Bronchopleural fistula (BPF) is a sinus tract between the main stem, lobar, or segmental bronchus and the pleural space. It can be a potentially catastrophic complication following pneumonectomy or other pulmonary resection. Morbidity ranges between 25% and 71%, and diagnosis and management is often a challenge for physicians. Other common causes of BPF include pulmonary infection causing necrosis, persistent spontaneous pneumothorax, chemotherapy or radiotherapy from malignancy, and tuberculosis. Treatment for BPF ranges from medical management to bronchoscopic procedures for critically ill patients and surgical intervention for those deemed at highest risk. There is a lack of consensus regarding optimal management due to varying therapeutic success. Varoli et al. described the time of onset following surgical intervention to classify fistulas as early (1 to 7 days), intermediate (8 to 30 days), or late (more than 30 days). Although fistulas almost always occur within 3 months after surgery,  BPF following pleuropulmonary infection can occur at any point.
The most common risk factors associated with BPF in the postoperative setting include right-sided pneumonectomy and right lower lobectomy. The fistula is commonly found on the stump beside the residual lobe due to the increased risk of ischemic necrosis or pooling of secretions leading to bacterial overgrowth and colonization. The increased risk of BPF associated with right pneumonectomy is due to the more extensive resection required. Other causes include:
BPF is most commonly encountered after lung resection surgery (pneumonectomy, lobectomy, segmentectomy), with a frequency ranging from 4.5% to 20% after pneumonectomy and 0.5% to 1% after lobectomy.
Post-operative BPF may be classified as acute, subacute and chronic. The acute form is caused by surgical dehiscence and requires prompt surgical intervention. When acute, BPF can be life-threatening due to tension pneumothorax or asphyxiation from pulmonary flooding. Patients present with the sudden appearance of dyspnea, hypotension, subcutaneous emphysema, cough with expectoration of purulent fluid, tracheal or mediastinal shift, persistent air leak, and a reduction or disappearance of pleural effusion on the chest radiograph. The subacute and chronic forms are primarily related to infection and are often seen in immunocompromised or debilitated patients with multiple comorbidities. The subacute presentation is more insidious and is characterized by wasting, malaise and fever. The chronic form is associated with an infectious process and fibrosis of the pleural space.
A BPF can occur any time during the postoperative period but more often occurs within 8 to 12 days after surgery. If seen within the first 4 postoperative days, the BPF is likely secondary to a mechanical failure of bronchial stump closure and requires reexploration. A BPF also may also occur after a suppurative pneumonia, massive pulmonary infarction, or spontaneously. Empyema may occur alone or may be associated with a BPF. A BPF can occur without associated empyema, and the fluid in the pleural space in these cases is sterile. Displacement of the mediastinum to the opposite side occurs because of air on the operative side. Clearing of fluid from the pleural space and coughing up of fluid and blood suggest a BPF. The sudden reappearance of air in an obliterated space suggests either a BPF or a gas-forming infectious process. If a fistula appears in nonsurgical cases or in the delayed postoperative period, the diagnosis should be suspected when fever, productive cough, and new or increasing air-fluid levels are seen on the chest radiograph in the pleural space. Esophagotracheal fistulas are associated with coughing and dyspnea during eating and drinking. Nonresolving pneumonias should be endoscopically investigated.
Patients with BPF can present with symptoms that range from acute symptoms of tension pneumothorax to subacute symptoms of empyema. Most patients present in the first two weeks following lung resection. BPF should be suspected in the postoperative lung resection patient who presents with sudden onset of dyspnea, chest pain, hemodynamic instability, and subcutaneous emphysema. Symptoms may be less abrupt in those in whom the chest tube is still in place, and a large persistent or new air leak through the chest tube drainage system may be the only sign present. Examination findings are typically nonspecific but may reveal reduced air entry on the affected side and tracheal deviation if a tension pneumothorax is present.
Patients who present in the late postoperative period (more than 14 days), or patients with BPF from other causes including infection or malignancy, present with symptoms and signs of empyema. These include fever, malaise, muscle wasting, cough with purulent sputum, reduced air entry and dullness to percussion on the affected side. When the BPF is associated with empyema that is not adequately drained, the infection can erode through the chest wall and a pleurocutaneous opening with drainage of mucopurulent material may be seen; it is known as empyema necessitans.
Persistent air leak after pulmonary injury may represent either disruption of a bronchus or rupture of an overdistended alveolus. Several methods have been used to diagnose BPFs, including the instillation of methylene blue into the pleural space and bronchography. In addition, small metallic probes introduced through the working channel of the bronchoscope, and changes in gas concentration in the pneumonectomy cavity after inhaling different concentrations of oxygen and N2O are also useful. Bronchoscopic exploration allows proper evaluation of the stump and attempts to localize the fistula. It also helps exclude tuberculosis or other infectious etiologies and allows the introduction of sealants into the fistulous tract. While a large BPF is more likely to be seen on bronchoscopy, sequential balloon occlusion of the bronchi is sometimes used, particularly for the localization of a smaller or segmental BPF. The diagnosis of BPF is made using a combination of clinical, radiographic, and bronchoscopic findings that confirm an air leak from a major, lobar, or segmental bronchus to the pleural space. Although there are no specific laboratory findings, some patients with an infected pleural space may exhibit leukocytosis or elevated erythrocyte sedimentation rate.
Radiological features that suggest the presence or the development of a BPF include an increase in intrapleural airspace, the appearance of a new air-fluid level, changes in an already present air-fluid level, development of tension pneumothorax, and a drop in the air-fluid level exceeding 2cm . With computed tomography (CT), apart from the demonstration of pneumothorax, pneumomediastinum, and underlying lung pathology, there may be evidence of actual fistulous communication in some patients. In those who have not undergone lung resection, features of the underlying etiology such as malignancy may be demonstrated via CT with a cavitating mass and air-fluid levels.
Using standard and thin section non-contrast CT scans, Westcott and Volpe could demonstrate a fistula in 10 out of 20 patients. In almost all of these cases, the fistulous tract was located peripherally and in only one case was there an air-leak in the setting of lung resection. Three-dimensional reconstruction using the spiral technique has also been used successfully to demonstrate BPF in its entirety. The presence of BPF is suggested by the return of continuous bubbles during bronchial washing. A bronchoscope with inflation of a balloon-tipped catheter into selected airways can aid in localization, as balloon occlusion decreases or eliminates an air leak. Capnography can be used to identify the bronchial segment related to BPF. End-tidal carbon dioxide is measured by connecting a capnograph to a polyethylene catheter and has a reported a sensitivity of 83% and a specificity of 100% in the diagnosis of BPF.
Several authors have tried to localize a BPF using radiolabeled aerosol inhalation with planar and single photon emission tomography (SPECT) imaging. These imaging modalities require substantial time and a patient’s cooperation. This technique has been criticized because the aerosol tends to deposit in areas of turbulence and may lead to false-positive results in patients with obstructive airways. In addition, the estimation of the size of the BPF may be inaccurate as it is an indirect measurement from the kinetics of tracer gas during different phases of the study.
Computed tomography bronchography (CTB) has been utilized in the diagnosis of a difficult BPF. Following bronchoscopy, Sarkar et al.  injected 20 to 30 mL of a water-based nonionic low osmolar iodinated contrast medium iohexol at the suspected fistula site either through a catheter or directly through the working channel of the bronchoscope. CT was performed immediately with a targeted reconstruction of images in different planes. The fistula was easily visualized using standard axial and sagittal sections.
First-line therapy should address any immediate, life-threatening conditions, for example, endobronchial contamination, pulmonary flooding and tension pneumothorax. Since most BPFs occur early in the postoperative period and are not infected, patients can undergo surgical repair with excellent success. Bronchoscopic approaches have variable success rates and are appropriate for those who are not suitable for surgical intervention.
The patient should be placed with the affected side dependent, and adequate pleural drainage is performed. The first intervention for BPF should be drainage of air and fluid from the pleural space by chest tube thoracostomy. Pleural fluid should be sent for complete blood cell count (CBC), pH, total protein, lactate dehydrogenase, glucose, cytology, triglycerides, gram stain, and culture to assess for pleural infection. Although integral for drainage, the chest tube itself can function as a foreign body and predispose to infection at the insertion site and in the pleural space. In patients who are mechanically ventilated, the chest tube can be used to add positive intrapleural pressure during the expiratory phase or occlusion during the inspiratory phase. The aim of these interventions is to decrease the air leak during expiration to maintain positive end-expiratory pressure (PEEP) and to decrease BPF flow during inspiration. The chest tube can also be used to apply sclerosing agents, such as talc and bleomycin, to promote pleurodesis. Broad-spectrum intravenous antibiotics against gram-positive, gram-negative, and anaerobic microorganisms should be given to all patients until gram stain, cultures, and sensitivities are available. Postural drainage can be initiated after specimens have been obtained, as long as a patient can expectorate, chest cavity and chest tube drainage is less than 30 mL per day, and there is concomitant pleural irrigation.
Suture re-closure of the bronchial stump with vascularized flap coverage is curative for the fistula presenting acutely, normally less than 2 weeks after surgery. In most cases, a video-assisted thoracoscopic surgical (VATS) approach is performed, but rarely a thoracotomy is needed. Surgical closure of the fistula is done by an anterior, trans-pericardial approach thoracotomy with muscle flap to fill the pleural space or with a muscle flap coverage of the fistula with a limited thoracoplasty. Patients who present with a BPF developing late or those who develop the fistula as a complication of suppurative pleuropulmonary diseases are initially managed medically. Medical management includes dependent drainage and reduction of the pleural space, antibiotic treatment, nutritional supplementation, and adequate ventilator management.
The disease course is more complicated for patients with BPF who are mechanically ventilated. Maintaining airway pressures at or below the critical opening pressure of the fistula to promote healing, but yet provide adequate alveolar ventilation for sufficient gas exchange can present a significant challenge. Air-leaks through bronchopleural fistulae may range from less than 1 to 16 L per minute. Adverse effects of BPF in mechanically-ventilated patients include incomplete lung expansion, loss of effective tidal volume, or positive end-expiratory pressure, inability to remove carbon dioxide, and prolonged ventilatory support. The large air leak via BPF can also result in auto-triggering of the ventilator, leading to severe hyperventilation and inappropriately large doses of sedatives and/or neuromuscular blockers administered to reduce spontaneous respiration. Mean airway pressure should be reduced as much as possible including minimal PEEP, low peak airway pressures, and reducing the proportion of minute ventilation provided by the ventilator. Intermittent mandatory ventilation modes with low tidal volumes, respiratory rates, and shorter inspiratory times are appropriate.
Other techniques described to decrease an air leak include independent lung ventilation with two ventilators, differential lung ventilation, and high-frequency jet ventilation (HFJV). Selective intubation of the unaffected lung, the use of double-lumen tube intubation with differential lung ventilation, or the use of independent lung ventilation and patient positioning are all advocated. HFJV with permissive hypercapnia avoids the need for a double lumen tube (DLT) or single lumen tube with a bronchial blocker. HFJV avoids barotrauma to the unaffected lung and decreases the air leak. If HFJV is not available, BPF is one of the strict indications for split lung ventilation. A DLT is preferred because it allows physiologic separation of the lungs if an air leak becomes severe enough to require varied ventilator modes or settings for each lung. According to the airway device used, low-frequency jet ventilation or low tidal volume ventilation is applied during resection. ECMO, either veno-venous or veno-arterial, is another alternative strategy for gas exchange.
Methods to limit air flow across BPFs include direct closure, decortication, thoracoplasty, omental or muscle transposition, and completion pneumonectomy. Localization and size of the fistula may be the decisive factors in choosing between surgical and endoscopic procedures. If a patient is severely debilitated or life expectancy is limited, palliation can sometimes be provided by a surgically created pleuro-cutaneous tract to vent the pleural space on a permanent or temporary basis. Patients should be assessed and treated for potential empyema. The development of post-pneumonectomy empyema (PPE) occurs at a rate between 2% and 16%. In the setting of PPE, BPFs are present in approximately 60% to 80% of patients and carry a high mortality rate ranging from 21% to 71%.
Patients with BPF from causes other than lung resection, including malignancy and infection, are treated with bronchoscopic methods because the bronchial stump is affected by disease and stump revision is not feasible. Bronchoscopic management is primarily useful for temporary fistula closure but can be used as a bridge to curative surgery if the underlying cause is reversible. A flexible bronchoscope provides superior and precise access to a greater portion of the bronchial tree than the rigid bronchoscope. Outcomes are variable with rates of successful closure ranging from 30% to 80%. Bronchoscopic intervention is a viable first option in small bronchopleural fistulas that are less than 5mm in diameter. Endoscopic closure of a BPF is lower cost, has a reduced rate of trauma, and can be performed in critically ill patients. The BPF can be closed endoscopically if there is no evidence of infection in the pleural cavity.
In patients with BPFs equal to 8 mm, fistula closure with airway stents, coils, or Amplatzer devices is a viable option. The largest case series of 31 patients with BPF reported that the Amplatzer device was effective in 96% of cases, lasting for up to 18 months. Silicone and covered metallic stents for BPF closure involving the central airway have also been described. Angiographic coils, alone or in combination with other occlusive materials, can also successfully treat BPF. Occlusive agents act first as a plug mechanically sealing the leak and then later induce an inflammatory process with mucosal proliferation and fibrosis, creating a permanent seal. Fistula repair also occurs by the organization of granulation tissue.
There are several case series and reports of patients with BPF who were treated with occlusive materials, none of which have been compared. Bronchoscopic placement of glutaraldehyde sterilized lead shot, gel foam and tetracycline, autologous blood patch, fibrin glue, gelatin-resorcinol mixture, oxidized regenerated cellulose, albumin-glutaraldehyde tissue adhesive, cryoprecipitate fibrin glue and NN-butyl 2-cyanoacrylate has been attempted at the fistula site. The 2 component fibrin cryoprecipitate glue (calcium gluconate and cryoprecipitate along with topical thrombin [1000 IU/mL]) is delivered at the fistula site through a double lumen catheter inserted into the operative channel of the bronchoscope, creating a fibrin clot that occludes the fistula. Small BPFs can be occluded using an endobronchial injection of ethanol, polidocanol, and tetracycline followed by autologous blood. Although not often utilized, argon plasma coagulation and neodymium-doped yttrium aluminium garnet (Nd:YAG) laser have been described for patients with small BPFs. Transplanted bone marrow-derived mesenchymal stem cells, and dehydrated amniotic membrane allograft have been reported as successful in BPF closure.
Patients should be monitored after fistula closure for clinical symptoms of recurrence, chest tube output of air, and with imaging of the chest. Patients with a sealed fistula should not have an air leak, and imaging should demonstrate stability or resolution of air in the pleural space. Repeat bronchoscopy is not routine and only performed if fistula recurrence or a complication is suspected. If valves or stents are used, a chest CT scan and bronchoscopy are repeated at 6 weeks to assess for complications. For patients who fail surgical or bronchoscopic intervention, options include repeat surgery, an alternate bronchoscopic method, or in rare cases, an open window thoracostomy such as Eloesser flap thoracostomy or a Claggett window.
In patients who have had a pneumonectomy, acute findings of tension pneumothorax are essentially pathognomonic for BPF. Although other conditions such as postoperative tension chylothorax or hemorrhage into the pleural space may also cause the findings of tension pneumothorax, the affected hemithorax should fill with fluid rather than air. Tension pneumothorax may also be due to a displaced or obstructed chest tube; ensuring a patent chest tube while imaging is being obtained may distinguish this phenomenon from true BPF. In patients who present with empyema, BPF is likely if the effusion contains air and there is an air leak after chest tube placement. The culture of the effusion may help distinguish anaerobic infection from BPF. Bronchoscopic findings of a bronchial defect will distinguish BPF from other etiologies.
A BPF can cause significant morbidity, prolonged hospitalization and mortality. Mortality rates vary between 18% to 67%. The most common cause of death is aspiration pneumonia and subsequent acute respiratory distress syndrome or development of tension pneumothorax. Pierson et al. reported their experience with all cases of mechanical ventilation at a major trauma center during a 4-year period. They found that 39 of the 1700 patients receiving mechanical ventilation had BPFs lasting at least 24 hours. Overall mortality in these 39 patients was 67%, and Pierson et al. found that mortality was higher when BPF developed late rather than early in the illness (94% versus 45%). Patients with large air leaks also had a significant mortality compared to smaller leaks. They concluded that BPF during mechanical ventilation identifies patients with high mortality, but that unmanageable respiratory acidosis from this complication is rare. Reports using omental and thoracic flaps have shown decreased mortality. Sirbu et al. found the mortality rate of BPF patients to be 27.2% (6 of 22 patients).
BPF is one of the most severe complications after thoracic surgery, and no effective prophylactic procedure is available. Evidence for buttressing the bronchial stump leading to a reduction in the incidence of a bronchopleural fistula is conflicting. Level 1b evidence exists that an intercostal flap reduces the risk of BPF in diabetics. Thoracic surgeons should consider coverage of the bronchial stump by an intercostal muscle pedicled flap in patients at high risk of BPF.
In a prospective randomized controlled trial of pneumonectomy patients with diabetes mellitus, Sfyridis et al. reported that intercostal muscle flap coverage was effective. Sfyridis et al. performed a randomized controlled trial on 70 diabetic patients undergoing pneumonectomy. Patients were randomized 1:1 to either intercostal muscle flap reinforcement or no flap coverage; results revealed a significant reduction in BPF risk and empyema in patients with a flap.
Deschamps et al. analyzed 713 pneumonectomy patients. Univariate analysis revealed bronchial stump reinforcement was associated with an increased incidence of BPF. Surgeons utilized multiple techniques for flap coverage - including a combination of muscle flaps, parietal pleura and pericardium, serratus anterior and latissimus dorsi. Hamad et al. report a series of 50 patients undergoing bronchial stump coverage with pericardium following extrapleural pneumonectomy for malignant mesothelioma. Although 2 patients died secondary to cardiac complications, leading to 4% perioperative mortality, there were no cases of BPF. Lastly, Lardinois et al. conducted a non-randomized comparison study of 26 patients undergoing either intercostal muscle or diaphragm flaps (DF). Thirty-day mortality was zero in both groups. There was a higher incidence of complications including pneumonia, atelectasis, herniation, and BPF in the DF group, 38% versus 8% in the intercostal muscle flap group. Both methods were deemed effective by the authors.