Comparison of Second-GenerationStents for Application in the SuperficialFemoral Artery: An In Vitro Evaluation

Focusing on Stent DesignMuller-Hulsbeck S, Schafer PJ, Charalambous N,

Yagi H, Heller M, Jahnke TJournal of Endovascular Therapy 2010;17(6):767–776.

¤EXPERIMENTAL INVESTIGATION ¤

Comparison of Second-Generation Stents forApplication in the Superficial Femoral Artery:An In Vitro Evaluation Focusing on Stent Design

Stefan Muller-Hulsbeck, MD1; Philipp J. Schafer, MD2; Nikolas Charalambous, MD2;Hiroshi Yagi3; Martin Heller, MD2; and Thomas Jahnke, MD2

1Department of Diagnostic and Interventional Radiology/Neuroradiology, AcademicHospitals Flensburg, Germany. 2Department of Radiology, University HospitalsSchleswig-Holstein – Campus Kiel, Germany. 3Terumo, Tokyo, Japan.

¤ ¤Purpose: To examine and compare in an ex vivo study different nitinol stent designsintended for the superficial femoral artery (SFA) with regard to the appearance of fracture.Methods: Seven different 8-340-mm nitinol stents were evaluated (Misago, Absolute,Smart, Luminexx, Sentinol, Lifestent NT, and Sinus-Superflex). Finite element analysis(FEA) was used for digitalized stent design comparison; the strain during stent movementwas calculated for bending, compression, and torsion. Additional mechanical fatigue testsfor bending (70u), compression (40%), and torsion (twisted counterclockwise by 180u) wereperformed up to 650,000 cycles or until a fracture was observed.Results: The FEA bending test showed that only the Misago, LifeStent, and Absolute stentspresented no zones of high strain; in the torsion test, the Smart stent also had no zones ofhigh strain. Macroscopic evaluation after mechanical bending indicated that the LifeStentperformed the best (no stent fracture after 650,000 cycles). Misago and Absolute stentsshowed fractures at 536,000 cycles and 456,667 cycles, respectively (range 320,000–650,000cycles). After compression and torsion testing, Misago showed no stent fracture after650,000 cycles. The worst performing stent was Luminexx during all test cycles.Conclusion: The 7 SFA stents showed differences in the incidence of high strain zones,which indicates a potential for stent fracture, as demonstrated by the mechanical fatiguetests. Differences in stent design might play a major role in the appearance of stent strutfracture related to restenosis and reocclusion.

J Endovasc Ther. 2010;17:767–776

Key words: superficial femoral artery, stent, nitinol, experimental study, finite elementanalysis, stent design, fatigue testing, stent fracture, compression, bending, torsion

¤ ¤

Endovascular stenting is a rapidly evolvingtechnology for treatment of either arterial orvenous obstruction, as well aneurysmal dis-ease, in the circulatory system. Several clin-ical reports and trials evaluated the effect ofstents in the superficial femoral artery (SFA).

In this dedicated segment, balloon-expand-able stents were not superior to balloonangioplasty alone,1 and self-expanding stentsfailed to show a beneficial effect in shortlesions ,5 cm in length.2 For longer lesions,trials report a benefit for stents compared to

This study was supported by an educational grant from Terumo Corporation, Tokyo, Japan.

Hiroshi Yagi is an employee of Terumo Corporation. The other authors have no commercial, proprietary, or financialinterest in any products or companies described in this article.

Address for correspondence and reprints: Stefan Muller-Hulsbeck, MD, Department of Diagnostic and InterventionalRadiology/Neuroradiology, Academic Hospitals Flensburg, Ev.-Luth. Diakonissenanstalt zu Flensburg, Knuthstrasse 1,24939 Flensburg, Germany. E-mail: muehue@diako.de

J ENDOVASC THER2010;17:767–776 767

� 2010 by the INTERNATIONAL SOCIETY OF ENDOVASCULAR SPECIALISTS Available at www.jevt.org

¤EXPERIMENTAL INVESTIGATION ¤

Comparison of Second-Generation Stents forApplication in the Superficial Femoral Artery:An In Vitro Evaluation Focusing on Stent Design

Stefan Muller-Hulsbeck, MD1; Philipp J. Schafer, MD2; Nikolas Charalambous, MD2;Hiroshi Yagi3; Martin Heller, MD2; and Thomas Jahnke, MD2

1Department of Diagnostic and Interventional Radiology/Neuroradiology, AcademicHospitals Flensburg, Germany. 2Department of Radiology, University HospitalsSchleswig-Holstein – Campus Kiel, Germany. 3Terumo, Tokyo, Japan.

¤ ¤Purpose: To examine and compare in an ex vivo study different nitinol stent designsintended for the superficial femoral artery (SFA) with regard to the appearance of fracture.Methods: Seven different 8-340-mm nitinol stents were evaluated (Misago, Absolute,Smart, Luminexx, Sentinol, Lifestent NT, and Sinus-Superflex). Finite element analysis(FEA) was used for digitalized stent design comparison; the strain during stent movementwas calculated for bending, compression, and torsion. Additional mechanical fatigue testsfor bending (70u), compression (40%), and torsion (twisted counterclockwise by 180u) wereperformed up to 650,000 cycles or until a fracture was observed.Results: The FEA bending test showed that only the Misago, LifeStent, and Absolute stentspresented no zones of high strain; in the torsion test, the Smart stent also had no zones ofhigh strain. Macroscopic evaluation after mechanical bending indicated that the LifeStentperformed the best (no stent fracture after 650,000 cycles). Misago and Absolute stentsshowed fractures at 536,000 cycles and 456,667 cycles, respectively (range 320,000–650,000cycles). After compression and torsion testing, Misago showed no stent fracture after650,000 cycles. The worst performing stent was Luminexx during all test cycles.Conclusion: The 7 SFA stents showed differences in the incidence of high strain zones,which indicates a potential for stent fracture, as demonstrated by the mechanical fatiguetests. Differences in stent design might play a major role in the appearance of stent strutfracture related to restenosis and reocclusion.

J Endovasc Ther. 2010;17:767–776

Key words: superficial femoral artery, stent, nitinol, experimental study, finite elementanalysis, stent design, fatigue testing, stent fracture, compression, bending, torsion

¤ ¤

Endovascular stenting is a rapidly evolvingtechnology for treatment of either arterial orvenous obstruction, as well aneurysmal dis-ease, in the circulatory system. Several clin-ical reports and trials evaluated the effect ofstents in the superficial femoral artery (SFA).

In this dedicated segment, balloon-expand-able stents were not superior to balloonangioplasty alone,1 and self-expanding stentsfailed to show a beneficial effect in shortlesions ,5 cm in length.2 For longer lesions,trials report a benefit for stents compared to

This study was supported by an educational grant from Terumo Corporation, Tokyo, Japan.

Hiroshi Yagi is an employee of Terumo Corporation. The other authors have no commercial, proprietary, or financialinterest in any products or companies described in this article.

Address for correspondence and reprints: Stefan Muller-Hulsbeck, MD, Department of Diagnostic and InterventionalRadiology/Neuroradiology, Academic Hospitals Flensburg, Ev.-Luth. Diakonissenanstalt zu Flensburg, Knuthstrasse 1,24939 Flensburg, Germany. E-mail: muehue@diako.de

J ENDOVASC THER2010;17:767–776 767

� 2010 by the INTERNATIONAL SOCIETY OF ENDOVASCULAR SPECIALISTS Available at www.jevt.org

balloon angioplasty alone.3 In particular, thislatter data has opened the door to a moreliberal and widespread use of stents in theSFA. Unfortunately, with longer stented seg-ments, the incidence of stent fractures ap-pears to rise, though the reported fracturerates for today’s commercially available de-signs differ widely from 1.8% to 18%.4–8 Evenif clinically relevant complications of stentfractures rarely occur (other than restenosisor reocclusion), it is important to gather moreinformation about the potential risk of stentfracture. In addition to clinical data, theoret-ical and experimental studies are neededprior to the release of new products.

The optimization of material property, sur-face finish, and stent design are necessary forthe development of a durable stent. Manyself-expanding stents share the same basicconstruction: the stents are laser-cut from anickel-titanium alloy tube, and the surface ofthe stent is electropolished to make it smoothand corrosion-resistant. However, the stentdesign is quite variable among products andmay play a significant role in the durability ofthe stent. It is expected that an optimal stentdesign could reduce the potential risk of astent fracture. Therefore, we used two meth-ods of analysis, finite element analysis (FEA)and mechanical fatigue testing, to identify thepotential fracture risks of some currentlyavailable SFA stent designs.

METHODS

Seven different nitinol stents measuring8-340-mm were evaluated: Misago (Terumo,Tokyo, Japan), Absolute (Abbott Vascular,Santa Clara, CA, USA), Smart (Cordis Endo-vascular, Miami Lakes, FL, USA), Luminexx(C.R. Bard, Tempe, AZ, USA), Sentinol (Bos-ton Scientific, Natick, MA, USA), Lifestent NT(C.R. Bard), and Sinus-Superflex (Optimed,Ettlingen, Germany). At least one samplefrom each manufacturer was tested (Table 1).

Stent Design

The basic stent design consisted of arepeating pattern of zigzag units linked to-gether. The geometry of the zigzag or peak-to-valley design differed among models, as well

as the bridges or transition zones connectingthe zigzag elements. The cell size dependedon the number of parallel peak-to-valley areasand the number of connections; fewer bridg-es between zigzags made the stent designmore open and flexible.

Misago’s zigzags were not straight; theyincluded a short segment that included 2large angled bridges located in the center of azigzag to serve as a transition zone to the nextzigzag and cell area, respectively. The zigzagdesign of the Absolute stent was very similar:straight bridges from one peak to anotherpeak parallel to the stent axis connected thezigzags.

The zigzags and short peak-to-valley bridg-es of the Smart stent were in a slightlyelongated S-shaped curve not parallel to thestent axis. The Luminexx stent had a zigzagdesign that was similar to that of the Smartstent, but the peaks and valleys were con-nected directly without any bridges. Sentinolalso had short peak-to-valley bridges, withzigzags similar to the Smart and Luminexxstents. Lifestent was quite similar in terms ofdesign to the Smart stent. Sinus-Superflexconsisted of elongated zigzag elements withan additional elongation in the mid part of azigzag element; short peak-to-valley bridgesorientated vertical to the stent axis connectedthe zigzags.

Finite Element Analysis

FEA is a method of finding approximatesolutions to partial differential equations andintegral equations. In this stent comparison,digitized models of the stent designs wereanalyzed, and the strains during stent move-ment were calculated according to the 3-dimensional (3D) solid element and hexahe-dral 8-node element method (ADINA ver. 8.5;ADINA R&D, Watertown, MA, USA). To keepthe analysis runtime reasonable, a finite-element mesh consisting of 3 elements inthe stent strut width direction and 2 elementsin the stent wall thickness direction was used.The material properties of each stent designwere obtained from strain-stress curves ofdog bone–shaped samples produced by man-ufacturing processes equivalent to the stent(laser-cut, heat set, electropolish). These

768 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

material data were applied for all involvedstent samples so the analysis accuratelyrepresented the differences in the fracture-resistant characteristics associated with thestent design. The nitinol material propertiesand the surface finish were ignored, as thealloy was the same for all stents.

For bending, the stent was fixed at 15 mmfrom the proximal edge and the distal edgewas pressed down by 12 mm. To evaluatecompression, one end of the stent was fixedand the other end was compressed along thelong axis of the stent until the length became20% shorter than the original length. Finally,for torsion, one end of the stent was fixed andthe other end was twisted counterclockwiseby 60u. The strain that the stent receivedunder each of these conditions was calculat-ed; high strain zones (,0.9% strain or more)appeared in red on the simulations. For semi-quantitative evaluation, red zones greaterthan a stent strut thickness were counted asone high strain area. Stents with no red zonesas defined above were indicated by a ‘‘2’’,

stents with ,10 red zones were given a ‘‘+’’,and stents with $10 red zones were markedas ‘‘++’’.

Mechanical Fatigue Testing

All tests were done in a regulated atmo-sphere at 37uC on actual 8-340-mm stents.Each test described below was repeated up toa maximum of 650,000 cycles. The occurrenceof the stent fracture was checked at 1,000,5,000, 10,000, 15,000, 20,000 and 25,000cycles, then every 10,000 cycles from 30,000to 100,000 and every 20,000 cycles from100,000 to the 650,000 maximum. The testwas continued until a fracture was observedor the maximum was reached.

Bending. The center point of the stentsample was placed between two 6-mm-thicksupports spaced 6 mm apart. The proximalend of the stent was fixed, and the distal endat 15 mm from the center point was flexeddownward by 40u and flexed upward by 30u(Fig. 1).

¤ ¤TABLE 1

High Strain Zones During Finite Element Analysis of Bending, Compression,and Torsion for the 7 Nitinol Stents

DesignNumbersTested Bending Compression Torsion

Misago Open cell zigzag units, and thecenter of the zigzags areconnected with 2 large angledbridges

15 — — —(n55) (n55) (n55)

Absolute Open cell zigzag; peak-to-peakconnected with straightbridges

7 — — —(n53) (n53) (n51)

Smart Open cell zigzag; peak-to-valleyconnected with S-shapedbridge

7 ++ ++ —(n53) (n53) (n51)

Luminexx Open cell zigzag; peak-to-valleyconnected directly

7 ++ ++ ++(n53) (n53) (n51)

Sentinol Open cell zigzag; peak-to-valleyconnected with short straightbridge

3 ++ ++ ++(n51) (n51) (n51)

LifeStent Open cell zigzag; similar toSmart

1/1/1 + — —(n51) (n51) (n51)

Sinus-SuperFlex Open cell zigzag; peak-to-valleyconnected with short bridges

1/1/1 + ++ ++(n51) (n51) (n51)

¤ ¤Stents with no red (strain) zones as defined in the text were indicated by a ‘‘2’’, stents with ,10 red zones weregiven a ‘‘+’’, and stents with $10 red zones were counted as ‘‘++’’.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 769Muller-Hulsbeck et al.

balloon angioplasty alone.3 In particular, thislatter data has opened the door to a moreliberal and widespread use of stents in theSFA. Unfortunately, with longer stented seg-ments, the incidence of stent fractures ap-pears to rise, though the reported fracturerates for today’s commercially available de-signs differ widely from 1.8% to 18%.4–8 Evenif clinically relevant complications of stentfractures rarely occur (other than restenosisor reocclusion), it is important to gather moreinformation about the potential risk of stentfracture. In addition to clinical data, theoret-ical and experimental studies are neededprior to the release of new products.

The optimization of material property, sur-face finish, and stent design are necessary forthe development of a durable stent. Manyself-expanding stents share the same basicconstruction: the stents are laser-cut from anickel-titanium alloy tube, and the surface ofthe stent is electropolished to make it smoothand corrosion-resistant. However, the stentdesign is quite variable among products andmay play a significant role in the durability ofthe stent. It is expected that an optimal stentdesign could reduce the potential risk of astent fracture. Therefore, we used two meth-ods of analysis, finite element analysis (FEA)and mechanical fatigue testing, to identify thepotential fracture risks of some currentlyavailable SFA stent designs.

METHODS

Seven different nitinol stents measuring8-340-mm were evaluated: Misago (Terumo,Tokyo, Japan), Absolute (Abbott Vascular,Santa Clara, CA, USA), Smart (Cordis Endo-vascular, Miami Lakes, FL, USA), Luminexx(C.R. Bard, Tempe, AZ, USA), Sentinol (Bos-ton Scientific, Natick, MA, USA), Lifestent NT(C.R. Bard), and Sinus-Superflex (Optimed,Ettlingen, Germany). At least one samplefrom each manufacturer was tested (Table 1).

Stent Design

The basic stent design consisted of arepeating pattern of zigzag units linked to-gether. The geometry of the zigzag or peak-to-valley design differed among models, as well

as the bridges or transition zones connectingthe zigzag elements. The cell size dependedon the number of parallel peak-to-valley areasand the number of connections; fewer bridg-es between zigzags made the stent designmore open and flexible.

Misago’s zigzags were not straight; theyincluded a short segment that included 2large angled bridges located in the center of azigzag to serve as a transition zone to the nextzigzag and cell area, respectively. The zigzagdesign of the Absolute stent was very similar:straight bridges from one peak to anotherpeak parallel to the stent axis connected thezigzags.

The zigzags and short peak-to-valley bridg-es of the Smart stent were in a slightlyelongated S-shaped curve not parallel to thestent axis. The Luminexx stent had a zigzagdesign that was similar to that of the Smartstent, but the peaks and valleys were con-nected directly without any bridges. Sentinolalso had short peak-to-valley bridges, withzigzags similar to the Smart and Luminexxstents. Lifestent was quite similar in terms ofdesign to the Smart stent. Sinus-Superflexconsisted of elongated zigzag elements withan additional elongation in the mid part of azigzag element; short peak-to-valley bridgesorientated vertical to the stent axis connectedthe zigzags.

Finite Element Analysis

FEA is a method of finding approximatesolutions to partial differential equations andintegral equations. In this stent comparison,digitized models of the stent designs wereanalyzed, and the strains during stent move-ment were calculated according to the 3-dimensional (3D) solid element and hexahe-dral 8-node element method (ADINA ver. 8.5;ADINA R&D, Watertown, MA, USA). To keepthe analysis runtime reasonable, a finite-element mesh consisting of 3 elements inthe stent strut width direction and 2 elementsin the stent wall thickness direction was used.The material properties of each stent designwere obtained from strain-stress curves ofdog bone–shaped samples produced by man-ufacturing processes equivalent to the stent(laser-cut, heat set, electropolish). These

768 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

material data were applied for all involvedstent samples so the analysis accuratelyrepresented the differences in the fracture-resistant characteristics associated with thestent design. The nitinol material propertiesand the surface finish were ignored, as thealloy was the same for all stents.

For bending, the stent was fixed at 15 mmfrom the proximal edge and the distal edgewas pressed down by 12 mm. To evaluatecompression, one end of the stent was fixedand the other end was compressed along thelong axis of the stent until the length became20% shorter than the original length. Finally,for torsion, one end of the stent was fixed andthe other end was twisted counterclockwiseby 60u. The strain that the stent receivedunder each of these conditions was calculat-ed; high strain zones (,0.9% strain or more)appeared in red on the simulations. For semi-quantitative evaluation, red zones greaterthan a stent strut thickness were counted asone high strain area. Stents with no red zonesas defined above were indicated by a ‘‘2’’,

stents with ,10 red zones were given a ‘‘+’’,and stents with $10 red zones were markedas ‘‘++’’.

Mechanical Fatigue Testing

All tests were done in a regulated atmo-sphere at 37uC on actual 8-340-mm stents.Each test described below was repeated up toa maximum of 650,000 cycles. The occurrenceof the stent fracture was checked at 1,000,5,000, 10,000, 15,000, 20,000 and 25,000cycles, then every 10,000 cycles from 30,000to 100,000 and every 20,000 cycles from100,000 to the 650,000 maximum. The testwas continued until a fracture was observedor the maximum was reached.

Bending. The center point of the stentsample was placed between two 6-mm-thicksupports spaced 6 mm apart. The proximalend of the stent was fixed, and the distal endat 15 mm from the center point was flexeddownward by 40u and flexed upward by 30u(Fig. 1).

¤ ¤TABLE 1

High Strain Zones During Finite Element Analysis of Bending, Compression,and Torsion for the 7 Nitinol Stents

DesignNumbersTested Bending Compression Torsion

Misago Open cell zigzag units, and thecenter of the zigzags areconnected with 2 large angledbridges

15 — — —(n55) (n55) (n55)

Absolute Open cell zigzag; peak-to-peakconnected with straightbridges

7 — — —(n53) (n53) (n51)

Smart Open cell zigzag; peak-to-valleyconnected with S-shapedbridge

7 ++ ++ —(n53) (n53) (n51)

Luminexx Open cell zigzag; peak-to-valleyconnected directly

7 ++ ++ ++(n53) (n53) (n51)

Sentinol Open cell zigzag; peak-to-valleyconnected with short straightbridge

3 ++ ++ ++(n51) (n51) (n51)

LifeStent Open cell zigzag; similar toSmart

1/1/1 + — —(n51) (n51) (n51)

Sinus-SuperFlex Open cell zigzag; peak-to-valleyconnected with short bridges

1/1/1 + ++ ++(n51) (n51) (n51)

¤ ¤Stents with no red (strain) zones as defined in the text were indicated by a ‘‘2’’, stents with ,10 red zones weregiven a ‘‘+’’, and stents with $10 red zones were counted as ‘‘++’’.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 769Muller-Hulsbeck et al.

Compression. The proximal end of the stentsample was fixed; the other end was com-pressed along the long axis of the stent untilthe overall stent length was 40% shorter thanthe original length and then returned to theoriginal position (Fig. 2).

Torsion. The proximal end of the stentsample was fixed and the other end wastwisted counterclockwise by 90u and thentwisted clockwise by 180u (90u from theoriginal position; Fig. 3).

Statistical Analysis

The data is presented as mean and range;due to the small sample sizes, no statisticalevaluation was performed.

RESULTS

Finite Element Analysis

In the bending tests (Table 1 and Fig. 4A),Misago and Absolute presented no zones ofhigh strain. LifeStent and Sinus-Superflexhad ,10 zones of high strain during bending.Sentinol (Fig. 4D), Smart (Fig. 4E), and Lumi-nexx (Fig. 4F) had $10 high strain areas. Inthe compression test (Fig. 4B), Misago, Ab-solute, and LifeStent presented no zones ofhigh strain. Sinus-Superflex had ,10 zones ofhigh strain during bending, while Smart,Sentinol, and Luminexx showed $10 highstrain areas. In the torsion tests (Fig. 4C),Misago, Absolute, Smart, and LifeStent dem-onstrated no zones of high strain, whileSentinol, Sinus-Superflex, and Luminexxshowed $10 high strain areas.

Fatigue Testing

In the mechanical bending tests (Table 2),macroscopic evaluation of the LifeStentshowed no stent fracture after 650,000 cycles(only 1 sample tested). One of 5 Misago stentsamples showed fracture after a mean536,000 cycles (range 80,000–650,000); 4 sam-ples had no fracture at 650,000 cycles. TheAbsolute stent had fractures at a mean456,667 cycles (range 320,000–650,000). Frac-tures appeared much earlier in the Luminexxstent (mean 2,415 cycles, range 2,000–3,000),Sinus-Superflex (at 5,000 cycles), Sentinol(mean 16,400 cycles, range 8,200–25,000),and Smart (mean 41,667 cycles, range25,000–50,000).

In the compression tests, macroscopicevaluation of Misago showed no stent frac-

Figure 1¤Schematic of the setup for the bendingtest.

Figure 2¤Schematic of the setup for the com-pression test.

Figure 3¤Schematic of the setup for the torsiontest.

770 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

ture after 650,000 cycles (5 samples tested).Again, fractures appeared earlier with Lumi-nexx (mean 1,000 cycles; all 3 samples withfractures), Smart (mean 3,182 cycles, range2,000–4,545), Sinus-Superflex (4,445 cycles),Sentinol (mean 6,515 cycles, range 4,545–10,000), Absolute (mean 23,333 cycles, range10,000–40,000), and LifeStent (48,995 cycles).

In the torsion tests, macroscopic evaluationof Misago showed no stent fracture after650,000 cycles (5 samples tested), whilefractures appeared earlier in the individualsamples of Luminexx (1,000 cycles), Sinus-Superflex (5,000 cycles), Smart (10,000cycles), Absolute (60,000 cycles), Sentinol(60,000 cycles), and LifeStent (60,000 cycles).

DISCUSSION

The poor performance of early SFA stents andthe complexity of the SFA have prompted thedevelopment of new stent designs, so-calledsecond-generation SFA stents. With thesededicated SFA stent designs, stent fractureshould no longer be an issue. The latest dataindicated low fracture rates ,5% during6 months of follow-up.3,7 Potential reasonsfor stents fractures have been put forth inseveral clinical evaluations that demonstratedchanges of the SFA during leg movement.Computerized fluoroscopy image evaluationof changes in the femoropopliteal segmentbetween the straight-leg (SL) and crossed-leg (CL) positions demonstrated significantchanges in length, curvature, and rotation inthe popliteal artery and significant but moremodest changes in length and rotation in theSFA during movement from the SL to the CLposition. The data showed mean shorteningof 6.1% and 15.8%, respectively, for the SFAand popliteal artery. The mean rotationangles for the SFA and popliteal artery were45.6u627.9u and 61.1u631.9u, respectively,and the mean flexion angles were 20.1u61.7uand 20.2u614.8u, respectively. The authorsconcluded that these data had importantimplications for endovascular therapies inthe femoropopliteal segment.

In other studies, Cheng et al.10 examinedhealthy young gymnasts and found the meanshortening of the SFA was 13%611% and amean rotation angle of 60u634u between the

supine and fetal positions. A study of oldersubjects showed mean shortening values of5.9%63.0%, 6.7%62.1%, and 8.1%62.0% inthe top, middle, and bottom of the SFA,respectively.11 The mean values of theseparameters are reasonable benchmark condi-tions for testing. However, these data arebased on the non-stented vessel; the investi-gator should consider the post-implant mor-phological changes in establishing appropri-ate boundary conditions.

The need for compression analysis wasdemonstrated by the assessment of magneticresonance angiography (MRA) datasets offemoropopliteal compression during isomet-ric thigh contraction.12 A 3D balanced steady-state free precession sequence was used toimage a 15- to 20-cm segment of the vascu-lature during relaxation and voluntary iso-metric thigh contraction. MRA of the femoro-popliteal segment during thigh contractiondemonstrated focal compression of the arte-rial segment in the distal adductor canalregion, which may also help explain the highstent failure rate and the high likelihood ofatherosclerotic disease in the adductor canal.

In our study, the main focus was oncomparing the performance limit to theappearance of fracture for each stent design,so more aggressive boundary conditionswere applied. In current mechanical fatiguetests, walking is generally considered afatigue cycle condition; 10 million cyclessimulates 10 years of walking. The fatiguecycles for more severe motion, such as sittingand stair climbing, are not standardized,but there are assumptive calculations13 onwhich we based our fatigue testing: sitting 5(365 d/y)3(12 h/d)3(60 min/h)3(1 cycle/15minutes)310 years 5 175,200 cycles; stairclimbing 5 (365 d/y)3(8 flights/d)3(32 stairs/flight)31 cycle/2 stairs)310 years 5 467,200cycles. For 10 years, the total for sitting andstair climbing is 642,400 cycles, so themaximum in our study was 650,000 cycles.

As shown in the DURABILITY trial, stentfracture rates may be operator dependent,i.e., a stent implanted in an elongated manner($10% over the nominal length) is more likelyto develop fractures.14 However, the long-term clinical significance of the fracturesremains unknown, but it is obvious that as

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 771Muller-Hulsbeck et al.

Compression. The proximal end of the stentsample was fixed; the other end was com-pressed along the long axis of the stent untilthe overall stent length was 40% shorter thanthe original length and then returned to theoriginal position (Fig. 2).

Torsion. The proximal end of the stentsample was fixed and the other end wastwisted counterclockwise by 90u and thentwisted clockwise by 180u (90u from theoriginal position; Fig. 3).

Statistical Analysis

The data is presented as mean and range;due to the small sample sizes, no statisticalevaluation was performed.

RESULTS

Finite Element Analysis

In the bending tests (Table 1 and Fig. 4A),Misago and Absolute presented no zones ofhigh strain. LifeStent and Sinus-Superflexhad ,10 zones of high strain during bending.Sentinol (Fig. 4D), Smart (Fig. 4E), and Lumi-nexx (Fig. 4F) had $10 high strain areas. Inthe compression test (Fig. 4B), Misago, Ab-solute, and LifeStent presented no zones ofhigh strain. Sinus-Superflex had ,10 zones ofhigh strain during bending, while Smart,Sentinol, and Luminexx showed $10 highstrain areas. In the torsion tests (Fig. 4C),Misago, Absolute, Smart, and LifeStent dem-onstrated no zones of high strain, whileSentinol, Sinus-Superflex, and Luminexxshowed $10 high strain areas.

Fatigue Testing

In the mechanical bending tests (Table 2),macroscopic evaluation of the LifeStentshowed no stent fracture after 650,000 cycles(only 1 sample tested). One of 5 Misago stentsamples showed fracture after a mean536,000 cycles (range 80,000–650,000); 4 sam-ples had no fracture at 650,000 cycles. TheAbsolute stent had fractures at a mean456,667 cycles (range 320,000–650,000). Frac-tures appeared much earlier in the Luminexxstent (mean 2,415 cycles, range 2,000–3,000),Sinus-Superflex (at 5,000 cycles), Sentinol(mean 16,400 cycles, range 8,200–25,000),and Smart (mean 41,667 cycles, range25,000–50,000).

In the compression tests, macroscopicevaluation of Misago showed no stent frac-

Figure 1¤Schematic of the setup for the bendingtest.

Figure 2¤Schematic of the setup for the com-pression test.

Figure 3¤Schematic of the setup for the torsiontest.

770 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

ture after 650,000 cycles (5 samples tested).Again, fractures appeared earlier with Lumi-nexx (mean 1,000 cycles; all 3 samples withfractures), Smart (mean 3,182 cycles, range2,000–4,545), Sinus-Superflex (4,445 cycles),Sentinol (mean 6,515 cycles, range 4,545–10,000), Absolute (mean 23,333 cycles, range10,000–40,000), and LifeStent (48,995 cycles).

In the torsion tests, macroscopic evaluationof Misago showed no stent fracture after650,000 cycles (5 samples tested), whilefractures appeared earlier in the individualsamples of Luminexx (1,000 cycles), Sinus-Superflex (5,000 cycles), Smart (10,000cycles), Absolute (60,000 cycles), Sentinol(60,000 cycles), and LifeStent (60,000 cycles).

DISCUSSION

The poor performance of early SFA stents andthe complexity of the SFA have prompted thedevelopment of new stent designs, so-calledsecond-generation SFA stents. With thesededicated SFA stent designs, stent fractureshould no longer be an issue. The latest dataindicated low fracture rates ,5% during6 months of follow-up.3,7 Potential reasonsfor stents fractures have been put forth inseveral clinical evaluations that demonstratedchanges of the SFA during leg movement.Computerized fluoroscopy image evaluationof changes in the femoropopliteal segmentbetween the straight-leg (SL) and crossed-leg (CL) positions demonstrated significantchanges in length, curvature, and rotation inthe popliteal artery and significant but moremodest changes in length and rotation in theSFA during movement from the SL to the CLposition. The data showed mean shorteningof 6.1% and 15.8%, respectively, for the SFAand popliteal artery. The mean rotationangles for the SFA and popliteal artery were45.6u627.9u and 61.1u631.9u, respectively,and the mean flexion angles were 20.1u61.7uand 20.2u614.8u, respectively. The authorsconcluded that these data had importantimplications for endovascular therapies inthe femoropopliteal segment.

In other studies, Cheng et al.10 examinedhealthy young gymnasts and found the meanshortening of the SFA was 13%611% and amean rotation angle of 60u634u between the

supine and fetal positions. A study of oldersubjects showed mean shortening values of5.9%63.0%, 6.7%62.1%, and 8.1%62.0% inthe top, middle, and bottom of the SFA,respectively.11 The mean values of theseparameters are reasonable benchmark condi-tions for testing. However, these data arebased on the non-stented vessel; the investi-gator should consider the post-implant mor-phological changes in establishing appropri-ate boundary conditions.

The need for compression analysis wasdemonstrated by the assessment of magneticresonance angiography (MRA) datasets offemoropopliteal compression during isomet-ric thigh contraction.12 A 3D balanced steady-state free precession sequence was used toimage a 15- to 20-cm segment of the vascu-lature during relaxation and voluntary iso-metric thigh contraction. MRA of the femoro-popliteal segment during thigh contractiondemonstrated focal compression of the arte-rial segment in the distal adductor canalregion, which may also help explain the highstent failure rate and the high likelihood ofatherosclerotic disease in the adductor canal.

In our study, the main focus was oncomparing the performance limit to theappearance of fracture for each stent design,so more aggressive boundary conditionswere applied. In current mechanical fatiguetests, walking is generally considered afatigue cycle condition; 10 million cyclessimulates 10 years of walking. The fatiguecycles for more severe motion, such as sittingand stair climbing, are not standardized,but there are assumptive calculations13 onwhich we based our fatigue testing: sitting 5(365 d/y)3(12 h/d)3(60 min/h)3(1 cycle/15minutes)310 years 5 175,200 cycles; stairclimbing 5 (365 d/y)3(8 flights/d)3(32 stairs/flight)31 cycle/2 stairs)310 years 5 467,200cycles. For 10 years, the total for sitting andstair climbing is 642,400 cycles, so themaximum in our study was 650,000 cycles.

As shown in the DURABILITY trial, stentfracture rates may be operator dependent,i.e., a stent implanted in an elongated manner($10% over the nominal length) is more likelyto develop fractures.14 However, the long-term clinical significance of the fracturesremains unknown, but it is obvious that as

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 771Muller-Hulsbeck et al.

772 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

the grade of the fracture worsens, from singlestrut fracture to multiple fractures and finallyloss of stent integrity, the incidences of .50%restenosis and occlusion may rise.

The progression of stent fracture is sup-posed to be a dynamic process that maybecome symptomatic in a case of completetransverse linear fracture. As the SFA travers-es the adductor canal, a particularly complexcombination of forces comes into play, inparticular, external compression, torsion,elongation, and flexion. These forces, togeth-er with manufacturing processes, will beresponsible for stent failure. Any potentialstent fracture would have tremendous impli-cations to the patient because stent platformsare so integral to cardiovascular treatment;clinicians and industry, therefore, have anobligation to look for stent refinements.Unfortunately, there is little standardizationin the processing, laser cutting, etching,

electropolishing, surface finishing, and fa-tigue testing of nitinol stents. Because of this,we attempted to gather more informationabout the influence of stent design onbending, compression, and torsion, as evalu-ated during FEA and mechanical fatigue tests.Physicians currently have no other way togather more information about stents andtheir potential behavior in terms of strutfracture. In order to obtain this missinginformation to complement clinical data ob-tained after stent implantation, identicalmethods for ex vivo testing would be helpful.

Nikanorov et al.15 also used in vitro fatiguetesting to characterize the types and ranges ofstent distortion theoretically produced byextremity movement; they used these rangesas parameters for in vitro long-term fatiguetesting of commercially available self-expand-ing nitinol stents. They placed stents (ProtegeEverFlex, Smart, Luminexx, LifeStent, Xceed,

r

Figure 4¤ Finite element analyses of all stents during (A) bending, (B) compression, and (C)torsion. Legend: 1: Misago, 2: Absolute, 3: Smart, 4: Luminexx, 5: Sentinol, 6: Lifestent NT,and 7: Sinus-Superflex. (D1) Finite element analysis of the Sentinol stent during bending,indicating a high strain zone at the zigzags adjacent to the short bridges, which show no highstrain areas. (D2) Photograph of a fracture in the corresponding stent area. (D3) The strutfracture cannot be seen on magnified high-resolution fluoroscopy. (E1) Finite elementanalysis of the Smart stent during bending, indicating high strain zones at the zigzagsadjacent to the short S-shaped bridges, which show no high strain areas. (E2) Photograph of afracture in the corresponding stent area. (E3) The strut fracture cannot be seen on magnifiedhigh-resolution fluoroscopy. (F1) Finite element analysis of the Luminexx stent duringbending, indicating high strain zones at the zigzags’ peaks and valley. (F2) Photograph of afracture in the corresponding stent area. (F3) The strut fracture cannot be seen on magnifiedhigh-resolution fluoroscopy.

¤ ¤TABLE 2

Mean Cycles to Stent Fracture During Mechanical Bending, Compression, and Torsion Testing

Numbers Tested: Bending/Compression/ Torsion Bending Compression Torsion

Misago 5/5/5 536,000 650,000* 650,000*Absolute 3/3/1 456,667 23,333 60,000Smart 3/3/1 41,667 3,182 10,000Luminexx 3/3/1 2,415 1,000 1,000Sentinol 3/3/1 16,400 6,515 60,000LifeStent 1/1/1 650,000* 48,995 60,000Sinus-SuperFlex 1/1/1 5,000 4,445 5,000¤ ¤

* Not broken at 650,000 cycles.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 773Muller-Hulsbeck et al.

772 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

the grade of the fracture worsens, from singlestrut fracture to multiple fractures and finallyloss of stent integrity, the incidences of .50%restenosis and occlusion may rise.

The progression of stent fracture is sup-posed to be a dynamic process that maybecome symptomatic in a case of completetransverse linear fracture. As the SFA travers-es the adductor canal, a particularly complexcombination of forces comes into play, inparticular, external compression, torsion,elongation, and flexion. These forces, togeth-er with manufacturing processes, will beresponsible for stent failure. Any potentialstent fracture would have tremendous impli-cations to the patient because stent platformsare so integral to cardiovascular treatment;clinicians and industry, therefore, have anobligation to look for stent refinements.Unfortunately, there is little standardizationin the processing, laser cutting, etching,

electropolishing, surface finishing, and fa-tigue testing of nitinol stents. Because of this,we attempted to gather more informationabout the influence of stent design onbending, compression, and torsion, as evalu-ated during FEA and mechanical fatigue tests.Physicians currently have no other way togather more information about stents andtheir potential behavior in terms of strutfracture. In order to obtain this missinginformation to complement clinical data ob-tained after stent implantation, identicalmethods for ex vivo testing would be helpful.

Nikanorov et al.15 also used in vitro fatiguetesting to characterize the types and ranges ofstent distortion theoretically produced byextremity movement; they used these rangesas parameters for in vitro long-term fatiguetesting of commercially available self-expand-ing nitinol stents. They placed stents (ProtegeEverFlex, Smart, Luminexx, LifeStent, Xceed,

r

Figure 4¤ Finite element analyses of all stents during (A) bending, (B) compression, and (C)torsion. Legend: 1: Misago, 2: Absolute, 3: Smart, 4: Luminexx, 5: Sentinol, 6: Lifestent NT,and 7: Sinus-Superflex. (D1) Finite element analysis of the Sentinol stent during bending,indicating a high strain zone at the zigzags adjacent to the short bridges, which show no highstrain areas. (D2) Photograph of a fracture in the corresponding stent area. (D3) The strutfracture cannot be seen on magnified high-resolution fluoroscopy. (E1) Finite elementanalysis of the Smart stent during bending, indicating high strain zones at the zigzagsadjacent to the short S-shaped bridges, which show no high strain areas. (E2) Photograph of afracture in the corresponding stent area. (E3) The strut fracture cannot be seen on magnifiedhigh-resolution fluoroscopy. (F1) Finite element analysis of the Luminexx stent duringbending, indicating high strain zones at the zigzags’ peaks and valley. (F2) Photograph of afracture in the corresponding stent area. (F3) The strut fracture cannot be seen on magnifiedhigh-resolution fluoroscopy.

¤ ¤TABLE 2

Mean Cycles to Stent Fracture During Mechanical Bending, Compression, and Torsion Testing

Numbers Tested: Bending/Compression/ Torsion Bending Compression Torsion

Misago 5/5/5 536,000 650,000* 650,000*Absolute 3/3/1 456,667 23,333 60,000Smart 3/3/1 41,667 3,182 10,000Luminexx 3/3/1 2,415 1,000 1,000Sentinol 3/3/1 16,400 6,515 60,000LifeStent 1/1/1 650,000* 48,995 60,000Sinus-SuperFlex 1/1/1 5,000 4,445 5,000¤ ¤

* Not broken at 650,000 cycles.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 773Muller-Hulsbeck et al.

and Absolute) in the SFA of cadavers. Lateralradiographs were obtained with the limb invarious degrees of hip and knee flexion. Thedegree of axial shortening and bending of thestent were estimated by planimetry and usedfor in vitro fatigue testing. In contrast to ourfatigue test method, the stents were mountedin elastic silicone tubing, bathed in phosphatebuffered saline at 37u62uC, and examined forfracture after 10 million cycles of chronicdeformation. For unstented arteries, the distalSFA/proximal popliteal artery had the great-est axial compression (23%) versus the mid-dle SFA (9%) or popliteal artery (14%) at 90u/90u knee/hip flexion. For stented arteries, thepopliteal artery exhibited the most axialcompression (11%) versus the middle SFA(3%) or distal SFA/proximal popliteal artery(6%) at 90u/90u knee/hip flexion. Axial com-pression of the stented popliteal artery at 70u/20u knee/hip flexion was 6%, with a deflectionangle of 33u. Based on these parameters,chronic in vitro fatigue testing produced arange of responses in commercially availablestents. Chronic 5% axial compression result-ed in high rates of fracture in Luminexx (80%)and LifeStent (50%), with lower fracture ratesfor Absolute (3%), Protege EverFlex (0%), andSmart (0%). Chronic 48u bending deformationresulted in high rates of fracture in ProtegeEverFlex (100%), Smart (100%), and Luminexx(100%), with lower rates in Absolute (3%) andLifeStent (0%). Even with lower bendingangle (70u), higher compression rate (40%),and an additional torsion component (180u,90u clockwise), our fatigue data match Nika-norv’s result for bending and torsion forSmart, Absolute, Luminexx, and Lifestent.This observation indicates that commerciallyavailable stents exhibit a variable ability towithstand chronic deformation in vitro, andtheir response is highly dependent on thetype of deformation applied.

In addition, we used computer-based anal-ysis to predict high strain areas during stentmovement (bending, compression, torsion).Our FEA findings match our mechanicalfatigue test results except for the Absolute(compression and torsion) and Smart (tor-sion) stents. Early fractures appeared withthese stents (in contrast to the FEA findings,which indicated no high strain areas). An

explanation for this mismatch might be thesmall number of test samples. Nevertheless,from our computer-simulated analysis ofbending and axial compression, it was clearthat the strain was concentrated in the zigzagunit with the bridge (Fig. 4D1). In the me-chanical fatigue bending test, fractures werealso observed at these points (Fig. 4D2). Withthe Smart stent, the strain was concentratedat the S-shaped bridges between the struts(Fig. 4E). The actual fractures (Fig. 4F) in theSmart stents were observed at that point inthe mechanical fatigue test as well, whichsupports the authenticity of the data. Thefractures observed near the bridge in bendingand axial compression conditions had ahigher risk of stent separation owing to lossof axial stent integrity, i.e., type 3 fracture.16

Therefore, it seems that bending and axialcompression could impart more critical stressto the stent than torsion.

With respect to design, a stent without thebridge on which the strain concentrateswould likely survive without failure becausedynamic loads are distributed equally to allstruts. As a result, Misago could be expectedto have a reduced risk of complete transversefracture since it does not have the connec-tions or bridges found in other stent designs.Optimizing stent link design would aid in thedevelopment of a dedicated stent for thepopliteal artery, where the devices are sub-jected to a more severe bending and/or axialcompression load than in the SFA.

Limitations

We are unaware of other reports utilizingFEA for evaluating stent design during simu-lated movement (stress), but other computerprograms used to undertake similar evalua-tions may produce different results. However,our study should be seen as only an attemptto gain insight into the connection betweenstent design and the potential for fracture invivo.

There are other limitations associated withthe design and construction of our ex vivosetting: a larger angle for bending, evaluatinglonger stent lengths, and stents not implantedinto a tube or another medium that mimics anin vivo situation. Finally, the fatigue tests

774 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

were performed under non-standardized con-ditions because these standardized methodsdo not exist. Another shortcoming was thesmall sample size for mechanical tests. Therewas also no consideration of material prop-erty or surface finish.

To our knowledge, pulsatile durability test-ing is standardized only by the AmericanSociety for Testing and Materials, but noindustry standard for in vitro stent fracturefatigue testing in the SFA exists; although itmay be difficult to reproduce the dynamics ofthe SFA, it is imperative that improvedstandards should be developed in all aspectsof nitinol stent processing and testing tobetter identify the failure modes of SFAstenting. Physicians should apply some pres-sure on the medical device industry tostandardize testing of stents and endopros-theses before marketing approval and clinicaluse.

In spite of these limitations, these simpletests offered a reproducible method of testingthe differences in stent design; the acquireddata gave us an indication of stent perfor-mance that might be expected in vivo. Evenwhen the FEA results match those of themechanical fatigue tests, the data have to beextrapolated with the greatest care beforebeing applied to a clinical situation. More-over, many stent strut structures are missedduring even high resolution fluoroscopy, aswe found in this study. Therefore, the cur-rently reported low fracture rates with sec-ond-generation SFA stents should be inter-preted cautiously.

Conclusion

The 7 nitinol SFA stents tested showeddifferences in the incidence of high strainzones that indicate the potential for stentfracture, which can be revealed only byfunctional investigation. Differences in stentdesign may play a major role in the occurrenceof restenosis and reocclusion related to stentstrut fractures. Mechanical bending has to beconsidered in future studies to assess theinfluence of differences in stent design, mate-rial, or even postinterventional drug treatmenton the long-term patency of stents and endo-prostheses in the femoropopliteal segment.

REFERENCES

1. Grimm J, Muller-Hulsbeck S, Jahnke T, et al.Randomized study to compare PTA aloneversus PTA with Palmaz stent placement forfemoropopliteal lesions. J Vasc Interv Radiol.2001;12:935–942.

2. Krankenberg H, Schluter M, Steinkamp HJ,et al. Nitinol stent implantation versus percu-taneous transluminal angioplasty in superficialfemoral artery lesions up to 10 cm in length:the femoral artery stenting trial (FAST). Circu-lation. 2007;116:285–292.

3. Schillinger M, Sabeti S, Loewe C, et al. Balloonangioplasty versus implantation of nitinolstents in the superficial femoral artery. N EnglJ Med. 2006;354:1879–1888.

4. Duda SH, Pusich B, Richter G, et al. Sirolimus-eluting stents for the treatment of obstructivesuperficial femoral artery disease: six-monthresults. Circulation. 2002;106:1505–1509.

5. Duda SH, Bosiers M, Lammer J, et al. Siroli-mus-eluting versus bare nitinol stent for ob-structive superficial femoral artery disease: theSIROCCO II trial. J Vasc Interv Radiol. 2005;16:331–338.

6. Schlager O, Dick P, Sabeti S, et al. Long-segment SFA stenting—the dark sides: in-stentrestenosis, clinical deterioration, and stentfractures. J Endovasc Ther. 2005;12:676–684.

7. Schulte KL, Muller-Hulsbeck S, Cao P, et al.MISAGO 1: first-in-man clinical trial with Misagonitinol stent. EuroIntervention. 2010;5:687–691.

8. Scheinert D, Scheinert S, Sax J, et al. Preva-lence and clinical impact of stent fractures afterfemoropopliteal stenting. J Am Coll Cardiol.2005;45:312–315.

9. Klein AJ, Chen SJ, Messenger JC, et al. Quanti-tative assessment of the conformational changein the femoropopliteal arterywith legmovement.Catheter Cardiovasc Interv. 2009;74:787–798.

10. Cheng CP, Wilson NM, Hallett RL, et al. In vivoMR angiographic quantification of axial andtwisting deformations of the superficial femo-ral artery resulting from maximum hip andknee flexion. J Vasc Interv Radiol. 2006;17:979–987.

11. Cheng CP, Choi G, Herfkens RJ, et al. The effectof aging on deformations of the superficialfemoral artery resulting from hip and kneeflexion: potential clinical implications. J VascInterv Radiol. 2010;21:195–202.

12. Brown R, Nguyen TD, Spincemaille P, et al. Invivo quantification of femoral-popliteal com-pression during isometric thigh contraction:assessment using MR angiography. J MagnReson Imaging. 2009;29:1116–1124.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 775Muller-Hulsbeck et al.

and Absolute) in the SFA of cadavers. Lateralradiographs were obtained with the limb invarious degrees of hip and knee flexion. Thedegree of axial shortening and bending of thestent were estimated by planimetry and usedfor in vitro fatigue testing. In contrast to ourfatigue test method, the stents were mountedin elastic silicone tubing, bathed in phosphatebuffered saline at 37u62uC, and examined forfracture after 10 million cycles of chronicdeformation. For unstented arteries, the distalSFA/proximal popliteal artery had the great-est axial compression (23%) versus the mid-dle SFA (9%) or popliteal artery (14%) at 90u/90u knee/hip flexion. For stented arteries, thepopliteal artery exhibited the most axialcompression (11%) versus the middle SFA(3%) or distal SFA/proximal popliteal artery(6%) at 90u/90u knee/hip flexion. Axial com-pression of the stented popliteal artery at 70u/20u knee/hip flexion was 6%, with a deflectionangle of 33u. Based on these parameters,chronic in vitro fatigue testing produced arange of responses in commercially availablestents. Chronic 5% axial compression result-ed in high rates of fracture in Luminexx (80%)and LifeStent (50%), with lower fracture ratesfor Absolute (3%), Protege EverFlex (0%), andSmart (0%). Chronic 48u bending deformationresulted in high rates of fracture in ProtegeEverFlex (100%), Smart (100%), and Luminexx(100%), with lower rates in Absolute (3%) andLifeStent (0%). Even with lower bendingangle (70u), higher compression rate (40%),and an additional torsion component (180u,90u clockwise), our fatigue data match Nika-norv’s result for bending and torsion forSmart, Absolute, Luminexx, and Lifestent.This observation indicates that commerciallyavailable stents exhibit a variable ability towithstand chronic deformation in vitro, andtheir response is highly dependent on thetype of deformation applied.

In addition, we used computer-based anal-ysis to predict high strain areas during stentmovement (bending, compression, torsion).Our FEA findings match our mechanicalfatigue test results except for the Absolute(compression and torsion) and Smart (tor-sion) stents. Early fractures appeared withthese stents (in contrast to the FEA findings,which indicated no high strain areas). An

explanation for this mismatch might be thesmall number of test samples. Nevertheless,from our computer-simulated analysis ofbending and axial compression, it was clearthat the strain was concentrated in the zigzagunit with the bridge (Fig. 4D1). In the me-chanical fatigue bending test, fractures werealso observed at these points (Fig. 4D2). Withthe Smart stent, the strain was concentratedat the S-shaped bridges between the struts(Fig. 4E). The actual fractures (Fig. 4F) in theSmart stents were observed at that point inthe mechanical fatigue test as well, whichsupports the authenticity of the data. Thefractures observed near the bridge in bendingand axial compression conditions had ahigher risk of stent separation owing to lossof axial stent integrity, i.e., type 3 fracture.16

Therefore, it seems that bending and axialcompression could impart more critical stressto the stent than torsion.

With respect to design, a stent without thebridge on which the strain concentrateswould likely survive without failure becausedynamic loads are distributed equally to allstruts. As a result, Misago could be expectedto have a reduced risk of complete transversefracture since it does not have the connec-tions or bridges found in other stent designs.Optimizing stent link design would aid in thedevelopment of a dedicated stent for thepopliteal artery, where the devices are sub-jected to a more severe bending and/or axialcompression load than in the SFA.

Limitations

We are unaware of other reports utilizingFEA for evaluating stent design during simu-lated movement (stress), but other computerprograms used to undertake similar evalua-tions may produce different results. However,our study should be seen as only an attemptto gain insight into the connection betweenstent design and the potential for fracture invivo.

There are other limitations associated withthe design and construction of our ex vivosetting: a larger angle for bending, evaluatinglonger stent lengths, and stents not implantedinto a tube or another medium that mimics anin vivo situation. Finally, the fatigue tests

774 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776

were performed under non-standardized con-ditions because these standardized methodsdo not exist. Another shortcoming was thesmall sample size for mechanical tests. Therewas also no consideration of material prop-erty or surface finish.

To our knowledge, pulsatile durability test-ing is standardized only by the AmericanSociety for Testing and Materials, but noindustry standard for in vitro stent fracturefatigue testing in the SFA exists; although itmay be difficult to reproduce the dynamics ofthe SFA, it is imperative that improvedstandards should be developed in all aspectsof nitinol stent processing and testing tobetter identify the failure modes of SFAstenting. Physicians should apply some pres-sure on the medical device industry tostandardize testing of stents and endopros-theses before marketing approval and clinicaluse.

In spite of these limitations, these simpletests offered a reproducible method of testingthe differences in stent design; the acquireddata gave us an indication of stent perfor-mance that might be expected in vivo. Evenwhen the FEA results match those of themechanical fatigue tests, the data have to beextrapolated with the greatest care beforebeing applied to a clinical situation. More-over, many stent strut structures are missedduring even high resolution fluoroscopy, aswe found in this study. Therefore, the cur-rently reported low fracture rates with sec-ond-generation SFA stents should be inter-preted cautiously.

Conclusion

The 7 nitinol SFA stents tested showeddifferences in the incidence of high strainzones that indicate the potential for stentfracture, which can be revealed only byfunctional investigation. Differences in stentdesign may play a major role in the occurrenceof restenosis and reocclusion related to stentstrut fractures. Mechanical bending has to beconsidered in future studies to assess theinfluence of differences in stent design, mate-rial, or even postinterventional drug treatmenton the long-term patency of stents and endo-prostheses in the femoropopliteal segment.

REFERENCES

1. Grimm J, Muller-Hulsbeck S, Jahnke T, et al.Randomized study to compare PTA aloneversus PTA with Palmaz stent placement forfemoropopliteal lesions. J Vasc Interv Radiol.2001;12:935–942.

2. Krankenberg H, Schluter M, Steinkamp HJ,et al. Nitinol stent implantation versus percu-taneous transluminal angioplasty in superficialfemoral artery lesions up to 10 cm in length:the femoral artery stenting trial (FAST). Circu-lation. 2007;116:285–292.

3. Schillinger M, Sabeti S, Loewe C, et al. Balloonangioplasty versus implantation of nitinolstents in the superficial femoral artery. N EnglJ Med. 2006;354:1879–1888.

4. Duda SH, Pusich B, Richter G, et al. Sirolimus-eluting stents for the treatment of obstructivesuperficial femoral artery disease: six-monthresults. Circulation. 2002;106:1505–1509.

5. Duda SH, Bosiers M, Lammer J, et al. Siroli-mus-eluting versus bare nitinol stent for ob-structive superficial femoral artery disease: theSIROCCO II trial. J Vasc Interv Radiol. 2005;16:331–338.

6. Schlager O, Dick P, Sabeti S, et al. Long-segment SFA stenting—the dark sides: in-stentrestenosis, clinical deterioration, and stentfractures. J Endovasc Ther. 2005;12:676–684.

7. Schulte KL, Muller-Hulsbeck S, Cao P, et al.MISAGO 1: first-in-man clinical trial with Misagonitinol stent. EuroIntervention. 2010;5:687–691.

8. Scheinert D, Scheinert S, Sax J, et al. Preva-lence and clinical impact of stent fractures afterfemoropopliteal stenting. J Am Coll Cardiol.2005;45:312–315.

9. Klein AJ, Chen SJ, Messenger JC, et al. Quanti-tative assessment of the conformational changein the femoropopliteal arterywith legmovement.Catheter Cardiovasc Interv. 2009;74:787–798.

10. Cheng CP, Wilson NM, Hallett RL, et al. In vivoMR angiographic quantification of axial andtwisting deformations of the superficial femo-ral artery resulting from maximum hip andknee flexion. J Vasc Interv Radiol. 2006;17:979–987.

11. Cheng CP, Choi G, Herfkens RJ, et al. The effectof aging on deformations of the superficialfemoral artery resulting from hip and kneeflexion: potential clinical implications. J VascInterv Radiol. 2010;21:195–202.

12. Brown R, Nguyen TD, Spincemaille P, et al. Invivo quantification of femoral-popliteal com-pression during isometric thigh contraction:assessment using MR angiography. J MagnReson Imaging. 2009;29:1116–1124.

J ENDOVASC THER2010;17:767–776

SFA STENT DESIGN 775Muller-Hulsbeck et al.

13. Dension A. Axial and bending fatigue resis-tance of nitinol stents. Proceedings of theInternational Conference on Shape Memoryand Superelastic Technologies, Baden-Baden,Germany. 2004 Oct:95–102.

14. Bosiers M, Torsello G, Gissler HM, et al. Nitinolstent implantation in long superficial femoralartery lesions: 12-month results of the DURABIL-ITY I study. J Endovasc Ther. 2009;16:261–269.

15. Nikanorov A, Smouse HB, Osman K, et al.Fracture of self-expanding nitinol stentsstressed in vitro under simulated intravascularconditions. Vasc Surg. 2008;48:435–340.

16. Jaff M, Dake M, Pompa J, et al. Stan-dardized evaluation and reporting of stentfractures in clinical trials of noncoronarydevices. Catheter Cardiovasc Interv. 2007;70:460–462.

776 SFA STENT DESIGNMuller-Hulsbeck et al.

J ENDOVASC THER2010;17:767–776