Chapter 18
Airway fistulas
Christophe Dooms and Jonas Yserbyt
Dept of Respiratory Diseases, University Hospitals KU Leuven, Leuven, Belgium.
Correspondence: Christophe Dooms, Dept of Respiratory Diseases, University Hospitals KU Leuven, 3000 Leuven, Belgium. E-mail: christophe.dooms@uzleuven.be
Airway fistulas are categorised by their localisation (tracheo-oesophageal, bronchopleural or alveolopleural) and by aetiology (spontaneous or as the result of an intervention). The mainstay in the treatment of nonmalignant fistulas is surgical repair, although endoscopic treatment plays a potential role when surgery is considered to be morbid or technically unfeasible. In general, the grade of evidence for endoscopic interventions is low, with expert opinion guiding clinical practice, with the exception of the use of unidirectional valves to treat alveolopleural fistulas. The latter have been subject of several cohort studies. Spontaneous airway fistulas most often originate from neoplastic disorders, infection, or macro- or microscopic deformities of the subpleural region. In neoplastic disorders, the role of surgical repair is limited, since the occurrence of the airway fistula is often a consequence of locoregional advanced disease in which endoscopic treatment plays an important role in palliation.
Cite as: Dooms C, Yserbyt J. Airway fistulas. In: Herth FJF, Shah PL, Gompelmann D, eds. Interventional Pulmonology (ERS Monograph). Sheffield, European Respiratory Society, 2017; pp. 264–275 [https://doi.org/10.1183/2312508X.10004017].
Airway fistulas are relatively uncommon but are associated with significant morbidity and mortality. Clinical signs or symptoms are related to the anatomical localisation. Tracheo-oesophageal fistulas (TOFs) present predominantly with respiratory symptoms, such as cough while swallowing. Broncho- or alveolopleural fistulas can be assessed whenever a chest tube is present, for example a continuous air leak in large bronchopleural fistulas or in patients on mechanical ventilation, or a (forced) expiratory air leak in alveolopleural fistulas.
Tracheo-oesophageal fistulas
An acquired TOF might be caused by either a malignancy or a nonmalignant entity. Regardless of aetiology, the symptoms and clinical signs are similar and predominantly respiratory, such as cough while swallowing, increased secretions, recurrent respiratory infection and inadequate nutrition with weight loss. In order to achieve a successful outcome, one must take into account the aetiology, accurate anatomical assessment and potential definitive treatment modalities. A complete evaluation consists of a clinical examination, contrast-enhanced chest CT, bronchoscopy with or without methylene blue through the oesophagoscope in order to facilitate the identification of a very small TOF (figure 1a), and oesophagogastroscopy.
Figure 1. a) Bronchoscopy after methylene blue injection into the oesophagus with identification of a small 1 mm fistula at the posterior wall of the left upper-lobe bronchus (indicated by an arrow); b) upper gastrointestinal endoscopy with a transparent plastic cap attached to the distal tip showing a fistula of 3–4 mm with clean edges located at 30 cm from incisors at 9 am; c) successful closure of a small chronic broncho-oesophageal fistula with an over-the-scope clip system.
Most malignant TOFs occur spontaneously due to tumour invasion or as a complication during or after radiochemotherapy for oesophageal, lung or tracheal cancer. The choice of the appropriate treatment depends on the characteristics of the fistula, the patient’s clinical state and the prognosis of the underlying disease. Most of these patients have an incurable cancer with limited life expectancy and mostly benefit from palliation by inserting either a fully covered oesophageal SEMS, or a tracheal silicone stent or fully covered SEMS. Stenting of TOFs in the oesophagus is preferred whenever a stenosis is present to secure the position of the stent, seal the fistula and also palliate dysphagia. In the absence of dysphagia or in the presence of a tracheal stenosis, the endoscopic placement of a tracheal stent (silicone or fully covered metallic stent depending on tracheal diameter and shape) is preferred. Ideally, the chosen stent diameter should guarantee a firm sitting, and the coated part of the stent should cover at least 1.5–2 cm of healthy wall proximal and distal to the fistula margins. The use of double (oesophageal and tracheal) stents is discouraged as this might cause pressure necrosis to the walls of both the oesophagus and the trachea between the two stents, which results in an enlargement of the TOF. Whenever the TOF is close to the main carina, a silicone or self-expandable Y-stent should be considered.
The aetiology of an acquired nonmalignant TOF can be a tracheal post-intubation injury, or prior oesophageal or tracheal surgery. Spontaneous closure in these cases is rare. Surgical correction is considered the gold standard for a nonmalignant TOF and typically requires fistula division with oesophageal repair, and tracheal repair or tracheal resection with airway reconstruction for small or large TOFs, respectively [1]. Surgery comes with an operative mortality of 3–5% and a recurrence rate as high as 11%, even in high-volume centres [2, 3]. Technically, an endoluminal stent, either in the oesophageal or the endotracheal position, might seal a nonmalignant TOF, but a mature TOF cannot heal with a stent in place. As a result, stenting is not regarded as a long-term solution. A stent can be considered temporarily as a bridge to surgical repair for very selected cases in which an improvement of the patient’s pulmonary and/or metabolic status is deemed necessary prior to surgery (figure 2). A recently introduced alternative to small TOFs is an endoscopic over-the-scope clipping. The fistula and surrounding oesophageal tissue are suctioned into a cap fitted on to the tip of the endoscope (figure 1b) and a clip is deployed (figure 1c). This clip provides circumferential approximation by encircling the defect, folding and compressing the gathered tissue towards the centre with all of the tissue edges approximated within the clip perimeter [4, 5].
Figure 2. a) Bronchosocpy performed 2.5 years after minimally invasive oesophagectomy for oesophageal cancer showing a trachea-neogastric fistula with clean edges of size 8 mm and located at the pars membranacea 5 cm proximal to the main carina; b) a temporary tracheal silicone stent of 50×18 mm inserted successfully to cover the fistula; c) after a 6-month interval, the stent has been removed, and a one-stage tracheal transection and resection of the fistula with gastric and tracheal repair has been performed.
Bronchopleural fistulas
Dehiscence of an end-to-end surgical anastomosis
Lobar resection with bronchial reconstruction and bronchial resection without lobar resection are often performed as lung-sparing surgical procedures, having in common the performance of bronchoplasty with bronchial re-anastomosis or reconstruction [6]. Overall, bronchial sleeve resections can be performed with a low risk of bronchial anastomotic complications, but a bronchopleural fistula might occur as a result of ischaemia or excessive tension. A recent review on the subject showed that the prevalence of bronchopleural fistulas or dehiscence after sleeve lobectomy ranges from 1.1% to 9.3% [7]. In the early postoperative period, the clinical suspicion of bronchopleural fistula is raised whenever a postoperative air leak persists or subcutaneous emphysema is increasing. When respiration or haemodynamics are compromised, this can potentially be life-threatening. Less often, bronchopleural fistulas can occur in the long term, most frequently presenting as a pleural infection.
The management of anastomotic problems after sleeve resections is often very challenging. Some authors recommend a surveillance bronchoscopy early after such bronchoplasty in order to identify potential threats. Surgical revision of the anastomosis is considered the standard treatment [7]. Endobronchial treatment strategies are considered less frequently applicable, as stenting of the anastomotic area after sleeve lobectomy is often technically challenging due to a short remaining main bronchus and/or segmental bronchi located very close below the anastomosis. A recently developed silicone OKI-stent (Novatech GSS) or small-sized silicone Y-stent (Novatech GSS) might be a solution in selected cases [8].
A retrospective series of 218 patients subjected to sleeve lobectomy for nonsmall cell lung cancer reported a bronchial anastomotic complication in 14 patients: seven patients required re-operation, three required airway stenting, two were managed conservatively and two were fatal [9]. Another retrospective series of 108 patients who underwent a bronchial sleeve resection reported dehiscence in seven patients of whom all but one healed conservatively, and bronchopleural fistulas in three patients of whom two were covered by a silicone bronchial stent and one healed conservatively [10].
Since its introduction in the early 1960s, lung transplantation has evolved from an experimental surgical intervention with excess morbidity and mortality to the treatment of reference for end-stage respiratory diseases. As the transplantation volume increases, anastomotic complications and more specifically dehiscence or (less frequently) bronchopleural fistulas are important determinants of morbidity and even mortality [11–13]. Anastomotic complications have classically been described in the Shennib and Massard classification, a classification of stages of healing taking into account the macroscopy of ischaemia/necrosis and the phenomena of healing (granulation, stricture, malacia), without taking into account the fact that different stages can coexist [14]. The Groupe d’Endoscopie de Langue Française proposed the MDS classification (currently under prospective validation), which overcomes this limitation, taking into account the macroscopic aspect, the diameter of the anastomosis and potential suture dehiscence [15]. In general, most cases of limited disruption of the anastomosis may evolve favourably without any intervention [13]. There is a role for protective ventilator strategies, early extubation and anti-infective/antifungal treatment, since anastomotic defects may be rapidly covered with fibrin or granulation tissue. Severe dehiscence is often life-threatening, and urgent surgical re-intervention is mandatory, although the risk of transplantectomy is considerable [11]. In all other cases of dehiscence after lung transplantation, a thorough clinical and multidisciplinary assessment is important before deciding to intervene. The use of bronchial stents to treat dehiscence after lung transplantation is less obvious than their use in the case of anastomotic stenosis or malacia [16]. The introduction of a rigid bronchoscope in the anastomotic region comprises a risk for increasing dehiscence, especially when the region is ischaemic or infected. Concerning the use of airway stents, silicone stents are preferred over a partially covered SEMS for all benign bronchial complications, although the insertion of a SEMS in the case of anastomotic dehiscence is technically easier than the insertion of a silicone stent [17–20]. The use of a fully covered SEMS might combine the technical feasibility of a partially covered SEMS with the full covering feature of silicone stents, but sufficient data on their use in lung transplantation are currently lacking. In our own experience, only four cases of a bronchopleural fistula after lung transplantation were treated with stent insertion over an 8-year time period. In two cases, a silicone stent was successfully removed after 1 and 3 years, respectively [8]. In one case, further dehiscence occurred after SEMS insertion necessitating transplantectomy, and in the other case, two attempts to extract the SEMS led to re-insertion of a third covered SEMS that was left in place permanently. Specific treatments for bronchopleural fistulas have been described in case series and consist of bronchoscopic treatment or surgery, whereas other reports only consider surgical re-interventions in the case of anastomotic dehiscence [21–23].
Dehiscence of a surgical stump
A bronchopleural fistula is a serious complication after lung resection, with an incidence of <1% after lobectomy and 5% after pneumonectomy [24]. The treatment should focus on both the fistula itself and the health status of the patient. The management of a bronchopleural fistula includes conservative treatment and various bronchoscopic or surgical procedures. As randomised or controlled studies comparing these options are lacking, the best treatment to be adopted depends largely on the size, site and morphology of the bronchopleural fistula in combination with expert opinion and knowledge of retrospective case series. Conservative treatment allows a spontaneous closure, which is obtained during chest-tube drainage, and should certainly be considered in very small fistulas, while some series have reported on success with this conservative treatment in larger bronchopleural fistulas [25–27]. Bronchoscopic treatment can be added to this conservative approach and can consist of the closure of a 1–3 mm fistula using biological (e.g. fibrin) or synthetic (e.g. cyanoacrylate) glues with adhesive and sealer properties, with the highest success rate in fistulas of 1 mm [28]. For the definitive bronchoscopic treatment of bronchopleural fistulas ranging from 3 to 6 mm, the use of spongy calf bone or polyvinyl alcohol sponge and glue are available strategies [29, 30]. In bronchopleural fistulas, temporary or definitive fistula coverage has been reported with placement of endoscopic devices such as silicone or a fully covered SEMS or atrial septal occluder devices such as the Amplatzer septal occluder (St Jude Medical, Minneapolis, MN, USA) (figure 3) [28, 31–33]. Definitive closure of a bronchopleural fistula of ≥6 mm by a bronchoscopic intervention represents a rarity, and surgical repair is therefore the preferred approach whenever feasible.
