Multidisciplinary, minimally invasive approach for oesophageal perforations with delayed presentation.

Background: The present study aims to report the outcomes of a multidisciplinary, minimally invasive approach to treating patients with delayed presentation of oesophageal perforation. Patients and
Methods: The present study is a retrospective analysis of prospectively maintained data at a tertiary care centre. All patients with oesophageal perforation presenting over 48 h after the onset of symptoms and without oesophageal obstruction were included in the study. Self-expanding Metallic Stent (SEMS) or endoscopic clip placement was performed in all the patients, followed by video-assisted thoracoscopic surgery (VATS) debridement and decortication of pleural cavity collection. 'Success' was defined as, discharge without the need of oesophageal diversion and complete healing of leak site at 8 weeks with successful removal of the stent.
Results: Between March 2012 and December 2019, 12 patients (10 males, median age of 55 years- range of 39-71 years) with oesophageal perforation and delayed presentation underwent treatment with this approach. Ten patients had spontaneous perforation (83.3%) and one patient each had upper gastrointestinal endoscopy-induced and post-traumatic perforation. The median duration of symptoms was 8 days (range 3-31 days). SEMS was placed in ten patients and, in two patients, an over-the-scope clip was used. VATS decortication was done in ten patients (83.3%) and the remaining two (16.7%) underwent VATS debridement. One patient required oesophageal diversion and another patient expired due to sepsis. The overall success with this approach was 83.3%.
Conclusion: This multidisciplinary, minimally invasive approach is feasible in patients with thoracic oesophageal perforation and delayed presentation, with a high success rate.

INTRODUCTION

Oesophageal perforation is an emergency, associated with high morbidity and mortality. Boerhaave's syndrome, iatrogenic injury and foreign body-induced trauma are the major causes,[1] and thorax is the most common site of oesophageal perforation.[234] Leakage of oesophageal and gastric contents into the mediastinum and pleural cavities initiates intense inflammatory reaction that results in severe sepsis, multiorgan dysfunction syndrome and often death.[56] Even with aggressive treatment, reported mortality in patients with oesophageal perforations could be as high as 20%.[789] As it is an uncommon condition with no specific symptoms, the diagnosis is often delayed, resulting in continuing mediastinal and pleural sepsis, further increasing the mortality.[10]

Primary surgical repair, or diversion procedures, and even oesophagectomy performed by thoracotomy have been the conventional modalities of management. However, the trauma of a major surgical procedure in these sick patients with sepsis adds to the insult, further jeopardising the outcome. With technical advancements, there has been a paradigm shift in the approach to oesophageal perforations from open surgery to a minimally invasive approach, wherein endoscopic interventions are used to seal the leak in conjunction with intercostal tube drainage for pleural collections.[1112] In most of these studies, however, patients presented ‘early’ after the symptom onset.

In patients with ‘delayed’ presentation, the margins of the perforation are unhealthy and inflamed, thus more difficult to heal. They also have a quantitatively larger pleural collection that is multiloculated, organised and often contains food materials. Thus, a simple intercostal tube insertion without aggressive debridement does not achieve complete clearance of septic material in the chest. Moreover, prolonged pleural collection often leads to collapse of the lung with thick peel formation (trapped lung), which leads to non-expansion of lung even after intercostal tube drainage, causing persistence of potential space and continuation of sepsis, leading to higher mortality.[13] Therefore, removal of peel from the lung surface (decortication) is essential in such a situation to achieve lung expansion, eliminate the potential space and completely clear the sepsis.

Based on this knowledge, we devised a multidisciplinary, minimally invasive approach to treat patients with delayed presentation of oesophageal perforation. The key components of this approach are closure of oesophageal leak with self-expandable metal stent (SEMS) or clips, placement of naso-jejunal (NJ) tube for feeding and complete clearance of pleural collections (debridement/decortication) by video-assisted thoracoscopic surgery (VATS). This study presents the results of our initial experience with this multidisciplinary, aggressive but minimally invasive approach.

PATIENTS AND METHODS

This is a retrospective analysis of a prospectively maintained database from March 2012 to December 2019 at a tertiary care centre in New Delhi. Data of all the patients with oesophageal perforation managed at our institute including demographic, clinical, imaging, endoscopic and surgical details were collected. This study was approved by the institutional ethics committee.

Inclusion and exclusion criteria

All patients with thoracic oesophageal perforation presenting more than 48 h (delayed presentation) after the symptom onset were included in the study. Patients with cervical oesophageal perforation and thoracic perforations with early (<48 h) presentation were excluded from the study. Patients with oesophageal obstruction due to various causes such as corrosive stricture, malignancy and achalasia cardia were also excluded from the study.

Management protocol

After haemodynamic stabilisation, the patients were started on broad-spectrum antibiotics and proton pump inhibitors. Diagnosis of oesophageal perforation was confirmed by review of imaging and endoscopic findings if available. All patients underwent computed tomography (CT) chest and upper abdomen with oral contrast, to document the site, size and type (free flowing or contained) of oesophageal leak. The pleural cavities and mediastinum were carefully evaluated for quantity of pleural effusion, any loculation inside the pleural cavity and status of the underlying lungs.

Multidisciplinary, minimally invasive step-up approach

We followed a ‘multidisciplinary, minimally invasive approach’ for thoracic oesophageal perforations with delayed presentation. This approach aimed to control further leakage from the oesophagus, and eliminate septic collections from the pleural cavity and the mediastinum by ensuring complete expansion of lungs. This was achieved by a combination of endoscopy and minimally invasive thoracic surgery.

Endoscopic therapy (performed immediately after presentation)

An upper gastrointestinal endoscopy (UGIE) was performed in all cases, and the leakage site was assessed. All the endoscopies were performed in the endoscopy suite under moderate sedation with midazolam and pentazocine. The choice of endoscopic accessory to close the leak was left to the discretion of the endoscopist. In general, the following algorithm was followed. Oesophageal leaks <2 cm in size and healthy margins were closed with over-the-scope clip application (11 mm, 12/6t, OTSC system, Ovesco Endoscopy, Tubingen, Germany). For larger lesions and with unhealthy margins, a fully covered removable self-expandable metal stent (18-mm diameter, 23- and 25-mm flare, Wallflex, Boston Scientific, MA, USA) was placed, and stent length was chosen so as to extend more than 3 cm on both the sides of leak. Various techniques were used to prevent the stent migration such as (1) mucosal flap technique, where a stent was fixed to the adjacent oesophageal mucosa with a clip [Figure 1][14] and (2) loop-and-clip technique where an endoloop was fixed to the stent and oesophageal wall with the help of clips [Figure 2] and 3. Thread fixation technique, where upper end of stent was attached to a thread and it was brought out from the nares and plastered to the ear. A protective soft catheter was passed over the thread to avoid the trauma. An NJ tube was placed simultaneously in all the patients after the endoscopy. No patient underwent feeding gastrostomy or jejunostomy. NJ tube helps in the drainage of gastric contents and allows early jejunal feeding. An oral contrast study with Gastrografin was performed once the patient was stable and was willing to accept orally, usually after 5–7 days of stent placement. Oral intake was resumed initially with a liquid diet if the contrast study demonstrated no leak. The stents were kept in situ for at least 8 weeks.

Figure 1

Mucosal Flap Technique of Stent Fixation (a) Raising the mucosal flap, (b) Fixing stent with the flap

Figure 2

Loop-and-clip technique of stent fixation, (a) An endoloop was fixed to the stent and oesophageal wall with the help of clips, (b) Endo-Loop and clips in-situ

Surgical therapy: Thoracoscopic debridement/decortication (performed within 24 h after endoscopic therapy)

After the stent placement, all the patients underwent VATS with the drainage of collections and debridement (removal of pus and debris). Additional decortication (removal of peel from the visceral surface of lung) was performed if required. CT imaging was reviewed in detail before surgery to have a roadmap for surgical treatment. All the procedures were performed under general anaesthesia. Initial diagnostic thoracoscopy was performed through a 5-mm telescope. Once the cavity was entered, all the loculi were broken and the infected fluid along with food particles was suctioned, and the sample was sent for culture and sensitivity. Thorough debridement of the infected tissue was done. After completion of debridement, the status of the underlying lung was assessed. If any thickened visceral peel was found on the surface of the lung, restricting its complete expansion, removal of this peel (visceral pleurectomy, i.e. decortication) was performed to achieve complete lung expansion. Another critical area was to clear the area between the under surface of the lung and the diaphragm, where collections often persist. Complete removal of parietal pleura (parietal pleurectomy) was also performed in all cases as inflamed and infected parietal pleura is often a source of continuing sepsis and excessive pleural drainage. The entire procedure was done thoracoscopically. Meticulous haemostasis was achieved. Two chest drains were placed (one at the lung base and another at the lung apex) and connected to a digital negative suction device (Thopaz; Medela AG, Baar, Switzerland) to assist in complete lung expansion.

In patients with continuing mediastinal sepsis, a Gastrografin study was done on the 5th post-operative day. If a persistent oesophageal leak was found despite a correctly positioned stent, diversion (cervical-end oesophagostomy, and stapled occlusion of gastro-oesophageal junction) and feeding jejunostomy were performed as a rescue surgery in addition to thoracoscopic debridement.

Post-operative care

Patients were extubated at the end of the surgery whenever possible. Those who required post-operative ventilation were shifted to the intensive care unit. Supervised chest physiotherapy was continued after the surgery to maintain good lung expansion. The antibiotic choice was guided by the culture sensitivity report. Special attention was paid to ensure high-calorie, high-protein liquid diet through NJ tube. In the initial period, blood and albumin transfusions were given liberally to maintain haemoglobin above 10 g % and serum albumin more than 3 g %. Effective pain relief was achieved by epidural analgesia supplemented by intravenous medications. Drains were removed once the drain fluid was non-purulent, <100 ml in 24 h and without any evidence of air leak in the drain. Patients were discharged once there was no evidence of sepsis and they were stable on oral or tube feed and medications.

SEMS was removed at 8 weeks after the placement; a contrast study was performed at the same time to demonstrate the healing of the leak site.

The primary end point was success of the minimally invasive approach, which was defined as: (1) discharge of the patient in a healthy state without the need of diversion procedure or oesophageal repair and (2) complete healing at leak site, demonstrated by successful removal of stent at 8 weeks without any leak on contrast study.

Statistical analysis

Continuous variables were presented as mean ± standard deviation or median interquartile range. Categorical variables were expressed as frequencies and percentages. IBM SPSS Statistics for Windows, version 23 (IBM Corp., Armonk, New York, USA) was used for statistical analysis.

RESULTS

Demographic characteristics

During the study period, 17 patients underwent treatment for oesophageal perforation. Two patients with cervical perforation and three patients with thoracic perforation with early presentation were excluded from the study [Figure 3]. Twelve patients (10 males, 83.3%) with thoracic perforation who presented after 48 h of symptom onset, were included in the final analysis. The demographic and baseline details are mentioned in Table 1. The mean age of the patients was 53.8 years (range, 19–72 years). Ten patients had spontaneous perforation (83.3%). One patient each had UGIE-induced and post-traumatic perforation. The median time interval between symptom onset and presentation to our institute was 8 days (range 3–31 days). Intercostal drain tubes were already in situ in all the patients.

Figure 3

Flowchart of management and results of oesophageal perforation in the study population

Table 1

Demographic details and disease characteristics in the study population (n=12)

Characteristics	Frequency
Male (%)	10 (83.3)
Female (%)	2 (16.7)
Median age in years (range)	55 (39-71)
Median duration of symptoms in days (range)	8 (3-31)
Comorbidities (%)
Hypertension	3 (25)
Diabetes mellitus	3 (25)
Aetiology of perforation (%)
Spontaneous (Boerhaave’s syndrome)	10 (83.3)
Iatrogenic (post-upper GI endoscopy)	1 (8.3)
Post-traumatic	1 (8.3)

GI: Gastrointestinal

Endoscopy, operative and peri-procedural details

All procedural details and outcomes were mentioned in Table 2. Endoscopic intervention was performed in all the patients. The site of perforation could be localised in all the patients. SEMS was placed in ten patients and, in two patients, an over-the-scope clip was used. SEMS was fixed with mucosal flap technique in three patients, loop-and-clip technique in three and by nasal thread technique in two patients. In two patients, the stent was not fixed. The median operative time was 136 min (range, 58–268 min), and the median blood loss was 300 ml (range, 200–500 ml). All the patients underwent VATS-guided drainage and debridement, and ten (83.3%) patients required simultaneous VATS decortication. None of the patients required conversion of VATS to open surgery. Ten patients were extubated in the operation theatre, and two required ventilator for 48 h.

Table 2

Endoscopic and peri-operative details of the study cohort (n=12)

Characteristics	Frequency
Type of endoscopic intervention (%)
‘Over-the-scope’ clip application	2 (16.7)
Self-expanding metallic stent placement	10 (83.3)
Stent migration (needed repositioning)	2 (16.7)
Type of surgical intervention (%)
Thoracoscopic debridement	2 (16.7)
Thoracoscopic decortication	10 (83.3)
Median operative time (range) in min	136 (58-268)
Median blood loss (range) in ml	300 (200-500)
Median post-operative ICD duration (days) (range)	10 (5-24)
Median hospital stay (days) (range)	9 (6-18)
Post-operative complications (%)
Pleural recollection	1 (8.3)
Atrial fibrillation	1 (8.3)
Left lower lobe pneumonia	1 (8.3)
Death	1 (8.3)
Failure of approach (%)
Need to perform oesophageal diversion	1 (8.3)
Death	1 (8.3)

ICD: Intercostal Drain

Post-operative complications included left lower lobe pneumonia (n = 1, 8.3%), pleural re-collection (n = 1, 8.3%) and atrial fibrillation (n = 1, 8.3%). Pleural re-collection was drained by ultrasound-guided pig tail insertion, and atrial fibrillation was managed with IV amiodarone therapy. Two (16.7%) patients had migration of SEMS during the hospital stay, for which endoscopy and repositioning of the stent were done. In both these initial patients, fixation technique was not used. There was one (8.3%) death due to sepsis and multiorgan dysfunction. This 68-year-old patient with multiple comorbidities was misdiagnosed outside as hydropneumothorax and was so managed for 1 month, with continued oral intake before being referred to us. At presentation, he was in septic shock, on vasopressors, and succumbed 24 h after the procedure. Another patient (8.3%) with multiple oesophageal perforations as a result of iatrogenic injury during failed attempts at endoscopic removal of an impacted denture required cervical oesophagostomy (diversion procedure) because of continuation of oesophageal leak and persistence of sepsis despite SEMS placement. He has recovered completely and awaits definitive bypass surgery. SEMS could be removed in all the ten patients at 8 weeks with complete healing of leak site.

Thus, 11 out of 12 patients (91.7%) presenting late after oesophageal perforation could be discharged from the hospital, and 10 (83.3%) had a successful outcome by the proposed minimally invasive approach. Therefore, overall survival with this approach is 91.7% at a median follow-up of 41 months (range 9–73 months).

DISCUSSION

Oesophageal perforation is a challenging clinical situation with 5%–40% mortality.[1516] Delay in the diagnosis and management increases the mortality further. Brinster et al.[10] reviewed the outcome of oesophageal perforation in 390 patients and found significant relationship between the treatment delay and mortality. Mortality increased from 14% to 28% if treatment was delayed by 24 h. Similarly, in the study of 52 patients by White and Morris,[17] mortality increased from 13% to 31%, and Wright et al.[18] reported increased mortality from 0% to 31% if treatment was delayed for more than 24 h after oesophageal perforation. Delay in the diagnosis is likely due to lack of awareness and non-specific symptoms.

Two important aspects of management of oesophageal perforation are stopping the leak from the oesophagus and complete clearance of infected collections from the pleural cavity and mediastinum. In the earlier reported series, both the goals were achieved by open surgery. However, open surgical approach could have a negative effect on patient recovery in view of surgical trauma. This factor along with the technical advancements has prompted a shift towards non-operative approach in selected patients. The non-operative approach consists of nil by mouth, broad-spectrum antibiotics, parenteral nutrition, aspiration of gastric contents and drainage of collection with the tubes in the pleural cavity. In this approach, the oesophageal leak site was neither repaired nor bridged. Drainage of collections without surgical stress was an essential part of algorithm, which was thought to be the reason behind the reduced mortality with the non-operative approach, however with longer hospital stay.[19]

Meanwhile, there was significant advancement in the field of endoscopy, with the development of self-expandable metal stents to bridge the leak site. With this approach, hospital stay was further reduced. Freeman et al. reported a high success rate with hybrid approach for iatrogenic and spontaneous perforation. In a study of 17 patients with early iatrogenic perforation,[20] the patients were treated with simultaneous placement of oesophageal plastic stent, drainage of collection and feeding gastrostomy. No patient died, and the mean duration of hospital stay was 8 ± 9 days. Decortication was performed in only one patient. The mean time from perforation to stent placement was 39 ± 39 h. In contrast, in the present study, due to delayed presentation, 83.3% of the patients required thoracoscopic decortication because of multiloculated fluid collection with thickened visceral as well as parietal pleura. The same authors published another study in patients with spontaneous perforation.[12] Nineteen patients with early spontaneous oesophageal perforations were managed with similar hybrid approach, wherein the site of leak was healed in 17 patients (89%) without surgery. The mean time from symptom onset to intervention was 22 ± 33 h with a hospital length of stay of 9 ± 12 days. Five (26%) patients required decortication, and all the patients survived. These two studies emphasised excellent outcome with hybrid approach in early oesophageal perforations. Ben-David et al.[21] published data of 76 consecutive patients of oesophageal perforation treated with SEMS placement with or without drainage procedure. All the patients were treated within 24 h of initial presentation and only one patient died. These studies suggest that combined approach of stent placement and drainage procedure is highly effective with reduced mortality and hospital stay, provided patients present early after the symptom onset. At this early stage, the pleural contamination is minimal, collections are small and not yet loculated, and there is no question of lung being trapped. In this situation, tube drainage is enough to contain pleural sepsis.

However, the outcome of this approach in patients with delayed presentation has not been studied. Management of patients with delayed presentation (>48 h) is challenging in view of more severe mediastinal and pleural contamination (often gross food particles lying in the pleural cavity) and multiloculated pleural collections which are not amenable to simple drainage procedures. If the diagnosis is too delayed, due to continuous infection and inflammation, the visceral pleura gets thickened and forms a peel over the lung surface, which limits lung expansion. Moreover, oesophageal wall at the site of leak may be infected and necrosed, and hence the effect of stenting may be different on the final outcome of leak closure. With the existing data of poor outcome in patients with delayed presentation, we planned to evaluate the outcome of a multidisciplinary, minimally invasive approach in our patients with delayed presentation.

Our multimodality, minimally invasive approach incorporates aggressive debridement/decortication in addition to the already-established hybrid approach. This is because we realised that leaving even a small focus of infection in the pleural cavity and mediastinum in this group of sick patients can be catastrophic. In this approach, SEMS was left for a longer duration of 8 weeks to ensure the complete healing of leak site. Chest tube which was placed previously was not effective in our patients, because of multiloculated pleural collection with continuation of oral feed and oesophageal leak. Therefore, these patients had incomplete drainage and partially expanded lungs with thickened visceral peel, causing persistence of sepsis.

The side of approach was decided based on whether patient was having unilateral or bilateral collections. If the patient was having only right-sided multiloculated collection/empyema, right-sided thoracoscopic debridement/decortication was performed and vice versa. However, if the patient was having bilateral multiloculated collections, the worse side was operated first followed by the better side, either in the same sitting or different sittings based on the clinical scenario. In this series, 4/12 (33.3%) patients underwent thoracoscopic surgery on the right side alone, 5/12 (41.7%) patients had on the left side alone and 3/12 (25%) patients had surgical intervention on both sides.

In all cases, we performed the surgery by a three-port approach. Multiloculated collections were drained completely by thorough debridement and if there was peel over the lung (trapped lung), removal of the visceral peel (decortication) was also done. It is mandatory to free the lung from the diaphragm also, except in the area of perforation. Complete parietal pleurectomy was also performed in all cases as it is often a source of continuing pleural sepsis. At the end of the procedure, complete lung expansion was achieved with clearance of all necrotic and infected tissues from the pleural cavity. There was no need for conversion to open surgery in our series.

By our approach of endoscopic stenting and aggressive VATS decortication, we could achieve excellent outcome in these difficult group of patients. We had 8.3% of mortality in the study group, and this patient had already developed multiorgan dysfunction at presentation. This signifies the importance of early detection and aggressive treatment before the organ failure sets in. Oesophageal leaks also healed successfully in 91.7% of patients, except in 8.3% who required diversion procedure. In oesophageal leak, as there is no structure to hold back the stent, migration of stent remains an important issue, which needs to be addressed.[222324]

Endoluminal vacuum therapy is a new emerging therapy for the management of oesophageal perforations. This method is reported to be more appropriate in large perforations which are not amenable for endoscopic clipping or not adequately covered with self-expanding metallic stents (SEMS). In this therapy, the wound VAC sponge is cut to match the size of the perforation, which is connected to a Ryle's tube. This procedure requires multiple sittings and facilitates healing of the perforation over a 2–3-week period.

Limitations and strengths

We accept certain limitations of our study. First being retrospective in nature, the study could miss important information. The second limitation is the small sample size; hence, we cannot have definitive conclusions with high degree of confidence. Another limitation is lack of comparative arm; hence, we could not compare the outcome of our algorithm with that of the conventional treatment.

The strength of our study is the assessment of this modified hybrid approach in sick patients with delayed presentation, where currently limited data are available. The study provides rationale for utilising hybrid approach, and also proposes an algorithm [Figure 4] for the management of these comparatively sick patients. Although a similar study with 25 patients was reported recently from India, 17 patients in this group were anastomotic leaks.[25] Therefore, we propose a multicentric prospective study to acquire data with larger sample size from multiple institutes to find out the efficacy of this approach.

Figure 4

’Institute of Chest Surgery, Medanta’ protocol for management of thoracic oesophageal perforation with late presentation

CONCLUSION

In patients with thoracic oesophageal perforation, with delayed presentation, a ‘multidisciplinary, minimally invasive approach’ (endoscopic stenting and thoracoscopic debridement/decortication) has the potential to improve success rates. Multicentric studies with larger number of patients are required to further evaluate this approach.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.