Matrix metalloproteinase-9 and tissue inhibitor of metalloproteinase-1 in hypertension and their relationship to cardiovascular risk and treatment: A substudy of the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT)

Abstract

Background:

Hypertension results in structural changes to the cardiac and vascular extracellular matrix (ECM). Matrix metalloproteinases (MMP) and their inhibitors (TIMP) may play a central role in the modulation of this matrix. We hypothesized that both MMP-9 and TIMP-1 would be abnormal in hypertension, reflecting alterations in ECM turnover, and that their circulating levels should be linked to cardiovascular (CHD) and stroke (CVA) risk scores using the Framingham equation. Second, we hypothesized that treatment would result in changes in ECM indices.

Methods:

Plasma MMP-9 and TIMP-1 were measured before and after treatment (median 3 years) from 96 patients with uncontrolled hypertension participating in the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT). Pretreatment values were compared to circulating MMP-9 and TIMP-1 levels in 45 age- and sex-matched healthy controls.

Results:

Circulating pretreatment MMP-9 and TIMP-1 levels were significantly higher in patients with hypertension than in the normotensive controls (P = .0041 and P = .0166, respectively). Plasma MMP-9 levels decreased, and TIMP-1 levels increased after treatment (P = .035 and P = .005, respectively). Levels of MMP-9 correlated with CHD risk (r = 0.317, P = .007) and HDL cholesterol (r = −0.237, P = .022), but not CVA risk. There were no significant correlations between TIMP-1 and CVA or CHD scores.

Conclusions:

Increased circulating MMP-9 and TIMP-1 at baseline in patients with hypertension could reflect an increased deposition and retention of type I collagen at the expense of other components of ECM within the cardiac and vascular ECM. After cardiovascular risk management, MMP-9 levels decreased and TIMP-1 levels increased. Elevated levels of MMP-9 also appeared to be associated with higher Framingham cardiovascular risk scores. Our observations suggest a possible role for these surrogate markers of tissue ECM composition and the prognosis of cardiovascular events in hypertension. Am J Hypertens 2004;17:764–769 © 2004 American Journal of Hypertension, Ltd.

Extracellular matrix, matrix metalloproteinase-9, tissue inhibitor of metalloproteinase-1, hypertension, Framingham risk score

Structural changes in hypertension affecting the left ventricular myocardium and vascular tree are mediated in part by cell hypertrophy and hyperplasia, but these important processes are accompanied by changes in extracellular matrix (ECM) turnover. The ECM is vital for maintaining tissue integrity and in defining the physical properties of these structures. A large range of matrix metalloproteinases (MMPs) exist within the ECM controlling the dynamic deposition and turnover of a variety of its structural components. Some of these enzymes have natural inhibitors of activity, the tissue inhibitors of metalloproteinases (TIMPs).¹ It is assumed that the prevailing concentration and activity of both MMPs and (where relevant) TIMPs determine ECM composition at any given time.

Collagen types I and III predominate in the cardiovascular ECM. The former has high tensile strength and rigidity because of thicker strands, whereas the latter has a weave-type pattern and provides elasticity.² A variety of studies have demonstrated altered collagen composition in the heart and vascular tree in hypertension, favoring the presence of type I collagen (by increased deposition and reduced breakdown) over the other subtypes and an increase in type I/III collagen ratio.^3,4 These changes are related to the increased tissue stiffness and reduced vascular compliance found in hypertensive heart disease. Indeed, circulating levels of TIMP-1 are raised in hypertension, and may be associated with myocardial stiffness or diastolic heart failure.⁵ Matrix metalloproteinase-1, which degrades collagen type I and is inhibited by TIMP-1, is reduced in the blood of patients with hypertension.⁶ These findings are consistent with a mechanism for reduced turnover of collagen type I in hypertension.

At present little is known about the presumed role of circulating MMP-9, a gelatinase in hypertension. This MMP may have a role as a prognostic marker in patients with coronary artery disease.⁷ We therefore hypothesized that both MMP-9 and TIMP-1 would be abnormal in hypertension, and that their circulating levels are linked to cardiovascular and stroke risk scores using the Framingham equation (an internationally accepted scoring system to determine cardiovascular and stroke risk). Second, we hypothesized that a “package of care” of cardiovascular risk management in hypertension would result in changes in circulating concentrations of MMP-9 and TIMP-1. To test these hypotheses, we measured MMP-9 and TIMP-1 levels in a cross-sectional and longitudinal study of hypertensive subjects.

Methods

The patients studied were derived from 96 patients screened within the Anglo-Scandinavian Cardiac Outcome Trial (ASCOT) attending City Hospital, Birmingham, United Kingdom. The details of this study are published elsewhere.⁸ Briefly, the inclusion criteria were patients with blood pressures (BPs) of >160/100 mm Hg if untreated or if treated, >140/90 mm Hg—in association with risk factors. Blood pressure was measured in triplicate, seated, in the nondominant arm using an OMRON (Omron Healthcare, Milton Keynes, UK) semiautomated device after a 10-min period of rest in a quiet room. The average of the latter two readings was used.

Routine clinical and laboratory assessments included a full medical history of relevant vascular disease; the presence of electrocardiographic left ventricular hypertrophy (LVH; according to the Sokolow-Lyon criteria); screening for diabetes (according to the World Health Organization guidelines); definition of Framingham risk scores⁹ documenting medical history; sex; age (>55 years); screening for microalbuminuria/proteinuria; current smoking status; nonfasting lipid profile (including total and fractionated cholesterol and triglycerides); family history of premature (<50 years) coronary artery disease, and peripheral vascular disease according to the Edinburgh Claudication Questionnaire.¹⁰ Patients were excluded from the study if they had a history of malignant or secondary hypertension, congestive cardiac failure, fasting serum triglycerides >4.5 mmol/L, major concomitant noncardiovascular disease, or were taking warfarin (excluding patients in chronic atrial fibrillation). Patients who had had an acute cardiovascular or cerebrovascular event in the preceding 3 months were also excluded. The 10-year cardiovascular (CHD) and stroke (CVA) risk scores were calculated using the Framingham equation.⁹ A repeat blood sample was taken after a median treatment period of 3 years (interquartile range 2.5 to 4.0), using a “package of care” of cardiovascular risk management, including antihypertensive therapy, lipid-lowering therapy, and lifestyle modification (eg, smoking cessation).

A matching control group was gathered comprised of 45 healthy, age- and sex-matched volunteers. They were asymptomatic, had no clinical evidence of vascular, neoplastic, metabolic, or inflammatory disease (assessed by history, examination, and routine laboratory analysis). These subjects were normotensive (supine BP <140/90 mm Hg on repeated readings measured as above), all had normal baseline rest electrocardiograms (ECGs), and were free of drug therapy. Baseline demography is presented in Table 1.

Table 1.

Baseline characteristics

	Controls (n = 45)	Hypertensives (n = 96)	P
Age (y)	60 ± 11	62 ± 8	.275
Male (%)	73	79	.441
SBP (mm Hg)	126 ± 13	178 ± 21	< .0001
DBP (mm Hg)	78 ± 9	94 ± 12	< .0001
Cholesterol (mmol/L)	5.6 ± 0.8	6.1 ± 1.0	.003
Creatinine (mmol/L)	87 (78–96)	101 (92–111)	< .0001
Smoker (%)	0	21.3	< .0001
Diabetes	0	22.3	< .0001
ECG LVH (%)	0	11.7	< .0001
CHD score	—	24.4 (17.1–34.5)
CVA score	—	8.8 (5.6–14.7)
MMP-9 (ng/mL)	70 (56–110)	110 (71–175)	.0041
TIMP-1 (ng/mL)	360 (300–412)	400 (310–490)	.0166

CHD = Framingham coronary heart disease; CVA = Framingham cerebrovascular; DBP = diastolic blood pressure; ECG = electrocardiogram; LVH = left ventricular hypertrophy; MMP = matrix metalloproteinase; SBP = systolic blood pressure; TIMP = tissue inhibitor of metalloproteinase.

Data expressed as mean ± SD, median (interquartile range), or as percentage.

Sampling and laboratory analyses

Patients and controls fasted for a minimum of 8 h and blood samples were collected from the antecubital vein. Samples were separated into aliquots for the preparation of serum and plasma. Serum was analyzed in the hospital laboratory for cholesterol, triglycerides, and HDL cholesterol. Citrated blood was centrifuged at 4°C, and frozen at −70°C until analysis. Plasma was separated, aliquoted, and frozen at −70°C for later analysis of metalloproteinases by ELISA.

The MMP and TIMP assays were validated with in-house methods, to assess total concentration of enzyme or inhibitor in plasma. Briefly, monoclonal primary and biotinylated secondary antihuman antibody for MMP-9 and TIMP-1 were tested against appropriate recombinant human MMP and TIMP (MAB911, BAF911, 911-MP; MAB970, BAF970, 970-TM, respectively; R&D Systems Europe Ltd., Abingdon, UK) and against plasma to determine the optimal concentration of primary (2 μg/mL and 4 μg/mL, respectively) and secondary antibody (0.1 μg/mL and 0.1 μg/mL, respectively) sample dilution (1/100) and standard curve setup (top standards 20 ng/mL). Subsequent lower standards were made by double diluting in the assay buffer (phosphate buffered saline with Tween 20 [Sigma-Aldrich, Dorset, UK]). The mean intra-assay and mean interassay coefficients of variation for the assays were <5% and <10%, respectively. The lower limits of detection for MMP-9 and TIMP-1 were 16 and 16 ng/mL, respectively.

Power calculation and statistical analyses

We hypothesized a minimum difference of 0.5 of a standard deviation in a normally distributed index between any two groups (eg, cases and controls). To achieve this at 2 P < .025 and 1-β = 0.85, we needed reliable data from 44 subjects per group. We therefore recruited consecutively until this was achieved. We also hypothesized that treatment would result in a change of at least 0.25 of a standard deviation in the patients. To achieve this, again at 2 P < .025 and 1-β = 0.85, we needed reliable data from 88 subjects, and therefore, as before, we recruited slightly in excess of this figure for added confidence.

Normality of continuous data was tested by the Anderson Darling Test. The MMP-9 and TIMP-1 levels were not normally distributed, and values are expressed as median and interquartile range. The Mann Whitney test was used to compare nonparametric data between the controls and the pretreatment group. We used the paired Wilcoxon rank-sum test to compare pretreatment and post-treatment levels of circulating MMP-9 and TIMP-1. Pretreatment circulating MMP-9 and TIMP-1 levels were correlated with Framingham CHD and CVA risk scores using the Pearson correlation method (Spearman method, when not normally distributed). Statistical analysis was performed using the MINITAB Statistical Software package (release 13.30, State College, PA). A P value of < .05 was taken as statistically significant.

Results

Hypertensive patients and controls were matched for age and sex. Predictably casual clinic BP, fasting serum total cholesterol, and creatinine were significantly higher in the patients with hypertension (Table 1).

Circulating concentrations of MMP-9 and TIMP-1 at baseline were significantly higher in patients with hypertension than in the normotensive controls (Table 1, Fig. 1). When baseline data on treated and untreated hypertensives were compared, there were no statistically significant differences between the baseline characteristics and circulating MMP-9 and TIMP-1, apart from more diabetics in the untreated versus the treated group (P = .05) (Table 2).

Circulating matrix metalloproteinase (MMP)-9 and tissue inhibitor of metalloproteinase (TIMP)-1 in controls and pretreatment at baseline. Thick black line indicates group median; open circles indicate individual control and patient results.

Figure 1.

Table 2.

Effects of treatment at baseline

	Treated* Hypertensives (n = 84)	Untreated Hypertensives (n = 12)	P
Age (y)	62 ± 8	59 ± 9	.269
Males (%)	80	75	.704
SBP (mm Hg)	177 ± 22	182 ± 11	.228
DBP (mm Hg)	93 ± 12	97 ± 9	.286
Cholesterol (mmol/L)	6.1 ± 1.0	5.9 ± 0.8	.533
Creatinine (mmol/L)	101 (93–112)	91 (83–106)	.055
Smokers (%)	20	27	.605
Diabetes (%)	19	45	.050
ECG LVH (%)	13	0	.199
Statin use (%)	20	0	—
ACE inhibitor use (%)	18	0	—
CHD score	24.4 (17.1–34.4)	29.6 (15.3–42.2)	.561
CVA score	8.8 (4.9–14.7)	11.1 (6.1–25.8)	.434
MMP-9 (ng/mL)	100 (70–174)	115 (106–189)	.170
TIMP-1 (ng/mL)	403 (311–486)	369 (263–453)	.321

ACE = angiotensin-converting enzyme; CHD = Framingham coronary heart disease; CVA = Framingham cerebrovascular; DBP = diastolic blood pressure; ECG = electrocardiogram; LVH = left ventricular hypertrophy; MMP = matrix metalloproteinase; SBP = systolic blood pressure; TIMP = tissue inhibitor of metalloproteinase.

Data expressed as mean ± SD, median (interquartile range), or as percentage.

*

Patients taking antihypertensive drugs had to have a baseline blood pressure of >140/90 mm Hg to be included in the study.

There was a significant decrease in diastolic and systolic BP after treatment for hypertension (Table 3) The percentage of patients who had ECG LVH post-treatment did not change significantly compared to baseline (11.2%). Circulating MMP-9 levels also decreased, and circulating TIMP-1 levels increased after treatment (Table 3).

Table 3.

Effect of treatment

	Pretreatment (n = 96)	Post-treatment (n = 96)	P
SBP (mm Hg)	178 ± 21	139 ± 15	< .0001
DBP (mm Hg)	94 ± 12	78 ± 10	< .0001
MMP-9 (ng/mL)	110 (71–175)	80 (51–150)	.035
TIMP-1 (ng/mL)	400 (310–490)	415 (356–518)	.005

DBP = diastolic blood pressure; MMP = matrix metalloproteinase; SBP = systolic blood pressure; TIMP = tissue inhibitor of metalloproteinase.

Data expressed as mean ± SD or median (interquartile range).

Correlations

There were no statistically significant correlations between MMP-9 or TIMP-1 with age, Sokolow Lyon scores, creatinine, cholesterol, or body mass index in the pretreated hypertensive cohort. However, there was a negative association between measured MMP-9 and HDL cholesterol (r = −0.237, P = .022) and a positive association between MMP-9 and TIMP-1 (r = 0.238, P = .020).

There was a statistically significant correlation between MMP-9 and CHD risk (r = 0.317, P = .007), but not CVA risk (r = 0.166, P = .167). There were no significant correlations between TIMP-1 and CHD (r = 0.045, P = .708) and CVA (r = 0.151, P = .209) scores.

Discussion

The role of circulating MMP-9 in hypertension has not been precisely defined. In the present study, we have shown evidence of elevated circulating levels of MMP-9 at baseline in untreated hypertension compared to normotensive control subjects. In deoxycorticosterone acetate (DOCA)-salt hypertensive rats, myocardial levels of MMP-9 activity have been shown to be increased,¹¹ suggesting that there may be increased ECM turnover in the myocardium. Increased circulating MMP-9 from our patients could therefore be coming from the myocardium. We also show for the first time that circulating MMP-9 levels decreased after high BP is treated, although it is possible that this effect might be the result of aggressive intervention, as there was little change in MMP-9 between treated and untreated hypertensives at baseline (Table 2). This finding is consistent with our hypothesis of a possible reduction in deleterious vascular remodeling, and hence, some prognostic relevance.

Hypertension is a systemic disease with a major impact on the arterial tree, and ECM abnormalities may originate in the vascular tree rather than cardiac changes per se. Although increasing arterial stiffness is well recognized and associated with alterations in the ECM, these are principally focused on an increased deposition of type I collagen at the expense of the other constituents.^12,13 For these changes in the ECM composition to occur, there must be a shift in the turnover of ECM. Collagen type I is not a major substrate for MMP-9,¹ and hence increased tissue activity and concentrations may reflect the breakdown of components other than collagen type I, to facilitate collagen changes mediated by other MMPs. Circulating concentrations of MMP-1 (involved principally in the breakdown of collagen type I) appear to be seen in lower levels in patients with hypertension, and this will favor the maintenance of a stiffer and more fibrous myocardial or arterial ECM.⁶

Cardiac and arterial remodeling in hypertension is regarded as pathologic and links powerfully to the frequency of many of the acute vascular complications of hypertension. For example, electrocardiographic LVH carries a generally worse prognosis of vascular events than mild LVH, only evident on echocardiography.¹⁴

Changes in tissue structure seem likely to require changes in the concentration and activity of a range of matrix enzymes over and above a decrease in collagenase activity. In the case of MMP-9, higher tissue levels would be expected and should be reflected in circulating levels. The measurement of these matrix enzymes could be a valuable tool to assess the state of tissue composition and turnover, and this may link to clinical outcomes and prognosis. For this to be possible the relationship between tissue levels and circulating concentrations has to be established in more detail. Certainly, there are little data available at present to confirm this assumption. One study of MMP-9 has shown a link between circulating MMP-9 levels and tissue concentrations of MMP-9 in atherosclerotic abdominal aortic aneurysms.¹⁵

In our sample, we observed a modest correlation between levels of MMP-9 and calculated Framingham CHD risk scores. This supports a suggestion that MMP-9 levels may be of prognostic value in reflecting the degree of tissue turnover and vascular stiffening associated with vascular complication of hypertension and even normal aging. This is also in keeping with the association of circulating levels of MMP-9 with outcomes in acute coronary syndromes.^16,17 Furthermore, the negative correlation observed between MMP-9 and HDL cholesterol is interesting, especially because the relationship of HDL cholesterol on MMP-9 is uncertain.^18,19 The relationship may be affected by the interaction between lipoproteins, the endothelium, and the matrix composition of vascular subintima or myocardial vessels. Higher levels of HDL cholesterol may be able to downregulate MMP-9 production within the vasculature, and prevent deleterious remodeling and plaque formation or expansion.²⁰

It has been hypothesized that circulating TIMP-1 may be a marker of left ventricular fibrosis or hypertrophy (as defined by echocardiography) and diastolic dysfunction or heart failure.^5,21 A proposed mechanism of this action could be by inhibiting MMP-1 activity (which primarily breaks down collagen type I), therefore increasing tissue collagen type I content. There is indirect evidence from human and animal studies using circulating markers of collagen carboxy-terminal telopeptide of type I collagen (increased levels of which reflect collagen breakdown and have a negative correlation with TIMP-1) and carboxy-terminal propeptide of procollagen type I (increased levels of which reflect collagen type I synthesis, and show a positive correlation with TIMP-1) to support a role for TIMP-1 in favoring reduced collagen type I turnover and increased deposition in vascular tissue.^5,6 This would correlate with a mechanism for vascular stiffness. Furthermore, the effect of antihypertensive treatment on TIMP-1 levels was rather surprising, in that the levels of the circulating marker increased after treatment. This could be explained by the fact that increased levels of TIMP-1 may reflect “positive” remodeling in vascular and target organs (eg, heart and kidney) during treatment, and the inhibition of MMPs, which propagate deleterious remodeling, for example, MMP-9. Furthermore, this effect may be related to intensive BP control as in the case of MMP-9 (vide supra).

Limitations and future work

Tissue sampling in human studies is increasingly problematic. Ideally, correlations should be established between arterial or cardiac tissue activity and circulating MMP-9 and TIMP-1 levels. Animal studies may be required in established models of hypertension or atherosclerosis to look more closely at the links between structure, function, and local as well as circulating MMP and TIMP composition.

The link between cardiovascular risk and MMP-9 is interesting, and this needs to be defined further in terms of a larger study with cardiovascular end point data. The larger sample size would also provide data to account for risk factors in the regression analysis (eg, diabetes, cholesterol). In the present study, our patients received a “package of care” of cardiovascular risk reduction therapy. Dissecting out the benefits of an individual drug is therefore not possible, as all of the patients are “high risk” and required multiple drug intervention. To do this, a study with patients with just isolated hypertension is required. However, in the age group that we dealt with, this is almost impossible as a patient will seldom require just one drug.

In our sample of patients with (relatively uncontrolled) hypertension at baseline, many of the patients did not have ECG LVH or target organ damage, and perhaps similar studies should focus on collecting a more selected and higher risk group of patients with hypertension where overt target organ damage is more advanced. Given that ECG and echocardiographic LVH both carry differing levels of adverse prognosis,²² it may be that the variance in ECG LVH measurement was simply too large to establish any such relationship.

In conclusion, we found that both circulating MMP-9 and TIMP-1 levels were increased at baseline in patients with hypertension. These changes could facilitate an increased deposition and retention of type I collagen at the expense of other components of ECM within the cardiac and vascular ECM. Higher levels of circulating MMP-9 appeared to be associated with higher Framingham cardiovascular risk scores. After treatment of elevated BP, MMP-9 levels decreased and TIMP-1 levels increased. Our observations suggest a possible role for these surrogate markers of tissue ECM composition in the prognosis of cardiovascular events in hypertension.

Acknowledgments

We thank the Pulse Trust and the City Hospital Research & Development Programme for the Haemostasis Thrombosis and Vascular Biology Unit. We also thank Drs. Charles Spencer and Dirk Felmeden for their help with data collection. The ASCOT Investigators are listed in Sever et al.⁸