Evaluation of Abdominal Aortic Aneurysm Stent Graft Failure.-Plain Radiograph, Axial 2D and 3D Volume-Rendered Multidetector Row CT

Leo P. Lawler, MD, FRCR1, G. Melville Willliams MD2,
Elliot K. Fishman, MD, FACR1.

1 The Russell H. Morgan Department of Radiology and Radiological Sciences, Johns Hopkins Medical Institutions.
2The Division of Vascular Surgery, Department of Surgery , Johns Hopkins Medical Institutions.

INTRODUCTION

Ruptured abdominal aortic aneurysms are the 10th leading cause of death in men over 55. Elective surgical repair carries a mortality risk ranging from 1.4 to 6.5% [1] with risks of 5.7% to 31% in those with other co-morbid conditions [2]. Endovascular stents have become an accepted alternative to open surgery for repair of abdominal aortic aneurysm since Parodi [3] reported the first series in 1991.The purported advantages over an open surgery are decreased peri-operative mortality and blood loss, reduced hospital stay and decreased convalesence[4]. Up to 80% of patients referred for abdominal aortic aneurysm repair are deemed suitable for such endovascular therapy [5]. Successful endovascular deployment can be achieved in 97% of these patients with 30-day mortality less than 3% [6].As with any developing therapy the success they have enjoyed has been followed with close scrutiny of any complications and failure particularly with longer follow-up.

• A description of the techniques for pre- and post-operative evaluation of infrarenal aneurysms with particular attention to multidetector row CT and three-dimensional volume rendering and the features pertinent for endovascular stent placement.
• A discussion and illustration of the features seen pre- and post-operatively that may predict or are signs of endoprosthesis failure using plain radiograph, axial two dimensional planar and three-dimensional volume rendered multidetector row CT.

IMAGING EVALUATION

Non-invasive imaging plays a fundamental role both in the appropriate selection of patients and their stent design. Likewise it plays a key part in follow-up of patients for the early detection and characterization of complications and failure and in planning salvage procedures. Even with the significant growth in the advanced technologies of CT and MRI the plain film remains the simple easily performed exam of great utility to clinicians. Its main role is demonstrating gross morphology and high resolution fine detail of the wire components of the stent and on occasion demonstrating the early features of abdominal complications such as bowel ischemia [7]. We routinely use an anteroposterior view with right and left oblique images to better appreciate the stent alignment and to view the wire mesh unobscured by vertebral bodies.
Currently a CT evaluation of an abdominal aortic aneurysm should provide a comprehensive study that furnishes all the information required for pre- and post-operative evaluation. Measurement accuracy and full appreciation of the aneurysm and its stent demand contemporary studies yielding high quality data acquisition and a three-dimensional approach that is of high fidelity to the original data set. Single detector spiral CT was a significant advance over sequential scanning by allowing volume data acquisition with faster z-axis coverage and is largely responsible for CT replacing conventional angiography for abdominal applications [8, 9]. Helical single detector CT has been shown effective in stent graft evaluation and is more sensitive for perigraft channels than digital subtraction angiography [10, 11]. Multidetector row spiral CT has sought to make further progress with shorter gantry rotations, faster coverage and near isotropic or isotropic data sets [12]and has been applied successfully in pre- and post-operative evaluation [13, 14].

MULTIDETECTOR ROW AND VOLUME RENDERING TECHNIQUES

The pre- and post-operative CT techniques for evaluation of an abdominal aortic aneurysm are similar. Our volume of coverage routinely involves from above the celiac axis to the femoral vessels (T10 to femoral lesser trochanters). We use negative oral contrast to enhance depiction of the aorta and branch vessels with volume rendering. We have not found this causes any loss of significant diagnostic information. The need for non-contrast studies is debatable. With the speed of multidetector row coverage non-contrast studies to carefully minimize the longitudinal distance are not required pre stent placement. Post stent placement non-contrast images are always obtained to better discriminate small amounts of contrast, hemorrhage or calcium between a stent and the aortic wall. Intravenous contrast is administered through an 18 or 20 gauge needle in an antecubital vein. We use an iodine concentration of 350 mg/L in a non-ionic solution. Power injection is used with an empiric 30-second delay. Although we may vary the delay somewhat according to perceived cardiac status, we don't use test bolus or tracking technique and yet find consistently high quality images. Delayed imaging may also be of value to detect a leak [11] but with multidetector row CT and greater heat loading capacity one must be careful to leave an adequate time between phases so contrast may leak out.
For CT angiography narrow collimation and overlapping reconstructions are advantageous. With aortic studies we will typically use the 1mm detectors and collimate to slice widths of 1.25mm, which will be reconstructed every 1mm. Even with this narrow collimation multidetector row imaging of the aorta will allow high pitch values (6-12.5) for the 50cm coverage required without widening of the slice sensitivity profile or increase in noise. Larger detectors and slice widths can be chosen to gain faster coverage in more unstable patients or those who are poor breath holders. The following table illustrates a typical aortic aneurysm protocol;

Notes.
o Pitch=Table increment per gantry rotation/single detector collimation
o For larger and faster coverage or coverage for those who are poor breatholders 2.5mm collimation with 3mm slice widths and 2mm reconstructions can be used.

To have a credible replacement to conventional angiography for surgeons and interventionalists high quality three-dimensional reconstructions are required. Volume rendering techniques use the high resolution multidetector row data sets to produce images comprised of voxels maintaining high fidelity to all of the originally acquired pixel values [15, 16]. Infinite permutations of width, level, opacity and brightness provide the tools to enhance the visualization of the aneurysm internal anatomy (e.g. thrombus, calcification and stent) and its relationship to external structures and relevant viscera. Likewise limitless clip planes provide fast slab-editing capability and real-time manipulation of perspectives permit evaluation of the aneurysm or stent with views that are customized to an individual patient's anatomy or pathology. Similarly once stents have been placed trapezoids can be applied that optimally depict the stent in relation to the aneurysm. More accurate measurements are obtained as imaging planes that are truly orthogonal to what is often a tortuous aorta can be fashioned. Volume rendering workstations do not preclude use of two-dimensional axial, multiplanar and maximum intensity projection images,which may on occasion be complimentary. We find the high speed of the rendering technique allows practical real-time clinical consultation with referring physicians even when handling 512 X 512 data sets.
Although these principles apply in general terms these images and parameters reflect our experience which is largely with a Siemens Volume Zoom and Siemens 3D Virtuoso workstation (Siemens Medical Systems, Iselin, NJ).

THE STENT GRAFT

Endovascular stent grafts in general are a fabric sleeve supported by a wire frame. Some suprarenal stents have no fabric or have openings to allow continued patency of the aortic branch vessels such as the renal arteries. The wire frame is easily seen and best resolved with plain radiography and CT. Small CT collimation allows accurate depiction of mesh morphology on three-dimensional imaging. The fabric will not be seen on CT unless there is contrast on both sides of it. Through a femoral cut-down stents are inserted and are either self-expanding or are expanded by balloon. The aneurysm is thus excluded form systemic arterial flow. There are three broad categories of endovascular stent for the aorta [17]

• Bifurcated or tubular unibody devices
• Modular multicomponent devices
• Aorto-uni-iliac grafts with contralateral occluder and surgical femoral-femoral bypass

ENDOVASCULAR STENT PLACEMENT

Ultimately the placement of a stent graft to treat an abdominal aneurysm has failed if the aneurysm continues to enlarge and ruptures or if the patient has suffered such complications as to warrant open-surgery or additional salvage interventional procedures. Open surgical repair carries a risk of 1.4% to 6.5% [1] with higher risks noted in patients with other comorbid medical conditions [2]. The idea of a minimally invasive approach is appealing but not without its own complications [18-20]. Endovascular repair has a peri-operative mortality 2.7% and a primary conversion rate of 5% [21]. Careful patient selection, close attention to planning the procedure and continued post-procedure vigilance are the only way to avoid such failures. Technical success not only means the lack of complications and successful deployment but complete exclusion of the aneurysm. Proximal perigraft leak and migration requiring redo operations all increase morbidity and mortality.

PRE-OPERATIVE EVALUATION TO IDENTIFY RISK FACTORS

There are two goals pre-operatively

• to triage those patients who are candidates for endovascular therapy,
• to customize the graft components and their placement to that individual patient for optimal outcome.

ANEURYSM NECK

• Neck dimensions- The neck is that portion between the renal arteries and the start of the aneurysm sac and is the proximal attachment site. There is concern regarding long term stability if proximal fixation is not secure [22]. Radial force, friction or transmural fixation are required to maintain the proximal seal. The preferred dimensions include a length of 1.5cm-2cm long [23] with a maximum diameter 2.8 to 3.2cm depending on the system used.. Necks less than 1.5cm long may require stents that are suprarenal with openings for the renal arteries. Grafts should be oversized 2mm at the neck. The neck should be free of significant thrombus or atheroma, be cylindrical and relatively straight. Measurements should be orthogonal from wall to wall and should include mural thrombus.
• Neck angulation-angulation can decrease the amount of graft in contact with the wall and can predispose to migration. Angulation over 60 degrees is excessive. There is debate regarding the shape of the neck and its effect.
• A short distal neck is required

ANEURYSM

• Diameter-Aneurysm diameters including patent lumen and thrombosed lining measurements are required. These measurements must be perpendicular to the lumen. The chances of endoluminal repair being successful are inversely related to the size of the aneurysm itself. The preoperative dimensions are useful to set a baseline to assess late the response to therapy.
• Length of aneurismal sac.
• Large thrombus/atheroma burden in the aneurysm may impede a true seal and aneurysm exclusion.
• Angulation and tortuosity of the aneurysm can make deployment difficult and can limit the intimacy of the seal.
• RENAL ARTERIES-The number and site of renal arteries and accessory renal arteries is required to evaluate proximal stent attachment and stent design to minimize the risk of renal vascular insult. In these predominantly atherosclerotic patients any renal artery stenoses should also be documented.
• COELIAC,SUPERIOR AND INFERIOR MESENTERIC ARTERIES. The patency of coeliac and mesenteric vasculature may help estimate the risk of bowel ischemia should the inferior mesenteric artery be compromised.
• EXTERNAL ILIAC, INTERNAL ILIAC,COMMON FEMORAL ARTERIES. Iliac arteries must be large enough to accommodate the sheath and small enough to allow distal attachment. Receiving iliac arteries 2.2-2.4cm are favored. Pronounced angulation and calcification are risk factors for adverse outcome as they cause difficulties with deployment though these vessels can be straightened somewhat during the procedure. Documented stenoses can be angioplastied before the stent procedure is started when recognized.
• LUMBAR VESSELS. Though implicated in post-procedure endoleaks, documentation of lumbar vessels pre-procedure will rarely change management.

POST-OPERATIVE COMPLICATIONS AND STENT FAILURE

A true success is a patient who has successful treatment of the aneurysm with a morbidity and hospital stay less than conventional open surgery. Follow-up CT should document the completeness of perigraft thrombosis which is a measure of the completeness of exclusion of the aneurysmal sac from the systemic pressures. CT is the preferred method for post-procedure evaluation after stent-graft placement with sensitivity and specificities quoted at 92% and 90% for perigraft leak compared with 63% and 77% for conventional angiography [24]. The goal of CT is to diagnose the presence of leak, evaluate its site, etiology and extent and assess for complications arising from it. Either seeing the leak itself or aneurysm enlargement indicates incomplete exclusion of the aneurysm. CT also aids planning of second procedures for further treatment. Bulging of the fabric may simulate a leak but three-dimensional evaluation may help discriminate this.

Failure to treat the aneurysm.
In its simplest form an increase in size of aneurysm or rupture indicates treatment failure and a decrease in size implies clot retraction [10]. In most cases where a stent has been successfully placed without complication the presumed decreased pressure on the aneurysm sac is deemed a therapeutic success whether the aneurysm remains stable in size or fails to enlarge. Therefore a stent failure is thought to be related to some failure to obtain a tight proximal or distal seal with resultant lack of stent surface contact with the aneurysmal sac. There are multiple factors involved that are inter-related. Failure of stent placement may be classified as follows;
1. Failure with an intact stent
2. Failure with a disrupted stent
3. Failure related to complications independent of the stent.

1. Failure to treat the aneurysm with an intact stent.

• Failure of proximal or distal attachment. 13% of 314 patients have been found to have proximal neck dilatation at one year follow-up [25]. Smaller necks dilate more than larger ones in both open and endovascular repair but neck enlargement may have no relation to neck length, endoleaks or aneurysm size. Enlargment of the neck may lead to loss of seal and thus leak, migration, aneurysm enlargement and rupture [26]. The most common complication at the distal site is incomplete occlusion with iliac leaks reported in 6%. This occurs in distal aorta with a tube graft or in the iliac vessels if a bifurcated or aorto-uniiliac stent and is usually due to a mismatch of sizes.
• Endoleak. An endoleak is the persistence of blood flow beyond the graft but within the aneurysmal sac [27] and is the most prevalent complication with rates of 2-45% noted [24, 27, 28]. An endoleak is synonymous with failure to completely exclude the aneurysmal sac but does not necessarily imply failure to treat the aneurysm. An endoleak may certainly lead to aneurysm enlargement which may cause migration or rupture in 12-44% of patients [29] but they can spontaneously close. Endoleak is the main indication for conversion [6],which has a 24 % mortality which is six times higher than primary open repair. Endoleaks may be classified by CT [28] and there are various classification persepectives-
• Primary or secondary leaks. Primary leakage occurs within the peri-operative period (within 30 days) and secondary leaks occur as a late event but either exposes the aneurysmal sac to systemic pressures and the potential for aneurysmal growth and rupture [30]. Primary leaks are usually related to an angulated proximal neck, short (<1.5cm) or noncircular attachment, mural thrombus, severe calcifications at the attachment or misplacement. Secondary leaks include these factors but also subsequent displacement of the graft due to ineffective seal at graft ends or between overlapping graft segments, membrane defect, where limbs that are too short or there is a size mismatch.
• Graft related or non graft related leaks. Non-graft endoleaks are related to backflow from aorto-iliac branches such as lumbar arteries or IMA and most resolve spontaneously [25].
• Leaks classified by site. Endoleaks classified as proximal, middle or distal site of origin [17, 23, 27]. Distal endoleaks are less dangerous than proximal leaks [6]. Proximal or distal leaks are due to incomplete fixation and migration [18] and mid graft leaks are related to graft defects or retrograde flow.
• Stent migration. Migration usually means distal movement of the proximal portion of the stent. It may be thought of as a failure of attachment with an unchanged aorta or as a failure related to change in aneurysm morphology. The stent force and aortic recoil may lead to inferior displacement causing proximal detachment [31]. If the neck is too short, the attachment is not solid or if there is insufficient columnar strength distal migrations may occur [13, 14]. Migration is associated with neck dilatation in 50% of cases. On plain radiographs eccentric compression at the proximal stent on plain radiograph has been associated with an increased risk of stent migration [32]. Radiography is less useful than CT for assessing short distances of migration [32].
• Stent occlusion. Stent occlusion due to kinking, angulation or thrombosis is relatively rare [18-20]. Mild changes in shape over time are acceptable but severe distortion, kinking or twisting can predict failure of the graft. When aneurysms decrease in diameter in response to stent placement they also may decrease in length [27, 33, 34], which may lead to kinking. Angulation may lead to stenosis or thrombosis or disconnection of the secondary inserted limb. Failure at the attachment sites and migration may also lead to angulation abnormalities [35]. Kinking can cause blood inflow or outflow velocity slowing. Such morphology abnormalities may be difficult to infer from two-dimensional axial CT but can be appreciated on multiple view plain radiographs and volume rendered images. These changes can help early diagnosis and plan salvage therapy. Although some peripheral crescentic thrombus may be seen in 3-19% of cases [19] this does not predict occlusion in everyone though continued surveillance is recommended to exclude progression. Limb thrombosis is seen in 1.5-6% cases.

3. Complications leading to poor outcome despite a successfully placed stent.

• Bowel or visceral vascular insult. The patient population addressed here often already have aorta atherosclerotic branch vessel lumen compromise. Branch vessel thrombosis is associated with stent placement. The plain radiograph may provide some of the earliest signs of bowel ischemia. Colon necrosis has been documented in association with endovascular stents[36]. Renal ischemia may appreciated by altered perfusion of the kidneys and diminished vessel opacification by CT. Shower emboli occur more commonly after endovascular stent than conventional open surgery [37] with rates of 4-28% and may often be rapidly fatal. It is thought to occur after disruption of thrombus in an atherosclerotic vessel. Non-occlusive decreased perfusion related to hypotension may occur during prolonged procedures and again may be related this patient group with increased cardiovascular risk factors.
• Thrombus penetration, dissection or pseudoaneurysm of aorta or iliac vessels. Thrombus penetration [38] or dissection [39] usually occurs during delivery. Pseudoaneurysms can be caused by mechanical disruption of the vessel or be related to secondary infection. Infection carries high rates of morbidity and mortality and may necessitate open surgical treatment. Aortoduodenal fistula may occur in association with pseudoaneurysms. Enlargement of the distal attachment site of the graft has been documented on plain film in a case with pseuodaneurysm formation related to infection [38]. Soft tissue thickening seen on CT does not always imply infection [40] and may be a transient phenomenon post placement.
• Access site complications-hematoma, lymphocele, femoral neuralgia, infection, pseudoaneurysm

REFERENCES

1. Ernst, C.B., N Engl J Med, 1993. 328(16): p. 1167-72.
2. Hollier, L.H., et al., J Vasc Surg, 1986. 3(5): p. 712-7.
3. Parodi, J.C., J.C. Palmaz, and H.D. Barone, 1991. 5(6): p. 491-9.
4. Zarins, C.K., et al., J Vasc Surg, 1999. 29(2): p. 292-305; discussion 306-8.
5. Matsumura, J.S. and W.H. Pearce, Surg Clin North Am, 1999. 79(3): p. 529-40.
6. Vallabhaneni, S.R. and P.L. Harris, Eur J Radiol, 2001. 39(1): p. 34-41.
7. May, J., et al., Eur J Vasc Endovasc Surg, 1997. 13(2): p. 202-6.
8. Rubin, G.D., M.D. Dake, and C.P. Semba, Radiol Clin North Am, 1995. 33(1): p. 51-70.
9. Rubin, G.D., et al., J Vasc Surg, 1993. 18(4): p. 656-64; discussion 665.
10. Rozenblit, A., et al., AJR Am J Roentgenol, 1995. 165(6): p. 1473-9.
11. Golzarian, J., L. Dussaussois, and J. Struyven, 20(1): p. 16-24.
12. Prokop, M., Eur J Radiol, 2000. 36(2): p. 86-96.
13. Rydberg, J., et al., AJR Am J Roentgenol, 2001. 177(3): p. 607-14.
14. Rydberg, J., et al., J Comput Assist Tomogr, 2001. 25(4): p. 580-6.
15. Johnson, P.T., et al., 1996. 167(3): p. 581-3.
16. Lawler, L.P. and E.K. Fishman, Radiographics, 2001. 21(5): p. 1257-73.
17. Dorffner, R., et al., Radiology, 1997. 204(1): p. 79-86.
18. Dorffner, R., AJR Am J Roentgenol, 1999. 172(6): p. 1690-1.
19. Blum, U., et al., Radiology, 1996. 198(1): p. 25-31.
20. May, J., G.H. White, and J.P. Harris, Eur J Radiol, 2001. 39(1): p. 16-21.
21. Zarins, C.K., et al., J Vasc Surg, 2000. 32(1): p. 90-107.
22. Wain, R.A., et al., J Vasc Surg, 1998. 27(1): p. 69-78; discussion 78-80.
23. Armerding, M.D., et al., Radiology, 2000. 215(1): p. 138-46.
24. Makaroun, M.S. and D.H. Deaton, J Vasc Surg, 2001. 33(2 Suppl): p. S39-45.
25. Resch, T., et al., J Vasc Interv Radiol, 1999. 10(3): p. 257-64; discussion 265-6.
26. White, G.H., et al., J Endovasc Surg, 1997. 4(2): p. 152-68.
27. Gorich, J., et al., Radiology, 1999. 213(3): p. 767-72.
28. Cuypers, P., et al., Eur J Vasc Endovasc Surg, 1999. 17(6): p. 507-16.
29. Lumsden, A.B., et al., Am J Surg, 1995. 170(2): p. 174-8.
30. Kaufman, J.A., et al., AJR Am J Roentgenol, 2000. 175(2): p. 289-302.
31. Malina, M., et al., J Endovasc Surg, 1997. 4(1): p. 23-30.
32. Tillich, M., et al., Radiographics, 1999. 19(6): p. 1573-83.
33. Wolf, Y.G., et al., J Vasc Surg, 2001. 33(3): p. 481-7.
34. Chuter, T.A., et al., Radiology, 1999. 210(2): p. 361-5.
35. Harris, P., et al., J Endovasc Surg, 1999. 6(1): p. 11-6.
36. Thompson, M.M., et al., J Vasc Surg, 1997. 25(1): p. 179-86.
37. Mita, T., et al., Radiographics, 2000. 20(5): p. 1263-78.
38. Dake, M.D., et al., J Thorac Cardiovasc Surg, 1998. 116(5): p. 689-703; discussion 703-4.
39. Sapoval, M.R., et al., AJR Am J Roentgenol, 1995. 164(5): p. 1271-3.