INTRODUCTION<\/strong><\/p>\n Prediabetes is defined as an elevation of plasma glucose values above the normal range but below the diagnostic criteria for diabetes.1<\/sup> The prediabetic state is characterized by insulin resistance and is often associated with inflammation and the presence of additional atherosclerotic risk factors linked to impaired endothelial function.2<\/sup> Endothelial damage and dysfunction are the early signs in the development of atherosclerosis and predictors of cardiovascular events.3<\/sup><\/p>\n During the last two decades, several studies have focused on circulating endothelial progenitor cells (EPCs) as a novel biomarker of endothelial function. EPCs originate primarily from bone marrow and were first identified in peripheral blood by Asahara and colleagues in 1997.4<\/sup> EPCs play a prominent role in neovascularization, incorporate into\u00a0sites of endothelial damage, help to maintain endothelial integrity and contribute to accelerated reendothelization and vascular homeostasis.5<\/sup> Decreased amounts of EPCs, as a possible biological marker, are correlated with atherosclerotic disease progression and increased cardiovascular risk.6,7<\/sup><\/p>\n To date, studies have focused on the associations between EPCs and sociodemographic, anthropometric as well as cardiometabolic risk factors either in healthy populations or in patients with diabetes.8,9<\/sup> Age, sex, exercise, smoking, lipid profile, insulin resistance, inflammation, hypertension and genetic background have emerged as meaningful factors correlated with EPCs.8,9 <\/sup>It has been shown that the prediabetic state is negatively associated with EPCs.10<\/sup> Nevertheless, examination of the aforementioned risk factors, especially in patients with prediabetes, remains an unexplored field. The present study aims to investigate and elucidate possible associations between cardiometabolic parameters of EPC levels in individuals with prediabetes in the absence of cardiovascular disease.<\/p>\n MATERIALS AND METHODS <\/strong><\/p>\n Study design and population <\/em><\/strong><\/p>\n The study population of this cross-sectional study consisted of general population subjects recruited in the outpatient diabetes clinic and internal medicine departments of Tzanio General Hospital, Piraeus, Greece. The study was approved by the Local Ethics Committee and performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants. The sample population consisted of 39 voluntary adults with prediabetes free of overt cardiovascular disease at study initiation. None of the participants had been diagnosed with alterations in carbohydrate metabolism prior to enrollment. Cardiovascular disease was excluded via a complete medical history, comprehensive physical examination, electrocardiogram, echocardiogram, treadmill exercise test and thallium 201 cardiac scintigraphy when applicable. Prediabetes was defined after performing 75g OGTT according to the 2011 ADA position statement.1<\/sup> Individuals were classified as having impaired fasting glucose (IFG) (fasting plasma glucose levels: 5.6 mmol\/l to 6.9 mmol\/L) and\/or impaired glucose tolerance (IGT) (2-h glucose levels in the oral glucose tolerance test: 7.8 mmol\/l to 11 mmol\/L). The newly adopted criterion for defining prediabetes by using HbA1c (5.7% to 6.4%) could not be used in our study since the HbA1c measurements in our laboratory were not in accordance with the National Glycohemoglobin Standardization Program (NGSP). All participants were free of overt renal and hematologic disease. Exclusion criteria also included: history of stroke, peripheral arterial disease, chronic heart failure (NYHA class III and IV), secondary hypertension, history of malignancy, history of infection in the last month, autoimmune disease, hyper- or hypothyroidism, excess alcohol intake (>14 drinks\/wk for women and >21 drinks\/wk for men), hormone replacement therapy and intake of oral contraceptives. Participants were classified as physically active if they were engaged in moderate physical activity (a minimum of 150 minutes of moderate aerobic activity per week) or otherwise sedentary. Vigorous exercise was not reported by any participant.<\/p>\n Measurements <\/em><\/strong><\/p>\n Medical history and anthropometric and biochemical parameters were obtained from all participants. Waist circumference was measured midway between the lowest border of the ribs and the iliac crest in a horizontal plane and hip circumference at the widest level over the greater trochanters. Visceral and subcutaneous abdominal fat thickness were assessed according to an ultrasonographic protocol.11<\/sup> Subcutaneous fat was defined as the depth between the cutaneous boundary and the linea alba. Visceral fat was defined as the depth between the peritoneum and the lumbar spine at the end of a quiet expiration without distortion of the abdominal cavity. In addition, fasting venous blood samples were collected from all participants for the determination of EPCs, lipid profile, glucose, insulin and high sensitive C-reactive protein (hs-CRP). The levels of hs-CRP in the serum samples were measured with high-sensitivity methods using nephelometry (BN II Nephelometer-Siemens).Insulin resistance was estimated by the homeostasis model assessment of insulin resistance (HOMA-IR) as described by Matthews et al.12<\/sup><\/p>\n Quantification of circulating EPCs<\/em><\/strong><\/p>\n EPCs were detected with flow cytometryand defined by the surface expression of CD34, CD133 and VEGFR-2. The antibody cocktail consisted of CD34+FITC (Beckman Coulter, Fullerton, CA, USA), anti-CD133PerCPeFluor 710 (eBioscience, San Diego, CA, USA) and anti-VEGFR-2 (R&D Systems). As the CD34FITC antibody concentration was not specified in the respective leaflet, they were mixed and titrated in umbilical cord blood in order to define the dilutions that minimized non-specific staining. Following these preliminary experiments, 5ul of anti-CD34FITC, 3ul of anti-CD133PerCPeFluor 710 and 3ul of anti-VEGFR-2-PE were mixed and added to 50 ul of whole blood. The fixed amount of blood was added to BD Trucount tubes\u2122 (BD Biosciences, San Jose, CA, USA), which contain known numbers of microbeads, allowing us to measure the absolute number of EPCs, as described in the literature.13<\/sup> A blood specimen was added to the antibody mix by reverse pipetting to improve accuracy and reproducibility.\u00a0Following incubation for 15 min at room temperature in the dark, 450 ul of NH4Cl were added and incubated for 15min at room temperature in the dark. Each specimen was stained in duplicate, 8000 beads of each sample were acquired on FacsCan (BD Biosciences) and control samples were analyzed using BD Cell-Quest\u2122 software. Control samples stained with isotype-matched antibodies, were used to exclude unspecific staining. The gated CD34+ cells, based on ISHAGE protocol,14<\/sup> were examined for the dual expression of CD133 and VEGFR-2 antigens. The absolute number of EPCs was expressed as the number of EPCs per mL of blood and calculated according to the formula: number of events in region containing cell population\/number of events in absolute count bead region X number of beads per test\/test volume.<\/p>\n The most frequently used and satisfactory method to define EPCs is by determination of co-expression of CD34+CD133+VEGFR-2+, or even dual expression CD34+VEGFR-2+. In our study sample, CD34+CD133+VEGFR-2+ and CD34+VEGFR-2+ concentrations were closely associated with each other (Spearman\u2019s rho=0.905, p<0.0001); given their close collinearity, CD34+CD133+VEGFR-2+ were only examined as the dependent variable so as to avoid unnecessary duplication of findings.<\/p>\n Statistical analysis <\/em><\/strong><\/p>\n Descriptive statistics were performed; categorical variables were presented as frequency (%), normally distributed variables as mean\u00b1SD, whereas non-normally distributed variables were summarized as per median and range. The Shapiro-Wilk test was performed for the evaluation of deviation from normality. Subsequently, for the evaluation of the factors associated with EPCs concentration, a standard two-step approach was followed: univariate and multivariate analysis. In univariate analysis, non-parametric tests were appropriately implemented given that EPCs concentration significantly deviated from normality. Specifically, Spearman\u2019s rank correlation coefficient (SR) was calculated when the independent variable was continuous or ordinal; the Mann-Whitney-Wilcoxon test for independent samples (MWW) was conducted when the independent variable was binary, whereas the Kruskal-Wallis test (KW) was performed when the independent variable was categorical with three or more subgroups.<\/p>\n In multivariate analysis, ordinal logistic regression was performed with backward selection of variables. The concentration of EPCs (dependent variable) was converted to an ordinal variable corresponding to the four levels\/quartiles. The four levels were as follows: 1: minimum value (0 cells\/mL) to 25th percentile (51 cells\/mL); 2: 25th percentile (51 cells\/mL) to median (158 cells\/mL); 3: median (158 cells\/mL) to 75th percentile (363 cells\/mL); 4: 75th percentile (363 cells\/mL) to maximum value (1133 cells\/mL). As appropriate, factors proven significant in univariate analysis were tested in the stepwise multivariate model as independent variables. The level of statistical significance was set at 0.05. Statistical analysis was performed using STATA 8.0 statistical software (Stata Corporation, College Station, TX, USA).<\/p>\n RESULTS<\/strong><\/p>\n The study population consisted of 17 males and 22 females, aged 49.5\u00b111.4 years (median: 49, range: 32-67). 28.2% of the study sample were current smokers (n=11). In relation to exercise status, 18 subjects (46.2%) declared physical exercise, whereas 21 subjects (53.8%) did not. The sample consisted mainly of patients with IFG (69.2%) and secondarily of patients with IFG and IGT (20.5%), as well as subjects with IGT (10.3%). HbA1c ranged between 5.5 and 6.3% (mean\u00b1SD: 5.93\u00b10.21, median: 5.9%). EPCs ranged between 0 and 1133cells\/mL (mean\u00b1SD: 247\u00b1273, median: 158 cells\/mL). The baseline clinical and metabolic characteristics of the 39 participants are summarized in Table 1<\/a>.<\/p>\n Univariate association analyses were performed to determine which variables were most closely associated with EPCs levels. Prediabetic status (p=0.910, KW), age (p=0.270, SR), sex (p=0.842, MWW), waist to hip ratio (p=0.744, SR), subcutaneous fat (p=0.497, SR), exercise (p=0.380, MWW), smoking status (p=0.490, MWW), statin use (p=0.078, MWW), renin-angiotensin-aldosterone system inhibitors (p=0.497, MWW), fasting glucose (p=0.560, SR), insulin (p=0.465, SR), HbA1c (p=0.093, SR), HDL-cholesterol (p=0.228, SR), triglycerides (p=0.257, SR) were not significantly associated with EPC levels. It should be noted that 40% of the participants in the quartile with the highest levels of EPCs were receiving statins; accordingly, a marginal trend towards an association between statin consumption and higher EPC levels was observed (322\u00b1154 EPCs\/mL vs. 233\u00b1289 for participants not receiving statins; p=0.078, MWW). In contrast, EPC levels were significantly inversely correlated with BMI (rho=-0.660, p<0.001), waist circumference (rho=-0.407, p=0.017), visceral fat, (rho=-0.339, p=0.041), HOMA-IR (rho=-0.400, p=0.031), systolic (rho=-0.400, p=0.019) and mean blood pressure (rho=-0.440, p=0.009), total cholesterol (rho=-0.471, p=0.003), LDL cholesterol (rho=-0.373, p=0.023) and hs-CRP (rho=-0.442, p=0.006). Variables shown to be statistically associated with EPC levels after performing univariate analyses are summarized in Table 2<\/a>.<\/p>\n Continuous variables have been presented as <median and >median for purely descriptive reasons; their continuous nature has been appropriately taken into account in the univariate tests.<\/p>\n