Annals of Biomedical Engineering(Accepted Postprint)

The Effect of Valve-in-Valve Implantation Height on Sinus Flow

Prem A Midha middot Vrishank Raghav middot IkechukwuOkafor middot Ajit P Yoganathan

Received date Accepted May 3 2016

Abstract Valve-in-valve transcatheter aortic valve replacement (VIV-TAVR) has proven to bea successful treatment for high risk patients with failing aortic surgical bioprostheses Howeverthrombus formation on the leaflets of the valve has emerged as a major issue in such proceduresposing a risk of restenosis thromboembolism and reduced durability In this work we attemptedto understand the effect of deployment position of the transcatheter heart valve (THV) on thespatio-temporal flow field within the sinus in VIV-TAVR Experiments were performed in anin vitro pulsatile left heart simulator using high-speed Particle Image Velocimetry (PIV) tomeasure the flow field in the sinus region The time-resolved velocity data was used to understandthe qualitative and quantitative flow patterns In addition a particle tracking technique wasused to evaluate relative thrombosis risk via sinus washout The velocity data demonstrate thatimplantation position directly affects sinus flow patterns leading to increased flow stagnationwith increasing deployment height The particle tracking simulations showed that implantationposition directly affected washout time with the highest implantation resulting in the leastwashout These results clearly demonstrate the flow pattern and flow stagnation in the sinusis sensitive to THV position It is therefore important for the interventional cardiologist andcardiac surgeon to consider how deployment position could impact flow stagnation during VIV-TAVR

Keywords TAVR middot particle tracking middot sinus washout middot thrombosis

Nomenclature

TAV R transcatheter aortic valve replacementTHV transcatheter heart valveV IV Valve-in-valve

1 Introduction

Aortic stenosis is the most prevalent valvular heart disease in developed countries [17] andhigh mortality is associated with untreated severe aortic stenosis [4] Patients diagnosed withmoderate or severe aortic stenosis undergo surgical aortic valve replacement approximately67500 surgeries are performed annually in the US [5] In the past 20 years surgical aortic valvereplacement has seen a dramatic shift toward the use of bioprosthetic over mechanical heartvalves [2] However tissue bioprostheses suffer from the same failure modes as native aorticvalves namely calcification-induced stenosis With the increasing use of surgical bioprostheses

Ajit YoganathanDepartment of Biomedical Engineering Georgia Institute of TechnologyE-mail ajityoganathanbmegatechedu

it is expected that the number of patients diagnosed with a degenerated bioprosthesis will onlyincrease [8] Though a second surgical procedure is often not an option due to high surgicalrisk valve-in-valve transcatheter aortic valve replacement (VIV-TAVR) has emerged as a viablenon-surgical treatment [11] Recently the Food and Drug Administration approved the SAPIEN(Edwards Lifesciences Irvine CA USA) and CoreValve (Medtronic Dublin Ireland) valvesfor VIV-TAVR in the US Large databases of retrospective patient data have been utilizedto demonstrate the feasibility of this technique as a successful solution [89] Yet while VIV-TAVR may restore valve function and improve symptoms adverse events such as elevated post-procedural gradients (284) coronary obstruction (35) device malpositioning (150) andvalve leaflet thrombosis (4) have been reported [1710146] Poor understanding of optimalVIV deployment locations and transcatheter heart valve (THV) hemodynamics may explainsome of these complications

The primary motivation of this study was to understand a specialized case of VIV-TAVRin which a patient presented with a failing surgical bioprosthesis smaller than what currentlyavailable THVs can accommodate Here the interventional cardiologist could perform an off-label procedure using the smallest available THV size and risk residual stenosis due to anincompletely deployed THV In this particular case the post-procedural pressure gradients areexpected to be much higher for two primary reasons a) the patient has a small bioprosthesissize to begin with and b) implanting a transcatheter valve further reduces the orifice area Aviable solution to this contraindicated case is to implant the THV supra-annular with respectto the recommended position and create a flared downstream opening (flower pot geometry) asseen in Figure 2 Our recent study [15] investigated the risks and benefits of the supra-annularimplantation of a balloon-expandable THV The results of that study demonstrated the capacityfor supra-annular valve positioning to lower post-procedural gradients from 33 mmHg (normaldeployment) to 13 mmHg (extreme supra-annular deployment) On the other hand this wasobserved to obstruct the flow in the sinus of Valsalva which could lead to decreased coronaryflow andor increased stagnation-induced thrombosis risk Based on Virchowrsquos triad it is wellaccepted that blood flow stagnation after contact with cardiovascular devices can result inthrombus formation [131824] highlighting the need for a more complete profiling of stagnationzones in and around THVs It should be noted that a direct relationship between flow stagnationin the sinus and valve thrombosis is yet to be established in the literature however evaluationof this risk is critical especially in the light of recent evidence of sub-clinical leaflet thrombosison THVs [146]

In this work we attempted to understand in detail the effect of supra-annular positioningon the spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed Particle Image Velocimetry (PIV) was used to measure the flow patterns in the sinus regionfor this case The temporally resolved velocity field data was used to understand the qualitativeand quantitative flow patterns in the sinus region Furthermore a particle tracking techniquewas used to analyze particle residence times within the sinus region to evaluate thrombosis risk

2 Materials and Methods

21 Left Heart Simulator

The Georgia Tech Left Heart Simulator (Figure 1) is a validated pulsatile flow loop that sim-ulates physiological and pathophysiological hemodynamic conditions of the heart [25] Devicesbeing tested were mounted in an idealized rigid acrylic aortic chamber based on published av-erage anatomical measurements (Figure 3) [1221] The chamber simulates the aortic root andproximal ascending aorta of a healthy adult however these average anatomical measurementsfall within the range of patients who have undergone a valve replacement procedure The choiceof a healthy patient anatomy was made due to the lack of a true ldquoaveragerdquo diseased patientanatomy Physiologic pressures and flows were achieved through adjustment of systemic resis-tance and compliance elements as well as a system of solenoids that control the contraction andrelaxation of a bladder pump A custom LabVIEW VI (v120 2012 National Instruments Cor-poration Austin TX) controlled the triggering of the solenoids and recorded the instantaneous

2

flow rate measured by an electromagnetic flow probe (600 series Carolina Medical ElectronicsEast Bend NC) Pressures were measured with Deltran DPT-200 pressure transducers (UtahMedical Products Inc Midvale UT) on either side of the test section The loop was tuned to amean arterial pressure of 100 mmHg cardiac cycle length of 856 ms (sim13 systole) and meancardiac output of 50 Lmin The working fluid was a 35 cSt saline-glycerine solution (36glycerin by volume in water) to match the kinematic viscosity of blood

Fig 1 The Georgia Tech Left Heart Simulator is a validated pulsatile flow loop that simulates physiological conditions ofthe left heart [25]

Fig 2 A schematic representation of the VIV deployment positions The surgical valve frame is shown in white andleaflets shown in dark brown The THV leaflets are shown in light brown and are attached to a metallic stent

22 Valve Deployment

A 19 mm PERIMOUNT tissue bioprosthesis (Edwards Lifesciences) was tethered into the aorticchamber by 6 interrupted sutures evenly spaced around the suture cuff The internal diameterof the PERIMOUNT is 17 mm Subsequently a previously unused 23 mm balloon-expandableSAPIEN XT (Edwards Lifesciences Irvine CA USA) was deployed in each of 4 positions

3

Fig 3 The region of interest (ROI) in the study is denoted by the red dashed line in panel A The green line in panel Bindicates the plane in which PIV data was captured Panel B is a 90 radial rotation of panel A

(Figure 2) The THV commissures were aligned approximately 60 degrees out of phase from thesurgical valve commissures (illustrated in Figure 3) The 23 mm SAPIEN XT is indicated forannulus diameters between 18 and 22 mm Due to a lack of VIV deployment recommendationsprovided by manufacturers the normal position was defined by the ViV Aortic phone application- a tool commonly used by cardiologists as a reference for valve sizing and positioning In anormal deployment the bottom of the THV stent is aligned with the bottom of the surgicalbioprosthesis sewing ring The remaining positions were 3 mm 6 mm and 8 mm supra-annularrelative to the normal position Due to the geometric constraints of the deployment each VIVconfiguration has a different degree of supra-annular flaring (ldquoflower potrdquo geometry)

23 Particle Image Velocimetry Instrumentation

High-speed PIV was used to measure the velocity field in the sinus region of the aortic flowchamber These 2-dimensional 2-component time-resolved measurements of the sinus flow ve-locities were used as a means of assessing relative risk of thrombosis induced by flow stagnationA diode-pumped continuous wave solid-state laser (Shenzhen Optolaser 2W 532 nm) emittinga 2 mm beam was used as the light source The laser beam was converted to a high frequencypulsed laser sheet by using a scanning mirror (rotating mirror) setup provided by LaVision(GmbH Goettingen Germany) The flow was seeded with fluorescent polymeric rhodamine-Bparticles (Dantec Dynamics Denmark) with a mean diameter of approximately 10microm A CMOScamera (Vision Research Phantom Miro MRLC123) was used to image the particles in thecentral plane of the sinus region (Figure 3) The particle size in the camera image ranged from2 to 4 pixels The camera was fitted with a macro lens system of focal length 60 mm and theaperture was set at f4 To improve the signal-to-noise ratio of the acquired data a high-passfilter (cut-off wavelength of 580 nm) was used to minimize laser reflections from the sinus regionAny unavoidable laser reflections were masked out during the post-processing steps

Data were acquired at approximately 900 Hz and processed using DaVis 80 (LaVisionGmbH Goettingen Germany) A series of 3 cycles of data were acquired continuously andrepeated 3 times such that a total of 9 cycles of data were acquired The images were acquiredas a time-series rather than the traditional image pairs from a frame-straddling acquisition tech-

4

nique After background subtraction and masking the velocities were calculated from spatialtime-series cross-correlation of the images An interrogation window (64 by 64 pixels) overlapof 50 and a second interrogation pass with a reduced window size (32 by 32 pixels) was usedto increase the signal-to-noise ratio of the correlation peak This yielded a vector resolution ofapproximately 039 mm which was around 2 of the distance between the valve annulus andthe sino-tubular junction An applied vector range a median filter and light smoothing (3 by3) were used as post-processing steps of the vector images

24 Particle Washout

The vector fields were bin-averaged effectively down-sampling the data from 760 framescycleto 152 framescycle Particle paths were calculated using Tecplot 360 EX 2015 R1 (Tecplot IncBellevue WA) as follows Velocity magnitudes were computed and regions with zero velocitywere eliminated from the ROI This limited the seeding area to locations with non-zero velocitiesFive hundred massless virtual particles were uniformly seeded within the sinus region Time-varying particle locations were computed by the time-integral of d

dtx(t) = v(x(t) t) Over atime step of 1 ms each particle experienced the spatially and temporally interpolated velocityvector at its particular location After each time step a new velocity vector was imposed on theparticle based on its updated location

The particle paths were exported to MATLAB for analysis Particle washout was quantifiedby counting the number of particles in the ROI at each time step At each time step particlelocations were compared with their final position in the simulation If the location did notchange that particle was artificially terminated at the first static time point This typicallyoccurred in very low velocity regions near the wall and can be aggravated by coarse spatialor temporal resolution In this case it is due to unreliable near-wall velocities due to opticaldistortion and laser reflections This workflow is summarized in Figure 4

Fig 4 Overview of the particle tracking workflow

25 Uncertainty Estimation

The uncertainty of the in-plane velocity measurements was computed using methodology dis-cussed in Raffel et al [20] The probability density function of the instantaneous velocity datawas used to determine the bias error in the PIV measurements It was determined that thebias error (εb) was less than 0001 ms and the probability density function did not reveal anypeak-locking error The relaxation time of the particle (of 10 microm diameter) was considered tocompute the lag error during the PIV measurements and compared to the characteristic timescale in the flow field The analysis showed that the characteristic time was several orders ofmagnitude larger than the relaxation time available for the particle indicating that the par-ticle lag error was insignificant Random error during the PIV experiments was computed byεr = c times dp where εr is the random error dp the particle diameter and c is an empirical con-stant that usually lies between 005 and 010 [19] In these experiments the random error rangedfrom 01 pixels in the best case to 04 pixels in the worst case The random error in velocity

5

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

it is expected that the number of patients diagnosed with a degenerated bioprosthesis will onlyincrease [8] Though a second surgical procedure is often not an option due to high surgicalrisk valve-in-valve transcatheter aortic valve replacement (VIV-TAVR) has emerged as a viablenon-surgical treatment [11] Recently the Food and Drug Administration approved the SAPIEN(Edwards Lifesciences Irvine CA USA) and CoreValve (Medtronic Dublin Ireland) valvesfor VIV-TAVR in the US Large databases of retrospective patient data have been utilizedto demonstrate the feasibility of this technique as a successful solution [89] Yet while VIV-TAVR may restore valve function and improve symptoms adverse events such as elevated post-procedural gradients (284) coronary obstruction (35) device malpositioning (150) andvalve leaflet thrombosis (4) have been reported [1710146] Poor understanding of optimalVIV deployment locations and transcatheter heart valve (THV) hemodynamics may explainsome of these complications

The primary motivation of this study was to understand a specialized case of VIV-TAVRin which a patient presented with a failing surgical bioprosthesis smaller than what currentlyavailable THVs can accommodate Here the interventional cardiologist could perform an off-label procedure using the smallest available THV size and risk residual stenosis due to anincompletely deployed THV In this particular case the post-procedural pressure gradients areexpected to be much higher for two primary reasons a) the patient has a small bioprosthesissize to begin with and b) implanting a transcatheter valve further reduces the orifice area Aviable solution to this contraindicated case is to implant the THV supra-annular with respectto the recommended position and create a flared downstream opening (flower pot geometry) asseen in Figure 2 Our recent study [15] investigated the risks and benefits of the supra-annularimplantation of a balloon-expandable THV The results of that study demonstrated the capacityfor supra-annular valve positioning to lower post-procedural gradients from 33 mmHg (normaldeployment) to 13 mmHg (extreme supra-annular deployment) On the other hand this wasobserved to obstruct the flow in the sinus of Valsalva which could lead to decreased coronaryflow andor increased stagnation-induced thrombosis risk Based on Virchowrsquos triad it is wellaccepted that blood flow stagnation after contact with cardiovascular devices can result inthrombus formation [131824] highlighting the need for a more complete profiling of stagnationzones in and around THVs It should be noted that a direct relationship between flow stagnationin the sinus and valve thrombosis is yet to be established in the literature however evaluationof this risk is critical especially in the light of recent evidence of sub-clinical leaflet thrombosison THVs [146]

In this work we attempted to understand in detail the effect of supra-annular positioningon the spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed Particle Image Velocimetry (PIV) was used to measure the flow patterns in the sinus regionfor this case The temporally resolved velocity field data was used to understand the qualitativeand quantitative flow patterns in the sinus region Furthermore a particle tracking techniquewas used to analyze particle residence times within the sinus region to evaluate thrombosis risk

2 Materials and Methods

21 Left Heart Simulator

The Georgia Tech Left Heart Simulator (Figure 1) is a validated pulsatile flow loop that sim-ulates physiological and pathophysiological hemodynamic conditions of the heart [25] Devicesbeing tested were mounted in an idealized rigid acrylic aortic chamber based on published av-erage anatomical measurements (Figure 3) [1221] The chamber simulates the aortic root andproximal ascending aorta of a healthy adult however these average anatomical measurementsfall within the range of patients who have undergone a valve replacement procedure The choiceof a healthy patient anatomy was made due to the lack of a true ldquoaveragerdquo diseased patientanatomy Physiologic pressures and flows were achieved through adjustment of systemic resis-tance and compliance elements as well as a system of solenoids that control the contraction andrelaxation of a bladder pump A custom LabVIEW VI (v120 2012 National Instruments Cor-poration Austin TX) controlled the triggering of the solenoids and recorded the instantaneous

2

flow rate measured by an electromagnetic flow probe (600 series Carolina Medical ElectronicsEast Bend NC) Pressures were measured with Deltran DPT-200 pressure transducers (UtahMedical Products Inc Midvale UT) on either side of the test section The loop was tuned to amean arterial pressure of 100 mmHg cardiac cycle length of 856 ms (sim13 systole) and meancardiac output of 50 Lmin The working fluid was a 35 cSt saline-glycerine solution (36glycerin by volume in water) to match the kinematic viscosity of blood

Fig 1 The Georgia Tech Left Heart Simulator is a validated pulsatile flow loop that simulates physiological conditions ofthe left heart [25]

Fig 2 A schematic representation of the VIV deployment positions The surgical valve frame is shown in white andleaflets shown in dark brown The THV leaflets are shown in light brown and are attached to a metallic stent

22 Valve Deployment

A 19 mm PERIMOUNT tissue bioprosthesis (Edwards Lifesciences) was tethered into the aorticchamber by 6 interrupted sutures evenly spaced around the suture cuff The internal diameterof the PERIMOUNT is 17 mm Subsequently a previously unused 23 mm balloon-expandableSAPIEN XT (Edwards Lifesciences Irvine CA USA) was deployed in each of 4 positions

3

Fig 3 The region of interest (ROI) in the study is denoted by the red dashed line in panel A The green line in panel Bindicates the plane in which PIV data was captured Panel B is a 90 radial rotation of panel A

(Figure 2) The THV commissures were aligned approximately 60 degrees out of phase from thesurgical valve commissures (illustrated in Figure 3) The 23 mm SAPIEN XT is indicated forannulus diameters between 18 and 22 mm Due to a lack of VIV deployment recommendationsprovided by manufacturers the normal position was defined by the ViV Aortic phone application- a tool commonly used by cardiologists as a reference for valve sizing and positioning In anormal deployment the bottom of the THV stent is aligned with the bottom of the surgicalbioprosthesis sewing ring The remaining positions were 3 mm 6 mm and 8 mm supra-annularrelative to the normal position Due to the geometric constraints of the deployment each VIVconfiguration has a different degree of supra-annular flaring (ldquoflower potrdquo geometry)

23 Particle Image Velocimetry Instrumentation

High-speed PIV was used to measure the velocity field in the sinus region of the aortic flowchamber These 2-dimensional 2-component time-resolved measurements of the sinus flow ve-locities were used as a means of assessing relative risk of thrombosis induced by flow stagnationA diode-pumped continuous wave solid-state laser (Shenzhen Optolaser 2W 532 nm) emittinga 2 mm beam was used as the light source The laser beam was converted to a high frequencypulsed laser sheet by using a scanning mirror (rotating mirror) setup provided by LaVision(GmbH Goettingen Germany) The flow was seeded with fluorescent polymeric rhodamine-Bparticles (Dantec Dynamics Denmark) with a mean diameter of approximately 10microm A CMOScamera (Vision Research Phantom Miro MRLC123) was used to image the particles in thecentral plane of the sinus region (Figure 3) The particle size in the camera image ranged from2 to 4 pixels The camera was fitted with a macro lens system of focal length 60 mm and theaperture was set at f4 To improve the signal-to-noise ratio of the acquired data a high-passfilter (cut-off wavelength of 580 nm) was used to minimize laser reflections from the sinus regionAny unavoidable laser reflections were masked out during the post-processing steps

Data were acquired at approximately 900 Hz and processed using DaVis 80 (LaVisionGmbH Goettingen Germany) A series of 3 cycles of data were acquired continuously andrepeated 3 times such that a total of 9 cycles of data were acquired The images were acquiredas a time-series rather than the traditional image pairs from a frame-straddling acquisition tech-

4

nique After background subtraction and masking the velocities were calculated from spatialtime-series cross-correlation of the images An interrogation window (64 by 64 pixels) overlapof 50 and a second interrogation pass with a reduced window size (32 by 32 pixels) was usedto increase the signal-to-noise ratio of the correlation peak This yielded a vector resolution ofapproximately 039 mm which was around 2 of the distance between the valve annulus andthe sino-tubular junction An applied vector range a median filter and light smoothing (3 by3) were used as post-processing steps of the vector images

24 Particle Washout

The vector fields were bin-averaged effectively down-sampling the data from 760 framescycleto 152 framescycle Particle paths were calculated using Tecplot 360 EX 2015 R1 (Tecplot IncBellevue WA) as follows Velocity magnitudes were computed and regions with zero velocitywere eliminated from the ROI This limited the seeding area to locations with non-zero velocitiesFive hundred massless virtual particles were uniformly seeded within the sinus region Time-varying particle locations were computed by the time-integral of d

dtx(t) = v(x(t) t) Over atime step of 1 ms each particle experienced the spatially and temporally interpolated velocityvector at its particular location After each time step a new velocity vector was imposed on theparticle based on its updated location

The particle paths were exported to MATLAB for analysis Particle washout was quantifiedby counting the number of particles in the ROI at each time step At each time step particlelocations were compared with their final position in the simulation If the location did notchange that particle was artificially terminated at the first static time point This typicallyoccurred in very low velocity regions near the wall and can be aggravated by coarse spatialor temporal resolution In this case it is due to unreliable near-wall velocities due to opticaldistortion and laser reflections This workflow is summarized in Figure 4

Fig 4 Overview of the particle tracking workflow

25 Uncertainty Estimation

The uncertainty of the in-plane velocity measurements was computed using methodology dis-cussed in Raffel et al [20] The probability density function of the instantaneous velocity datawas used to determine the bias error in the PIV measurements It was determined that thebias error (εb) was less than 0001 ms and the probability density function did not reveal anypeak-locking error The relaxation time of the particle (of 10 microm diameter) was considered tocompute the lag error during the PIV measurements and compared to the characteristic timescale in the flow field The analysis showed that the characteristic time was several orders ofmagnitude larger than the relaxation time available for the particle indicating that the par-ticle lag error was insignificant Random error during the PIV experiments was computed byεr = c times dp where εr is the random error dp the particle diameter and c is an empirical con-stant that usually lies between 005 and 010 [19] In these experiments the random error rangedfrom 01 pixels in the best case to 04 pixels in the worst case The random error in velocity

5

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

flow rate measured by an electromagnetic flow probe (600 series Carolina Medical ElectronicsEast Bend NC) Pressures were measured with Deltran DPT-200 pressure transducers (UtahMedical Products Inc Midvale UT) on either side of the test section The loop was tuned to amean arterial pressure of 100 mmHg cardiac cycle length of 856 ms (sim13 systole) and meancardiac output of 50 Lmin The working fluid was a 35 cSt saline-glycerine solution (36glycerin by volume in water) to match the kinematic viscosity of blood

Fig 1 The Georgia Tech Left Heart Simulator is a validated pulsatile flow loop that simulates physiological conditions ofthe left heart [25]

Fig 2 A schematic representation of the VIV deployment positions The surgical valve frame is shown in white andleaflets shown in dark brown The THV leaflets are shown in light brown and are attached to a metallic stent

22 Valve Deployment

A 19 mm PERIMOUNT tissue bioprosthesis (Edwards Lifesciences) was tethered into the aorticchamber by 6 interrupted sutures evenly spaced around the suture cuff The internal diameterof the PERIMOUNT is 17 mm Subsequently a previously unused 23 mm balloon-expandableSAPIEN XT (Edwards Lifesciences Irvine CA USA) was deployed in each of 4 positions

3

Fig 3 The region of interest (ROI) in the study is denoted by the red dashed line in panel A The green line in panel Bindicates the plane in which PIV data was captured Panel B is a 90 radial rotation of panel A

(Figure 2) The THV commissures were aligned approximately 60 degrees out of phase from thesurgical valve commissures (illustrated in Figure 3) The 23 mm SAPIEN XT is indicated forannulus diameters between 18 and 22 mm Due to a lack of VIV deployment recommendationsprovided by manufacturers the normal position was defined by the ViV Aortic phone application- a tool commonly used by cardiologists as a reference for valve sizing and positioning In anormal deployment the bottom of the THV stent is aligned with the bottom of the surgicalbioprosthesis sewing ring The remaining positions were 3 mm 6 mm and 8 mm supra-annularrelative to the normal position Due to the geometric constraints of the deployment each VIVconfiguration has a different degree of supra-annular flaring (ldquoflower potrdquo geometry)

23 Particle Image Velocimetry Instrumentation

High-speed PIV was used to measure the velocity field in the sinus region of the aortic flowchamber These 2-dimensional 2-component time-resolved measurements of the sinus flow ve-locities were used as a means of assessing relative risk of thrombosis induced by flow stagnationA diode-pumped continuous wave solid-state laser (Shenzhen Optolaser 2W 532 nm) emittinga 2 mm beam was used as the light source The laser beam was converted to a high frequencypulsed laser sheet by using a scanning mirror (rotating mirror) setup provided by LaVision(GmbH Goettingen Germany) The flow was seeded with fluorescent polymeric rhodamine-Bparticles (Dantec Dynamics Denmark) with a mean diameter of approximately 10microm A CMOScamera (Vision Research Phantom Miro MRLC123) was used to image the particles in thecentral plane of the sinus region (Figure 3) The particle size in the camera image ranged from2 to 4 pixels The camera was fitted with a macro lens system of focal length 60 mm and theaperture was set at f4 To improve the signal-to-noise ratio of the acquired data a high-passfilter (cut-off wavelength of 580 nm) was used to minimize laser reflections from the sinus regionAny unavoidable laser reflections were masked out during the post-processing steps

Data were acquired at approximately 900 Hz and processed using DaVis 80 (LaVisionGmbH Goettingen Germany) A series of 3 cycles of data were acquired continuously andrepeated 3 times such that a total of 9 cycles of data were acquired The images were acquiredas a time-series rather than the traditional image pairs from a frame-straddling acquisition tech-

4

nique After background subtraction and masking the velocities were calculated from spatialtime-series cross-correlation of the images An interrogation window (64 by 64 pixels) overlapof 50 and a second interrogation pass with a reduced window size (32 by 32 pixels) was usedto increase the signal-to-noise ratio of the correlation peak This yielded a vector resolution ofapproximately 039 mm which was around 2 of the distance between the valve annulus andthe sino-tubular junction An applied vector range a median filter and light smoothing (3 by3) were used as post-processing steps of the vector images

24 Particle Washout

The vector fields were bin-averaged effectively down-sampling the data from 760 framescycleto 152 framescycle Particle paths were calculated using Tecplot 360 EX 2015 R1 (Tecplot IncBellevue WA) as follows Velocity magnitudes were computed and regions with zero velocitywere eliminated from the ROI This limited the seeding area to locations with non-zero velocitiesFive hundred massless virtual particles were uniformly seeded within the sinus region Time-varying particle locations were computed by the time-integral of d

dtx(t) = v(x(t) t) Over atime step of 1 ms each particle experienced the spatially and temporally interpolated velocityvector at its particular location After each time step a new velocity vector was imposed on theparticle based on its updated location

The particle paths were exported to MATLAB for analysis Particle washout was quantifiedby counting the number of particles in the ROI at each time step At each time step particlelocations were compared with their final position in the simulation If the location did notchange that particle was artificially terminated at the first static time point This typicallyoccurred in very low velocity regions near the wall and can be aggravated by coarse spatialor temporal resolution In this case it is due to unreliable near-wall velocities due to opticaldistortion and laser reflections This workflow is summarized in Figure 4

Fig 4 Overview of the particle tracking workflow

25 Uncertainty Estimation

The uncertainty of the in-plane velocity measurements was computed using methodology dis-cussed in Raffel et al [20] The probability density function of the instantaneous velocity datawas used to determine the bias error in the PIV measurements It was determined that thebias error (εb) was less than 0001 ms and the probability density function did not reveal anypeak-locking error The relaxation time of the particle (of 10 microm diameter) was considered tocompute the lag error during the PIV measurements and compared to the characteristic timescale in the flow field The analysis showed that the characteristic time was several orders ofmagnitude larger than the relaxation time available for the particle indicating that the par-ticle lag error was insignificant Random error during the PIV experiments was computed byεr = c times dp where εr is the random error dp the particle diameter and c is an empirical con-stant that usually lies between 005 and 010 [19] In these experiments the random error rangedfrom 01 pixels in the best case to 04 pixels in the worst case The random error in velocity

5

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

Fig 3 The region of interest (ROI) in the study is denoted by the red dashed line in panel A The green line in panel Bindicates the plane in which PIV data was captured Panel B is a 90 radial rotation of panel A

(Figure 2) The THV commissures were aligned approximately 60 degrees out of phase from thesurgical valve commissures (illustrated in Figure 3) The 23 mm SAPIEN XT is indicated forannulus diameters between 18 and 22 mm Due to a lack of VIV deployment recommendationsprovided by manufacturers the normal position was defined by the ViV Aortic phone application- a tool commonly used by cardiologists as a reference for valve sizing and positioning In anormal deployment the bottom of the THV stent is aligned with the bottom of the surgicalbioprosthesis sewing ring The remaining positions were 3 mm 6 mm and 8 mm supra-annularrelative to the normal position Due to the geometric constraints of the deployment each VIVconfiguration has a different degree of supra-annular flaring (ldquoflower potrdquo geometry)

23 Particle Image Velocimetry Instrumentation

High-speed PIV was used to measure the velocity field in the sinus region of the aortic flowchamber These 2-dimensional 2-component time-resolved measurements of the sinus flow ve-locities were used as a means of assessing relative risk of thrombosis induced by flow stagnationA diode-pumped continuous wave solid-state laser (Shenzhen Optolaser 2W 532 nm) emittinga 2 mm beam was used as the light source The laser beam was converted to a high frequencypulsed laser sheet by using a scanning mirror (rotating mirror) setup provided by LaVision(GmbH Goettingen Germany) The flow was seeded with fluorescent polymeric rhodamine-Bparticles (Dantec Dynamics Denmark) with a mean diameter of approximately 10microm A CMOScamera (Vision Research Phantom Miro MRLC123) was used to image the particles in thecentral plane of the sinus region (Figure 3) The particle size in the camera image ranged from2 to 4 pixels The camera was fitted with a macro lens system of focal length 60 mm and theaperture was set at f4 To improve the signal-to-noise ratio of the acquired data a high-passfilter (cut-off wavelength of 580 nm) was used to minimize laser reflections from the sinus regionAny unavoidable laser reflections were masked out during the post-processing steps

Data were acquired at approximately 900 Hz and processed using DaVis 80 (LaVisionGmbH Goettingen Germany) A series of 3 cycles of data were acquired continuously andrepeated 3 times such that a total of 9 cycles of data were acquired The images were acquiredas a time-series rather than the traditional image pairs from a frame-straddling acquisition tech-

4

nique After background subtraction and masking the velocities were calculated from spatialtime-series cross-correlation of the images An interrogation window (64 by 64 pixels) overlapof 50 and a second interrogation pass with a reduced window size (32 by 32 pixels) was usedto increase the signal-to-noise ratio of the correlation peak This yielded a vector resolution ofapproximately 039 mm which was around 2 of the distance between the valve annulus andthe sino-tubular junction An applied vector range a median filter and light smoothing (3 by3) were used as post-processing steps of the vector images

24 Particle Washout

The vector fields were bin-averaged effectively down-sampling the data from 760 framescycleto 152 framescycle Particle paths were calculated using Tecplot 360 EX 2015 R1 (Tecplot IncBellevue WA) as follows Velocity magnitudes were computed and regions with zero velocitywere eliminated from the ROI This limited the seeding area to locations with non-zero velocitiesFive hundred massless virtual particles were uniformly seeded within the sinus region Time-varying particle locations were computed by the time-integral of d

dtx(t) = v(x(t) t) Over atime step of 1 ms each particle experienced the spatially and temporally interpolated velocityvector at its particular location After each time step a new velocity vector was imposed on theparticle based on its updated location

The particle paths were exported to MATLAB for analysis Particle washout was quantifiedby counting the number of particles in the ROI at each time step At each time step particlelocations were compared with their final position in the simulation If the location did notchange that particle was artificially terminated at the first static time point This typicallyoccurred in very low velocity regions near the wall and can be aggravated by coarse spatialor temporal resolution In this case it is due to unreliable near-wall velocities due to opticaldistortion and laser reflections This workflow is summarized in Figure 4

Fig 4 Overview of the particle tracking workflow

25 Uncertainty Estimation

The uncertainty of the in-plane velocity measurements was computed using methodology dis-cussed in Raffel et al [20] The probability density function of the instantaneous velocity datawas used to determine the bias error in the PIV measurements It was determined that thebias error (εb) was less than 0001 ms and the probability density function did not reveal anypeak-locking error The relaxation time of the particle (of 10 microm diameter) was considered tocompute the lag error during the PIV measurements and compared to the characteristic timescale in the flow field The analysis showed that the characteristic time was several orders ofmagnitude larger than the relaxation time available for the particle indicating that the par-ticle lag error was insignificant Random error during the PIV experiments was computed byεr = c times dp where εr is the random error dp the particle diameter and c is an empirical con-stant that usually lies between 005 and 010 [19] In these experiments the random error rangedfrom 01 pixels in the best case to 04 pixels in the worst case The random error in velocity

5

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

nique After background subtraction and masking the velocities were calculated from spatialtime-series cross-correlation of the images An interrogation window (64 by 64 pixels) overlapof 50 and a second interrogation pass with a reduced window size (32 by 32 pixels) was usedto increase the signal-to-noise ratio of the correlation peak This yielded a vector resolution ofapproximately 039 mm which was around 2 of the distance between the valve annulus andthe sino-tubular junction An applied vector range a median filter and light smoothing (3 by3) were used as post-processing steps of the vector images

24 Particle Washout

The vector fields were bin-averaged effectively down-sampling the data from 760 framescycleto 152 framescycle Particle paths were calculated using Tecplot 360 EX 2015 R1 (Tecplot IncBellevue WA) as follows Velocity magnitudes were computed and regions with zero velocitywere eliminated from the ROI This limited the seeding area to locations with non-zero velocitiesFive hundred massless virtual particles were uniformly seeded within the sinus region Time-varying particle locations were computed by the time-integral of d

dtx(t) = v(x(t) t) Over atime step of 1 ms each particle experienced the spatially and temporally interpolated velocityvector at its particular location After each time step a new velocity vector was imposed on theparticle based on its updated location

The particle paths were exported to MATLAB for analysis Particle washout was quantifiedby counting the number of particles in the ROI at each time step At each time step particlelocations were compared with their final position in the simulation If the location did notchange that particle was artificially terminated at the first static time point This typicallyoccurred in very low velocity regions near the wall and can be aggravated by coarse spatialor temporal resolution In this case it is due to unreliable near-wall velocities due to opticaldistortion and laser reflections This workflow is summarized in Figure 4

Fig 4 Overview of the particle tracking workflow

25 Uncertainty Estimation

The uncertainty of the in-plane velocity measurements was computed using methodology dis-cussed in Raffel et al [20] The probability density function of the instantaneous velocity datawas used to determine the bias error in the PIV measurements It was determined that thebias error (εb) was less than 0001 ms and the probability density function did not reveal anypeak-locking error The relaxation time of the particle (of 10 microm diameter) was considered tocompute the lag error during the PIV measurements and compared to the characteristic timescale in the flow field The analysis showed that the characteristic time was several orders ofmagnitude larger than the relaxation time available for the particle indicating that the par-ticle lag error was insignificant Random error during the PIV experiments was computed byεr = c times dp where εr is the random error dp the particle diameter and c is an empirical con-stant that usually lies between 005 and 010 [19] In these experiments the random error rangedfrom 01 pixels in the best case to 04 pixels in the worst case The random error in velocity

5

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

measurement ranged between 0002 ms and 0008 ms The total measurement error with a

95 confidence interval was computed as εm = kradicεb2 + εr2 and was determined to be 0016

ms where k=196 for a 95 confidence interval

Fig 5 Simulated long exposures of the raw PIV data at peak systole demonstrate the drastic reduction in sinus fluidmotion in the highest deployment conditions

3 Results

31 Qualitative Flow Features and Velocity Fields

High-speed PIV was used to capture flow patterns within the sinus region of the model Asliding average over ten frames (sim11 ms) of the data was implemented to create pathline imageswhich provide a qualitative understanding of the flow in the sinus Figure 5 illustrates the flow

6

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

Fig 6 Maximum velocity experienced within the sinus at each spatial location throughout the cardiac cycle

in the sinus at peak systole for different implantation positions of the THV In the case ofthe PERIMOUNT valve only the particle pathlines clearly illustrate a predominant vorticalstructure in the sinus region Streamwise flow directed towards the sino-tubular junction canalso be observed as indicated by the pathlines In contrast as THV deployment height increasesthe vortical structure and streamwise flow towards the sino-tubular junction diminishes until itis absent in the +6 mm and +8 mm deployments At the highest implantation positions (+6 and+8 mm) the lengths of the pathlines are smaller (negligible particle motion over ten frames)when compared to the control normal and + 3mm cases This is a clear indication of reducedflow velocities in the sinus region at the highest implantation positions

To gain an understanding of the magnitude of flow within the sinus the maximum velocityexperienced by each spatial location throughout the entire cardiac cycle was plotted Figure 6shows that the implantation height of the THV affects the maximum velocity experiencedthroughout the sinus Based on previous work regions with a maximum velocity below 005ms are considered stagnation regions [22] For the control case every spatial location in thesinus experiences sim015 ms of flow velocity at some time point within the cardiac cycle Thiswas also observed for the normal implantation case However as implantation height increasedfrom +3 mm to + 8 mm the sinus region closer to the annulus experienced almost no fluidmotion ( 0025 ms) It is important to note that the sinus area of the normal condition wasobserved to be larger than that of other conditions because the surgical valve leaflets are notpushed into the sinus as is the case with all of the VIV deployments Also the THV stentwhen expanded takes a flower pot shape which further reduces the sinus area

32 Particle Washout

A video of the particle tracking simulation results can be found in the supplementary materialThis video shows the decreasing particle motion at the base of the sinus under increasing THVdeployment height Figure 7 shows the percentage of particles left inside the sinus as a functionof cardiac cycles The particles are seeded (only once) at the beginning of the first cycle Astime progresses it can be observed that in the control case all the particles wash out within 4cycles At the end of the 9th cycle the control normal +3 mm +6 mm and +8 mm conditionshad 0 0 1 12 and 28 of particles leftover in the sinus respectively The initial rate ofparticle washout for the first 50 of particles was similar among all cases except for the +8 mmdeployment which shows a much lower rate of wash out throughout the 9 cardiac cycles Theparticle half-life was approximately 1 cardiac cycle for each case except the +8 mm deploymentwhich was approximately 3 cardiac cycles After the initial 50 of particles are washed away

7

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

the ejection rates reduced for the VIV deployments and differences were observable betweenconditions

Fig 7 Ejection rate of particles out of the sinus as a function of cardiac cycle number Each cycle is 856 ms

4 Discussion

In this work we have studied in detail the effect of supra-annular positioning of a THV onthe spatio-temporal flow field within the sinus region in a contraindicated VIV-TAVR High-speed PIV was used to understand the qualitative and quantitative flow patterns within thesinus region Particle residence time has been shown to correlate with thrombus formation [24]therefore particle washout was quantified within the sinus region at each implantation position

A qualitative understanding of the sinus flow field was gained by evaluating a sliding aver-age view of the raw PIV images The qualitative observations in the control case agree withprevious observations where a predominant vortical structure is seen [16] However when aTHV was implanted the vortical structure in the sinus was observed to shrink in size andorcompletely disappear at the higher implantation locations (+6mm and +8mm) with a corre-sponding decrease in outward flow towards the sino-tubular junction In addition we observedthat as the implantation height of the THV increased flow stagnation close to the base of thesinus dramatically increased Figure 6 shows that for the +6 mm and +8 mm cases there isalmost no fluid motion close base of the sinus throughout the entire cardiac cycle

These findings were corroborated by the particle tracking analyses performed on each ofthe conditions Only in the control and normal cases were all the particles within the sinuswashed out in under 9 cycles and in the case of the +6 mm and +8 mm deployments particlesstill remained after 9 cycles The remaining particles congregated around the base of the sinusand when compared to the maximum velocity plots these results indicate that the very lowmax velocity regions do in fact correspond to long-term stagnation within those regions Asblood flow stagnation after contact with cardiovascular devices has previously been shown toresult in thrombus formation [131824] this highlights the need for a more complete profilingof stagnation zones in and around THVs This indicates that as the THV implantation height

8

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

increases the risk of thrombus formation may correspondingly increase as well however a directlink between sinus flow stagnation and clinical valve thrombosis is yet to be established

The particle washout data appear to follow exponential decay trends of the form aeminusbtwhere b is the rate constant The half-life of the particles in the sinus were derived from theexponential curve fit and are summarized in Table 1 While the data are specific to the aorticand VIV geometries used in this study an increase in deployment height results in a decreasein the rate constant (less washout)

Table 1 Exponential decay curve fit comparisons for each deployment condition

Rate Constantb (sminus1)

R2 Half-Life (car-diac cycles)

Control 097 098 083VIV Normal 068 092 119VIV +3 064 094 127VIV +6 034 076 237VIV +8 013 087 640

The valves used in this study were previously unused however the number of times a valveis crimped can affects the stent frame in its deployment final dimensions and performanceWe attempted to minimize these effects by never crimping the valve beyond what was requiredto insert it into the surgical valve (17 mm) Rotational orientation of the valve and its impacton washout was not addressed in this study While rotational alignment in TAVR or VIVprocedures is not often considered in the clinic we recognize that leaflet kinematics couldimpact sinus flow fields (or vice versa) and that out-of-phase deployment could alter washoutto some degree [23] The in vitro aortic chamber model did not incorporate factors such asaortic distensibility or coronary flow and does not account for patient specific anatomy thatcould alter sinus obstruction andor thrombosis risk In addition the velocity fields capturedin this study are 2-dimensional representations of 3-dimensional scenarios While these factorsmay improve sinus washout particularly during diastolic coronary flow this study presents aconsistent worst-case scenario similar to what occurs in the non-coronary sinus However asthe intention of this study was to understand the relative differences in sinus flow patterns withrespect to various THV deployment positions we feel these limitations do not diminish thevalue of the results

Thrombosis in the cardiovascular system is a complex interplay between fluid flow foreignmaterials and biochemistry as described by Virchowrsquos triad While it is known that prostheticheart valves can have an adverse effect on platelet activation flow stagnation regions are neces-sary for these activated platelets to deposit aggregate and form a thrombus [3] The results inthis study clearly demonstrate that flow stagnation in the sinus is highly sensitive to THV de-ployment height Hence it is important for the interventional cardiologist and cardiac surgeonto consider this patient-specific risk of decreased coronary flow andor increased stagnation-induced thrombosis risk during a VIV-TAVR procedure For instance a patient with a narrowor short sinus may experience extreme sinus flow stagnation at a lower deployment positionthan was seen in this study

As shown in previous work the risk of embolization also increases with supra-annular de-ployment due to decreased contact between the THV and surgical bioprosthesis [15] Whilehigh THV implantation in VIV can result in more favorable hemodynamics [15] the impacton embolization risk and sinus flow stagnation should be heavily considered Supra-annular de-ployment leads to increased geometric obstruction of the sinus and our study indicates thatthis results in decreased sinus velocity maxima Particle tracking simulations show that the re-duction in velocities in the sinus result in increased stagnation of blood elements near the baseof the sinus These results indicate that THV deployment greater than +6 mm is inadvisable interms of flow stagnation and potentially increased thrombosis risk in the sinus region howeverfuture studies are needed to build on these findings

9

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

Acknowledgements The authors would like to acknowledge the members of the Cardiovascular Fluid Mechanics Labo-ratory for their assistance and feedback The authors are grateful to our collaborators at Emory University Hospital - DrsVinod Thourani Vasilis Babaliaros Stamatios Lerakis and Jose Condado for providing the valve models and assistancewith placing the work in the context of clinical relevance The glycerin used in this study was generously provided by Proc-ter amp Gamble This study at the CFM lab at the Georgia Institute of Technology was made possible through discretionaryfunds available to the Principal Investigator including the Wallace H Coulter Endowed Chair

References

1 Azadani AN Jaussaud N Matthews PB Ge L Chuter TAM Tseng EE Transcatheter aortic valves in-adequately relieve stenosis in small degenerated bioprostheses Interactive cardiovascular and thoracic surgery 11(1)70ndash7 (2010) DOI 101510icvts2009225144 URL httpicvtsoxfordjournalsorgcontent11170short

2 Brown JM OrsquoBrien SM Wu C Sikora JAH Griffith BP Gammie JS Isolated aortic valve replacementin North America comprising 108687 patients in 10 years changes in risks valve types and outcomes in the Societyof Thoracic Surgeons National Database The Journal of thoracic and cardiovascular surgery 137(1) 82ndash90 (2009)DOI 101016jjtcvs200808015 URL httpwwwncbinlmnihgovpubmed19154908

3 Cannegieter SC Rosendaal FR Briet E Thromboembolic and bleeding complications in patients with me-chanical heart valve prostheses Circulation 89(2) 635ndash641 (1994) DOI 10116101CIR892635 URL httpcircahajournalsorgcgidoi10116101CIR892635

4 Carabello BA Paulus WJ Aortic stenosis The Lancet 373(9667) 956ndash966 (2009) DOI 101016S0140-6736(09)60211-7 URL httpdxdoiorg101016S0140-6736(09)60211-7httplinkinghubelseviercomretrievepiiS0140673609602117

5 Clark MA Duhay Thompson Keyes Svensson Bonow Stockwell Cohen D Clinical and eco-nomic outcomes after surgical aortic valve replacement in Medicare patients Risk Management andHealthcare Policy 5 117 (2012) DOI 102147RMHPS34587 URL httpwwwdovepresscomclinical-and-economic-outcomes-after-surgical-aortic-valve-replacement-peer-reviewed-article-RMHP

6 De Marchena E Mesa J Pomenti S Marin y Kall C Marincic X Yahagi K Ladich E Kutz R AgaY Ragosta M Chawla A Ring ME Virmani R Thrombus Formation Following Transcatheter Aortic ValveReplacement Journal of the American College of Cardiology Cardiovascular Interventions 8(5) 728ndash739 (2015)DOI 101016jjcin201503005 URL httplinkinghubelseviercomretrievepiiS1936879815003441

7 Dvir D Barbanti M Tan J Webb JG Transcatheter aortic valve-in-valve implantation for patients with degen-erative surgical bioprosthetic valves Current problems in cardiology 39(1) 7ndash27 (2014) DOI 101016jcpcardiol201310001 URL httpwwwncbinlmnihgovpubmed24331437

8 Dvir D Webb JG Bleiziffer S Pasic M Waksman R Kodali S Barbanti M Latib A Schaefer U Rodes-Cabau J Treede H Piazza N Hildick-Smith D Himbert D Walther T Hengstenberg C Nissen H Bek-eredjian R Presbitero P Ferrari E Segev A de Weger A Windecker S Moat NE Napodano M WilbringM Cerillo AG Brecker S Tchetche D Lefevre T De Marco F Fiorina C Petronio AS Teles RC TestaL Laborde JC Leon MB Kornowski R Transcatheter Aortic Valve Implantation in Failed Bioprosthetic Sur-gical Valves JAMA 312(2) 162 (2014) DOI 101001jama20147246 URL httpwwwncbinlmnihgovpubmed25005653httpjamajamanetworkcomarticleaspxdoi=101001jama20147246

9 Dvir D Webb JG Piazza N Blanke P Barbanti M Bleiziffer S Wood Da Mylotte D Wilson AB TanJ Stub D Tamburino C Lange R Leipsic J Multicenter evaluation of transcatheter aortic valve replacementusing either SAPIEN XT or CoreValve Degree of device oversizing by computed-tomography and clinical outcomesCatheterization and Cardiovascular Interventions 00(January) nandashna (2015) DOI 101002ccd25823 URL httpdoiwileycom101002ccd25823

10 Hansson N Leetmaa T Leipsic JA Jensen K Andersen HR Jensen JM Webb J Blanke P TangM Noslashrgaard B TCT-665 Early Aortic Transcatheter Heart Valve Thrombosis Diagnostic Value of Contrast-Enhanced Multislice Computed Tomography Journal of the American College of Cardiology 64(11) B193ndashB194 (2014)DOI 101016jjacc201407735 URL httpcontentonlinejaccorgarticleaspxarticleid=1904473httplinkinghubelseviercomretrievepiiS0735109714052759

11 Harjai KJ Paradis JM Kodali S Transcatheter aortic valve replacement game-changing innovation for patientswith aortic stenosis Annual review of medicine 65 367ndash83 (2014) DOI 101146annurev-med-010813-102251 URLhttpwwwncbinlmnihgovpubmed24160938

12 Knight J Kurtcuoglu V Muffly K Marshall W Stolzmann P Desbiolles L Seifert B Poulikakos D AlkadhiH Ex vivo and in vivo coronary ostial locations in humans Surgical and radiologic anatomy SRA 31(8) 597ndash604(2009) DOI 101007s00276-009-0488-9 URL httpwwwncbinlmnihgovpubmed19288041

13 Ku DN Blood flow in Arteries Annual Review of Fluid Mechanics 29(1) 399ndash434 (1997) DOI 101002ar109141041414 Makkar RR Fontana G Jilaihawi H Chakravarty T Kofoed KF de Backer O Asch FM Ruiz CE Olsen

NT Trento A Friedman J Berman D Cheng W Kashif M Jelnin V Kliger Ca Guo H Pichard ADWeissman NJ Kapadia S Manasse E Bhatt DL Leon MB Soslashndergaard L Possible Subclinical LeafletThrombosis in Bioprosthetic Aortic Valves New England Journal of Medicine p 151005110046004 (2015) DOI101056NEJMoa1509233 URL httpwwwnejmorgdoiabs101056NEJMoa1509233

15 Midha PA Raghav V Condado JF Arjunon S Uceda DE Lerakis S Thourani VH Babaliaros VYoganathan AP How Can We Help a Patient With a Small Failing Bioprosthesis JACC Cardiovascular Interventions8(15) 2026ndash2033 (2015) DOI 101016jjcin201508028 URL httplinkinghubelseviercomretrievepiiS1936879815015277

16 Moore B Dasi LP Spatiotemporal complexity of the aortic sinus vortex Experiments in Fluids 55(7) 1770 (2014)DOI 101007s00348-014-1770-0 URL httpwwwncbinlmnihgovpubmed25067881

17 Nkomo VT Gardin JM Skelton TN Gottdiener JS Scott CG Enriquez-Sarano M Burden of valvular heartdiseases a population-based study The Lancet 368(9540) 1005ndash1011 (2006) DOI 101016S0140-6736(06)69208-8URL httplinkinghubelseviercomretrievepiiS0140673606692088

10

18 Petschek H Adamis D Kantrowitz AR Stagnation flow thrombus formation Transactions - American Societyfor Artificial Internal Organs 14 256ndash60 (1968) URL httpwwwncbinlmnihgovpubmed5701540

19 Prasad AK Particle Image Velocimetry Experimental Fluid Mechanics vol 79 Springer Berlin Hei-delberg Berlin Heidelberg (2007) DOI 101007978-3-540-72308-0 URL httpdxdoiorg101007978-3-540-73528-1$delimiter026E30F$nhttplinkspringercom101007978-3-540-72308-0httplinkspringercom101007978-3-540-72308-0

20 Raffel M Willert C Wereley S Kompenhans J Particle Image Velocimetry Experimental Fluid MechanicsSpringer Berlin Heidelberg Berlin Heidelberg (2007) DOI 101007978-3-540-72308-0 URL httplinkspringercom101007978-3-540-72308-0

21 Reul H Vahlbruch A Giersiepen M Schmitz-Rode T Hirtz V Effert S The geometry of the aortic root inhealth at valve disease and after valve replacement Journal of Biomechanics 23(2) 181ndash191 (1990) DOI 1010160021-9290(90)90351-3 URL httpwwwsciencedirectcomsciencearticlepii0021929090903513

22 Saikrishnan N Gupta S Yoganathan AP Hemodynamics of the Boston Scientific Lotus Valve An In VitroStudy Cardiovascular Engineering and Technology 4(4) 427ndash439 (2013) DOI 101007s13239-013-0163-5 URLhttplinkspringercom101007s13239-013-0163-5

23 Saikrishnan N Yoganathan A TRANSCATHETER VALVE IMPLANTATION CAN ALTER THE FLUID FLOWFIELDS IN THE AORTIC SINUSES AND ASCENDING AORTA AN IN VITRO STUDY Journal of the AmericanCollege of Cardiology 61(10) E1957 (2013) DOI 101016S0735-1097(13)61957-9 URL httpcontentonlinejaccorgarticleaspxarticleid=1666166httplinkinghubelseviercomretrievepiiS0735109713619579

24 Wootton DM Ku DN Fluid Mechanics of Vascular Systems Diseases and Thrombosis Annual Review of Biomed-ical Engineering 1(1) 299ndash329 (1999) DOI 101146annurevbioeng11299 URL httpwwwncbinlmnihgovpubmed11701491httpwwwannualreviewsorgdoiabs101146annurevbioeng11299

25 Yap CH Saikrishnan N Tamilselvan G Yoganathan AP Experimental measurement of dynamic fluid shearstress on the aortic surface of the aortic valve leaflet Biomechanics and Modeling in Mechanobiology 11(1-2) 171ndash182(2012) DOI 101007s10237-011-0301-7 URL httplinkspringercom101007s10237-011-0301-7

11