Soluble Flt-1 as a diagnostic marker of pre-eclampsia

Authors


  • This research was undertaken under the auspices of a grant provided by the ANZCOG.

: Professor Annemarie Hennessy, University of Western Sydney, Locked Bag 1797, Penrith South DC, NSW 1797, USA. Email: an.hennessy@uws.edu.au

Abstract

Backgound: Serum levels of soluble fms-like tyrosine kinase (sFlt-1) increase in pre-eclampsia (PE).

Aims: To determine whether concentrations of serum sFlt-1 can differentiate PE or superimposed PE (SPE) from gestational hypertension (GH) or chronic hypertension (CH).

Methods: Blood was collected from pregnant women being investigated for hypertension (blood pressure of  > 140 and/or 90 mmHg). Normotensive (NP) and pre-eclamptic (PE-C) control ranges were measured.

Results: Patients with evolving hypertension in pregnancy eventually fell into four groups: GH (n = 14), PE (n = 7), CH (n = 9) and SPE (n = 9). Patients who later developed pre-eclampsia had a higher sFlt-1 (PE: 2.61 ng/mL and SPE: 2.77 ng/mL, respectively) than GH (P < 0.001) or CH (1.05 ng/mL, P = 0.11). Women with established PE at recruitment (PE-C; (n = 18) (3.13 ng/mL; interquartile range (IQR): 2.14–4.17 ng/mL) had a median sFlt-1 higher than NP (n = 18) (0.47 ng/mL; IQR: 0.11–0.89) (P < 0.0008). Patients with GH compared to NP had a slight increase (1.33 ng/mL, P < 0.003). Using a sFlt-1 cut-off of ≥ 1.9 ng/mL yielded a sensitivity of 94% (95% confidence interval (CI) 73–100%) and specificity of 78% (95% CI 64–82%).

Conclusions: sFlt-1 was elevated in women with PE compared to NP. The sFlt-1 also differentiated women destined to develop PE among those who presented with a diagnostic rise in maternal blood pressure. The sFlt-1 test is a useful diagnostic test for PE.

Background

Pre-eclampsia affects around 5% of pregnancies causing substantial maternal and fetal morbidity and mortality,1 although the aetiology and pathogenesis of pre-eclampsia remain largely unknown. It has been hypothesised that the initial cause is abnormal implantation of the placenta resulting in impaired placental blood flow.1 The hypoxic placenta releases soluble factors into the maternal circulation, which induces systemic endothelial dysfunction. This gives rise to the clinical features of hypertension, proteinuria, oedema and coagulation abnormalities.1

The circulating factor or factors secreted by the placenta that are responsible for the widespread endothelial dysfunction in pre-eclampsia have been difficult to elucidate despite extensive efforts. Maynard et al. and others discovered the placenta of pregnant women with pre-eclampsia produced increased levels of soluble fms-like tyrosine kinase 1 receptor (sFlt-1).2,3 They also demonstrated that exogenous sFlt-1 administered to pregnant rats produces hypertension, proteinuria and histological glomeruloendotheliosis, which are all features of human pre-eclampsia. Makris et al. have recently proved that elevated sFlt-1 is caused by placental ischaemia.4

sFlt-1 acts by binding to the receptor binding domains of vascular endothelial growth factor (VEGF) and placental growth factor (PlGF), another member of the VEGF family that is made predominantly by the placanta.2 VEGF is known to have potent angiogenic properties and also promotes vasodilation.5 VEGF exerts its biological effects through Flt-1, a membrane-bound receptor tyrosine kinase. sFlt-1 is the soluble secreted variant of this receptor which lacks the transmembrane and cytoplasmic domains and antagonises VEGF and PlGF.6 Increased levels of sFlt-1 in the maternal circulation will therefore result in reduced levels of free VEGF and free PlGF, creating an antiangiogenic state that may be responsible for the hypertension and proteinuria of pre-eclampsia.2

Levine et al. demonstrated that increased levels of sFlt-1 and reduced levels of PlGF predict the development of pre-eclampsia up to five weeks before the onset of clinical symptoms in stored blood samples.7

This study aimed to determine if sFlt-1 levels at the time of first elevation in blood pressure could be used to differentiate pre-eclampsia from gestational hypertension (GH) and escalating chronic hypertension (CH). Women who have CH develop pre-eclampsia with increased frequency and severity and it is often difficult to diagnose pre-eclampsia due to the similarity of signs.8 An accurate diagnostic test for pre-eclampsia would be especially useful for this group of patients. We hypothesised that sFlt-1 would be elevated in patients who were subsequently proven on clinical grounds to have progressed to pre-eclampsia.

Methods

Patients were selected prospectively from ward or delivery suite admissions, day stay assessment or antenatal clinic at Royal Prince Alfred Women and Babies, a large teaching hospital in Sydney, Australia. Subjects were recruited consecutively by a single clinician. Patients were being investigated for a new rise in blood pressure (systolic blood pressure of ≥ 140 mmHg or diastolic blood pressure of ≥ 90 mmHg on at least two occasions > 4 h apart). They were assessed in four groups according to their final diagnosis: GH (n = 14), pre-eclampsia (n = 7), CH (n = 9), and superimposed pre-eclampsia (SPE, n = 9). Data were established for normotensive controls (NP, n = 18) and women with pre-eclampsia at presentation (pre-eclampsia controls: PE-C, n = 18). Results for CH and gestational hypertension were also compared to NP.

Pre-eclampsia was defined as hypertension with at least one other feature of pre-eclampsia. These included proteinuria (one dipstick measurement ≥ 2+ or ≥ 300 mg in a 24-h urine collection), reduced platelet count < 150 &times; 109/L, elevated liver function tests (alanine aminotransferase > 35 iu/L) or birthweight below the third centile, according to the classification system of the Australasian Society for the Study of Hypertension in Pregnancy (ASSHP).8 Chronic hypertension was defined as blood pressure requiring treatment with antihypertensive medication outside pregnancy or an antenatal blood pressure at less than 20 weeks gestation of ≥ 140/90 mmHg on more than two occasions at least four hours apart which was treated and controlled to below 140/90 prior to inclusion in the study. A final diagnosis was made post-partum as to whether patients had developed additional defining features of pre-eclampsia.

Normotensive controls were matched by age (within two years) and gestation (within one week) to control patients with pre-eclampsia, that is, patients with pre-eclampsia at the time of recruitment. Normal pregnancies were not complicated by medical or obstetric complications and delivered an infant of greater than 2500 g at term. Pre-eclampsia controls had diagnostic features of pre-eclampsia, as described above at the time of presentation.

All women provided written informed consent prior to collection of serum samples, which were collected between March 2005 and January 2006. This study was approved by the Ethics Committee of Royal Prince Alfred Hospital in accordance with National Health and Medical Research Council Guidelines for Human Research.

Venepuncture, using an 18-gauge needle and vacutainer system at the cubital fossa, was performed at the time of recruitment and blood collected into tubes containing EDTA. The samples were centrifuged within four hours of collection and the serum stored at –70&deg;C. The concentrations of sFlt-1 were measured with the use of enzyme-linked immunosorbent assay (ELISA) from BD (San Jose, CA, USA) by a single operator. The person who performed the assay was blinded to the clinical diagnoses of the patients. All measurements were made in duplicate on 1:2 dilution of the sera. The microtitre plates were read with a programmable spectrophotometer.

Data were analysed using Minitab 10, utilising Kruskal–Wallis and Mann–Whitney tests for non-parametric data. Data are presented as medians and interquartile ranges. Significance was assumed for a P-value of < 0.05. Sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio and negative likelihood ratio were calculated using an arbitrary cut-off of serum sFlt-1 at 1.9 ng/mL.

Results

There were no significant differences in mean gestation and maternal age between the four groups (Table 1). Serum samples were taken at a median of 35 weeks and three days gestation (range 25–41 weeks) and a median of 2.6 weeks before delivery (range 0–11 weeks).

Table 1.  Clinical characteristics of women with recently elevated blood pressure
Final diagnosisGHPECHSPE*P-value
  1. CH, chronic hypertension; GH, gestational hypertension; IQR, interquartile range; NS, not significant; PE, certain pre-eclampsia; SPE, superimposed pre-eclampsia.

Maternal age, years (IQR)29.5 (26–35)31 (28–33)32 (28–35)34 (30–36)NS
Gestation at trial entry, weeks (IQR)36 (29–39)34 (32–35)36 (29–39)34 (29–38) 
Delivery gestation, weeks (IQR)37 (36–39)35 (33–38)37 (36–40)36 (30–39)NS
Birthweight, g (IQranger)3188 (2973–3768)2530 (2108–2579)2828 (1568–3328)2500 (1111–3196)P < 0.008

At recruitment in the CH and PE groups, 83% and 94% of patients, respectively, were being treated with antihypertensive medications, while only 24% of patients with new onset hypertension in the GH group were being treated. These medications included clonidine (81%n = 29), hydralazine (31%n = 11), oxprenolol (14%n = 5), labetalol (8%n = 3) and methyldopa (6%n = 2). At delivery all patients were being treated with antihypertensive medications and 19% (n = 13) of patients were on more than one type of antihypertensive medication. All patients in the PE-C group and no patients in the NP group were on blood pressure lowering medications (Table 2).

Table 2.  Clinical characteristics of controls
Final diagnosisNPPE-C*P-value
  1. IQ, interquartile; NP, normal pregnancy; NS, not significant; PE-C, pre-eclampsia controls.

Maternal age, years (IQR)32 (28–39)32 (28–38)NS
Delivery gestation, weeks (IQR)40 (39–40)36 (33–38)NS
Birthweight, g (IQR)3463 (3131–3866)2562 (2006–3279)P < 0.008

Of the 21 patients with an uncertain diagnosis at recruitment, 33% (n = 7) developed features of pre-eclampsia, while 50% (n = 9) of patients with CH progressed to SPE. Of the seven pre-eclamptic patients, one patient was being treated with intravenous antihypertensive medication and one patient was treated with magnesium sulphate as an anticonvulsant. Of the nine patients with SPE, two patients required intravenous therapy and one patient was being treated with magnesium sulphate. In the PE-C group, 59% (n = 10) of patients were being treated with more than one antihypertensive agent at delivery and 23% (n = 4) were on magnesium sulphate.

Figure 1 shows the concentrations of sFlt-1 according to diagnosis as confirmed post-partum. Women with pre-eclampsia at recruitment (PE-C) had a median serum concentration of sFlt-1 higher than women with NP (PE-C: median 3.13 ng/mL interquartile range (IQR) 2.14–4.17 ng/mL and NP: 0.47 ng/mL IQR 0.11–0.89 ng/mL; P < 0.0008).

Figure 1.

Serum sFlt-1 concentrations in normal pregnancy and in patients with pre-eclampsia at presentation. Individual patient data; median and interquartile ranges shown. **P < 0.001 compared to normal pregnancy.

Patients with elevated blood pressure at recruitment (Fig. 2) who later developed features of pre-eclampsia had a higher median serum concentration of sFlt-1 than GH or CH (PE: median 2.61 ng/mL, IQR 1.9–2.69 ng/mL < 0.00 and SPE: 2.77 ng/mL, IQR 2.45–3.65 ng/mL P < 0.001). Patients with GH compared to normal pregnancy had a slight increase (1.33 ng/mL, IQR 1.06–1.77 ng/mL P < 0.003), whereas those with a final diagnosis of CH were not different to normal (1.05 ng/mL IQR 0.38–1.83 ng/mL, P = 0.11).

Figure 2.

Serum sFlt-1 concentrations at the time of recent elevation in blood pressure according to final diagnosis. Individual patient data; median and interquartile ranges shown. **P < 0.001 compared to gestational hypertension and chronic hypertension.

Using a sFlt-1 cut-off of ≥ 1.9 ng/mL this test yielded a sensitivity of 94% (95% confidence interval (CI) 73–100%) and specificity of 78% (95% CI 64–82%). Of the 16 women who developed pre-eclampsia after recruitment, nine (56%) had an elevated sFlt-1 level, ≥ 1.9 ng/mL, up to eight weeks before the onset of pre-eclampsia. The positive likelihood ratio for this test is 4.3 (95% CI 1.762–8.553) and the negative likelihood ratio is 0.08 (95% CI 0.018–0.540) (Table 3).

Table 3.  Serum sFlt-1 as a diagnostic marker for pre-eclampsia for sFlt-1 concentration ≥ 1.9 ng/mL
 95% CI
  1. CI, confidence interval.

Sensitivity94%73–100%
Specificity78%64–82%
Positive predictive value75%59–80%
Negative predictive value95%77–100%
Positive likelihood ratio  4.31.97–9.46
Negative likelihood ratio  0.080.018–0.540

There were five patients included in the data analysis with particularly high levels of sFlt-1 at recruitment. These values were 7.34, 5.45, 5.16, 4.82 and 4.78 ng/mL. These patients all had evidence of severe disease. All five patients were being treated with intravenous antihypertensive therapy for blood pressure ≥ 170/110 and four women were being treated with magnesium sulphate at delivery. All five babies were delivered prematurely at 28, 31, 33, 34 and 35 weeks gestation because of worsening pre-eclampsia.

Discussion

Our findings confirm that the maternal serum sFlt-1 concentration is markedly increased in women with pre-eclampsia when compared with controls. We have also shown that in women with a new onset rise in blood pressure, serum sFlt-1 concentrations can ‘diagnose’ pre-eclampsia a maximum of eight weeks before the onset of additional clinically defining features. This study has also found that very high levels of sFlt-1 were present in patients with severe pre-eclampsia.

This study extends prior observations by examining serum sFlt-1 concentrations in patients with CH. Our data show that sFlt-1 levels may be a useful test for this group of women at high risk of developing SPE.

No single test exists to identify which women with new onset ‘gestational’ hypertension or escalating CH will develop pre-eclampsia. Serum uric acid concentration has been suggested as one of the most sensitive indicators of pre-eclampsia.9 Decreased renal tubular excretion is the most likely mechanism for uric acid being increased in pre-eclampsia.10,11 Oxidative stress may also give rise to the hyper-uricaemia seen in pre-eclampsia.12 Serum uric acid levels used for the diagnosis of pre-eclampsia gave a sensitivity of 69% and specificity of 51% with a cut-off of 0.33 mmol/L.13 There may be some correlation between elevated serum uric acid levels and both the severity of pre-eclampsia and the neonatal morbidity.14–16 The clinical utility of serum uric acid values in differentiating various hypertensive diseases of pregnancy, however, appears to be limited.17

Serum creatinine levels have also been used to help confirm a diagnosis of pre-eclampsia. A level greater then 106 µmol/L in pregnancy supports the diagnosis of pre-eclampsia;18 however, levels may not help differentiate CH because of underlying renal disease.9

Some investigators have retrospectively looked at the usefulness of Down syndrome screening markers for the prediction of pre-eclampsia. Second trimester serum levels of alpha-fetoprotein (AFP), human chorionic gonadotrophin (hCG), unconjugated estriol (uE3) and inhibin A (inh A) have been used individually and in combination for risk assessment of pre-eclampsia. The most promising of these markers seems to be inh A, which has been reported to be significantly increased in the second trimester in women destined to develop pre-eclampsia with a sensitivity of up to 48% for a specificity of 90%.19 inhA may be useful for the prediction of pre-eclampsia as early as ten weeks gestation;20 however, levels appear to be more markedly increased closer to disease onset in pre-eclamptic women.21 In a small study, Wald et al. reported that combining the values of inh A, hCG and uE3 from second trimester maternal screening would detect women who would subsequently develop pre-eclampsia in their pregnancy with a sensitivity of 55% and specificity of 95%.22

Second trimester serum markers combined with other maternal indices show greater predictive value. For example maternal age, body mass index and parity with free βhCG predicted 70% of women who developed pre-eclampsia with a specificity of 71%.23 inh A with Doppler ultrasound predicted 71% of cases of pre-eclampsia with a specificity of 93%.24

Longitudinal blood sampling has shown that leptin concentrations are increased in established pre-eclampsia, as well as before pre-eclampsia is clinically evident from 20 weeks gestation until delivery.25 Serum leptin and placental growth factor in combination have been used as a screening test for pre-eclampsia at 24 weeks gestation with reported sensitivity of 67% for a specificity of 100%. These authors also showed a decrease in insulin-like growth factor binding factor-1 prior to the onset of pre-eclampsia.26

Fibronectin levels, a marker of endothelial dysfunction, are elevated up to four weeks before the onset of pre-eclampsia.27,28 A longitudinal study of 378 women showed as early as nine to 12 weeks of gestation, there was a difference in fibronectin levels between women who developed pre-eclampsia in their pregnancy and controls. At 22–26 weeks gestation the sensitivity of fibronectin for predicting pre-eclampsia was 73% with a specificity of 87%.29

Similarities between pre-eclampsia and insulin resistance syndrome have led to theories that measurements of glucose tolerance and insulin resistance may predict pre-eclampsia. A recent study has shown that fasting insulin sensitivity indexes can predict in early (16–20 weeks) and late (26–30 weeks) pregnancy the subsequent development of pre-eclampsia with a sensitivity of 85% and 88%, respectively, for a specificity of 97%.30

Placental sFlt-1 has been implicated as being involved in the pathophysiology of pre-eclampsia by disrupting maternal endothelial cell function through antagonism of VEGF. sFlt-1 concentrations in pre-eclamptic women were more than threefold over the control which helps to confirm this hypothesis. The findings of this study are consistent with previous studies which have also found elevated concentrations of sFlt-1 in patients with pre-eclampsia at the time of diagnosis.2,31,32 A longitudinal study of eight women who developed pre-eclampsia, six with hypertension only (four CH and two gestational hypertension) and nine with normal pregnancy showed that sFlt-1 was significantly higher at 25–28 weeks in women who developed pre-eclampsia compared to women with hypertension alone or normal pregnancy. The group with hypertension alone appeared not to have substantially higher levels than women with normal pregnancy.33

A larger longitudinal analysis of serum sFlt-1 concentrations in normal pregnancy and pre-eclampsia showed that sFlt-1 levels were increased only within five weeks before the onset of hypertension and proteinuria.7

Pre-eclampsia causes significant morbidity and methods for early identification and streamlined management of women destined to develop pre-eclampsia are needed urgently. In this cohort of women there was a clear elevation in sFlt-1 at presentation with pre-eclampsia compared to normotensive controls. The sFlt-1 concentration also differentiated women destined to develop pre-eclampsia from those with CH and those who had presented with a diagnostic rise in maternal blood pressure with no other feature of pre-eclampsia (gestational hypertension).

This test has important clinical utility as it targets patients at high risk of developing pre-eclampsia, that is, those women with uncertainty in their diagnosis. Serum sFlt-1 measurements could be used to guide important management decisions regarding whether or not admission to hospital is required, closer monitoring in a day-stay unit or consideration of delivery. To reduce the risks that are associated with the development of severe pre-eclampsia, early detection and timely assessment of the disease are essential. Serum sFlt-1 appears to be a useful tool as a one-off test in the third trimester for patients with clinical uncertainty of the cause of hypertension in pregnancy.

Ancillary