Introduction

Pleurobiliary fistula (PBF) is a rare condition that can be congenital or occur as a complication of liver hydatid cyst, liver abscess, biliary tract obstruction, traumatic wounds, or iatrogenesis (liver resection, radiofrequency catheter ablation, biliary stenosis, radiation, biliary drainage, etc.)^1,2. The first report of PBF was described by Graham in 1897³. It represents a rare complication (0-2%) of penetrating trauma with associated liver injury¹¹. As most publications correspond to case reports there are no clear recommendations for its treatment. In 1994, Rothberg et al.¹ reported that only 34 cases had been published so far. Andrade-Alegre and Ruiz-Valdés² added 33 more cases considering those of their case series. The hallmark is the presence of postraumatic pleural effusion with bilious fluid. Determination of bilirubin in pleural fluid confirms the diagnosis of PBF². Computed tomography (CT) scan and ultrasound are the methods of choice to rule out associated complications, such as pleural, biliary or subphrenic abscesses and collections, and serve to guide percutaneous drainage, if necessary^2,3,5. However, both methods are not useful for defining the exact anatomical location of the fistula, which is necessary to determine the therapeutic approach. Endoscopic retrograde cholangiopancreatography (ERCP) is the method of choice to evaluate the biliary tree⁴. The traditional treatment of PBF involves an aggressive surgical approach, with exploratory laparotomy, biliary drainage, lavage of the abdominal cavity and thoracotomy with repair of the diaphragm. However, with the advent of endoscopy, interventional radiology and minimally invasive techniques, treatment has evolved and has turned out to be a challenge for trauma surgeons6,8. We present a case series to analyze the treatments performed to patients with PBF due to penetrating thoracoabdominal injury and their results, compare them with the published literature, and propose a treatment algorithm.

Material and methods

For this descriptive and retrospective study, we reviewed the medical records of patients admitted with penetrating trauma due to stab wound or gunshot to Hospital José María Cullen, Santa Fe, Argentina, Centro de Trauma Nivel 1 between January 2010 and January 2016. The demographic data, kinematics of trauma, diagnosis, treatment and patient progress were analyzed. Liver injuries were classified according to the American Association for the Surgery of Trauma liver injury scale based on CT scan or laparotomy findings. All the patients were evaluated and resuscitated in the shock room following the advanced trauma life support (ATLS) guidelines. The initial management was decided by the on-call trauma surgeon and general surgery residents. The subsequent management was decided by all the members of the department of general surgery. All trauma patients with hemodynamic stability undergo full-body or thoracoabdominal CT-scan as usual practice. As specialists in diagnostic imaging were not on call at the hospital, the lesions were classified by the on-call trauma surgeon. The diagnosis of PBF was made by observing bilious fluid through the pleural drainage and was confirmed by the presence of bilirubin in pleural fluid. The treatments proposed were pleural drainage (PD), PD and endoscopic or percutaneous biliary drainage, and thoracotomy or laparotomy. Length of hospital stay, morbidity and mortality were also evaluated.

Results

From total of 5688 clinical records reviewed, 614 had involvement of the thoracic and abdominal wall. Of these, 7 patients presented PBF, corresponding to an incidence of 1.1%. All the cases were confirmed with determination of bilirubin in pleural fluid. Sex, causes and mechanism of injury are summarized in Table 1. Mean age was 26 years (range 18-32). In 4 cases, pleural drainage was the initial management after trauma. Two patients with suspected associated lesions underwent PD plus laparotomy and 1 patient required urgent thoracotomy with suture of the diaphragm followed by PD. All the patients presented bilious fluid through the PD 2 and 6 (median 4.2) days after the intervention. Five patients with bilious fluid through the PD for more than 72 hours underwent ERCP between 3 and 5 days after the diagnosis was made (Fig. 1). In all these patients a wide sphincterectomy was performed during the procedure and a plastic biliary stent was placed in 2 of them. Amylase levels increased in 1 patient after the procedure who remained asymptomatic with favorable outcome; amylase level returned to normal values within 4 days. The biliary stents were removed 8 weeks after the fistula resolved, without complications. Of the 5 cases that underwent ERCP, 2 patients required ultrasound-guided percutaneous drainage during hospitalization due to right subphrenic collections.

Table 1 

Figure 1 Endoscopic retrograde colangiopancrea tography showing biliary leaks. 

Two patients of the entire series required thoracotomy due to persistent biliary leak and sepsis despite the treatments indicated and underwent decortication, washing of the pleural cavity and suture of the laceration of diaphragm. One of these patients required open pleurostomy due to chronic empyema associated with sepsis.

None of the patients died and mean time to resolution was 7 to 45 days, but 2 patients required prolonged antibiotic treatment. Data are summarized in Tables 2 and 3.

Table 2 

Table 3 

Discussion

In our experience, PBF occurred as a complication of penetrating thoraco-abdominal injuries. They are considerably rare after trauma in general and result from the combination of hepatic duct injury, inadvertent or inappropriately treated diaphragmatic injury, and bile leakage into the pleural space or bronchial system facilitated by the pressure gradient between the abdomen and the pleural space. Anderson found only 3 cases of thoracobiliary fistula in a review of 1767 cases of thoracoabdominal wounds from World War I and the Korean conflict. Carroll et al. reported only 5 cases from 254 patients with traumatic hepatic injury in Vietnam, and Trunkey et al. described only 1 case from 811 consecutive patients treated at a major trauma center^1,4,7. We have found 71 cases of traumatic PBF in the literature. In our case series, the incidence was 7 cases in 614 patients with shotgun injuries and stab wound injuries (1.1%). Ivatury et al. reported an incidence of 0-2%10 (Table 4). The complications associated with liver trauma, as biliary fistulae, hepatic hematoma, intraperitoneal bile collection and hemoperitoneum occur in 3-5% of patients and are considered predisposing factors for the development of PBF¹³. In addition, traumatic diaphragmatic injuries may occasionally remain inadvertent^4,9,10,13. Post-traumatic PBF are more common in young men and after penetrating wounds, as in our series^1,2,10. The spectrum of post-traumatic PBF varies from small, bilious pleural effusions that resolve after PD as described by Avraham⁶ and in 2 patients in our series, to large, symptomatic, infected, persistent fistulae, requiring more complex management. Bile leakage to pleural space can occur at any time depending on the grade of the injury and is more commonly observed in grade III injuries, according to our series. They usually occur within a week after the injury, as in our cases, but several authors have described a later onset, ranging between 2 and 73 days¹⁰. The diagnosis is made by the macroscopic visualization of bile through the PD. The determination of bilirubin in pleural fluid is an objective finding to confirm the diagnosis. Nevertheless, and although bilirubin was measured in all our patients, we believe this is not strictly necessary as most reports have shown.

Table 4 

Most publications have reported that PBF can be managed with a surgical or conservative approach; yet, we prefer the terms aggressive surgical approach and less or minimally invasive surgical approach. The first publications recommended an initial aggressive surgical approach. In fact, in 1967 Ferguson and Burford¹² published the necessary steps for a successful treatment of PBF: aggressive surgical treatment with early thoracotomy, closure of the diaphragm with nonabsorbable suture, decortication, lobectomy in case of bronchobiliary fistula and prophylactic decompression of the biliary tree. But in the last decade of the 20th century, some reports supporting and endorsing less invasive treatment emerged. Then, percutaneous and endoscopic drainage and decompression of the biliary tree gained importance as first line treatment in patients with PBF. Three concepts are important in the management of PBF: location of the leak, adequate drainage of pleural fluid and subphrenic collections, and decompression of the biliary tree^7,9. The best imaging test for defining biliary duct trauma is cholangiography, either intraoperatively, endoscopically (ERCP) or by a percutaneous transhepatic approach⁹. Singh et al.³ demonstrated that magnetic resonance cholangiopancreatography and hepatobiliary scintigraphy have high sensitivity to identify biliary leakages. Ultrasound and CT scan are used to visualize the pleural space, monitor the hepatic injury and possible intra-abdominal collections, and guide those that need to be drained. Pleural effusions should always be drained in case of PBF. But in case of subphrenic collections, these should also be drained to ensure resolution of FBP as the rate of failure of PD alone is 38%^1,4. Decompression of the biliary tree, either by conventional surgery, percutaneously or endoscopically, also plays an important role in the treatment. The percutaneous and endoscopic approach are part of the less invasive treatment, but the endoscopic approach is the best option as it is useful to identify the leak and biliary anatomy including third-order bile ducts and allows for therapeutic interventions such as sphincterotomy and implantation of a biliary stent or a nasobiliary tube. Nasobiliary intubation is not commonly used in the endoscopic approach but has been described to avoid a second endoscopy and for follow-up cholangiography¹³. There is no optimal timing for ERCP. Singh et al.³ performed ERCP within 72-96 hours. In a series of 22 patients (11 treated without surgery and 11 with urgent surgical exploration), Burmeister et al.¹³ reported that ERCP was performed at a mean time of 13.2 days after liver injury. In our case, ERCP was done within 3 and 5 days after the diagnosis was made. Sphincterotomy is justified as a therapeutic procedure, understanding the physiology of the sphincter of Oddi. Geenen et al.¹⁶ observed that the normal pressure in the common bile duct was 10 ± 2 mm Hg and the mean baseline pressure in the sphincter of Oddi was 8 ± 3 mm Hg. The pressure gradient between the common bile duct and the duodenum before sphincterotomy was 10 ± 1 mm Hg. Given this pressure gradient, in an intact ductal system bile flows along an open fistula tract following the path of the least resistance. After sphincterotomy, the mean basal sphincter of Oddi pressure decreases to 1 ± 1 mm Hg, favoring the flow of bile into the duodenum rather than into the fistula tract^4,16. We have performed sphincterotomy to all the patients who underwent ERCP, with favorable outcome. Complications are uncommon and include pancreatitis, cholangitis, bleeding, duodenal perforation and stent occlusion or migration; in fact, only one patient had elevated serum amylase level, but remained asymptomatic. The use of a temporary stent has demonstrated a significant decrease in morbidity and mortality compared to surgical intervention¹⁷. Stents are removed once the fistula has resolved, usually within 6 weeks, which corresponds to the accepted patency period for small-caliber stents⁹. In this study, patients who underwent stenting resolved the fistula with conservative management, but stents were removed at 8 weeks. Several recent reports of some case series have indicated a success rate of 97% for the treatment of PBF combining ERCP and stent placement, but we emphasize the small number of cases in these series¹⁰. As an example, Krige et al.¹⁴ reported 20 cases of post-traumatic PBF successfully treated with ERCP and sphincterotomy with stent placement in 18 patients and only with sphincterotomy in 2 patients. Percutaneous embolization of the peripheral intrahepatic bile duct close to the origin of the fistula under fluoroscopic guidance is another minimally invasive treatment that was described by Giménez et al.18 for a biliobronchial fistula and by Burmesteir et al.¹³ for a PBF. This technique could be safe and feasible but requires advanced interventional skills and should be considered only when conventional methods have failed and the appropriate endoscopic or radiological expertise is available.

Aggressive surgical treatment is only indicated in complex PBF complicated with chronic pleural empyema, persistent fluid drainage and/or sepsis. Some authors prefer early surgical approach via thoracotomy, describing a success rate of 97%^1,4. Thoracotomy is the approach most commonly used, since it allows the combination of pleural decortication, extraperitoneal drainage of the liver, fistulectomy and suture of the diaphragm^2,10. We have used this approach in 1 case as initial treatment and in 2 cases to treat complications developed after the diagnosis. Laparotomy is indicated in cases with major injuries of the extrahepatic biliary tree and for patients with hemodynamic instability during conservative treatment of liver trauma. Fistula closure with the greater omentum is a possible surgical method to consider¹¹.

Morbidity and mortality vary depending on the cause and associated injuries. The clinical suspicion, imaging tests and early detection of biliary leak are the key to patients’ outcome. There were no deaths in our series, and 100% of cases were cured. When we compare our series with those reviewed, PBF resolved with either the less invasive or aggressive surgical approach.

We propose an algorithm for the management of PBF (Fig. 2) suggesting chest and abdominal ultrasound or CT scan according to availability and accessibility after the PBF has been diagnosed. If a subphrenic collection is visualized, we indicate percutaneous drainage to reduce pressure and corrosion of the diaphragm generated by the presence of bile on the diaphragm. Drainage should be monitored for 72 hours, and in case of persistent drainage, ERCP and sphincterotomy with/or stent placement is indicated. If the condition is not resolved or the patient deteriorates, surgery via thoracotomy indicated.

Figure 2 Proposed management algorithm 

Conclusion

There is no consensus about the therapeutic management of these fistulae. In current surgical practice, the wide availability of interventional endoscopy has improved the minimally invasive approach for the management of PBF. Our data demonstrate that ERCP is an accurate method for identifying post-traumatic PBF and that endoscopic treatment can be safe and effective and improves patient morbidity, mortality and outcome. Pleural drainage associated with endoscopic biliary drainage was the most used combination in this paper. Surgery should be limited to fistulae not responding lo less invasive treatments or to complicated fistulae.