Analysis of Unsuccessful Esophageal Stent Placements for Esophageal Perforation, Fistula, or Anastomotic Leak

Transcription

1 Analysis of Unsuccessful Esophageal Stent Placements for Esophageal Perforation, Fistula, or Anastomotic Leak Richard K. Freeman, MD, Anthony J. Ascioti, MD, Theresa Giannini, ARNP, and Raja J. Mahidhara, MD Department of Thoracic and Cardiovascular Surgery, St. Vincent Hospital, Indianapolis, Indiana Background. Esophageal stent placement for the treatment of a, anastomotic leak, or fistula has been adopted by some thoracic surgeons. Results have been reported for this technique, but little discussion has focused on treatment failures. This analysis reviews patients in whom stent placement was not successful in an attempt to identify factors that may increase the likelihood of failure of this technique. Methods. Patients undergoing stent placement for the treatment of an, anastomotic leak, or fistula in which the stent failed to adequately seal the leak were identified from a single institution s database. The anatomic location, chronicity, and cause of the leak were recorded using a newly developed classification system. Comparison was made to patients in whom stent placement was successful. Results. Over a 7-year period, 187 patients had an stent placed for leaks. Fifteen (8%) of these patients required traditional operative repair when the stent failed to resolve the leak after an average of 3 days. A comparison of the 2 patient groups found that stent failure was significantly more frequent in patients who had an leak of the proximal cervical esophagus, 1 that traversed the gastro junction, an injury longer than 6 cm, or an anastomotic leak associated with a more distal conduit leak (p < 0.05). Malignancy or previous radiation therapy was not associated with treatment failure. Conclusions. This investigation identified 4 factors that significantly reduce the effectiveness of stent placement for the treatment of, fistula, or anastomotic leak. These potential contraindications should be considered when developing a treatment plan for individual patients and may prompt traditional operative repair as initial therapy. (Ann Thorac Surg 2012;94:959 65) 2012 by The Society of Thoracic Surgeons Endoluminal stent placement for the treatment of, fistula, or intrathoracic anastomotic leak has been reported by our group and others [1]. When used as part of a comprehensive approach in selected patients, it has been shown to be a safe and effective technique, avoiding operative repair in the majority of patients. Such encouraging initial results have led to the adoption of this technique by some thoracic surgeons. As would be expected with the dispersion of a new technique, a variation in the reported results has occurred when compared with initial publications. This would include treatment failures. However it is often not possible for individual treatment failures to gain notoriety. Analyzing such treatment failures could identify commonalities that would be useful in modifying the indications and contraindications of an emerging technique. This investigation summarizes the results of an Accepted for publication May 11, Presented at the Forty-eighth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 28 Feb 1, Address correspondence to Dr Freeman, 8433 Harcourt Rd, Indianapolis, IN 46260; rfreeman@corvascmds.com. analysis of patients whose, fistula, or anastomotic leak was treated with an stent and who went on to require a traditional operative repair. Material and Methods Patients undergoing stent placement for the treatment of an intrathoracic anastomotic leak after esophagectomy or an or fistula were identified from a comprehensive general thoracic surgery database at a single institution. The institution s institutional review board approved this protocol and waived individual patient consent for this investigation. In contrast to our previous reports, patients with a or fistula involving an or gastric malignancy were included in this investigation. Excluded from analysis were patients receiving an stent for a recalcitrant stricture, patients with a benign or malignant obstruction without a leak, and patients undergoing stent placement at an outside facility and transferred for further care. Patients with an leak from any cause with a distal obstruction not 2012 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc

2 960 FREEMAN ET AL Ann Thorac Surg UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 2012;94: Table 1. Esophageal Leak Classification System Characteristic Location Cervical Intrathoracic Gastro junction Mechanism Iatrogenic Spontaneous (Boerhaave s syndrome) Anastomotic leak Fistula Injury size 6cm 6cm Associated malignancy Distal stricture at presentation Symbol C T G I B A F X Y M S hours after stent placement or when the patient was able to participate in the examination. In the absence of a continued leak, a soft mechanical diet is initiated. It was the intention to remove all patients stents after a sufficient amount of time to allow the leak to seal. Stent removal was again carried out in the operating room using general anesthesia. Flexible esophagoscopy was performed before and after stent removal. An esophagram was obtained after stent removal before discharge. Analysis of data was carried out using GraphPad Prism software 4.02 (GraphPad Software, San Diego, CA) for Windows (Microsoft Corp, Redmond WA). Continuous data are expressed as mean standard deviation except when otherwise indicated. Differences between categorical variables were evaluated by Fisher s exact test. Differences between continuous variables were measured by the 2-tailed Student s t test or the Mann-Whitney test for nonnormally distributed data. A value of p 0.05 was considered significant. able to be bridged with an stent are generally not offered stent placement in our program. A retrospective analysis was performed after eligible patients were segregated into 2 groups; patients in whom stent placement successfully sealed the leak (stent group) and patients in whom the stent was considered unsuccessful (failure group). Failure of the stent was defined by the need for operative repair of the or anastomotic leak after stent placement. Stent migration requiring replacement or repositioning was not considered a failure. A comparative analysis of the 2 groups was conducted. The location of anastomoses, chronicity, and cause of the leak were categorized using a classification system developed to facilitate the comparison of patients receiving an stent (Table 1). Also compared were patient demographic data, causes of stent failure, complications related to the use of an stent, stent migration, morbidities, and operative mortality. Patient Evaluation and Stent Placement The presence of an anastomotic leak, fistula, or was documented and localized by Gastrografin (sodium amidotrizoate and meglumine amidotrizoate) or barium esophagography, or both, before any treatment. To be considered a significant leak eligible for treatment other than observation, contrast had to be seen leaving the lumen of the esophagus, with extravasation into the mediastinum or pleural space. Additionally, all patients being considered for stent placement underwent computed tomographic imaging of the neck, chest, and abdomen. All stents were placed in the operating room by a thoracic surgeon using general anesthesia and fluoroscopy after flexible esophagoscopy. Adequate drainage of infected areas was also simultaneously achieved either by video-assisted thoracoscopy or imageguided percutaneous drainage. Leak occlusion was confirmed by contrast esophagography a minimum of 24 Results Between 2003 and January 2010, 236 patients underwent stent placement at the general thoracic surgery service of the study institution. Of these patients, 187 met the entrance criteria for this investigation (Table 2). This included 37 anastomotic leaks, 48 chronic fistulas, and 104 s. The group included 43 spontaneous s and 61 iatrogenic s. Of the iatrogenic s, 14 were in proximity to an (12) or gastric (2) malignancy. A Polyflex stent (Boston Scientific, Natick, MA) was placed in 172 (92%) patients, with the remaining patients treated with an AliMAXX-E stent (Alveolus Inc, Charlotte, NC). Stent choice was at the discretion of the surgeon. Included in the stent group were 11 (6%) patients whose initial esophagram at 24 hours after stent place- Table 2. Patients Included in This Investigation Variable Stent patients 187 Successful 172 Anastomotic leak 31 Chronic fistula 38 Perforation 103 Spontaneous 42 Iatrogenic 61 Associated with malignancy 14 Failed Anastomotic leak 15 Chronic fistula 3 Perforation 3 Spontaneous 9 Iatrogenic 4 N

3 Ann Thorac Surg FREEMAN ET AL 2012;94: UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 961 ment displayed a small amount of contrast that traversed the proximal margin of the stent and flowed between the outside of the stent and the mucosal surface of the esophagus (Fig 1). This included 6 iatrogenic s, 3 spontaneous s, and 2 intrathoracic anastomotic leaks. With no clinical evidence of a continued leak, esophagography was repeated 48 hours later and the contrast extravasation had resolved in all patients. These patients were then allowed to begin oral hydration and nutrition. Fifteen patients (8%) who underwent stent placement for an, fistula, or anastomotic leak during the study period required traditional transthoracic (11 patients) or transabdominal (4 patients) operative repair when the stent failed to resolve the leak at a mean of 3 days (median, 2 days; mode, 2 days) after initial stent placement (Fig 2). Reasons for failure of the stents were a leak site that was too proximal (4 patients) or distal (5 patients) for the stent to seat properly, an injury greater than 6 cm in length (3 patients), or a leak site other than the esophagogastrostomy after esophagectomy (3 patients). No significant differences were found in demographic data between the stent and failure groups, including previous external beam radiation therapy to the chest Fig 2. Computed tomographic image. An stent is seen in the lumen with continued pneumomediastinum and a left pleural effusion, implying that the stent has not sealed the. (Table 3). Mean length of stay did differ significantly between the stent and failure groups. When the length of stay in the failure group was adjusted to remove days from stent placement to failure recognition, patients requiring traditional operative repair still had a significantly longer mean length of stay than did patients treated with an stent. Stent migration requiring repositioning or replacement occurred in 29 (17%) patients in the stent group. Morbidities were found to occur significantly more frequently in patients requiring a surgical repair (p ) (Table 4). There were no complications associated with stent placement or removal in either group. No significant difference in the frequency of mortality was found between the stent and failure groups (p 0.6). Causes of death included exsanguination from an infected thoracic aortic graft (1 patient) and an aorto fistula (1 patient) in the stent group and adult respiratory distress syndrome (1 patient), multisystem organ failure after sepsis (1 patient), and a myocardial infarction (1 patient) in the failure group. Subgroup analysis of patients with a or fistula associated with an or proximal gastric malignancy found a higher rate of significant complica- Fig 1. Contrast esophagram showing oral contrast flowing between the outside of the stent and the mucosal surface of the esophagus. Table 3. Patient Demographics Variable Stent Group Failure Group p N Age (mean, y) Male sex 92 (53%) 8 (60%) 1 Radiation therapy 27 (16%) 4 (27%) 0.28 Hospital stay (mean, d) Adjusted hospital stay (mean, d) Adjusted hospital stay mean hospital stay after hospital days between admission and stent failure were removed in the failure group.

4 962 FREEMAN ET AL Ann Thorac Surg UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 2012;94: Table 4. Patient Morbidities Outcome tions when compared with the entire stent treatment group. Mean indwelling stent time was also significantly longer in this subgroup. Stent migration rates, however, were similar. No patients in this subgroup required conversion to operative therapy, nor were there any mortalities. A comparison of the stent and failure patient groups found that stent failure was significantly more frequent in patients who had an leak of the proximal cervical esophagus, an leak that traversed the gastro junction, an injury greater than 6 cm in length, or an anastomotic leak associated with a more distal conduit leak (Table 5). Comment Stent Group (%) Failure Group (%) Stent migration 29 (17) 0 Morbidity Deep venous thrombosis 3 (2) 1 (7) Pneumonia 3 (2) 2 (13) Respiratory failure 6 (4) 3 (20) Renal failure 2 (1) 2 (13) Atrial fibrillation 5 (3) 1 (7) Myocardial infarction 1 (0.6) 2 (13) Prolonged ileus 3 (2) 1 (7) Operative mortality 2 (1) 3 (20) A disruption of gastro continuity in the form of an, fistula, or intrathoracic anastomotic leak presents a challenge for even the most experienced thoracic surgeon. The lack of a serosal investment can make the surgical repair of these conditions difficult, a dilemma that is only compounded in the face of inflammation and infection. Thus despite significant advances in antimicrobial therapy, critical care, medical imaging, and surgical technique, operative morbidity and mortality after operative repair of an leak remains excessive and is associated with significant failure rate [2, 3]. An ideal treatment for an leak would be one that minimized the negative impact of the treatment on the patient while fulfilling the traditional goals of therapy: preventing further contamination of the mediastinum by direct closure of the, eliminating infection in the mediastinum and pleural spaces, maintaining the integrity of the gastrointestinal tract, and using enteral nutritional. The ability to produce a plastic prosthesis coated with silicone that is easy to insert and remove and can rapidly form an occlusive seal within the lumen of the esophagus led some investigators to implant these stents in selected patients as a temporary measure to treat an, anastomotic leak, or fistula [4]. The evolution of stent use in our practice began in patients who either were exceedingly high risk for the transthoracic repair of an leak or had undergone a previous operative repair that had failed [5]. The encouraging results of this report caused us to consider whether a similar approach would be feasible in the treatment of patients with an acute iatrogenic of the esophagus [6]. We then assessed the viability of using an stent in the treatment of patients with spontaneous or Boerhaave s syndrome [7]. Our most recent report outlines the results of using this same technique in patients with acute intrathoracic anastomotic leaks after esophagectomy [8]. Others have also reported their experiences using nonmetallic stenting in patients with, fistula, or anastomotic leak [1, 9, 10]. Such series reinforce the value of endoluminal stent placement for indications such as, fistula, or anastomotic leak. However the majority of these series are either small or combine patients with different indications for stent placement into a single group, which may obscure the results of this technique. Despite the aforementioned publications and the occasional discussion of stent failures in regional and national meetings, only 2 published reports describing nonmetallic stent failures used in the treatment of leaks are found in the thoracic surgery literature [11, 12]. Each is valuable in its dissemination of a lesson learned and suggestions to avoid a similar outcome in the future. The report by Odell and DeVault [11] outlines the clinical course of a 64-year-old man whose spontaneous was treated with multiple stents despite a persistent leak. Of concern is the fact that the initial stent placed was not a covered stent, making it unlikely to be of benefit in such a patient. The authors, who did not provide the patient s initial care, rightly recommended an exit strategy for patients in whom initial stent placement is not successful in rapidly sealing an leak. The second publication by Whitelocke and colleagues [12] reports the case of a patient treated with an stent for a persistent anastomotic leak after esophagectomy. Six weeks after esophagectomy, the patient was found to have a fistula between the thoracic aorta and the intrathoracic gastric conduit. Despite a successful operative repair, the patient eventually died of multisystem organ failure. This report highlights the radial force that an stent can exert on adjacent structures. Table 5. Comparison of Failure Group Characteristics Characteristic Stent Group Failure Group p Proximal/cervical location Gastro junction location Injury 6 cm Conduit leak Associated malignancy

5 Ann Thorac Surg FREEMAN ET AL 2012;94: UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 963 A similar event in a patient cared for at our center and reported in this investigation, has caused us to significantly reduce the time we leave stents in place. Failure analysis, the practice of reviewing unsuccessful methods or outcomes, is a common tool in some engineering disciplines. It has been found to recognize flawed or unsafe processes or products earlier. Analysis of failures also allows adjustments that improve safety and outcomes in a more rapid fashion. This investigation sought to identify common themes of failure within a relatively large group of patients undergoing stent placement for, fistula, or intrathoracic anastomotic leak. It found that stent placement for these indications is safe and effective in the vast majority of patients. Failures were attributable to 4 mechanical factors and were not related to failures of the integrity of the stent, incorrect placement, complications related to the stent, or stent migration. Some important differences between the stent group and the failure group were recognized. Mean length of hospital stay was significantly longer in patients who required a traditional operative repair when compared with patients treated with stent placement. This remained true even when length of stay in patients in whom stent placement failed was reduced to account for stent placement and failure. Similarly, significant morbidities were more common in the stent failure group as well. Although this is by no means an optimal comparison, it is the first comparison of stent placement and surgical treatment for these disorders in the literature. When the costs of operation and additional length of stay are considered, it is likely that successful stent placement is a lower cost option for the treatment of these disorders when compared with traditional operative therapy even when the costs of stent migration are included. The classification system of injury and leaks adopted in this report was of benefit in this analysis. It was developed specifically to allow better comparisons of patients with, fistula, or anastomotic leak. The classification system was based on the authors experience with a relatively large number of these patients as well as an understanding of the difficulty of comparing series of similar patients in the recent surgical literature. It is built on the foundation of previous work by Dr Paul Samson [13], as well as a classifica- No Perforation Further evaluation Cervical Proximal Primary operative repair and drainage Distal Failure of repair Suspected Esophageal Perforation History and physical examination CT neck/chest/abdomen Esophagram (Gastrograf in and/or Barium) Intrathoracic (< 6 cm in length) Hybrid Stent Repair Intrathoracic (> 6 cm in length) Transthoracic repair Failure of repair Intra abdominal and/or gastric Transabdominal repair & jejunostomy Fig 3. Treatment algorithm for. (CT computed tomography.)

6 964 FREEMAN ET AL Ann Thorac Surg UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 2012;94: tion system developed for injury from foreign bodies [14]. Any measure of success in treating patients with an injury found in this review results from our adherence to several basic principles. First and foremost, the thoracic surgeon must be involved in the care decisions for patients with. This is important both to identify contraindications for stent placement, in which immediate operative repair is undertaken, and to rapidly identify and treat stent failures. It is also important to ensure that these patients receive the level of care their condition requires, including aggressive antimicrobial and nutritional therapy, surveillance for continued areas of infection, and critical care support. Although it represents a relatively large number of patients with leak treated with stents, this investigation has some weaknesses. Specifically, although significant differences were found between the stent and failure groups, this is a retrospective study performed at a single institution. Furthermore, because of the small number of treatment failures, insufficient power exists to allow regression analysis to be performed. In conclusion, this investigation found that stent placement was safe and successful in treating the vast majority of patients presenting with, fistula, or anastomotic leak. However, 4 factors were identified that significantly reduce the effectiveness of stent placement in treating these disorders. These potential contraindications should be considered when developing a treatment plan for individual patients and may prompt traditional operative repair as initial therapy (Fig 3). References 1. Leers JM, Holscher AH. Stenting for Esophageal Perforation and Anastomotic Leak. In: Ferguson MK, ed. Difficult Decisions in Thoracic Surgery: An Evidence-Based Approach. 2nd ed. London, England: Springer-Verlag;2011: Brinster CJ, Singhal S, Lee L, Marshall MB, Kaiser LR, Kucharczuk JC. Evolving options in the management of. Ann Thorac Surg 2004;77: Minnich DJ, Yu P, Bryant AS, Jarrar D, Cerfolio RJ. Management of thoracic s. Eur J Cardiothorac Surg 2011;40: Mumtaz H, Barone GW, Ketel BL, Ozdemir A. Successful management of a nonmalignant with a coated stent. Ann Thorac Surg 2002;74: Freeman RK, Ascioti AJ, Wozniak TC. Postoperative leak management with the Polyflex stent. J Thorac Cardiovasc Surg 2007;133: Freeman RK, Van Woerkom JM, Ascioti AJ. Esophageal stent placement for the treatment of iatrogenic intrathoracic. Ann Thorac Surg 2007;83: Freeman RK, Van Woerkom JM, Vyverberg A, Ascioti AJ. Esophageal stent placement for the treatment of spontaneous s. Ann Thorac Surg 2009;88: Freeman RK, Vyverberg A, Ascioti AJ. Esophageal stent placement for the treatment of acute intrathoracic anastomotic leak following esophagectomy. Ann Thorac Surg 2011; 92: Blackmon SH, Santora R, Schwarz P, Barroso A, Dunkin BJ. Utility of removable covered self-expanding metal stents for leak and fistula management. Ann Thorac Surg 2010;89: Dubecz A, Watson TJ, Raymond DP, et al. Esophageal stenting for malignant and benign disease: 133 cases on a thoracic surgical service. Ann Thorac Surg 2011;92: Odell JA, DeVault KR. Extended stent usage for persistent leak: should there be limits? Ann Thorac Surg 2010;90: Whitelock D, Maddaus M, Andrade R, D Cunha J. Gastroaortic fistula: a rare and lethal complication of stenting after esophagectomy. J Thorac Cardiovasc Surg 2010;140:e Samson PC. Injuries and wounds of the esophagus; a classification. Calif Med 1954;80: Chang S, Cheng BC, Huang J, Mao ZF, Wang TS, Xia J. Classification and surgical treatment of intrathoracic injury caused by foreign body [article in Chinese]. Zhonghua Wai Ke Zhi 2006;44: DISCUSSION DR DAVID COOKE (Sacramento, CA): I have 2 questions. I assume these are self-expanding partially-covered metallic stents. DR FREEMAN: The majority of these are actually Polyflex polyester silicone-coated stents. About 8% to 10% of them are metallic-coated stents. DR COOKE: At least in the metallic stents, the 1 thing I have noticed is stent migration. Do you have any tricks or clinical pearls to avoid stent migration? DR FREEMAN: The only thing I would tell you is that I have found that the Polyflex stent, although harder to put in, migrates less because we re able to oversize it. I m not paid by any particular company, but we find that it migrates less when it s used in an oversized fashion, both in diameter and length, and we tend not to leave them very long now either. DR DONALD LOW (Seattle, WA): I enjoyed your presentation. We found much the same thing regarding the evolution in the treatment of these perforated patients with more of them being appropriate for a nonsurgical approach, stent placement being 1 option. There are 2 issues I would like you to talk about. Most of your stents are Polyflex. Are you using some of the selfexpanding metal stents? Some of them are completely covered and some of them are partially covered, and there are perceived advantages regarding the pros and cons of the 2 regarding migration. Also, are you using any fixation techniques, such as clips, et cetera? Second, the patient population that we have found most challenging has been the Boerhaave s patients who have a normal esophagus without a stricture and we find we just can t get stents to stay in place. Maybe you can give us some guidance about that patient population. DR FREEMAN: Well, to the first question, we have several stents available. It s really surgeon preference and condition preference. I don t feel comfortable placing a metallic stent across an anastomosis, for instance. I like the plastic stent for that. We don t generally use clips or any other fixation

7 Ann Thorac Surg FREEMAN ET AL 2012;94: UNSUCCESFUL ESOPHAGEAL STENT PLACEMENT 965 devices. We do oversize in length, as we said, and I think that that helps. As far as the Boerhaave s patients, we did report a series of those patients, and they were challenging patients. They are often very sick when they get to us. They do have a higher migration rate in our experience. Some of those patients we will say out of hand are not stent candidates if we can tell by the esophagram that the or the tear extends significantly onto the stomach. Now, if it does not, we have had some success in placing a long stent, of which the majority is in the esophagus, and giving those patients about a week with that in place. DR KAMAL KHALIL (HOUSTON, TX): I have 2 questions. We had a case of neoadjuvant chemotherapy, for adenocarcinoma of the middle third, that had a stent placed midway in the radiotherapy course of 45 Gy. When we went to do the Ivor-Lewis procedure, we found a big mediastinal abscess that the stent has been sitting in, with more or less dissolution of the wall. Have you had this kind of problem, and how do you deal with it? DR FREEMAN: We have seen it. My guess is that that probably was not a covered stent. DR KHALIL: It was a Polyflex-covered stent. DR FREEMAN: It was a Polyflex? DR KHALIL: Yes, sir. DR FREEMAN: I have not seen that, mainly because if we place a Polyflex for dysphagia, and we have placed quite a few of both kinds of stents in patients who are undergoing neoadjuvant therapy for dysphagia, we do not leave them in very long, and that seems to be a common theme I ve heard from other centers who have had a small collection of patients transferred to them who developed TE fistulas, for example, after esophagogastrectomy. Again, I think that that has to do with the length that that stent is in place, because it does have significant radial force, so at that point, there is not much you can do except deal with what you ve found, but those all go to our continuing learning curve and leaving these in for as little a time as possible. We re not trying to let the esophagus heal. We re simply letting it seal to whatever is juxtaposed against it. DR KHALIL: The second question, when we do operate on patients who have s of more than 24 hours, we end up picking pieces of onion peels and food fragments and all kinds of other things. What adjuvants would you recommend in addition to sealing the with a stent? DR FREEMAN: We are very up front about saying that those patients need a hybrid approach and should be cared for by a thoracic surgeon, because we place the stent and then we do a video-assisted pleural drainage procedure, open the mediastinum, and try and clean out as much of that material as we can.