First-Trimester Screening for HELLP Syndrome—Prediction Model Based on MicroRNA Biomarkers and Maternal Clinical Characteristics

We evaluated the potential of cardiovascular-disease-associated microRNAs for early prediction of HELLP (hemolysis, elevated liver enzymes, and low platelets) syndrome. Gene expression profiling of 29 microRNAs was performed on whole peripheral venous blood samples collected between 10 and 13 weeks of gestation using real-time RT-PCR. The retrospective study involved singleton pregnancies of Caucasian descent only diagnosed with HELLP syndrome (n = 14) and 80 normal-term pregnancies. Upregulation of six microRNAs (miR-1-3p, miR-17-5p, miR-143-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) was observed in pregnancies destined to develop HELLP syndrome. The combination of all six microRNAs showed a relatively high accuracy for the early identification of pregnancies destined to develop HELLP syndrome (AUC 0.903, p < 0.001, 78.57% sensitivity, 93.75% specificity, cut-off > 0.1622). It revealed 78.57% of HELLP pregnancies at a 10.0% false-positive rate (FPR). The predictive model for HELLP syndrome based on whole peripheral venous blood microRNA biomarkers was further extended to maternal clinical characteristics, most of which were identified as risk factors for the development of HELLP syndrome (maternal age and BMI values at early stages of gestation, the presence of any kind of autoimmune disease, the necessity to undergo an infertility treatment by assisted reproductive technology, a history of HELLP syndrome and/or pre-eclampsia in a previous gestation, and the presence of trombophilic gene mutations). Then, 85.71% of cases were identified at a 10.0% FPR. When another clinical variable (the positivity of the first-trimester screening for pre-eclampsia and/or fetal growth restriction by the Fetal Medicine Foundation algorithm) was implemented in the HELLP prediction model, the predictive power was increased further to 92.86% at a 10.0% FPR. The model based on the combination of selected cardiovascular-disease-associated microRNAs and maternal clinical characteristics has a very high predictive potential for HELLP syndrome and may be implemented in routine first-trimester screening programs.


Identification of Risk Factors for the Development of HELLP Syndrome
The clinical characteristics of cases (HELLP-syndrome pregnancies) and controls (normal-term pregnancies) are outlined in Table 2.
The following independent risk factors for the development of HELLP syndrome with or without PE were identified at early gestational stages: the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, the presence of mutations in trombophilic genes, and the positivity of first-trimester PE/FGR screening by the Fetal Medicine Foundation (FMF) algorithm.

Altered Expression Profiles of MicroRNAs during the First Trimester of Gestation in Pregnancies Developing HELLP Syndrome
Whole peripheral blood first-trimester expression profiles of microRNAs were compared between pregnancies that developed HELLP-syndrome and normal-term pregnancies.

The Prediction Model for HELLP Syndrome Based on Selected Maternal Clinical Characteristics Only
The prediction model for HELLP syndrome based on the combination of six selected maternal clinical characteristics only (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes) identified pregnancies developing HELLP syndrome with a relatively high accuracy (AUC 0.862, p < 0.001, 71.43% sensitivity, 97.50% specificity, cut-off >0.1244). A total of 71.43% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) also revealed 71.43% of HELLP pregnancies with a 10.0% FPR (AUC 0.849, p < 0.001, 71.43% sensitivity, 97.50% specificity, cutoff >0.1406) ( Figure 3). ROC analysis: the combination of six microRNAs only (miR-1-3p, miR-17a-5p, miR-143a-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p). With a 10.0% FPR, 78.57% of pregnancies developing HELLP syndrome were correctly identified. CI, confidence interval; +LR, positive likelihood ratio; -LR, negative likelihood ratio; FPR, false-positive rate.

The Prediction Model for HELLP Syndrome Based on Selected Maternal Clinical Characteristics Only
The prediction model for HELLP syndrome based on the combination of six selected maternal clinical characteristics only (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes) identified pregnancies developing HELLP syndrome with a relatively high accuracy (AUC 0.862, p < 0.001, 71.43% sensitivity, 97.50% specificity, cut-off >0.1244). A total of 71.43% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) also revealed 71.43% of HELLP pregnancies with a 10.0% FPR (AUC 0.849, p < 0.001, 71.43% sensitivity, 97.50% specificity, cut-off > 0.1406) ( Figure 3).    sensitivity, 97.50% specificity, cut-off >0.1244). A total of 71.43% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) also revealed 71.43% of HELLP pregnancies with a 10.0% FPR (AUC 0.849, p < 0.001, 71.43% sensitivity, 97.50% specificity, cutoff >0.1406) (Figure 3).

The Full Prediction Model for HELLP Syndrome Based on the Combination of Six MicroRNAs and Selected Maternal Clinical Characteristics
The full prediction model for HELLP syndrome based on the combination of six mi-croRNAs (miR-1-3p, miR-17a-5p, miR-143a-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) and six selected maternal clinical characteristics (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes) identified pregnancies developing HELLP syndrome with relatively high accuracy (AUC 0.979, p < 0.001, 100.0% sensitivity, 86.25% specificity, cut-off > 0.0494). A total of 85.71% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) revealed 92.86% of HELLP pregnancies at a 10.0% FPR (AUC 0.975, p < 0.001, 92.86% sensitivity, 92.50% specificity, cut-off > 0.1110) (Figure 4). fied pregnancies developing HELLP syndrome with relatively high accuracy (AUC 0.979, p < 0.001, 100.0% sensitivity, 86.25% specificity, cut-off >0.0494). A total of 85.71% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) revealed 92.86% of HELLP pregnancies at a 10.0% FPR (AUC 0.975, p < 0.001, 92.86% sensitivity, 92.50% specificity, cut-off >0.1110) (Figure 4). . ROC analysis: the combination of six microRNAs (miR-1-3p, miR-17a-5p, miR-143a-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) and six maternal clinical characteristics (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes). In the case of the combination of six microRNA biomarkers and seven maternal clinical characteristics, the positivity of first-trimester PE/FGR screening by FMF was added. With a 10.0% FPR, 85.71% and 92.86% of pregnancies developing HELLP syndrome were correctly identified. CI, confidence interval; +LR, positive likelihood ratio; -LR, negative likelihood ratio; BMI, body mass index; PE, pre-eclampsia; FGR, fetal growth restriction; FMF, Fetal Medicine Foundation; FPR, false-positive rate. . ROC analysis: the combination of six microRNAs (miR-1-3p, miR-17a-5p, miR-143a-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) and six maternal clinical characteristics (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes). In the case of the combination of six microRNA biomarkers and seven maternal clinical characteristics, the positivity of first-trimester PE/FGR screening by FMF was added. With a 10.0% FPR, 85.71% and 92.86% of pregnancies developing HELLP syndrome were correctly identified. CI, confidence interval; +LR, positive likelihood ratio; -LR, negative likelihood ratio; BMI, body mass index; PE, pre-eclampsia; FGR, fetal growth restriction; FMF, Fetal Medicine Foundation; FPR, false-positive rate.

Discussion
Initially, we focused on the identification of risk factors associated with later development of HELLP syndrome occurring separately or accompanying severe PE.
We identified an increased incidence of HELLP syndrome in patients with already diagnosed autoimmune diseases, such as SLE, APS, SS, RA, T1DM, and coeliac disease, which has not yet been reported.
Furthermore, we observed a higher incidence of HELLP syndrome in patients undergoing an infertility treatment by assisted reproductive technology (ART), which was also reported as an independent risk factor for the onset of hypertensive disorders during pregnancy, such as GH or PE [52][53][54] and FGR [55].
Qin et al. [56] pointed to the fact that singleton pregnancies undergoing assisted reproductive technologies are at a higher risk of adverse outcomes and recommended that they be managed as high-risk pregnancies. The ART singleton pregnancies had a significant risk of pregnancy-induced hypertension, GDM, placenta previa, placental abruption, antepartum hemorrhage, postpartum hemorrhage, polyhydramnios, oligohydramnios, cesarean sections, preterm birth, small for gestational age, perinatal mortality, and congenital malformation [56].
Similarly to other researchers [32][33][34][35][36][37], we confirmed that a history of HELLP syndrome and/or PE in a previous gestation represents a risk factor predisposing to the development of HELLP syndrome.
Furthermore, we demonstrated that the presence of trombophilic gene mutations is more frequent in pregnancies developing HELLP syndrome. This finding is congruent with the observations of Muetze et al. [39], who reported that mutation in factor V Leiden is associated with HELLP syndrome in women of Caucasian descent.
All of the maternal risk factors identified by our research group were placed in a model together with maternal age and BMI values at early gestational stages to assess their common predictive potential for later development of HELLP syndrome. This prediction model for HELLP syndrome identified pregnancies developing HELLP syndrome with relatively high accuracy, since it was able to reveal 71.43% of cases with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) did not yield a better detection rate, since it was able to detect the same proportion of pregnancies with HELLP syndrome (71.43% of cases with a 10.0% FPR).
Our model based on maternal risk factors only showed better performance than the logistic regression model demonstrated previously by Oliveira et al. [36]. This model was based on racial origin, nulliparity, and the occurrence of HELLP syndrome and PE in a previous gestation, and reached a detection rate of 55.0% of cases only, with a 10.0% FPR.
A recently developed neuro-fuzzy model for HELLP syndrome identification and prediction [47] also showed a lower discrimination power than our novel model based on six or seven basic maternal clinical characteristics, including maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, and the presence of mutations in trombophilic genes, or the positivity of first-trimester PE/FGR screening by FMF [57][58][59][60].
Afterwards, we were interested to see whether there was any other potential way to improve the detection rate of our novel HELLP predictive model based on maternal clinical characteristics only. Therefore, we evaluated the predictive potential of microRNAs that play a crucial role in the development and maintenance of homeostasis in the cardiovascular system and in the pathophysiology of cardiovascular and cerebrovascular diseases (Table 1) [51].
Gene expression of preselected microRNAs associated with the cardiovascular system was retrospectively studied in peripheral blood during the first trimester in pregnancies subsequently developing HELLP syndrome and in normal-term pregnancies selected as a matched control group based on the equality of the period of biological sample storage and gestational age at sampling.
Currently, the upregulation of six microRNAs associated with the cardiovascular system (miR-1-3p, miR-17-5p, miR-143-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) was observed during the early gestational stages in pregnancies developing HELLP syndrome. The combination of these six microRNA biomarkers only identified pregnancies developing HELLP syndrome with relatively high accuracy. A total of 78.57% of pregnancies destined to develop HELLP syndrome was revealed at early gestational stages with a 10.0% FPR. This is an optimistic result if we take into consideration that any knowledge of maternal clinical characteristics is required to achieve such a high discrimination power. Nevertheless, the availability of early screening for HELLP syndrome in clinical practice depends on the successful validation of microRNA biomarkers in consecutive prospective cohort studies and the acquisition of CE and IVDR certifications. The advantage of such a screening based on microRNA biomarkers only is that additional information about maternal characteristics is not needed.
Interestingly, apart from HELLP-syndrome pregnancies, miR-1-3p also showed an altered early expression profile in pregnancies destined to develop SGA [49] and GDM [51]. MiR-143-3p displayed an aberrant expression profile in pregnancies developing PE [48]. Besides HELLP syndrome, an altered early expression profile of miR-146a-5p appeared in pregnancies developing PE [48] and FGR or SGA [49]. An aberrant early expression profile of miR-181a-5p was present in most pregnancies regardless of the type of pregnancyrelated complication: GH or PE [48], FGR or SGA [49], GDM [51], or HELLP syndrome. MiR-499a-5p expression profiles were also observed to be dysregulated in early stages of gestation in pregnancies destined to develop GDM [51].
When these six cardiovascular-disease-associated microRNAs were added to the predictive model based on six maternal clinical characteristics identified by our research group as risk factors for the onset of HELLP syndrome, the discrimination power increased to 85.71% with a 10.0% FPR. The addition of another maternal clinical characteristic to the prediction model (the positivity of first-trimester PE/FGR screening by FMF) did not increase the AUC but significantly increased the sensitivity with a 10.0% FPR. Using this approach, 92.86% of HELLP pregnancies were finally detected.
To our knowledge, no studies on the early prediction of HELLP syndrome during the first trimester through screening of extracellular microRNAs in maternal body fluids (plasma/serum) or peripheral blood samples are currently available.
Just one study describing the identification of differentially expressed microRNAs in serum samples of patients with clinical manifestations of HELLP syndrome is available. Upregulation of miR-122, miR-758, and miR-133a was detected in a group of patients with HELLP syndrome [61]. Concerning miR-122 and miR-758, we did not examine their expression levels in maternal peripheral venous blood leukocytes. In addition, we studied miR-133a-3p, whose expression levels did not differ in early stages of gestation between pregnancies developing HELLP syndrome and those with normal courses of gestation delivering at term.

Patient Cohort
The peripheral blood sampling was performed within the framework of the firsttrimester prenatal screening between 10 and 13 gestational weeks within the period November 2012-May 2018. In total, 4187 samples were collected from Caucasian singleton pregnancies. In the end, 3028 pregnant women delivered on site. In all, 14 of 3028 pregnant women were diagnosed with HELLP syndrome.
Complete (full) HELLP syndrome developed in 3 cases and incomplete (partial) HELLP syndrome was diagnosed in 11 cases.
The most severe form of HELLP syndrome (also termed class 1) was present in 3 cases, the moderate form (class 2) in 4 cases, and the mild form (class 3) in 7 cases.
Seven pregnancies were diagnosed with HELLP syndrome only, and seven pregnancies had HELLP syndrome associated with severe PE.
Clinical management guidelines issued by the American College of Obstetricians and Gynecologists (ACOG) including diagnostic criteria for pre-eclampsia were followed [62].
The selection of controls was performed with respect to the equality of gestational age at the time of sample collection and the period of storage of biological samples. The control group consisted of 80 normal-term pregnancies. The control group delivered healthy newborns after 37 gestational weeks with weights over 2500 g.
In detail, leukocyte lysates were prepared from 200 µL peripheral blood samples using the QIAamp RNA Blood Mini Kit (Qiagen, Hilden, Germany) and were stored in a mixture of RLT buffer and β-mercaptoethanol (β-ME) at −80 • C.
The MirVana microRNA Isolation kit (Ambion, Austin, USA) was used to isolate RNA fractions highly enriched for small RNAs.
RT and real-time qPCR analyses were performed via TaqMan MicroRNA Assays (Applied Biosystems, Branchburg, NJ, USA) on a 7500 Real-Time PCR System under standard TaqMan PCR conditions. The relative expression of microRNA genes was assessed using the delta-delta Ct method [63]. The endogenous controls (RNU58A and RNU38B) were used to normalize microRNA gene expression data [64,65].

Statistical Analysis
MicroRNA gene expression was compared between cases and controls using the Mann-Whitney test.
To select the best microRNA combinations, logistic regression with subsequent ROC curve analyses was applied (MedCalc Software bvba, Ostend, Belgium). This statistical approach was also applied to develop novel early predictive models for HELLP syndrome based on the combination of appropriate microRNAs only, maternal clinical characteristics only, and the combination of appropriate microRNAs and maternal clinical characteristics (MedCalc Software bvba, Ostend, Belgium).

Conclusions
Consecutive, large-scale retrospective analyses have to be performed to verify the reliability of our novel early predictive model for HELLP syndrome based on the combination of microRNAs associated with the cardiovascular system (miR-1-3p, miR-17a-5p, miR-143a-3p, miR-146a-5p, miR-181a-5p, and miR-499a-5p) and maternal clinical characteristics (maternal age and BMI values at early gestational stages, the presence of any autoimmune disease, an infertility treatment by assisted reproductive technologies, the occurrence of HELLP syndrome and/or PE in a previous gestation, the presence of mutations in trombophilic genes, and, alternatively, the positivity of first-trimester PE/FGR screening by FMF).
The model based on the combination of six cardiovascular-disease-associated mi-croRNA biomarkers and six maternal clinical characteristics has a very high discrimination power. It is able to detect 85.71% of cases with a 10.0% FPR. The addition of another maternal clinical characteristic to this particular prediction model (the positivity of first-trimester PE/FGR screening by FMF) may significantly increase the detection rate to 92.86% of cases.
In addition, consecutive prospective cohort studies are needed to validate the suggested early predictive model for HELLP syndrome. The availability of early screening for HELLP syndrome in clinical practice depends on the successful validation of microRNA biomarkers in consecutive prospective cohort studies and the acquisition of CE and IVDR certifications. Only high-risk pregnancies identified firstly by the early predictive model based on maternal characteristics may be further screened using microRNA biomarkers to achieve a reasonable cost-benefit ratio for the early predictive model for HELLP syndrome.

Patents
National patent application, Industrial Property Office, Czech Republic (patent No. PV 2022-505). Institutional Review Board Statement: Approval of the study was gained initially from the Ethics Committee of the Third Faculty of Medicine, Charles University (Implication of placental specific microRNAs in maternal circulation for diagnosis and prediction of pregnancy-related complications, date of approval: 7 April 2011). Ongoing approvals for the study were gained from the Ethics Committee of the Third Faculty of Medicine, Charles University (Long-term monitoring of complex cardiovascular profiles in mother, foetus, and offspring descending from pregnancy-relatedcomplications, date of approval: 27 March 2014) and the Ethics Committee of the Institute for the Care of the Mother and Child, Charles University (Long-term monitoring of complex cardiovascular profiles in mother, foetus, and offspring descending from pregnancy-related-complications, date of approval: 28 May 2015, approval No.: 1/4/2015). All procedures were in compliance with the Helsinki Declaration of 1975, as revised in 2000.

Informed Consent Statement:
Informed consent for participation in the study was gained from patients during the first trimester of gestation when the collection of peripheral blood samples for the first-trimester screening was held. Informed consent was signed by all pregnant women involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to rights reserved by the funding supporters.