Complications of Fenestrated Endovascular Aneurysm Repair

Stent Graft Design

Many of the complications of endovascular aneurysm repair result from a mismatch between the demands of the pathology and the capabilities of the device, and many can be avoided by proper patient selection and device selection, based on a thorough understanding of the differences between various alternatives.

The first branched stent grafts were of unibody design: inserted whole with the branches already attached. The implantation of an entire unibody multibranched stent graft was achieved using a complex system of catheters, all of which had to work in perfect synchrony. The resulting degree of irreducible complexity made the unibody approach unforgiving and complication prone.

These days, most multibranched stent grafts are of modular design: inserted in pieces and assembled in situ. Modular stent grafts fall into two main groups, fenestration-based and cuff-based, depending on the type of branch attachment site.

A fenestration-based branch exits the trunk of the stent graft through a small wire-reinforced hole ( Fig. 14.1A ). The entire ring of contact between the two is barely longer than the thickness of the graft wall. Secure hemostatic attachment depends on the transaxial orientation of the branch and the presence of a slightly overdilated flange at the proximal end of a balloon-expanded covered stent.

A cuff-based branch exits the trunk of the stent graft through a short axially oriented branch ( Fig. 14.1B ) and runs between the wall of the stent graft and the wall of the aorta before turning into the lumen of the corresponding target artery. A 14–18 mm long zone of contact between the inner aspect of the cuff and the outer aspect covered stent helps stabilize the inter-component connection. The curved path of the branch calls for some flexibility in the covered stent, which is usually of the self-expanding variety.

The first step in branch deployment is the same regardless of whether the branch attachment site is a fenestration or a cuff: a catheter is directed from the lumen of the stent graft, through the perigraft space, and into the lumen of the target artery. This bridging catheter is then exchanged, over a stiff guidewire, for the delivery system of a covered stent (the branch). However, there the similarities end. Fenestration-based and cuff-based multibranched stent grafts are selected differently, planned differently, and implanted differently. They fail in different ways, and their failure modes require different remedies. Although most users have a distinct preference for one or the other, the choice often depends on arterial anatomy. As a rule, nondilated, or mildly dilated, segments of the aorta call for fenestration-based branches, whereas truly aneurysmal segments of the aorta call for cuff-based branches. Most experienced centers plan complex endovascular repairs accordingly.

Failure to Insert the Trunk of a Modular Stent Graft

The original multibranched stent grafts were bulky devices, especially those that had long helical cuffs. It used to be relatively common for the delivery sheath to be too wide, or too inflexible, to traverse the external iliac arteries, especially in women. This has become less of a problem since thin-walled fabrics, nitinol stents, and low-profile cuffs have reduced the caliber of the delivery system ( Fig. 14.2A,B ). The low-profile version of the most widely used multibranched stent graft (t-branch, Cook Medical Inc., Bloomington, Indiana) now has an 18-French sheath.

Nevertheless, very small, calcified, stenotic external iliac arteries sometimes impede or prevent transfemoral insertion of even a low-profile delivery system. Under these circumstances, a surgically created route of access is sometimes needed in the form of iliofemoral bypass. We prefer to do this at a separate operation prior to definitive aneurysm repair, thereby minimizing the blood loss that would otherwise be associated with prolonged heparin anticoagulation (during the remainder of a long complicated multibranched stent graft insertion) and allowing the patient to recover in the interval between the two operations. Before embarking on surgical bypass, a close look should be taken at the state of the common iliac arteries, which may be too narrow to admit the delivery system or too calcified to clamp. When the obstructing lesion is confined to the external iliac artery, it may be possible to perform an endovascular bypass of the external iliac artery by aggressive balloon dilatation of the offending artery with a covered stent already in place in case of iliac artery rupture. This is fairly straightforward if the internal iliac artery is already occluded. Otherwise, the proximal end of the covered stent has to be positioned very carefully to preserve flow through internal iliac collaterals to the spine.

Failure to Position the Trunk of the Stent Graft at the Correct Level

Even when the iliac artery disease is not severe enough to prevent insertion, it can still complicate deployment by generating friction between the iliac arteries and the delivery system. Applying caudally direct traction to the entire aortoiliac segment results in high deployment of the stent graft.

Cuff-based stent grafts are quite forgiving of small errors in stent graft position because they run down the outside of the aorta for a variable distance. High deployment of the stent graft just lengthens the necessary stent graft. Low deployment is a different matter. If the outer orifice of a caudally oriented cuff comes to lie below the orifice of the corresponding artery, it can be difficult, if not impossible, to turn the tip of the catheter upward.

Fenestration-based stent grafts are less forgiving of stent graft malposition because the branches run a short distance straight across the perigraft space in a transaxial plane, leaving little room for corrective rerouting.

Failure to Catheterize the Target Artery

In an ideal world, all stent grafts would be customized to achieve a perfect match between the distribution of branch attachment sites on the stent graft and the distribution of branches on the aorta. In reality, this is often difficult to achieve. Branch attachment sites are confined to the spaces between the struts of the stent graft. They cannot always go exactly where wanted. Moreover, it is not always possible to predict the exact orientation of the stent graft, especially in cases of aortic tortuosity. Sometimes the stent graft deploys in alignment with the aorta, sometimes in alignment with the delivery system.

In fenestration-based repairs, the catheter is usually introduced into the stent graft through an access site in the femoral artery and directed out of the stent graft through each fenestration, and constraining ties prevent full stent graft expansion. The resulting preservation of a limited perigraft space allows the catheter enough freedom of movement to accommodate moderate degree of misalignment (see earlier description of failure to orient the stent graft). In addition, the partially deployed stent graft is free to move a little in response to externally applied torque and traction. Most experienced users have learned how to tweak stent graft position during deployment and how to use a range of catheters, in conjunction with a repertoire of catheter manipulations, to bridge the gap in case of less than perfect alignment. The VanSchie range of catheters (numbers 1–5, Cook Medical Inc., Bloomington, Indiana), with its range of acutely angled radiopaque tips, was developed specifically for this purpose. When a particularly wide degree of misalignment cannot be corrected by reorienting the stent graft, it may be necessary to reform a recurved (visceral selective or Sos) catheter in the perigraft space. The preoperative CT will show the proper viewing angle. In most cases, the best angle is at right angles to the arterial ostium, but intermittent views down the long axis of the branch artery help to show the relative positions of the catheter tip, the fenestration, and the arterial ostium.

In cases where some (usually the renal) branches arise from a nonaneurysmal segment of the aorta, the perigraft space can be so compressed as to limit the free movement of a catheter. Under these circumstances the catheter, guidewire, and sheath need to be stiff enough to push the offending stent graft aside. Any tendency for the sheath to form redundant loops at bends in often tortuous pathways from the brachial artery to the cuff can generally be overcome by the tension in a small (0.014) brachiofemoral buddy wire ( Fig. 14.3A ).

In patients who have undergone renal stenting with a portion of the uncovered stent sticking out into the lumen, the catheter tip sometimes finds its way between the fabric and a stent attached to its outer surface; although the guidewire slips through this gap into the target artery, the catheter will not follow. The only remedy is to withdraw both the catheter and guidewire to the outer orifice of the cuff and start again. Under these circumstances, the wire passes easily into the artery through the wall, not the end, of the stent. Catheterization is even more difficult if the end of the stent protrudes from the target artery orifice and impinges on the wall of the stent graft. Under these circumstances, it is sometimes possible to deflect the tip of the wire through the short acutely angled catheter off the wall of the stent graft into the lumen of the stent.

Most experienced users have learned a range of more radical maneuvers for use when all else fails. The piggy-back technique of Ferreira, is an example in which, in which a transbrachial guidewire is carried by a snare over a previously placed transfemoral wire into the target artery. Alternatively, a wire can be introduced into the surgically exposed distal renal artery, retrieved by a snare, and brought out through the usual transfemoral or transbrachial access site.

Failure to Deploy the Branch

The covered stent and its delivery system face two main impediments to smooth trauma-free insertion: the angle between the long-axis of the stent graft (or cuff) and the long-axis of the target artery; and areas of compression, kinking, or stenosis that obstruct the path from the lumen of the stent graft to the lumen of the target artery.

In fenestration-based repairs, the line of insertion has to make a turn of at least 90 degrees to achieve the proper alignment for secure hemostatic attachment between the balloon-expanded covered stent and the margin of the fenestration, unless the vessel is cranially oriented, in which case it will be less than 90 degrees. Assuming the typical transfemoral line of insertion, caudally oriented branches and large aneurysms require the support of stiff guidewires and/or sheaths. Unfortunately, the stiffer the wires and sheaths, the more likely they are to bow and fall out. It may be difficult to insert a stiff wire without the support of a stiff catheter and difficult to insert a stiff catheter (or sheath) in the absence of a stiff wire. There is no universal solution to this particular catch 22. An incremental approach often has to be adopted, using a succession of wires and sheaths of gradually increasing stiffness. Directional sheaths provide robust support and controllable angulation. They are available in an ever-increasing range of lengths and tip curvatures for the insertion of Endoanchors and in the electrophysiology lab for cardiac ablation.

Sometimes, in cases of 3- and 4-branch fenestration-based repair, a large occlusion balloon above the branches will serve to buttress the transfemoral line of insertion. When the renal arteries are very caudally oriented, it may be preferable to use a transbrachial approach (rather than the usual transfemoral approach). A scallop in the proximal margin of the stent graft provides access to the stent graft catheter with the assistance of an indwelling catheter.

Branch Artery Injury

The wires that guide branch insertion often have to be very stiff, because the line of insertion is tortuous, and the arteries, especially the renal arteries, are short. Any uncontrolled movement of such a wire has the potential to perforate distal renal branches or dissect proximal renal branches, with dire consequences. Distal arterial perforation, in a heavily anticoagulated patient, may cause sufficient bleeding to require coil embolization, reversal of heparin, and cessation of the procedure. Although extravasation is usually confined by the perinephric capsule, the resulting capsular pressurization can cause a complete, if temporary, loss of renal function. Arterial dissection occurs when the tip of the guidewire moves back and forth through the primary branch point. Left untreated, the resulting dissection can cause total permanent loss of renal function. The best remedy involves the placement of a short self-expanding stent over a guidewire in the true lumen, which is one reason to perform completion angiograms through a 5-French sheath with a guidewire in place.

As always, an ounce of prevention is worth a pound of cure. If the renal artery is long enough and wide enough, a Rosen guidewire (Cook Medical Inc., Bloomington, Indiana), with its atraumatic tip, can be substituted for some of the more dangerous straight-tip alternatives. Regardless of guidewire choice, the real culprit is uncontrolled wire tip movement, resulting from poor fluoroscopic guidance, an unstable sheath, or a highly curved (or otherwise obstructed) path of branch insertion.

For proper fluoroscopic control of guidewire position, the tip of the wire has to be in the field of view and the X-ray beam directed at right angles to the trunk of the renal artery. This is not necessarily the same beam angle used for catheterization of the cuff and renal orifice. The right renal artery, for example, often originates from the anterolateral aorta (requiring a left anterior oblique [LAO] view), before turning posteriorly and laterally (requiring a right anterior oblique [RAO] view).

Kinking or Compression of the Branch

Kinking or compression of the branch is easy to correct at the time of implantation, less easy to correct once the guidewire has been removed from the branch, and difficult, or impossible, to correct months later when the branch occludes. Kinking occurs more often in cuff-based repairs, in which the proximal end of a self-expanding branch must share the same axial orientation as the cuff and the distal end must share the same axial orientation as the target artery, resulting in an unavoidable bend. The branches of a fenestration-based branch are less susceptible to kinking because they are straight at the time of balloon-driven deployment and robust enough to stay that way, at least in the short to medium term.

Visceral Ischemia

The modular approach to multibranched aneurysm repair is designed to avoid interruption of flow to the visceral arteries within the field of repair. Both fenestrations and cuffs provide a path for blood flow through the wall of the stent graft into the perigraft space, and through the perigraft space into any as-yet-vacant target arteries. Under normal circumstances, this pathway maintains visceral perfusion right up to the point at which branch deployment bridges the gap, excludes the perigraft space, and provides a more direct source of perfusion through the lumen of the branch itself. Furthermore, in fenestration-based repair, the stent graft is maintained in a state of partial compression by a series of constraining ties until all the target arteries have been catheterized and prompt branch deployment assured.

The proximal superior mesenteric often gives rise to a couple of indispensable arteries: the middle colic and right hepatic. Careful preoperative computerized tomography angiography (CTA), intraoperative angiography, and precise deployment are mandatory to avoid inadvertent coverage.

Early fenestrated stent grafts usually had one stented fenestration for each renal artery and one unstented scallop for the superior mesenteric artery. The renal fenestrations were positioned over the corresponding renal orifices using balloon-expanded stents, first covered then uncovered. This pattern remains the norm in the United States, mainly because of the glacial pace of regulatory approval. Elsewhere, there has been a shift toward the use of stent grafts with fenestrations and covered stents for the superior mesenteric and celiac arteries. One factor driving this shift is the potential for the margins of the unstented scallop to cover part, if not all, of the superior mesenteric artery, a cause of late-occurring superior mesenteric occlusion.

Embolism

Postoperative CTA reveals splenic and renal infarcts, presumably the result of intraoperative embolism, in many cases of multibranched endovascular aneurysm repair. This is hardly surprising, given the complexity of aortic instrumentation and the prevalence of extensive mural thrombus in many of these patients. What is surprising is the low overall rate of serious embolic complications such as stroke, paraplegia, renal failure, and intestinal perforation. Some of this disparity may be attributed to aspects of standard implantation technique that isolate much of the aortic wall from the disruptive effects of endovascular instrumentation. Nearly all the aorta within the field of the repair is covered early in the operation by the deployment of a stent graft, and any uncovered aorta between the left subclavian artery and the top of the stent graft is protected by a sheath. Once the sheath is in place and its position stabilized by a brachiofemoral wire, no guidewire, catheter, or branch delivery sheath touches the thoracic aorta.

The disparity between the rate of embolism seen on postoperative CT versus the rate of clinical complications is partly attributable to organ tolerance for small isolated emboli. CT proven embolism to the small intestine rarely causes ischemic necrosis and perforation. Often the only manifestation of localized embolism to the small intestine is a self-limiting ileus. Although the central nervous system is generally considered to be intolerant of ischemia, embolism-related stroke and paraplegia are rare, even when postoperative CTA shows localized areas of hypoperfusion. Lower body embolism, as evidenced by the typical cutaneous mottling, rarely causes serious morbidity unless associated with embolism to upstream intraabdominal organs. The typical course is relatively benign with spontaneous resolution within days. Even in the presence of digital gangrene, a policy of benign neglect results in the least amount of tissue loss.