Professional Documents
Culture Documents
Journal Pre-Proof: American Journal of Obstetrics and Gynecology
Journal Pre-Proof: American Journal of Obstetrics and Gynecology
Placental growth factor predicts time to delivery in women with signs or symptoms of
early preterm preeclampsia: a prospective multicenter study.
John R. Barton, M.D., M.S., Doug A. Woelkers, M.D., Roger B. Newman, M.D., C.
Andrew Combs, M.D., PhD., Helen Y. How, M.D., Kim A. Boggess, M.D., James
N. Martin, Jr., M.D., Kenneth Kupfer, PhD., Baha M. Sibai, M.D., For the PETRA
(Preeclampsia Triage by Rapid Assay) Trial
PII: S0002-9378(19)31108-1
DOI: https://doi.org/10.1016/j.ajog.2019.09.003
Reference: YMOB 12874
Please cite this article as: Barton JR, Woelkers DA, Newman RB, Combs CA, How HY, Boggess KA,
Martin Jr. JN, Kupfer K, Sibai BM, For the PETRA (Preeclampsia Triage by Rapid Assay) Trial, Placental
growth factor predicts time to delivery in women with signs or symptoms of early preterm preeclampsia:
a prospective multicenter study., American Journal of Obstetrics and Gynecology (2019), doi: https://
doi.org/10.1016/j.ajog.2019.09.003.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
Presented at the 36th Annual Clinical Meeting of the Society for Maternal–Fetal
Medicine, Atlanta, GA, February 1-6, 2016.
2
preeclampsia.
angiogenic factors and the time to delivery in women presenting with suspected
B. As compared to women with normal placental growth factor, women with a low
placental growth factor level (< 100 pg/mL) have a hazard ratio of 7.17 (5.08,
10.13) in Cox regression for time-to-delivery after adjusting for both gestational
growth factor level was strongly correlated with preterm delivery independent of a
hypertension in pregnancy
3
Abstract
between the measured angiogenic factors and the time to delivery in women
remains to be clarified.
centers. Blood was collected at presentation for PlGF, and subjects were
measured retrospectively on the Alere, Inc. Triage platform. A normal PlGF was
defined as >100 pg/ml and the assay’s limit of detection is 12 pg/ml. 2x2 tables
regression.
4
Results. 753 subjects were enrolled; 538 (71%) had a final diagnosis of
preeclampsia. 542 (72%) delivered <37 weeks and 358 (47%) <34 weeks.
Among the 279 (37%) women with a normal PlGF at presentation, the NPV for
preeclampsia delivered within 14 days, or within 7 days was 90% and 93%,
respectively. As compared to women with normal PlGF, women with PlGF < 100
pg/mL have a hazard ratio of 7.17 (5.08, 10.13) in Cox regression for time-to-
delivery after adjusting for both gestational age at enrollment and the final
diagnosis of preeclampsia. The PlGF levels of normal (>100 pg/mL), low (12 to
100 pg/mL), and very low (<12 pg/mL), have well separated distributions of time-
to-delivery with median values of 45, 10, and 2 days, respectively. Subjects with
PlGF < 100 pg/mL have a perinatal death rate of 5.7% and an SGA rate of
51.7%, while subjects with PlGF > 100 pg/mL have a perinatal death rate of 0%
final diagnosis of preeclampsia. This suggests that PlGF levels are superior to
suspected preeclampsia.
5
Introduction
tragic, but relatively uncommon. In fact, the majority of perinatal morbidity and
traditional diagnosis and staging of preeclampsia are difficult and despite recent
preeclampsia.3
potential of these factors has been reported.3 Indeed, Levine et al5 demonstrated
lower mean PlGF concentrations in subjects who later develop preeclampsia and
Using separate assays, two recent studies from European cohorts4,6 with
rates of confirmed disease within 1 and 2 weeks of screening of 18% and 55%
respectively, reported high sensitivity and high negative predictive values using
association between the measured angiogenic factors and the time to delivery in
objectives of this study are to examine the relationship between PlGF and time-
blocks and venous blood samples for this study were collected at enrollment
(estimated fetal weight less than the 10th percentile for gestational age), or
panel with the final diagnosis of preeclampsia according to the 2013 ACOG Task
were labeled and transported to the laboratory where they were immediately
centrifuged at 1300 g for 10min at 2-8 C and the supernatants were kept frozen
at -80 C until assayed. Plasma was analyzed for PlGF retrospectively in batch
using the Triage PlGF Test in a central laboratory facility (Alere, San Diego,
PlGF for PlGF quantification and contains chemistries for on-board positive and
negative controls systems to ensure that the quantitative PlGF result is valid.
Briefly, 230 ul of thawed plasma (room temperature) is pipetted into the sample
port of a new test cartridge. The cartridge is inserted into the meter, and the
cutoff value of 12 pg/ml represented the lower detection limit of the assay.4,9
Chappell et al.4 also showed that PlGF did not perform well for subjects enrolled
after 35 weeks of gestation, even when using the gestational age dependent
single cutoff (100 pg/mL) in the cohort of women presenting prior to 35 weeks of
gestation.
dichotomous PlGF together with gestational age and established clinical markers
standard of care clinical biomarkers was via the final adjudicated diagnosis of
PE. The final diagnosis represents a composite that includes the entire sequence
information collected during enrollment and throughout the clinical course of the
subject. Adjusting for final diagnosis demonstrates that a single PlGF value
The minimum total sample size was calculated based on the width of the
95% confidence interval of the sensitivity and the specificity with goal of
subjects. This minimum total sample size was exceeded to achieve a minimum of
weeks gestation, noting that the prevalence of preeclampsia was not controlled
and 215 non-preeclamptic subjects. All laboratory staff were masked to clinical
managed and delivery timing and mode were determined per local protocols.
Kruskal Wallis test. Cox Proportional Hazards assessment was performed with
MATLAB coxphfit for the prediction of TTD censored at 14 days using univariate
models incorporating either PlGF alone (positive < 100 mg/mL vs negative > 100
mortality was defined as stillbirth or neonatal death. Small for gestational age
(SGA) was defined as birth weight less than the 10th percentile according to
Alexander et al.10
Results
1,258 subjects were initially enrolled over the study period from November
2010 to January 2012. 24 subjects were ineligible or did not follow the protocol,
10
and 11 were withdrawn or lost to follow up, leaving 1,223 subjects. Six enrollment
samples could not be evaluated on the Triage device, leaving 1,217 subjects in
the entire cohort. 753 women were enrolled at < 35 weeks, and these constitute
the subjects of the current analysis and are presented in the concert flowchart.
(Figure 1).
preexisting hypertension.
Most (58.2%) were obese; 680 pregnancies (90.3%) were singleton gestations.
BMI was included as a covariate in the Cox proportional hazards regression. The
hazard ratio for BMI is weak and the adjusted hazard ratio for PlGF is the same
regardless of the BMI. Therefore, the results are independent of BMI. The study
125 subjects (16.6%) did not ultimately meet diagnostic criteria for any
heterogeneity, however in sub-analysis the odds ratio for PlGF predicting preterm
delivery in this subset of 125 subjects is 33.5 (95%CI 9.35 to 120), consistent
with the odds ratio of 26.0 (16.8, 40.3) reported for the overall cohort (Table 6).
The median gestational age at delivery was 34.1 weeks, with 72% delivering < 37
and 47% delivering < 34 weeks. The median gestational age at delivery varied by
the final diagnosis and was significantly earlier for the 538 subjects with a final
diagnosis of preeclampsia (33.1 weeks) versus the other 215 subjects (37.6
week categories of < 24+0, 24+0 to 27+6, 28+0 to 31+6, and 32+0 to 34+6 was
and those with superimposed preeclampsia were more likely to deliver within 2
days, 7 days, and within 14 days as compared to the other groups (two-tailed
Fisher exact p-value < 0.0001 for the contingency table comparing two groups of
versus all other, by two delivery intervals, 0-14 days, versus >14 days)..
(12-100), or very low (<12 pg/ml). Within the group of 280 subjects that had a
very low PLGF level (< 12 pg/mL), delivery within two days was 5.6 fold more
likely than delivery after 14 days following enrollment (174 subjects versus 31
the final diagnosis of preeclampsia. The PlGF levels of normal, low, and very low
have well separated distributions of TTD with median values of 45, 10, and 2
days, respectively.
calculating the hazard ratio in Cox regression. The test positive group (PlGF <
100 pg/mL) as compared to the test negative group (PlGF > 100) has a hazard
regression. The hazard ratio is 9.21 (6.58, 12.91) when adjusted for the
gestational age at enrollment. The hazard ratio is 7.55 (5.35, 10.65) when
adjusted for both the gestational age at enrollment and the presence of
proteinuria at enrollment. The hazard ratio is 7.17 (5.08, 10.13) when adjusted for
both gestational age at enrollment and the final diagnosis of preeclampsia. The
regression, where it has a hazard ratio of 2.43 (1.75, 3.36). These hazard ratios
importance; perinatal mortality and SGA (birth weight <10th %tile) infant.
delivery within or after 2 weeks from enrollment, a PlGF value of <100 pg/ml is
associated with the higher risk of subsequent perinatal death (1.3 to 11.4%)
and/or delivery of a SGA infant (48.9 to 63.2%). To simplify, subjects with PlGF <
100 pg/mL have a perinatal death rate of 5.7% (95% confidence interval of 3.8 to
13
8.2%) and an SGA rate of 51.7% (47.1 to 56.3%), while subjects with PlGF > 100
pg/mL have a perinatal death rate of 0% (0 to 1.3%) and an SGA rate of 16.8%
(12.6 to 21.8%).
the other hand, for ruling out deliveries within 7 or 14 days of enrollment, with or
greater and the NPVs are high (approximately 90%). In addition, for prediction of
preterm delivery before 37 weeks, a PlGF < 100 pg/ml shows good specificity
and sensitivity (85% and 82%, respectively) and a high PPV of 93.5%. In the 280
patients with a very low PlGF (<12 pg/ml) as compared to the rest of the cohort,
the prediction of preterm delivery before 37 weeks is very strong, with a PPV of
Comment
Principal Findings
weeks’ gestation, PlGF levels (dichotomized about 100 pg/mL) had a sensitivity
strongly and independently predict the time to delivery within 7 and 14 days
and an SGA infant. Moreover, PlGF levels may be useful to identify women with
confirmed preeclampsia that are not at risk for preterm birth or other adverse
perinatal outcomes.
Results
associated with a wide spectrum of obstetric complications that can include one
findings reveal that extremely low PlGF values can be a marker for the Great
Obstetrical Syndromes.
our study, Chappell et al4 examined the diagnostic accuracy of a low plasma
PlGF concentration (<5th centile for gestation, <100 pg/ml) in 287 women
they reported that a PlGF <5th centile had a high sensitivity (0.96; 95%
15
confidence interval, 0.89–0.99) and a high negative predictive value (0.98; 0.93–
0.995) for developing preeclampsia within 14 days; however, the specificity was
lower (0.55; 0.48–0.61). The authors4 concluded that for women who present
before 35 weeks’ gestation with suspected preeclampsia, low PlGF level has
high sensitivity and high negative predictive value for preeclampsia developing
within 14 days and that low PlGF levels are superior to other currently used
SGA infant with a birthweight less than the 3rd percentile for gestational age. As
in our study, they found that low PlGF values could be a helpful adjunct for
identifying those at high risk for an SGA infant who could benefit from close
amount of its soluble receptor, FMS like tyrosine kinase receptor-1 (sFlt-1), which
is synthesized and secreted by the placenta.5 The ratio of sFlt-1 to PlGF has
36+6 weeks using an sFlt-1:PlGF ratio cutoff of 38, Zeisler, et al6 reported a
negative predictive value of 99.3% for preeclampsia and delivery within one week
preeclampsia in this cohort was only 17.8%, much less than the rate of 71.4% in
16
our cohort, demonstrating that PlGF retains a clinically useful sensitivity and
negative predictive value in a high prevalence and high risk early preterm
population.
Clinical Implications
This study was designed to collect specimens for the validation of the
PlGF assay. The statistical analysis plan for the validation was created following
Chappell et al4 which was the initial study to propose (1) a fixed 100 pg/mL cutoff
for women enrolled prior to 35 weeks gestation and (2) an endpoint of delivery
(or delivery for confirmed preeclampsia) within 14 days. Chappell and colleagues
based on medical decision logic using PlGF at the 100 pg/mL.15 Although the
RCT reports excellent and important results, the medical decision logic (using
PlGF at the 100 pg/mL cutoff) is based on a single observational cohort, the
original Chappell 2013 study. Our study verifies the original observations that
cohort was 71.4%. This finding is likely the consequence of our stratified
enrollment from all gestational ages, rather than consecutively enrolled patients
Chappell et al4 regarding the sensitivity (92.5%) of low PlGF levels predicting
90.3%
17
whereas the reported incidence by Chappell et al4 was 47%. In addition, the
incidence of SGA was 43.4% in those with a final diagnosis of preeclampsia, but
was 17.8% in those without preeclampsia. It is important to note that the rate of
an SGA infant was substantially higher in those with low PlGF level irrespective
of the presence of preeclampsia. Furthermore, a low PlGF level was more likely
to be associated with preterm delivery and perinatal mortality. Indeed, all of the
perinatal losses in this study were associated with a low PlGF level.
cutoffs for effective use of PlGF after 35 weeks are unclear. This limitation was
studied in healthy pregnancy where it was shown that healthy subjects have
rapidly dropping PlGF values as they approach term.9 These values are often
less than 100 pg/mL, making PlGF difficult to interpret for GA ≥ 35 weeks.
Furthermore, a test with applicable predictive value for 14 days would allow a
clinical sites where the study was performed. Statistical control of these variables
that the dominant variables that influence time to delivery were gestational age at
18
enrollment and final diagnosis, which itself is a composite of all the clinical
Research Implications
who never developed the disease. Our results would also suggest that low PlGF
levels might identify those at high risk for adverse pregnancy outcome, including
et al16, and others that randomized trials are needed to determine the potential
women.
that results were obtained with rigorous quality control. All final diagnoses were
Study limitations include test results were not validated in a repeat sample
required on why these data are delayed in their submission for publication.
Although study enrollment was conducted between November 2010 and January
2012, the specimens were banked and the PlGF analysis was not complete until
August 2016. The project was suspended for business reasons and in August
publication.
Conclusions
weeks, a low PlGF is strongly correlated with the need for preterm delivery
preeclampsia, of which, 92% had a low PlGF level with high rates of adverse
delivered within 2 weeks of enrollment had a normal PlGF but these subjects had
no perinatal mortality and a low rate of SGA. This suggests that in women
presenting with suspected preeclampsia, low PlGF levels identify those at high
preeclampsia.
21
References
1. Roberts JM, August PA, Bakris G, Barton JR, Bernstein IM, Druzin M, et al.
2013;122:1122-31.
2. Stevens W, Shih T, Incerti D, Ton TGN, Lee HC, Peneva D,,et al. Short-term
Gynecol 2017;217:237-248.
3.Cerdeira AS, Agrawal S, Staff AC, Redman CW, Vatish M. Angiogenic factors:
1395.
5. Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, et al. Circulating
395.
22
8. Asvold BO, Vatten LJ, Romundstad PR, Jenum PA, Karumanchi SA, Eskild A.
Angiogenic factors in maternal circulation and the risk of severe fetal growth
Health 2013;3:124-132.
10. Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United States
Gynecol 2011;204:193-201.
12. Benton SJ, McCowan LM, Heazell AEP, Grynspan D, Hutcheon JA, Senger
blood test at 24-28 weeks: a role for angiogenic index-1 (PlGF/sVEGFR-1 ratio).
2018;51:387-395.
15. Duhig KE, Myers J, Seed PT, Sparkes J, Lowe J, Hunter RM, et al. Placental
Lancet 2019;393:1807-18.
16. Seely EW, Solomon CG. Improving the prediction of preeclampsia. N Engl J
Med 2006;374:83-4.
24
Lupus 9 (1.2)
*Of the 292 satisfying the definition of SGA, 234 are classified as
features
PE = preeclampsia
HTN = hypertension
28
TTD n (%) Diagnosis n (%) PlGF Category n (%) Death n (%) SGA n (%)
Test + defined as PlGF < 100 pg/ml, Test - defined as PlGF > 100 pg/ml
PlGF levels: normal (>100 pg/ml), low (12-100 pg/ml), very low (<12 pg/ml)
PlGF levels: normal (>100 pg/ml), low (12-100 pg/ml), very low (<12 pg/ml)
31
Appendix 1
Appendix 2
The Alere Triage® PlGF Test (San Diego, CA, USA) is a fluorescence
immunoassay used with the Triage® Meter for the quantitative determination of
PlGF in EDTA-anticoagulated plasma specimens. The test procedure involves
the addition of several drops (230ul) of plasma to the sample port of the single-
use disposable plastic cartridge. The test uses fluorescently-labeled monoclonal
antibodies against PlGF for PlGF quantification and contains chemistries for on-
board positive and negative controls systems to ensure that the quantitative PlGF
result is valid. Calibration information is supplied by the manufacturer in the form
of a lot-specific EPROM chip that is provided with each kit of devices. The
measurable range of the assay is 12–3000 pg/mL. The total precision
(coefficient of variation) on plasma controls for PlGF at concentrations of 85 and
1300 pg/ml is 12.8% and 13.2%, respectively. The concentration of PlGF is
displayed on the meter screen approximately 15-20 minutes from the addition of
specimen.
Study Enrollment
1258
Enrolled
24 ineligible or
off protocol
1234
11 withdrawal or
lost to f/u
1223
6 unevaluable PlGF
1217
464 delivered >350 wks
753*
<350 wks