OUP user menu

Extent of flow recirculation governs expression of atherosclerotic and thrombotic biomarkers in arterial bifurcations

Jordi Martorell, Pablo Santomá, Kumaran Kolandaivelu, Vijaya B. Kolachalama, Pedro Melgar-Lesmes, José J. Molins, Lawrence Garcia, Elazer R. Edelman, Mercedes Balcells
DOI: http://dx.doi.org/10.1093/cvr/cvu124 37-46 First published online: 19 May 2014

Abstract

Aims Atherogenesis, evolution of plaque, and outcomes following endovascular intervention depend heavily on the unique vascular architecture of each individual. Patient-specific, multiscale models able to correlate changes in microscopic cellular responses with relevant macroscopic flow, and structural conditions may help understand the progression of occlusive arterial disease, providing insights into how to mitigate adverse responses in specific settings and individuals.

Methods and results Vascular architectures mimicking coronary and carotid bifurcations were derived from clinical imaging and used to generate conjoint computational meshes for in silico analysis and biocompatible scaffolds for in vitro models. In parallel with three-dimensional flow simulations, geometrically realistic scaffolds were seeded with human smooth muscle cells (SMC) or endothelial cells and exposed to relevant, physiological flows. In vitro surrogates of endothelial health, atherosclerotic progression, and thrombosis were locally quantified and correlated best with an quantified extent of flow recirculation occurring within the bifurcation models. Oxidized low-density lipoprotein uptake, monocyte adhesion, and tissue factor expression locally rose up to three-fold, and phosphorylated endothelial nitric oxide synthase and Krüppel-like factor 2 decreased up to two-fold in recirculation areas. Isolated testing in straight-tube idealized constructs subject to static, oscillatory, and pulsatile conditions, indicative of different recirculant conditions corroborated these flow-mediated dependencies.

Conclusions Flow drives variations in vascular reactivity and vascular beds. Endothelial health was preserved by arterial flow but jeopardized in regions of flow recirculation in a quasi-linear manner. Similarly, SMC exposed to flow were more thrombogenic in large recirculating regions. Health, thrombosis, and atherosclerosis biomarkers correlate with the extent of recirculation in vascular cells lining certain vascular geometries.

  • Flow
  • Endothelium
  • Geometry
  • Atherosclerosis
  • Thrombosis

1. Introduction

Endothelial cells (EC) are exposed to a complex biomechanical milieu and are responsible for relaying biochemical messages to underlying tissue. Disrupted flow patterns1,2 and, more specifically, flow recirculation exacerbate low-density lipoprotein internalization,3 cell adhesion molecules expression,4,5 and monocyte adhesion6,7 to the endothelium. Such events, in conjunction with other environmental factors,8 are early determinants for atherosclerosis progression.9 Though flow disruptions are evident in every vascular bifurcation, their magnitude depend heavily on the geometries specific to every patient and vessel.

The clinical treatment for atherosclerosis has tremendously evolved over many decades, with percutaneous interventions such as stents and/or grafts becoming almost a commodity in patient care.10 Device deployment, however, creates local, deep injury11 in the already fragile diseased vessel. Endothelial denudation12 exposes subintimal smooth muscle cells (SMC) to blood flow, and a cascade of reactions is elicited that modulates vessel patency and may trigger the extrinsic coagulation cascade.

Unravelling the molecular mechanisms behind vascular cell response to physiological and pathological flow characteristics has been for decades a matter of intense research.13 Both clinical data, obtained with tremendous advances in the field of bioimaging,8,1416 and animal models have helped us understand how haemodynamics affect the vessel wall.17,18 Parallel in vitro work1921 with idealized bench-top model systems has enabled the cellular and molecular examination of EC biological response to isolated flow descriptors (average flow, maximum amplitude, and flow frequency) in the presence and absence of SMC.

We now test the hypothesis that subtle variations in flow arising in different bifurcation settings are the most powerful predictors of biological markers crucial to our understanding of atherothrombotic disease. We examined how vessel-like, bench-top constructs derived from specific patient geometries create a more precise view of the underlying relationships between flow disruptions and local expressions of atherogenic and thrombotic markers. Using invasive22 and non-invasive imaging techniques,23 we reconstructed arterial geometries of different patients for use in a computational model, casted in vitro constructs based on the extracted geometry features, and seeded human vascular cells mimicking the arterial wall components. Computational fluid dynamic (CFD) simulations predicted physiological metrics of interest including velocity and quantified regions of flow recirculation. Bench-top-derived measurements of atherogenic and thrombotic markers and their geometry-specific variations were correlated with computational model-based predictions. A scalar metric, defined to capture the extent of recirculation for a specific geometry, correlated with oxidized low-density lipoprotein (Ox-LDL) uptake and localized monocyte adhesion to EC. Furthermore, SMC seeded in regions with a larger extent of recirculation increased their tissue factor (TF) expression. These observations point to the importance of accounting for patient-specific geometry variations and flow derangements to extend derived biological inferences beyond idealized cell culture models to real-world settings.

2. Methods

2.1 Arterial replication platform

Geometrical representations and flow wave forms of the left main coronary artery (LM) bifurcating into left anterior descending (LAD) and left circumflex (LCX) (Figure 1A and D), and the common carotid artery (CCA) bifurcating into the external (ECA) and internal (ICA) carotid arteries (Figure 1B and E), were needed as input values for in silico and in vitro experiments. Design of arterial mimics was performed using a modification of a previously developed computational framework24 (Figure 1C). EC (Figure 1F) and SMC (Figure 1G) were seeded on polydimethylsiloxane (PDMS) scaffolds. Scanning electron microscopy (SEM) revealed complete cellular coverage along length and circumference of the constructs (Figure 1H), and histological examination confirmed multilayer structure of SMC-coated constructs (Figure 1K). SMC (Figure 1I) and EC (Figure 1J) retained their cytoskeletal organization and alignment under flow. Vascular geometries and flow parameters, computational platform design, in silico models, and scaffold casting techniques are further detailed in Supplementary material online, Methods.

Figure 1

Computational platform to design personalized vessel-like scaffolds. Real data from patients may be obtained with angiographic images (A) or eco-Doppler measurements (B) and a physician defines the co-ordinates that represent arterial dimensions and that are introduced in a custom-designed Visual Basic® interface (C). The interface encodes four different macrofiles for CATIA® that are used to drive formulation of Teflon® moulds that will be used to manufacture PDMS scaffolds (D and E). A cell-lined tubular structure is created from EC (F) [CD31, yellow] or SMC (G) [SMC α-actin, red]. Scaffolds were fixed, sputter-coated, and SEM imaged (H) or H & E stained (K) to confirm seeding homogeneity, coverage, and morphology. SMC (I) and EC (J) align their tubulin (red) filaments after 24 h exposure to arterial-like flow.

2.2 Recirculation metrics

The tangential component of the velocity vector (νTan), calculated as the dot product of local velocity vector and normalized direction vector representing the vessel centreline, was used to measure the velocity of the fluid flow (see Supplementary material online, Figure S2). This strategy is optimal in curved vessels since it always accounts for local recirculation. Based on νTan, extent of flow recirculation was classified into three categories: severe, mild, or negligible. If maximal backward velocity (νTanMIN) was above 20% of the inlet velocity during the pulse, extent of recirculation was considered severe. If νTanMIN was between 5 and 20%, extent was considered mild, and was negligible if the former criteria were not met. νTan was evaluated at each node and all tetrahedrons containing at least one node with negative νTan were summed to estimate the ‘total recirculating volume’ (VolRec) for each branch (Eq. 1). The parameter ‘normalized νTan·VolRec’ (NVV) was calculated multiplying VolRec by νTan, dividing it by the volume of the branch, and normalizing it to the value of the straight segment of the bifurcation (Eq. 2). NVV was the parameter found to best correlate disease biomarkers and flow disruptions (see Supplementary material online, Figure S3). $$mathtex$$\hbox{VolRec} = \sum {({\rm tetrahedron}{s_{\nu{\rm Tan \lt 0}}})}$$mathtex$$ (1) $$mathtex$$\hbox{NVV} = \displaystyle{{\nu \hbox{Tan} \cdot \hbox{VolRec}/\hbox{Vol Branch}} \over {(\nu\hbox{Tan} \cdot \hbox{VolRec}/\hbox{Vol Branch) Straight Branch}}}$$mathtex$$ (2)

2.3 Biological assessment of cell-seeded scaffolds

All biological results shown are extracted from the regions of interest. In straight-tube experiments, biomarkers of interest were analysed in the central region of the tube. In bifurcated experiments, data were extracted from the regions of recirculation indicated by CFD simulations (ECA, ICA, LAD, and LCX) and the centre of the straight upstream segments (CCA and LM). All results were normalized to their respective arterial flow (AF) values in single-channel experiments and to the straight sections in the bifurcations.

Vascular cell culture, seeding of scaffolds, perfusion bioreactor, Ox-LDL uptake, monocyte adhesion, vascular cell adhesion molecule 1 (VCAM-1), TF, phosphorylated endothelial nitric oxide synthase (p-eNOS) and Krüppel-like factor 2 (KLF-2) expression, KLF-2 mRNA expression, thrombogenicity, radioimmunoprecipitation assay extraction, western blot analysis, microscopic examination, ex vivo detection of macrophage presence in murine carotid bifurcations, and antibodies used are further detailed in Supplementary material online, Methods. All imaging analysis was done using the FIJI imaging platform.25

2.4 Human data and samples

All human data (angiographic images) were deidentified, and biological samples (EC, SMC, and blood) were from anonymous donors. Samples and data were treated following the guidelines of the Declaration of Helsinki.

2.5 Statistics

All in vitro experiments described were performed on triplicate specimens and repeated two separate times. In graphical presentations, data are expressed as average ± standard error of mean. Non-parametric Kruskal–Wallis test, followed by a Scheffé post hoc analysis of the original measured values normalized to their corresponding controls, was conducted to determine statistical differences between values. Values of P < 0.05 were considered statistically significant.

3. Results

3.1 Impact of flow patterns on markers of atherogenesis, and inflammation in EC, and thrombosis in SMC in straight constructs

In our idealized, straight constructs, we observed a protective effect on EC health under conditions of pulsatile unidirectional arterial flow in contrast to bidirectional oscillatory flow (OF) and static (STA) conditions. p-eNOS and KLF-2 expression were highest under AF, and were reduced significantly under OF or STA. KLF-2 mRNA expression was also muted in the oscillatory and static cases (data not shown) when compared with the arterial flow case.

VCAM-1 expression (Figure 2A), Ox-LDL uptake (Figure 2C), and cytokine-induced monocyte adhesion (Figure 2D) were highest under oscillatory or static conditions, and were reduced significantly under arterial flow. Activation by tumour necrosis factor alpha (TNF-α) amplified the signal in all cases. The normalized difference in expression of VCAM-1 was muted in activated EC (Figure 2B), but Ox-LDL uptake remained lower in the presence of arterial flow even in the face of cytokine stimulation. As expected, monocyte adhesion was below detection limits without TNF-α activation.

Figure 2

Oscillatory flow jeopardizes vascular health triggering inflammatory response in EC and thrombogenic response in SMC. EC exposed for 24 h to low shear oscillatory flow reduce their p-eNOS (A) and KLF-2 (B) expression when compared with EC exposed to coronary arterial flow. EC increase VCAM-1 expression when quiescent (C), but lose their sensitivity when exposed to inflammatory conditions (D). Ox-LDL uptake by EC (E) doubled in cells exposed to OF vs. those exposed to AF. Monocyte adhesion to EC (F) doubled in cells exposed to OF vs. those exposed to AF under inflammatory conditions. SMC exposed for 24 h to OF up-regulate TF expression by 2.5-fold (G) vs. those exposed to AF (IK). The previously described26 thrombogenicity apparatus was used to test SMC after 24 h exposure to AF, OF, or STA (H). Thrombogenicity (LN) was evaluated as haemoglobin concentration in the cell lysate and increased up to three-fold in SMC exposed to OF, correlating with TF expression. n = 6 for each marker described.

Similarly, arterial flow was protective of SMC lacking endothelial shielding. Basal TF levels rose 2.5-fold (Figure 2G and IK) under oscillatory flow; in correlation, thrombus formation increased in our model to a similar extent (Figure 2H and LN). TF expression by SMC was not affected by TNF-α activation. Thus, not only can disrupted flow exacerbate arterial injury but also the integrity of intact arterial flow may play an important protective role as well.

3.2 Predicting flow patterns in complex geometries

CFD calculations predicted recirculation downstream of bifurcation branches. Recirculation was observed in the two derivative branches of the carotid artery. Severe recirculation (νTanMIN = −195 mm/s) was observed in the ICA and mild recirculation (νTanMIN = −85 mm/s) in the ECA (Figure 3A). The extent of recirculation was four-fold higher in the ICA vs. the CCA (Figure 3C), while ECA's increase was around two-fold. An idealized model of the bifurcation of the LM was also studied. The two derivative branches, the LAD and the left circumflex (LCX) branches, presented recirculation that was mild, both in terms of flow reversal (Figure 3B) and extent of recirculation (Figure 3D). Simulation results were validated by tracking circulating injected latex microparticles.10

Figure 3

Computational fluid dynamics. Flow recirculations are geometry-dependent. CFD simulations were performed to calculate blood flow profiles in arterial bifurcations. (A) The bifurcation of the CCA to the ICA with a minor aneurysm and the ECA. An idealized LM bifurcation into the LAD and LCX (B) was also simulated. Results in (A and B) show negative tangential velocity profiles (νTan) in mm/s at 0.06 s after the pulse. Recirculation areas were observed after a pulse in both branches of each studied bifurcation, with a minimal speed of −195 mm/s for the carotid bifurcation and −65 mm/s for the idealized coronary bifurcation. NVV (C and D) is measured to establish the correlation between recirculation and disease biomarkers. Recirculation is four-fold (ICA) and two-fold (ECA) more intense than in CCA in the carotid bifurcation (C) and about two-fold in the LAD and LCX vs. LM in the coronary bifurcation (D).

3.3 Expression of atherogenesis and thrombosis biomarkers correlates with the extent of flow recirculation in arterial bifurcations

Arterial flow appeared to protect EC from inflammatory factors in straight tubes seeded with vascular cells while oscillatory flow (as an idealized model of flow disruption) increased expression of cardiovascular disease biomarkers (Figure 2). The same array of tests was run in cell-seeded, 1 : 1 scale bifurcated models (Figures 4 and 5).

Figure 4

Endothelial inflammatory response and SMC thrombogenic response to geometry-induced flow disruptions in the coronary bifurcation. EC and SMC exposed for 24 h to pulsatile coronary-like flow exhibit different levels of p-eNOS (A), KLF-2 (B), Ox-LDL uptake (D), monocyte adhesion (E), and TF (F) as a function of the extent of recirculation in each branch of the coronary bifurcation. p-eNOS and KLF-2 expressions showed significant 40 and 50% respective decreases in regions of recirculation. Ox-LDL uptake by EC had a significant 50% increase for LAD and LCX vs. LM. Similarly, monocyte adhesion in LAD and LCX increased ∼40% vs. LM. TF expression in SMC increased in a range between 35 and 50% for LAD and LCX vs. LM. VCAM-1 expression (C) was not flow-sensitive. n = 6 for each marker described.

Figure 5

Endothelial inflammatory response and SMC thrombogenic response to geometry-induced flow disruptions in the carotid bifurcation. EC and SMC exposed for 24 h to pulsatile carotid-like flow exhibit different levels of p-eNOS (A), KLF-2 (B), Ox-LDL uptake (D), monocyte adhesion (E), and TF (F) as a function of the extent of recirculation in each branch of the carotid bifurcation. p-eNOS levels significantly decreased up to 50% and KLF-2 expression was further reduced over 60% in the ICA. Ox-LDL uptake by EC was three-fold higher in the ICA than in the CCA and about two-fold in the ECA. Monocyte adhesion to EC rose two-fold in cells present in the ICA and 1.5-fold in the ECA when compared with the CCA. TD expression in SMC increased about 2.2-fold in the ICA, but a very discrete 1.3-fold in the ECA when compared with CCA. VCAM-1 expression (C) was not significantly flow-sensitive. n = 6 for each marker described.

Phosphorylated eNOS (Figures 4A and 5A) and KLF-2 (Figures 4B and 5B) expression by EC were inversely correlated with NVV, with a 50% reduction for p-eNOS and up to 70% decrease for KLF-2. Contrarily, Ox-LDL uptake by EC (Figures 4D, 5D, see Supplementary material online, Figure S4AC, and Table 1), monocyte adhesion to activated EC (Figures 4E, 5E, see Supplementary material online, Figure S4DF, and Table 1), and TF expression in SMC (Figures 4F, 5F, see Supplementary material online, Figure S4GI, and Table 1) correlated directly, and most powerfully, with the extent of recirculation in each branch of the studied bifurcations.

View this table:
Table 1

Comparative biomarkers expression as a function of the extent of recirculation

Recirculation level/biomarkerNegligible (AF, LM, and CCA)Mild (LAD, LCX, and ECA)Severe (ICA)Oscillatory flow
p-eNOS↓↓
KLF-2↓↓
Ox-LDL=↑↑↑↑
Monocyte=↑↑
TF=↑↑↑↑
VCAM-1

Ox-LDL uptake by EC increased with the intensity of recirculation, up to three-fold higher than upstream of the bifurcation divider (LM and CCA). Monocyte adhesion to activated EC also grew with the intensity of recirculation, doubling the recruitment by EC free of recirculation. VCAM-1 levels in activated EC, which were blunted in the straight-tube constructs, did not correlate with the observed flow regimes in our bifurcating models (Figures 4C, 5C, and Table 1) and were not detectable in the short sections analysed in the absence of TNF-α activation.

In SMC, TF levels were also sensitive to flow recirculation. SMC seeded in areas with severe recirculation showed a 2.5-fold increase in TF expression when compared with those seeded in regions exposed to negligible recirculation levels. The presence of macrophage was tested in carotid bifurcations of mice fed with a high-fat diet (see Supplementary material online, Figure S5). The extracted carotid geometry was simulated, and regions of flow stagnation near the vessel walls correlated fairly with macrophage recruitment.

4. Discussion

4.1 Flow patterns dictate EC and SMC reactivity

Vascular disease biomarkers and flow patterns are strongly correlated.5,2729 In 1983, Zarins et al.27 observed that plaque formed predominantly in regions of flow separation and hypothesized that distinct wall shear stress profiles were responsible for such plaque localization. Dai et al.29 discovered how waveforms could be atheroprotective or atheroprone for EC, and could coexist simultaneously in different regions of the carotid bifurcation. In particular, cDNA array profiling proved that the two waveforms differentially stimulated genes involved in signal transduction, transcriptional regulation, inflammation, angiogenesis, coagulation, and lipid metabolism. They observed how atheroprotective waveforms enhanced KLF-2 and subsequently up-regulated eNOS, while atheroprone waveforms promoted up-regulation of interleukin-8, chemokine receptor type 4, and pentraxin 3 expression, which are biomarkers of atherosclerosis. Disturbed, atheroprone flow regimes usually comprise phases of reversal flow that generate oscillatory wall shear stresses, an attribute notably absent in atheroprotective flow regions.30 Oscillatory flow has been shown to up-regulate easily detectable cell adhesion molecules such as VCAM-1 or ICAM-1.31 We chose VCAM-1 as one of the atherogenesis biomarkers instead of ICAM-1, since VCAM-1 seems to be more critical at the beginning of lesion formation.32 We also chose Ox-LDL uptake and monocyte adhesion, which are fundamental players in atherosclerosis progression.33 Ox-LDL binds to scavenger receptors CD3634 and LOX-1,35 and ligation of LOX-1 induces NF-κB in EC.36 Transcriptional activation of genes encoding for LDL receptor, with the consequent enhancement of LDL uptake, are increased in EC in disturbed flow regions.37 Oscillatory flow increases native38 and Ox-LDL mediated monocyte chemotactic protein-1 expression in EC and subsequent monocyte adhesion.39 In the same study, high shear stress prevented both Ox-LDL uptake and monocyte adhesion. Depending on the degree of oxidation, LDL induces expression of P-selectin in human umbilical vein endothelial cells, which increases adhesion of monocytes.6

The literature is equivocal regarding the effects of haemodynamic forces on the coagulation cascade. While certain groups have reported an increase of TF expression,40 gene transcription and increased Factor Xa concentration41 in EC exposed to high shear stress, others affirmed that high shear stress provokes release of the TF pathway inhibitor from EC,42 reducing TF activity, and hence risk of thrombosis. Other authors reported that TF activity was up-regulated by oscillatory shear stress,43 and some noted that TF is of only modest importance in EC.44 While the effects of haemodynamic forces on TF in EC have been deeply explored, the influence of flow on SMC function is poorly characterized.45 Indeed, the endothelium is denuded after angioplasty46 and SMC are regularly exposed to blood flow. We therefore exposed single cultures of SMC to pulsatile flow and oscillatory flow along straight tubes and bifurcations, and analysed TF levels to elucidate risk of thrombosis as a function of flow. Although the increase of thrombogenicity may not only be caused by the increase of TF levels, in our system, TF expression in SMC correlated with ex vivo clot formation.

4.2 The extent of recirculation simplifies the relationship between geometry and flow and correlates with early markers of EC and SMC diseases

Geometrical and dimensional parameters are often used to define the state of human vasculature24,27 and link to vascular disease.47,48 Yet, the number of parameters required to define complex geometries prohibits rigorous characterization and produces a vast array of dimensions and possible interactions. We propose an alternative strategy that integrates dimensions and physiology into a single flow parameter based on the extent of recirculation in the vessel under study. Instead of limiting the evaluation of risk to a specific number of factors at specific locations in the vessel, we scanned the complete architecture of the vessel tree and determine the risk using a single variable that reflects how geometry affects recirculation. Our data suggest that the distinct geometry of each patient's vascular tree dictates different flow regimes, and that these regimes are the critical players in defining each patient's vascular phenotype. The coronary models, however, did not include the effect of forces caused by the myocardium during systole and diastole. Such forces further increase the oscillatory shear index in the LAD, but do not alter the localization of the recirculation sites.49

Indeed, Ox-LDL uptake by EC and monocyte adhesion to activated EC correlated directly with the extent of recirculation in each branch of the studied bifurcations. Furthermore, p-eNOS and KLF-2 expressions were higher in regions of lowest recirculation. The implication of these findings is that endothelial cells function differently when exposed to flows with different levels of recirculation.

The full sequence of atherogenesis was activated in regions of flow recirculation and stagnation. Recirculation was also a powerful predictor of SMC biology where TF expression tracked with the amount of recirculation precisely. In this context, the two elements of the paper come together nicely to explain the power of flow: arterial flow is protective of vascular injury and inductive of repair but subject to disruption by recirculation. Atherosclerosis and thrombosis are, however, highly multifactorial events. The inter-patient variability due to disease progression or after any of a range of vascular interventions can also originate from procedure trauma and release of anti-proliferative drugs, both responsible of potential thrombosis. Micro-recirculations caused at strut level by stent thickness and malapposition also increase the risk of thrombosis.26 In any case, restoration of flow after a stent deployment should reduce the risk of thrombus linked to blood recirculation.

In vivo correlations between flow recirculation and markers of vascular damage have been established in the past. In 2004, Richter et al.21 proved that regions of recirculation in a porcine model of the ilio-femoral bifurcation correlated with leucocyte recruitment and intimal hyperplasia levels. Gimbrone50 reported analogous findings in murine and rabbit aortic arches in 2010. Very recently, Assemat et al.51 found comparable results correlating in vivo atheroma with regions of bi-directional low shear stress in an aortic murine branch calculated by CFD. In this work, we tested the presence of macrophage in carotid bifurcations in mice fed with a high-fat diet. Regions of flow stagnation near the vessel walls were able to predict regions where macrophage recruitment was more dynamic.

4.3 The arterial replication platform is a multidisciplinary tool to study the local impact of flow disruptions

Ever since Gimbrone et al. designed the first in vitro flow chamber, researchers have tried to find accurate models of arterial flow disruptions. Though precise control over flow and geometry makes these models excellent tools for basic research, these systems do not generally simultaneously expose cells in a single construct to different flow regimes. Single-channel ex vivo and in vitro models are limited to one flow regime per system and lack the spatial resolution of different regions of a bifurcated artery.

Our system now allows exact replication of vascular architecture,10 simultaneous exposure of vascular cells to a physiological range of flow regimes within the same construct and quantitative measure of cell- and flow-specific responses. We can examine local recirculation and compare response of genetically identical cells,2 exposed to the same solutes and <1 cm apart. Cells that might be thought to be identical can now be shown to express significantly different levels of disease biomarkers under the control only of different degrees of recirculation. This spatial control over stimulus is critical in examining the cellular response to local flow alterations. While all inflammation and thrombogenic biomarkers generally rose when cells were exposed to oscillatory flow and static conditions, these values showed a clear correlation with the local extent of recirculation for each branch on each arterial bed. The three-dimensional approach localizes better the impact of flow patterns, but poses novel challenges at the same time. The studied biomarkers presented levels near the detection thresholds of the analytical methods applied, especially in small sections like recirculation regions of the bifurcated constructs.

Integrated frameworks such as the one we described here could be useful in basic research and clinical practice. As an in vitro tool, the platform allows for parametrical and mechanistic studies of the impact of patient-specific features in biological outcomes, not only for inflammation or coagulation but also for angiogenesis or cellular migration among many. Our platform has already been proved useful to determine the thrombogenicity and impact of different anti-proliferative treatments on endothelial recovery. The findings might well suggest a different focus in performing procedures—one that focuses on restoration of intact fluid flow rather than where stent struts are located. Increasingly flow is emerging as a quantifiable interventional parameter, and our results will place flow within a greater context. One could even envision extending this work to other flow domains such as the respiratory or urinary tracks.20 In the future, a parameter such as NVV may be provided to directly guide intervention based on individual patient state and vascular geometry of specific patients to prevent flow- and drug-related post-implantation complications.

4.4 Conclusion

Vascular cells attain a range of phenotypes that are most responsive to their flow environment. Endothelial health is preserved by arterial flow, but jeopardized in regions of flow recirculation in a quasi-linear manner. Similarly, smooth muscle cells exposed to flow are more thrombogenic in large recirculating regions. Thrombotic and inflammatory biomarkers correlate with the extent of recirculation that vascular cells are exposed to when lining certain vascular geometries. The integration of engineering of fluidics with vascular biology is the only means to examine these complex interactions.

Funding

This work was supported by the National Institute of Health/National Institute of General Medical Science (RO1/GM049039 to E.R.E.); Ministerio de Ciencia e Innovación (BFU 2009-09804 to M.B., J.J.M., and E.R.E.); MIT International Science and Technology Initiatives (MIT-Spain Seed Fund 2010 to E.R.E. and J.J.M.); Generalitat de Catalunya (2013FI_B2 00093 to J.M. and MOBINT 2011 to P.S.), Banco Santander (Fórmula Santander 2011 to P.S.) and POSIMAT (PS2010-2013 to M.B.).

Acknowledgements

The authors thank the following people and institutions for their support: Philip Seifert for scanning electronic microscopy and histopathology at CBSET, Nicki Watson and Wendy Salmon for spinning confocal disk microscopy at Whitehead Institute, and Dr Jeffrey Wyckoff for multiphoton imaging at Koch Institute. We also acknowledge Dr Caroline C. O'Brien, Dr Vipul Chitalia, Dr Andrés A. García-Granada, Jay Wang, Elisabet Rosàs, and Fernando García-Polite for their technical support and intellectual discussion.

Conflict of interest: none declared.

References

View Abstract