Journal of Surgical Oncology 2015;111:855–861
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment ROBERT C.G. MARTIN, MD, PhD, RUSSELL W. FARMER, MD, R. CHARLES ST. HILL, MD, KELLY M. MCMASTERS, MD, PhD, AND CHARLES R. SCOGGINS, MD, MBA Department of Surgery, Division of Surgical Oncology, University of Louisville, Louisville, Kentucky
Background: Esophageal anastomotic leak is associated with high morbidity and mortality and potentially lethal if not recognized promptly and treated aggressively. While many studies have attempted to delineate the surgical techniques associated with lower rates of anastomotic leak, few have attempted to determine its long term effects on adjuvant therapy and patient quality of life. Methods: We reviewed our prospective 350 patient esophageal-gastric database and found 194 esophageal anastomoses performed from 1994 to 2013. Leaks were classified based on timing, severity, and location. We then compared their postoperative courses of adjuvant treatment, morbidity/ mortality, and quality of life measured by Karnofsky performance status and EORTC score. Statistical correlations were calculated with x2,T-test, Kaplain–Meier, ANOVA, and Cox Regression analyses as appropriate. Results: Of 194 patients receiving esophageal anastomoses for cancer, 35 (18%) developed clinically diagnosed anastomotic leak 27 from esophagogastric and 8 from esophagojejunal. These groups were similar in preoperative history, preoperative staging, and type of tumor. Type of operation and variations in operative technique did not significantly affect leak rate. Patients with a leak were more likely to require intraoperative transfusion (47.1% vs. 24.1%, P ¼ 0.013). As expected, they had a greater prevalence of perioperative complications to include pneumonia (38.6% vs. 16.3%, P ¼ 0.001), pulmonary embolus (11.3% vs. 4.3%, P ¼ 0.043), ileus (11.4% vs. 1.6%, P ¼ 0.006%), empyema (11.4% vs. 0%, P > 0.001), and catheter related blood stream infections (8.6% vs. 0%, P ¼ 0.001). Despite this increase in perioperative morbidity, there was no statistically significant difference in 90 day peri-operative mortality (2.8% vs. 2.3%) with similar ability to receive adjuvant therapy (38.6% vs. 48.0%, P ¼ 0.303), quality of life scores (93.2 vs. 93.1, P ¼ 0.9), and survival at 12 months (93% vs. 94%, P ¼ 0.3). Conclusion: Anastomotic leak after oncologic resection does not preclude adjuvant therapy and, when managed appropriately, does not affect long term performance status or survival.
J. Surg. Oncol. 2015;111:855–861. ß 2015 Wiley Periodicals, Inc.
KEY WORDS: anastomotic leak; esophageal cancer; adjuvant treatment; morbidity; quality of life
INTRODUCTION Esophageal anastomosis is a necessary component of surgical treatment of many types of foregut cancer. Esophageal cancer alone accounts for 14,000–15,000 new cases in the US every year [1]. While many various techniques of esophageal reconstruction are employed [2–4], esophagogastric and esphagoenteric anastomoses are the most common. Anastomotic leak is a dreaded complication following esophageal surgery, with rates reported as low as 3.5% and as high as 26% [5]. These patients have been shown to at worse have in increase in 30-day mortality to at best no change in Overall Survival (OS) [6]. Much has been published regarding the risk factors for esophageal anastomotic leak (EAL) [7–10] (Table I), with much focus placed on the technical methods of anastomoses [5,11,12]. Similar controversy exists even in the diagnosis of this complication and the differentiation of a clinically relevant versus a radiographicaly apparent leak [13,14]. However, little has been published on the effect of EAL long term on the patient population with regards to adjuvant treatment as well as overall quality of life (QOL). The goal of this study is to determine the effect of esophageal anastomotic leak on patients’ ability to receive adjuvant therapy as well as to compare quality of life in patients with leak to those without leak following esophageal anastomoses for cancer. Risk factors for esophageal anastomosis were analyzed as a secondary goal.
METHODS We reviewed our single-center, IRB approved prospective database of 350 esophageal-gastric patients from 1994 to 2009 identifying 194
ß 2015 Wiley Periodicals, Inc.
esophageal anastomoses. These included anastomoses performed for the treatment of both esophageal and gastric cancer. Patients with isolated gastric resection were excluded. Factors affecting outcome were compared between those patients developing anastomotic leak versus those without. Esophageal anastomotic leak was diagnosed via extravasation of contrast on postoperative gastrograffin swallow and stratified according to grading system. We utilized prior classification of leaks based on three parameters: time of appearance after surgery, magnitude or clinical severity, and location of the leaks [15–18]. Thus, early leaks were classified as those that appeared 1–4 days after surgery; intermediate leaks those that appeared 5–9 days after surgery, and late leaks those that appeared 10 or more days after surgery. Furthermore, type I or subclinical are those that appear as a localized leak, without spillage or dissemination, with few clinical manifestations and easy to treat medically. Type II leaks are those with dissemination or diffusion into the abdominal or pleural cavity, by way of an irregular pathway, with the appearance of contrast medium (methylene blue, radiological contrast) or food through any of the abdominal drain, with severe clinical consequences. In this study, the exact day of the appearance of
*Correspondence to: Robert CG Martin II, MD, PhD, FACS, Director, Division of Surgical Oncology, 315 E Broadway, Ste 311, Louisville 40206, KY. Fax: þ502-629-3030. E-mail: [email protected] Received 15 November 2014; Accepted 7 January 2015 DOI 10.1002/jso.23902 Published online 13 April 2015 in Wiley Online Library (wileyonlinelibrary.com).
856
Martin et al.
TABLE I. Published studies on Esophageal Anatasmoses Author
Year published
Study period
Study type Determine the influence of elevated body mass index on outcomes after esophagectomy for cancer.
Melis et al.
2010
Median follow-up of 25 months, and a 5 years overall and disease -free survivals follow-up.
Kim el al.
2010
Literature review.
Escofet X et al.
2010
Scipione et al.
2007
N/A
Morgan et al.
2007
Prospective studies: Patients were reviewed every 3 months for the first year and every 6 months thereafter. Patients were followed up for a minimum of 5 years or until death. One hundred and forty-one patients (66%) were followed up until death or 5 years.
Griffin et al.
2001
N/A
Karl et al.
2000
N/A
Comparing different methods of esophagogastric anastomoses. Determine the Patients undergoing contemporary esophagectomy were prevalence, outcome, reviewed every and survival after 3 months for the first esophagogastric year, and every anastomotic leakage 6 months thereafter for (EGAL) following at least 5 years or until esophagectomy. death.
Journal of Surgical Oncology
# of patients & demography A total of 490 patients were analyzed; 166 were obese (33.9%), 176 overweight (35.9%), 148 (30.2%); 420 men, and 70 women. N/A
% leaks
% received adjuvant therapy
9 (5.4%) of obese patients had leaks, 7 (4.0%) of the overweight patients and; 11 (7.4%) were normal.
47.6% of obese patients; 54.5% of overweight patients; 66.2% normal pts.
N/A
N/A
105 neoadjuvant Twenty patients Two hundred forty chemotherapy, 49 developed EGAL consecutive patients chemoradiotherapy). (8.3%, 15 managed underwent conservatively, 5 esophagectomy over a reoperations). 10-year period (median age 61 [31–79] years, 147 transthoracic and 93 trans hiatal esophagectomy, 105 neoadjuvant chemotherapy, 49 chemoradiotherapy). N/A Evaluation of the impact 266 (133 paired sets): 204 Total: Obese ¼ 22 (17%), malignant (Obese- 75 of obesity on the Non-obese ¼ 19 men and 27 women; outcomes of (14%); Malignant: non-obese- 77 men and esophagectomy to Obese ¼ 19 (19%), 25 women), 62 benign determine whether Non-obese ¼ 13 (Both- 18 men, 13 profound obesity (13%); Benign: women). should contraindicate Obese ¼ 3 (10%), the transhiatal Non-obese ¼ 6 (19%). approach. 8 (8%) of normal or low Of the 215 patients who A total of 250 pts.; 97 Determine the role of underwent resection, 62 BMI pts. and 6(5%) of (45%) had low or body mass index (BMI) had neoadjuvant High BMI Pts. normal BMIs, 86 in a Western population chemo-therapy (37 (40%) were on outcomes after with high BMI, 25 with overweight, and a esophagectomy for normal or low outcome further 32 (15%) were cancer. after esophagectomy obese. related to BMI 31 BMI) and 49 had neoadjuvant chemoradiation (37 with high BMI, 29 with normal or low BMI). The remainder of patients had surgery without neoadjuvant treatment. 13 patients had an N/A Evaluate the diagnosis, Some 291 consecutive isolated leak. management and patients underwent two-phase subtotal outcome of mediastinal oesophagectomy with leaks following radical gastric interposition for oesophagectomy with a malignancy. stapled intrathoracic anastomosis. Examine the safety of 143 patients treated by Five patients (3.5%) had Preoperative neoadjuvant transthoracic a documented treatment was delivered TTE; 127 men and 16 esophagogastrectomy anastomotic leak; to 28 patients. women. (TTE) in a multidisciplinary cancer center and to determine which clinical parameters influenced survival and the rates of death and complications.
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment
857
TABLE I. (Continued) Author Beitler et al.
Year published 1998
Study period Literature review
# of patients & demography
Study type Comparison of stable and N/A hand-sewn esophagogastric anastomosis
symptoms, the diagnosis of the presence of a leak, the medical or surgical treatment, the evolution, and the day of closure were carefully recorded. The management of leaks at our institution is based on first adequate drainage/control/management of a leak–defined by a CT of the chest that does not demonstrate any un-drained fluid. Once control is confirmed, then our bias in patients >10 days out from anastomosis is to utilize esophageal stenting to exclude the leak, which we have have demonstrated can lead to quicker drainage tube removal and quicker time to discharged and possible PO intake [19,20]. In the leak patients, only once the leak was controlled with esophageal stenting (controlled defined as no drainage tubes of catheters) or healed was adjuvant therapy initiated. Patient characteristics were assessed by routine history and physical, as well as preoperative biopsy of suspected lesions, complete metabolic panel, and imaging. Preoperative imaging including CT scans of the chest, abdomen, and pelvis at minimum with PET scanning if CT demonstrated suspicious lymphadenopathy. Neo-adjuvant therapy was utilized in all stage 3 patients and in high-risk stage 2 patients. We have predominantly utilized three forms of therapy–1st weekly carbotaxol with XRT per the CROSS trial [21], 2nd neo-adjuvant EpirubicinCisplatin-Fluoruracil per the MAJIC trial [22], and 3rd Cisplatin– Fluoruracil with XRT based on the: (i) Tumor location–gastic versus gastro-esophageal versus esophageal; (ii) histology–adenocarcinoma versus squamous cell; (iii) Degree of lymph node involvement at diagnosis based on CT and CT-PET scanning. Factors potentially associated with esophageal anastomotic leak were identified by literature search as well as anecdotal evidence. Complications were graded on a scale previously presented [23] with grades 1–2 requiring no invasive intervention, 3–4 requiring either percutaneous drainage, reoperation, ICU readmission, or grade 5 representing death. These data were accrued prospectively over the course of each patient’s course. Initial postoperative follow-up occurred between 2–3 weeks with 3–6 month follow up for the next 2 years. At each follow up visit, patients completed the Karnofsky Performance Status evaluation as a proxy for quality of life [24] as well as the Global Rating Scale [25,26]. Post-hoc univariate analyses were performed using x2, independent T-Test, Fisher’s Exact Test, and Log Rank testing. Multivariate analysis was performed using ANOVA. Survival
% leaks N/A
% received adjuvant therapy N/A
comparisons were performed in the method of Kaplan and Maier. All statistical analyses were performed using PASW/SPSS (IBM, Somers, NY). An a of .05 was considered statistically significant.
RESULTS Within the cohort of patients undergoing surgery for gastric and esophageal cancers over a 10 year period, 194 patients received esophageal anastomoses as part of total gastrectomy (57, 29%) or IvorLewis esophagogastrectomy 137 (71%). In those patients, 35 (18.3%) developed esophageal leak based on clinical and radiographic diagnoses. In the EAL patients eight developed this after total gastrectomy with the remainder 27 after esophagogastrectomy. There were no early leaks, 12 were intermediate time interval, and 23 were late. The diagnosis was Type 1 in 28 patients with Type 2 in seven patients. The median severity grade of EAL was two with the majority of these (60%) requiring observation and continued drainage, 29% required intervention (either interventional radiographic treatment or thoracostomy), and 11% of leak patients having only incidental findings on gastrograffin swallow without clinical signs or symptoms. 90-day mortality was similar in the leak and no leak group (2.8% vs. 2.3%). The median age was similar in the EAL versus the non-leak group (64 vs. 61, P ¼ 0.9, Table II), as was the gender make-up (68.6% male vs. 80.4% male, P ¼ 0.3). The past medical history of both groups was similar with regards to: cardiac history (22.9% vs. 23.3%, P ¼ 0.96), pulmonary risk factors (14.3% vs. 10.1% P ¼ 0.5), and diabetes (20% vs. 14.0% P ¼ 0.4). The rate of substance abuse was also similar for both tobacco (62.9% vs. 58.1% P ¼ 0.615), median pack-years (43.3 vs. 48.0, P ¼ 0.7), and frequent ethanol use (20.9% vs. 14.3% P ¼ 0.4). A similar percentage experienced preoperative weight loss (45.7% vs. 53.5% P ¼ 0.4). There was no difference in median preoperative albumin level (CI 0.36 to 0.06). The use of neoadjuvant chemotherapy did not affect the development of esophageal leak (34.3% vs. 38.8% P ¼ 0.6). The histologic origin of the tumors was conserved between groups as well with similar proportions of patients with adenocarcinoma (77.1% vs. 78.3%, P ¼ 0.8). For patients with esophageal cancer, patients that
TABLE II. Preoperative Patient Characteristics
Age (median) Gender (%Male) Cardiac Dz Pulmonary Diabetes Alcohol Tobacco Mean pack years Squamous cell Adenocarcinoma Weight Loss Albumin level (median) Preop chemo Rx
Journal of Surgical Oncology
Leak N ¼ 35 (18%)
No leak N ¼ 159 (82%)
P value
64 68.6% 8 (23%) 5 (14%) 7 (20%) 5 (21%) 22 (63%) 43.3 7 (20%) 27 (77%) 16 (46%) 3.7 12 (34%)
61 80.4% 37 (23%) 15 (10%) 22 (14%) 25 (14%) 92 (58%) 48 31 (19%) 128 (81%) 84 (53%) 3.6 62 (39%)
0.914 0.254 0.960 0.480 0.378 0.3790 0.615 0.729 0.842 0.842 0.414 0.164 0.638
858
Martin et al.
experienced leak did not have a statistically significant different location of tumor from those without (P ¼ 0.3). Preoperative T stage (P ¼ 0.6) and N stage (P ¼ 0.4) were the same in the leak and no-leak groups. The presence of leak was not dependent upon the procedure performed, either esophagogastrectomy or total gastrectomy (P ¼ 0.6, Table III). There was no difference in the rate of leak based on operative technique. Neither the use of a stapler (48.5% vs. 43%, P ¼ 0.6), nor the type of hand-sewn anastomosis were risk factors for development of esophageal leak (P ¼ 0.1). Intraoperatively, there was no difference in mean estimated blood loss (SEM ¼ 60.5, P ¼ 0.7); however, a significantly larger proportion of those patients developing esophageal leak required red blood cell transfusion (57.1% vs. 34.1%, P ¼ 0.013). Of those patients that required transfusion, the amount was not increased in those patients developing a leak with both groups having a median transfusion requirement of 2 units (P ¼ 0.7, SEM ¼ 0. 31). All patients were admitted to the ICU postoperatively, with leak patients having a significantly longer median ICU Length of Stay (LOS, 10 vs. 4.5, P ¼ 0. 006, Table III). As expected patients with a leak had higher rates of concomitant postoperative complications including pneumonia (48.6% vs. 16.3%, P < 0.001), pleural effusion (14.3% vs. 4.3%, P ¼ 0.043), ileus (11.4% vs. 1.6%, P ¼ 0.006), empyema (11.4% vs. 0, P < 0.001), and central venous catheter related blood stream infections (8.6% vs. 0% P ¼ 0.001). Other postoperative complications showed neither statistical significance nor trend toward significance (Table III). Mortality up to 90 days post-op was similar with 0% mortality in patients with leak compared to 2.3% mortality (3/129). Oncologic factors did not predict esophageal leak with no difference found in our cohort between T stage (P ¼ 0.4), N stage (P ¼ 0.4), M stage (P ¼ 0.8), margin positivity (P ¼ 0.6) or number of positive lymph nodes (P ¼ 0.2). Of those factors associated with statistically significant adverse outcome in the leak group, only pneumonia, effusion, ileus, empyema, and central line infection remained statistically significant on ANOVA analysis. Those patients developing EAL were divided into two groups: lowgrade [1–2] EAL, treated only with medical therapy and observation, and high-grade (Grade 3) EAL, which required intervention. The majority of leaks were low grade (60%) with high grade (29%). There was a trend toward increasing grade of leak with esophagogastrectomy rather than total gastrectomy, but this was not statistically significant.
There was no difference as well in type of anastomosis (P ¼ 0.3), EBL, transfusion requirement (P ¼ 0.2), or ICU LOS. Many of the complications seen were in those patients with high-grade leaks, but the divisions of EAL patients into these groups lead to insufficient power for statistical significance (Table III). Though a greater percentage of low-grade EAL patients received adjuvant therapy, this was not statistically significant. The Karnofsky score of both low and high-grade leak patients was the same at 6 months, with each group showing a mean of 90 or greater. Patient who suffered a leak had a worse KPS at 6 weeks post op (median 70 for leak and 90 for no leak), however the median score for both a 3 months was 90%, and remained similar, with the only major decreases in KPS being related to adjuvant chemotherapy and or Chemo-radiation therapy. Similar differences were seen in contrast, when all of the patients completed their global quality of life scores by the QLQ-C30. There were statistical differences see in overall QOL at the 6 week post-operative visit with the most dramatic changes occurring in relation to questions: “How would you rate your overall health during the past week” and “How would you rate your overall quality of life during the past week,” both on a seven point scale, there was a significant loss of overall QOL from baseline (Fig. 1) (P ¼ 0.02 for 6 weeks post-operative) Patients with esophageal leak underwent adjuvant chemotherapy at rates comparable to those patients without leak (28.6% vs. 38.0%, P ¼ 0.3, Table IV). The type of adjuvant therapy was decide upon based on what form of neo-adjuvant therapy was utilized. If they received three cycles of Epirubin-Cisplatin-Fluorouracil (ECF), then they recived another three cycles of ECF, restage and if they had a >20% lymph node positive to lymph node negative ration then they received adjuvant 5FU and XRT to the nodal bed. If they had neoadjuvant Carbo-taxol with XRT and had a moderate to poor response based on pathology and had >20% LN ratio then they received four cycles of ECF based chemotherapy or FOLFOX based chemotherapy. If they had a good response, but with >20% lymph node ration then they received FOLFOX adjuvant therapy. The median time to initiate adjuvant therapy was 48 days (approximately 6 weeks) with a range of 35–90 days. In total 75% of all patients received their planned adjuvant therapy with no difference in successful completion of adjuvant therapy in the leak (72%) and the no leak group (79%). They also developed similar rates of dysphagia (5.7% vs. 8.5%, P ¼ 0.6), and stricture (8.6% vs. 7.8% P ¼ 0.9). Both groups of patients had a minority that required esophageal dilation for intractable
TABLE III. Operative Factors and Complications Compared for Affect on Leak
Esophagogastrectomy total gastrectomy Stapled hand-sewn EBL (median, mean) Blood transfusion # of Units (median) Length of stay–ICU (median,mean) Pneumonia Pleural Effusion New arrhythmia (afib/svt) Ileus beyond expected Fever Confusion Respiratory failure Empyema CVC related BSI Acute renal failure
Journal of Surgical Oncology
No leak N ¼ 159 (%)
Leak N ¼ 35 (%)
Grade 1–2 leaks N ¼ 24 (71%)
Grade 3 leaks N ¼ 10 (29%)
P value*
119 (75%) 40 (25%)
0% 4 (17%)
6 (60%) 4 (40%)
0.574
69 (43%) 90 (57%) 350, 488.3 54 (34%) 2
28 (80.0%) 7 (20%) 16 (46%) 19 (54%) 400, 543.3 20 (57%) 2
3 (12%) 1 (6%) 516 8 (33%) 2.5
4 (40%) 6 (60%) 498 6 (60%) 2.3
4.5, 10.3 21 (13%) 6 (4%) 17 (11%) 2 (1%) 9 (6%) 6 (4%) 10 (7%) 0 0 3 (2%)
10, 10.6 17 (49%) 5 (14%) 7 (20%) 4 (11%) 4 (11%) 2 (6%) 6 (17%) 4 (11%) 3 (9%) 2 (6%)
10.0 8 (33%) 0 (0%) 4 (17%) 8 (33%) 4 (17%) 4 (17%) 0 0 0 0
0.569 0.741 0.013, 0.195 .695 SEM ¼ 0.31 0.006 >0.001 0.043 0.311 0.006 0.336 0.796 0.097 >0.001 0.001 0.301
5 2 2 1 1 1 2 1 1 1
11.7 (50%) (20%) (20%) (10%) (10%) (10%) (20%) (10%) (10%) (10%)
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment
859
Fig. 1. Global quality of life score at baseline, 6 weeks, 3 months, 6 months, 9 months, and 12 months after esophagectomy for patients who developed a leak versus patients who did not develop leak. dysphagia postoperatively (5.7% vs. 1.6% P ¼ 0.2). The average Karnofsky Score was the same for both groups at 6 months (93.2 vs. 93.1 (P ¼ 1). Furthermore, the percentage of patients presenting with a Karnofsky Score of 80 or greater, representing that level at which no assistance is needed for daily activities, was conserved between patients with and without esophageal leak (89.5% vs. 92.9%, P ¼ 0.6). The presence of esophageal leak did not affect 12 month overall survival (93% vs. 94%, P ¼ 0.3).
DISCUSSION We present a review of our experience with 194 esophageal anastomoses performed for oncologic resection. In this patient cohort, representing 10 years of cases at our region’s major cancer center, there was no association between the development of esophageal leak and overall survival. The major caveat to this of course, remains that it is difficult to tell the true impact of esophageal leak on OS given the
TABLE IV. Follow Up Evaluation No leak N ¼ 159 (78.7) Stricture Dilation Dysphagia Adjuvant Rx Karnofsky Score (avg @ 6 mos.) K. Score > 80 þ (independent) AWD/NED @ 12 months *
14 (9%) 10 (6%) 10 (6%) 46 (29%) 93.2 143 (90%) 93%
EAL N ¼ 35 (21.3%) 3 2 11 12
(8%) (6%) (31%) (38%) 93.1 33 (92.9%) 94%
Grade 1–2 EAL N ¼ 24 (71%)
Grade 3 EAL N ¼ 10 (29%)
P value*
8 (33%) 4 (17%) 8 (33%) 12 (50%) 90.0 16 (66%) 95%
1 (10%) 1 (10%) 0 2 (20%) 93.2 5 (50%) 75%
0.874, 0.017 0.158 0.586, 0.001 0.303 0.984 0.627 0.273
Second listed P-value represents comparison between low-grade and high-grade EAL. If no second P-value is listed, it is not significant.
Journal of Surgical Oncology
860
Martin et al.
propensity of progression of most stage three esophageal and gastric cancers. In addition, the development of esophageal leak had no effect on the ability of patients in our cohort to receive adjuvant therapy. Anastomotic leak can be a devastating complication when not managed appropriately, one that depletes the patient physiologically and mentally. Indeed, leak led to many of the more dreaded complications seen in this cohort postoperatively. The anecdotal ideal of treating physicians would be to avoid chemotherapy and radiotherapy in a patient that has recently undergone such an insult. However, the similarity of survival in patients with and without leak as stratified by the receipt of adjuvant therapy contradicts this particular prejudice (Fig. 1). Much has been made about the diagnosis of esophageal leak: how best to diagnose it, its manifestations, the difference between a clinical relevant leak, and a non-relevant one. However, relatively little has been published as to the effects of esophageal leak on the patient in the long term. In these 164 patients, no difference was noted in average Karnofsky score at 6 months for all patients. There was no difference in proportion of patients that were functional on a daily basis at home between patients with and without leak at 6 month follow up as well. As such, esophageal leak acutely affects QOL, but does not appear to do so in the long term. The clinicopathologic factors analyzed here are consistent with those previously postulated to be potentially associated with esophageal leak. The only factor in this cohort that consistently predicted the development of esophageal leak was the requirement of RBC transfusion (P ¼ 0.013). No one operative technique was superior, nor were the AJCC staging criteria associated with leak development. Our rate of leak, 21% is commensurate with that of currently published studies [7], with other studies describing a lower rate. However, Whooley and colleagues [27] report a significant difference in hospital mortality among 475 esophagectomy patients that was related to leak (35% vs. 9%). This is corroborated by Ramirez et al in review of 276 patients [8]. While not specifically analyzed for hospital or 30-day mortality, our data likely differs from that of these studies by virtue of leak classification. With fewer, 3.5–5% of patients developing an anastomotic leak coupled with the much higher mortality for these patients, the sensitivity of leak diagnosis and inclusion in survival analysis appear to be higher in our patient population. Wright et al recently published their experience with 2,315 esophagectomies over multiple institutions [10]. Their analysis revealed a higher leak rate more commensurate with our own (11.3%). Additionally, they analyzed factors associated with anastomotic leak. However, they included all “major” morbidities associated with esophagectomy in their analysis. As such, the numerous significant factors they associate with the development of leak such as history of CHF, peripheral vascular disease, and African American race cannot truly be extrapolated solely to that group of patients developing anastomotic leak. Escofet and colleagues [6] published a series of 240 cases, 105 of which underwent neoadjuvant therapy. Their rate of leak was similar to ours, approximately 20%. However, in contrast to our analysis, they associated the use of neoadjuvant treatment with the development of leak. The other factors they identified, low Body Mass Index and surgeon experience, were not analyzed in our review. Karl et al [9] reviewed 143 esophagectomy patients with a very low leak rate (3.5%). The only factor identified with increased risk of leak as the presence of diabetes mellitus. AJCC stage, transfusion requirement, and presence of positive lymph nodes were not associated with leak development. This is consistent with our findings with the exception of the transfusion requirement. A recent meta-analysis of leaks was published by Schaheen et al, which confirmed our management and findings in that Intraoperative anastomotic drain placement was associated with earlier identification and resolution of anastomotic leak (mean 23.4 vs. 80.7 days). In addition, reinforcement of the anastomosis with omentoplasty may reduce the incidence of anastomotic leak by nearly 50%. Endo-scopic stent placement was associated with leak resolution in 72%. Journal of Surgical Oncology
CONCLUSIONS We have shown that the development of esophageal leak, when managed conservatively and well, does not affect overall survival at 6 months. In addition and novelly, esophageal leak does not preclude patients from undergoing postoperative adjuvant therapy. Finally, there is no difference between those patients with leak and without regarding performance status at 6 month follow up. Additional research is needed to determine the overall effect of esophageal leak on patients undergoing esophageal anastomoses.
REFERENCES 1. SEER, SEER Database, 2007, NCI. 2. Mine S, et al.: Colon interposition after esophagectomy with extended lymphadenectomy for esophageal cancer. Ann Thorac Surg 2009;88:1647–1653. 3. Stein HJ, et al.: Limited surgical resection and jejunal interposition for early adenocarcinoma of the distal esophagus. Semin Thorac Cardiovasc Surg 2007;19:72–78. 4. Visbal AL, et al.: Ivor Lewis esophagogastrectomy for esophageal cancer. Ann Thorac Surg 2001;71:1803–1808. 5. Kim RH, Takabe K. Methods of esophagogastric anastomoses following esophagectomy for cancer: A systematic review. J Surg Oncol 2010;101:527–533. 6. Escofet X, et al.: Prevalence and outcome of esophagogastric anastomotic leak after esophagectomy in a UK regional cancer network. Dis Esophagus 2010;23:112–116. 7. Atkins BZ, et al.: Reducing hospital morbidity and mortality following esophagectomy. Ann Thorac Surg 2004;78:1170–1176, discussion 1170-1176. 8. Junemann-Ramirez M, et al.: Anastomotic leakage postesophagogastrectomy for esophageal carcinoma: retrospective analysis of predictive factors, management and influence on longterm survival in a high volume centre. Eur J Cardiothorac Surg 2005;27:3–7. 9. Karl RC, et al.: Factors affecting morbidity, mortality, and survival in patients undergoing Ivor Lewis esophagogastrectomy. Ann Surg 2000;231:635–643. 10. Wright CD, et al.: Predictors of major morbidity and mortality after esophagectomy for esophageal cancer: A Society of Thoracic Surgeons General Thoracic Surgery Database risk adjustment model. J Thorac Cardiovasc Surg 2009;137:587–595,discussion 596. 11. Beitler AL, Urschel JD. Comparison of stapled and handsewn esophagogastric anastomoses. Am J Surg 1998;175:337– 340. 12. Okuyama M, et al.: Hand-sewn cervical anastomosis versus stapled intrathoracic anastomosis after esophagectomy for middle or lower thoracic esophageal cancer: A prospective randomized controlled study. Surg Today 2007;37:947–952. 13. Tirnaksiz MB, et al.: Effectiveness of screening aqueous contrast swallow in detecting clinically significant anastomotic leaks after esophagectomy. Eur Surg Res 2005;37:123–128. 14. Upponi S, et al.: Radiological detection of post-oesophagectomy anastomotic leak - a comparison between multidetector CT and fluoroscopy. Br J Radiol 2008;81:545–548. 15. Csendes A, Burgos Braghetto AM, Braghetto I, et al.: Classification and management of leaks after gastric bypass for patients with morbid obesity: A prospective study of 60 patients. Obes Surg 2012;22:855–862. 16. Csendes A, et al.: Management of leaks after laparoscopic sleeve gastrectomy in patients with obesity. J Gastrointest Surg 2010;14:1343–1348. 17. Csendes A. Conservative management of anastomotic leaks. Obes Surg 2006;16:375–376,author reply 376. 18. Csendes A, et al.: Conservative management of anastomotic leaks after 557 open gastric bypasses. Obes Surg 2005;15:1252–1256. 19. Martin R, et al.: The use of self-expanding silicone stents in esophageal cancer care: optimal pre-, peri-, and postoperative care. Surg Endosc 2009;23:615–621.
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment 20. Martin RC, et al.: The use of self-expanding silicone stents in esophagectomy strictures: Less cost and more efficiency. Ann Thorac Surg 2008;86:436–440. 21. van Hagen P, et al.: Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012;366: 2074–2084. 22. Cunningham D, et al.: Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 2006;355:11–20. 23. D’Angelica M, et al.: Major hepatectomy with simultaneous pancreatectomy for advanced hepatobiliary cancer. J Am Coll Surg 2004;198:570–576.
Journal of Surgical Oncology
861
24. O’Toole DM, Golden AM. Evaluating cancer patients for rehabilitation potential. West J Med 1991;155:384–387. 25. Brown RE, et al.: A prospective phase II evaluation of esophageal stenting for neoadjuvant therapy for esophageal cancer: Optimal performance and surgical safety. J Am Coll Surg 2011;212:582– 588,discussion 588-589. 26. Mbah N, et al.: Impact of post-operative complications on quality of life after pancreatectomy. JOP 2012;13:387–393. 27. Whooley BP, et al.: Critical appraisal of the significance of intrathoracic anastomotic leakage after esophagectomy for cancer. Am J Surg 2001;181:198–203.
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment ROBERT C.G. MARTIN, MD, PhD, RUSSELL W. FARMER, MD, R. CHARLES ST. HILL, MD, KELLY M. MCMASTERS, MD, PhD, AND CHARLES R. SCOGGINS, MD, MBA Department of Surgery, Division of Surgical Oncology, University of Louisville, Louisville, Kentucky
Background: Esophageal anastomotic leak is associated with high morbidity and mortality and potentially lethal if not recognized promptly and treated aggressively. While many studies have attempted to delineate the surgical techniques associated with lower rates of anastomotic leak, few have attempted to determine its long term effects on adjuvant therapy and patient quality of life. Methods: We reviewed our prospective 350 patient esophageal-gastric database and found 194 esophageal anastomoses performed from 1994 to 2013. Leaks were classified based on timing, severity, and location. We then compared their postoperative courses of adjuvant treatment, morbidity/ mortality, and quality of life measured by Karnofsky performance status and EORTC score. Statistical correlations were calculated with x2,T-test, Kaplain–Meier, ANOVA, and Cox Regression analyses as appropriate. Results: Of 194 patients receiving esophageal anastomoses for cancer, 35 (18%) developed clinically diagnosed anastomotic leak 27 from esophagogastric and 8 from esophagojejunal. These groups were similar in preoperative history, preoperative staging, and type of tumor. Type of operation and variations in operative technique did not significantly affect leak rate. Patients with a leak were more likely to require intraoperative transfusion (47.1% vs. 24.1%, P ¼ 0.013). As expected, they had a greater prevalence of perioperative complications to include pneumonia (38.6% vs. 16.3%, P ¼ 0.001), pulmonary embolus (11.3% vs. 4.3%, P ¼ 0.043), ileus (11.4% vs. 1.6%, P ¼ 0.006%), empyema (11.4% vs. 0%, P > 0.001), and catheter related blood stream infections (8.6% vs. 0%, P ¼ 0.001). Despite this increase in perioperative morbidity, there was no statistically significant difference in 90 day peri-operative mortality (2.8% vs. 2.3%) with similar ability to receive adjuvant therapy (38.6% vs. 48.0%, P ¼ 0.303), quality of life scores (93.2 vs. 93.1, P ¼ 0.9), and survival at 12 months (93% vs. 94%, P ¼ 0.3). Conclusion: Anastomotic leak after oncologic resection does not preclude adjuvant therapy and, when managed appropriately, does not affect long term performance status or survival.
J. Surg. Oncol. 2015;111:855–861. ß 2015 Wiley Periodicals, Inc.
KEY WORDS: anastomotic leak; esophageal cancer; adjuvant treatment; morbidity; quality of life
INTRODUCTION Esophageal anastomosis is a necessary component of surgical treatment of many types of foregut cancer. Esophageal cancer alone accounts for 14,000–15,000 new cases in the US every year [1]. While many various techniques of esophageal reconstruction are employed [2–4], esophagogastric and esphagoenteric anastomoses are the most common. Anastomotic leak is a dreaded complication following esophageal surgery, with rates reported as low as 3.5% and as high as 26% [5]. These patients have been shown to at worse have in increase in 30-day mortality to at best no change in Overall Survival (OS) [6]. Much has been published regarding the risk factors for esophageal anastomotic leak (EAL) [7–10] (Table I), with much focus placed on the technical methods of anastomoses [5,11,12]. Similar controversy exists even in the diagnosis of this complication and the differentiation of a clinically relevant versus a radiographicaly apparent leak [13,14]. However, little has been published on the effect of EAL long term on the patient population with regards to adjuvant treatment as well as overall quality of life (QOL). The goal of this study is to determine the effect of esophageal anastomotic leak on patients’ ability to receive adjuvant therapy as well as to compare quality of life in patients with leak to those without leak following esophageal anastomoses for cancer. Risk factors for esophageal anastomosis were analyzed as a secondary goal.
METHODS We reviewed our single-center, IRB approved prospective database of 350 esophageal-gastric patients from 1994 to 2009 identifying 194
ß 2015 Wiley Periodicals, Inc.
esophageal anastomoses. These included anastomoses performed for the treatment of both esophageal and gastric cancer. Patients with isolated gastric resection were excluded. Factors affecting outcome were compared between those patients developing anastomotic leak versus those without. Esophageal anastomotic leak was diagnosed via extravasation of contrast on postoperative gastrograffin swallow and stratified according to grading system. We utilized prior classification of leaks based on three parameters: time of appearance after surgery, magnitude or clinical severity, and location of the leaks [15–18]. Thus, early leaks were classified as those that appeared 1–4 days after surgery; intermediate leaks those that appeared 5–9 days after surgery, and late leaks those that appeared 10 or more days after surgery. Furthermore, type I or subclinical are those that appear as a localized leak, without spillage or dissemination, with few clinical manifestations and easy to treat medically. Type II leaks are those with dissemination or diffusion into the abdominal or pleural cavity, by way of an irregular pathway, with the appearance of contrast medium (methylene blue, radiological contrast) or food through any of the abdominal drain, with severe clinical consequences. In this study, the exact day of the appearance of
*Correspondence to: Robert CG Martin II, MD, PhD, FACS, Director, Division of Surgical Oncology, 315 E Broadway, Ste 311, Louisville 40206, KY. Fax: þ502-629-3030. E-mail: [email protected] Received 15 November 2014; Accepted 7 January 2015 DOI 10.1002/jso.23902 Published online 13 April 2015 in Wiley Online Library (wileyonlinelibrary.com).
856
Martin et al.
TABLE I. Published studies on Esophageal Anatasmoses Author
Year published
Study period
Study type Determine the influence of elevated body mass index on outcomes after esophagectomy for cancer.
Melis et al.
2010
Median follow-up of 25 months, and a 5 years overall and disease -free survivals follow-up.
Kim el al.
2010
Literature review.
Escofet X et al.
2010
Scipione et al.
2007
N/A
Morgan et al.
2007
Prospective studies: Patients were reviewed every 3 months for the first year and every 6 months thereafter. Patients were followed up for a minimum of 5 years or until death. One hundred and forty-one patients (66%) were followed up until death or 5 years.
Griffin et al.
2001
N/A
Karl et al.
2000
N/A
Comparing different methods of esophagogastric anastomoses. Determine the Patients undergoing contemporary esophagectomy were prevalence, outcome, reviewed every and survival after 3 months for the first esophagogastric year, and every anastomotic leakage 6 months thereafter for (EGAL) following at least 5 years or until esophagectomy. death.
Journal of Surgical Oncology
# of patients & demography A total of 490 patients were analyzed; 166 were obese (33.9%), 176 overweight (35.9%), 148 (30.2%); 420 men, and 70 women. N/A
% leaks
% received adjuvant therapy
9 (5.4%) of obese patients had leaks, 7 (4.0%) of the overweight patients and; 11 (7.4%) were normal.
47.6% of obese patients; 54.5% of overweight patients; 66.2% normal pts.
N/A
N/A
105 neoadjuvant Twenty patients Two hundred forty chemotherapy, 49 developed EGAL consecutive patients chemoradiotherapy). (8.3%, 15 managed underwent conservatively, 5 esophagectomy over a reoperations). 10-year period (median age 61 [31–79] years, 147 transthoracic and 93 trans hiatal esophagectomy, 105 neoadjuvant chemotherapy, 49 chemoradiotherapy). N/A Evaluation of the impact 266 (133 paired sets): 204 Total: Obese ¼ 22 (17%), malignant (Obese- 75 of obesity on the Non-obese ¼ 19 men and 27 women; outcomes of (14%); Malignant: non-obese- 77 men and esophagectomy to Obese ¼ 19 (19%), 25 women), 62 benign determine whether Non-obese ¼ 13 (Both- 18 men, 13 profound obesity (13%); Benign: women). should contraindicate Obese ¼ 3 (10%), the transhiatal Non-obese ¼ 6 (19%). approach. 8 (8%) of normal or low Of the 215 patients who A total of 250 pts.; 97 Determine the role of underwent resection, 62 BMI pts. and 6(5%) of (45%) had low or body mass index (BMI) had neoadjuvant High BMI Pts. normal BMIs, 86 in a Western population chemo-therapy (37 (40%) were on outcomes after with high BMI, 25 with overweight, and a esophagectomy for normal or low outcome further 32 (15%) were cancer. after esophagectomy obese. related to BMI 31 BMI) and 49 had neoadjuvant chemoradiation (37 with high BMI, 29 with normal or low BMI). The remainder of patients had surgery without neoadjuvant treatment. 13 patients had an N/A Evaluate the diagnosis, Some 291 consecutive isolated leak. management and patients underwent two-phase subtotal outcome of mediastinal oesophagectomy with leaks following radical gastric interposition for oesophagectomy with a malignancy. stapled intrathoracic anastomosis. Examine the safety of 143 patients treated by Five patients (3.5%) had Preoperative neoadjuvant transthoracic a documented treatment was delivered TTE; 127 men and 16 esophagogastrectomy anastomotic leak; to 28 patients. women. (TTE) in a multidisciplinary cancer center and to determine which clinical parameters influenced survival and the rates of death and complications.
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment
857
TABLE I. (Continued) Author Beitler et al.
Year published 1998
Study period Literature review
# of patients & demography
Study type Comparison of stable and N/A hand-sewn esophagogastric anastomosis
symptoms, the diagnosis of the presence of a leak, the medical or surgical treatment, the evolution, and the day of closure were carefully recorded. The management of leaks at our institution is based on first adequate drainage/control/management of a leak–defined by a CT of the chest that does not demonstrate any un-drained fluid. Once control is confirmed, then our bias in patients >10 days out from anastomosis is to utilize esophageal stenting to exclude the leak, which we have have demonstrated can lead to quicker drainage tube removal and quicker time to discharged and possible PO intake [19,20]. In the leak patients, only once the leak was controlled with esophageal stenting (controlled defined as no drainage tubes of catheters) or healed was adjuvant therapy initiated. Patient characteristics were assessed by routine history and physical, as well as preoperative biopsy of suspected lesions, complete metabolic panel, and imaging. Preoperative imaging including CT scans of the chest, abdomen, and pelvis at minimum with PET scanning if CT demonstrated suspicious lymphadenopathy. Neo-adjuvant therapy was utilized in all stage 3 patients and in high-risk stage 2 patients. We have predominantly utilized three forms of therapy–1st weekly carbotaxol with XRT per the CROSS trial [21], 2nd neo-adjuvant EpirubicinCisplatin-Fluoruracil per the MAJIC trial [22], and 3rd Cisplatin– Fluoruracil with XRT based on the: (i) Tumor location–gastic versus gastro-esophageal versus esophageal; (ii) histology–adenocarcinoma versus squamous cell; (iii) Degree of lymph node involvement at diagnosis based on CT and CT-PET scanning. Factors potentially associated with esophageal anastomotic leak were identified by literature search as well as anecdotal evidence. Complications were graded on a scale previously presented [23] with grades 1–2 requiring no invasive intervention, 3–4 requiring either percutaneous drainage, reoperation, ICU readmission, or grade 5 representing death. These data were accrued prospectively over the course of each patient’s course. Initial postoperative follow-up occurred between 2–3 weeks with 3–6 month follow up for the next 2 years. At each follow up visit, patients completed the Karnofsky Performance Status evaluation as a proxy for quality of life [24] as well as the Global Rating Scale [25,26]. Post-hoc univariate analyses were performed using x2, independent T-Test, Fisher’s Exact Test, and Log Rank testing. Multivariate analysis was performed using ANOVA. Survival
% leaks N/A
% received adjuvant therapy N/A
comparisons were performed in the method of Kaplan and Maier. All statistical analyses were performed using PASW/SPSS (IBM, Somers, NY). An a of .05 was considered statistically significant.
RESULTS Within the cohort of patients undergoing surgery for gastric and esophageal cancers over a 10 year period, 194 patients received esophageal anastomoses as part of total gastrectomy (57, 29%) or IvorLewis esophagogastrectomy 137 (71%). In those patients, 35 (18.3%) developed esophageal leak based on clinical and radiographic diagnoses. In the EAL patients eight developed this after total gastrectomy with the remainder 27 after esophagogastrectomy. There were no early leaks, 12 were intermediate time interval, and 23 were late. The diagnosis was Type 1 in 28 patients with Type 2 in seven patients. The median severity grade of EAL was two with the majority of these (60%) requiring observation and continued drainage, 29% required intervention (either interventional radiographic treatment or thoracostomy), and 11% of leak patients having only incidental findings on gastrograffin swallow without clinical signs or symptoms. 90-day mortality was similar in the leak and no leak group (2.8% vs. 2.3%). The median age was similar in the EAL versus the non-leak group (64 vs. 61, P ¼ 0.9, Table II), as was the gender make-up (68.6% male vs. 80.4% male, P ¼ 0.3). The past medical history of both groups was similar with regards to: cardiac history (22.9% vs. 23.3%, P ¼ 0.96), pulmonary risk factors (14.3% vs. 10.1% P ¼ 0.5), and diabetes (20% vs. 14.0% P ¼ 0.4). The rate of substance abuse was also similar for both tobacco (62.9% vs. 58.1% P ¼ 0.615), median pack-years (43.3 vs. 48.0, P ¼ 0.7), and frequent ethanol use (20.9% vs. 14.3% P ¼ 0.4). A similar percentage experienced preoperative weight loss (45.7% vs. 53.5% P ¼ 0.4). There was no difference in median preoperative albumin level (CI 0.36 to 0.06). The use of neoadjuvant chemotherapy did not affect the development of esophageal leak (34.3% vs. 38.8% P ¼ 0.6). The histologic origin of the tumors was conserved between groups as well with similar proportions of patients with adenocarcinoma (77.1% vs. 78.3%, P ¼ 0.8). For patients with esophageal cancer, patients that
TABLE II. Preoperative Patient Characteristics
Age (median) Gender (%Male) Cardiac Dz Pulmonary Diabetes Alcohol Tobacco Mean pack years Squamous cell Adenocarcinoma Weight Loss Albumin level (median) Preop chemo Rx
Journal of Surgical Oncology
Leak N ¼ 35 (18%)
No leak N ¼ 159 (82%)
P value
64 68.6% 8 (23%) 5 (14%) 7 (20%) 5 (21%) 22 (63%) 43.3 7 (20%) 27 (77%) 16 (46%) 3.7 12 (34%)
61 80.4% 37 (23%) 15 (10%) 22 (14%) 25 (14%) 92 (58%) 48 31 (19%) 128 (81%) 84 (53%) 3.6 62 (39%)
0.914 0.254 0.960 0.480 0.378 0.3790 0.615 0.729 0.842 0.842 0.414 0.164 0.638
858
Martin et al.
experienced leak did not have a statistically significant different location of tumor from those without (P ¼ 0.3). Preoperative T stage (P ¼ 0.6) and N stage (P ¼ 0.4) were the same in the leak and no-leak groups. The presence of leak was not dependent upon the procedure performed, either esophagogastrectomy or total gastrectomy (P ¼ 0.6, Table III). There was no difference in the rate of leak based on operative technique. Neither the use of a stapler (48.5% vs. 43%, P ¼ 0.6), nor the type of hand-sewn anastomosis were risk factors for development of esophageal leak (P ¼ 0.1). Intraoperatively, there was no difference in mean estimated blood loss (SEM ¼ 60.5, P ¼ 0.7); however, a significantly larger proportion of those patients developing esophageal leak required red blood cell transfusion (57.1% vs. 34.1%, P ¼ 0.013). Of those patients that required transfusion, the amount was not increased in those patients developing a leak with both groups having a median transfusion requirement of 2 units (P ¼ 0.7, SEM ¼ 0. 31). All patients were admitted to the ICU postoperatively, with leak patients having a significantly longer median ICU Length of Stay (LOS, 10 vs. 4.5, P ¼ 0. 006, Table III). As expected patients with a leak had higher rates of concomitant postoperative complications including pneumonia (48.6% vs. 16.3%, P < 0.001), pleural effusion (14.3% vs. 4.3%, P ¼ 0.043), ileus (11.4% vs. 1.6%, P ¼ 0.006), empyema (11.4% vs. 0, P < 0.001), and central venous catheter related blood stream infections (8.6% vs. 0% P ¼ 0.001). Other postoperative complications showed neither statistical significance nor trend toward significance (Table III). Mortality up to 90 days post-op was similar with 0% mortality in patients with leak compared to 2.3% mortality (3/129). Oncologic factors did not predict esophageal leak with no difference found in our cohort between T stage (P ¼ 0.4), N stage (P ¼ 0.4), M stage (P ¼ 0.8), margin positivity (P ¼ 0.6) or number of positive lymph nodes (P ¼ 0.2). Of those factors associated with statistically significant adverse outcome in the leak group, only pneumonia, effusion, ileus, empyema, and central line infection remained statistically significant on ANOVA analysis. Those patients developing EAL were divided into two groups: lowgrade [1–2] EAL, treated only with medical therapy and observation, and high-grade (Grade 3) EAL, which required intervention. The majority of leaks were low grade (60%) with high grade (29%). There was a trend toward increasing grade of leak with esophagogastrectomy rather than total gastrectomy, but this was not statistically significant.
There was no difference as well in type of anastomosis (P ¼ 0.3), EBL, transfusion requirement (P ¼ 0.2), or ICU LOS. Many of the complications seen were in those patients with high-grade leaks, but the divisions of EAL patients into these groups lead to insufficient power for statistical significance (Table III). Though a greater percentage of low-grade EAL patients received adjuvant therapy, this was not statistically significant. The Karnofsky score of both low and high-grade leak patients was the same at 6 months, with each group showing a mean of 90 or greater. Patient who suffered a leak had a worse KPS at 6 weeks post op (median 70 for leak and 90 for no leak), however the median score for both a 3 months was 90%, and remained similar, with the only major decreases in KPS being related to adjuvant chemotherapy and or Chemo-radiation therapy. Similar differences were seen in contrast, when all of the patients completed their global quality of life scores by the QLQ-C30. There were statistical differences see in overall QOL at the 6 week post-operative visit with the most dramatic changes occurring in relation to questions: “How would you rate your overall health during the past week” and “How would you rate your overall quality of life during the past week,” both on a seven point scale, there was a significant loss of overall QOL from baseline (Fig. 1) (P ¼ 0.02 for 6 weeks post-operative) Patients with esophageal leak underwent adjuvant chemotherapy at rates comparable to those patients without leak (28.6% vs. 38.0%, P ¼ 0.3, Table IV). The type of adjuvant therapy was decide upon based on what form of neo-adjuvant therapy was utilized. If they received three cycles of Epirubin-Cisplatin-Fluorouracil (ECF), then they recived another three cycles of ECF, restage and if they had a >20% lymph node positive to lymph node negative ration then they received adjuvant 5FU and XRT to the nodal bed. If they had neoadjuvant Carbo-taxol with XRT and had a moderate to poor response based on pathology and had >20% LN ratio then they received four cycles of ECF based chemotherapy or FOLFOX based chemotherapy. If they had a good response, but with >20% lymph node ration then they received FOLFOX adjuvant therapy. The median time to initiate adjuvant therapy was 48 days (approximately 6 weeks) with a range of 35–90 days. In total 75% of all patients received their planned adjuvant therapy with no difference in successful completion of adjuvant therapy in the leak (72%) and the no leak group (79%). They also developed similar rates of dysphagia (5.7% vs. 8.5%, P ¼ 0.6), and stricture (8.6% vs. 7.8% P ¼ 0.9). Both groups of patients had a minority that required esophageal dilation for intractable
TABLE III. Operative Factors and Complications Compared for Affect on Leak
Esophagogastrectomy total gastrectomy Stapled hand-sewn EBL (median, mean) Blood transfusion # of Units (median) Length of stay–ICU (median,mean) Pneumonia Pleural Effusion New arrhythmia (afib/svt) Ileus beyond expected Fever Confusion Respiratory failure Empyema CVC related BSI Acute renal failure
Journal of Surgical Oncology
No leak N ¼ 159 (%)
Leak N ¼ 35 (%)
Grade 1–2 leaks N ¼ 24 (71%)
Grade 3 leaks N ¼ 10 (29%)
P value*
119 (75%) 40 (25%)
0% 4 (17%)
6 (60%) 4 (40%)
0.574
69 (43%) 90 (57%) 350, 488.3 54 (34%) 2
28 (80.0%) 7 (20%) 16 (46%) 19 (54%) 400, 543.3 20 (57%) 2
3 (12%) 1 (6%) 516 8 (33%) 2.5
4 (40%) 6 (60%) 498 6 (60%) 2.3
4.5, 10.3 21 (13%) 6 (4%) 17 (11%) 2 (1%) 9 (6%) 6 (4%) 10 (7%) 0 0 3 (2%)
10, 10.6 17 (49%) 5 (14%) 7 (20%) 4 (11%) 4 (11%) 2 (6%) 6 (17%) 4 (11%) 3 (9%) 2 (6%)
10.0 8 (33%) 0 (0%) 4 (17%) 8 (33%) 4 (17%) 4 (17%) 0 0 0 0
0.569 0.741 0.013, 0.195 .695 SEM ¼ 0.31 0.006 >0.001 0.043 0.311 0.006 0.336 0.796 0.097 >0.001 0.001 0.301
5 2 2 1 1 1 2 1 1 1
11.7 (50%) (20%) (20%) (10%) (10%) (10%) (20%) (10%) (10%) (10%)
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment
859
Fig. 1. Global quality of life score at baseline, 6 weeks, 3 months, 6 months, 9 months, and 12 months after esophagectomy for patients who developed a leak versus patients who did not develop leak. dysphagia postoperatively (5.7% vs. 1.6% P ¼ 0.2). The average Karnofsky Score was the same for both groups at 6 months (93.2 vs. 93.1 (P ¼ 1). Furthermore, the percentage of patients presenting with a Karnofsky Score of 80 or greater, representing that level at which no assistance is needed for daily activities, was conserved between patients with and without esophageal leak (89.5% vs. 92.9%, P ¼ 0.6). The presence of esophageal leak did not affect 12 month overall survival (93% vs. 94%, P ¼ 0.3).
DISCUSSION We present a review of our experience with 194 esophageal anastomoses performed for oncologic resection. In this patient cohort, representing 10 years of cases at our region’s major cancer center, there was no association between the development of esophageal leak and overall survival. The major caveat to this of course, remains that it is difficult to tell the true impact of esophageal leak on OS given the
TABLE IV. Follow Up Evaluation No leak N ¼ 159 (78.7) Stricture Dilation Dysphagia Adjuvant Rx Karnofsky Score (avg @ 6 mos.) K. Score > 80 þ (independent) AWD/NED @ 12 months *
14 (9%) 10 (6%) 10 (6%) 46 (29%) 93.2 143 (90%) 93%
EAL N ¼ 35 (21.3%) 3 2 11 12
(8%) (6%) (31%) (38%) 93.1 33 (92.9%) 94%
Grade 1–2 EAL N ¼ 24 (71%)
Grade 3 EAL N ¼ 10 (29%)
P value*
8 (33%) 4 (17%) 8 (33%) 12 (50%) 90.0 16 (66%) 95%
1 (10%) 1 (10%) 0 2 (20%) 93.2 5 (50%) 75%
0.874, 0.017 0.158 0.586, 0.001 0.303 0.984 0.627 0.273
Second listed P-value represents comparison between low-grade and high-grade EAL. If no second P-value is listed, it is not significant.
Journal of Surgical Oncology
860
Martin et al.
propensity of progression of most stage three esophageal and gastric cancers. In addition, the development of esophageal leak had no effect on the ability of patients in our cohort to receive adjuvant therapy. Anastomotic leak can be a devastating complication when not managed appropriately, one that depletes the patient physiologically and mentally. Indeed, leak led to many of the more dreaded complications seen in this cohort postoperatively. The anecdotal ideal of treating physicians would be to avoid chemotherapy and radiotherapy in a patient that has recently undergone such an insult. However, the similarity of survival in patients with and without leak as stratified by the receipt of adjuvant therapy contradicts this particular prejudice (Fig. 1). Much has been made about the diagnosis of esophageal leak: how best to diagnose it, its manifestations, the difference between a clinical relevant leak, and a non-relevant one. However, relatively little has been published as to the effects of esophageal leak on the patient in the long term. In these 164 patients, no difference was noted in average Karnofsky score at 6 months for all patients. There was no difference in proportion of patients that were functional on a daily basis at home between patients with and without leak at 6 month follow up as well. As such, esophageal leak acutely affects QOL, but does not appear to do so in the long term. The clinicopathologic factors analyzed here are consistent with those previously postulated to be potentially associated with esophageal leak. The only factor in this cohort that consistently predicted the development of esophageal leak was the requirement of RBC transfusion (P ¼ 0.013). No one operative technique was superior, nor were the AJCC staging criteria associated with leak development. Our rate of leak, 21% is commensurate with that of currently published studies [7], with other studies describing a lower rate. However, Whooley and colleagues [27] report a significant difference in hospital mortality among 475 esophagectomy patients that was related to leak (35% vs. 9%). This is corroborated by Ramirez et al in review of 276 patients [8]. While not specifically analyzed for hospital or 30-day mortality, our data likely differs from that of these studies by virtue of leak classification. With fewer, 3.5–5% of patients developing an anastomotic leak coupled with the much higher mortality for these patients, the sensitivity of leak diagnosis and inclusion in survival analysis appear to be higher in our patient population. Wright et al recently published their experience with 2,315 esophagectomies over multiple institutions [10]. Their analysis revealed a higher leak rate more commensurate with our own (11.3%). Additionally, they analyzed factors associated with anastomotic leak. However, they included all “major” morbidities associated with esophagectomy in their analysis. As such, the numerous significant factors they associate with the development of leak such as history of CHF, peripheral vascular disease, and African American race cannot truly be extrapolated solely to that group of patients developing anastomotic leak. Escofet and colleagues [6] published a series of 240 cases, 105 of which underwent neoadjuvant therapy. Their rate of leak was similar to ours, approximately 20%. However, in contrast to our analysis, they associated the use of neoadjuvant treatment with the development of leak. The other factors they identified, low Body Mass Index and surgeon experience, were not analyzed in our review. Karl et al [9] reviewed 143 esophagectomy patients with a very low leak rate (3.5%). The only factor identified with increased risk of leak as the presence of diabetes mellitus. AJCC stage, transfusion requirement, and presence of positive lymph nodes were not associated with leak development. This is consistent with our findings with the exception of the transfusion requirement. A recent meta-analysis of leaks was published by Schaheen et al, which confirmed our management and findings in that Intraoperative anastomotic drain placement was associated with earlier identification and resolution of anastomotic leak (mean 23.4 vs. 80.7 days). In addition, reinforcement of the anastomosis with omentoplasty may reduce the incidence of anastomotic leak by nearly 50%. Endo-scopic stent placement was associated with leak resolution in 72%. Journal of Surgical Oncology
CONCLUSIONS We have shown that the development of esophageal leak, when managed conservatively and well, does not affect overall survival at 6 months. In addition and novelly, esophageal leak does not preclude patients from undergoing postoperative adjuvant therapy. Finally, there is no difference between those patients with leak and without regarding performance status at 6 month follow up. Additional research is needed to determine the overall effect of esophageal leak on patients undergoing esophageal anastomoses.
REFERENCES 1. SEER, SEER Database, 2007, NCI. 2. Mine S, et al.: Colon interposition after esophagectomy with extended lymphadenectomy for esophageal cancer. Ann Thorac Surg 2009;88:1647–1653. 3. Stein HJ, et al.: Limited surgical resection and jejunal interposition for early adenocarcinoma of the distal esophagus. Semin Thorac Cardiovasc Surg 2007;19:72–78. 4. Visbal AL, et al.: Ivor Lewis esophagogastrectomy for esophageal cancer. Ann Thorac Surg 2001;71:1803–1808. 5. Kim RH, Takabe K. Methods of esophagogastric anastomoses following esophagectomy for cancer: A systematic review. J Surg Oncol 2010;101:527–533. 6. Escofet X, et al.: Prevalence and outcome of esophagogastric anastomotic leak after esophagectomy in a UK regional cancer network. Dis Esophagus 2010;23:112–116. 7. Atkins BZ, et al.: Reducing hospital morbidity and mortality following esophagectomy. Ann Thorac Surg 2004;78:1170–1176, discussion 1170-1176. 8. Junemann-Ramirez M, et al.: Anastomotic leakage postesophagogastrectomy for esophageal carcinoma: retrospective analysis of predictive factors, management and influence on longterm survival in a high volume centre. Eur J Cardiothorac Surg 2005;27:3–7. 9. Karl RC, et al.: Factors affecting morbidity, mortality, and survival in patients undergoing Ivor Lewis esophagogastrectomy. Ann Surg 2000;231:635–643. 10. Wright CD, et al.: Predictors of major morbidity and mortality after esophagectomy for esophageal cancer: A Society of Thoracic Surgeons General Thoracic Surgery Database risk adjustment model. J Thorac Cardiovasc Surg 2009;137:587–595,discussion 596. 11. Beitler AL, Urschel JD. Comparison of stapled and handsewn esophagogastric anastomoses. Am J Surg 1998;175:337– 340. 12. Okuyama M, et al.: Hand-sewn cervical anastomosis versus stapled intrathoracic anastomosis after esophagectomy for middle or lower thoracic esophageal cancer: A prospective randomized controlled study. Surg Today 2007;37:947–952. 13. Tirnaksiz MB, et al.: Effectiveness of screening aqueous contrast swallow in detecting clinically significant anastomotic leaks after esophagectomy. Eur Surg Res 2005;37:123–128. 14. Upponi S, et al.: Radiological detection of post-oesophagectomy anastomotic leak - a comparison between multidetector CT and fluoroscopy. Br J Radiol 2008;81:545–548. 15. Csendes A, Burgos Braghetto AM, Braghetto I, et al.: Classification and management of leaks after gastric bypass for patients with morbid obesity: A prospective study of 60 patients. Obes Surg 2012;22:855–862. 16. Csendes A, et al.: Management of leaks after laparoscopic sleeve gastrectomy in patients with obesity. J Gastrointest Surg 2010;14:1343–1348. 17. Csendes A. Conservative management of anastomotic leaks. Obes Surg 2006;16:375–376,author reply 376. 18. Csendes A, et al.: Conservative management of anastomotic leaks after 557 open gastric bypasses. Obes Surg 2005;15:1252–1256. 19. Martin R, et al.: The use of self-expanding silicone stents in esophageal cancer care: optimal pre-, peri-, and postoperative care. Surg Endosc 2009;23:615–621.
Esophageal Anastomotic Leak Does Not Affect Ability to Receive Adjuvant Treatment 20. Martin RC, et al.: The use of self-expanding silicone stents in esophagectomy strictures: Less cost and more efficiency. Ann Thorac Surg 2008;86:436–440. 21. van Hagen P, et al.: Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012;366: 2074–2084. 22. Cunningham D, et al.: Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 2006;355:11–20. 23. D’Angelica M, et al.: Major hepatectomy with simultaneous pancreatectomy for advanced hepatobiliary cancer. J Am Coll Surg 2004;198:570–576.
Journal of Surgical Oncology
861
24. O’Toole DM, Golden AM. Evaluating cancer patients for rehabilitation potential. West J Med 1991;155:384–387. 25. Brown RE, et al.: A prospective phase II evaluation of esophageal stenting for neoadjuvant therapy for esophageal cancer: Optimal performance and surgical safety. J Am Coll Surg 2011;212:582– 588,discussion 588-589. 26. Mbah N, et al.: Impact of post-operative complications on quality of life after pancreatectomy. JOP 2012;13:387–393. 27. Whooley BP, et al.: Critical appraisal of the significance of intrathoracic anastomotic leakage after esophagectomy for cancer. Am J Surg 2001;181:198–203.
