Aortic Graft–Enteric Fistula

Presentation

Close approximation of the vascular prosthesis and the aorta is required for an SAEF to develop. The most commonly involved GI tract segment is the third or fourth portion of the duodenum. That segment of the duodenum is retroperitoneal, and it is also relatively fixed in position by the ligament of Treitz and by the scar tissue that forms after open aortic surgery. The communication between the GI tract and the graft can involve the graft–aorta suture line (anastomotic fistula, or truly an aortoenteric fistula [AEF]) or only the body or limb of a graft (prosthesis–enteric fistula or erosion [PEF]). Anastomotic fistulas can also be associated with a pseudoaneurysm. SAEFs commonly occur late (24 to 36 months or longer after initial graft implantation) because there is a mechanical component to the etiology. The graft, anastomotic line or pseudoaneurysm rubs against the fixed adjacent duodenum and this process requires time to develop.

The majority of patients present with evidence of GI bleeding (70%), either alone (24%) or in combination with symptoms and signs of infection (46%). The pattern of bleeding results from the type of communication with the GI tract. PEFs tend to manifest with episodic subacute or chronic low-grade bleeding because the source of bleeding is the eroded edge of the bowel wall. AEFs tend to manifest with acute bleeding, including the classically described herald bleeding episode, because the source of bleeding is the aortic lumen itself. However, many patients (27%) only manifest signs of infection without any bleeding episodes (or even guaiac positive stool), and occasionally an AEF or PEF is entirely occult, manifesting as graft or graft limb thrombosis.

The symptoms and signs of infection are varied, but the most common are fever (67%) and leukocytosis (61%). Both abnormalities are often mild (mean T _max = 38.2 ± 0.9, and mean white cell count = 13.7 ± 5.2 10 ⁹ cells/L). When an AEF or PEF is present, bacteremia and sepsis occur more often than they occur in the setting of prosthetic graft infection without involvement of the GI tract, but the overall incidence is still low (<20%). Because many involved aortic grafts are entirely intraabdominal, most patients (75%) have no superficial changes to suggest an underlying graft problem. Occasionally infection tracks along the limb of an aortofemoral graft and manifests as a groin wound infection. Bleeding causes hypotension more often (24%). Infection alone rarely produces hypotension (sepsis) (9%). Surprisingly, septic embolization occurs relatively often (27%). Malaise affects a little more than one third of patients (36%).

Preoperative Evaluation

Any patient with GI bleeding in the presence of a prosthetic aortic graft should be assumed to have an AEF or PEF until proved otherwise. The most common preoperative assessment consists of esophagogastroduodenoscopy (EGD) and computed tomography angiography (CTA). EGD should be performed promptly in any patient with a suspected SAEF because it is the only direct means of diagnosing an SAEF. EGD should be performed by an experienced endoscopist and should always include the third and fourth portions of the duodenum, the most likely location of the fistula (73%). It may be necessary to use a pediatric colonoscope to complete the examination to this level. Ideally, the surgeon is also in attendance to observe the findings. The presence of another source of GI bleeding, such as gastritis, should not result in termination of the study before the entire duodenum is thoroughly examined. Despite a careful approach by experienced endoscopists, the false-negative rate for EGD can be high (47% in the author’s institution’s earlier experience). Although EGD is abnormal more than half of the time, it is only rarely diagnostic of SAEF (visible graft material, 12%).

CTA is performed to establish the diagnosis and to provide the information about the arterial anatomy needed to plan the revascularization portion of the treatment. Therefore the CTA should image all relevant anatomic regions. This can mean imaging the chest to provide information about the subclavian and axillary arteries and the thoracic aorta if any of those sites may be used as the source of inflow to the legs, as well as imaging the infrainguinal arteries to provide information about the runoff, particularly if the patient has known peripheral occlusive disease. It is obvious that the CTA should at least image the entire abdomen and pelvis (including the femoral arteries), particularly to define the anatomy of the renal and visceral arteries and determine if aortic branch reconstruction will be necessary.

Although it is actually uncommon for the CTA to be diagnostic of an SAEF (33.3% of the time), it is almost always abnormal (91.7%). The only finding that strongly suggests an SAEF is perigraft air present more than 2 to 3 months after aortic reconstruction. More commonly, the CTA shows changes consistent with graft infection, but not specific for SAEF: perigraft fluid, an increase in perigraft soft tissue, proximal anastomotic false aneurysm formation, loss of the continuous aneurysm wrap around the graft (where relevant), and thickening of the adjacent bowel wall.

The evaluation of the patient thought to have an SAEF must be rapidly completed. If the patient is actively bleeding, it may only be possible to perform one test while resuscitating the patient and then proceed directly to the operating room. If only one test can be performed, a CTA is preferred. If the patient has had a herald bleeding episode or has only had subacute or chronic blood loss, there is probably time to complete a more thorough evaluation. Because one can never be sure when life-threatening bleeding might begin, it is important to proceed expeditiously through the diagnostic evaluation and preoperative assessment.

A definitive diagnosis of an SAEF is difficult. For the most part, the surgeon decides to proceed with operative treatment based on findings that strongly suggest, but are not diagnostic for, an erosion or fistula. In the author’s series, the diagnosis was definitively established preoperatively in only one third of patients, but the presence of an SAEF and infection was strongly suspected in an additional half of the patients. It is an error to delay treatment while performing multiple additional diagnostic tests that are likely to be futile at securing a diagnosis. The surgeon must have enough experience and confidence to proceed based on suggestive data alone.

Once the diagnosis has been established or suspected, prompt operative treatment should follow. The patient should receive broad-spectrum antibiotics preoperatively. Although Staphylococcus epidermidis is the most common pathogen in prosthetic vascular graft infections overall, it is very rare for it to be a solitary infecting organism in the setting of an SAEF (3.3%). In fact, infection by multiple organisms, usually including enteric and gram-negative species (75%), is the rule. Therefore, antibiotic coverage should remain broad until the intraoperative culture results are available.

Management