- Email: [email protected]
Type 2 Endoleaks after Abdominal Aortic Aneurysm Stent Grafting with Systematic Mesenteric and Lumbar Coil Embolization
Type 2 Endoleaks after Abdominal Aortic Aneurysm Stent Grafting with Systematic Mesenteric and Lumbar Coil Embolization Maureen K. Sheehan, MD,1 Ryan T. Hagino, MD,1 Edith Canby, MD,1 Michael H. Wholey, MD,2 Darren Postoak, MD,2 Rajeev Suri, MD,2 and Boulos Toursarkissian, MD,1,2 San Antonio, Texas
We evaluated the results of our policy of systematic coil embolization of the inferior mesenteric artery (IMA) and/or lumbar arteries (LAs) prior to endovascular abdominal aortic aneurysm (AAA) repair (EVAR). We retrospectively reviewed all patients undergoing EVAR over a 4-year period at one hospital. Results were analyzed using uni- and multivariate analyses. Fifty-five male patients with an average age of 71 years were evaluated. Follow-up averaged 15 ± 13 months. The IMA was either coiled or occluded in 30 cases. One or more LAs were coiled in 29 patients. An average of 1.3 LAs per patients were coiled (range 0-6). There were no immediate or late complications from coiling. At last follow-up, 14 AAAs showed no change in diameter, one increased by 2 mm, and the remainder (n = 40) decreased by 7.5 ± 6 mm in maximal diameter. Only five (9%) type 2 endoleaks were detected during follow-up. Three were associated with AAA size increase. Four of the five were treated with additional coiling, with good results. By logistic regression, neither endoleak occurrence nor AAA shrinkage correlated with LA or IMA coiling. However, by multivariate analysis, completeness of lumbar coiling correlated negatively with aneurysm shrinkage (p = 0.04) and IMA coiling correlated positively with aneurysm shrinkage (p = 0.04). Coil embolization of the IMA and/or LAs prior to EVAR can be safely accomplished in a large number of cases and is associated with a low incidence of type 2 endoleaks. We cannot at present demonstrate a benefit to LA embolization in terms of endoleak prevention or AAA shrinkage. However, IMA embolization may be of benefit in terms of AAA shrinkage.
INTRODUCTION Endoleaks remain a common problem following endovascular abdominal aortic aneurysm (AAA) repair (EVAR). The significance of type 2 endoleaks remains to some degree controversial. While most
1 Division of Vascular Surgery, University of Texas Health Science Center, San Antonio, TX, USA. 2 Division of Interventional Radiology, University of Texas Health Science Center, San Antonio, TX, USA. Correspondence to: Boulos Toursarkissian, MD, Division of Vascular Surgery, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78229, USA, E-mail: [email protected]
Ann Vasc Surg 2006; 20: 458-463 DOI: 10.1007/s10016-006-9103-2 Ó Annals of Vascular Surgery Inc. Published online: June 24, 2006
458
would agree that a type 2 endoleak in the face of a stable AAA size can be safely watched, the additional follow-up during observation can be anxietyprovoking to both patient and physician, especially when the AAA is sizable. Moreover, the treatment of a type 2 endoleak when required can be difficult, requiring supraselective vessel catheterization and/ or translumbar sac puncture. In an effort to reduce the incidence of type 2 endoleaks in our clinical practice, we began a number of years back to preoperatively embolize, whenever feasible, any patent inferior mesenteric artery (IMA) or lumbar arteries (LAs) seen on preoperative angiography. The purpose of this report is to review our experience with this approach. Since a number of patients did not undergo embolization (for a variety of reasons discussed
Vol. 20, No. 4, 2006
later), we also compared the results between the group that did and the one that did not. We specifically wanted to describe the incidence of type 2 endoleak in our practice and to identify any factors that might be predictive of its occurrence.
Type 2 endoleaks after AAA stent grafting 459
Table I. Patency presentation
of
aortic
side
branches
at
Vessel
Yes
No
IMA patent LAs patent
39 (3 with stenosis) 52
16 3 (unclear)
PATIENTS AND METHODS All patients who underwent EVAR at the Audie L. Murphy Veterans Hospital in San Antonio, Texas, between June 2001 and June 2005 were identified from a query of our surgical registry. Medical records, angiogram reports, and operative notes were reviewed in the computerized patient record system (CPRS). Angiographic films were reviewed. We noted the presence or absence of a patent IMA, its coil embolization if patent, the number of patent LAs at the L1-L4 levels, as well as the number and levels of those embolized and those remaining patent. Follow-up information was obtained from vascular clinic visit notes in the CPRS as well as review of follow-up computed tomographic (CT) scans and reports. Any secondary intervention was noted. Statistical analysis was carried out using the Stata software program (StataCorp, College Station, TX). Tests used included t-tests, simple logistic regression, as well as multivariate analysis. For analysis purposes, a decrease in AAA size was defined as an at least 5 mm decrease in maximum diameter. Our protocol for AAA management has been as follows. All patients presenting to the Vascular Surgery Service with an AAA warranting repair are considered for possible stent graft implantation. All patients undergo thin slice (3 mm) spiral CT scan, followed by arteriography. The arteriography is carried out with a marker pigtail and includes anteroposterior, lateral, and oblique views. On the basis of these studies, if the patient is deemed an EVAR candidate, coil embolization is carried out at either the same setting or a later setting. The decision is usually based on contrast load use and other practical scheduling considerations. Contrast load is limited to 150 cc per session. If the embolization requires >150 cc, the patient returns for a second session to complete the embolization. If the celiac and superior mesenteric arteries are patent, the IMA is selectively cannulated with a 5F diagnostic catheter and 0.018-inch tornado-shaped coils are used, preferably 4-2 mm tapering coils. Every effort is made to embolize the IMA close to its origin while keeping its bifurcation free of coils. For the LAs, following initial catheterization with a
Table II. Patency of LAs at presentation Minimum number of patent LAs on angiography
Number of patients
0 1 2 3 4 5 6 7 8 Unclear
0 5 10 3 23 3 6 1 1 3
5 or 4F catheter, a renegade microcatheter (Boston Scientific, Natick, MA) is used to deliver coils to the proximal part of the vessel. As many LAs are embolized as is safely possible.
RESULTS A total of 55 patients received stent grafts during the period. All patients were men, with a mean age of 71 ± 8 years. The average AAA size was 5.1 ± 0.6 cm (range 3.5-7.0, with the smaller ones having associated large iliac aneurysms) at initial presentation. The angiographic details on these 55 patients are shown in Tables I and II. Of the 39 patients with an angiographically patent IMA, 14 underwent IMA coil embolization. Thus, the IMA remained patent (not coiled or occluded) in 25 patients, while it was effectively occluded in 30. Of the 52 patients with patent LAs, 29 underwent coil embolization. The average number of LAs embolized per patient was 1.3. In patients who actually underwent embolization, the number of LAs coiled was 2.6. The resulting angiographic picture for all patients is depicted in Tables III and IV. The most frequent reason for nonembolization of any vessels was scheduling difficulties. Technical difficulty, meaning inability to safely cannulate a vessel, either because of tortuosity in the access
460 Sheehan et al.
Annals of Vascular Surgery
Table III. LAs left patent after embolization Number of LAs left patent
Number of patients
0 1 2 3 4 5 6 Undetermined
7 9 11 6 11 4 1 6
route or more frequently small size of the target vessel in the face of a large aneurysm or a large AAA thrombus load, was the second most frequent reason. It is important to note that no patient suffered embolic or other complications as a result of the coiling procedures. One patient developed back pain after LA embolization without evidence of rupture. He was treated with a stent graft within 24 hr of the embolization procedure, with no untoward sequelae. Eleven patients had one hypogastric artery embolized as well, while another had an accessory renal artery coiled to allow its coverage with the stent graft. The types of stent graft used included AneuRx (Medtronic, Santa Rosa, CA) in 27 cases, Ancure (Guidant, Menlo Park,CA) in nine cases, Zenith (Cook, Bloomington, IN) in 17 cases, and Excluder (initial design; Gore, Flagstaff, AZ) in two patients. On initial completion angiography, endoleaks were noted in 17 patients, two type 1, 14 type 2, and one type 3. Types 1 and 3 endoleaks were immediately addressed in all cases with the use of adjunctive balloon angioplasty, and no cuffs were needed. Of the type 2 endoleaks observed, only one, through the LAs, persisted on follow-up (patient 4 in Table V). During follow-up, which averaged 15 ± 13 months (range 1-41) following EVAR, five patients were noted to have a type 2 endoleak (four new and one persistent). Details on these patients are shown in Table V. As a matter of policy, all patients with any type of endoleak on follow-up CT scan were studied by selective angiography, regardless of AAA size at the time of endoleak detection. No type 1 endoleaks were noted during follow-up. Two other patients had type 3 endoleaks. One of these was treated with simple angioplasty, and the other required a stent graft cuff at the junction of two limbs. At last follow-up, 14 AAAs showed no change in diameter, one increased by 2 mm, and the remainder (n = 40) decreased by 7.5 ± 6 mm in maximal diameter.
By simple logistic regression, occurrence of type 2 endoleak did not correlate with LA coil embolization (p = 0.48) or with IMA status (occluded or coiled vs. patent) prior to EVAR (p = 0.81), nor did AAA shrinkage correlate with LA coil embolization (p = 0.37) or with IMA status (occluded or coiled vs. patent) prior to EVAR (p = 0.96). Patients with an occluded status of the IMA before EVAR (occluded or coiled) had slightly more AAA diameter shrinkage, although the difference fell short of statistical significance (0.6 vs. 0.4 mm in the group with patent IMA; p = 0.066, t-test). There was no association of endoleak or shrinkage with device type or initial aneurysm size. Multivariate analysis failed to reveal additional associations predictive of endoleak occurrence. We then used multiple logistic regression to determine what variables might predict aneurysm shrinkage. We developed a model based on completeness of lumbar embolization, use of any lumbar embolization, use of IMA embolization, and final patency of the IMA (occluded or coiled vs. patent). Using backward variable selection, we found three significant predictors of aneurysm shrinkage. Completeness of lumbar coiling correlated negatively with aneurysm shrinkage (p = 0.04). A patent IMA (noncoiled or nonoccluded) correlated negatively with aneurysm shrinkage (p = 0.02). Finally, IMA coiling correlated positively with aneurysm shrinkage (p = 0.04).
DISCUSSION A number of investigators have studied the issue of whether type 2 endoleaks can be predicted based on aortic side branch anatomy and patency. Gorich et al.1 noted that the presence of four or more patent LAs preoperatively was associated with type 2 endoleaks that persisted for at least 6 months. In their study, 71% of cases with type 2 endoleaks had four or more patent LAs. Fritz et al.,2 in a study of 89 patients with 6 months or more follow-up postEVAR, noted that a larger number of patent LAs before stenting were associated with a significantly higher rate of type 2 endoleak. Similarly, Broeders et al.3 also noted that patients with endoleaks had significantly more patent aortic side branches than patients who did not. Additionally, using a multiple logistic regression model, Sampaio et al.4 noted that the number of patent aortic side branches was an independent predictor of type 2 endoleaks. Finally, Velasquez et al.5 reported that significantly more patients with a patent IMA pre-EVAR develop a persistent type 2 endoleak compared to patients with an occluded IMA (24% vs. 3%).
Vol. 20, No. 4, 2006
Type 2 endoleaks after AAA stent grafting 461
Table IV. Final angiographic result before EVAR Number of patients with a patent IMA and patent LAs
Number of patients with a patent IMA but no LAs
Number of patients with no patent IMA and no patent LAs
Number of patients with patent LAs but not IMA
Undetermined
17
3
4
25
6
None of the four patients with no patent side branches developed an endoleak.
Table V. Patients with type 2 endoleak during follow-up
Patient
IMA patent at time of EVAR
LAs coiled (n)
LAs patent (n)
1
Yes
0
2
4
2
Yes
1
?
3
No
1
4a
No
5
Yes
Time endoleak noted (months)
Source of endoleak
Change in AAA diameter
IMA/LA
Increase 0.2 cm
11
IMA
3
40
LAs
Initial decrease of 0.5 cm, then return to baseline None
3
1
24
LAs
None
0
4
21
IMA
Initial decrease of 0.3 cm, then return to baseline
Intervention
Outcome
IMA and LA coiling IMA coiling
Stabilized
Four LAs coiling None
Decrease in size
IMA coiling
Stabilized
Stabilized (endoleak resolved on its own) Stabilized
a
This patient is the only one who had a type 2 endoleak on completion angiogram post-EVAR. Endoleak persisted until 24 months follow-up and then resolved.
It would therefore seem that decreasing the number of patent aortic side branches might decrease the incidence of type 2 endoleaks. However, reports have been conflicting. Some have reported a benefit. Indeed, Parry et al.,6 who followed a policy of systematic LA and IMA embolization, noted a 62% incidence of delayed-onset type 2 endoleak in patients with patent LAs, as opposed to 0% in patients who underwent embolization. Similarly, Axelrod et al.7 reported on their experience with IMA embolization before stent grafting. IMA embolization reduced the incidence of type 2 endoleak at 6 months from 48% to 17% and resulted in greater AAA shrinkage in their series. In a series of 22 patients treated with embolization before EVAR, Bonvini et al.8 reported a very low 4.5% late incidence of type 2 endoleak. From these reports, preoperative embolization of side branches seems to negatively correlate with the incidence of type 2 endoleak on follow-up. On the other hand, Walker et al.9 did not find the number of patent LAs and IMA to be predictive of type 2 endoleaks in a series of 47 patients, al-
though the duration of follow-up was not specified in that study. Back et al.10 also noted that type 2 endoleak incidence could not be predicted from preoperative side branch patency. Likewise, not all studies have found that preoperative embolization affects postoperative endoleak incidence. Gould et al.11 found no benefit to preoperative side branch embolization, noting a 20% incidence of type 2 endoleak in 20 patients with successful or partially successful embolization vs. 23.5% in the group without preoperative embolization. However, in that same study, none of the nine patients with all LAs as well as the IMA occluded on angiography developed any endoleak. There are a number of possible explanations for the variance in findings. Most of the series listed above, as well as our own in this report, are small in the number of patients, rendering statistical analysis difficult, and all the studies including our own are retrospective. Some studies, such as the one by Gould et al.,11 included incomplete embolization in the ‘‘embolization’’ group, thereby skewing the results. In many cases, assessing the
462 Sheehan et al.
patency of the LAs or IMA on angiography in the presence of a large aneurysm can be challenging. For instance, Walker et al.9 noted only five patent IMAs in 47 patients, a relatively small percentage. In that same study, LAs were subjectively graded as small, medium, or large. Finally, the time at which the endoleaks are detected may be of importance; primary type 2 endoleaks may be more or less likely to be affected by coil embolization. For instance, when looking at immediate poststenting results, Fan et al.12 noted a 50% incidence of type 2 endoleaks in patients with more than six LAs, as opposed to 13% in those with 0-3 LAs patent. Our own incidence of immediate endoleak was much higher than what we saw during follow-up. Our results suggest that IMA embolization prior to stent grafting is beneficial in terms of aneurysm shrinkage. This is in agreement with the results reported by Axelrod et al.7 and by Velasquez et al.5 LA embolization did not appear to be of benefit, as suggested by Gould et al.11 This may be related to the fact that very few patients had all lumbar vessels effectively treated, reflecting a technical limitation. Surprisingly, incomplete LA embolization seemed beneficial as opposed to complete embolization in terms of AAA shrinkage. Whether this is a true finding or a statistical aberration is difficult to prove. LA embolization may have a different effect on sac size as opposed to endoleak occurrence. One may speculate that incomplete LA coiling may allow AAA sac ‘‘decompression.’’ Our incidence of type 2 endoleak was too low (9%) to allow meaningful statistics. Perhaps one reason for our low incidence has to do with the coiling performed. Since preoperative coiling has been our routine, we do not have a historical control group who did not undergo preoperative coiling with which to compare endoleak rates Whether change in aneurysm size can be used as a valid surrogate end point can be debated. In general, few AAAs will decrease in size in the presence of an endoleak, and incidence of expansion in the presence of a type 2 endoleak ranges 13-19% in some series;13-15 also, it is generally agreed that a decrease in AAA size is a reliable sign of a successful outcome. We found it difficult to embolize all LAs seen on angiography. Limitations exist in terms of patient comfort, contrast load required, technical issues in terms of ability to achieve cannulation, as well as the length of time required for complete embolization. We tended to concentrate on the largest LAs, which may or may not be responsible for most type 2 endoleaks.16 Only seven of our patients were left with no single LA patent before EVAR, and none of these seven patients developed a type 2
Annals of Vascular Surgery
endoleak, similar to the findings by Gould et al.11 Interestingly, three of these seven patients had patent IMAs. Thus, while complete LA embolization may be desirable, it may not be a practical goal. IMA coil embolization is a less cumbersome procedure but is not always possible in the presence of an aneurysm. In our experience, IMA embolization was successful in 14 of the 32 cases where it was attempted (44%). We continue, based on our data, to attempt IMA embolization whenever possible. There were no complications from the embolization procedures. There were no embolic, renal, neurological, or gastrointestinal complications in our experience, although they have been reported in the literature.17 To minimize such occurrences, it is important to deploy the coils as close to the vessel’s origin from the aorta as possible. One patient experienced pain and needed urgent stent grafting; it is not clear whether this was related to the embolization procedure, although one must assume a relationship. There are currently no standard accepted ways of treating type 2 endoleaks when deemed necessary, and several techniques exist; translumbar hypogastric, superior mesenteric artery, and transsac cannulations are all utilized but can be difficult and are not free of complications. Hypogastric and superior mesenteric artery cannulations frequently require navigating a tortuous pathway through diminutive vessels, the translumbar approach requires the patient to lay in a prone position, and transsac cannulations involve the risk of graft dislodgement or creation of a type 1 endoleak. In general, type 2 endoleaks follow a benign course, when not associated with sac expansion; however, no methods exist to predict which type 2 endoleaks will cause expansion and which will not. There are several algorithms available for treatment or prevention of type 2 endoleaks. If preoperative work-up only includes a CT angiogram, then persistent type 2 endoleaks with sac expansion can be interrogated and treated when deemed necessary postoperatively. If angiogram is performed preoperatively, the patient can undergo IMA coiling at that time. Likewise, if a patent IMA is seen on CT angiogram, the patient can undergo IMA coiling preoperatively or at the time of operation. Being able to avoid the sometimes difficult task of accessing the endoleak after the stent graft is in place makes the option of potential prevention attractive, at least until effective treatment methods can be developed. However, to date, a strong argument in favor of embolization exists only with regard to the IMA.
Vol. 20, No. 4, 2006
Type 2 endoleaks after AAA stent grafting 463
REFERENCES 1. Gorich J, Rilinger N, Sokiranski R, et al. Endoleaks after endovascular repair of aortic aneurysm: are they predictable? Initial results. Radiology 2001;218:477-480. 2. Fritz GA, Deutschmann HA, Schoellnast H, et al. Frequency and significance of lumbar and inferior mesenteric artery perfusion after endovascular repair of abdominal aortic aneurysms. J Endovasc Ther 2004;11:649-658. 3. Broeders IA, Blankensteijn JD, Eikelboom BC. The role of infrarenal aortic side branches in the pathogenesis of endoleaks after endovascular aneurysm repair. Eur J Vasc Endovasc Surg 1998;16:419-426. 4. Sampaio SM, Panneton JM, Mozes GI, et al. Aneurysm sac thrombus load predicts type 2 endoleaks after endovascular aneurysm repair. Ann Vasc Surg 2005;19:302-309. 5. Velasquez OC, Baum RA, Carpenter JP, et al. Relationship between preoperative patency of the inferior mesenteric artery and subsequent occurrence of type II endoleak inpatients undergoing endovascular repair of abdominal aortic aneurysms. J Vasc Surg 2000;32:777-788. 6. Parry DJ, Kessel DO, Robertson I, et al. Type 2 endoleaks: predictable, preventable, and sometimes treatable? J Vasc Surg 2002;36:105-110. 7. Axelrod DJ, Lookstein RA, Guller J, et al. Inferior mesenteric artery embolization before endovascular aneurysm repair: technique and initial results. J Vasc Interv Radiol 2004; 15:1263-1267. 8. Bonvini R, Alerci M, Antonucci F, et al. Preoperative embolization of collateral side branches: a valid means to reduce type 2 endoleaks after endovascular AAA repair. J Endovasc Ther 2003;10:227-232. 9. Walker SR, Halliday K, Yusuf SW, et al. A study on the patency of the inferior mesenteric and lumbar arteries in
10.
11.
12.
13.
14.
15.
16.
17.
the incidence of endoleak following endovascular repair of infrarenal aortic aneurysms. Clin Radiol 1998;53:593595. Back MR, Bowser AN, Johnson BL, et al. Patency of infrarenal aortic side branches determines early aneurysm sac behavior after endovascular repair. Ann Vasc Surg 2003; 17:27-34. Gould DA, McWilliams R, Edwards RD, et al. Aortic side ranch embolization before endovascular aneurysm repair: incidence of type 2 endoleak. J Vasc Interv Radiol 2001; 12:337-341. Fan CM, Rafferty EA, Geller SC, et al. Endovascular stent graft in abdominal aortic aneurysms: the relationship between patent vessels that arise from the aneurysmal sac and early endoleak. Radiology 2001;218:176-182. Tolia AJ, Landis R, Lamparello P, et al. Type II endoleaks after endovascular repair of abdominal aortic aneurysms. Radiology 2005;235:683-686. Steinmetz E, Rubin BG, Sanchez LA, et al. Type II endoleak after endovascular abdominal aortic aneurysm repair: a conservative approach with selective intervention is safe and cost-effective. J Vasc Surg 2004;39:306312. van Marrewijk CJ, Fransen G, Laheij RJF, et al. Is a type II endoleak after EVAR a harbinger of risk? Causes and outcomes of open conversion and aneurysm rupture during follow-up. Eur J Vasc Endovasc Surg 2004;27:128137. Solis MM, Ayerdi J, Babcock GA, et al. Mechanism of failure in the treatment of type 2 endoleak with percutaneous coil embolization. J Vasc Surg 2002;36:485-491. Miller FJ, Minneau DE. Transcatheter arterial embolization. Major complications and their prevention. Cardiovasc Intervent Radiol 1983;6:141-149.
We evaluated the results of our policy of systematic coil embolization of the inferior mesenteric artery (IMA) and/or lumbar arteries (LAs) prior to endovascular abdominal aortic aneurysm (AAA) repair (EVAR). We retrospectively reviewed all patients undergoing EVAR over a 4-year period at one hospital. Results were analyzed using uni- and multivariate analyses. Fifty-five male patients with an average age of 71 years were evaluated. Follow-up averaged 15 ± 13 months. The IMA was either coiled or occluded in 30 cases. One or more LAs were coiled in 29 patients. An average of 1.3 LAs per patients were coiled (range 0-6). There were no immediate or late complications from coiling. At last follow-up, 14 AAAs showed no change in diameter, one increased by 2 mm, and the remainder (n = 40) decreased by 7.5 ± 6 mm in maximal diameter. Only five (9%) type 2 endoleaks were detected during follow-up. Three were associated with AAA size increase. Four of the five were treated with additional coiling, with good results. By logistic regression, neither endoleak occurrence nor AAA shrinkage correlated with LA or IMA coiling. However, by multivariate analysis, completeness of lumbar coiling correlated negatively with aneurysm shrinkage (p = 0.04) and IMA coiling correlated positively with aneurysm shrinkage (p = 0.04). Coil embolization of the IMA and/or LAs prior to EVAR can be safely accomplished in a large number of cases and is associated with a low incidence of type 2 endoleaks. We cannot at present demonstrate a benefit to LA embolization in terms of endoleak prevention or AAA shrinkage. However, IMA embolization may be of benefit in terms of AAA shrinkage.
INTRODUCTION Endoleaks remain a common problem following endovascular abdominal aortic aneurysm (AAA) repair (EVAR). The significance of type 2 endoleaks remains to some degree controversial. While most
1 Division of Vascular Surgery, University of Texas Health Science Center, San Antonio, TX, USA. 2 Division of Interventional Radiology, University of Texas Health Science Center, San Antonio, TX, USA. Correspondence to: Boulos Toursarkissian, MD, Division of Vascular Surgery, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78229, USA, E-mail: [email protected]
Ann Vasc Surg 2006; 20: 458-463 DOI: 10.1007/s10016-006-9103-2 Ó Annals of Vascular Surgery Inc. Published online: June 24, 2006
458
would agree that a type 2 endoleak in the face of a stable AAA size can be safely watched, the additional follow-up during observation can be anxietyprovoking to both patient and physician, especially when the AAA is sizable. Moreover, the treatment of a type 2 endoleak when required can be difficult, requiring supraselective vessel catheterization and/ or translumbar sac puncture. In an effort to reduce the incidence of type 2 endoleaks in our clinical practice, we began a number of years back to preoperatively embolize, whenever feasible, any patent inferior mesenteric artery (IMA) or lumbar arteries (LAs) seen on preoperative angiography. The purpose of this report is to review our experience with this approach. Since a number of patients did not undergo embolization (for a variety of reasons discussed
Vol. 20, No. 4, 2006
later), we also compared the results between the group that did and the one that did not. We specifically wanted to describe the incidence of type 2 endoleak in our practice and to identify any factors that might be predictive of its occurrence.
Type 2 endoleaks after AAA stent grafting 459
Table I. Patency presentation
of
aortic
side
branches
at
Vessel
Yes
No
IMA patent LAs patent
39 (3 with stenosis) 52
16 3 (unclear)
PATIENTS AND METHODS All patients who underwent EVAR at the Audie L. Murphy Veterans Hospital in San Antonio, Texas, between June 2001 and June 2005 were identified from a query of our surgical registry. Medical records, angiogram reports, and operative notes were reviewed in the computerized patient record system (CPRS). Angiographic films were reviewed. We noted the presence or absence of a patent IMA, its coil embolization if patent, the number of patent LAs at the L1-L4 levels, as well as the number and levels of those embolized and those remaining patent. Follow-up information was obtained from vascular clinic visit notes in the CPRS as well as review of follow-up computed tomographic (CT) scans and reports. Any secondary intervention was noted. Statistical analysis was carried out using the Stata software program (StataCorp, College Station, TX). Tests used included t-tests, simple logistic regression, as well as multivariate analysis. For analysis purposes, a decrease in AAA size was defined as an at least 5 mm decrease in maximum diameter. Our protocol for AAA management has been as follows. All patients presenting to the Vascular Surgery Service with an AAA warranting repair are considered for possible stent graft implantation. All patients undergo thin slice (3 mm) spiral CT scan, followed by arteriography. The arteriography is carried out with a marker pigtail and includes anteroposterior, lateral, and oblique views. On the basis of these studies, if the patient is deemed an EVAR candidate, coil embolization is carried out at either the same setting or a later setting. The decision is usually based on contrast load use and other practical scheduling considerations. Contrast load is limited to 150 cc per session. If the embolization requires >150 cc, the patient returns for a second session to complete the embolization. If the celiac and superior mesenteric arteries are patent, the IMA is selectively cannulated with a 5F diagnostic catheter and 0.018-inch tornado-shaped coils are used, preferably 4-2 mm tapering coils. Every effort is made to embolize the IMA close to its origin while keeping its bifurcation free of coils. For the LAs, following initial catheterization with a
Table II. Patency of LAs at presentation Minimum number of patent LAs on angiography
Number of patients
0 1 2 3 4 5 6 7 8 Unclear
0 5 10 3 23 3 6 1 1 3
5 or 4F catheter, a renegade microcatheter (Boston Scientific, Natick, MA) is used to deliver coils to the proximal part of the vessel. As many LAs are embolized as is safely possible.
RESULTS A total of 55 patients received stent grafts during the period. All patients were men, with a mean age of 71 ± 8 years. The average AAA size was 5.1 ± 0.6 cm (range 3.5-7.0, with the smaller ones having associated large iliac aneurysms) at initial presentation. The angiographic details on these 55 patients are shown in Tables I and II. Of the 39 patients with an angiographically patent IMA, 14 underwent IMA coil embolization. Thus, the IMA remained patent (not coiled or occluded) in 25 patients, while it was effectively occluded in 30. Of the 52 patients with patent LAs, 29 underwent coil embolization. The average number of LAs embolized per patient was 1.3. In patients who actually underwent embolization, the number of LAs coiled was 2.6. The resulting angiographic picture for all patients is depicted in Tables III and IV. The most frequent reason for nonembolization of any vessels was scheduling difficulties. Technical difficulty, meaning inability to safely cannulate a vessel, either because of tortuosity in the access
460 Sheehan et al.
Annals of Vascular Surgery
Table III. LAs left patent after embolization Number of LAs left patent
Number of patients
0 1 2 3 4 5 6 Undetermined
7 9 11 6 11 4 1 6
route or more frequently small size of the target vessel in the face of a large aneurysm or a large AAA thrombus load, was the second most frequent reason. It is important to note that no patient suffered embolic or other complications as a result of the coiling procedures. One patient developed back pain after LA embolization without evidence of rupture. He was treated with a stent graft within 24 hr of the embolization procedure, with no untoward sequelae. Eleven patients had one hypogastric artery embolized as well, while another had an accessory renal artery coiled to allow its coverage with the stent graft. The types of stent graft used included AneuRx (Medtronic, Santa Rosa, CA) in 27 cases, Ancure (Guidant, Menlo Park,CA) in nine cases, Zenith (Cook, Bloomington, IN) in 17 cases, and Excluder (initial design; Gore, Flagstaff, AZ) in two patients. On initial completion angiography, endoleaks were noted in 17 patients, two type 1, 14 type 2, and one type 3. Types 1 and 3 endoleaks were immediately addressed in all cases with the use of adjunctive balloon angioplasty, and no cuffs were needed. Of the type 2 endoleaks observed, only one, through the LAs, persisted on follow-up (patient 4 in Table V). During follow-up, which averaged 15 ± 13 months (range 1-41) following EVAR, five patients were noted to have a type 2 endoleak (four new and one persistent). Details on these patients are shown in Table V. As a matter of policy, all patients with any type of endoleak on follow-up CT scan were studied by selective angiography, regardless of AAA size at the time of endoleak detection. No type 1 endoleaks were noted during follow-up. Two other patients had type 3 endoleaks. One of these was treated with simple angioplasty, and the other required a stent graft cuff at the junction of two limbs. At last follow-up, 14 AAAs showed no change in diameter, one increased by 2 mm, and the remainder (n = 40) decreased by 7.5 ± 6 mm in maximal diameter.
By simple logistic regression, occurrence of type 2 endoleak did not correlate with LA coil embolization (p = 0.48) or with IMA status (occluded or coiled vs. patent) prior to EVAR (p = 0.81), nor did AAA shrinkage correlate with LA coil embolization (p = 0.37) or with IMA status (occluded or coiled vs. patent) prior to EVAR (p = 0.96). Patients with an occluded status of the IMA before EVAR (occluded or coiled) had slightly more AAA diameter shrinkage, although the difference fell short of statistical significance (0.6 vs. 0.4 mm in the group with patent IMA; p = 0.066, t-test). There was no association of endoleak or shrinkage with device type or initial aneurysm size. Multivariate analysis failed to reveal additional associations predictive of endoleak occurrence. We then used multiple logistic regression to determine what variables might predict aneurysm shrinkage. We developed a model based on completeness of lumbar embolization, use of any lumbar embolization, use of IMA embolization, and final patency of the IMA (occluded or coiled vs. patent). Using backward variable selection, we found three significant predictors of aneurysm shrinkage. Completeness of lumbar coiling correlated negatively with aneurysm shrinkage (p = 0.04). A patent IMA (noncoiled or nonoccluded) correlated negatively with aneurysm shrinkage (p = 0.02). Finally, IMA coiling correlated positively with aneurysm shrinkage (p = 0.04).
DISCUSSION A number of investigators have studied the issue of whether type 2 endoleaks can be predicted based on aortic side branch anatomy and patency. Gorich et al.1 noted that the presence of four or more patent LAs preoperatively was associated with type 2 endoleaks that persisted for at least 6 months. In their study, 71% of cases with type 2 endoleaks had four or more patent LAs. Fritz et al.,2 in a study of 89 patients with 6 months or more follow-up postEVAR, noted that a larger number of patent LAs before stenting were associated with a significantly higher rate of type 2 endoleak. Similarly, Broeders et al.3 also noted that patients with endoleaks had significantly more patent aortic side branches than patients who did not. Additionally, using a multiple logistic regression model, Sampaio et al.4 noted that the number of patent aortic side branches was an independent predictor of type 2 endoleaks. Finally, Velasquez et al.5 reported that significantly more patients with a patent IMA pre-EVAR develop a persistent type 2 endoleak compared to patients with an occluded IMA (24% vs. 3%).
Vol. 20, No. 4, 2006
Type 2 endoleaks after AAA stent grafting 461
Table IV. Final angiographic result before EVAR Number of patients with a patent IMA and patent LAs
Number of patients with a patent IMA but no LAs
Number of patients with no patent IMA and no patent LAs
Number of patients with patent LAs but not IMA
Undetermined
17
3
4
25
6
None of the four patients with no patent side branches developed an endoleak.
Table V. Patients with type 2 endoleak during follow-up
Patient
IMA patent at time of EVAR
LAs coiled (n)
LAs patent (n)
1
Yes
0
2
4
2
Yes
1
?
3
No
1
4a
No
5
Yes
Time endoleak noted (months)
Source of endoleak
Change in AAA diameter
IMA/LA
Increase 0.2 cm
11
IMA
3
40
LAs
Initial decrease of 0.5 cm, then return to baseline None
3
1
24
LAs
None
0
4
21
IMA
Initial decrease of 0.3 cm, then return to baseline
Intervention
Outcome
IMA and LA coiling IMA coiling
Stabilized
Four LAs coiling None
Decrease in size
IMA coiling
Stabilized
Stabilized (endoleak resolved on its own) Stabilized
a
This patient is the only one who had a type 2 endoleak on completion angiogram post-EVAR. Endoleak persisted until 24 months follow-up and then resolved.
It would therefore seem that decreasing the number of patent aortic side branches might decrease the incidence of type 2 endoleaks. However, reports have been conflicting. Some have reported a benefit. Indeed, Parry et al.,6 who followed a policy of systematic LA and IMA embolization, noted a 62% incidence of delayed-onset type 2 endoleak in patients with patent LAs, as opposed to 0% in patients who underwent embolization. Similarly, Axelrod et al.7 reported on their experience with IMA embolization before stent grafting. IMA embolization reduced the incidence of type 2 endoleak at 6 months from 48% to 17% and resulted in greater AAA shrinkage in their series. In a series of 22 patients treated with embolization before EVAR, Bonvini et al.8 reported a very low 4.5% late incidence of type 2 endoleak. From these reports, preoperative embolization of side branches seems to negatively correlate with the incidence of type 2 endoleak on follow-up. On the other hand, Walker et al.9 did not find the number of patent LAs and IMA to be predictive of type 2 endoleaks in a series of 47 patients, al-
though the duration of follow-up was not specified in that study. Back et al.10 also noted that type 2 endoleak incidence could not be predicted from preoperative side branch patency. Likewise, not all studies have found that preoperative embolization affects postoperative endoleak incidence. Gould et al.11 found no benefit to preoperative side branch embolization, noting a 20% incidence of type 2 endoleak in 20 patients with successful or partially successful embolization vs. 23.5% in the group without preoperative embolization. However, in that same study, none of the nine patients with all LAs as well as the IMA occluded on angiography developed any endoleak. There are a number of possible explanations for the variance in findings. Most of the series listed above, as well as our own in this report, are small in the number of patients, rendering statistical analysis difficult, and all the studies including our own are retrospective. Some studies, such as the one by Gould et al.,11 included incomplete embolization in the ‘‘embolization’’ group, thereby skewing the results. In many cases, assessing the
462 Sheehan et al.
patency of the LAs or IMA on angiography in the presence of a large aneurysm can be challenging. For instance, Walker et al.9 noted only five patent IMAs in 47 patients, a relatively small percentage. In that same study, LAs were subjectively graded as small, medium, or large. Finally, the time at which the endoleaks are detected may be of importance; primary type 2 endoleaks may be more or less likely to be affected by coil embolization. For instance, when looking at immediate poststenting results, Fan et al.12 noted a 50% incidence of type 2 endoleaks in patients with more than six LAs, as opposed to 13% in those with 0-3 LAs patent. Our own incidence of immediate endoleak was much higher than what we saw during follow-up. Our results suggest that IMA embolization prior to stent grafting is beneficial in terms of aneurysm shrinkage. This is in agreement with the results reported by Axelrod et al.7 and by Velasquez et al.5 LA embolization did not appear to be of benefit, as suggested by Gould et al.11 This may be related to the fact that very few patients had all lumbar vessels effectively treated, reflecting a technical limitation. Surprisingly, incomplete LA embolization seemed beneficial as opposed to complete embolization in terms of AAA shrinkage. Whether this is a true finding or a statistical aberration is difficult to prove. LA embolization may have a different effect on sac size as opposed to endoleak occurrence. One may speculate that incomplete LA coiling may allow AAA sac ‘‘decompression.’’ Our incidence of type 2 endoleak was too low (9%) to allow meaningful statistics. Perhaps one reason for our low incidence has to do with the coiling performed. Since preoperative coiling has been our routine, we do not have a historical control group who did not undergo preoperative coiling with which to compare endoleak rates Whether change in aneurysm size can be used as a valid surrogate end point can be debated. In general, few AAAs will decrease in size in the presence of an endoleak, and incidence of expansion in the presence of a type 2 endoleak ranges 13-19% in some series;13-15 also, it is generally agreed that a decrease in AAA size is a reliable sign of a successful outcome. We found it difficult to embolize all LAs seen on angiography. Limitations exist in terms of patient comfort, contrast load required, technical issues in terms of ability to achieve cannulation, as well as the length of time required for complete embolization. We tended to concentrate on the largest LAs, which may or may not be responsible for most type 2 endoleaks.16 Only seven of our patients were left with no single LA patent before EVAR, and none of these seven patients developed a type 2
Annals of Vascular Surgery
endoleak, similar to the findings by Gould et al.11 Interestingly, three of these seven patients had patent IMAs. Thus, while complete LA embolization may be desirable, it may not be a practical goal. IMA coil embolization is a less cumbersome procedure but is not always possible in the presence of an aneurysm. In our experience, IMA embolization was successful in 14 of the 32 cases where it was attempted (44%). We continue, based on our data, to attempt IMA embolization whenever possible. There were no complications from the embolization procedures. There were no embolic, renal, neurological, or gastrointestinal complications in our experience, although they have been reported in the literature.17 To minimize such occurrences, it is important to deploy the coils as close to the vessel’s origin from the aorta as possible. One patient experienced pain and needed urgent stent grafting; it is not clear whether this was related to the embolization procedure, although one must assume a relationship. There are currently no standard accepted ways of treating type 2 endoleaks when deemed necessary, and several techniques exist; translumbar hypogastric, superior mesenteric artery, and transsac cannulations are all utilized but can be difficult and are not free of complications. Hypogastric and superior mesenteric artery cannulations frequently require navigating a tortuous pathway through diminutive vessels, the translumbar approach requires the patient to lay in a prone position, and transsac cannulations involve the risk of graft dislodgement or creation of a type 1 endoleak. In general, type 2 endoleaks follow a benign course, when not associated with sac expansion; however, no methods exist to predict which type 2 endoleaks will cause expansion and which will not. There are several algorithms available for treatment or prevention of type 2 endoleaks. If preoperative work-up only includes a CT angiogram, then persistent type 2 endoleaks with sac expansion can be interrogated and treated when deemed necessary postoperatively. If angiogram is performed preoperatively, the patient can undergo IMA coiling at that time. Likewise, if a patent IMA is seen on CT angiogram, the patient can undergo IMA coiling preoperatively or at the time of operation. Being able to avoid the sometimes difficult task of accessing the endoleak after the stent graft is in place makes the option of potential prevention attractive, at least until effective treatment methods can be developed. However, to date, a strong argument in favor of embolization exists only with regard to the IMA.
Vol. 20, No. 4, 2006
Type 2 endoleaks after AAA stent grafting 463
REFERENCES 1. Gorich J, Rilinger N, Sokiranski R, et al. Endoleaks after endovascular repair of aortic aneurysm: are they predictable? Initial results. Radiology 2001;218:477-480. 2. Fritz GA, Deutschmann HA, Schoellnast H, et al. Frequency and significance of lumbar and inferior mesenteric artery perfusion after endovascular repair of abdominal aortic aneurysms. J Endovasc Ther 2004;11:649-658. 3. Broeders IA, Blankensteijn JD, Eikelboom BC. The role of infrarenal aortic side branches in the pathogenesis of endoleaks after endovascular aneurysm repair. Eur J Vasc Endovasc Surg 1998;16:419-426. 4. Sampaio SM, Panneton JM, Mozes GI, et al. Aneurysm sac thrombus load predicts type 2 endoleaks after endovascular aneurysm repair. Ann Vasc Surg 2005;19:302-309. 5. Velasquez OC, Baum RA, Carpenter JP, et al. Relationship between preoperative patency of the inferior mesenteric artery and subsequent occurrence of type II endoleak inpatients undergoing endovascular repair of abdominal aortic aneurysms. J Vasc Surg 2000;32:777-788. 6. Parry DJ, Kessel DO, Robertson I, et al. Type 2 endoleaks: predictable, preventable, and sometimes treatable? J Vasc Surg 2002;36:105-110. 7. Axelrod DJ, Lookstein RA, Guller J, et al. Inferior mesenteric artery embolization before endovascular aneurysm repair: technique and initial results. J Vasc Interv Radiol 2004; 15:1263-1267. 8. Bonvini R, Alerci M, Antonucci F, et al. Preoperative embolization of collateral side branches: a valid means to reduce type 2 endoleaks after endovascular AAA repair. J Endovasc Ther 2003;10:227-232. 9. Walker SR, Halliday K, Yusuf SW, et al. A study on the patency of the inferior mesenteric and lumbar arteries in
10.
11.
12.
13.
14.
15.
16.
17.
the incidence of endoleak following endovascular repair of infrarenal aortic aneurysms. Clin Radiol 1998;53:593595. Back MR, Bowser AN, Johnson BL, et al. Patency of infrarenal aortic side branches determines early aneurysm sac behavior after endovascular repair. Ann Vasc Surg 2003; 17:27-34. Gould DA, McWilliams R, Edwards RD, et al. Aortic side ranch embolization before endovascular aneurysm repair: incidence of type 2 endoleak. J Vasc Interv Radiol 2001; 12:337-341. Fan CM, Rafferty EA, Geller SC, et al. Endovascular stent graft in abdominal aortic aneurysms: the relationship between patent vessels that arise from the aneurysmal sac and early endoleak. Radiology 2001;218:176-182. Tolia AJ, Landis R, Lamparello P, et al. Type II endoleaks after endovascular repair of abdominal aortic aneurysms. Radiology 2005;235:683-686. Steinmetz E, Rubin BG, Sanchez LA, et al. Type II endoleak after endovascular abdominal aortic aneurysm repair: a conservative approach with selective intervention is safe and cost-effective. J Vasc Surg 2004;39:306312. van Marrewijk CJ, Fransen G, Laheij RJF, et al. Is a type II endoleak after EVAR a harbinger of risk? Causes and outcomes of open conversion and aneurysm rupture during follow-up. Eur J Vasc Endovasc Surg 2004;27:128137. Solis MM, Ayerdi J, Babcock GA, et al. Mechanism of failure in the treatment of type 2 endoleak with percutaneous coil embolization. J Vasc Surg 2002;36:485-491. Miller FJ, Minneau DE. Transcatheter arterial embolization. Major complications and their prevention. Cardiovasc Intervent Radiol 1983;6:141-149.