A risk model of prenatal screening markers in first trimester for predicting hypertensive disorders of pregnancy

Abstract

Background

We aimed to construct a risk model to assess the diagnostic value of predicting hypertensive disorders of pregnancy (HDPs) by screening a range of prenatal markers, including pregnancy-associated plasma protein A (PAPP-A), free beta-human chorionic gonadotropin (free β-hCG), and fetal nuchal translucency (NT).

Method

We analyzed 902 women, classified into four groups: healthy gravidas (n = 680, controls), gravidas with gestational hypertension (n = 61; GH), gravidas with preeclampsia (n = 90; PE), and gravidas with severe preeclampsia (n = 71, SPE). We then compared the multiple of median (MoM) of PAPP-A, free β-hCG, and NT. A risk model was constructed and receiver operating characteristic curve (ROC) analysis was used to diagnose HDPs.

Results

Levels of PAPP-A and free β-hCG levels in the GH, PE, and SPE groups were significantly lower than those in the control group (χ² = 7.522, P = 0.001; χ² = 17.775, P < 0.001). NT did not differ significantly when compared across all four groups (χ² = 1.592, P > 0.05). When the cut-off values for PAPP-A and free β-hCG were 0.795 MoM and 1.185 MoM, the corresponding sensitivities and specificities were 0.514 and 0.635, and 0.734 and 0.450, respectively. The best risk calculation featured PAPP-A, free β-hCG, and NT; this model exhibited the highest diagnostic value in the SPE group, followed by the GH group and then the PE group.

Conclusion

The use of prenatal screening markers during early pregnancy can identify fetal aneuploidy and can also predict HDPs. The development of innovative screening strategies for gravidas and the targeted prevention of HDPs in high-risk gravidas are essential for perinatal care and early intervention, thus creating significant opportunities for predictive and preventive personalized medicine. In our study, we found that the combination of a series of prenatal screening markers in early pregnancy is better than a single marker; our data clearly demonstrate the diagnostic value of combining PAPP-A, free β-hCG, and NT for patients with SPE.

Introduction

Hypertensive disorders of pregnancy (HDPs), including gestational hypertension (GH), preeclampsia (PE), and eclampsia, are common and affect approximately 10% of pregnant women [1]. HDPs are one of the leading causes of perinatal death, although the etiology and pathogenesis underlying the condition have yet to be elucidated. PE is arguably the most representative type of HDP and is associated with hypertension, proteinuria, or other multisystem involvement, after 20 weeks of gestation with signs of damage in the target organ; the detection rate for PE is 5–8% [2–5]. Of the patients diagnosed with PE, 6.2% progress to develop severe preeclampsia (SPE) [6], 9.5% develop eclampsia, and 1% of patients die [7].

Previous research has shown that placental aging may be a potential risk factor of PE and fetal growth restriction [8] and that levels of angiogenic factors and coagulation disorders might play key roles in the mechanisms underlying PE [4]. Therefore, prenatal screening for HDPs during early pregnancy plays an important role in preventing birth defects and maternal perinatal death. Because there are different therapeutic methods for differing severities of PE and SPE, it is vital that we correctly diagnose and classify HDPs [3]. Combining early detection methodology with appropriate therapeutic interventions is vital to prevent the clinical manifestation of these diseases. Consequently, there is a clear need for clinicians to move away from the perspective of relying on a delayed interventional approach to predictive, preventive, and personalized medicine (PPPM) [9–11]. The inclusion of PPPM for maternal and neonatal health screening would provide critical information for the early detection of HDPs, including GH and PE, and allow us to improve diagnostic, preventative, and treatment methods. This practice will allow us to create a personalized approach to preventing and treating HDPs.

Serum levels of maternal pregnancy–associated plasma protein A (PAPP-A), free beta-human chorionic gonadotropin (free β-hCG), and fetal nuchal translucency (NT) are routinely screened during the first trimester of pregnancy in order to detect fetal aneuploidy [12–14]. According to recent studies, the scope for prenatal screening should be expanded to include PE and other complications in pregnancy. Focusing on individualized HDP treatment might represent the best form of care for the mother while complying with the standard treatment plan with regard to fetal safety; collectively, the combination of these two strategies would provide us with a new PPPM method for HDPs. Low doses of soluble aspirin have been recommended as a preventive treatment prior to 16 weeks of gestation, if necessary [15]. Bernardes et al. [16] reported that we should consider immediate delivery for all pregnant females with HDPs in order to reduce the risk of maternal complications; these authors also reported that the potential impact of HDPs on newborns is dependent on gestational age [16]. PAPP-A is very useful for predicting PE; by linking clinical risk factors with biochemical indicators, we might be able to improve the success rate of predicting PE [4, 17]. However, it has yet to be ascertained whether free β-hCG has predictive value for PE [18, 19]. Furthermore, while some reports claim that NT can predict PE, other reports do not [20, 21]. Previous studies have also failed to consider whether HDPs can be predicted by collectively considering PAPP-A, free β-hCG, or NT.

In the present study, we investigated serum levels of PAPP-A and free β-hCG level, along with and fetal NT thickness, in 222 pregnant women with HDPs and 680 pregnant women without HDPs. We then compared these two groups in terms of multiple of median (MoM) and area under curve (AUC) values to determine the diagnostic value of PAPP-A, free β-hCG, and fetal NT thickness during the first trimester of pregnancy. In particular, we focused on the predictive ability of different combinations of these parameters. The findings of this study are expected to increase our knowledge of the pathogenesis of HDPs and create a window of opportunity for PPPM-specific measures, including risk assessment, screening programs, and targeted prevention [9, 10].

Methods

Study design/study participants

This was a retrospective case-control study that was conducted in Hangzhou Women’s Hospital (Hangzhou Maternity and Child Health Care Hospital). We collected data from 29,096 pregnant women from January 2015 to April 2019. After excluding repeated test results, a total of 902 pregnant women were selected, including 680 HDP-free pregnant women (these women had a single live birth by spontaneous vertex delivery) and 222 pregnant women with HDPs (61 cases of GH, 90 cases of PE, and 71 cases of SPE). All subjects had a single pregnancy without complications such as primary hypertension, diabetes, or liver and kidney diseases. All subjects signed an informed consent form prior to the examinations. This study was approved by the medical ethics committee of the hospital (Approval number: 2018-004-01).

Diagnostic criteria

Diagnosis was performed based on the “Guidelines for Diagnosis and Treatment of Hypertension during Pregnancy (2015)” [22]. GH was considered if the woman was first found to have hypertension (systolic blood pressure ≥ 140 mmHg, and/or diastolic blood pressure ≥ 90 mmHg, 1 mmHg = 0.133 kPa) after 20 weeks of gestation but recovered within 20 weeks of delivery, and also if she was negative for urinary protein. PE was diagnosed if women had a SBP ≥ 140 mmHg, and/or DBP ≥ 90 mmHg after 20 weeks of gestation, accompanied by any one of the following symptoms: urinary protein ≥ 0.3 g/24 h, urinary protein/creatinine ratio ≥ 0.3, or random urinary protein ≥ (+); an abnormal change in the important organs including the heart, lung, liver, and kidney; or an abnormal change in the circulatory, digestive or nervous systems, or the involvement of any organ or system (e.g., placental-fetal unit) without proteinuria. PE was classified into either the mild form or the severe form.

Exclusion criteria

The main purpose of our study was to exclude potential interference on the detection of PAPP-A, free β-hCG, and NT caused by other diseases and factors. We therefore excluded pregnant women from our analysis for the following reasons: twin or multiple pregnancy; chronic disease history such as hypertension, heart disease, kidney disease, diabetes, hyperthyroidism, connective tissue diseases, or blood disease; smoking; in vitro insemination; a pregnancy outcome including chromosome 21-, 18-, or 13-trisomy syndrome or other birth defects; a history of immunotherapy and blood transfusion or the use of special drugs; and pregnant women with incomplete or mismatched data.

Reagents and apparatus

Throughout this study, we used a 1235 Automatic Immunoassay System (time-resolved) (PerkinElmer, Shelton, USA) equipped with PAPP-A and free β-hCG assay kits, enhancement solution and washing liquor, quality control samples, a range of standards (Wallac Oy, Turku, Finland), and a Voluson E8 ultrasonic system (GE, Boston, USA).

Sampling and screening indicators

All of the pregnant women enrolled in this study were required to undergo routine prenatal examinations, including ultrasonography. At 11 to 13 weeks and 6 days of gestation, 2 to 3 mL of venous blood was collected on an empty stomach at the designated hospital and centrifuged at 2500 rpm for 10 min after standing for 30 min. Then, the serum sample was separated, stored at 2 to 8 °C, and submitted for testing within 1 week. The serum levels of PAPP-A and free β-hCG were then assayed by time-resolved fluoroimmunoassay (DELFIA) method in accordance with the manufacturer’s guidelines. NT thickness was determined and screened according to the standards issued by the Fetal Medicine Foundation [23, 24]: a midsagittal view of the fetus was taken and in a natural posture, the maximum thickness of the translucent space between the skin and the soft tissues overlying the cervical spine was measured by magnifying the image to display the head and upper chest of the fetus only. The fetus was normal when the NT thickness was < 2.5 mm but was considered abnormal if ≥ 2.5 mm. The serum levels of PAPP-A and free β-hCG level, along with the NT thickness, were all expressed using the MoM corrected by maternal weight and gestational age.

Constructing different risk prediction models for HDP

We used the likelihood ratio construction method and Python 3.7 software (Google, USA) to construct a risk model featuring PAPP-A, free β-hCG, or NT, alone and with different combinations of these variables. The MoM values of PAPP-A, free β-hCG, and NT obeyed the multivariate normal distribution, and according to calculated risk, the modeling method used parameters that corresponded to the distribution of each index [25]. We then calculated the parameters corresponding to each indicator distribution. Then, the likelihood for the distribution was estimated using the model constructed to finally determine the risk of HDP. Using this principle, we created five different models: model 1: PAPP-A MoM alone; model 2: free β-hCG MoM alone; model 3: NT MoM; model 4: PAPP-A + free β-hCG; model 5: PAPP-A + free β-hCG + NT.

Statistical analysis

All statistical analysis was carried out with SPSS version 21.0 statistical software (IBM-SPSS, Corp, NY, USA). The one-sample Kolmogorov-Smirnov test was used to test normality. Serum levels of PAPP-A and free β-hCG were shown to exhibit a skewed distribution and were expressed as medians and percentiles [M (P_2.5, P_97.5)]. Maternal age and NT values were normally distributed and expressed as means ± standard deviations ($\overline{x}$± s). Skewed data were compared between or within groups using the Mann-Whitney U test or the Kruskal-Wallis H (K) test. Normally distributed data were compared between or within groups using the independent t test or by analysis of variance (ANOVA). Maternal weight was analyzed for risk by using multivariate logistic regression analysis. The cut-off and AUC were determined from receiver operating characteristic (ROC) curves to assess the diagnostic value of PAPP-A, free β-hCG, and NT; we then calculated the optimal cut-off, the AUC, and Youden’s index. P < 0.05 was considered to be statistically significant. The risk model with the maximal AUC and the highest sensitivity was considered to exhibit the most outstanding diagnostic value.

Results

Basic demographic indicators

The maternal weights in the GH, PE, and SPE groups were significantly higher than those in the control group (Z = 5.391, Z = 5.677, Z = 4.580, respectively; all P < 0.001). Multivariate logistic regression analysis showed that the risk factors for maternal weight in the GH, PE, and SPE group were as follows: odds ratio [OR] = 1.103 (95% confidence interval [CI]: 1.073–1.134, P < 0.001), OR = 1.093 (95%CI: 1.070–1.123, P < 0.001), and OR = 1.080 (95%CI: 1.051–1.110, P < 0.001), respectively. Maternal age and gestational age in the GH, PE, and SPE groups were not significantly different from those in the control group (χ² = 2.065, χ² = 3.799, P > 0.05) (see Table 1).

Table 1.

Basic demographic data of pregnant women in different groups at the screening during the first trimester of pregnancy

Group	n	Maternal age (year)	Maternal weight (kg)	Gestational age (day)
Control	680	28.43 ± 2.80 (28.22–28.64)	52.00 (42.21–68.99)	90.00 (79.00–97.00)
GH	61	28.77 ± 3.34 (27.92–28.64)	58.00 (43.55–88.44)	90.00 (76.20–97.00)
PE	90	28.89 ± 2.72 (28.32–29.46)	57.60 (40.41–96.18)	89.00 (80.00–97.00)
SPE	71	29.15 ± 2.68 (28.52–29.79)	56.90 (43.00–82.06)	91.00 (78.80–97.00)
χ2 or F		2. 065	69.036	3.799
P		0.103	〈0.001	0.284
Control	680	28.43 ± 2.80 (28.22–28.64)	52.00 (42.21–68.99)	90.00 (79.00–97.00)
GH + PE + SPE	222	28.94 ± 2.88 (28.56–29.32)	57.45 (43.00–87.83)	90.00 (80.00–97.00)
Z or t		2.357	8.265	0.426
P		0.019	< 0.001	0.670

Independent t test (bootstrap)/Mann-Whitney U test (Monte Carlo). Data are presented as mean ± standard deviation or median (th2.5–th97.5), as appropriate. GH, gestational hypertension; PE, preeclampsia; SPE, severe preeclampsia

Correction of MoM (replacing the original concentration) by maternal weight and gestational age (days)

$$\mathrm{MoM}\text{was defined}\mathrm{by}\text{the following formula:}\mathit{MoM}=\frac{\mathit{\text{Original Conj}}.}{\mathit{\text{Median}}}$$

“Original Conj.” represents the original value for PAPP-A, free β-hCG, and NT. “Median” represents the median of the original concentration of the corresponding indicator [26]. In order to reduce the deviation caused by gestational age and maternal weight, we ensured that we corrected the MoM value. To do this, we used the equation for the median of gestational age and maternal weights (derived by Hangzhou Women’s Hospital):

$${\mathit{GA}}_{-}\mathit{Med}=10\left(10.6589-04597\times \mathit{GA}+0.007377\times {\mathit{GA}}^{2}0.000048822\times {\mathit{GA}}^3+0.000000116899\times {\mathit{GA}}^4\right)$$

[GA: gestational age (days): Med: median]

$$\mathit{\text{maternal Weight}}\_\mathit{Med}=0.43391+\frac{37.643}{\mathit{\text{maternal weight}}}$$

The MoM value, adjusted according to the equation above, was then used to calculate the prediction risk model:

$$\mathit{\text{Adjuste}}{d}_{-}M\mathit{oM}=\frac{\mathit{MoM}}{{\mathit{GA}}_{-}\mathit{Med}\times \mathit{\text{maternal weight}}\_\mathit{Med}}$$

Comparing PAPP-A, free β-hCG, and fetal NT thickness across the four groups

The MoM for PAPP-A in the GH, PE, and SPE groups was 0.79 (0.21 to 2.84), 0.87 (0.25 to 2.22), and 0.71 (0.12 to 2.61), respectively; these MoMs were all significantly lower than that in the control group 0.95 (0.33 to 2.38; (χ² = 7.522, P = 0.001). The MoM of the SPE group was significantly lower than that in the control group (Z = 3.798, P < 0.001) and the PE group (Z = 2.132, P = 0.033). There were no statistical differences between the other groups (P > 0.05). Please refer to Table 2 and Fig. 1a.

Table 2.

PAPP-A, free β-hCG, and NT in different groups at the screening during the first trimester of pregnancy

Group	n	PAPP-A (MoM)	Free β-hCG (MoM)	NT (MoM)
Control	680	0.95 (0.33–2.38)	1.09 (0.33–3.60)	0.92 ± 0.23 (0.90–0.94)
GH	61	0.79 (0.21–2.84)	0.87 (0.36–2.92)	0.95 ± 0.24 (0.88–4.01)
PE	90	0.87 (0.25–2.22)	0.98 (0.32–3.63)	0.96 ± 0.28 (0.90–4.03)
SPE	71	0.71 (0.12–2.61)	0.82 (0.31–2.68)	0.88 ± 0.19 (0.83–0.93)
X2or F		7.522	17.775	1.592
P		0.001	〈0.001	0.190
Control	680	0.95 (0.33–2.38)	1.09 (0.33–3.60)	0.92 ± 0.23 (0.90–0.94)
GH + PE + SPE	222	0.79 (0.22–2.37)	0.88 (0.34–3.03)	0.93 ± 0.24 (0.90–0.97)
Z or t		3.599	3.968	0.747
P		< 0.001	< 0.001	0.455

Independent t-test (bootstrap)/Mann-Whitney U test (Monte Carlo). Data are presented as mean ± standard deviation or median (th2.5–th97.5), as appropriate. PAPP-A, pregnancy-associated plasma protein A; free β-hCG, free beta-human chorionic gonadotropin; NT, nuchal translucency; GH, gestational hypertension; PE, preeclampsia; SPE, severe preeclampsia; MoM, multiple of the median

Fig. 1.

Comparison of PAPP-A, free β-hCG, and fetal NT MoM among the four groups. a PAPP-A MoM, b Free β-hCG MoM. c Fetal NT MoM. PAPP-A, pregnancy-associated plasma protein A; free β-hCG, free beta-human chorionic gonadotropin; NT, nuchal translucency; GH, gestational hypertension; PE, preeclampsia; SPE, severe preeclampsia

The MoM of free β-hCG in the GH, PE, and SPE groups was 0.87 (0.36 to 2.92), 0.98(0.32 to 3.63), and 0.82 (0.31 to 2.68), respectively; these were all significantly lower than the MoM of the control group (1.09; 0.33 to 3.60; χ² = 17.775, P < 0.001). The MoMs of the GH and SPE groups were significantly lower than that in the control group (Z = 2.871, Z = 3.106, P < 0.05) but were not significantly different when compared between the other groups (P > 0.05) (see Table 2 and Fig. 1b).

As shown in Table 2 and Fig.1c, compared with the control group, the NT thickness was significantly higher in the GH and PE group, and was lower in the SPE group; however, no statistical differences were identified (χ² = 1.592, P > 0.05).

Constructing a risk prediction model using single or multiple markers

Likelihood ratio was calculated using the probability density function of the normal distribution as the score of each sample to predict the risk of HDP. The lifecycle-like risk value calculation method was used to construct our models [25] as follows:

The maternal age equation for woman at the expected date of confinement [27]:

$${\text{risk}}_{\mathrm{age}}=0.000627+{\mathit{exp}}^{-16.2395+0.286\ast \left(\mathit{age}-0.5\right)}$$

The likelihood ratio represents the likelihood for the distribution in HDP pregnant women divided by the distribution in HDP-free pregnant women, where risk_age represents the risk for maternal age at the expected date of confinement; age refers to the maternal age at the expected date of confinement.

The likelihood of a one-dimensional normal distribution was calculated as follows:

$$f\left(x\right)=\frac{1}{\sqrt{2\pi }{\sigma }^2}exp\left(-\frac{{\left(x-\mu \right)}^2}{2{\sigma }^2}\right)$$

The likelihood for a two-dimensional normal distribution was calculated as follows:

Let χ be the vector of a two-dimension normal distribution χ = (χ₁, χ₂)^T

$$f\left(x,y\right)=\frac{1}{2\pi {\sigma }_X{\sigma }_Y\sqrt{1-{\rho }^2}}exp\left(-\frac{1}{2\left(1-{\rho }^2\right)}\left[\frac{{\left(x-{\mu }_X\right)}^2}{{\sigma }_X^2}+\frac{{\left(y-{\mu }_Y\right)}^2}{{\sigma }_X^2}-\frac{2\rho \left(x-{\mu }_X\right)\left(y-{\mu }_y\right)}{{\sigma }_X{\sigma }_Y}\right]\right)$$

where σ is the standard deviation of the corresponding indicator, and ρ is the correlation coefficient between indicators, and μ is the average. χ refers to the logarithm of PAPP-A MoM while y is the logarithm of free β-hCG MoM [25, 28].

The likelihood for a three-dimensional normal distribution was calculated as follows:

Let χ be the vector of two-dimension normal distribution: χ = (χ₁, χ₂, χ₃)^T

$$f\left(x\right)=\frac{1}{\sqrt{{\left(2\pi \right)}^3·\left|\sum \right|}}exp\left(-\frac{1}{2}\left[{\left(x-\mu \right)}^{\mathrm{T}}·{\sum }^{-1}·\left(x-\mu \right)\right]\right)$$

where |Σ| is the determinant of the covariance matrix of χ, Σ⁻¹ represents the inverse matrix of the covariance matrix of χ, and μ is the average. χ refers to the logarithm of the MoM value of the corresponding indicator.

Finally, the ultimate risk of GH was calculated, as follows:

$${\mathit{\text{risk}}}_{\mathit{HDP}}=\frac{1}{{\mathit{LR}}_{\mathit{\text{multinorm}}}\times {\mathit{\text{risk}}}_{\mathit{age}}}$$

Comparing the clinical value of a single marker and multiple markers for the prediction of GH, PE, and SPE

Table 2 shows that when predicting GH, only free β-hCG alone had a certain diagnostic value; the AUC was 0.595 (95% CI: 0.527–0.664, P = 0.013). When the cut-off value was 1.185 MOM, the sensitivity and specificity for free β-hCG alone were 0.803 and 0.420, respectively. PAPP-A and NT showed no significant diagnostic value (P > 0.05). When considering different combinations, we found that the combination of PAPP-A, free β-hCG, and NT had the maximal diagnostic value (P < 0.05), followed by the combination of PAPP-A and free β-hCG (P < 0.05), as shown in Table 3 and Fig. 2a.

Table 3.

Diagnostic value of PAPP-A, free β-hCG, or NT alone and different combinations among them for GH, PE, SPE, or HDP

Group or screening indicators	n	AUC	95%CI	P	Cut-off	Sensitivity	Specificity	Youden index
GH	61
PAPP-A		0.560	0.489–0.631	0.118	0.825	0.574	0.585	0.159
Free β-hCG		0.595	0.527–0.664	0.013	1.185	0.803	0.420	0.223
NT		0.538	0.455–0.622	0.362	1.005	0.451	0.672	0.123
PAPP-A + free β-hCG		0.597	0.530–0.664	0.011	1.045	0.672	0.532	0.204
PAPP-A+ free β-hCG+ NT		0.666	0.591–0.741	< 0.001	0.716	0.882	0.366	0.249
PE	90
PAPP-A		0.521	0.453–0.590	0.505	0.445	0.178	0.916	0.094
Free β-hCG		0.537	0.473–0.601	0.249	1.155	0.667	0.427	0.094
NT		0.552	0.482–0.622	0.134	0.825	0.701	0.393	0.095
PAPP-A + free β-hCG		0.574	0.511–0.637	0.022	1.245	0.311	0.833	0.144
PAPP-A+ free β-hCG+ NT		0.605	0.540–0.671	0.002	0.921	0.552	0.600	0.152
SPE	71
PAPP-A		0.627	0.553–0.701	< 0.001	0.595	0.451	0.794	0.245
Free β-hCG		0.598	0.531–0.665	0.006	1.065	0.746	0.492	0.239
NT		0.554	0.482–0.627	0.180	0.865	0.582	0.557	0.139
PAPP-A + free β-hCG		0.650	0.576–0.723	< 0.001	1.154	0.479	0.799	0.278
PAPP-A+ free β-hCG+ NT		0.714	0.642–0.787	< 0.001	1.124	0.636	0.713	0.350
HDP = GH + PE + SPE	222
PAPP-A		0.580	0.535–0.625	< 0.001	0.795	0.514	0.635	0.149
Free β-hCG		0.589	0.546–0.631	< 0.001	1.185	0.734	0.450	0.184
NT		0.519	0.471–0.568	0.434	0.825	0.661	0.398	0.059
PAPP-A + free β-hCG		0.613	0.570–0.656	< 0.001	1.032	0.459	0.722	0.182
PAPP-A+ free β-hCG+ NT		0.626	0.580–0.673	< 0.001	1.034	0.503	0.684	0.187

PAPP-A, pregnancy-associated plasma protein A; free β-hCG, free beta-human chorionic gonadotropin; NT, nuchal translucency; GH, gestational hypertension; PE, preeclampsia; SPE, severe preeclampsia; HDP, hypertensive disorders of pregnancy

Fig. 2.

ROC curve for the diagnosis of GH, PE, or SPE. a ROC curve for the diagnosis of GH. b ROC curve for the diagnosis of PE. c ROC curve for the diagnosis of SPE. d ROC curve for the diagnosis of GH + PE + SPE. PAPP-A, pregnancy-associated plasma protein A; free β-hCG, free beta-human chorionic gonadotropin; NT, nuchal translucency; GH, gestational hypertension; PE, preeclampsia; SPE, severe preeclampsia

As shown in Table 3 and Fig. 2b, when considered alone, PAPP-A, free β-hCG, and NT alone had no predictive value for PE. However, we did note significant predictive value for the combination of PAPP-A and free β-hCG (P < 0.05), and the combination of PAPP-A, free β-hCG, and NT (P < 0.05). Of these, the combination of all three parameters showed a more superior AUC (0.605).

When used as predictors for SPE, PAPP-A or free β-hCG alone had a significant diagnostic value, with AUCs of 0.627 (95% CI: 0.553–0.701, P < 0.001) and 0.598 (95% CI: 0.531–0.665, P = 0.006), respectively. When the cut-off value was 0.595 and 1.065 MOM, the corresponding sensitivities and specificities were 0.451 and 0.794, and 0.746 and 0.492, respectively. NT did not show any significant diagnostic value (P > 0.05). Of the different combinations, the combination of PAPP-A, free β-hCG, and NT showed the maximal diagnostic value (AUC = 0.714; P < 0.001), followed by the combination of PAPP-A and free β-hCG (P < 0.001), as shown in Table 3 and Fig. 2c.

An ROC curve was generated from 222 cases of HDPs; the AUCs for PAPP-A and free β-hCG to predict HDPs were 0.580 (95% CI: 0.535–0.625, P < 0.001) and 0.589 (95%CI: 0.546–0.631, P < 0.001), respectively. When the cut-off values were 0.795 and 1.185 MOM, the corresponding sensitivities and specificities were 0.514 and 0.635, and 0.734 and 0.450, respectively. NT had no significant diagnostic value (P = 0.434). The combination of PAPP-A and free β-hCG, and the combination of PAPP-A, free β-hCG, and NT both showed significant diagnostic value for the prediction of HDPs (P < 0.001) in the following order of effect: PAPP-A + free β-hCG + NT > PAPP-A + free β-hCG > free β-hCG > PAPP-A > NT. The combination of PAPP-A, free β-hCG, and NT had the highest diagnostic value (AUC = 0.626) and was most effective for SPE, followed by GH, and then PE (see Table 3 and Fig. 2d).

Discussion

The prenatal screening markers used for detecting aneuploidy during the first trimester of pregnancy predominantly include PAPP-A, free β-hCG, and fetal NT [12, 24]. Currently, there are very few risk models that use PAPP-A, free β-hCG, or NT to predict maternal PE while screening for aneuploid fetuses. According to recent studies, the scope for prenatal screening should be expanded to include PE and other pregnancy complications [29], thus guiding the clinical use of soluble low-dose aspirin for preventive and therapeutic intervention before 16 weeks of pregnancy and reducing the incidence of perinatal complications [15].

This was a retrospective case-control study based on prenatal screening in early pregnancy. The DELFIA method was used to detect maternal serum levels of PAPP-A and free β-hCG in the three study groups (GH, PE, and SPE) and the control group. In order to compare the traditional prenatal screening protocols involving PAPP-A and free β-hCG, we established a multi-index model involving NT, PAPP-A, and free β-hCG, and calculated the risk for PE in pregnant women. Our results showed that the levels of PAPP-A and free β-hCG levels in the GH, PE, and SPE groups were significantly lower than the control group (P < 0.05). We also found that NT had no diagnostic value for the prediction of GH and PE (P > 0.05). The risk calculation model constructed using a combination of NT, PAPP-A, and free β-hCG was more effective than the original method of using NT, PAPP-A, or free β-hCG MOM alone. Furthermore, it would increase the detection rate of GH and PE and prevent the occurrence of PE. This new risk model could be used to guide medical staff with regard to the provision of low-dose aspirin as an interventional method as early as possible so as to reduce the rate of birth defects and perinatal maternal death. The prenatal screening markers that are used to detect aneuploidy during the first trimester of pregnancy are also predictive for maternal PE; consequently, this reflects the principles for early prediction and individualized intervention in PPPM [30]. Kaijomaa M. et al. [31] and Morris et al. [32] found that the OR values were 1.940 (95%CI: 1.630–2.300) and 10.9 (95%CI: 4.300–27.600) for PE when the maternal serum PAPP-A level was low during the first trimester of pregnancy. Ozdamar et al. [33] also indicated that when predicting PE, PAPP-A levels measured during the first trimester of pregnancy may be useful but free β-hCG and NT were not; there was no statistical difference in free β-hCG when compared between the PE and SPE groups. Mikat et al. [34] found that the serum levels of free β-hCG in the PE group were lower in the PE group, although no statistical differences were noted with regard to PAPP-A level (P > 0.05); these findings differed from our present findings.

Very few reports are available on the use of NT to predict PE; these existing studies tend to be controversial. During the first trimester of pregnancy, NT has been associated with abortion, fetal growth restriction, premature delivery, low birth weight, and a remarkably increased risk of PE. Moreover, the severity of HDP was consistent with a gradual increase in the average NT MoM [20]. During the second trimester of pregnancy, fetal NT cannot be used as a single predictive indicator for subsequent HDPs in pregnant women [21]. In the present study, we found that NT had no diagnostic value for screening GH and PE, although a prediction model involving PAPP-A, free β-hCG, and NT increased the AUC and sensitivity of the prediction model, thereby improving the prediction effect. This novel finding signifies that NT can represent a useful tool for PPPM by predicting the early risk of HDP.

According to our results, the use of PAPP-A alone only showed diagnostic value for SPE, with AUC of 0.627 (95% CI: 0.553–0.701, P < 0.001), but not for GH and PE (AUC = 0.560 and 0.521, respectively, P > 0.05). We also found that free β-hCG had diagnostic value for GH and SPE (AUC = 0.595 and 0.598, P < 0.05); NT did not exhibit any diagnostic value for GH, PE, and SPE (AUC = 0.538, 0.552 and 0.554, P > 0.05). Therefore, it is important to promote an opportunity for personalized medicine in PPPM in which different screening markers are used to predict different subtypes of HDPs. A previous study found that the levels of PAPP-A were negatively correlated with delayed PE (AUC = 0.751, P = 0.003) and early-onset PE (P = 0.020) with a cut-off of 0.805 MOM [35]. When the 10 th percentile of PAPP-A level, as measured during the first trimester of pregnancy, was used as the cut-off value, the levels of PAPP-A level were able to predict 70.73% of PE cases [36]; these findings were similar to those acquired during the present study. The determination of PAPP-A and free β-hCG levels helped to illuminate the clinical features and medical history of a puerpera and thus help us to improve our ability to predict PE [19]. Low levels of PAPP-A or high levels of hCG level are known to increase the risk of early-onset SPE by threefold, with an RR of 4.200 (95% CI: 3.000–5.900) and an RR of 3.300 (95% CI: 2.100–5.200); furthermore, the regular detection of PAPP-A and hCG during the first trimester of pregnancy may provide unique risk information for early-onset SPE [37].

In view of the low sensitivity of PAPP-A and its limited value for screening PE [38], this particular parameter should be combined with other sensitive markers [39] to create an effective screen. In the present study, we constructed risk models using PAPP-A, free β-hCG, or NT, either alone or in combination. Our results showed that the combination of PAPP-A, free β-hCG, and NT showed maximal predictive effect, followed by the combination of PAPP-A and free β-hCG, free β-hCG alone, PAPP-A alone, and NT alone. The risk calculation model constructed with a combination of PAPP-A, free β-hCG, and NT was the best model and had optimal diagnostic value for SPE, followed by GH and PE. Hence, by integrating the combination of PAPP-A, free β-hCG, and NT as a screening tool for HDP during antenatal care, it will be possible to generate a new approach for the early identification of normotensive pregnant women who are likely to develop PE or SPE. The requirement for biomarkers in PPPM includes the selection of predictive markers that must be carefully assessed by comparing important parameters such as sensitivity and specificity. Unfortunately, biomarkers with ideal levels of specificity and sensitivity are often difficult to find. One potential solution is to use the combined power of a large number of biomarkers which are not satisfactory when used alone [30, 40]. Intervention with PPPM will promote patient surveillance, risk stratification, optimal diagnosis, and the prediction of adverse drug to drug interactions, and will allow us to identify diseases earlier [9, 10, 40].

A previous study found that when compared with pregnant women with a normal BMI, obese pregnant women are more likely to suffer from PE (OR = 2.360, 95% CI: 1.200–4.650) [41]. In the present study, we found that the maternal weights of pregnant women in the GH, PE, and SPE groups were all significantly higher than weight of pregnant women in the control group (P < 0.001). Our multivariate logistic regression analysis identified maternal weight as a risk factor for GH, PE and SPE, with ORs of 1.103, 1.093, and 1.080, respectively (P < 0.001); however, their relative effects were limited (the 95% CIs for OR ranged from 1.000 to 1.100). Moreover, all MoM values used in this study were corrected by gestational age and maternal weight. Consequently, maternal weight had little effect on predictive ability when using PAPP-A and free β-hCG.

In a previous study, we found that low levels of maternal serum PAPP-A during the first trimester of pregnancy were predictive of PE [42] and that there was a statistically significant difference between the PE and SPE groups in terms of free β-hCG (P < 0.05). These previous findings were not consistent with those of the present study, possibly because of the lack of GH cases, an insufficient number of PE and SPE cases, and the control population (not limited to those who had experienced a single live birth by spontaneous vertex delivery).

Conclusions and expert recommendations

According to the results of this study, the levels of PAPP-A and free β-hCG in the GH, PE, and SPE groups were significantly lower than those in the control group. PAPP-A alone showed diagnostic value only for SPE, and not for GH and PE. Free β-hCG had diagnostic value for GH and SPE. NT did not have any diagnostic value for GH, PE, and SPE when used alone. The combination of PAPP-A, free β-hCG, and NT had the maximal predictive effect, followed by the combination of PAPP-A and free β-hCG, free β-hCG alone, PAPP-A alone, and NT alone. The risk calculation model constructed with the combination of PAPP-A, free β-hCG, and NT was identified as the best model and exhibited optimal diagnostic value for SPE, followed by GH and PE. NT was not diagnostic for GH and PE, although a predictive model combining PAPP-A, free β-hCG, and NT increased the AUC and sensitivity of the prediction model, thus improving the prediction effects.

Based on maternal serum PAPP-A, free β-hCG, and NT in the first trimester of pregnancy, it should be possible to screen for both fetal aneuploidy and maternal HDP at the same time. Furthermore, it should be possible to develop an innovative screening program for pregnant women in association with targeted prevention strategies for high-risk pregnant women. This will allow us to better adapt to the needs of healthcare throughout the perinatal period, and to carry out intervention in advance, thereby creating opportunities for prediction, prevention, and personalized medicine. In our study, the combination of prenatal screening markers for early pregnancy screening was better than a single marker. In particular, we found that the combination of PAPP-A, free β-hCG, and NT created a predictive model with the best diagnostic value for SPE. In the future, the prevention and treatment of HDPs must be personalized. Treatment should be developed according to the patient’s situation. However, we should be cautious to remember that this disease poses serious risk to the mother and it is therefore vital that we maximize treatment effects and minimize harmful effects on the fetus.

Abbreviations

PAPP-A

pregnancy-associated plasma protein A

free β-hCG

free beta-human chorionic gonadotropin

NT

nuchal translucency

HDP

hypertensive disorders of pregnancy

GH

gestational hypertension

PE

preeclampsia

SPE

severe preeclampsia

MoM

multiple of the median

ROC

receiver operating characteristic

AUC

area under curve

PPPM

predictive, preventive and personalized medicine

OR

odds ratio

Authors’ contribution

Yiming Chen (corresponding author) performed laboratory measurements, statistical analysis, and wrote the first draft of the manuscript. Zhen Xie, Xue Wang, Yiming Chen, and Qingxin Xiao contributed to experimental design and participated in laboratory analyses. Xiao Lu and Yezhen Shi were involved in protocol development, patient recruitment, data acquisition, and data analysis. Sha Lu and Shaolei Lv provided intellectual guidance and critically reviewed the manuscript. Yiming Chen conceived and designed the study and critically reviewed the manuscript. All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

Funding information

This study is funded by the Natural Science Foundation of Zhejiang (Grant number LGF19H040006), Zhejiang Medicine and Health Scientific Research Project (disciplinary platform-2018ZD036), Hangzhou Medicine and Health Scientific Research Project (2017A052), and Hangzhou Yuhang District Science and Technology Bureau Research Project (Medical Treatment and Public Health-2018008)

Compliance with ethical standards

Conflicts of interest

The authors declare that they have no conflict of interest.