CLINICAL RESEARCH STUDIES
`
`Zenith abdominal aortic aneurysm endovascular
`graft
`
`Roy K. Greenberg, MD,a Timothy A. M. Chuter, MD,b Richard P. Cambria, MD,c
`W. Charles Sternbergh III, MD,d and Neal E. Fearnot, PhD,e Cleveland, Ohio; San Francisco, Calif; Boston,
`Mass; New Orleans, La; and West Lafayette, Ind
`
`Purpose: The safety and efficacy of the Zenith (Cook Inc, Bloomington, Ind) endovascular graft was assessed based on the
`United States multicenter trial through 5 years of follow-up.
`Methods: Between 2000 and 2003, the pivotal study enrolled patients to open surgery (control) or the Zenith endovascular
`graft (endovascular). A separate continued access study arm enrolled endovascular patients using the same inclusion/exclusion
`criteria. Both studies were designed for 2-year follow-up, and the pivotal endovascular patients had the option of extending the
`study follow-up through 5 years. All endovascular patients were stratified by physiologic risk into high-risk and standard-risk
`groups to assess overall mortality, rupture, conversion, endoleaks, secondary interventions, and sac enlargement. The entire
`endovascular cohort was pooled to assess device integrity, limb occlusion, component separation, and migration. The
`suboptimal endovascular result (SER) was established as an end point to assess late adverse outcomes. Statistical analyses
`included Kaplan-Meier estimations and Cox regression to assess factors contributing to sac enlargement and SER.
`Results: The study enrolled 739 endovascular patients (352 pivotal, 387 continued access); 158 patients in the pivotal study
`reconsented to be followed up for 5 years. For the patients at standard and high risk at 5 years, the respective survival estimate
`was 83% and 61%, aneurysm-related death was 2% and 4%, and freedom from rupture was 100% and 99.6%, respectively.
`Cumulative risk of conversion, limb occlusion, migration >10 mm, or component separation was <3% at 5 years. Cumulative
`risk of late endoleak was 12% to 15%, representing the primary indication for secondary interventions which occurred in 20%
`of standard-risk patients and 25% of high-risk patients through 5 years. Sac enlargement was very rare and associated with
`advanced age and larger aneurysms. SER was predicted by advanced age and internal iliac artery occlusion.
`Conclusion: These middle- and long-term data support long-term durability of the Zenith endovascular graft. Risk of
`aneurysm-related death or rupture was exceptionally low, and complications of migration, limb occlusion, and device
`integrity issues were uncommon. Incidence of late endoleaks and association of endoleaks with sac growth underscore the
`need for long-term follow-up of patients treated with endovascular grafts, although the sequelae of such events are
`unknown. ( J Vasc Surg 2008;48:1-9.)
`
`Long-term results after endovascular aneurysm repair
`are essential to affirm durability of the intent to diminish
`risk of rupture and aneurysm-related death. The continu-
`ous evolution of implant design during the course of clin-
`
`From the Department of Vascular and Cardiothoracic Surgery and Biomed-
`ical Engineering, Cleveland Clinic Foundation, Cleveland;a the Depart-
`ments of Vascular Surgery at University of San Francisco, San Francisco,b
`Massachusetts General Hospital, Boston,c The Oschner Clinic, New
`Orleansd; and MED Institute, West Lafayette.e
`Competition of interest: Dr Greenberg has received grant and research
`support, consulting fees, and licensed intellectual property to Cook Inc.
`Dr Chuter has received research support, consulting fees, and licensed
`intellectual property to Cook Inc. Dr Sternbergh and Dr Cambria have
`received research support from Cook Inc for their Zenith and TX2
`endograft trials. Dr Fearnot is employed by MED Institute, a Cook Group
`company.
`Additional material for this article may be found online at www.jvascsurg.
`org.
`Correspondence: Roy Greenberg, MD, Director of Endovascular Research,
`The Cleveland Clinic Foundation, Desk S-40, Cleveland, OH 44195
`(e-mail: greenbr@ccf.org).
`0741-5214/$34.00
`Copyright © 2008 by The Society for Vascular Surgery.
`doi:10.1016/j.jvs.2008.02.051
`
`ical trials may affect the relevance of long-term data. Unlike
`most other commercially available stent grafts, the Zenith
`device (Cook Inc, Bloomington, Ind) has undergone re-
`markably few changes since its introduction in 1997. Nev-
`ertheless, there is a paucity of published information on the
`long-term performance of the current device. Reports from
`Australia have included data on predicate designs,1 and
`only two centers in the United States (US) had access to the
`Zenith stent graft before the US pivotal Zenith Multicenter
`Trial (ZMT) in 2000.2 The primary end point of this study
`was a comparison of the morbidity between the surgical
`control arm and the standard-risk endovascular arm at 30
`days and at 1 year. The initial analysis of the pivotal trial was
`reported in 2004, shortly after the device was approved for
`commercial use in the United States.3
`In addition to the initial ZMT report, several analyses
`have further delineated the device performance in a variety
`of patient populations. These include an assessment of late
`complications,4 the effects of the suprarenal stent on renal
`function,5 stent graft oversizing on neck dilation,6 gender
`on overall outcome,7 large iliac limb diameters on iliac
`1
`
`TMT 2099
`Medtronic v. TMT
`IPR2021-01532
`
`

`

`2 Greenberg et al
`
`JOURNAL OF VASCULAR SURGERY
`July 2008
`
`dilatation,8 and the management of type II endoleaks.9 The
`effect of physiologic risk and challenging anatomic factors
`were also evaluated with regard to morphologic out-
`comes,10 and other reports assessed a variety of factors and
`late outcomes
`in the context of other endovascular
`grafts.11-13
`The infrarenal device used in the ZMT construct has
`also become a platform to develop endovascular grafts to
`treat more complex aneurysms, including devices with fen-
`estrations intended to allow the treatment of juxtare-
`nal14,15 and thoracic aneurysms,16 and branched devices
`used to treat thoracoabdominal aneurysms.17-19 Design
`modifications have allowed for the ability to preserve ante-
`grade internal iliac arterial flow in the setting of common
`iliac aneurysms.20-22 We analyzed later outcomes (up to 5
`years) of infrarenal aneurysms treated in the US pivotal
`study with the initial infrarenal device construct and also in
`a continued access study that enrolled patients using iden-
`tical inclusion/exclusion criteria and stratification regimen.
`Patients in the pivotal study were monitored for up to 5
`years, and patients in the continued access study were
`monitored to 2 years. In this article we report the extended
`follow-up data from both the pivotal and continued access
`studies.
`
`MATERIAL AND METHODS
`The US Zenith abdominal aortic aneurysm (AAA) En-
`dovascular Graft pivotal study was a 2-year controlled trial
`that enrolled 432 patients who were treated with open
`surgery (control) or the Zenith endovascular graft (endo-
`vascular); the latter group was stratified by physiologic risk
`into high-risk and standard-risk groups. The details of the
`study design, high-risk criteria, and inclusion and exclusion
`criteria were previously published.3 The primary study end
`point was a comparison of the morbidity between the
`surgical control arm and the standard-risk endovascular
`arm at 30 days and 1 year. The device received US Food
`and Drug Administration (FDA) approval in May 2003.
`Patient enrollment after the pivotal cohort was allowed
`within a separate continued access study arm, within which
`patients were enrolled with the same inclusion and exclu-
`sion criteria and allocated into high- and standard-risk
`groups, in a manner identical to that in the pivotal study.
`However, sites participating in the continued access arm of
`the trial were not required to submit preoperative films for
`review and sizing to a centralized site, providing an avenue
`for greater implanting physician independence. After de-
`vice approval, pivotal patients were given the option of
`extending the study follow-up through 5 years, whereas
`continued-access patients were monitored for only 2 years.
`All participating patients were required to sign an in-
`formed consent document approved by the respective in-
`stitutional review boards. A core laboratory was responsible
`for the independent assessment of all imaging studies, and
`a clinical events committee was responsible for review and
`adjudication of all events reported during the course of the
`trial. A data safety monitoring board reviewed the results
`
`Zenith AAA Endovascular Graft Clinical Trial
`Combined Pivotal, Continued Access and Female Registry Cohorts
`
`High Risk
`(n = 287 enrolled,
`of which 286
`received a device)
`
`Standard Risk
`(n = 452 enrolled, of
`which 450 received
`a device)
`
`Surgical Control
`(n = 80)
`
`Approval based on
`1-year Pivotal study
`results of Standard
`Risk vs. Surgical
`
`Continued Access
`(n = 120)
`
`Continued Access
`(n = 181)
`
`Pivotal (n = 90)
`
`Pivotal (n = 194)
`
`Patients available for
`follow-up at 2 years
`
`n = 76
`
`n = 39
`
`n = 183
`
`n = 119
`
`Pivotal patients eligible for
`5-year follow-up study
`
`Patients agreeing to
`participate in the 5-year
`follow-up study
`
`Fig 1. Enrollment in the United States Zenith Multicenter Trial
`(ZMT). The pivotal portion of the trial
`included 80 surgical
`controls, 200 standard-risk endovascular patients, 100 high-risk
`endovascular patients, and 52 patients allocated to the roll-in arm.
`After pivotal enrollment was completed, continued access was
`provided through a separate study arm. At the conclusion of the
`2-year follow-up, pivotal patients were given the opportunity to
`participate in an extended follow-up study carrying out annual
`assessments through 5 years. The various categories of the trial
`resulted in three fundamental patient groups: surgical controls,
`standard-risk endovascular, and high-risk endovascular. For analy-
`sis purposes, the roll-in, pivotal study, and continued access groups
`were combined into conglomerate standard- and high-risk endo-
`vascular groups, with a variable degree of follow-up. This flow
`chart shows enrollment and follow-up details. AAA, Abdominal
`aortic aneurysm.
`
`according to an enrollment schedule and allowed the study
`to progress to completion.
`Enrollment criteria for the pivotal and continued access
`studies were identical. Moreover, a statistical analysis
`showed that data from the two studies were poolable
`(Table I, online only). Patients in both studies were cate-
`gorized into standard- and high-risk groups by fitness for
`surgery (physiologic risk). For the purpose of data analysis,
`it was assumed that a patient’s physiologic state would have
`no bearing on outcomes such as device integrity, limb
`occlusion, and component separation and migration, which
`definitively relate to long-term device performance. These
`outcomes were assessed using pooled data from both risk
`categories.
`Outcomes considered physiologic risk– dependent in-
`cluded death, rupture, conversion, endoleak, secondary
`interventions, and sac enlargement. Indeed, differences in
`outcomes with this device between patients at high and
`standard physiologic risk have been previously published
`with respect to survival,23 sac behavior, and endoleaks.10
`For risk-dependent outcomes, the standard- and high-risk
`groups were analyzed separately; however, data were still
`combined from the pivotal and continued access studies.
`Specific endovascular outcomes are reported in accor-
`dance with the most recent version of the endovascular
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 48, Number 1
`
`Greenberg et al 3
`
`reporting standards document,24 except when detailed.
`Aneurysm-related deaths included any patients who died
`ⱕ30 days of the primary procedure or any secondary pro-
`cedure, and patients outside of that window whose deaths
`were considered by the clinical events committee to be
`potentially related to the procedure or device.
`The methods for determining device migration are
`detailed in a previously published article.25 For complete-
`ness, both 5-mm and 10-mm cutoffs are reported. Device
`integrity issues refer to holes in the fabric, separation (frac-
`ture) of the barbs, and fractures of the z-stents. Endoleaks
`were considered late occurring only if no endoleak had
`been identified on postprocedural cross-sectional imaging
`studies before the 6-month examination.
`For this study, a conglomerate end point, termed the
`suboptimal endovascular result (SER), was created with the
`intention of assessing the risk that a potentially adverse
`outcome would develop after endovascular repair. This end
`point was intended to assess which patients derived any
`benefit from scheduled follow-up visits or interventions.
`This end point included AAA-related death, rupture, con-
`version, migration ⬎10 mm, limb thrombosis, the devel-
`opment of a late endoleak, or the need for any aneurysm-
`related secondary intervention. The early identification of
`patients who are at risk for reaching this end point has
`significant implications on the optimal follow-up protocols.
`Data were managed by MED Institute, a Cook Group
`company. All statistical analyses were performed using SAS
`8.2 software (SAS Institute, Cary, NC). Standard tabula-
`tions of data included means, standard deviations, and
`percentages, where applicable. Poolability was determined
`by comparing the continued access groups with the pivotal
`trial groups. Kaplan-Meier life-tables were constructed to
`assess for differences between the standard- and high-risk
`patient cohorts and, where appropriate, for clinical and
`imaging outcome variables.
`Patients are censored at their last known date of follow-
`up. For outcomes other than death, the last known date of
`follow-up includes the date of death. A Cox regression
`model was developed to determine the relationship be-
`tween the time and occurrence of maximum measured
`aneurysm enlargement and the relevant covariates of inter-
`est included in the standard- and high-risk patient cohorts.
`A similar Cox regression analysis was used to assess preop-
`erative patient characteristics as potential predictors of the
`time and occurrence of an SER. For both Cox analyses,
`individual factors were first analyzed by a univariate model.
`Individual factors with a value of P ⬍ .15 from the univar-
`iate analysis were included in an additional multivariate
`step-wise Cox regression analysis. A significance level of
`P ⫽ .05 was used to identify significant results from both
`univariate and multivariate analyses.
`
`RESULTS
`The study began in January 2000 and was completed in
`June 2003 after the enrollment of 819 patients, of which 80
`were standard-risk surgical controls. Of the 739 enrolled
`endovascular patients, 736 underwent successful implanta-
`
`tion of a Zenith AAA stent graft, and three implantations
`were aborted due to iliac artery morphology not appreci-
`ated before the procedure. Two of these patients were
`successfully treated with open repair, and one patient
`elected to have no intervention for the AAA. At 2 years, 610
`patients were alive and participating in the trial. Of the 88
`patients who died, 7 had been converted, 1 did not receive
`the device, and 33 were lost to follow-up. From the survi-
`vors, 259 patients were candidates for the 5-year study, and
`158 patients provided voluntary consent (Fig 1). The con-
`tinued access group was considered to be poolable with the
`pivotal trial group based on statistical analyses that demon-
`strated only two differences between the groups, consisting
`of a slightly higher incidence of myocardial infarctions and
`lower incidence of prior aortic surgery in the pivotal study
`group (Table I, online version only).
`Death, rupture, and conversion. Table II provides a
`detailed analysis of freedom from death, rupture, and con-
`version through 5 years of follow-up. Fig 2 depicts Kaplan-
`Meier plots for all-cause mortality. Not surprisingly, the
`all-cause mortality rate was significantly higher in the high-
`risk group (P ⬍ .001). This was the only outcome in Table
`II to show a significant difference between the physiologic
`risk groups, although it is possible that differences were
`obscured by the rarity of many of these events.
`Only one rupture occurred in the entire study cohort.
`The patient was successfully converted, as previously re-
`ported.3 There were seven other conversions in the study,
`two of which occurred after failed stent graft insertions (as
`described previously). Two more conversions were per-
`formed to treat infection detected between 1 and 2 years,
`one conversion was for an additional visceral aortic aneu-
`rysm, and one was done to address a persistent proximal
`endoleak at 6 months. The only conversion that occurred
`⬎2 years resulted from late proximal neck dilation at 4 years
`of follow-up.
`Migration, component separation, limb occlusion,
`and device integrity. Given that these adverse events were
`considered to be independent of physiologic risk, all endo-
`vascular patients were viewed as a conglomerate group.
`Device migration was categorized into movement of 5 mm
`and 10 mm, the latter of which relates to the endovascular
`reporting standards and most other device trials.24 These
`results are detailed in Table III. Only two cases of migration
`of ⬎10 mm occurred, both in the continued access group
`at their concluding follow-up of 2 years. No migrations
`⬎10 mm occurred in the pivotal study throughout the 5
`years. Of the 19 patients where migration between 5 and 10
`mm was noted at any time point, none underwent second-
`ary procedures associated with the migration, continued to
`have migration, or had associated adverse events.
`Component separation was rare, and only three pa-
`tients experienced graft limb separation through 5 years. In
`two cases, disconnection of the uncovered top stent from
`the graft material necessitated placement of proximal exten-
`sions with new uncovered proximal fixation systems.
`Limb occlusions were also rare, with a cumulative risk
`of 2.6% throughout the study course. Of note, all limb
`
`

`

`4 Greenberg et al
`
`JOURNAL OF VASCULAR SURGERY
`July 2008
`
`Table II. Freedom from adverse endovascular events categorized by physiologic risk group (standard vs high risk) and
`expressed as Kaplan-Meier estimates, with standard errors listed in the parentheses
`
`Group/exam period
`
`Rupture
`
`Conversion
`
`AAA-death
`
`Freedom from adverse event
`
`High risk
`1-month
`
`12-month
`
`24-month
`
`36-month
`
`48-month
`
`60-month
`
`Standard risk
`1-month
`
`12-month
`
`24-month
`
`36-month
`
`48-month
`
`60-month
`
`1.000 (0)
`(n ⫽ 283)
`(e ⫽ 0)
`(c ⫽ 4)
`0.996 (0.004)
`(n ⫽ 252)
`(e ⫽ 1)
`(c ⫽ 34)
`0.996 (0.004)
`(n ⫽ 210)
`(e ⫽ 1)
`(c ⫽ 76)
`0.996 (0.004)
`(n ⫽ 39)
`(e ⫽ 1)
`(c ⫽ 247)
`0.996 (0.004)
`(n ⫽ 39)
`(e ⫽ 1)
`(c ⫽ 247)
`0.996 (0.004)
`(n ⫽ 26)
`(e ⫽ 1)
`(c ⫽ 260)
`
`1.000 (0)
`(n ⫽ 447)
`(e ⫽ 0)
`(c ⫽ 4)
`1.000 (0)
`(n ⫽ 422)
`(e ⫽ 0)
`(c ⫽ 29)
`1.000 (0)
`(n ⫽ 375)
`(e ⫽ 0)
`(c ⫽ 76)
`1.000 (0)
`(n ⫽ 119)
`(e ⫽ 0)
`(c ⫽ 332)
`1.000 (0)
`(n ⫽ 116)
`(e ⫽ 0)
`(c ⫽ 335)
`1.000 (0)
`(n ⫽ 79)
`(e ⫽ 0)
`(c ⫽ 372)
`
`0.997 (0.003)
`(n ⫽ 283)
`(e ⫽ 1)
`(c ⫽ 3)
`0.993 (0.005)
`(n ⫽ 252)
`(e ⫽ 2)
`(c ⫽ 33)
`0.988 (0.007)
`(n ⫽ 210)
`(e ⫽ 3)
`(c ⫽ 74)
`0.988 (0.007)
`(n ⫽ 39)
`(e ⫽ 3)
`(c ⫽ 245)
`0.988 (0.007)
`(n ⫽ 39)
`(e ⫽ 3)
`(c ⫽ 245)
`0.988 (0.007)
`(n ⫽ 26)
`(e ⫽ 3)
`(c ⫽ 258)
`
`0.998 (0.002)
`(n ⫽ 447)
`(e ⫽ 1)
`(c ⫽ 3)
`0.993 (0.004)
`(n ⫽ 422)
`(e ⫽ 3)
`(c ⫽ 26)
`0.991 (0.005)
`(n ⫽ 375)
`(e ⫽ 4)
`(c ⫽ 72)
`0.991 (0.005)
`(n ⫽ 119)
`(e ⫽ 4)
`(c ⫽ 328)
`0.991 (0.005)
`(n ⫽ 116)
`(e ⫽ 4)
`(c ⫽ 331)
`0.982 (0.010)
`(n ⫽ 79)
`(e ⫽ 5)
`(c ⫽ 367)
`
`0.990 (0.006)
`(n ⫽ 283)
`(e ⫽ 3)
`(c ⫽ 1)
`0.968 (0.010)
`(n ⫽ 252)
`(e ⫽ 9)
`(c ⫽ 26)
`0.964 (0.011)
`(n ⫽ 210)
`(e ⫽ 10)
`(c ⫽ 67)
`0.964 (0.011)
`(n ⫽ 39)
`(e ⫽ 10)
`(c ⫽ 238)
`0.964 (0.011)
`(n ⫽ 39)
`(e ⫽ 10)
`(c ⫽ 238)
`0.964 (0.011)
`(n ⫽ 26)
`(e ⫽ 10)
`(c ⫽ 251)
`
`0.993 (0.004)
`(n ⫽ 447)
`(e ⫽ 3)
`(c ⫽ 1)
`0.989 (0.005)
`(n ⫽ 422)
`(e ⫽ 5)
`(c ⫽ 24)
`0.984 (0.006)
`(n ⫽ 375)
`(e ⫽ 7)
`(c ⫽ 69)
`0.978 (0.009)
`(n ⫽ 119)
`(e ⫽ 8)
`(c ⫽ 324)
`0.978 (0.009)
`(n ⫽ 116)
`(e ⫽ 8)
`(c ⫽ 327)
`0.978 (0.009)
`(n ⫽ 79)
`(e ⫽ 8)
`(c ⫽ 364)
`
`n, patients at risk; e, cumulative events; c, cumulative censored; n/a, not applicable.
`
`occlusions occurred ⱕ2 years of insertion; no new events
`were seen between 2 and 5 years of follow-up.
`Endoleaks, sac enlargement, and secondary inter-
`ventions. Endoleak rates were analyzed separately for the
`standard- and high-risk groups owing to potential differ-
`ences between the physiologic risk groups10 and the poten-
`tial for physiologic risk to influence the likelihood that a
`patient would undergo a secondary intervention. The re-
`
`ported incidence of primary endoleaks was exceptionally
`low, and most were resolved by the 2-year time point.3
`Details regarding late endoleaks are listed in Table II and
`Fig 3. Most of these were type II endoleaks, and no
`difference existed between the groups at standard and high
`physiologic risk.
`Kaplan-Meier estimates were used to assess the likeli-
`hood of freedom from sac enlargement and freedom from
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 48, Number 1
`
`Table II. Continued
`
`Greenberg et al 5
`
`Death
`
`Late endoleak
`
`Secondary intervention
`
`Aneurysm enlargement
`
`Freedom from adverse event
`
`0.990 (0.006)
`(n ⫽ 283)
`(e ⫽ 3)
`(c ⫽ 1)
`0.912 (0.017)
`(n ⫽ 252)
`(e ⫽ 25)
`(c ⫽ 10)
`0.815 (0.023)
`(n ⫽ 210)
`(e ⫽ 51)
`(c ⫽ 26)
`0.682 (0.039)
`(n ⫽ 39)
`(e ⫽ 66)
`(c ⫽ 182)
`0.682 (0.039)
`(n ⫽ 39)
`(e ⫽ 66)
`(c ⫽ 182)
`0.612 (0.048)
`(n ⫽ 26)
`(e ⫽ 70)
`(c ⫽ 191)
`
`0.993 (0.004)
`(n ⫽ 447)
`(e ⫽ 3)
`(c ⫽ 1)
`0.964 (0.009)
`(n ⫽ 422)
`(e ⫽ 16)
`(c ⫽ 13)
`0.908 (0.014)
`(n ⫽ 375)
`(e ⫽ 40)
`(c ⫽ 36)
`0.860 (0.020)
`(n ⫽ 119)
`(e ⫽ 51)
`(c ⫽ 281)
`0.853 (0.021)
`(n ⫽ 116)
`(e ⫽ 52)
`(c ⫽ 283)
`0.830 (0.024)
`(n ⫽ 79)
`(e ⫽ 55)
`(c ⫽ 317)
`
`n/a
`
`0.964 (0.013)
`(n ⫽ 172)
`(e ⫽ 7)
`(c ⫽ 92)
`0.959 (0.014)
`(n ⫽ 111)
`(e ⫽ 8)
`(c ⫽ 152)
`0.959 (0.014)
`(n ⫽ 26)
`(e ⫽ 8)
`(c ⫽ 237)
`0.920 (0.040)
`(n ⫽ 22)
`(e ⫽ 9)
`(c ⫽ 240)
`0.859 (0.070)
`(n ⫽ 13)
`(e ⫽ 10)
`(c ⫽ 248)
`
`n/a
`
`0.972 (0.008)
`(n ⫽ 344)
`(e ⫽ 11)
`(c ⫽ 88)
`0.947 (0.012)
`(n ⫽ 251)
`(e ⫽ 19)
`(c ⫽ 173)
`0.917 (0.019)
`(n ⫽ 94)
`(e ⫽ 23)
`(c ⫽ 326)
`0.907 (0.022)
`(n ⫽ 76)
`(e ⫽ 24)
`(c ⫽ 343)
`0.879 (0.029)
`(n ⫽ 46)
`(e ⫽ 26)
`(c ⫽ 371)
`
`0.972 (0.010)
`(n ⫽ 276)
`(e ⫽ 8)
`(c ⫽ 3)
`0.906 (0.018)
`(n ⫽ 230)
`(e ⫽ 26)
`(c ⫽ 31)
`0.867 (0.021)
`(n ⫽ 151)
`(e ⫽ 35)
`(c ⫽ 101)
`0.836 (0.031)
`(n ⫽ 30)
`(e ⫽ 37)
`(c ⫽ 220)
`0.752 (0.054)
`(n ⫽ 27)
`(e ⫽ 40)
`(c ⫽ 220)
`0.752 (0.054)
`(n ⫽ 18)
`(e ⫽ 40)
`(c ⫽ 229)
`
`0.971 (0.008)
`(n ⫽ 434)
`(e ⫽ 13)
`(c ⫽ 4)
`0.900 (0.014)
`(n ⫽ 381)
`(e ⫽ 44)
`(c ⫽ 26)
`0.876 (0.016)
`(n ⫽ 282)
`(e ⫽ 54)
`(c ⫽ 115)
`0.822 (0.024)
`(n ⫽ 98)
`(e ⫽ 62)
`(c ⫽ 291)
`0.805 (0.026)
`(n ⫽ 93)
`(e ⫽ 64)
`(c ⫽ 294)
`0.805 (0.026)
`(n ⫽ 61)
`(e ⫽ 64)
`(c ⫽ 326)
`
`1.000 (0)
`(n ⫽ 210)
`(e ⫽ 0)
`(c ⫽ 0)
`0.990 (0.007)
`(n ⫽ 195)
`(e ⫽ 2)
`(c ⫽ 13)
`0.959 (0.015)
`(n ⫽ 120)
`(e ⫽ 7)
`(c ⫽ 83)
`0.877 (0.041)
`(n ⫽ 29)
`(e ⫽ 12)
`(c ⫽ 169)
`0.784 (0.063)
`(n ⫽ 22)
`(e ⫽ 15)
`(c ⫽ 173)
`0.784 (0.063)
`(n ⫽ 14)
`(e ⫽ 15)
`(c ⫽ 181)
`
`1.000 (0)
`(n ⫽ 372)
`(e ⫽ 0)
`(c ⫽ 0)
`0.997 (0.003)
`(n ⫽ 354)
`(e ⫽ 1)
`(c ⫽ 17)
`0.972 (0.009)
`(n ⫽ 257)
`(e ⫽ 9)
`(c ⫽ 106)
`0.952 (0.014)
`(n ⫽ 104)
`(e ⫽ 13)
`(c ⫽ 255)
`0.933 (0.019)
`(n ⫽ 90)
`(e ⫽ 15)
`(c ⫽ 267)
`0.922 (0.022)
`(n ⫽ 48)
`(e ⫽ 16)
`(c ⫽ 308)
`
`any secondary intervention and are listed in Table II. The
`presumed causes of sac enlargement and the indication for
`any secondary intervention are listed in Table IV. Certain
`differences were noted with respect to morphologic out-
`come stratified by physiologic risk. A log-rank test showed
`a greater risk of late sac enlargement in high-risk patients.
`Specific factors that may have contributed to this effect
`were then assessed with a Cox regression model (Table V,
`
`online version only). Individual factors conducive to in-
`creased risk of sac enlargement (P ⬍ .05) included ad-
`vanced age, lower body weight, female sex, larger initial
`aneurysm size, presence of cancer, and inclusion in the
`high-risk cohort. Anticoagulation status (aspirin, clopi-
`dogrel or warfarin) was not a factor associated with an
`increased risk of enlargement. An additional analysis with a
`multivariate step-wise Cox regression model revealed that
`
`

`

`6 Greenberg et al
`
`JOURNAL OF VASCULAR SURGERY
`July 2008
`
`model showed that advanced age and internal iliac artery
`occlusion are joint predictors of an increased risk of SER.
`
`DISCUSSION
`The initial report of the ZMT provided evidence that
`endovascular repair of AAA with amenable anatomy was
`superior to open surgery for both associated morbidities
`and aneurysm-related death.3 The conclusions were tem-
`pered by the expectation that endovascular repair would
`require more detailed follow-up than open repair and
`might not provide as durable a result. The data presented in
`this article extend our confidence regarding longer-term
`durability of endovascular repair with the Zenith AAA
`device through 5 years. The observation in 736 implants of
`only a single rupture (at 6 months), a freedom from open
`surgical conversion of ⬎98%, and freedom from aneurysm-
`related death of 98% in standard-risk patients and 96% in
`high-risk patients support the long-term durability of this
`device. Yet as with most evaluations of endovascular treat-
`ments, it is the detailed analysis that will provide clues of
`whether a sustained benefit can be expected.
`Regretfully, we have no data on the surgical control
`group beyond 1 year, precluding any long-term compari-
`sons against surgical controls. Several previously reported
`adverse endovascular outcomes, such as limb thrombosis,
`stent fracture, and component separation, were so uncom-
`mon that further statistical analyses seeking causative fac-
`tors or effect were not fruitful. However, other observa-
`
`Fig 2. Kaplan-Meier graph shows mortality stratified by stan-
`dard-risk (black line) and high-risk (red line) physiologic groups.
`Error bars represent 95% confidence intervals at each time point.
`
`advanced patient age and larger initial aneurysm size are
`joint predictors of an increased risk of sac enlargement.
`Suboptimal endovascular result analysis. Individual
`factors related to increased risk of SER were evaluated with
`a Cox regression model (Table VI, online only). The factors
`predictive of increased risk of SER (P ⬍ .05) included
`advanced age, presence of iliac involvement in the aneu-
`rysm, internal iliac artery occlusion, and smaller neck diam-
`eter. An additional multivariate step-wise Cox regression
`
`Table III. Pooled data (standard- and high-risk groups) for complications relating to the stent graft expressed as
`Kaplan-Meier estimates, with standard errors listed in the parentheses
`
`Exam period
`
`1-month
`
`12-month
`
`24-month
`
`36-month
`
`48-month
`
`60-month
`
`Freedom from event
`Limb occlusion Migration ⬎5 mm Migration ⬎10 mm Barb separation
`
`Stent fracture
`
`Component separation
`
`0.988 (0.004)
`(n ⫽ 721)
`(e ⫽ 9)
`(c ⫽ 9)
`0.981 (0.005)
`(n ⫽ 662)
`(e ⫽ 14)
`(c ⫽ 63)
`0.974 (0.006)
`(n ⫽ 494)
`(e ⫽ 18)
`(c ⫽ 227)
`0.974 (0.006)
`(n ⫽ 155)
`(e ⫽ 18)
`(c ⫽ 566)
`0.974 (0.006)
`(n ⫽ 152)
`(e ⫽ 18)
`(c ⫽ 569)
`0.974 (0.006)
`(n ⫽ 106)
`(e ⫽ 18)
`(c ⫽ 615)
`
`n/a
`
`n/a
`
`0.995 (0.003)
`(n ⫽ 568)
`(e ⫽ 3)
`(c ⫽ 145)
`0.977 (0.007)
`(n ⫽ 400)
`(e ⫽ 12)
`(c ⫽ 304)
`0.955 (0.012)
`(n ⫽ 128)
`(e ⫽ 17)
`(c ⫽ 571)
`0.955 (0.012)
`(n ⫽ 110)
`(e ⫽ 17)
`(c ⫽ 589)
`0.935 (0.019)
`(n ⫽ 71)
`(e ⫽ 19)
`(c ⫽ 626)
`
`0.999 (0.001)
`(n ⫽ 568)
`(e ⫽ 1)
`(c ⫽ 147)
`0.999 (0.001)
`(n ⫽ 400)
`(e ⫽ 1)
`(c ⫽ 315)
`0.996 (0.003)
`(n ⫽ 128)
`(e ⫽ 2)
`(c ⫽ 586)
`0.996 (0.003)
`(n ⫽ 110)
`(e ⫽ 2)
`(c ⫽ 604)
`0.996 (0.003)
`(n ⫽ 71)
`(e ⫽ 2)
`(c ⫽ 643)
`
`1.000 (0)
`(n ⫽ 705)
`(e ⫽ 0)
`(c ⫽ 26)
`0.994 (0.003)
`(n ⫽ 583)
`(e ⫽ 4)
`(c ⫽ 144)
`0.980 (0.006)
`(n ⫽ 402)
`(e ⫽ 11)
`(c ⫽ 318)
`0.965 (0.010)
`(n ⫽ 132)
`(e ⫽ 15)
`(c ⫽ 584)
`0.935 (0.018)
`(n ⫽ 115)
`(e ⫽ 19)
`(c ⫽ 597)
`0.918 (0.021)
`(n ⫽ 69)
`(e ⫽ 21)
`(c ⫽ 641)
`
`1.000 (0)
`(n ⫽ 705)
`(e ⫽ 0)
`(c ⫽ 26)
`1.000 (0)
`(n ⫽ 587)
`(e ⫽ 0)
`(c ⫽ 144)
`0.998 (0.002)
`(n ⫽ 410)
`(e ⫽ 1)
`(c ⫽ 320)
`0.993 (0.006)
`(n ⫽ 137)
`(e ⫽ 2)
`(c ⫽ 592)
`0.985 (0.010)
`(n ⫽ 123)
`(e ⫽ 3)
`(c ⫽ 605)
`0.968 (0.015)
`(n ⫽ 72)
`(e ⫽ 5)
`(c ⫽ 654)
`
`1.000 (0)
`(n ⫽ 705)
`(e ⫽ 0)
`(c ⫽ 26)
`1.000 (0)
`(n ⫽ 587)
`(e ⫽ 0)
`(c ⫽ 144)
`0.994 (0.03)
`(n ⫽ 408)
`(e ⫽ 3)
`(c ⫽ 320)
`0.991 (0.04)
`(n ⫽ 137)
`(e ⫽ 4)
`(c ⫽ 590)
`0.977 (0.011)
`(n ⫽ 123)
`(e ⫽ 6)
`(c ⫽ 602)
`0.977 (0.011)
`(n ⫽ 72)
`(e ⫽ 6)
`(c ⫽ 653)
`
`n, patients at risk; e, cumulative events; c, cumulative censored; n/a, not applicable.
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 48, Number 1
`
`Greenberg et al 7
`
`Fig 3. The incidence of both new and persistent endoleaks by type and time of onset is shown for high-risk (red) and
`standard-risk (black) groups. The Kaplan-Meier plot shows the cumulative risk for a late endoleak of any type. Error
`bars represent 95% confidence intervals at each time point. No statistical differences were noted between the high- and
`standard-risk groups, although the numbers of patients at late follow-up time points were relatively small.
`
`tions such as endoleak, sac enlargement, and secondary
`interventions merit further discussion.
`The Zenith endovascular graft has shown good long-
`term device durability in this study. Component separation
`was rare: Only three patients experiencing graft limb sepa-
`ration through 5 years. In two cases, the uncovered top
`stent had disconnected from the graft material, which
`required placement of proximal extensions with new un-
`covered proximal fixation systems. These occurrences,
`along with a small number of cases in Europe and Australia,
`were responsible for a design modification in 2002 to
`double the suture-mediated attachment between the top
`stent and the proximal margin of the graft. None of the
`modified devices in the continued access arm have shown
`signs of top stent disconnection.
`Barb separations were also noted and previously reported3
`and still have not been associated with any clinical sequelae.
`Subsequent to this study, the barbs were increased in
`diameter—from 0.009 inches to 0.011 inches—to reduce
`the likelihood of separations.
`
`Minor device changes occurred after the conclusion of
`the US trial and included an increased gap between the first
`three z-stents to improve neck conformability, the addition
`of a 36-mm-diameter device, and the implementation of a
`hydrophilic-coated sheath and a modified valve (Captor
`Valve, Cook) in an effort to improve delivery.
`The overall incidence of endoleaks, sac enlargement,
`and migration were favorable in comparison with other
`multicenter device trials,26,27 with a low incidence of pri-
`mary endoleaks and an overall incidence of late endoleaks of
`⬍4% at every annual time point. The risk of sac enlarge-
`ment was approximately 1.6% for the standard-risk patients
`at each annual time point, and the cumulative risk for
`migration was ⬍1% for all patients.
`Type I and III endoleaks were notably absent after
`correction of problems relating to the initial implantation.
`As calculated from life-table estimates, 12% to 14% of the
`patients are at risk for the development of late endoleaks ⱕ5
`years of implantation. Reassuringly, most late leaks in this
`study were type II in nature, but the mechanism by which
`
`

`

`8 Greenberg et al
`
`JOURNAL OF VASCULAR SURGERY
`July 2008
`
`Table IV. Data for patients that had evidence of
`aneurysmal sac size increase (⬎5 mm) or any secondary
`intervention
`
`Event
`
`Aneurysm enlargement
`Potential cause
`Total patients, No.
`Persistent endoleak
`Type II
`Other
`Graft infection
`Proximal neck dilation
`Unidentified
`Secondary interventions
`Total
`Conversion
`Endoleak
`Proximal type I
`Distal type I
`Type IIa
`Type III
`Multiple
`Graft kink, limb stenosis or occlusion
`Top stent detachment
`Renal (angiography, angioplasty or stenting)
`Distal embolization
`Other
`Peripheral vascular
`Thoracic dissection
`Thrombosis of dialysis graft
`Diagnostic Angiogram for Endoleak
`
`No.
`
`31
`
`20
`5
`2
`1
`3
`
`153
`8
`96
`5
`10
`67
`11
`3
`21
`2
`14
`2
`10
`7
`1
`1
`1
`
`aType II endoleaks were the most frequent etiology of enlargement and the
`most common secondary intervention. The need for such interventions and
`method by which they were done was left to the discretion of the treating
`physician.
`
`they occur is uncertain. It is possible that late endoleaks
`occurred in an intermittent manner or simply were not
`appreciated on earlier imaging studies, but those are un-
`likely explanations for all late leaks. Alternatively, late type
`II endoleaks may represent spontaneous recanalization of
`lumbar or the inferior mesenteric arteries.
`Although many endovascular enthusiasts consider type
`II endoleaks entirely benign, ruptures relating to type II
`endoleaks13 and the incontrovertible link between en-
`doleaks and sac enlargement11 cannot be ignored. Of the
`31 patients who experienced sac enlargement during the
`trial, 25 were related to endoleaks (early or late), 80% of
`which were type II in nature. Not surprisingly, the treat-
`ment of early and late type II endoleaks accounts for about
`two-thirds of all secondary interventions. The association
`of sac enlargement with high physiologic risk has been
`previously noted10 and is confirmed in this report.