Background

Pregnancy complications, such as gestational diabetes mellitus (GDM), intrahepatic cholestasis of pregnancy (ICP), pre-eclampsia (PE), and pregnancy induced hypertension (PIH), affect more than one fifth of all pregnancies and jeopardize the health of mothers and their infants in a short and/or long term besides adverse perinatal outcomes [1]. GDM complicates 7–13% of pregnancies and women who experienced GDM have a 32% increased risk of venous thrombosis, a 69% increased risk of cardiovascular disease (CVD), a 89% increased risk of hypertension, a 2.1-fold increase in risk of cardiovascular and metabolic morbidity, a 4.5-fold increase in dyslipidemia risk, and a nearly tenfold higher risk of type 2 diabetes mellitus when compared to those with a normoglycemic pregnancy [2,3,4,5]. The offspring of mothers with GDM appear to be at a greater risk of developing overweight in adolescence; female infants born to the affected mothers are particularly likely to develop GDM during their pregnancies [6, 7]. ICP occurs in 6% of pregnancies, which has been associated with adverse birth outcomes for the fetus and subsequent hepatobiliary disease for the mothers [8, 9]. PE affects 2–8% of pregnancies and causes a significant proportion of maternal and neonatal morbidity and mortality [10]. Women with a history of PE have a 2.5-fold increase in risk of coronary heart disease (CHD) and a 4.2-fold increase in risk of incident heart failure [11]. PIH complicates 1–6% of pregnancies and contributes to a greater risk of CVD, CHD, and heart failure by 81%, 83%, and 71%, respectively [12]. Therefore, targeted screening and risk-reduction strategies for women at high risk of pregnancy complications may help decrease the harm to mothers and their infants.

Platelets undergo physiological changes in number and function during pregnancy. While an important role for platelet in the pathogenesis of CVD among the general population has been demonstrated, evidence is conflicting regarding the associations of pregnancy complications with maternal platelet quantity and function. For example, two studies from China and Iran reported that higher platelet count (PC) at 4–20 and 24–28 weeks of gestation was an independent predictor of GDM [13, 14]. Another two studies from China and Turkey showed that PC at 11–13 and 24–28 weeks of gestation in GDM group was significantly higher than that in non-GDM group, but was not independently associated with GDM [15, 16]. Evidence from other different ethnic studies found no differences in PC between the two groups in the first and second trimesters [17,18,19,20,21]. Similar to the case of GDM, the association and predictive efficiency between PC and PE/PIH have also been greatly inconsistent in different studies [22, 23]. In addition, a limited number of case–control studies investigated platelet indices in ICP patients and controls, and observed elevated mean platelet volume (MPV) values rather than PC in those with ICP [24,25,26,27].

A number of studies have reported that high total homocysteine (tHcy) was associated with multiple pregnancy complications, including PE, GDM, placental abruption, recurrent pregnancy loss, preterm delivery, and fetal growth restriction. However, findings from these studies also lack consistency [28]. And no studies have investigated the effects of high tHcy on ICP prevalence. It is plausible that tHcy and platelets might mutually enhance and thereby jointly affect the incidence of pregnancy complications, since they are independent risk factors of endothelial injury involved in pathogenesis of these complications [29, 30]. However, to date, little attention has been paid to evaluate the joint association of platelet and tHcy with the incidence of pregnancy complications. Therefore, large observational cohort studies focusing on the impacts of platelet and tHcy on pregnancy complications are urgently needed. The aim of this study was to examine whether an elevated frequency of pregnancy complications was evident among women with low platelet counts and high tHcy levels in a homogeneous population.

Materials and methods

Study design and data collection

An observational cohort study was conducted on 13,275 consecutive pregnant women who delivered at Changzhou Maternal and Child Health Care Hospital (a 3-A-Class Specialized Hospital) between 2016 and 2017. Women who fulfilled the following criteria were included: detailed medical records, singleton pregnancy, live birth with no birth defects, and complete laboratory tests. Women were excluded if they met these conditions: smoking or use of alcohol and illicit drugs during pregnancy; a history of pre-pregnancy diseases that affect the PC, tHcy level and pregnancy outcomes, including diabetes mellitus (type 1 or 2), chronic hypertension, chronic heart, liver and kidney diseases, immune rheumatic disease, syphilis, and thrombocytopenia. Among the 13,275 initial subjects to observe, 1722 pregnant women who presented with pre-pregnant diseases (n = 488), multiple gestation (n = 335), absence of live birth (n = 96), without platelet count and tHcy level (n = 803) were excluded from final analysis. Baseline data and pregnancy outcomes were downloaded from electronic medical record of the hospital, including maternal age, height, body weight, gravidity, parity, blood pressure, use of illicit drugs and alcohol, medical history, pregnancy complications, neonatal gender, height, body weight, and gestational age. None of the women smoked, drank alcohol, and used illegal drugs during pregnancy. At the time of admission to the hospital, blood samples were collected and transferred to the laboratory for whole blood cell analysis and biochemical test including hepatic and renal function, blood lipid, high sensitive C-reactive protein (hs-CRP), folic acid, vitamin B12, and tHcy. The results of these tests were obtained from the laboratory database. In this study, hepatic and renal function, blood lipids, hsCRP, tHcy, folic acid, and vitamin B12 levels were determined by specific automated analyzers with matching reagents, respectively (for hepatic, renal function, and blood lipids: AU5800, Beckman Coulter Inc., Japan; for hsCRP and tHcy: BN II System, Siemens Diagnostics Inc., Germany; for folic acid and vitamin B12: UniCel DxI 800 Access, Beckman Coulter Inc., USA). The whole blood cell counts were analyzed by hematology analyzer (XN550, Sysmex INC., Japan). Inter- and intra-assay coefficient of variation (CV) values for the tests in the laboratory were as follows: <2%/<2% for red blood cell (RBC), <2%/<5% for white blood cell (WBC), <5%/<8% for platelet, and <5%/<10% for tHcy, hs-CRP, folic acid, vitamin B12, hepatic and renal function, and blood lipids.

Definitions

Advance age, overweight, and obesity were defined as an age ≥ 35 years, a BMI ≥ 25 and < 30 kg/m2, and a BMI ≥ 30 kg/m2, respectively [3, 31]. Hyperhomocysteinemia, folic acid deficiency, and vitamin B12 insufficiency were defined as a tHcy > 15 μmol/L, a folic acid < 10 nmol/L, and a vitamin B12 < 148 pmol/L, respectively [32, 33]. SGA/AGA/LGA were defined as a birthweight < the 10th percentile, a birthweight ≥ the 10th and ≤ 90th percentiles, and a birthweight > the 90th percentile of gestational age specific cutoff value from the cohort, respectively [1]. PTB was diagnosed as a birth at < 37 gestational weeks [34].

As detailed from a previous study, the primary outcome to observe was pregnancy complications occurring at the time of admission for labor, including GDM, ICP, PE, and PIH [35]. All cases of pregnancy complications were adjudicated by experts in obstetrics according to a previous report [36].

Statistical analysis

Data are described as frequency (%) for categorical variables and as mean ± standard deviation (SD) for continuous variables by platelet count quartiles and two categories of tHcy levels. The odds ratios (ORs) and 95% confidence intervals (CIs) for pregnancy complications associated with PC and tHcy level were calculated using logistic regression models by adjusting for pertinent confounding factors, including maternal age, BMI, gravidity, parity, assisted reproduction, neonatal sex, and laboratory results. Similarly, the odds ratios (ORs) and 95% confidence intervals (CIs) for the specific complications across each subgroup defined by PC and tHcy level were assessed and their interactions were evaluated. Additionally, the potential effect modifications on the association between hyperhomocysteinemia and ICP due to different subgroups defined by maternal characteristics and their interactions were estimated. Data analysis was carried out using Empower software (X&Y Solutions, Inc. Boston, Massachusetts) and R statistical package (http://www.R-project.org). A P < 0.05 was denoted to be statistical significance in the analysis.

Results

Characteristics of the study population

The flow diagram of the subject to observe is presented in Fig. 1. Of the 13,275 consecutive pregnant women in the initial dataset, the final analyses were limited to 11,553 participants with the measurement of PC and tHcy level at the time of admission for labor. Of these, 1.7% (201/11,553) were defined as hyperhomocysteinemia. The prevalence of GDM, ICP, PE, and PIH in the study population was 8.4% (965/11,553), 6.2% (711/11,553), 3.4% (394/11,553), and 2.1% (245/11,553), respectively.

Fig. 1
figure 1

Flow diagram

Full size image

As listed in Table 1, the participants’ demographic characteristics and laboratory data are shown by maternal PC quartiles (Q): Q1: < 164 × 109/L; Q2: ≥ 164 to ≤ 198 × 109/L; Q3: ≥ 199 to ≤ 236 × 109/L; and Q4: > 236 × 109/L. Except for a step-wise increase in BMI, there were significant decrement trends in maternal age, the prevalence of GDM and ICP, ALT and AST levels from PC Q1 to Q4. With increasing quartile of PC, MPV and the levels of vitamin B12, folic acid, total bilirubin and creatinine increased significantly, while WBC count and hs-CRP level decreased. Significant differences without stable trends were found for systolic BP, the prevalence of PE and PTB, RBC count, and the levels of tHcy, total bile acid (TBA) total bilirubin, direct bilirubin, urea nitrogen, total cholesterol, and LDL-C between platelet count quartiles. When compared to the non-hyperhomocysteinemia group (tHcy ≤ 15 μmol/L), diastolic BP, the prevalence of ICP, PE, and PTB, platelet count, the levels of TBA, direct bilirubin, urea nitrogen, creatinine, and LDL-C, were significantly higher in the hyperhomocysteinemia group, whereas age, BMI, assisted reproduction rate, GDM prevalence, fetal birth length and weight, RBC count, the levels of vitamin B12, folic acid, hs-CRP, ALT, and HDL-C were significantly lower in the hyperhomocysteinemia group (Table 2).

Table 1 The characteristics of study population by the quartiles of maternal platelet count (n = 11,553)
Full size table
Table 2 The characteristics and laboratory results of pregnant women with and without hyperhomocysteinemia (n = 11,553)
Full size table

Associations between PC, tHcy and pregnancy complications

The associations of the prevalence of pregnancy complications with PC and tHcy level were investigated in unadjusted and adjusted logistic regression models (Table 3). Total participants were divided into four groups according to the quartiles of PC and tHcy level, respectively. Compared with the women in Q1 of platelet, women in Q3 and Q4 had substantially lower risks of GDM and ICP. After adjusted for maternal clinical characteristics and relative laboratory results, women in Q4 had still a 24% decreased risks of GDM (OR: 0.76; 95% CI: 0.62, 0.94), and a 32% decreased risks of ICP (OR: 0.68; 95% CI: 0.53, 0.87), respectively. On the contrary, adjusted models showed that women in Q4 of tHcy (> 9.38 μmol/L) had significantly increased risks of ICP, PE, and PIH by 1.9-fold (95% CI: 1.48, 2.45), 2.8-fold (95% CI: 1.95, 3.90), and 2.0-fold (95% CI: 1.31, 2.99), respectively, when compared to those in Q1. In addition, when compared to non-hyperhomocysteinemia women, hyperhomocysteinemia women had higher risks of ICP (OR: 1.69; 95% CI: 1.03, 2.78) and PE (OR: 5.09; 95% CI: 3.05, 8.47) in adjusted models.

Table 3 ORs and 95% CIs for pregnancy complications with the categories of maternal platelet count and tHcy level
Full size table

Joint influence of PC and tHcy on pregnancy complications

Table 4 quantifies the modification effects of platelet count on associations of hyperhomocysteinemia with GDM, ICP, and PE. For those with hyperhomocysteinemia (compared with non-hyperhomocysteinemia group), the ICP prevalence increased from 8.4% to 28.6% in the low PC quartile (Q1), representing an absolute risk increment of 20.2% and a relative risk increment of 3.3-fold in the adjusted model (OR: 3.34; 95% CI: 1.55, 7.17; P = 0.002). In contrast, a moderate reduction in the ICP risk was observed for the high PC quartiles (Q2– Q4) (OR: 0.70; 95% CI: 0.34, 1.41; P = 0.313). A interaction test between PC and hyperhomocysteinemia on ICP was statistically significant (P for interaction = 0.014). With regards to the GDM risk, a significant interaction between PC and hyperhomocysteinemia was observed only in the crude models (P for interaction = 0.042). The GDM prevalence decreased from 10.5% in the non-hyperhomocysteinemia women with PC Q1 to 0.6% in the hyperhomocysteinemia women with PC Q2–Q4, suggesting an absolute risk decline of 9.9% and a relative risk reduction of 95% in the unadjusted model (OR: 0.05; 95% CI: 0.01, 0.39; P = 0.004). However, the interaction test did not remain significant in the adjusted model (P for interaction = 0.09).

Table 4 Effect modification of platelet on associations between hyperhomocysteinemia and pregnancy complications
Full size table

Subgroup analysis by important covariates

To further verify that the results in Table 4 are steady to potential confounding factors, a stratified analysis of subgroups categorized by major covariates was performed, including age, BMI, parity, folic acid and vitamin B12 levels. All analysis was adjusted for age, gravidity, parity, BMI, assisted reproduction, RBC and WBC count, hs-CRP, total bilirubin, direct bilirubin, ALT, AST, urea nitrogen, creatinine, HDL-C, LDL-C, total cholesterol, triglyceride, folic acid, and vitamin B12 levels, except for the covariate that was stratified. Table 5 reveals a highly consistent model: among participants with PC Q1, regardless of subgroup, hyperhomocysteinemia resulted in a significant increment in ICP risk, with ORs ranging from 1.64 to 7.66. On the contrary, among women with PC Q2–Q4, the efficacy of hyperhomocysteinemia was greatly attenuated, with ORs ranging from 0.39 to 1.35.

Table 5 Subgroup analysis of the effect modification of platelet on associations between hyperhomocysteinemia and ICP
Full size table

Discussion

Main findings

So far as we know, this is the largest hospital-based study to report retrospective associations of pregnancy complications with platelet count and tHcy levels in a Chinese population. We observed negative associations of PC quartiles with GDM and ICP prevalence, which was 10.4% and 8.7% in Q1, and decreased to 8.6% and 5.7%, 7.7% and 5.6%, 6.7% and 4.7, in Q2, Q3, and Q4, respectively. The highest rate of GDM (10.5%) was observed in the low PC Q1 and low tHcy (tHcy ≤ 15 μmol/L) subgroup, whereas the lowest rate (0.6%) was observed in the high PC (Q2–Q4) and high tHcy subgroup. We also found a remarkable differences in the efficacy of hyperhomocysteinemia across subgroups. The greatest risk increment in ICP for those with hyperhomocysteinemia was observed in the low PC Q1 group (from 8.4% to 28.6%), a risk increment of 3.3-fold (OR: 3.34; 95% CI: 1.55, 7.17; P = 0.002). In contrast, hyperhomocysteinemia had no effect on ICP in the high PC group. Taken together, our findings suggest that pregnant Chinese women with both low PC and hyperhomocysteinemia are at highest risk for ICP, while those with high PC and hyperhomocysteinemia are at lowest risk for GDM. These results, if confirmed, could help identify those pregnant women who are at high risk of ICP and low risk of GDM.

Platelet and pregnancy complications

Previous studies have evaluated potential platelet-associated alterations in pregnancy complications. A prospective case–control study from Turkey (40 cases of ICP and 40 controls) found that MPV values were higher in patients with ICP and positively associated with D-dimer in all participants during the third trimester of pregnancy [24]. Results from another Turkish prospective case–control study (117 cases of ICP and 100 controls) reported that patients with ICP had significantly higher MPV and platelet distribution width (PDW) values and an elevated MPV was related to PTB [25]. The third retrospective case–control study from Turkey (84 cases of ICP and145 age-matched controls) further indicated that MPV can be used as an indicator of disease severity in patients with ICP [26]. In addition, Silva et al. in USA expanded the previous studies on association between platelet indices and ICP in both early and late pregnancy from 33 patients with ICP and 33 controls matched for age, parity, and race [27]. However, they found no significant differences in platelet indices (MPV/PDW/PC) between ICP and control in the first trimester and between mild and severe ICP in the third trimester. In the present cohort study, we found that women with ICP had a higher MPV (11.6 vs. 11.1 fl, P < 0.001) and a lower PC (187 vs. 200 109/L, P < 0.001), when compared to women without pregnancy complications, and there were significant decreased trends in the prevalence of ICP from PC Q1 to Q4 (8.7% vs. 5.7% vs. 5.6% vs.4.7%, P < 0.001). Similar findings were observed in women with GDM (MPV: 11.2 vs. 11.1 fl, P = 0.003; PC: 191 vs. 200 109/L, P < 0.001; 10.4% vs. 8.6% vs. 7.7% vs.6.7%, P < 0.001). In addition, our study revealed there is a significant increment in MPV and a significant decrement in PC among women with PE during late pregnancy (MPV: 11.4 vs. 11.1 fl, P = 0.003; PC: 194 vs. 200 109/L, P < 0.001), which is agreement with the findings from a recent meta-analysis [23]. Our results suggest that pregnant women with late pregnancy complications such as GDM, ICP, and PE exhibit signs of low-grade activation in the coagulation system, as evidenced by changes in platelet morphology and function. [23]. Our study also found that women with PIH had a significant increase only in MPV but not in PC when compared to women without pregnancy complications (MPV: 11.3 vs. 11.1 fl, P = 0.004; PC: 202 vs. 200 109/L, P = 0.336). This is perhaps because PE and PIH are manifestations of different severity of the same pathophysiology [37].

Hyperhomocysteinemia and pregnancy complications

Elevated tHcy level may lead to DNA damage and endothelial dysfunction by increasing oxidative stress and inflammatory response, and has been associated with numerous diseases [38]. Hyperhomocysteinemia exerts a wide range of pathological effects on maternal endothelial injury and placental vascular dysfunction by increasing the release of inflammatory cytokines from vascular endothelial cells and promoting the proliferation of vascular smooth cells. It has inconsistent associations with placenta-mediated complications [28]. In our study, pregnant women with hyperhomocysteinemia had a 5.1-fold greater risk of PE (OR: 5.09; 95% CI: 3.0, 8.4; P < 0.001) and a 1.7-fold increased risk of ICP (OR: 1.69; 95% CI: 1.03, 2.78; P = 0.038) compared to those with non-hyperhomocysteinemia. Our results are in agreement with a recent meta-analysis that revealed an increased risk of PE associated with elevated homocysteine levels [39]. However, other studies found no association [40, 41]. In addition, we showed that pregnant women with hyperhomocysteinemia had a significantly lower prevalence of GDM (2.0% vs. 8.5%, P = 0.001) compared to those with non-hyperhomocysteinemia, which is similar with the previous study by López-Quesada et al., who reported that higher tHcy levels were associated with decreased odds of GDM [42]. However, a meta-analysis and a recent case–control study showed that GDM women exhibited elevated tHcy levels, and GDM risk was 1.79-fold higher in women with high tHcy (≥ 7.29 μmol/L) relative to those with low tHcy (< 5.75 μmol/L) [43, 44]. In summary, although a number of studies have investigated the associations of pregnancy complications with tHcy levels and PC, the results remain inconclusive and inconsistent, especially for GDM and PE. The reason for discrepancy could be the difference in population frequency of the MTHFR (677 C > T) polymorphism, and in cut-off criteria of hyperhomocysteinemia in various studies. MTHFR polymorphism might contribute to moderate elevation of tHcy level [45]. Different cut-off values, including quartiles, tertiles, percentiles, and tHcy > 10 or 15 μmol/L were used to define hyperhomocysteinemia in previous studies [28]. In addition, this discrepancy could also be explained by epidemiological study designs, sample sizes, geographical location and ethnicity, the timing of sample collection, gestational weeks of study participants, and adjustment for confounding factors.

Possible mechanism linkages

The mechanism by which maternal low PC and high tHcy could jointly increase ICP prevalence remains unknown. Over the past 40 years, there is a growing evidence to document the role of hyperhomocysteinemia in causing vascular damage and promote thrombosis that triggers a coagulation process contributing to platelet activation and consumption [46]. Our results appear to be in agreement with the previous study by Kong et al., who reported a joint effect of high tHcy and low PC on increased first stroke risk [47]. Our findings support one speculation that maternal/placental endothelial damage and thrombosis were involved in the pathogenesis of ICP [9]. The elevated MPV found in women with low PC group (Q1) in the present study further supports this speculation, since the large platelet is more reactive, production of prothrombotic factors, and aggregation [29]. The findings on a joint effect of high tHcy and low PC on elevated ICP prevalence also support that a combination of platelet and tHcy could also be a marker for endothelial injury and thrombosis [47].

Strengths and limitations

This study contribute new information to the literature on the adverse effect of hyperhomocysteinemia on ICP, which could be further modified by platelet count. This large hospital-based observational cohort study ensure us to correct various important confounders, including blood lipids and hs-CRP levels, hepatic and renal function. Importantly, we adjusted the status of folic acid and vitamin B12. Finally, the present study is also one of the few to evaluate the relation between platelet count and ICP during prenatal period. However, some limitations of the present study should also be mentioned. First, this is a single-center, post-hoc analysis that can prove an association but not a causal relationship and its generalizability for other centers needs to be confirmed. Second, although multivariate-adjusted logistic regression models were performed, some uncollected data or undetected variables might affect the prevalence of pregnancy complications. For example, we did not have data on whether the participants underwent sex-selection abortions during their previous pregnancy, which may affect their platelet counts and overall health status [48, 49]. Hence, the influence of residual confounders should be considered. Third, this was a retrospective cohort study, which may introduce bias. In addition, the present study did not investigate the underlying mechanisms for the observed associations, which may necessitate additional research.

Conclusion

Among 11, 553 Chinese pregnant women in a 3-A-Class Specialized Hospital, we revealed that the subgroup with low PC and high tHcy level at the time of admission for labor has the highest risk of ICP, while the other subgroup with high PC and high tHcy level has the lowest risk of GDM. These finding, if confirmed, enable us to identify these individuals who are at high risk of ICP or low risk of GDM with a combination of platelet count and tHcy level (both tests are easy to get). Our findings would be helpful in the screening and management of pregnancy complications in Chinese populations.