METHODS

We studied 111 patients (85 male, 26 female, mean (SD) age 61.6 (10) years) attending for elective day case coronary angiography. All patients had a history of stable angina. We excluded patients with recent myocardial infarction (within the last six weeks), unstable angina, stroke, or congestive heart failure, as well as those with renal or liver impairment or on warfarin treatment. Following the procedure all angiograms were analysed for disease severity (see below). The West Birmingham ethics committee passed the protocol, and informed consent was obtained.

Baseline results were compared with 34 age and sex matched healthy controls (24 male, 10 female, mean age 59.4 (12) years), recruited from hospital staff and preoperative clinics for minor procedures, including hernia repairs, cataract surgery, and so on. All healthy controls were “healthy” on the basis of careful clinical history and examination, as well as basic blood screening tests. These subjects were included to provide a perspective (that is, what should be “healthy” values) for the patient data; no direct case–control comparison is intended.

Blood samples and analysis

Citrated plasma samples were taken before angiography and immediately placed in ice before being centrifuged at 1000 g for 20 minutes at 4°C. They were then stored at −70°C until the time of analysis. Samples were analysed by an in-house ELISA for VEGF and sFlt-1 (R&D Systems, Abingdon, UK) (previously described in detail⁷), vWf (Dako, Ely, UK), and TF (Axis-Shield, Dundee, UK). The lower limits of detection by ELISA were 10 pg/ml for VEGF and TF, 0.1 ng/ml for sFlt, and 2 IU/dl for vWf. The interassay and intra-assay variabilities were < 5% and 10%, respectively, for all assays.

Coronary angiography and analysis

Coronary angiography was undertaken by the percutaneous transfemoral approach and images were recorded digitally. All angiograms were analysed by a single experienced cardiologist who was blinded to the clinical details of the patients. Disease severity was assessed in two ways: first, by the number of vessels (0–3) with at least one significant stenosis (>50% stenosis); and second, by an atheroma score (the coronary atheroma score, as previously described¹⁷) to indicate disease severity. In brief, 15 proximal segments of the major coronary arteries were examined, as described by an ad hoc committee on grading of coronary artery disease of the Council on Cardiovascular Surgery, American Heart Association.¹⁸ Atheromatous lesions in each segment were evaluated for extension (number of plaques) and size (% of vessel diameter involved). The scores for extension and size were multiplied together for each segment and the total score was calculated as the sum of individual segment scores divided by the number of segments analysed. Segments distal to a total occlusion were not analysed. Thus patients could be classed as having “0 vessels diseased” with no stenoses of >50%, but still have an atheroma score above 0, as the coronary atheroma score includes lesions causing less than 50% stenosis in the analysis.

Power calculations

Our primary hypothesis was a significant correlation between the coronary atheroma score (a continuously variable index) and any one of the plasma markers. Given the variability of the measured indices and the coronary atheroma score, we believe that a Spearman correlation coefficient of 0.35 is meaningful. To achieve this, for a two sided probability of < 0.01 and 1−β of 0.85, 100 datasets are required.¹⁹ Thus to be fully confident in our data we recruited in excess of this figure. The number of diseased arteries is distributed as a discrete but ordered dataset. We hypothesised increasing differences in vWf, VEGF, and TF, but decreasing sFlt-1, as disease became more severe. Such differences were sought using the Kruskal-Wallis test. The number of subjects with coronary artery disease compared with controls provided the power to detect a difference of 0.5 of a standard deviation at p < 0.05 and 1−β  =  0.80.

Statistical methods

Statistical analyses were done using the statistical program SPSS 10.0 for Windows. Parametric results are expressed as mean (SD) and differences between groups were compared by unpaired Student’s t test. Non-categorical data were compared by the χ² test. Non-parametric results are expressed as medians with interquartile ranges (IQR) and comparisons made using Mann-Whitney and Kruskal-Wallis tests. Multiple regression analyses were used to determine predictors for the research indices. Correlations were examined using Spearman’s rank correlation. The level of significance was taken to be p < 0.05.

RESULTS

Cross sectional data

Demographic details, risk factors, and drugs prescribed for the patients with coronary artery disease are summarised in table 1. vWf, VEGF, sFlt-1, and TF data are presented in table 2. As expected, all indices were increased in the patients with coronary artery disease compared with the healthy controls except for sFlt-1, which was lower in the patients.

Relation to disease severity

The distribution of coronary atheroma score was not normally distributed (median 1.77 (IQR 0.23−4.5). No significant Spearman correlations were found between coronary atheroma score and concentrations of vWf (r  =  0.21, p  =  0.83), VEGF (r  =  0.11, p  =  0.27), and TF (r  =  −0.04, p  =  0.68). However, there was an inverse correlation between sFlt-1 and coronary atheroma score (r  =  −0.19, p  =  0.049).

When coronary angiograms were analysed by the number of vessels with significant stenoses, the cohort was divided as follows:

• 37 patients (33%) had no vessels significantly diseased (with > 50% stenosis)

• 27 patients (24%) had one vessel diseased

• 22 patients (20%) had two vessels diseased

• 25 patients (23%) had three vessels diseased.

There were no significant differences in plasma concentrations of vWf, VEGF, sFlt-1, and TF between these four groups (table 3). Note that patients with no vessels with significant stenoses had a median coronary atheroma score of 1.42 (IQR 0.67−2.40) and concentrations of vWf, VEGF, and TF were higher in these patients than in the control group (p < 0.001, p  =  0.014, and p  =  0.020, respectively).

Subgroups and multiple regression analyses

The patients with coronary artery disease were then subdivided according to the presence or absence of the risk factors diabetes mellitus, hypertension, and hyperlipidaemia. The presence or absence of diabetes or hyperlipidaemia made no significant difference to plasma concentrations of VEGF, sFlt-1, TF, or vWf (data not shown). Only patients with coronary artery disease and hypertension had significantly higher concentrations of TF (p  =  0.009) when compared with normotensive patients (data not shown).

When adjusting for clinical indices in the coronary artery disease patients using standard multiple regression analyses with plasma concentrations of VEGF, sFlt-1, TF, or vWf as dependent variables, age was found to be a predictor for VEGF concentrations (β  =  −0.190, p  =  0.045), and HDL values a predictor for sFlt-1 concentrations (β  =  0.300, p  =  0.003). However, the relatively small numbers involved in these analyses must be borne in mind when interpreting their relevance.

Relations between indices

Concentrations of VEGF correlated with TF in the patients with coronary artery disease (r  =  0.25, p  =  0.008), while concentrations of sFlt-1 correlated with VEGF (r  =  0.36, p < 0.001) and TF (r  =  0.42, p < 0.001). No other correlations reached significance.

DISCUSSION

This study confirms previous observations of higher plasma VEGF, TF, and vWf in patients with coronary artery disease compared with healthy controls.^8,13,20–23 We found no relation between the coronary atheroma score and concentrations of VEGF, sFlt-1, TF, or vWf; furthermore, there were no significant differences in concentrations of indices according to subgroups, based on the number of vessels with > 50% stenosis in the patient group. Furthermore, the (very) weak Spearman correlation coefficient of −0.19 between coronary atheroma score and sFlt (p  =  0.049) has to be interpreted with caution. The possibility arises that the finding of higher concentrations of vWf, VEGF, and TF in patients undergoing angiography with no coronary artery stenoses of more than 50% compared with healthy controls may be explained by the presence of risk factors (hypertension, diabetes, and hypercholesterolaemia) in this patient group.

We are unaware of other studies investigating the relation between coronary atheroma score or peripheral plasma VEGF concentrations and coronary artery disease severity. The lack of correlation between plasma VEGF and coronary artery disease severity may reflect the more generalised nature of thrombogenesis, angiogenesis, and endothelial disturbance, rather than that occurring in specific vascular beds. In one study looking at coronary collateral flow and intracoronary VEGF concentrations, Fleisch and colleagues found a trend toward higher concentrations of intracoronary VEGF with more extensive coronary artery disease, as assessed by the number of diseased coronary arteries (stenosis > 50%) on coronary angiography.²⁴ However, concentrations of VEGF were measured on serum samples, and there is increasing evidence that VEGF should be measured on plasma samples, as in our study.^25,26

In the present study, we did not find any significant relation between vWf and coronary artery disease severity. We accept that there are many possible indices of endothelial damage/dysfunction, and in our study we chose to measure vWf, the most well established index of endothelial damage/dysfunction.²⁰ Indeed, in another study,²⁷ Yildirir and colleagues measured concentrations of E-selectin, VCAM-1, and ICAM-1 as markers of endothelial cell activation²⁰ in 83 consecutive patients attending for coronary angiography. They used patients with no coronary artery stenoses of > 50% as their control group and found significantly lower concentrations of VCAM-1 and E-selectin in their controls compared with patients with one or more stenoses of > 50%. However, included among their subjects were patients with unstable angina (n  =  23), whereas in the present study we excluded all patients with episodes of unstable angina within the previous six weeks. A significant correlation between severity of coronary lesions, as graded using the Gensini score, and E-selectin concentrations was only found in this small subgroup of patients with unstable angina, and there were no significant correlations between ICAM-1 or VCAM-1 and disease severity.

We are unaware of any studies to date that have investigated concentrations of TF in relation to angiographic disease severity, although no significant relation was present. Nonetheless, we have shown a correlation between TF and concentrations of the angiogenic markers VEGF and sFlt-1, suggesting a link between angiogenesis and thrombogenesis. However, the relevance of this observation should be interpreted with caution given the low values of the Spearman correlation coefficients (r  =  0.225−0.42). Further studies with larger sample sizes are needed to confirm the hypothesis that angiogenesis and thrombogenesis are closely related in vascular disease, as appears to be the case in embryonic studies²⁸ and cancer.^29–33

Study limitations

Our study was limited by its cross sectional design, but we have attempted to relate our research indices to coronary artery disease severity using two different methods of scoring disease severity. In addition, we recognise that our markers may simply reflect generalised vascular disease or risk factors such as diabetes or hypertension (rather than coronary artery disease per se, or concomitant treatment), and this would be a limitation common to all studies examining plasma markers such as VEGF, TF, and vWf in patients with coronary artery disease compared with healthy controls.^8,13,20–23 Furthermore, acute episodes of ischaemia could influence the production of angiogenic factors,^34–37 but we only included patients with chronic stable symptoms.

Conclusions

Despite evidence of abnormal angiogenesis (VEGF and sFlt-1), thrombogenesis (TF), and endothelial damage/dysfunction (vWf) in our patients with coronary artery disease, there was no meaningful relation between VEGF, sFlt-1, vWf, or TF and angiographically defined coronary artery disease severity.