Aortic stenosis (AS) is the most common valvular heart disease in westernised societies. Severe AS affects up to 3% of individuals over the age of 65, and it is the third most important cardiovascular disease after coronary artery disease (CAD) and hypertension.1 2 AS progresses over several decades during which the valve leaflets undergo extensive fibrocalcific tissue remodelling. For years, AS was believed to be a degenerative disease due to a “wear and tear” process. However, it was recently demonstrated that AS is a disease process akin to atherosclerosis.3 AS has been linked to several risk factors for coronary artery disease or peripheral arterial disease. In addition, histological analyses of explanted valves have revealed that lipoprotein deposition, chronic inflammation, and active calcification play a determinant role in the disease process.4^–6

Over the recent years several studies have highlighted the pivotal role of ox-LDL as activator of inflammation and atherosclerosis development. In atherosclerotic plaques, ox-LDL promotes the recruitment of inflammatory cells and the production of cytokines, which are involved in the tissue remodelling process and disease progression.7 In AS valves, valvular infiltration by ox-LDL has been well documented, and has been shown to colocalise with inflammatory infiltrates and calcific nodules.4 5 These previous studies therefore support the concept that AS development may be, at least in part, influenced by ox-LDL. Recently, several studies have reported an association between high plasma levels of ox-LDL and CAD and have demonstrated its usefulness as an independent risk marker for coronary events.8^–10 We therefore hypothesised that an increased level of circulating ox-LDL enhances the fibrocalcific remodelling of AS valves.

MATERIAL AND METHODS

Patients and tissue collection

We examined 105 AS valves that were explanted from patients at the time of aortic valve replacement for a stenosis. Valves with an aortic regurgitation grade >2+ and valves of patients with a history of rheumatic disease, endocarditis or an inflammatory disease were excluded. All patients underwent a comprehensive Doppler echocardiographic examination preoperatively. Doppler echocardiographic measurements included the left ventricular stroke volume and transvalvular gradients using modified Bernoulli equation. Patients with reduced left ventricular ejection fraction (LVEF) (<40%) were excluded. Two segments of the aortic valve were partially decalcified in Cal-Ex (Fisher, Nepean, Ont, Can) for 24 h and then one segment was fixed in formaldehyde 10% for histological processing and the other one was embedded in optimum cutting temperature (OCT) compound (TissueTek, Miles Laboratories, Elkhart, Indiana, USA) and frozen in liquid nitrogen for immunohistochemical analyses.

Histological analysis and tissue remodelling score

Decalcified and fixed tissues were processed for paraffin embedding. Valve samples were excised vertically to the base at the midpoint. 5 μm-thick sections were obtained and stained with haematoxylin-eosin (H&E). Histological sections were analysed and the degree of valvular tissue remodelling and calcification was assessed using a scoring system adapted and modified from Warren et al.11 Score 1: mild fibrous thickening, structural integrity of the cusps is maintained; Score 2: moderate valve thickening and early nodular calcification with preservation of the fibrosa; Score 3: severe thickening with many calcified nodules and a distorted fibrosa; Score 4: severe thickening and distortion with many calcified nodules, important fibrosis and destruction of most structural components with disruption of elastic tissue. The remodelling score was attributed by an experienced cardiovascular pathologist (CC) blinded to clinical and echocardiographic data.

Determination of valvular calcium concentration

A segment of valve tissue was kept in liquid nitrogen until determination of the calcium content. Leaflets were homogenised and treated with HCl 6N at 95°C for 24 h. Then treated tissues were centrifuged at 4400 rpm for 30 min and supernatants were collected. Calcium content was determined by the o-cresolphthalein complexone method. Results were expressed as mg of calcium per wet weight of tissue (Ca mg/g ww).

Lipid profile

Overnight fasting plasma was collected and immediately processed by the laboratory for the measurement of glucose, total cholesterol, low-density cholesterol (LDL-C), high-density cholesterol (HDL-C), and triglyceride levels. After centrifugation, one plasma specimen was kept and stored at −80°C until measurement of C-reactive protein (CRP), LDL particle size, HDL particle size and circulating levels of ox-LDL. Methods used to measure CRP, LDL and HDL particle size have been detailed previously.12 13 Briefly, LDL particle size was determined using non-denaturing 2–16% polyacrylamide gradient gel electrophoresis. LDL peak particle size corresponded to the estimated diameter of the major peak in each patient. The relative proportion of small LDL particles (termed %LDL_<255Å) was determined by computing the relative area of the densitometric scan <255Å. Cholesterol levels in small LDL particles (termed LDL-C _<255Å) were estimated by multiplying the total LDL-C levels by the relative proportion of %LDL_<255Å. HDL particle size was also determined by non-denaturing polyacrylamide gel electrophoresis (4–30%). HDL size was extrapolated from the relative migration of eight standards of known diameters. Plasma apoA-I and apoB concentrations were measured by nephelometry as previously described.14

Measurement of circulating oxidised LDL

Plasma ox-LDL was measured by sandwich ELISA with the monoclonal antibody 4E6 (Mercodia, Uppsala, Sweden) directed against the modified apoB-100 of ox-LDL. The test was conducted according to the manufacturer’s instructions and optical density was read at 450 nm. Results were expressed as units per litre (U/l).

Statistical analysis

Continuous data were expressed as mean (SEM) and compared using a Wilcoxon test to examine the effect of group. A Tukey’s test was used for pairwise comparisons. Categorical data were expressed as a percentage and compared using the χ² test. Correlation analyses were determined using Spearman’s coefficients. We used a multiple linear regression analysis to identify the variables independently associated with the transvalvular gradients and with the plasma ox-LDL level and a multiple logistic regression analysis to identify those associated with the valve remodelling score. Relevant variables were first tested on univariate analysis and those with a p value <0.05 were then entered in the multivariate models. Statistical analysis was performed with a commercially available software package (JMP IN 5.1).

RESULTS

Demographic data and clinical risk factors

The clinical data and risk factors are presented in table 1. In the whole cohort of patients (n = 105), there was a predominance of male patients (68%) with a mean age of 70 (SEM 2) years. The average peak transvalvular gradient was 71.6 (SEM 2.1) mm Hg (range 21–124 mm Hg), and the average aortic valve area (AVA) was 0.77 (SEM 0.02) cm² (range 0.4–1.6 cm2). The stenosis was severe (AVA ⩽1.0 cm²) in 90% of the patients. The predominant risk factors were hypertension (61%) and diabetes (23%). A high proportion of patients were on statin treatment (69%). In this series with predominantly severe AS, all patients had a score of at least 2, with the following distribution: score 2 (n = 29; 28 %), score 3 (n = 38; 36%), and score 4 (n = 38; 36%). More than half of the population (56%) had an associated coronary artery disease (CAD), which was equally distributed among the different valve remodelling scores. Patients with a grade 3 remodelling score had a significantly higher prevalence of hypertension (p = 0.03), whereas the distribution of diabetes was similar among the different groups. Although not significant, there was a trend for a higher proportion of men in patients with higher remodelling score (p = 0.05).

Relationship between valve tissue remodelling and stenosis haemodynamic severity

The valve remodelling score correlated with the valvular calcium content (r = 0.32; p = 0.001). The average calcium content was 45.6 mg/g ww, 58.4 (SEM 5.8) mg/g ww, and 76.6 (SEM 6.0) mg/g ww in valves with aortic remodelling score 2, 3, and 4, respectively (p = 0.003) (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Aortic valve calcium content in relation to the valve remodelling score.*p<0.05 compared with remodelling score of 2.

Patients with a valve remodelling score of 4 had higher peak and mean transvalvular gradients and lower aortic valve area than those with a lower score (table 1). The valve remodelling score correlated with the peak (r = 0.36; p<0.001) and mean (r = 0.32; p = 0.001) gradients and with the aortic valve area (r = −0.24; p = 0.04). The aortic valve calcium content also correlated, but to a lesser extent, with the mean gradient (r = 0.20; p = 0.04) and AVA (r = −0.22; p = 0.01). There was no significant correlation between calcium content and peak gradient (r = 0.15; p = 0.09) (table 2). After correction for age and gender, we found that only the aortic remodelling score was significantly associated with the peak aortic transvalvular gradient (p<0.001) and the mean aortic transvalvular gradient (p<0.001) (table 3).

Relationship between plasma lipid profile and valve remodelling score

Patients with a grade 2 valve remodelling score had higher HDL-C than those with a higher score (p = 0.03). The level of apoB had a tendency to be higher in valves with the highest remodelling score (p = 0.07), whereas levels of LDL-C were similar among the three groups. The size of HDL and LDL particles, including the proportion of small size LDL subfraction (<255 Å), were also similar among groups of remodelling score. In addition, the levels of CRP were also similar between groups. However, plasma ox-LDL levels were significantly elevated in patients with the highest valve remodelling scores (score 2: 27.3 (SEM 2.6) U/l; score 3: 32.2 (SEM 2.3) U/l; score 4: 38.3 (SEM 2.3) U/l; p = 0.02). After adjustment for age, gender, hypertension, and HDL-C, plasma ox-LDL level remained significantly associated with the aortic remodelling score (p<0.001) (table 4). There was no significant association between plasma ox-LDL level and transvalvular gradients or AVA.

Factors associated with the ox-LDL plasma level

Ox-LDL levels were similar in patients with and without coronary artery disease (CAD). The mean ox-LDL value was 33.6 (SEM 1.8) U/l and 35.8 (SEM 1.9) U/l respectively for patients with and without angiographically proven CAD (p = 0.40). Circulating ox-LDL level was not affected by gender (p = 0.99), diabetes (p = 0.75) or hypertension (p = 0.48). Patients who were active smokers had a tendency towards having higher plasma ox-LDL levels: 43.9 (SEM 6.2) U/l vs 33.9 (SEM 1.3) U/l (p = 0.08). Patients on statin treatment had significantly lower plasma ox-LDL levels: 33.2 (SEM 1.6) U/l vs 39.5 (SEM 3.13) U/l (p = 0.04). Circulating ox-LDL level also correlated significantly with the following plasma variables: apoB (r = 0.59; p<0.001), LDL-C (r = 0.40; p<0.001), triglyceride (r = 0.39; p<0.001), LDL-C_<255Å (r = 0.22; p = 0.02) and apoA-I (r = 0.23; p = 0.01). Ox-LDL levels were not associated with CRP, waist circumference or BMI. After adjustment for covariables in a model explaining 36% of the variation of circulating ox-LDL, we found that apoB (p<0.001) and triglyceride levels (p = 0.01) were the only variables significantly associated with plasma ox-LDL level. In a second model where smoking status and statin treatment were introduced into the model (r² adjusted = 0.38; p<0.001), apoB (p = 0.002), triglyceride levels (p = 0.02) and smoking status (p = 0.01) were significantly associated with plasma ox-LDL levels, while statin treatment (p = 0.39) was not (table 5).

DISCUSSION

The main finding of this study is that increased plasma ox-LDL is independently associated with more advanced aortic valve remodelling.

Aortic valve remodelling

AS development involves complex interactions between extracellular matrix, valve interstitial cells (ICs) and inflammatory infiltrates. Calcification of vascular cells and valve ICs has been ascribed to an active process analogous to the ossification process.15 16 Previous studies have documented the presence of bone-associated proteins along with extensive tissue remodelling in explanted AS valves.6 17 18

Valve thickening and calcification are the main factors leading to reduced leaflet mobility and obstruction of the valve orifice in AS. Recently, Roberts et al19 reported that the weight of the explanted aortic valve correlates with the peak transvalvular gradient. In the present study, we found that a histological score of tissue remodelling, which takes into account presence and extent of calcification as well as the modifications of the extracellular matrix, correlates significantly with transvalvular gradients and AVA. Moreover, the valve remodelling score correlates better than the valve calcium content with the indices of stenosis haemodynamic severity. These observations suggest that, beyond the production of calcium, remodelling of the extracellular matrix may contribute to valve cusp stiffening.

In a recent investigation, Messika-Zeitoun et al20 have shown that there is an excellent agreement between aortic valve calcification score measured by electron-beam computed tomography (EBCT) and direct biochemical measurement of calcium in explanted valves. Moreover, the calcification measured by EBCT in patients with AS as well as in normal subjects correlates well with both peak transvalvular gradient and AVA measured by Doppler echocardiography. In the present study, we also found a significant correlation between the amount of valvular calcium and AVA. However, the correlation (r = −0.22) was much weaker than that observed in the study of Messika-Zeitoun et al (r = −0.79). This may be explained by the fact that the relationship between valve calcium content and AVA is curvilinear and becomes relatively flat for values of AVA <1.0 cm². Ninety per cent of the patients in our series had an AVA <1.0 cm², whereas in the series of Messika-Zeitoun et al the AVA ranged from 4.0 to 0.5 cm². When analysed collectively, these findings support the concept that calcification is probably a predominant determinant of AS progression in the earlier stage of the disease, whereas in the more advanced stage of the disease extracellular remodelling with progressive fibrosis may make, in addition to calcium deposition, an important contribution to disease progression.

Factors associated with aortic valve remodelling

Lipoprotein oxidation is thought to play a central role in the development of atherosclerosis.21 Indeed, ox-LDLs are potent inductors of inflammation and are actively involved in the formation of foam cells. Similarly to vascular atherosclerosis, calcification of the aortic valve has been associated with tissue infiltration by oxidatively modified lipoproteins.4 5 Elevated circulating levels of ox-LDL have been linked to CAD and subclinical atherosclerosis.8 10 However, one recent study reported that circulating ox-LDL level is not an independent predictor of CAD.22

The present study is the first one to investigate the relationship between the plasma level of ox-LDL and calcific aortic valve disease. Among different plasma variables analysed in this study, only circulating ox-LDL was independently associated with the aortic valvular remodelling process. In healthy men high plasma ox-LDL levels predicted the development of cardiovascular events.23 Recently, it has been reported that elevated plasma ox-LDL level is associated with activation of inflammation in circulating leucocytes through the NF-κB pathway, suggesting that the circulating fraction of ox-LDL is biologically active.9 In line with this observation, previous investigations have found a positive association between circulating ox-LDL and CRP levels, suggesting that ox-LDL are possibly involved in the activation of immune-mediated reactions.8 However, in the present study CRP levels were not associated either with ox-LDL levels or with the aortic remodelling process. Nevertheless, we cannot exclude the possibility that plasma ox-LDL is associated with other inflammatory pathways in patients with AS.

Clinical implications

This study, as well as other previous studies, has linked plasma ox-LDL levels with some features of the metabolic syndrome.24^–26 In the present study, we have observed that apoB and triglyceride levels are independently associated with plasma ox-LDL. Interestingly, we have recently reported that AS progression rate was faster in patients with the metabolic syndrome.27 Although a causal relationship cannot be confirmed from this study, the level of circulating ox-LDL may, at least in part, represent a significant metabolic risk marker of an active aortic valve remodelling process. Nonetheless, it is also possible that plasma ox-LDL represents a biologically active factor which activates cellular receptors. Indeed, the lectin-like ox-LDL receptor-1 (LOX-1) is expressed by monocytes and the endothelium.9 Thus, through LOX-1 receptors circulating ox-LDL would have the ability to activate valvular endothelium, and thereby contributing to enhanced valve inflammation and possibly the remodelling process. Therefore, although still unproven, it is possible that specific behavioural or pharmacological interventions targeting plasma ox-LDL levels would contribute to decrease aortic valve remodelling and thereby slow the progression of AS. In this regard, previous studies have demonstrated that statin therapy may significantly reduce plasma ox-LDL levels.28 Although the role of statins in patients with AS is not clearly established,29 30 the effect of a lipid lowering therapy on the plasma ox-LDL level and its relation with AS progression rate remains to be studied. In the present study, after correction for covariates, statin therapy was not independently associated with lower plasma ox-LDL levels, whereas smoking was associated with a significant increase in circulating ox-LDL. Thus, it remains to be determined whether appropriate lifestyle interventions and/or pharmacological treatment leading to reduction in plasma ox-LDL level would be beneficial in patients with AS.

Limitations

This cross-sectional study has some inherent limitations in so far as it included patients with an advanced pathological process necessitating an aortic valve replacement. In order to have a more complete assessment of the mechanisms implicated in the fibrocalcific remodelling of the aortic valve, it would have been necessary to perform serial evaluations of the histological architecture of the valve at different stages of the disease, which was not possible in this study. Thus, histopathological changes that have occurred from a prepathological state, referred to as remodelling in this study, imply that gradual changes took place for years. In light of this inherent limitation, the results and conclusions of this study should be restricted to end-stage AS and cannot be directly extrapolated to the whole spectrum of the disease process.

Also, clinical data were collected at the time of the operation and some important data were missing, such as the duration of statin treatment prior to aortic valve replacement. This lack of information may have limited our ability to detect a potential significant effect of lipid-lowering therapy on either plasma ox-LDL levels or the valve remodelling process.

CONCLUSION

There is now strong evidence supporting the concept that AS development and progression is an atherosclerotic-like process. Precise mechanisms by which AS develops and progresses, however, remain to be elucidated. The present study is the first to report that plasma level of oxidised LDL is independently associated with aortic valve remodelling in patients with severe AS. Further studies are needed to determine whether circulating ox-LDL is a risk marker for a highly atherogenic profile and/or a circulating molecule which is actively involved in the pathogenesis of calcific aortic valve disease.