Between November 2007 and May 2008, 261 consecutive patients referred for multidetector computed tomography (MDCT) for non-invasive assessment of CAD due to symptoms suggestive of ischemia were prospectively screened for study eligibility in a Portuguese hospital (Centro Hospitalar de Vila Nova de Gaia). Patients with known CAD (n=16), patients requiring different protocols for MDCT imaging, such as simultaneous request for aorta evaluation or combined assessment of coronary, aorta and pulmonary arteries (“triple rule-out”) (n=7), patients with very unstable rhythms (n=5), known anemia (Hb ≤8.5) (n=2), known iodine-based contrast allergy (n=1) and renal insufficiency (n=2), or with severe mobility limitations that could preclude accurate anthropometric measurements (n=5), and patients who refused or were unable to give written informed consent (n=8), were excluded. The final population consisted of 215 symptomatic patients (132 male, 83 female) without known CAD. Written informed consent was obtained from all participants.

All the 215 subjects in the final study population underwent the same protocol.

After written informed consent was obtained, a brief clinical interview was performed and clinical information (including cause of referral, symptoms, previous clinical history, known risk factors, previous examinations and medication) was obtained using a standardized health questionnaire and referral letters. Hypertension was defined as a history of known elevated blood pressure diagnosed and treated with medication, diet, and/or exercise for at least one year or requirement for antihypertensive medication. Diabetes mellitus was defined as requirement for insulin or oral hypoglycemic drugs. Dyslipidemia was defined as a history of hypercholesterolemia or hypertriglyceridemia diagnosed and/or treated by a physician or the use of lipid-lowering agents. Smoking status was defined in three categories: current smoker — smoking exposure within the last 12 months; ex-smoker — life smoking exposure greater than 2 pack-years with no smoking in the last year; non-smoker — no smoking within the last 12 months and a life consumption of less than 2 pack-years. Family history of CAD was defined as having a first-degree relative with a history of myocardial infarction, coronary revascularization or sudden death at a young age (men <55 years; women <65 years). Metabolic syndrome was diagnosed according to the Adult Treatment Panel III (ATP III) criteria12.

Abdominal circumference was measured according to standard methods13, and body mass index (BMI) calculated from simultaneous measurements of height and weight.

Prior to administration of MDCT, pre-medication, measurements of sitting heart rate and blood pressure were obtained and peripheral blood samples collected after overnight fasting.

All patients underwent an MDCT scan using a 64-slice CT scanner (Somatom Sensation 64, Siemens Medical Solutions, Forchheim, Germany), including three different acquisitions: one for abdominal fat quantification, the second for quantification of coronary artery calcification (CAC) and the third for coronary angiography plus epicardial adipose tissue quantification.

To assess abdominal fat, a single-slice abdominal CT scan was performed between L4 and L5, according to the method described by Borkan et al.14. The scan parameters were 120kV and 216 mAs with 5mm thickness. This resulted in an estimated radiation exposure of 0.06mSv. On the scan obtained, a cursor pointer was used to trace the abdominal visceral fat area, and the data were processed using a histogram-based statistical program according to the previously described method14. One expert, unaware of the patient's details, results for calcium score or CT angiography, measured abdominal fat distribution. The CT value of body fat ranged from -150 to -50 Hounsfield units (HU). Total abdominal fat area was measured and subcutaneous fat area was obtained by subtracting abdominal visceral fat from the total abdominal fat area. The ratio of visceral to subcutaneous fat (V/S ratio) was also calculated.

All patients underwent a low-dose scan to assess CAC. The scan parameters for this acquisition were: collimation 24×1.2mm; gantry rotation time 330ms; pitch 0.2; tube voltage 120kV and tube current 190 mAs. Estimated radiation exposure was 2.1±0.4mSv. Image reconstruction of the calcium score acquisition was performed using an effective slice thickness of 3mm. CAC was reported as the mean Agatston score and was calculated using a detection threshold of 130 HU with semi-automated software (Syngo Calcium Scoring, Siemens Medical Solutions) as described previously15.

Following the CAC acquisition, a CT angiography (CTA) acquisition was performed (collimation of 64×0.6mm; tube current 850 mAs; all other parameters similar to CAC acquisition scan). Tube current modulation with electrocardiographic pulsing was used to decrease the radiation dose, with full tube current applied at 60-65% of the RR interval. In patients with body weight <70kg, the tube voltage was reduced to 100kV. This resulted in an estimated radiation dose of 6.6±2.2mSv. Depending on the scan time, a bolus of 80 to 100ml of contrast (Ultravist®, iopromide 370mg/ml, Bayer Schering Pharma AG, Berlin, Germany) was injected at 5cc/min via a power injector (Stellant D®, Medrad Inc, Warrendale, PA, USA) followed by a 40-ml saline “chaser”, using a dedicated antecubital vein 18-gauge access catheter. A bolus-tracking technique was used, with a region of interest placed in the ascending aorta, set to detect a predefined threshold of 150 HU.

For assessment of CAD, multiphase sets of the reconstructed CTA images were processed on a dedicated workstation and analyzed for detection of at least one luminal diameter narrowing of >50% in any coronary artery segment16. Severely calcified segments (concentric vessel wall calcification precluding lumen assessment) were classified as positive for CAD.

Blood samples from all the participants were collected in tubes with EDTA and processed within 24h after extraction in a single reference center (Centro de Histocompatibilidade do Norte, Porto, Portugal). EPC quantification was performed according to the method described by Fadini et al. in 200517: 100ml of peripheral blood was incubated with monoclonal antibodies, FITC-conjugated anti-hCD34 (Pharmingen, Becton Dickinson) and PE-conjugated hKDR (R&D Systems). Unstained cells were used as controls. After incubation, the cells were washed with PBS, fixed with CellFix (Becton Dickinson) and subsequently counted in FACSCalibur (Becton Dickinson). The count was analyzed using fluorescence-1/fluorescence-2 dot-plot quadrant statistics and manual gating (Cell Quest Pro software, Becton Dickinson) (Figure 1).

Figure 1.

Flow cytometry analysis of a representative blood sample. EPC detection was based on detection of CD34 and KDR surface markers in cell populations of low cytoplasmic granularity.

% Gated - % of CD34/KDR in region 1; % Total: total % of CD34/KDR counts; FL1: fluorescence 1 = CD34-FITC; FL2: fluorescence 2 = KDR-PE; FSC-H: forward scatter — height; SSC-H: side scatter — height.

(0.12MB).

Ejection fraction was calculated using dedicated software (Circulation®, Siemens Medical Solutions) for CT, based on the data acquired for CTA. For this purpose, multiphase reconstructions with 20 phases (5% increments) of the R-R cycle were obtained. Systolic and diastolic phases were suggested by the software and corrected by the operator. Semi-automatic detection of the endocardial border was used under operator control for end-systolic and end-diastolic volume quantification. For left ventricular mass quantification, the epicardial contour was manually traced on the end-diastolic phase.

Mean radiation exposure was estimated by the method proposed by the European Working Group for Guidelines on Quality Criteria in CT18.

The effective radiation doses for the CAC and CTA acquisitions were calculated by the product of the chest coefficient (0.014mSv/mGy per cm averaged between male and female models) and the dose-length product (DLP) obtained during each scan18. For the abdominal fat acquisition the abdominal coefficient (0.015mSv/mGy per cm averaged between male and female models)19 was used.

Continuous variables were expressed as means and standard deviations. Distributions of continuous variables were examined using density plots and the Kolmogorov-Smirnov test. Associations between CD34+/KDR+ cells and potential predictors were examined with regression models. Models were defined as appropriate if the assumption was fulfilled for linear regression (normal distribution, heteroskedasticity and independence of residuals). Poisson regression was used for count data and non-parametric regression was used in other cases. The respective SAS codes used for linear regression were “proc glm”, for Poisson regression “proc genmod”, and for non-parametric tests “proc npar1way”. The type I error level was taken to be p=0.05. All calculations were carried out using the statistical software packages SAS version 9.1 (SAS Institute Inc., Cary, NC, USA).

EPCs appear to be the cornerstone of endogenous vascular regeneration after ischemic injury. Monitoring circulating EPCs (CD34+/KDR+) could therefore be useful for identifying patients at risk. This concept is driving research in this field as therapies aimed at stimulating endogenous vascular regeneration and repair become a reality10,11. Stents that “capture” EPCs onto their surface have been recently developed using bioengineering techniques. This approach aims to accelerate vascular regeneration after coronary angioplasty, promoting rapid re-endothelialization. This may have the advantage of reducing restenosis and intrastent thrombosis33-35. However, the application of therapies based on mobilization of endogenous EPCs requires a deeper knowledge of possible inter-individual variations in circulating EPC levels. It is thus important to recognize the normal distribution patterns of these cells in different populations and to identify factors that may influence their concentration and mobilization.

Our study aimed to investigate whether EPC counts could be predicted with any degree of certainty on the basis of patients' baseline characteristics. The results show that, at least in this population of patients with low-to-intermediate CV risk without known CAD, only very limited prediction is possible: all the previously described associations (pre-specified in our study) were found to be weak or even non-significant.

Studies on the impact of obesity on EPC concentrations in peripheral blood are contradictory8,36,37. In our study no correlation was detected between EPCs and BMI and no significant differences were found between obese and non-obese patients. Similarly, no significant relation was found between EPC levels in peripheral blood and abdominal circumference, abdominal fat or epicardial fat.

Some authors have proposed that, as circulating EPC levels may fall in the preclinical stage of the atherosclerotic process, quantification of EPCs in peripheral blood could provide useful additional information to classical risk factors for global cardiovascular risk assessment9,38,39. In our study no significant correlation was shown between EPC counts and calcium score (as a marker of atherosclerotic burden). Similarly, and in contrast to a smaller study by Wang et al.6, we found no correlation between the presence or severity of CAD and the number of circulating EPCs. This possible relation is still the subject of debate, as contradictory data have been published40,41.

Only patients without known CAD referred for MDCT were included in this single-center exploratory analysis, and so this sample may not represent the general population. The potential value of quantifying circulating EPCs for cardiovascular risk prediction was not tested, since there was no follow-up of events. Furthermore, only the number of circulating EPCs in peripheral blood — and not EPC activity or mobility — was analyzed in our study; this should be taken in consideration when comparing our results with studies in which these were measured.

Finally, the lack of a standardized, verified, and universally accepted EPC phenotype is an important limitation that is common to all studies focusing on vasculogenesis42. To minimize this, the cell population (CD34+/KDR+) chosen and the quantification method used in our study are among the most widely used in publications concerning EPC characterization.