The role of maternal sFlt-1/PlGF ratio and uterine Doppler ultrasound in the third-trimester prediction of late-onset hypertensive disorders of pregnancy – diagnosis and complications in a high-risk population

Introduction

Hypertensive disorders of pregnancy (HDP) represent a significant cause of maternal mortality and fetal complications. The American College of Obstetricians and Gynecologists (ACOG)⁽¹⁾ estimated that HDP accounted for 16% of maternal deaths in highly developed countries. Approximately 4.6% of pregnancies are impaired by preeclampsia throughout the world, and its incidence is currently rising^(2,3). Among HDP, preeclampsia is the most important health issue, as it constitutes a major risk factor for maternal long-term cardiovascular disease, as well as life-threatening fetal complications such as fetal growth restriction (FGR), impaired fetal brain development, antepartum fetal demise, preterm delivery and, also, long-term neonatal complications, like cerebral palsy or chronic pulmonary hypertension of the newborn^(4-6).  

Preeclampsia is regarded as early-onset if diagnosed before 34 weeks, and late-onset if detected at or after 34 weeks of pregnancy⁽⁷⁾. Traditional risk factors may be classified into anamnestic factors (personal history of preeclampsia in a previous pregnancy, thrombophilia or first-degree relative with preeclampsia) and clinical factors (in vitro fertilization, multifetal pregnancy, obesity, diabetes mellitus, advanced maternal age, above 40 years old, systemic lupus erythematosus, preexisting chronic hypertension or chronic kidney disease, significantly increased serum uric acid or serum uric acid to creatinine ratio)^(8-11). However, the majority of preeclampsia cases affect healthy nulliparous women with none of the previously mentioned traditional risk factors⁽¹⁾. Consequently, prediction strategies using clinical judgement based on the aforementioned traditional risk factors has displayed modest results, with only a 29% detection rate for late-onset preeclampsia⁽⁸⁾. However, as both early-onset and late-onset preeclampsia constitute an important cause of severe, sometimes life-threatening maternal-fetal complications, there is a need for more accurate screening strategies. Mild preeclampsia can sometimes unpredictably evolve into the severe form^(7,12).

Developing more accurate screening strategies for HDP – mainly preeclampsia and its associated complications – represents a continuous preoccupation in the recent literature. For this purpose, the biochemical markers soluble fms-like tyrosine kinase 1 (sFlt-1) and placental growth factor (PlGF), as well as Doppler ultrasound parameters – the uterine artery pulsatility index (PI) and resistivity index (RI) – have been proven valuable screening tools when it comes to risk assessment of both early-onset and late-onset preeclampsia, and for major complications: FGR, preterm delivery and intrapartum fetal demise. sFlt-1, as an antagonist of vascular endothelial growth factor (VEGF) and PLGF, normally acts as a physiologic angiogenesis inhibitor in many tissues throughout the body. However, when produced in excess, it induces pathological vascular changes in the placenta which may cause preeclampsia at a later stage of pregnancy⁽¹³⁾. Typically, sFlt-1 overexpression occurs at 21-24 weeks, thus being a valuable screening tool for preeclampsia in the second and third trimesters, whereas abnormally decreased serum levels of PlGF appear at an earlier stage of pregnancy, and it may be used for preeclampsia prediction in the first trimester as well⁽¹⁴⁾. The imbalance of the two biomarkers’ serum concentration generates a higher sFlt-1/PlGF ratio, which has been widely regarded as a reliable third-trimester preeclampsia predictor in women at high risk based on clinical judgement^(14,15). Furthermore, sFlt-1/PlGF plays an important role in the third trimester screening of FGR, antepartum fetal demise and preterm delivery under 32 and 37 weeks⁽¹⁴⁾. As previously mentioned, beside sFlt-1/PlGF ratio, additional value in the third trimester HDP prediction is provided by uterine artery Doppler ultrasound^(1,16). The pathological vascular changes in the placenta generate an increased resistance to normal blood flow in the uterine artery, a status which is revealed by high uterine artery PI and RI values. The guidelines elaborated by the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG)⁽¹⁶⁾ established the 95^th centile of mean uterine artery PI to be 1.17 at 30-34 weeks when calculated by transabdominal ultrasound. This cutoff value correctly anticipated 54% of preeclampsia cases with clinical onset before 37 weeks, a considerably better performance when compared to clinical judgement alone⁽⁸⁾. Moreover, the addition of sFlt-1/PlGF ratio improves the prediction performance of Doppler ultrasound⁽¹⁶⁾. The sFlt-1/PlGF ratio and the uterine artery Doppler parameters represent such a valuable preeclampsia screening tool due to the fact that their abnormally increased values usually precede the clinical onset of the condition with at least four or five weeks^(14,16).   

The aim of the present study is to develop a poten­tial third-trimester prediction algorithm of late-onset HDP occurrence and severity, particularly late-onset preeclampsia and its complications (late-onset FGR, preterm delivery), based on a combination of sFlt-1/PlGF ratio values and uterine artery Doppler ultrasound parameters PI and RI, in a population of pregnant patients initially judged as high-risk by the presence of traditional risk factors.    

Materials and method

1. Study population

We performed a prospective observational study through­out 21 months, from January 2021 until September 2022, in the Obstetrics and Gynecology Department of the Mureş County Clinical Hospital. We enrolled 60 patients with singleton pregnancy, at 32 weeks of gestation, initially considered as high-risk for later onset of HDP due to the presence of traditional risk factors. Therefore, the inclusion criteria in the study group consisted of the presence of at least one of the following known risk factors: diagnosis of preeclampsia in a previous pregnancy, personal history of type 2 diabetes mellitus or obstructive sleep apnea, advanced maternal age over 40 years old, obesity defined as Body Mass Index over 30 prior to pregnancy, or a first-degree relative diagnosed with preeclampsia. On the other hand, the exclusion criteria from the study were represented by multiple pregnancy, any diagnosis of HDP in the actual pregnancy, prior screening of preeclampsia by sFlt-1/PlGF ratio assessment or uterine artery Doppler examination and any other gestational age than 32 weeks in the moment of study enrollment. sFlt-1/PlGF ratio values were determined for each study participant at 32 weeks, in the moment of study enrollment. Moreover, all pregnant women underwent uterine artery Doppler ultrasound scan when included in the study, at 32 weeks, using the examination technique recommended by 2018 ISUOG Practice Guidelines. After that, the patients were periodically followed-up until delivery by clinical and ultrasound scans which included the assessment of umbilical and middle cerebral arteries PI, RI and the cerebroplacental ratio (CPR). Any possible onset of HDP and fetal morbidity, such as late-onset FGR, preterm delivery or antepartum fetal demise, was actively searched for during the periodic follow-up. The criteria provided by 2020 ISUOG Practice Guidelines⁽¹⁷⁾ were used for the positive diagnosis of late-onset FGR: ultrasound estimated fetal weight <3^rd centile or fetal weight <10^th centile combined with mean uterine artery PI >95^th centile, excluding congenital anomalies, at or after 32 weeks of pregnancy. Pregnant patients who remained normotensive until delivery formed the control group, whereas women who developed hypertension after 32 weeks (blood pressure values ≥140/90 mmHg), with no other diagnostic criteria for preeclampsia, formed the gestational hypertension group. Furthermore, pregnancies in which hypertension was followed by one other diagnostic criterion for preeclampsia, as defined by ACOG, after 32 weeks⁽¹⁾ (proteinuria revealed by a protein concentration of at least 30 mg/dl in the urine sample, thrombocytopenia: platelet count <100,000/microliter, or renal insufficiency: serum creatinine levels higher than 1.1 mg/dl), formed the preeclampsia group. For a better assessment of the fetal morbidity risk among preeclamptic women, the preeclampsia group was subdivided into two subgroups: the severe preeclampsia group, which consisted of patients with blood pressure values ≥160/110 mmHg alongside thrombocytopenia or renal insufficiency, as well as FGR causing preterm birth or antepartum fetal demise, and the mild preeclampsia group, formed by patients whose blood pressure values ranged between 140/90 mmHg and 160/110 mmHg, and who presented only proteinuria after 32 weeks, with no evidence of the maternofetal complications present among the severe preeclampsia group.  

2. sFlt-1 and PlGF assessment

For the purpose of biochemical markers assessment, a 5-ml sample of venous blood was obtained from each patient and stored in a 6-ml biochemistry vacuum blood tubes with clot activator, in the moment of study enrollment, when the patient presented for the routine consultation in the Obstetrics Department. Each specimen was centrifuged in the attached laboratory within 30 minutes from the collection. After that, the serum was stored in 1.5 ml Eppendorf tubes and later freezed at -20°C for a couple of days. For the study purpose, as two biomarkers had to be determined, two serum samples were required from each patient. After a few days, the specimens were transported to the Advanced Center of Medical and Pharmaceutical Research of the “George Emil Palade” University of Medicine, Pharmacy, Sciences and Technology, Târgu-Mureş, Romania, where they were freezed at -80°C. After maximum six months, the two biomarkers’ concentrations were determined by using an Elisa Dynex DSX fully-automated ELISA analyzer, through sandwich ELISA technique. 

3. Doppler examination of the uterine arteries at 32 weeks of gestation

Uterine arteries Doppler examinations were performed with the aid of a Voluson E8 BT18 (General Electric Healthcare, Chicago, IL, USA), equipped with a RAB-6D convex volumetric abdominal probe. The transabdominal approach was used, and the ultrasound technique was based on the recommendations provided by ISUOG Practice Guidelines⁽¹⁶⁾: after the identification of the uterus and cervical canal in the sagittal section, the transducer was moved towards either the left or right iliac fossae and the position of placenta was identified; then, color flow mode was selected, and the uterine artery was identified at the crossover with the external iliac artery; the sampling gate was set at 2 mm and placed on the uterine artery, 1 cm away from the aforementioned crossover; while maintaining an insonation angle under 50°, pulsed wave Doppler mode was set, and three identical waveforms of the uterine artery were obtained. The same steps were used for the Doppler scan of the contralateral uterine artery. Finally, uterine artery PI and RI were calculated, and the arithmetic mean for each parameter was determined. Based on recommendations provided by Cavoretto et al.⁽¹⁸⁾, mean uterine PI values were considered normal in the range of 0.78±0.23, whereas mean uterine PI values higher than 1.17 were regarded as a risk factor for later onset of preeclampsia, as advised by ISUOG Practice Guidelines⁽¹⁶⁾.

4. Statistical analysis

The statistical analysis consisted initially of descriptive statistics. Mean, standard deviation (SD), median and range were calculated for three continuous variables: sFlt-1/PlGF ratio values, mean uterine artery PI values, and mean uterine artery RI values. The frequency in number and percentages was calculated for the five categorical variables, represented by the studied groups: control, gestational hypertension, preeclampsia, mild preeclampsia, and severe preeclampsia groups. After that, the distribution of the three continuous variables was assessed with the help of Shapiro-Wilk test, and the median comparison between these variables was performed with the help of the non-parametric Kruskal-Wallis test. In this study, p values below 0.05 were considered statistically significant. Furthermore, Dunn’s test was later used for multiple comparison of the median values between the five groups. The area under the ROC curve (AUC) was used to assess the performance of the cutoff values over the range of possible cutpoints for sFlt-1/PlGF ratio, mean uterine PI and mean uterine RI for the prediction of preeclampsia, severe preeclampsia and gestational hypertension respectively. The cutoff value displaying the highest AUC was considered the best predictor for each of the three parameters. IBM SPSS software in trial version, GraphPad Prism software in trial version and Microsoft Excel were used for the statistical analysis.

Consequently, three prediction algorithms – for late-onset preeclampsia, late-onset severe preeclampsia (complicated by FGR with preterm birth or antepartum fetal demise) and, also, for gestational hypertension, based on combinations between sFlt-1/PlGF ratio values, on the one hand, and mean uterine PI and RI values, on the other hand – were developed in this study.

Results

Among the 60 patients displaying traditional (clinical or anamnestic) risk factors for preeclampsia, who formed the study group, 15 developed high blood pressure ≥140/90 mmHg with no evidence of proteinuria after 32 weeks of pregnancy, thus being categorized as gestational hypertension group, whereas 15 patients remained normotensive until delivery, consequently being included in the control group. Moreover, 30 patients were diagnosed with late-onset preeclampsia, the clinical onset of symptoms appearing after the biomarkers and ultrasound screening at 32 weeks. Out of these patients, 16 developed the mild form of preeclampsia, whereas 14 women matched the criteria for severe preeclampsia. The prevalence of these five study groups among the study population can be visualized in Table 1 and Table 2. It is also worth pointing out that only 50% of the pregnant women initially regarded as carrying a high risk for preeclampsia based on the traditional risk factors were finally diagnosed with the condition.
 

While all severe preeclampsia cases were impaired by late-onset FGR, 13 of these pregnancies ended in premature delivery due to acute fetal distress, and one pregnancy was impaired by antepartum fetal demise. 

Regarding uterine artery Doppler parameters’ ability to predict the later onset of preeclampsia, a significant difference concerning median PI values between the preeclampsia, control and gestational hypertension group could be noticed by applying Kruskal-Wallis test (1.45 versus 1.05 versus 1.10; p<0.0001), as depicted in Table 3. In addition, Dunn’s test revealed that pregnant women who later developed preeclampsia displayed significantly higher median PI values at the 34-week screening compared to gestational hypertension and control group (Table 3, Figure 1).  
 

Furthermore, uterine artery RI followed the same pattern as PI values in the third trimester. Kruskal-Wallis test emphasized a significant difference concerning median RI values between the preeclampsia, control and gestational hypertension groups (0.75 versus 0.54 versus 0.62; p<0.0001), as shown in Table 4.

Additionally, Dunn’s test revealed that pregnant women who later developed preeclampsia displayed significantly higher median RI values at the 32-week screening compared to gestational hypertension and control group (Table 4, Figure 2).
 

Next, it is worth mentioning that ROC analysis revealed that, in our study, the most accurate mean uterine artery PI cutoff at 32 weeks for preeclampsia prediction was 1.15 (Figure 3), as it displayed the highest sensitivity and specificity over the range of possible cutpoints. Furthermore, the most accurate mean uterine artery RI cutoff for third-trimester preeclampsia prediction was 0.65 (Figure 4), a value which displayed slightly lower sensitivity (93.3%) compared to mean uterine artery PI cutoff.
 

Regarding sFlt-1/PlGF ratio’s ability to predict at the 32-week screening the later occurrence of preeclampsia, it displayed similar accuracy as mean uterine artery PI. Kruskal-Wallis test revealed a significant difference of the ratio’s median values between the preeclampsia, control and gestational hypertension groups (237.1 versus 9.07 versus 52.4; p<0.0001), as depicted in Table 5 and Figure 5. Dunn’s test revealed that pregnant women from preeclampsia group displayed significantly higher sFlt-1/PlGF ratio values compared to both the gestational hypertension and control groups (Table 5).
 

ROC analysis revealed 64.76 as the most accurate sFlt-1/PlGF ratio cutoff value at 32 weeks for preeclampsia prediction (Figure 6).
 

When comparing the accuracy of the three cutoff values described before, the sFlt-1/PlGF ratio and mean uterine artery PI proved to display the same prediction performance of later preeclampsia, with a sensitivity of 100% and the same AUC value, as shown in Table 6. No pregnant women enrolled in the study who developed preeclampsia had sFlt-1/PlGF values below 64.76 or mean uterine artery values below 1.15. In the meantime, mean uterine artery RI showed a reduced ability at 32 weeks to predict preeclampsia when compared to sFlt-1/PlGF ratio and mean uterine artery PI, displaying a sensitivity of 93.3%. Two pregnant patients screened at 32 weeks recorded mean uterine RI values below 0.65, and they were later diagnosed with preeclampsia.
 

One of the most challenging parts regarding the screening of preeclampsia would be the prediction of the severe late-onset form, complicated by FGR and preterm delivery or antepartum fetal demise. For this purpose, distinct cutoff values of the biomarkers and uterine artery Doppler parameters are needed.

First of all, Kruskal-Wallis and Dunn’s tests showed that women affected by severe late-onset preeclampsia displayed significantly higher median uterine PI values at 32 weeks compared to mild preeclampsia (1.73 versus 1.43; p<0.0001), as presented in Table 7 and Figure 7.
 

In the same manner, women who later developed severe late-onset preeclampsia displayed significantly higher mean uterine artery RI values at the 32-week screening compared to the other three groups (0.81 versus 0.73 for the mild form of late-onset preeclampsia; p<0.0001), as depicted in Table 8 and Figure 8. 
 

ROC analysis revealed the most accurate cutoff value for preeclampsia prediction at 32 weeks to be 1.48 in case of mean uterine artery PI (Figure 9) and 0.76 in case of mean uterine artery RI (Figure 10).
 

The two serum biomarkers’ degree of imbalance influenced the severity of late-onset preeclampsia. sFlt-1/PlGF ratio median values differed significantly between the four groups (411.3 versus 227.8 versus 52.4 versus 9.07; p<0.0001), as depicted in Table 9. Pregnant women later affected by severe late-onset preeclampsia displayed significantly higher sFlt-1/PlGF median values at 32 weeks compared to mild preeclampsia group, gestational hypertension group and control group (Table 9 and Figure 11). 
 

Moreover, ROC analysis established 239.09 to be the most accurate sFlt-1/PlGF cutoff at 32 weeks for late-onset severe preeclampsia prediction (Figure 12).
 

It is worth mentioning that the three cutoff values for the prediction of severe preeclampsia displayed the same sensitivity and specificity (92.9% and 100%, respectively). As expected, women from the severe preeclampsia group had the highest median PI, RI and sFlt-1/PlGF ratio values at the screening time among the study sample. However, among the 14 cases of late-onset severe preeclampsia, 12 cases displayed all three parameters above the cutoff at the screening time, whereas one pregnant woman had the sFlt-1/PlGF value above the cutoff, while the two Doppler parameters displayed values under the cutoff, and another pregnant woman later affected by severe preeclampsia presented mean uterine artery PI and RI values above the cutoff, while the sFlt-1/PlGF value was under the cutoff.

Regarding the prediction of gestational hyperten­sion, the affected women displayed significantly lower median uterine PI values at the screening time compared to late-onset preeclampsia group (1.10 versus 1.45; p<0.0001), as shown in Table 3, as well as significantly lower median sFlt-1/PlGF values compared to late-onset preeclampsia group (52.40 versus 237.1; p<0.0001), as shown in Table 5. In addition, pregnant women later affected by gestational hypertension displayed significantly higher median PI values than the control group (1.05 versus 1.10; p=0.0132), as represented in Table 3, as well as significantly higher median sFlt-1/PlGF values than the control group (52.40 versus 9.07; p<0.0001). ROC analysis enabled the development of predictive cutoffs for gestational hypertension at the 32-week screening time: 64.76 in case of sFlt-1/PlGF ratio (100% sensitivity, 66.7% specificity), 1.15 in case of mean uterine artery PI (100% sensitivity, 66.7% specificity), and 0.65 in case of mean uterine artery RI (100% sensitivity, 62.2% specificity). The cutoffs for gestational hypertension displayed lower predictive performance compared to the three markers’ cutoff for late-onset preeclampsia and late-onset severe preeclampsia. 

Finally, as previously mentioned, it is worth pointing out that only 50% of the pregnant women initially regarded as carrying a high risk for preeclampsia based on the traditional risk factors finally developed the disease (Table 1). Meanwhile, if we evaluate the risk of late-onset preeclampsia and late-onset severe preeclampsia based on sFlt-1/PlGF ratio, mean uterine artery PI and mean uterine artery RI cutoff values, all 30 cases of pregnant women who were later affected by late-onset preeclampsia displayed at least one out of three parameters above the cutoff at the 32-week screening, thus the predictive algorithm based on a combination of the three cutoff values had a sensitivity of 100% in preeclampsia prediction.

The same would apply to the predictive algorithm with specific cutoffs for late-onset severe preeclampsia, which displayed a sensitivity of 100% compared to 23.33% if the prediction was based on traditional clinical and anamnestic risk factors.

Discussion

A special focus was placed in the recent years on the prediction of preeclampsia and HDP-related complications due to the fact that screening based on traditional risk factors has shown modest results⁽⁸⁾. For this pur­pose, biochemical markers as well as Doppler ultrasound parameters have been tested in various studies, with promising results.

The recommended approach would be to start assessing the risk of HDP in the first trimester. Consequently, Fetal Medicine Foundation has developed a screening algorithm consisting on a combination of traditional risk factors, mean arterial pressure, mean uterine artery PI and PlGF, which are evaluated at 11-14 weeks of gestation⁽¹⁴⁾. Meanwhile, pregnancy-associated plasma protein A (PAPP-A) has also emerged as a potential useful biomarker⁽¹⁹⁾. HDP screening performed as early as possible in pregnancy brings certain advantages, such as proper risk stratification of patients with superior accuracy compared to screening only by clinical judgement, the possibility to implement personalized, more intensive maternal-fetal monitoring programs in case of high-risk pregnancies, or to commence early prophylaxis with low-dose aspirin^(1,14), early diagnosis of HDP, early initiation of antihypertensive drug where needed, the possibility to avoid or reduce the severity of maternal cardiovascular complications and pregnancy complications such as antepartum fetal demise or premature delivery, better fetal monitoring, and a better management of FGR by properly planning the moment and way of delivery⁽²⁰⁾.  

However, many high-risk pregnancies do not benefit from a standard first-trimester screening. Moreover, it is notable that preeclampsia mainly affects healthy nulliparous women with none of the traditional clinical and anamnestic risk factors⁽¹⁾. In addition, even the apparently mild forms of preeclampsia may unpredictably evolve into the severe form^(7,12), as previously mentioned. Therefore, the implementation of second- and third-trimester screening strategies is reasonable. In the third trimester, women considered as high-risk, based on sFlt-1/PlGF cutoff values or mean uterine artery PI and RI cutoff values at 32 weeks, certainly need intensive obstetric monitoring programs in the last part of the pregnancy, in order to early diagnose late-onset preeclampsia or gestational hypertension, and to properly manage the maternal-fetal complications. Proper early diagnosis of late-onset FGR in these pregnancies would allow rigorous clinical judgement in order to carefully choose the time of delivery and to minimize the risk of antepartum fetal demise. Even though there is a higher chance of maternofetal complications occurrence in early-onset preeclampsia⁽²¹⁾, the risk of late-onset FGR, antepartum fetal demise and preterm delivery is still high in late-onset preeclampsia, compared to gestational hypertension or normotensive pregnant women.

The sFlt-1/PlGF ratio plays a major role in the third-trimester preeclampsia screening strategies. Panaitescu et al.⁽²²⁾ stated in their study that the addition of sFlt-1 and PlGF to traditional risk factors improved the detection rate of preeclampsia from 28% to 70% at the 35-37 weeks’ screening. However, uterine artery PI did not appear to improve the screening performance. We reported similar results, with sFlt-1/PlGF ratio improving the detection rate of preeclampsia from 50% to 100% in high-risk women. Moreover, in our study, even though uterine artery PI did not improve the prediction performance of late-onset preeclampsia as a whole, it proved to be particularly useful in the prediction of late-onset severe preeclampsia, so it can be regarded as a marker of disease severity.

Insight into sFlt-1/PlGF ratio’s ability to predict adverse outcomes related to preeclampsia was provided by Poon et al.⁽²³⁾, who found sFlt-1/PlGF ratio to be significantly higher in case of preeclamptic women who developed complications such as FGR, fetal death, indicated delivery compared to uncomplicated pregnancies (47 versus 10.8; p<0.001). When added to hypertension and proteinuria, sFlt-1/PlGF ratio also raised the AUC from 0.84 to 0.93 (p<0.001) regarding the prediction of adverse events induced by preeclampsia. Furthermore, a sFlt-1/PlGF cutoff of 85 significantly predicts delivery within the next two weeks in the third trimester. Yang et al.⁽²⁴⁾ obtained the same ratio cutoff of 85 for the prediction of severe preeclampsia complicated by preterm delivery. By comparison, we reported a higher sFlt-1/PlGF cutoff (239.09) at 32 weeks in case of pregnancies later affected by complications such as FGR, fetal death and preterm delivery. 

The ROBUST study was another recent major re­search⁽²⁵⁾ that focused on sFlt-1/PlGF ratio’s potential to predict preeclampsia and its related obstetric complications in the third trimester. In this study, the authors also used sFlt-1/PlGF cutoff of 85 to predict a significant risk of later onset of severe preeclampsia complicated by preterm delivery: 90.9% compared to 8% (p<0.001) if sFlt-1/PlGF ratio values were under 33 between 28 and 37 weeks of gestation. The mean gestational age at delivery was only 32.6 weeks in pregnancies with maternal sFlt-1/PlGF values above 85, compared to 37.4 weeks if sFlt-1/PlGF ratio’s values were under 33. It is worth nothing that sFlt-1/PlGF ratio’s cutoff used in our research for the prediction of severe late-onset preeclampsia complicated by preterm delivery was 239.09, which is considerably higher than 85, the cutoff proposed by Poon et al.⁽²³⁾, Yang et al.⁽²⁴⁾ and Soundararajan et al.⁽²⁵⁾

The INSPIRE study⁽²⁶⁾, a major randomized controlled trial which included pregnant patients between 24 and 37 weeks of gestation, established that women who displayed sFlt-1/PlGF values above a cutoff of 38 carried a high risk of preeclampsia development in the following week. The sensitivity for this cutoff was 100%. The cutoff provided in this study is slightly smaller than sFlt-1/PlGF cutoff of 64.76 established in our study; however, the sensitivity of the cutoff is the same. In the PROGNOSIS study⁽²⁷⁾, the same cutoff of 38 was used to predict preeclampsia occurrence within the next four weeks after the assessment, but with a reduced sensitivity (66.2%), and also to predict premature delivery.

Ciciu et al.⁽²⁸⁾ highlighted sFlt-1/PlGF ratio’s ability to discern between preeclampsia and other forms of HDP, mainly gestational hypertension, discovering significantly higher mean sFlt-1/PlGF values among patients diagnosed with preeclampsia compared to gestational hypertension (209.2 versus 46.8; p<0.001). However, a notable difference would be that this study focused on sFlt-1/PlGF ratio’s specific values in different forms of HDP, while we provided predictive cutoff values for preeclampsia as well as gestational hypertension. Never­theless, it is worth mentioning that we reported similar median values in our study: 237.1 in case of preeclampsia versus 52.4 for gestational hypertension (p<0.0001). Although the risk of FGR complicated by preterm delivery or antepartum fetal demise is considerably lower in gestational hypertension, pregnant women with high blood pressure ≥140/90 mmHg in the third trimester certainly need closer obstetric monitoring to exclude later onset of preeclampsia. Thus, we consider implementing specific cutoffs for the prediction of gestational hypertension in the third trimester to be reasonable.  

As far as the role of ultrasound parameters in the third trimester prediction of preeclampsia is concerned, Adekanmi et al.⁽²⁹⁾ reported significantly increased mean uterine PI values among pregnant women later affected by preeclampsia compared to the control group (1.38 versus 0.75; p<0.001), as well as significantly increased mean uterine RI values (0.59 versus 0.5; p=0.002). In comparison, the mean uterine PI and RI values reported in our study were slightly higher: 1.45 versus 0.54 and 0.75 versus 0.54, respectively. Moreover, the 95^th centile of mean uterine PI provided by ISUOG Guidelines⁽¹⁶⁾ for preeclampsia prediction in the third trimester, at 30-34 weeks, was 1.17, which is very close to 1.15, the cutoff we developed at 32 weeks of pregnancy.

Valuable insight into the potential role of a combina­tion of sFlt-1/PlGF ratio and Doppler ultrasound markers in the prediction of late-onset preeclampsia was provided by Graupner et al.⁽³⁰⁾, who enrolled singleton pregnancies already diagnosed with late-onset preeclampsia. The authors reported that a combination of sFlt-1/PlGF ratio value above 110 and mean uterine artery PI higher than the 95^th centile was useful for the prediction of postnatal small-for-gestational-age babies (birth weight under the 3^rd centile). Significant negative correlations between sFlt-1/PlGF and CPR, sFlt-1/PlGF and birthweight, sFlt-1/PlGF and low gestational age at delivery were also reported. The findings in this study came to support our own conclusions, as we also highlighted the increased accuracy of a combination of sFlt-1/PlGF ratio above 239.09 with mean uterine PI>1.48 for the prediction of late-onset preeclampsia complicated by FGR compared to the ratio or PI alone. Moreover, an evident negative correlation between sFlt-1/PlGF and low gestational age at delivery can be emphasized in our study as well.   

Strengths and limitations

The major strength of our study would be the implementation of the three predictive algorithms based on the combination between biochemical parameters (sFlt-1 and PlGF) and uterine artery Doppler parameters (PI and RI). These algorithms brought significant added value to the prediction of HDP and their related complications based solely on traditional clinical and anamnestic risk factors in the third trimester, or based only on biochemical parameters or Doppler screening. Moreover, due to the development of specific cutoffs for three distinct clinical outcomes – gestational hypertension, uncomplicated late-onset preeclampsia, as well as severe preeclampsia complicated by FGR, preterm delivery or antepartum fetal demise –, these three algorithms may become valuable tools for clinicians for the proper risk assessment of patients initially categorized as high-risk based on clinical judgement. On the other hand, the main limitation of the study would be the relatively small sample size, which may have impacted the results.

Conclusions

The predictive algorithms based on a combination of sFlt-1/PlGF ratio and mean uterine artery Doppler ultrasound PI and RI specific cutoffs are significantly better predictive tools in the third trimester for late-onset preeclampsia, compared to clinical judgement based solely on traditional risk factors in case of patients initially regarded as high-risk. While the addition of mean uterine artery Doppler parameters did not increase the late-onset preeclampsia prediction accuracy compared to prediction based on sFlt-1/PlGF alone, combining the screening based on sFlt-1/PlGF with mean uterine PI and RI specific cutoffs in the third trimester proved to be particularly useful in the prediction of the severe form of late-onset preeclampsia, complicated by FGR and preterm delivery or antepartum fetal demise. Moreover, uterine artery Doppler appeared to be less useful at 32 weeks for the screening of late-onset uncomplicated gestational hypertension in the third trimester and to differentiate it from late-onset preeclampsia. These predictive algorithms offer clinicians a better risk assessment compared to clinical judgement based solely on traditional risk factors, thus enabling the implementation of more intensive monitoring programs in the third trimester of pregnancy for patients considered as high-risk based on sFlt-1/PLGF ratio and uterine Doppler PI and RI specific cutoffs, early HDP diagnosis and proper anticipation and management of preeclampsia-related complications, such as FGR, antepartum fetal demise and preterm delivery.  

Corresponding author: Claudiu Mărginean, e-mail: marginean.claudiu@gmail.com

Conflict of interests: none declared.

Financial support: none declared.

This work is permanently accessible online free of charge and published under the CC-BY.