be identified. Few studies specifically look at different methods of stapling in a controlled manner with similar neoadjuvant therapy and compare them against the HS technique in the chest. Because the SSS technique has only been recently applied to the intrathoracic anastomosis, we evaluated our experience with this novel technique for intrathoracic anastomoses and compared the outcome with CS or HS techniques.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Institutional Review Board approval was obtained to review the data. Consent for the study was waived. Database and chart reviews were used to collect data on demographics, comorbidities, stage, indications, anastomotic technique, histology, and adjuvant therapy. End points included leak, dysphagia, stricture, number of dilations, other major complications, and overall survival.
Dysphagia is graded on a 5-point scale from 0 to 4 [17]. A score of 0 was interpreted as no clinical dysphagia, and 4 included a complete inability to swallow anything (Table 1). Any patient with a score of 1 more was determined to have dysphagia. Severe dysphagia included those patients with a score of 3 or more.
|
Intrathoracic Anastomotic Techniques
Stapled side-to-side
The SSS technique is also referred to as the modified Collard [18] or Orringer anastomotic technique [5, 19]. The gastric conduit is prepared in the usual fashion through an abdominal incision along with a pyloroplasty or pyloromyotomy. The thoracic portion of the esophageal mobilization completed, the cancer is resected, and the proximal esophagus is placed medial to the proximal limb of the gastric tube. The stapler is inserted through a gastrotomy on the greater curvature of the conduit, aligning the superior edge of the stapler into the opened distal esophagus (Autosuture Endo GIA 30-3.5; United States Surgical Corp, Norwalk, CT; Fig 1). This side-to-side anastomosis provides a larger luminal diameter than the direct end-to-side method [18] and becomes a functional end-to-end anastomosis. The nasogastric tube is then positioned, and the remaining hood is either sutured or stapled to complete the anastomosis. The chest is closed in the usual fashion.
|
|
|
Statistical Analysis
Direct comparisons of the leak, dysphagia, and stricture rates among the three anastomotic groups were performed using the Fisher exact test. Univariate analyses were then performed using all preoperative variables available from the database predicting postoperative leak, dysphagia, or stricture. Those nearing statistical significance were entered into a multivariable logistic regression analysis with forward selection and backward elimination using leak, dysphagia, or stricture as an end point.
Propensity scores were used to generate triplets of matched patients from each of the three groups (SSS, CS, and HS). All collected preoperative clinical variables were used to generate the propensity score model. Both multinomial logistic regression and cross-tabulations with 
2 analysis for categoric variables and analysis of variance for continuous variables were used to determine which variables predicted patient membership to one of the three groups. None of the preoperative variables predicted group membership, so 14 variables were chosen to create the propensity score for each patient. This propensity score was used to create 23 patient triplets, which were then matched and compared. A large number of postoperative variables were used to compare these triplet groups.
Statistical comparison between the groups was performed using the Cochran Q test for triplet comparison of dichotomous variables and the Friedman test for triplet comparison of continuous variables. All data entry and analysis was performed with SPSS 13.0 software (Apache Software Foundation, Chicago, IL).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
For unmatched Fisher exact test analysis of the 214 patients, the leak, dysphagia, severe dysphagia, stricture, and severe stricture rates for patients having a SSS, CS, or HS anastomosis are listed in Table 1. There was no difference in leak or severe dysphagia, but dysphagia, stricture, and severe stricture were significantly increased among the HS patients.
Univariate Analyses
A univariate analysis of variance identified 10 covariates that were significantly related to postoperative leak, six covariates that were significantly related to postoperative dysphagia, and six covariates that were significantly related to postoperative stricture, Table 2.
|
Multivariable logistic regression analysis of all variables was performed using dysphagia as the end point. The only statistically significant factor was the type of anastomosis performed (p = 0.02). With the CS anastomosis used as the referent, there was no statistically significant difference compared with the SSS group (HR, 0.92; 95% CI, 0.43 to 2.00; p = 0.84), and HS anastomosis patients had an increased hazard ratio indicating a significantly higher rate of dysphagia (HR, 3.60; 95% CI, 1.46 to 8.87; p = 0.01).
Multivariable logistic regression analysis of all variables was performed using stricture as the end point. Again, the only statistically significant factor was the type of anastomosis (p = 0.02). With the CS anastomosis used as the referent, there was no statistically significant difference compared with the SSS group (HR, 0.64; 95% CI, 0.21 to 1.97; p = 0.43), and HS anastomosis patients had an increased hazard ratio indicating a significantly higher rate of stricture (HR, 3.39; 95% CI, 1.27 to 9.02; p = 0.02). For a summary of these data, please see Table 2.
Patient Characteristics for Propensity-Matched Groups
Of the 214 patients included in the initial study, 69 patients were propensity matched into triplets, with each group having 23 patients. Variable frequencies after matching are summarized by group in Table 3. The preoperative variables among the three patient populations were similar and not statistically different after matching, with the exception of body mass index (BMI), which was 26.7 ± 4.7 for HS, 26.4 ± 4.8 for SSS, and 26.0 ± 4.1 for CS (p = 0.019). The 13 additional variables, listed at the bottom of Table 3, not included in the propensity matching, had no significant differences between the three groups after they were matched (data not shown). Mean and median follow-up times for patients in the study were 25.2 ± 20.4 and 17.7 months, respectively.
|
Postoperative Dysphagia
Postoperative dysphagia of any degree not attributed to recurrence was significantly higher in HS anastomosis patients (SSS, 6 [26.1%]; CS, 5 [21.7%]; and HS, 13 [56.5%]; p = 0.04; Table 1). Postoperative severe dysphagia was not significantly different (SSS, 2 [8.7%], CS, 0; HS, 2 [8.7%]; p = 0.26).
Postoperative Anastomotic Stricture and Dilations
Stricture identified by endoscopy was also increased in the HS group with eight (34.8%), compared with two (8.7%) with SSS and two (8.7%) with CS (p = 0.04; Table 1). Three times as many patients requiring three or more dilations were in the HS group compared with the SSS or CS groups (6 [26.1%], 1 [4.3%], and 2 [8.7%], respectively; p = 0.10), but this was not statistically significant.
Other Outcomes Derived From Propensity-Matched and Unmatched Patient Populations
No difference was noted in perioperative mortality, length of stay, long-term survival, or locoregional recurrences between techniques (Table 4). Locoregional recurrence was not different among the unmatched or triplet matched groups. Those with a SSS, CS, or HS anastomosis in the unmatched group had a 23.3%, 19.2%, and 39.1% locoregional recurrence rate. Five-year survival was not statistically significantly different between SSS, CS, and HS, respectively, 0.23, 0.69, and 0.47 (p = 0.32). Because this technique was implemented in 2002, the SSS group does not have 5-year follow-up.
|
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Leak
We have shown that all three techniques have a low leak rate, both in our propensity-matched group analysis and in our entire patient population. The overall leak rate was 7% (n = 15) in the 214 patients in our study, which is similar to other recent published series ranging from 0% to 24% [5, 8, 16, 20–22]. In cervical anastomosis, a lower leak rate in the stapled anastomotic group compared to a HS group was recently reported by Behzadi [16] and Ercan [8] and their colleagues. Perhaps our focus on only intrathoracic anastomoses makes technique less significant for leak [14, 15]. The technique coupled with a higher location and possibly different tension on the anastomosis may accentuate differences between cervical and thoracic anastomoses.
Dysphagia
The development of postoperative dysphagia was doubled in the HS group compared with the stapling techniques. Only one patient with dysphagia also had a postoperative leak, and this was in the SSS group. When this patient is excluded from analysis, a statistically significant difference remains between the two stapled groups and the HS group (data not shown).
Postoperative dysphagia was defined in this study as any complaint of dysphagia. This may have overestimated clinically significant dysphagia and is the reason why we separately analyzed for severe dysphagia. Postoperative severe dysphagia was not different among the three groups.
The 22% to 56% incidence of any type of dysphagia in our study is comparable with the 23.7% to 73.3% range reported in other studies [21, 23]. Some studies reported dilating patients who had dysphagia, and because they did not all have endoscopy to confirm the presence of a stricture, the dysphagia and stricture data appear to have some degree of overlap. One patient with dysphagia in our cohort underwent dilation, but a stricture was not documented.
Martin and colleagues [24] indicated most patients after esophagectomy will have some degree of dysphagia. Some data appear to favor dilating patients with dysphagia even though they do not have anatomic or objective evidence of dysphagia [25]. The wide variability the published reports is partially the result of poor definitions of dysphagia and a lack of standardization in documentation.
Stricture
The development of anastomotic stricture was four times greater in patients having a HS anastomosis. The mean number of anastomotic dilations per patient with a stricture was 2.5 after a SSS, 4 after CS, and 3.5 after HS anastomosis (p = 0.12). Each stricture required an average of more than two dilations, but the difference in the number of dilations per stricture was not significant.
One of the more severe strictures did not respond to dilation until steroids were injected. Another severe stricture resulted from a postoperative leak. It is known that patients experiencing a postoperative leak will have an increased stricture rate [26]. Only one patient with a stricture also had a postoperative leak, and this was in the SSS group. Of note, this was the same patient who also reported postoperative dysphagia. If this patient is excluded from analysis, a statistically significant difference still remains between the groups for stricture.
All patients with an anastomotic stricture underwent at least one dilation, but other studies recorded a stricture only if patients required two or more dilations. If this criterion were applied to our entire cohort of 214 patients, 2 patients would be excluded from each stapled group (CS and SSS). This would increase the disparity between HS and stapled strictures in the unmatched analysis.
There was concern that the SSS technique was implemented more recently than other two groups, and this might result in a falsely low postoperative stricture rate. There was no statistical difference in the duration of follow-up documented when the HS group was compared with the CS or SSS groups (p = 0.07). Furthermore, most studies report most strictures appearing within the first 3 to 4 months after the procedure [5, 8, 27]. The 13.6% stricture rate reported from CS anastomoses in this study appears to be much lower than in published reports, which can be as high as 28% to 40% [10, 28].
Technique
One of the inherent difficulties in any study measuring the quality of a hand-sewn anastomosis is the variation in technique. Some institutions advocate a single-layer technique [8, 29] and others are in favor of a double-layer technique [16, 30, 31]. Nonrandomized studies in which both techniques were used have found no difference [32]. Other factors unrelated to the anastomosis can directly affect the quality of the blood flow to that area. Issues such as gastric conduit trauma, technique of stapling, and tension are also important uncontrolled factors in this study [33]. Poor vascular supply at the time of operation has also been found to predict postoperative stricture [34].
When the entire CS group was evaluated, the anvil size did not predict leak, dysphagia, or stricture. This finding is different from other published series, including two studies from Hong Kong [10, 27], finding the smaller CS anvil size to correlate with a higher stricture rate. A third study by Berrisford and colleagues [23] also reported the smaller staplers correlated with a higher stricture rate and found no significant difference between the two major manufacturers of these staplers.
In addition to surgical technique, neoadjuvant therapy is also an important consideration. The thoracic anastomosis, which is often closer to the field of radiation, becomes even more critical considering our rate of induction therapy is slightly higher than most reported in the literature [8, 16, 35]. Currently, there are reports of induction therapy not increasing morbidity or mortality [36], but this remains controversial.
Limitations
Inherent limitations with retrospective studies include loss to follow-up and small numbers in our matched comparison groups. By propensity-matching the patients and controlling for background characteristics between groups, we were able to create a more robust comparison than a linear regression analysis of retrospective data alone. It is important to note the propensity scores can only adjust for observed confounding covariates and not unobserved ones.
To prevent excluding patients from analysis, we balanced patients with missing data against each other, placing them in a similar category for those variables that were missing. A randomized prospective study would be superior to this, but in the absence of such a current study in the literature, this analysis may aid the clinician in deciding which type of anastomosis to create for intrathoracic reconstruction until such trials are conducted.
Techniques in esophageal surgery continue to change. A recent publication of 35 hybrid intrathoracic anastomoses performed with a minimally invasive approach reported an acceptable 6% leak rate [35]. The computerized Powermed stapling device (Powermed, Inc, Washington Crossing, PA), which can now be placed through the mouth, allows the surgeon to perform an intrathoracic anastomosis without device entry through the chest. Complications such as leak, dysphagia, and stricture as well as their association with technique and surgical approach still need to be evaluated in a prospective manner. The development of new techniques that may decrease the number of complications and the severity of such complications also needs to be continually evaluated.
Conclusion
In this carefully matched group of patients, intrathoracic use of the SSS esophagogastric anastomosis in esophageal cancer patients is safe and effective. Postoperative dysphagia and need for stricture dilation was lower using a stapled technique compared to a traditional HS anastomosis in our study.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
As thoracic surgeons, we are always trying to strive to do the best for our patients, especially those with esophageal cancer. As you know, the majority of the time we do not get the opportunity to operate on these patients because of advanced disease. But when we do, we must perform the operation with the utmost attention to detail to maintain the patients’ quality of life, especially their ability to eat. Therefore, the esophageal reconstruction, especially the anastomosis, component of the procedure is the most important part that guarantees the patient a near-normal quality of life.
I have several questions for you. First of all, I was surprised at the small percentage of hand-sewn anastomosis performed at your institution, which was only 10% in your retrospective series. I would think that percentage would be higher, somewhere between 70% and 80% in your earlier patients. Has the majority of your anastomosis always been stapled at Anderson, or has it changed because one of your more senior surgeons, Dr Putnam, took all of the silk sutures with him when he moved to Nashville?
Second question. There was no difference in the leak rate based on your anastomotic techniques, but the hand-sewn patients had an almost fourfold increase in stricture rate. How do you explain this significant increase even though the numbers of dilations required between the groups were the same?
Third, the hand-sewn patients also had significantly more dysphasia, but then again, there was no difference in the requirement for dilatation. How do you explain this increased dysphagia rate?
And finally, when you talk to the patients about dysphagia, is it related to the anastomosis or to the function of the gastric conduit? A lot of patients will have feelings of dysphagia and it is not related to anastomotic problems but related to gastric dysfunction. Do you have any information in regards to the medications they required for conduit problems, such as Reglan, erythromycin, and/or if they had any pyloric problems?
Thank you very much.
DR BLACKMON: Thank you, Dr Miller, for your comments. With the first question, regarding the low number of hand-sewn anastomoses, I think that has more to do with the time period that we chose to study. I tried to pick patients that had the most up-to-date data and the most complete data, and by using a more contemporary set of data, I excluded some of the earlier patients who probably would have been the majority of the hand-sewn patients. We had a feeling within the group that perhaps the stricture rate was high in the hand-sewn patients, and that was one of the reasons why the stapled side-to-side technique was implemented in our institution, and so there was a shift in paradigm as the stapled side-to-side technique was implemented in a more recent era.
And then regarding your comments about Dr Putnam and taking silk suture out of MD Anderson, when I went back and looked through all the different surgeons performing different techniques, Dr Putnam preferred to perform the transhiatal approach, and in our patients he performed only two of the surgeries in this study, one hand-sewn and one circular stapled, and from memory I don’t believe that he had any complications from either one of those, but he did in the majority of his cases do transhiatal.
For your second question regarding the leak rate being relatively the same and the stricture rate being so high in the hand-sewn groups and why I think that is the case, there was a significant increased number of dilations for the hand-sewn group both overall and in the unmatched group for severe strictures. I think part of the reason could be the conduit, and part of the reason could be from the actual double-layered sutured anastomotic technique. Explaining why there was no difference between leak rates, and yet stricture was different, is probably due to the double layer technique. Not all strictures were due to leaks. Clearly we had no differences in our leak rate, and so you have to have a tradeoff: if you do a double layer, you perhaps have a lower leak rate than if you did a single layer but perhaps also have a higher stricture rate.
Regarding the dysphagia differences between the groups, you are right, the severe dysphagia was not different among our matched population, but was different among our unmatched population. I think dysphagia was worse in the hand-sewn patients because of a bulkier anastomosis and more scar tissue.
And then the final dysphagia question, we didn’t have any known pyloric problems from our study that affected dysphagia, but I can go back and look at that again. We didn’t specifically look at that in our study. We made a point to quantify stricture as being seen on endoscopy or barium study and qualify it as the number of dilations being performed. Dysphagia was specifically graded by personal account and not whether or not it required dilation. We also did not specifically look at reflux. Thank you for your comments.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. E. Kight Nutrition Considerations in Esophagectomy Patients Nutr Clin Pract, October 1, 2008; 23(5): 521 - 528. [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 |
