Introduction

Coronary heart disease is projected to remain the worldwide leading cause of death until 2030.¹ Coronary heart disease is a major cause of morbidity and reduced quality of life with enormous economic consequences.^2,3 Atherosclerosis, a multifocal lipid-driven inflammatory process, is the principal underlying pathology in patients with coronary heart disease, which commonly presents clinically with symptoms secondary to luminal narrowing of an epicardial coronary artery or an acute coronary syndrome. The latter is a major cause of coronary heart disease death and most commonly results from rupture at the site of a thin cap fibroatheroma (TCFA) leading to coronary thrombosis.⁴

Clinical Perspective on p 1012

The precise environmental cues that lead plaques toward an advanced and high-risk phenotype are not yet fully elucidated, but disturbed blood flow is thought to play a central role in both lesion initiation and progression.⁵ Disturbed flow is most frequently quantified by metrics of shear stress, which is the frictional force imposed by blood flowing over the endothelial surface, and association between these metrics and coronary atherosclerotic lesion stage have been demonstrated in vivo in both animal models^6–9 and patients.^10,11 However, few studies have investigated the impact of prevalent shear conditions on the longitudinal development of advanced coronary plaques, wherein they report discrepant results. Previous experimental studies^6–8,12 have reported TCFA development in arterial segments with persistently low time-averaged wall shear stress (WSS) in diabetic hypercholesterolemic pigs. By contrast, clinical studies¹¹ suggested that regions of high WSS were associated with the development of features of plaque vulnerability. This work also showed an association between low WSS and regional plaque progression, consistent with the findings of the Prediction of Progression of Coronary Artery Disease and Clinical Outcome Using Vascular Profiling of Shear Stress and Wall Morphology (PREDICTION) trial.¹⁰ Although informative, these important studies do not confirm a causal role for metrics of disturbed WSS in inducing the formation and progression of advanced coronary plaques.

In the current study, we hypothesized that: (1) inducing disturbed blood flow initiates the development of advanced coronary plaque types, including TCFA, and (2) different longitudinal patterns of low, multidirectional, and high WSS metrics determine plaque morphology. To test these hypotheses, we developed a stenotic shear modifying stent (SMS)¹³ that could be placed percutaneously to induce regions of persistent blood flow perturbations within the coronary arteries of D374Y-PCSK9 hypercholesterolemic transgenic minipigs.¹⁴ Our results provide evidence for a causal role of disturbed flow in the development of advanced coronary plaques and extend our previous work demonstrating that inducing disturbed blood flow in a carotid artery of hypercholesterolemic mice^15,16 causes the development of TCFA-like plaques.

Methods

Detailed methods are available in the online-only Data Supplement, and the study protocol is summarized in Figure I in the online-only Data Supplement. All animal experiments were performed in accordance with the ethical and welfare regulations of the University of Aarhus and approved by the Danish Animal Experiments Inspectorate. In brief, SMSs were implanted into either the left anterior descending or left circumflex arteries of hypercholesterolemic D374Y-PCSK9 transgenic minipigs (n=6), with the unstented artery serving as a control.¹⁴ Serial intracoronary frequency-domain optical coherence tomography (FD-OCT) and Doppler flow velocity measurements were obtained at 0 (baseline), 0+ (immediately poststent), 19 weeks, and 34 weeks. From these data, the 3-dimensional (3-D) geometry of each artery was reconstructed, and computational fluid dynamics (CFD) was performed to compute established and custom WSS metrics of disturbed flow. To correlate computed WSS metrics to plaque morphology, vessel segments were excised proximal and distal to the SMS in the instrumented artery and in a comparable region of the uninstrumented control artery at the final study time point. The segments were serially sectioned, and alternating sections were stained with hematoxylin and eosin, picrosirius red (collagen), and muramidase (pan macrophage). Stained histological sections from each pig were imaged, contoured manually, and accurately coregistered to the corresponding segments of the in vivo 3-D reconstructed arteries of that pig over all time points, to evaluate on a local basis (3.6° circumferentially and ≈105 μm axially) the time course of the change in magnitude of each WSS metric within each plaque type. Plaques were independently classified by expert histopathologists, who were blinded to CFD results, as TCFA, fibrous cap atheroma (FCA), pathological intimal thickening (PIT), xanthoma (XA), intimal thickening (IT), or normal vessel wall (NOR). A color-coding system was used to segment plaque type continuously around the circumference of each hematoxylin and eosin–stained section (Figure II in the online-only Data Supplement). The primary readout was the mean of each WSS metric within each of the plaque types identified in each histological section, averaged over all sections containing that plaque type across all pigs. The accuracy of this workflow was confirmed in a formal error analysis (see Methods in the online-only Data Supplement).

Statistical Analysis

Variables (stenosis area, blood velocity, and total cholesterol) are reported as mean±standard deviation. Some variables (WSS metrics of disturbed flow and metrics of plaque size) were log-transformed and reported as geometric mean (GM) with 95% confidence interval (CI). All statistical analyses were performed by using a linear mixed-effects model to compare continuous dependent variables and various independent categorical variables, including: plaque type or group (fixed effect), time (fixed effect), and subject (random effect attributable to repeated measures of multiple histological sections within each pig). Comparisons between normally distributed variables are expressed as an arithmetic mean difference with 95% CI, and log-transformed distributed variables are expressed as a geometric mean ratio (GMR) with 95% CI. Both outputs come directly from the mixed model with an associated P value. All P values generated from the linear mixed-effect model and reported in the article have been systematically Bonferroni corrected by multiplying each P value by the appropriate number of comparisons. Statistical significance was based on these corrected P values being <0.05. All statistical analyses were performed in Stata version 13 (StataCorp, College Station, TX).

Results

One pig died of acute stent thrombosis immediately after SMS implantation. Of the remaining 5 pigs, 4 animals completed the entire study protocol, with 1 animal dying after developing stent thrombosis following the completion of intracoronary imaging at the 19-week time point. A total of 5957 FD-OCT frames were contoured manually to allow reconstruction and CFD modeling for the calculation of WSS in 38 vessels composed of both instrumented and control arteries over all time points in 5 pigs. Histology sections were coregistered to the corresponding 3-D reconstructed arteries. Local plaque type was manually segmented in hematoxylin and eosin–stained sections (n=538). Regional overlaps between these histological measures and WSS metrics of disturbed flow were quantified.

SMS Induced Atherogenesis and Advanced Plaque Development

We observed accelerated atherosclerosis resulting from SMS placement. In comparison with control regions, the segment downstream of the SMS had moderately increased external elastic lamina area (GM, 2.1; 95% CI, 2.0–2.2 mm² versus GM, 1.4; 95% CI, 1.3–1.5 mm² and GMR, 1.6; 95% CI, 1.4–1.8; P<0.001) and large increases in plaque area (GM, 1.3; 95% CI, 1.1–1.5 mm² versus GM, 0.2; 95% CI, 0.2–0.3 mm² and GMR, 6.0; 95% CI, 4.7–8.0; P<0.001). The segment upstream of the SMS had similar increases in external elastic lamina area (GM, 2.1; 95% CI, 1.9–2.3 mm² versus GM, 1.4; 95% CI, 1.3–1.5 mm² and GMR, 1.5; 95% CI, 1.3–1.7; P<0.001) but plaque area only increased modestly (GM, 0.3; 95% CI, 0.3–0.4 mm² versus GM, 0.2; 95% CI, 0.2–0.3 mm² and GMR, 1.5; 95% CI, 1.1–2.0; P=0.012), which was significantly less than in the downstream segment (GMR, 4.0; 95% CI, 3.3–5.0 in downstream versus upstream plaque area; P<0.001). Strikingly, plaque burden was ≈3-fold higher in the segment downstream of the SMS than in both the SMS upstream (GMR, 3.4; 95% CI, 2.9–4.1; P<0.001) and control artery segments (GMR, 3.5; 95% CI, 2.8–4.4; P<0.001; Figure 1A).

• Download figure
• Download PowerPoint

Sections from the segment downstream of the SMS exhibited plaques at varying stages of advancement (Figure 1B), including NOR (34%), IT (20%), XA (17%), PIT (22%), FCA (4%), and TCFA (3%). FCA and TCFA lesions contained a large necrotic core, macrophage infiltration, and a fibrous cap (≤65 μm for TCFA) with newly deposited collagen, similar to human lesions (Figure 1B). The following plaque types developed in the region upstream of the SMS: NOR (36%), IT (29%), XA (10%), PIT (20%), and FCA (5%). No TCFA was observed. Sections from segments of the uninstrumented artery exhibited NOR (41%), IT (30%), XA (5%), and PIT (24%), but no FCA or TCFA was observed (Figure III in the online-only Data Supplement).

SMS Implantation Caused Persistent WSS Perturbations

The arithmetic mean initial maximum areal stenosis created by the SMS was 62.6±7.0% (n=5). There was a progressive increase in stenosis severity to 83.9±3.1% at 34 weeks (arithmetic mean difference, 20.9; 95% CI, 14.5–27.4% at 34 weeks versus immediately post-SMS; P<0.001) because of neointimal hyperplasia within the SMS (Figure 2A and 2B). The stenosis resulted in a reduction of mean blood velocity at the inlet of the instrumented vessel from 147.8±41.0 at baseline to 90.7±19.2 mm/s at 34 weeks (arithmetic mean difference, –57.1; 95% CI, –93.7 to –20.3 mm/s at 34 weeks versus baseline; P=0.018). The change in blood velocity in the control vessel over time was not statistically significant (P=0.55; Figure 2C). The reduced blood velocity caused a concomitant reduction in WSS in the instrumented vessel (Figure 2D). Notably, in the region downstream of the SMS, the accelerated blood flow exiting the SMS was directed toward the outer curvature of the vessel in a concentrated stream, causing regions of markedly increased flow localized to the outer curvature. This high kinetic energy in the concentrated stream induced flow separation and vortex formation, causing large regions with low and multidirectional flow on the inner curvature (Figure 2E). WSS metrics with the largest positive change from baseline values over time were the low shear index (LSI), indicating regions with lowered WSS; high shear index (HSI), indicating regions with increased WSS; and a variant of the transverse wall shear stress (tSS; Figure 2D), indicating multidirectional flow. We therefore focused on these metrics for the remainder of the study.

• Download figure
• Download PowerPoint

Advanced Plaques Were Initiated by Low and Multidirectional WSS

After observing that placement of the SMS increased LSI, HSI, and tSS, we next evaluated whether any of these metrics of disturbed flow preferentially colocalized to regions of particular plaque types immediately after SMS implantation (t=0+) to explore the initiating stimulus for those plaques. Plaques were initially grouped as advanced lesions (TCFA, FCA, and PIT), early lesions (XA and IT), or normal vessel wall (NOR) in the segments upstream and downstream of the SMS, and the uninstrumented control artery of all pigs, as well. In all histological sections, we quantified the geometric mean of each WSS metric within each plaque group at t=0+ (Figure 3). LSI within advanced lesions located in the downstream segment (n=125 sections) was significantly higher than those in the upstream (GMR, 29.3; 95% CI, 20.7–41.6; n=86; P<0.002) and control segments (GMR, 17.9; 95% CI, 10.9–29.3; n=33; P<0.002). LSI, which is a measure that increases in value as WSS reduces, was also significantly greater in advanced than in early lesions (GMR, 5.5; 95% CI, 4.1–7.3; P<0.002) and normal vessel wall (GMR, 6.9; 95% CI, 5.2–9.1; P<0.002) in the instrumented downstream segment (Figure 3A). By comparison, HSI, which is a measure that increases in value as WSS increases, exhibited a reverse pattern in the downstream segment with reduced values in advanced lesions (except FCA at 19 weeks, see below) that were significantly lower in comparison with early lesions (GMR, 0.62; 95% CI, 0.51–0.76; P<0.002) and normal vessel (GMR, 0.46; 95% CI, 0.38–0.56; P<0.002; Figure 3B). Similar to LSI in the downstream segment, tSS had larger values in advanced lesions than in both early lesions (GMR, 1.7; 95% CI, 1.5–1.9; P<0.002) and normal vessel (GMR, 1.8; 95% CI, 1.6–2.0; P<0.002; Figure 3C). Overall, the striking colocalization of advanced lesions to regions of lowered WSS (LSI) and multidirectional WSS (tSS) within the downstream segment, where plaque burden was ≈3-fold higher in comparison with other regions, suggests that both patterns of disturbed flow contributed to their initiation.

• Download figure
• Download PowerPoint

Time Course and Magnitude of Disturbed WSS Determined Final Plaque Type

Next, we determined whether each plaque type experienced a different time course of disturbed flow in the segment downstream of the SMS (Figure 4). TCFA, FCA, and PIT were observed in 2 of 5, 3 of 5, and 5 of 5 animals, respectively. Over all postinstrumentation times, TCFA regions demonstrated higher LSI values (Figure 5A) in comparison with FCA (GMR, 1.8; 95% CI, 1.3–2.5; P<0.005) and PIT (GMR, 1.4; 95% CI, 1.1–1.9; P=0.25), although only the comparison with FCA was statistically significant. LSI in TCFA was also significantly higher in comparison with XA (GMR, 1.94; 95% CI, 1.47–2.56; P<0.005), IT (GMR, 2.75; 95% CI, 2.10–3.58; P<0.005), and normal vessel wall (GMR, 2.66; 95% CI, 2.05–3.46; P<0.005). In addition, LSI values were of greater magnitude in PIT in comparison with values in XA (GMR, 1.4; 95% CI, 1.2–1.6; P<0.005), IT (GMR, 1.9; 95% CI, 1.7–2.2; P<0.005), and normal vessel wall (GMR, 1.9; 95% CI, 1.7–2.1; P<0.005). LSI was also higher in FCA than in IT (GMR, 1.5; 95% CI, 1.2–1.9; P<0.005) and normal wall (GMR, 1.5; 95% CI, 1.2–1.9; P<0.005), but not XA (GMR, 1.1; 95% CI, 0.9–1.4; P=0.90). Interestingly, FCA was also the only advanced plaque with increased HSI (Figure 5B), wherein, at 19 weeks, FCA was increased in comparison with PIT (GMR, 3.2; 95% CI, 1.8– 5.6; P<0.005) and TCFA (GMR, 2.5; 95% CI, 0.9–5.6; P=0.79), although the latter comparison was not significant. Finally, analysis of tSS (Figure 5C) over all postinstrumentation times was not statistically higher between any of the advanced plaque types, and only TCFA and PIT showed any statistically higher values of tSS in comparison with IT (GMR, 1.4; 95% CI, 1.2–1.7 and GMR, 1.3; 95% CI, 1.1–1.4, respectively; P<0.005) and normal vessel wall (GMR, 1.4; 95% CI, 1.2–1.7 and GMR, 1.3; 95% CI, 1.2–1.4, respectively; P<0.005; Figure 5C).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Discussion

Previous studies have reported the association between naturally occurring disturbed blood flow and the development of atherosclerotic lesions.^6–11 These investigations have supported the hypothesis that persistent, low WSS may initiate and advance the development of high-risk plaque features, but do not prove its causal role. In the current study, we imposed a persistent perturbation of local WSS by implantation of a SMS. The main findings of our study were, first, that SMS implantation induced a heterogeneous pattern of flow disturbance that caused accelerated atherogenesis and formation of advanced plaque types in the coronary arteries of D374Y-PCSK9 hypercholesterolemic minipigs. Second, we observed that the majority of advanced plaques were located in the segment downstream of the SMS (Figure 1) within regions of initially lowered WSS (quantified by LSI) and also multidirectional WSS (quantified by tSS; Figure 3). In contrast, the upstream SMS and control artery segments exhibited little overlap between WSS metrics of disturbed flow and any plaque type (Figure IV in the online-only Data Supplement). Third, we observed that, in comparison with other advanced plaque types, TCFA developed in those locations downstream of the SMS exposed to the largest and most persistent reduction in WSS over the duration of the study (Figure 5A), consistent with previous observations.⁸ Finally, FCA developed in locations subjected to greater degrees of high WSS (quantified by HSI), particularly at 19 weeks (Figure 5B), which supports the hypothesis that high WSS may promote a stable plaque phenotype. Collectively, these findings demonstrate a causal role for disturbed flow in the formation of advanced plaques, including TCFA, within coronary arteries.

Building on initial studies that suggested a proatherogenic role for low and oscillatory WSS,^17,18 we previously reported that placing a constrictive, external cuff to perturb blood flow in the carotid artery of Apo-E knockout mice led to the formation of TCFA upstream and a stable plaque downstream of this device. This approach provided the first evidence for a causal role of disturbed blood flow in vulnerable and stable plaque formation.¹⁶ However, their small size and different cardiovascular physiology impose limitations on the study of atherosclerosis progression and predilection sites as it relates to humans. Furthermore, Reynolds and Dean numbers scale with size and are significantly lower in rodent models in comparison with values in humans.¹⁹ Recent data also suggest limitations of murine models for studying acute inflammatory disease processes.²⁰ We were therefore motivated to overcome these limitations by studying the coronary arteries of a transgenic hyperlipidemic pig model which, in contrast to murine models, develops human-like advanced plaques.¹⁴ Our results are consistent with previous observations that lowering WSS induces TCFA.^15,16 We now demonstrate that, in addition to colocalization with regions of early atherosclerotic lesion formation,²¹ multidirectional WSS may act synergistically with lowered WSS to initiate the development of advanced coronary plaques including TCFA, which has not been reported previously. In addition, our observations that early exposure to raised WSS, quantified by HSI, significantly overlapped with the formation of early plaque types (Figure 3B), and that exposure to high WSS was associated with the development of FCA (Figure 5B), supports further investigation of the hypothesis that high WSS may be protective by inducing the formation of stable plaque phenotypes.

The marked variation in plaque type observed in both the axial and circumferential directions of each instrumented vessel (Figure 4 and Figure II in the online-only Data Supplement) motivated our novel approach to determine how temporal changes in WSS may influence final plaque morphology on a local, high-resolution scale. This type of analysis is important because plaque rupture is likely determined by variations in the biomechanical properties, and possibly endothelial cell mechanobiology, on this scale.²² This approach was enabled by using FD-OCT rather than intravascular ultrasound for 3-D reconstruction of the vessel lumen and its curvature. We performed a detailed error propagation analysis that confirmed an error of <10% for computed WSS despite the use of a commercial FD-OCT system that could not provide ECG-gated acquisitions. In addition, we performed a detailed error analysis of the 3-D histology approach developed for coregistering WSS metrics from CFD to histological data, which demonstrated a circumferential error of 265 μm and an axial error of 340 μm that represent a significant improvement to previous reports from both others^{6–8,12,23,24} and ourselves.^25–29 Importantly, this 3-D histology error analysis included the assessment of the reproducibility of identification of the same vessel segment of interest from the FD-OCT images through time. Therefore, it accounts for small changes in vessel size that may occur over time owing to remodeling or acquisition of the FD-OCT at different average positions within the cardiac cycle (see Methods in the online-only Data Supplement).

Our data show that development of advanced lesions in the SMS-treated artery directly resulted from induced changes in local blood flow. Two new metrics were developed to quantify those changes relative to the instrumented vessel at baseline (ie, before SMS placement) over a continuous range, namely, the LSI and HSI. These metrics require no a priori assumption of a threshold value of WSS for determining nonnormal WSS magnitudes and are applicable to the study of animals of different sizes and humans. This approach may facilitate the evaluation of changes in WSS over time by overcoming potential issues arising from using a global threshold for defining low and high WSS, as previously reported in the literature.^{6–8,23,30–32} Interestingly, the GM of the absolute values of WSS within regions of LSI was 0.7 (95% CI, 0.6–0.8) Pa (Figure V in the online-only Data Supplement), which is lower than the binary threshold values of 1 to 1.2 Pa reported previously.^{6–8,23,30–32}

The finding that low and multidirectional flow accelerated the initiation and development of all advanced plaque types is consistent with previous experimental data demonstrating that such flow disturbances promote a proatherogenic endothelial cell type, which is characterized by the increased expression of proinflammatory adhesion molecules,^33,34 impaired nitric oxide production,³⁵ increased production of reactive oxygen species, augmented permeability to low-density lipoprotein,³⁶ and enhanced attraction and adhesion of monocytes. Furthermore, studies in hyperlipidemic mice,³⁷ rabbits,³⁸ pigs,¹² and humans¹⁰ consistently demonstrate that atherosclerotic lesions preferentially occur in regions with natural low and multidirectional WSS conditions, including curved vessel segments and bifurcations.³⁹

We studied D374Y-PCSK9 transgenic hypercholesterolemic Yucatan minipigs that have growth curves that suit long-term serial invasive follow-up studies and exhibit chronically elevated cholesterol when fed a cholate-free high-fat, high-cholesterol diet, resulting from a gain-of-function mutation in the proprotein convertase subtilisin/kexin type 9 gene resulting in impaired low-density lipoprotein cholesterol clearance. When fed the high-fat, high-cholesterol diet used in this study, these animals typically achieve a total cholesterol of ≈20 mmol/L (≈773 mg/dL) and low-density lipoprotein cholesterol of ≈11 mmol/L (≈425 mg/dL),¹⁴ which is similar to the cholesterol levels of animals in the current study (Figure VI in the online-only Data Supplement).

Study Limitations

Several limitations of this study should be considered. First, a small cohort of 5 pigs was studied and human-like fibroatheromata developed in only 3 of 5 animals, including TCFA in 2 of 5 animals. Even with this sample size, the high-resolution analysis provided sufficient power to allow identification of statistically significant associations between WSS metrics and final plaque morphology. However, further studies in a larger cohort are needed to provide sufficient sample size to assess the relative magnitude, sensitivity, and specificity of LSI, HSI, and tSS, or their combination, for precise prediction of the development of each advanced plaque type. Second, the temporal evolution of WSS metrics could be related to plaque morphology only at the final time point, when the animals were euthanized for histology. Further work is needed to evaluate the time course of development of each advanced plaque. Extending the duration of follow-up and high-fat, high-cholesterol diet with further histological evaluation of the proximal portion of the coronary arteries is also needed to enable more detailed comparison of the determinants of advanced plaque development in the uninstrumented and SMS-treated arteries. Our results suggest the utility of this model for longitudinal studies of the development of advanced human-like plaques, including TCFA, especially because advanced plaques develop at a predictable location within the coronary artery tree in a practical timeframe. Finally, in the current study, we principally considered the effect of induced local flow disturbance through placement of the SMS. Although this intervention may induce an altered arterial remodeling response in comparison with that induced by naturally occurring flow disturbances, we suggest that this approach enables direct testing of the causal role of different patterns of flow disturbances on advanced coronary atherosclerotic lesion development and thus complements natural history studies.

Conclusions

We have established a novel model of accelerated atherogenesis and formation of advanced human-like atherosclerotic plaques, including TCFA. This model combined with FD-OCT–derived CFD and 3-D histology provides a new means of studying the biomechanics and mechanobiology of human-like advanced coronary plaques. Our results demonstrate a causal role for lowered and multidirectional WSS within the coronary arteries as determinants of final plaque type. Persistently lowered WSS appears to be the principal hemodynamic disturbance needed for the formation of TCFA. Further investigation to specify the relative contributions of persistently decreased, increased, and altered multidirectional shear stress on the development of each of the different advanced coronary atherosclerotic plaque types is warranted.

Acknowledgments

We thank Zahra Nasr, Lisa Maria Røge, and Dorte Qualmann for technical assistance.

Footnotes