Gestational diabetes mellitus is associated with antenatal hypercoagulability and secondary hyperfibrinolysis: a case control study of Chinese women

Background: To determine the relationship between gestational diabetes mellitus (GDM) and coagulation/fibrinolysis abnormality in antenatal Chinese women. Methods: Case control study. Fifty women had GDM and 132 did not (the NGDM group) grouping by the International Association of Diabetes and Pregnancy Study Groups (IADPSG) criteria. Maternal plasma biochemistry and previous medical history were collected from perinatal health records. Antenatal coagulation/fibrinolysis activity(CFA) parameters was assessed using thromboelastography and routine CFA parameters respectively. Univariate and multiple regression analyses were used to evaluate the associations between GDM and CFA parameters. Results: The women with GDM were significantly older than those without GDM (30.3 vs. 28.6 years, P=0.012). Compared with the NGDM group, the GDM group had a significantly higher prevalence of cesarean delivery (56.0% vs. 37.9%, P=0.027) and higher values of fibrinogen (FIB) (4.7vs. 4.3 g/L P=0.001), activated partial thromboplastin time (APTT) (30.9 vs. 29.5 seconds P=0.010).There were no significant differences in the prevalence of maternal thrombotic events or neonatal events.GDM was significantly associated with higher APTT (β 1.41seconds, 95% CI 0.29–2.53), FIB (β 0.38g/L, 95% CI 0.14–0.61), and percentage reduction in clot lysis after 30 min(LY30)(β 1.14%, 95% CI 0.15–2.13) after adjustment for potential confounding factors. Conclusions: GDM is significantly associated with hypercoagulability and secondary hyperfibrinolysis in these antenatal Chinese women.


Background
Coagulation/fibrinolysis activity (CFA) represents a dynamic balance that is of vital importance to safe childbirth. Coagulability gradually increases from the first to third trimesters [1] in pregnant women, becoming higher than in normal women [2] . Consistent with this, late pregnancy is characterized not only by changes in hormone secretion and liver metabolism, but higher coagulation indices [3] . Moreover, this hypercoagulability during pregnancy is aggravated in the presence of diabetes [4,5] . Excessive hypercoagulability is associated with adverse pregnancy outcomes, including stroke and deep venous thrombosis (DVT) of the lower extremities [6] . However, neither a nationwide study conducted in Denmark [7] , nor another conducted in the Arabian Gulf [8] , identified GDM as a risk factor for stroke or DVT. Therefore, we hypothesized that secondary hyperfibrinolysis might coexist with, and gradually worsen, hypercoagulability in GDM.
Many methods have been used to assess CFA. Thromboelastography (TEG) is a relatively new technique that evaluates whole-blood hemostatic properties in real time, and has been shown to be a reliable method of urgently assessing CFA in pregnant women [9] . In addition, fibrinogen (FIB), activated partial thromboplastin time (APTT), prothrombin time (PT), international normalized ratio of prothrombin time (PT-INR), and D-Dimer (DD) are conventionally used to assess CFA. In view of the changes in CFA during late pregnancy, we speculated that there might be differences in a number of these parameters between women with GDM and normal pregnant women. However, the results of platelet activation, fibrinolytic activity, and conventional assays of CFA have rarely been compared in women with GDM [10] , and there has been no assessment of the relationship between GDM and CFA, assessed using TEG, in pregnant Chinese women. Therefore, in this study, we aimed to evaluate the relationship between GDM and measures of CFA, after adjustment for potential confounding factors.

Study design and study population
The study was a case control study. All procedures performed in the study involving the Two hundred women who had never been smokers were included, 18 of whom were excluded because of severe liver disease (three), diabetes mellitus diagnosed before pregnancy (eleven), receive heparin treatment (two), and thrombocytopenia (two). Thus, 182 individuals remained for analysis (a flow chart is presented in Figure 1).

Materials and methods
Baseline information, including maternal weight before pregnancy, weight gain during pregnancy, and previous medical history, were collected from the patients' health records.
Before delivery, the participants underwent a CFA assessment, which comprised the measurement of conventional CFA parameters (FIB, PT, APTT, PT-INR, and DD) and TEG.
Venous blood was collected in the morning, on an empty stomach, and anti-coagulated with 0.38% sodium citrate. The blood was then centrifuged at 1500g (2500 rpm) for 20 min to prepare plasma for APTT, PT, TT, FIB, and DD analysis, which was performed using a Stago STAR-R analyzer (Diagnostica STAGO, France). The remaining blood was placed into EDTA tubes for platelet counting using a Huma-Count hematology analyzer (Human GmbH, Germany).
TEG was performed using TEG-5000 Hemostasis analyzers (Haemonetics Corp, Braintree, MA) according to the manufacturer's guidelines [12][13][14] , by two certified laboratory technicians. The TEG-5000 machines underwent quality control assessments every 8 hours. The parameters measured were R (sec) ; K (sec); α-angle; MA (mm) and LY30 (%) ( Figure 2). R represents "reaction time", an indicator of clotting time, and is the time to initial fibrin formation (to 2 mm amplitude). K represents "kinetic time", an indicator of clot kinetics, and is the speed at which specific clot strength is reached (period for amplitude to increase from 2 to 20 mm). The α-angle is a measure of clot kinetics, indicative of the rate of fibrin accumulation and cross-linking. MA represents "maximum amplitude", an indicator of clot strength. LY30 is the percentage reduction in amplitude 30 min post-MA, which is a measure of the degree of fibrinolysis.

Statistical analysis
No multiple imputation was performed because < 5% of the data was missing.
The procedure of statistical analysis included three steps. Firstly, baseline characteristics of participants were presented based on the following principles (cases were categorized into GDM and NGDM): (1) continuous variable was expressed as mean ± standard deviation (normal distribution) or medium (inter-quartile range) (Skewed distribution P=0.010) were significantly higher in the GDM than those in the NGDM group. Other parameters including PT, TT, PT-INR, and DD didn't show the significant differences between the two groups, as were the TEG parameters, including LY30.

Univariate analysis
The results of the univariate analyses were shown in Table 2. GDM was shown to be significantly positively associated with LY30, FIB, and APTT. Furthermore, Platelet, ALT, ALP, GGT, and Maternal FPG significantly positively correlated with FIB. Maternal weight gain during pregnancy, a previous history of abortion, and a previous history of full-term birth were negatively associated with FIB. Finally, ALT significantly positively correlated with LY30.

Analysis of the multiple regression equations for the relationships between FIB,APTT,LY30 and GDM
We constructed three models to analyze the independent effects of GDM on FIB,APTT,LY30, after adjustment for potential confounding factors. The effect values (β) and 95% confidence intervals (CIs) of the two models are shown in

Discussion
In this case control study, both the conventional CFA measures FIB and APTT, and the newer measure indicator LY30, were found to be significantly associated with GDM, and in the fully-adjusted regression model, these associations remained stable. In women with GDM, APTT was 1.41 sec longer (β1.41, 95% CI 0.29-2.53), FIB was 0.38 g/L higher (β0.38, 95% CI 0.14-0.61), and LY30 was 1.14% higher (β1.14, 95% CI 0.15-2.13), which is indicative of both hypercoagulability and secondary hyperfibrinolysis in this group of Chinese women with GDM.
Maintenance of a balance between coagulation and fibrinolysis is of great importance for the perinatal safety of both mothers and infants [17] . A physiologic hypercoagulable state has been demonstrated during pregnancy and the peripartum period [17,18] , which helps to protect women from excessive bleeding during childbirth. Nevertheless, in developed nations, the leading cause of maternal death is thromboembolic disease [19][20][21] .
Thrombotic events during pregnancy, including maternal DVT(deep venous thrombosis) and pregnancy-related cerebral venous sinus thrombosis, are responsible for increasing levels of maternal mortality worldwide [22,23] . Previous studies have shown that this hypercoagulability is exacerbated in pregnant women with GDM [4,24,25] and that there is a significantly higher frequency of DVT in pregnant women with type 1 diabetes mellitus [26] . However, in a nationwide study conducted in Denmark, an unadjusted model showed a higher risk of thrombosis in patients with GDM, but in a model adjusted for confounding factors this association disappeared [7] . In previous studies [7,27] and our study, no significant differences were identified in the frequency of maternal thrombotic events in GDM group compared with NGDM group, which seems to be inconsistent with the hypercoagulability identified.
The balance between coagulation and fibrinolysis determines whether hemorrhage, thrombosis, or neither occurs during the perinatal period. FIB is one of the conventional measures of coagulability during the perinatal period, as well as being a biomarker of inflammation, and is a complex glycoprotein synthesized by hepatocytes [28] . FIB increases during normal pregnancies, but also in pregnancy-related complications such as GDM [29,30] , reaching significantly higher values in patients with GDM than those in healthy pregnant women [24,30] . In the present study, FIB was also higher in GDM patients than that in the NGDM group, confirming a tendency towards hypercoagulability in the former.
The shortening of APTT in type 2 diabetic patients has been described as indicating a high risk of a hypercoagulable state [31] . It has also been shown that the increase in fibrinogen and the reduction in fibrinolytic activity in GDM are more similar to the changes present in type 2 than in type 1 diabetic patients [32] , and it is thought that a shorter APTT is a marker of venous thromboembolism risk [33,34] . Furthermore, PT and APTT, measured between the 20th and 24th gestational weeks, were significantly lower in patients with GDM than in healthy people [29] . In another study, the mean APTTs of non-diabetic, treated, and untreated type 2 diabetic patients were 32.8 ± 4.12 sec, 34.4 ± 5.3 sec, and 25.4 ±8.5 sec, respectively, which showed that APTT was lower in diabetic patients and that this defect was normalized by diabetes treatment [31] . This might help to explain the significantly higher APTT in GDM patients identified in the present study. The prenatal maternal FPG (4.8±1.2 mmol/l) and HbA1c (6.5±0.7%) values in the GDM group show that glucose levels were well controlled in the present study. In addition, the significantly lower maternal weight gain during pregnancy in the GDM group is indicative of good weight control in our recruited subjects. Finally, the characteristic increases in FIB and APTT, indicating hypercoagulability and secondary hyperfibrinolysis, might help to partially explain why there were no significant differences in the prevalence of some adverse maternal events, such as maternal thrombosis, between these "healthy" GDM women and NGDM women.
Fibrinolytic activity is lower during normal pregnancies and systemic fibrinolysis returns rapidly to normal after delivery, according to previous studies [35,36] . After delivery, both the coagulatory and fibrinolytic systems of pregnant women are activated for at least 2 weeks [37] . However, few studies have focused on the fibrinolytic activity in GDM. Recently, TEG has been used to monitor the CFA of pregnant women in real time, to study the human CFA process and the sequence of events involved [38] . The TEG parameters measured in the present study have further indicated the secondary hyperfibrinolysis.
There were no significant differences in TEG.R, TEG.K, TEG.MA, LY30, or α-angle between the GDM and NGDM groups, which is consistent with the findings of a previous report that showed no significant differences in thromboelastographic parameters between 50 GDM patients and a control group(75 women of normal pregnancies at the same period) [39] .
However, after adjustment for potential confounders, GDM was significantly associated with higher value of LY30, which further suggested the existence of secondary hyperfibrinolysis in these pregnant women with GDM. The coexistence of hypercoagulability and secondary hyperfibrinolysis might help to partially explain why there was no significant increase in the number of thrombotic events in these GDM patients. However, the mechanism involved and its impact on the children of these patients requires further study.
The clinical value of this study is two-fold. Firstly, we have shown significant associations between GDM and higher APTT, LY30, and FIB. Secondly, CFA of the pregnant women was assessed by using both conventional CFA measures and TEG at the same time point and the results were consistent. The findings of this study might be helpful in guiding further research regarding biomarkers that could sensitively predict the risk of thrombotic events in GDM patients. Additionally, our study was a single-center study, indicating that the study population was more homogeneous. The maternity center is located in a small city in China that has relatively few population migrations, and most of the participants in this study were local residents. As previously reported [34] , the influence of ethnicity on coagulation might exist that could be excluded in our study.
However, this study still had a number of limitations. Firstly, because the data were collected in a single local population, the conclusion should be confirmed by further research and multi-center study. Secondly, we excluded pregnant women that had undergone heparin treatment and that had a history of other diseases, such as severe liver disease, which might have affected CFA. Thus, the conclusions of this study do not apply to pregnant women with these kinds of diseases. Thirdly, in our study, IADPSG criteria was adopted which was recently widely used in China [40] . However, it was reported recently that an increase in the prevalence of GDM when the IADPSG 1-step approach was adopted without an increase of maternal or neonatal outcomes compared with ACOG 2-step strategy [41] .

Consent for publication
During the analysis, the data were anonymized, and therefore the requirement for informed consent was waived by the local ethics committee with no impact on health outcome.

Availability of data and material
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
All authors declare that they have no competing interests.

Funding
The development of the analytic methods and software used in this work was supported by