Successful emergency management of a bleeding tracheoinnominate fistula

Abstract

In this case, we describe a novel approach to achieving temporary haemostasis in acute massive haemorrhage from a bleeding tracheoinnominate fistula. We report the case of a 42-year-old man admitted to hospital after suffering 80% body surface area burns. Thirty days following the percutaneous insertion of a tracheostomy, spontaneous massive haemorrhage occurred via the tracheostomy stoma, the tracheostomy tube and the mouth. After hyperinflation of the tracheostomy cuff which controlled airway contamination, effective tamponade was achieved using a hyperinflated balloon on a Foley catheter that was introduced by direct laryngoscopy into the upper larynx above the tracheotomy stoma. This provided temporary control of the bleeding until definitive management through ligation of the innominate artery via median sternotomy.

Background

Tracheoinnominate fistula is a rare but devastating complication of tracheostomy. The incidence varies from 0.3% to 0.79%,¹ with a reported mortality in excess of 92%.¹ Early identification and management is paramount, with almost all patients dying without early diagnosis and prompt surgical intervention.²

Case presentation

A 42-year-old man was admitted to Royal North Shore Hospital with 80% body surface area burns. He had no significant comorbidities. He underwent immediate resuscitation before multiple surgical procedures for burn debridement, dressing (including cadaveric skin dressings) and grafting over the next several weeks.

A little over a week after admission, a tracheostomy was placed to facilitate long-term airway access in view of the need for ongoing mechanical ventilatory support and multiple surgical procedures. Percutaneous tracheostomy was performed with bronchoscopic guidance: the position of the needle and the insertion of the wire were confirmed as satisfactory on bronchoscopy via the in situ endotracheal tube before a size 8 Tracoe dual lumen device (TRACOE medical GmbH, Germany) was inserted. The procedure was not difficult or complicated and there was no bleeding. It was, however, noted that extra air had to be introduced into the cuff to maintain airway occlusion. Chest X-ray confirmed a satisfactory position of the tracheostomy tube in the trachea.

Over the following 3 days, a significant cuff leak was intermittently problematic requiring high cuff pressures. This leak was exacerbated when the patient’s respiratory function deteriorated requiring increased Positive End Expiratory Pressure (PEEP) and airway pressures to maintain oxygenation. In response, the tracheostomy tube was upsized to a 9 mm Tracoe twist: a translaryngeal endotracheal tube was reintroduced, the old tracheostomy removed over a guidewire and the new tube introduced over the in-situ wire as for a primary tracheostomy. Again, bronchoscopy was used throughout the procedure and it was specifically noted that no cause for the persistent leak was identified. The tube change appeared to have ameliorated the cuff leak problem with reduced cuff pressures.

The patient’s respiratory function subsequently improved, and the patient tolerated multiple surgical procedures for dressings and grafting.

Thirty days following the insertion of the original tracheostomy, small amounts of blood-stained secretions and tracheal aspirates were noted for the first time. Later that evening, spontaneous massive haemorrhage via the tracheostomy stoma, the tracheostomy tube and the mouth occurred.

Acute resuscitation

The acute haemorrhage was first managed with hyperinflation of the tracheostomy cuff that evidently ceased airway contamination and enabled satisfactory ventilation. The hospital’s massive transfusion protocol was initiated, and balanced blood product resuscitation (in 1:1:1 ratio) was commenced.

With a view to translaryngeal intubation, direct laryngoscopy was attempted and the mouth and pharynx cleared of massive blood clots. It became apparent that some control of the ongoing haemorrhage was afforded by anterior displacement of the larynx with the laryngoscope.

Effective tamponade was then achieved using a hyperinflated balloon on a Foley catheter that was introduced by direct laryngoscopy into the upper larynx above the tracheotomy stoma. With control of the bleeding, transport to the operating theatre was then undertaken for definitive management.

Definitive surgical management

In the operating theatre, massive haemorrhage again occurred with deflation of the Foley catheter. Haemostasis was re-established with reinflation and with laryngeal manipulation. A median sternotomy was undertaken and the proximal innominate artery ligated. This did not immediately control the bleeding until the distal artery was also ligated suggesting very good backflow down the distal innominate artery.

A 1.5 cm area of pressure necrosis was reported on the upper part of the innominate artery and a large defect in the trachea was noted on the operation report, evidently the site of the tracheostomy. It was reported that the tracheostomy entry point was low in the trachea and the tracheostomy tube was compressing the trachea for some segments above the tracheal entry point. There was no suggestion of any point of erosion through the tracheal wall—the blood was flowing into the trachea via the tracheostomy site.

Outcome and follow-up

The subsequent course was satisfactory. The tracheostomy was later reinserted surgically via the same site. The reintroduction of the tracheostomy was originally undertaken using a Shiley dual cannula device (Medtronic, USA). This was changed to a Portex Unipec tracheostomy tube (Smiths Medical, USA) with an adjustable flange and a flexible, wire-reinforced shaft. Figure 1 demonstrates the differences between the three devices used. This tube was positioned bronchoscopically to be 1 cm above the carina. It was then visualised using the bronchoscope introduced via the larynx; this identified that the cuff was partially inflated within the stoma itself. The tube was positioned deeper in the trachea so that the cuff was circumferential within the distal trachea. This suggests very strongly that for both the Tracoe device originally employed and for the Shiley tube used for the reestablishment of the tracheostomy that the cuff could not have been fully contained within the trachea and it seems likely that this factor contributed to the pressure erosion of the innominate artery.

Figure 1.

Comparison of three tracheostomy devices used in this case. The Portex Uniperc adjustable flange tracheostomy (A) is made of a soft reinforced tube, and does not have a rigid curve as seen in the Tracoe twist (B) or Shiley (C) dual cannula devices.

The patient recovered fully from the event and was later decannulated successfully with satisfactory respiratory function and with a near normal voice. Despite his ligated innominate artery, he maintained adequate cerebral and right upper limb perfusion via retrograde flow in his Anterior Communicating Artery and right common carotid. He was subsequently discharged home and continues to recover from his significant burn injury.

Discussion

Tracheoinnominate fistula is a rare complication of tracheostomy though it is not clear if the development of the dilatational technique for tracheostomy has had an impact on this incidence.³ Haemorrhage usually occurs between 3 days and 6 weeks of the procedure, with a peak incidence at 7–14 days.³ Massive haemorrhage is heralded by a smaller sentinel bleed in up to 50% of cases.⁴ While bleeding within 48 hours of tracheostomy placement is usually associated with venous injury, coagulopathy or suction trauma, any bleeding occurring between 3 days and 6 weeks of the procedure should be carefully investigated further for potential tracheoinnominate fistula,³ usually with bronchoscopy or CT angiography.

Three mechanisms of fistula formation have been reported, all involving pressure erosion of the artery by different parts of the tracheostomy tube: tip erosion, shoulder erosion and cuff erosion.¹ Although the innominate artery usually crosses the trachea between the sixth and 10th tracheal rings there is considerable variability in its anatomical course.³ Factors thought to be associated with tracheoinnominate fistula are included in Box 1.

Box 1. Factors associated with formation of tracheoinnominate fistulae² .

• Low insertion site

• High position of innominate artery

• Excessive neck movement

• Radiotherapy

• Long duration of tracheostomy (although most occur between 3 days and 6 weeks of insertion)

• High pressure tracheostomy cuff

• Type of tube especially for low insertion and shoulder erosion

For the patient reported here, the type of tube may have contributed, given that the Tracoe tube is rigid. Certainly, the insertion site was lower than optimal and lower than was planned. This is probably related to the angle of the needle insertion into the trachea. The interaction of the tube type and the low insertion site is likely to be important given the later observations of cuff position with the flexible Portex Uniperc tracheostomy. The type of tracheostomy tube is only likely to be relevant if there is a very high innominate crossing point of the trachea or a low insertion site. Identifying the crossing point is not possible using ultrasound and other techniques are probably not justifiable in view of the rarity of this complication and of the nature of the illness of the patients undergoing this procedure.

Bronchoscopic visualisation of the tracheostomy tube cuff position during the original procedure may have aided in identifying the low insertion site and this may have led to the use of a softer, more deeply positioned tube. While this may have been beneficial, visualisation from above can be a difficult procedure and is not generally recommended.

The persistent cuff leak and requirement for additional insufflation of the tracheostomy cuff may have been an early indication of tube malposition. Furthermore, given that high cuff pressures are associated with mucosal ischaemia,⁵ it is possible that these excessive pressures contributed to erosion into the artery.

In this case, we describe a novel approach to achieving haemostasis in the acute phase. The initial priority in resuscitation is to limit airway soiling and facilitate ongoing oxygenation. Fortunately, this was achieved with hyperinflation of the tracheal cuff. This is considered first-line management and is said to temporarily control acute haemorrhage in up to 85% of cases.^{1 2}

If overinflation of the tracheostomy cuff fails to control bleeding, the ‘Utley manoeuvre’ has been described.⁶ Here, the tracheostomy is first replaced with a translaryngeal endotracheal tube. With the tracheostomy removed, the fascial plane anterior the trachea is entered through the tracheostomy wound with blunt digital dissection, and a finger is advanced posteroinferiorly towards the carina along the anterior surface of the trachea. The innominate artery is then digitally compressed against the posterior surface of the sternum,⁶ as illustrated in figure 2.

Figure 2.

Digital compression of the bleeding innominate artery against the posterior sternum as described by Utley (1972). Reproduced with permission of Elsevier from Ailawadi.⁷

While the ‘Utley manoeuvre’ is recommended as the next step in clinical management if cuff hyperinflation fails to control bleeding,³ it may only be required if there is persistent bleeding into the lungs with ongoing airway contamination. If bleeding is thereafter isolated to the upper airway and mouth, less invasive measures such as the approach we report, may be appropriate.

Definitive surgical management generally involves ligation of the innominate artery via a median sternotomy. Surgical repair of the arterial defect is associated with a high failure rate,² and while cases of endoluminal stenting have been reported, this approach remains uncommon given the risk of exposing graft material to the potentially contaminated fistula. Fortunately, neurological or vascular complications of innominate artery ligation are not commonly reported.³

Learning points.

• Even with the use of bronchoscopic guidance, inadvertent low placement of percutaneous tracheostomy is possible.

• Early significant cuff leak should prompt review of the tracheostomy position, and consideration of changing both the size and type of tracheostomy tube. Cuff pressures should be monitored and maintained <20 cmH₂O to avoid mucosal ischaemia.

• Tamponade of acute haemorrhage may be achieved by translaryngeal placement of a Foley catheter above the tracheostomy in order to allow stabilisation before definitive surgical management.