sFlt-1/PlGF ratio as a predictor of pregnancy outcomes in twin pregnancies: a systematic review

Abstract

Objective

To review the usefulness of the sFlt-1/PlGF ratio to detect adverse pregnancy outcomes related to placental dysfunction in twin pregnancies.

Methods

A systematic review in Pubmed-Medline, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Web of Science, and National Guideline was performed. Studies were selected if they were published in the last 10 years, included a sample size equal to or greater than 10 twin gestations, determined the sFlt-1/PIGF ratio, and revealed the pregnancy outcome of the included patients.

Results

A total of 11 studies were selected. Outcomes related to the association between sFlt-1/PlGF ratio throughout pregnancy and perinatal outcome, particularly related to placental dysfunction (early and late-onset preeclampsia and FGR), were collected. The vast majority of studies showed an increased sFlt-1/PlGF ratio in twin pregnancies complicated with preeclampsia or other adverse perinatal outcomes compared with uneventful pregnancies. The included articles revealed promising results when evaluating the usefulness of the sFlt-1/PlGF ratio to rule out preeclampsia. The scarce available data regarding FGR suggests that the sFlt-1/PlGF ratio is a promising tool for detecting this pregnancy complication. Data concerning other aspects of the sFlt-1/PlGF ratio, such as its evolution during healthy twin pregnancies or variations according to chorionicity, is limited.

Conclusion

The sFlt-1/PlGF ratio in twin pregnancies is useful to detect, and particularly to rule out adverse pregnancy outcomes related to placental dysfunction, such as preeclampsia or FGR.

Keywords:

• Twin pregnancy
• sFlt-1/PlGF ratio
• placental dysfunction
• preeclampsia
• fetal growth restriction
• adverse outcome

Introduction

Over the past decades, the rate of multiple pregnancies has grown notoriously [1–3]. Although it is mainly due to fertility treatments, the increase in multiple pregnancies has also been influenced by other factors, such as the increase in maternal age [4] or obesity [5]. Twin pregnancies associate a higher risk of preterm delivery [6], infant mortality [7], postpartum hemorrhage, hypertensive disorders of pregnancy [8], or the need for a cesarean section [9], compared to singleton pregnancies.

Pregnancy complications related to placental dysfunction, such as preeclampsia or fetal growth restriction (FGR), are widely known to be more prevalent in multiple pregnancies than in singletons [10,11]. Indeed, gestational hypertension and preeclampsia affect 5–6% of singleton pregnancies and up to 13% of twin gestations [12,13]. This incidence varies according to risk factors such as assisted reproductive technology, maternal age, nulliparity, or associated comorbidities [14–16]. Hypertensive disorders of pregnancy are key causes of maternal-fetal morbidity and mortality, being responsible for more than 70,000 maternal deaths and 500,000 fetal deaths every year [17]. Fortunately, late-onset preeclampsia, which affects pregnant women over 34 weeks of gestation, is more common than early-onset preeclampsia [18]. According to most authors, FGR is defined as an estimated fetal weight by ultrasound below the 10^th percentile [19]. Its prevalence in twin pregnancies is higher than in singletons, reaching percentages up to 47% vs 8% respectively [11,19,20], especially in monochorionic (20–45%) compared to dichorionic twins (11–24%) [21]. Impaired Doppler study suggests severity, defines staging, and therefore contributes to select the optimal time of delivery [19].

Furthermore, twin pregnancies entail specific complications related to growth restriction: selective growth restriction and growth discordance [22,23]. In almost 60% of discordant twins, the smaller fetus is under the 10^th percentile [24]. This may be explained by anomalous patterns of vascular anastomoses and unequal placental sharing, leading to unbalanced access to nutrients [23].

Placental dysfunction is characterized by an anomalous trophoblastic invasion of the myometrial segment of the spiral arteries, leading to placental ischemia [25,26]. Consequently, this hypoxic environment promotes the discharge of antiangiogenic factors into the maternal circulation, such as soluble fms-like tyrosine kinase-1 (sFlt-1), and reduces the bioavailability of proangiogenic factors, such as placental growth factor (PlGF) [26]. The sFlt-1/PlGF as an indicator of placental dysfunction has been widely studied during the last decade [26–30].

Abnormal levels of sFlt-1 and PlGF can be detected in maternal blood secondary to placental release in cases of placental dysfunction [26,29]. Antiangiogenic factor sFlt-1 leads to peripheral vasoconstriction with the aim of recruiting extra blood volume and, secondarily, reducing ischemia [31]. Considering sFlt-1 acts by adhering to the receptor-binding domains of PlGF and vascular endothelial growth factor (VEGF), it is logical that it leads to decreased levels of free PIGF [32–35]. Variations of these angiogenic markers can be detected before the clinical manifestations appear. Some studies report an increase in maternal sFlt-1 levels 5 weeks before the onset of symptoms [29,30,36].

Scarce published evidence can be found related to the sFlt-1/PlGF ratio in twin gestations. Its development and validation are rather recent, as the first published articles mentioning its relevance go back to 2004. Nevertheless, in just less than a decade, it has become a fundamental tool for preeclampsia diagnosis and exclusion. To the best of our knowledge, this is the first systematic review regarding the sFlt/PlGF ratio in twin pregnancies. For this reason, the aim was to collect the existing data and summarize the evidence respecting the use of sFlt-1/PlGF ratio to detect placental dysfunction in twin pregnancies.

Methods

Information sources

This systematic review was developed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guide. Publications on the following sources were included: Pubmed-Medline, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Web of Science, and National Guideline. Plus, reference lists from the selected articles were also consulted. All articles following the inclusion criteria and no excluding criteria, published from 2012 to November 2022, were submitted to our study selection.

Eligibility criteria

All studies comprehending sFlt-1/PIGF ratio determination in twin gestations in order to predict fetal and maternal outcomes were considered adequate for this review. Subsequently, exclusive criteria were defined: (1) sample under 10 twin gestations or not involving twin gestations, (2) published before 2012, (3) not evaluating maternal-fetal outcomes, (4) not measuring sFlt-1/PIGF ratio, (5) studies with no full text available.

Search strategy

In order to define our search, a PICO question was defined: P (patient = twin gestations), I (intervention = sFlt-1/PIGF ratio measurement), C (comparison = singleton or no intervention), O (outcome = fetal and maternal outcome). Then, MeSH terms were defined: preeclampsia, twin, pregnancy, ratio, sFlt-1/PIGF, biomarkers, fetal growth restriction (FGR), abruptio placentae, and placental abruption.

Data collection process

The literature search was executed by one author (E.S.P). Formerly, titles and abstracts of all articles were screened independently by two authors (E.S.P and A.M.V). Both authors carefully selected articles that accomplished the criteria established previously. Finally, a discussion was performed between collaborators, and a final selection was made by consensus. The selection process is shown in the following flow diagram, Figure 1.

Figure 1. PRISMA diagram Reflecting the selection process.

Data synthesis

Given the heterogeneity of the data reported meta-analysis was not appropriate. Measures of statistical significance, if reported, are recorded in the tables. Nevertheless, most results are reflected in the discussion.

Bias assessment

In order to reduce the risk of bias in this systematic review, the described eligibility criteria were strictly defined and applied. An extensive research was launched in multiple information sources and including articles in English, French, and Spanish. This scrutiny was performed in an objective and reproducible process. Furthermore, and as previously detailed, two independent authors analyzed and screened the selected articles with the aim of contrasting and refining the final selection.

Once the articles were selected, the Newcastle-Ottawa Scale (NOS) was used to assess their quality. This scale is optimal for nonrandomized studies, including case-control and cohort studies, and establishes a star system judged on three broad perspectives: the selection of the study groups, the comparability of the groups, and the ascertainment of either the exposure or outcome of interest for case-control or cohort studies respectively.

Referring to the existence of missing data from studies, the NOS scale includes a specific category that assesses the follow-up of the exposed and non-exposed cohorts to ensure that losses are not related to either the exposure or the outcome. Our evaluation was limited by the statements and information given by the authors of each publication. Table 2 details the scores given to all articles in each category.

Table 2. Detailed quality assessment using NOS schedule. Scores 9–7, 6–4, under 4 are considered low, intermediate and high risk, respectively.

Study	Selection				Comparability	Outcome			Total stars (maximum 9)	Risk of bias
	Representa-tiveness of the exposed cohort	Selection of the non exposed cohort	Ascertainment of exposure	Demonstration that outcome of interest was not present at start of study	Comparability of cases and controls on the basis of the design or analysis	Assessment of outcome	Follow-up long enough for outcomes to occur	Adequacy of follow up of cohorts
M de la Calle et al. 2021	b (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	a (1)	7	Low
J. Hoffman et al. 2017	c (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	a (1)	7	Low
A Martínez Varea et al. 2022	a (1)	a (1)	a (1)	a (1)	0*^1	a (1)	a (1)	a (1)	7	Low
J Binder et al. 2020	c (0)	a (1)	a (1)	a (1)	0*^2	0*^3	b (0)*^4	a (1)	4	Intermediate
JM Faupel-Badget et al. 2015	a (1)	a (1)	a (1)	a (1)	0*^5	a (1)	a (1)	a (1)	7	Low
S Rana et al. 2012	a (1)	a (1)	a (1)	a (1)	0*^6	a (1)	a (1)	a (1)	7	Low
A Karge et al. 2021	b (1)	a (1)	a (1)	a (1)	0	0*^7	a (1)	a (1)	7	Low
L Saleh et al. 2018	b (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	b (1)	7	Low
L Dröge et al. 2022	a (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	a (1)	7	Low
R Choi et al. 2020	a (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	b (1)	7	Low
S Shinohara et al. 2021	a (1)	a (1)	a (1)	a (1)	0	a (1)	a (1)	a (1)	7	Low

*¹ differences were not statistically significant; *² age difference was not statistically significant; *³ clinicians evaluating women with suspected preeclampsia were not blinded for the sFlt-1/PlGF ratio; *⁴ there were some outlier cases with high sFlt-1/PlGF values in the delivery without pre- eclampsia group. A few of these patients had an assessment to delivery interval shorter than a week; *⁵ adjusted for gestational age, maternal age and artificial reproductive technologies used (all p < .001); *⁶ age difference was not statistically significant; *⁷ clinicians were not blinded for the sFlt-1/PlGF ratio.

In the furtherance of reducing heterogenicity between samples, the inclusion criteria were strict. Only studies regarding the usefulness of the ratio in twin pregnancies to detect or to rule out placental dysfunction, and with information regarding perinatal outcomes, were included. It is essential to recognize that the objective was to launch a qualitative systematic review regarding the usefulness of the ratio in detecting placental dysfunction in twin pregnancies. The included studies do not comprise an intervention, yet consist of descriptive evaluations. Therefore, the available data from each study regarding the usefulness of sFlt-1/PlGF for the diagnosis of placental dysfunction in twin pregnancies were collected. Given that the included studies were scarce, and the ratio was measured at different gestational ages, a meta-analysis was not performed.

Results

Search results

As seen in Figure 1, 287 articles were collected in four different databases. A total of 242 records were excluded after reading the title and abstract, as the main subject was not related to the investigated matter. A total of 21 reports were assessed for eligibility and submitted to an intense evaluation after dismissing duplicates. Finally, 11 documents were selected to constitute our review. Figure 1 details justifications for article exclusion.

Characteristics of included studies

Altogether, 994 twin pregnancies were gathered and analyzed. A total of 5 studies compared twin pregnancies to a singleton cohort, and so up to 3009 singleton pregnancies were recruited. Most studies had retrospective cohort designs. One systematic review was found, but it was not included in the final selection of articles as it coincided with most of the included articles and therefore didn’t provide additional data. Nevertheless, the information recruited was included in the discussion. The mean total sample size was 90.36 twin pregnancies per article, but it should be highlighted that the sample size was rather discordant (from 21 up to 269).

All the selected articles include sFlt-1/PIGF ratio measurement. Nevertheless, 5 articles include healthy mothers with no suspected preeclampsia, 3 of them exclusively (35.85% of the total twin sample) [37,38], while the rest of the recruited samples include patients with suspected or confirmed preeclampsia.

The gestational age established to measure the sFlt-1/PIGF ratio is heterogeneous, and so determinations from week 10 to week 41 are analyzed. However, only 2 articles record measurements during the first trimester of pregnancy [38,39]. Moreover, 4 articles distinguish between early and late-onset preeclampsia [37,40–42], while just 6 consider fetal growth restriction or small for gestational age. Detailed study characteristics are described in Table 1.

Table 1. Included articles and their basic characteristics.

Study	Design	Sample	Suspected or confirmed vs healthy	Gestational age measurment	Monochorionic vs dichorionic	Ratio and cutoff	PlGF	Sflt-1	Compared to singleton	PE (%)	EO PE vs LOPE	FGR or SGA	Combined to other factors
M de la Calle et al. 2021 [40]	Data analyzed from 3 prospective studies	269 twin	Healthy + 62 suspected PE	10 to >37 (7 windows)	Yes	Yes (>38)	Yes	Yes	Yes	Yes	No	Yes	No
J. Hoffman et al. 2017 [41]	Retrospective cohort study	16 twin − 67 singleton	Suspected PE	< 34 = 13, <37 = 29, ≥37 = 41	No	Yes 85 (< 34w) and 110 (≥34 w)	No	No	Yes	Yes 2 (12.5)	Yes	Yes	No
A Martinez Varea et al. 2022 [38]	Prospective cohort study	108 twin	Healthy	24w	Yes	Yes (>17)	Yes	Yes	No	Yes	Yes	Yes	Yes
J Binder et al. 2020 [43]	Retrospective cohort study	164 twins	Suspected PE	33.6 weeks (interquartile range, 30.0–35.2)	Yes	Yes (>38)	Yes	No	Yes	Yes 71 (43)	No	Yes	Yes
JM Faupel.Badget et al. 2015 [39]	Prospective cohort study	132 twins − 2255 singleton	Healthy	10, 18, 26, and 35 (n 91) - third trimester (31-39) and at delivery (33-41) (n = 41).	Yes	Yes (No cutoff)	Yes	Yes	Yes	No	No	No	No
S Rana et al. 2012 [42]	Prospective cohort study	79 twins	Suspected PE	33.9 (31.9–36.0)	No	Yes (85)	Yes	Yes	No	Yes 46 (58)	Yes	No	Yes
A Karge et al. 2021 [44]	Prospective cohort study	49 twins	Suspected PE	Not detailed	Yes	Yes (> 53)	No	No	No	34 (69)	Yes	Yes	?
L Saleh et al. 2018 [45]	Secondary analysis of prospective cohort study	21 twin − 21 singletons	Suspected PE	30 (median)	No	Yes (>38)	Yes	Yes	Yes	Yes 13 (62)	No	No	No
L Dröge et al. 2022 [46]	Case control study	49 twins − 292 singletons	49 twins: 18 PE (13 severe + 5 mild) + 31 healthy and 292 single: 45 PE (27 severe + 27 mild) + 238 healthy	33 + 3 to 33 + 5	No	Yes (>53)	Yes	Yes	Yes	Yes 18 (36.7)	No	Yes	No
R Choi et al. 2020 [47]	Retrospective cohort study	374 single − 29 twin	21 healthy + 3 healthy in initial assessment but finally suspected + 5 suspected	31.1 (28.8 − 32.8)	No	Yes (>53)	No	No	Yes	No	No	No	No
S Shinohara et al. 2021 [48]	Retrospective cohort study	78 twin	Healthy	28 + 0 to 30 + 6	No	Yes (22)	No	No	No	Yes 10 (12.9)	No	No	No

^aPE: preeclampsia; LOPE: late-onset preeclampsia; EOPE: early-onset preeclampsia; FGR: fetal growth restriction; SGA: small for gestational age.

Quality of included studies

The Newcastle-Ottawa Scale (NOS) was used. Table 2 gathers detailed information concerning the included studies and their individual score on each element of the NOS.

Evolution of sFlt-1/PIGF ratio in twin pregnancies

Compared to singleton pregnancies

Four groups studied sFlt-1/PIGF ratio levels along twin pregnancies and compared their evolution with singleton gestations [38,39,43,45]. When analyzing a sample of 269 twin pregnancies before 29 weeks, median, 5th, and 95th percentiles did not differ between healthy twin and singleton mothers. Nevertheless, from 29 weeks to delivery, median, 5th, and 95th percentiles were higher in twin (10.2, 38.7, and 38.7) versus singleton (3.75, 9.03, and 19.6) pregnancies [39].

On the contrary, other prospective cohort studies report higher sFlt-1/PIGF ratio levels during the entire pregnancy in healthy twins (26 vs. 3; p = .361) [43] with the greatest differences at week 35 (168.4 vs 29.0; p < .0001) [38]. In the same line, a reviewed case-control study also describes higher median ratio levels in twins vs singles with normal outcomes (13.29 vs 4.94; p ≤ .001) [45]. Nevertheless, when studying women with diagnosed placental dysfunction, these same authors find unmatching results: one obtains lower ratio values than in singletons (49 vs. 158; p = .002) [43], while the other does not report differences between both groups [44].

Comparing healthy twin pregnancies with those with preeclampsia

Seven of the included articles compare levels of sFlt-1/PIGF ratio in healthy twin pregnancies to those with a confirmed diagnosis of preeclampsia. Most authors (71.4%) agree that patients who developed preeclampsia had statistically significantly higher sFlt-1/PIGF ratios [37,40,44,45,47].

An article published in 2017 describes a slight but notable difference between both groups, achieving median levels of 69.8 (IQR 24.7) in preeclamptic pregnancies compared to 54.4 (IQR 30.9, p = .005) in healthy ones [41]. A later publication released in 2022 agreed with this finding by associating sFlt-1/PIGF ratios over 17 on week 24 to an increase in preeclampsia diagnosis (odds ratio, 37.13 [95% CI 4.78–288.25]; p = .002) and higher mean values compared to healthy pregnancies at this week (20.3 vs. 4.3, p = .002) [37].

J Binder et al. studied the probability of prompt delivery due to preeclampsia and detected significantly higher ratio levels in pregnancies ending within 1 week compared with those who did not (98.9 vs 27.6) and within 2 weeks compared with those who did not (84.2 vs 23.5; p < .001) [44]. These findings are in concordance with the exposed by other authors [48]: after analyzing 10 twin women with preeclampsia diagnosed in a 78-sample-sized retrospective study, a higher mean ratio was found compared to healthy twins (46.2 [22.2–64.6] vs 4.0 [0.5–70.7]; p < .001).

Ultimately, a European case-control study published in the past year not only describes increased levels in preeclamptic pregnancies compared with those with an uneventful pregnancy 164.22 vs 13.29 (p ≤ .001), but it also subdivides between severe and mild cases, reporting statistically significant differences in both groups (168.67 and 145.58 vs 13.29; p ≤ .001 and p = .037, respectively). Nonetheless, two other articles (28,5%) did not find significant differences between patients with suspected and proven preeclampsia [41,42].

Early onset vs late-onset preeclampsia

The evidence published referring to the diagnosis of early onset (EOPE) vs late-onset preeclampsia (LOPE) is rather scarce. Only three of the included articles make differentiation in this sense when analyzing preeclamptic cohorts, all of them obtaining similar results [37,40,41].

In their retrospective cohort study, including 16 twin pregnancies and 67 singletons with suspected preeclampsia, J Hoffman et al. detail that 13 women developed EOPE and 70 LOPE. In their investigation, the sFlt-1/PlGF ratio increased significantly in women with LOPE diagnosis who were suspected before 34 weeks (77.5 [71.3–79.1] vs. 181.0 [163.6–464.5]). Compared to this group, the sFlt-1/PlGF ratio at the last measurement was significantly lower in patients with suspected LOPE, tested over week 34. These results go in line with the observations made by S Rana et al. who found that differences in sFlt-1/PlGF ratio levels were more striking among women presenting preeclampsia prior to 34 weeks (n = 40), (97.7 [76.6–178.1] vs 31.7 [6.5–48.7] p = .001).

In a prospective cohort study, A Martinez-Varea et al. eventually detected 2 cases of EOPE (one of them associated with FGR) and 3 cases of LOPE (one of them with a prior diagnosis of early FGR) among the 108 healthy twin pregnancies analyzed in week 24. The mean ratio at 24 weeks was not statistically different between EOPE and LOPE (61 vs. 9, p = .11), probably due to the small sample size.

Fetal growth restriction (FGR)

Six groups collected results concerning fetal growth, although only three interpret results independently for twins.

Among a group of pregnant women composed of 16 twins and 67 singletons, 10 developed FGR. The sFlt-1/PlGF ratio did not differ in growth restricted (64.1 (IQR 21.3) and normal weighty neonates (61.4 [IQR 35.4]; p = .527), although no analysis was made independently for twins [40].

A similar limitation happens with a bigger sample-sized group, as patients suffering from fetal growth restriction are analyzed altogether with those suffering from preeclampsia or any other kind of fetal compromise (such as birthweight discordance, selective fetal growth restriction, and neonatal care unit admission). Even so, results display significantly lower mean ratios in FGR-diagnosed pregnancies than in those with indicated delivery after 1 week and 2 weeks of assessment due to preeclampsia (mean of 23.5 [5.6–49.7] vs 98.9 [63.4–171.6] and 84.2 [59.7–132.2], respectively) [44].

Other articles do independently analyze twin pregnancies. They conclude that a sFlt-1/PlGF ratio ≥17 at 24 weeks is associated with a significant increase in the frequency of FGR (OR 39.58 [95% CI, 6.31–248.17]; p < .001) [37], and support its predictive value for FGR (AUC = 0.755) [42].

Furthermore, information related to each biomarker’s independent use has also been evaluated in a prospective cohort study including 79 twin pregnancies. Lower sFlt-1 median values in small for gestational age (SGA) (14 714.0 [8096.0 –18 205.0]) than in those with no adverse outcome (7495.0 [3498.0–10482.0]) were determined. Moreover, the median PlGF for no adverse outcome was 224.0 (156.0 − 449.0), while SGA grounded up to 104.0 (96.0 − 187.0). Median sFlt-1/PlGF levels for no adverse outcome were also significantly lower (36.2 [7.1–71.3]) compared to SGA (105.1 [67.5–233.6]) [41].

Predicting adverse perinatal outcome and time until delivery

Comparing the evidence collected is fairly complicated, as each article defines adverse perinatal outcome (APO) disparately. Hence, the preceding results for each author are exposed.

J Hoffman et al. classify severe adverse outcomes as the diagnosis of severe preeclampsia/HELLP syndrome/eclampsia, any severe, life-threatening maternal or fetal disease, extremely preterm birth (<1000 g/32 gestational week), severe neonatal growth restriction (birth weight <5th percentile) and maternal or fetal death occurrence. It should be considered that no difference is made between twin and singleton pregnancies when results are analyzed. Severe adverse outcomes complicated 30/83 (36.1%) of their cases. Eclampsia and maternal death did not occur in their patient group. sFlt-1/PlGF ratio was not different in patients with or without adverse outcomes (54.4 (IQR 37.0) vs. 66.2 (IQR 20.8); p = .204). Nevertheless, in patients with an intermediate sFlt-1/PlGF ratio (33–85 [< 34 weeks] or 33–110 [>34 weeks], the overall preterm birth rate (< 37 weeks) was 27.7% (23/83), with an early preterm delivery rate (< 34 weeks) of 8.4% (7/83) and an extreme preterm birth rate (< 32 weeks) of 5/83 (6.0%). Plus, the median pregnancy duration after the first sFlt-1/PlGF ratio measurement was 5 (IQR 14) days and correlated negatively with the gestational age (r = 0.424; p < .001). Therefore, they conclude that an intermediate sFlt-1/PlGF ratio indicates a risk for preterm birth, independent of the occurrence of preeclampsia [40].

J Binder et al. study the event of delivery due to preeclampsia or adverse outcome (defined as maternal morbidity including intensive care unit admission, pulmonary edema, postpartum hemorrhage, or eclampsia) within 1 or 2 weeks of blood sampling. The area under the curve of sFlt-1/PlGF was significantly higher than for PlGF alone (mean 0.88 and 0.88 versus 0.81 and 0.80) for predicting delivery because of preeclampsia within 1 and 2 weeks of blood sampling (p = .055 and .001, respectively). Standalone predictive accuracy of angiogenic markers for maternal adverse outcomes was suboptimal with AUC values below 0.70 [44].

On their behalf, S Rana et al. define adverse outcomes as hemolysis, elevated liver enzymes, and low platelets syndrome; disseminated intravascular coagulation; abruption; pulmonary edema; cerebral hemorrhage; maternal, fetal, and neonatal death; eclampsia; acute renal failure; small for gestational age; and indicated delivery. Comparing women who experimented with adverse outcomes in 2 weeks, the median sflt-1 level was significantly elevated (p = .0004). Indicated delivery occurred in 47 women (90.4%), 27.7% occurred before week 34, and 39 deliveries before week 37. A higher sFlt-1/PlGF ratio was associated with a shorter duration of pregnancy (p = .03) for the full cohort and (p = .03) for <34 weeks [41].

Adverse perinatal outcome was defined by A Karge et al. as the presence of at least one of the following outcomes: respiratory distress syndrome, intubation, admission to neonatal intensive care unit, and arterial umbilical cord pH value < 7.10. APO occurred in 10 of 49 pregnancies in at least one of two twins (20.4%). The median sflt-1/PIGF ratio was not significantly different in patients with APO compared to those without (89.45 vs. 62.00, p = .669) [42]. Mean time until delivery was significantly negatively correlated with the ratio in < 34 weeks, while in ≥ 34 weeks, no significant differences were found.

L Saleh and his group defined maternal complications as the occurrence after admission of preeclampsia, HELLP, eclampsia, subcapsular hematoma of the liver, pulmonary edema, placental abruption, renal insufficiency, cerebral edema, cerebral hemorrhage, and vision disorders. Fetal and neonatal complications consisted of admission to the Neonatal Intensive Care Unit, treatment with an endotracheal tube, a birth weight percentile <10, respiratory distress syndrome, bronchopulmonary dysplasia, sepsis, an intracerebral abnormality such as bleeding, and fetal and neonatal death. In twin pregnancies, maternal complications occurred in 4 of the 12 patients (33%) with a ratio >38 at inclusion, but no difference in adverse fetal/neonatal outcomes was observed between patients with a ratio ≤38 or >38. No significant difference in gestational age at delivery was observed in twin pregnancies with a ratio ≤38 or >38 [43].

Finally, S Shinohara et al. describe that women with an sFlt-1/PlGF ratio of <22.2 delivered later than those with an sFlt-1/PlGF ratio of ≥22.2 (median duration of pregnancy after measuring serum angiogenic markers (4.7 vs. 7.3 weeks; p < .001). Moreover, 38.9% of women with an sFlt-1/PlGF ratio of ≥22.2 and 0.0% of women with an sFlt-1/PlGF ratio of <22.2 (p < .001) delivered within 4 weeks of preeclampsia onset [47].

Establishing an optimal cutoff: predictive values, sensitivity, and specificity

Most articles included aim to establish a cutoff value with optimal predictive values for twin pregnancies (83%). However, each author considers diverse thresholds due to the scarce evidence published until the date.

Three studies perform ROC analysis to establish the optimal cutoff for twin pregnancies (all sensitivities and specificities are detailed in Table 3 below). By comparing 49 twin pregnancies and 292 matched singleton pregnancies, 53 were established as an optimal cutoff in a multicenter case-control study [46], while other authors fix a cutoff value of 22.2 to be the best to predict preeclampsia onset within 4 weeks [48]. Moreover, determined levels of sFlt-1/PIGF ratio on 108 twin pregnancies on week 24 proved that women at risk of developing preeclampsia could be early identified by detecting ratio figures over 17 [37].

Table 3. Negative predictive values and positive predictive values for one week and two weeks after sFlt-1/PlGF ratio determination, specificity, and sensibility for the diagnosis of preeclampsia event.

%	Cutoff	Week	NPV 1w		NPV 2w	PPV 1w	PPV 2w	Spec	Sens
M de la Calle	38	Throughout pregnancy	91.9		83.8	20.0	44.	–	–
J Binder	38	33.6	98.8		96.4	–	–	–	–
J Hoffman	33		–
A Martinez Varea	17	24	98.0			42.9		96	60
S Shinohara	22	28–30	100			58.8		88.2	100
S Rana	85	33.9	–		–	–	–	90.9	61.1
	75		–		–	–	–	86.4	77.8
L Dröge	53		–		–	–	–	74.2	94.4
	33		–		–	–	–	67.7	100
	85		–		–	–	–	80.6	83.3
A Karge		No predictive value for sflt-1/PIGF and APO (AUC = 0.618, 95% CI: 0.387–0.849, p = .254).
R Choi		No predictive value for sflt-1/PIGF
L Saleh	38	30		Concludes sflt-1/PIGF ratio not applicable to twin pregnancies

NPV: negative predictive value; PPV: positive predictive value; Spec: specificity; Sens: sensibility.

The other five publishers examine sFlt-1/PIGF ratio ranges previously established in singletons with contradictory results. Data from three prospective studies (PROGNOSIS, STEPS, and a multicenter case-control study), with a total sample of 269 twin pregnancies, demonstrates that classic cutoffs used with single pregnancies can be transferred to twin pregnancies [39]. On the same line, by evaluating 164 twin pregnancies with suspected preeclampsia, another group concludes that an sFlt-1/PlGF ratio cutoff of 38 can be used to rule out delivery due to placental dysfunction in 1 week [44]. Likewise, the specificities and sensibilities of using a cutoff of 85 were evaluated by following 79 women with suspected preeclampsia in the 3rd trimester, obtaining significant and promising ciphers [41].

A retrospective analysis of 83 pregnant women with intermediate values for the sFlt-1/PIGF ratio verified its capacity to exclude or diagnose preeclampsia. The diagnosis was ruled out if the sFlt-1/PlGF ratio was < 33 and confirmed if the sFlt-1/PLGF ratio was 85 with < 34 weeks or 110 with 34 weeks, respectively. Nevertheless, there no independent evaluation for their twin sample (n = 16) was made, nor did they calculate predictive values for each cutoff [46].

Aversely, other publishers investigate ratio evolution and screen-positive rates for preeclampsia risk assessment with a cutoff of 58, but no screening positive worth was identified in twin pregnancies [41]. Similar results are found in a multicenter study that compared 21 twin and 21 singleton pregnancies: it is concluded that a cutoff ratio of 38 is not applicable to twin pregnancies [43].

Use of PlGF or sFlt-1 value as independent markers

The selected articles also independently evaluate sFlt-1 and PlGF levels to describe their normal evolution in healthy twin pregnancies. Additionally, some authors examine the biomarkers’ capacity to predict placental dysfunction.

PlGF

Three authors study its evolution in healthy twin pregnancies. In one study, PlGF levels were measured along 269 twin pregnancies (over 75% are healthy, the rest with suspected preeclampsia). Both groups’ concentration increased from 10 to 15 weeks until 29 weeks. Nevertheless, in women with singleton pregnancies, PlGF values then plateaued before decreasing until delivery. By contrast, in women with twin pregnancies, it decreased until 37 weeks before increasing sharply from 37 weeks until delivery [39]. Moreover, a prospective cohort study published in 2015 compared 132 twins to 2255 singletons by measuring biomarkers from week 10 to delivery and concluded that maternal PlGF concentrations were lower in twin than singleton pregnancies at week 35 only (219.2 vs 350.2 pg/ml; p < .0001) [38], with similar results than those published by others published in 2018 (228 vs. 440 pg/ml p = .479). Contrarily, other authors do not find differences in median PlGF serum levels when comparing twin with singleton pregnancies with a normal outcome (403.00 pg/ml vs 346.50 pg/ml (p = .095) [45].

Additionally, six articles explore PIGF’s progression and predictive value in twin pregnancies with poor outcomes. One author finds higher PlGF values in preeclamptic twin pregnancies compared to those in singleton (185 vs. 33 pg/ml, p < .001) and (p < .001 Wilcoxon rank-sum test), respectively [43]. When comparing levels in twin gestations with preeclampsia to twin controls, three studies discover significantly lower levels in those with adverse outcomes (138.80 vs 403.00 pg/ml (p ≤ .001) [38,41,45]. One of the mentioned studies, however, finds lower but not statistically significant differences in women with preeclampsia diagnosis (162.5 pg/ml [98.0–226.5] vs 224.0 pg/ml [156.0–449.0], p = .005), (p = .23) [41]. No significant predictive power is found when sFlt-1 or PlGF values are tested at 24 weeks gestation [37].

sFlt-1

Across all gestational age windows, higher sFlt-1 concentrations are detected in twin versus singleton, with a steeper increase after week 29. From then onward, levels are described up to 2–3 times higher (6377 vs 1732 pg/ml, p = .008) [38,39,45].

In twin pregnancies with adverse outcomes, median sFlt-1 was elevated (11,461.5 pg/ml vs 7495.0 pg/ml, p = .0004 [41] and 20 011.50 pg/ml vs 4503.00 pg/ml; p ≤ .001 [46]) and those with preeclampsia diagnose had a significantly higher median sFlt-1 level (p = .01) [41]. These findings were corroborated by other authors who found similar results, especially in the third trimester and around the delivery [38]. One study found higher but not statistically different values of sFlt-1 (9134 vs 8625 pg/ml) than singleton preeclamptic pregnancies [43].

Association to other parameters

Martinez-Varea et al. hold that although the addition of the mean pulsatility index of uterine arteries to the ratio does not improve the identification of patients that develop preeclampsia, it slightly enhances the selection of patients who develop FGR. Adding maternal characteristics and mean pulsatility index ratio enhances the identification of patients who develop preeclampsia or FGR [37].

J Binder’s group exposes that the predictive accuracy of PlGF for delivery within one week because of preeclampsia is improved by adding baseline characteristics. On the contrary, the addition of baseline characteristics (mean arterial pressure and gestational age at assessment) did not significantly improve the AUC values of 1- or 2-week models (p = .540 and .222, respectively) [44].

Finally, S Rana et al. concludes that the most favorable results are obtained with the combination of the sFlt-1/PlGF ratio and gestational age (AUC 0.83), as well as the model that included these two predictors along with the highest systolic blood pressure and proteinuria (AUC 0.85) [41].

Chorionicity

Five studies considered chorionicity in their twin sample. At least 66 monochorionic pregnancies and 271 dichorionic pregnancies were independently analyzed. However, two of the mentioned studies did not specify the number of patients in each group.

Most investigators agree that maternal concentrations of the sFlt-1/PIGF ratio are independent of chorionicity, corroborating its predictive accuracy in both groups [37,39,42,44].

On the contrary, one group found that maternal prenatal concentrations of sFlt-1 (10,721 vs 6169 pg/ml; p = .03) and the sFlt-1/PlGF ratio (47.0 vs 15.7 pg/ml; p = .03) were higher in monochorionic than dichorionic after adjustment for gestational age. Furthermore, placental weight was inversely correlated with the sFlt-1/PIGF ratio at the end of pregnancy, positively with PIGF, and had no association with sFlt-1 [38].

Discussion

In this systematic review concerning predicting maternal and perinatal outcomes associated with placental dysfunction in twin pregnancies, the sFlt/PlGF ratio appears to have a solid and promising role.

Preeclampsia syndrome is as twice as frequent in twin gestations. Additionally, it often occurs in more severe forms, present atypically, and its onset is usually earlier than in singleton pregnancies [16;50]. Thus, developing diagnostic tests to help obstetricians detect this syndrome in twin pregnancies is crucial.

Four of the included studies try to determine the physiological evolution of sFlt-1/PlGF ratio along pregnancy in twin pregnancies [38,39,43,45], as recent theories postulate that larger placental mass, higher inflammatory response, and association with ART procedures may determine different levels than those found in singletons [49]. Two of these studies find higher levels (one from week 29 onwards, the other during the whole pregnancy). However, the two groups obtained unclear and unmatching results when comparing pregnancies complicated with placental dysfunction, with lower levels found in twin pregnancies vs no significant differences compared to singletons. This aspect remains unclear and further investigations must be conducted.

Although significant in only five studies [40–42,44,47], an elevated ratio was found in groups of women who suffered preterm or indicated delivery secondary to preeclampsia diagnosis. Furthermore, two groups related an elevated ratio to greater rates of severe adverse maternal-fetal outcomes [41,44]. Along with these findings, most publishers (71.4%) agree that patients developing preeclampsia have statistically significantly higher sFlt-1/PIGF ratios [37,40,44,45,47]. Two other publishers describe higher mean ratios in their series, although not statistically significant. Unfortunately, only four studies consider EOPE and LOPE cases independently. Nevertheless, all of them find higher mean ratios in EOPE, and only one does not achieve statistical significance (probably due to the small sample included) [37].

One of the principal objectives in most of the included studies was to define an effective cutoff to rule out or diagnose preeclampsia in twins. Most authors evaluate the existing cutoffs for singletons (particularly 38 and 85). This finding should be considered promising, as 80% of the published results report a positive predictive value.

Fetal growth restriction is one of the significant causes of morbidity and mortality rates in twin pregnancies. Slower growth rates have been described compared to singletons, especially from weeks 30-32, probably due to placental and umbilical cord affection [50].

Significantly elevated ratios were found in 3 studies when evaluating fetal growth restriction [37,41,42], suggesting its determination could also be a compelling tool in the diagnosis of this syndrome. Although two other studies did not find significant results, this may be because no independent analysis was made for twin pregnancies [40,44].

Isolated information about abruptio placentae and HELLP syndrome was assembled, mainly considered adverse maternal outcomes, but with no independent evaluation. As both phenomena constitute complications secondary to placental dysfunction and are more frequent in twin pregnancies, further investigations should consider them separately [49,51].

The major area for improvement in most studies was the de consistency of comparison groups. Although most claim no existing differences between groups or that differences were not statistically significant, this is not sufficient for establishing comparability, as most evaluation scales request matched characteristics in the design and adjusted confounders in the analysis. Thus, most results are of limited clinical value due to the described variation in comparison groups. Also, the definition of feto-maternal outcome was considerably heterogeneous (as detailed along the collected results, each group considered diverse adverse events), just as the study design (especially gestational age chosen to determine sFlt/PlGF ratio). For this reason, no metanalysis could be made. Furthermore, many studies did not give independent results for twin populations but included them in the same groups as singletons. In addition, we were unable to evaluate the role of sFlt/PlGF ratio in the first trimester of pregnancy, as only one study includes determinations in this period.

The most significant limitation when evaluating evidence concerning sFlt-1/PlGF ratio in twin pregnancies was the scarce number of scientific publications regarding the topic. Only the twelve included articles, mostly retrospective, and one single systematic review [52] that discussed the subject were found. Also, no metanalysis could be completed due to the high heterogenicity between each study concerning design, gestational age for ratio measurement, the definition of the feto-maternal outcome, and every other parameter.

Finally, it must be highlighted that certain potential aspects and uses of the sFlt-1/PlGF ratio are scantly explored, such as its role in chorionicity and placentation during the first months of pregnancy, as just a few investigators have determined its levels along these stages.

Further investigations must be made in the interest of understanding twin pregnancies and their complications. Placental dysfunction-associated syndromes entail a high proportion of the potential complications in twin pregnancies, some leading to severe consequences in both newborns and mothers. The pursuance of developing effective tools to diagnose and therefore treat and prevent adverse outcomes must motivate new investigations, as well as the aim to establish reference values of the sFlt-1/PlGF to help obstetricians make clinical decisions.

In conclusion, the results of this systematic review are promising, though limited. A scarce number of publications were finally found to accomplish inclusion criteria, and therefore the studies included in this systematic review were scarce. Nonetheless, most studies agree on the existing differences between sFlt-1/PlGF ratio levels between twins and singletons, healthy twins vs those diagnosed with placental dysfunction-related syndromes, mainly preeclampsia and FGR, as well as recognizing the predictive capability of the sFlt-1/PlGF. Statistically significant numbers are scarce due to the mentioned limitations, but it should be highlighted that all results point in the same direction: sFlt-1/PlGF ratio in twin gestation is a promising tool to select pregnancies with placental dysfunction. Additionally, the sFlt-1/PlGF ratio has demonstrated great clinical utility in singleton gestations, and it should encourage investigators to find evidence that supports its application and usefulness in twins. Interestingly, the findings of this systematic review are consistent with the results of the only existing systematic review contemplating this topic [52], which was expected, as some of the included studies are in common [39,44,45].

Ethical approval

The present is a non-interventional, retrospective review study and does not need ethics approval.

Acknowledgments

Not applicable.

Disclosure statement

The authors report there are no competing interests to declare.

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials.

Additional information

Funding

No funding was needed.