OUP user menu

Collagenase matrix metalloproteinase-8 expressed in atherosclerotic carotid plaques is associated with systemic cardiovascular outcome

Wouter Peeters, Frans L. Moll, Aryan Vink, Peter J. van der Spek, Dominique P.V. de Kleijn, Jean-Paul P.M. de Vries, Jan H. Verheijen, Andrew C. Newby, Gerard Pasterkamp
DOI: http://dx.doi.org/10.1093/eurheartj/ehq517 2314-2325 First published online: 2 February 2011


Aims Atherosclerotic plaque rupture and subsequent thrombus formation are the major cause of acute cardiovascular events. Local plaque markers may facilitate detection of the vulnerable plaque and help identify the patient at risk for cardiovascular events. Matrix metalloproteinases (MMPs) are prevalent in the arterial wall throughout the arterial system and are associated with local plaque destabilization. We hypothesized that local MMP plaque levels are predictive for atherosclerotic cardiovascular events in other vascular territories.

Methods and results Atherosclerotic plaques were obtained from 543 patients undergoing carotid endarterectomy (CEA). Plaques were analysed for the presence of macrophages, lipid-core, smooth muscle cells, collagen, calcification, and presence of plaque haemorrhage. MMP-2, MMP-8, and MMP-9 levels were assessed within the plaque. Following CEA, all patients underwent follow-up during 3 years. The primary outcome was defined as the composite of vascular death, non-fatal vascular event, and surgical or percutaneous vascular intervention. In contrast with MMP-2 plaque levels, MMP-8 and MMP-9 levels in the plaque were associated with an unstable carotid plaque composition and clinical presentation at baseline. Increased plaque MMP-8 level (>4.58) was associated with an increased risk for the occurrence of secondary manifestations of atherosclerotic disease during follow-up [hazard ratio = 1.76, 95% CI (1.25–2.48)] (P= 0.001), whereas plaque MMP-2 and MMP-9 levels were not predictive for systemic cardiovascular events.

Conclusion In contrast with MMP-2, increased carotid MMP-8 and MMP-9 plaque levels are associated with an unstable plaque phenotype. High collagenase MMP-8 levels in the carotid plaque are associated with the occurrence of systemic cardiovascular outcome during follow-up.

  • Matrix metalloproteinase
  • Carotid
  • Plaque
  • Cardiovascular outcome


Atherosclerotic plaque rupture and subsequent luminal thrombosis are considered the crucial step in the development of acute cardiovascular events, such as myocardial or cerebral infarction. Patients who have initially suffered from acute manifestations of atherosclerotic disease are at increased risk for secondary clinical manifestations.

There is sufficient evidence suggesting that matrix metalloproteinases (MMPs) play a central role in degradation of the extracellular matrix (ECM), resulting in destabilization of the atherosclerotic plaque, which may contribute to plaque rupture.1,2 The integrity of the ECM depends on the balance of degradation and repair of collagen and other matrix components. Collagenase (MMP-8) and gelatinases (MMP-2, MMP-9) have been recognized as proteases contributing to atherosclerotic plaque rupture and clinical events by degenerating structural components of plaque matrix.36 In a smaller cohort, we previously showed that the MMP-2 level was associated with a stable fibrous plaque phenotype, whereas levels of MMP-8 and MMP-9 are higher in lipid-rich inflammatory plaques. Matrix metalloproteinases have also been considered circulating biomarkers that are related with systemic manifestations of atherosclerotic disease.710 It was shown that the local atherosclerotic plaque hides molecular targets associated with histological features of unstable plaques. Since atherosclerosis is a systemic arterial disease, we hypothesized that local plaque markers may be related with acute manifestations of atherosclerotic disease in other vascular territories and that local atherosclerotic plaque composition may serve as a fingerprint of atherosclerotic plaque progression in other arterial beds.11 The association of local plaque MMP levels with the occurrence of adverse events has not been investigated previously. In the present study, we examined the association of local carotid plaque MMP levels with clinical cardiovascular events during follow-up. We report that local plaque MMP-8 is associated with a higher incidence of secondary cardiovascular manifestations of atherosclerotic disease during follow-up.


Athero-Express biobank

Athero-Express is a longitudinal ongoing biobank study comprising atherosclerotic specimens obtained from patients who underwent carotid endarterectomy (CEA). The study design and the standardized protocol with respect to plaque processing were reported previously.12 After the operation, the patients underwent 3 years clinical follow-up including duplex scanning. All patients undergoing CEA in the two participating centres (St Antonius Hospital Nieuwegein and University Medical Center Utrecht) were asked to participate in the study and provided written informed consent. The medical ethics committees of both participating hospitals approved the study. For the present study, we included 543 plaques from consecutive patients operated between 24 March 2002 and 2 February 2006.

Patient population

The indications for CEA for asymptomatic and symptomatic patients were based on the outcomes of the Asymptomatic Carotid Atherosclerosis Study (ACAS), the Asymptomatic Carotid Surgery Trial (ACST), the North American Symptomatic Carotid Endarterectomy Trial (NASCET), and the European Carotid Surgery Trial (ECST).13 A multidisciplinary vascular team reviewed all CEA indications. Patients were examined pre-operatively and post-operatively by a neurologist to assess the cerebrovascular symptom status and to document any new neurologic deficits after CEA. The grade of stenosis of both carotid arteries was determined with duplex ultrasound pre-operatively. The patients filled out questionnaires concerning cardiovascular risk factors, medical history, and medication use. Clinical data were obtained from patient charts. The definitions of hypercholesterolaemia and hypertension were restricted to those cases requiring medical treatment. Serum levels of C-reactive protein were measured at the time of inclusion. Patients who suffered from stroke, transient ischaemic attack (TIA), or amaurosis fugax (AFX) were considered as being symptomatic. Patients operated due to a significant stenosis (>70%) without symptoms were considered asymptomatic. For the comparison of MMP levels between symptomatic and asymptomatic patients, patients who suffered from AFX were analysed separately, since the plaque composition from AFX lesions corresponded strongly with the composition of asymptomatic lesions.14

Follow-up and outcome

Follow-up protocol

All patients underwent a 3-year clinical follow-up after CEA. Patients completed a questionnaire informing if they had experienced any vascular event or had been hospitalized in the past year at 1, 2, and 3 years after surgery. If any of the questions was answered positively, further research was performed to validate the clinical outcome. According to a standard scheme discharge letters, and if needed, laboratory measurements and results of additional studies such as electrocardiograms or imaging studies were collected from the institution where the potential event occurred. If patients did not respond to the follow-up questionnaire, the general practitioner was contacted. Per potential outcome event, all available information was assessed by two members of the outcome assessment committee. If the outcome event between the members did not correspond, a third opinion was requested.

Definition of outcome

The primary outcome was defined as a composite of endpoints including, any death of vascular origin (fatal stroke, fatal myocardial infarction, sudden death, and other vascular death), non-fatal stroke, non-fatal myocardial infarction, and any arterial vascular intervention that had not already been planned at the time of inclusion (e.g. carotid surgery or angioplasty, coronary artery bypass, percutaneous coronary artery intervention, peripheral vascular surgery, or angioplasty). In addition, we distinguished three subgroups regarding the clinical outcome in different vascular territories; ‘coronary outcome’ was defined as (non) fatal myocardial infarction, coronary artery bypass, coronary artery intervention, and sudden death. ‘Stroke’ outcome defined as non-fatal and fatal stroke. Peripheral interventions comprised, leg amputation and peripheral arterial intervention during follow-up that had not been planned at the time of inclusion. A composite of ‘major outcome’ was defined as (non) fatal myocardial infarction, (non) fatal stroke, coronary artery bypass, coronary artery intervention, and sudden death.

Tissue examination

Determination of plaque phenotype

The atherosclerotic plaques were divided into segments of 5 mm thickness according to a standardized protocol as described previously.12 The segment with the greatest plaque burden was considered the culprit lesion and subjected to histological examination. Macrophage infiltration (CD68) and smooth muscle cell infiltration (alpha-actin) were quantitatively scored using computerized analyses (AnalySIS 3.2, Soft Imaging Systems GmbH, Münster, Germany) as well as semi-quantitatively as ‘no/minor’ or ‘moderate/heavy’. The amount of collagen (Picro Sirius Red) and calcification [Haematoxylin and eosin (H&E)] were semi-quantitatively scored as (1) ‘no/minor’ or ‘moderate/heavy’ staining. Collagen content was estimated with polarized light.

The size of the lipid core was visually estimated at 40 time magnification. The lipid core size was expressed as the percentage of total plaque area and scored in two categories: <40 and >40% of the total plaque area. Haematoxylin and eosin sections were used and the lipid core comprised predominantly cholesterol clefts.15

Plaque haemorrhages (fibrin stainings), including the combination of plaque haemorrhage at the luminal side and intraplaque haemorrhage, were scored as being ‘absent’ or ‘present’. Intraplaque vessel density (CD34) was determined by the average number of vessels of three hotspots within every single plaque. A hotspot was defined as one high power field at 40× magnification. For counting the vessels per hotspot, we used a grid (100 × 100 μm) overlying every hotspot to improve the reproducibility and to avoid counting vessels twice. The vessels that were crossed by a bar of the grid within the selected hotspots were counted. Increased vessel density was defined as an average vessel count per hotspot higher than the median (=8) of the cohort.

The histological examinations were performed by two independent observers, who were blinded for clinical outcome. The histological examinations showed good inter-observer and intra-observer reproducibility. In addition, the stainings for macrophages and smooth muscle cells were scored quantitatively using computer-based analyses, which revealed an excellent correlation with the semi-quantitative analyses.16

Matrix metalloproteinase 2, 8, and 9 activity measurements

Segments adjacent to the culprit lesion served for protein isolation. Protein isolation was performed according to a standardized protocol using Tris.12 Total MMP levels, the sum of active MMP and activatable pro-MMP, within the atherosclerotic specimens were quantified with the following specific Biotrak activity assays: MMP-2 RPN-2631, MMP-8 RPN-2635, and MMP-9 RPN-2634 (GE Healthcare LifeSciences, Buckinghamshire, UK) (detailed description including references—supplementary file Online). Matrix metalloproteinase levels were corrected for the total protein amount and are expressed as arbitrary units (AU). For the survival analyses, the median was arbitrarily chosen as the cut-off value, since the MMP levels were not normally distributed (MMP-2; ≤4.0 AU vs. >4.0 AU, MMP-8; ≤4.58 AU and >4.58 AU, MMP-9; ≤0.65 AU and >0.65 AU).

Staining of MMP-8 and double-staining of MMP-8 and CD68

To determine the localization of MMP-8 expression, 25 plaques were stained for MMP-8 as follows. Ethylenediaminetetraacetic acid pre-treated sections were stained with a monoclonal mouse anti-human pro/active MMP-8 antibody (dilution 1:800; R&D Systems, Minneapolis, MN, USA) followed by incubation with Powervision poly AP anti-mouse IgG ready-to-use (Immunologic, Duiven, The Netherlands). The signal was visualized with liquid permanent red (Dako, Glostrup, Denmark). A mouse monoclonal antibody of the same isotype was used as negative control.

Statistics and data analysis

For statistical analyses, SPSS 15.0 was used (SPSS Inc, Chicago, Illinois). Matrix metalloproteinase levels were related with categorized baseline variables by the non-parametric Man–Whitney U-test. Considering the histological plaque composition, characteristics were associated with MMP-8 level by the Man–Whitney U-test. The relation of MMP expression levels with computerized quantified continuous histological parameters were analysed by Spearman Bivariate correlation analyses. For survival analyses, we performed Kaplan–Meier survival analyses to estimate the cumulative event rate at 1, 2, and 3 years after CEA. To determine that the MMP level in relation to clinical outcome was independent from cardiovascular risk factors and medication use, we used a multivariate Cox-regression backward model, by leaving variables with a P-value > 0.1 step-wise out. Statistically significant associations with clinical outcome were defined as a 95% confidence interval not including 1.


Between 24 March 2002 and 2 February 2006, we included 543 consecutive patients who underwent CEA (asymptomatic 101, AFX 72, TIA 229, and stroke 141).

Matrix metalloproteinase levels and patient characteristics

The baseline characteristics of the population in relation to MMP levels are outlined in Table 1. The MMP-8 level was increased in male patients [median with interquartile range (IQR), 5.13 AU (3.02–9.63)] in comparison with female patients [3.87 AU (2.13–8.35), P = 0.001]. On the other hand, the MMP-2 level was increased in female patients compared with male patients [4.49 (3.12–6.38) vs. 4.03 (2.84–5.82), P = 0.04]. Patients with hypertension revealed decreased MMP-2 plaque level [with hypertension; 4.57 (3.08–6.40) vs. patients without hypertension; 4.02 (2.82–5.72), P = 0.02] and patients with an increased body mass index (>25 kg/m2) showed increased MMP-9 level [<25 kg/m2; 0.70 (0.36–1.40) vs. >25 kg/m2; 0.87 (0.43–1.55), P = 0.05]. Carotid atherosclerotic lesions from patients who suffered from TIA or stroke showed increased MMP-8 and MMP-9 levels in comparison with asymptomatic patients [5.26 (2.86–10.33) vs. 3.54 (2.53–6.59), P = 0.002 and 0.58 (0.35–1.27) vs. 0.91 (0.48–1.55), P = 0.007]. Remaining baseline characteristics were not associated with different plaque MMP levels, including the time elapsed between event and operation.17

View this table:
Table 1

Baseline characteristics in relation to matrix metalloproteinase-8 activity

Patient characteristicsMMP-2, median (IQR)P-valueMMP-8, median (IQR)P-valueMMP-9, median (IQR)P-value
Age (years) (mean) (range)67.537.5–91.1
 Male389/54371.6%4.03 (2.84–5.82)5.13 (3.02–9.63)0.71 (0.35–1.41)
 Female154/54328.4%4.49 (3.12–6.38)0.043.87 (2.13–8.35)0.0010.86 (0.45–1.49)0.10
Current smoker
 No387/52873.3%4.18 (2.89–6.00)4.57 (2.57–9.05)0.76 (0.43–1.45)
 Yes141/52826.7%4.00 (2.94–5.70)0.605.31 (2.95–10.30)0.220.88 (0.42–1.52)0.69
Diabetic status
 Non-diabetes442/54181.7%4.15 (2.92–6.02)4.76 (2.71–9.40)0.80 (0.43–1.51)
 Diabetes99/54118.3%4.14 (2.92–5.50)0.695.09 (2.51–10.05)0.870.76 (0.38–1.41)0.39
 No189/54035.0%4.57 (3.08–6.40)4.41 (2.49–9.11)0.88 (0.42–1.55)
 Yes351/54065.0%4.02 (2.82–5.72)0.024.99 (2.72–9.53)0.530.79 (0.43–1.42)0.62
 No220/54240.6%4.12 (2.83–6.05)4.93 (2.98–9.01)0.82 (0.39–1.50)
 Yes322/54249.4%4.18 (2.99–5.95)0.754.66 (2.51–9.96)0.630.76 (0.43–1.44)0.77
Body mass index
 BMI <25176/48636.2%4.16 (2.88–5.92)4.50 (2.46–8.94)0.70 (0.36–1.40)
 BMI >25310/48663.8%4.13 (2.92–5.95)0.954.86 (2.76–9.61)0.320.87 (0.43–1.55)0.05
History of vascular intervention
 No362/54366.7%4.22 (2.89–6.19)4.93 (2.67–9.61)0.84 (0.42–1.53)
 Yes181/54333.3%3.91 (2.96–5.31)0.104.54 (2.56–8.77)0.580.76 (0.41–1.35)0.30
History myocardial infarction
 No426/54178.7%4.15 (2.89–6.0)4.88 (2.57–9.45)0.80 (0.42–1.48)
 Yes115/54121.3%4.03 (2.94–5.59)0.584.41 (2.93–9.07)0.850.75 (0.44–1.45)0.90
Bilateral stenosis
 No315/54358.0%4.08 (2.92–5.95)4.43 (2.43–8.95)0.82 (0.43–1.45)
 Yes228/54342.0%4.18 (2.90–6.0)0.885.0 (3.04–9.65)0.060.76 (0.42–1.48)0.65
Statin use
 No145/54126.8%4.12 (2.84–5.90)4.28 (2.62–9.06)0.77 (0.37–1.44)
 Yes396/54173.2%4.15 (2.94–6.0)0.925.0 (2.68–9.64)0.750.82 (0.43–1.47)0.38
Aspirin use
 No56/54110.4%3.89 (3.08–5.72)5.07 (2.69–11.11)0.83 (0.41–1.54)
 Yes485/54189.6%4.16 (2.89–6.0)0.474.81 (2.65–9.25)0.410.79 (0.42–1.47)0.71
Oral anti-coagulants
 No467/54386.0%4.15 (2.92–5.97)4.68 (2.61–9.53)0.80 (0.42–1.48)
 Yes76/54314.0%4.07 (2.88–5.97)0.815.54 (3.02–9.21)0.370.76 (0.44–1.37)0.87
 No516/54095.5%4.15 (2.93–6.0)4.84 (2.66–9.52)0.80 (0.42–1.48)
 Yes24/5404.5%4.09 (2.41–5.82)0.444.54 (2.50–8.94)0.730.67 (0.37–1.34)0.60
Serum levels
 C-reactive protein (mean) (range)7.80.17–190
 HDL (mmol/L) (mean) (range)1.130.28–2.45
 LDL (mmol/L) (mean) (range)2.840.65–7.68
Clinical presentation
 Asymptomatic101/5434.15 (2.97–5.93)3.54 (2.53–6.59)0.58 (0.35–1.27)
 Amaurosis Fugax72/5434.18 (3.06–6.19)4.31 (2.44–8.65)0.90 (0.44–1.55)
 TIA229/5434.12 (2.77–6.01)5.01 (2.82–10.53)0.93 (0.52–1.54)
 Stroke141/5434.05 (2.62–5.53)5.54 (2.90–10.11)0.68 (0.40–1.32)
 Asymptomatic101/4714.15 (2.97–5.93)3.54 (2.53–6.59)0.58 (0.35–1.27)
 Symptomatic (TIA and Stroke)370/4714.15 (2.97–6.09)0.975.26 (2.86–10.33)0.0020.91 (0.48–1.55)0.007
  • Matrix metalloproteinase activities are expressed in arbitrary units (AU) as median with interquartile range (IQR).

  • BMI, body mass index, FU, follow-up; IR, interquartile range; SD, standard deviation; TIA, transient ischaemic attack.

Matrix metalloproteinase levels and plaque phenotype

Table 2 shows data with respect to the MMP levels in relation to plaque composition. In short, increased MMP-8 and MMP-9 levels corresponded with an unstable plaque phenotype, characterized by increased plaque inflammation, and an increased lipid core size. In addition, smooth muscle cell and collagen content were inversely related with MMP-8 and MMP-9 levels. Presence of plaque haemorrhage was positively associated with MMP-8 level and inversely related with MMP-2 level. Vessel density was also inversely related with MMP-2 level within the atherosclerotic plaque. These results support previous observations that MMP-8 and MMP-9 are associated with an unstable plaque phenotype and that MMP-2 is associated with a relatively more stable plaque phenotype, since plaque haemorrhages and increased vessel density are determinants of unstable plaques.18

View this table:
Table 2

Matrix metalloproteinase level in relation to histological plaque composition of atherosclerotic carotid plaques

Characteristic/categoryMacrophages, median (IQR)SMC, median (IQR)Calcification, median (IQR)Collagen, median (IQR)Vessel density, median (IQR)Plaque haemorrhage, median (IQR)Lipid core size, median (IQR)
 No/minor4.24 (2.87–6.01)3.87 (2.82–5.60)4.19 (3.15–6.14)4.15 (3.09–6.47)4.19 (3.04–6.37)Absent4.24 (3.10–6.12)0–40%4.15 (2.86–5.95)
 Moderate/heavy4.05 (2.93–5.85)4.19 (2.94–6.02)4.07 (2.82–5.82)4.13 (2.88–5.83)4.05 (2.70–5.59)Present3.77 (2.72–5.39)>40%4.12 (3.06–6.12)
 QuantitativeR = 0.04R = 0.04
 No/minor3.81 (2.41–7.60)6.99 (3.23–12.03)4.68 (2.57–10.07)5.48 (3.16–10.75)4.72 (2.58–9.57)Absent4.39 (2.50–9.35)0–40%4.05 (2.40–7.80)
 Moderate/heavy5.66 (3.21–10.80)4.27 (2.53–8.34)4.84 (2.66–8.75)4.50 (2.53–9.04)5.04 (2.77–9.53)Present6.89 (2.56–11.53)>40%6.91 (3.52–11.41)
 QuantitativeR = 0.19R = −0.20
 No/minor0.62 (0.34–1.19)1.12 (0.57–1.86)0.91 (0.42–1.51)1.02 (0.54–1.73)0.91 (0.45–1.54)Absent0.76 (0.41–1.41)0–40%0.64 (0.34–1.24)
 Moderate/heavy0.94 (0.53–1.56)0.72 (0.39–1.26)0.76 (0.43–1.41)0.76 (0.40–1.37)0.76 (0.40–1.35)Present0.89 (0.48–1.56)>40%1.13 (0.62–1.86)
 QuantitativeR = 0.22R = −0.21
  • Matrix metalloproteinase activities are expressed in arbitrary units (AU) as median with the interquartile range (IQR). Categorized histological plaque characteristics were associated with MMP-8 level by the Mann–Whitney U-test. Correlations of the histological plaque characteristics, quantified by computerized analyses, with continuous MMP activities were analysed by the Spearman bivariate correlation test. Differences with a P-value <0.05 were regarded as being significant.

  • *Plaque haemorrhage included haemorrhage at the luminal side and intraplaque haemorrhage, which was scored as absent or present. Lipid core size was expressed as <40 and >40% of the total plaque area.

We have performed histology stainings that supported the cross-sectional observations, showing co-localization of increased MMP-8 expression with a subset of macrophage infiltration (Figure 1A and B). These findings support previous observations by Molloy et al.5 and Laxton et al.19

Figure 1

Matrix metalloproteinase-8 expression in a subset of macrophages. (A) MMP-8 expression (in red) in the shoulder of an atherosclerotic plaque (Bar = 200 μm). (B) Consecutive slide stained for macrophages (in brown).

Matrix metalloproteinase levels and clinical follow-up

In total 141 patients reached a primary endpoint during a mean follow-up of 2.4 years (SD = 0.97 maximum = 3 years) (Table 3). Atherosclerotic lesions from patients who reached a primary endpoint (event) during follow-up revealed increased MMP-8 level [5.92 (3.37–9.59)] compared with control patients [4.25 (2.48–9.40), P = 0.002] that did not reached a primary endpoint (Table 3). Kaplan–Meier estimates of the MMP-8 plaque level in relation to clinical outcome during follow-up after surgery revealed a 3-year risk of 20.7% in patients with low MMP-8 plaque level and 32.8% in patients with increased MMP-8 level [hazard ratio (HR) = 1.76, 95% confidence interval 1.25–2.48, P = 0.001] (Table 4, Figure 1). After adjusting for general cardiovascular risk factors, C-reactive protein levels and pre-operative medication use, carotid plaque MMP-8 level was independently associated with primary outcome [1.68 (1.19–2.37), P = 0.003] (Table 5). Estimates of the 3-year risk to experience a major vascular endpoint was 11.0 and 18.3% [HR = 1.83, 95% CI (1.14–4.04, P = 0.01)] for, respectively, decreased and increased MMP-8 level (Table 6, Figure 1). Regarding vascular event in specific arterial beds, the estimates were 6.7 and 13.9% [HR = 2.24, 95% CI (1.25–4.04), P = 0.007], 7.0 and 9.5% [1.48 (0.80–2.72), P = 0.21], and 12.2 and 20.3% [HR = 1.77, 95% CI (1.13–2.78), P = 0.01] for coronary events, recurrent stroke, and peripheral interventions, respectively (Table 6, Figure 1). Local plaque levels of MMP-2 or MMP-9 were not associated with primary outcome (Tables 3 and 4, Figure 2) and vascular endpoints in different vascular territories. Additional analyses examining continuous MMP levels with clinical outcome supported the observations [MMP-2 HR 0.97 95% CI (0.91–1.03), MMP-8 HR 1.02 (1.01–1.04), MMP-9 HR 1.03 (0.88–1.21)].

View this table:
Table 3

Characteristics with respect to follow-up and matrix metalloproteinase activities

Baseline characteristics
 Mean follow-up (years) (SD, maximum)2.4(0.97, 3.0)
 Primary outcome (3 years of FU)141/54326.0%
Primary clinical outcome
 Non-fatal myocardial infarction/coronary intervention34/14124.1%
 Non-fatal stroke31/14122.0%
 Peripheral vascular intervention54/14139.3%
Primary clinical outcome and MMP-2 levelP-value
 Control, median (IR)4.16 (2.87–6.0)
 Event, median (IR)4.04 (3.05–5.95)0.95
Primary clinical outcome and MMP-8 level
 Control, median (IR)4.25 (2.48–9.40)
 Event, median (IR)5.92 (3.37–9.59)0.002
Primary clinical outcome and MMP-9 level
 Control, median (IR)0.76 (0.40–1.42)
 Event, median (IR)0.95 (0.46–1.54)0.14
  • Baseline characteristics with respect to clinical outcome and MMP activities.

View this table:
Table 4

Matrix metalloproteinase activities in relation to cardiovascular outcome

Event rateCumulative risk (KM estimate)Risk differenceP-value
1 year2 years3 yearsHR (95% CI)1 year2 years3 years
All patients141/543 (26.0%)76/543 (14.0%)36/543 (6.6%)29/543 (5.3%)
 Primary outcome
  Below median (≤4.0)68/253 (26.9%)16.2%21.5%27.9%Ref.Ref.Ref.Ref.
  Above median (>4.0)73/290 (25.2%)12.1%20.2%26.1%0.90 (0.65–1.25)−4.1%−1.3%−1.8%0.54
 Primary outcome
  Below median (≤4.48)52/264 (19.7%)9.9%15.7%20.7%Ref.Ref.Ref.Ref.
  Above median (>4.58)89/279 (31.9%)18.3%25.7%32.8%1.76 (1.25–2.48)8.4%10.0%12.1%0.001
 Primary outcome
  Below median (≤0.65)55/229 (24.0%)12.2%18.0%25.4%Ref.Ref.Ref.Ref.
  Above median (>0.65)86/314 (27.4%)15.3%22.9%28.2%1.18 (0.84–1.65)3.1%4.9%2.8%0.35
  • Matrix metalloproteinase activities binomially expressed in relation to clinical outcome during 3 years of follow-up.

View this table:
Table 5

Multivariate model

 Risk of primary outcome
Univariate analyses HR (95% CI)P-valueMultivariate analyses HR (95% CI)P-value
Matrix metalloproteinase 8 level1.76 (1.25–2.48)0.0011.68 (1.19–2.37)0.003
Gender1.95 (1.27–3.0)0.0021.78 (1.15–2.74)0.009
Age1.02 (1.0–1.04)0.041.02 (1.0–1.04)0.09
Smoker1.28 (0.88–1.84)0.19
Diabetes1.04 (0.68–1.59)0.86
Hypertension0.99 (0.70–1.40)0.95
Hypercholesterolaemia1.25 (0.89–1.77)0.20
Body mass index (BMI) >25 kg/m21.03 (0.71–1.51)0.86
History of vascular intervention1.48 (1.06–2.07)0.021.29 (0.91–1.81)0.15
History myocardial infarction1.43 (0.98–2.07)0.071.03 (0.67–1.56)0.91
Bilateral carotid stenosis1.92 (1.38–2.67)<0.0011.82 (1.31–2.54)<0.001
Statin use1.28 (0.87–1.91)0.21
Aspirin0.57 (0.36–0.90)0.020.61 (0.38–0.97)0.04
Oral coagulants0.98 (0.92–1.05)0.56
C-reactive protein (mg/L)1.0 (0.99–1.01)0.72
Symptomatic carotid lesion1.07 (0.70–1.65)0.73
  • This table demonstrates the hazard ratios (HR) of clinical variables in relation to MMP-8 level as a binomial variable (cut-off value 4.58). All analyses were performed in subgroups with no missing values of any variable. Univariate significant variables were included in a multivaritie Cox-regression model.

View this table:
Table 6

Matrix metalloproteinase-8 level in relation to cardiovascular outcome in different vascular territories

Event rateCumulative risk (KM estimate)Risk differenceP-value
1 year2 years3 yearsHR (95% CI)1 year2 years3 years
All patients141/543 (26.0%)76/543 (14.0%)36/543 (6.6%)29/543 (5.3%)
Primary outcome
 Below median (≤4.48)52/264 (19.7%)9.9%15.7%20.7%Ref.Ref.Ref.Ref.
 Above median (>4.58)89/279 (31.9%)18.3%25.7%32.8%1.76 (1.25–2.48)8.4%10.0%12.1%0.001
Major vascular outcome
 Below median (≤4.48)27/264 (10.2%)5.3%7.3%11.0%Ref.Ref.Ref.Ref.
 Above median (>4.58)50/279 (17.2%)10.0%14.5%18.3%1.83 (1.14–2.91)4.7%7.2%7.3%0.012
 Below median (≤4.48)16/264 (6.1%)2.3%3.9%6.7%Ref.Ref.Ref.Ref.
 Above median (>4.58)36/279 (12.9%)4.4%9.4%13.9%2.24 (1.25–4.04)2.1%5.5%7.2%0.007
 Below median (≤4.48)17/264 (6.4%)3.8%5.0%7.0%Ref.Ref.Ref.Ref.
 Above median (>4.58)26/279 (12.9%)7.2%%9.1%9.5%1.48 (0.80–2.72)3.4%4.1%2.5%0.211
Peripheral intervention
 Below median (≤4.48)30/264 (11.4%)4.6%9.0%12.2%Ref.Ref.Ref.Ref.
 Above median (>4.58)52/279 (18.6%)9.2%15.4%20.3%1.77 (1.13–2.78)4.6%6.4%8.1%0.013
Figure 2

Relation between matrix metalloproteinase (MMP)-8 plaque levels and clinical outcome in different vascular territories. The figures show the percentage of patients who suffer an event during three years of follow-up after carotid endarterectomy. The red line represents plaques with MMP-8 expression above the median and the blue line plaques with MMP-expression below the median.


Besides the observations that carotid plaque levels of MMP-2, MMP-8, and MMP-9 are associated with stable and unstable plaque characteristics, the present study shows that just MMP-8 plaque level is associated with systemic cardiovascular events during 3 years of follow-up in patients with carotid atherosclerotic disease. In addition, local MMP-8 plaque levels were specifically associated with vascular events in different arterial territories.

Matrix metalloproteinases are widely expressed in atherosclerotic lesions in multiple vascular territories and are recognized as prominent ECM degrading enzymes, which contribute to thinning of the fibrous cap and cap rupture.36

The integrity of the ECM depends on a balance of degradation and repair of collagen and other matrix components. Matrix metalloproteinase (MMP)-2, MMP-8, and MMP-9 degenerate structural components of plaque matrix such as connective tissue collagens type I, II, III, IV, V, VII, X, XII and elastin. In atherosclerotic lesions, MMPs are frequently co-localized with traditional vulnerable plaque characteristics, such as increased macrophage presence and neutrophil infiltration, suggesting that proteases contribute to plaque destabilization.2

In the present study, we confirmed previous observations that increased MMP-2 level is associated with a more stable plaque phenotype and that increased MMP-8 and MMP-9 levels correspond with unstable plaque characteristics.18 Symptomatic lesions demonstrated higher MMP-8 and MMP-9 levels, suggesting that these proteases are related with clinical manifestations of atherosclerotic disease. In addition, carotid lesions from men were associated with increased MMP-8 levels and on the other hand, plaques from female patients were related with increased MMP-2 level, which is consistent with previous observations that women have more stable atherosclerotic lesions in comparison with men.

Recent reports supported the findings that MMP-8 and MMP-9 levels are associated with an unstable plaque composition and symptomatic lesions.35 Kunte et al.3 and Molloy et al.5 showed that increased MMP-8 and MMP-9 carotid plaque levels were significantly increased in symptomatic lesions. In addition, experimental studies supported current observations and showed that MMP-8 and MMP-9 are associated with atherosclerotic plaque development, increased plaque inflammation, and decreased smooth muscle cell content.19,20

The main aim of the present study is based on an emerging concept that local plaque protein levels are associated with the risk for future systemic cardiovascular events. Since atherosclerosis has been considered a systemic arterial disease21,22 and plaque composition correlates between different arterial segments within individuals,21,23 we hypothesized that the local atherosclerotic carotid plaque composition reflects disease progression in other arterial territories and holds targets that enable risk stratification and identification of the vulnerable patients. In a previous study, Hellings et al.15 have shown that the atherosclerotic plaque composition holds markers that are associated with future restenosis and vascular remodelling. We hypothesized that the local atherosclerotic plaque serves as a fingerprint of atherosclerotic plaque progression in the whole arterial tree, which would facilitate risk stratification of patients at risk for systemic cardiovascular events. We show for the first time that increased MMP-8 levels within carotid atherosclerotic lesions are associated with adverse systemic cardiovascular events. Survival analyses revealed a 76% increased event rate during 3-year follow-up of patients with MMP-8 plaque levels above the median (>4.58) compared with patients with MMP-8 levels below the median (≤4.58). Besides gender and the presence of bilateral carotid stenosis, the relation of increased local plaque MMP-8 levels and future vascular events was independent from general risk factors and pre-operative medication.24

Question may raise whether MMP-8's specific relation with events is restricted to the MMP, or that it can be demonstrated with the expression of any inflammatory marker that has been characterized with plaque vulnerability. We feel that we provide supportive evidence that this is not the case. We have data showing that MMP-8 is associated with outcome independent of atheromatous or fibrous plaque composition. Moreover, not just other inflammatory parameters such as MMP-2 and MMP-9 but also histological plaque characteristics that characterize the vulnerable plaque (e.g. macrophage content and lipid core size) did not show a relation with event rate.25

The following explanations for this observation can be suggested. In contrast with macrophage infiltration, we have recently shown that MMP levels do not change over time following an ischaemic event,16 which indicates that the local carotid plaque holds stable markers that are related with manifestations of atherosclerotic plaque progression in other arterial territories. MMP-8 levels remain stable in vulnerable lesions of the vulnerable patient after carotid surgery, contributing to an increased risk of adverse systemic cardiovascular events.

Matrix metalloproteinase-8 is also expressed by neutrophils and the relative contribution of neutrophils to MMP-8 expression could alter expression levels compared with MMP-9. However, also neutrophil presence has been found to be unrelated with adverse events during follow-up making it unlikely that MMP-8 release by neutrophils would explain the observed higher event rate when MMP-8 expression is increased.26 Another explanation could be the presence of a variety of macrophage phenotypes derived from monocyte subsets in response to mediators of innate and acquired immunity that have been found in plaques which may have different impacts on plaque development. Although unproven we could discuss that MMP-8 and MMP-9 expression levels may differ in macrophage subtypes.27 This should be investigated in future experiments.

Local carotid plaque levels of gelatinases MMP-2 and MMP-9 were not associated with adverse systemic cardiovascular events. In contrast with MMP-2, local MMP-8 levels were associated with increased presence of intraplaque haemorrhage, suggesting that MMP-8 indeed is associated with local vulnerability. Another explanation that local collagenase MMP-8 and not MMP-2 or MMP-9 is related with adverse cardiovascular events may be that the predominant structural components of the atherosclerotic plaque cap, collagens I, II, and III, are predominantly degraded by collagenase MMP-8.28,29 Cleavage of these fibrous elements may dangerously weaken the atherosclerotic plaque cap. MMP-2, but not MMP-9, has also been recognized as a collagen I, II, III degrading enzyme in a similar manner as the collagenase MMP-8; however, the collagenolytic effect of MMP-2 has been considered much weaker in comparison with MMP-8.28 In addition, Herman et al.29 have shown that MMP-8 has a three-fold increased enzymatic activity against collagen type I, which is the major load bearing collagen of the fibrous cap. Degradation of collagen I may contribute to threatened plaque destabilization.

The current study provides opportunities for diagnostic imaging modalities. Since recent imaging studies have succeeded to identify MMPs in vulnerable atherosclerotic lesions, the outcome of the current study provides new opportunities for prognostic molecular imaging modalities to identify the vulnerable patient and to prevent acute clinical atherosclerotic manifestations.3033 Matrix metalloproteinase expression has been visualized in vivo in animal studies and very recently Wallis de Vries et al.33 have identified MMP expression by near-infrared fluorescence imaging in excised human CEA specimens. The observations of the present study provide more rationale to consider MMPs as imaging target to stratify patients at risk for systemic cardiovascular events.

The current study holds several limitations. This is the first biobank study investigating the association of MMP plaque levels with systemic cardiovascular outcome. Therefore, clinical application of the observations needs to be considered carefully, since validation studies are currently lacking. Experimental studies revealed genetic polymorphisms of MMPs34 and suggest that different genotypes may be associated with a part of the pathogenic process or particular clinical outcome, such as myocardial infarction or stroke. It reasonable that genetic polymorphism might have affected the association of MMP-2 and MMP-9 with clinical outcome in the present study;31 however, we were not able to exclude genetic variations in this study. In addition, MMPs, once activated, are capable of degrading the ECM. Therefore, it is essential that tissue inhibitors of metalloproteinases (TIMPs) carefully control MMP activity.35 In the present study, we have not taken TIMP expression into account, which might have influenced the outcome. New studies are needed to clarify the association of the TIMP/MMP ratio in relation clinical outcome. Statins have been associated with decreased MMP secretion. Although local MMP-8 levels might be affected by pre-operative statin use, we did not observe different MMP levels between statin using and non-statin using patients.3,24 Prospective studies are essential to clarify the association of statin use and plaque MMP levels. Finally, the current study comprised a selection of MMPs (MMP-2, -8 and -9). These proteases are all expressed by macrophage foam cells, but MMP-2 is also strongly associated with the presence of smooth muscle cells;18 we cannot exclude if other macrophages related MMPs may be predictive targets.

In conclusion, in contrast with MMP-2, increased carotid plaque MMP-8 and MMP-9 levels are associated with an unstable plaque phenotype, but only specifically collagenase MMP-8 levels are associated with the occurrence of systemic cardiovascular outcome during follow-up.


The reported work was supported by the European Union (grant EU OIF21773). This study was funded by the University Medical Center of Utrecht (UMCU) and the Interuniversity Cardiology Institute of the Netherlands (ICIN).

Conflicts of interest: the concept of the predictive value of local plaque markers in relation to clinical outcome has been incorporated in a start-up company, CAVADIS. Professor Pasterkamp, Professor Moll and Dr de Kleijn are co-founders of Cavadis B.V.


The authors thank Louise Catanzariti and Evelyn Velema for technical assistance. The UMCU and the ICIN had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation or approval of the manuscript.


View Abstract