ology 134 (2009) 25–32www.elsevier.com/locate/ijcard

International Journal of Cardi

Relationship of shear stress with in-stent restenosis: Bare metal stentingand the effect of brachytherapy

Michail I. Papafaklis a, Christos V. Bourantas a, Panagiotis E. Theodorakis b,Christos S. Katsouras c, Dimitrios I. Fotiadis a,d, Lampros K. Michalis a,c,⁎

a Michaelideion Cardiac Center, University of Ioannina, Ioannina, Greeceb Department of Materials Science and Engineering, University of Ioannina, Ioannina, Greece

c Department of Cardiology, Medical School, University of Ioannina, Ioannina, Greeced Department of Computer Science, University of Ioannina, Ioannina, Greece

Received 27 August 2007; received in revised form 15 January 2008; accepted 27 February 2008Available online 16 June 2008

Abstract

Background: The association of shear stress (SS) with in-stent restenosis after bare metal stenting is not clear. We investigated thesignificance of SS on predicting areas with neointima thickness (NT) in humans and the effect of vascular brachytherapy (VBT) followingcoronary artery stenting on the relationship of SS with NT.Methods: By using coronary angiography and intravascular ultrasound, we performed three-dimensional (3D) artery and stent reconstructionin 14 patients at 8-month follow-up after bare metal stenting with (stent&VBT group; 7 patients) or without (stent group; 7 patients)adjunctive beta-VBT. In-stent SS was calculated by applying computational fluid dynamics to the stent reconstruction and NT wasdetermined in 3D space.Results: NT was significantly increased in the stent group (0.2±0.16 mm vs. 0.07±0.18 mm, pb0.001). The estimated intercepts and slopesregarding the relationship of NT with SS were 0.28 mm [95% confidence interval (CI): 0.20 to 0.36 mm, pb0.001] and −0.057 mm/Pa(95% CI: −0.062 to −0.052 mm/Pa, pb0.001) respectively for the stent group, and 0.15 mm (95% CI: −0.02 to 0.31 mm, p=0.082)and −0.017 mm/Pa (95% CI: −0.023 to −0.010 mm/Pa, pb0.001) respectively for the stent&VBT group. After pooling all data together, asignificant effect of VBT on the relationship between NT and SS was found [estimate of interaction term (group×SS) for stent&VBT vs. stentgroup: 0.04 mm/Pa; 95% CI: 0.032 to 0.049 mm/Pa, pb0.001].Conclusions: In-stent restenosis is inversely related to SS after coronary artery stenting and VBT diminishes the inverse relationship betweenNT and SS.© 2008 Elsevier Ireland Ltd. All rights reserved.

Keywords: Stents; Intravascular ultrasound; Three-dimensional reconstruction; Computational fluid dynamics; Hemodynamics; Radiation therapy

1. Introduction

Restenosis constitutes the major drawback after percuta-neous transluminal coronary angioplasty, occurring in ap-

⁎ Corresponding author. Department of Cardiology, Medical School,University of Ioannina, University Avenue, Ioannina GR 45110, Greece.Tel.: +30 26510 97710; fax: +30 26510 97017.

E-mail address: lmihalis@cc.uoi.gr (L.K. Michalis).

0167-5273/$ - see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.ijcard.2008.02.006

proximately 40 to 60% of patients within the first 6 monthsafter the procedure [1], and all research efforts during the last2 decades were targeted in reducing this phenomenon.Coronary stenting has proven to be beneficial in blockingvascular contraction and reducing restenosis rates; however,in-stent restenotic lesions appear in relatively high rates of 20to 30% [2,3]. Radiation therapy emerged in the mid-90s as atherapeutic modality, which could diminish neointimal pro-liferation and hyperplasia, and thus restenosis. The applica-tion of beta- or gamma-radiation therapy in both animal

26 M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

studies [4] and randomized clinical trials proved the efficacyof vascular brachytherapy (VBT) in preventing clinical andangiographic recurrence of in-stent restenosis [5,6].

Neointimal hyperplasia, which is the major contribut-ing pathophysiologic mechanism of in-stent restenosis, isinfluenced by clinical characteristics, such as diabetes [7],and other parameters, such as residual stenosis, number ofstents, stent length [8] and plaque burden [9]. Recently,shear stress (SS) has also been implicated in the processof in-stent restenosis, given the long known causativecorrelation of atherosclerotic disease with the overlyingflow pattern and hemodynamic parameters [10]. However,two early studies on the relationship of SS with in-stentneointimal hyperplasia in humans reported contradictoryresults [11,12]; one found a significant inverse correlationbetween SS and neointima thickness (NT) after bare metalstenting, while the other reported a limited role of SS in in-stent restenosis.

In the current study, we provide additional data on thevalue of SS in predicting NT at follow-up after bare metalstent implantation using realistic three-dimensional (3D)coronary artery reconstruction and computational fluiddynamics. Moreover, we investigate the effect of beta-VBT, used in the recent past for reducing in-stent restenosis,on the relationship of NT with SS.

2. Materials and methods

2.1. Patient population

Fourteen patients were studied 8 months after coronaryartery stenting with or without adjunctive beta-VBT; 7 non-consecutive patients were included in each group (stentgroup and stent&VBT group). Patients were eligible if theyhad underwent angiographically successful bare metal stentimplantation for de novo lesions in native coronary arteries(stent length≥12 mm), while exclusion criteria includedacute myocardial infarction as an indication for the initialintervention, ejection fraction b50% and serum creatinineN1.5 mg/dl.

Patients included in the stent&VBT group were part of alarger study investigating the acute and long-term outcomefollowing beta-VBT [13]. Vascular brachytherapy had beenperformed using a 90Strontium/90Yttrium beta source storedin a transfer device and delivered intravascularly via a non-centering 5Fr catheter (Novoste Beta-Cath™); only caseswith successful brachytherapy, defined as complete (N90%)delivery of the prescribed radiation dose including doseinterruption and resumption, were included.

All patients received post-intervention dual anti-platelettherapy; aspirin 100 mg/day indefinitely and clopidogrel75 mg/day for 1 month after bare metal stent implantationand indefinitely after bare metal stenting with adjunctivebeta-VBT. The study was approved by the institutionalboard of our hospital and each patient provided informedconsent.

2.2. Imaging and three-dimensional reconstruction ofcoronary arteries

Three-dimensional reconstruction of the coronary arterieswas performed using a method based on the fusion of biplaneangiographic and intravascular ultrasound data as we havepreviously reported [14]. In brief, the IVUS (Boston Scimed,MA, USA) catheter was advanced in the coronary arteryunder fluoroscopic control distally from the stented segmentand following intracoronary administration of nitroglycerinangiographic images from two different views were obtainedjust after the start of the pullback during diluted contrast agentinjection. IVUS was performed using a motorized pullback ata speed of 0.5 mm/s. The ECG was recorded in both theangiographic and IVUS sequence. The biplane angiographicimages at end-diastole (at the peak of the R wave), a timeduring which the heart is relatively stable, were used toreconstruct the catheter path in 3D space. The obtained IVUSsequence was digitized and the lumen and stent borders in theend-diastolic IVUS frames were detected using a previouslydescribed and well validated semi-automatic segmentationalgorithm [15,16]. The algorithm is based on the basicprinciples of deformable models (snakes) and incorporates aHopfield neural network combined with a simulated anneal-ing schema for fast energy minimization. Initially, an expertobserver identified the regions of interest (lumen and stentborders) in the first IVUS frame. In following frames theregions of interest were identified in the neighbourhood ofthe previously detected borders and were defined at the re-gions where the model's energy is minimized. Then, thelumen contours were placed perpendicularly onto the recon-structed 3D catheter path. Their relative axial twist was esti-mated by the sequential triangulation algorithm, while theabsolute orientation of the first contour was determined af-ter comparing the resultant reconstructed artery and path withthe catheter path and the silhouette of the lumen, as thesewere depicted in the biplane angiographies.

Two 3D reconstructions were obtained for each coronaryartery. In the first (baseline) reconstruction, the lumencontours for the non-stented region of the artery and thestent contours (determined by the leading edge of the stentstruts) for the stented region were used to represent thecoronary lumen at the moment of stent implantation as it hasbeen previously implemented [11]. A second reconstructionrepresenting the lumen at follow-up was also performedbased on the lumen contours in both the stented and non-stented region of the artery. SS was computed at the stentsurface using the baseline 3D reconstruction. Fig. 1 displaysan example of the coronary angiogram and the 3D re-constructed coronary artery and stent.

2.3. Computational fluid dynamics

The 3D reconstructed models enabled the generation of afinite volume mesh using a software package (ICEM CFD 5,Ansys, PA, USA). The mesh consisted of hexahedral

Fig. 1. Angiogram of a right coronary artery (A) with a region of in-stentrestenosis (arrows) and three-dimensional reconstruction (B) of the arteriallumen (red) and stent (green).

27M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

elements with a resolution of ≈0.5 mm along the axialdirection (Fig. 2A). Around the circumference there were 72equal intervals and each cross-section contained 684hexahedrons, each covering a cross-sectional area rangingfrom 0.004 mm2 to 0.07 mm2 with the highest resolutionnear the wall (Fig. 2B).

The detailed characteristics of the intra-coronary flowfield and specifically the SS calculations at the stent surfacewere obtained by solving the incompressible 3D Navier–Stokes equations in each node of the computational mesh(CFX 10, Ansys, PA, USA). We modeled blood as a

Fig. 2. (A) Detailed view of a stented arterial segment (green colour represents thesurface of the hexahedral elements of the computational mesh with a resolutioncomputational mesh. (C) Calculation of neointima thickness around the vessel circuthe green line represents the stent border, the red line the lumen border, the orange a

homogeneous and Newtonian fluid with a dynamic viscosityof 0.0035 Pa∙s and a density of 1050 kg/m3, while the arterialwall was considered to be rigid. Laminar flow was assumedand at the inlet of the artery a steady flow of 60 ml/min with aparabolic profile was specified in all cases according toliterature data about average coronary blood flow [17].Steady instead of pulsatile flow was used, since averageresults from a pulsatile flow calculation are similar to thosefrom a steady one as previously reported [18]. In addition,the relationship between NT and SS was investigated in arelative sense, i.e. the difference in NT values between lowand high SS regions, making the application of patientspecific flow data less important. Zero pressure conditionswere imposed at the outlet and the no-slip condition wasapplied at the wall. SS at the stent surface was computed asthe product of the gradient of blood velocity at the wall andviscosity.

2.4. Analysis and statistics

The analysis was performed only for the stented arterialsegments. The location of in-stent side branches wasidentified by IVUS and the corresponding cross-sections,as well as adjacent segments with a length equal to thediameter of the side branch, were excluded from our dataset.In addition, a part of the segment at the entrance and exit ofthe stent, covering a length of 2.5 mm, was also excluded.The aforementioned restrictions were applied in order to: 1)reduce the influence of side branches, which were notconsidered in our 3D models, on the flow field; 2) minimizethe impact of the proximal/distal edges of a stent; 3) avoidthe inclusion of any edge restenosis (especially after beta-VBT) in our results.

NT was calculated as the difference between the stent(baseline) and the lumen (follow-up) model in 3D space usingin-house developed algorithm implemented in Visual Fortran(Compaq Computer Corporation, Texas, USA). Negative NTdenoted that the lumen border was outside the extrapolated

stent and red colour the restenotic lumen). Green lines delineate the externalof ≈0.5 mm along the axial direction. (B) Cross-sectional view of themference along the line (white) passing through the luminal center of mass;rea the neointima (in-stent restenosis) and the blue area the tissue regression.

Table 2Area and volume measurements in the two patient groups.

Parameter Bare metalstents

Bare metal stentsand brachytherapy

pvalue

(n=7) (n=7)

Mean stent area (mm2) 6.87±1.28 8.97±2.37 0.061Mean lumen area (mm2) 5.52±1.69 7.82±1.01 0.009Mean NIH area burden a (%) 20.66±12.47 10.05±11.40 0.123Stent volume (mm3) 112.07±31.23 213.95±41.22 0.001Lumen volume (mm3) 89.18±15.97 194.35±52.99 0.001NIH volume (mm3) 22.89±17.87 19.60±27.42 0.795NIH volume burden b (%) 18.54±10.48 10.16±14.23 0.234

NIH: neointimal hyperplasia.a NIH area burden=(stent area− lumen area) / stent area.b NIH volume burden=(stent volume− lumen volume) / stent volume.

28 M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

stent border and thus, lumen enlargement was present. Indetail, for each cross-section NT was determined at 36locations around the vessel circumference as the distancefrom the stent to the lumen border along the line passingthrough the luminal center of mass (Fig. 2C); SS wascomputed at the same locations. After excluding all theabovementioned in-stent segments from the dataset, therewas an average of 954 points (range 252–1512) per stentedarterial segment, which were available for investigating therelationship of NT with SS. Furthermore, the area of eachcross-section was measured, while volume measurementswere derived from the 3D reconstructed models.

Categorical variables (presented as counts and percen-tages) were compared between groups using Fisher's exacttest. The unpaired t-test was used for comparing continuousvariables (presented as the mean±standard deviation). Thecorrelation between NT and SS for each patient wasinvestigated using linear regression analysis. After poolingall data a linear mixedmodel controlling for patient-to-patientvariation was used for estimating the overall relationship ofNT with SS and the influence of VBT (stent&VBT vs. stentgroup) on this relationship; the parameters entered in thismodel were SS, intervention group and the interaction ofSS with intervention group. All tests were two-tailed anda p valueb0.05 was considered significant. The SPSS sta-tistical software package (version 13.0 for Windows, SPSSInc. Chicago, USA) was used for the analysis.

3. Results

Three-dimensional reconstruction of the coronary arteriesand stents was performed successfully in all patients in bothgroups. The comparative baseline demographic and proce-dural characteristics of the two patient groups are presentedin Table 1; mean stent length was bigger in the stent&VBTgroup reflecting our interventional practice to use VBTmainly in diffuse lesions (lesion length≥20 mm) with higher

Table 1Baseline clinical and procedural characteristics of the two patient groups.

Characteristic Bare metalstents

Bare metal stentsand brachytherapy

pvalue

(n=7) (n=7)

Age 63.0±11.4 65.6±12.5 0.695Male gender 7 (100%) 6 (85.7%) 1.000

Cardiovascular risk factorsDiabetes 5 (71.4%) 3 (42.9%) 0.592Hypertension 5 (71.4%) 2 (28.6%) 0.286Hypercholesterolemia 4 (57.1%) 5 (71.4%) 1.000Smoking 2 (28.6%) 4 (57.1%) 0.592Vessels treated 1.000Left anterior descending 5 (71.4%) 4 (57.1%)Right coronary artery 2 (28.6%) 3 (42.9%)Stent length (mm) 17.4±4.8 24.1±6.4 0.047Stent diameter (mm) 3.1±0.2 3.3±0.4 0.430

restenotic risk. The stent designs, which were implanted, hadsimilar characteristics (e.g. strut thickness) and were thefollowing: AVE (Medtronic, MN, USA) in 3 patients,Multilink Tetra (Guidant, CA, USA) in 2 patients, JoStentFlex (Jomed, Ragendingen, The Netherlands) in 1 patientand Express (Boston Scimed, MA, USA) in 1 patient in thestent group, and AVE in 3 patients, Multilink Tetra in 2patients, JoStent Flex in 1 patient and Coroflex (B.Braun,PA, USA) in 1 patient in the stent&VBT group.

Mean area and volume measurements were significantlyhigher both for the stent and the lumen at follow-up in thestent&VBT group (Table 2) reflecting the higher meanlength and diameter of the stents implanted in these patients.Neointimal hyperplasia was more prominent in the stentgroup with approximately double neointimal hyperplasiaarea and volume burden. Furthermore, NT was significantlyincreased in the stent group (0.2±0.16 vs. 0.07±0.18 mm,pb0.001). Lumen enlargement (negative NT) was morefrequently observed in the stent&VBT group (38% of allmeasured stent locations; minimum value: −0.13 mm)compared to the stent group (7% of all measured stentlocations; minimum value: −0.10 mm). SS was significantlyhigher in the stent group (1.98±0.98 Pa vs. 1.25±0.52 Pa,pb0.001).

3.1. Relation of NT to SS

The results of the linear regression analysis for each pa-tient in both intervention groups are summarized in Table 3.In the majority of patients with bare metal stents NT wasfound to be significantly correlated to SS with an inverserelationship; the correlation coefficient (r) varied from −0.63to −0.16. Fig. 3 displays an example of a stented segmentwith regions of low baseline SS corresponding to increasedNT. Two patients showed a non-significant relationship.After applying a linear mixed model for all data in the stentgroup the estimated intercept and slope were 0.28 mm [95%confidence interval (CI): 0.20 to 0.36 mm, pb0.001] and−0.057 mm/Pa (95% CI: −0.062 to −0.052 mm/Pa,pb0.001) respectively.

Table 3Correlations between neointima thickness and shear stress for all patients inthe two patient groups.

Interventiongroup

Patient r Slope(mm/Pa)

95% CI(mm/Pa)

p value

Bare metalstents

#1 −0.16 −0.061 −0.086 to −0.036 b0.001#2 −0.63 −0.087 −0.098 to −0.076 b0.001#3 −0.46 −0.085 −0.094 to −0.076 b0.001#4 −0.01 −0.001 −0.004 to 0.003 0.858#5 −0.56 −0.126 −0.141 to −0.111 b0.001#6 0.11 0.01 −0.001 to 0.021 0.076#7 −0.17 −0.027 −0.035 to −0.019 b0.001

Bare metalstentsandVBT

#8 0.15 0.025 0.015 to 0.035 b0.001#9 −0.17 −0.241 −0.382 to −0.099 b0.001#10 0.07 3.6E−04 1.2E−04 to 5.9E−04 0.003#11 0.07 0.01 4.9E−04 to 0.02 0.039#12 0.05 0.003 4.4E−04 to 0.005 0.019#13 −0.58 −0.99 −1.085 to −0.884 b0.001#14 −0.21 −0.018 −0.024 to −0.012 b0.001

VBT: vascular brachytherapy.

29M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

In the stent&VBT group the results per patient variedfrom a quite strong inverse relationship (r=−0.58) betweenNT and SS to a weak positive relationship (r=0.15). Theestimated intercept and slope for all patients in this groupwere 0.15 mm (95% CI: −0.02 to 0.31 mm, p=0.082) and−0.017 mm/Pa (95% CI: −0.023 to −0.010 mm/Pa,pb0.001) respectively.

When pooling all data from both groups a statisticallysignificant linear relation between NTand SS with a negativeslope (−0.057 mm/Pa; 95% CI: −0.062 to −0.053 mm/Pa,pb0.001) was found. VBT was also found to significantlyalter the relationship of NT with SS [estimate of interactionterm (group×SS) for stent&VBT vs. stent group: 0.04 mm/Pa; 95% CI: 0.032 to 0.049 mm/Pa, pb0.001], while the

Fig. 3. Three-dimensional colour-coded maps demonstrating (A) the shear stress (Sthickness (NT). Areas (circles) with increased NT correspond to areas with low S

intervention group per se was not found to be a significantparameter for predicting NT (estimated parameter forstent&VBT vs. stent group: −0.13; 95% CI: −0.30 to 0.04,p=0.114).

4. Discussion

The present study provides data about the relationship ofSS with the subsequent formation of in-stent neointimalhyperplasia in humans and is the first to investigate the effectof intravascular radiation therapy on this relationship. Weobserved an inverse correlation between NT and SSindicating the development of most profound neointima inregions of low SS. However, this correlation seems to beaffected (diminished) by VBT.

4.1. The role of SS in neointimal proliferation

Although atherosclerosis is a systemic disease, its mani-festations are focal and eccentric [19]. Within each patient,each coronary obstruction progresses, regresses, or remainsquiescent in an independent manner. Neointima formationcausing in-stent restenosis is often observed at specificlocations in the stented segment [20]. Local hemodynamicfactors that create a unique environment seem to be a majordeterminant of the behavior of atherosclerosis and restenosis ina susceptible individual. It has been observed that regions ofmoderate to high SS, where flow remains unidirectional andaxially aligned, are spared of intimal thickening, while focallesions develop only in areas of low and recirculating flow[10].

The pathophysiology of in-stent restenosis involvesaccumulation of new tissue within the arterial wall and thebasis of neointima formation is a cascademechanism involving

S) distribution on the surface of a stent and (B) the corresponding neointimaS and vice versa. Different views (C, D) are also provided.

30 M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

platelet, leukocyte and macrophage aggregation leading tomedial smooth muscle cell migration and proliferation in theintima [21]. Normal or high SS seems to exert protectivefunctions on the endothelial layer by increasing nitric oxidesynthase [22] and reducing endothelial cell death and pro-liferation [23], while low SS induces the production of in-flammatory molecules and reactive oxygen species [24]. SSalso acts through the endothelium to modulate smooth musclecell gene expression and proliferation [25].

Our findings, derived from patients with mild neointimalhyperplasia (mean NT: 0.2 mm) following bare metal stentimplantation, indicated an inverse correlation of NT to SS(r range: −0.63 to −0.01) in all patients but one. Previousdata on the relationship of SS with the process of in-stentrestenosis in humans are limited and not clear. The firstpublished results based on patient data after Wallstent im-plantation demonstrated a significant negative correlation(r range: −0.65 to −0.04) between SS and NT (mean:0.44 mm) in 9 out of the 14 patients studied [11]. In a laterstudy [12], however, neointimal hyperplasia in 6 stentedcoronary segments appeared to occur to some degree in allcategories of SS values resulting in a limited role of SS inin-stent restenosis. More recently, a negative correlation(r range: −0.36 to −0.21) of in-stent NT (mean: 0.29 mm)to SS was also reported in 5 out of the 7 patients studied[26].

Based on the available literature data and our currentresults, the role of SS in neointimal proliferation as dem-onstrated in experimental models also applies to humans andin-stent restenosis. However, SS cannot predict the exactlocations of neointimal hyperplasia in all cases, since thethickening does not necessarily occur in all regionswhere SS islow and not all regions where SS is high are spared fromneointimal hyperplasia as demonstrated in Fig. 3. This reflectsthe multiple pathophysiological mechanisms that are involvedin the restenotic process. Therefore, it would be better toconsider regions with low SS as regions with a higher prob-ability for neointimal hyperplasia to occur and be moreprofound. The incorporation of additional parameters in the SSprediction model could increase its accuracy and presumablyprove more useful in the future.

4.2. Effect of VBT

Vascular radiation exerts its anti-restenotic effect by directlydamaging the cell's DNA and causing its death [27]; the mainradiation target cell is thought to be the vascular smoothmuscle cell [28]. Application of VBT in experimental modelsleads to significant reduction in intimal proliferation [4,29],and clinical randomized trials demonstrated low restenosisrates in the irradiated patients [30,31]. However, the initialgood results were followed by increasing reports about itscomplications, such as late occlusion especially with the use ofstents for de novo lesions [6,32], edge restenosis/geographicalmiss [33], and lumen enlargement/late stent malapposition[34], which limited its use.

In the current report, in-stent neointimal hyperplasia wasless prominent in the irradiated group, as expected, withdecreased mean NT, neointima area and volume burden.However, in some cases of the stent&VBT group there waslocally highly increased neointimal response (maximum NTfor stent&VBT group: 1.17 mm vs. 0.83 mm). These non-uniform results could be explained by the fact that applicationof VBTcan be non-uniform intrarterially, especially in curvedsegments, when using non-centering catheters and this resultsin reduced dose delivery and anti-restenotic effect [35].Neointima regression (lumen enlargement) behind the stentstruts was more frequent in the irradiated stent segments,even if no stent malapposition was observed.

Irradiated patients showed a high diversity in the re-lationship between NT and SS; in 3 patients a significantnegative correlation of medium strength (r range: −0.58 to−0.17) was observed, while in the other 4 a very weakpositive correlation (r range: 0.05 to 0.15) was found. Afterpooling all data, VBT was found to affect the relationshipbetween NT and SS, while the intervention group per se wasnot a significant predictor of NT.

The aforementioned variety is thought to reflect the min-imal neointimal hyperplasia in most in-stent locations withrelatively few local spots of increased neointimal response inthe irradiated patients. Our hypothesis is that in arterial regions,where radiation dose is efficient and NT is minimal, there is noeffect of SS on in-stent restenosis, since the cells that SS affectshave been previously damaged by VBT; in arterial regions,where radiation dose is low or non-efficient, the biologicalresponse of the endothelium and smooth muscle cells to SS isretained. Whether the altered relationship between NT and SSis mainly due to smooth muscle cell mortality or a dysfunctionof endothelial cells– and indirectly of smoothmuscle cells– tosense SS is not known; probably both mechanisms contribute.This presumptive scenario could also be of value in the case ofdrug-eluting stents, which nowadays have dominated as ananti-restenotic modality due to the deleterious effect they havein the cell cycle and thus, the reduction of cell proliferation.Based on an initial small-sized study and preliminary resultsfrom our group [36,37], it seems that NT after sirolimus-eluting stent implantation is also inversely correlated to SS;further research is needed in that direction.

4.3. Limitations

Our study is limited by the small number of patients ineach group. However, the large number of locations (points)in each stented segment both in the axial and the cir-cumferential direction, where NTand SS measurements weremade and interrelated, strengthens the validity of our report.

For the 3D reconstruction of coronary arteries and stents,no baseline data were available and therefore, the stentborders in the stented segment at follow-up (any minimalpost-PCI stent malapposition that could be identified onlywith post-stenting IVUS interrogation cannot be excluded)and the lumen borders in the non-stented segment (assuming

31M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

that there was no progress in the atherosclerotic process andthus, no change in lumen morphology) were used to ap-proximate arterial morphology at baseline. Since our anal-ysis was restricted only in the stented segment (excludingalso a part of the segment at the entrance and exit of thestent), it is not likely that these assumptions have influencedthe results.

We also acknowledge the assumptions of Newtonianviscosity and rigid arterial walls (ignoring wall complianceand the effects of cardiac motion) in the computational fluiddynamics analysis. However, it has been previously reportedthat these assumptions do not create significant errors in theanalysis [38–40]. Furthermore, wall compliance is actuallynot an issue in stented arterial segments which are rigid.

5. Conclusions

The present study provides additional evidence of the roleof hemodynamics in the process of in-stent restenosis basedon real patient coronary artery reconstructions and bloodflow simulations. Profound neointimal hyperplasia mostlypresents in regions with low SS. The effect of VBT on therelationship of NTwith SS was investigated for the first timeand it was found that VBT diminished the negative impact ofSS on NT.

References

[1] Detre K, Holubkov R, Kelsey S, et al. Percutaneous transluminalcoronary angioplasty in 1985–1986 and 1977–1981. TheNational Heart,Lung, and Blood Institute Registry. N Engl J Med 1988;318:265–70.

[2] Serruys PW, de Jaegere P, Kiemeneij F, et al. A comparison of balloon-expandable-stent implantation with balloon angioplasty in patientswith coronary artery disease. Benestent Study Group. N Engl J Med1994;331:489–95.

[3] Kimura T, Abe K, Shizuta S, et al. Long-term clinical and angiographicfollow-up after coronary stent placement in native coronary arteries.Circulation 2002;105:2986–91.

[4] WaksmanR,RobinsonKA,Crocker IR, GravanisMB, CipollaGD,KingIII SB. Endovascular low-dose irradiation inhibits neointima formationafter coronary artery balloon injury in swine. A possible role for radiationtherapy in restenosis prevention. Circulation 1995;91:1533–9.

[5] Teirstein PS, Massullo V, Jani S, et al. Three-year clinical and angi-ographic follow-up after intracoronary radiation: results of a randomizedclinical trial. Circulation 2000;101:360–5.

[6] Serruys PW, Wijns W, Sianos G, et al. Direct stenting versus directstenting followed by centered beta-radiation with intravascularultrasound-guided dosimetry and long-term anti-platelet treatment:results of a randomized trial: Beta-Radiation Investigation with DirectStenting and Galileo in Europe (BRIDGE). J Am Coll Cardiol2004;44:528–37.

[7] Abizaid A, Kornowski R, Mintz GS, et al. The influence of diabetesmellitus on acute and late clinical outcomes following coronary stentimplantation. J Am Coll Cardiol 1998;32:584–9.

[8] Kasaoka S, Tobis JM, Akiyama T, et al. Angiographic andintravascular ultrasound predictors of in-stent restenosis. J Am CollCardiol 1998;32:1630–5.

[9] Hibi K, Suzuki T, Honda Y, et al. Quantitative and spatial relation ofbaseline atherosclerotic plaque burden and subsequent in-stent neointi-mal proliferation as determined by intravascular ultrasound.Am JCardiol2002;90:1164–7.

[10] Malek AM, Alper SL, Izumo S. Hemodynamic shear stress and its rolein atherosclerosis. Jama 1999;282:2035–42.

[11] Wentzel JJ, Krams R, Schuurbiers JC, et al. Relationship betweenneointimal thickness and shear stress after Wallstent implantation inhuman coronary arteries. Circulation 2001;103:1740–5.

[12] Stone PH, Coskun AU, Kinlay S, et al. Effect of endothelial shear stresson the progression of coronary artery disease, vascular remodeling, andin-stent restenosis in humans: in vivo 6-month follow-up study.Circulation 2003;108:438–44.

[13] Nikas DN, Kalef-Ezra J, Katsouras CS, et al. Long-term clinicaloutcome of patients treated with beta-brachytherapy in routine clinicalpractice. Int J Cardiol 2007;115:183–9.

[14] Bourantas CV, Kourtis IC, Plissiti ME, et al. A method for 3D re-construction of coronary arteries using biplane angiography andintravascular ultrasound images. Comput Med Imaging Graph2005;29:597–606.

[15] Plissiti ME, Fotiadis DI, Michalis LK, Bozios GE. An automatedmethod for lumen and media-adventitia border detection in a sequenceof IVUS frames. IEEE Trans Inf Technol Biomed 2004;8:131–41.

[16] Bourantas CV, Plissiti ME, Fotiadis DI, et al. In vivo validation of anovel semi-automated method for border detection in intravascularultrasound images. Br J Radiol 2005;78:122–9.

[17] Nichols WW, O'Rourke MF. McDonald's blood flow in arteries:theoretical, experimental and clinical principles. 4th ed. London:Arnold; 1998.

[18] Feldman CL, Ilegbusi OJ, Hu Z, Nesto R, Waxman S, Stone PH.Determination of in vivo velocity and endothelial shear stress patternswith phasic flow in human coronary arteries: a methodology to predictprogression of coronary atherosclerosis. Am Heart J 2002;143:931–9.

[19] Asakura T, Karino T. Flow patterns and spatial distribution of athero-sclerotic lesions in human coronary arteries. Circ Res 1990;66:1045–66.

[20] Thury A, Wentzel JJ, Vinke RV, et al. Images in cardiovascularmedicine. Focal in-stent restenosis near step-up: roles of low andoscillating shear stress? Circulation 2002;105:e185–7.

[21] Carter AJ, Laird JR, Farb A, Kufs W, Wortham DC, Virmani R.Morphologic characteristics of lesion formation and time course ofsmooth muscle cell proliferation in a porcine proliferative restenosismodel. J Am Coll Cardiol 1994;24:1398–405.

[22] Nadaud S, Philippe M, Arnal JF, Michel JB, Soubrier F. Sustainedincrease in aortic endothelial nitric oxide synthase expression in vivo ina model of chronic high blood flow. Circ Res 1996;79:857–63.

[23] Cho A, Mitchell L, Koopmans D, Langille BL. Effects of changes inblood flow rate on cell death and cell proliferation in carotid arteries ofimmature rabbits. Circ Res 1997;81:328–37.

[24] Passerini AG, Polacek DC, Shi C, et al. Coexisting proinflamma-tory and antioxidative endothelial transcription profiles in a disturbedflow region of the adult porcine aorta. Proc Natl Acad Sci U S A2004;101:2482–7.

[25] Chiu JJ, Chen LJ, Chen CN, Lee PL, Lee CI. A model for studying theeffect of shear stress on interactions between vascular endothelial cellsand smooth muscle cells. J Biomech 2004;37:531–9.

[26] Sanmartin M, Goicolea J, Garcia C, et al. [Influence of shear stress onin-stent restenosis: in vivo study using 3D reconstruction andcomputational fluid dynamics]. Rev Esp Cardiol 2006;59:20–7.

[27] Rubin P, Soni A,Williams JP. The molecular and cellular biologic basisfor the radiation treatment of benign proliferative diseases. SeminRadiat Oncol 1999;9:203–14.

[28] Brenner DJ, Miller RC, Hall EJ. The radiobiology of intravascularirradiation. Int J Radiat Oncol Biol Phys 1996;36:805–10.

[29] Wiedermann JG, Marboe C, Amols H, Schwartz A, Weinberger J.Intracoronary irradiation markedly reduces restenosis after balloonangioplasty in a porcine model. J Am Coll Cardiol 1994;23:1491–8.

[30] Raizner AE, Oesterle SN, Waksman R, et al. Inhibition of restenosiswith beta-emitting radiotherapy: Report of the Proliferation Reductionwith Vascular Energy Trial (PREVENT). Circulation 2000;102:951–8.

[31] Verin V, Popowski Y, de Bruyne B, et al. Endoluminal beta-radiationtherapy for the prevention of coronary restenosis after balloon

32 M.I. Papafaklis et al. / International Journal of Cardiology 134 (2009) 25–32

angioplasty. The Dose-Finding Study Group. N Engl J Med2001;344:243–9.

[32] Costa MA, Sabate M, van der GiessenWJ, et al. Late coronary occlusionafter intracoronary brachytherapy. Circulation 1999;100:789–92.

[33] Sianos G, Kay IP, Costa MA, et al. Geographical miss during catheter-based intracoronary beta-radiation: incidence and implications inthe BRIE study. Beta-Radiation In Europe. J Am Coll Cardiol2001;38:415–20.

[34] Kozuma K, Costa MA, Sabate M, et al. Late stent malappositionoccurring after intracoronary beta-irradiation detected by intravascularultrasound. J Invasive Cardiol 1999;11:651–5.

[35] Wahle A, Lopez JJ, Pennington EC, et al. Effects of vessel geometryand catheter position on dose delivery in intracoronary brachytherapy.IEEE Trans Biomed Eng 2003;50:1286–95.

[36] Gijsen FJ, Oortman RM, Wentzel JJ, et al. Usefulness of shear stresspattern in predicting neointima distribution in sirolimus-eluting stentsin coronary arteries. Am J Cardiol 2003;92:1325–8.

[37] Papafaklis MI, Bourantas CV, Theodorakis PE, Katsouras CS, FotiadisDI, Michalis LK. Shear stress and prediction of neointima distribution:sirolimus-eluting stents versus bare metal stents [Abstract]. Am JCardiol 2006;98:S174.

[38] Friedman MH, Bargeron CB, Duncan DD, Hutchins GM, Mark FF.Effects of arterial compliance and non-Newtonian rheology on correla-tions between intimal thickness and wall shear. J Biomech Eng1992;114:317–20.

[39] Delfino A, Moore Jr JE, Meister JJ. Lateral deformation and movementeffects on flow through distensible tube models of blood vessels.Biorheology 1994;31:533–47.

[40] Zeng D, Ding Z, FriedmanMH, Ethier CR. Effects of cardiac motion onright coronary artery hemodynamics. Ann Biomed Eng 2003;31:420–9.