Introduction

Aortoenteric fistula remains one of the most difficult acute care and vascular surgery challenges. It is an entity that often presents emergently, allowing limited time to plan, and requires complex surgical skill from two different specialties (GI and vascular surgery). Furthermore, these patients carry multiple comorbidities both chronic (athlerosclerotic disease) and acute (malnutrition) and the perioperative period often necessitates advanced critical care support and input from multiple subspecialties (e.g., infectious disease).

Etiology

An aortoenteric fistula may form between any level of the GI track that is able to be directly apposed to the vessel (or graft that has been placed). It is most common within the abdomen. The third and fourth portions of the duodenum are particularly at risk as this segment of the GI track drapes over the aorta as is courses along the retroperitoneum. Aneurysmal dilatation of the aorta displaces the duodenum anteriorly while also exerting pressure along its posterior wall. The virgin duodenum is fixed in the retroperitoneum at this point between its course posterior to the SMA/SMV and the ligament of Treitz. Similarly, the duodenum remains in this proximity once an aortic procedure is performed; the raw tissue from the dissection guaranteeing proximity of these tissues via adhesion formation. More distally, postoperative adhesions may draw both small and large bowels into proximity of the graft (both tube and limb grafts).

Fistulas may form to aortic grafts from open replacements, endovascular-stented aneurysms as well as primary aortic aneurysms [1]. As will be seen later in this chapter, this has significant effect on one’s approach to the repair. The formation of the fistula may occur from several different inciting events. An infectious episode: For example, a diverticular abscess overlying the left iliac limb of an aorto-bi-iliac graft results in an aorto-colonic fistula. An iatrogenic event may occur such as catching bowel with a suture while closing the retroperitoneum following graft placement. Most frequently, in the case of aorto-duodenal fistula, simple tensioned apposition of the peristalsing duodenal segment over the foreign material of graft eventually leads to erosion of the back wall of the duodenum (Fig. 12.1).

Fig. 12.1
figure 1

a Anterior displacement of the duodenum by tortuous aortic graft placement resulting in aorto-duodenal fistula (b)

Full size image

The fistulas are usually all characterized structurally as being stuck to the aorta or graft with a full thickness violation of the bowel wall, the vessel, or graft essentially “patching” the defect. This pathologic structure is at the heart of the presentation of aortoenteric fistulas (Fig. 12.2).

Fig. 12.2
figure 2

a Exposed, stained aortic graft from an aorto-duodenal fistula (b) from a double fistula to both distal limbs of an aorto-bifemoral graft, the fistula from a segment of small bowel. c Fistula to exposed endostent

Full size image

Presentation

Aortoenteric fistulae essentially demonstrate two distinctive patterns of presentation. The first is defined by sudden onset of an episode of massive gastrointestinal bleeding. This is painless, often with bright red blood per rectum and initially self-limited. In the setting of primary aneurysmal fistula, it can occur because of connection at any point along the aneurysmal sac. In the setting of prior graft replacement, the fistula usually has formed at one of the suture lines between the graft and the native aorta; at times, the infection of the graft can spread along its course, “un-incorporating” the Dacron material but bleeding still will occur from suture line failure (we have not seen dissolution of the graft material itself). As with other bleeds associated with infectious etiology, the first bleed is often known as a “heralding bleed,” stopping from a combination of hypotension (reducing the pressure at the suture line) allowing acute coagulation. Aggressive resuscitation in the setting of diagnostic uncertainty combined with clot lysis (likely accelerated by active infection and exposure to the digestive ferocity of succus) leads to subsequent episodes of massive hemorrhage (less self-limited). Healthy suspicion, combined with rapid diagnosis, is vital if patients with this presentation are to be successfully treated.

Resuscitation of the patient who presents with gastrointestinal bleeding from an aortoenteric fistula is challenging. At times, massive ongoing bleeding results in severe hemodynamic instability and the patient requires full replenishment of intravascular volume to stabilize. Once the diagnosis is made and intervention is underway, a strategy of judicious hypotensive resuscitation can be employed. This decreases fluid administration and avoids the subsequent disastrous edema that results from excessive fluid administration. Additionally, passive hypotension theoretically reduces the pressure-based risk of disrupting a clot and decreases the rapidity and volume of ongoing hemorrhage. We utilize a strategy of balanced hemostatic resuscitation in our approach with low ratios of transfusion of units packed red blood cell to fresh frozen plasma and platelets. For these patients, we frequently enact our standard institutional massive transfusion protocol.

The second pattern of presentation is more insidious, and likely more common [2, 3]. Slow erosion of the bowel wall results in the continual bathing of a patch of graft with succus or stool. This continual exposure results in episodic bacteremia with poly-microbial exposure and no possibility for sterilization with antibiotic administration, unlike an uncomplicated mono-microbial graft infection (if there really is such a thing). There is often suspicion of a graft infection but not necessarily a proven etiology (imaging findings can be subtle). It is not uncommon for these patients to undergo months of several courses of escalating antibiotic therapy suffering chronic malaise and malnutrition.

Diagnosis

When the presentation is acute, expeditious diagnosis is vital. Confirmation of fistula then allows management planning (this can be a terminal event as not all patients are appropriate operative candidates). CT scanning and endoscopy are the primary modalities to confirm aortoenteric fistula [4, 5]. CT is least invasive and may be immediately diagnostic. Inflammation is generally seen at the fistula site with air, extraluminal to the GI track, as seen along the wall of the aorta or aortic graft [6] (Figs. 12.3 and 12.4). EGD or colonoscopy will often demonstrate the fistula, almost shockingly, by visualizing obvious graft material comprising a large portion of the bowel wall. This too is immediately diagnostic (Fig. 12.5).

Fig. 12.3
figure 3

CT of graft with massive amount of air around prosthetic graft (a); b coronal reconstruction; c fistulous connection with overlying sigmoid colon

Full size image
Fig. 12.4
figure 4

CT of aortic graft with subtle amount of periduodenal inflammation and a spicule of air in the graft bed (right lower frame)

Full size image
Fig. 12.5
figure 5

EGD view of posterior wall of fourth portion of the duodenum with visualized prosthetic from the aortic graft

Full size image

While the CT may suggest some inflammation along the course of the aorta, it is sometimes quite difficult to confirm a fistula (Fig. 12.6). The duodenum is almost always collapsed in its third and fourth portions. Radiographically this can hide wall thickening along the posterior aspect of the duodenum. Further, there is seldom an associated pocket of purulence as the connection to the bowel effectively drains internally and in thin patients, the paucity of fat frequently precludes identification of significant inflammation.

Fig. 12.6
figure 6

Subtle periduodenal inflammation in a patient with an endostent; confirmed aortoenteric fistula

Full size image

There is usually no pain associated with aortoenteric fistula formation (sometimes achy back discomfort) and few digestive symptoms. Chronic unwellness is nonspecific but continued elevation of inflammatory indicators, episodic illness, polymicrobial, and shifting bacteremias (especially enteric flora) should heighten suspicion and guide more aggressive diagnostic efforts.

Management

No aortoenteric fistula will heal spontaneously. Operative management is the only solution for correction of aortoenteric fistula. A decision to manage conservatively with ongoing antibiotic regimens is reasonable if the patient is unfit for operative repair, does not want to risk repair, or carries a prognosis from other comorbidities that govern life span (such as metastatic cancer). This is purely a palliative option; the natural history can be expected to be a septic or hemorrhagic death if due to the fistula.

The goals of operative therapy are to (1) assure arterial flow and adequate distal perfusion, (2) detach the connection between the GI track and the arterial system, (3) re-establish GI track integrity, (4) minimize risk of recurrence, and (5) definitively manage the systemic infection. These operations are long, technically challenging and high risk. We have chosen a multidisciplinary approach combining the efforts of both acute care and vascular surgeons with critical care anesthesia and where possible, perfusionists.

Whether the presentation is by bleeding or other, adequate access and availability of blood products are keys. Appropriate broad-spectrum antibiotics are given and redosed as per protocol. Excellent intraoperative communication is also vital as we have found that at times it is most appropriate to expect to have to truncate the repair (stage in a damage control manner) to optimize the patient’s physiologic reserve. The operative plan is also of paramount importance. The patients are positioned in the supine position and the operative field that is prepared is from the neck to the mid-thighs with the axillae and femoral regions exposed. Adequate venous access for resuscitation is assured.

Arterial Flow and Distal Perfusion

With respect to arterial reconstruction, there are two overarching approaches to the management of infected aortic grafts: extra-anatomic bypass versus in situ graft placement. The advantages of extra-anatomic bypass (axillofemoral plus femoral–femoral grafts) include minimal invasive approach which can be performed prior to graft excision, removing prosthetic from an infected graft bed, assured distal perfusion to allow time to dissect and if aortic clamping is needed during fistula takedown. Disadvantages include a plan to over-sew the aortic stump in the infected field, and thus to degenerate and rupture fatally, as well as a short patency of the small diameter extra-anatomic bypass circuits. Conversely, in situ reconstruction has a longer patency and preserves native flow but leaves graft in the soiled bed, may be a longer and more complex operation, and may require a longer duration of systemic antibiotics.

In situ reconstruction of the infected aorta is gaining favor over staged reconstruction; advocates of in situ reconstruction note a durable anastomosis can be achieved and debridement of the infected graft removes the nidus of infection to reduce the risk of reinfection [7]. In situ reconstruction has better patency versus extra-anatomic bypass, and when the patient’s presentation includes septic shock, limb salvage based off axillary inflow alone may be insufficient to perfuse the distal legs.

Fistula Takedown

At entry into the abdomen, it is important to preserve the omentum (the reason will be apparent later). As these cases are frequently reoperative adventures, slow and careful dissection is required. A thorough lysis of adhesions is undertaken exposing the retroperitoneum. This ultimately exposes the aortoenteric fistula. These connections are invariably characterized by dense adhesion of the bowel to the aortic structure. This is often left as the last area of dissection.

A decision must be made early in the procedure about the strategy of aortic control. For fistulas to primary aortic aneurysms, early aortic control, often supra-celiac and at the diaphragmatic hiatus, is prudent to guard against uncontrolled hemorrhage during dissection. Endovascular positioning of a Coda balloon devise is also an option. Similar consideration should be taken for patients with fistulas that present with massive bleed (indicating the connection is at a compromised aortic suture line) and in fistulas arising in the setting of endostents (Fig. 12.7). In the latter, the fixation of the proximal tines is compromised and there is a concern for disruption on the dissection of the enteric connection to the aorta. For patients with more common presentation of recurrent infection, disruption of the aortic integrity is less likely and aortic control can be deferred until the fistula is resected and the aortic dissection is performed at the stage of graft excision. With the necessary precautions, dissection of the fistulous adhesion may be undertaken confidently because by definition one wall of the bowel has a hole and “sealed” by the graft. This interface defines the dissection plane.

Fig. 12.7
figure 7

Aortoenteric fistula with peri-graft air as well as contrast extravasation into the aneurysmal sac. This patient presented with large intermittent GI bleeding

Full size image

Gastrointestinal Integrity

We feel that the involved bowel segment in these cases ultimately should be managed with segmental resection rather than primary repair, no matter how small the connection that is identified. The reasons for this are multifactorial. First the bowel proximal and distal to the fistula is often inflamed and more compromised than it appears. The second principle is demonstrated in the case of the duodenal fistulas. Primary repair of the duodenal defect anatomically leaves the suture line directly anterior to the aortic replacement or an aortic stump and within the infected field. This sets this area up for the risk of recurrent adhesion and, in turn, recurrence of fistula [3].

In the case of duodenal resection, we recommend segmental resection of the fourth portion of the duodenum (and only as much of the third as is necessary), with duodeno-jejunal reconstruction to the right of the root of the SMA/SMV. We favor a hand-sewn side to side duodeno-jejunostomy. Full kocherization of the duodenum eases the technical challenge of this anastomosis and as with all anastomoses in this location; constant awareness of the location of the Ampulla of Vater is vital to avoid injury to the common bile duct. Occasionally, the reconstruction needs to be performed to the second portion of the duodenum.

The reconstruction is performed after the aortic portion of the case is completed. This avoids both needless traction on fresh anastomoses and the risk of anastomotic bleed following heparin administration given for the vascular work. The foramen posterior to the mesenteric vessels where the duodenum previously ran is closed by reapproximating the retroperitoneal tissue, further separating the beds of healing tissues.

Colonic fistulae are best managed in the acute repair with colostomy. These patients are often chronically ill and in an inflammatory state, malnourished and in the setting of acute bleed, subject to critical stress, massive transfusion, and hemodynamic lability, all potential risks for colonic anastomotic failure. The colostomy is then reversed at interval following recovery from the initial resection.

Minimize Risk of Recurrence

All patients receive broad-spectrum antibiotics and antifungals preoperatively and intraoperatively, with redosing based on blood loss of 1500 ml. At a minimum, it is vital to remove all unincorporated graft tissue; incorporated tissue can be retained. Tube grafts may be more easily excised in total than bi-limbed grafts. Similarly, tube endostents may be more readily removed as well. In addition to anatomic separation of the gastrointestinal and vascular healing tissues, it is important to cover the resection tissue. Any prior aneurysmal sac should also be debrided at this time; its blood supply is poor and it will be by definition contaminated tissue and a liability. Additionally, any pockets of purulence that are identified need to be drained, debrided, and irrigated. Adjacent psoas abscesses and infected vertebral disks are examples of such entities. The omentum that has been carefully preserved at abdominal entry is dissected from its transverse colonic attachment and an omental pedicle flap is created; the vascular supply is preserved along the greater curve of the stomach. It is brought through the transverse mesocolon to the left of the middle colic artery and the in situ replaced graft is wrapped circumferentially (not just draped). If the omentum is foreshortened, it can be split on a line radially perpendicular to the greater curve of the stomach to allow both anterior and posterior coverages of the graft. Additional length can be achieved by bringing the flap through the mesocolon as posteriorly as possible; the omental flap is then sutured in place to the retroperitoneal tissue. If the aortic stump is to be oversewn, this too should be covered with a healthy omental flap [7]. It is our practice to prepare the in situ replacement graft (Dacron) by soaking it in Rifampin for 1 h prior to placement in a solution of 1200 mg Rifmapin in 200 cc of saline. Studies have demonstrated rifampin can be detected on the graft up to 7 days postoperatively [8].

Managing Systemic Infection

Unlike infected grafts that result from hematogenous contamination, by definition, the aortoenteric fistula will be polymicrobial and enteric bacteria and fungus must be covered. Late definitive diagnosis may have resulted in episodic antibiotic exposure so perioperative broad-spectrum antibiotics are required. Culture of the excised graft and tissue is important as it may help guide and narrow appropriate antibiotic choice. We favor an aggressive antibiotic regimen with an anticipated course of at least 6–12 weeks of intravenous antibiotics, or longer until the ESR and CRP levels have normalized. At that time, oral antibiotics are started and ESR and CRP levels are checked monthly. An elevation of these proinflammatory markers should prompt a CT scan to evaluate for new retroperitoneal inflammation.

Physiologic Considerations

By definition, these are operatively complex, technically challenging, lengthy cases that potentially impose severe physiologic stress on the patient. As with all aortic operations, EBL is usually significant and intraoperative resuscitation strategies should be managed to avoid excessive crystalloid infusion. If the patient is unstable (on pressors, acidotic, and cold) following the vascular portion of the case, then a damage control approach should be employed deploying a temporary abdominal closure prior to gastrointestinal reconstruction. Once the physiologic derangements have been corrected, the patient is returned to the OR from the surgical intensive care unit.

Cell saver autotransfusion may serve to reduce non-autogenous transfusion but should not be employed until the infected graft and debridement is completed; given the substantial friability of the inflamed tissue in the periaortic region or neighboring lumbosacral discitis, substantial bleeding should be anticipated.

Clinical scenario

70-year-old woman is s/p endovascular repair of an abdominal aortic aneurysm 8 years prior to presentation. She was being followed by her primary care provider for occult GI Bleed and anemia. She is transferred from and outside hospital after a self-limited massive upper and lower gastrointestinal hemorrhage. Esophagogastroduodenoscopy demonstrated visualization of the endovascular aortic stent at the junction of D3–D4.

Author’s comment:

This is the overt, dramatic presentation of aorto-duodenal fistula with a massive “heralding bleed.” Though she appears to be stable at the moment, she will likely rebleed at any time and as such there is significant urgency for planning repair. Extra-anatomic vascular bypass remains an option as to the discussion or palliative care for the elderly patient with significant comorbidity. In our current practice, we would arrange a multidisciplinary procedure with both a vascular surgeon and a general surgeon (acute care). The approach for us is most frequently a midline laparotomy with the plan to excise the fistula, replace the aortic graft in situ with a rifampin-soaked Dacron graft, and re-establish GI continuity following segmental resection of the duodenum. Key operative points include: (1) An effort to preserve omentum on entry to the abdomen. In this case, the previous procedure was an EVAR so it is possible that there was no prior abdominal surgery. This will be used as a vascularized pedicle to wrap the new graft. (2) Early vascular control, prior to dissection of the fistula. As this was an EVAR the risk of dislodgement of the endostent is high, especially given that this presentation was a bleeding event suggesting possible endoleak. (3) Confident dissection of the fistula once the vascular control is secure. The safest approach is from the left of the fourth portion of the duodenum, following the mobilization of the ligament of Treitz. Once the adhered duodenum is off of the graft, the segmental blood supply is reliably spirals superiorly as the surgeon follows the duodenum proximally as it crosses under the superior mesenteric artery and vein. (4) Performing a staged, “damage control” procedure as necessary. This is well tolerated and appropriate for patients that undergo large volume resuscitation. (5) Gastrointestinal reconstruction following vascular repair. This too may be performed in a staged manner pursuant to the patient’s physiology.

Key Questions

  1. 1.

    What is the preferred operative approach to taking down an aorto-duodenal fistula?

  2. 2.

    Is there a role for a duodenal exclusion procedure?