Introduction

Thrombin acts as a pivotal enzyme for generating fibrin clots and activates platelets, vascular endothelial cells, vascular smooth muscle cells, macrophages and fibroblasts to enhance procoagulation,1 chemoattraction,2 mitogenesis3 and proliferation4 of these cells. Thrombin exerts its physiological and pathological actions in these cells through the proteolytic processing of specific cell-surface receptors known as protease-activated receptors (PARs).5, 6 Nelken et al.7 demonstrated that PAR-1 was widely expressed in regions where macrophages, cardiomyocytes, cardiac fibroblasts, vascular smooth muscle cells and mesenchymal-appearing intimal cells are abundantly present. Thus, excessive PAR-1 activation promotes cardiovascular disorders, including cardiac remodeling, arterial thrombosis and atherosclerosis. From the point of view of PAR-1 modulation in cardiovascular remodeling, the use of antithrombin agents and antagonists of the thrombin receptor might be useful approaches for the treatment and prevention of these disorders. Antithrombin (AT), a major serine protease inhibitor (serpin), inhibits thrombin action at the intravascular lumen and it exerts its optimal AT actions by binding to heparan sulfate and heparan sulfate proteoglycans on the luminal surface of endothelial cells. In addition, heparin cofactor II (HCII), which is also a serpin, is a plasma glycoprotein with a molecular weight of 65.6 kDa that is synthesized by the liver and circulates in plasma at a concentration of 1.0 μmol l−1. HCII potently inhibits thrombin action by forming a bimolecular complex with dermatan sulfate proteoglycans under the endothelial layer in mammalians.8, 9 Although AT inhibits not only the thrombin action, but also the actions of several proteases involved in blood coagulation or fibrinolysis, HCII only inactivates thrombin and has no inhibitory effect on the action of any other proteases. As HCII can counteract the actions of thrombin at injured vascular walls, we, and others, have investigated and confirmed the protective role of HCII against atherosclerosis in clinical examinations and studies using HCII-deficient mice.9, 10, 11, 12, 13, 14, 15

As the development of vascular remodeling, including atherosclerosis, has been shown to be closely associated with cardiac remodeling in humans and experimental animal models, we hypothesized that HCII is involved in the process of not only atherosclerosis, but also cardiac remodeling. In order to clarify this issue, we investigated the relationships between plasma HCII activity and surrogate markers with respect to cardiac remodeling in elderly subjects with cardiovascular risk factors. We found that plasma HCII activity is inversely associated with cardiac remodeling in subjects without systolic heart failure (HF).

Methods

Subjects for cross-sectional study

We consecutively recruited 304 Japanese subjects (169 males and 135 females) who were outpatients with lifestyle-related diseases and subjects older than 35 years of age were recruited consecutively from the Department of Medicine and Bioregulatory Sciences and Department of Cardiovascular Medicine at Tokushima University Hospital, Tokushima, Japan between April 2007 and September 2009. All subjects underwent a standardized interview and a physical examination. Current smokers were defined as subjects who had smoked within the past year. Body mass index was calculated as an index of obesity. Blood pressure was measured twice and averaged. Hypertensive patients were defined as those with systolic blood pressure (SBP) ⩾140 mm Hg and/or diastolic blood pressure (DBP) ⩾90 mm Hg or those receiving antihypertensive agents. Pulse pressure was calculated as SBP−DBP. Patients who were diagnosed with white coat hypertension were not categorized as having hypertension. Hyperlipidemic patients were defined as those with low-density lipoprotein cholesterol ⩾140 mg −1dl and/or triglyceride level ⩾150 mg dl−1 or those receiving lipid-lowering agents. Patients were classified as diabetics by their use of insulin and/or oral hypoglycemic agents or by glycosylated hemoglobin A1c (HbA1c) >6.5%. In this study, the criteria for cardiovascular risk factor(s) included current smoking, hypertension, hyperlipidemia and diabetes mellitus. The exclusion criteria included subjects with overt left ventricular systolic dysfunction (ejection fraction <50%), history of previous episodes of congestive HF, moderate to severe valvular disease and atrial fibrillation, known malignancy, renal failure, liver dysfunction and malnutrition. Our study followed the institutional guidelines of the University of Tokushima and was approved by the Institutional Review Board. Prior informed consent was obtained from all patients according to the Declaration of Helsinki.

Biochemical analyses

Before noon, overnight fasting blood samples were collected from the antecubital vein and were assayed immediately for HbA1c and serum lipid parameters, including low-density lipoprotein cholesterol, high-density lipoprotein cholesterol and triglyceride level. Serum levels of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol and triglyceride level were measured using the enzymatic method. HbA1c was measured using high-performance liquid chromatography.

Measurements of plasma HCII and AT activities

Blood was drawn as described above, collected into a tube containing 1/10 volume of 3.8% sodium citrate and centrifuged at 2000 × g for 20 min. Plasma was stored at −80 °C until use. Plasma HCII and AT activities were measured as previously described.16

Echocardiography

The ultrasound instrument used in this study was a Toshiba Aplio 80 with a 2.5-MHz transducer (Toshiba Medical Corporation, Tokyo, Japan). Left atrial size was quantified as left atrial volume (LAV) because LAV is a more accurate estimate of left atrial size than M-mode or 2D left atrial diameters and is a better predictor of cardiovascular events.17, 18, 19 Left atrial (LA) dimensions were measured in three orthogonal planes: parasternal long axis, lateral and superoinferior (Figure 1).20 Those dimensions were recorded at the end phase of ventricular systole. As suggested by current guidelines,19 LAV was calculated by using the length diameter ellipsoid method computed at ventricular end-systole by the following equation: V=4π/3 × (parasternal long axis/2) × (lateral/2) × (superoinferior/2). Finally, left atrial volume index (LAVI) was indexed by body surface area. Left ventricular mass index (LVMI) was estimated using Devereux formula21 and was calculated as an index of body surface area. Relative wall thickness (RWT) was calculated as (septal wall thickness+posterior wall thickness in the end-diastolic phase)/(left ventricular end−diastolic diameter). Doppler echocardiographic assessment, including measurements of peak velocities of E and A waves and deceleration time (DcT), was carried out in all patients who underwent tissue Doppler imaging. Spectral pulsed wave Doppler tissue interrogation of longitudinal mitral annular velocity was recorded throughout the cardiac cycle at the septal annulus in the apical four-chamber view. The ratio of peak E velocity to early diastolic mitral annulus velocity (E/e' ratio) was calculated. Other routine echocardiographic examinations, including measurements of left ventricular fractional shortening (FS%) and ejection fraction, were also performed.

Figure 1
figure 1

Representative images of measurement of LAV. PLAX was taken in the parasternal long-axis view. The LAT and SI dimensions were both taken from the apical four-chamber view using the inner edge-to-inner edge measurement. LAV was calculated by using the length diameter ellipsoid method, applying the following equation: V=4π/3 × (PLAX/2) × (LAT/2) × (SI/2). LAV, left atrial volume, PLAX, parasternal long axis, LAT, lateral, SI, superoinferior.

Full size image

Statistical analysis

Statistical analyses were performed using the StatView statistical package (StatView 5.0; SAS Institute, Japan Ltd., Tokyo, Japan). Continuous variables were averaged and values were expressed as the mean±s.d. or as a percentage for categorical parameters. Male gender and the presence of hypertension, diabetes mellitus, hyperlipidemia and current smoking status were coded as dummy variables. The degrees of association between independent variables including sex, age, body mass index, SBP, serum lipid parameters, HbA1c, plasma AT and HCII activities, history of current smoking, hypertension, diabetes mellitus and hyperlipidemia were determined by means of multiple regression analysis.

Results

Characteristics of subjects

The physical and laboratory characteristics of the subjects enrolled in this study are shown in Table 1. The high-density lipoprotein cholesterol levels and plasma AT activity were higher in females than in males. On the other hand, males showed higher levels of serum creatinine and more were current smokers. The mean plasma HCII activity in all of the participants was 95.8±17.0% and there was no difference between the mean plasma HCII activity in males and females. In addition, there were no significant gender differences in the age and body mass index, SBP, pulse pressure, low-density lipoprotein cholesterol, triglyceride level and HbA1c values. No significant gender differences were observed between the prevalences of hypertension, diabetes mellitus and hyperlipidemia and the use of medications for cardiovascular treatment in males and females.

Table 1 Clinical characteristics of subjects and echocardiographic measurements
Full size table

Echocardiographic measurements

Table 1 also shows the results of the echocardiographic examinations. Although the left atrial dimension index (LADI) in females was significantly larger than that in males, there was no statistically significant gender difference in LAVI. Males manifested higher LVMI values than females. In contrast, the left ventricular systolic functions indicated by FS% and ejection fraction were both slightly higher in females than in males. Although there was no gender difference in the value of E/A, DcT and E/e' ratio were significantly larger in females than in males.

HCII is an independent and negative determinant of left atrial size

Age and sex-adjusted scatter plots between HCII and LADI and between HCII and LAVI indicated significant linear associations (Figure 2). Multiple regression analysis showed that age was a positive contributor for an increase in LADI and LAVI (Table 2). Conversely, HCII was found to be an independent and negative contributor for an increase in LADI and LAVI (Table 2). The significance of the relationship between HCII and LA size markers was more accurate in LAVI than in LADI (Figure 2 and Table 2).

Figure 2
figure 2

Age and sex-adjusted scatter plots between plasma HCII activity and cardiac geometrical and functional parameters. LADI, left atrial dimension index, LAVI, left atrial volume index, RWT, relative wall thickness, LVMI, left ventricular mass index, DcT, deceleration time.

Full size image
Table 2 Multiple regression analysis for determinants of left atrial size
Full size table

Plasma HCII activity is inversely associated with concentric left ventricular remodeling

As increased RWT was recognized as a concentric change of the cardiac left ventricle, we evaluated the independent determinants for increasing RWT using simple and multiple regression analyses. Age and sex-adjusted scatter plots between HCII and RWT demonstrated a significant linear relationship as shown in Figure 2. Although age and body mass index were independent contributors for an increase in RWT, HCII was the sole negative contributor for an increase in RWT (Table 3). Although male gender, pulse pressure and the presence of hypertension were independent contributors for increase in LVMI, plasma HCII activity did not have any significant association with LVMI (Figure 2 and Table 3). Taken together, these results suggest that reduced plasma HCII activity is associated with the phenotype of cardiac concentric remodeling without increased LVMI.

Table 3 Multiple regression analysis for determinants of RWT and LVMI
Full size table

Plasma HCII activity is inversely associated with left ventricular diastolic dysfunction

As not only left ventricular systolic dysfunction, but also left ventricular diastolic dysfunction were shown to be closely associated with all-cause mortality in the general population,22 we evaluated the relationship between plasma HCII activity and left ventricular diastolic function. Although age, SBP and the presence of hypertension were independent contributors for decrease in E/A ratio, plasma HCII activity did not have any significant association with E/A ratio (Figure2 and Table 4). On the other hand, significant linear associations were found in age and sex-adjusted scatter plots between HCII and DcT, and between HCII and E/e' value as shown in Figure 2. We then evaluated the independent determinants of DcT and E/e' ratio and found that HCII was an independent and negative contributor for increases in DcT and E/e' value (Table 4). These results indicated that HCII is the sole protective factor against left ventricular diastolic dysfunction.

Table 4 Multiple regression analysis for determinants of E/A ratio, DcT and E/e' ratio
Full size table

Discussion

Enlarged LAD reflects left atrial pressure and volume overload in response to cardiac dysfunction associated with cardiovascular diseases, including atrial fibrillation.23, 24 There is accumulating evidence that the association between cardiovascular disease and LAV/body surface area (LAVI) is stronger than that observed with LAD/body surface area (LADI) after adjustment for age and gender in subjects with sinus rhythm.17, 23, 24, 25 Therefore, we assessed the relationships between plasma HCII activity and the LA size parameters. We found a significant inverse correlation between plasma HCII activity and LAVI, as well as LADI even after adjustment for other confounding cardiovascular risk factors, suggesting that HCII independently counteracts LA enlargement. Enlarged LAV has been shown to be associated with inflammation and atherosclerosis.26, 27 As cuff-tube placement around the femoral artery caused increases in the gene expression levels of inflammatory cytokines and chemokines, such as interleukin-1β and -6 and monocyte chemoattractant protein-1 in HCII-deficient mice compared with the levels in wild-type mice,14 there is a possibility that HCII prevents LA enlargement partly through its anti-inflammatory potency leading to attenuation of left atrial remodeling. As LAV is an indicator of the burden of diastolic dysfunction, even in patients without atrial fibrillation or significant valvular heart disease,28 our results indicated that plasma HCII activity could be involved in the pathogenesis of left ventricular diastolic dysfunction. Therefore, we estimated association between plasma HCII activity and left ventricular (LV) diastolic function in humans and observed the significant relationship between plasma HCII activity and ventricular diastolic dysfunction indicated by the DcT and E/e' ratio. As it is well known that prolonged DcT indicates abnormal LV relaxation in patients with early diastolic abnormality,29 a significant inverse relationship between plasma HCII activity and DcT suggests that HCII has potency to counteract abnormal LV relaxation in subjects without systolic LV dysfunction. Diastolic tissue Doppler velocities reflect myocardial relaxation and, in combination with conventional Doppler measurements, the ratios (transmitral early diastolic velocity/mitral annular early diastolic velocity (E/e' ratio)) have been used to non-invasively estimate LV filling pressure, as well as pulmonary capillary wedge pressure.30, 31, 32 As pulmonary capillary wedge pressure is a prognostic indicator in patients with HF, E/e' value is a similarly powerful predictor of prognosis in patients with various cardiac diseases.33 In the present study, plasma HCII activity was independently and inversely associated with the value of E/e', suggesting that HCII preserves compliance of the left ventricular wall.

Warfarin is a synthetic thrombin inhibitor and is frequently used for the prevention of cardiogenic thrombosis. AT is also an endogenous thrombin inhibitor in the intravascular lumen. As warfarin has never been documented to have a pharmacological effect on anticardiac remodeling and as the present study has shown that AT was not associated with any cardiac remodeling phenotype, thrombin inactivation at the intravascular lumen and/or intracardiac chamber might be unable to attenuate cardiac remodeling. Conversely, as HCII is thought to inhibit thrombin action by forming a bimolecular complex with dermatan sulfate proteoglycans that are deposited at vascular smooth muscle cells and fibroblasts, HCII may also exert tissue thrombin inactivation in cardiomyocytes and/or cardiac fibroblasts in concert with binding to the deposited dermatan sulfate.

PAR-1 is expressed in the heart by cardiomyocytes and cardiac fibroblasts34, 35 and a recent study demonstrated that PAR-1 expression was increased in the hearts of patients with ischemic and idiopathic dilated cardiomyopathy.36 PAR-1 expression is increased in the LV of a mouse model for chronic HF37 and Pawlinski et al.38 demonstrated that PAR-1 overexpression by cardiomyocytes induced cardiac hypertrophy in MHC-PAR-1 mice. These hypertrophic changes of cardiomyocytes by PAR-1 activation may be partly explained by the mechanism of cleavage of PAR-1 resulting in activation of Gq, G12/13 and Gi, as well as downstream signaling pathways, including the MAPK pathways ERK 1/2 and ERK5.39, 40 Therefore, activation of PAR-1 in the heart promotes hypertrophic growth and/or influences the survival of cardiomyocytes.34 These findings are consistent with the assumption that PAR-1 activation in the heart, including cardiomyocytes and cardiac fibroblasts, accelerates cardiac remodeling, leading to reduced elasticity of the left ventricular wall. From these previous observations, we hypothesized that HCII counteracts cardiac remodeling through inactivation of the tissue thrombin–PAR-1 pathway. As there is a possibility that activation and expression of (pro) thrombin and its major receptor PAR-1 axis is modulated according to the condition of cardiac stress in each subject, it is difficult to clarify the exact interplay between HCII and PAR-1 activation in such subjects. However, our results suggest that reduced plasma HCII activity is one of several major causes of LA enlargement and LV diastolic dysfunction regardless of PAR-1 expression levels.

The results of the present clinical study were further corroborated by a study using our HCII-deficient mice.41 In that study, infusion of angiotensin II prominently accelerated cardiac remodeling, including concentric LV changes, enlargement of LAV and exaggeration of cardiac fibrosis in HCII-deficient mice when compared with littermate wild-type mice.41 The study using HCII-deficient mice demonstrated that HCII protects against angiotensin II-induced cardiac remodeling through the suppression of the NAD(P)H oxidase–TGF-β1 pathway.41 Therefore, we speculate that HCII is also capable of attenuating oxidative stress in the human heart, as well as in the murine heart through suppression of the NAD(P)H oxidase–TGF-β1 pathway.

As LV hypertrophy causes LA enlargement and the LA enlargement may impair LV function, it is important to clarify whether HCII primarily influences LA or LV remodeling. Although further examinations are needed to clarify this issue, results of our animal studies using HCII-deficient mice indicate that abnormal changes, including angiotensin II-induced LV concentric change and LA enlargement with acceleration of cardiac fibrosis, seem to occur simultaneously after starting angiotensin II infusion (data not shown in reference 41), and it is therefore possible that HCII directly and independently influences LA enlargement and LV dysfunction.

The results of the present clinical study cannot be extended to the general population because we only enrolled patients with cardiovascular risk factors and we previously reported that subjects without cardiovascular risk factors had higher levels of plasma HCII activity than those in subjects with one or more cardiovascular risk factor.11 Thus, large-scale investigations and cohort studies are required to assess and clarify the prognostic value of plasma HCII activity for cardiac remodeling in the general population. In addition, it is crucial to compare plasma HCII activities in subjects with and without LV systolic dysfunction for understanding the pathophysiological roles of HCII in cardiac remodeling. Therefore, further examinations focusing on the relationship between HCII and systolic HF are needed. In summary, plasma HCII activity is independently and inversely associated with the development of cardiac remodeling including concentric cardiac changes, LA enlargement and LV diastolic dysfunction. These results suggest that the inactivation of thrombin in cardiac tissue by HCII might be a novel and valuable therapeutic approach to prevent cardiac remodeling and atherosclerosis.