Association of growth-differentiation factor-15 with the number of circulating proangiogenic endothelial progenitor cells in patients with type 2 diabetes mellitus

1State Medical University, Internal Medicine Department, Zaporozhye, Ukraine 2State Medical University, Clinical Pharmacology Department, Zaporozhye, Ukraine 3Institute of Histology and Embryology, University of Ljubljana, Faculty of Medicine, Slovenia 4Victor Babes University of Medicine and Pharmacy, Department of Functional Sciences, Timişoara, Romania 5State Medical University, Student of Medicine Faculty, Zaporozhye, Ukraine

Numerous clinical studies have shown the lowered number and weak function of CD34+ EPCs in patients with pre-diabetes and T2DM [20][21][22]. Moreover, vascular complications of DM including retinopathy and atherosclerosis, muscular related to declined circulating the number of CD34+ EPCs with high proliferative capacity and ability to restore vascular integrity and function [23]. Thus, GDF-15 and CD34+EPCs could coordinate vascular repair and restore endothelial function in diabetics. This study aimed to investigate the relationship between levels of GDF-15 and circulating number of EPCs with proliferative and angiopoietin phenotypes: CD34 + CD14 + CD309 + , and CD34+CD14 + CD309 + Tie2 + in patients with T2DM without known CV diseases.

Methods
The study cohort consisted of 76 patients with type 2 DM (41 males) aged 38 to 55 years who were retrospectively involved between March 2014 and July 2017. All DM patients included in the study have no known CV diseases including angina pectoris, previous myocardial infarction/stroke, heart failure, and asymptomatic atherosclerosis (defined by the negative result of the contrast-enhanced multiple spiral tomography angiography). Apart from established CV disease, the criteria of non-inclusion were acute infections; active inflammation; pulmonary edema; tachyarrhythmia; valvular heart disease; thyrotoxicosis; ischemic stroke; intracranial hemorrhage; surgery; trauma, autoimmune disease, malignancy, before the study entry. As a control cohort, we enrolled 30 healthy volunteers matched for age and sex with type 2 DM patients.

(a) Ethical declaration
All the patients have given their written informed consent for participation in the study. The study was approved by the Local Ethical Committee. The investigators followed strictly all the requirements to clinical trials in conformity with the World Medical Association Declaration of Helsinki, 1964, good clinical practice provided by International Conference on Harmonization, Council of Europe Convention for the Protection of Human Rights and Dignity of the Human Being in view of using achievements in biology and medicine, convention on Human Rights and Biomedicine, including Additional Protocol to the Convention on Human Rights and Biomedicine, concerning Biomedical Research, and legislation of Ukraine.
Diagnosis of T2DM was checked and confirmed according to the current recommendation of ADA (2018) [24]. Patients with type 2 DM were treated according to current clinical guideline [6]. All DM patients were treated with metformin in individually adjusted daily doses under continuous control of fasting glycemia, the daily profile of glucose concentration and glycated hemoglobin level (HbAc1). No insulin given patients with T2DM were selected in the study. Twenty-five T2DM individuals with mild-to-moderate arterial hypertension were treated with angiotensin-II receptor antagonist valsartan in daily doses 80 mg to 160 mg depending on office systolic and diastolic BP values. Dyslipidemic T2DM patients have treated with statins predominantly atorvastatin in averaged doses 40-80 mg/daily.
(b) Demographic data, smoking status, and anthropometric measurements Demographic factors such as age, gender, height, weight, body mass, body mass index, waist circumference, and waist-to-hip ratio past medical and medication history were collected at baseline. Current smoking was defined as consumption of one cigarette daily for three months [25].
Anthropometric data were measured by professional health attendants with the participants standing without shoes and heavy outer garments with a wall-mounted stadiometer (OMRON, Japan). Body mass index (BMI) was calculated by the staff person as weight (kg) divided by height  [26,27].

(c) Cardiac ultrasound and Doppler procedures
Transthoracic echocardiography was performed on ACUSON ultrasound system (Siemens, Germany) in B-mode regimen and Tissue Doppler Imaging (TDI) regimen from parasternal, subcostal, and apical positions over the short and long axis using 5 MHz phased transducers. Left ventricular ejection fraction (LVEF) was measured by the modified Simpson's method [28]. Left ventricular (LV) mass was estimated using the formula recommended by American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group [28]. LV hypertrophy (LVH) was defined as an LV mass/body surface area (BSA) ≥ 96 g/m 2 , for women, and ≥ 116 g/m 2 , for men [29]. To measure peak systolic (Sm), early diastolic (Em), and late diastolic (Am) myocardial velocities TDI was carried out according to the American Society of Echocardiography [29].
(d) Calculation of glomerular filtration rate CKD-EPI formula was used to calculate Glomerular Filtration Rate (GFR) [30].

(e) Blood sampling
Blood samples were drawn in the morning following overnight fasting (at 7-8 a.m.) into barcoded silicone test tubes (Thermo Fisher Scientific, Waltham, MA, USA) wherein two mL of 5% Trilon B solution were added. Then samples were centrifuged upon permanent cooling at 6,000 rpm for 3 minutes, and then plasma was collected to be immediately refrigerated. Each aliquot was stored at a temperature -70 o C. The GDF-15 level was measured by ELISA assay using commercial kit manufacturing by LifeSpan BioSciences (USA) on Elecsys 1010 analyzer (Roche, Mannheim, Germany). Detection range was 31.2 to 2,000 pg/mL.
High-sensitivity C-reactive protein (C-RP) levels were measured by the nephelometric technique with the commercial kit (Eagle Biosciences, Nashua, NH, USA) and obtained with "AU640 Analyzer" (Olympus Diagnostic Systems Group, Japan).
High-performance liquid chromatography method was performed to determine hemoglobin A1c (HbA1c) in 5% Trilon B anticoagulated blood samples.

(f) Sample preparation for isolating peripheral blood mononuclear cells
We used a standard method for isolating of peripheral blood mononuclear cells from collected blood samples by means density gradient centrifugation using "Lympholyte" solution (Cedarlane Laboratories, Burlington, ON, Canada). Each prepared sample contained 5 mL of peripheral blood were previously exposed to 1:2 dilution with Phosphate-Buffered Saline (PBS, Gibco TM PBS buffers, Thermo Fisher Scientific, Waltham, MA, USA) and stratified onto 5 mL of "Lympholyte" solution. All received samples were centrifuged 30 min at room temperature at 3,000 rpm without a brake. After separation, white blood cells were recollected, diluted with PBS+0.02% Tween and centrifuged at 1,200 rpm for 5 min at room temperature. For erythrocytes removal pellet was treated with RBC lysis buffer and washed 6X with 300µL in PBS before analysis.

(h) Statistical Analysis
Statistical analysis of the results obtained was carried out in SPSS system for Windows, Version 20 (SPSS Inc, Chicago, IL, USA). The data were expressed as mean (M) and error of the mean (±m) or a 95% confidence interval (CI); the median (Me) and the interquartile range (IQR). Categorical variables were reported as numerous (n) and percentages (%). Shapiro-Wilk test and Kolmogorov-Smirnov test were used to assay the normality of continuous variables. To compare the main parameters of patients' groups (subject to the type of distribution of the parameters analyzed), one-tailed Student t-test or Mann-Whitney U-test were used. The two-tailed version of the Wilcoxon test was used for paired comparison of parameter values inside the group. Categorical variables between groups were compared with Chi 2 test (χ2) and Fisher F exact test. The factors, which could be associated potentially with the number of circulating EPCs, were determined by means of univariate analysis of variance (ANOVA); and then, the identified factors with P< 0.1 also were studied by means of multivariate analysis of variance (MANOVA). The odds ratio (OR), Wald x 2 and 95% CI were calculated for all the independent predictors of declining of the circulating number of EPCs with angiogenic phenotypes. The calculated difference of P<0.05 was considered statistically significant. Biological markers in individuals participating in the study are reported in Table 2. There were no significant differences between both cohorts in estimated GFR as well as in levels of creatinine. Fasting glucose, HbA1c, hs-CRP, and GDF-15 in blood were sufficiently higher in people with diabetes to healthy volunteers. Additionally, lipid profile abnormality was found in people with diabetes rather than in healthy volunteers. Total cholesterol (TC), cholesterol of lowdensity lipoproteins (LDL) and triglycerides (TG) were higher, but cholesterol of high-density lipoproteins (HDL) was lower in people with diabetes in comparison with healthy individuals.

Discussion
This is the first study that determines the inverse association between levels of GDF-15 in peripheral blood and circulating number of EPCs with angiopoietin capacity labeled CD45 − CD34 + CD14 + CD309 + and CD45 − CD34 + CD14 + CD309 + Tie 2+ in type 2 DM patients without established CV diseases. We confirmed that elevated levels of GDF-15 in diabetic population positively related to age, BMI, smoking, and hs-CRP and that circulating number of EPCs in people with diabetes was significantly lower to healthy volunteers. In this study, we have been hypothesized that deficiency of EPCs with high proliferative and angiopoietin activity in patients with type 2 DM associating with altered vascular repair could be compensated with the protective capability of GDF-15. Indeed, previous studies have revealed that elevated GDF-15 level could be a protector from injury of numerous tissues, such as heart, adipose tissue, and endothelium, by inhibiting c-Jun N-terminal kinase, Bcl-2-associated death promoter, and epidermal growth factor receptor and activating Smad, eNOS, PI3K, and AKT signaling pathways [32][33][34]. However, there are severe controversies in this issue, because few reports in opposite revealed that lowered GDF-15 levels are beneficial against endothelial injury and microvascular inflammation [35,36]. In contrast, there is evidence that successful glycemic control and treatment of dyslipidemia with statins in moderate-to-high doses did not cause changes in plasma levels of GDF-15 [37,38]. Although traditionally achieving of full control for hyperglycemia and hyperlipidemia is considered a predictor of better clinical outcomes in people with diabetes, lack of GDF-15 dynamic requires to be clarified. We suggest that GDF-15 as stressinduced cytokine interplays PI3K / AKT signaling in EPCs supporting their proliferative capacity.
On the other hand, inhibiting c-Jun N-terminal kinase and Bcl-2-associated death promoter with GDF-15 could protect EPCs epigenetically modified by LDL cholesterol, TG, pro-inflammatory cytokines, phospholipases, reactive oxygen species and improve survival of proangiogenic EPCs. Therefore, activating Smad / eNOS in early growth EPCs may accompany by eliminating apoptotic EPCs from circulation that prevent vascular injury [39]. Nevertheless, stimulating of Tie2/PI3K/Akt/eNOS signaling in late growth EPCs supports their re-endothelialization capacity and survival [40]. The final effect of GDF-15 on EPCs is high likely positive and associate with vascular reparation and protection of resident cells from "metabolic memory" phenomenon [41]. Additionally, there is a large body of evidence regarding that other CV risk factors corresponding to DM development and progression could directly influence on vascular wall and endothelium. Indeed, hypertension, dyslipidemia, hyperglycemia, and inflammation are established risk factors for multiple focus atherosclerosis and vascular calcification [6]. Whether GDF-15 could be involved as the protector from atherosclerotic injury with similar molecular mechanisms as mentioned above is not sufficiently clear. Although some scientists reported that levels of GDF-15 in people with diabetes demonstrated the close positive association with traditional CV risk factors, such as LDL cholesterol, BMI, age, inflammatory activity, but not with TG [14,15]. The results of our study well correspond with these findings, and we yielded that GDF-15 could be a component of the endogenous vascular repair system that maintains vascular integrity and function. In fact, the role of GDF-15 in DM is not entirely understood and requires to be investigated in the large clinical trials focusing on protective mechanisms of GDF-15 on EPCs and other components of vascular repair system.

(a) Study limitations
There were several limitations of the study predominantly relating to a small number of patients and no randomized design. Although it was a small sample size in this study, statistical power was adequate. To note, the duration of T2DM for each patient included in the study was not known and hence. It can be significant because there was a strong negative correlation between the number of circulating angiogenic EPCs and severity of the disease, as well as fluctuation of fasting glucose and the level of HbAc1. Therefore, in the study, we included T2DM patients without established CV disease including asymptomatic atherosclerosis. Probably, an association between GDF-15 and circulating number of angiogenic EPCs requires to be compared in people with diabetes with known atherosclerosis and/or CV disease. Additionally, our findings regarding protective role of GDF-15 remains to be confirmed via validation in external cohorts, particularly in large clinical trials.

Conclusions
The results of the study clarified that GDF-15 was an independent factor influenced on number of endothelial precursors with high angiopoetic activity in T2DM patients. These findings could take into consideration in risk stratification and individualized care of the T2DM patients.