Predictive values of various serum biomarkers in women with suspected preeclampsia: A prospective study

Abstract Background Preeclampsia (PE) prediction has been shown to improve the maternal and fetal outcomes in pregnancy. We aimed to evaluate the PE prediction values of a series of serum biomarkers. Methods The singleton pregnant women (20–36 gestational weeks) with PE‐related clinical and/or laboratory presentations were recruited and had the blood drawn at their first visits. The following markers were tested with the collected serum samples: soluble fms‐like tyrosine kinase 1 (sFlt‐1), placental growth factor (PlGF), thrombomodulin (TM), tissue plasminogen activator inhibitor complex (tPAI‐C), complement factors C1q, B, H, glycosylated fibronectin (GlyFn), pregnancy‐associated plasma protein‐A2 (PAPP‐A2), blood urea nitrogen (BUN), creatinine (Cre), uric acid (UA), and cystatin C (Cysc). Results Of the 196 recruited subjects, 25% (n = 49) developed preeclampsia before delivery, and 75% remained preeclampsia negative (n = 147). The serum levels of sFlt‐1, BUN, Cre, UA, Cysc, and PAPP‐A2 were significantly elevated, and the PlGF level was significantly decreased in the preeclampsia‐positive patients. In the receiver operating characteristics (ROC) analyses, the area under the curves were listed in the order of decreasing values: 0.73 (UA), 0.67 (sFlt‐1/PlGF), 0.66 (Cysc), 0.65 (GlyFn/PlGF), 0.64 (PAPP‐A2/PlGF), 0.63 (BUN), 0.63 (Cre), and 0.60 (PAPP‐A2). The positive predictive values of these serum markers were between 33.1% and 58.5%, and the negative predictive values were between 80.9% and 89.5%. Conclusions The serum markers investigated in current study showed better performance in ruling out than ruling in PE. Absence of pre‐defined latency period between blood draw and the onset of PE limits the clinical utility of these markers.


| INTRODUC TI ON
Preeclampsia, one of the most common complications during pregnancy, is estimated to have an incidence rate of 2-8% worldwide 1 and can lead to serious maternal or fetal morbidity and mortality if not managed properly. 2 Extensive studies have been performed to reveal the clinical value of preeclampsia prediction, 3 of which three major beneficial effects may be concluded: identifying highrisk patients who require close monitoring to decrease potential complications, reducing the necessity of antenatal care of low-risk populations, and promoting the development of early therapeutic interventions of preeclampsia. 4 According to a systematic review published in 2019 about preeclampsia prediction models, 5 most previous studies (87%) have conducted risk assessments during the first trimester of pregnancy.
In first trimester preeclampsia screening, the prediction model is recommended to combine maternal background risk factors, imaging tests, and serum biomarkers to increase sensitivity and reduce the false detection rate. 5 Due to relatively low positive predictive values (PPV) (8-33%) during first trimester screening for preeclampsia, 6 false-positive patients who do not develop preeclampsia may be exposed to unnecessary tests and prophylactic interventions with no benefit.
Recently, some researchers evaluated preeclampsia predictive markers in patient groups >20 gestational weeks and identified high-risk factors or clinical/laboratory signs of preeclampsia. 7,8 This type of testing scenario with patients suspected of preeclampsia development has been proven effective, especially in the studies on soluble fms-like tyrosine kinase-1 (sFlt-1) and placental growth factor (PlGF). 7,9 For example, a sFlt-1/PlGF ratio of 38 was found to effectively exclude preeclampsia with a negative predictive value (NPV) of 99.3% 7 ; however, a similar prospective study with Chinese pregnant women has yet to be published. Interestingly, a recent longitudinal study conducted in Singapore suggested significant differences in the PlGF and sFlt-1 concentrations during pregnancy between different Asian ethnicities (Chinese, Malay, and Indian). 10 In a retrospective study with 118 Chinese singleton pregnancies who had been diagnosed with preeclampsia, the sFlt-1/PlGF ratio was shown to be an efficient marker in differentiating preeclampsia and predicting the timing of delivery for preeclampsia pregnancies. 11 In addition to sFlt-1/PlGF, a series of other serum biomarkers have been shown to be associated with the occurrence or outcomes of preeclampsia. For instance, the maternal pregnancy-associated plasma protein-A2 (PAPP-A2) serum concentration was found to be upregulated in preeclampsia patients, resulting in local activation of insulin-like growth factor (IGF) signaling pathways. 12 This finding implied that PAPP-A2 may be upregulated in preeclampsia to compensate for IGF binding protein 5-mediated pathway. 12 The maternal serum glycosylated fibronectin (GlyFn) was reported to be elevated in all three trimesters of preeclampsia patients; the test was further recommended as a point-of-care biomarker to quickly determine risk for preeclampsia and for poor maternal and fetal outcomes among preeclamptic patients. 13 In uteroplacental thrombosis, which is one of the major mechanisms of preeclampsia, several thrombotic and fibrinolytic factors including circulating soluble thrombomodulin (TM) and tissue plasminogen activator (tPA) were found to be elevated in PE and correlated with the severity of proteinuria. 14,15 The relative changes of these coagulation factors reflected endothelial disturbance in preeclampsia, and they were recommended for future evaluation as potential risk biomarkers. 14,15 The dysregulation of complement pathways also contributes to the development of preeclampsia. The differential expression of complement factors C1q, B, and H were found in specific trimesters of severe preeclampsia patients, suggesting promising values as diagnostic markers for severe preeclampsia. 16 The presence of proteinuria, which is a hallmark in preeclampsia, indicates that renal deficiency contributes significantly to the pathogenesis of preeclampsia. 17 The renal function markers, such as uric acid (UA), blood urea nitrogen (BUN), creatinine (Cre), and cystatin C (Cysc), have been found to be disturbed in preeclampsia patients 17 and their performance in predicting preeclampsia after 20 weeks of gestation still lacks validation studies.
In summary, even though the panel of serum markers described above have been studied in a broad context of preeclampsia, whether or not these biomarkers will add value in preeclampsia pre-

| Subjects
The enrollment criteria for women with suspected preeclampsia are described as follows. The recruited singleton pregnant women were at least 18 years old and between 20 and 36 gestational weeks (GWs), as pregnancies of >36 GWs are likely to be subjected to delivery of fetus if preeclampsia is diagnosed or the blood pressure (BP) is severely elevated. In addition, one of the following recruiting criteria had to be met for patient enrollment: new onset of hypertension (systolic BP >120 and <160 mmHg and/or diastolic BP Beijing Obstetrics and Gynecology Hospital, with follow-up for the presence ("preeclampsia-positive" group) or absence ("preeclampsianegative" group) of preeclampsia until delivery.
The preeclampsia diagnosis was determined with the diagnostic criteria proposed by the 2019 ACOG Practice Bulletin, 6 in which preeclampsia was defined as gestational hypertension (systolic/diastolic blood pressure ≥140/90 mmHg) in previously normotensive women accompanied by proteinuria (urine protein ≥300 mg/24 h) or endorgan damage after 20 weeks of gestation.

| Serum samples, reagents, and methods
The maternal blood from each participant (3 mL) was drawn when they were enrolled and left to clot for 30 min followed by centrifugation for 10 min at 2300 g. The serum aliquots (1 mL) were separated and stored at −80°C until being tested.

| Statistical analysis
Data analysis was performed using statistical software SPSS 23.0.
The Kolmogorov-Smirnov test was used to evaluate the normality of the data distribution. Numerical values were expressed as the mean and standard deviation (SD) for variables with normal distribution and as the median and percentiles for nonnormally distributed F I G U R E 1 Schematic diagram depicting patient recruitment and study design. BUN, blood urea nitrogen; Cre, creatinine; Cysc, cystatin C; GlyFn, glycosylated fibronectin; PAPP-A2, pregnancy-associated plasma protein-A2; PlGF, placental growth factor; sFlt-1, soluble fms-like tyrosine kinase 1; TM, thrombomodulin; tPAI-C, tissue plasminogen activator inhibitor complex; UA, uric acid data. Comparisons between the two groups were performed using the t test (for normal distribution) or Mann-Whitney U test (for non-normal distribution). Categorical variables were expressed as frequencies and proportion; comparisons between the two groups were tested by chi-square test. The receiver operating characteristics (ROC) curve was used to analyze the predictive values of the markers for preeclampsia. The comparison of before and after adjusted area under curves (AUCs) was assessed using the algorithm developed by DeLong et al. 19 The adjusted factors included age, prepregnancy BMI, parity, and underlying chronic diseases in the ROC analyses. Sensitivity, specificity, and cutoff values were reported when Youden's index was at the maximum or specificity was fixed at 90%. A binary logistic regression analysis was performed including age, prepregnancy BMI, parity, underlying chronic diseases, and each of the markers. All statistical tests were two-sided, and p < 0.05 was considered statistically significant.

| RE SULTS
A flowchart depicting patient recruitment and the study design is presented in Figure 1 preeclampsia negative for remainder of the pregnancy. Since the focus was narrowed down on the patients with relevant symptoms, the prevalence of preeclampsia (25%) in the current cohort was much higher than that of the overall pregnancies (5.2%) during the study period in our institute. The demographic data of all the recruited subjects are available in Table S1. As shown in Table S2 Table 1, except for underlying chronic diseases, no significant difference was found in maternal age, prepregnancy BMI, blood sampling gestational weeks (GW), gravidity, or parity between the two groups; however, patients with  The underlying chronic diseases that may be considered as risk factors for preeclampsia development were presented as percentages. The number in the parentheses after percentage figures indicated the number of the subjects that had the corresponding underlying chronic disease when they were enrolled.

TA B L E 1 Demographic data for the recruited subjects
underlying diseases (hypertension being the most prominent) that are associated with preeclampsia development were more likely to develop preeclampsia during pregnancy (p = 0.010, Table 1). On average, the time interval from serum collection to preeclampsia occurrence was 7 weeks in the preeclampsia-positive patients (Table 1).
To evaluate the preeclampsia predictive values of the selected markers in present study, we first compared their serum concentrations that were determined using our laboratory devices and platforms. As shown in Table 2 then subjected to ROC analyses. As shown in Figure 2 and Table 3,    (Table S5). Moreover, the NPV (94.4%) was close to that previously reported, 7 the sensitivity (40.0%), specificity (83.4%), and PPV (16.7%) were much lower with our cohort (Table S5), suggesting that ethnicity may be a confounding factor for the application of the sFlt-1/PlGF ratio and the cutoff value needs to be further optimized for Chinese populations before clinical implementation. The hemostatic factors such as TM and tPAI-C were found to be related to the incidence and severity of PE decades ago. 21 In preeclampsia patients, significant endothelial disturbance and procoagulant potential, along with aberrant expression of these hemostatic factors, were reported in previous studies 14,15 ; however, whether they could be useful in preeclampsia prediction has yet to be investigated. With our cohort, no difference was observed between the preeclampsia-positive and negative groups, indicating their limited values in preeclampsia prediction ( Table 2).
The excessive activation and poor regulation of the complement system at the maternal-fetal interface contributes to the development of preeclampsia. 22 More importantly, a recent study by Jia et al. showed that the complement factors C1q, B, and H were able to diagnose early-onset severe preeclampsia with AUCs of 0.81, 0.74, and 0.68, respectively. To further evaluate their potential utility in preeclampsia prediction, the circulating levels were measured in the present study. No significant difference was found between the preeclampsia-positive and preeclampsia-negative groups ( Table 2).
The two glycoproteins that were included in our testing panel, GlyFn and PAPP-A2, have been widely studied in preeclampsia. In a 2020 study by Huhn et al., 8 the GlyFn level in a prospective cohort identified with preeclampsia-specific high-risk factors was reported to show satisfactory preeclampsia prediction with an AUC of 0.94 in the ROC analysis. In our study, GlyFn was also increased, although not significantly, in the patient group that developed preeclampsia. This apparent discrepancy may be introduced by differences in the sample size, patient recruiting criteria or testing methodology of the two studies. The glycoprotein PAPP-A2, involved in cleaving insulin-like growth factor binding protein in the placenta, was found to be helpful in diagnosing 12 and predicting preeclampsia. 8 In our study, the PAPP-A2/ PlGF ratio (p = 0.003) was found to be a better marker than PAPP-A2 (p = 0.032) alone (Table 2), with an adjusted AUC of 0.72 ( Table 3).
The common renal function tests such as BUN, Cre, UA, and Cysc were shown to be potentially valuable in preeclampsia diagnosis and prediction. For example, the BUN 23 and BUN/Cre ratio 24 were both found increased in the preeclampsia patients compared with those in the normal controls. Cysc was also found to be elevated in preeclampsia patients 25 and was able to predict preeclampsia in combination with neutrophil gelatinase-associated lipocalin (AUC = 0.88). 26 In a prospective study with a relatively large cohort (n = 9522) by Rezk et al., 27 the serum UA level during the second trimester was found to be a useful preeclampsia predictor for women at moderate or low risk. More interestingly, an elevated UA level was later reported to be a risk factor for women with gestational hypertension to develop preeclampsia and deliver small-for-gestational-age infants. 28 We observed similar findings in which UA was the most promising predictor with the greatest AUCs (0.73 and 0.77, before and after adjustment, respectively) in the ROC analyses ( Figure 2 and Table 3).
In conclusion, in a prospective cohort suspected of preeclampsia development, the angiogenic modulators sFlt-1 and PlGF; the renal function markers BUN, Cre, UA, and Cysc; and the glycoprotein PAPP-A2 were significantly altered between the two groups. Last but not least, the serum markers invested in our study showed better performance in ruling out than ruling in preeclampsia. Absence of pre-defined latency period between blood draw and the onset of preeclampsia or delivery significantly limited the clinical utility of these markers.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
All authors have certified the author list and the contribution description. All authors have read and approved the submitted TA B L E 3 Comparison of AUCs before and after adjusted for demographic data of recruited subjects

DATA AVA I L A B I L I T Y S TAT E M E N T
The baseline characteristics of the recruited subjects are provided as Table S1. Alternatively,