Mechanical failure of prosthetic human implants:
`A 10-year experience with aortic stent graft
`devices
`
`Tikva S. Jacobs, MD, Jamie Won, BS, Edwin C. Gravereaux, MD, Peter L. Faries, MD,
`Nicholas Morrissey, MD, Victoria J. Teodorescu, MD, Larry H. Hollier, MD, and Michael L.
`Marin, MD, New York, NY
`
`Objective: The first endovascular stent graft was implanted to treat an abdominal aortic aneurysm more than a decade ago.
`This technique has evolved dramatically with the growing understanding of metallurgic and fabric sciences and improved
`device designs. However the potential for stent graft material failure remains. This investigation describes the incidence
`of material failure, potential modes of device fatigue, and the clinical significance of these failures.
`Methods: Six hundred eighty-six endovascular stent grafts were used to treat patients with aortic aneurysms. Device fatigue
`in the form of stent, suture fracture, or graft wear was identified with an analysis of follow-up radiographs and explanted
`stent grafts. A review of patient clinical histories, spiral computed tomographic scan studies, scanning electron
`microscopy, and energy dispersion spectroscopy of explanted devices was conducted to evaluate the modes and
`consequences of failure.
`Results: Sixty patients were identified with device fatigue, 49 of whom had abdominal endovascular repairs and 11 of
`whom had thoracic repairs. Of the 60 patients with stent graft fatigue, 43 patients had metallic stent fractures, 14 had
`suture disruptions, and three had graft holes. These material failures occurred within seven distinct stent graft designs.
`The average time to the recognition of failure was 19 months, with a mean follow-up period of 8 months since the event
`was identified. Eleven patients died, and one was lost to follow-up 2 years after identification of a stent fracture. The
`remaining patients are presently being followed eoyj physical examination, plain film radiograph, and computed
`tomographic scans for clinical sequelae of device fatigue.
`Conclusion: Endovascular stent graft fatigue has been recognized in numerous devices after aortic implantation. Fatigue
`may take the form of stent, graft, or suture failure, with certain modes unique to specific stent graft devices. The clinical
`significance of stent graft material failure remains uncertain. (J Vasc Surg 2003;37:16-26.)
`
`The first endovascular stent graft repair to treat an
`abdominal aortic aneurysm was performed by Parodi,
`Palmaz, and Barone1 more than a decade ago. An estimated
`greater than 25,000 aortic stent grafts have since been
`deployed worldwide, and although preliminary results have
`been promising, problems with deployment, stent graft
`migration, endoleak, material failure, and aneurysm rup-
`ture have all been reported.2-7 Many of these problems
`were seen with first generation stent grafts. Fundamental
`mechanical and technical device problems have been ad-
`dressed, and individual
`implants have been improved.
`However, as midterm results of second-generation endo-
`vascular grafts are reported, new problems are discovered.
`
`From the Division of Vascular Surgery, Department of Surgery, Mount Sinai
`School of Medicine.
`Competition of interest: nil.
`Presented at the Fiftieth Annual Meeting of the American Association for
`Vascular Surgery, Boston, Mass, Jun 9-12, 2002.
`Reprint requests: Dr Michael L. Marin, Division of Vascular Surgery,
`Department of Surgery, Mount Sinai Medical Center, 5 E 98th St, Box
`1259, New York, NY 10029-6574
`(e-mail: michael.marin@
`mountsinai.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.2003.58
`16
`
`Device fatigue remains one of the most concerning
`modes for potential procedure failure, encompassing the
`breakdown of the intrinsic mechanical parts of the stent
`graft. It is often difficult to identify device fatigue because
`patients are typically asymptomatic at the time of presenta-
`tion. This difficulty has made a true understanding of the
`magnitude of the problem challenging. Moreover, the clin-
`ical significance of many identified material failures is un-
`known. The purpose of this study was to analyze a large
`single experience of endovascular aortic repairs to identify
`the incidence of device fatigue and the potential modes of
`fatigue and to try and determine the clinical significance of
`these material failures.
`
`METHODS
`Study population
`Six-hundred eighty-six patients underwent endovascu-
`lar aortic aneurysm repair over a 10-year period, 39 of
`which were implanted by the senior author while attending
`at the Montefiore Medical Center and the remaining 647
`performed at the Mount Sinai Medical Center. Four hun-
`dred four of these patients had a full compliment of fol-
`low-up analyses for review and form the study set for this
`investigation. A total of 60 patients (15%) of this subset had
`stent graft fatigue as identified with radiographic studies or
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 1
`
`Jacobs et al 17
`
`Table I. Distribution of stent graft fatigue by device
`
`Device for aortic
`aneurysm repair
`
`Total
`implanted
`
`Radiographs
`reviewed
`
`Total fatigue/
`fracture
`
`Average time to fracture/
`fatigue (mo; range)
`
`Average follow-up since
`fracture/fatigue (mo; range)†
`
`Abdominal
`Vanguard
`Talent
`Modified Parodi
`EVT/Ancure
`AneuRx
`Gore
`Teramed
`Thoracic
`Gore TAG
`Talent
`Total
`
`26
`337
`164
`9/20
`39
`18
`10
`
`22
`41‡
`686
`
`22 (85%)
`232 (69%)
`24 (15%)
`7/6
`33 (85%)
`18 (100%)
`10 (100%)
`
`19 (86%)
`33 (80%)
`404
`
`16 (72%)
`24 (10%)
`5 (21%)
`1/0 (14%)
`3 (10%)
`0
`0
`
`7 (37%)
`4 (12%)
`60
`
`*Excluding patient with acute conversion.
`†Excluding those patients who underwent open conversion and stent graft explanation.
`‡Including emergent use not part of clinical study.
`
`Table II. Failure mode analysis by device
`
`26* (3-48)
`13 (1-31)
`38 (33-48)
`8
`10 (1-24)
`
`24 (3-38)
`9.5 (1-24)
`19
`
`13 (1-39)
`5 (1-12)
`6 (1-8)
`24 and then lost to follow-up
`3 (1-6)
`
`12 (1-42)
`4 (2-7)
`8
`
`Device
`
`AAA
`Talent
`
`Vanguard
`
`EVT
`Modified
`Parodi
`
`AneuRx
`TAA
`Gore
`
`Talent
`
`No.
`fatigued
`
`Location
`
`No.
`
`Average time to fatigue
`(mo; range)
`
`49
`24
`
`16
`
`1
`5
`
`3
`11
`7
`
`4
`
`Graft hole
`Metal stent fracture
`Longitudinal bar
`Aortic body: proximal
`Aortic body: distal
`Iliac
`Proximal transrenal stent
`Z-shaped body stent
`Suture disruption
`Row separation
`Body separation
`Graft hole
`Hooks and shanks
`Metal stent fracture
`
`Top row of stent
`Second row
`Metal stent fracture
`
`Metal stent fracture
`Isolated longitudinal bar
`Longitudinal bar and z-stent
`Z-stent
`Metal stent fracture
`Z-stent
`Longitudinal bar
`
`1
`23
`14
`7
`1
`6
`5
`4
`14
`5
`9
`2
`1
`5
`
`4
`1
`3
`
`7
`3
`2
`2
`4
`3
`1
`
`16
`13 (1-31)
`
`25 (3-48)
`
`Intraoperative & 36
`8
`38 (33-48)
`
`10 (1-24)
`
`24 (3-38)
`
`9.5 (1-24)
`
`AAA, Abdominal aortic aneurysm; TAA, thoracic aortic aneurysm.
`
`the analysis of an explanted stent graft device. Patients were
`followed at 1, 3, 6, and 12 months and annually thereafter,
`in accordance with the specific US Food and Drug Admin-
`istration protocols for each device, with physical examina-
`tion, plain film radiographs, duplex ultrasonography, and
`spiral computed tomographic (CT) scans. The median
`follow-up period was 30 months, with a range from 1 to 52
`months. The information obtained from these visits and the
`results of the radiologic studies were analyzed for evidence
`
`of device fatigue. Device fatigue was defined as an actual
`change in the mechanical nature of the device, such as
`suture disruption, metal fracture, or fabric erosion. Proce-
`dure failure was defined as a repressurization of the aneu-
`rysm sac or rupture of the aneurysm.
`
`Techniques for endovascular repair
`The three techniques used for aortic aneurysm repair
`included: tubular (aortoaortic), aortouniiliac, bifurcated-
`
`

`

`18 Jacobs et al
`
`JOURNAL OF VASCULAR SURGERY
`January 2003
`
`Fig 1. Clinical examples of fatigue in Talent endovascular system. A, After persistent type I endoleak, 70-year-old man
`had endovascular graft explanted and conventional repair completed. Explanted graft showed signs of graft wear with
`frayed fabric yarns and creation of defined hole (arrow). B, Proximal longitudinal bar fracture in Talent endovascular
`graft (arrow). C, Distal aortic longitudinal bar fracture (arrow). D, Ipsilateral limb longitudinal bar fracture in Talent
`endovascular graft. (Continued.)
`
`modular, and bifurcated-unibody. All of these techniques
`have been demonstrated in detail previously.8-13
`
`Devices for abdominal aortic aneurysm and thoracic
`aortic aneurysm repair
`Seven different devices were used to treat patients with
`abdominal aortic aneurysms, and two for thoracic aortic
`aneurysms. Extensive descriptions have been provided in
`the literature of the nine devices used: Talent,14,15 Van-
`
`guard,14,16 Parodi-Modified Endovascular stent graft,13
`EVT,10,17 AneuRx,18 Gore Excluder,19 Cordis Quantum
`LP,20 and the Talent and Gore thoracic devices.21
`
`Stent graft analysis
`Radiographic studies. Plain film radiographs and CT
`scans were performed at 1, 6, and 12 months and annually
`thereafter. CT scanning consisted of a noncontrast exami-
`nation followed by a dynamic CT angiogram to evaluate for
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 1
`
`Jacobs et al 19
`
`Fig 1, continued. E, Midbody z stent fracture (arrow). F, Proximal transrenal stent fracture in Talent endovascular
`graft. Note fracture has occurred adjacent to site of nitinol wire crimp, which contains two ends of z-configured stent.
`
`endoleaks or changes in aneurysm size. Plain films were
`used to analyze the stent components of the stent graft.
`Initially two views, anteroposterior and lateral, were used;
`however, over the past 2 years, right posterior oblique and
`left anterior oblique views have been added to improve
`visualization of the structural components of the device
`Explant analysis. Five grafts with device fatigue were
`available for explant analysis after open surgical conversion.
`The device was first inspected macroscopically for evidence
`of stent fracture, fabric tears, or suture breaks, and then
`representative regions of the metal stent were studied with
`scanning electron microscopy (SEM).
`Preparation of specimens for SEM. Explanted stent
`grafts were rinsed of remaining debris and fixed in buffered
`formaldehyde. Each specimen was rinsed in distilled water
`and air dried for 24 hours. Representative regions of the
`metallic stent were ultrasonicated in a mild cleaning agent
`for 10 minutes, rinsed in distilled water, and ultrasonicated
`in absolute alcohol for another 10 minutes. The specimens
`were mounted on SEM stubs with silver paint. A Hitachi
`S530 SEM was used to evaluate the metal surface for
`defects and to analyze the site of fracture for evidence of
`fatigue. Energy dispersion spectroscopy (EDS) analysis was
`used at 15 kV on selected specimens to determine the
`surface composition of the metal and to assess for metal
`leaching.
`
`RESULTS
`Sixty patients (49 men and 11 women) were identified
`with device fatigue, 55 on radiographic analysis and five on
`explantation. Forty-nine of the devices with fatigue were
`seen in abdominal aortic aneurysm stent grafts, and 11 were
`found in thoracic aortic stent grafts. The 60 fatigued stent
`grafts in this study were distributed among seven different
`
`devices (Table I). The average time to fatigue for all devices
`combined was 19 months (range, 1 to 48 months), with an
`average follow-up period since fatigue identification of 8
`months. A marked variation in the time to failure was seen
`depending on the type of fatigue (ie, suture fracture, graft
`holes, or stent fractures; Table II). Eleven of the 60 patients
`followed with device fatigue have died, and one patient was
`lost to follow-up 2 years after fatigue was identified. Of the
`11 deaths, three were device related, and the remaining eight
`patients died of causes unrelated to their aortic aneurysms.
`Excluding one early postoperative death, the average time to
`death was 29 months (range, 17 to 52 months) after implan-
`tation and 13 months (range, 2 to 26 months) after fatigue
`was first identified. The remaining patients continue to be
`followed for clinical sequelae of their stent graft fatigue.
`
`Stent graft fatigue analyzed by device
`Abdominal aorta stent grafts. Twenty-four of the 60
`patients with device fatigue had Talent grafts inserted (Ta-
`ble II). Fatigue was recognized within several different
`regions of the stent graft. One patient had a wear hole
`detected in an explanted prosthesis at the site of graft to
`stent fixation (Fig 1, A). The remaining Talent endografts
`had fractured stents. Fourteen fractures occurred along the
`longitudinal bar of the graft (Fig 1, B to D), and nine in the
`serpentine shaped nitinol (Fig 1, E, F).
`Device fatigue was identified in 16 Vanguard abdomi-
`nal aortic aneurysm stent grafts; 14 had evidence of suture
`disruption, and two had fabric wear holes (Fig 2, A to D).
`The remaining nine fatigues were metal fractures: one EVT
`device (Fig 3), five modified Parodi endovascular stent
`grafts (Fig 4), and three AneuRx devices (Fig 5). Metal
`fatigue within AneuRx grafts was difficult to conclusively
`identify on plain film examination because of overlapping
`
`

`

`20 Jacobs et al
`
`JOURNAL OF VASCULAR SURGERY
`January 2003
`
`Fig 2. Clinical examples of fatigue in Boston Scientific Vanguard endovascular graft. Sixty-nine-year-old man had
`tube graft inserted for repair of abdominal aortic aneurysm. A, Proximal row separation was defined on this graft along
`with distal endoleak caused by retraction of prosthesis. B, Graft was explanted, and conventional repair was done at 36
`months. Arrow points to site of proximal row separation. C, High-powered magnification of graft depicted in Fig 9, B,
`shows fabric fatigue and wear holes on prosthesis surface. D, Seventy-nine-year-old man had endovascular tube graft
`placed for repair of abdominal aortic aneurysm. On surveillance abdominal radiographs, suture fractures in body of graft
`are detected (arrow). This patient remains in surveillance program.
`
`metal densities of the stent and the relatively close diamond
`pattern.
`Thoracic aorta stent grafts. Eleven of the 60 cases
`with device fatigue were metal fractures in thoracic aortic
`stent graft devices: seven Gore TAG and four Talent.
`Although the seven TAG devices had fractures in either the
`longitudinal bar, the z-shaped stent, or both (Fig 6), the
`
`four Talent thoracic aortic aneurysm devices had fractures
`in the z-shaped stents only.
`
`Endoleaks
`Endoleaks were identified in 24 of the 60 patients
`(40%) with device fatigue at some point during the fol-
`low-up period (Table III). Five (21%) had evidence of
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 1
`
`Jacobs et al 21
`
`Fig 3. Attachment system of EVT endovascular grafting system.
`Note fracture of hook shank at site of its laser weld on z-configured
`attachment system (arrow).
`
`fatigue before the endoleak. One patient had proximal and
`distal metallic fractures in a first generation EVT stent graft,
`and the second patient had a row separation caused by
`suture fracture of a Vanguard device that later developed a
`type I endoleak from dislodgment of the distal left limb.
`The third patient also had a row separation of a Vanguard
`tube device and 7 months later was noted to have a type I
`endoleak from the proximal and the distal attachment sites.
`The fourth patient was noted to have a body separation of
`a bifurcated Vanguard device that went on to develop a
`type III endoleak from a separation of the modular iliac
`limb. The final patient was noted to have a type II endoleak
`18 months after a fracture of the third row of the metallic
`stents in the AneuRx device was identified.
`Eight patients (33%) had stent graft fatigue develop
`after the endoleak was diagnosed, and 11 fatigued grafts
`(46%) were identified at the same time as an endoleak.
`Sixteen patients needed treatment for their endoleaks, two
`had spontaneous resolution, and six are currently being
`monitored (Table III).
`Explants
`Five of the six patients who underwent open conversion
`had the stent grafts explanted. One patient underwent
`conversion at the time of the primary stent graft procedure
`for an inadvertent iliac artery rupture with insertion of a
`second device to seal a type III endoleak. The remaining
`four patients had a new endoleak and an enlarging aortic
`aneurysm. The average time to explantation was 28 months
`(range, 15 to 36 months).
`
`SEM and EDS analysis
`SEM analysis was conducted on all five explanted pros-
`theses. Special attention was given to regions of the stent
`graft where corrosion would be expected to form (ie,
`crimped area of the ends of the nitinol wire stent in the
`Talent device, the platinum-nitinol interface of the Van-
`guard stent graft, and the high stressed angled regions of
`zig-zag nitinol stents; Fig 7, A to C). There was minimal
`evidence of corrosion on all the specimens examined. The
`
`Fig 4. Palmaz balloon expandable proximal stent attachment de-
`vice of handmade Parodi-Palmaz system. Note fracture of proximal
`diamond row (arrow).
`
`stent wire was covered with a relatively uniform oxide
`surface layer (Fig 7, D) except for an occasional pit. These
`pits were analyzed with EDS, and an apparent decrease in
`the nickel concentration on the metal surface was noted.
`The expected balanced ratio of titanium to nickel (51%:
`49%) was commonly seen in the area immediately subjacent
`to the regions with defects (Fig 8, A, B) One explanted
`TAG device with a metal fracture was analyzed; the fracture
`surface showed cleavage lines within the radius of the stent
`typical of stress fatigue (Fig 9).
`
`DISCUSSION
`Since the first endovascular stent graft was introduced by
`Parodi, Palmaz, and Barone1 over a decade ago, advance-
`ments in stent graft materials, designs, and deployment tech-
`niques have all contributed to the rapid growth and utility of
`these devices. Despite enhanced understanding of the metal-
`lurgic and fabric properties of endovascular stent grafts, mate-
`rial failure continues to be a potential problem. The inherent
`properties of the materials (strength and corrosion resistance)
`combined with extrinsic forces contribute significantly to the
`risk of device fatigue. Before deployment, and while still in the
`delivery sheath, the metallic stent may experience increased
`risk for failure based on damage during loading and subse-
`quent confinement in the delivery catheter. Once implanted,
`the device is then subjected to additional extrinsic forces
`imposed by the geometry of the tortuous aorta and the impact
`of continuous, high-pressure blood flow. In this study of 60
`patients with stent graft fatigue, 43 of the patients had metallic
`stent fractures, 14 had suture-stent disruptions, and three had
`graft holes. It is useful to analyze these three failure modes
`before speculating on the overall clinical significance of stent
`graft fatigue.
`
`

`

`22 Jacobs et al
`
`JOURNAL OF VASCULAR SURGERY
`January 2003
`
`Fig 5. Abdominal radiograph of patient who underwent AneuRx
`endografting of abdominal aortic aneurysm. Arrow denotes site of
`fracture in one of nitinol rings on stent strut.
`
`Fig 6. High-resolution explant radiograph of Gore TAG stent
`graft that was used to repair thoracic aortic aneurysm. This patient
`was seen 32 months after endovascular repair of thoracic aneurysm
`with new endoleak. Endoprosthesis was explanted, and conven-
`tional reconstruction was successfully accomplished. Explanted
`examination of stent graft with plain film radiograph shows z-
`configured wire fractures (arrowheads) and longitudinal bar frac-
`ture (arrow).
`
`Table III. Incidence of endoleaks in grafts with device fatigue
`
`Device
`
`AAA
`Vanguard
`
`Talent
`
`EVT
`Modified
`Parodi
`AneuRx
`TAA
`Gore
`
`Talent
`
`Time of diagnosis in relation to fatigue
`
`Endoleak #
`
`Before
`
`After
`
`Same time
`
`Management
`
`20
`9
`
`7
`
`1
`1
`
`1
`4
`2
`
`2
`
`2
`
`5
`
`1
`
`3
`
`1
`
`1
`
`4
`
`2
`
`2
`
`2
`
`5-Endovascular repair
`2-Open repair
`2-Observation
`1-Embolization of type II endoleak
`3-Open repair
`2-Resolved spontaneously
`1-Observation
`1-MEGS A-I-F
`1-Endovascular repair
`
`1-Observation
`
`1-Open repair
`1-Observation
`1-Endovascular repair
`1-Refused surgery, died of rupture
`
`AAA, Abdominal aortic aneurysm; TAA, thoracic aortic aneurysm; MEGS A-I-F, Mount Sinai endovascular stent graft aorto-iliac-femoral.
`
`Metallic fracture
`The 43 metallic stent fractures analyzed in this study
`occurred in devices fabricated by five different manufactur-
`ers. Thirty-seven fractures were documented in superelastic
`nitinol (nickel and titanium) stents, five fractures in stain-
`less steel (316L) devices, and one in an elgiloy (cobalt-
`chromium, nickel) stent.
`
`Elgiloy fractures were initially reported in the hooks of
`first generation EVT devices during the early phase of the
`EVT trial.17 Even though the hooks were remodeled to a
`more gradual angle to decrease the stress applied across the
`hook, Najibi et al22 reported two cases of hook fractures
`identified 36 months after implantation of the remodeled
`Ancure device. Although the EVT device in this study had
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 1
`
`Jacobs et al 23
`
`Fig 7. SEM of explanted endovascular grafts for abdominal aortic aneurysm. A, Vanguard endovascular graft has
`platinum wire wound around proximal and distal portions of attachment system to improve radiograph visualization
`(40⫻). B, Despite suspected increased potential for metal fatigue and corrosion at interface of these two metals, no such
`fatigue was found in nitinol in this 36-month-old implant (200⫻). C, SEM of Talent stent explanted 15 months after
`insertion. Site depicts crimped region of two ends of metal bar. No significant wear or pitting was detected on surface
`of nitinol stent at this presumably high risk area for fatigue (40⫻). D, High-powered (4K) magnification of surface of
`Talent nitinol spring after explantation. Rough-appearing, uniform oxide layer is discerned.
`
`similar fractures to the ones reported in the literature, there
`was no evidence of fractures in the remodeled Ancure
`system. However, the mean follow-up for patients with
`Ancure devices in this study was only 13 months.
`Stress fatigue is just one of the causes postulated for metal
`fractures in these devices. Metal corrosion has been identified
`on SEM studies of explanted Stentor grafts, with more severe
`corrosive irregularities detected in those devices implanted for
`longer period of time.23,24 However, SEM failed to show
`significant signs of corrosion in the nitinol stents of this study
`and, in fact, revealed a relatively uniform oxide layer on the
`surface of the metal stents, which may provide resistance to
`corrosion. The lack of corrosion in this study compared with
`the results of the earlier Stentor models may reflect the im-
`proved understanding of nitinol processing and the impor-
`tance of surface treatments, such as electropolishing,25 tita-
`nium nitride annealing,26 heat treatments, and nitric acid
`passivation27 to create a thin, uniform oxide layer to help
`prevent corrosion and increase biocompatibility and long-
`term durability of the metal.28
`Overall structural
`integrity of these devices can be
`affected by morphologic characteristics of the recipient
`vessel, changes in the aneurysm shape over time, or cyclic
`
`loading from aortic pulsations.29,30 Kinking of the graft can
`occur from these stresses and has been associated with limb
`occlusion and device or modular component migration in
`the Stentor/Vanguard model29 and with metal stent frac-
`ture of the longitudinal bars in the Talent device.30 In this
`report, 18 fractures occurred in the longitudinal nitinol
`bars of Talent or Gore stent grafts. These events were
`associated with tortuosity of the implant vessels with pre-
`sumed increased stress across the nitinol wire.
`The remaining metal stent fractures occurred in Palmaz
`stainless steel stents (n ⫽ 5) and high stressed angled
`regions of nitinol z-shaped stents (n ⫽ 19). The cyclic load
`from the aortic pulsations can generate enough stress to
`cause microcracks, material irregularities on the surface of
`metals produced during laser cutting, to propagate and
`eventually fracture. Unfortunately, limited in vivo informa-
`tion exists on fatigue crack propagation for endovascular
`stents. Studies to date have focused on orthopedic or heart
`valve implants, which are larger devices and have different
`loading patterns.31 Although the metals used today have
`high crack propagation thresholds, the strut widths are
`thin, approximating 250 ␮m, thereby leaving a small dis-
`tance necessary for a microcrack to travel before fracture.32
`
`

`

`24 Jacobs et al
`
`JOURNAL OF VASCULAR SURGERY
`January 2003
`
`Fig 8. EDS of explanted Talent graft. A, Marked decrease in nickel peak is discerned in area containing defect in
`relatively uniform oxide layer. B, In subjacent region to that irregular area described in A, normal titanium to nickel
`distribution is identified.
`
`Fabric fatigue
`All five explanted stent grafts had evidence of fabric
`fatigue. Although degradation of conventional polyester
`grafts has been reported, it usually occurs 10 to 20 years
`after implantation.33 The fabric wear seen in endovascular
`stent grafts occurs much earlier. The pulsatile flow of the
`aorta and the configuration of the stent graft allows for
`micromotion of the individual metal stents against the
`fabric leading to eventual breakdown and graft wear. Fabric
`fatigue was one of the first reported causes of aneurysm
`rupture and stent graft failure in the earlier Stentor graft.2
`External abrasions of the fabric on the metal stent and frank
`graft holes have been described.24,33,34
`In the Stentor and Vanguard devices, the stent is at-
`tached to the graft at the proximal and distal ends only
`allowing for continuous motion between the stent and the
`graft. Fewer examples of graft fenestrations have been
`reported with the Talent design, whose stent is completely
`fixed to the graft with sutures, thereby reducing the move-
`ment between the graft and the nitinol stent. In the one
`patient whose device was explanted immediately on open
`conversion and in which fabric fatigue was seen, micromo-
`tion of the Vanguard device is unlikely to have caused the
`
`failure. Fatigue had to have occurred during manufacturing
`or packaging of the device into the deployment catheter.
`Theoretically, if the stent graft is drawn into the catheter
`from the opposite direction from which it will be deployed,
`the angled points of the stents can press firmly against the
`graft fabric, potentially causing a small tear or hole.
`
`Suture breakage
`Polypropylene sutures that are used to assemble the Sten-
`tor/Vanguard device may also be subject to fatigue. Micro-
`motion of the individual nitinol stents causes friction and wear
`of the sutures, with ultimate suture fracture and stent row
`separation. As of January 2001, there has been a 21% preva-
`lence rate of row separation reported in the Vanguard mod-
`el.35 Of the 14 suture disruptions that were identified in this
`report, five were row separations and nine occurred in the
`body; this finding is in support of Riepe’s hypothesis that there
`is more motion in the larger frames of the body stents.3
`Besides the intrinsic properties of the materials used for
`stent graft fabrication and the extrinsic hostile environment
`of the native aorta, other factors could play a role in device
`fatigue. Twenty-four of the 60 patients (40%) in this report
`were diagnosed with endoleaks during the follow-up pe-
`
`

`

`JOURNAL OF VASCULAR SURGERY
`Volume 37, Number 1
`
`Jacobs et al 25
`
`asymptomatic and there is no evidence of aneurysm enlarge-
`ment, rupture, or a type I or III endoleak, observation of the
`stent graft fatigue is acceptable in the setting of increased graft
`surveillance.
`
`CONCLUSION
`Endovascular grafts have been clinically used to treat
`aortic aneurysms for more than 10 years. A significant growth
`in the understanding of the failure modes of the materials used
`to fabricate these devices and a respect for the relatively harsh
`environment into which they must function has led to the
`development of improved stent graft designs. Endovascular
`graft devices may experience metal fracture, fabric erosion,
`and ultimate device fatigue as a result of these two compo-
`nents. Stent graft packaging into delivery catheters, steriliza-
`tion, and shelf aging may all further impact the occurrence and
`timing of device integrity failure. Improved manufacturing
`practices can decrease the frequency of these fatigue events;
`however, clinical implications of many forms of device fatigue
`remains to be defined.
`
`We thank Mr Norman Katz for technical assistance
`with the scanning electron microscopy portion of this in-
`vestigation and Mr Joseph Samet for help with the photo-
`graphic preparation of this manuscript.
`
`REFERENCES
`
`1. Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft im-
`plantation for abdominal aortic aneurysms. Ann Vasc Surg 1991;5:491-9.
`2. Beebe HG, Cronenwett JL, Katzen BT, Brewster DC, Green RM.
`Results of an aortic endograft trial: impact of device failure beyond 12
`months. J Vasc Surg 2001;33:S55-63.
`3. Riepe G, Heilberger P, Umschield T, Chakfe N, Raithel D, Stelter W, et
`al. Frame dislocation of body middle rigs in endovascular stent tube
`grafts. Eur J Vasc Endovasc Surg 1999;17:28-34.
`4. Bohm T, Soldner J, Rott A, Kaiser WA. Perigraft leak of an aortic stent
`graft due to material fatigue. AJR Am J Roentgenol 1999;172:1355-7.
`5. Norgren L, Jernby B, Engellau L. Aortoenteric fistula caused by a
`ruptured stent-graft: a case report. J Endovasc Surg 1998;5:269-72.
`6. Maleux G, Rousseau H, Otal P, Colombier D, Glock Y, Joffre F.
`Modular component separation and reperfusion of abdominal aortic
`aneurysm sac after endovascular repair of the abdominal aortic aneu-
`rysm: a case report. J Vasc Surg 1998;28:349-52.
`7. Holzenbein TJ, Kretschmer G, Thrunher S, Schoder M, Aslim E,
`Lammer J, et al. Midterm durability of abdominal aortic aneurysm
`endograft repair: a word of caution. J Vasc Surg 2001;33:S46-54.
`8. Parodi JC, Marin ML, Veith FJ. Transfemoral, endovascular stented graft
`repair of an abdominal aortic aneurysm. Arch Surg 1995;130:549-52.
`9. Marin ML, Hollier LH, Avrahami R, Parsons R. Varying strategies for
`endovascular repair of abdominal and iliac artery aneurysms. Surg Clin
`North Am 1998;78:631-45.
`10. Moore WS. The EVT tube and bifurcated endograft systems: technical
`considerations and clinical summary. J Endovasc Surg 1997;4:182-94.
`11. Dake MD, Miller C, Semba CP, Mitchell S, Walker PJ, Liddell RP.
`Transluminal placement of endovascular stent-grafts for the treatment of
`descending thoracic aortic aneurysm. N Engl J Med 1994;331:1729-34.
`12. Chuter TA, Green RM, Ouriel K, Fiore WM, Deweese JA. Transfemo-
`ral aortic graft placement. J Vasc Surg 1993;18:185-97.
`13. Faries PL. Endovascular grafts for the treatment of abdominal aortic
`aneurysm. In: Marin ML, Hollier LH, editors. Endovascular grafting:
`advanced treatment for vascular disease. Armonk (NY): Futura Publish-
`ing Co, Inc; 2000. p. 73-94.
`14. May J, White GH, Harris JP. Devices for aortic aneurysm repair. Surg
`Clin North Am 1999;79:507-27.
`
`Fig 9. Free surface of a fractured nitinol wire from Gore TAG
`supporting stent explanted 32 months after thoracic aortic aneu-
`rysm repair. Note cleavage planes and irregular surface at step-off
`point of this fatigue fracture (200⫻).
`
`riod. The precise time of device failure and endoleak for-
`mation is often impossible to determine; however, only five
`patients were found to have metallic fractures before the
`onset of the endoleak. The remaining 19 patients had an
`endoleak develop followed by metallic fracture or had both
`events simultaneously discovered. Endoleaks have been
`shown to maintain pressure and turbulent flow in the
`aneurysm sac, leading to eventual enlargement of the an-
`eurysm and risk for rupture.6,37 The residual aneurysm sac
`pressure may be communicated to the stent graft itself in
`the form of increased device pulsatility that can lead to
`metallic fracture, suture disruption, and fabric wear, result-
`ing in stent graft device fatigue.
`Although most authors agree in the significance of a type
`I or type III endoleak contributing to aneurysm dilatation and
`risk of eventual rupture, the role a type II endoleak has is less
`clear.38-40 In this study, there were seven type II endoleaks, six
`of which were diagnosed before the stent graft fracture. All six
`of these endoleaks were observed on average 20 months
`(range, 6 to 32 months) before diagnosis of the stent graft
`fracture. One can hypothesize that although a type II en-
`doleak may not generate enough pressure to cause aneurysm
`dilatation and rupture, that there is enough pulsatile flow to
`expose the stent graft to excess micromotion, which in turn
`could lead to device fatigue.
`
`Clinical significance
`Although fractures were found in 60 patients with endo-
`vascular stent grafts, most of the pat