The surgical management of non-malignant aerodigestive fistula

Operative technique

When considering surgery as treatment for ADF, several factors should be specifically assessed for. It is imperative that a careful search for malignancy is performed prior to surgery, particularly as many patients will have a preceding history of proximal oesophageal squamous cell carcinoma (SCC) treated with chemo-radiotherapy. If active malignancy is present in the context of an ADF, this represents locally advanced disease with poor outcome, rarely amenable to curative surgical intervention. In these patients, endocopic treatment should be considered to alleviate symptoms, coupled with chemo-radiotherapy if appropriate. The physiological state of the patient must also be thoroughly assessed, to ensure that the risks of major morbidity and mortality after surgery are minimised, and that the patient would be able to recover from such intervention. Patients should be carefully optimised, and where indicated, the pre-operative placement of a feeding jejunostomy and a venting gastrostomy to improve the nutritional and metabolic state, and to minimise continued soiling of the airway, should be performed.

Once a patient is deemed to have an ADF curable by surgery, secondary factors relating to the ADF and surrounding tissue become important considerations. A larger defect, a history of previous local radiotherapy and endoscopic intervention are all factors which make surgery more challenging. Also, the location of the ADF is important, as more proximally sited fistulae are amenable to repair through a neck incision, yet for distal ADF a thoracotomy is mandated, carrying a greater risk of major morbidity and mortality. If there has been significant local contamination, then it may be prudent not perform a synchronous reconstruction, as the likelihood of an anastomotic dehiscence increases. In these patients, a delayed reconstruction confers improved chances of better recovery. However, given the heterogenous aetiology of ADF, each case should be considered with a view to an individualised treatment plan.

At induction, for tracheo-oesophageal fistulae (TOF), it is important that the endotracheal tube balloon is sited distal to the fistula. This will avoid inadvertent damage to the cuff whilst dissecting and exposing the fistula, and negate the possibility of ventilatory embarrassment intra-operatively. Furthermore, this manoeuvre minimises further contamination of the respiratory tract by manipulation of the affected structures during surgery.

ADF exploration and repair

This may involve either an incision in the neck for proximal fistulae, or a thoracotomy for more distal ADFs. In the neck, dissection must proceed to mobilise the thyroid with careful identification and preservation of the recurrent laryngeal nerves and parathyroid glands. The oesophagus should be circumferentially mobilised, as this manoeuvre will allow the pharyngo-laryngeal complex to be gently pulled superiorly and away from the thoracic inlet, to provide good access to the fistula. Once the fistula has been identified, it can be dissected free and a primary repair of the oesophagus and trachea with absorbable sutures can be performed. It is critical that the fistula is accessible from both sides of the neck to ensure complete control of the airway during the repair, whilst also facilitating a pedicled strap muscle interposition flap. This reinforces the repair by providing a physical barrier between the two suture lines.

In the thorax, a similar approach is used with an intercostal flap which is carefully prepared at the time of thoracotomy. Once the fistula has been identified, again, it is dissected free and a primary repair performed, with the intercostal flap placed between the suture lines.

Exclusion

Exclusion surgery involves isolating the oesophagus from alimentary tract continuity, both proximal and distal to the fistula. This involves an incision in the neck to access the proximal oesophagus, where, once circumferentially mobilised, it is transected above the fistula and brought to the skin as an oesophagostomy. If the fistula is very proximal, then the superior oesophagus may be left in situ, and a large T-tube placed in the lumen with the distal limb of the tube brought to the skin.

Next, a laparotomy is performed where the oesohago-gastric junction (OGJ) is mobilised and the stomach transected below this, from the lesser curve through to the fundus. This manoeuvre excludes the oesophagus from the GI tract entirely, whilst preserving the majority of the stomach for future reconstruction. The small stomach remnant attached to the OGJ is brought to the abdominal wall, where a generous gastrostomy is fashioned. This allows retrograde access to the excluded oesophagus, for both endoscopic surveillance and therapy, and facilitates venting of oesophageal mucous.

Our unit policy is to defer reconstruction as a second, staged procedure. This allows the patient a period of recovery, whilst respiratory and nutritional optimisation continues. Furthermore, by fashioning an anastomosis at the index operation in a potentially contaminated surgical field, there is a higher chance of a leak. If this were to occur, there is substantial risk of fistula recurrence. Where possible, the stomach is used a conduit, and is brought to the neck through the retrosternal space, thus avoiding the need for a repeat thoracotomy. If there is insufficient proximal oesophagus, the stomach may be anastomosed directly to the inferior pharyngeal constrictors.

Resection

This is normally reserved for large or recurrent fistulae. For proximally sited ADF - those affecting the trachea, this will involve resection of the oesophagus, via a transthoracic approach. The fistula is identified, and the oesophagus dissected away around it. However, the oesophageal tissue intimately involved with the fistula is left in situ, thus avoiding direct dissection of the trachea and minimising the risk of an air leak. The tracheal defect with the overlying oesophageal tissue is then primarily closed, with the latter acting as a buttress reinforcing the tracheal repair. Typically, a gastric conduit is utilised for reconstruction, necessitating a laparotomy.

For more distal ADF, those affecting the bronchus intermedius and more distal, a thoracotomy is performed to identify the fistula. A segmentectomy or lobectomy of the lung can be performed, dependent on the size of the defect and the quality of the surrounding parenchyma. Thus, the affected distal airway and the fistula are excised en bloc. The oesophageal defect can be repaired primarily, utilising an intercostal flap to reinforce the repair, or an oesophageal resection is performed if the defect is very large and unlikely to heal. In these instances, given the anastomosis will be at a distinct site from the ADF, a synchronous reconstruction can be performed safely.

In our experience, tracheal resection is a very challenging operation, with the risk of significant short and long-term complications [2, 3]. Owing to the limited vascularity of the trachea, healing, particularly in this cohort of patients, may be protracted, necessitating prolonged mechanical ventilation. Thus, we have preferred to avoid such an operative intervention. However, for very large TOFs, or those where a circumferential injury to the trachea is present (such as cuff related fistulae), or where other intervention has failed, tracheal resection and reconstruction may be indicated. Mathisen et al provide an operative description and experience of this technique [3].

Preceding history and previous intervention

The median age at diagnosis of ADF was 56 years (range 29–73 years). Three patients had a previous diagnosis of oesophageal malignancy; all were treated with chemo- and/or radiotherapy, and one with surgical resection in addition. Two patients had a prior diagnosis of Tuberculosis (TB) and had received anti-microbial therapy in the past, and one patient previously had surgical intervention following a post-emetic oesophageal leak. In two patients, no obvious cause of ADF was identified, likely representing congenital fistulae that had persisted into adulthood. Of these cases, two patients (1 and 6) developed oesophageal strictures after their initial treatment. Patient 1 had undergone several balloon dilations and stent placements, with the stent subsequently eroding into the airway (Fig. 1). Similarly, patient 6 also received a stent which directly caused the fistula. Table 1 summarises key patient factors.

Patient	Age at ADF Diagnosis (years)	Sex	Preceding Diagnosis	Preceding Intervention	Chemo-Radiotherapy	Preceding Treatment Related Complication	Previous Endotherapy
1	60	F	SSCa Proximal Oesophagus	Definitive chemo-radiotherapy	Chemo-radiotherapy	Radiotherapy related stricture	-×2 stents -×3 dilations
2	57	M	TBb	Medical therapy	–	–	–
3	59	M	Adenocarcinoma Distal Oesophagus	Ivor-Lewis Oesophagectomy	Neo-adjuvant chemotherapy	Anastomotic leak	–
4	73	M	Unknown	–	–	–	–
5	32	F	TBb	Medical Therapy	–	–	–
6	33	M	Boerhaave’s Syndrome	Repair of leak	–	Re-leak; stricture	-× 1 stent
7	55	F	Oesophageal B Cell lymphoma	Chemotherapy	Chemotherapy	–	–
8	29	F	Unknown	–	–	–	–

^aSCC- Squamous Cell Carcinoma; ^b TB- Tuberculosis

ADF characteristics

The two patients with an unknown cause of fistula had a very long history of symptoms, and had been managed in the community with a diagnosis of asthma (Table 2). The median time to ADF development for the three patients with a malignancy was 15 months (range 3–21), with the shortest time affecting a patient who had an oesophageal lymphoma (Patient 7). She had a complete response to chemotherapy, with a residual fistula persisting (Fig. 2). Both patients with TB had a long interval after curative medical therapy, although they had suggestive symptoms for some time prior to referral. Most patients presented with recurrent chest infections and symptoms suggestive of aspiration. Of these, one patient (4) presented with acute respiratory failure owing to overwhelming infection caused by aspiration. Of interest, he had a fistula affecting the very proximal trachea (Fig. 3).

Patient	Time to ADF Development (months)	Fistula Site	Fistula Size (mm)	Main Symptoms	Endotherapy to Treat ADF
1	15	Proximal trachea 20 cm	12	Aspiration	×3 stents
2	> 30 years	Oesophagus-right bronchus intermedius 30 cm	12	Recurrent chest infections	–
3	21	Gastric conduit-lung 25 cm	16	Recurrent chest infections	Endoclip
4	> 30 years	Proximal trachea 17 cm	5	Aspiration; Respiratory embarrassment	–
5	144	Distal oesophagus-lung 38 cm	15	Haemoptysis	–
6	3	Oesophagus-carina 24 cm	15	Recurrent chest infections	×3 stents
7	3	Oesophagus- left main bronchus 26 cm	5	Recurrent chest infections	–
8	> 20 years	Proximal trachea 17 cm	3	Recurrent chest infections	–

The size of the fistula ranged from 3 to 16 mm, with the larger defects affecting those who had a prior diagnosis of carcinoma or who underwent surgical treatment. The location of the fistula in relation to the airway too was variable, reflecting the site of underlying disease. Those thought to be congenital were very proximal. Those secondary to TB were both distal, involving the smaller bronchi and lung parenchyma at the original Ghon focus. Patient 6 presented with Boerhaave’s Syndrome, and was initially managed with surgery to repair the oesophageal defect. However, he subsequently re-leaked, which again was managed surgically with a repair over a T-tube, but then developed a stricture at the site of injury, which was treated with an oesophageal stent. This eroded into the airway at its proximal extent, causing a fistula at 24 cm, with subsequent referral to our unit (Fig. 4).

After the diagnosis of ADF, three patients underwent further endotherapy in an attempt to manage the fistula prior to referral to our unit. Patient 1 had 3 oesophageal stents placed, but given the proximal location of the ADF, these all slipped distally. Patient 3, who had developed an ADF between the airway and a gastric conduit, most likely after a sub-clinical leak, had endoclips placed via flexible gastroscopy which failed to close the ADF. Patient 6 too had a stent^s placed to treat the fistula without the desired effect.

Surgical intervention for ADF treatment

Two patients (1 and 6) underwent an oesophageal exclusion operation. These were both performed as staged procedures with delayed reconstruction. The reason for performing this operation was that in both patients there was sufficient concern regarding the state of tissue. Patient 1 had previous radiotherapy for an oesophageal SCC, on a background of achalasia requiring a myotomy via thoracotomy several years previously. The degree of tissue inflammation, scarring and adhesions precluded a transthoracic resection. Thus, through a neck incision, the oesophagus was transected above the fistula and an oesophagostomy fashioned with a synchronous repair of the fistula. Distally, the oesophagus and OGJ was transected and a venting gastrostomy fashioned. After a period of optimisation and treatment of longstanding respiratory disease, the patient underwent reconstruction utilising a gastric conduit through the retrosternal space. Patient 6 too had severe inflammation and adhesions in the chest following his surgical management of Boerhaave’s syndrome. Exclusion and subsequent reconstruction with a colonic conduit (he had previously undergone a distal gastrectomy for benign ulcer disease) was performed. In both cases, the native tissues were poor enough to carry a high risk of leak with primary anastomosis at index surgery.

Patient 2 had a distal ADF, approached through a thoracotomy. The right bronchus intermedius was involved, and the ADF and affected parenchyma was excised as a segmentectomy, with oesophageal repair. Patient 5 underwent an exploration of the fistula through a thoractomy with a lobectomy. The fistula, along with associated necrotic parenchyma was excised en bloc, with a subsequent suture repair of the oesophagus. Patient 7, after chemotherapy for lymphoma, underwent an oesophageal resection. Again, severe residual inflammation was noted at the time of the surgery precluding repair of a small (5 mm) fistula. In addition, our oncology colleagues felt there was a reasonable risk of future recurrence of lymphoma, thus a resection was advocated. A primary repair of the left main bronchus was performed, utilising a flap muscle graft. A single-phase operation was performed as the anastomosis was at a distinct site to the fistula.

Morbidity and mortality

Three patients suffered from complications following surgery. Patient 2 developed a severe respiratory infection on Day 4 following surgery, requiring bronchoscopic washout. Patient 7 suffered a small anastomotic leak, necessitating prolonged nil oral alimentation. Her nutrition was maintained with jejunal feeding, and a contrast swallow at 7 weeks confirmed complete healing of the leak, after which gradual oral feeding was commenced.

Patient 4, who presented with respiratory failure, was affected by severe post-operative recurrent chest infections, and an inability to wean off mechanical ventilation. This had been anticipated, and hence a tracheostomy had been placed at the time of surgery. Despite 4 months of intensive management, he passed away with respiratory failure and multiple organ dysfunction. He represents the only mortality in this series.

Normal oral alimentation following surgery was achieved in all patients, bar patient 4. The time to attain this milestone ranged from 2 to 3 weeks in all cases, with the exception of patient 7 who had suffered a leak. She required supplemental enteral feeding at home for a short period. We reserve the use of contrast studies and formal swallow assessment for the very proximal fistulae, where the risk of leak and aspiration is highest. At the time of last follow-up, no patient demonstrated clinical evidence of recurrent ADF.

Hospital episodes statistics (HES)