`procedures after endovascular repair of abdominal
`aortic aneurysm
`
`Ellis S. K. Sampram, MD,a Matthew T. Karafa, MS,b Edward J. Mascha, MS,b Daniel G. Clair, MD,a
`Roy K. Greenberg, MD,a Sean P. Lyden, MD,a Patrick J. O’Hara, MD,a Timur P. Sarac, MD,a
`Sunita D. Srivastava, MD,a Brett Butler, MD,a and Kenneth Ouriel, MD,a Cleveland, Ohio
`
`Objectives: Endovascular stent grafting offers a potentially less invasive option for treatment of abdominal aortic
`aneurysm. Clinical benefit has been demonstrated with respect to early parameters such as blood transfusion, return of
`gastrointestinal function, and length of hospital stay. Endovascular repair, however, has been criticized on the basis of
`inferior long-term outcome. Secondary procedures may be necessary to address durability issues such as migration,
`high-pressure endoleak, graft limb thrombosis, and degeneration of the stent-fabric structure itself, issues that may
`compromise the primary goal of aneurysm repair, protection from rupture.
`Methods: Between 1996 and 2002, 703 patients underwent endovascular treatment of infrarenal abdominal aortic
`aneurysm at The Cleveland Clinic Foundation. During this time, five devices were used: Ancure, AneuRx, Excluder,
`Talent, and Zenith. Outcome was assessed with physical examination, lower extremity arterial studies, plain abdominal
`radiography, and computed tomography at discharge, at 1, 6, and 12 months postoperatively, and annually thereafter.
`Secondary procedures were defined as any procedure, exclusive of diagnostic angiography, performed after stent graft
`implantation, directed at treatment of aneurysm-related events. Multivariable statistical techniques for censored data
`(Cox proportional hazards modeling) were used to determine baseline parameters associated with need for secondary
`procedures over follow-up, with calculation of hazards ratio (HR) and 95% confidence interval (CI).
`Results: Patient follow-up averaged 12.2 ⴞ 11.7 months. Patient survival was 90% ⴞ 1.4% at 1 year, 78% ⴞ 2.6% at 2 years,
`and 70% ⴞ 3.8% at 3 years. Aneurysm rupture occurred in 3 patients (0.4%), accounting for rupture risk of 1.4% over the
`first 2 years of follow-up (Kaplan-Meier method). Overall, 128 secondary procedures were required in 104 patients
`(15%), with a cumulative risk of 12% ⴞ 1.5% at 1 year, 24% ⴞ 2.8% at 2 years, and 35% ⴞ 4.4% at 3 years after stent graft
`implantation. Among the secondary procedures, new stent grafts and extensions were placed in 34 patients (27%),
`embolization of endoleak was performed in 33 patients (26%), and open surgical conversion was undertaken in 11
`patients (9%). Periprocedural mortality of secondary procedures was 8% overall, but was 18% for patients undergoing
`open surgical conversion. Multivariable modeling identified the date the procedure was performed (HR, 1.53 per
`3-month period of study; CI, 1.22-1.92; P < .001) and aneurysm size (HR, 1.35 per centimeter of minor axis; CI,
`1.13-1.60; P < .001) as independent predictors of need for secondary procedures.
`Conclusions: Current endovascular devices are associated with a relatively high rate of complications over mid-term
`follow-up, culminating in frequent need for secondary remedial procedures. With strict follow-up imaging compliance,
`however, risk for rupture and aneurysm-related death remain exceedingly low. Newer technology may achieve improved
`durability and a lower requirement for secondary procedures, while maintaining the minimally invasive nature of
`presently available devices. (J Vasc Surg 2003;37:930-7.)
`
`Endovascular abdominal aortic aneurysm (AAA) repair
`provides substantial clinical benefit for the patient.1 The
`physiologic insult is clearly reduced with endovascular tech-
`niques compared with a traditional approach, culminating
`
`From The Departments of Vascular Surgerya and Biostatistics,b The Cleve-
`land Clinic Foundation.
`Competition of interest: RKG is a consultant for Cook, Inc, and KO is a
`consultant for Medtronic AVE. The Cleveland Clinic received research
`funding from Guidant, Medtronic AVE, WL Gore & Associates, and
`Cook, Inc.
`Presented at the Twenty-sixth Annual Meeting of the Midwestern Vascular
`Surgical Society, Madison, Wis, Sep 12-14, 2002.
`Reprint requests: Kenneth Ouriel, MD, Chairman, Department of Vascular
`Surgery, The Cleveland Clinic Foundation, Desk S40, 9500 Euclid Ave,
`Cleveland, OH 44195 (e-mail: ourielk@ccf.org).
`Copyright © 2003 by The Society for Vascular Surgery and The American
`Association for Vascular Surgery.
`0741-5214/2003/$30.00 ⫹ 0
`doi:10.1067/mva.2003.281
`930
`
`in decreased use of the intensive care unit, diminished
`length of hospital stay, and early return to normal activi-
`ties.2 In contrast, longer term outcome may be inferior in
`patients who receive stent -grafts, with a greater require-
`ment for hospital readmission to treat device-related prob-
`lems.3,4 The need for secondary procedures reached 33%
`within 3 years of endovascular aneurysm repair in patients
`entered into the EUROSTAR registry.5 Persistent en-
`doleak, device migration, and limb thrombosis were com-
`mon causes for these remedial procedures, approximately
`one fourth of which were open surgical procedures and
`three fourths were endovascular procedures.3
`A variety of issues may underlie the low durability of
`present endovascular devices. Some of these issues may
`relate to anatomy of the aorta and iliac vessels, parameters
`best assessed with precise anatomic measurements such as
`those accessible from core laboratory data. Other concerns
`
`TMT 2104
`Medtronic v. TMT
`IPR2021-01532
`
`
`
`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 5
`
`Sampram et al 931
`
`Table I. Follow-up protocol comprising institutional standard of care
`
`Item
`
`Timing
`
`History and physical examination
`Four-view abdominal plain radiograph films*
`Helical CT scans
`
`Preoperative; at discharge; 1, 6, 12 months; yearly thereafter
`At discharge; 1, 6, 12 months; yearly thereafter
`Within 3 months of repair; at discharge; 1, 6, 12 months; yearly thereafter
`
`*Before 1999, abdominal plain x-ray films were obtained in two planes alone, anteroposterior and lateral.
`
`may derive from clinical variables such as demographic
`parameters, aneurysm size, or brand of device implanted.
`The present investigation was undertaken to characterize
`the nature and frequency of secondary interventions after
`endovascular aneurysm repair, with analysis of baseline
`variables predictive of remedial interventions.
`
`METHODS
`Over 6 years ending in 2002, 703 patients underwent
`endovascular repair of infrarenal AAA. Five devices were
`used over this period: Ancure (Guidant, Menlo Park, Calif),
`AneuRx (Medtronic/AVE, Santa Rosa, Calif), Excluder
`(W. L. Gore, Flagstaff, Ariz), Talent (Medtronic/AVE),
`and Zenith (Cook Inc, Bloomington, Ind). Among the
`703 patients, 555 patients (79%) received treatment in
`investigational device exemption (IDE) trials: 362 (51%) as
`part of multicenter corporate-sponsored clinical evalua-
`tions, and 193 (27%) under single-institution sponsor-
`investigator IDE studies. The Institutional Review Board
`of The Cleveland Clinic Foundation approved the study
`protocols and the informed consent forms. Commercially
`available devices were placed in the remaining 148 patients
`(21%), and informed consent was obtained from each of
`these patients. Patients who received Zenith devices were
`categorized as those in whom devices were placed as part of
`a multicenter clinical trial (pivotal and continuing access
`phases, Zenith-MCT) and those who received treatment as
`part of a sponsor-investigator IDE (Zenith-SIT). This sub-
`division was deemed necessary because of the broadly dif-
`fering anatomic eligibility criteria in the two subgroups.
`The follow-up protocol was similar for patients in IDE
`trials and patients who underwent implantation of com-
`mercially available devices, representing the standard of
`care for patients receiving treatment at the institution (Ta-
`ble I). Preoperative helical computed tomography (CT)
`was performed with 3 mm axial reconstruction. Angiogra-
`phy and intravascular ultrasound studies were also per-
`formed when measurements were deemed inaccurate on
`the basis of CT scans, in the presence of suspected renal or
`iliac occlusive disease, or when the endograft was placed as
`part of a clinical trial that mandated these studies. Except in
`patients with contraindications such as renal insufficiency
`or history of severe allergy to contrast material, postopera-
`tive CT scans included non-contrast-enhanced, contrast-
`enhanced, and 3-minute to 5-minute delayed post-con-
`trast-enhanced images. Delayed scans are included in our
`imaging protocol to improve sensitivity for detecting en-
`doleak, with careful comparison of sac appearance on pre-
`contrast-enhanced scans and delayed scans.
`
`Data were collected from a review of radiology imaging
`reports and physician chart notes. Any significant discrep-
`ancy between these two data sources prompted an interro-
`gation of the source imaging studies. Outcome reporting
`adhered to the standards outlined by the Ad Hoc Commit-
`tee for Standardized Reporting Practices in Vascular Sur-
`gery of The Society for Vascular Surgery/American Asso-
`ciation for Vascular Surgery (SVS/AAVS).4 Late deaths
`were classified as aneurysm-unrelated or aneurysm-related,
`where aneurysm-related deaths included those deaths that
`occurred as a result of aneurysm rupture or after any pri-
`mary or secondary procedure directed at treating the aneu-
`rysm or complications thereof.
`Secondary procedures were defined as any subsequent
`procedure, whether percutaneous or open surgical, related
`to aneurysm repair or complications thereof. While proce-
`dures performed because of wound complications were
`tabulated for descriptive purposes, they were excluded from
`statistical analysis.
`Endoleak was classified on the basis of serial imaging
`studies.5,6 CT findings, although most influential, com-
`posed only one factor. Presumed type I leak observed on
`CT scans was invariably followed up with angiographic
`confirmation. A leak was considered type II when the
`contrast collection was posterior or at the orifice of the
`inferior mesenteric artery. Type II and IV endoleaks noted
`only on an intraoperative or predischarge imaging study
`were excluded from analysis. Leakage through enlarged
`suture holes or fabric tears was classified as type III en-
`doleak, as were defects related to separation of modular
`components.7 An endoleak was suspected to be type III
`when it directly abutted the graft fabric or was associated
`with obvious disunion of components; half of these were
`confirmed at angiography or at open surgical conversion.
`Migration was defined with clinical and radiographic pa-
`rameters, as suggested by the SVS/AAVS document on
`endovascular reporting standards.4 Migration included
`caudal movement of the proximal attachment site or cranial
`movement of a distal attachment site. A device was consid-
`ered to have migrated when at least 10 mm of movement
`was noted relative to anatomic landmarks, when the patient
`experienced symptoms, or when an intervention was un-
`dertaken to treat migration, irrespective of distance. Aneu-
`rysm shrinkage or growth was determined with a preproce-
`dural CT scan performed 3 months or less before the date
`of the procedure as the baseline. A predischarge imaging
`study was used as the reference scan when a preoperative
`study was not available. Size measurements were made on
`the CT scan with the greatest minor sac dimension on any
`
`
`
`932 Sampram et al
`
`JOURNAL OF VASCULAR SURGERY
`May 2003
`
`remaining length of sealing was deemed inadequate, usu-
`ally when reduced to less than 10 mm.
`Statistical analysis was performed with methods appro-
`priate for censored data. Kaplan-Meier survival analysis was
`used to express patient survival, freedom from aneurysm-
`related death, and freedom from secondary procedures.
`Freedom from such sentinel events as endoleak, device
`migration, graft limb occlusion, and post-repair sac shrink-
`age or enlargement were also expressed with Kaplan-Meier
`curves. Cox analysis was used to evaluate time to secondary
`procedure for specific baseline variables including proce-
`dure date, patient demographic parameters, aneurysm size,
`and procedural details such as brand of device implanted
`and use of adjuvant procedures including placement of
`renal or aortic stents, hypogastric embolization, and use of
`iliac conduits for access. Hazard ratio (HR) was calculated,
`as was 95% confidence interval (CI). Multivariable Cox
`proportional hazards modeling was used to define indepen-
`dent predictors of secondary procedures. Values are speci-
`fied as mean ⫾ SD, except for Kaplan-Meier analysis, in
`which data are expressed as mean ⫾ SE. Significance was
`assumed at P ⬍ .05.
`
`RESULTS
`Patient ages ranged from 48 to 100 years (mean, 75 ⫾
`8.1 years); 605 (86%) were men and 98 (14%) were women.
`The diameter of the aneurysm sac averaged 54 ⫾ 10 mm in
`minor dimension and 58 ⫾ 11 mm in major dimension.
`With a 5.5 cm threshold for minor and major sac axis, 283
`aneurysms (40%) and 391 aneurysms (56%), respectively,
`would be considered large. Pperative mortality (⬍30 days)
`was 1.7% overall (12 patients), 1.0% in those undergoing
`elective treatment (7 patients), and 19% in those undergo-
`ing urgent treatment (5 patients). Of the 12 operative
`deaths, 10 (83%) were procedure-related and occurred as a
`result of perioperative medical complications, whereas 2
`(17%) were device-related and occurred as a direct result of
`complications from the device. Mean duration of follow-up
`was 12.2 ⫾ 11.7 months (range, 0-65 months). Patient
`survival was 90% ⫾ 1.4% at 1 year, 78% ⫾ 2.6% at 2 years,
`and 70% ⫾ 3.8% at 3 years (Kaplan-Meier method; Fig 1).
`Aneurysm-related deaths, including deaths related to the
`initial procedure, occurred in 12 patients (1.7%) and ac-
`counted for 16% of 74 deaths. Risk for aneurysm-related
`death was 2.7% ⫾ 0.5% at 1 year and 3.6% ⫾ 1.2% at 2 and
`3 years postoperatively.
`There were 24 graft limb occlusions in 20 patients
`(2.8%). Risk for limb occlusion was 2.7% ⫾ 0.7% at 1 year,
`4.1% ⫾ 1.2% at 2 years, and 5.5% ⫾ 1.9% at 3 years after
`repair (Fig 2). Late limb occlusion was rare; no new limb
`occlusion was observed after 30 months. Migration was
`noted in 25 patients (3.6%) and, unlike limb occlusion,
`appeared to increase in frequency over longitudinal follow-
`up. Examination of the Kaplan-Meier curve revealed low
`risk for migration during the first year after stent graft
`implantation, estimated at 1.0% ⫾ 0.5% at 1 year but rising
`to 12% ⫾ 3.1% after 3 years of follow-up. Endoleak, of any
`type and observed at any point after discharge, was docu-
`
`Fig 1. Patient survival, expressed as all-cause mortality and aneu-
`rysm-related death.
`
`Fig 2. Freedom from migration or graft limb occlusion over
`follow-up.
`
`axial image. Aneurysm shrinkage was defined as decrease of
`5 mm or more in the minor dimension of the sac; enlarge-
`ment was defined as increase of 5 mm or more in this
`dimension.
`The individual clinician responsible for the patient’s
`care determined the need for secondary procedures. Cer-
`tain clinical events mandated intervention, including post-
`implant aneurysm rupture or symptomatic graft limb occlu-
`sion. Most, however, were less catastrophic, and the need
`for secondary intervention was subjective. Basic treatment
`paradigms were, however, standard at the institution.
`Treatment of types I and III endoleaks was always recom-
`mended. Type II endoleak was treated when the aneurysm
`sac was observed to enlarge over time. As well, patients with
`type II leak received treatment when the aneurysm failed to
`contract despite observation for more than 12 months after
`the initial endovascular repair. Type II leak in patients with
`a shrinking sac was not treated. Device migration was
`treated when it was associated with a type I leak or when the
`
`
`
`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 5
`
`Sampram et al 933
`
`Fig 3. Freedom from endoleak over follow-up.
`
`Fig 4. Freedom from aneurysm sac shrinkage or growth, defined
`by change 5 mm or more in diameter compared with baseline index
`CT scan.
`
`mented in 162 patients (23.0%). At Kaplan-Meier analysis,
`risk for endoleak was 30% ⫾ 2.3% at 1 year, 42% ⫾ 3.4% at
`2 years, and 56% ⫾ 5.5% at 3 years after implantation (Fig
`3). Most endoleaks were type II (35% ⫾ 3.2% at 2 years);
`fewer were device-related (type I or III, 10% ⫾ 2.6% at 2
`years). Sac shrinkage of 5 mm or more occurred in 39% ⫾
`2.7% of patients at 1 year, 60% ⫾ 3.2% of patients at 2 years,
`and 68% ⫾ 3.6% of patients at 3 years after aneurysm repair
`(Fig 4). Sac enlargement was observed in 3.5% ⫾ 1.0% of
`patients at 1 year, 11% ⫾ 2.5% of patients at 2 years, and
`21% ⫾ 4.5% of patients at 3 years after repair.
`Three post-implantation aneurysm ruptures occurred,
`4, 7, and 19 months after implantation, respectively, for a
`rupture-free probability of 98.6% ⫾ 0.9% at 2 years, ac-
`counting for a 0.7% annualized linear risk for rupture.
`Cause of rupture was thought to be displacement of an iliac
`limb from its attachment site and sudden development of a
`distal type I endoleak in 1 patient and disunion of a proxi-
`
`Fig 5. Freedom from secondary procedures over follow-up.
`
`mal extension cuff from the main body in a second patient.
`Cause of rupture remains undefined in the third patient,
`but occurred in an aortic pseudoaneurysm that developed
`at the juxtaposition of the proximal aspect of a stent graft
`and a previously placed renal stent. Among the three pa-
`tients with aneurysm rupture, two underwent conversion
`to open surgery, and 1 of these patients died periopera-
`tively. The third patient underwent successful endovascular
`insertion of a proximal aortic extension cuff.
`Overall, 128 secondary procedures were performed
`in 104 patients (15%). Among secondary procedures, 4
`were directed at treatment of infectious groin wound
`complications and were excluded from subsequent anal-
`yses, leaving 124 procedures performed in a 100- patient
`cohort available for hazard analysis. Cumulative risk for
`secondary procedures was 12% ⫾ 1.5% at 1 year, 24% ⫾
`2.8% at 2 years, and 35% ⫾ 4.4% at 3 years of follow-up
`(Fig 5). Eight deaths occurred as a result of secondary
`intervention, for a periprocedural mortality rate of 6.3%.
`In total, conversion to open surgical repair was required
`in 12 patients (1.9%); 1 was immediate and was excluded
`from further study, and the remainder were tabulated as
`secondary procedures. Among the 11 patients in whom
`the secondary procedure was conversion to open sur-
`gery, 2 patients died (18%). There were six major indi-
`cations for secondary procedures: endoleak (n ⫽ 63),
`endograft migration (n ⫽ 13), graft limb occlusion (n ⫽
`24 in 19 patients), postoperative hemorrhage (n ⫽ 12),
`inadvertent partial or complete renal artery coverage
`(n ⫽ 8), and aneurysm rupture (n ⫽ 3). Interventions for
`each of these indications are listed in Table II. Of note,
`among 34 type II endoleaks treated, 20 (59%) were
`treated because of sac enlargement and 10 (29%) because
`of failure to contract over more than 12 months of
`observation. In the remaining 4 patients (12%), angiog-
`raphy was performed because of suspected device-related
`(type I or III) leak, but type II leak was found at
`angiography and were embolized at that time.
`Univariable analysis identified four baseline variables
`associated with a significantly increased risk for subse-
`
`
`
`934 Sampram et al
`
`JOURNAL OF VASCULAR SURGERY
`May 2003
`
`Table II. Number of complications and nature of interventions performed
`
`Complication
`
`Procedures*
`
`Covered
`stent†
`
`Bare
`stent
`
`Embolization
`(branch/sac)
`
`Bypass or
`thrombectomy
`
`Surgical
`conversion
`
`Other‡
`
`Total
`events
`
`Endoleak
`Type I proximal§
`Type I distal
`Type II
`Type III, disunion
`Type III, microleak
`Migration
`Proximal neck㛳
`Distal attachment site
`Limb occlusion
`Hemorrhage
`Wound infection
`Post-implant sac rupture
`Renal compromise
`Distal embolization
`Total
`
`17
`3
`34
`4
`5
`
`6
`6
`24
`12
`4
`3
`8
`2
`128
`
`10
`3
`1§
`4
`1
`
`4
`6
`
`4
`
`1
`
`3
`
`1
`
`2
`
`8 (2)
`
`33 (1)
`
`19 (1)
`
`4 (1)
`
`4
`
`1
`
`2 (1)
`
`34
`
`14
`
`33
`
`19
`
`11
`
`17
`3
`131
`6
`10
`25
`13
`12
`24
`12
`4
`3
`8
`2
`270
`
`3
`8 (1)
`4 (1)
`
`2
`17
`
`Numbers in parentheses represent periprocedural deaths, when they occurred.
`*Total complications, excluding those that were not treated. Number exceeds number of patients; many patients had more than one complication and more
`than one secondary procedure.
`†Includes placement of aortic extension cuffs (10), iliac extension limbs (18), and new stent grafts (6).
`‡Evacuation of hematoma with (4) or without (4) repair of artery, incision and drainage of groin wound (4), fasciotomy (3), toe amputation (2).
`§Extension limb placed to occlude backflow from hypogastric artery into sac.
`㛳Patients with proximal type I endoleak due to device migration (13) are categorized under endoleak heading and do not appear again under migration.
`
`Table III. Summary of univariable risk in secondary
`procedures after endovascular aneurysm repair
`
`Hazard
`ratio
`
`95% Confidence
`interval
`
`Male
`Age (per year)
`Procedure date (by quarter)
`Minor sac axis
`Major sac axis
`Renal stent
`Aortic stent
`Hypogastric embolization
`Iliac conduit
`Device
`AneuRx (n ⫽ 203)
`Ancure (n ⫽ 63)
`Talent (n ⫽ 39)
`Zenith-MCT (n ⫽ 144)
`Zenith-SIT (n ⫽ 181)
`Other devices (n ⫽ 73)
`
`1.101
`1.001
`1.549
`1.364
`1.373
`0.946
`2.928
`1.243
`1.027
`
`1.000
`1.217
`2.161
`1.824
`1.446
`1.572
`
`(0.60, 2.02)
`(0.98, 1.03)
`(1.24, 1.94)
`(1.15, 1.62)
`(1.16, 1.62)
`(0.38, 2.33)
`(1.35, 6.36)
`(0.66, 2.34)
`(0.32, 3.25)
`
`Reference
`(0.53, 2.78)
`(0.99, 4.73)
`(0.94, 3.54)
`(0.82, 2.56)
`(0.84, 2.96)
`
`P
`
`.76
`.95
`⬍.001
`⬍.001
`⬍.001
`.90
`.007
`.50
`.96
`.32
`
`quent, secondary procedures (Table III). Secondary pro-
`cedures were more common in patients with larger major
`and minor aneurysm sac axes, in patients who received a
`large aortic stent because of proximal endoleak evident at
`initial aneurysm repair, and in patients who received
`treatment later in the course of the review. There were
`no significant differences in risk for secondary proce-
`dures according to brand of device (Fig 6). Multivariable
`analysis identified two independent predictors of second-
`ary procedures: date of primary operation (HR, 1.53;
`95% CI, 1.22-1.92 per quarter; P ⬍ .001) and minor
`
`Fig 6. Freedom from secondary procedures by brand of device
`implanted. Follow-up is truncated at 24 months because of small
`numbers of uncensored patients at this time, with resultant in-
`crease in standard error of the estimate.
`
`aneurysm axis (HR, 1.35; 95% CI, 1.13-1.60 per centi-
`meter; P ⬍ .001).
`
`DISCUSSION
`More than 25 years ago, Dr Juan Parodi first envisaged
`endovascular techniques for repair of AAA after he ob-
`served the morbidity and mortality of traditional open
`surgical repair when performed in patients with compro-
`mised medical status.8 Reasoning that morbidity was di-
`rectly proportional to invasiveness of the intervention, Dr
`Parodi developed a transfemoral technique for aneurysm
`repair. Early on, however, durability was a problem. The
`initial aorto-aortic devices did not include a distal stent.
`
`
`
`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 5
`
`Sampram et al 935
`
`Distal type I endoleak developed over time, resulting in
`recurrent pressurization of the aneurysm sac.9 Even when
`later iterations used proximal and distal stents to affix the
`stent graft to the aorta, late endoleak ensued in an unac-
`ceptably great proportion of patients who received nonbi-
`furcated devices.10
`Traditional open surgical repair is associated with a
`moderate degree of periprocedural morbidity but sets a
`high standard for long-term durability. While selected re-
`views document graft-related complications in as many as
`10% of patients,11,12 more recent series attest to lower
`long-term risks.13 The Cleveland Clinic 10-year experience
`documented an extremely low risk for graft-related compli-
`cations, approximating 0.4% in more than 1135 consecu-
`tive patients undergoing elective open surgical infrarenal
`aortic aneurysm repair.13 In fairness to endovascular aneu-
`rysm repair, however, studies of traditional aneurysm sur-
`gery may be criticized on the basis of less rigorous long-
`term imaging follow-up.
`In the present analysis of approximately 700 endovas-
`cular procedures performed at a single institution over 6
`years, secondary procedures were deemed necessary in a
`high proportion of patients. With Kaplan-Meier analysis,
`risk for a secondary remedial procedure was 35% within 3
`years of implantation. It must be emphasized, however,
`that most secondary procedures were percutaneous. Open
`surgical conversion was rarely necessary. This feature may in
`part account for the low rate of aneurysm-related death
`after endovascular repair, estimated at 3.6% over a 3-year
`follow-up, inclusive of initial perioperative deaths. More-
`over, death from aneurysm rupture occurred in only 2 of
`703 (0.3%) patients followed up for a mean of 1 year.
`Univariable analysis enabled identification of three vari-
`ables associated with increased risk for asecondary proce-
`dures: date the procedure was performed, use of a large
`aortic stent for proximal endoleak at implantation, and
`aneurysm size (minor and major axes). Although the par-
`ticular brand of device did not achieve statistical signifi-
`cance as a predictor of secondary procedures, confidence
`intervals for several of the devices, particularly the Talent
`and Zenith-MCT, suggest that type II statistical error is
`great. Of interest, only two variables, ie, procedure date and
`aneurysm size, remained significant in the multivariable
`model.
`The subjectivity of secondary procedures is the main
`limitation of our analysis. It is likely that physicians were
`more apt to intervene to treat an endoleak, for instance,
`when the aneurysm was large. Equally likely, however, is
`the association between challenging anatomy and larger
`aneurysm.14 The proximal aortic neck shortens as the an-
`eurysm enlarges,15 increasing potential for migration and
`type I endoleak. As the aneurysm enlarges, longitudinal
`dimensions change concurrently.16 Angulation of the aor-
`tic neck and tortuosity of the iliac arteries would also be
`expected to correlate with aneurysm size. Had these ana-
`tomic parameters been available for analysis in the present
`study, they might have been found to underlie the associ-
`ation between secondary procedures and sac diameter.
`
`However, precise baseline anatomic measurements were
`unavailable for most patients included in the study.
`Increasing risk for endoleak over follow-up was ex-
`pected for device-related endoleak, specifically type I and
`III leaks. By contrast, the observation of increasing risk for
`type II leak was not expected. In part, this finding can be
`explained by the cumulative manner in which we chose to
`express the data. As well, however, a few patients demon-
`strated new type II endoleak that became evident many
`months after endograft implantation. It is quite possible
`that these leaks were present from the start but were of
`sufficiently low flow to remain below the threshold of
`detection with conventional CT technology. Over time,
`some type II leaks may act in a manner reminiscent of
`arteriovenous fistula, ie, flow increases with decreased out-
`flow resistance and dilatation of the inflow channels.
`The second independent predictor of secondary proce-
`dures was related to chronology of the initial endovascular
`repair. Patients who received treatment later in the course
`of the review were at markedly increased risk for secondary
`procedures. There are several potential explanations for this
`finding, each of which relates to practice change over time.
`The most obvious explanation centers on a time-related
`change in the aggressiveness with which clinicians treated
`such events as endoleak, aneurysm enlargement, and device
`migration. As well, clinicians may have begun to treat more
`anatomically challenging cases during the later years of
`study. While it is tempting to ascribe time-related changes
`in the distribution of endograft brands to increased require-
`ment for secondary procedures, the multivariable design of
`the analysis suggests that the effect of variations in other
`practice-associated parameters was stronger than the effect
`of the brand of device.
`Studies of endovascular aneurysm repair have been
`criticized with respect to failure to demonstrate true pro-
`tection from rupture. Collin and Murie17 noted that the
`observed rupture risk of 1% per year after endovascular
`repair was not remarkably different from the natural history
`of most small aneurysms followed up without intervention.
`It is appealing to compare the results of the present study
`with natural history studies to determine whether endovas-
`cular aneurysm repair truly protects against rupture and
`whether the method does so to an extent equivalent to that
`of traditional open surgical repair. While comparisons be-
`tween our own series and the two multicenter randomized
`small aneurysm trials are neither statistically nor scientifi-
`cally robust, some element of protection from rupture and
`aneurysm-associated death appears to be conferred by en-
`dovascular repair. The UK Small Aneurysm Trial docu-
`mented death from rupture in 31 of 563 (5.5%) aneurysms
`in the surveillance group over a mean of 8 years of follow-
`up.18 This finding was apparent in spite of the relatively
`small size of the aneurysms and the fact that in 62% of these
`patients open surgical repair was ultimately necessary. Sim-
`ilar observations were noted in the Aneurysm Detection
`and Management Trial.19 Although 62% of patients in the
`surveillance group also eventually required open repair,
`2.6% risk for aneurysm-related death was evident over mean
`
`
`
`936 Sampram et al
`
`JOURNAL OF VASCULAR SURGERY
`May 2003
`
`follow-up of 4.9 years, with observed 0.6% annual risk for
`rupture. The present endovascular series documents mid-
`term risks comparable with or better than those of the
`randomized trials. Moreover, these results were achieved in
`a population in which approximately half of the patients
`had large aneurysms that would have excluded them from
`the randomized small aneurysm trials. The relatively low
`mid-term survival rate in our series is most likely related to
`our policy of advocating endovascular repair for patients
`with medical comorbid conditions and recommending
`open surgical repair for younger, healthier patients.
`In summary, endovascular aneurysm repair in its present
`state appears to be associated with durability that is inferior to
`that we have become accustomed to with traditional open
`surgical techniques. Unlike open repair, endovascular repair is
`associated with a relatively frequent requirement for secondary
`procedures designed to address problems including endoleak,
`migration, and limb thrombosis. Despite these problems, the
`technology does appear to protect patients from aneurysm
`rupture and aneurysm-associated death. Here is where the
`goals of endovascular repair attain relevance. Above all, endo-
`vascular methods must protect patients from aneurysm rup-
`ture and aneurysm-related death, at least to the extent accom-
`plished with standard surgical repair. As well, these results
`should be achieved with a lesser degree of patient discomfort,
`shorter hospital stay, and more rapid return to normal activi-
`ties. While protection from rupture and death appear to be
`attainable with presently available endovascular technology, a
`high rate of secondary procedures may negatively affect pa-
`tient satisfaction. We expect that introduction of newer, more
`durable devices presently under development might diminish
`the rate of secondary procedures after endovascular aneurysm
`repair. Until then, it seems reasonable to limit use of stent graft
`repair to those patients with compromised medical status and
`inordinate operative risk or to those patients willing to accept
`the follow-up imaging protocols and who acknowledge the
`higher rate of secondary procedures that are part and parcel of
`present technology.
`
`REFERENCES
`
`1. Hill BB, Wolf YG, Lee WA, Arko FR, Olcott C, Schubart PJ, et al. Open
`versus endovascular AAA repair in patients who are morphological
`candidates for endovascular treatment. J Endovasc Ther 2002;9:255-
`61.
`2. Zarins CK, White RA, Schwarten D, Kinney E, Diethrich EB, Hodgson
`KJ, et al. AneuRx stent graft versus open surgical repair of abdominal
`
`aortic aneurysms: multicenter prospective clinical trial. J Vasc Surg
`1999;29:292-305.
`3. Carpenter JP, Baum RA, Barker CF, Golden MA, Velazquez OC,
`Mitchell ME, et al. Durability of benefits of endovascular versus con-
`ventional abdominal aortic aneurysm repair. J Vasc Surg 2002;35:
`222-8.
`4. Chaikof EL, Blankensteijn JD, Harris PL, White GH, Zarins CK,
`Bernhard VM, et al. Reporting standards for endovascular aortic aneu-
`rysm repai