Utility of placental growth factor for preeclampsia prediction in pregnancies complicated by sickle cell disease

Key Points

• •Low PlGF in pregnancies of individuals without SCD can predict preeclampsia; upregulation of PlGF in SCD, rendered its role in preeclampsia prediction unclear.
• •PlGF of <100 pg/mL at 20 to 24 weeks gestation has 100% sensitivity and specificity for prediction of early-onset preeclampsia in context of SCD.

Visual Abstract

Abstract

Outside of pregnancy, placental growth factor (PlGF), is produced by erythroid cells in typically undetectable levels. In pregnancy, PlGF is strongly expressed by the trophoblast layer covering the placental villi. PlGF levels rise progressively due to placental growth, peak at 28 to 30 weeks gestation, and then slowly decline toward term. Low PlGF has emerged as a powerful diagnostic test for preterm preeclampsia. However, its interpretation in context of sickle cell disease (SCD) is potentially confounded by upregulation of cellular PlGF expression in nonpregnant individuals with SCD, and higher third trimester circulating PlGF levels documented in healthy Black compared with White individuals. Primary objectives were to determine the distribution of PlGF at midtrimester in pregnant individuals with SCD compared with unaffected Black controls and to explore the diagnostic accuracy of PlGF in the context of suspected preeclampsia in pregnancies of individuals with SCD. Secondary objective was to examine the relationship between low PlGF and placental disease in pregnancies of individuals with SCD. Pregnant individuals with SCD at Mount Sinai Hospital in Canada (January 2017 to September 2021) with at least 1 PlGF measurement 20+0 to 35+6 weeks gestation, and pregnant Black controls without SCD with suspected preeclampsia or growth restriction, were included in this retrospective study. Maternal and neonatal outcomes were extracted from medical records. For early-onset, but not late-onset, preeclampsia, a PlGF cutoff of <100 pg/mL demonstrated 100% sensitivity and specificity at 20 to 24 weeks gestation. This study is, to our knowledge, the first to demonstrate the utility of PlGF in predicting early-onset preeclampsia in pregnancies of individuals with SCD, allowing clinicians to anticipate and mitigate adverse pregnancy outcomes.

Introduction

Sickle cell disease (SCD) is a prevalent hemoglobinopathy characterized by distortion of red cells into rigid, sickle shapes under hypoxic conditions. This process leads to arteriolar microvascular occlusion, ischemia-reperfusion tissue injury, hemolysis, and endothelial dysfunction; ultimately this vascular pathology can result in widespread end-organ damage and place pregnant individuals at increased risk of mortality.^1,2

Placental growth factor (PlGF), is a proangiogenic factor produced by erythroid cells, and is typically expressed at near undetectable levels in healthy nonpregnant tissues. However, its production is upregulated in proinflammatory disease states, such as SCD, which is characterized by increased red cell, platelet, and leukocyte activation, as well as complement activation and endothelial dysfunction,^3,4 in which PlGF plays a counter-regulatory role to balance proinflammatory signals.⁵

In pregnancy, a unique tissue, namely the trophoblast compartment covering the floating placental villi, is a potent source of circulating PlGF, as the villi are in direct contact with maternal blood for nutrient and gaseous exchange.⁶ In healthy White individuals, PlGF levels rise progressively and peak at 28 to 30 weeks gestation.⁷ Racial differences may influence circulating PlGF levels, with Black individuals having higher circulating PlGF levels than other racial groups in the third trimester.⁸

Preeclampsia is a hypertensive clinical syndrome whose origins are typically traced back to abnormal early placental development.⁹ It is often classified as early-onset (developing before 34 weeks gestation) and late-onset (developing at, or after, 34 weeks gestation) disease.¹⁰ The most common underlying type of placental dysfunction begins with impaired transformation of the uteroplacental arterioles feeding the placenta,¹¹ a pathologic process, that is, more common in persons of African-American descent and further amplified by the chronic sickling process. The net effect of chronic placental ischemia is to prevent the physiologic rise in circulating PlGF during the second trimester).¹² Because the principal actions of PlGF include systemic endothelium-dependent vasodilation, normal circulating PlGF levels provide strong protection against the development of early-onset preeclampsia because this diseases is characterized by intense vasoconstriction.¹³ The impaired production and release of PlGF subsequently mediates a pathologic excess production of soluble Fms-like tyrosine kinase 1 (sFlt-1),¹⁴ typically after 24 weeks of gestation, which acts in concert with low PlGF to mediate severe early-onset preeclampsia/fetal growth restriction (FGR) and ultimately the need for iatrogenic preterm birth.¹⁵

Using the traditional definition, which includes blood pressure (BP) of ≥140/90, the incidence of preeclampsia in individuals with SCD is increased twofold to threefold over those without SCD.¹⁶ However, this convention may underestimate the incidence of preeclampsia in this population, given that our previous work has shown that pregnant people with SCD (particularly those with hemoglobin SS [HbSS]/HbSβ genotype) tend to have lower BP than pregnant people without SCD,¹⁷ and the current BP thresholds used to diagnose hypertensive disorders of pregnancy (HDP) lack sensitivity in the SCD population (HbSS/HbSβ0 genotype [systolic BP, 21% sensitive; diastolic BP, 5.3% sensitive]; HbSS/HbSβ⁺ genotype [systolic BP, 10% sensitive; diastolic BP, 0% sensitive]).¹⁸

A growing body of literature, summarized recently by Gladstone et al,⁶ demonstrates that, in pregnancies unaffected by SCD, there are clear links between low PlGF levels and adverse pregnancy outcomes, as well as an association between low PlGF and underlying placental maternal vascular malperfusion (MVM) disease, the most common pattern of placental injury associated with preeclampsia and FGR.¹¹ Furthermore, MVM has been shown to be the predominant pathology in SCD-affected pregnancies that experienced an adverse outcome. To date, the capacity of low PlGF to predict pregnancy complications in SCD is largely unknown, and due to the higher PlGF levels observed in the third trimester of healthy Black individuals, for whom currently used reference ranges, established mostly in White individuals, may render a high rate of false-negative diagnostic test results.

Therefore, the primary objectives of this study were to determine the distribution of PlGF levels at midtrimester in pregnancies of individuals with SCD (who in this cohort were Black) in comparison with Black controls, and to explore the role of PlGF in preeclampsia prediction in individuals with SCD. The secondary objectives were to evaluate the applicability of PlGF thresholds established for preeclampsia prediction in the general population for use in individuals with SCD, and to examine the relationship between low PlGF levels and underlying placental disease in individuals with SCD.

Materials and methods

Study design

This retrospective, single-center cohort study was conducted at Mount Sinai Hospital, Toronto, Canada, which is the largest provider of referral maternal-fetal medicine services in Canada, with 7200 births per year. The study evaluated PlGF levels in pregnant individuals with SCD compared with a Black control population without SCD, with testing completed for suspicion of placental insufficiency (preeclampsia or FGR).¹⁹ The study cohort included singleton pregnancies in individuals affected by SCD, from 1 January 2017 to 30 September 2021, with at least 1 PlGF measurement between 20+0 and 35+6 weeks gestation. Since 2017, PlGF measurements were established as standard clinical practice in individuals with suspected preeclampsia or other clinical manifestations of placental dysfunction. Therefore, PlGF tests were also progressively incorporated into the care pathway for pregnant individuals with SCD. Hospital central laboratory PlGF testing is available 24 hours, 7 days a week, on a daily basis, using fresh plasma, as previously reported.⁷ Control PlGF samples were made available from hospital records, having been collected between 20+0 and 35+6 weeks gestation, and obtained from March 2017 to December 2019.¹⁹

Data collected from medical records included maternal demographics (age, ethnicity, body mass index [BMI], and medical history), laboratory tests (PlGF levels, Hb, hematocrit, and other relevant biomarkers), placental ultrasound findings (measurements and assessments of placental function and structure), delivery details (gestational age at delivery, mode of delivery), maternal outcomes (incidence of HDP, subclassified into early-onset [<34 weeks gestation] and late-onset [≥34 weeks gestation] preeclampsia as per accepted definitions),²⁰ and neonatal outcomes (birth weight; small for gestational age [SGA] size, defined as <10th percentile for gestational age²¹; preterm birth, defined as <37 weeks of gestation; and stillbirth).

Thie study was approved by the Mount Sinai Hospital research ethics board (number 23-0041-C).

Statistical analysis

Categorical data are presented as percentages, whereas continuous data are expressed as means with standard deviations or medians with interquartile ranges [IQRs], depending on the normality of their distribution. Categorical variables were analyzed using the χ² test or the Fisher exact test/likelihood ratio as appropriate, whereas continuous variables were examined using Student t tests or Mann-Whitney U tests, depending on the distribution of the data. The Kruskal-Wallis test was used to compare a continuous variable across multiple groups, if the data were not normally distributed. Multivariable linear regression was used for continuous outcomes, whereas binary logistic regression was used for categorical outcomes to account for potential confounders. Significance was defined as P value <.05.

The analysis aimed to balance high sensitivity with optimal specificity, focusing on prediction of HDP in individuals with SCD. Receiver operating characteristic (ROC) curves were constructed to explore optimal PlGF cutoffs, balancing sensitivity and specificity. The positive (PPV) and negative predictive values (NPV) of the selected PlGF cutoffs were examined. Given its predictive performance for HDP outside of SCD,²² a cutoff value of <100 pg/mL was explored for its applicability to individuals with SCD.²³

Results

Pregnancy characteristics in individuals with SCD and Black controls

Table 1 depicts pregnancy characteristics of individuals with SCD and control pregnancies, stratified by the presence of preeclampsia. No cases of gestational hypertension were observed within the cohort. Most patients with SCD had HbSS/HbSβ0-thalassemia genotype (56/82 [67.5%]). Of those with preeclampsia, 8 of 11 (7.3%) had HbSS/HbSβ0-thalassemia genotype and 3 of 11 (2.7%) HbSC/HbSβ⁺-thalassemia genotype.

Table 1.

Pregnancy and PlGF characteristics of pregnant individuals with SCD and controls of Black ancestry, stratified by preeclampsia status

Pregnancies of individuals with SCD						Control pregnancies of Black ancestry
	Overall SCD cohort (n = 83)	Preeclampsia (n = 11)		No preeclampsia (n = 72)	P value∗	Overall control cohort (149)	Preeclampsia (n = 19)		No preeclampsia (n = 130)	P value∗
		Early onset (n = 4)	Late onset (n = 7)				Early onset (n = 3)	Late onset (n = 16)
SCD genotype, n (%)
HbSS	54 (65.0)	3 (75.0)	5 (71.4)	46 (63.9)	.57	N/A	N/A	N/A	N/A	N/A
HbSC	24 (28.9)	0 (0.0)	2 (28.6)	22 (30.6)	.40	N/A	N/A	N/A	N/A	N/A
HbSβ0-thalassemia	1 (1.2)	0 (0.0)	0 (0.0)	1 (1.4)	1.00†	N/A	N/A	N/A	N/A	N/A
HbSβ+-thalassemia	4 (4.8)	1 (25.0)	0 (0.0)	3 (4.2)	.44†	N/A	N/A	N/A	N/A	N/A
Maternal age, median (IQR), y	31.0 (27.0-35.0)	34.5 (31.8-38.0)	35 (30.0-40.0)	31.0 (26.3-34.0)	.032	33.0 (29.8-36.3)	34.0 (30.0-35.0)	34.5 (30.8-38.0)	33.0 (29.0-36.0)	.13
BMI,‡ median (IQR), kg/m2	23.7 (20.2-27.0)	23.2 (18.6-26.2)	25.9 (24.8-27.6)	23.3 (20.1-26.9)	.37	31.2 (28.0-37.3)	42.6	36.5 (31.9-47.4)	31.0 (27.5-35.9)	.006
Nulliparity, n (%)	36 (43.4)	2 (50.0)	5 (71.4)	29 (40.3)	.196	53 (35.6)	0 (0.0)	6 (37.5)	47 (36.2)	1.0
Antenatal low-dose ASA, n (%)	80 (96.4)	4 (100.0)	7 (100.0)	69 (95.8)	1.00†	-	-	-	-	-
Hb in T1, median (IQR), g/L	89.0 (78.0-101.0)	84.0 (78.5-95.5)	86.0 (68.0-93.0)	91.0 (78.6-101.0)	.18	-	-	-	-	-
PAPP-A MoM, median (IQR)	0.82 (0.51-1.3)	0.71 (0.45-1.28)	0.71 (0.44-2.10)	0.82 (0.55-1.3)	.40	0.92 (0.58-1.35)	0.83 (0.3-2.41)	1.35 (0.54-1.80)	0.90 (0.58-1.23)	.36
PlGF, median (IQR), pg/mL
Lowest value	357 (186-522)	51.0 (36.8-309.8)	146.0 (99.0-413.0)	386.5 (214.8-544.5)	.02	555 (340.3-883.0)	95.0 (29.3-142.0)	317.5 (126.5-627.3)	416.5 (246.3-591.8)	<.001
GA at lowest value	23.7 (22.7-28.6)	26.9 (24.9-30.1)	26.6 (26.1-35.9)	23.4 (22.5-28.3)	.72	26.2 (24.5-28.4)	30.0 (24.6-32.6)	24.9 (21.4-31.8)	26.3 (22.6-29.4)	.944

Values set in bold represent a statistically significant result at a P value < 0.05.

GA, gestational age; MoM, multiples of median; N/A, not applicable; PAPP-A, pregnancy-associated plasma protein A; T1, first trimester.

^∗P values for comparison of the SCD group with and without preeclampsia; for continuous variables, normal distribution was assessed using the Shapiro-Wilk test. If normally distributed, comparisons were made using the independent samples t test; if not, the Mann-Whitney U test was used. Categorical data were analyzed using the χ² test unless otherwise marked.

^†Categorical data were analyzed using Fisher exact test.

^‡BMI at time of first or “booking” visit for obstetrical care.

Maternal age at delivery was significantly older in the SCD cohort with preeclampsia (median age, 34.5 years) compared with those without (median age, 31.0 years; P = .032), whereas there was no difference between these groups in the control cohort. In the SCD cohort, early pregnancy BMI was low and did not differ significantly between those with and without preeclampsia. Conversely, in the control cohort, individuals with preeclampsia had a significantly higher BMI (median, 36.5 kg/m²) than those without preeclampsia (median, 31.0 kg/m²; P = .006).

The use of antenatal low-dose aspirin for primary prevention of preeclampsia was nearly universal in the SCD cohort. Data on aspirin use were not available in the control cohort. The incidence of preeclampsia in the SCD cohort was not influenced by first trimester Hb level. First trimester pregnancy-associated plasma protein A level was not discriminatory for preeclampsia in either the SCD or the control cohort (Table 1).

PlGF levels in pregnancies of individuals with SCD and Black controls according to gestational age and preeclampsia status

A comparison of PlGF levels in pregnancies of individuals with SCD and Black controls is presented in Table 2. At 20 to 24 weeks, pregnancies of individuals with SCD and early-onset preeclampsia had significantly lower median PlGF levels (78 pg/mL [IQR, 69-87]) than pregnancies with late-onset preeclampsia (158 pg/mL [IQR, 114-617]; P = .02), or no preeclampsia (435 pg/mL [IQR, 285-561]; P = .018). This pattern paralleled that of the controls, for whom median PlGF was lower in early-onset preeclampsia (55 pg/mL [IQR, 29-87]) than late-onset preeclampsia (448 pg/mL; IQR, 357-676; P < .001) and no preeclampsia (322 pg/mL [IQR, 259-417]; P < .001). The findings were consistent across gestational ages until 32 weeks. Beyond 32 weeks gestation, the median PlGF level in pregnancies of individuals with late-onset preeclampsia was significantly lower than in individuals without preeclampsia, but there was considerable overlap of values between the late-onset preeclampsia and no preeclampsia groups. This held true for individuals with SCD (late-onset preeclampsia, 138 pg/mL [IQR, 37-223] vs no preeclampsia, 147 pg/mL [IQR, 95-517]; P = .011), and for controls (late-onset preeclampsia, 919 pg/mL [IQR, 244-1344] vs no preeclampsia, 738 pg/mL [IQR, 526-1121]; P = .003).

Table 2.

Comparison of PlGF levels in pregnancies of individuals with SCD and controls of Black ancestry according to preeclampsia status

Gestational age, wk	Preeclampsia status	Pregnancies of individuals with SCD			Controls of Black ancestry			P value
		PlGF, median (IQR), pg/mL	P value		PlGF, median (IQR), pg/mL	P value
20-24	None	435 (285-561)		.018	322 (259-417)		<.001	.20
	Early onset	78 (69-87)	.002		55 (29-87)	<.001		.66
	Late onset	158 (114-617)			448 (357-676)			.43
24-28	None	563 (358-799)		.02	614 (414-869)		0.006	.42
	Early onset	60 (53-313)	.032		87 (47-90)	.011		.88
	Late onset	217 (51-603)			360 (218-994)			.41
28-32	None	594 (208-964)		.015	841 (588-122)		.005	.14
	Early onset	60 (51-69)	.043		142 (88-159)	.034		.07
	Late onset	233 (60-405)			251 (102-989)			.33
>32	None	147 (95-517)	.011		738 (526-1121)	.003		.07
	Late onset	138 (37-223)			919 (244-1344)			.28

Predictive capacity of PlGF for preeclampsia in pregnancies of individuals with SCD

Prediction of early-onset preeclampsia

PlGF levels demonstrated robust predictive capacity for early-onset preeclampsia across gestational age subgroups, with diagnostic test accuracy varying by gestational age (Table 3). At 20 to 24 weeks gestation, a PlGF level of 87 pg/mL achieved 100% sensitivity and specificity, with a ROC area under the curve (AUC) of 1.000 (P < .001), indicating perfect discriminatory ability. Using a cutoff of <100 pg/mL at this gestational age also maintained 100% sensitivity, specificity, PPV, and NPV (Figure 1).

Table 3.

Predictive capacity of PlGF for preeclampsia in pregnancies of individuals with SCD

Test characteristics	20-24 wk			24-28 wk			28-32 wk			>32 wk
	Early-onset preeclampsia	Late-onset preeclampsia	All preeclampsia	Early-onset preeclampsia	Late-onset preeclampsia	All preeclampsia	Early-onset preeclampsia	Late-onset preeclampsia	All preeclampsia	Late-onset preeclampsia	All preeclampsia
PlGF (pg/mL)	87	832	834	396	692	734	69	413	418	67	299
ROC AUC	1.000	0.715	0.796	0.925	0.763	0.844	0.929	0.821	0.875	0.962	0.700
ROC AUC P value	<.001	.193	.02	<.001	.077	<.001	<.001	.031	<.01	<.001	.133
Specificity (%)	100.0	7.4	7.4	70	30.0	30.0	85.7	64.3	64.3	94.3	46.2
False-positive rate (%)	0.0	92.6	92.6	30	70.0	70.0	14.3	35.7	35.7	7.7	53.8
False-negative rate	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Sensitivity (%)	100.0	20.0	42.9	75.0	50.0	62.5	100	50.0	75.0	100	20.0
Specificity (%)	100.0	100.0	100.0	95.0	95.0	95.0	85.7	85.7	85.7	76.9	76.9
False-positive rate (%)	0.00	0.0	0.0	5.0	5.0	5.0	4.3	4.3	4.3	3.1	3.1
False-negative rate	0.0	80.0	57.1	25.0	50.0	37.5	0.0	50.0	25.0	0.0	80.0
PPV	100.0	100.0	100.0	75.0	66.7	83.3	50.0	33.3	60.0	25.0	25.0
NPV	100.0	93.1	100.0	95.0	90.5	86.4	100.0	92.3	92.3	100.0	71.4

Figure 1.

ROC curves of PlGF at 20-24 weeks as predictor of preeclampsia. (A) Early-onset preeclampsia (AUC 1.0), (B) late-onset preeclampsia (AUC 0.715), and (C) all preeclampsia (AUC 0.796).

At 24 to 28 weeks gestation, the PlGF level required for 100% sensitivity rose to 396 pg/mL. A cutoff of <100 pg/mL resulted in a reduced sensitivity of 75%, while retaining 100% specificity. The ROC AUC was 0.925 (P < .00001), reflecting strong discriminatory ability.

At 28 to 32 weeks gestation, a PlGF level of 69 pg/mL achieved 100% sensitivity. The cutoff of <100 pg/mL provided 100% sensitivity, with 85.7% specificity, a PPV of 50%, and an NPV of 100%. The ROC AUC was 0.929 (P = .000), highlighting continued strong discriminatory ability.

Beyond 32 weeks gestation, the required PlGF level for 100% sensitivity was 67 pg/mL. The cutoff of <100 pg/mL maintained 100% sensitivity, with 76.9% specificity, and a NPV of 100% (ROC AUC, 0.962; P < .0001).

Prediction of late-onset preeclampsia

For late-onset preeclampsia prediction at 20 to 24 weeks gestation, a PlGF level of 832 pg/mL achieved 100% sensitivity. Using a cutoff of <100 pg/mL at this gestational age resulted in low sensitivity of 20%, with 100% specificity, a PPV of 100%, and a NPV of 93.1%. The ROC AUC was 0.715 (P = .193), indicating fair discriminatory ability, although it did not reach statistical significance (Figure 1).

At 24 to 28 weeks gestation, a PlGF level of 692 pg/mL was required for 100% sensitivity. A cutoff of <100 pg/mL resulted in 50% sensitivity and 95% specificity, with a PPV of 66.7% and an NPV of 90.5%. The ROC AUC was 0.763 (P = .077), reflecting good discriminatory ability, although, again, not statistically significant.

At 28 to 32 weeks gestation, a PlGF level of 413 pg/mL achieved 100% sensitivity. The cutoff of <100 pg/mL showed low sensitivity (50%) and moderate specificity (85.7%), with a PPV of 33.3% and an NPV of 92.3%. The ROC AUC was 0.821 (P = .031), indicating good discriminatory ability.

Beyond 32 weeks gestation, PlGF lacked discriminatory ability to predict late-onset preeclampsia.

Pregnancy outcomes

Details of pregnancy outcomes and placental pathology are presented in Table 4. Delivery occurred progressively earlier based on the presence and severity of preeclampsia in both the SCD and control cohorts. Among pregnancies in individuals with SCD, median gestational age at delivery was significantly lower in early-onset preeclampsia (31.8 weeks) than in pregnancy with no preeclampsia (38.1 weeks; P = .002). A similar pattern was observed in the control cohort, with median gestational age at delivery of 31.4 weeks in early-onset preeclampsia vs 39.1 weeks without preeclampsia. Cesarean deliveries were more frequent in cases with preeclampsia, particularly for early-onset preeclampsia, in both cohorts.

Table 4.

Pregnancy outcomes and placental pathology in individuals with SCD stratified by hypertensive disorder of pregnancy status

Pregnancies of individuals with SCD						Control pregnancies (Black ancestry)
	Overall SCD cohort (n = 83)	Preeclampsia (n = 11)		No preeclampsia (n = 72)	P value∗	Overall control cohort (149)	Preeclampsia (n = 19)		No Preeclampsia (n = 130)	P value∗
		Early onset (n = 4)	Late onset (n = 7)				Early onset (n = 3)	Late onset (n = 16)
GA at delivery, median (IQR), wk	38.0 (36.6-38.7)	31.8 (29.8-34.1)	36.6 (36.1-37.3)	38.1 (37.1-38.9)	<.001	38.9 (37.9-39.9)	31.4 (27.4-33.7)	37.8 (36.0-38.8)	39.1 (38.1-40.0)	<.001
Mode of delivery, n (%)
Spontaneous vaginal delivery	37 (44.6)	1 (25.0)	3 (42.9)	33 (45.8)	.59	88.0 (59.1)	1 (33.3)	5 (31.3)	82.0 (63.1)	.009
Operative vaginal delivery	1 (1.2)	0 (0.0)	0 (0.0)	1 (1.4)	-	6 (4.0)	0 (0.0)	0 (0.0)	6 (4.6)	1.0
Cesarean delivery (elective)	27 (32.5)	1 (25.0)	2 (28.6)	24 (33.3)	.67	56 (37.6)	2 (66.6)	11 (68.8)	43 (33.1)	.0046
Cesarean delivery (emergency)	18 (21.7)	2 (50.0)	2 (28.6)	14 (19.4)	.22
Neonatal outcomes
Live birth, n (%)	81 (97.6)	4 (100.0)	7 (100.0)	70 (97.2)	1.00†	145 (97.3)	3 (100.0)	15 (97.7)	127 (97.7)	.083
Birth weight, median (IQR), g	2780 (2520-3160)	1605 (1027-2070)	2520 (2190-2780)	2930 (2555-3225)	<.001	3265 (2910-3543)	1440 (610-1520)	3070 (2665-3420)	3310 (2960-3590)	.022
Birth weight (centile)‡, median (IQR)	27.8 (12.0-51.0)	18.25 (0.03-65.0)	7.30 (2.0-17.9)	22.35 (8.6-43.4)	.086	37.0 (14.8-59.0)	1.0 (0.0-5.0)	37.5 (11.8-87.0)	37.0 (15.0-59.0)	.48
SGA§, mean (SD)	28 (33.7)	2 (50.0)	4 (57.1)	22 (30.6)	.12	29 (19.5)	1 (33.3)	4 (25.0)	22 (16.9)	.058
Placental pathology,|| abnormal, n (%)	32/44 (72.7)	3/4 (75.0)	5/5 (100.0)	24/35 (68.6)	.22	10/29 (34.5)	1 (33.3)	1/8 (12.5)	8/18 (44.4)
Type of placental pathology,||n (%)24
MVM	23/44 (43.2)	3 (66.7)	4 (60.0)	16 (58.3)	.26	5/29 (17.2)	2 (66.6)	1/8 (12.5)	2/18 (11.1)	.34
FVM	10/44 (13.6)	1 (0.0)	1 (0.0)	8 (25.0)	.66	4/29 (13.8)	0.0 (0)	0/8 (0.0)	4/18 (22.2)	.27
MVM + FVM¶	4/44 (9.0)	1 (33.3)	1 (20.0)	2 (8.3)	.25†	1/29 (3.4)	0.0 (0)	0/8 (0.0)	1/18 (5.5)	1.00
Other#	3/44 (6.8)	0 (0.0)	1 (20.0)	2 (8.3)	1.00†	0/29 (0.0)	0.0 (0)	0/8 (0.0)	0/18 (0.0)	-

Values set in bold represent a statistically significant result at a P value < 0.05.

FVM, fetal vascular malperfusion; SD, standard deviation.

^∗P values for comparison of the SCD group with and without preeclampsia; for continuous variables, normal distribution was assessed using the Shapiro-Wilk test. If normally distributed, comparisons were made using the independent samples t test; if not, the Mann-Whitney U test was used. Categorical data were analyzed using the χ² test unless otherwise marked.

^†Categorical data were analyzed using Fisher exact test.

^‡Birth weight centile was calculated as described by Kramer et al.²¹

^§Birth weight of <10th centile as described by Kramer et al.²¹

^||Placental pathology examination was completed in 44 pregnancies (4/4 with early-onset preeclampsia, 5/7 with late-onset preeclampsia, and 35/72 without preeclampsia).

^¶Specific characteristics of MVM and FVM reported in this category also reported in their respective aforementioned categories.

^#Other included acute chorioamnionitis, villitis of unknown etiology, and laminated thrombohematoma.

Live-birth rates were high in both SCD (97.6%) and control groups (97.3%). In the SCD cohort, preeclampsia was associated with lower median birth weights (1605 g in early-onset preeclampsia vs 2930 g without preeclampsia; P = .001). Similarly, in controls, birth weights were significantly lower in early-onset preeclampsia cases (1440 g vs 3310 g; P = .022). Birth weight centiles showed a trend toward lower values in preeclampsia, particularly in early-onset cases, although not all differences reached statistical significance. SGA rates trended higher in pregnancies complicated by preeclampsia in both cohorts, without reaching statistical significance.

Placental pathology

Placental pathology was available in 44 pregnancies of individuals with SCD and 31 Black control pregnancies (Table 4). The incidence of placental pathology was high across all groups and dominated by MVM. The risk of MVM disease was especially high in individuals with SCD, with greater frequency in the presence of preeclampsia. In pregnancies of individuals with SCD, placental pathology was observed in 75% of early-onset preeclampsia and 100% of late-onset preeclampsia cases, and in 69% of cases without preeclampsia. Among controls, rates were 67%, 86%, and 47%, respectively. MVM was the predominant pathology in both cohorts.

Lower PlGF levels were consistently associated with MVM across all gestational ages in pregnancies of individuals with SCD. At 24 to 28 weeks gestation, median PlGF levels were significantly lower in pregnancies with MVM (300 pg/mL) than those without (540 pg/mL; P = .002). This trend persisted at 28 to 32 weeks (P = .015) and beyond 32 weeks (P = .015) gestation. Similar patterns were observed in the control group, with lower PlGF levels correlating with MVM, although the PlGF cutoff of <100 pg/mL did not significantly correlate with MVM in either cohort.

No significant association of fetal vascular malperfusion with low PlGF levels was observed in either individuals with SCD or in controls at any gestational age.

Discussion

The principal findings of this study are threefold. First, as in the pregnant population without SCD, low circulating PlGF may be used as a diagnostic test for suspected preeclampsia, especially early-onset disease. Second, individuals with SCD are highly vulnerable to the most common underlying type of placental disease, described as MVM. Finally, when individuals with SCD have low circulating PlGF, this is most likely caused by significant placental disease, necessitating enhanced care, surveillance, and preparation ahead of potential iatrogenic preterm birth. Pregnant individuals with SCD have unacceptably high rates of adverse pregnancy outcomes. The inclusion of PlGF testing into care pathways for such people has the capacity to enhance safety and improve perinatal outcomes.

PlGF, a proangiogenic factor that belongs to the vascular endothelial growth factor (VEGF) family, has the capacity to indirectly promote the binding of the potent VEGF to the functional VEGF receptor 2, which activates endothelium-dependent vaso-relaxation of the systemic arteriolar vessels and thus suppresses systemic vascular resistance. PlGF is typically expressed at very low or undetectable levels in healthy nonpregnant tissues, but its production is upregulated in proinflammatory disease states, such as SCD, which is characterized by increased red cell, platelet, and leukocyte activation, as well as complement activation and endothelial dysfunction,^3,4 in which PlGF plays a counter-regulatory role to balance proinflammatory signals.⁵ PlGF produced by erythroid cells is expressed at higher levels in individuals with more severe SCD disease phenotypes, in comparison with those with milder phenotypes and with unaffected controls.²⁵ PlGF is further increased in the context of hemolysis, correlating with elevated hemolytic markers (reticulocyte count, lactate dehydrogenase, and uric acid), as well as in the context of high ferritin values.^26,27

In the context of pregnancy, PlGF stimulates neoangiogenesis, contributing to the transformation of the utero-placental vasculature into a low-resistance system.¹⁵ During a low-risk pregnancy, PlGF increases as gestation advances, peaking at 30 weeks, and declining thereafter, alongside a concurrent third trimester parallel rise in sFlt-1 until delivery.²⁸ In contrast, in pregnancies affected by preeclampsia, early pregnancy angiogenesis is suboptimal, with impaired transformation of the spiral arteries, owing largely to the angiogenic imbalance between PlGF and sFlt-1,²⁹ and with low PlGF levels observed in early second trimester, well ahead of clinical evidence of disease.²² Interestingly, contrary to the inverse association of PlGF and sFlt-1 demonstrated in preeclampsia, in SCD, concentrations of sFlt-1 are elevated at baseline and do not increase further during acute vaso-occlusive events.³⁰

In a meta-analysis assessing the predictive accuracy of PlGF in asymptomatic pregnancies, low PlGF levels had a predictive odds ratio of 9. Subgroup analysis of PlGF thresholds demonstrated the highest predictive odds ratio of 25 for PlGF levels of 80 to 120 pg/mL, with a sensitivity of 0.78, specificity of 0.88, positive likelihood ratio of 6.3, and a negative likelihood ratio of 0.26.¹⁵ In another study, PlGF of <100 pg/mL was associated with more severe hypertension, worse maternal outcomes, and higher rates of preterm birth and SGA infants, whereas identification of levels 12 to 100 pg/mL with attendant higher level care allowed mediation of outcome severity.²³

The differences in test performance between early- and late-onset preeclampsia underscore the previously described distinct pathophysiological mechanisms that underlie these conditions, previously often thought to span a continuum of disease. In that respect, evolving understanding suggests that early-onset preeclampsia (developing <34 weeks gestation) occurs secondary to defective placentation, including shallow trophoblast invasion and failure of the spiral adaptation to low-resistance flow, whereas late-onset preeclampsia (≥34 weeks gestation) is likely a reflection of the interplay between normal placental senescence and a maternal genetic predisposition to cardiometabolic disease.³¹ In line with findings from pregnancies of individuals without SCD, those with early-onset preeclampsia consistently exhibited the lowest PlGF levels, reflecting a more severe degree of placental dysfunction. This aligns with the established pathogenesis of preeclampsia, driven by poor placentation, inadequate angiogenesis, and remodeling of uterine spiral arteries.³¹ The impetus for the study was the observation that baseline PlGF levels are often elevated in individuals with SCD due to chronic inflammation and endothelial dysfunction inherent in the disease,3, 4, 5 potentially confounding the use of PlGF for prediction of preeclampsia in this group. Despite observations of elevated PlGF levels in nonpregnant cohorts of individuals with SCD, in our study PlGF levels in early-onset preeclampsia were comparable or slightly lower than those in Black controls with early-onset preeclampsia, underscoring the additional placental burden imposed by SCD.

The prevalence of early-onset preeclampsia was significantly higher in pregnancies of individuals with SCD than Black controls, with 4.9% of pregnancies of individuals with SCD developing early-onset preeclampsia vs 1.5% of Black controls (P = .03). This suggests a compounding effect of SCD-related vascular dysfunction on placental pathology. Chronic hemolysis, vascular inflammation, and endothelial injury in SCD are likely to exacerbate the risk of severe early-onset placental dysfunction, contributing to the observed higher prevalence. These findings align with previous research, including that by Oteng-Ntim et al, who reported an odds ratio of 2.24 for overall preeclampsia in pregnancies of individuals with SCD compared with pregnancies of individuals without SCD.¹⁶

Low-dose acetylsalicylic acid (ASA) has been shown to effectively reduce the risk of preterm preeclampsia in pregnancies at increased risk.³² Although conducted in the non-SCD population, findings have been extrapolated and SCD guidelines now recommend the use of low-dose ASA for primary prevention of preeclampsia in this population.³³ Although most of the SCD group in our study were prescribed low-dose ASA, its use in our control group has not been recorded. Interestingly, ASA has not been shown to have any significant effects on PlGF trajectories when compared with placebo in individuals (without SCD) at risk of preterm preeclampsia.³⁴

The PlGF cutoff <100 pg/mL plays a distinct role in preeclampsia prediction.²² In our study, for early-onset preeclampsia, this cutoff demonstrated exceptionally high predictive performance, achieving 100% sensitivity and 100% specificity at 20 to 24 weeks gestation, indicating that a PlGF level of <100 pg/mL at this gestational age would have high utility in predicting all cases of early-onset preeclampsia in patients with SCD.

In contrast, the same PlGF cutoff of <100 pg/mL showed markedly lower sensitivity for predicting late-onset preeclampsia, with a reduced sensitivity of 20% to maintain a 100% specificity, essentially necessitating a much higher PlGF threshold for accurate prediction. Specifically, a PlGF level of 832 pg/mL at 20 to 24 weeks gestation was required in individuals with SCD to achieve 100% sensitivity for prediction of late-onset preeclampsia, at a cost of a specificity of only 7%.

In a cohort of healthy nulliparous women, low PlGF levels in the early second trimester were linked with characteristics of MVM on placental pathology and its related adverse outcomes,³⁵ and low PlGF was shown to be useful in the identification of MVM-associated placental FGR.³⁶ These data are in keeping with our previous data in patients without SCD, in whom placental disease (especially MVM) was near universal in early-onset preeclampsia, and for whom the cause of stillbirth with low maternal PlGF was characterized by severe placental disease and evidence of MVM, whereas stillbirths with normal maternal PlGF demonstrated minimal placental pathology.¹⁹ This study likewise demonstrated an association between low PlGF and placental pathology, particularly MVM, further emphasizing its utility as a marker of placental health. MVM was the predominant pathology in both cohorts and was linked to low PlGF levels across all gestational ages. This observation is consistent with a recent study linking MVM as the principal pathologic lesion in individuals with SCD, strongly associated with adverse pregnancy outcomes, which, in that study, included intrauterine fetal death, early neonatal death, preterm birth, SGA, and HDP.³⁷

More recently, we have conducted a large-scale PlGF screening study measuring PlGF in tandem with testing for gestational diabetes at 24 to 28 weeks gestation, in which the same cutoff of 100 pg/mL showed high test precision to predict iatrogenic preterm delivery before 34 weeks gestation, mostly driven by early-onset preeclampsia and/or FGR.³⁸

From a pragmatic, patient-management perspective, determination of the risk of preeclampsia via measurement of PlGF levels allows for identification of a potential need (maternal and/or fetal) for iatrogenic preterm delivery,³⁸ which then prompts the following approach:

• 1.Continued management in a tertiary care center
• 2.Antenatal neonatology/pediatrics consult
• 3.
  Administration of a course of steroids for fetal lung maturation

  • •
    Owing to reports of increased vaso-occlusive events and hospitalizations in nonpregnant patients with SCD treated with corticosteroids,^39,40 this intervention may need to be accompanied by maternal education, encouragement to proactively institute measures used typically to manage vaso-occlusive event pain, provision of analgesics to be readily available for use, and potential consideration of maternal admission for the duration of steroid administration depending on the complexity of SCD history.
• 4.High-level surveillance for severe preeclampsia/HELLP (hemolysis, elevated liver enzymes, and low platelet levels) syndrome and FGR, including clinical, laboratory, and ultrasound assessment including Doppler-based fetal surveillance
• 5.Counseling regarding the potential for, and mode of, early delivery

We understand from the national Canadian Neonatal Network data that planned delivery (as opposed to emergency care) increases the use of both magnesium sulfate for fetal and maternal neuroprotection at birth, together with the use of delayed cord clamping and the aforementioned pragmatic approach, triggered by incorporation of PlGF assessment, can aid in maximizing uptake of these important interventions.

Future studies should focus on validating these findings in larger, multicenter cohorts of pregnant individuals with SCD and exploring the integration of PlGF with other biomarkers, to enhance its predictive accuracy. Investigation of the interplay between SCD-specific factors (such as vaso-occlusive episodes and homolysis) and placental dysfunction may provide further insights into the unique pathophysiology of preeclampsia in this population. Exploration of the potential of low molecular weight heparin to repress the development of severe placental MVM disease in this population may also be of interest. Finally, our data did not allow for exploration of the impact of transfusions, hydroxyurea, or anticoagulation, and this too represents a prime target for future work.

Conclusions

This study demonstrates, to our knowledge, for the first time, the critical role of PlGF in predicting early-onset preeclampsia in pregnancies of individuals with SCD. Specifically, a PlGF cutoff of <100 pg/mL proved highly discriminative in predicting early-onset preeclampsia, with 100% sensitivity and specificity at 20 to 24 weeks gestation. However, the same cutoff was inadequate for predicting late-onset preeclampsia, in which much higher PlGF levels were required for accurate prediction. By refining predictive strategies, clinicians can better anticipate and mitigate adverse pregnancy outcomes, improving maternal and fetal health. Future research should continue to explore the mechanisms underlying the distinct pathogenesis of early- and late-onset preeclampsia in SCD and validate these predictive thresholds in larger cohorts.

Conflict-of-interest disclosure: K.H.M.K. reports research funding from Agios Pharmaceuticals and Pfizer; consultancy fees from Alexion Pharmaceuticals, Agios Pharmaceuticals, Biossil, Bristol Myers Squibb, Forma Therapeutics, Pfizer, Novo Nordisk, and Vertex Pharmaceuticals; consultancy for, and honoraria from, Agios Pharmaceuticals and Bristol Myers Squibb; and serves on the data safety monitoring board for Sangamo. The remaining authors declare no competing financial interests.