Lethal aorto-oesophageal fistula 16 years after aortic prosthetic reconstruction: Delayed rupture as cause of death

Introduction

Aorto-oesophageal fistula (AOF) is an abnormal communication between the oesophagus and aorta. It has numerous possible causes, including thoracic aortic aneurysm, foreign body ingestion, invasion of a malignant oesophageal tumour, oesophageal reflux, tuberculosis, congenital abnormalities, traumatic injuries and iatrogenic forms following surgical interventions. ¹ An example of the latter is aortic replacement, which can be performed through both open and endovascular approaches. Focusing on potential iatrogenic causes, a symptom-free interval occurs between the surgical intervention and the clinical manifestations of AOF. This delay does not appear to depend upon the particular surgical approach (i.e. thoracic endovascular aortic repair (TEVAR) or open surgery). ¹ , ² The main symptoms of AOF are represented by Chiari’s triad, consisting of mid-thoracic pain, sentinel haematemesis and lethal massive haematemesis. ³ The latter can occur with a variable delay from the sentinel haemorrhage, ranging from a few minutes to several weeks. ⁴ This delay may be due to various causes, such as aortic wall vasospasm, hypotension caused by initial bleeding or fistula closure due to the formation of a haematoma. ⁵ Moreover, other signs and symptoms include dysphagia, caused by extra-luminal compression of the oesophagus, and an oesophageal wall ulcer located near the surgical site. When the fistula does not cause death because of acute and abundant bleeding, the passage of food bolus and/or saliva through it could provoke local infection (i.e. mediastinitis) or general infection (i.e. sepsis). AOF is rarely clinically investigated because the massive bleeding leads to death before diagnosis. ⁶ In cases where fistula is suspected and the patient is haemodynamically stable, the gold standard imaging is a computed tomography scan of the chest with contrast. The treatment has three key aims: bleeding control, avoidance of the onset of mediastinitis and carrying out an oesophageal repair. ⁷ Regarding the available therapeutic strategies, several authors have reported that in situ allograft replacement in association with subtotal oesophagectomy could be a valid option. ⁸ , ⁹ The difficulty in the diagnosis and the severity and suddenness of the clinical manifestations explain the high mortality rate, even when the fistula is correctly detected. Akashi et al. reported a mortality rate at 18 months of 70.3% after diagnosis and consequent surgery. ¹⁰ This work reports a case of massive gastroesophageal haemorrhage caused by AOF. This followed a surgical repair performed 16 years prior to the death of the patient.

Case report

The case under examination concerns a 58-year-old Caucasian male, who suffered polytrauma due to falling from a height of 12 metres while undertaking his job. Following the above-mentioned injury, he presented with left hemi-diaphragm rupture, multiple fractures of the splanchocranium, rib fractures, haemomediastinum and intima laceration of the descending thoracic aorta at the isthmic level. For this reason, the resection of the torn aortic component and its replacement with a 22 mm Dacron prosthesis (i.e. a synthetic polyester material which replaces natural body tissue) was performed the day after the injury. Following the surgical intervention, the subject appeared in good health during the succeeding years, although he did describe fatiguability, occasionally associated with general malaise, nausea and vomiting. Sixteen years after surgery, he was admitted to an emergency department, where he presented with non-specific signs and symptoms, such as agitation, nausea and vomiting. On medical examination, he presented with only rhonchi and fever. His vital signs, as recorded by physicians, were: temperature 38°C, blood pressure 170/100 mmHg, heart rate 145 bpm and oxygen saturation 98% with room air. He was treated with 1 g paracetamol and discharged. However, he died four days later from hypovolemic shock provoked by profuse haematemesis. Thirty-four days after his death, a medico-legal autopsy was performed at the request of the District Attorney. The most significant finding regarding the oesophagus was a reddish-coloured mucosa in the middle and distal third (Figure 1), with the presence of a black-red-coloured material of gelatinous consistency, compatible with blood, in the latter. The aforementioned material was also found in the stomach. A full-thickness wound of the mucosa was observed at the border between the middle and distal third of the oesophageal wall. Investigation of the aorta showed a full-thickness loss of substance, with sharp borders and smooth edges, in the peri-prosthetic tissue, immediately close to the prosthetic tract of biocompatible material (Dacron). The location of the lesion in the aorta was in close contact with the oesophagus. The examiner detected a fistulous passage between the oesophagus and the thoracic descending aorta, gently probing the path (Figure 2). This lesion has been considered primum movens of hypovolemic shock due to profuse haemorrhage.

OPEN IN VIEWER

Figure 1.

Blood material in the distal third of the oesophagus.

OPEN IN VIEWER

Figure 2.

Fistulous passage between the thoracic descending aorta and the oesophagus after formalin fixation.

Discussion

This report describes a rare case of a delayed presentation of AOF following open vascular surgery – that is, the replacement of torn aortic component with a Dacron prosthesis that led the patient’s death due to massive oesophagogastric bleeding. This complication occurred 16 years after the surgical intervention.

Aortic replacement can be performed using both open and endovascular approaches. Successful open surgery of thoracic aortic graft replacement was first described in 1953. ¹¹ However, nowadays, this technique is not commonly used due to the introduction of an innovative and less risky endovascular repair technique. Although the first description on the latter procedure dates back to 1994, ¹² the US Food and Drug Administration) approved the first commercial device for TEVAR in 2005. Even if the mortality rate associated with the endovascular approach is lower in comparison to the open procedure, several complications, such as embolisation, rupture or thrombosis, are shared between the two procedures. Rare and late complications include aortic dissection and AOF or aorto-bronchial fistula. ¹³ AOF is detected in 1–5% of patients undergoing prosthetic replacement in open vascular surgery, ¹ while it occurs in 1.9% of those undergoing endovascular repair of the descending thoracic aorta. ² Most AOF cases described in the literature developed after an average delay of about one year following surgery, regardless of the surgical technique. In 1991, Hollander et al. evaluated several cases of open surgery and observed that the average time interval between the prosthetic vascular replacement surgery and the occurrence of signs/symptoms was 14.8 months, ¹ while in 2013 Uno et al. reported a delay of 1–16 months ² with respect to the TEVAR procedure.

This work introduces some considerations regarding the pathogenesis of AOF development and its delayed onset. There are three main hypotheses underlying the occurrence of this phenomenon described in the scientific literature. The first is the chronic pulsation of the peri-prosthetic aortic tissue against the oesophageal wall, which may cause its erosion or necrosis due to the compression of the vessels that supply it. The second is a local infectious process of the prosthetic graft, with rupture of the abscess into the oesophagus. The third is the development of a pseudoaneurysm, where the portion of the suture has the lowest tensile strength. ⁵ The main limitation in this case is that the autopsy was performed 34 days after death on an unburied and poorly preserved body, precluding histological investigations that could have helped to verify the pathogenetic hypothesis of AOF. Regardless of the lack of histological features, among the aforementioned mechanisms, we can easily rule out a pseudoaneurysm due to the absence of necroscopic characteristic signs. We are not able to exclude the infectious mechanism, but it is our belief that this is quite unlikely, since we were not able to detect any elements supporting an infectious process, after considering all the available data. Consequently, we suggest that the most feasible AOF pathogenesis could be the chronic pulsation of the peri-prosthetic aortic tissue against the oesophageal wall. The peculiarity of this case is the 16-year gap observed between the surgical intervention and the clinical manifestation of AOF. No specific symptoms were observed during this 16-year period, apart from a few non-specific ones (e.g. fatiguability, general malaise, nausea and vomiting). According to the scientific literature, there are no significant differences in pathogenetic mechanisms of AOF and its incidence between the open and the endovascular approach. Considering the uncertain time interval between surgery and the onset of the symptoms that rapidly lead to death, it is suggested that before proceeding with the autopsy, the medical examiner has to review the patient’s medical history carefully, in particular when encountering a case of sudden death. Additional information on previous vascular surgery could have a great impact on the autopsy approach, regardless of the elapsed time from the surgical operation. Any kind of stress to the vascular walls could lead to structural and anatomical consequences on the mediastinal organs, with both acute and chronic mechanisms of damage. It is concluded that previous open vascular surgery has to be considered as a potential cause of sudden death due to the possibility that clinical manifestations of AOF, in the absence of specific symptoms, may occur, even with a particularly long delay (up to 16 years, as in the present case).