Perforations of the upper gastrointestinal tract
Enders K.W. Ng
Introduction
Foregut perforations are challenging surgical emergencies and the keys to success in managing these potentially life-threatening situations are timely diagnosis and appropriate intervention. This chapter summarises the latest evidence and clinical experience concerning the investigation and therapeutic options for perforation in different parts of the upper gastrointestinal tract. Owing to the potential implications for prognosis and management planning, perforations are conventionally classified according to the anatomical position and causative mechanism.
Peptic ulcer perforation
The estimated lifetime risk of perforation in peptic ulcer disease ranges from 2% to 10%.1–3 Typical presentation includes a sudden onset of epigastric pain, which rapidly generalises to other parts of the abdomen. Some patients may report a history of dyspepsia or peptic ulcer, but such findings are absent in about one-third of the patients.4 ‘Board-like rigidity’ is the most commonly mentioned physical sign, referring to the diffuse peritonitis revealed on abdominal examination. However, in the elderly and the critically ill who develop ulcer perforation during hospitalisation, the presentation can often be atypical and subtle. Diagnosis under such circumstances requires a high index of suspicion. Though subdiaphragmatic free gas on a plain erect chest X-ray (Fig. 6.1) is diagnostic, it is seen in no more than 70% of patients.5 A lateral decubitus X-ray, water-soluble contrast meal, ultrasound scan and computed tomography (CT) of the abdomen are thus invaluable additional investigations when the diagnosis is in doubt (see also Chapter 5). However, it is noteworthy that the role of CT may be limited if the onset of symptoms is less than 6 hours.6 Pneumogastrogram was once advocated for being able to increase the diagnostic yield from 66% on plain erect X-ray to 91%.7 However, such a practice is now abandoned because forceful intragastric air insufflation is likely to reopen some of the spontaneously sealed-off perforations. For patients with profound peritoneal signs and a serum amylase level not diagnostic of acute pancreatitis, a laparoscopy or laparotomy after resuscitation is indeed the most time-saving and efficient investigation. It has the advantage of being able to execute immediate therapeutic intervention, either laparoscopically or via an open laparotomy, once the pathology is verified.
FIGURE 6.1 Erect chest radiograph showing right subphrenic free gas shadow associated with a perforated peptic ulcer.
Prognosis
Perforation accounts for the majority of deaths related to peptic ulcer disease.8,9 Spillage of gastroduodenal contents through the perforation initiates a chemical peritonitis, which will rapidly be superimposed by bacterial infection if left untreated. This leads in turn to the systemic inflammatory response syndrome and multi-organ dysfunction associated with the bacterial translocation across the peritoneal surface. Despite surgical intervention, mortality rates remain around 5–15%.9–11 The prognostic indicators for ulcer perforation have been widely studied, yet most published series consist of relatively small numbers of patients with marked heterogeneity in both demographics and treatment methods.
Before the era of laparoscopic surgery, significantly higher rates of mortality were reported by Boey et al. in patients with an ulcer perforation who were admitted with major medical illness, preoperative shock and long-standing perforation (more than 24 hours). Those with no, one, two and all three risk factors at presentation were noted to have mortality rates of 0%, 10%, 45.5% and 100%, respectively.12 This study underscores the importance of stratifying patients with ulcer perforation according to their risk of mortality, which is of relevance to the prognosis and choice of surgical intervention.
Similar findings have been reported by other multivariate analyses. Some also identified that age > 65 years and perforation during hospital stay are additional predictive factors of death.13,14 In 2001, Lee et al. compared the APACHE II score with the Boey parameters in a cohort of over 400 patients.15 An APACHE II score > 5 was found to have predictive value for increasing postoperative complications and death, whereas the Boey score only predicted mortality but not morbidity. Interestingly, in the same study, a higher Boey score was found to be associated with an increased chance of conversion when laparoscopic repair was attempted. Albeit a precise and useful tool for research purposes, the APACHE II system is cumbersome to employ in daily practice, not to mention that the calculated score may vary with different time points of assessment.
It is widely accepted that patient comorbidity, age > 65 years, the presence of preoperative shock and long-standing perforation are associated with an increase in mortality and morbidity following perforated peptic ulcer.
A Danish group has recently published a revised prediction model, the Peptic Ulcer Perforation (PULP) score, based on a multicentre study including more than 2600 patients with perforated gastric or duodenal ulcers treated surgically over a 6-year period.16 It comprises eight variables including (1) age over 65 years, (2) active malignancy or human immunodeficiency virus (HIV) infection, (3) cirrhosis, (4) steroid use, (5) presentation more than 24 hours after symptom onset, (6) preoperative shock, (7) serum creatinine higher than 130 μM/L and (8) the four levels of ASA score (2–5). In the study, PULP score was reported to predict postoperative 30-day mortality with an area under curve (AUC) of 0.83, which was significantly better than the conventional Boey score (AUC = 0.70) and ASA score alone (AUC = 0.78).
Treatment
Some 50% of perforated peptic ulcers have spontaneously sealed at the time of admission to hospital.17 Such an observation has led to the sporadic advocacy of non-operative management, especially for older patients with poor premorbid status.18 However, the non-operative approach has not been widely accepted, with the exception of a few centres. Donovan et al. were among the first to advocate use of diatrizoate meglumine (Hypaque) water-soluble contrast meal for confirmation of spontaneous sealing of the perforation.19 In the absence of duodenal scarring and contrast extravasation, patients with subphrenic free gas were managed non-operatively with nasogastric suction and intravenous antibiotics. By implementing this preset policy, these authors were able to report an overall mortality rate of only 4.6% in a series of 249 patients.20
The non-operative approach to the treatment of perforated peptic ulcers has been studied in a randomised trial consisting of 83 patients, of whom 40 were assigned to non-operative management while the rest underwent immediate laparotomy.21 After 12 hours, 28% of the patients in the non-operative group failed to improve and required surgery. Though mortality rates were comparable between the two approaches, the incidence of intra-abdominal collections and sepsis was much higher in the non-operative group. Most strikingly, patients over 70 years of age were found to be less responsive to the non-operative approach when compared to the younger patients. In concluding the study, the authors did not recommend routine use of non-operative management for peptic ulcer perforation. However, they recognised that the results suggested that patients (a) did not need to be rushed to the operating theatre and time could be more usefully spent in better preoperative resuscitation, and (b) in selected patients the non-operative approach has a role to play.
Though the majority of surgeons now advocate immediate surgical intervention for perforated peptic ulcers, there remains debate as to what constitutes the most appropriate operation. The omental repair technique first described by Graham in 1937 involves patching the perforation with a piece of detached omentum.22 It is no longer a common practice nowadays and has been largely replaced by the pedicle omentopexy which is usually secured in place by two to three tie-over sutures (Fig. 6.2a, b). For a time such a simple omentopexy repair was associated with a high subsequent ulcer relapse rate.23–25 This resulted in the popularity of adding a definitive acid-reducing procedure, such as distal gastrectomy or vagotomy, to lower ulcer recurrence.26–29 In a prospective follow-up study, 107 selected patients with perforated pyloroduodenal ulcers undergoing omental patch closure and parietal cell vagotomy were evaluated up to 21 years later. The operative mortality was only 0.9% and the recurrent ulcer rate by life table analysis was 7.4%, with a re-operative rate of only 1.9%.30As a result, emergency vagotomy was once a standard of care recommended for peptic ulcer perforation.
FIGURE 6.2 (a) A small perforation at the juxtapyloric area. (b) Pedicle omental patch repair on the perforation site secured with absorbable sutures.
The addition of a truncal or highly selective vagotomy may cause little early morbidity or mortality apart from prolonging the time of anaesthesia. However, resectional surgery in the form of emergency antrectomy or distal gastrectomy has been shown to increase mortality in patients with perforated peptic ulcers, and therefore should be avoided wherever possible.31
All these ‘definitive’ surgical approaches, however, have changed over the last two decades following better understanding about the pathogenesis of peptic ulcer disease; most surgeons now prefer a much less aggressive ‘damage control’ tactic when treating ulcer perforations.32,33 The availability of Helicobacter pylori eradication regimens and potent proton-pump inhibitors provides excellent means to alleviate patients' ulcer diathesis without the need for a definitive acid-reducing procedure. A simple omental patch repair and thorough peritoneal toileting are the only key manoeuvres needing to be done during the emergency operation. The only exception is probably those suffering from giant duodenal perforations (> 2 cm), which may not be amenable to a simple patch repair (see below).34
Role Of Helicobacter Pylori And Non-Steroidal Anti-Inflammatory Drugs (NSAIDs)
While the relationship between Helicobacter pylori infection and uncomplicated peptic ulcers is indisputable, the role of the bacteria in ulcer perforation was once enigmatic. Reported infection rates in earlier series varied markedly from 0% to 100%.35–37 The difference in the number, timing and types of diagnostic test used to diagnose H. pylori infection in these studies may have accounted for the widely different findings. Furthermore, the frequency of NSAID intake can be an important confounding factor. In one consecutive series of 73 patients with perforated duodenal ulcers, intraoperative antral biopsies for rapid urease test and histopathology were taken. These revealed an H. pylori infection rate of 70%,38 the figure rising to 80% if NSAID users were excluded. Some studies also reported a high proportion of NSAID use among patients with perforated gastric ulcers, especially in those without H. pylori infection, suggesting that these drugs may represent a factor of relevance to H. pylori-negative ulcer perforation.39,40
The more affirmative evidence confirming a causal relationship between H. pylori infection and non-NSAID peptic ulcer perforation comes from long-term follow-up studies. A study of 163 consecutive patients with perforated peptic ulcer managed by simple closure confirmed the strong association between persistent H. pylori infection and ulcer.41 Of the 163 patients, 47.2% were found to have H. pylori infection during endoscopy at a mean follow-up of 6 years. Male gender and H. pylori were the only two independent predictive factors of ulcer relapse in the study.
In a randomised study, 99 patients with perforated duodenal ulcers and confirmed H. pylori infection received either 1 week of anti-Helicobacter therapy or 4 weeks of omeprazole.42Endoscopic surveillance after 1 year showed that ulcer recurrence rate decreased significantly from 38% in the control group to 4.8% in the eradication group. A recent meta-analysis has further consolidated the efficacy of H. pylori eradication in the prevention of ulcer relapse after simple repair of duodenal ulcer perforation.43 The pooled incidence of 1-year ulcer relapse from three prospectively randomised trials was only 5.2% in patients treated with H. pylori eradication, which was significantly lower than that of the control group (35.2%).
These two studies42,43 confirm that medical eradication of H. pylori results in resolution of ulcer diathesis without the need for long-term antacid therapy or definitive surgery in patients who have undergone simple omental patch closure for perforated duodenal ulcers. The majority of patients with perforated peptic (pyloroduodenal) ulcers should be managed by simple omental patch closure, thorough peritoneal lavage, and subsequent treatment with proton inhibitors and eradication of H. pylori as necessary.
The Role Of Cyclo-Oxygenase-2 (COX-2)-Specific Inhibitors
Selective COX-2 inhibitors have been approved for symptomatic treatment of arthritis for nearly two decades. In contrast to conventional NSAIDs, COX-2 inhibitors provide anti-inflammatory effects without suppressing synthesis of prostaglandins essential for maintenance of mucosal integrity.44 In the Celecoxib Long-term Arthritis Safety Study (CLASS), a double-blind randomised controlled trial, selective COX-2 inhibitor was confirmed to be associated with a significantly lower annualised incidence of peptic ulcer complications when compared to non-selective NSAIDs (0.44% vs 1.27%, P = 0.04).45 However, such a protective effect was negated if the patient used aspirin concomitantly.45,46 In another population-based study, a 44% increase in the annual number of prescriptions for NSAIDs (both selective and non-selective) had been recorded since the introduction of COX-2 inhibitors, but the annual hospitalisation rates for perforated peptic ulcer decreased from 17 per 100 000 person-years to 12 per 100 000 person-years during the study period.47
The impact of COX-2 inhibitor and traditional NSAIDs on perforation-related death was recently determined in a population cohort study.48 While the observed 30-day mortality of the entire cohort of 2061 patients with ulcer perforation was 25%, that among NSAID and COX-2 inhibitor users was much higher, amounting to 35%. The increase in mortality associated with use of COX-2 inhibitors was similar to that of traditional NSAIDs, with an adjusted mortality rate ratio (compared to non-users of NSAIDs) of 2.0 and 1.7, respectively. COX-2 inhibitors, although being able to reduce the overall incidence of perforated ulcers, do not confer any advantage in clinical outcomes if perforation occurs. Such an observation may be related to the poor underlying comorbid condition of the patients who require extended use of NSAIDs or COX-2 inhibitor therapy.
As a result of knowing that the mortality risk is high once perforation occurs in the long-term NSAID users, protective measures against development of peptic ulcer in this group of patients have been extensively investigated. In an economic evaluation study using the Markov model and data extracted from a systematic review, the prescription of a proton-pump inhibitor (PPI) was shown to be cost-effective for people with osteoarthritis no matter whether they are taking a traditional NSAID or COX-2 selective inhibitor.49 Importantly, the cost-effectiveness of adding a PPI remained valid even in patients at low risk of gastrointestinal adverse events.
Laparoscopic Versus Open Repair
As the ulcer diathesis is readily corrected by H. pylori eradication and cessation of NSAID usage, most patients with ulcer perforation can now be treated with a simple patch repair together with a thorough peritoneal washing. These operations used to mandate a midline laparotomy, but developments in minimally invasive surgery have revolutionised the surgical approach. The first laparoscopic closure of peptic ulcer perforation was reported in 1992.50 Since then several series have been published.51–53 However, interpretation of these data is difficult due to marked heterogeneity in the study designs, patients' demographics and the laparoscopic techniques used. With the exception of the two truly randomised trials from Hong Kong, all the others are either single-centre series or retrospective non-randomised comparative studies.54,55 Most of them reported significantly better, albeit marginal, outcomes in the laparoscopic group, which could be due to selection bias when choosing the surgical approach for individual patients.
In a systematic review of 15 publications, the laparoscopic closure was found to be associated with significantly less analgesic use, shorter hospital stay, less wound infection and lower mortality rate compared to open surgery.56 However, shorter operating time and less suture-site leakage were clear advantages of the open technique. Interestingly, in a more updated review comprising data extracted from 56 studies, apart from confirming the above-mentioned findings, the issue of conversion to open surgery was addressed.57 The overall conversion rate was 12.4%, with the main reason being the size of perforation and inadequate localisation of the perforated pathology. The authors also identified that patients with a Boey score of 3, age over 70 years and symptoms persisting longer than 24 h were associated with a higher morbidity and mortality, and recommended these factors to be relative contraindications for laparoscopic intervention.
Laparoscopic repair of perforated peptic ulcers results in significantly less analgesic use, shorter hospital stay, less wound infection and lower mortality rate compared to open surgery at the expense of a longer operating time and higher incidence of suture-site leakage.56 Elderly patients and those with a Boey score of 3 and symptoms for > 24 hours should undergo open surgery.57
Giant Duodenal Ulcer Perforation
Giant duodenal perforation (> 2 cm in diameter) remains a challenge to most surgeons because failure of omental patch repair is not uncommon (2–10%), which can lead in turn to increased morbidity rates and a resultant higher mortality (10–35%).58,59 Converting the perforation into a controlled fistula by suturing the perforated site around an indwelling Foley catheter or T-tube was once advocated as a salvage measure. However, leakage remains a major concern with this technique and it should be reserved for patients with overtly unstable physiology to which the operative time is limited. For patients who are haemodynamically stable, a distal gastrectomy incorporating resection of the perforation-bearing duodenum is considered to be a viable option by some surgeons. However, due to pre-existing scarring and recent tissue loss, the duodenal stump thus created can be difficult to manage. Various procedures have been described, keeping in mind the friability of tissues. Nissen's double-layer closure technique,60 catheter duodenostomy (Fig. 6.3),61 lateral duodenostomy through the third part of the duodenum,62 and making a duodenojejunostomy with a Roux-en-Y segment of jejunum63 have all been mentioned. Nevertheless, it is noteworthy that the best way to avoid a difficult duodenal closure is to avoid an unnecessary gastrectomy, even in the presence of giant duodenal perforation. The alternatives to gastrectomy in these situations include converting the large perforation into a Finney pyloroplasty, use of omental plugging technique64 or fashioning a controlled tube duodenostomy after primary closure of the perforation.65It cannot be overemphasised that most of the technical advocates are based on level III evidence only because of the rare nature of giant perforations. However, if there is no option but to proceed with a distal gastrectomy, a Roux-en-Y reconstruction is preferable as this will permit enteral nutrition in the postoperative period while the (inevitable, but hopefully controlled) duodenal fistula is allowed to close.
FIGURE 6.3 A catheter duodenostomy for managing a difficult duodenal stump.
Perforated Gastric Ulcers
Reports pertaining to gastric ulcer perforation only are scarce.66 On a statistical basis, gastric perforation tends to cluster among the elderly and is more likely to be associated with use of NSAIDs, but other possible causes include carcinoma, lymphoma and gastrointestinal stromal tumours. Perforation in the older age group carries a less favourable prognosis, partly related to their impaired physiological response to the septic insult, and more so to the delay in presentation, which is not uncommon. While simple omental patch remains the best option (after biopsying the lesion – see below), most surgeons prefer an ulcer excision if this can easily be carried out and without compromise to the stomach lumen. Formal gastric resection is not indicated unless either of these options is not possible, and this is rare.
Because a small minority of gastric perforations are malignant, and the possibility of gastric lymphoma needs to be remembered, it is essential to biopsy the ulcer edge, if it is not being excised.67 Definitive surgery, if then required, can be carried out at a later date once the pathology has been reviewed, full staging investigations carried out, and after a full and frank discussion with the patient regarding prognosis and alternative treatment options. This is particularly true for gastric lymphoma, which can now be managed almost entirely by chemotherapy, radiotherapy or a combination of the two.
Though surgery for a perforated gastric cancer was once considered to be invariably palliative in nature, recent evidence suggests that long-term survival is still achievable in selected patients (stage I and II). A systematic review based on nine published articles encompassing 127 patients surgically treated for gastric cancer perforation reveals that the diagnosis of malignancy was known in 14–57% during the preoperative and intraoperative phases.68 While mortality rates for emergency gastrectomy ranged from 0 to 50%, a two-stage approach was adopted in five of the nine series, and patients able to receive an R0 gastrectomy during the second-stage operation demonstrated significantly better long-term survival (median 75 months, 50% 5-year) compared with patients who had only simple closure. It highlights the importance of not taking a too pessimistic stance if the biopsy of a repaired gastric ulcer turns out to be carcinoma.
Simple omental patch closure with biopsies of the ulcer edge or ulcer excision is the best option for perforated gastric ulcers. The former works surprisingly well, even for large lesions and is to be preferred where possible over resection. Gastric resection, often at night, in difficult circumstances on a sick patient and increasingly by surgeons who may not be experienced in upper gastrointestinal surgery should only be undertaken as a last resort, when all other methods of closure are considered doomed to failure.
Oesophageal perforation
Classification
Perforation of the oesophagus can be broadly classified into iatrogenic and spontaneous. Iatrogenic perforations represent the most common cause of oesophageal perforation, with most occurring during diagnostic or therapeutic endoscopy, while some are related to para-oesophageal operations, such as fundoplication, cardiomyotomy, bariatric procedures, etc. The reported incidence of perforation for rigid oesophagoscopy is 0.11% and for fibre-optic endoscopy varies from 0.018% to 0.03%.69,70 Therapeutic endoscopy is associated with a much higher frequency of perforation (1–10%).71,72 Spontaneous rupture of the oesophagus (Boerhaave's syndrome) accounts for approximately 15% of perforations and classically occurs after forceful retching or vomiting, often related to an episode of binge eating and/or drinking.73 The perforation is typically located just proximal to the oesophagogastric junction and more commonly opens into the left pleural space. Other less common causes of oesophageal perforation include foreign body penetration, traumatic endotracheal intubation, nasogastric catheterisation and the swallowing of corrosives.
Diagnosis
The overall mortality following oesophageal perforation was estimated to be 18% in a recent review covering 726 published patients.74 Paramount in the management of this highly lethal emergency is expedient diagnosis and judicious clinical management. However, the presentation can be non-specific and may easily be confused with other disorders such as spontaneous pneumothorax/pneumomediastinum, myocardial infarction, acute pancreatitis and pneumonia. The essential attribute is to maintain a high index of suspicion, especially if a normal laparotomy has been performed. Pain is the most common symptom, which may occur anywhere from the neck to the abdominal region, depending on the location, the cause and the time elapsed between onset and presentation. Less often, dysphagia, odynophagia, dyspnoea, cyanosis and fever are other possible complaints. A preceding history of forceful vomiting, oesophageal instrumentation, para-oesophageal surgery and chest trauma should always be taken into consideration. Physical examination and plain radiograph may reveal surgical emphysema over the neck and upper chest wall (Fig. 6.4). Hydrothorax or pneumothorax may also be found on erect chest X-ray (Fig. 6.5). Hamman's sign (a crunching sound heard on chest auscultation) is found less commonly than suggested in textbooks.
FIGURE 6.4 Plain radiograph of a patient with surgical emphysema in the neck due to a mid oesophageal perforation following endoscopic ultrasound examination and transmural biopsy.
FIGURE 6.5 Left hydropneumothorax revealed on decubitus chest radiograph in a patient with Boerhaave's syndrome.
The role of ancillary investigations for oesophageal perforation depends largely on the condition of the patient. If a patient is relatively stable and requires no endotracheal intubation, a water-soluble contrast swallow is a relatively easy way to confirm the diagnosis, determine the location, and assess the severity of the perforation (Fig. 6.6). However, for patients with profound sepsis mandating intubation and ventilatory support, a contrast-enhanced CT scan of the suspected region will be required to help establish the diagnosis (Fig. 6.7). A careful flexible oesophagoscopy is a very useful technique when the diagnosis is in doubt after these other investigations and often before surgery, as it confirms the exact site and length of perforation.
FIGURE 6.6 A water-soluble contrast swallowing showing massive leak at the lower oesophagus in Boerhaave's syndrome.
FIGURE 6.7 Pneumomediastinum and left pleural effusion revealed by CT scan in a patient with lower oesophageal perforation.
The management of oesophageal perforation starts with aggressive resuscitation and close monitoring of the physiological parameters. Further contamination of the mediastinum is minimised by making the patient ‘nil by mouth’. Although insertion of a nasogastric tube reduces the potential for ongoing leakage from refluxed gastric contents, in the presence of a perforation that might be managed non-operatively (see below), this will need to be inserted either under direct vision (endoscopically) or using X-ray guidance. Vigorous intravenous fluid replacement and broad-spectrum antibiotics are prescribed for control of systemic infection, and antifungal therapy should also be considered. Chest drainage should be established if there is hydrothorax or hydropneumothorax and most patients require intensive care support. Subsequent management should be individualised according to the patient's clinical condition, the presence of comorbidities, the underlying oesophageal pathology, the location of perforation, and sometimes the time lapse between perforation and diagnosis. Nowadays, a multidisciplinary team approach involving medical, endoscopic and surgical management is increasingly adopted in the care of such patients.
Non-Operative Management
Candidates for conservative treatment, as defined by Cameron and colleagues in 1979, include those with a small perforation that does not breach the mediastinal pleura, together with mild symptoms and no evidence of systemic sepsis.75,76 Preferably, the perforation is accompanied by a re-entry of extravasated contrast medium into the oesophageal lumen. Any sympathetic pleural effusion has to be drained, with the effluent regularly inspected to exclude fistulation. Nutritional support, either by the parenteral route or increasingly by the operative insertion of a feeding jejunostomy tube, is commenced early.
Previous experience suggests that patients who respond to non-operative management are those with small cervical tears following oesophagoscopy or traumatic endotracheal intubation, well-circumscribed intramural dissection after pneumatic dilatation for achalasia and small anastomotic leaks following oesophageal surgery.77,78 Non-operative treatment of spontaneous rupture (Boerhaave's syndrome) and iatrogenic perforation of the thoracic oesophagus with pleural contamination used to have a high failure rate in previously reported series and such patients were often managed by early surgical intervention. However, all this has changed over the last decade with the introduction of aggressive conservative strategy and the increasing interest in endoscopic stenting technology.
Vogel et al. were the first to advocate an ‘aggressive conservative’ approach in a series of 47 patients with oesophageal perforation (10 proximal and 37 thoracic).79 Of these 47 patients, 34 were managed non-operatively with repetitive radiographs, aggressive image-guided chest drainage and operative intervention when indicated (only needed in 30% of the patients). The authors reported no mortality and an admirably high oesophageal healing rate in the series, even among patients with Boerhaave's syndrome. In another cohort study by a tertiary referral centre of 81 consecutive patients with acute oesophageal perforation managed over a 20-year period, there was a marked temporal increase in the frequency of utilising CT scan for diagnosis and treatment monitoring of the patients. The percentage of patients being treated non-operatively also increased significantly from 0% in the first 5-year cohort to 75% in the last one.80 Parallel to these results was a marked reduction in complication rate and mean hospital stay. There was no difference in final mortality rates between patients treated with a non-operative approach and those who underwent immediate surgery.
One hindrance in promulgating wider use of the non-operative approach for oesophageal perforation lies in the difficulty of selecting appropriate candidates for this treatment while those who will benefit from early surgery are not delayed unnecessarily. The criteria set by Cameron et al. were far from evidence based, and the subsequent modification by Altorjay et al. also restricted non-operative management to patients with almost microperforation only.81 Recently, an oesophageal perforation severity scoring system has been suggested by the Pittsburgh group.82 Based on 10 clinical variables (old age, tachycardia, leucocytosis, pleural effusion, fever, non-contained leak, respiratory compromise, delay in diagnosis, cancer and hypotension), a maximum score of 18 can be derived at the time of admission. The prevalence of complications, mortality and the length of hospital stay were found to be significantly varied among groups with different scores (less than 2, 3–5 and more than 5). Notably the data suggested that patients with low clinical scores had worse outcomes if treated operatively when compared to those managed by the non-operative approach. These authors concluded that patients with a low clinical score, especially those with minimal mediastinal contamination and no respiratory distress, should be tried with aggressive non-operative treatment and surgical repair reserved for those with signs of deterioration.
Endoscopic Stenting
An important adjunct now available in dealing with these ‘minor’ perforations is endoluminal stenting.83,84 The recent development of a retrievable type of silicone-coated stent has shown some promising results (Fig. 6.8). In a prospective series of 18 patients with iatrogenic perforation of the oesophagus, 17 were treated with the Polyflex (Boston Scientific, Natick, MA, USA) stent in addition to chest drainage.85 Except for one patient with a persistent leak who required operative repair, the remaining 16 had successful resolution of their perforations as confirmed by oesophagram. Most of the patients were able to resume oral intake in 72 hours, and the mean time before the stent was removed was 52 days. Alternatively, elderly and frail patients who suffer perforation following attempted dilatation of a malignant stricture can be managed by insertion of a covered permanent stent.
FIGURE 6.8 Contrast radiography of the oesophagus. (a) Barium swallow demonstrating a tight malignant stricture. (b) Water-soluble contrast swallow after dilatation demonstrating a perforation. (c) Water-soluble contrast study in the same patient following insertion of an expandable metal stent, demonstrating no further leak.
The effectiveness of self-expandable covered stents in non-operative treatment of oesophageal rupture has been confirmed by a recent systematic review based on 25 selected studies encompassing 267 patients.86 Due to a lack of randomised controlled trials, this study is already the best current evidence supporting the use of endoscopic stenting in treating oesophageal perforation that is not associated with profound sepsis.
The technical success rate of stenting was 99% and clinical success was achieved in 85% of patients. Though 34% of the patients had a stent-related complication (mainly stent migration), surgical intervention was required in only 13%, in whom the rupture site failed to heal completely after 6–8 weeks of stenting. These stents should be removed as soon as possible to prevent longer-term problems.
Surgical Intervention
Early surgical intervention should be seriously considered if patients have profound sepsis upon admission, if adequate chest drainage cannot be achieved, or when clinical deterioration is observed after a period of aggressive non-operative management. In addition to thorough drainage of the mediastinum and the pleural space, three surgical manoeuvres for control of leakage may be applied:
Oesophageal exclusion and diversion
The technique of oesophageal exclusion and diversion was described several decades ago. It involves division and closure of the oesophagus proximal and distal to the site of injury, with creation of an end-cervical oesophagostomy. Subsequently it has evolved to a side cervical oesophagostomy for proximal diversion of saliva and a staple transection of the oesophagogastric junction to prevent reflux of gastroduodenal contents back into the oesophageal lumen. Nevertheless, such an incomplete diversion often ends up with continuous soiling in the mediastinum, leading to persistent thoracic sepsis. With time, it proves to be a suboptimal treatment and is now reserved mainly for patients who are too unstable to undergo definitive repair or resection.
Primary repair
Primary repair is an appropriate option for perforations occurring less than 24 hours after the injury, because surrounding tissues are neither excessively inflamed nor ischaemic. As the laceration on the mucosal side can be much longer/wider than is appreciated from the adventitial surface, a cautious longitudinal extension of the muscular defect is often required for better assessment of the extent of the mucosal injury. Following careful debridement of the necrotic tissues along the perimeter of the perforation, the edges are then closed with interrupted absorbable sutures (Fig. 6.9). It may be possible, in very early perforations with very healthy tissue, to close the defect wound in two layers: first the mucosa/submucosal layer, followed by the muscular layer. However, in the majority of circumstances a single all-layer suture is adequate. Enrolment of reinforcing autogenous vascular pedicle flap remains controversial.87,88 Most advocates of reinforcement flaps suggest an onlay patch rather than a wrap, to minimise the possibility of stricture at the repair site. A major concern with the wrap technique is the induction of an obstructive component distal to the repair site, rendering it more prone to leakage.
FIGURE 6.9 Primary repair with interrupted absorbable stitches for a lower oesophageal rupture.
Though technically feasible, primary repair is associated with a high failure rate and many surgeons now advocate insertion of a T-tube into the perforation, which is closed around it (after debridement), producing a controlled fistula. This can be removed at a later date, often many weeks/months later, once the patient has recovered and the surrounding leakage (managed by appropriately placed chest drains) has subsided.
Immediate resection
Several series have pointed out that primary repair resulted in greater morbidity and even mortality than immediate oesophageal resection.89–91 However, this may well relate to the underlying cause of perforation and whether the cause remains.
In a series of 41 patients, of whom 25 patients underwent surgical repair of oesophageal perforation, about one-third had continued swallowing difficulty that mandated regular oesophageal dilation after an average of 3.7 years.92 A high incidence of functional or structural deficiency was observed after primary repair of oesophageal perforation, which indeed affected the quality of life of the patients.
Oesophagectomy is therefore a reasonable option (if the surgeon is experienced in this procedure) in patients with underlying disease (including carcinoma, if the lesion was considered operable before the perforation occurred) and if the diagnosis has been made early and the patient is in a good condition. Occasionally it may also be required in patients presenting late, in whom simple closure over a T-tube is not possible. This scenario is, however, very uncommon. (The reader is referred to the Oesophagogastric Surgery volume in this Companion to Specialist Surgical Practice series for more detailed information.)
Primary repair with or without T-tube drainage is the best option for the management of oesophageal perforations depending on time from occurrence to diagnosis. Patients with an underlying condition related to the perforation, in whom the diagnosis has been made early (< 24 hours) and who are in a good condition, can be considered for resection if the appropriate surgical expertise is available.
Corrosive perforation
Corrosive injury to the upper gastrointestinal tract is an entity very different from other mechanical types of oesophageal perforation. It is a notoriously difficult situation associated with dreadful rates of morbidity and mortality. While more than 80% of accidental corrosive ingestion occurs in children, injuries in adults are usually intentional, and thus more severe in extent.93 The mortality rate ranges from 10% to 78% in cases of attempted suicide.94,95 The severity of damage depends on the type, concentration, volume and duration of contact of the agent with the mucosal surface. Making a distinction between acid and alkali as to the extent and severity of oesophagogastric damage has been elusive. It is well known that acid induces a coagulating burn injury, which tends to be self-limiting with the coagulum formation. In contrast, alkali causes a liquefactive necrosis to the tissue, leading to dissolution of protein and collagen, saponification of fats, and thrombosis of blood vessels. All these incur a deeper damage and even transmural perforation.
The acute management of corrosive ingestion entails a quick and efficient assessment of the patient's vital signs. Laryngeal oedema can be life threatening and establishing a patent airway is of paramount importance. Either an early endotracheal intubation or tracheostomy may be required in difficult cases before further resuscitation and management. It is generally agreed that early gentle endoscopic examination is of value for both prognostic consideration and decision-making.96,97 Surgical intervention, such as thoracotomy, drainage, mediastinal debridement, proximal diversion of saliva with a cervical oesophagostomy and exclusion from gastric reflux, are better performed early if signs of full-thickness perforation are evident on radiological or endoscopic investigations. Reconstruction should be deferred until sepsis has subsided. More often than not, reconstruction is done by fashioning a jejunal or colonic interposition loop through an extra-anatomical plane (e.g. the substernal or presternal routes).
Perforation after endotherapy for mucosal/submucosal tumours
Since its introduction in the 1980s, endoscopic mucosal resection (EMR) has been increasingly practised for upper gastrointestinal lesions identified at early stages.98–100 With the snaring technique, en bloc resection used to be a challenging task as the average size of lesion removed is limited (< 15 mm in diameter).101 The recent development of new endoscopic accessories, such as the insulated-tip knife and hook knife, has opened a new horizon for such purposes.102,103 The novel endoluminal procedure, endoscopic submucosal dissection (ESD), has virtually no limitation on the size of resection, provided that the lesion is superficial with a low chance of nodal spread.104 However, ESD is a technically demanding procedure with a significantly higher risk of perforation compared with conventional EMR.105,106 As most ESD-induced perforations can be recognised intraoperatively, attempts to close the defect by various techniques have been reported.107 Fujishiro et al. described a 100% success rate of closing ESD perforations using endoscopic clips.108 The average duration of hospital stay was only 12 days and none of the 27 patients in the series required surgical salvage. More importantly, as reported in an intermediate-term follow-up study by Ikehara et al., perforation during ESD for early oesophageal and gastric cancers has not been associated with increased risk of dissemination of the malignancy.109
Duodenal and jejunal perforations during endoscopic retrograde cholangiopancreatography (ERCP)
Visceral perforations due to ERCP are not uncommon, with an incidence reported to range between 0.5% and 2.1%.110,111 The injury carries a death rate over 15%. Risk factors include older age of patients, suspected sphincter of Oddi dysfunction, a dilated duct, performance of sphincterotomy and longer duration of the procedure.112 ERCP-related perforation can be further subclassified according to the causative mechanism and location because management strategy may vary accordingly. Stapfer et al. first described a typing system in 2000 to guide the treatment for the situation (type I: duodenal wall injury; type II: peri-Vaterian injury related to sphincterotomy; type III: distal bile duct injury probably related to guidewire or basket injury; and type IV: retroperitoneal air alone).113 While perforation of the oesophagus and stomach due to insertion of the side-view duodenoscope can be managed as discussed in the previous sections, type I injury with lateral or medial duodenal wall tear is often related to difficult positioning of the endoscope in the attempt to obtain access to the papilla. Vulnerability further increases if there is anatomical deformity such as a stricture or tumour compression. This kind of perforation tends to be intraperitoneal, which is unlikely to be self-contained, and early surgical repair with debridement of devitalised tissue is recommended. Small perforations less than 1 cm in diameter can be closed primarily with transverse sutures in one or two layers. For perforations of a larger size, or when the diagnosis has been delayed, jejunal serosal patch, duodenal diversion or conversion to a Billroth II gastrectomy are alternative options, but to some extent the choice depends mainly on the patient's clinical condition and intraoperative findings.
Another prominent risk factor for ‘scope’ perforation is a history of previous Billroth II gastrectomy, because retrograde insertion of the endoscope along the afferent loop may overstretch and rupture the jejunum around the relatively fixed duodenojejunal flexure. The reported incidence under such circumstances can be as high as 10% even in expert centres and early surgical repair is usually necessary.114,115
In contrast, perforations resulting from endoscopic sphincterotomy (type II) or guidewire/basket (type III) are mostly retroperitoneal with a small size of leakage and therefore more amenable to non-operative therapy. Although some perforations can be identified at the time of the ERCP, many are not suspected until after the procedure. Clearly, early suspicion is raised by the presence of abdominal pain post-procedure, but the observation of a paraduodenal or perinephric gas shadow on plain abdominal radiographs is suggestive of retroperitoneal perforation. In such cases, and even if the plain X-rays are unremarkable, a CT scan of the abdomen with oral and intravenous contrast should be organised to confirm the diagnosis and extent of the perforation. Small perforations with minimal extravasation of contrast can be successfully managed by non-operative means, including nasogastric drainage, antibiotics and parenteral nutrition. Some surgeons also advocate biliary decompression by either an internal plastic stent or external nasobiliary drain. Radiologically guided percutaneous drainage is indicated if a sizeable collection is shown by the initial imaging. However, surgical intervention should be offered to patients with extensive extravasation of contrast or clinical deterioration despite initial non-operative treatment. This can be a formidable undertaking and is associated with significant morbidity and mortality. Thankfully they are relatively rare, but when they occur each patient will need to be managed according to first principles: debridement of ischaemic and damaged tissue, primary repair of the defect if possible, diversion of biliary contents (by means of a T-tube) and consideration given to duodenal exclusion (closure of pylorus and gastrojejunostomy). In such circumstances a feeding jejunostomy should also be inserted for long-term nutritional support.
It has been shown in various series that one of the key prerequisites for successful non-operative management of ERCP perforations is the early recognition and diagnosis of the situation.116,117
Key points
References