Modified Multiple Marker Aneuploidy Screening As A Primary Screening Test For Preeclampsia

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Huang et al.

BMC Pregnancy and Childbirth (2022) 22:190


https://doi.org/10.1186/s12884-022-04514-4

RESEARCH Open Access

Modified multiple marker aneuploidy


screening as a primary screening test
for preeclampsia
Tianhua Huang1,2,3*, H. Melanie Bedford1,4, Shamim Rashid1, Evasha Rasasakaram1, Megan Priston1,
Ellen Mak‑Tam1, Clare Gibbons1,5, Wendy S. Meschino1,4, Howard Cuckle6 and Elad Mei‑Dan3,7

Abstract
Background: Abnormal levels of maternal biochemical markers used in multiple marker aneuploidy screening have
been associated with adverse pregnancy outcomes. This study aims to assess if a combination of maternal charac‑
teristics and biochemical markers in the first and second trimesters can be used to screen for preeclampsia (PE). The
secondary aim was to assess this combination in identifying pregnancies at risk for gestational hypertension and
preterm birth.
Methods: This case-control study used information on maternal characteristics and residual blood samples from
pregnant women who have undergone multiple marker aneuploidy screening. The median multiple of the median
(MoM) of first and second trimester biochemical markers in cases (women with PE, gestational hypertension and
preterm birth) and controls were compared. Biochemical markers included pregnancy-associated plasma protein A
(PAPP-A), placental growth factor (PlGF), human chorionic gonadotropin (hCG), alpha feto-protein (AFP), unconju‑
gated estriol (uE3) and Inhibin A. Logistic regression analysis was used to estimate screening performance using dif‑
ferent marker combinations. Screening performance was defined as detection rate (DR) and false positive rate (FPR).
Preterm and early-onset preeclampsia PE were defined as women with PE who delivered at < 37 and < 34 weeks of
gestation, respectively.
Results: There were 147 pregnancies with PE (81 term, 49 preterm and 17 early-onset), 295 with gestational hyper‑
tension, and 166 preterm birth. Compared to controls, PE cases had significantly lower median MoM of PAPP-A (0.77
vs 1.10, p < 0.0001), PlGF (0.76 vs 1.01, p < 0.0001) and free-β hCG (0.81 vs. 0.98, p < 0.001) in the first trimester along
with PAPP-A (0.82 vs 0.99, p < 0.01) and PlGF (0.75 vs 1.02, p < 0.0001) in the second trimester. The lowest first trimester
PAPP-A, PlGF and free β-hCG were seen in those with preterm and early-onset PE. At a 20% FPR, 67% of preterm and
76% of early-onset PE cases can be predicted using a combination of maternal characteristics with PAPP-A and PlGF in
the first trimester. The corresponding DR was 58% for gestational hypertension and 36% for preterm birth cases.
Conclusions: Maternal characteristics with first trimester PAPP-A and PlGF measured for aneuploidy screening
provided reasonable accuracy in identifying women at risk of developing early onset PE, allowing triage of high-risk
women for further investigation and risk-reducing therapy. This combination was less accurate in predicting women
who have gestational hypertension or preterm birth.

*Correspondence: [email protected]
1
Genetics Program, North York General Hospital, 4001 Leslie Street,
Toronto, ON M2K 1E1, Canada
Full list of author information is available at the end of the article

© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​
mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 2 of 14

Keywords: Multiple marker screening, Preeclampsia, Gestational hypertension, Preterm birth, Pregnancy-associated
plasma protein A, Placental growth factor, Human chorionic gonadotropin, Alpha feto-protein, Unconjugated estriol
and Inhibin A

Background also easy to implement, as it requires minimal change to


Maternal multiple marker screening for fetal aneuploidy current aneuploidy screening. Women who are identi-
is part of routine prenatal care. Over the past decades, fied as high-risk in the first trimester can be triaged for
multiple marker screening tests, comprising of biochem- MAP and UTPI or be considered for prophylactic ther-
ical markers and the ultrasound marker nuchal translu- apy (e.g. Aspirin) initiated before 16 weeks of gestation
cency (NT), have evolved from second to first trimester to reduce the risk for developing PE [6, 10]. The results
screening. In recent years, there has been an increasing from the second trimester screening might be useful for
uptake of cell free fetal DNA (cffDNA) in aneuploidy modifying the first trimester risk, assessing the effective-
screening. For screening programs that use a contin- ness of preventive prophylactic treatment and optimiz-
gent cffDNA approach, multiple marker screening has ing the management of at-risk pregnancies. Our study
been used as a first-tier screen to identify pregnancies at assessed the first stage of this contingent approach by
increased risk of fetal aneuploidy followed by cffDNA or using maternal characteristics with first and/or second
diagnostic testing [1, 2]. Maternal biochemical markers trimester biochemical markers to predict PE, in addition
used in multiple marker screening have long been asso- to other adverse pregnancy outcomes including gesta-
ciated with adverse pregnancy outcomes such as preec- tional hypertension and preterm birth.
lampsia (PE), preterm birth and intrauterine growth
restriction (IUGR) [3, 4]. Results from recent studies Methods
have shown that about 65% of early-onset PE can be Study population
predicted using first trimester maternal serum placental A retrospective case-control study was carried out using
growth factor (PlGF), pregnancy-associated plasma pro- frozen residual serum samples from women who had
tein-A (PAPP-A) and maternal demographics and his- undergone multiple marker screening and delivered at a
tory at a false positive rate (FPR) of 5% [5]. The accuracy tertiary center in Toronto, Ontario, Canada between Jan-
increased to 90% for early-onset PE and 80% for pre- uary 1, 2014 and October 31, 2017.
term PE with the addition of mean arterial blood pres- A combined dataset was created by linking multiple
sure (MAP) and uterine artery pulsatility index (UTPI) marker screening records with maternal and newborn
at a 5% FPR [5]. Moreover, studies have shown that records using patient identifiers and expected date of
the administration of nightly low-dose aspirin before delivery (EDD). Women who had a first and/or second
16 weeks of gestation, to women who are identified as trimester serum sample available were included in this
high-risk for PE, can prevent about 90% of early-onset study. The exclusion criteria included multiple preg-
PE cases [6]. Second trimester biochemical markers nancies, pregnancies with a fetal anomaly, and those
were also found to be helpful in identifying women at with unavailable or unmeasurable residual samples.
risk of developing PE, allowing for close surveillance and Cases were identified from maternal records of women
a timely intervention [7–9]. who had PE, gestational hypertension, or preterm birth.
The ideal screening test for PE should include maternal PE cases were further classified as preterm (women
characteristics, medical and family history, and biochem- who developed PE and gave birth at < 37 weeks of ges-
ical and biophysical markers [10]. In reality, biophysical tation), early-onset (women who developed PE and
markers such as MAP and especially UTPI might not be gave birth at < 34 weeks of gestation) or all PE. Women
readily accessible, especially to women in remote areas. with early-onset PE were included in the subgroup of
Since PlGF has recently been added to some screening preterm PE cases. Early-onset, preterm and term PE
programs to improve the accuracy of first trimester aneu- cases were included in the group of all PE cases. Each
ploidy screening, maternal characteristics (questionnaire) affected pregnancy was matched with three controls by
combined with biochemical markers (blood test) could date of blood sample draw (within 30 days), gestational
be used as a first-tier test to identify women for second- age at first blood sample draw (within 7 days), mater-
ary MAP (physical test) and UTPI (ultrasound scan), nal age (within 5 years), maternal ethnicity and amount
allowing for the expansion of current aneuploidy screen- of residual sample. Controls were selected from sin-
ing programs to include PE screening. A contingent PE gleton pregnancies that had a live full term birth and
screening approach would not only be cost effective but were not complicated by fetal anomalies, PE, pregnancy
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 3 of 14

hypertension, preterm birth, preterm premature rup- Statistical analysis


ture of the membranes and fetal growth restriction. The differences in the median of continuous variables
During the study period, most of the women were including maternal age at EDD, maternal weight and
screened using integrated prenatal screening. Oth- body mass index (BMI), gestational age at birth, gesta-
ers were screened using first trimester screening, tional age at first and second sample among cases and
enhanced first trimester screening or second trimester controls were tested using Mann-Whitney U-tests. Cat-
serum screening (QUAD). Integrated prenatal screen- egorical variables such as ethnicity, pregnancy outcome,
ing consisted of NT and PAPP-A measured between parity, gravidity, chronic hypertension, Type 1 diabetes,
11 + 0 and 13 + 6 weeks, and alpha fetoprotein (AFP), Type 2 diabetes, smoking and in vitro fertilization (IVF)
unconjugated estriol (uE3) and intact human chorionic were compared using chi-square or Fisher’s exact tests.
gonadotropin (hCG) measured between 15 + 0 and The concentrations of biochemical markers in this
18 + 6 weeks. Integrated prenatal screening results were study were expressed as Multiple of the Median (MoM)
reported after the second blood test was completed. calculated by dividing marker concentrations by the
First trimester screening consisted of NT, PAPP-A and expected median values at a particular gestational age.
free β-hCG measured between 11 + 0 and 13 + 6 weeks. The expected median was generated by weighted regres-
Enhanced first trimester screening was introduced in sion analysis on marker concentrations of all the con-
April 2016 which includes all markers used for first tri- trols. The MoM values were then adjusted for maternal
mester screening and at the same gestational window weight and were further corrected for ethnicity, parity
with the addition of PlGF and AFP. Second trimester and smoking status where applicable. The adjustment
QUAD consisted of AFP, uE3, total hCG and inhibin A factors for ethnicity, parity and smoking status were gen-
measured between 15 + 0 to 20 + 6 weeks. As all first erated based on MoM values calculated from the current
trimester samples had PAPP-A measured and all sec- study. As routinely performed in our multiple marker
ond trimester samples had AFP, uE3 and total hCG aneuploidy screening laboratory, for pregnancies with
measured, the concentrations of these markers from missing maternal weight, a weight was imputed based on
routine aneuploidy screening records were used in ethnic-specific median weight calculated from the study
this study. Other markers were measured using frozen population. No adjustment was made for pregnancies
residual samples. In addition, first trimester inhibin A with missing ethnicity, parity, smoking, insulin depend-
and second trimester PAPP-A and PlGF that have not ent diabetes mellitus (IDDM) or IVF status. For example,
been used for aneuploidy screening were also measured a pregnancy with missing ethnicity and smoking sta-
using frozen residual samples for this study. tus was treated as a Caucasian non-smoker. Multiple of
the median values of markers that were taken from the
routine screening records were already adjusted for eth-
Biochemical assays nicity, parity and smoking status using factors that were
All samples were processed and assayed in a local implemented in the screening software Alpha 8.0 (Logi-
genetic laboratory. Frozen serum samples had been cal Medical Systems Ltd, London). All analyses going
stored at -20 °C immediately after an initial assay for forward were based on adjusted MoM values, unless oth-
routine screening and at -70 °C four weeks later for long erwise specified. To assess if there were significant varia-
term storage. All except 33 samples had one freeze-thaw tions in biochemical markers in pregnancies with adverse
cycle, and none had been used in other studies. Samples outcomes, median MoM values of biochemical markers
having two freeze-thaw cycles were from women who from cases and controls were compared using Mann-
had originally enrolled for integrated prenatal screen- Whitney U-tests.
ing but did not provide a second sample in time to com- Of women who had both the first and second trimes-
plete the test. Therefore, the first sample was converted ter serum samples available, a subsequent data set was
to first trimester screening and was measured for free created to assess the change in biochemical marker con-
β-hCG. All assays were performed on the AutoDELFIA centrations and MoM values between the second and
analyzer (PerkinElmer). The AutoDELFIA routine dic- first trimester samples in the case and control groups.
tates a solid phase, two-site fluorometric assay using The pair-wise changes were calculated by subtracting the
two monoclonal antibodies directed against two sepa- value of the first trimester sample from the second tri-
rate antigenic sites on the molecule. One antibody is mester sample. The median changes in cases and controls
labelled with a fluorescent marker, and after incubation, were compared using Mann-Whitney U-tests.
the europium or samarium forms fluorescent chelates. Logistic regression was used to assess if the risk of
The fluorescence is proportional to the concentration of developing PE, gestational hypertension, and preterm
analyte in the sample. birth can be predicted using a combination of maternal
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 4 of 14

biochemical markers and maternal characteristics. Logis- biochemical markers seen in case and control groups.
tic regression analysis with backward selection was run The median MoM of first trimester PAPP-A, PlGF, and
with all first trimester biochemical markers to determine free-β hCG were significantly reduced among cases
the best predictive models for each case group using only compared to controls. Of women who developed PE,
first trimester markers. This was repeated using second the lowest median MoM of PAPP-A and PlGF were
trimester biochemical markers. For models with mater- seen in those with preterm PE (0.63 and 0.73), and
nal characteristics, an initial logistic regression with early-onset PE (0.54 and 0.72), respectively. First tri-
backward selection was run to identify significant mater- mester median MoM values of AFP increased in the all
nal characteristics for each case group. Varying biochem- PE, gestational hypertension and preterm birth groups.
ical marker combinations were then added to significant Free-β hCG decreased in the PE, gestational hyperten-
maternal characteristics to develop predictive models sion and preterm birth groups. No change in the first
with a combination of maternal characteristics and bio- trimester inhibin A median MoM was observed in case
chemical markers in each case group. Receiver operating groups.
characteristic (ROC) curves were created to determine In the second trimester, median MoM values of PlGF
detection rates (DR) at a 5%, 10%, and 20% FPR, respec- were lower in all PE, gestational hypertension and pre-
tively. The positive and negative likelihood ratios (LR) at term birth groups compared to controls. MoM values
each FPR were also estimated. All statistical analysis was of PAPP-A were lower in the PE group but not in the
carried out using SAS 9.4. gestational hypertension and preterm birth groups.
The study was approved by the Research Ethics Board Median MoM of hCG was higher in PE groups. The low-
of North York General Hospital on May 17, 2017. All est median MoM of PlGF (0.42) and PAPP-A (0.80) was
methods were performed in accordance with the relevant seen in women with early-onset PE and preterm PE,
guidelines and regulations (Declaration of Helsinki). respectively. Median MoM of PlGF was lower and AFP
was higher in women with preterm birth, compared to
Results controls.
In total, 608 cases and 1,815 controls were identified from Figure 1 illustrates the increase in PlGF concentrations
15,640 singleton pregnancies. The case group included between first and second trimester samples; revealing
147 pregnancies with PE (81 term, 49 preterm and 17 smaller increases among all PE (43.70 pg/ml), preterm PE
early-onset), 295 with gestational hypertension, and 166 (32.85 pg/ml), early-onset PE (21.40 pg/ml), and preterm
preterm birth. All but eight cases were matched to three birth cases (51.50 pg/ml) compared to controls (64.40 pg/
controls. Eight cases had 1-2 controls because the con- ml, p < 0.01) (Fig. 1). Trends among other biomarkers
trol samples were either unusable or unavailable. Samples were not statistically or clinically significant.
were available from both the first and second trimesters Table 3 gives the coefficients of logistic regressions
for 459 cases and 1358 controls. which indicate the degree of the associations between
Table 1 compares the maternal characteristics, medical biochemical markers, maternal characteristics and the
history, and pregnancy outcomes among case and control risk of developing PE, gestational hypertension, and
groups. Maternal age and ethnicity were similar among preterm birth. When maternal characteristics were not
case and control groups. Women with PE delivered two included in the model, first and second trimester PlGF
weeks earlier, and those with gestational hypertension were associated with all PE, preterm PE and early-onset
delivered one week earlier than the control group. The PE. First trimester PAPP-A, free-β hCG and inhibin
PE group included a greater proportion of women with a A were associated with all PE and preterm PE. When
history of chronic hypertension, higher BMI and of nul- maternal characteristics were included in the model, the
liparous women compared to controls. Live birth preg- associations between PlGF and all PE, preterm PE and
nancies were lowest among preterm birth cases (72%), early-onset PE remained statistically significant.
followed by early-onset PE (94%), preterm PE (98%), and Table 4 shows the estimated DR for PE, gestational
all PE cases (99%). Median gestational age was 88 days for hypertension and preterm birth at a FPR of 5%, 10%,
first trimester samples and 114 days for second trimes- and 20% using different biochemical marker combina-
ter samples in both case and control groups. There were tions with and without maternal characteristics. Detec-
more nulliparous women in the case groups. The smok- tion rates for preterm PE and early-onset PE were higher
ing status was similar in cases and controls. A greater than for all PE cases. Adding maternal characteristics to
proportion of IVF pregnancies was seen in the case group any combinations yielded better DR values than mater-
with gestational hypertension. nal characteristics or biochemical marker combination
Table 2 gives the median and interquartile range alone. Detection rates for both gestational hypertension
of MoM of first trimester and second trimester and preterm birth, using different biochemical marker
Table 1 Maternal characteristics, medical history, and pregnancy outcomes among case and control groups
Maternal Preeclampsia Gestational hypertension Preterm birth Controls
Characteristics (n = 295) (n = 166) (n = 1815)

All (n = 147) Preterm (n = 49) Early-onset (n = 17)

n Median n Median n Median n Median n Median n Median


(5th,95th percentile) (5th,95th percentile) (5th,95th percentile) (5th,95th percentile) (5th,95th percentile) (5th,95th percentile)

/Proportion(%) / Proportion (%) / Proportion(%) / Proportion(%) / Proportion(%) / Proportion(%)

NS NS NS NS NS
Maternal age, years 147 34.1 49 34.9 17 34.4 295 33.5 166 33.9 1815 33.5
(26.5-40.9) (26.0-42.8) (28.1-41.6) (26.4-41.4) (24.9-41.7) (26.3-39.9)
Maternal weight, kg 146 66.0* 49 65.4% 17 65.5NS 282 71.5* 161 60.5NS 1772 60.5
(48.6-113.2) (47.7-98.6) (42.3-97.7) (50.8-110.0) (45.5-92.7) (46.8-88.5)
Maternal BMI, kg/M2 124 25.7* 37 25.2% 13 24.8NS 264 26.3* 134 23.3NS 1560 22.4
(18.7-43.1) (18.6-36.0) (17.9-32.4) (19.5-38.5) (17.6-35.3) (17.9-32.9)
Huang et al. BMC Pregnancy and Childbirth

Gestational age at birth, weeks 147 37.0* 49 35.0* 17 33.0* 295 38.0* 166 31.0* 1815 39.0
(32.0-40.0) (28.0-36.0) (23.0-33.0) (36.0-40.0) (20.0-33.0) (37.0-41.0)
Gestational age 1st sample scan, days 143 88.0NS 48 87.0NS 17 87.0NS 286 88.0NS 158 87.0NS 1769 88.0
(81.0-93.0) (81.0-93.0) (81.0-93.0) (82.0-92.0) (81.0-93.0) (82.0-93.0)
nd NS NS NS NS NS
Gestational age 2 sample scan, days 125 114.0 43 114.0 15 117.0 223 114.0 128 115.0 1404 114.0
(107.0-128.0) (107.0-131.0) (107.0-134.0) (107.0-130.0) (106.0-129.0) (107.0-128.0)
(2022) 22:190

Ethnicity
Caucasian 36 24.5NS 8 16.3NS <6 -NS 111 37.6NS 50 30.1NS 589 32.5
Asian 85 57.8 35 71.4 12 70.6 154 52.2 90 54.2 983 54.2
Black 16 10.9 <6 - 0 0 19 6.4 11 6.6 137 7.5
Other/ 10 6.8 <6 - <6 - 11 3.7 15 9.0 106 5.8
Unknown
Medical History
Chronic hypertension 40 27.2* 11 22.4* <6 -* 20 6.8* 0 0 <6 -
NS
Diabetes Type 1 <6 0 0 0 0 0 0 <6 -NS <6 -
Diabetes <6 -% 0 0 0 0 <6 -NS 0 0 6 0.33
Type 2
APS/SLE 0 0 0 0 0 0 <6 -% <6 -NS 0 0
Smoking <6 -NS 0 0 0 0 10 3.4NS <6 -NS 31 1.7
Graviditya,b
1 65 45.1^ 22 46.8% 8 47.1NS 116 40^ 51 32.9NS 534 30.0
Parityb, c
Nulliparous 90 62.5* 27 57.4% 9 52.9NS 164 56.6* 80 51.6 ¶ 729 41.0
NS NS ¶
IVF <6 - <6 - 0 0 11 3.7 <6 -NS 25 1.4
Pregnancy Outcome d
Live Birth 146 99.3NS 48 98.0% 16 94.1 ¶ 295 100 120 72.3* 1815 100
*
p < 0.0001; ^ p < 0.001; ¶ p < 0.01; % p < 0.05; NS Not significant. The significance levels of the differences are reported in respect to the control group
Comparison between case and control groups: Chi-Square or Fisher’s exact test for categorical variables; Mann-Whitney U test for continuous variables
BMI, Body Mass Index
a
Gravidity is a binary variable categorized as 1 or > 1
b
Parity and Gravidity were missing among cases and controls: 3 missing among all preeclampsia cases(n = 144), 2 missing among preterm preeclampsia cases (n = 47), 5 missing among gestational hypertension cases(n = 290), 11
missing among preterm birth cases(n = 155), and 35 missing among controls(n = 1780)
c
Parity is a binary variable categorized as nulliparous or parous
d
Page 5 of 14

Pregnancy Outcome is a binary variable categorized as live birth or stillbirth/pregnancy loss


Table 2 Median multiple of the medians (MoMs) of biochemical markers among case and control groups
Marker Tested Preeclampsia Gestational Preterm birth Controls
hypertension

All Preterm Early-onset


n Median MoM (IQR) n Median MoM (IQR) n Median MoM (IQR) n Median MoM (IQR) n Median MoM (IQR) n Median MoM (IQR)

First trimester (T1)


PAPP-A Routinea 141 0.77 * 48 0.63 * 17 0.54 ¶ 283 0.93* 157 0.86 ^ 1746 1.10
Huang et al. BMC Pregnancy and Childbirth

(0.54,1.17) (0.49,0.96) (0.44,0.85) (0.61,1.30) (0.59,1.42) (0.74,1.59)


free-β hCG Studyb 141 0.81 ^ 48 0.82% 17 0.81 NS 284 0.88% 157 0.88 NS 1751 0.98
(0.53,1.13) (0.52,1.12) (0.52,1.12) (0.63,1.30) (0.60,1.31) (0.66,1.51)
PlGF Study 141 0.76 * 48 0.73 * 17 0.72 ^ 284 0.92 * 157 0.91 ^ 1751 1.01
(0.62,1.09) (0.57,0.97) (0.51,0.79) (0.69,1.18) (0.70,1.15) (0.80,1.31)
(2022) 22:190

AFP Study 141 1.16 ¶ 48 1.17 NS 17 0.92 NS 284 1.11 ^ 157 1.27 * 1751 1.00
(0.86,1.53) (0.87,1.54) (0.86,1.49) (0.87,1.46) (0.90,1.64) (0.78,1.33)
Inhibin A Study 141 1.03 NS 48 1.07 NS 17 0.98 NS 284 1.08 NS 157 0.95 NS 1751 1.00
(0.79,1.41) (0.83,1.53) (0.78,1.50) (0.79,1.46) (0.72,1.36) (0.77,1.35)
Second trimester (T2)
AFP Routine 126 1.03 NS 43 1.01 NS 15 1.01 NS 223 1.04 NS 130 1.18 * 1402 1.01
(0.83,1.28) (0.83,1.25) (0.83,1.28) (0.83,1.28) (0.94,1.47) (0.83,1.22)
 ­uE3 Routine 91 1.00 NS 31 0.92% 10 0.74 NS 163 1.01 NS 101 0.98 NS 1056 1.01
(0.80,1.15) (0.74,1.08) (0.67,1.06) (0.89,1.16) (0.83,1.15) (0.86,1.18)
hCG Routine 91 1.17% 31 1.30% 10 1.62 ¶ 163 1.14 NS 101 1.09 NS 1056 1.06
(0.89,1.79) (1.01,1.97) (1.27,1.97) (0.85,1.63) (0.81,1.67) (0.75,1.48)
PAPP-A Study 110 0.82 ¶ 36 0.80% 13 0.87 NS 189 0.90 NS 97 0.98 NS 1140 0.99
(0.57,1.26) (0.53,1.09) (0.55,1.26) (0.60,1.37) (0.60,1.35) (0.68,1.41)
PlGF Study 113 0.75 * 37 0.74 * 13 0.42* 197 0.89 ¶ 103 0.79 ^ 1187 1.02
(0.53,1.08) (0.42,0.88) (0.28,0.74) (0.66,1.23) (0.61,1.21) (0.76,1.37)
Inhibin ­Ac Routine 6 1.45% <6 1.51 NS 0 - 12 0.95 NS 9 1.36 NS 75 0.98
(1.19,1.74) (-) (0.76,1.21) (0.98,2.12) (0.73,1.40)
*
p < 0.0001; ^ p < 0.001; ¶ p < 0.01; % p < 0.05; NS Not significant. The significance levels of the differences are reported in respect to the control group
MoM: Multiple of the median; IQR: Interquartile range
PAPP-A pregnancy-associated plasma protein A, hCG human chorionic gonadotropin, PlGF placental growth factor, AFP alpha feto-protein, uE3 unconjugated Estriol
a
Routine- marker measurement from routine screening tests
b
Study- marker (re)/measured specifically for this study
c
The IQR of inhibin A for preterm PE and early-onset PE cannot be computed due to small sample size
Page 6 of 14
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 7 of 14

Fig. 1 Box and whisker plots for changes in PlGF concentration between the first and second trimester samples among preeclampsia (PE),
gestational hypertension, preterm birth cases and controls. Notes: The ends of the box are the upper and lower quartiles of the change. The
horizontal line inside the box marks the median change. The two vertical lines outside the box are the whiskers extending to the greatest and
smallest changes. The white circles indicate the outliners and the diamond represents the mean change

combinations with and without maternal characteristics, Discussion


were low. Our study assessed whether first and second trimes-
Table 5 gives the positive and negative LR and their ter biochemical markers used in prenatal aneuploidy
95% confidence intervals at a FPR of 5%, 10%, and 20% screening could accurately identify pregnancies at risk of
using different marker combinations for PE, gestational developing adverse pregnancy outcomes, mainly PE. We
hypertension and preterm birth. For each case group, found that the combination of maternal characteristics
the highest positive LR and lowest negative LR were with biochemical markers in either trimester can pro-
observed using maternal characteristics with second vide reasonable accuracy in identifying women at risk of
trimester serum markers, followed by maternal char- developing preterm and early-onset PE but not women at
acteristics with first trimester serum markers. With the risk for gestational hypertension or preterm birth. PAPP-
latter marker combination, at a FPR of 20%, there was a A and PlGF in the first and second trimesters were the
4.1-fold increase in the odds of developing early-onset biochemical markers most reliably associated with and
PE in women with a positive test result. A negative result predictive of adverse pregnancy outcomes, particularly
would decrease the odds of having the condition by 4.6- early-onset PE.
fold. With the same marker combination, at the same Our results have confirmed previously reported asso-
FPR, the highest positive LR and lowest negative LR were ciations between individual biochemical markers and
observed for early-onset PE. PE and preterm birth [7–9]. Similar to previous stud-
Figure 2 demonstrates the ROC curves for PE, gesta- ies, we found the performance of PE prediction to be
tional hypertension and preterm birth using maternal improved by using multiple biochemical markers in
characteristics and PlGF and PAPP-A in the first trimes- combination with maternal characteristics [5, 11, 12].
ter. The area under the ROC curve (AUC) was largest The most commonly used marker combination of first
among preterm PE (0.83) and early-onset PE (0.82) cases trimester PAPP-A and PlGF along with maternal char-
followed by all PE (0.78), gestational hypertension (0.75) acteristics can predict 65% and 76% of early-onset PE
and preterm birth cases (0.61). The best screening perfor- with a FPR of 10% and 20% respectively. Second tri-
mance for a FPR < 25% was seen for early-onset PE fol- mester biochemical markers alone can identify 88% of
lowed by preterm PE and all PE. the pregnancies at risk of developing early-onset PE
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 8 of 14

Table 3 Logistic regression coefficient and standard error for the prediction of adverse pregnancy outcomes using maternal
characteristics and biochemical markers
Marker Preeclampsia Gestational Preterm birth
hypertension

All Preterm Early-onset


Coefficient SE Coefficient SE Coefficient SE Coefficient SE Coefficient SE

T1 Biochemical Markersa
Intercept -1.1692¶ 0.3565 -1.1523NS 0.5985 -1.6729% 0.7381 -1.1012* 0.2421 -1.9447* 0.3125
PAPP-A -0.7193^ 0.1853 -1.5537* 0.3874 - - -0.4694* 0.1200 -0.3024% 0.1464
free-β hCG -0.4488¶ 0.1551 -0.6470% 0.2718 - - -0.2657¶ 0.1020 - -
PlGF -0.9420^ 0.2780 -1.3867¶ 0.5224 -3.3491^ 0.9404 -0.3503% 0.1767 -0.5498% 0.2334
AFP 0.1868% 0.0898 - - - - - - 0.3657^ 0.1094
Inhibin A 0.5241^ 0.1566 0.8701^ 0.2274 - - 0.4195¶ 0.1290 -
T2 Biochemical Markersb
Intercept -1.5231* 0.3796 -2.5203* 0.6174 -2.8837¶ 1.0162 -2.2967* 0.2990 -3.9344* 0.3311
^
AFP - −
- - - - 0.7503 0.2113 1.2020* 0.2441
− −
 ­uE3 - - - - - - - -
* ¶ % −
hCG 0.5895 0.1483 0.5332 0.1899 0.6212 0.2678 - - -
% % −
PAPP-A -0.6405 0.2501 - - - - -0.3987 0.1841 -
^ ¶ % −
PlGF -1.0464 0.3104 -1.9338 0.6259 -3.1643 1.2555 - - -
c
Maternal Characteristics + T1 PlGF + T1 PAPP-A
Intercept -6.9419* 1.0096 -5.2600¶ 1.6327 -5.7611% 2.4117 -6.2080* 0.7445 -1.7145* 0.2782
^ ¶ ¶ NS
PlGF -0.9925 0.2829 -1.7875 0.5428 -2.8433 0.9518 -0.3273 0.1913 -0.4626NS 0.2366
PAPP-A -0.9265* 0.1927 -1.5843* 0.4034 -1.1291NS 0.6044 -0.6066* 0.1303 -0.4302¶ 0.1575
Maternal Characteristicsc + All T1 Biochemical Markers a
Intercept -7.4496* 1.0422 -5.9671^ 1.7274 -6.2326% 2.5740 -6.5085* 0.7625 -2.0900* 0.3673
PAPP-A -0.9335* 0.1960 -1.6698* 0.4025 -1.1970% 0.5981 -0.6177* 0.1337 -0.3836% 0.1607
free-β hCG -0.3955% 0.1607 -0.5637% 0.2787 -0.2477NS 0.3872 -0.1426NS 0.1038 -0.0709NS 0.1302
PlGF -0.7713¶ 0.2817 -1.2278% 0.5345 -2.4894% 0.9844 -0.2676NS 0.1912 -0.4342NS 0.2397
AFP 0.2180% 0.1004 0.1505NS 0.2163 -0.0942NS 0.4392 0.1697% 0.0863 0.3204¶ 0.1102
Inhibin A 0.4649¶ 0.1652 0.8114^ 0.2419 0.5278NS 0.3822 0.2345NS 0.1443 0.0016NS 0.1835
Maternal Characteristicsc + All T2 Biochemical Markers b
Intercept -5.5285* 1.3954 -1.1855NS 2.3472 -3.1794NS 3.9327 -7.2465* 1.1403 -2.9484* 0.7141
NS NS NS % ¶
AFP -0.4117 0.3438 -0.7740 0.5826 -1.1304 1.1195 0.4014 0.2567 0.9651 0.3051
 ­uE3 -0.4744NS 0.5340 -0.9717NS 0.9421 -1.9243NS 1.6075 -0.0260NS 0.4512 -0.5001NS 0.5564
hCG 0.6030^ 0.1700 0.9158^ 0.2644 0.7432NS 0.4389 0.2084NS 0.1527 0.1450NS 0.2074
PAPP-A -0.8083¶ 0.2598 -0.8946% 0.4279 0.3250NS 0.5517 -0.6117¶ 0.2133 -0.5368 NS 0.3105
PlGF -1.0086¶ 0.3186 -1.4560% 0.5983 -3.0874% 1.2967 -0.3119NS 0.2161 -0.3736NS 0.2901
*
p < 0.0001; ^ p < 0.001; ¶ p < 0.01; % p < 0.05; NS Not significant, The significance levels of the differences are reported in respect to the control group
SE Standard error
PAPP-A pregnancy-associated plasma protein A, hCG human chorionic gonadotropin, PlGF placental growth factor, AFP alpha feto-protein, uE3 unconjugated Estriol
a
First Trimester (T1) Biochemical Markers: PAPP-A, free-β hCG,PlGF, AFP, Inhibin A
b
Second Trimester (T2) Biochemical Markers: AFP, uE3, hCG, PAPP-A, PlGF
c
Through backward selection significant maternal characteristics for each case group were identified. Significant maternal characteristics for each case group
include—All PE: age, weight, ethnicity, nulliparous; Preterm PE: age, weight, ethnicity, gravidity; Early-onset PE: age, gravidity; Gestational Hypertension: age, weight,
ethnicity, nulliparous; Preterm Birth: nulliparous

with a FPR of 20%. However, this will require the addi- control group, particularly in those affected by early-
tion of PAPP-A and PlGF which are not currently used onset PE.
in second trimester aneuploidy screening. In addition, Low first trimester PAPP-A and PlGF have been asso-
we found that the increase of PlGF with gestational age ciated with adverse pregnancy outcomes, especially PE
was smaller in women affected by PE compared to the in multiple studies [5, 11–13]. The case-control study
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 9 of 14

Table 4 Model-predicted screening performance using maternal characteristics and different serum marker combinations
Marker Preeclampsia Gestational Preterm birth
hypertension
All Preterm Early-onset

T1 Biochemical Markers a
AUC​ 0.70 (0.654,0.753) 0.80 (0.734,0.867) 0.76 (0.635,0.890) 0.63 (0.593,0.663) 0.64 (0.588,0.685)
DR% (CI) for 5% FPR 23 (15.9,29.8) 25 (12.8,37.3) 35 (12.6,58.0) 13 (9.5,17.4) 17 (11.3,23.1)
DR% (CI) for 10% FPR 33 (25.1,40.6) 42 (27.7,55.6) 35 (12.6,58.0) 19 (14.8,24.0) 25 (18.1,31.6)
DR% (CI) for 20% FPR 55 (46.8,63.2) 73 (60.4,85.5) 65 (42.0,87.4) 35 (29.8,40.9) 40 (32.5,47.8)
T2 Biochemical Markers b
AUC​ 0.69 (0.620,0.755) 0.76 (0.645,0.873) 0.83 (0.608,1.000) 0.58 (0.531,0.638) 0.65 (0.581,0.725)
DR% (CI) for 5% FPR 23 (13.2,32.1) 29 (11.0,47.4) 50 (15.4,84.6) 12 (6.7,18.0) 19 (9.8,28.5)
DR% (CI) for 10% FPR 31 (20.2.41.1) 42 (21.9,61.4) 63 (29.0,96.0) 20 (13.1,26.9) 29 (18.6,40.2)
DR% (CI) for 20% FPR 44 (32.8,55.2) 58 (38.6,78.1) 88 (64.6,100.0) 28 (20.7,36.2) 38 (26.9,49.8)
Maternal Characteristics Only c
AUC​ 0.70 (0.661,0.745) 0.71 (0.632,0.779) 0.68 (0.558,0.806) 0.74 (0.707,0.771) 0.55 (0.510,0.594)
DR% (CI) for 5% FPR 18 (11.9,24.5) 15 (4.7,25.0) 18 (0,35.8) 22 (17.1,26.9) 7 (3.2,11.5)
DR% (CI) for 10% FPR 31 (23.2,38.3) 30 (16.7,42.9) 18 (0,35.8) 34 (28.7,39.9) 15 (9.6,21.1)
DR% (CI) for 20% FPR 45 (36.6.52.9) 47 (32.5,61.1) 41 (17.8,64.6) 54 (48.3,60.0) 25 (18.4,32.3)
Maternal Characteristics c + T1 PlGF + T1 PAPP-A
AUC​ 0.78 (0.740,0.821) 0.83 (0.778,0.890) 0.82 (0.702,0.944) 0.75 (0.723,0.787) 0.61 (0.560,0.660)
DR% (CI) for 5% FPR 30 (22.2,37.6) 35 (21.0,48.6) 41 (17.8,64.6) 23 (18.2,28.4) 15 (9.3,20.9)
DR% (CI) for 10% FPR 42 (33.3,49.9) 52 (37.7,66.6) 65 (42.0,87.4) 39 (33.6,45.4) 21 (14.6,27.9)
DR% (CI) for 20% FPR 60 (51.7,68.1) 67 (53.8,80.9) 76 (56.3,96.6) 58 (52.0,63.8) 36 (28.5,44.1)
Maternal Characteristics c + All T1 Biochemical Markers a
AUC​ 0.79 (0.749,0.828) 0.84 (0.780,0.897) 0.84 (0.725,0.960) 0.76 (0.728,0.791) 0.64 (0.589,0.691)
DR% (CI) for 5% FPR 30 (22.2,37.6) 37 (23.0,50.9) 47 (23.3,70.8) 24 (18.6,28.8) 19 (12.8,25.6)
DR% (CI) for 10% FPR 45 (36.9,53.6) 59 (44.5,72.9) 71 (48.9,92.3) 40 (34.3,46.1) 29 (21.4,36.1)
DR% (CI) for 20% FPR 59 (50.9,67.4) 72 (58.7,84.8) 82 (64.2,100.0) 59 (52.7,64.6) 40 (31.8,47.7)
Maternal Characteristics c + All T2 Biochemical Markers b
AUC​ 0.78 (0.723,0.836) 0.80 (0.703,0.901) 0.83 (0.594,1.000) 0.76 (0.713,0.807) 0.70 (0.621,0.779)
DR% (CI) for 5% FPR 27 (17.2,37.6) 39 (19.2,59.0) 75 (45.0,100.0) 29 (20.8,37.3) 18 (7.7,27.4)
DR% (CI) for 10% FPR 38 (27.2,49.5) 48 (27.4,68.2) 75 (45.0,100.0) 40 (31.3,49.1) 37 (24.3,49.4)
DR% (CI) for 20% FPR 63 (51.9,74.1) 65 (45.8,84.7) 75 (45.0,100.0) 56 (47.4,65.4) 58 (45.1,70.7)
DR Detection Rate, FPR False Positive Rate, AUC​ Area under the Curve, CI 95% confidence interval
a
First Trimester (T1) Biochemical Markers: PAPP-A, free-β hCG,PlGF, AFP, Inhibin A. Refer to Table 3 for specific markers used in each case group
b
Second Trimester (T2) Biochemical Markers: AFP, uE3, hCG, PAPP-A, PlGF. Refer to Table 3 for specific markers used in each case group
c
Significant maternal characteristics for each case group include- All PE: age, weight, ethnicity, nulliparous; Preterm PE: age, weight, ethnicity, gravidity; Early-onset
PE: age, gravidity; Gestational Hypertension: age, weight, ethnicity, nulliparous; Preterm Birth: nulliparous

by Keikkala et al. (2016) investigating maternal serum described association between first trimester inhibin A
samples from 8-13 weeks of gestation found lower and PE was not confirmed in our study [14, 15].
median MoM values for PlGF and PAPP-A in women Second trimester PlGF and PAPP-A median MoM val-
with PE compared to controls [13]. Consistent with ues were lower in our study than those reported by other
our study, the lowest median MoM values were seen in studies [9, 16]. However, this was mainly true for PlGF
cases with preterm and early-onset PE. As with previ- and not significant for PAPP-A in women with preterm
ous studies, our study found median first trimester AFP or early-onset PE, likely due to the small number of preg-
MoM values to be higher and median free β-hCG MoM nancies in this group. We found an increase in total hCG
values to be lower in women with PE, although the sig- in PE pregnancies, as reported in previous studies [17].
nificance of these changes was smaller in comparison In contrast, no change in Inhibin A was seen in our study
to those of PAPP-A and PlGF [8, 9]. The previously although reported as an optimal second trimester marker
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 10 of 14

Table 5 The positive and negative likelihood ratios using maternal characteristics and different serum marker combinations
Marker Preeclampsia Gestational hypertension Preterm birth

All Preterm Early-onset


LR + (CI) LR- (CI) LR + (CI) LR- (CI) LR + (CI) LR- (CI) LR + (CI) LR- (CI) LR + (CI) LR- (CI)

T1 Biochemical Markersa
5% FPR 4.53 0.81 4.95 0.79 6.99 0.68 2.66 0.91 3.40 0.87
(3.14,6.53) (0.74,0.89) (2.91,8.42) (0.67,0.93) (3.56,13.73) (0.48,0.97) (1.86,3.81) (0.87,0.96) (2.29,5.08) (0.81,0.94)
10% FPR 3.27 0.75 4.15 0.65 3.52 0.72 1.94 0.90 2.47 0.84
(2.49,4.31) (0.66,0.84) (2.89,5.97) (0.51,0.82) (1.82,6.79) (0.51,1.02) (1.47,2.55) (0.84,0.95) (1.82,3.34) (0.76,0.92)
20% FPR 2.75 0.56 3.64 0.34 3.23 0.44 1.77 0.81 2.00 0.75
(2.30,3.28) (0.47,0.68) (2.92,4.32) (0.21,0.54) (2.25,4.65) (0.23,0.84) (1.47,2.12) (0.74,0.88) (1.62,2.48) (0.66,0.85)
T2 Biochemical Markersb
5% FPR 4.53 0.81 5.83 0.75 10.00 0.53 2.40 0.92 3.74 0.85
(2.72,7.55) (0.72,0.92) (2.93,11.61) (0.58,0.96) (4.72,21.20) (0.26,1.05) (1.43,4.24) (0.86,0.99) (2.12,6.59) (0.76,0.96)
10% FPR 3.07 0.77 4.17 0.65 6.25 0.42 2.00 0.89 2.84 0.79
(2.06,4.55) (0.66,0.90) (2.49,6.97) (0.46,0.91) (3.52,11.09) (0.17,1.02) (1.34,2.98) (0.81,0.97) (1.87,4.32) (0.67,0.92)
20% FPR 2.20 0.70 2.92 0.52 4.38 0.16 1.42 0.89 1.86 0.78
(1.65,2.94) (0.57,0.86) (2.03,4.20) (0.32,0.84) (3.26,5.87) (0.03,0.98) (1.05,1.93) (0.80,1.00) (1.34,2.58) (0.64,0.94)
Maternal Characteristics Only
5% FPR 3.63 0.86 2.98 0.90 3.53 0.87 4.40 0.82 1.47 0.98
(2.42,5.44) (0.80,0.93) (1.46,6.08) (0.80,1.01) (1.23,10.05) (0.70,1.08) (3.26,5.95) (0.77,0.88) (0.80,2.68) (0.93,1.02)
10% FPR 3.08 0.77 2.98 0.78 1.78 0.91 3.43 0.73 1.53 0.94
(2.32,4.08) (0.69,0.86) (1.88,4.72) (0.65,0.94) (0.63,5.03) (0.73,1.14) (2.76,4.25) (0.67,0.80) (1.03,2.29) (0.88,1.01)
20% FPR 2.24 0.69 2.34 0.67 2.04 0.74 2.71 0.57 1.27 0.93
(1.82,2.75) (0.60,0.80) (1.70,3.22) (0.51,0.87) (1.15,3.63) (0.50,1.10) (2.34,3.12) (0.50,0.65) (0.95,1.69) (0.85,1.03)
Maternal Characteristicsc + T1 PlGF + T1 PAPP-A
5% FPR 5.96 0.74 6.92 0.69 8.20 0.62 4.64 0.81 3.00 0.89
(4.28,8.29) (0.66,0.82) (4.43,10.83) (0.56,0.85) (4.48,15.01) (0.42,0.92) (3.43,6.27) (0.76,0.86) (1.94,4.65) (0.83,0.96)
10% FPR 4.14 0.65 5.19 0.53 6.44 0.39 3.93 0.67 2.11 0.88
(3.24,5.29) (0.56,0.75) (3.80,7.09) (0.39,0.72) (4.41,9.41) (0.21,0.75) (3.20,4.83) (0.61,0.74) (1.50,2.98) (0.80,0.95)
20% FPR 2.98 0.50 3.36 0.41 3.80 0.29 2.89 0.53 1.81 0.80
(2.52,3.52) (0.41,0.62) (2.69,4.20) (0.27,0.62) (2.88,5.04) (0.13,0.69) (2.51,3.33) (0.46,0.61) (1.43,2.29) (0.70,0.90)
Maternal Characteristicsc + All T1 Biomarkersa
5% FPR 5.96 0.74 7.36 0.66 9.37 0.56 4.71 0.80 3.82 0.85
(4.28,8.29) (0.66,0.82) (4.78,11.32) (0.53,0.83) (5.44,16.16) (0.36,0.87) (3.49,6.36) (0.75,0.86) (2.58,5.65) (0.79,0.92)
10% FPR 4.50 0.61 5.84 0.46 7.03 0.33 4.00 0.66 2.87 0.79
(3.57,5.69) (0.52,0.71) (4.41,7.74) (0.33,0.65) (5.01,9.86) (0.16,0.68) (3.26,4.91) (0.60,0.73) (2.14,3.84) (0.71,0.88)
20% FPR 2.95 0.51 3.58 0.35 4.11 0.22 2.93 0.52 1.98 0.75
(2.49,3.49) (0.42,0.63) (2.92,4.40) (0.22,0.56) (3.24,5.23) (0.08,0.62) (2.55,3.36) (0.45,0.60) (1.59,2.48) (0.66,0.86)
Maternal Characteristicsc + All T2 Biomarkersb
5% FPR 5.46 0.76 7.80 0.64 14.86 0.26 5.79 0.75 3.48 0.87
(3.39,8.81) (0.66,0.88) (4.32,14.08) (0.46,0.89) (9.04,24.42) (0.08,0.87) (3.84,8.74) (0.66,0.84) (1.84,6.56) (0.77,0.98)
10% FPR 3.82 0.69 4.77 0.58 7.43 0.28 4.00 0.67 3.65 0.70
(2.68,5.46) (0.57,0.82) (2.97,7.66) (0.39,0.86) (4.74,11.63) (0.08,0.92) (2.96,5.41) (0.57,0.77) (2.46,5.42) (0.58,0.86)
20% FPR 3.14 0.46 3.25 0.44 3.73 0.31 2.81 0.55 2.88 0.53
(2.51,3.92) (0.34,0.63) (2.34,4.51) (0.25,0.76) (2.45,5.70) (0.09,1.04) (2.28,3.47) (0.44,0.67) (2.23,3.74) (0.39,0.72)
LR ( +) Positive Likelihood Ratio, LR (-): Negative Likelihood Ratio, CI 95% confidence Interval
a
First Trimester (T1) Biochemical Markers: PAPP-A, free-β hCG,PlGF, AFP, Inhibin A. Refer to Table 3 for specific markers used in each case group
b
Second Trimester (T2) Biochemical Markers: AFP, uE3, hCG, PAPP-A, PlGF. Refer to Table 3 for specific markers used in each case group
c
Significant maternal characteristics for each case group include- All PE: age, weight, ethnicity, nulliparous; Preterm PE: age, weight, ethnicity, gravidity; Early-onset
PE: age, gravidity; Gestational Hypertension: age, weight, ethnicity, nulliparous; Preterm Birth: nulliparous

by others [18]. We suspect this is due to the small num- gestational hypertension and preterm birth cases was
ber of early PE pregnancies involved in our study. Lastly, consistent with other studies in general although the
the change in the first and second trimester markers in magnitude of the change was variable [19, 20].
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 11 of 14

Fig. 2 ROC curves for models including maternal characteristics and first trimester PAPP-A + PlGF for preterm delivery (AUC 0.61), gestational
hypertension (AUC 0.75), all preeclampsia (PE) (AUC 0.78), preterm preeclampsia (AUC 0.83), and early-onset preeclampsia (AUC 0.82) cases

In recent years, the results of several large clinical tri- combination of maternal characteristics and biochemical
als have suggested that the performance of PE screening markers in the first trimester can predict the risk of PE
can be improved by using multiple biochemical mark- with reasonable accuracy, the current aneuploidy screen-
ers together with maternal characteristics and biophysi- ing could be expanded to include PE screening to identify
cal markers [5, 12, 21]. A prospective study by Akolekar women at increased risk of developing PE, with biophysi-
et al. on 58,884 singleton pregnancies at 11–13 weeks cal markers to be followed as a second line or contingent
reported a 50.5% DR at a 10% FPR with maternal charac- screen. In our study, the PE screening performance was
teristics alone for early-onset PE, 74.3% with the addition comparable to previous studies when using maternal
of PlGF and PAPP-A and 89.7% with the addition of bio- characteristics in combination with first trimester PAPP-
physical markers, MAP and UTPI [5]. The DR improved A and PlGF. At a FPR of 20%, the DR was 76%, 67% and
to 96.3% when maternal characteristics and biochemical 60% for early-onset PE, preterm PE and all PE, respec-
and biophysical markers were combined [12].Similarly, tively. This suggests that for 20% of women who screen
in a prospective study of 35,948 singleton pregnan- positive for early-onset PE, a contingent PE screening
cies at 11-13 weeks’ gestation, O’Gorman et al. found strategy using MAP and UTPI followed right after the
that combined screening obtained a DR of 75% and 47% biochemical screen can predict up to 76% of early-onset
at a 10% FPR for preterm PE and term PE pregnancies, PE. As with the one-time screening approach, our con-
respectively [12]. When such modelling combinations tingent screen can yield final PE screening results before
were applied to the ASPRE trial, similar DR values were 16 weeks of gestation in order to initiate prophylactic
observed. First-trimester screening for preterm PE with a therapy with Aspirin.
risk cut-off of 1 in 100 detected 76.7% of preterm PE and Although a contingent approach requires minimal
43.1% of term PE pregnancies, at a screen-positive rate of change to the current screening program, there is a
10.5% and a FPR of 9.2% [21]. chance of missing the optimal time window for prophy-
While the best performance can be achieved by com- lactic therapy if there is a delay in measuring MAP and
bining multiple variables (e.g., maternal characteristics, UTPI. Based on our local experience of utilizing MAP
medical and family history, biochemical and biophysical and UTPI, the gestational age at first trimester screening
markers), MAP and especially UTPI might not be read- and the test turnaround time, we anticipated a substantial
ily accessible, especially for women in remote areas. If the proportion of women could receive the final result before
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 12 of 14

16 weeks of gestation. When timely MAP and UTPI The strengths of our study include the identification of
examinations are not feasible, maternal characteristics cases and controls from a routine unselected screening
and first trimester biochemical markers only can also be population, representing a true sample of women under-
used as a ‘mini’ PE screening test to identify women who going prenatal screening in Ontario. Also unique to our
might benefit from prophylactic therapy. Nevertheless, study is the availability of both first and second trimes-
the addition of secondary MAP and UTPI would improve ter serum samples for most cases and controls. Having
the test accuracy by reducing its FPR. samples from both trimesters enabled us to investigate
First trimester screening aims to identify those at risk the change in biochemical markers between the first and
of PE within the target window for treatment with Aspi- second trimesters, which might provide additional infor-
rin. Second trimester screening, on the other hand, is mation for PE screening and monitoring. The limitations
useful for patient triage, by identifying pregnancies that of this study include the transfer of some women to other
need close surveillance and more urgent medical atten- obstetrical centres, potentially lowering the incidence
tion. For example, numerous studies have focused on sec- of PE in our population. Additionally, our local popula-
ond trimester PlGF or soluble fms-like tyrosine kinase-1 tion includes a greater proportion of women of Asian
(sFlt-1)/PlGF ratio [22]. In our study, in addition to the ancestry compared with other studies. Varying ethnicity
PE screening performance using second trimester mark- between studies may impact PE prevalence; in particular,
ers, we compared the changes in biochemical mark- a lower prevalence of PE among Asian women has been
ers between the first and second trimester samples. For noted previously [26]. Our study lacked complete infor-
women who developed PE, the increase in PlGF concen- mation on all maternal characteristics and as such, the
trations was significantly smaller compared to controls. accuracy of PE screening using maternal characteristics
These results have been confirmed by the findings from cannot be directly compared to studies where maternal
other studies [23–25]. Theoretically, for women who are characteristics were explicitly collected for PE screening.
identified to be at high risk for PE in the first trimester, However, the contribution of biochemical markers and
a repeat test for PlGF and PAPP-A in the second trimes- overall PE screening performance using the combina-
ter might provide useful information which can be used tion of maternal characteristics and biochemical mark-
for risk modification. Nevertheless, a calculation of bio- ers were consistent with previous studies. In a real-life
marker changes between trimesters did not yield a better clinical setting, it will be possible to collect all maternal
prediction for PE than using second trimester markers characteristics missed in this study. Since information
alone in our study. Further investigation of this trend may on MAP and UTPI was not available to our study, we
provide clinicians with valuable information for monitor- were not able to assess the final performance of a contin-
ing and early detection of high-risk women. gent PE screening strategy. The contingent approach we
In our study, the preterm birth screening performance described in this study lacked validation. Implement of
was less optimal than previously reported [20]. A pos- this screening strategy would require close collaborations
sible reason for the lower DR is that unlike some previ- of a multidisciplinary team. In addition to timely multi-
ous studies, our preterm birth group excluded all women ple marker screening, MAP and UTPI results, a quality
with PE and gestational hypertension. Incomplete mater- assurance scheme for MAP and UTPI should also be in
nal characteristics data may have also contributed to the place before the test can be adopted for clinical utiliza-
lower-than-expected DR values for preterm birth. tion. Two studies are currently underway in our program
Our study assessed the accuracy of a PE screening to validate the performance of the first-tier PE screen-
approach originating from established multiple marker ing in a different population, and to assess the feasibility,
aneuploidy screening. For screening programs that use and patient and provider acceptance of the contingent
primary cffDNA test, a separate system that involves screening approach. Nevertheless, we achieved our goal
clinical, biochemical and ultrasound expertise would be of assessing the first-tier of a contingent screening strat-
required in order to introduce PE screening. While the egy, one that provides a reasonable performance, war-
patient pathway and screening process might be different ranting expansion of current aneuploidy screening to
in a program specifically designed for PE screening, our include preeclampsia.
findings relating to how biochemical markers may assist
in identifying women at risk of developing PE can be Conclusions
generalized. Although an ideal PE screening test should Our study, based on a routine aneuploidy screening,
incorporate all test components, the contingent approach showed that the combination of maternal character-
we have described in this study provides an option to istics and first trimester serum PAPP-A and PlGF can
programs that are unable to offer MAP and UTPI to all provide reasonable performance for PE screening.
screened women due to limited resources. If our study results can be validated by prospective
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 13 of 14

studies, the current aneuploidy screening program can Consent for publication
Not applicable.
be expanded to include identification of women at risk
of developing PE, particularly early-onset PE. As a con- Competing interests
tingent strategy, it can provide first-tier PE screening The reagent kits used in the study were provided by PerkinElmer free of
charge. Professor Howard Cuckle is a consultant of PerkinElmer. Dr. Elad Mei-
with minimal associated costs and minimal change to Dan is an Editorial Board Member of BMC Pregnancy & Childbirth. The rest of
the current workflow. the authors declare to have no competing interest.

Author details
1
Abbreviations Genetics Program, North York General Hospital, 4001 Leslie Street, Toronto,
AFP: α-Fetoprotein; AUC​: Area Under the Curve; BMI: Body mass index; ON M2K 1E1, Canada. 2 Prenatal Screening Ontario, Better Outcomes Registry
cfDNA: Cell-free fetal DNA; DR: Detection rate; EDD: Expected date of deliv‑ & Network (BORN) Ontario, Ottawa, ON, Canada. 3 Department of Obstetrics
ery; FPR: False positive rate; hCG: Human chorionic gonadotrophin; IUGR​ and Gynecology, University of Toronto, Toronto, ON, Canada. 4 Department
: Intrauterine growth restriction; IVF: In vitro fertilization; MAP: Mean arterial of Paediatrics, University of Toronto, Toronto, ON, Canada. 5 Department
blood pressure; MoM: Multiple of the median; NT: Nuchal translucency; of Molecular Genetics, University of Toronto, Toronto, ON, Canada. 6 Faculty
PAPP-A: Pregnancy-associated plasma protein A; PE: Preeclampsia; PlGF: of Medicine, Tel Aviv University, Tel Aviv, Israel. 7 Maternal and Newborn Pro‑
Placental growth factor; QUAD: Second trimester serum screening; ROC: gram, North York General Hospital, Toronto, ON, Canada.
Receiver operating characteristic; sFlt-1: Soluble fms-like tyrosine kinase-1;
T1: First trimester; T2: Second trimester; uE3: Unconjugated estriol; UTPI: Received: 4 May 2021 Accepted: 16 February 2022
Uterine artery pulsatility index.

Acknowledgements
The authors thank PerkinElmer for providing assay kits used in this study free
of charge. We wish to acknowledge staff of the Maternal Multiple Marker References
Screening Laboratory at North York General Hospital for preparing study 1. Committee Opinion No. 640: Cell-Free DNA Screening For Fetal Ane‑
samples and performing laboratory assays. uploidy. Obstet Gynecol. 2015;126(3):e31–7.
2. Sánchez-Durán MÁ, Bernabeu García A, Calero I, Ramis Fossas J, Illescac
Authors’ contributions T, Aviles MT, et al. Clinical application of a contingent screening strategy
TH contributed to conception and design, performed data analysis, inter‑ for trisomies with cell-free DNA: A pilot study. BMC Pregnancy Childbirth.
preted the data, drafted the manuscript and agreed to act as a guarantor of 2019;19:274.
the work presented in this article. SR performed laboratory assay and revised 3. Litwińska E, Litwińska M, Oszukowski P, Szaflik K, Litwinski W, Korcz M,
the manuscript critically. HC performed data analysis, interpreted the data, et al. Biochemical markers in screening for preeclampsia and intrauterine
and revised the manuscript critically. ER performed data analysis and drafted growth restriction. Ginekol Pol. 2015;86(8):611–5.
the manuscript. EM, HMB, MP, EMT, CG and WSM and contributed substantially 4. Cohen JL, Smilen KE, Bianco AT, Moshier EL, Ferrara LA, Stone JL.
to the interpretation of data, and revised the manuscript critically. All authors Predictive value of combined serum biomarkers for adverse pregnancy
have given the final approval of version to be published. outcomes. Eur J Obstet Gynecol Reprod Biol. 2014;181:89–94.
5. Akolekar R, Syngelaki A, Poon L, Wright D, Nicolaides KH. Competing risks
Funding model in early screening for preeclampsia by biophysical and biochemi‑
The study was supported by Imperial Order Daughters of the Empire (IODE) cal markers. Erratum appears in Fetal Diagn Ther. 2013;34(1):43.
Genetics Program Research Fund of North York General Hospital. 6. Park F, Russo K, Williams P, Pelosi M, Puddephatt R, Walter M, et al.
Prediction and prevention of early-onset pre-eclampsia: Impact of
Availability of data and materials aspirin after first-trimester screening. Ultrasound Obstet Gynecol.
All data generated or analysed during this study are included in this published 2015;46:419–23.
article. 7. Tsiakkas A, Duvdevani N, Wright A, Wright D, Nicolaides KH. Serum
placental growth factor in the three trimesters of pregnancy: effects of
maternal characteristics and medical history. Ultrasound Obstet Gynecol.
Declarations 2015;45:591–8.
8. Bredaki FE, Mataliotakis M, Wright A, Wright D, Nicolaides KH. Maternal
Ethics approval and consent to participate serum alpha-fetoprotein at 12, 22 and 32 weeks’ gestation in screening
The Research Ethics Board of North York General Hospital has approved the for pre-eclampsia. Ultrasound Obstet Gynecol. 2016;50:807.
study on May 17, 2017. The Informed consent was waived by NYGH’s REB. The 9. Wright A, Guerra L, Pellegrino M, Wright D, Nicolaides KH. Maternal serum
informed consent was waived because our study meets the requirements PAPP-A and free β-hCG at 12, 22 and 32 weeks’ gestation in screening for
specified in Canada’s Tri-council policy statements C, Article 12.3 (https://​eth‑ pre-eclampsia. Ultrasound Obstet Gynecol. 2016;47:762–7.
ics.​gc.​ca/​eng/​tcps2-​eptc2_​chapt​er12-​chapi​tre12.​html) for following reasons: 10. Poon LC, Shennan A, Hyett JA, Kapur A, Hadar E, Divakar H, et al. The
1) The study is retrospective in nature. It is impracticable to seek consent from International Federation of Gynecology and Obstetrics (FIGO) initiative
each individual patient. At the time of the sample collection, we did not know on pre-eclampsia: A pragmatic guide for first-trimester screening and
if a pregnancy will have an adverse pregnancy outcome (therefore will be prevention. Int J Gynaecol Obstet. 2019;145(Suppl 1):1–33 (Erratum in: Int
include in the study). As screening tests were ordered by thousands of physi‑ J Gynaecol Obstet. 2019;146:390-391).
cians across the province of Ontario, Canada, we do not have direct contact 11. Poon LC, Nicolaides KH. First-trimester maternal factors and biomarker
with patients. screening for preeclampsia. Prenat Diagn. 2014;34:618–27.
2) This study cannot be carried out without using residual samples. 12. O’Gorman N, Wright D, Syngelaki A, Akolekar R, Wright A, Poon LC, et al.
3) Using residual serum samples without patient consent will not have Competing risks model in screening for preeclampsia by maternal fac‑
an adverse effect on patients because all subjects in the study have been tors and biomarkers at 11–13 weeks gestation. Am J Obstet Gynecol.
managed using current standard of care guidelines for detection, follow-up, 2016;214:103.e1-103.e12.
confirmation and treatment. 13. Keikkala E, Koskinen S, Vuorela P, Laivuori H, Romppanen J, Heinonen S,
4). We have de-identified the study dataset and study samples to protect the et al. First trimester serum placental growth factor and hyperglycosylated
privacy of individuals and safeguard residual blood samples. No patients will human chorionic gonadotropin are associated with pre-eclampsia: a case
be contacted. Only aggregate data will be published. control study. BMC Pregnancy Childbirth. 2016;16:378.
All methods were performed in accordance with the relevant guidelines and 14. Cuckle H, Sehmi I, Jones R. Maternal serum inhibin A can predict pre-
regulations (Declaration of Helsinki). eclampsia. BJOG An Int J Obstet Gynaecol. 1998;105:1101–3.
Huang et al. BMC Pregnancy and Childbirth (2022) 22:190 Page 14 of 14

15. Spencer K, Cowans NJ, Nicolaides KH. Maternal serum inhibin-A and
activin-A levels in the first trimester of pregnancies developing pre-
eclampsia. Ultrasound Obstet Gynecol. 2008;32:622–6.
16. Wright D, Silva M, Papadopoulos S, Wright A, Nicolaides KH. Serum preg‑
nancy-associated plasma protein-A in the three trimesters of pregnancy:
Effects of maternal characteristics and medical history. Ultrasound Obstet
Gynecol. 2015;46:42–50.
17. Barjaktarovic M, Korevaar TIM, Jaddoe VWV, de Rijke YB, Peeters RP,
Steegers EAP. Human chorionic gonadotropin and risk of pre-eclampsia:
prospective population-based cohort study. Ultrasound Obstet Gynecol.
2019;54:477–83.
18. Broumand F, Salari Lak S, Nemati F, Mazidi A. A study of the diagnostic
value of Inhibin A Tests for occurrence of preeclampsia in pregnant
women. Electron Physician. 2018;10:6186–92.
19. Lakhi N, Govind A, Moretti M, Jones J. Maternal serum analytes as markers
of adverse obstetric outcome. Obstet Gynaecol. 2012;14(4):267–73.
20. Jelliffe-Pawlowski LL, Baer RJ, Blumenfeld YJ, et al. Maternal characteristics
and mid-pregnancy serum biomarkers as risk factors for subtypes of
preterm birth. BJOG An Int J Obstet Gynaecol. 2015;122:1484–93.
21. Rolnik DL, Wright D, Poon LCY, et al. ASPRE trial: performance of screening
for preterm pre-clampsia. Ultrasound Obstet Gynecol. 2017;50:492–5.
22. Caillon H, Tardif C, Dumontet E, Winer N, Masson D. Evaluation of sFlt-1/
PlGF ratio for predicting and improving clinical management of pre-
eclampsia: Experience in a specialized perinatal care center. Ann Lab
Med. 2018;38:95–101.
23. Myatt L, Clifton RG, Roberts JM, Spong CY, Wapner RJ, Thorp JM Jr, et al.
Can changes in angiogenic biomarkers between the first and second
trimesters of pregnancy predict development of pre-eclampsia in a
low-risk nulliparous patient population? BJOG An Int J Obstet Gynaecol.
2013;120:1183–91.
24. Vatten LJ, Eskild A, Nilsen TIL, Jeansson S, Jenum PA, Staff AC. Changes in
circulating level of angiogenic factors from the first to second trimester
as predictors of preeclampsia. Am J Obstet Gynecol. 2007;196(239):e1-6.
25. Erez O, Romero R, Espinoza J, Fu W, Todem D, Kusanovic JP, et al. The
change in concentrations of angiogenic and anti-angiogenic factors in
maternal plasma between the first and second trimesters in risk assess‑
ment for the subsequent development of preeclampsia and small-for-
gestational age. J Matern Fetal Neonatal Med. 2008;21:279–87.
26. Xiao J, Shen F, Xue Q, et al. Is ethnicity a risk factor for developing preec‑
lampsia? An analysis of the prevalence of preeclampsia in China. J Hum
Hypertens. 2014;28:694–8.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub‑
lished maps and institutional affiliations.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission


• thorough peer review by experienced researchers in your field
• rapid publication on acceptance
• support for research data, including large and complex data types
• gold Open Access which fosters wider collaboration and increased citations
• maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions

You might also like