Prediction of preterm birth in growth‐restricted and appropriate‐for‐gestational‐age infants using maternal PlGF and the sFlt‐1/PlGF ratio—A prospective study

To assess the utility of placental growth factor (PlGF) levels and the soluble fms‐like tyrosine kinase‐1/placental growth factor (sFlt‐1/PlGF) ratio to predict preterm birth (PTB) for infants with fetal growth restriction (FGR) and those appropriate for gestational age (AGA).


| I N TRODUC TION
Preterm birth (PTB) and fetal growth restriction (FGR) are significant contributors to the worldwide burden of disease.Globally almost 20 million infants are born with low birthweight-many of whom are growth restricted, 1 while 15 million are born preterm, [2][3][4] the majority in low-and middle-income countries. 5Stillbirth and severe neonatal morbidity rates are highest in these infants [6][7][8][9] and survivors are at increased risk of cerebral palsy, social and cognitive problems, anxiety, depression, childhood/adult obesity, and cardiovascular and metabolic disease. 10Furthermore, women from culturally and linguistically diverse groups, migrants, and refugees, socioeconomically disadvantaged backgrounds, adolescents, and those who live in remote/ rural communities have the highest rates of small or suboptimally grown infants and PTB, and experience a range of health inequities-this is an unmet clinical need and a key area of maternal and child health research. 11ormal placental development requires synchrony of several highly regulated processes involving angiogenic factors, hormones, transcription factors, cytokines and cell adhesion molecules. 124][15] Furthermore, FGR and pre-eclampsia often necessitate medically indicated PTB. 3,16However, despite many similarities in abnormalities of placental structure and function in women with pre-eclampsia and FGR, the relationship between the extent of failure of spiral artery conversion, gestational age at onset, type of disease, and maternal and infant phenotype, as well as clinical outcomes, remains poorly understood. 17,18lacental dysfunction is associated with an imbalance in circulating placentally derived angiogenic and antiangiogenic factors. 14,19,205][26][27][28] However, the optimal predictive thresholds for these biomarkers for PTB in women with FGR remain unknown. 29As gestational age at birth is a key determinant of short-and long-term outcomes, a test that reliably predicts PTB in this cohort would be a significant addition to the obstetric armamentarium for clinicians.This is because one of the main challenges faced by healthcare providers when managing pregnancies complicated by FGR is the degree of surveillance required to prevent stillbirth while balancing the risks of neonatal mortality and severe morbidity that come from iatrogenic PTB. 30 Currently, many international guidelines [31][32][33] recommend using fetal biometry and growth velocity and multivessel fetal Doppler ultrasound parameters 34,35 to monitor infants with FGR.However, as the rate of deterioration of fetal well-being can be variable and sometimes difficult to predict, 36 better tools for monitoring and prediction of adverse outcomes are required.8][39][40] The aim of this prospective cohort study was therefore to assess the association between low PlGF levels and elevated sFlt-1/PlGF ratio for any PTB (<37 weeks of gestation) with stratification for spontaneous and medically indicated PTB in women with early and late FGR.

| M ET HODS
This was a prospective, observational, cohort study conducted at the Mater Mother's Hospital in Brisbane, Australia.Ethics and Governance approvals were provided by the Mater Misericordiae Limited Human Research Ethics Committee (HREC/MML/66263) and the Mater Governance Office, respectively.Women were recruited, after providing written informed consent, from a dedicated fetal growth and assessment clinic within the Mater Centre for Maternal and Fetal Medicine from May 2022 to May 2023.Only women with singleton fetuses without known genetic syndromes or aneuploidy or major structural malformations were included.Gestational age was estimated based on first-trimester crown-rump length measurement. 41t enrolment, all women had their blood pressure, weight and height measured.Fetal biometry (biparietal diameter, head circumference, abdominal circumference (AC), femur length) and deepest liquor pocket were measured and estimated fetal weight (EFW) was calculated using Hadlock's formula. 42The International Fetal and Newborn Growth Consortium (INTERGROWTH-21st) growth centiles charts 43 were used for fetal biometry measurements.In addition to fetal biometry, Umbilical Artery Pulsatility Index (UA PI), Middle Cerebral Artery Pulsatility Index (MCA PI), Cerebroplacental Ratio (CPR), Ductus Venosus Pulsatility Index (DV PI), Ductus Venosus Peak Velocity Index for veins (DV PVIV) and Mean Uterine Artery Pulsatility Index (UtA PI) were measured.Reference charts from Flatley et al., 44 Kessler et al. 45 and Cavoretto et al. 46 were used for the various Doppler indices.Assessment of fetal growth was undertaken every 2-4 weeks depending on the severity of the fetal condition.All ultrasound examinations were performed using the Voluson™ E10 (GE HealthCare, Chicago, IL, USA) ultrasound platform.Blood samples (taken in BD Vacutainer EDTA tubes; Becton Dickinson Labware, Franklin Lakes, NJ, USA) for PlGF, sFlt-1 and calculation of the sFlt-1/PlGF ratio were obtained at the first visit and then repeated 4-weekly.PlGF and sFlt-1 levels were measured using the BRAHMS KRYPTOR PLUS system (Thermo Fisher Scientific, BRAHMS GmbH, Hennigsdorf, Germany).
All study information (biomarker measurements, maternal demographic, current and previous pregnancy information, pregnancy outcomes and other relevant clinical details) were recorded in an electronic form using REDCap™.Women who were screened high risk from firsttrimester pre-eclampsia screening or those with known risk factors for placental dysfunction were advised to commence low-dose aspirin (100-150 mg daily) before 16 weeks of gestation. 47Ultrasound data were stored in a dedicated ultrasound reporting and archiving system (Viewpoint®; GE HealthCare).
Early and late onset FGR were defined using the Delphi consensus. 48Early FGR (<32 +0 weeks): (1) AC or EFW <3rd centile or (2) absent end-diastolic flow in umbilical artery or (3) AC or EFW <10th centile in combination with UtA PI >95th centile and/or UA PI >95th centile.Late FGR (≥32 +0 weeks): (1) AC or EFW <3rd centile or (2) at least two out of three of the following: AC or EFW <10th centile, AC or EFW crossing more than two growth quartiles and CPR <5th centile or UA PI >95th centile.A fetus appropriate for gestational age (AGA) was defined as one with AC and EFW >10th centile and UA PI <95th centile for gestational age.
At each visit, ultrasound findings were reviewed by a Maternal Fetal Medicine specialist who was blinded to the PlGF, sFlt-1 levels and the sFlt-1/PlGF ratio.All obstetric management decisions including timing of birth were based purely on fetal ultrasound parameters and/ or the overall maternal clinical condition, consistent with the International Society of Ultrasound in Obstetrics & Gynaecology (ISUOG) practice guidelines 33 and stagedbased management protocol for FGR by Figueras and Gratacos. 36A course of antenatal corticosteroids-intramuscular betamethasone 12 mg 24 hours apart was administered if PTB between 24 +0 and 34 +6 weeks of gestation was anticipated. 49Maternal magnesium sulphate prophylaxis (intravenous infusion of 4-g bolus over 20 minutes followed by 1 g per hour for up to 24 hours) for fetal neuroprotection was given if birth before 30 weeks was likely to occur. 50

| Statistical analysis
The primary study outcome was PTB (delivery before 37 weeks (259 days) of gestational age).We also investigated the relative risk (RR) for abnormal biomarkers resulting in PTB within 1, 2 and 3 weeks and the RR for abnormal biomarkers resulting in spontaneous and medically indicated PTB.Abnormal biomarkers were defined as PlGF <100 ng/L 51 and sFlt-1/PlGF ratio >5.78 for gestational age if <28 weeks and >38 for gestational age ≥28 weeks. 24ample-size calculations were based on abnormal PlGF levels (equivalent to ≤100 ng/L) resulting in approximately seven times the hazard of PTB (hazard ratio [HR] 7.17, 95% CI 5.08-10.13) 51and reported probabilities of PTB of 0.10 52 in normal-risk and 0.41 53 in high-risk populations.For 90% power and α of 5%, 111 participants were required to detect an HR of 7, using the more conservative PTB probability of 0.10.
The distribution of continuous variables was assessed using histograms.Maternal demographic characteristics were compared using Fisher's exact tests, chi-square tests, t tests, Wilcoxon Rank Sum tests, analysis of variance (ANOVA) or Kruskal-Wallis ANOVA, according to the variable type, distribution and number of comparison groups.Log rank tests and univariable Cox proportional hazards models were used to evaluate the effect of abnormal biomarkers on PTB.
Relevant variables considered for potential inclusion in multivariable Cox proportional hazards models included pre-eclampsia, fetal growth restriction, intrauterine infection, maternal age, antepartum haemorrhage, maternal body mass index (BMI), smoking, illicit drugs and alcohol, preterm prelabor rupture of membranes, previous PTB, diabetes mellitus and antiphospholipid syndrome.Of these, pre-eclampsia and FGR status were deemed clinically relevant, that is causally related to the exposure and the outcome.We stratified by FGR status rather than adjusted for this variable, because FGR status could also potentially act as an effect measure modifier as well as a confounding variable.In all models, pre-eclampsia changed the unadjusted coefficient by a minimum of 10% for all models. 54The analysis period was defined as the time from sample to the time of delivery or censoring.A multiple record format, with each test considered a separate observation, was used to enable explanatory variables (biomarkers) to be analysed at multiple time points throughout pregnancy.Failure was defined as the occurrence of PTB.Models accounted for clustering at the patient level due to repeated observations (multiple blood tests).The proportional hazards assumption was checked using scaled Schoenfeld residuals plotted over time. 54oodness of fit was confirmed by plotting the Cox Snell residuals as failure times against the Nelson Aalen cumulative hazard function. 54Potential influential observations were checked by plotting Df-β residuals against time. 54Risk estimates were presented as HR for the time to PTB.Incidence rates and median survival times were calculated from the number of failures (PTB) and the time from maternal blood test (i.e.PlGF and sFlt-1/PlGF ratio measurement) to PTB or right censoring (persondays at risk [PDAR]) from the maternal blood test.The RR of abnormal biomarkers and delivery within 1, 2 and 3 weeks and the RR of medically indicated PTB compared with spontaneous PTB was calculated from the number of events divided by the PDAR.Significance was set at α = 0.05 for all statistical tests.
The reporting of this study conforms to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement. 55Statistical analyses were performed using Stata 18® (Statacorp LLC, College Station, TX, USA).
Table 1 lists the demographic characteristics of the study population.Other than higher rates of nulliparity, smoking, hypertension and pre-eclampsia in women with FGR there were no other demographic differences.However, compared with women with late FGR, those in the early FGR cohort had higher median BMI, higher mean arterial pressure at recruitment and higher rates of hypertension, pre-eclampsia and aspirin use.Median PlGF levels were also significantly lower (54 versus 229 ng/L, p < 0.001) and median sFlt-1 levels (2774 ng/L versus 2096 ng/L, p < 0.001) and sFlt-1/PlGF ratio (35 versus 10, p < 0.001) were significantly higher in the early FGR group.The median gestational age at delivery was significantly lower in the early FGR group (240 versus 265 days, p < 0.001) when compared with the late FGR group.Other pregnancy outcome details are presented in Table A1.
The RR for birth within 1, 2 and 3 weeks associated with abnormal biomarker levels are presented in Table 3. PlGF <100 ng/L significantly increased the risk of delivery within

| Main findings
The key findings of this large prospective study are, first, a low maternal PlGF level or raised sFlt-1/PlGF ratio strongly predicts PTB in both FGR and AGA pregnancies.The highest HRs were present when PlGF levels were <100 ng/L at diagnosis.Second, we found that women with PlGF levels <100 ng/L or elevated sFlt-1/PlGF ratio were at increased risk of medically indicated PTB within 3 weeks of diagnosis-these risks are greatest in the late FGR cohort.Our results also show that the interval from diagnosis to PTB was shortest in women with PlGF levels <100 ng/L with the lowest median duration in the early FGR cohort.

| Strengths and limitations
The main strength of this study is its prospective design and large cohort of well characterised cases of FGR.We used simple and well validated cutoffs for PlGF 51 and the sFlt-1/PlGF ratio 24 and compared their performance in early and late FGR cohorts as well as in an AGA control group.7][58] We also analysed our data by adjusting for pre-eclampsia to reduce its potential confounding effect on outcomes.Furthermore, clinicians were blinded to PlGF and sFlt-1/PlGF ratio results and hence decisions for delivery were not influenced by these factors.Our study population included women from diverse ethnic backgrounds, increasing the generalisability of our results.We used the sFlt-1/ PlGF ratio cutoff value of >5.78 and >38 used by Gaccioli et al. 24 for FGR instead of the more widely used ratio of 85 because the latter threshold was described for pregnancies complicated by pre-eclampsia.Limitations of our study include its single-centre focus.We also acknowledge that different immunoassay systems reported in the literature to   | 1097 PREDICTION OF PRETERM BIRTH USING PLGF AND 1/PLGF RATIO measure PlGF and sFlt-1 concentrations may yield varying results, 59 thus limiting comparisons of our findings with that of other investigators.Women in the AGA cohort were also not necessarily 'low risk' despite having an AGA infant.Many of these women were referred for serial monitoring of fetal well-being and growth because of risk factors (such as previous obstetric outcome, medical co-morbidities, age, BMI).We were unable to perform competing risk analyses of medically indicated and spontaneous PTB due to the low numbers of spontaneous PTB in our cohort.

| Interpretation
Our results are consistent with a study by Quezada et al., 56 who showed that a high sFlt-1/PlGF ratio was not only associated with an increased risk of PTB but was also related to shortened interval to delivery.Our findings show that the relative risk of PTB is greatest within 1 week from when biomarkers are measured and although this risk subsequently declines as gestation progresses, it remains elevated.Another study by Barton et al. 51 demonstrated that maternal PlGF ≤100 pg/mL was strongly associated with PTB independent of pre-eclampsia or gestational age and a more recent publication by Palma Dos Reis et al. 60 showed that in a cohort of early FGR, a high sFlt-1/PlGF ratio was strongly correlated (HR 9.87, 95% CI 5.06-19.24)with, and was predictive of, a shorter latency for earlier delivery or fetal demise.Our findings are also consistent with earlier observational studies. 61,62Benton et al. 61 reported that a maternal PlGF level <5th centile for gestational age outperformed gestational age, fetal AC and UA Dopplers for predicting FGR and a cutoff of <12 pg/mL predicted a shorter latency to delivery (13 versus 29.5 days, p < 0.001) whereas Reddy et al. 63 found in women with pre-eclampsia, the sFlt-1/PlGF ratio better predicted adverse perinatal outcome compared with EFW.
5][66][67] Current evidence suggests that a PlGF threshold of 100 ng/L (equivalent to 100 pg/mL) 51,62 or <5th centile 61 for gestational age are not only associated with high sensitivity and specificity for a diagnosis of FGR but also associated complications such as PTB.Indeed, our results show that this cutoff was strongly correlated with PTB for early FGR (HR 11.02, 95% CI 3.24-37.50)and late FGR (HR 34.53, 95% CI 4. 42-269.88)and more than ten times and four times the risk of PTB within 1 and 2 weeks of diagnosis in both cohorts, respectively.
It is now well established that pre-eclampsia and FGR are strongly associated with abnormal uteroplacental perfusion-imaging, histological (evidence of maternal vascular malperfusion and hypoxia-mediated injury to the placental villous tree) and biomarker data show that placental dysfunction results in or contributes to the maternal syndrome and/ or suboptimal fetal growth. 61,68From a biomarker perspective, although there are numerous molecules that are associated with placental dysfunction, currently the two most promising biomarkers are PlGF and sFlt-1 (and their ratio). 20hese proteins are stable at room temperature in blood samples rendering them suitable for batch processing. 69,70PlGF in particular has been shown to be a good screening test for pre-eclampsia because of its utility as a 'rule-out' test. 26,31,64,71here is evidence that when the increase in sFlt-1 and sFlt-1/ PlGF ratio is more rapid, the anti-angiogenic shift is reflective of acute placental failure and a more pro-inflammatory state that might trigger the onset of spontaneous labour. 9,72A recent review by Stepan et al. 22 reported that PlGF, sFlt-1 and sFlt-1/PlGF ratio are useful clinical tools for screening, diagnosing, predicting and monitoring placenta-related complications in pregnancy.In particular, maternal PlGF ≤100 pg/ mL predicted PTB with a sensitivity of 81.7% and specificity of 85.3%.Our results show that both low PlGF levels and high sFlt-1/PlGF ratio are strongly associated with PTB, although PlGF outperformed sFlt-1/PlGF ratio for this outcome.

| CONCLUSIONS
Low maternal PlGF levels and raised sFlt-1/PlGF ratio are strongly associated with an increased risk of PTB, including T A B L E 4 Relative risks for abnormal placental growth factor and soluble fms-like tyrosine kinase-1/placental growth factor (sFlt-1/PlGF) ratio levels resulting in medically indicated rather than spontaneous preterm birth.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared.

DATA AVA I L A BI L I T Y S TAT E M E N T
Data used to produce the results in this article will be available to any researcher provided appropriate ethics approval, inter-institutional data sharing agreements and other regulatory requirements are in place.A PPE N DI X 1

F I G U R E 3
Multivariable Cox regression model from time to recruitment to preterm delivery in days by soluble fms-like tyrosine kinase-1/placental growth factor (sFlt-1/PlGF) ratio.

HR p value aHR a p value HR p value aHR a p value HR p value aHR a p value HR p value Adj HR a p value
Hazard ratios for preterm birth according to placental growth factor level and soluble fms-like tyrosine kinase-1/placental growth factor (sFlt-1/PlGF) ratio.

PlGF level <100 ng/L -RR (95% CI) sFlt-1/PlGF ratio >5.78 if <28 weeks or >38 if >28 weeks -RR (95% CI) FGR (N = 300) n = 144 n = 133
Medically indicated preterm birth 35.20 (11.48, 175.46) 10.75 (5.26, 24.80) : AGA, appropriate for gestational age; FGR, fetal growth restriction; n, number of abnormal blood tests; N, number of blood tests; PlGF, placental growth factor; RR, relative risk; sFlt-1, soluble fms-like tyrosine kinase-1.medicallyindicated in FGR and AGA pregnancies, but PlGF performs better than the sFlt-1/PlGF ratio, especially in pregnancies complicated by late FGR.Collectively, our results support the incorporation of PlGF and sFlt-1 into prenatal management algorithms when FGR is present.Their predictive utility for PTB may allow transfer of women to more appropriate facilities, administration of antenatal corticosteroids and magnesium sulphate for neuroprotection as well as help to contextualise parental counselling.SK and JH conceptualised the study.JH and EC were responsible for data curation.KC and JH performed the data analysis.JH, KC, FSC and SK contributed to data interpretation.SK provided supervision and resources for the study.JH and SK wrote the original draft with KC contributing to the Methods section.SK made substantial revisions to the manuscript.All authors edited and approved the final version of the draft before submission for publication.Prof. Kumar is supported by the Mater Foundation and receives research funding from the Australian Medical Research Future Fund and the National Health and Medical Research Council.Jesrine Hong is supported by the University of Queensland and Mater Research.None of the funding parties had any in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript and decision to submit the manuscript for publication.We thank Jennifer Hong for her contribution in data curation.Open access publishing facilitated by The University of Queensland, as part of the Wiley -The University of Queensland agreement via the Council of Australian University Librarians.
AbbreviationsAC K NO W L E D GE M E N T S

p value a Early onset FGR Late onset FGR p value b N = 320 N = 141 N = 179 N = 83 N = 96
Pregnancy outcomes of the study population.Data are presented as median (interquartile range) for continuous measures, and n (%) for categorical measures.Significant at p < 0.05.Abbreviations: AGA, appropriate for gestational age; AEDF, absent end diastolic flow; FGR, fetal growth restriction; GA, gestational age; IVH, intraventricular haemorrhage; LBW, low birthweight; MAS, meconium aspiration syndrome; NEC, necrotising enterocolitis; NICU, neonatal intensive care unit; RDS, respiratory distress syndrome; REDF, reversed end diastolic flow; SCN, special care nursery; TTN, transient tachypnoea of newborn; UA, umbilical artery.a p value comparing categories of AGA and FGR.
ORC I DJesrine Hong https://orcid.org/0000-0001-8585-2357SaileshKumarhttps://orcid.org/0000-0003-0832-4811REFERENC E S SU PP ORT I NG I N FOR M AT IONAdditional supporting information can be found online in the Supporting Information section at the end of this article.| 1101DICTION OF PRETERM BIRTH USING PLGF AND SFLT-RATIOA L E A 1Note: b p value comparing categories of AGA, early and late FGR.