Diabetes Mellitus and Cardiovascular Risk Assessment in Mothers with a History of Gestational Diabetes Mellitus Based on Postpartal Expression Profile of MicroRNAs Associated with Diabetes Mellitus and Cardiovascular and Cerebrovascular Diseases.

Mothers with a history of gestational diabetes mellitus (GDM) have an increased risk of developing diabetes in the future and a lifelong cardiovascular risk. Postpartal expression profile of cardiovascular/cerebrovascular disease associated microRNAs was assessed 3-11 years after the delivery in whole peripheral blood of young and middle-aged mothers with a prior exposure to GDM with the aim to identify a high-risk group of mothers at risk of later development of diabetes mellitus and cardiovascular/cerebrovascular diseases who would benefit from implementation of early primary prevention strategies and long-term follow-up. The hypothesis of the assessment of cardiovascular risk in women was based on the knowledge that a series of microRNAs play a role in the pathogenesis of diabetes mellitus and cardiovascular/cerebrovascular diseases. Abnormal expression profile of multiple microRNAs was found in women with a prior exposure to GDM (miR-1-3p, miR-16-5p, miR-17-5p, miR-20a-5p, miR-20b-5p, miR-21-5p, miR-23a-3p, miR-24-3p, miR-26a-5p, miR-29a-3p, miR-100-5p, miR-103a-3p, miR-125b-5p, miR-126-3p, miR-130b-3p, miR-133a-3p, miR-143-3p, miR-145-5p, miR-146a-5p, miR-181a-5p, miR-195-5p, miR-199a-5p, miR-221-3p, miR-342-3p, miR-499a-5p, and-miR-574-3p). Postpartal combined screening of miR-1-3p, miR-16-5p, miR-17-5p, miR-20b-5p, miR-21-5p, miR-23a-3p, miR-26a-5p, miR-29a-3p, miR-103a-3p, miR-133a-3p, miR-146a-5p, miR-181a-5p, miR-195-5p, miR-199a-5p, miR-221-3p, and miR-499a-5p showed the highest accuracy for the identification of mothers with a prior exposure to GDM at a higher risk of later development of cardiovascular/cerebrovascular diseases (AUC 0.900, p 0.001, sensitivity 77.48%, specificity 93.26%, cut off >0.611270413). It was able to identify 77.48% mothers with an increased cardiovascular risk at 10.0% FPR. Any of changes in epigenome (upregulation of miR-16-5p, miR-17-5p, miR-29a-3p, and miR-195-5p) that were induced by GDM-complicated pregnancy are long-acting and may predispose mothers affected with GDM to later development of diabetes mellitus and cardiovascular/cerebrovascular diseases. In addition, novel epigenetic changes (upregulation of serious of microRNAs) appeared in a proportion of women that were exposed to GDM throughout the postpartal life. Likewise, a previous occurrence of either GH, PE, and/or FGR, as well as a previous occurrence of GDM, is associated with the upregulation of miR-1-3p, miR-17-5p, miR-20a-5p, miR-20b-5p, miR-29a-3p, miR-100-5p, miR-125b-5p, miR-126-3p, miR-130b-3p, miR-133a-3p, miR-143-3p, miR-145-5p, miR-146a-5p, miR-181a-5p, miR-199a-5p, miR-221-3p, and miR-499a-5p. On the other hand, upregulation of miR-16-5p, miR-21-5p, miR-23a-3p, miR-24-3p, miR-26a-5p, miR-103a-3p, miR-195-5p, miR-342-3p, and miR-574-3p represents a unique feature of aberrant expression profile of women with a prior exposure to GDM. Screening of particular microRNAs may stratify a high-risk group of mothers with a history of GDM who might benefit from implementation of early primary prevention strategies.

We showed that previous occurrence of either GH, PE, and/or FGR might predispose a proportion of women to later development of cardiovascular and cerebrovascular diseases due to the presence of alterations in the expression of cardiovascular and cerebrovascular disease-associated microRNAs in their whole peripheral blood [1][2][3].
Gestational diabetes mellitus (GDM) is defined as carbohydrate intolerance that develops during pregnancy, usually during second or third trimester of gestation [4]. Women with GDM have an increased risk of developing diabetes (predominantly type 2 diabetes) later in life. It is estimated that up to 70% of women with GDM will develop diabetes within 22-28 years after pregnancy [4][5][6][7].
The incidence of GDM increases with the same risk factors seen for type 2 diabetes such as obesity, sedentary lifestyle, and increasing reproductive age of women [4][5][6][7].
The initial criteria for diagnosis of GDM were established more than 40 years ago to identify women at a high risk for development of diabetes after pregnancy [8][9][10]. The two-step screening approach to testing for GDM is currently recommended by The American College of Obstetricians and Gynecologists [4] and The International Association of Diabetes and Pregnancy Study Groups (IADPSG) [10]. Overall, using the proposed IADPSG criteria, 18% of pregnant women in the United States are identified as having GDM [10].
The goal of the current study was to evaluate the risk of later development of diabetes mellitus, and cardiovascular and cerebrovascular diseases based on an epigenetic profile of cardiovascular/cerebrovascular disease associated microRNAs in whole peripheral blood of young and middle-aged mothers with a history of gestational diabetes mellitus (GDM) 3-11 years after the delivery. The hypothesis of the assessment of cardiovascular risk in women was based on the knowledge that a series of microRNAs play a role in the pathogenesis of diabetes mellitus and cardiovascular/cerebrovascular diseases (Table 1)  .
MicroRNAs, small noncoding RNAs, are known to regulate gene expression at the posttranscriptional level by blocking translation or degrading of target messenger RNA [171].

Results
MicroRNA gene expression was compared between mothers after normal and GDM-complicated pregnancies in maternal whole peripheral blood 3 to 11 years postpartum. Just the results, when the complicated cases reached a statistical significance, are presented below.
The ROC curve analysis confirmed a significant upregulation of these particular microRNAs for mothers with a history of GDM when the comparison to the controls was performed ( Figure 1).

Expression Profile of MicroRNAs Associated with Diabetes Mellitus and Cardiovascular/Cerebrovascular Diseases in Mothers after GDM Pregnancies with Regard to the Treatment Strategies (Diet Only and/or Diet and Therapy)
Concurrently, it was observed that the expression of miR-1-3p (p < 0.001, p = 0.013), miR-16-5p (p < 0.001, p = 0.057), miR-17-5p (p < 0.001, p = 0.035), miR-20b-5p (p < 0.001, p = 0.002), miR-21-5p , and miR-499a-5p (p < 0.001, p < 0.001) differed significantly or showed a trend toward statistical significance between the groups of mothers affected with GDM regardless of the treatment strategies (diet only and/or the combination of diet and therapy) and the controls ( Figure 4). Nevertheless, miR-133a-3p was upregulated just in the group of patients previously affected with GDM on diet only (p = 0.001) (Figure 4).
Due to a low number of women on the diet and therapy (n = 18) in our studied group of patients previously affected with GDM, we decided not to perform ROC curve analysis for particular groups of patients regarding the GDM treatment strategies (diet only and/or combination of diet and therapy).

Information on MicroRNA-Gene-Biological Pathway Interactions
The extensive file of predicted targets of all microRNAs aberrantly expressed in whole peripheral blood of mothers with a history of GDM indicates that a large group of genes is involved in biological pathways related to insulin signaling, type 1 diabetes mellitus, and type 2 diabetes mellitus (Tables 2-4).    The miRWalk database and the predicted target module were used to provide information on predicted interactions between appropriate microRNAs and specific genes involved in human biological pathways such as insulin signaling, type 1 diabetes mellitus, and type 2 diabetes mellitus.

Discussion
Postpartal expression profile of cardiovascular/cerebrovascular disease associated microRNAs was assessed 3-11 years after the delivery in whole peripheral blood of young and middle-aged mothers with a prior exposure to GDM with the aim to identify a high-risk group of mothers at risk of later development of diabetes mellitus and cardiovascular and cerebrovascular diseases who would benefit from implementation of early primary prevention strategies and long-term follow-up. The hypothesis of the assessment of cardiovascular risk in women was based on the knowledge that a serious of microRNAs play a role in the pathogenesis of diabetes mellitus and cardiovascular/cerebrovascular diseases (Table 1)  .
Screening based on the combination of these particular microRNAs was superior over using individual microRNAs, since it showed the highest accuracy for mothers with a history of GDM (AUC 0.900, p < 0.001, sensitivity 77.48%, specificity 93.26%, cut off > 0.611270413). It was able to identify 77.48% mothers with an increased cardiovascular risk at 10.0% FPR.
Subsequently, epigenetic profile was compared between groups with regard to the treatment strategy (GDM on diet only, GDM on the combination of diet and therapy). The upregulation or trend towards upregulation of almost all studied microRNAs (with the exception of miR-133a-3p) was present in both groups of mothers affected with GDM regardless of the treatment strategies.
Previously, we demonstrated that pregnancy-related complications, such as gestational hypertension, preeclampsia, and/or fetal growth restriction, were associated with postpartum alterations in gene expression of cardiovascular/cerebrovascular disease-associated microRNAs [1][2][3].
Moreover, the compilation of data resulting from our previous and current studies showed that the upregulation of miR-17-5p, miR-20b-5p, miR-29a-3p, and miR-126-3p was a mutual phenomenon observed in women with a prior exposure to GH, severe PE, and/or GDM [1][2][3].
Alike, as in women affected with severe PE and/or early PE, upregulation of miR-133a-3p was also present in women affected with GDM. Moreover, a history of severe PE and/or GDM was associated with upregulation of miR-130b-3p [1][2][3].
Moreover, a set of circulating microRNAs associated with cardiovascular/cerebrovascular diseases (miR-16-5p, miR-17-5p, miR-20a-5p, miR-29a-3p, miR-125b-5p, and miR-195-5p) was reported by several independent studies [20,21,80,92,143,172,173] to be upregulated during various gestational ages in serum or plasma of mothers with already-diagnosed GDM or mothers who were destined to develop GDM later during pregnancy. The data resulting from these prenatally ongoing studies and interaction network analysis [20,21,80,92,98,125,143,172,173] may support our current finding referring to upregulated expression profile of these particular microRNAs in whole peripheral blood of women with a prior exposure to GDM.
In addition, from the portfolio of microRNAs we tested, just miR-16-5p, miR-17-5p, miR-20a-5p, and miR-29a were identified by other investigators as the most promising biomarkers of GDM or the best predictors of GDM, respectively [21,173], which is consistent with our current finding, since during the course of postpartum screening, miR-16-5p (20.72%), miR-17-5p (21.62%), and miR-29a-3p (21.62%) upregulation persisted in a larger proportion of women with a prior exposure to GDM.
In compliance with our study, which showed a low sensitivity of miR-20a-5p (10.81%) to select women with a higher cardiovascular risk after GDM gestation based on its upregulated profile, miR-20a-5p also added little value as the predictor of GDM due to its low predictive value based on its upregulated profile (ROC, AUC just 0.740) [21,173].
It is obvious that changes in epigenome induced by GDM, which are persistently present in maternal circulation, may cause later development of diabetes mellitus and cardiovascular and cerebrovascular diseases in women with a prior exposure to GDM.
However, most of microRNAs, which were the subjects of our interest, have not yet been observed by other investigators to be dysregulated in maternal circulation (peripheral blood, plasma, or serum samples) of women with clinically established GDM or before the onset of GDM [20,21,37,80,92,143,149,[172][173][174][175][176]. So, it is feasible that epigenetic profiles of a series of microRNAs have also been changing with time by force of various circumstances as a result of the interaction between genetic and environmental factors [177].

Participants
The study included a prospectively collected cohort of Caucasian mothers with a history of GDM (n = 111) and age-matched mothers after normal course of gestation (n = 89). In-person visit was conducted 3-11 years after the pregnancy ended. Of the 111 GDM pregnancies, 93 were on diet only and 18 were on the combination of diet and therapy (17 patients required insulin administration and in 1 patient metformin, an oral hypoglycemic agent, was prescribed).
The clinical characteristics of mothers after GDM-complicated pregnancies are presented in Tables 5-7. Gestational diabetes mellitus is defined as any degree of glucose intolerance with onset or first recognition during pregnancy [10, 178,179]. The International Association of Diabetes and Pregnancy Study Groups' (IADPSG) recommendations on the diagnosis and classification of hyperglycemia in pregnancy were followed [10]. The first screening phase detects, during the first trimester of gestation, women with overt diabetes (fasting plasma glucose level is ≥7.0 mmol/L) and women with GDM (fasting plasma glucose level ≥ 5.1 mmol/L < 7.0 mmol/L). The second screening phase, 2-h 75-g OGTT, at 24-28 weeks of gestation is done in all women not previously found to have overt diabetes or GDM, and identifies GDM if fasting plasma glucose level is ≥5.1 mmol/L or 1-h plasma glucose is ≥10.0 mmol/L or 2-h plasma glucose is ≥8.5 mmol/L [10].
Patients demonstrating other pregnancy-related complications, such as gestational hypertension, preeclampsia, fetal growth restriction, premature rupture of membranes, in utero infections, fetal anomalies or chromosomal abnormalities, and fetal demise in utero or stillbirth, were not involved in the study.
Written informed consent was provided for all participants included in the study. The study was approved by the Ethics Committee of the Institute for the Care of the Mother and Child, Prague, Czech Republic (grant no. AZV 16-27761A, long-term monitoring of complex cardiovascular profile in the mother, fetus, and offspring descending from pregnancy-related complications, date of approval: 28 May 2015) and by the Ethics Committee of the Third Faculty of Medicine, Prague, Czech Republic (grant no. AZV 16-27761A, long-term monitoring of complex cardiovascular profile in the mother, fetus, and offspring descending from pregnancy-related complications, date of approval: 27 March 2014).  Angina or heart attack in a first degree relative before the age of 60 years 2 (2.22%) 6 (6.45%) 2 (11.11%) --Atrial fibrillation 0 (0%) 0 (0%) 0 (0%) --DM type I 0 (0%) 0 (0%) 0 (0%) --DM type II 0 (0%) 0 (0%) 0 (0%) --  Data are presented as mean ± SE for continuous variables and as number (percent) for categorical variables. Statistically significant results are marked in bold. Continuous variables were compared using Mann-Whitney test. p-value 1 , the comparison among normal pregnancies and GDM on diet; p-value 2 , the comparison among normal pregnancies and GDM on diet and therapy, respectively. Categorical variables were compared using a chi-square test. GA, gestational age; SBP, systolic blood pressure; DBP, diastolic blood pressure; CS, caesarean section. Data are presented as mean ± SE for continuous variables and as number (percent) for categorical variables. Statistically significant results are marked in bold. Continuous variables were compared using Mann-Whitney test. p-value 1 , the comparison among normal pregnancies and GDM on diet; p-value 2 , the comparison among normal pregnancies and GDM on diet and therapy, respectively. Categorical variables were compared using a chi-square test. GA, gestational age; SBP, systolic blood pressure; DBP, diastolic blood pressure; CS, caesarean section.

Processing of Samples
Homogenized cell lysates were prepared immediately after collection of whole peripheral blood samples (EDTA tubes, 200 µL) using QIAamp RNA Blood Mini Kit (Qiagen, Hilden, Germany, no: 52304).
Total RNA was extracted from homogenized cell lysates stored at −80 • C using a mirVana microRNA Isolation kit (Ambion, Austin, USA, no: AM1560) and followed by an enrichment procedure for small RNAs. To minimize DNA contamination, the eluted RNA was treated for 30 min at 37 • C with 5 µL of DNase I (Thermo Fisher Scientific, CA, USA, no: EN0521). A RNA fraction highly enriched in short RNAs (< 200 nt) was obtained. The concentration and quality of RNA was assessed using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, USA). If the A(260/280) absorbance ratio of isolated RNA was 1.8-2.0 and the A(260/230) absorbance ratio was greater than 1.6, the RNA fraction was pure and used for the consecutive analysis.

Reverse Transcriptase Reaction
Individual microRNAs were reverse transcribed into complementary DNA (cDNA) in a total reaction volume of 10 µL using microRNA-specific stem-loop RT primers, components of TaqMan MicroRNA Assays (Table 8)  TaqMan PCR conditions were set up as described in the TaqMan guidelines for a 7500 Real-Time PCR system. All PCRs were performed in duplicates with the involvement of multiple negative controls such as NTC (water instead of cDNA sample), NAC (nontranscribed RNA samples), and genomic DNA (isolated from equal biological samples), which did not generate any signal during PCR reactions. The samples were considered positive if the amplification signal occurred at Ct < 40 (before the 40th threshold cycle).
The expression of particular microRNA was determined using the comparative Ct method [180] relative to normalization factor (geometric mean of two selected endogenous controls) [181]. Two noncoding small nucleolar RNAs (RNU58A and RNU38B) were optimal for qPCR data normalization in this setting. They demonstrated equal expression between children descending from normal and complicated pregnancies. RNU58A and RNU38B also served as positive controls for successful extraction of RNA from all samples and were used as internal controls for variations during the preparation of RNA, cDNA synthesis, and real-time PCR.
A reference sample, RNA fraction highly enriched for small RNAs isolated from the fetal part of one randomly selected placenta derived from gestation with normal course, was used throughout the study for relative quantification.

Statistical Analysis
Data normality was assessed using the Shapiro-Wilk test [182]. Since our experimental data did not follow a normal distribution, microRNA levels were compared between groups using the Kruskal-Wallis one-way analysis of variance with post hoc test for the comparison among multiple groups. The significance level was established at a p-value of p < 0.05.
Receivers operating characteristic (ROC) curves were constructed to calculate the area under the curve (AUC) and the best cut-off point for particular microRNA was used in order to calculate the respective sensitivity at 90.0% specificity (MedCalc Software bvba, Ostend, Belgium). For every possible threshold or cut-off value, the MedCalc program reports the sensitivity, specificity, likelihood ratio positive (LR+), likelihood ratio negative (LR−).
To select the optimal combinations of microRNA biomarkers, logistic regression was used (MedCalc Software bvba, Ostend, Belgium). The logistic regression procedure allowed us to analyses the relationship between one dichotomous dependent variable and one or more independent variables. Another method to evaluate the logistic regression model makes use of ROC curve analysis. In this analysis, the power of the model's predicted values to discriminate between positive and negative cases is quantified by the area under the ROC curve (AUC). To perform a full ROC curve analysis, the predicted probabilities are first saved and next used as a new variable in ROC curve analysis. The dependent variable used in logistic regression then acts as the classification variable in the ROC curve analysis dialog box.
Box plots encompassing the median (dark horizontal line) of log-normalized gene expression values for particular microRNAs were generated using Statistica software (version 9.0; StatSoft, Inc., USA). The upper and lower limits of the boxes represent the 75th and 25th percentiles, respectively. The upper and lower whiskers indicate the maximum and minimum values that are no more than 1.5 times the span of the interquartile range (range of the values between the 25th and the 75th percentiles). Outliers are marked by circles and extremes by asterisks.

Information on MicroRNA-Gene-Biological Pathway Interactions
MiRWalk database (available: http://www.umm.uni-heidelberg.de/apps/zmf/mirwalk/) and the Predicted Target module were used to provide information on predicted targets of those microRNAs that have been found to be dysregulated in whole peripheral blood of patients with a history of GDM. Only those predicted targets involved in particular human biological pathways (insulin signaling pathway, type 1 diabetes mellitus pathway, and type 2 diabetes mellitus pathway) are reported [183].
MiRWalk is a comprehensive database that provides information on microRNAs from human, mouse, and rat on their predicted and/or validated target genes. MiRWalk2.0 not only documents miRNA binding sites within the complete sequence of a gene, but also combines this information with a comparison of binding sites resulting from 12 existing miRNA-target prediction programs (DIANA-microTv4.0, DIANA-microT-CDS, miRanda, mirBridge, miRDB4.0, miRmap, miRNAMap, DoRiNA, PicTar2, PITA, RNA22v2, RNAhybrid2.1, and Targetscan6.2) to build novel comparative platforms of binding sites for the promoter (4 prediction datasets), cds (5 prediction datasets), 5'-(5 prediction datasets), and 3'-UTR (13 prediction datasets) regions. Information on miRNA-target interactions on 2035 disease ontologies (DO), 6727 human phenotype ontologies (HPO) and 4980 OMIM disorders is available. It provides possible interactions between microRNAs and genes associated with 597 KEGG, 456 Panther, and 522 Wiki pathways.
Any changes in epigenome (upregulation of miR-16-5p, miR-17-5p, miR-29a-3p, and miR-195-5p) that were induced by GDM-complicated pregnancy are long-acting and may predispose mothers affected with GDM to later development of diabetes mellitus and cardiovascular/cerebrovascular diseases. In addition, novel epigenetic changes (upregulation of a series of microRNAs) appeared in a proportion of women that were exposed to GDM throughout the postpartal life.
Overall, previous occurrence of GDM, therefore, predisposes a proportion of affected individuals to later development of diabetes mellitus and cardiovascular/cerebrovascular diseases. Screening of particular microRNAs may stratify a high-risk group of mothers with a history of GDM that might benefit from implementation of early primary prevention strategies.
Consecutive large-scale studies, including the groups of mothers on diet only and on the combination of diet and therapy, are needed to verify the findings resulting from this particular pilot study.

Patents
National patent granted-Industrial Property Office, Czech Republic (Patent no. 308178). International patent filed-Industrial Property Office, Czech Republic (PCT/CZ2019/050051).  Acknowledgments: All procedures were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. We thank the staff of the Institute for the Care of Mother and Child (Katerina Hamplova, Katerina Mackova, Radek Cabela, and Adam Krasny) for assistance with collection of patients´anamnesis. We also thank Jana Kumprichtova, Sarka Stranska, Veronika Marvanova, and Andrea Semencova for assistance with biological sample collection and processing.

Conflicts of Interest:
The authors declare no conflict of interest.