Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Current concepts in the management of esophagueal perforations


Published on

  • Be the first to comment

  • Be the first to like this

Current concepts in the management of esophagueal perforations

  1. 1. Current Concepts in the Management ofEsophageal Perforations: A Twenty-Seven YearCanadian ExperiencePankaj Bhatia, MD, Dalilah Fortin, MD, Richard I. Inculet, MD, andRichard A. Malthaner, MDDivision of Thoracic Surgery, The University of Western Ontario, London, Ontario, Canada GENERAL THORACIC Background. Perforation of the esophagus remains a or greater with 19.6 (95% CI 4.8 to 84.9), the presence of achallenging clinical problem. pulmonary comorbidity with 13.9 (95% CI 2.9 to 97.4), Methods. A retrospective review was performed of and sepsis with 3.1 (95% CI 1.0 to 10.1). A wait time ofpatients diagnosed with an esophageal perforation ad- greater than 24 hours was not associated with an in-mitted to the London Health Sciences Centre from 1981 creased risk of mortality (p ‫.)25.0 ؍‬to 2007. Univariate and multivariate logistic regression Conclusions. Malignant perforations, sepsis, mechani-was used to determine which factors had a statistically cal ventilation at presentation, a higher overall burden ofsignificant effect on mortality. comorbidity, and a pulmonary comorbidity have a sig- Results. There were 119 patients; 15 with cervical, 95 nificant impact on the overall survival. Time to treatmentwith thoracic, and 9 with abdominal perforations. Fifty- is not as important. Restoration of intestinal continuity,one percent of all the perforations were iatrogenic and either by primary repair or by excision and reanastomo-33% were spontaneous. Multivariate logistic regression sis can be attempted even in patients with a greater timeanalysis revealed that patients with preoperative respira- from perforation to treatment with respectable morbiditytory failure requiring mechanical ventilation had a mor- and mortality rates.tality odds ratio of 32.4 (95% confidence interval [CI] 3.1to 272.0), followed by malignant perforations with 20.2 (Ann Thorac Surg 2011;92:209 –15)(95% CI 5.4 to 115.6), a Charlson comorbidity index of 7.1 © 2011 by The Society of Thoracic SurgeonsP erforation of the esophagus remains a challenging clinical problem in the world of thoracic surgery.The management options for esophageal perforations continue to have esophageal leaks postoperatively that can often be managed nonoperatively [1]. Obviously, this adds to the debate as to the optimal management ofvary depending on the location of the perforation, its esophageal perforations.etiology, the time from its occurrence to the intervention, In the contemporary era, primary repair of esophagealand the overall condition of the patient [1– 6]. Delay in the perforations is becoming widely accepted as the treat-diagnosis and treatment can significantly influence the ment of choice, and several studies [2, 11–14] haveoverall outcomes of patients sustaining perforations, as demonstrated respectable mortality and morbidity ratescan the location of the perforation and the presence of with benign perforations, even with delayed presenta-underlying esophageal pathology such as a malignancy. tion. Obviously, if the esophagus is devitalized or con-This multivariate influence on the natural history and tains malignant disease, alternative methods of treatmentoutcome of esophageal perforations is reflected in the other than primary repair should be undertaken. Thesewide range of mortality rates reported in various studies patients have been shown to have a worse outcome [3, 11]examining the outcomes of patients sustaining a disrup- in a number of retrospective reviews, including a reviewtion [2, 5, 7–10]. The evolution of management of esoph- in 2004 [2]. This has led us to postulate that time toageal perforations has taken us from routine diversion to treatment is perhaps not as critical a factor in patientesophageal resection to primary repair and stenting of mortality as many textbooks suggest. The overall presen-perforations, yet controversy continues to exist with re- tation of the patient, be it hemodynamically stable tospect to the best way to treat the aforementioned sub- sepsis to florid shock, is far more predictive of outcome.types of esophageal rupture. It is also accepted that The fact that several options exist for managing thedespite the surgical procedure used, 30% of patients will various types of esophageal perforations has promptedAccepted for publication March 29, 2011. several authors to review their experience with this challenging and potentially deadly entity. We presentAddress correspondence to Dr Malthaner, Divisions of Thoracic Surgery here our contemporary experience in the hopes of an-and Surgical Oncology, London Health Sciences Centre, 800 Commis-sioners Rd E, Ste E2-124, London, Ontario, Canada N6A 5W9; e-mail: swering some of the questions surrounding the ment of this important surgical problem. Our hypotheses© 2011 by The Society of Thoracic Surgeons 0003-4975/$36.00Published by Elsevier Inc doi:10.1016/j.athoracsur.2011.03.131
  2. 2. 210 BHATIA ET AL Ann Thorac Surg ESOPHAGEAL PERFORATION 2011;92:209 –15 are that time to treatment of an esophageal perforation is mortality. Categoric variables were analyzed using the not as critical a predictor of mortality and morbidity as Fisher exact test and continuous variables were analyzed previously thought, that malignant perforations of the using the Wilcoxon two-sample rank test. Statistically esophagus carry with them a higher mortality and mor- significant variables (variables with p values less than or bidity than benign perforations, and that the nature of equal to 0.1) in the univariate analysis were then in- the patient’s presentation is a more critical predictor of cluded in a multivariate logistic-regression analysis to mortality than time to treatment. determine the magnitude of their influence on mortality in the form of odds ratios and the associated 95% confi- dence interval (CI). Patients and Methods Approval for our study was obtained from the ERB (Ethics Review Board) of London Health Sciences Centre ResultsGENERAL THORACIC and the University of Western Ontario. Individual pa- One hundred nineteen patients were treated for esoph- tient consents were waived by the ERB because individ- ageal perforations between 1981 and 2007. Of these 119 ual patients were not identified in the manuscript. patients, 15 suffered cervical perforations, 95 patients We conducted a retrospective review of our prospec- sustained thoracic perforations, and 9 sustained abdom- tive thoracic database of patients diagnosed with an inal perforations (Fig 1). esophageal perforation admitted to the London Health In our series, 51% (n ϭ 61) of all the perforations were Sciences Centre from 1981 to 2007. We excluded esoph- iatrogenic in nature. Spontaneous perforations ac- ageal perforations discovered intraoperatively during counted for 33% (n ϭ 39), whereas malignant perfora- other thoracic surgery procedures such as lobectomies tions accounted for 15% (n ϭ 18) of the total. There was a and repaired by the surgeon at the time of this first. The single perforation because of an ingestion of a caustic included patients were subsequently tabulated according substance. Perforations because of an ingested foreign to the location of the perforation (cervical, thoracic, body resulting in attempted instrumentation to remove it abdominal), and the etiology of the perforation (iatro- constituted 4% (n ϭ 5) of all perforations and are in- genic, malignant, spontaneous, foreign body). The time cluded in the iatrogenic group, as the endoscopic proce- from onset of symptoms to treatment, as well as the dure resulted in the perforation in all cases. method of treatment (primary repair, diversion, surgical Patients with abdominal perforations tended to be drainage, conservative) were analyzed for each patient. considerably younger, with an average age of 50.5 years Mortality, defined as death within 30 days of admission (median, 42 years; range, 22 to 84 years), as compared to hospital or during the same admission for an esopha- with thoracic (mean age ϭ 65.5 years; median, 64 years; geal perforation, length of stay, and reoperation were range, 8 to 92 years) and cervical (mean age of 67.1 years; examined for each location and treatment group. The median, 69 years; range, 39 to 85 years) (Table 2). There initial presentation of patients and their underlying co- was a predominance of males (74%) in the thoracic morbid medical issues were also examined. An empiric perforation category. Patients with malignant thoracic comorbidity score was given for each patient, with pa- perforations had the highest comorbidity score (mean tients receiving one point if a comorbidity was present in 2.5), American Society of Anesthesiologists (ASA) score each of 9 domains (cardiac, pulmonary, hepatic, renal, (3.7), and Charlson comorbidity index (6.3) among all the smoking, diabetic, alcohol abuse, malnutrition, steroid groups. use) for a total score out of a possible nine points. In The average time of perforation to treatment was 37 addition, we also calculated the Charlson comorbidity hours (median, 11 hours; range, 0.8 to 240 hours) in the index for each patient as outlined by Charlson and cervical group, 19 hours (median, 8 hours; range, 0.5 to 72 colleagues in 1987 [15]. hours) in the abdominal group, 129 hours (median, 24 With respect to outcome measures, we defined mortal- hours; range, 0.5 to 480 hours) in the malignant thoracic ity as any death occurring within 30 days of the primary group, and 65 hours (median, 27 hours; range, 0 to 480 treatment or at any point during the index hospital hours) in the benign thoracic group (Table 3). Of note, the admission. Esophageal leak was defined as any second- vast majority of benign thoracic perforations (72%), ab- ary violation of the remnant esophageal conduit at any dominal (67%), and cervical (67%) were repaired primar- point during the hospital admission after resolution of ily (Table 3). Conservative treatment, classified as treat- the initial perforation by the primary treatment. ment of an esophageal perforation with a closed-chest Follow-up was available for all patients and consisted tube thoracostomy, expandable stent, or compassionate of a single postoperative visit approximately 4 weeks care accounted for 27% of cervical perforations, 0% of all after discharge from hospital. Further visits were sched- abdominal perforations, 31% of all malignant thoracic uled based on patient need (underlying esophageal pa- perforations, and 19% of all benign thoracic perforations. thology, management of postoperative complications). Diversion was only performed twice in the abdominal In an attempt to ascertain the determinants of mortal- perforation group (one of these was due to a severe ity for our patient population, a number of predictor caustic injury), twice in the malignant thoracic group, and variables, listed in Table 1, were then analyzed using 7 times in the benign thoracic group. No diversions were univariate and multivariate logistic regression to deter- performed in the cervical perforation group. mine which factors had a statistically significant effect on With respect to perioperative mortality, there were
  3. 3. Ann Thorac Surg BHATIA ET AL 2112011;92:209 –15 ESOPHAGEAL PERFORATIONTable 1. Univariate Analysis Factor No Mortality (n ϭ 97) Mortality (n ϭ 22) p Value Mean age 60.8 65.7 p ϭ 0.99 Median age (range) 60 (8–92) 63.5 (46–85) p ϭ 0.99 Female gender 28 (28.9%) 7 (31.8%) p ϭ 0.80 Perforation type Malignant 7 (7.2%) 10 (45.5%) p ϭ 0.0008a Iatrogenic 35 (36.1%) 4 (18.2%) p ϭ 0.32 Spontaneous 46 (47.4%) 8 (36.4%) p ϭ 0.66 Postoperative 9 (9.2%) 0 (0%) p ϭ 1 Preop morbidity GENERAL THORACIC Cardiac 18 (18.6%) 4 (18.1%) p ϭ 1 Pulmonary 14 (14.4%) 10 (45.5%) p ϭ 0.02a Smoker 46 (47.4%) 16 (72.3%) p ϭ 0.26 Renal failure 2 (2.1%) 1 (4.5%) p ϭ 0.47 Diabetes 7 (7.2%) 2 (9.1%) p ϭ 0.68 Hepatic disease 3 (3.1%) 3 (13.6%) p ϭ 0.11 Alcohol abuse 23 (23.7%) 5 (22.7%) p ϭ 1 Malnutrition 12 (12.3%) 13 (59.1%) p ϭ 0.001a Steroid use 7 (7.2%) 2 (9.1%) p ϭ 0.68 Charlson comorbidity index (mean) 2.8 7.1 p ϭ 0.0004a Presentation Sepsis 45 (46.4%) 18 (81.2%) p ϭ 0.08a Hemodynamic instability 14 (14.4%) 4 (18.1%) p ϭ 0.75 Mechanical ventilation 16 (16.5%) 13 (59.1%) p ϭ 0.006a Mean time to treatment 54.3 153.3 N/A Median time to treatment (range) 28.5 (0–480) 36 (0.5–480) p ϭ 0.52 Treatment Primary repair 69 (71.1%) 8 (36.4%) p ϭ 0.2 Resection and reconstruction 4 (4.1%) 5 (22.7%) p ϭ 0.9 Diversion 10 (10.3%) 1 (4.5%) p ϭ 0.9 Drainage alone 5 (5.2%) 3 (13.6%) p ϭ 0.9 Conservative 9 (9.3%) 5 (22.7%) p ϭ 0.9a p Ͻ 0.1.Univariate analysis of esophageal perforations based on various factors’ influence on mortality. Categoric variables were analyzed using the Fisher exacttest, and continuous variables were analyzed using the Wilcoxon two-sample test.N/A ϭ not applicable; Preop ϭ preoperative.only 2 deaths in the cervical perforation group (8%), one who leaked underwent a reoperation, as 4 of thein the abdominal group (11%), and 10 patients in the patients and their families elected not to undergo abenign thoracic group (13%). The perioperative 30-day second operation and desired compassionate care. Themortality of patients with malignant thoracic perforations median time from perforation to initial treatment forwas 63%. those patients subsequently presenting with a second- The mean and median lengths of stay were consid- ary leak was considerably higher in the malignanterably higher in both the malignant and benign tho- thoracic perforation group (168 hours), although theracic perforations as compared with the cervical and number of patients was quite small. In reviewing theabdominal perforations (Table 4). The median number initial treatment strategy for those patients who had aof critical care unit days, and median number of days secondary esophageal leak, the leak in the cervicalof mechanical ventilation during their hospital stay perforation group followed an initial resection andwas lowest in the cervical perforation group (4.5 and diversion procedure (distal staple line dehiscence).1.5 days, respectively) and highest in the malignant The primary treatments for the 18 leaks in the nonma-thoracic perforation group (9.5 and 8.5 days, respec- lignant thoracic perforation group were the following:tively). Additionally, postoperative leak rates were 9 primary repairs, surgical drainages with decortica-highest in the malignant thoracic perforation group tions, 3 resection and diversions, 1 exclusion and(43.7%), and lowest in the abdominal perforation group diversion, and 2 chest-tube drainages without surgical(0%). Only 3 of the 7 malignant thoracic perforations decortication. The primary treatments for the 7 leaks in