Diagnostic accuracy of repeat placental growth factor measurements in women with suspected preeclampsia: A case series study

Abstract Introduction Preeclampsia affects about 3% of singleton pregnancies and is characterized by placental dysfunction. It is associated with significant maternal and perinatal morbidity and mortality. The diagnosis of preeclampsia remains a challenge, and the clinical course can develop for weeks before a diagnosis is confirmed. National guidelines have approved placental growth factor (PlGF) testing to rule out suspected preeclampsia, but the utility of repeated PlGF measurement is unknown. The aim of this case series analysis was to evaluate the test performance of repeated PlGF sampling in women presenting with suspected preeclampsia, and to describe relevant clinical outcomes. Material and methods Women who presented to maternity services with suspected preeclampsia between 20+0 and 36+6 weeks’ gestation who underwent repeat PlGF sampling with a minimum test interval of 7 days were assessed. The outcomes were delivery for preeclampsia within 14 days of sampling, the proportion changing PlGF categories, and time to delivery. Results In total, 289 women with suspected preeclampsia undergoing repeat PlGF sampling were included. PlGF <100 pg/mL had a high sensitivity (87.5%, 95% confidence interval [CI] 67.6%‐97.3%) and a negative predictive value (97.7%, 95% CI 93.5%‐99.5%) at the initial test (receiver operating characteristic [ROC] area 0.79, 95% CI 0.68‐0.89). Similar test performance was seen for PlGF <100 pg/mL when undertaken as a repeat test (sensitivity 90.7%, 95% CI 85.2%‐95.9%, negative predictive value 92.2%, 95% CI 85.3‐96.6%). Overall, 25.6% of women changed PlGF category between the first and second PlGF tests. For each PlGF category, determination of time to delivery was similar for first and second tests. Conclusions Repeat PlGF measurement demonstrates high negative predictive value for determining preeclampsia requiring delivery in 14 days. Repeat testing may be clinically useful to risk stratify women with ongoing symptoms of disease. Confirmation of the impact of these findings is required in further studies.


| INTRODUC TI ON
Preeclampsia affects 2.8% of singleton pregnancies and is characterized by placental dysfunction. 1 It is associated with significant maternal and perinatal morbidity and mortality. 2 One in 20 stillbirths without a congenital anomaly are attributed to, or associated with preeclampsia 3 and it is a leading cause of iatrogenic preterm birth. 4 The prediction and diagnosis of preeclampsia is an important component of antenatal maternity care; however, the diagnosis of preeclampsia remains a clinical challenge. Presentation is widely variable, and women with preeclampsia may be asymptomatic, even in the presence of severe disease. The clinical disease course can progress for weeks before a diagnosis is confirmed. The diagnostic criteria for preeclampsia, as defined by the International Society for the Study of Hypertension in Pregnancy, relies on the assessment of new-onset hypertension and one or more features of multi-organ disease. 5 These signs and symptoms may be subject to substantial observer error 6 and are frequently associated with poor test accuracy. 5,7,8 The clinical uncertainty associated with diagnosing preeclampsia leads to the extensive use of laboratory investigations, and is frequently associated with admissions to hospital for inpatient monitoring. Advances in our understanding of the pathophysiology of preeclampsia have highlighted a role for placentally derived angiogenic factors as relevant disease biomarkers. 9 Placental growth factor (PlGF) is a member of the vascular endothelial growth factor (VEGF) family, that in pregnancy is expressed predominantly in placental tissue, and correlates well with placental function. In normal pregnancies, PlGF is synthesized in abundance in the syncytiotrophoblast. Circulating PlGF concentrations rise throughout pregnancy until 30 weeks' gestation, thereafter declining towards term. It is thought that placental hypoxia, oxidative stress, the release of pro-inflammatory cytokines and reduced utero-placental blood contributing to endothelial activation in preeclampsia causes a release of antiangiogenic factors and a subsequent reduction in maternal serum PlGF. 10 A diagnostic accuracy study published in 2013 demonstrated that in women presenting with suspected preeclampsia, low circulating maternal PlGF concentrations (<100 pg/mL) had high sensitivity (96%; 95% CI 89-99%) and negative predictive value (98%; 95% CI 93%-99.5%) for diagnosing preeclampsia requiring delivery within 14 days. 11 In the PARROT cluster randomized controlled trial, single PlGF measurements were revealed to clinicians alongside a clinical management algorithm to be used as a diagnostic adjunct in suspected preeclampsia. When compared with standard practice, there was a reduction seen in the median time to diagnosis of preeclampsia from 4.1 to 1.9 days (time ratio 0.36, 95% CI 0.15-0.87) with a concurrent reduction in severe maternal adverse outcomes (5.4-3.8%, adjusted odds ratio 0.32, 95% CI 0.11-0.96). 12 Of women who tested "normal" for PlGF (>100 pg/mL) in the PARROT trial, 10.9% went on to receive a diagnosis of preeclampsia in the pregnancy. In 2016, there was one small study of 100 women published which looked at repeated PlGF measurements in women presenting with suspected preeclampsia. This gave only limited descriptive outcomes of the participants, and no test performance characteristics. 13 In 2019, Zeisler and colleagues published a post hoc analysis of participants in the PROGNOSIS study, in which repeated soluble fmslike tyrosine kinase 1 (sFlt-1)/PlGF measurements were undertaken on a weekly basis after initial presentation. This showed that there was a significantly larger median increase in sFlt-1/PlGF ratio in women who subsequently developed preeclampsia or adverse fetal outcomes as compared with those who did not (mean difference 21.22 vs 1.40, P < .001). These repeated measurements were taken at fixed timepoints in all participants, not specifically in women with an ongoing suspicion of preeclampsia. The authors did not report test performance statistics for the repeat tests, which would enable confirmation that PlGF-based tests retain similar test performance when used sequentially. 14 We hypothesize that repeat PlGF sampling has high diagnostic accuracy in predicting the need for delivery for preeclampsia within 14 days. Repeat PlGF testing provides clinically important information in women in whom a diagnosis of preeclampsia remains uncertain following an initial test.
The aim of this analysis was to assess repeat sampling in a case series of women with suspected preeclampsia, to determine test performance of the first and second test for predicting preeclampsia requiring delivery within 14 days (as a marker of clinical disease severity) and to describe relevant clinical parameters of patients by the repeated test results.

| MATERIAL AND ME THODS
The data for this case series were obtained from pregnant women undergoing repeat PlGF testing from the PELICAN, 11 MAPPLE (UK centers only), 15

Key message
Repeat placental growth factor measurements demonstrate high negative predictive value for preeclampsia, and can risk stratify women with ongoing symptoms of disease.

K E Y W O R D S
Diagnostic test, placental growth factor, preeclampsia suspected preeclampsia which included one or more of the following signs or symptoms: hypertension, dipstick proteinuria, headache with visual disturbances, epigastric pain, and fetal growth restriction between 20 +0 and 36 +6 weeks' gestation. Women recruited with fetal growth restriction needed to have further signs or symptoms that led to clinical suspicion of preeclampsia. PlGF samples were repeated if women presented to antenatal services with an ongoing clinical suspicion of disease, and all second PlGF samples were taken 7 days or more from the first test. Women were excluded from this analysis if the repeat PlGF measurement was taken less than 7 days from the first test. Management decisions were left to the treating clinician's discretion.
Blood samples were taken from women into bottles containing ethylenediamine tetra-acetic acid, and tested for PlGF con- Anonymized clinical data were collected in all cases in real time.
Pregnancy outcome data were captured from the electronic patient records after birth. Preeclampsia was defined by the International Society for the Study of Hypertension in Pregnancy 2014 statement. 18 All data were checked for completeness and adjudicated by the lead investigators of each study. Missing data from electronic records were retrieved from case notes where possible.

| Statistical analyses
Women were classified according to the gestation of the test (<35 and 35-36 +6 weeks' gestation).
Women were categorized by their measured PlGF concentration into the following predetermined groups: 1. PlGF ≥100 pg/mL-determined as "normal" 2. PlGF 12-99 pg/mL, equivalent to <5th centile for gestation and determined as "low" 3. PlGF <12 pg/mL, the lowest limit of detection for the assay and determined as "very low" These categorical groups were used based on the evidence that in those presenting <35 weeks' gestation, "low" PlGF has a high diagnostic accuracy (0.96; 95% CI 0.89-0.99) and negative predictive value (0.98; 95% CI 0.93-0.995) of determining preeclampsia requiring delivery in 14 days in a prospective observational cohort study, 11 and "very low" PlGF is the lowest limit of detection of the assay.
We have previously reported that a PlGF threshold of <100 pg/mL predicted preeclampsia requiring delivery within 14 days or before 37 weeks' gestation (whichever was sooner) with sensitivity and negative predictive values similar to diagnostic accuracy estimates obtained by using a <5th centile cut-off. The study reflects national guidance, which recommends that PlGF is implemented as a rule-out test up to 34 + 6 weeks' gestation. 11 For further sub-classification of normal and low values, visual inspection of the data was undertaken and clinically applicable thresholds explored.
A positive test was PlGF <100 pg/mL. Test performance for the first test, and the repeat test for the primary endpoint of preeclampsia requiring delivery within 14 days (or before 37 weeks' gestation in those recruited between 35 +0 and 36 +6 ) was calculated using sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios, and receiver operating characteristic (ROC) areas, all with 95% confidence intervals. Test performance for the repeat test following an initial normal (PlGF >100 pmg/mL) test was also calculated. Median and interquartile ranges for time from second PlGF test to delivery were calculated, stratified by the result of the first test. Birthweights are presented as birthweight centiles. 19 An additional sensitivity analysis of the test performance was performed on samples where the results were not revealed to clinicians in order to avoid treatment paradox. All statistical analyses were undertaken using STATA version 14.2 (StataCorp, College Station, TX).   We investigated using the absolute difference in PlGF concentration, and the ratio of the values, and found that these measures of deterioration (eg, differences <0 and ratios <1) gave areas under the ROC curve of 0.47 and 0.53, respectively.

| RE SULTS
Time to delivery in days by the second PlGF result, stratified by the initial PlGF sample, is given in Figure 3 (with accompanying values in Table 4). The time to delivery was similar whether the test was on repeat or initial presentation, and trends across the sub-categories remained similar regardless of initial test category. pregnancy. Women with the most severe preeclampsia may not benefit from repeat sampling, as they may reach a diagnosis quickly or may need medically indicated delivery before a second test is taken.

| D ISCUSS I ON
Given the heterogeneity in clinical presentations of suspected preeclampsia and the poor diagnostic accuracy of our current methods of clinical assessment, these data show that PlGF may be a useful diagnostic adjunct in women where there is ongoing diagnostic uncertainty. Our finding that the high sensitivity and negative predictive value appears to be maintained at repeat sampling, suggests that there may be benefit in repeated testing for women in whom a diagnosis remains uncertain. The relatively low positive predictive value in both initial and repeat testing may lead to over-investigation of women, but as this is a cohort at risk of serious maternal and perinatal adverse events, this is likely to be acceptable to women and clinicians.
The results of the PARROT trial demonstrated no difference in perinatal death rates with revealed PlGF testing, in contrast to other case series in the literature. 15 It may be that in order to prevent avoidable stillbirths, repeated PlGF testing is indicated alongside ultrasound fetal surveillance as a means of disease monitoring.
Further work is required to assess the impact of repeat PlGF testing on important maternal and perinatal outcomes, to ascertain rates of clinical outcomes in those changing categories at the second test, and the cost-effectiveness of repeat PlGF testing.
Women testing closer to the threshold values of 12 or 100 pg/ mL at the initial test are more likely to change category at the second test. Further work should therefore determine whether PlGF should be considered a continuous variable, as with other markers such as platelet count or serum creatinine, and whether absolute values or gestation referenced centiles could lead to improved risk discrimination beyond the three currently utilized categories of PlGF level. A prospective study of repeat PlGF sampling could also explore the different scenarios in which repeat testing may be indicated, for example in all women to play a role in ongoing risk stratification, in those without a definitive diagnosis but whom there is an ongoing clinical suspicion of disease, or even in those with confirmed preeclampsia for prognosis of pregnancy outcome.

| CON CLUS ION
Suspected preeclampsia is a clinical challenge encountered daily by clinicians in maternity triage settings. The diagnostic accuracy of detection of preeclampsia requiring delivery within 14 days remains high with repeat PlGF testing. Repeat testing may be clinically useful to risk stratify women presenting suspected preeclampsia if they present with ongoing symptoms of disease.

CO N FLI C T O F I NTE R E S T
The authors have stated explicitly that there are no conflicts of interest in connection with this article.