Unusual cause of gastrointestinal bleeding in an 84-year-old woman: a miraculous survival from an aortoduodenal fistula repair

Abstract

An 84-year-old woman presented acutely with dizziness, fatigue and a total of 800 mL of fresh per rectum (PR) bleeding. The significant history of abdominal aortic aneurysm repair 5 years ago included multiple episodes of endovascular leak around the stent associated with abscess of left psoas major, left abdominal wall abscess with sinus formation, appendicitis with abscess formation, and acute pancreatic and chronic cholecystitis with multiple gallstones in the 7 months prior to this presentation. During the preceding 7 months, the patient was stabilised with an intravenous proton pump inhibitor, blood transfusions and Intensive Care Unit (ICU) management for the assumed diagnosis of stress ulcers over multiple hospital admissions. Imaging with CT scan of the abdomen made the more accurate diagnosis of acute gastrointestinal haemorrhage caused by a fistula between the distal duodenum and aorta, which was later surgically confirmed. Removal of infected stents and axillobifemoral bypass were performed with a successful recovery.

Background

Aortoduodenal fistula (ADF) is an extremely rare and fatal complication of aortic graft surgery. It is usually a postmortem diagnosis due to the massive upper gastrointestinal (GI) bleeding that presents after sudden and significant haemodynamic compromise. This case had an outstanding risk factor; a clear history of abdominal aortic aneurysm (AAA) repair without regular post-stent follow-up for 5 years due to patient non-adherence. However, there was a confounding factor in this case; multiple hospital admissions for treatment of significant comorbidities include an endovascular leak, pointing to the more relevant clinical diagnosis of acute GI haemorrhage. The exact cause of the GI bleed was not identified until an endoscopy and an abdominal CT scan were performed, allowing a potentially fatal and out-of-control GI exsanguination to be avoided. Consequently, a semielective surgery including removal of infected stents and axillary-bilateral femoral artery bypass was performed with a positive outcome of full recovery. A comprehensive literature review showed that there were a few case reports demonstrating the diagnostic and surgical salient learning points, but none of them demonstrating the unique diagnostic features of stress ulcers as the confounding factor and the innovative axillary-bilateral femoral artery bypass in this case. This case illustrates the importance of regular follow-up after aortic graft surgery to foresee treatable and fatal complications. The salient point of this case is that any patient with a history of aortic surgery or a stent graft who develops an upper GI haemorrhage should be considered to have an aortoenteric fistula (AEF) until proven otherwise.

Case presentation

Presenting report

An 84-year-old female patient presented with persistent right upper abdominal pain. There was no nausea, vomiting, fever or jaundice.

Medical history

The patient had hypertension, which had been well controlled for the past 10 years with nifedipine 30 mg once a day and telmisartan 80 mg once a day. She also had a history of coronary artery disease with a percutaneous coronary intervention in 2014. Previously, she also had untreated biliary and renal calculi for years.

Seven years ago, the patient’s AAA was discovered incidentally on abdominal CT scan when she presented with abdominal pain due to biliary calculi. A 2014 abdominal CT scan showed the AAA had increased in diameter and was subsequently repaired with endovascular stent. The patient was supposed to present for regular postoperative check-up yearly over the next 5 years but this did not eventuate, as she did not have health insurance that would cover the cost of check-ups and imaging.

During the 7 months prior to this presentation, the patient repeatedly presented with episodes of abdominal pain. Imaging (mainly abdominal CT scan) found a left psoas abscess, left abdominal wall abscess with sinus formation, chronic cholecystitis with multiple gallstones, choledocholithiasis, acute exacerbation of chronic appendicitis and acute pancreatitis on different occasions and with overlapping time frames.

These were managed with various interventions including a psoas abscessectomy and drainage, debridement and suture of the abdominal wall and endoscopic retrograde cholangiopancreatography. At one point, the patient spent 3 weeks intubated in the Intensive Care Unit (ICU), with intravenous esomeprazole and later oral esomeprazole as ulcer prophylaxis. During this time, a type III endoleak was also discovered on abdominal CT scan at the AAA repair site, and was managed conservatively because it did not cause any clinical issues.

Laboratory findings

The white cell count was 12.1×10⁹/L.

Imaging

An abdominal CT scan confirmed cholecystitis with multiple gallstones, and exudation from the wall of a previously repaired AAA, as well as visualisation of the known endoleak between the AAA wall and the endovascular stent.

Subsequent management

The patient was transferred to hepatobiliary surgery, where she was given spasmolysis, analgesia, antibiotic prophylaxis and antacids. The abdominal pain was gradually relieved but returned in the lower abdomen 2 days later. An emergent abdominal CT scan confirmed the endoleak again but there was no acute rupture. No endoscopy was performed as it was deemed too risky in light of the patient’s GI bleeding risk. A multidisciplinary meeting was held with the consensus management plan of endoleak repair by changing the endovascular stent.

One day before the planned surgery, the patient suddenly experienced increased abdominal distention and discomfort, passed frank blood PR (per rectum) three times (total volume estimated at 800 mL) and reported feeling dizzy and fatigued. Her extremities were cold and damp, and she appeared in hypovolaemic shock on clinical assessment. The immediate diagnosis was acute GI haemorrhage most likely caused by stress ulcers in view of her recent months of hospital admission with multiple surgeries. However, the clinicians also suspected a fistula between the small intestine and the aortic endoleak.

The patient was immediately given both blood transfusion and plasma transfusion. She remained fasting and was also given octreotide to slow the haemorrhage. A vascular screen was performed to rule out the possibility of a vascular lesion, such as an arteriovenous malformation. The patient was diagnosed with an infected AAA stent, complicated with endoleak and intestinal fistula.

Differential diagnosis

Two of the major differential diagnoses for this case were stress ulcers and gastric malignancy. Gastric malignancy is another condition which can cause abdominal pain and upper GI bleeding.¹ The possibility of ulcer-related malignancy should be considered in a patient who presents with symptoms of a GI bleed at this age. However, it is also likely to cause several other symptoms, such as chronic weight loss, early satiety and dysphagia,¹ none of which the patient reported. Furthermore, an abdominal CT showed no abnormal gastric thickening, which made it a less likely diagnosis for our case.

‘Stress ulcers’ are ulcerations of the GI tract which occur as a complication of hospitalisations, particularly ICU admission.² Since the patient had experienced multiple abdominal and GI reports over the preceding 6 months and had also been intubated in the ICU for a 3-week period, stress ulcers initially seemed to be the most likely diagnosis. However, the presence of the AAA endoleak on several of the abdominal CT scans including multiplanar reconstruction images (figure 1) raised the possibility of an ADF with CT scan showing duodenal adhesion to the AAA endoleak (figure 2). As the sequelae of an ADF can be life-threatening, surgical exploration was decided as the immediate course of action.

Figure 1.

Multiplanar reconstruction CT scan images which show the endoleak, arrows illustrating the leakage site.

Figure 2.

CT slice which shows duodenum adhesion to the aneurysm wall, arrow illustrating the possible site of aortoduodenal fistula.

Treatment

A stent removal and axillary-bilateral femoral artery bypass were planned immediately. Due to the patient’s age, there was concern about the increased cardiac workload which would be caused by clamping of the abdominal aorta. With sufficient antibiotic prophylaxis (intravenous tazocin) and intraoperative inotropes support, an axillary-bilateral femoral artery bypass with polytetrafluoroethylene artificial vessel was performed. The infected stents and infected aneurysmal tissue were removed.

During the operation, the aorta was clamped proximally below the renal arteries and the common iliac arteries were clamped distally. Then the aortic aneurysm was exposed and cut through, revealing a large amount of infected thrombi. The infected aneurysm, thrombus and the stents were removed with exposure of aneurysm sac (figure 3). The aneurysm had strong adherence to the descending segment of duodenum proximally (figure 4), and the right iliac artery and vein were heavily adhered with the appendix.

Figure 3.

After removal of the infected stents, arrow illustrating the aneurysm sac.

Figure 4.

Intraoperative picture, arrow illustrating the duodenum and appendix adhesion to the aneurysm wall.

Blood lost was significant from both the right iliac artery and the right iliac vein, hence large blood transfusion and fluid support were provided with compression and sutures were applied to stop the bleeding. A fistula (figure 5) between the aneurysm and the descending segment of the duodenum was confirmed and the adhesion between the duodenum and the aneurysm was separated (figure 5) in conjunction with the fistula repair. Finally, the surgical field was repeatedly washed with hydrogen peroxide and copious normal saline. A jejunal nutrition tube was placed with two drainage tubes in the retroperitoneal and pelvic cavities.

Figure 5.

After intraoperative separation of duodenum, appendix and aneurysm wall, blue arrow illustrating the fistula orifice of duodenum, green arrow illustrating clamped aneurysm wall.

The total combined loss of blood and fluid was 28 000 mL. Urine volume throughout the operation was 1000 mL. Eighty units of packed Red Blood Cells (RBCs) were infused, along with 6550 mL of plasma, 27.5 units of cryoprecipitate and 1 unit of platelets.

Outcome and follow-up

The patient was sent to the ICU after surgery and was successfully extubated 3 days later with discharge out of ICU 5 days after surgery.

Enteral nutrition began on day 9 postoperatively. Throughout the postoperative period, the patient was given antibiotic prophylaxis, pulmonary and cardiac rehabilitation therapy. The drainage tube and nutrition tube were gradually removed in the next 2 weeks.

The patient’s condition gradually improved and she commenced oral feeding 1 week before discharge. She made a full recovery and was discharged from hospital 7 weeks after surgery. She was followed up again at the vascular surgical outpatient department on a regular basis.

Discussion

ADFs are the most frequent type of AEFs, making up 80%.³ This particular ADF was classed as secondary, as it was caused by the endovascular repair of an AAA.³ The pathogenesis is thought to involve Staphylococcus epidermidis biofilm infection on grafts, insufficient retroperitoneal tissue or inadequate materials used for prosthetics.³ This leads to inflammation and destruction of the aortic wall.⁴ Over time, the increased pressure exerted on the intestinal tract by the aneurysm sac leads to fistula formation.⁵

There are three symptoms which make up the classic AEF presentation: upper GI bleeding, a pulsatile abdominal mass and abdominal pain.⁶ This particular constellation is known as Cooper’s triad.⁶ Our patient bled from the upper GI tract and experienced abdominal pain, but had no pulsatile mass. However, it should be noted that these three symptoms only occur together in 11% of patients with AEF.⁶ It is of vital importance not to over-rely on Cooper’s triad for a diagnosis,⁶ as time is critical for diagnosis⁷ and the exsanguination caused by an untreated ADF is always fatal.⁸

ADF is diagnosed through focal inflammation and thickening of the GI tract wall, periaortic haematoma, gas shadow being found around the graft and intravenous contrast in the lumen of the GI tract.⁶ An abdominal CT and endoscopy make up the usual imaging workup for diagnosing ADF.⁹ CT is considered a good diagnostic modality for patients with ADF, as its sensitivity and specificity are 90%.¹⁰ However, an endoscopy has less diagnostic value than a CT and may not be tolerated by a patient who is haemodynamically unstable, like our patient.¹¹ Time is crucial in diagnosing a suspected ADF,¹¹ as delay in diagnosis has been linked to adverse outcomes.⁷

Diagnosing ADF can be very difficult when the presentation is not typical or is obscured by confounding factors, as it was in this case.¹² A strong index of clinical suspicion and exploration in theatre are key to starting effective management.¹²

Although our patient experienced GI bleeding, the blood loss was initially 800 mL, which while significant is not necessarily life-threatening. This brings to mind a phenomenon associated with AEFs, known as the ‘herald bleed’. A ‘herald bleed’ is an initial episode of blood loss from the AEF which is often temporarily sealed by a thrombus until the high blood pressure of the aorta dislodges it and causes massive exsanguination.⁶ It is possible that our patient experienced a herald bleed, which possibly bought us some time to surgically explore and manage the ADF prior to massive blood loss.

The total amount of blood and fluid lost was 28 000 mL, which is a significant amount. This was most likely due to systemic inflammatory response syndrome (SIRS), caused by the multiple instances of intra-abdominal inflammation leading up to the herald bleed, and the subsequent tissue injury.

Tissue injury has the potential to release antigens and proinflammatory mediators, which are collectively known as alarmins.¹³ The presence of alarmins and fragments of injured tissue in the bloodstream has a procoagulant effect, causing platelets to agglutinate and form thromboses in small vessels.¹⁴ It also activates the fibrinolytic enzyme system.¹⁴ This is known as disseminated intravascular coagulation, and leads to significant platelet depletion and coagulopathy.¹⁴ This in turn leads to massive rapid volume depletion.

Patients who lose a massive amount of fluid also often experience hypothermia due to the significant volume depletion, which impairs the activity of platelets and clotting factors.¹⁵ This temperature-induced coagulopathy further perpetuates blood loss, leading to a ‘vicious circle’ of continuing hypothermia, coagulopathy and volume depletion.¹⁵ This is the probable reason why there was such a significant intraoperative blood loss.

A considerable amount of literature has been published about AEFs secondary to AAA repairs. Our patient displayed the same symptoms as many other patients with AEF; a herald bleed followed by significant blood loss and resulting shock. In a pooled data analysis of 823 patients who had experienced secondary AEF, 72% of them presented with a GI haemorrhage and 33% of them had signs of hypovolaemic shock. Forty per cent also had sepsis due to stent infection.⁸

Looking at the literature also yielded other statistics. Two other studies reported a median intraoperative fluid loss of 2016 mL¹⁶ and 2719 g,¹⁷ respectively. Our patient lost considerably more (28 000 mL). In-hospital mortality rates for secondary AEF are quite high, with a study of 59 patients reporting 10.2% in-hospital mortality¹⁸ and the 823 patients’ data analysis reporting 30.7%.⁸ The data analysis also reported increased risk of patient mortality with increased patient age.⁸

Our patient had a type III endoleak, which occurs when an endovascular repair (EVAR) graft fails; either due to disjunction of components or a defect in the graft fabric.¹⁹ Many factors contribute to type III endoleak formation, including aneurysm anatomy, postoperative dilatation and remodelling of aneurysm tissue and the degree of stent graft migration.²⁰ This results in direct arterial flow into the perigraft space, putting the aneurysm sac under high pressure.²¹ In our patient’s case, the resulting expanded aneurysm sac and turbulent flow may have further fuelled her coagulopathy.²²

Our patient also experienced infection, most likely S. epidermidis, at the EVAR site. Aneurysm or graft infection can be caused by spread from a contiguous site or haematogenous spread.²³ Staphylococcus and Salmonella spp are the most common causes of infected aneurysms, as they have high affinity for the arterial wall.²⁴ The infection spreads along the vascular wall in the direction of blood flow, weakening intimal and muscular tissue as it does so.²⁵ This increases the risk of organ penetration and fistula formation, as occurred in this case.²⁵

The major differences which set our patient with secondary AEF apart from many others are her multiple episodes of abdominal inflammation and 3-week stay in the ICU prior to the case’s presenting report. The intra-abdominal inflammation likely contributed to SIRS and the excess fluid loss experienced in this case, while her prior intubation in the ICU made us initially think that she had stress ulcers.

Stress ulcers are a common complication of ICU admission and are thought to occur due to the hypoperfusion and compromised microcirculation caused by cardiogenic/hypovolaemic shock, as well as the patient’s severely unwell state promoting widespread GI inflammation.² The risk of stress ulcers is increased if the patient has coagulopathies, compromised renal or hepatic function, or is ill enough to require mechanical ventilation.²⁵ Proton pump inhibitors are often given to ICU patients on admission as ulcer prophylaxis, as occurred with our patient.²⁶

As our patient had a confounding factor (the ICU stay) which made the ADF diagnosis more difficult and was in a proinflammatory state that contributed to massive fluid loss, it is remarkable that she made a full recovery, considering her age and the mortality rate for secondary AEF. We consider ourselves fortunate that the case had a positive outcome.

Currently, there are no clear guidelines specifically for the management of ADF, as the formation of a fistula between the aorta and the duodenum is quite rare.³ The most important goals for AEF surgical management are to prevent fatal haemorrhage, prevent infection and reconstruct the GI tract.¹⁰ It was for this reason that our patient received blood products and antibiotic prophylaxis at various points in the management.

Patient’s perspective.

When she initially got her tummy pain on the right side, a friend told her she might have gallstones. Her CT scan confirmed her gallbladder infection with a few gallstones. She was told by the doctor that a CT had also found a leak between her ballooned artery and the repaired plastic tube in the artery. She was treated mainly with antibiotics and other stuff, and the pain gradually got better. She did not realise how serious this leak was until she had lower tummy pain and lots of blood in her poo. Another CT scan found her repaired balloon artery being connected to her intestine. If the doctor did not stop the connection, she could suddenly die anytime. She was told that a special and new kind of surgery to rebuild her artery would be required to save her life. Even then, she was told before the operation there would be only a slim chance of her survival through this operation. It was through the grace of God that she pulled through the special operation after losing so much fluid. She was extremely grateful to the surgeon and her team of doctors. She is recovering very well and is almost completely back to normal life and now she always takes the advice of the doctors to have regular check-ups. Finally, she must say that the doctors have given her a second life, which she’ll cherish. She hopes they can use her case story to teach and train future doctors to match their competency.

Learning points.

• Patients who receive aortic graft surgery should always receive regular follow-up to foresee treatable and fatal complications.

• Any patient with a history of aortic surgery or a stent graft who develops a gastrointestinal haemorrhage should be suspected of having an aortoenteric fistula (AEF) until proven otherwise.

• Stress ulcers (due to history of pancreatitis, appendiceal abscess and psoas abscess) can be a confounding factor in AEF detection and diagnosis.

Ethics statements

Patient consent for publication

Next of kin consent obtained.