Introduction
The Endovascular Aneurysm Repair (EVAR) 1 trial and Dutch Randomised Endovascular Aneurysm Management (DREAM) trials have established the safety and efficacy of endovascular repair using commercially available devices.^1–3 However, many abdominal aortic aneurysms (AAA) are not suitable for endoluminal grafting, often for anatomical reasons. The most common of these is a compromised proximal neck anatomy. The reason is often that the proximal neck is dilated or conical, indicating that it is part of the aneurysmal process. In most trials of endovascular aneurysm repair, an increase in aortic diameter of 10% signifies the end of the proximal neck, and thus, the use of standard endovascular devices in patients who have necks which fall outside these strict limits is unproven. Placing standard endovascular devices into aneurysms with parallel necks shorter than 15 mm has the potential to lead to increased risk of migration and type 1 endoleak.⁴

The concept of fenestrations is to create holes in a stentgraft, which can be aligned with intended branch vessels, thus, allowing the proximal fixation and sealing site of the graft to be moved upwards into a more healthy section of the aorta while maintaining perfusion of the branch vessels. This allows the technology of endovascular grafting to be applied to juxtarenal aneurysms (Figure 1), an area previously amenable only to open surgery, which often involved significant risk of renal failure and mesenteric ischemia.⁵

There are 2 means by which branch arteries can be preserved. They include conventional (usually reinforced) fenestrations and branches. Fenestrations can be further subdivided into small fenestrations, which are 6 mm wide and range from 6–8 mm high. These can be created at least 15 mm inferior to the proximal aspect of the graft, so as to remain free from any crossing struts, a prerequisite for inserting a mating balloon-expandable stent. Alternatively, a larger fenestration with a diameter between 8 and 12 mm can be created at least 10 mm below the top of the fabric. A portion of the large fenestration may be traversed by one of the struts of the proximal sealing Z-stent and, thus, is not typically used with an additional visceral vessel stent. Finally, a scallop allowing the incorporation of one or more vessels can be carved out of the proximal end of the fabric, with a nominal width of 10 mm and a height ranging from 6–12 mm. The location of the fenestrations and scallops are customized to fit the individual patient’s anatomy (Figure 2).

Indications
The generally accepted indications for fenestrated grafts are for aneurysms with short necks (3–15 mm) (i.e., juxtarenal aneurysms). It is important that the fenestrated vessels emerge from the nondilated aorta, as the graft must be opposed to the aortic wall at the branch point to prevent leaks. Shorter necks (and dilated necks) are generally felt to be unsuitable for fenestrated grafts and are usually selected for reinforced fenestration or for directional branched grafts. In general, rather than traversing a long segment of an artery at a right angle to the direction of flow (using a reinforced fenestrated design coupled with a balloon expandable stent-graft), if the gap between the stent-graft and visceral branch is more than 10 mm, a directional branch is preferred. Furthermore, in the setting of marked angulation, as with all internal iliac branches, directional branches are generally used.

Sizing
Each endograft is generally customized to fit the patient’s anatomy, based upon preoperative, aortic, high-resolution CT scanning incorporating the thoracic and abdominal aorta. Angiography, magnetic resonance imaging, and intravascular ultrasound scanning can be used in selected patients, but are not used routinely. Standard measurements are obtained (proximal and distal neck lengths and diameters, angulation, and aneurysm morphology). The use of a workstation with the availability of centerline of flow analysis (e.g., AquariusWS, Terarecon, San Francisco, California) simplifies the planning process to such a degree that it is almost compulsory.

Anatomical requirements include a neck diameter from 19–31 mm, angulation of less than 45 degrees, a 30-mm distal iliac sealing zone, and access vessels of 7.5 mm or greater for delivery of the device. The device is supplied as a proximal tubular component containing the fenestrations, and a distal bifurcate component similar to the conventional Zenith device into which may be placed one or more limb extensions. This configuration has several advantages. The tubular component is easier than a bifurcate component to manipulate into accurate apposition with the branch vessels. The modular design also ensures that forces on the bifurcate component are uncoupled from the tubular component, thus protecting the fenestrations from the effects of any migratory forces on the bifurcate component.

As with conventional devices, the outer diameter of the proximal component is oversized, however, not as extensively as with infrarenal devices (10–15% maximum). Next, the fenestrated/seal zone must be planned. Clearly, it is here that accuracy is of paramount importance, otherwise the fenestrations will not line up with their corresponding branches. The lowest aortic segment, which will provide an adequate seal is chosen. It is optimal if the sealing zone is greater then 15 mm, or at least longer than 50% of the proximal sealing stent length to create a parallel configuration. Small fenestrations are generally used for the lowest vessels in a short neck (usually the renal arteries). Scallops and large fenestrations are used for more proximal vessels. The ostial diameters of each vessel, their relative distances from the superior mesenteric artery, and orientations from which they arise from an aortic cross-section are recorded. Device design is intended to maximize the proximal sealing zone and accommodate native arterial angulation, thus providing durable fixation. The areas of overlap between the tubular and bifurcate sections are intentionally long, preferably greater than 4 stents in length, so that if caudad migration of the bifurcate component occurs, separation of the two components can occur without complete disruption and a type 3 endoleak.

The first reports of the use of fenestrated devices were published in 1999.⁶ A number of case reports with various designs have been published, most with successful outcomes. There is a single, commercially available fenestrated endograft, based on the Cook Zenith platform (Bloomington, Indiana). The material (polyester and stainless steel Gianturco stents) as well as the delivery system are similar to the standard Zenith graft. As it has not yet received FDA approval, its use remains confined to the realms of investigational device exemptions in the United States. The device has a CE mark and is commercially available in Europe, Australia, and New Zealand.

Technique
After femoral artery exposure, patients are heparinized to maintain activated clotting times greater than 300 seconds for the duration of the procedure. A stiff wire is advanced into the aortic arch through the femoral artery on the intended side of delivery. A large sheath is inserted into the contralateral femoral artery. A flush catheter is positioned immediately above the celiac artery through the contralateral femoral artery. Angled catheters are placed by puncturing the valve of the large sheath on the contralateral side and left at the level of the aortic bifurcation. The first component is oriented using radio-opaque markers to accommodate the incorporated renal and visceral ostia and then inserted over the stiff wire. Partial expansion of the device is then accomplished by sheath withdrawal to reveal 2 covered stents (Figure 3). A further angiogram is performed at this point, and the device is more accurately oriented. The graft is then fully exposed by complete withdrawal of the sheath. Diameter-reducing ties cause posterior tethering and prevent complete expansion of the prosthesis after sheath withdrawal. This allows additional adjustment of fenestration position by rotational and longitudinal movement.

Access to the partially expanded endograft is achieved through the contralateral femoral sheaths with the use of steerable catheter-guidewire systems. A minimum of 2 visceral vessels are then cannulated through the respective fenestrations from within the prosthesis. Guiding catheters (8 Fr Multipurpose B Lumex Guiding Catheter, Cook, Inc.) or sheaths (Ansel, Cook, Inc.) are inserted over Rosen wires into both of the accessed fenestrations (Figure 4). In the setting of significant ostial stenosis, other techniques may be required to gain renal access. Once at least 2 branch vessels are cannulated, the graft material is then fully expanded by removing the wire tethering the posterior aspect of the prosthesis. The top cap is then deployed. Relatively long balloon-expandable stents or stent-grafts (Jostent or Jomed covered stent graft, Abbott, Abbott Park, Illinois) are then used to stent the branch vessels. This is done so that at least 15–16 mm of the stent is lodged within the visceral vessel and 3–4 mm extends out into the aorta. The visceral and renal stenting technique is modified to account for early bifurcations, ostial stenoses, and severe angulation. The aortic component of the balloon-expanded visceral stents is flared by further dilatation with a 10 mm balloon and then selectively flared with a compliant latex balloon. This maneuver “rivets” the stent-graft to the aortic wall. The top cap is retrieved while access to both stented vessels is maintained with the guiding catheters. The guiding catheters are removed after further angiography. The second (bifurcate) component of the system is then inserted through the ipsilateral femoral artery, oriented, and deployed such that the contralateral limb expands immediately above the aortic bifurcation. Contralateral access is then obtained in a standard fashion through the contralateral sheath, and the remainder of the deployment sequence is similar to the infrarenal Zenith system. Compliant balloon inflation at all joints and distal sealing zones precede completion angiography.

Cleveland Clinic Data
From 2001 onwards, high-risk patients in the Cleveland Clinic with compromised neck morphology were recruited to a trial of the Zenith fenestrated device. The studies were prospective and non-randomized in nature. Follow-up studies included clinical, laboratory, and imaging studies at 1 month, 6 months, 12 months, and annually thereafter. The complete inclusion and exclusion criteria have been published elsewhere.⁷

In a manner similar to the EVAR trials,^1,2 the definition of “high risk” was left to the treating physician with some guidelines regarding anesthetic risk, cardiopulmonary dysfunction, and renal insufficiency provided. As well as physiologic factors, patients with multiple prior abdominal or aortic surgeries, inflammatory aneurysms, and other situations, which may have resulted in unfavorable open surgical results were potentially considered “high-risk.” Preoperative, intra-operative and postoperative data collection was extensive, particularly in relation to risk factors and other physiologic information. Postprocedural complications were stratified into those that occurred within the first 30 days and those that occurred after the first 30 days. The complications were also categorized by body systems as follows: cardiac, respiratory, vascular, neurologic, renal, and gastrointestinal and those complications related to cancer and diabetes. Survival data were supplemented by querying the Social Security Death Index on a quarterly basis.
Data analysis. All data captured were entered and stored in a remote database (Oracle Clinical v 4.03, Redwood Shores, California). Statistical analyses were performed using S+ version 6.1 for Unix (Insightful Corporation, Seattle, Washington).

Results
The most recent CCF fenestrated experience was reported in 2006.8 This included 119 patients. There were 98 men and 21 women, with a mean age of 75 years. They were a high-risk group with 49% having significant coronary artery disease and a 26% having renal insufficiency. Comorbidities are described in Table 1. Although we have made several attempts to categorize these patients into their relevant high-risk factors, many patients had specific issues that did not lend themselves to grouping. The technical success rate was 100%. Conduits were used for access in 6 patients. In addition to proximal fenestrations, 8 patients also had hypogastric branch devices inserted to preserve antegrade internal iliac flow in at least one internal iliac artery.^9,10

Mortality. The mean follow-up was 19 months (range 0–48 months). Sixteen patients died during the follow-up period, 1 within 30 days, 7 within the first year, and 8 patients after 12 months of follow-up. Kaplan–Meier estimates of survival at 1, 12, 24, and 36 months are 0.99, 0.92, 0.83, and 0.79, respectively. There were no late aneurysm-related deaths and no conversions to open procedures.

Endovascular outcomes. The mean diameter of the proximal neck was 26 mm (range, 17–34 mm). The mean proximal neck length was 8 mm (range, 3–18 mm). The proximal neck length was < 10 mm in 70 patients, and between 10 and 18 mm in 49 patients, all of which had morphologic factors implying compromised sealing or fixation. A total of 302 visceral vessels were incorporated in the prosthesis design. Endoleaks were depicted on the post-procedural CT scan as detailed in Table 2. Four of six patients with proximal type I endoleaks underwent post-operative secondary procedures prior to hospital discharge. There were no mortalities associated with secondary procedures.

Discussion
Endovascular repair has been embraced by vascular surgeons since its introduction in 1991. Criticisms of the technique have included the potentially inferior durability of endovascular repair compared to open repair and the requirement for more assiduous follow-up. However, despite the increased requirement for secondary intervention, most of these procedures (all in our series) can be performed endoluminally with minimal morbidity.^11,12 Given the relative complexity of the planning and execution of the technique using fenestrated grafts, experience has been limited to centers with extensive experience with endovascular grafting.^6,8,13–16

Most of the centers performing appreciable numbers of fenestrated grafts are now at the stage of reporting their intermediate-term results. The intermediate-term results of fenestrated grafts seem to support their continued use, especially in patients with anatomic contraindications for standard EVAR. Close surveillance is important for early identification of visceral or branched vessel stenosis and preocclusion. Failure, although uncommon, as a result of death, secondary interventions, branch vessel patency, and complications seem to occur most commonly during the first year and then diminish.¹⁵ As the procedure matures, long-term results and randomized clinical trials will ultimately be required to determine the safety, efficacy, and stability of the procedure itself as well as devices.

The definition of complications also merits discussion. In the CCF series, there were 30 endoleaks out of 119 patients.⁸ All of those that underwent a secondary intervention had them carried out endoluminally. Historically, most type 2 endoleaks do not require intervention. When they do, the procedures are done with a miniscule incidence of morbidity and mortality. Thus, it is debatable whether type 2 endoleaks should be classified as complications of endovascular repair. Secondary interventions have long been considered the bane of EVAR. However, all of the interventions in our series of complex grafts were achieved with endovascular means without recourse to open surgery. The magnitude of the intervention and its associated morbidity is more important than its mere existence. In EVAR 2, there were 32 secondary interventions out of 178 patients, including 3 conversions to open surgery. It seems unlikely that secondary interventions contributed appreciably to the mortality rate in our series or in EVAR 2 as the 30-day mortality rate following secondary intervention was 0 in EVAR 2 and 1 patient in our series.

One might ask why the technology has taken so long to disseminate. There are several reasons why this may be so. First, as the devices are not yet FDA approved, each patient must be part of an investigational device exemption in the United States. Commercial use of the devices in Europe and Australia is typically accompanied with a proctor or knowledgeable device specialist. There is also a perception that the technical aspect of inserting a fenestrated endovascular graft is very complex. There is no doubt that it is technically more demanding than conventional endografting. Designing and implanting a device that will accommodate the visceral vessels of the aorta is an intricate task, and one that should only be attempted by an experienced interventionist. However, the fact that over 1500 fenestrated grafts have now been implanted worldwide by over 150 physicians is testament to the fact that the relative complexity of device planning and deployment can be overcome and skepticism about the dissemination of the procedure is rapidly lessening.

Conclusions
Available data today suggest that fenestrated endovascular grafting is relatively safe and feasible with a low morbidity and mortality. It is clearly an option for the high-risk patient with compromised proximal neck morphology, particularly in those high-risk patients who undoubtedly have high morbidity and mortality associated with complex aortic repair. The relative proportion of aneurysms involving the renal arteries is small, yet up to 40% of all AAA are precluded from conventional endovascular repair, many because of inadequate neck length. It is likely that fenestrated techniques will allow for endovascular repair to be conducted in these patients.