A 77-year-old man presented with five episodes of sepsis to different hospitals in the year after having a left lower lobectomy for squamous cell carcinoma. In the first two episodes, a rim-enhancing left pleural effusion was seen and drained under computed tomography guidance. In his third and subsequent episodes, the effusion was noted as being present but was considered too small and loculated to drain effectively. His admission chest radiograph at his fifth presentation, which showed nonspecific changes at the left base, is shown in Figure 1A. Computed tomography and bronchoscopy failed to demonstrate a bronchopleural fistula. Therefore, we combined both techniques. The patient underwent bronchoscopy in a near-upright position under conscious sedation with midazolam. At bronchoscopy, 15 ml of Niopam 300 (an iodine-containing contrast agent routinely used in computer tomography enhancement, angiography, and arthrography) was diluted with an equal volume of sterile normal saline to reduce its viscosity. Three 10-ml aliquots of diluted Niopam 300 were gently injected by hand over 5 minutes into the left lower bronchial stump under fluoroscopic guidance (Figure 1B). Contrast was seen pooling passively in a cavity as well as extending in an arboreal pattern into the bronchi (Figure 1C). Subsequent bronchoalveolar lavage was performed with 120 ml sterile normal saline and aspirated to dryness. Fluoroscopy demonstrated residual contrast in a pleural-based cavity (Figure 1D), whereas the arboreal pattern was lost, indicating washout of contrast from the airways. The patient was observed in the bronchoscopy recovery area for 2 hours and then sent to the radiology suite. Finally, computed tomography after the bronchoscopy provided definitive proof of a fistula (Figure 1E).

Figure 1. (A) Plain chest radiograph of the patient on admission, showing nonspecific changes at the left base. (B) Fluoroscopic image showing moment of contrast injection into left bronchial stump. The white arrow points at contrast entering the pleural cavity. (C) Fluoroscopic image showing contrast in the pleural cavity (thick white arrow) and also in the airways in an arboreal pattern. The bronchoscope is indicated by a thin white arrow. (D) Fluoroscopic image showing persistence of contrast in a pleural cavity after bronchoalveolar lavage while the arboreal pattern is lost (white arrow shows “washout” from airways). (E) Computed tomography image after the bronchoscopy with patient lying supine shows dependent pooling of contrast in a pleural-based cavity (white arrow).

Download Figure | Download Powerpoint

[More] [Minimize]

Bronchopleural fistula is a significant complication affecting ∼2% of postresection patients (1). There are no established guidelines for their management, and most clinicians focus on intensive antibiotic and nutritional support, empyema drainage, and closure of the tract by surgical or bronchoscopic means (2). Many patients are debilitated owing to multiple factors, such as surgery, repeated bouts of sepsis before recognition of the diagnosis, and preexisting comorbidities, and hence may not be suitable for further surgical intervention. Novel interventions include fistula occlusion with fibrin sealant (2), other sclerosants (3), or endobronchial valves/coils (4, 5). We describe a novel method of visualizing small bronchopleural fistulae that cannot be detected with conventional bronchoscopy or imaging. Our method requires access to bronchoscopy services, contrast media, and fluoroscopy, which are usually available in most district general hospitals. Proving the existence of a fistula then leads to the possibility of tract closure to end the vicious cycle of episodes of sepsis. We note that this method is dependent on the availability of fluoroscopy in the endoscopy suite. An alternative method in patients with indwelling pleural drains would be to inject methylene blue into the bronchial stump during bronchoscopy in a manner similar to that previously described. The appearance of methylene blue in the drain output then proves the existence of a fistula. Our patient was deemed too frail for surgery and is receiving antibiotics pending bronchoscopic glue injection. We recommend that clinicians maintain a high index of suspicion and be open to using novel approaches such as the one we describe if a bronchopleural fistula cannot be localized using conventional methods.