A H :

Hansol Kim
Gastrointestinal hemorrhage is a common cause of admission to the hospital with acute
upper GI hemorrhage accounting for >500,000 admissions each year. There are many
causes of GI hemorrhage and to better define, diagnose, and treat each disease processes,
it is categorized into two categories: an acute upper and lower GI hemorrhage. This
distinction lies based on whether the lesion lies prior to or post a suspensory muscular
band also known as ligament of Treitz which separates the duodenum from jejunum.
There are many causes of acute upper and lower GI bleeding. The three most common
causes of acute upper and lower GI bleeding include peptic ulcer disease, gastritis or
duodenitis, and esophageal varices, and upper GI bleed, angiodysplasia or AV
malformation, and diverticular disease. However, other causes are also identified:
gastric carcinoma, Mallory-Weiss tear, Dieulafoy lesions, hemorrhoids, ischemic bowel
disease, inflammatory bowel disease, colon cancer, intussusceptions, Meckel's
diverticulum, and aortoenteric fistulas.
Despite many efforts to prevent GI bleeding by treating H. pylori infections and by
implementing stress ulcer prophylaxis in post-operative and critically ill patients, the
common and everyday use of nonsteroidal anti-inflammatory agents (NSAIDS),
especially in the elderly has maintained the risk of developing both upper and lower GI
bleeding. In addition, it has also been shown certain calcium channel blockers such as
dilatiazem and verapamil as well as serotonin-reuptake inhibitors are associated with
increased incidence of GI hemorrhage.
Based on the location of GI bleed, the presentation varies. In acute upper GI
hemorrhage, hematemesis or melena may be the presenting symptom. Hematemesis is
defined as vomiting of bright red blood or with coffee ground/black blood. Melena is
defined as black or tarry stools, once described to occur with 50-80mL of blood in the
GI tract. If the bleeding is heavy and acute, patient may also present with hematochezia
or bright red blood per rectum. Acute lower GI bleeding manifests with melena or
hematochezia, which is thought to occur with profuse bleeding of up to 1,000mL of
blood in a short period of time.
Often times, history of present illness and past medical and surgical history play a vital
role in diagnosing a patient with GI bleed. In patients with history of NSAID use, peptic
ulcer disease may be suspected. In patients with history of abdominal aortic aneurysm
status post endovascular stent placement or bypass graft, the question of aortoenteric
fistula must be entertained. In order to distinguish between upper and lower GI bleed, a
nasogastric tube (NGT) may be placed. A negative trace of blood in NGT aspirate,
however, does not rule out an upper GI bleeding. The best diagnostic test and treatment
for GI hemorrhage can vary based on the cause of bleeding. However, the initial test of
sigmoidoscopy/colonoscopy/anoscopy which allows direct visualization of the
gastrointestinal tract. Other imaging tests include capsule endoscopy, double balloon
enteroscopy, small bowel imaging such as small bowel follow through, radionuclide
scanning such as 99m-Technicium scan, or angiography in cases of active rapid
There are limitations and contraindications of each study. Colonoscopy or EGD may be
useful if bleeding is <1mL/minute as heavy bleeding may not allow visualization and
there is a risk of perforating the bowel. In cases of radionuclide scans, it is often
nonspecific and cannot be used to determine the cause although it can be useful in rapid
bleeding. Angiography is also useful in rapid bleeders >0.5mL/min, however it requires
the use of IV contrast and cannot determine the cause of bleeding most of the time.
However, it may be advantageous as it has the potential to treat active bleeding with
embolizing materials such as gelfoam and vasoconstrictors. Of course, this carries an
inherent risk of infarcting the bowel.
Aortoenteric fistula is a rare cause of abdominal hemorrhage which carries 100% risk of
mortality left untreated. It is defined as a condition in which there is a fistulous
communication between the GI tract and the aorta. This condition is further broken
down into two categories: primary and secondary aortoenteric fistula (AEF) based on
history of abdominal aortic aneurysm (AAA) repair and presence of prosthetic materials.
Primary AEFs arise in patients without history of AAA repair and was first described by
sir Ashley Cooper in 1817. Overall, there are only 250 reported cases which is
equivalent to estimated incidence of 0.007/million/year5. Secondary AEFs are more
common in comparison to primary AEFs, first reported by R.C Brock in 1953 and first
successfully treated by G. Herberer in 1957. Secondary AEFs may arise at any time, as
little as 11 days to as many as 10 years post AAA repair with incidence of 0.6-2%5.
Despite the advances in technology, AEF continues to be difficult to diagnose and treat
and is associated with high mortality (65-100%) and morbidity.5 With increasing trend
of AAA repairs, it is imperative to have a high suspicion of AEF in patients presenting
with acute GI bleeding.
Although pathogenesis of AEF is unclear, it is thought to arise secondary to chronic low
grade infection with subsequent abscess formation and erosion through the graft and
aortic pulsations that transmit repetitive pressure on the intestine. Accordingly, signs of
aortic infection are present at surgery or autopsy5 and when cultured, it commonly
grows Escherichia coli, E. faecalis, Salmonella, Mycobacterium tuberculosis,
Clostridium septicum, Lactobacillus and Klebsiella species.11 In one study of 22
patients, 72% of patients were proven to have infection of the graft3. Along with
aneurysm and infection, radiotherapy, tumor, atherosclerotic disease, collagen vascular
disease, and foreign body ingestion have been shown to cause primary AEFs as well.11
The most common site of AEF is in the duodenum with 60% of cases arising in the 3rd
and 4th section of duodenum due to close proximity and direct contact with the
abdominal aorta. The other 40% can be found in other loops of small and large bowel
such as jejunum (12%), ileum (18%), cecum (8%), and appendix (4%)3. In addition to
AAA repair, there are other causes of secondary AEFs such as cases of aortogastric
fistula following nissen fundoplication procedure. There may be other fistulous
connections between arteries such as the iliac arteries and GI tract, or cases in which
multiple fistulae have formed in a single patient and therefore necessitates a careful
review to rule out other sites of fistulous communication in all patients.
Table 1. Presentation of primary AEF over 50 years (Saers and Scheltinga)
A patient with AEF may present with typical signs and symptoms of GI hemorrhage
with a classic triad of acute or chronic GI blood loss, pulsating abdominal mass, and
abdominal pain. Classic triad is a rather rare finding with one study showing such signs
in only 11% of all the patient population. In some patients, however, it may present
more uniquely with what is described as a herald hemorrhage in which there is an initial
brisk bleeding along with hematemesis and hypotension that spontaneously resolves for
hours to days before re-bleeding. According to Champion et al., the duration between
initial bleed to next may last from six hours to eight months with a mean of 25 days. In
70% of patients, this interval is more than 6 hours, in 50% over 24 hours and 29% more
than 1 week11. The herald bleeding is thought to occur either secondary to vasospasm
and thrombus formation or from mucosal ulceration and focal necrosis instead of from
the AEF itself. If the first thought is true, excessive hydration must be avoided as the
thrombus plug may be dislodged leading to overt bleeding and exsanguination. Other
presenting features included signs of sepsis, hematemesis or melena, and hemodynamic
Diagnosing a patient with AEF is a rather difficult task as many studies have a high
false negative rate. In 1980s, according to Champion et al., an endoscopic evaluation of
the GI tract proved superior to barium contrast studies. The signs suggestive of AEF
may include visualization of the bleeding site, adherent blood clot, ulcer combined with
a blood filled stomach, erosion with pulsating mass protruding through the duodenum.11
However, the brisk bleeding of the AEF may limit its use. In such cases, aortography
was recommended in which AEF would be suggestive with extravasation of contrast
agent into the bowel during active bleeding (Figure 1 and 2). However, only 25% of
AEF patients had a positive study11 and there is a risk of embolizing and exsanguinating
a patient as the catheter may be in close proximity to an unstable plug of thrombus.
Many studies such as these are retrospective chart reviews of single or few institutions
and may not best represent the population as a whole. In fact, the optimal investigative
study for suspected AEF has not yet been established and adds to the difficulty of
diagnosing AEF patients.
In 2006, Hughes et al., performed a retrospective study evaluating the sensitivity and
specificity of CT in diagnosing AEF. According to the study, CT had an overall
specificity of 100% and sensitivity of 50%, which was not comparable to similar studies
performed in 1980s-2000s which had shown sensitivities ranging 40-100% and
specificity of 33-100%. Compared to other studies such as upper GI endoscopy
(sensitivity of 20%), angiogram (sensitivity 33%) and nuclear studies such as
scintigraphy with SPECT using radiotracers gallium 67 citrate, 111In labeled white blood
cells or 99mTc hexametazime (sensitivity 0%), CT was by far the best test of choice.
MRI has similar sensitivity and specificity for diagnosing AEF but the urgency of the
nature and technical expertise that is required favors the use of CT. The most specific
sign on CT was ectopic gas (specificity 100%; sensitivity 40%; figure 4) and most
sensitive signs were peri-aortic fluid or soft tissue >5mm (sensitivity 90%, specificity
92%; figure 4), breach in aortic wall (sensitivity 89% specificity 75%; figure 5), and
loss of fat pad between bowel and aorta (sensitivity 90% specificity 33%; figure 6)
Although nuclear studies such as scintigram, SPECT, or PET scans are not widely used,
it can supplement CT findings especially when radiotracer is found in the bowel lumen
on scintigrams (Figure 8a). Because many other disease processes such as infectious
aortitis, infected aortic aneurysm, retroperitoneal fibrosis, and perigraft infection can
mimic AEF, it is important to rule out other disease processes.
As difficult AEF is to diagnose given the poor sensitivities of most diagnostic studies,
treating the disease is not an easy task. There are both surgical and interventional
procedures available such as simple surgical closure, placement of in situ graft,
placement of antibiotic impregnated in situ graft, extra-anatomic reconstruction,
endovascular stent grafting, and coiling but there is not enough data to conclude
superiority of one technique over another.
Since 1951, there has been an increasing trend to surgically treat patients in primary
AEF with decreasing mortality rate11(Table 5). According to the review by Saers and
Schlettinga, surgical closure of defect carried the highest mortality rate of 75% and
antibiotic impregnated in situ graft and embolic coiling was associated with lowest
mortality rate of 0% (Table 6).
However, long term results of these patients were not available. In comparison to
surgical in situ graft with omental pedicle covering the aorta and GI tract, endovascular
stent graft was associated with lower mortality rate (36% vs. 14%). However based on
recent study by Kakkos et al., endovascular reconstruction is only beneficial in the first
two years and is associated with similar morbidity and mortality rates over long term
period (Figure 9, Table 7).
An axillobifemoral bypass is a form of extra-anatomic reconstruction (EAR) and
remains as the gold standard in vascular surgery. However, there are conflicting reviews
of the best optimal treatment (EAR versus in situ reconstruction (ISR)) with some
suggesting EAR as the primary treatment and others suggesting EAR to be reserved
only for patients with severe peritonitis or with extensive local sepsis as it is inferior to
other methods with mortality rate of 40%11 (Table 6). According to one study, EAR is
not superior to ISR2. Given EAR and ISR are lengthy surgical procedures, both carry
risks of post-operative infection, wound dehiscence, anastomotic rupture, pulmonary
failure and acute renal failure among others. Other long term risks include graft
thromboses, femoral amputation in case of axillobifemoral bypass, graft stenosis, reinfection, and reformation of AEF. According to Batt et al., the rates of such
complications were similar between EAR and ISR.
Many of AEFs arise secondary to chronic low grade infection. Therefore, in patients
with primary AEFs status post treatment, antibiotic therapy is recommended for >1
week in culture negative patients, >4-6 weeks if cultures are positive and even longer
period in patients who had undergone endovascular operation11. Literature suggests
antibiotic therapy in patients with secondary AEFs, however, there is no consensus on
the length of treatment. Even though the re-infection rate has been similar between ISR
and EAR, use of autogenous veins or cryopreserved allografts in ISR has been shown to
be beneficial in prevention of re-infection in long term2.
The ambivalence in the scientific community regarding the best optimal treatment for
AEF creates confusion over mode of therapy. With reports of higher incidence of reinfection and recurrent bleeding post endovascular stent grafting, it raises the question
of when endovascular approach should be used. Endovascular stent grafting have many
advantages over open surgical procedure given that it is less invasive, less time
intensive, is associated with less post operative length of stay, and there is minimal risk
of reperfusion injury given there is no clamping of the aorta. Endovascular stent
grafting is primarily used in the elderly population found too risky to undergo an open
surgical procedure. Given endovascular repair is associated with lower morbidity and
mortality in the first two years post procedure, at which time such benefits are lost, it
may serve as a life saving procedure in the elderly and as a bridging therapy in younger
patients as more permanent open surgical repair can be performed when patient is at a
better health status.
Majority of AEFs arise secondary to chronic infection of prosthetic material placed
while treating aortic aneurysms. Given the difficulties of diagnosis and treatment of
AEFs with associated high morbidity and mortality, prevention becomes a key in
reducing the incidence of AEFs. The question of whether aortic aneurysms should be
treated with endovascular stent grafting or with open surgical techniques becomes vital.
Both endovascular and open surgical techniques are known to cause AEFs, however, to
the best of my knowledge, there is no comparative study evaluating the relative risks of
developing AEF following endovascular and open surgical treatment. A higher risk of
AEF may be expected in patients treated with open surgical therapy compared to
endovascular stent grafting due to higher risk of infection. If this is true, we would
expect to hopefully see a decreasing trend in incidence of AEFs as endovascular
treatment of aortic aneurysms are becoming the preferred mode of therapy over other
techniques. Other future areas of research may include the use of antibiotic impregnated
prosthetic graft in treatment of aortic aneurysms as a preventative measure for AEFs as
well as defining and assessing the role of endovascular repair as a bridging therapy.
1. Antoniou GA, Koutsias S, Antoniou SA, Georgiakis A, Lazarides MK, Giannoukas
AD. Outcome after endovascular stent graft repair of aortoenteric fistula: a systematic
review. J Vasc Surg 2009; 49(3): 782-9.
2. Batt M, et al., Early and Late Results of Contemporary Management of 37 Secondary
Aortoenteric Fistulae. Eur J Vasc Endovasc Surg (2011), doi:10.1016/j.ejvs.2011.02.020
3. Champion MC, Sullivan SN, Coles JC, Goldbach M, Watson WC. Aortoenteric
fistula: incidence, presentation recognition, and management. Ann Surg 1982;
4. Chenu C, Marcheix B, Barcelo C, Rousseau H. Aorto-enteric fistula after
endovascular abdominal aortic aneurysm repair: case report and review. Eur J Vasc
Endovasc Surg 2009; 37: 401-406.
5. Hughes FM, Kavanagh D, Barry M, Owens A, MacErlaine DP, Malone DE.
Aortoenteric fistula: a diagnostic dilemma. Abdom Imaging 2007; 32:398-402.
6. Kakkos SK, Antoniadis PN, Klonaris CN, Papazoglou KO, Giannoukas AD,
Matsagkas MI, et al. Open or endovascular repair of aortoenteric fistulas? A multicentre
comparative study. Eur J Vasc Endovasc Surg 2011; 41:625-34.
7. Karkos CD, Viachou PA, Hayes PD, Fishwick G, Bolia A, Naylor AR. Temporary
endovascular control of a bleeding aortoenteric fistula by transcatheter coil embolization.
J Vasc Interv Radiol 2005; 16: 867-871.
8. Krupnick AS, Lombardi JV, Engels FH, Kreisel D, Zhuang H, Alavi A, et al. 18Fluorodeoxyglucose positron emission tomography as a novel imaging tool for the
diagnosis of aortoenteric fistula and aortic graft infection: a case report. Vascular and
Endovascular Surgery 2003; 37(5): 363-366.
9. Lawlor DK, DeRose G, Harris KA, Forbes TL. Primary aorto/illiac-enteric fistula:
report of 6 new cases. Vascular and Endovascular Surg 2004; 38(3):281-286.
10. Mark AS, McCarthy SM, Moss AA, Price D. Detection of abdominal aortic graft
infection: comparison of CT and In-labeled white blood cell scans. AJR 1985; 144: 315318.
11. Saers SJF, Scheltinga MRM. Primary aortoenteric fistula. British Journal of Surgery
2005; 92: 143-152.
12. Tse DML, Thompson ARA, Perkins J, Bratby MJ, Anthony S, Uberoi R.
Endovascular repair of a secondary aorto-appendiceal fistula. Cardiovasc Intervent
Radiol 2011.
13. Vu QDM, Menias CO, Bhalla S, Peterson C, Wang LL, Balfe DM. Aortoenteric
fistulas: CT features and potential mimics. Radiographics 2009; 29(1):197-209.