- Email: [email protected]
Evolving Management Strategies in Esophageal Perforation: Surgeons Using Nonoperative Techniques to Improve Outcomes
Evolving Management Strategies in Esophageal Perforation: Surgeons Using Nonoperative Techniques to Improve Outcomes Madhan Kumar Kuppusamy, MD, Michal Hubka, MD, Chance D Felisky, MD, Philip Carrott, MD, Elizabeth M Kline, MD, Richard P Koehler, MD, FACS, Donald E Low, MD, FACS, FRCS(C) Management of acute esophageal perforation continues to evolve. We hypothesized that treatment of these patients at a tertiary referral center is more important than beginning treatment within 24 hours, and that the evolving application of nonsurgical treatment techniques by surgeons would produce improved outcomes. STUDY DESIGN: Demographics and outcomes of patients treated for esophageal perforation from 1989 to 2009 were recorded in an Institutional Review Board–approved database. Retrospective outcomes assessment was done for 5 separate time spans, including timing and type of treatment, length of stay (LOS), complications, and mortality. RESULTS: Eighty-one consecutive patients presented with acute esophageal perforation. Their mean age was 64 years, and 55 patients (68%) had American Society of Anesthesiologists levels 3 to 5; 59% of the study population was referred from other hospitals; 48 patients (59%) were managed operatively, 33 (41%) nonoperatively, and 10 patients with hybrid approaches involving a combination of surgical and interventional techniques; 57 patients (70%) were treated ⬍24 hours and 24 (30%) received treatment ⬎24 hours after perforation. LOS was lower in the early-treatment group; however, there was no difference in complications or mortality. Nonoperative therapy increased from 0% to 75% over time. Nonsurgical therapy was more common in referred cases (48% vs 30%) and in the ⬎24 hours treatment group (46% vs 38%). Over the period of study, there were decreases in complications (50% to 33%) and LOS (18.5 to 8.5 days). Mortality for the entire series involved 3 patients (4%): 2 operative and 1 nonoperative. CONCLUSIONS: Results from our series indicate that referral to a tertiary care center is as important as treatment within 24 hours. An experienced surgical management team using a diversified approach, including selective application of nonoperative techniques, can expect to shorten LOS and limit complications and mortality. (J Am Coll Surg 2011;213:164–172. © 2011 by the American College of Surgeons) BACKGROUND:
highly dependent on the experience of the managing physicians and institutional clinical resources. Management strategies vary from observation and supportive measures for early-diagnosed small-contained injuries to more complex multidisciplinary management in critically ill patients with established mediastinitis and sepsis. Historically, outcomes appear to be highly affected by the ability to rapidly and accurately diagnose the site and severity of perforation and initiate treatment within the first 24 hours. This hallmark time period dictated both the ability to perform primary repair and was found to be closely associated with posttreatment levels of morbidity and mortality.2 As a tertiary referral center for esophageal disease, we are referred a wide variety of acute perforations, many of which have had a delay in diagnosis and management or have undergone initial therapy in other institutions. Our ap-
Acute esophageal perforation has historically been associated with a high mortality rate ranging from 4% to 80%.1 This potentially life-threatening clinical condition has a diverse presentation, and its diagnosis and management are
Author Disclosure Information: Nothing to disclose. Editor Disclosure Information: Nothing to disclose. Presented at the Western Surgical Association 118th Scientific Session, Chicago, IL, November 2010. Received December 7, 2010; Revised January 15, 2011; Accepted January 21, 2011. From the Section of General Thoracic Surgery, Virginia Mason Medical Center, Seattle, WA. Correspondence address: Donald E Low, FACS, FRCS(C), Section of General Thoracic Surgery, Virginia Mason Medical Center, 1100 Ninth Ave. C6-SUR, Seattle, WA 98111. email: [email protected]
© 2011 by the American College of Surgeons Published by Elsevier Inc.
164
ISSN 1072-7515/11/$36.00 doi:10.1016/j.jamcollsurg.2011.01.059
Vol. 213, No. 1, July 2011
Kuppusamy et al
Evolving Management in Esophageal Perforation
165
Acute esophageal perforation
Unstable / Septic
Clinically stable
Water soluble / Barium contrast study (Observed by managing team)
If feasible
Non-Intubated
CT scan followed by EGD by managing team
Followed immediately by CT scan
Contained perforation
Decision favors observation
Contained perforation or Limited free perforation
Extensive contained perforation or free perforation
Decision favors endoscopic therapy
Decision favors surgical therapy
NPO Antibiotics / Anti-fungals ± NG Tube ± Nutritional support
Intubated
Decision favors endoscopic therapy or patient too unstable for surgery
EGD in O.R. prior to positioning Leave scope in place during operative procedure
If primary repair, review options for additional endoscopic therapy
Endoscopic clips / stents Trans-esophageal drainage Enteric feeding tubes ± Additional surgical or interventional radiology drains
Figure 1. Therapeutic algorithm for management of acute esophageal perforation. CT, computerized tomography; EGD, esophagogastroduodenoscopy; NG, nasogastric; NPO, nil per os; OR, operating room.
proach to the management of acute esophageal perforation has evolved over the period of study. Under surgical leadership, increasing, but selective, applications of nonsurgical modalities, both endoscopic and involving interventional radiology, have been used. We hypothesized that although early detection and treatment remains important, ultimate results are more closely associated with the experience of the initial management team rather than an arbitrary time period. We also contended that the appropriate increased use of nonsurgical and hybrid treatment approaches has contributed to improved outcomes over the study period.
METHODS Patients presenting to our tertiary referral residency training center with acute esophageal perforation between June 1989 and March 2009 were prospectively entered in an Institutional Review Board–approved and monitored database and retrospectively reviewed. This included patients initially diagnosed and treated at our institution as well as those referred after initial diagnosis of perforation elsewhere. All patients with acute iatrogenic or spontaneous
benign or malignant perforations were entered into the study. However, patients with spontaneous malignant perforations or fistulas were excluded. Patient demographics, pertinent history, timing and characteristics of presentation and diagnosis, management techniques, and outcomes were reviewed. American Society of Anesthesiologists (ASA) levels for all nonoperative patients were assigned by a single blinded anesthesiologist. All interventions, whether surgical, gastorintestinal, or involving interventional radiology, were recorded. Management was multidisciplinary but always under the supervision of the surgical team. This included endoscopic interventions, which were performed either by the surgical team or by a gastroenterologist with a member of the surgical team present. The therapeutic algorithm for management of acute esophageal perforation at our institution is depicted in Fig 1. Assessment of demographics and types of management was carried out over the entire study period, and results were also assessed separately in 5 4-year periods over the course of the study to assess the evolution of
166
Kuppusamy et al
Evolving Management in Esophageal Perforation
treatment techniques and outcomes over time. Treatment was assessed according to whether it involved surgical or nonoperative techniques; patients who received hybrid treatment approaches (involving both surgical and nonoperative modalities) were also reviewed but categorized as surgical procedures. Descriptive statistics were used to report continuous variables with range or standard deviation (SD), and categoric variables were reported as frequencies with percentages. Between-group comparisons of categoric outcomes were performed using Fisher exact test and chi square test as appropriate to the distribution of the data, with p values denoting levels of statistical significance (p ⫽ 0.05). Multivariate logistic regression was performed with length of stay and occurrence of complications as dependent variables. Statistical analyses were performed using SPSS (version 18) software package (SPSS, Chicago, IL).
RESULTS Between June 1989 and March 2009, 81 patients presented to Virginia Mason Medical Center with acute esophageal perforation. Their mean age was 63.6 years (range 22 to 96 years) and included 58 men and 23 women. Fifty-nine percent of patients were referred after initial diagnosis at another institution, with 8 of the 46 referred patients having initial surgical6 or endoscopic therapy2 at another institution. Sixty-eight percent of all patients were graded as ASA levels 3, 4, or 5 at the time of presentation. Etiology of perforation involved 51 iatrogenic (63%), 24 spontaneous Boerhaave-type barogenic perforations (30%), 4 spontaneous (5%), and 2 foreign body induced (2%). The iatrogenic perforations were most often associated with esophageal dilation (n ⫽ 27; 53%), with pneumatic dilation for achalasia being the most common dilation technique, involved in 14 patients. Therapeutic
J Am Coll Surg
endoscopy (eg, sclerotherapy, stenting, removing food impaction) accounted for 7 patients, diagnostic endoscopy for 5, nonesophageal surgical procedure accounted for 6, rigid esophagoscopy for 1, transesophageal echocardiography for 1, endoscopic ultrasound for 1, endoscopic retrograde cholangiopancreatography for 1, and self-dilation for 2. Spontaneous perforation, without a known etiology, was present in 4 patients. Iatrogenic endoscopic injuries resulting in perforation were identified at the time of injury in 64% of cases. Initial investigations involved upper gastrointestinal contrast studies in 75%. We recommend that a member of the surgical team witness this investigation first-hand to increase accuracy of interpretation and aid in treatment planning. Computerized tomographic (CT) scans were used in 46 patients (56%), and upper endoscopy was done in 64% at some point during their workup and management. Use of both CT and endoscopy increased over time. The application of endoscopy as a component of diagnosis or management of acute esophageal perforation increased throughout the study period, being used in 3 patients in the first 4 years (38%) but in 80% of cases in the last 5 years (Table 1). The most common location of injury was distal esophagus in 75% followed by cervical in 20% and midesophagus in 5%. Length of perforations averaged 2.7 cm (median 2 cm, range ⬍1 to 13 cm). Treatment was initiated within 24 hours in 57 patients (70%). Patients having treatment ⬎24 hours after perforation were more likely to be referred (79% vs 51%; p ⫽ 0.018) and have ASA levels 3–5 (83% vs 61%; p ⫽ 0.095; see Table 2). Median length of hospital stay was longer in patients treated after 24 hours compared with patients treated within 24 hours (20 vs 11 days; p ⫽ 0.003). However, there was no significant difference in complication
Table 1. Evolution of Management over the Study Period (1989 to 2009) n Operative treatment, n (%) Nonoperative treatment, n (%) Median LOS, d (range) ASA 3–5, n (%) Primary cases, n (%) Referred cases, n (%) UGI used in diagnosis, n (%) CT used in diagnosis, n (%) EGD used in management, n (%) ⬍24 h to Diagnosis, n (%) ⱖ24 h to Diagnosis, n (%)
1989–1992
1993–1996
1997–2000
2001–2004
2005–2009
Total
8 8 (100) 0 (0) 18.5 (9–81) 6 (75) 4 (50) 4 (50) 8 (100) 1 (12) 3 (38) 6 (75) 2 (25)
19 15 (78) 4 (22) 12 (4–87) 14 (74) 6 (32) 13 (68) 14 (74) 3 (16) 12 (63) 8 (43) 11 (57)
12 7 (58) 5 (42) 13.5 (4–86) 10 (83) 6 (50) 6 (50) 10 (83) 6 (50) 6 (50) 10 (83) 2 (27)
18 12 (67) 6 (33) 15 (3–73) 12 (67) 10 (56) 8 (44) 10 (56) 13 (72) 11 (61) 14 (78) 4 (22)
24 6 (25) 18 (75) 8.5 (1–41) 13 (54) 7 (29) 17 (71) 19 (79) 23 (96) 20 (80) 19 (79) 5 (21)
81 48 (59) 33 (41) 12 (1–87) 55 (68) 33 (41) 48 (59) 61 (75) 46 (57) 52 (64) 56 (69) 25 (31)
ASA, American Society of Anesthesiologists; LOS, length of stay.
p Value
⬍0.001 0.094 0.43 0.37 0.14 ⬍0.001 0.12 0.048
Vol. 213, No. 1, July 2011
Kuppusamy et al
Table 2. Patient Characteristics and Outcomes by Timing of Treatment Time to treatment
<24 h
>24 h
p Value
Patients, n (%) ASA 3–5, n (%) Referred, n (%) Median LOS, d Operative, n (%) Complications, n (%) Mortality
57 (70) 35 (61) 29 (51) 11 35 (61) 20 (37) 1
24 (30) 20 (83) 19 (79) 20 13 (54) 11 (41) 2
0.095 0.018 0.003 0.55 0.36 0.43
ASA, American Society of Anesthesiologists; LOS, length of stay.
rates (40.9% vs 37.2%; p ⫽ 0.36) or mortality (2 vs 1; p ⫽ 0.43). Mortalities in the 2 patients receiving treatment ⬎24 hours after perforation involved a 78-year-old patient transfered intubated and septic 2.5 days after iatrogenic perforation. She was treated with a hybrid approach involving right thoracotomy, mediastinal drainage, and decortications with the simultaneous insertion of an endoscopic esophageal stent. She recovered and left the intensive care unit, but died on postoperative day 19 of cardiac arrest secondary to complete heart block. The second patient was transfered to our hospital septic, with Fournier gangrene and Boerhaave syndrome, for hyperbaric oxygen therapy for the Fournier gangrene. Transfer was 15 days after the perforation; although a chest tube was placed on recognition of pneumothorax at the previous hospital, it had not been recognized as an esophageal perforation until a swallow study was performed at our institution. He died on hospital day 86 when care was withdrawn. The third mortality was in a 74-year-old patient on peritoneal dialysis, who experienced an episode of vomiting after an endoscopy for abdominal pain and underwent immediate laparotomy for free intraperitoneal air at another center. Attempted transabdominal repair at the other hospital led to splenectomy and 2-liter blood loss with an ongoing leak, which led to transfer to our institution. The patient recovered, but refused further dialysis; he ultimately died on hospital day 51. Operative treatment was used in 48 patients (59%) with 33 (41%) treated without surgery. Table 3 highlights the treatments used in both approaches. ASA levels 3–5 were more common in operative than in nonoperative patients (75% vs 58%; p ⫽ 0.099; Table 4). Primary repair was the most common surgical treatment, in 72%. Surgical approach was transthoracic in 59%, transabdominal in 37%, and combined thoracoabdominal in 4%. Hybrid approaches involving sequential or synchronous interventional and surgical procedures were carried out in 10 patients (21%). These procedures included endoscopic stents placed at the time of primary repair or stents (5) and endoscopically placed nasomediastinal drains (1) placed in asso-
Evolving Management in Esophageal Perforation
167
ciation with open or thoracoscopic pleural decortication and drainage. The evolution of presentation and treatment is shown in Table 4. Typically, ⬃50% of cases were referred, except for the last 5 years, when 71% came from other institutions. Table 3 demonstrates that nonoperative therapy was more common in referred cases than in primary cases (48% vs 30%) and in those whose treatment was delayed (⬎24 hours) compared with patients treated early (⬍24 hours; 46% vs 38%, respectively). Patients undergoing treatment ⬎24 hours after perforation have varied moderately over time from 21% to 57% of patients. Figure 2 demonstrates in graphic form that the issues that underwent the most pronounced changes over time were patients undergoing nonsurgical management, which increased from 0 to 75% (p ⬍ 0.001). Over the same 20-year period, complications decreased from 50% to 33% (p ⫽ 0.94) and median length of stay decreased from 18.5 to 8.5 days (p ⫽ 0.094). Table 5 demonstrates that there was no statistical difference in mortality, persistent leak, or major medical complications of sepsis, pneumonia, myocardial infarction, dysrhythmia, renal failure, or deep vein thrombosis between operative and nonoperative populations. Multivariate logistic regression was used to analyze Table 3. Details of Operative and Nonoperative Treatments Treatment
n
Operative treatment (n ⫽ 48) Intraoperative endoscopy Surgical drainage and stent Resection Early Late Primary repair Concurrent procedures Buttressed Antireflux procedures Intraoperative stent T-tube Nonoperative treatment (n ⫽ 33) Endoscopic stenting* Total parenteral nutrition Dobhoff Gastrostomy Feeding jejunostomy Radiologic drainage procedures Endoscopic mediastinal drainage Foreign body removal Endoscopic repair (clip repair-3; glue-2)
20 3 5 3 3 34 25 6 4 2 11 7 5 5 3 11 2 2 5
*Polyflex, 5; Ultraflex, 2; Dua, 1; Celestin, 1; Alimaxx, 1; self-expanding metal stent, 1.
168
Evolving Management in Esophageal Perforation
Kuppusamy et al
J Am Coll Surg
Table 4. Patient Characteristics and Outcomes by Treatment Group Operative treatment
Patients ASA 3–5 Referred Median LOS, d Treatment ⬎24 h Complications Mortality
n
%
48 36 25 15.5 13 21 1
59 75 52
Nonoperative treatment n %
33 19 23 10 11 10 2
27 44
41 58 70 33 30
p Value
0.099 0.11 0.002 0.54 0.22 0.74
ASA, American Society of Anesthesiologists; LOS, length of stay.
method and timing of treatment, presentation (ie, referral or primary), era, source of perforation, diagnostic methods, and ASA level with length of stay and occurrence of complications as the covariates. Of these, increased ASA level (p ⫽ 0.016), presentation ⬎24 hours (p ⫽ 0.038), use of endoscopy (p ⫽ 0.01), and occurrence of complications (p ⫽ 0.006) significantly accounted for variations in length of stay. Only ASA status was significant regarding occurrence of complications (p ⫽ 0.005).
management continue to evolve. Historically, authors have advocated the efficacy of either operative,3,4 or nonoperative management5 of esophageal injuries. In the present study, we found that, in our experience, the management of esophageal perforation in an era of rapid technologic progress continues to change. Part of this change is expressed in the increased application of hybrid-type procedures where the advantages of surgical and interventional endoscopy and radiology techniques are applied. We believe that improving outcomes, including lowering historic levels of mortality, morbidity, and ultimately cost, are dependent on the patient receiving appropriate work-up and tailored management by an experienced tertiary-care surgical team
DISCUSSION The optimal management of esophageal perforation remains controversial, and the options for intervention and 80
Complicaons Non-Operave Treatment Length of Stay (Days)
70
75%
60 50
50%
42% 37%
40
39% 42%
33%
30 20
33% 18.5 21%
15
13.5
8.5
10
12 0%
0 1989-1992 (n=8)
1993-1996 (n=19)
Complicaons (%) Non-Operave Treatment (%) Median LOS (Days)
1989-1992 (n=8) 50 0 18.5
1997-2000 (n=12)
2001-2004 (n=18)
1993-1996 1997-2000 (n=19) (n=12) 37 42 21 42 12.0 13.5
2005-2009 (n=24)
2001-2004 2005-2009 (n=18) (n=24) 39 33 33 75 15.0 8.5
p 0.94 < 0.001 0.094
Figure 2. Graphic description of management and complications over time. Nonoperative treatment has increased significantly, and complications and length of stay (LOS) have decreased, though not to a statistically significant degree.
Vol. 213, No. 1, July 2011
Kuppusamy et al
Table 5. Complications by Treatment Group
Complication
Pneumonia Dysrhythmias Deep vein thrombosis Persistent leak (clinical or radiologic evidence of leak at 14th day) Stent migration Sepsis Renal failure Death
Operative treatment (n ⴝ 47) n %
Nonoperative treatment (n ⴝ 34) n %
7 11 1
14 23 2.0
4 4 1
11 11 2.9
2 0 0 0 1
4.3 0.0 0.0 0.0 2.1
3 3 1 1 2
8.8 27.3 2.9 2.9 5.9
All p values ⬎0.05.
facile with all aspects of surgical and nonoperative management modalities. We have developed a treatment algorithm for esophageal perforation at our institution, which is now used routinely to aid in treatment decision making (Fig. 1). Appropriate work-up of esophageal perforation has typically included an upper gastrointestinal contrast study. We emphasize the importance of the presence of a surgical team member at the time of this study to ensure that the study is complete and provides the anatomic details often missed when the radiologist performs the study alone. This can have a major effect on the surgeon’s impression of the magnitude and location of the perforation and can affect treatment planning. This is also the justification we have used for repeating contrast studies when transfered patients arrive with ambiguous studies. CT scanning allows an additional evaluation of the severity of mediastinal or pleural space contamination. Table 1 illustrates that the use of CT scanning increased from selective (12%) to routine (96%) in medically stable patients. Iatrogenic esophageal injuries remain the most common etiology, which is consistent with earlier studies.1,6,7 Earlier reports have highlighted the importance of endoscopy in recognition of injury at the time of the event.6 In the early stages of the present study, endoscopy was rarely used; however, in the last 5 years, 80% of patients with esophageal perforation underwent upper endoscopy, and it is now a routine part of the work-up in almost all patients. We have not identified any significant complications or problems with using endoscopy in patients with acute esophageal perforation. We have also found endoscopic examination to be invaluable regarding initial treatment decision making by identifying the exact localization of the perforation and assessing secondary pathology and viability of esophageal mucosa. We have used upper endoscopy intraoperatively in
Evolving Management in Esophageal Perforation
169
44% of cases guiding surgical repair. This approach is particularly helpful because the mucosal and muscular defects can often be spatially separated and primary closure of the external esophageal layers does not result in mucosal approximation. Moreover, mediastinal contamination can obscure anatomy, making exact intraoperative visualization of the level and extent of perforation challenging, and the in-place endoscope can help guide the surgeon directly to the mucosal defect. Intraoperative endoscopy can also facilitate dilation of persisting strictures, removal of foreign bodies, and placement of endoscopic stents and drains when appropriate. In our series, the number of patients treated ⬎24 hours after injury has varied only moderately over the period of this study. In part, this is due to the high percentage of patients referred to our institution (50% to 75% over the past 20 years).Operative management has historically been the mainstay of treatment, with the first 24-hour period being the hallmark time metric of good outcome.1,2 In a study of 75 patients, Muir and colleagues3 found increased mortality from 5% to 44% in patients treated before and after this hallmark time period. However, our results suggest that the initial evaluation by an experienced team able to provide tailored management appears to be more important than an arbitrary time period related to onset of injury. We have found that patients presenting ⬎24 hours after perforation are more likely to be referred and be unstable, as reflected by a higher ASA score (Table 2). These patients demonstrate a longer length of stay; however, we did not find any differences in minor or major complications or in mortality compared with patients treated within the first 24 hours after injury. This finding supports the concept of transfering appropriate patients with perforations to highvolume esophageal centers, even when the transfer will further delay definitive treatment. Examination of the trends in management of esophageal perforation in the present study showed that nonoperative management increased dramatically (0 to 75%) over the study period. In association with this trend toward nonoperative management, lengths of stay and complications declined (Fig. 2). Improvement in results in a study spanning 20 years is to be expected. However, the largest single change in management approach has been the introduction and growth of nonoperative and hybrid treatment techniques. We first used endoscopic stents in 1999 (Polyflex; Boston Scientific), and have used a variety since that time (Table 3); typically we place fully covered removable stents. The safety and success of nonoperative management has been previously highlighted in reports by our group8 and Vogel and associates.5 Multidisciplinary care is not just about an experienced surgical team but is also related to the
170
Kuppusamy et al
Evolving Management in Esophageal Perforation
J Am Coll Surg
Table 6. Large Case Series of Esophageal Perforation First author
Study period
n
Nonoperative management
Mortality
Michel9 Bufkin6 Flynn7 Muir3 Lawrence10 Port11 Vogel5 Vallbohmer12 Abbas13 Kuppusamy (present report) Keeling14
1958–1978 1973–1993 1977–1988 1985–2000 1986–1996 1990–2001 1992–2004 1996–2008 1998–2008 1989–2009 1997–2008
72 66 69 75 30 26 47 44 119 81 97
19/72 (26%) 12/66 (18%) 10/69 (14%) 17/75 (23%) 5/30 (17%) 4/26 (15%) 34/47 (72%) 20/44 (45%) 28/119 (24%) 33/81 (40%) 25/97 (26%)
17% 24% 10% 16% 10% 3.8% 4.7% 6.8% 14% 3.7% 8%
availability of up-to-date endoscopic and radiologic services. Having these services available in a single referral center is one of the major reasons why specialty centers can provide more diversified treatment and better outcomes. The increasing use of CT scanning in conjunction with standard contrast esophagram allows periodic assessment of the severity of mediastinal and pleural contamination. Persistent uncontrolled leaks or inadequately drained collections should result in either modification of nonoperative approach or proceeding to operative intervention. Primary operative treatment is required when, in the judgment of the managing surgical team, the extent of damage or the presenting situation is not suited for endoscopic or interventional radiology techniques. Currently, the most common issues requiring operative management include large perforations in cases that involve widespread pleural and mediastinal contamination, especially when trapped lung requiring decortication is also present. In 10 patients (21% of surgical cases), a hybrid-type approach, involving open or thoracoscopic surgery in conjunction with endoscopic management, was used. This treatment approach is the clearest manifestation of how experience and diverse options for management can improve patient outcomes. We expect these types of procedures to become much more common in the future. The presence of underlying pathology often affects treatment decisions. In the present study, patients with spontaneous malignant perforations or fistulas were excluded; however, all patients with iatrogenic malignant or benign perforations were included. Underlying secondary pathology must be adequately evaluated and selectively addressed along with treating the acute perforation. Any undiagnosed mass should be biopsied; foreign bodies should be removed. Nonmalignant strictures distal to perforations should be dilated if no further intervention is planned or if primary repair is required. However, if stent placement is contemplated, then dilation should be limited to decrease
the incidence of stent migration. Iatrogenic perforations in patients with cancer should be resected when physiology and extent of disease are appropriate. The need to assess secondary pathology and the increased use of nonoperative approaches are two reasons we have evolved to assessing nearly all patients endoscopically at the time of presentation. The overall mortality rate of 3.7% related to the treatment of esophageal perforation over the past two decades is lower than historically reported in the literature. Table 6 highlights earlier large series evaluating the management and reporting outcomes of esophageal perforation. Studies reporting results over the past 20 years, including the present study, demonstrate nonoperative treatment approaches in 23% to 73% of cases, with mortality rates ranging from 3.7% to 14%. Our study is the third largest series from a single institution and reinforces the trend of decreased use of operative management and improving mortality. We have also highlighted the important place of hybrid approaches and verified that complications, length of stay, and therefore cost can dramatically improve when these patients are managed in a specialty center. Limitations of this study include the fact that data analysis was performed retrospectively, although recall and recording bias were minimized because all data were gathered in prospective fashion. That this is a heterogeneous group and the patients are between wide extremes of age and health status cannot be overstated. Our approach, however, has led to improvements in outcome over time and a mortality rate that is quite low for any cohort of esophageal perforations. Length of stay decreased significantly throughout the duration of the study. This could be due to advances in intensive care, in addition to increasing experience of the managing team. Furthermore, the advances in endoscopic techniques and stent technology advanced substantially throughout the study period.
Vol. 213, No. 1, July 2011
CONCLUSIONS We conclude that management of esophageal perforations in an experienced specialty center using a diversity of approaches is more important in producing good outcome than an arbitrary time period to treatment. We also believe that surgical leadership of the multidisciplinary team is critical to providing the best results and advocate that surgeons become familiar with, and be prepared to use, a wide variety of surgical, hybrid, and nonoperative approaches to management to minimize the necessity for defunctioning procedures and to improve outcomes, including mortality, morbidity, length of stay, and cost. Author Contributions
Study conception and design: Kuppusamy, Kline, Felisky, Low Acquisition of data: Kuppusamy, Kline, Felisky, Low Analysis and interpretation of data: Kuppusamy, Carrott, Hubka, Low Drafting of manuscript: Kuppusamy, Hubka, Carrott, Low Critical revision: Low, Carrott, Hubka, Koehler
REFERENCES 1. Brinster CJ, Singhal S, Lee L, et al. Evolving options in the management of esophageal perforation. Ann Thorac Surg 2004; 77:1475–1483. 2. Whyte RI, Iannettoni MD, Orringer MB. Intrathoracic esophageal perforation. The merit of primary repair. J Thorac Cardiovasc Surg 1995;109:140–144. 3. Muir AD, White J, McGuigan JA, et al. Treatment and outcomes of oesophageal perforation in a tertiary referral centre. Eur J Cardiothorac Surg 2003;23:799–804. 4. Richardson JD. Management of esophageal perforations: the value of aggressive surgical treatment. Am J Surg 2005;190: 161–165. 5. Vogel SB, Rout WR, Martin TD, Abbitt PL. Esophageal perforation in adults: aggressive, conservative treatment lowers morbidity and mortality. Ann Surg 2005;241:1016–1021. 6. Bufkin BL, Miller JI, Jr., Mansour KA. Esophageal perforation: emphasis on management. Ann Thorac Surg 1996;61:1447– 1451. 7. Flynn AE, Verrier ED, Way LW, et al. Esophageal perforation. Arch Surg 1989;124:1211–1214. 8. Karbowski M, Schembre D, Kozarek R, et al. Polyflex selfexpanding, removable plastic stents: assessment of treatment efficacy and safety in a variety of benign and malignant conditions of the esophagus. Surg Endosc 2008;22:1326–1333. 9. Michel L, Grillo HC, Malt RA. Operative and nonoperative management of esophageal perforations. Ann Surg 1981;194: 57–63. 10. Lawrence DR, Moxon RE, Fountain SW, et al. Iatrogenic oesophageal perforations: a clinical review. Ann R Coll Surg Engl 1998;80:115–118. 11. Port JL, Kent MS, Korst RJ, et al. Thoracic esophageal perforations: a decade of experience. Ann Thorac Surg 2003;75:1071– 1074.
Kuppusamy et al
Discussion
171
12. Vallbohmer D, Holscher AH, Holscher M, et al. Options in the management of esophageal perforation: analysis over a 12-year period. Dis Esophagus 2010;23:185–190. 13. Abbas G, Schuchert MJ, Pettiford BL, et al. Contemporaneous management of esophageal perforation. Surgery 2009;146:749– 755. 14. Keeling WB, Miller DL, Lam GT, et al. Low mortality after treatment for esophageal perforation: a single-center experience. Ann Thorac Surg 2010;90:1669–1673.
Discussion INVITED DISCUSSANT: DR JEFFREY PETERS (Rochester, NY): Successfully managing esophageal perforation has been among the most challenging acute surgical problems for nearly 100 years. Patients are often elderly with chronic illness, the presentation and/or diagnosis is all too frequently delayed, mediastinal and pleural contamination is significant, and the presence of underlying esophageal disease all complicate judgments. As such the mortality has historically been high. To these patient- and disease-related factors, technology has added further complexity. The surgeon should now be facile with interventional flexible endoscopy, including intraluminal esophageal stents and video-assisted thoracoscopy. Finally, the epidemiology and outcomes of esophageal perforation are changing; most are now iatrogenic after endoscopic interventions and the benefit of early treatment may not be as critical as in years past. Dr Low and colleagues highlight these changing complexities as they outline the treatment and outcomes of patients with esophageal perforation seen at a tertiary referral center over the past 20 years. Most were iatrogenic and involved the distal esophagus. Increasing numbers were referred from outside hospitals and there was a marked trend in nonoperative therapy. Three-quarters of the patients were treated without operation in the past 5 years. Importantly, outcomes also markedly improved, mortality was infrequent (3 patients), complications decreased by nearly half (50% to 33%), and length of stay improved by nearly a week. Progress indeed. The authors emphasize that the complexities of care have evolved to the point that referral to a tertiary care center is more important than treatment within 24 hours. Although this is likely true, the methodology of the study does not allow for such an overarching conclusion. I have a few questions. First, given that the majority of patients can now be treated nonoperatively without compromise to the principles of closure of the leak and drainage of the sepsis, when, if ever, is a thoracotomy or laparotomy indicated? My second question is, how should the size of the injury and the presence of underlying esophageal disease, such as refractory stricture of malignancy, be taken into account in the treatment decisions? Last, would you agree that we must be careful not to set too high a threshold for operative intervention? As such, when would you consider nonoperative treatment to have failed and pursue a salvage procedure? DR MICHAL HUBKA (Seattle, WA): I would like to point out that this study was not intended to advocate surgical or nonop-
highly dependent on the experience of the managing physicians and institutional clinical resources. Management strategies vary from observation and supportive measures for early-diagnosed small-contained injuries to more complex multidisciplinary management in critically ill patients with established mediastinitis and sepsis. Historically, outcomes appear to be highly affected by the ability to rapidly and accurately diagnose the site and severity of perforation and initiate treatment within the first 24 hours. This hallmark time period dictated both the ability to perform primary repair and was found to be closely associated with posttreatment levels of morbidity and mortality.2 As a tertiary referral center for esophageal disease, we are referred a wide variety of acute perforations, many of which have had a delay in diagnosis and management or have undergone initial therapy in other institutions. Our ap-
Acute esophageal perforation has historically been associated with a high mortality rate ranging from 4% to 80%.1 This potentially life-threatening clinical condition has a diverse presentation, and its diagnosis and management are
Author Disclosure Information: Nothing to disclose. Editor Disclosure Information: Nothing to disclose. Presented at the Western Surgical Association 118th Scientific Session, Chicago, IL, November 2010. Received December 7, 2010; Revised January 15, 2011; Accepted January 21, 2011. From the Section of General Thoracic Surgery, Virginia Mason Medical Center, Seattle, WA. Correspondence address: Donald E Low, FACS, FRCS(C), Section of General Thoracic Surgery, Virginia Mason Medical Center, 1100 Ninth Ave. C6-SUR, Seattle, WA 98111. email: [email protected]
© 2011 by the American College of Surgeons Published by Elsevier Inc.
164
ISSN 1072-7515/11/$36.00 doi:10.1016/j.jamcollsurg.2011.01.059
Vol. 213, No. 1, July 2011
Kuppusamy et al
Evolving Management in Esophageal Perforation
165
Acute esophageal perforation
Unstable / Septic
Clinically stable
Water soluble / Barium contrast study (Observed by managing team)
If feasible
Non-Intubated
CT scan followed by EGD by managing team
Followed immediately by CT scan
Contained perforation
Decision favors observation
Contained perforation or Limited free perforation
Extensive contained perforation or free perforation
Decision favors endoscopic therapy
Decision favors surgical therapy
NPO Antibiotics / Anti-fungals ± NG Tube ± Nutritional support
Intubated
Decision favors endoscopic therapy or patient too unstable for surgery
EGD in O.R. prior to positioning Leave scope in place during operative procedure
If primary repair, review options for additional endoscopic therapy
Endoscopic clips / stents Trans-esophageal drainage Enteric feeding tubes ± Additional surgical or interventional radiology drains
Figure 1. Therapeutic algorithm for management of acute esophageal perforation. CT, computerized tomography; EGD, esophagogastroduodenoscopy; NG, nasogastric; NPO, nil per os; OR, operating room.
proach to the management of acute esophageal perforation has evolved over the period of study. Under surgical leadership, increasing, but selective, applications of nonsurgical modalities, both endoscopic and involving interventional radiology, have been used. We hypothesized that although early detection and treatment remains important, ultimate results are more closely associated with the experience of the initial management team rather than an arbitrary time period. We also contended that the appropriate increased use of nonsurgical and hybrid treatment approaches has contributed to improved outcomes over the study period.
METHODS Patients presenting to our tertiary referral residency training center with acute esophageal perforation between June 1989 and March 2009 were prospectively entered in an Institutional Review Board–approved and monitored database and retrospectively reviewed. This included patients initially diagnosed and treated at our institution as well as those referred after initial diagnosis of perforation elsewhere. All patients with acute iatrogenic or spontaneous
benign or malignant perforations were entered into the study. However, patients with spontaneous malignant perforations or fistulas were excluded. Patient demographics, pertinent history, timing and characteristics of presentation and diagnosis, management techniques, and outcomes were reviewed. American Society of Anesthesiologists (ASA) levels for all nonoperative patients were assigned by a single blinded anesthesiologist. All interventions, whether surgical, gastorintestinal, or involving interventional radiology, were recorded. Management was multidisciplinary but always under the supervision of the surgical team. This included endoscopic interventions, which were performed either by the surgical team or by a gastroenterologist with a member of the surgical team present. The therapeutic algorithm for management of acute esophageal perforation at our institution is depicted in Fig 1. Assessment of demographics and types of management was carried out over the entire study period, and results were also assessed separately in 5 4-year periods over the course of the study to assess the evolution of
166
Kuppusamy et al
Evolving Management in Esophageal Perforation
treatment techniques and outcomes over time. Treatment was assessed according to whether it involved surgical or nonoperative techniques; patients who received hybrid treatment approaches (involving both surgical and nonoperative modalities) were also reviewed but categorized as surgical procedures. Descriptive statistics were used to report continuous variables with range or standard deviation (SD), and categoric variables were reported as frequencies with percentages. Between-group comparisons of categoric outcomes were performed using Fisher exact test and chi square test as appropriate to the distribution of the data, with p values denoting levels of statistical significance (p ⫽ 0.05). Multivariate logistic regression was performed with length of stay and occurrence of complications as dependent variables. Statistical analyses were performed using SPSS (version 18) software package (SPSS, Chicago, IL).
RESULTS Between June 1989 and March 2009, 81 patients presented to Virginia Mason Medical Center with acute esophageal perforation. Their mean age was 63.6 years (range 22 to 96 years) and included 58 men and 23 women. Fifty-nine percent of patients were referred after initial diagnosis at another institution, with 8 of the 46 referred patients having initial surgical6 or endoscopic therapy2 at another institution. Sixty-eight percent of all patients were graded as ASA levels 3, 4, or 5 at the time of presentation. Etiology of perforation involved 51 iatrogenic (63%), 24 spontaneous Boerhaave-type barogenic perforations (30%), 4 spontaneous (5%), and 2 foreign body induced (2%). The iatrogenic perforations were most often associated with esophageal dilation (n ⫽ 27; 53%), with pneumatic dilation for achalasia being the most common dilation technique, involved in 14 patients. Therapeutic
J Am Coll Surg
endoscopy (eg, sclerotherapy, stenting, removing food impaction) accounted for 7 patients, diagnostic endoscopy for 5, nonesophageal surgical procedure accounted for 6, rigid esophagoscopy for 1, transesophageal echocardiography for 1, endoscopic ultrasound for 1, endoscopic retrograde cholangiopancreatography for 1, and self-dilation for 2. Spontaneous perforation, without a known etiology, was present in 4 patients. Iatrogenic endoscopic injuries resulting in perforation were identified at the time of injury in 64% of cases. Initial investigations involved upper gastrointestinal contrast studies in 75%. We recommend that a member of the surgical team witness this investigation first-hand to increase accuracy of interpretation and aid in treatment planning. Computerized tomographic (CT) scans were used in 46 patients (56%), and upper endoscopy was done in 64% at some point during their workup and management. Use of both CT and endoscopy increased over time. The application of endoscopy as a component of diagnosis or management of acute esophageal perforation increased throughout the study period, being used in 3 patients in the first 4 years (38%) but in 80% of cases in the last 5 years (Table 1). The most common location of injury was distal esophagus in 75% followed by cervical in 20% and midesophagus in 5%. Length of perforations averaged 2.7 cm (median 2 cm, range ⬍1 to 13 cm). Treatment was initiated within 24 hours in 57 patients (70%). Patients having treatment ⬎24 hours after perforation were more likely to be referred (79% vs 51%; p ⫽ 0.018) and have ASA levels 3–5 (83% vs 61%; p ⫽ 0.095; see Table 2). Median length of hospital stay was longer in patients treated after 24 hours compared with patients treated within 24 hours (20 vs 11 days; p ⫽ 0.003). However, there was no significant difference in complication
Table 1. Evolution of Management over the Study Period (1989 to 2009) n Operative treatment, n (%) Nonoperative treatment, n (%) Median LOS, d (range) ASA 3–5, n (%) Primary cases, n (%) Referred cases, n (%) UGI used in diagnosis, n (%) CT used in diagnosis, n (%) EGD used in management, n (%) ⬍24 h to Diagnosis, n (%) ⱖ24 h to Diagnosis, n (%)
1989–1992
1993–1996
1997–2000
2001–2004
2005–2009
Total
8 8 (100) 0 (0) 18.5 (9–81) 6 (75) 4 (50) 4 (50) 8 (100) 1 (12) 3 (38) 6 (75) 2 (25)
19 15 (78) 4 (22) 12 (4–87) 14 (74) 6 (32) 13 (68) 14 (74) 3 (16) 12 (63) 8 (43) 11 (57)
12 7 (58) 5 (42) 13.5 (4–86) 10 (83) 6 (50) 6 (50) 10 (83) 6 (50) 6 (50) 10 (83) 2 (27)
18 12 (67) 6 (33) 15 (3–73) 12 (67) 10 (56) 8 (44) 10 (56) 13 (72) 11 (61) 14 (78) 4 (22)
24 6 (25) 18 (75) 8.5 (1–41) 13 (54) 7 (29) 17 (71) 19 (79) 23 (96) 20 (80) 19 (79) 5 (21)
81 48 (59) 33 (41) 12 (1–87) 55 (68) 33 (41) 48 (59) 61 (75) 46 (57) 52 (64) 56 (69) 25 (31)
ASA, American Society of Anesthesiologists; LOS, length of stay.
p Value
⬍0.001 0.094 0.43 0.37 0.14 ⬍0.001 0.12 0.048
Vol. 213, No. 1, July 2011
Kuppusamy et al
Table 2. Patient Characteristics and Outcomes by Timing of Treatment Time to treatment
<24 h
>24 h
p Value
Patients, n (%) ASA 3–5, n (%) Referred, n (%) Median LOS, d Operative, n (%) Complications, n (%) Mortality
57 (70) 35 (61) 29 (51) 11 35 (61) 20 (37) 1
24 (30) 20 (83) 19 (79) 20 13 (54) 11 (41) 2
0.095 0.018 0.003 0.55 0.36 0.43
ASA, American Society of Anesthesiologists; LOS, length of stay.
rates (40.9% vs 37.2%; p ⫽ 0.36) or mortality (2 vs 1; p ⫽ 0.43). Mortalities in the 2 patients receiving treatment ⬎24 hours after perforation involved a 78-year-old patient transfered intubated and septic 2.5 days after iatrogenic perforation. She was treated with a hybrid approach involving right thoracotomy, mediastinal drainage, and decortications with the simultaneous insertion of an endoscopic esophageal stent. She recovered and left the intensive care unit, but died on postoperative day 19 of cardiac arrest secondary to complete heart block. The second patient was transfered to our hospital septic, with Fournier gangrene and Boerhaave syndrome, for hyperbaric oxygen therapy for the Fournier gangrene. Transfer was 15 days after the perforation; although a chest tube was placed on recognition of pneumothorax at the previous hospital, it had not been recognized as an esophageal perforation until a swallow study was performed at our institution. He died on hospital day 86 when care was withdrawn. The third mortality was in a 74-year-old patient on peritoneal dialysis, who experienced an episode of vomiting after an endoscopy for abdominal pain and underwent immediate laparotomy for free intraperitoneal air at another center. Attempted transabdominal repair at the other hospital led to splenectomy and 2-liter blood loss with an ongoing leak, which led to transfer to our institution. The patient recovered, but refused further dialysis; he ultimately died on hospital day 51. Operative treatment was used in 48 patients (59%) with 33 (41%) treated without surgery. Table 3 highlights the treatments used in both approaches. ASA levels 3–5 were more common in operative than in nonoperative patients (75% vs 58%; p ⫽ 0.099; Table 4). Primary repair was the most common surgical treatment, in 72%. Surgical approach was transthoracic in 59%, transabdominal in 37%, and combined thoracoabdominal in 4%. Hybrid approaches involving sequential or synchronous interventional and surgical procedures were carried out in 10 patients (21%). These procedures included endoscopic stents placed at the time of primary repair or stents (5) and endoscopically placed nasomediastinal drains (1) placed in asso-
Evolving Management in Esophageal Perforation
167
ciation with open or thoracoscopic pleural decortication and drainage. The evolution of presentation and treatment is shown in Table 4. Typically, ⬃50% of cases were referred, except for the last 5 years, when 71% came from other institutions. Table 3 demonstrates that nonoperative therapy was more common in referred cases than in primary cases (48% vs 30%) and in those whose treatment was delayed (⬎24 hours) compared with patients treated early (⬍24 hours; 46% vs 38%, respectively). Patients undergoing treatment ⬎24 hours after perforation have varied moderately over time from 21% to 57% of patients. Figure 2 demonstrates in graphic form that the issues that underwent the most pronounced changes over time were patients undergoing nonsurgical management, which increased from 0 to 75% (p ⬍ 0.001). Over the same 20-year period, complications decreased from 50% to 33% (p ⫽ 0.94) and median length of stay decreased from 18.5 to 8.5 days (p ⫽ 0.094). Table 5 demonstrates that there was no statistical difference in mortality, persistent leak, or major medical complications of sepsis, pneumonia, myocardial infarction, dysrhythmia, renal failure, or deep vein thrombosis between operative and nonoperative populations. Multivariate logistic regression was used to analyze Table 3. Details of Operative and Nonoperative Treatments Treatment
n
Operative treatment (n ⫽ 48) Intraoperative endoscopy Surgical drainage and stent Resection Early Late Primary repair Concurrent procedures Buttressed Antireflux procedures Intraoperative stent T-tube Nonoperative treatment (n ⫽ 33) Endoscopic stenting* Total parenteral nutrition Dobhoff Gastrostomy Feeding jejunostomy Radiologic drainage procedures Endoscopic mediastinal drainage Foreign body removal Endoscopic repair (clip repair-3; glue-2)
20 3 5 3 3 34 25 6 4 2 11 7 5 5 3 11 2 2 5
*Polyflex, 5; Ultraflex, 2; Dua, 1; Celestin, 1; Alimaxx, 1; self-expanding metal stent, 1.
168
Evolving Management in Esophageal Perforation
Kuppusamy et al
J Am Coll Surg
Table 4. Patient Characteristics and Outcomes by Treatment Group Operative treatment
Patients ASA 3–5 Referred Median LOS, d Treatment ⬎24 h Complications Mortality
n
%
48 36 25 15.5 13 21 1
59 75 52
Nonoperative treatment n %
33 19 23 10 11 10 2
27 44
41 58 70 33 30
p Value
0.099 0.11 0.002 0.54 0.22 0.74
ASA, American Society of Anesthesiologists; LOS, length of stay.
method and timing of treatment, presentation (ie, referral or primary), era, source of perforation, diagnostic methods, and ASA level with length of stay and occurrence of complications as the covariates. Of these, increased ASA level (p ⫽ 0.016), presentation ⬎24 hours (p ⫽ 0.038), use of endoscopy (p ⫽ 0.01), and occurrence of complications (p ⫽ 0.006) significantly accounted for variations in length of stay. Only ASA status was significant regarding occurrence of complications (p ⫽ 0.005).
management continue to evolve. Historically, authors have advocated the efficacy of either operative,3,4 or nonoperative management5 of esophageal injuries. In the present study, we found that, in our experience, the management of esophageal perforation in an era of rapid technologic progress continues to change. Part of this change is expressed in the increased application of hybrid-type procedures where the advantages of surgical and interventional endoscopy and radiology techniques are applied. We believe that improving outcomes, including lowering historic levels of mortality, morbidity, and ultimately cost, are dependent on the patient receiving appropriate work-up and tailored management by an experienced tertiary-care surgical team
DISCUSSION The optimal management of esophageal perforation remains controversial, and the options for intervention and 80
Complicaons Non-Operave Treatment Length of Stay (Days)
70
75%
60 50
50%
42% 37%
40
39% 42%
33%
30 20
33% 18.5 21%
15
13.5
8.5
10
12 0%
0 1989-1992 (n=8)
1993-1996 (n=19)
Complicaons (%) Non-Operave Treatment (%) Median LOS (Days)
1989-1992 (n=8) 50 0 18.5
1997-2000 (n=12)
2001-2004 (n=18)
1993-1996 1997-2000 (n=19) (n=12) 37 42 21 42 12.0 13.5
2005-2009 (n=24)
2001-2004 2005-2009 (n=18) (n=24) 39 33 33 75 15.0 8.5
p 0.94 < 0.001 0.094
Figure 2. Graphic description of management and complications over time. Nonoperative treatment has increased significantly, and complications and length of stay (LOS) have decreased, though not to a statistically significant degree.
Vol. 213, No. 1, July 2011
Kuppusamy et al
Table 5. Complications by Treatment Group
Complication
Pneumonia Dysrhythmias Deep vein thrombosis Persistent leak (clinical or radiologic evidence of leak at 14th day) Stent migration Sepsis Renal failure Death
Operative treatment (n ⴝ 47) n %
Nonoperative treatment (n ⴝ 34) n %
7 11 1
14 23 2.0
4 4 1
11 11 2.9
2 0 0 0 1
4.3 0.0 0.0 0.0 2.1
3 3 1 1 2
8.8 27.3 2.9 2.9 5.9
All p values ⬎0.05.
facile with all aspects of surgical and nonoperative management modalities. We have developed a treatment algorithm for esophageal perforation at our institution, which is now used routinely to aid in treatment decision making (Fig. 1). Appropriate work-up of esophageal perforation has typically included an upper gastrointestinal contrast study. We emphasize the importance of the presence of a surgical team member at the time of this study to ensure that the study is complete and provides the anatomic details often missed when the radiologist performs the study alone. This can have a major effect on the surgeon’s impression of the magnitude and location of the perforation and can affect treatment planning. This is also the justification we have used for repeating contrast studies when transfered patients arrive with ambiguous studies. CT scanning allows an additional evaluation of the severity of mediastinal or pleural space contamination. Table 1 illustrates that the use of CT scanning increased from selective (12%) to routine (96%) in medically stable patients. Iatrogenic esophageal injuries remain the most common etiology, which is consistent with earlier studies.1,6,7 Earlier reports have highlighted the importance of endoscopy in recognition of injury at the time of the event.6 In the early stages of the present study, endoscopy was rarely used; however, in the last 5 years, 80% of patients with esophageal perforation underwent upper endoscopy, and it is now a routine part of the work-up in almost all patients. We have not identified any significant complications or problems with using endoscopy in patients with acute esophageal perforation. We have also found endoscopic examination to be invaluable regarding initial treatment decision making by identifying the exact localization of the perforation and assessing secondary pathology and viability of esophageal mucosa. We have used upper endoscopy intraoperatively in
Evolving Management in Esophageal Perforation
169
44% of cases guiding surgical repair. This approach is particularly helpful because the mucosal and muscular defects can often be spatially separated and primary closure of the external esophageal layers does not result in mucosal approximation. Moreover, mediastinal contamination can obscure anatomy, making exact intraoperative visualization of the level and extent of perforation challenging, and the in-place endoscope can help guide the surgeon directly to the mucosal defect. Intraoperative endoscopy can also facilitate dilation of persisting strictures, removal of foreign bodies, and placement of endoscopic stents and drains when appropriate. In our series, the number of patients treated ⬎24 hours after injury has varied only moderately over the period of this study. In part, this is due to the high percentage of patients referred to our institution (50% to 75% over the past 20 years).Operative management has historically been the mainstay of treatment, with the first 24-hour period being the hallmark time metric of good outcome.1,2 In a study of 75 patients, Muir and colleagues3 found increased mortality from 5% to 44% in patients treated before and after this hallmark time period. However, our results suggest that the initial evaluation by an experienced team able to provide tailored management appears to be more important than an arbitrary time period related to onset of injury. We have found that patients presenting ⬎24 hours after perforation are more likely to be referred and be unstable, as reflected by a higher ASA score (Table 2). These patients demonstrate a longer length of stay; however, we did not find any differences in minor or major complications or in mortality compared with patients treated within the first 24 hours after injury. This finding supports the concept of transfering appropriate patients with perforations to highvolume esophageal centers, even when the transfer will further delay definitive treatment. Examination of the trends in management of esophageal perforation in the present study showed that nonoperative management increased dramatically (0 to 75%) over the study period. In association with this trend toward nonoperative management, lengths of stay and complications declined (Fig. 2). Improvement in results in a study spanning 20 years is to be expected. However, the largest single change in management approach has been the introduction and growth of nonoperative and hybrid treatment techniques. We first used endoscopic stents in 1999 (Polyflex; Boston Scientific), and have used a variety since that time (Table 3); typically we place fully covered removable stents. The safety and success of nonoperative management has been previously highlighted in reports by our group8 and Vogel and associates.5 Multidisciplinary care is not just about an experienced surgical team but is also related to the
170
Kuppusamy et al
Evolving Management in Esophageal Perforation
J Am Coll Surg
Table 6. Large Case Series of Esophageal Perforation First author
Study period
n
Nonoperative management
Mortality
Michel9 Bufkin6 Flynn7 Muir3 Lawrence10 Port11 Vogel5 Vallbohmer12 Abbas13 Kuppusamy (present report) Keeling14
1958–1978 1973–1993 1977–1988 1985–2000 1986–1996 1990–2001 1992–2004 1996–2008 1998–2008 1989–2009 1997–2008
72 66 69 75 30 26 47 44 119 81 97
19/72 (26%) 12/66 (18%) 10/69 (14%) 17/75 (23%) 5/30 (17%) 4/26 (15%) 34/47 (72%) 20/44 (45%) 28/119 (24%) 33/81 (40%) 25/97 (26%)
17% 24% 10% 16% 10% 3.8% 4.7% 6.8% 14% 3.7% 8%
availability of up-to-date endoscopic and radiologic services. Having these services available in a single referral center is one of the major reasons why specialty centers can provide more diversified treatment and better outcomes. The increasing use of CT scanning in conjunction with standard contrast esophagram allows periodic assessment of the severity of mediastinal and pleural contamination. Persistent uncontrolled leaks or inadequately drained collections should result in either modification of nonoperative approach or proceeding to operative intervention. Primary operative treatment is required when, in the judgment of the managing surgical team, the extent of damage or the presenting situation is not suited for endoscopic or interventional radiology techniques. Currently, the most common issues requiring operative management include large perforations in cases that involve widespread pleural and mediastinal contamination, especially when trapped lung requiring decortication is also present. In 10 patients (21% of surgical cases), a hybrid-type approach, involving open or thoracoscopic surgery in conjunction with endoscopic management, was used. This treatment approach is the clearest manifestation of how experience and diverse options for management can improve patient outcomes. We expect these types of procedures to become much more common in the future. The presence of underlying pathology often affects treatment decisions. In the present study, patients with spontaneous malignant perforations or fistulas were excluded; however, all patients with iatrogenic malignant or benign perforations were included. Underlying secondary pathology must be adequately evaluated and selectively addressed along with treating the acute perforation. Any undiagnosed mass should be biopsied; foreign bodies should be removed. Nonmalignant strictures distal to perforations should be dilated if no further intervention is planned or if primary repair is required. However, if stent placement is contemplated, then dilation should be limited to decrease
the incidence of stent migration. Iatrogenic perforations in patients with cancer should be resected when physiology and extent of disease are appropriate. The need to assess secondary pathology and the increased use of nonoperative approaches are two reasons we have evolved to assessing nearly all patients endoscopically at the time of presentation. The overall mortality rate of 3.7% related to the treatment of esophageal perforation over the past two decades is lower than historically reported in the literature. Table 6 highlights earlier large series evaluating the management and reporting outcomes of esophageal perforation. Studies reporting results over the past 20 years, including the present study, demonstrate nonoperative treatment approaches in 23% to 73% of cases, with mortality rates ranging from 3.7% to 14%. Our study is the third largest series from a single institution and reinforces the trend of decreased use of operative management and improving mortality. We have also highlighted the important place of hybrid approaches and verified that complications, length of stay, and therefore cost can dramatically improve when these patients are managed in a specialty center. Limitations of this study include the fact that data analysis was performed retrospectively, although recall and recording bias were minimized because all data were gathered in prospective fashion. That this is a heterogeneous group and the patients are between wide extremes of age and health status cannot be overstated. Our approach, however, has led to improvements in outcome over time and a mortality rate that is quite low for any cohort of esophageal perforations. Length of stay decreased significantly throughout the duration of the study. This could be due to advances in intensive care, in addition to increasing experience of the managing team. Furthermore, the advances in endoscopic techniques and stent technology advanced substantially throughout the study period.
Vol. 213, No. 1, July 2011
CONCLUSIONS We conclude that management of esophageal perforations in an experienced specialty center using a diversity of approaches is more important in producing good outcome than an arbitrary time period to treatment. We also believe that surgical leadership of the multidisciplinary team is critical to providing the best results and advocate that surgeons become familiar with, and be prepared to use, a wide variety of surgical, hybrid, and nonoperative approaches to management to minimize the necessity for defunctioning procedures and to improve outcomes, including mortality, morbidity, length of stay, and cost. Author Contributions
Study conception and design: Kuppusamy, Kline, Felisky, Low Acquisition of data: Kuppusamy, Kline, Felisky, Low Analysis and interpretation of data: Kuppusamy, Carrott, Hubka, Low Drafting of manuscript: Kuppusamy, Hubka, Carrott, Low Critical revision: Low, Carrott, Hubka, Koehler
REFERENCES 1. Brinster CJ, Singhal S, Lee L, et al. Evolving options in the management of esophageal perforation. Ann Thorac Surg 2004; 77:1475–1483. 2. Whyte RI, Iannettoni MD, Orringer MB. Intrathoracic esophageal perforation. The merit of primary repair. J Thorac Cardiovasc Surg 1995;109:140–144. 3. Muir AD, White J, McGuigan JA, et al. Treatment and outcomes of oesophageal perforation in a tertiary referral centre. Eur J Cardiothorac Surg 2003;23:799–804. 4. Richardson JD. Management of esophageal perforations: the value of aggressive surgical treatment. Am J Surg 2005;190: 161–165. 5. Vogel SB, Rout WR, Martin TD, Abbitt PL. Esophageal perforation in adults: aggressive, conservative treatment lowers morbidity and mortality. Ann Surg 2005;241:1016–1021. 6. Bufkin BL, Miller JI, Jr., Mansour KA. Esophageal perforation: emphasis on management. Ann Thorac Surg 1996;61:1447– 1451. 7. Flynn AE, Verrier ED, Way LW, et al. Esophageal perforation. Arch Surg 1989;124:1211–1214. 8. Karbowski M, Schembre D, Kozarek R, et al. Polyflex selfexpanding, removable plastic stents: assessment of treatment efficacy and safety in a variety of benign and malignant conditions of the esophagus. Surg Endosc 2008;22:1326–1333. 9. Michel L, Grillo HC, Malt RA. Operative and nonoperative management of esophageal perforations. Ann Surg 1981;194: 57–63. 10. Lawrence DR, Moxon RE, Fountain SW, et al. Iatrogenic oesophageal perforations: a clinical review. Ann R Coll Surg Engl 1998;80:115–118. 11. Port JL, Kent MS, Korst RJ, et al. Thoracic esophageal perforations: a decade of experience. Ann Thorac Surg 2003;75:1071– 1074.
Kuppusamy et al
Discussion
171
12. Vallbohmer D, Holscher AH, Holscher M, et al. Options in the management of esophageal perforation: analysis over a 12-year period. Dis Esophagus 2010;23:185–190. 13. Abbas G, Schuchert MJ, Pettiford BL, et al. Contemporaneous management of esophageal perforation. Surgery 2009;146:749– 755. 14. Keeling WB, Miller DL, Lam GT, et al. Low mortality after treatment for esophageal perforation: a single-center experience. Ann Thorac Surg 2010;90:1669–1673.
Discussion INVITED DISCUSSANT: DR JEFFREY PETERS (Rochester, NY): Successfully managing esophageal perforation has been among the most challenging acute surgical problems for nearly 100 years. Patients are often elderly with chronic illness, the presentation and/or diagnosis is all too frequently delayed, mediastinal and pleural contamination is significant, and the presence of underlying esophageal disease all complicate judgments. As such the mortality has historically been high. To these patient- and disease-related factors, technology has added further complexity. The surgeon should now be facile with interventional flexible endoscopy, including intraluminal esophageal stents and video-assisted thoracoscopy. Finally, the epidemiology and outcomes of esophageal perforation are changing; most are now iatrogenic after endoscopic interventions and the benefit of early treatment may not be as critical as in years past. Dr Low and colleagues highlight these changing complexities as they outline the treatment and outcomes of patients with esophageal perforation seen at a tertiary referral center over the past 20 years. Most were iatrogenic and involved the distal esophagus. Increasing numbers were referred from outside hospitals and there was a marked trend in nonoperative therapy. Three-quarters of the patients were treated without operation in the past 5 years. Importantly, outcomes also markedly improved, mortality was infrequent (3 patients), complications decreased by nearly half (50% to 33%), and length of stay improved by nearly a week. Progress indeed. The authors emphasize that the complexities of care have evolved to the point that referral to a tertiary care center is more important than treatment within 24 hours. Although this is likely true, the methodology of the study does not allow for such an overarching conclusion. I have a few questions. First, given that the majority of patients can now be treated nonoperatively without compromise to the principles of closure of the leak and drainage of the sepsis, when, if ever, is a thoracotomy or laparotomy indicated? My second question is, how should the size of the injury and the presence of underlying esophageal disease, such as refractory stricture of malignancy, be taken into account in the treatment decisions? Last, would you agree that we must be careful not to set too high a threshold for operative intervention? As such, when would you consider nonoperative treatment to have failed and pursue a salvage procedure? DR MICHAL HUBKA (Seattle, WA): I would like to point out that this study was not intended to advocate surgical or nonop-