MicroRNAs modulation and clinical outcomes at 1 year of follow-up after excision of subcutaneous abdominal fat in overweight patients with pre-diabetes treated with metformin vs. placebo

Background: obese pre-diabetics have altered expression of cytokines, and sirtuin-1, that might influence myocardial function via microRNAs (miRs) expression. Objectives: to evaluate inflammatory/oxidative stress, miRs ’ expression and cardiovascular function in obese pre-diabetics randomly assigned to metformin therapy vs. placebo vs. normoglycemics at 12 months of follow-up. Materials and methods: eighty-three obese patients enrolled for abdominoplastic surgery, were divided in pre-diabetics (n 55), normo-glycemics (n 28), and assigned to hypocaloric diet. Pre-diabetics were randomly assigned to metformin (n 23) or to placebo (n 22) plus hypocaloric diet. Results: at enrollment, pre-diabetics obese vs. normo-glycemic presented higher values of glucose, insulin resistance (HOMA-IR), inflammatory/oxidative stress markers, miR-195 and miR-27, and lower values of sirtuin-1 (p<0.05). At 12 months of follow up, obese pre-diabetics with metformin vs. placebo experienced significant reduction IMT, MPI, and LVM (p<0.05). Obese pre-diabetics in metformin vs. placebo showed significant reduction of serum miR-195 and miR-27 (p<0.05). Obese pre-diabetics in metformin vs. normoglycemics showed higher expression of serum miR-195 and miR-27 ( p<0.05). Finally, we found inverse relation between IMT and insulin (R -0.351), HOMA-IR (R -0.340), miR-195 (R -0.355), miR-27 (R -0.181); between LVEF and Insulin (R -0.332), HOMA-IR (R -0.142), miR-195 (R -0.297) and miR-27 (R -0.163). We found inverse correlation between LVM and sirtuin-1 (R -0.272), Insulin (R -0.810), HOMA-IR (R-0.183), miR-195 (R -0.446) and miR-27 (R-0.433), and direct correlation with interleukin-6 (R 0.195). MPI inversely linked to miR-195 (R -0.260) and miR-27 (R -0.591). the miRNAs isolation. RNA concentration and quality was determined by Nano Drop 2000c spectrophotometer (Thermo Scientific).Mature miRs were converted in cDNAby Gene Amp PCR System 9700 (Applied Biosystems) and MiScript II Reverse Transcription Kit (218161, Qiagen).The triplicate determinations of hsa-miR-195-5p (MIMAT0000461) and hsa-miR-27a-3p (MIMAT0000084)(MIMAT000163) were evaluated through CFX96 Real-Time System C1000 Touch Thermal Cycler (BioRad Laboratories, Inc), by using miScript SYBR Green PCR kit (218073, Qiagen) and specific miScript primer Assays (MS00003703 andMS00003241,Qiagen). Ce_miR-39-3p (MIMAT0000010; MS00019789, Qiagen) was used to normalize miRs expression. The qRT-PCR data were analyzed by using 2^ - ΔΔCt method, where Inc., Chicago Illinois). We calculated a sample size with 15 participants for each group, with estimated 80% power to detect a change of 0.015 between the mean changes of LVM in the placebo-treated and actively treated groups, at a 5% level of significance. A 20% Loss due to early withdrawals and/or non-evaluable measurements was assumed and, combined with the effect of stratification on analysis, resulted in the requirement to recruit at last 18 patients per treatment group. finding and to evaluate the effects of metformin at molecular, cellular, epigenetic and clinical level. Thus, in future studies we could evaluate the effects of metformin therapy added to abdominoplastic surgery and hypocaloric diet on the abdominal fat regulation of SIRT1/miRs, and of cytokine blood levels and circulating miRs, and to translate these information in a clinical setting for pre-diabetics patients.

protective effects on cardiac remodeling (10,11). Notably and clinically relevant, as first there are not data about metformin effects on molecular inflammatory pathways and miRs' expression in obese patients with pre-diabetes.
Secondly, there are not data about metformin effects on cardiac remodeling via the regulation of molecular inflammatory pathways and miRs' expression in obese patients with pre-diabetes. Thus, our study hypothesis was that in overweight normoglycemics vs. pre-diabetics patients there was a significant difference of the adipose tissue expression of inflammatory/oxidative stress molecules, SIRT1, miR-195 and miR-27. Moreover, these patients could differently express circulating levels of inflammatory/oxidative stress factors, and of miR-195 and miR-27 at baseline. Thus, in the present study we evaluated the expression of inflammatory/oxidative stress molecules, SIRT1, miR-195 and miR-27 at baseline in the adipose tissue of overweight with pre-diabetes vs. normoglycemics, and the circulating expression of inflammatory/oxidative stress biomarkers and of miR-195, miR-27 at baseline, and selectively in normoglycemics vs. pre-diabetics randomly assigned to metformin vs. placebo therapy at follow-up of 12 months.
Finally, we correlated the values of these biomarkers to modifications of intima-media thickness (IMT), left ventricle mass (LVM), left ventricle ejection fraction (LVEF) and cardiac performance at 12 months of follow-up in overweight normoglycemics vs. overweight with pre-diabetes randomly assigned to metformin vs. placebo therapy.

Research design and Methods
We prospectively enrolled in a placebo-controlled study, conducted from January 2015 to January 2019 at University of Campania "Luigi Vanvitelli" , 83 obese patients with standard indications to receive an abdominoplastic surgery (6).
We defined obese the patients with body mass index (BMI) > 30 kg/m2 (7). All 83 patients underwent abdominoplastic surgery, and after treatment received a hypocaloric diet as previously reported (1). The full description of diet therapy is reported in supplementary file.
Patients with pre-diabetes were randomly divided in those (n =28) treated by hypocaloric diet added to metformin therapy vs. those (n =27) treated by hypocaloric diet plus placebo. However, the patients with pre-diabetes received metformin 850 mg twice a day vs. placebo. Twenty-eight patients were obese normo-glycemics and received hypocaloric diet.
All study groups volunteered for repeated clinical evaluations and laboratory analyses as well as echocardiography. In the follow-up period, patients were treated with a multidisciplinary approach consisting of diet, exercise, and behavioral and nutritional counseling as previously described (1). The enrolled patients were followed quarterly on an outpatient basis until 12 months.
Exclusion study criteria were: diagnosis of type 2 diabetes mellitus, cardiovascular disease, psychiatric problems, a history of alcohol abuse, smoking, or any hypoglycemic medication assumption. All patients had normal results for laboratory data (urea nitrogen, creatinine, electrolytes, liver function tests, uric acid, thyroxin, and complete blood count), chest x-rays, and electrocardiograms. All patients were evaluated at baseline, and at 12 months follow up.
Each patient provided informed written consent to participate in this study, which was approved by the institutional committee of ethical practice of our institution. The patients subscribed a separate informed consent to undergo abdominoplasty.

Echocardiography
Two experienced physicians (C.S, G.G) full trained in cardiovascular echography, performed a trans thoracic full 2dimensional and Doppler echocardiography assessment in independent way and blinded to study protocol and treatment groups. The exam was performed at baseline and after 12 months, according to the American Society of Echocardiography recommendations (12). For the echocardiography we used a Philips iE33 echocardiograph (Eindhoven, The Netherlands). LVM was calculated and normalized by both body surface area (BSA) and by height squared correct for the effect of overweight (13). As described previously, we calculated the LVEF, as index of cardiac pump (12). LVEF was calculated, dividing the stroke volume by the volume of blood collected in the left ventricle at the end of diastolic filling as end-diastolic volume (12). The stroke volume was the fraction of chamber volume ejected in systole as the difference between end diastolic volume and end systolic volume (12). Subsequently, we measured doppler velocities and time intervals from mitral inflow and left ventricular outflow recordings. However, mitral early diastolic flow deceleration time was measured as the time interval between the peak of early diastolic velocity and the end of early diastolic flow (12). The total systolic time interval was measured from the cessation of one mitral flow to the beginning of the following mitral inflow (12). The ratio of velocity time intervals of mitral early and late diastolic flows was then calculated (12). As previously reported (1), we evaluated as myocardial doppler indexes and intervals the left ventricle isovolumetric relaxation time (IRT), left ventricle ejection time (ET), and left ventricle isovolumetric contraction time (ICT) (1,8). Thus, IRT was the time interval from cessation of left ventricular outflow to onset of mitral inflow, and the ET was the left ventricle ejection time, as an interval from the onset and cessation of left ventricular outflow (1). However, we evaluated the ICT as interval calculated by subtracting ET and IRT from the total systolic time interval (1). Thereafter, from these indexes we obtained the myocardial performance index. In fact, myocardial performance index (MPI) was calculated by using the formula MPI= (IRT +ICT)/ET (1). The MPI, by including both systolic and diastolic time intervals, could be evaluated for assessment of cardiac dysfunction (1).
Finally, we calculated in all study population the LVM using 2-dimensional (2D) echocardiography, and via M-mode as recommended method (12). The used formula to estimate LVM from linear dimensions, based on the assumption of the left ventricle as a prolate ellipsoid of revolution, was to have linear measurements of interventricular septum wall thickness (IVST), as well as left ventricular internal diameter (LVID) and posterior wall thickness (PWT), from the parasternal acoustic window in end-diastole at the level of the LV minor axis (mitral valve leaflet tips) using 2Dtargeted M-mode or directly from 2D images (12).

Measuring the IMT of carotid artery
At baseline and at follow-up we evaluated for all study population the IMT of carotid artery by B-mode ultrasound imaging of the carotid arteries, using a Philips iE33 echocardiograph (Eindhoven, The Netherlands) with a 7-MHz linear array transducer used to clearly display both the blood-intima and media-adventitia boundaries on the far walls of the arteries. However, according to authors we used gain settings to optimize image, and to have best quality for visualization of the carotid arteries' lumen (14). However, we obtained the scanning of the right and left common carotid arteries longitudinally in the 5-to 20-mm segment proximal to the carotid bulb and in areas free of plaques (14). Than in the study protocol we used these projections for measuring carotid artery IMT, to locate the lumenintima and media-adventitia echographic boundaries, and to have IMT measurements offline on a personal computer (14). Finally, we diagnosed the atherosclerotic carotid plaques, as localized echo structures encroaching into the vessel lumen for which the distance between the media-adventitia interface and the internal side of the lesion was>1.5 mm (14). All scans and IMT measurements were performed by a two experienced physicians (G.G, R.M) full trained in vascular ultrasound, blinded to the study protocol, and the intra observer coefficient of variation for C-IMT was<3%.

Abdominal Dermolipectomy
As previously reported, the patients underwent conventional abdominoplasty surgical procedure, with umbiliculum transposition and cutaneous adipose mass tissue excision ranging from 200 grams (1). However, 24 h after the intervention the patients were mobilized; anti-inflammatory therapy (non-steroidal anti-inflammatory drugs) was suspended after 48 h and were discharged 72 h following with antibiotic therapy. From excision of adipose tissue authors (G.P, S.F) extracted the specimens of adipose tissue for further analysis.

Analyses of adipose tissue
After the abdominal surgery, the specimens were cut parallel to the long axis into four different parts for the different works-ups. Thus, a first part was frozen in liquid nitrogen for the following enzyme-linked immune-sorbent assay analysis. However, a portion of the other specimens was immediately immersion fixed in 10% buffered formalin.
Sections were serially cut at 5 µm, mounted on lysine-coated slides, and stained with hematoxylin/eosin. Finally, the specimens were analyzed by light microscopy.

Analyses of Blood Samples
Authors obtained for all the patients enrolled in the study the serum samples, to evaluate the inflammatory markers, cytokine levels and miRsì expression. The samples were stored at temperature under 80°C until assayed. Serum concentrations of tumor necrosis factor alpha (TNFα), interleukine 6 (IL6), and Nitrotyrosine were determined in duplicate using a highly sensitive quantitative sandwich enzyme assay (ELISA, Quantikine HS; R&D Systems, Minneapolis, MN). Venous blood samples were drawn for nitrotyrosine evaluation. Nitrotyrosine plasma concentration was assayed by ELISA, after an overnight fast, at breakfast time, and before the sensor insertion. Assays for serum total and high-density lipoprotein cholesterol, triglyceride, and glucose levels were performed in the hospital's chemistry laboratory. Authors assayed the plasma insulin levels by radioimmunoassay technique (Ares, Serono, Italy). However, we assessed the insulin resistance in the fasting state with homeostasis model assessment (HOMA) and calculated with the following formula: fasting plasma glucose (millimoles per liter) times fasting serum insulin (microunits per milliliter) divided by 25, as previously described (1).

MicroRNAs isolation and reverse-transcription Real Time PCR (qRT-PCR)
The miRs were isolated from adipose tissue by using Mirneasy mini kit (Qiagen, Milan, Italy), while serum miRs isolation was performed by using miRNeasy serum/plasma kit (Qiagen, Milan, Italy  (15).

Study endpoints
The study endpoints were:

Statistical analysis
The data were presented as group means ± SD. One-way analysis of variance (ANOVA) was used to compare baseline data, followed by Scheffe's test for pair wise comparisons. We used simple and partial correlation to evaluate relationships between variables. A linear regression analysis was performed to study the relation between the study variables (BMI, hypertension, WHR, HOMA-IR, glucose, miR 195, miR 27 etc) and the clinical study outcomes. In detail, these study variables were correlated to the delta values of study outcomes, that represented changes between follow up vs.
baseline values. The presented data were a combination of obese pre-diabetics in metformin arm, obese pre-diabetics in placebo arm and of obese normoglycemics, and we performed a partial correlation with groups and fasting glucose as a confounder. However, for each study outcome we calculated the R value and the p value. A value of P<0.05 was considered significant. Statistical analysis was performed using the SPSS software package for Windows 17.0 (SPSS Inc., Chicago Illinois). We calculated a sample size with 15 participants for each group, with estimated 80% power to detect a change of 0.015 between the mean changes of LVM in the placebo-treated and actively treated groups, at a 5% level of significance. A 20% Loss due to early withdrawals and/or non-evaluable measurements was assumed and, combined with the effect of stratification on analysis, resulted in the requirement to recruit at last 18 patients per treatment group.

Clinical characteristics of study population at enrollment.
At baseline, pre-diabetics obese in metformin and in placebo arm of treatment vs. normo-glycemics obese patients presented higher values of glucose (6.64 ± 0.14 vs. 5.31 ± 0.52 mmol/L, and 6.63 ± 0.18 vs. 5 This trend was confirmed as over inflammation/oxidative stress and miRs' expression also at level of adipose tissue.

Study outcomes
At follow-up end the enrolled patients experienced a modification in circulating miRs' values with significant difference comparing the three cohorts of study. Indeed, comparing pre-diabetics obese in metformin vs. prediabetics obese in placebo we found a significant reduction of miR    Table 3. For MPI we found an inverse relation with miR-195 (R -0.260, p=0.049) and miR-27 (R -0.591, p=0.026). Table   3.

Discussion
In the present study, the hyperglycemia and the insulin resistance in obese with pre-diabetes vs. obese and secretion of inflammatory molecules in human adipose tissue and in obese patients (17). Again, the hyperglycemia could increase miR-195 levels and decrease SIRT1 expression in a cultured cellular model of diabetes (18). To date, the transfection with miR-195 antagomir in these cells forced expression of SIRT1, preventing the SIRT1 mediated tissutal damage via hyperglycemia and insulin resistance (18). In this setting, miR-195 caused endothelial dysfunction and insulin resistance via over expression of inflammation/oxidative stress and down regulation of SIRT1 (19). In addition, in diabetic mice model the knockdown of miR-195 increased myocardial capillary density and improved maximal coronary blood flow, leading to the reduction of diabetic cardiomiopathy (20). Therefore, metformin therapy could cause down regulation of miR-195, beyond improving glycemic control and insulin resistance, and leading to cardio-protective effects (21). Similarly, the expression of miR-27 is increased in fat tissue of obese mice, and in patients with metabolic syndrome and type 2 diabetes mellitus (23). Indeed, miR-27 is implied in molecular pathways regulators of lipid metabolism, vascular signaling, and glucose homeostasis (23). Furthermore, in peripheral blood increased levels of miR-27 are linked to over expression of inflammatory markers as IL-6, TNF-α, and to reduced expression of SIRT1 (24). Notably, miR-27 has a regulatory role in increasing insulin sensitivity (25). Thus, in cellular lines with insulin resistance, authors found an altered expression of PPAR-γ, PI3K/Akt signaling pathway and GLUT4 expression after transfection with antagomiR-27 (25). However, the down regulation of miR-27 levels in cellular and mice model enhanced glucose uptake (glucose lowering effect) in time-and dose-dependent manners (25).
Therefore, taken together our study results might evidence, in obese patients with pre-diabetes, the implication of miR-195 and miR-27 as regulators of glucose homeostasis and insulin resistance, and of inflammation/oxidative stress pathways which are implied in cardiovascular remodeling (17)(18)(19)(20)(21)(22)(23)(24)(25). Notably and clinically relevant, the metformin in these patients could modulate these metabolic and inflammatory pathways, leading to cardio-protective effects.
Furthermore, miR-195 and miR-27 could become targets for the treatment of obesity and insulin resistance, and to revert the pathological shifting from over inflammation/oxidative stress to cellular hyperplasia and hypertrophy, as seen in obese with pre-diabetes. Indeed, here we provide for metformin therapy an important regulative function on the control of these pathogenic mechanisms, by significant reduction of inflammatory/oxidative stress markers, as CRP, IL6, TNFα, and nitrotyrosine (1,16), and by down regulation of circulating miR-195 and miR-27, (21,25). Finally, an inverse correlation exists between miR-195, miR-27 and cardiovascular parameters as IMT, LVEF, LVM and MPI.
Again, in our study lowest SIRT1 adipose tissue expression and highest circulating IL6 values linearly marked to highest LVM values. However, we might speculate that, patients with an induced over expression of SIRT1 and down regulation of IL6, could experience a significant reduction of LVM, as in the case of obese pre-diabetics under metformin therapy vs. placebo. Indeed, SIRT1 is implied in glucose metabolism, insulin resistance, and it plays an antiremodeling effect during adaptive cardiac hypertrophy (1). Intriguingly, metformin therapy could work as direct SIRT1 inductor (1). Therefore, we might say that metformin therapy reduces the inflammation/ oxidative stress, and secondary this might reduce myocardial wall thickness, and myocardial mass, leading to the improvement of the cardiac function in pre-diabetics obese patients (1). Therefore, the significant reduction of IL6, that could be seen as anti-inflammatory effect induced by metformin therapy, could linearly mark with reduction of LVM in obese prediabetics.

Conclusion
Abdominal fat tissue in pre-diabetics obese patients over expresses inflammatory/oxidative metabolites that are linked to reduced expression of SIRT1, and to increased expression of miR-195 and miR-27 at level of adipose abdominal tissue.
All these molecules could influence IMT, LVM, LVEF and MPI via systemic effects, that might be linked to adipose tissue and circulating expression of miR-195 and miR-27. Intriguingly, in pre-diabetics obese patients the metformin therapy might significantly reduce the inflammatory/oxidative stress metabolites, and the expression of circulating miR-195 and miR-27 at 12 months of follow-up. Notably, these effects are associated to the reduction of IMT, LVM, and MPI, and are linked to the amelioration of LVEF. Moreover, we might say that metformin therapy in addition to hypocaloric diet vs. placebo looks to be a relevant treatment to reduce hyperglycemia and insulin resistance, and to revert the systemic inflammation/oxidative stress in obese pre-diabetics via down regulation of circulating miR-195 and miR-27. Future studies are needed to further assess the effects of metformin in pre-diabetics obese patients, and its possible correlation with clinical outcomes via modulation of miRs.

Study limitations
The present study evidences few limitations, as the short duration of follow up that might affect the long term outcomes.
Again, the small sample size of pre-diabetics obese might affect the study results, and the loss of animal or cellular models could not test the human study results obtained by peripheral blood analysis (baseline and follow-up end analysis) and by direct analysis of samples by abdominal fat tissue biopsy (baseline analysis). Again, in the present study we did not use the magnetic cardiac resonance to evaluate the LVM, and so we did not compare this technique to echocardiography. On other hand, also in absence of this innovative technique for the analysis of LVM we could refer to the echocardiography, which did not show its inferiority to evaluate LVM as compared to magnetic heart resonance (12,13). Again, we did not use other techniques to evaluate the overall percentage of abdominal fat deposit for study population. On other hand, the evaluation of entire volumetry of fat tissue for each cohort of study is outside th scope of the present study. Therefore, we could conclude that further long-term studies in a larger population of pre-diabetics obese will be needed to confirm our finding and to evaluate the effects of metformin at molecular, cellular, epigenetic and clinical level. Thus, in future studies we could evaluate the effects of metformin therapy added to abdominoplastic surgery and hypocaloric diet on the abdominal fat regulation of SIRT1/miRs, and of cytokine blood levels and circulating miRs, and to translate these information in a clinical setting for pre-diabetics patients.