A, Picro-Sirius Red collagen staining, original magnification 20×. Representative histological appearance of a fibrous plaque with no lipid core. B, Picro-Sirius Red collagen staining, original magnification 20×. Representative histological appearance of a plaque with a large lipid core (>40% of plaque area). The inset shows a higher magnification of the same plaque, which reveals the cholesterol crystals in the lipid core. C, CD-68 staining, original magnification 100×. Representative histological appearance of a plaque with minor macrophage infiltration. D, CD-68 staining, original magnification 100×. Representative histological appearance of a plaque with substantial (moderate/heavy) macrophage infiltration (brown). CD-68 staining was performed using diaminobenzidine (DAB)
substrate as chromagen and hematoxylin as counterstaining.
Hellings WE, Moll FL, De Vries JPM, Ackerstaff RGA, Seldenrijk KA, Met R, Velema E, Derksen WJM, De Kleijn DPV, Pasterkamp G. Atherosclerotic Plaque Composition and Occurrence of Restenosis After Carotid Endarterectomy. JAMA. 2008;299(5):547-554. doi:10.1001/jama.299.5.547
Author Affiliations: Department of Vascular Surgery (Drs Hellings and Moll) and Experimental Cardiology Laboratory (Drs Met, Derksen, De Kleijn, and Pasterkamp, and Ms Velema),
University Medical Center, Utrecht, the Netherlands; and Departments of Vascular Surgery (Dr De Vries), Clinical Neurophysiology (Dr Ackerstaff),
and Pathology (Dr Seldenrijk), St Antonius Hospital, Nieuwegein, the Netherlands.
Context Previous studies have assessed the predictive value of clinical and angiographic parameters for development of restenosis after vascular interventions. The composition of the atherosclerotic plaque at the intervention site has not been evaluated as a marker for restenosis.
Objective To investigate the relationship between atherosclerotic plaque histology and the occurrence of restenosis after carotid endarterectomy.
Design, Setting, and Patients The Athero-Express study is a longitudinal vascular biobank study that includes the collection of atherosclerotic plaques of patients undergoing primary carotid endarterectomy. Five hundred patients were prospectively followed up between April 1, 2002, and March 14, 2006,
to assess carotid artery restenosis measured by duplex ultrasound 1 year after the intervention.
Main Outcome Measures Risk of carotid restenosis in relation to predefined histological characteristics (macrophage and smooth muscle cell infiltration, collagen,
calcifications, intraplaque bleeding, luminal thrombus, and lipid core size), adjusted for clinical characteristics (multivariate logistic regression analysis).
Results At 1 year, 85 patients (17%) developed 50% or greater restenosis,
including 40 patients (8%) who developed 70% or greater restenosis of the target vessel. Patients whose histological examination of the plaque revealed marked macrophage infiltration (n = 286)
had a lower risk than those with none or minor macrophage infiltration (n = 214) of developing 50% or greater restenosis (risk difference, 11.5% vs 24.3%; adjusted odds ratio [OR], 0.43; 95% confidence interval [CI], 0.26-0.72) and a lower risk of developing 70% or greater restenosis (risk difference, 4.5% vs 12.6%; adjusted OR, 0.36; 95%
CI, 0.17-0.74). Patients (n = 177) with a plaque having a large lipid core size (>40%) had a lower risk than those (n = 94)
with a plaque having a lipid core size of less than 10% of developing 50% or greater restenosis (risk difference, 11.3% vs 25.5%; adjusted OR, 0.40; 95% CI, 0.19-0.81) and a lower risk of developing 70% or greater restenosis (risk difference, 5.6% vs 14.9%; adjusted OR, 0.42;
95% CI, 0.17-1.04), independent of clinical characteristics.
Conclusions Plaque composition is an independent predictor of restenosis after carotid endarterectomy. The dissection of a lipid-rich, inflammatory plaque is associated with reduced risk of restenosis.
Restenosis is a drawback of catheter-based and surgical interventions in different vascular territories. In coronary and peripheral artery disease, diabetes has been associated with increased development of restenosis.1- 4 In addition, following carotid artery interventions, smoking, age,
and female sex have been reported as independent predictors of luminal renarrowing.5- 8 Furthermore, angiographic lesion characteristics, such as lesion length and decreased vessel diameter, are associated with increased risk of restenosis.9,10
In contrast with established risk factors and angiographic parameters,
the relationship between local composition of the vessel wall and development of restenosis has not been explored. Among many other factors, restenosis largely depends on local smooth muscle cell proliferation and migration, protease activity, and matrix deposition occurring at the lesion site after the intervention has been performed. The mechanisms underlying formation of neointima and geometrical vascular remodeling are influenced by the local homeostatic milieu.11,12 Therefore, it is reasonable to assume that factors related to the composition of the atherosclerotic lesion at the intervention site may be associated with restenosis development. Studies on atherosclerotic plaque composition in relation to restenosis are limited by their retrospective nature, because histology of the restenotic lesions was investigated rather than the characteristics of the primary target lesion in relation to future restenosis.13- 16
The lack of prospective evidence relating plaque characteristics with restenosis rate led to the design of the Athero-Express study,
which is a longitudinal vascular biobank study coupled with clinical follow-up and duplex follow-up for restenosis.17 Our primary goal in this study was to investigate the relationship between plaque histology at baseline and restenosis during follow-up among patients who were undergoing primary carotid endarterectomy and longitudinally studied for restenosis development.
The design of the Athero-Express study has been reported previously.17 Briefly, carotid plaques of patients undergoing primary carotid endarterectomy were collected and subjected to histological examination. Patients underwent duplex follow-up to assess patency of the target vessel at 1 year and clinical follow-up at 1 to 3 years after surgery. All patients who underwent primary carotid endarterectomy at 1 of the 2 participating hospitals were asked to participate. The medical ethics boards of both participating hospitals approved the study and all patients provided written informed consent.
The study was initiated on April 1, 2002. Based on a power calculation,
restenosis incidence of 10% vs 20%, 80% power, and α = .05,17 we estimated that a sample size of 500 patients would be sufficient to investigate the relationship between plaque histology and the occurrence of restenosis. The target of 500 one-year duplex follow-ups was reached on April 9, 2007.
All patients undergoing primary carotid endarterectomy between April 1, 2002, and March 14, 2006, at 1 of the 2 participating hospitals were considered for inclusion in the Athero-Express study. The criteria to perform carotid endarterectomy were based on the recommendations by the Asymptomatic Carotid Atherosclerosis Study and Asymptomatic Carotid Surgery Trial studies for asymptomatic patients and the North American Symptomatic Carotid Endarterectomy Trial and European Carotid Surgery Trial studies for symptomatic patients.18- 20 All patients were evaluated by a neurologist before primary carotid endarterectomy to document the cerebrovascular symptom status.21 At baseline, clinical parameters including cardiovascular risk factors and medication use were recorded. Peripheral vascular disease was defined as intermittent claudication as defined by the Edinburgh questionnaire.22 Coronary artery disease was defined as angina pectoris as defined by the Rose questionnaire, or history of myocardial infarction or coronary intervention.23
Before cross-clamping of the carotid artery, 5000 units of heparin was administered intravenously. The choice of closure method of the arteriotomy was at the discretion of the surgeon and could be either primary closure or patch closure with a Dacron or vein patch.
Directly after excision, the atherosclerotic plaque specimen was taken to the laboratory. The plaque was divided in segments of 5-mm thickness along the longitudinal axis. The segment with the greatest plaque burden was subjected to histological examination.17 Macrophage inflitration (CD-68), smooth muscle cell infiltration (alpha-actin), the amount of collagen (Picro-Sirius Red) and calcification (hematoxylin and eosin) were semiquantitatively scored as (1) none or minor or (2) moderate or heavy staining. The criteria for classification were defined as follows: for macrophages:
(1) absent or minor CD-68 staining with negative or few scattered cells or (2) moderate or heavy staining, clusters of cells with more than 10 cells present; for smooth muscle cells: (1) minor alpha-actin staining over the entire circumference with absent staining at parts of the circumference of the arterial wall or (2) positive cells along the circumference of the luminal border, with locally at least minor staining with few scattering cells; and for collagen staining: (1)
none or minor staining along part of the luminal border of the plaque or (2) moderate or heavy staining along the entire luminal border.
Luminal thrombus and intraplaque bleeding were examined in hematoxylin and eosin and Elastin von Gieson stainings and rated as being absent or present.
The size of the lipid core was visually estimated as a percentage of total plaque area using hematoxylin and eosin and Picro-Sirius Red stains, with a division in 3 categories of less than 10%, 10%
to 40%, and more than 40%. The histological examination was performed by 2 independent observers, who were blinded for clinical data. The interrater and intrarater reproducibility was assessed in 100 specimens.
Briefly, 100 specimens were assessed by 2 independent observers and the ratings of both observers were compared with κ statistics.
To assess intraobserver reproducibility, the second observer reassessed the specimens 2 months afterwards with blinding for the previous assessments of the plaques. Both interobserver and intraobserver reproducibility were found to be excellent (κ = 0.6-0.9).24
In protein isolated from adjacent plaque segments, interleukin (IL-8) levels were quantified by Fluorescent Bead Immunoassay (Bendermed,
All patients underwent follow-up with duplex ultrasound (Philips Medical Systems, Eindhoven, the Netherlands) 1 year after carotid endarterectomy. The occurrence of 50% or greater restenosis at the ipsilateral bifurcation was defined as a peak systolic velocity of at least 125 cm/s. Restenosis of 70% or greater was defined as a peak systolic velocity of more than 230 cm/s.25 Reocclusion was defined as the absence of flow with duplex, which was confirmed by another imaging modality (magnetic resonance angiography or angiography). Restenosis was assessed 1 year after the index procedure.26 Duplex measurements were performed by investigators who were blinded for data regarding plaque phenotype and baseline characteristics.
Clinical end points at 1 year included incidence of stroke,
myocardial infarction, and all-cause mortality. Transient ischemic attack was not included as a clinical end point.
We also performed a substudy in which carotid intima-media thickness was assessed in 50 patients who were included in the Athero-Express study between June 3, 2004, and January 1, 2006, and who underwent follow-up in 1 of the 2 centers where the carotid endarterectomy had been performed. The caliber of the vessel and the thickness of the intima-media complex were measured with B-mode ultrasound (7.5 MHz linear array transducer, Philips Medical Systems) preoperatively and 1 year after carotid endarterectomy. The carotid bifurcation was insonated such that optimal images were obtained along the longitudinal axis of the vessel. Still images were captured and calibrated for measurement of carotid intima-media thickness and vessel caliber (AnalySIS version 3.2, Soft Imaging GmbH, Münster, Germany).
The carotid intima-media thickness was measured at the near wall and far wall and averaged.27 The vessel diameter was defined as the distance from the adventitial edge of the intima-media complex of the near wall to the adventitial edge of the intima-media complex at the far wall. Three measurement sites were chosen as a representation of the endarterectomy area: the carotid bifurcation, the proximal internal carotid artery exactly 1 cm distal to the bifurcation, and the distal common carotid artery exactly 1
cm proximal to the bifurcation. The resulting measurements of carotid intima-media thickness and vessel caliber at the 3 sites were then averaged.
SPSS version 15.0 (SPSS Inc, Chicago, Illinois) was used for all statistical analyses. The discrete variable clinical presentation was rearranged as nominal variables to obtain 2 groups (symptomatic [transient ischemic attack and stroke] vs asymptomatic). Cross tables (2 × 2) were constructed to calculate the relative risk (RR) and risk difference with 95% confidence intervals (CIs)
for the development of restenosis (≥50% and ≥70% separately)
for each clinical and plaque characteristic. The accompanying P value was calculated with the χ2 statistic. Continuous variables were additionally compared between patients with and without restenosis during follow-up with the Mann-Whitney U test, to confirm the result of the χ2 test on the dichotomized variables.
The univariate analysis including baseline parameters served as the basis for a multivariate logistic regression model. Two models were constructed: 1 model with 50% or greater restenosis as the dependent variable (model 1) and 1 model with 70% or greater restenosis, which is a subgroup of the group with 50% or greater stenosis, as the dependent variable (model 2). Variables showing association (P < .10) with restenosis (either ≥50% or ≥70%)
in univariate analysis were included in the multivariate analysis.
Age, sex, and clinical presentation (symptomatic vs asymptomatic)
were included and retained in all multivariate models. The other variables were retained in the multivariate model based on the likelihood ratio (threshold: P = .10), with stepwise removal of nonsignificant variables. The adjusted odds ratios with 95% CIs are given for all variables in the final models. The models were characterized by the area under the receiver operating characteristic curve, with 95% CIs (0.5 = no predictive value; 1.0 = perfect prediction).28 The substudy patient group (n = 50) was divided in 2 groups based on the absence or presence of histological features (macrophages, lipid). Between these groups, differences in vessel diameter and carotid intima-media thickness with 95% CIs were calculated.
Among 685 patients considered for inclusion in the study, 81
were excluded due to malignant disease, permanent extramural care,
residence outside the Netherlands, and follow-up conducted at another hospital. An additional 39 patients did not undergo follow-up because of all-cause mortality, cerebral bleeding, or stroke within 1 year.
Sixty-five patients were considered lost to follow-up (9.4%). The baseline characteristics of the study population (n = 500)
are shown in Table 1. The patients who did not have follow-up were older than the patients included in the study (mean age: 69.2 vs 67.2 years; 95% CI of the difference,
0.30-3.34) and more frequently demonstrated intraplaque bleeding (66%
vs 58%; P = .04), whereas the other baseline and plaque characteristics did not differ.
The overall incidence of 50% or greater restenosis of the target vessel 1 year after carotid endarterectomy was 17%, including 8% with 70% or greater restenosis, of whom 3% had occlusion of the target vessel. Two clinical parameters associated with restenosis were identified in univariate analysis: hypercholesterolemia and asymptomatic clinical presentation (Table 2). Patch closure was associated with decreased incidence of restenosis during follow-up: the risk of 50% or greater restenosis for venous patch (n = 269), Dacron patch (n = 131), and primary closure (n = 100) were 12%, 21%, and 26%, respectively (patch vs no patch: RR, 0.56; 95% CI, 0.37-0.84; P = .006). Age, sex, and current smoking were not associated with restenosis during follow-up. There were no clinical parameters at baseline that were associated with occlusion of the target vessel during follow-up.
In univariate analysis, marked presence of plaque macrophages (moderate or heavy) (see Figure for representative image) was associated with reduced risk of restenosis during follow-up (Table 3).
Patients whose histological examination of the plaque revealed moderate or heavy macrophage infiltration had a lower risk of developing 50%
or greater restenosis (RR, 0.47; 95% CI, 0.32-0.71) and a lower risk of developing 70% or greater restenosis (RR, 0.36; 95% CI, 0.19-0.68)
compared with patients whose plaque histology showed none or minor macrophage infiltration. Carotid arteries with macrophage-rich plaques were less prone to occlusion during follow-up (1.1% vs 5.2%; RR, 0.20;
95% CI, 0.06-0.72; P = .006).
In addition, in univariate analysis, a large lipid core size was associated with lower restenosis rate during follow-up. Patients with a large lipid core size (>40% of plaque area) in the dissected carotid plaques had a lower risk of 50% or greater restenosis (RR,
0.44; 95% CI, 0.26-0.76) and a lower risk of 70% or greater restenosis (RR, 0.38; 95% CI, 0.18-0.82) compared with patients with a very small or absent lipid core (<10%). We observed no association between lipid core size and occlusion. Other histological plaque characteristics showed no relationship with either restenosis or occlusion. The relationship of macrophages and lipid core size with restenosis persisted when occlusion was excluded from the definition of restenosis.
The IL-8 plaque levels were lower when restenosis was observed,
further supporting the concept that features of unstable plaques are related with a lower incidence of restenosis (≥50%: 78 pg/L [95%
CI, 32-125] vs 247 pg/L [95% CI, 189-305]; P < .001; and ≥70%: 114 pg/L [95% CI, 40-188] vs 258
pg/L [95% CI, 195-321]; P = .006,
In multivariate logistic regression analysis, macrophage infiltration and lipid core size were independently associated with decreased risk of developing 50% restenosis. The initial variable set of the models consisted of age, sex, clinical presentation (symptomatic vs asymptomatic),
hypercholesterolemia, patch closure (primary, venous, Dacron), macrophage infiltration, and lipid core size. The adjusted odds ratios with 95%
CIs derived from the resulting models are shown in Table 4. The clinical parameter that remained significant in multivariate logistic regression analysis was patch closure. The area under the receiver operating characteristic curve for predicting restenosis was 0.73 (95% CI, 0.67-0.79) for model 1 and 0.75 (95% CI, 0.69-0.82) for model 2.
Within 1 year, the incidence of stroke, myocardial infarction,
and death was 9 of 482 patients (1.9%), 5 of 482 patients (1.0%),
and 0 of 482 patients (0%) who underwent duplex follow-up (n = 500),
and 13 of 164 patients (7.9%), 7 of 164 patients (4.3%), and 17 of 164 patients (10%) who were not included in the study (n = 185).
In Kaplan-Meier survival analysis, infiltration of macrophages and lipid core size were not related with clinical outcome in the first year after carotid endarterectomy (occurrence of stroke, myocardial infarction, or death), neither in the patients with duplex follow-up nor in the patients without duplex follow-up.
In the exploratory analysis to examine potential mechanisms of protection from restenosis in vessels with macrophage infiltration and lipid-rich plaques, 2 determinants were considered that may contribute to luminal renarrowing: increased carotid intima-media thickness and geometrical remodelling (eg, constrictive or adaptive remodeling12). The B-mode ultrasound substudy (n = 50)
revealed that at 1-year follow-up no single determinant was associated with this restenosis protection. Carotid intima-media thickness tended to be lower and vessel diameter larger in vessels with macrophage-rich plaques (n = 32), as observed before the endarterectomy procedure, compared with macrophage-poor plaques (n = 18)
(carotid intima-media thickness at 1 year: 1.22 mm [95% CI, 1.07-1.38]
vs 1.40 mm [95% CI, 1.18-1.62]; vessel diameter at 1 year: 10.64 mm [95% CI, 10.01-11.27] vs 10.13 mm [95% CI, 9.28-10.98]).
This is the first study to our knowledge that provides prospective evidence that the composition of the atherosclerotic plaque, low macrophage infiltration, and small or absent lipid core is associated with risk of restenosis after a vascular intervention.
Theoretically, assessment of atherosclerotic plaque composition using noninvasive imaging modalities may play a role in treatment stratification to help tailor treatment for the patient with carotid artery stenosis (carotid endarterectomy, carotid stenting, or medical treatment). Previously, it was shown that symptomatic clinical presentation is related to plaques with a large lipid core size and strong macrophage infiltration, and it was suggested that benefit of carotid endarterectomy might be less in plaques without these histopathological features.29 Our study shows that the vessels with these nonvulnerable plaques (low macrophage and lipid content) are more prone to develop restenosis after endarterectomy.
Stenting and endarterectomy of lipid-rich, echolucent plaques have previously been associated with an increased restenosis rate.13,30 Stenting of a lipid-rich vulnerable lesion elicits an inflammatory response that results in increased neointima formation.13 In our study, we examined restenosis when plaques had been dissected,
which makes comparison of these studies difficult. In another study,30 echolucent-rich plaques were associated with more restenosis following endarterectomy. However, in that study,
ultrasound parameters were used and histological validation was not performed.
Our exploratory substudy suggests that geometric vascular remodeling may partly explain the reduced occurrence of restenosis in vessels with plaques with high macrophage and lipid content. Hypothetically,
stenting could be an alternative treatment for vessels with plaques with low macrophage and lipid content, because the mechanism of stenting is based on the inhibition of constrictive remodeling.31 Present imaging techniques may be used to assess specific plaque characteristics preoperatively. The size of the lipid core can be determined with magnetic resonance imaging,
whereas in vivo imaging of plaque macrophages may be possible in the future.32,33 Our results could be helpful for developing strategies for incorporating imaging of atherosclerotic plaques into clinical trials.
Our results also implicate that the use of a patch for closure of the arteriotomy could be adjusted to the composition of the atherosclerotic plaque. Randomized evidence has shown that patch closure decreased the incidence of restenosis, especially when vein patches are used.34 Restricting patch closure to patients who are more likely to benefit might improve overall outcome by eliminating patch-related complications in patients who may not require patch closure (patients with plaques having high macrophage or lipid content).
In our patient cohort, the choice of patch closure was not randomized,
which may have biased these results. Studies that incorporate randomization of closure method coupled with imaging and plaque characterization may help elucidate if and how patch closure needs to be adjusted to the composition of the atherosclerotic lesion.
After surgical removal of an atherosclerotic lesion, the composition of the plaque is related to local restenosis. This relationship might be mediated via remainders of plaque at the endarterectomy site. However,
we consider this explanation unlikely because medial layers were visible in the majority of plaque specimens. The presence of an inflammatory plaque goes hand in hand with inflammation and protease activity in the media and adventitia,35,36 the layers remaining after endarterectomy. Leukocytes are highly infiltrated into the media and adventitia of inflammatory plaques,
giving rise to production of matrix metalloproteinases, which leads to thinning of the media and expansive remodeling.37 Our substudy on geometric vascular remodeling and neointima formation suggests that vessels with inflammatory plaques had larger vessel diameters at follow-up, even though vessel diameters were similar at baseline. It could be hypothesized that the plaque composition at baseline is related to a dynamic vascular remodeling process after the endarterectomy procedure.
Our study has several limitations. First, duplex ultrasound is not the gold standard for determining the degree of carotid stenosis.
However, duplex ultrasound has been extensively validated against angiography and pathology,38,39 and we standardized our data according to internationally accepted criteria. Second, we did not study plaque rupture. Assessment of plaque rupture was not always reliable in carotid endarterectomy specimens. The study by Lovett et al40 showed that plaque rupture in carotid endarterectomy specimens had relatively low reproducibility.
Third, postprocedure angiography was not performed to determine residual stenosis after carotid endarterectomy. Therefore, the stenosis grade measured at 1 year might be due to residual stenosis in some patients.
Finally, our findings are based on a small group of patients and our data need confirmation in larger studies in other populations at risk.
In conclusion, atherosclerotic plaque composition is associated with risk of restenosis after carotid endarterectomy. Macrophage infiltration and large lipid core size are associated with less restenosis.
Corresponding Author: Gerard Pasterkamp,
MD, PhD, Experimental Cardiology Laboratory, Division of Heart and Lungs, University Medical Center Utrecht, Heidelberglaan 100, Ste G02.523, 3584CX Utrecht, the Netherlands (firstname.lastname@example.org).
Author Contributions: Drs Hellings and Pasterkamp had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hellings,
Moll, De Kleijn, Pasterkamp.
Acquisition of data: Hellings, Moll,
De Vries, Ackerstaff, Seldenrijk, Met, Velema, Derksen, Pasterkamp.
Analysis and interpretation of data: Hellings, Seldenrijk, Velema, De Kleijn, Pasterkamp.
Drafting of the manuscript: Hellings,
Critical revision of the manuscript for important intellectual content: Moll, De Vries, Ackerstaff, Seldenrijk,
Met, Velema, Derksen, De Kleijn, Pasterkamp.
Statistical analysis: Hellings, Pasterkamp.
Obtained funding: Pasterkamp.
Administrative, technical, or material support: Hellings, Moll, De Vries, Ackerstaff, Seldenrijk, Met, Velema,
Derksen, De Kleijn, Pasterkamp.
Study supervision: Moll, De Vries,
De Kleijn, Pasterkamp.
Financial Disclosures: None reported.
Funding/Support: This study was funded by the University Medical Center, Utrecht, the Netherlands.
Role of the Sponsor: The University Medical Center Utrecht had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data;
or in the preparation, review, or approval of the manuscript.