Transthoracic esophagectomy is generally accepted as standard surgical treatment for esophageal cancer but it is also recognized as a complex surgical procedure associated with a high rate of postoperative complications of up to 60% even in specialized centres [1, 12]. As proven in several recent investigations, anastomotic leakage remains the predominant surgical complication with a reported rate of up to 16% for intrathoracic and 32% for cervical reconstruction [9, 18]. Several attempts have been made to establish a commonly accepted definition and classification of AL in order to grade the severity of this complication and to make it comparable across specialized centres [10]. Two classification systems suggested by Lerut et al. and Price et al. were based on the severity of clinical symptoms associated with AL but both grading systems failed to achieve wide acceptance within the esophageal community [19, 20].
Recently, a novel grading system has been proposed by the ,Esophagectomy Complication Consensus Group’ (ECCG) defining anastomotic leakage as a ,full thickness gastrointestinal defect involving esophagus, anastomosis, stapler line, or conduit irrespective of presentation or method of identification’ [11]. In contrast to previous classification systems, the novel ECCG classification is entirely based on the applied management of anastomotic leakage and differentiates between three distinct types as described above. The main advantage of this proposed grading system appears to be its striking simplicity and general applicability.
So far, several international registry analyses have applied the ECCG definition and classification system of AL to compare surgical outcome between single centres [9, 12]. In addition, in an international survey on the general management of AL the ECCG classification was used to assess the different treatment options [21]. To the knowledge of the authors, this is the first study analysing the proposed ECCG classification with respect to its clinical applicability. The main strength of the presented data is that it is based on large, homogenous cohort of patients undergoing a standardized hybrid IL esophagectomy with a well-established anastomotic technique currently practiced in many expert centres [9, 18, 22–24]. The overall leakage rate for this series is comparably low but reflects the outmost standardization of a long-time practiced surgical procedure in a well-recognized high-volume centre [3, 13–15] However, it is possible that the incidence of AL is underestimated if no examination (e.g. esophagography, upper endoscopy) is routinely performed during the postoperative course. This is particularly true for minor leaks which become clinically not apparent and could increase the number of ECCG type I leakages. In order to exclude the influence of learning curves as usually associated with the total minimally invasive including the robotic procedures we focused only on the hybrid IL esophgectomy. In addition, complete open IL procedures were also excluded since the impact of laparotomy on the development on AL is still matter of discussion. However, it has to be mentioned that for the patient cohort with hybrid IL esophagectomy the incidence of previous major abdominal surgery is not documented.
Several results of this retrospective analysis need to be discussed with respect to the ECCG definition and classification of AL.
As described above, ECCG defined AL as a full thickness defect irrespective of the clinical presentation. In this series, two patients underwent re-thoracotomy in the very early postoperative course due to suspicious fluid collection in the right thoracic drainage. Though all available intraoperative tests including an intraoperative endoscopy were performed, no (full thickness) defect could be detected and surgery was terminated as diagnostic procedure. Despite the lack of a clear full thickness defect but due to the obvious clinical symptom of bile leakage strongly indicating an AL, patients were classified and counted as AL type III. This suggests that the present ECCG definition may not cover all clinical scenarios of AL.
It also needs to be emphasized that the differentiation into the three AL types is strongly procedure related. For Ivor-Lewis esophagectomy, the data clearly show that a conservative management of AL (classified as ECCG type I) is currently not practiced but the vast majority of AL is successfully managed using some kind of endoscopic intervention (ECCG type II), predominantly the application of a self-expanding stent or an EVT. This obvious strategy reflects the fact that any intrathoracic AL is associated with some degree of mediastinitis including the potential burden of a consecutive sepsis with multiple organ failure. The AL management for the McKeown esophagectomy is completely different possibly resulting in a different distribution of the three AL types.
The publishing ECCG consensus group argued that the increasing invasiveness of AL management (conservative, interventional and surgical) clearly expresses the severity of the postoperative course associated with AL. In this patient series, surgical AL treatment was indeed associated with a significant higher rate of single or multiple organ failure as compared to AL managed by endoscopic means. Almost one out of four surgically managed patients developed a single or multiple organ failure (CD IVA/B) along the postoperative course and two patients died. Accordingly, the rate of ICU readmission and days of ICU treatment were significantly higher for AL type III predominantly due to an increased rate of respiratory failure. This data seems to be in favour of an endoscopic management when diagnosing a postoperative AL with a high success rate for this less invasive interventional management [16, 25, 26]. This might be due just to the second thoracotomy for AL type III which is a major surgical procedure and for itself contributes to morbidity and possibly mortality. However, it is notable that AL type III was diagnosed on average four days earlier compared to AL type II. Therefore, it is possible that AL type II and III reflect different underlying causes of anastomotic break down requiring different treatment strategies. In this patient cohort, AL type III has to be considered as complication of the very early postoperative course due to a simple technical failure or an impaired gastric perfusion as the main etiologic factors [27]. In this context, there is mounting evidence that patients with severe calcifications of the thoracic aorta, peripheral arterial disease or even higher grade stenosis of the celiac trunc are at an increased risk to developing an anastomotic leakage [Tzorzakakis, Cancers 2022, 14:1016; Brinkmann S, Schroeder W. Dis Esoph 2019, DOI: 10.1093/dote/doy107]. In contrast, AL type II usually becomes evident later in the postoperative course when oral feeding has already started or even completed depending on the recovery program of individual centres. The etiology of leakage in the later course is different and primarily driven by preoperative comorbidities or other accompanying infectious complications [27, 28]. These different causes of AL diagnosed in the very early and later postoperative course are the main trigger for different treatment strategies. Therefore, it is not a question of whether surgical or endoscopic management is to prefer for the AL management but both treatment options are necessary tools to successfully treat AL. It is also evident that this practiced treatment algorithm reflects the institutional preference and surgeons experience how to manage AL, but might be different in other esophageal expert centres. For this reason, the proposed ECCG classification of AL with three distinct types seem to discriminate sufficiently the severity of AL in individual departments but does not safely allow a comparison of leakages between different centres.
It also has to be mentioned that a sufficient discrimination depends on the selection of outcome variables as CD scores significantly differ but length of hospital stay and mortality were comparable for AL type II and III for the cohort under investigation. A strict differentiation between different AL types does not include all practiced treatment strategies since a clinically relevant patient cohort underwent an interventional and surgical management in changing chronological sequence. According to the ECCG criteria, all patients with a combined management have to be classified as AL type III that is considered as most invasive procedure. In addition, AL type II and III summarizes a broad variety of endoscopic and surgical techniques applied in order to restore continuity of the gastrointestinal tract.