Non-invasive prediction of fetal growth restriction in pre-eclampsia: integrating peripheral blood endothelial progenitor cell count with uterine artery Doppler velocimetry

Abstract

Objective

To investigate the predictive value of peripheral blood endothelial progenitor cells (EPCs) combined with uterine artery resistance index (RI) and pulsatility index (PI) for fetal growth restriction (FGR) in patients with pre-eclampsia (PE).

Methods

A retrospective study was conducted on 265 PE patients. Based on pregnancy outcomes, they were categorized into FGR and N-FGR groups. EPCs and RI/PI were analyzed using flow cytometry and Doppler ultrasonography. The correlations of EPCs with RI/PI in PE complicated with FGR, predictive value of EPCs, RI, PI and their combination for FGR in PE, and factors influencing FGR occurrence in PE were analyzed by Pearson correlation, ROC, and multivariate logistic regression analyses.

Results

In this PE cohort, the prevalence of FGR was 33.58% (89/265). Compared to the N-FGR group, the FGR group exhibited significantly lower EPCs and higher RI/PI. The Early-FGR group showed reduced EPCs and elevated RI/PI than the Late-FGR group. EPC count significantly inversely correlated with RI/PI in PE-FGR patients. The combination of EPCs, RI, and PI demonstrated superior predictive accuracy for FGR in PE. PE classification (OR = 5.501), RI (OR = 1.422), and peripheral blood EPCs (OR = 0.044) independently correlated with FGR occurrence, and PLT (OR = 0.986) and PlGF (OR = 0.942) independently correlated with reduced EPCs in PE patients.

Conclusion

Reduced peripheral blood EPCs and elevated uterine artery RI/PI are closely associated with FGR progression in PE, and their combination presents high predictive value for PE-related FGR. EPCs and RI are independently correlated with PE-related FGR, indicating their potentials for clinical prevention, diagnosis, and treatment of PE-related FGR.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00404-025-08123-2.

Introduction

Pre-eclampsia (PE), a pregnancy-specific hypertensive syndrome complicating approximately 4–6% of global pregnancies, represents a major contributor to maternal–fetal morbidity and mortality [1, 2]. This multisystem disorder typically manifests after 20 weeks of gestation through new-onset hypertension and proteinuria, primarily stemming from impaired placental development and maternal vascular dysfunction [3]. A critical consequence of PE involves fetal growth restriction (FGR), characterized by the fetus’s inability to achieve genetically predetermined growth parameters [4]. The underlying pathophysiology centers on abnormal placentation, particularly the inadequate remodeling of uterine spiral arteries by invasive trophoblasts [5]. Clinically, PE-associated FGR elevates risks of perinatal complications and predisposes offspring to neurodevelopmental impairments, metabolic syndromes, and cardiovascular disorders extending into later life stages [6], which underscores the urgency for early predictive strategies.

Emerging predictive approaches for PE-FGR integrate hemodynamic and cellular biomarkers [7]. Doppler ultrasonography (US) has gained prominence as a non-invasive modality for assessing uterine artery hemodynamics through resistance index (RI) and pulsatility index (PI) measurements [8, 9]. These indices reflect maternal vascular resistance alterations, with abnormal uterine artery waveforms predicting severe perinatal outcomes in hypertensive pregnancies, including FGR [10, 11]. Notably, uterine artery RI and PI demonstrate independent associations with adverse outcomes and correlate with disease severity in PE [12]. However, standardized criteria for interpreting these parameters in PE-FGR risk stratification remain elusive.

Beyond hemodynamic disturbances, endothelial dysfunction emerges as a cornerstone of PE-FGR pathogenesis. PE and FGR share origins in placental insufficiency from defective trophoblast invasion [13], compounded by circulating placental factors inducing maternal endothelial injury [14]. In addition, bidirectional vascular abnormalities, affecting both maternal decidual vessels and fetal placental vasculature, create a pathological synergy [15]. Endothelial progenitor cells (EPCs), bone marrow-derived precursors critical for vascular homeostasis [16], emerge as key regulators through their dual roles in vasculogenesis and endothelial repair via paracrine signaling [17]. Compromised EPC function may perpetuate spiral artery remodeling defects, impairing uteroplacental perfusion and potentially initiating PE [18]. Crucially, FGR pregnancies exhibit suppressed EPC mobilization and function in maternal peripheral blood and fetal circulation [19, 20], with gestational-age-matched comparisons revealing impaired EPC growth kinetics [15]. These findings position EPCs as promising biomarkers for pregnancy complications [21], though their interplay with Doppler parameters in PE-FGR prediction is unexplored.

Despite these advances, critical knowledge gaps persist. First, the regulatory hierarchy between cellular dysfunction (EPCs) and hemodynamic alterations (RI/PI) in PE-FGR development remains undefined. Second, existing studies predominantly examine these biomarkers in isolation, neglecting their potential synergistic predictive value. Third, no comprehensive model integrating EPC quantification with uterine artery Doppler profiling has been proposed for PE-FGR risk assessment.

To address these gaps, this study pioneers a dual-modal investigation combining flow cytometric EPC analysis with Doppler US in PE patients. We hypothesize that peripheral blood EPC depletion synergizes with elevated uterine artery RI/PI to predict FGR development in PE. By establishing quantitative relationships between cellular biomarkers and hemodynamic parameters, this work aims to: (1) develop a novel predictive algorithm for PE-FGR comorbidity; (2) provide mechanistic insights to guide targeted interventions. Our findings may redefine risk stratification paradigms and inform precision management strategies for this high-risk obstetric population.

Materials and methods

Subjects

A retrospective cohort study was conducted on 395 PE patients who delivered at The Affiliated Hospital of Xuzhou Medical University between January 2021 and June 2023. According to predefined inclusion and exclusion criteria, 265 PE patients were enrolled and categorized into 2 groups based on different pregnancy outcomes: a FGR group (N = 89) and a non-FGR (N-FGR) group (N = 176). This study was carried out complying with the ethical principles of the World Medical Association Declaration of Helsinki [22], relevant specifications and regulations for clinical research, as well as the guidelines from the Enhancing the QUAlity and Transparency Of Health Research (EQUATOR) network [23]. The study protocol received approval from the Ethics Committee of The Affiliated Hospital of Xuzhou Medical University.

Diagnostic criteria for PE

PE was diagnosed according to ISSHP criteria [24, 25]: systolic blood pressure (SBP) ≥ 140 mmHg and/or diastolic blood pressure (DBP) ≥ 90 mmHg after 20 weeks of gestation, accompanied by 24-h proteinuria (PRO) ≥ 0.3 g or random PRO test ( +). PE was generally classified as mild or severe. Severe PE was defined per the American College of Obstetricians and Gynecologists (ACOG, 2008) guidelines [26] as meeting one or more of the following criteria: SBP/DBP ≥ 160/110 mmHg on two or more occasions, 24-h PRO ≥ 3 g, visual impairments, oliguria, epigastric pain, acute kidney injury (creatinine ≥ 1 mg/dL), elevated liver enzymes (transaminase levels > 40 IU/L), or decreased platelet count (PLT; < 150 × 10⁹/L). Patients not meeting severe criteria were diagnosed with mild PE. Subsequently, based on the onset time of PE, we further divided PE into early-onset pre-eclampsia (EOPE; onset time < 34 weeks of gestation) and late-onset pre-eclampsia (LOPE; onset time ≥ 34 weeks of gestation) [27].

Diagnostic criteria for FGR

FGR was defined according to the Society for Maternal–Fetal Medicine (SMFM) guidelines [28] and Hadlock criteria as (1) abnormal fetal biometry (with Hadlock criteria as a reference curve for biometry), an estimated fetal weight (EFW) below the 10th percentile for gestational age (Grade 1B); (2) abnormal Doppler blood flow in the umbilical artery (meeting any of the following criteria): umbilical artery PI > 95th percentile of gestational age, loss of end diastolic blood flow, or reverse end diastolic blood flow. FGR was further classified as early onset if diagnosed before 32 weeks of gestation or late onset if diagnosed at or after 32 weeks [29].

Inclusion and exclusion criteria

Inclusion criteria: (1) met the diagnostic criteria for PE as defined in this study; (2) maternal age > 20 years at enrollment; (3) gestational age ≥ 20 weeks at diagnosis; (4) no history of substance/alcohol abuse or smoking; (5) complete clinical records.

Exclusion criteria: (1) concurrent pregnancy complications unrelated to PE or FGR, including postpartum hemorrhage, amniotic fluid embolism, pregnancy diabetes, and premature rupture of membranes; (2) thyroid disorders; (3) autoimmune diseases; (4) multifetal gestation (twins or higher-order multiples); (5) placental/umbilical cord abnormalities or structural fetal malformations detected on ultrasound; (6) chromosomal abnormalities or genetic syndromes; (7) intrauterine infection; and (8) statin use during pregnancy or recurrent miscarriage.

Sample and data collection

The data collected when the subjects were admitted for prenatal examination at 20 weeks of pregnancy included maternal age, DBP, SBP, body mass index (BMI), gravidity, history of hypertension, prior FGR, family history of FGR, PE severity, PE classification, aspirin intervention history, gestational age at diagnosis, fetal birth weight (FBW), daily dairy intake, and laboratory parameters [fasting blood glucose (FBG), triglyceride (TG), total cholesterol (TC), PLT, prothrombin time (PT), thrombin time (TT), uric acid (UA), alanine aminotransferase (ALT), PRO, and placental growth factor (PLGF)]. Peripheral venous blood (5 mL) was collected when the subjects were admitted for prenatal examination at 20 weeks of pregnancy under fasting conditions, and EPCs were quantified using flow cytometry.

Cytological experiment and flow cytometry

As described previously [30, 31], mononuclear cells (MNCs) were isolated from blood samples using density gradient centrifugation with Biocoll (Biochrom, Berlin, Germany) at 400 × g for 30 min. After washing with phosphate buffered saline (PBS, Biochrom, Berlin, Germany), MNCs were cultured in endothelial basal medium-2 (EBM-2, Clonetics, Cell Systems, St Katharinen, Germany) supplemented with endothelial growth mediator 2 (EGM-2) Sin single Quotes (Clonetics, Inc., San Diego, CA) containing fetal bovine serum (FBS), human vascular endothelial growth factor A (VEGF-A), human fibroblast growth factor B (FGF-B), human epidermal growth factor (EGF), insulin-like growth factor 1 (IGF-1), and ascorbic acid (AA) on human fibronectin (Sigma-Aldrich Chemie, Munich, Germany)-coated dishes. Before analysis, 400,000 MNCs were acquired from each sample using a FACSCalibur flow cytometer (Becton Dickinson). Data analysis was performed with FlowJo (TreeStar). EPCs were defined as CD34 + /CD45dim/KDR + cells.

Uterine artery Doppler US

Uterine artery RI and PI were assessed when the subjects were admitted for prenatal examination at 20 weeks of pregnancy using a Voluson E10 ultrasound system (GE Healthcare, USA) with a 3–5-MHz probe and fully automated spectral analysis software.

Statistical analysis

Statistical analyses were performed using SPSS 21.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 9.5 (GraphPad Software Inc., USA). Normality assessment was conducted via Kolmogorov–Smirnov/Shapiro–Wilk tests. Continuous variables with normal distribution were expressed as mean ± standard deviation (SD) and analyzed with independent sample t tests, while non-normally distributed data were presented as median (minimum, maximum) and evaluated using Mann–Whitney U tests. Categorical variables were described as n (%) and compared through Fisher’s exact/Chi-square tests. Pearson correlation coefficient (r) was used to quantify inter-variable correlations. Receiver-operating characteristic (ROC) curves were plotted to assess the predictive value of EPC count, RI, PI, and their combination for FGR in PE patients. The diagnostic efficacy was evaluated via MedCalc Software (MedCalc Software Ltd., Belgium), and the areas under the ROC curves (AUCs) were compared with the Delong test. Multivariate logistic regression was applied to identify independent risk factors for FGR in PE, and potential factors affecting the reduction of EPCs in PE patients. All tests were two-tailed, with statistical significance set at P < 0.05.

Results

Baseline characteristics

In the PE cohort of this study, the prevalence of FGR was 33.58% (89/265). Comparative analysis of baseline clinical data between the FGR group (N = 89) and the N-FGR group (N = 176) (Table 1) revealed significant disparities: the FGR cohort demonstrated higher proportions of severe PE, EOPE and EFW < the 10th percentile, as well as elevated UA (496.85 ± 25.36 vs. 432.81 ± 20.85 μmol/L) and PRO [3.10 (1.61, 4.73) vs. 2.54 (1.82, 4.60) g/24 h] compared to the N-FGR group. Conversely, the FGR group exhibited reduced PLT [148.43 (100.81, 205.99) vs. 175.75 (103.45, 332.10) × 10⁹/L], FBW (2.71 ± 0.32 vs. 3.28 ± 0.49 kg), lower ALT (28.33 ± 6.50 vs. 34.11 ± 6.47 U/L), and diminished PLGF (50.00 ± 9.05 vs. 96.36 ± 10.10 pg/mL) levels (all comparisons P < 0.01).

Table 1.

Baseline characteristics

	FGR group (N = 89)	N-FGR group (N = 176)	z/t/x2	P
Age (year)	32 (23, 47)	32 (24, 44)	0.558	0.577
DBP (mmHg)	96.90 (78.51, 126.54)	97.02 (68.07, 127.80)	0.977	0.329
SBP (mmHg)	156.44 (120.91, 178.98)	155.26 (125.81, 179.59)	0.654	0.513
BMI (kg/m2)	23.67 ± 2.36	23.35 ± 2.32	1.099	0.273
Gravidity (number of pregnancies)	3 (1, 4)	3 (1, 4)	0.885	0.376
History of hypertension (n, %)			1.478	0.288
Yes	8 (8.99)	9 (5.11)
No	81 (91.01)	167 (94.89)
History of FGR (n, %)			1.985	0.336
Yes	1 (1.12)	0 (0.00)
No	88 (98.88)	176 (100.00)
Family history of FGR (n, %)			3.985	0.112
Yes	2 (2.25)	0 (0.00)
No	87 (97.75)	176 (100.00)
PE severity (n, %)			5.581	0.019
Mild	40 (44.94)	106 (60.23)
Severe	49 (55.06)	70 (39.77)
PE classification (case, %)
EOPE	63 (70.79)	34 (19.32)	67.476	< 0.001
LOPE	26 (29.21)	142 (80.68)
Aspirin intervention history
Yes	12 (13.48)	35 (19.89)	1.661	0.197
No	77 (86.52)	141 (80.11)
EFW < the 10th percentile (case, %)
Yes	84 (94.38)	9 (5.11)	206.791	< 0.001
No	5 (5.62)	167 (94.89)
Gestational age (week)	33 (28, 39)	33 (28, 40)	0.645	0.519
FBW (kg)	2.71 ± 0.32	3.28 ± 0.49	11.523	< 0.001
Daily dairy intake (g)	432.62 ± 56.03	428.45 ± 50.13	0.614	0.540
Laboratory parameters
FBG (μmol/L)	4.18 (3.11, 5.07)	4.09 (3.10, 5.10)	1.355	0.175
TG (μmol/L)	0.84 ± 0.13	0.85 ± 0.11	0.584	0.560
TC (μmol/L)	4.04 ± 0.50	3.95 ± 0.46	1.426	0.155
PLT (× 109/L)	148.43 (100.81, 205.99)	175.75 (103.45, 332.10)	7.041	< 0.001
PT (s)	11.30 ± 2.48	11.15 ± 2.35	0.493	0.622
TT (s)	16.39 ± 3.91	16.21 ± 3.70	0.361	0.718
UA (μmol/L)	496.85 ± 25.36	432.81 ± 20.85	20.567	< 0.001
ALT (U/L)	28.33 ± 6.50	34.11 ± 6.47	6.855	< 0.001
PRO (g/24 h)	3.10 (1.61, 4.73)	2.54 (1.82, 4.60)	7.349	< 0.001
PlGF (pg/mL)	50.00 ± 9.05	96.36 ± 10.10	36.520	< 0.001

EOPE: early-onset preeclampsia; LOPE: late-onset preeclampsia; EFW: estimated fetal weight; FGR: fetal growth restriction; SBP: systolic blood pressure; DBP: diastolic blood pressure; BMI: body mass index; PE: pre-eclampsia; FBW: fetal birth weight; FBG: fasting blood glucose; TG: triglyceride; TC: total cholesterol; PLT: platelet count; PT: prothrombin time; TT: thrombin time; UA: uric acid; ALT: alanine aminotransferase; PRO: proteinuria; PLGF: placental growth factor

Normality was tested using the Kolmogorov–Smirnov test

Continuous variables with normal distribution were expressed as mean ± standard deviation (SD) and analyzed with independent sample t tests, while non-normally distributed data were presented as median (minimum, maximum) and evaluated using Mann–Whitney U tests. Categorical variables were described as n (%) and compared by Fisher’s exact/Chi-square tests

Peripheral blood EPC count reduces while RI and PI increase in PE patients with FGR

We analyzed the peripheral blood EPC status of all subjects using flow cytometry, and measured the uterine RI and uterine PI using a Doppler ultrasound diagnostic instrument. Subsequently, we compared the differences in peripheral blood EPCs, RI, and PI between severe PE (N = 119) and mild PE (N = 146), and between EOPE (N = 97) and LOPE (N = 168) to clarify the relationship between the grading and classification of PE patients and peripheral blood EPCs, RI, and PI. The results showed (Supplementary Fig. 1) that there was no significant difference in peripheral blood EPCs, RI, or PI between severe PE and mild PE patients (all P > 0.05); peripheral blood EPCs in EOPE were significantly lower than those of LOPE, and the PI was significantly higher than that of LOPE (all P < 0.05), while RI showed no significant difference between EOPE and LOPE (P > 0.05). We further investigated the relationship between the occurrence of FGR and peripheral blood EPCs, RI, and PI in PE patients; as shown in Fig. 1, compared to the N-FGR group, the FGR group showed a reduction in peripheral blood EPC count alongside elevated RI and PI (all P < 0.001). These findings indicate concomitant EPC depletion and uteroplacental hemodynamic dysfunction in PE complicated by FGR. In addition, aspirin is associated with the dysfunction of endothelial repair or nitric oxide system [32]. Therefore, we further investigated the effect of aspirin intervention on EPCs in PE patients and found that (Supplementary Fig. 2) there was no significant correlation between aspirin intervention history and EPC count (P = 0.208).

Fig. 1.

Comparison of peripheral blood EPCs, RI, and PI between FGR and N-FGR groups. A Comparison of peripheral blood EPCs; B RI comparison; C PI comparison. FGR group: N = 89; N-FGR group: N = 176. Normality was tested using the Kolmogorov–Smirnov test. Continuous variables with normal distribution were expressed as mean ± standard deviation (SD) and analyzed with independent sample t tests. ***P < 0.001. EPC: endothelial progenitor cell; RI: resistance index; PI: pulsatility index; FGR: fetal growth restriction

Early-onset FGR patients have lower peripheral blood EPC count but higher RI and PI than late-onset FGR patients

FGR is clinically classified into early-onset and late-onset subtypes [33]. Accordingly, the FGR cohort (N = 89) was further stratified into early-onset (Early-FGR, N = 39) and late-onset (Late-FGR, N = 50) groups for comparison of their peripheral blood EPC count, RI and PI. As shown in Fig. 2, the Early-FGR group exhibited lower peripheral blood EPC count but higher uterine artery RI and PI values than those in the Late-FGR group (all P < 0.05), suggesting that peripheral blood EPC count and Doppler indexes (RI/PI) in patients with PE complicated with FGR are closely related to FGR severity.

Fig. 2.

Comparison of peripheral blood EPCs, RI, and PI between early- and late-onset FGR. A Comparison of peripheral blood EPCs; B RI comparison; C PI comparison. Early-FGR group: N = 39; Late-FGR group: N = 50. Normality was tested using the Shapiro–Wilk test. Non-normally distributed continuous variables were presented as median (minimum, maximum) and evaluated using Mann–Whitney U tests. ***P < 0.001; *P < 0.05. EPC: endothelial progenitor cell; RI: resistance index; PI: pulsatility index; FGR: fetal growth restriction

Peripheral blood EPC count is negatively correlated with RI and PI in PE patients with FGR

The correlations of peripheral blood EPC count with RI and PI in patients with PE complicated with FGR were evaluated by Pearson correlation analysis, revealing that peripheral blood EPC count was significantly negatively correlated with RI (r = – 0.607) and PI (r = – 0.539) in PE-FGR patients (both P < 0.001), as displayed in Fig. 3.

Fig. 3.

Correlations of peripheral blood EPCs with RI and PI in PE patients with FGR. A Correlation between peripheral blood EPCs and RI; B correlation between peripheral blood EPCs and PI. The correlations of peripheral blood EPCs with RI and PI in patients with PE complicated with FGR were evaluated by Pearson correlation analysis, with correlation coefficient (r) calculated. EPC: endothelial progenitor cell; RI: resistance index; PI: pulsatility index; PE: pre-eclampsia; FGR: fetal growth restriction

Peripheral blood EPC count combined with RI and PI has higher predictive value for FGR in PE patients

Subsequently, ROC curve analysis was conducted to assess the diagnostic potential of peripheral blood EPC count, RI, PI, and their combined testing for predicting FGR in PE patients. The results (as detailed in Fig. 4, Table 2) showed that the predictive value of combined testing (AUC = 0.989) for FGR in PE patients was significantly higher than that of peripheral blood EPC count alone (AUC = 0.812), RI alone (AUC = 0.808), and PI (AUC = 0.917) alone (all P < 0.001).

Fig. 4.

Diagnostic efficiency for FGR in PE patients. ROC curve analysis was performed to evaluate the predictive value of peripheral blood EPCs, RI, PI, and their combined testing for FGR in PE patients. FGR: fetal growth restriction; PE: pre-eclampsia; ROC: receiver operating characteristic; EPC: endothelial progenitor cell; RI: resistance index; PI: pulsatility index

Table 2.

Diagnostic efficiency for FGR in PE patients

Index	Sensitivity (%)	Specificity (%)	AUC	P	95%CI	Cutoff value	Positive LHR	Negative LHR
PB EPCs	74.16	73.30	0.812	< 0.001	0.760–0.857	3.20	2.778	0.353
RI	62.92	83.52	0.808	< 0.001	0.755–0.854	85.00	3.818	0.444
PI	95.51	80.11	0.917	< 0.001	0.878–0.948	11.90	4.802	0.056
Combined testing	98.88	100.00	0.989	< 0.001	0.968–0.998	–	–	–
PB EPCs vs. Combined testing	P < 0.001
RI vs. Combined testing	P < 0.001
PI vs. Combined testing	P < 0.001

LHR: likelihood ratio

The areas under the POC curves (AUCs) were compared with the Delong test

FGR: fetal growth restriction; PE: pre-eclampsia; ROC: receiver operating characteristic; PB EPC: peripheral blood endothelial progenitor cell; RI: resistance index; PI: pulsatility index

Peripheral blood EPC count and RI are independently correlated with FGR in PE patients

The indicators in Table 1 and Fig. 1 with P < 0.001 were used as independent variables, and whether PE patients had concurrent FGR (FGR = 1, N-FGR = 0) as the dependent variable. Next, following exclusion of independent variable indicators (UA, ALT, PlGF, PI) with multiple linear relationships through multivariate linear analysis (Supplementary Table 1) and adjustment for potential confounders (excluding FBW, gestational age, and EFW < 10th percentile), we conducted a multivariate logistic regression analysis to identify the factors influencing the occurrence of FGR in PE patients. As listed in Table 3, PE classification (OR = 5.501), PRO (OR = 5.501), RI (OR = 1.422), and peripheral blood EPCs (OR = 0.044) were independently correlated with concurrent FGR in PE patients. Subsequently, we further investigated the potential factors associated with the reduction of EPCs. Based on the median score of EPCs in PE patients (3.376), PE patients were classified in to L-EPCs (≤ 3.376, N = 134) and H-EPCs (> 3.376, N = 131). Then we selected age, PLT, ALT, and PIGF as independent variables using the logistic univariate regression analysis. Then we excluded age and ALT indicators using the multivariate linear analysis (Supplementary Table 2). After that, we conducted a logistic multivariate regression analysis with changes in EPCs in PE patients as the dependent variable (L-EPCs = 1, H-EPCs = 0). The results showed that PLT (OR = 0.986) and PlGF (OR = 0.942) were independently correlated with the decrease in EPCs in PE patients (Supplementary Table 3).

Table 3.

Independent correlation of peripheral blood EPCs and RI with FGR in PE Patients

Factor	B	S.E	Wals	P	OR	95%CI
PE classification (EOPE = 1, LOPE = 0)	1.705	0.609	7.843	0.005	5.501	1.668–18.142
PLT (× 109/L)	-0.007	0.006	1.410	0.235	0.993	0.981–1.005
PRO (g/24 h)	1.132	0.458	6.124	0.013	3.103	1.266–7.607
PB EPCs (× 10−1n/mL)	-3.13	0.666	22.105	< 0.001	0.044	0.012–0.161
RI (× 10–2)	0.352	0.067	27.795	< 0.001	1.422	1.248–1.621

PB EPC: peripheral blood endothelial progenitor cell; RI: resistance index; FGR: fetal growth restriction; PE: pre-eclampsia; FBW: fetal birth weight; PRO: proteinuria; the bolded content has significant meaning

Discussion

This study presents novel insights into the synergistic predictive value of integrating cellular biomarkers (peripheral blood EPCs) and hemodynamic parameters (uterine artery RI/PI) for identifying FGR in women with pregnancies complicated by PE. Our findings advance the current understanding of PE-FGR pathophysiology by establishing a quantitative link between endothelial repair capacity and placental perfusion efficiency, thereby bridging critical gaps in risk stratification strategies.

In early pregnancy, trophoblasts invade and remodel the uterine spiral artery, ensuring placenta engineering and providing nutrients for the fetus [34]. However, in PE pregnant women, trophoblast invasion is insufficient, which leads to the disorder of uterine spiral artery remodeling and abnormal placental formation, laying a hidden danger for the occurrence and development of FGR [7]. RI and PI, as predictive tools for artery resistance, have been shown in a retrospective cohort study to be abnormal among fetuses with severe early-onset FGR [35]. A prospective study has also demonstrated that RI and PI are higher in pregnant woman with hypertensive disorders, and suggest that PI has the highest sensitivity for predicting FGR [36], supporting that RI and PI assessment provides prediction for pregnancy-related disorders. In addition, when the placental blood perfusion is insufficient, placental ischemia and hypoxia will promote the release of toxic substances into the blood, further causing vascular endothelial injury, which may also affect the distribution of the EPC pool and alter the number of circulating EPCs in pregnancy-related diseases [21]. Previous research has revealed that circulating EPC number in the umbilical cord blood significantly decreased in the PE group compared with the normal group, and circulating fetal EPCs in FGR are fewer in number [37]. All these findings are in good accordance with our results showing reduced peripheral blood EPC count while increased RI and PI in PE patients with FGR, confirming their predictive roles. However, a study by Ranjan Monga et al. has reported that exposure to PE and intrauterine growth restriction (IUGR) is associated with obvious changes in EPC population [38]. Therefore, whether the reduction in EPCs is a consequence of PE or whether pregnant women with low EPCs are at a higher risk of developing PE has not yet been determined. Moreover, it has been reported that aspirin inhibits the production of thromboxane A₂, promotes the release of nitric oxide and prostacyclin, thereby improving placental blood flow and endothelial function, and a healthy microenvironment may provide favorable conditions for the recruitment, homing, and differentiation of EPCs [39]. The results of this study showed no significant correlation between aspirin intervention history and EPC count (P > 0.05). This might be due to the limited number of patients with a history of aspirin intervention in this study. Therefore, in future research, we need to expand the sample size of PE patients with a history of aspirin intervention and conduct prospective studies on dynamic monitoring of the number and function of EPCs to further clarify the direct effect of aspirin on EPCs.

The principal innovation of this work lies in its dual-modal approach, which transcends conventional single-biomarker investigations. By demonstrating that EPC depletion synergizes with elevated uterine artery resistance to predict FGR, our model achieved diagnostic accuracy surpassing EPC count alone or isolated Doppler indices. This integrative strategy aligns with the multifactorial nature of PE-FGR pathogenesis, where endothelial dysfunction and impaired placentation coexist [40]. Notably, the inverse correlation between EPCs and RI/PI suggests a mechanistic interplay: spiral artery remodeling defects may amplify uteroplacental resistance, thus diminishing endothelial repair capacity [7]. This bidirectional relationship positions EPCs not merely as biomarkers but as active participants in disease progression, offering a novel therapeutic target for restoring vascular homeostasis.

Furthermore, our stratification of FGR into early- and late-onset subtypes revealed gradient reductions in EPCs and incremental elevations in RI/PI, underscoring the clinical relevance of these biomarkers in distinguishing FGR severity. EPCs have also been used as markers for the severity of multiple disorders, such as congenital heart disease, transient ischemic attack, and hypertrophic cardiomyopathy [41–43]. The identification of independent correlations between PLT, PlGF, and reduced EPC count through multivariate analysis provides a framework for personalized risk assessment, enabling clinicians to prioritize high-risk pregnancies for intensified monitoring or prophylactic interventions.

Nevertheless, several limitations warrant consideration. First, this study has a small sample size and single-center design. In addition, research has shown that statins can increase circulating EPC numbers, enhance their mobilization, migratory capacity, and survival [44]. It has also been suggested that pregnant women with specific FGR may benefit from statin therapy, although the feasibility of statin use during pregnancy remains highly debated [45]. However, our study did not include patients with a history of statin use, which needs further analysis in subsequent research. Furthermore, the study population comprised women aged 23–47 years, encompassing both women of appropriate age and advanced maternal age. The predictive value of combining peripheral blood EPC count with RI and PI for FGR in older PE patients was not analyzed separately. In addition, the cohort analyzed in this study (patients with PE) represents a highly selected high-risk population. In addition, in the PE cohort of this study, the prevalence rate of FGR was 33.58% (89/265), which was 10–15% higher than that of the general population in China (10.1016/j.genrep.2021.101170.). The above two reasons may lead to overestimation of the discriminability of peripheral blood EPCs combined with RI and PI detection in predicting PE complicated with FGR, and the generalization is questionable. In the future, further external prospective validation will be conducted to clarify the clinical applicability of peripheral blood EPCs combined with RI and PI detection for predicting PE complicated with FGR. Furthermore, as a retrospective analysis, this study cannot fully establish the causal relationship between EPC deficiency and FGR. Although the causal relationship between EPC deficiency and FGR is uncertain, a previous study has shown that EPCs affect fetal nutrition supply by promoting placental angiogenesis [46]. Meanwhile, this study confirms that EPCs are closely related to the severity of FGR (Fig. 2), suggesting that the deficiency of EPCs may be involved in the pathological mechanism of FGR. However, some studies have also shown that the hypoxic microenvironment of the placenta induced by FGR leads to an impaired EPC function, forming a vicious cycle [15, 20]. Therefore, in future research, it is necessary to further validate through prospective cohorts or intervention experiments, and to differentiate causal time series by dynamically monitoring changes in EPCs. Regardless of the causal relationship, EPC detection plays a certain auxiliary role in early identification of FGR. In summary, this study reveals that EPC deficiency is associated with concurrent FGR in PE, which may assist in early identification of FGR, but the specific mechanism needs to be further explored. Moreover, flow cytometry may vary from laboratory to laboratory and its applicability may be limited. Therefore, in future research, more popular and convenient detection methods such as RT-PCR for the detection of endothelial specific genes (such as endothelial nitric oxide synthase-EPC functional core markers) can be further used to optimize the detection of EPCs.

Therefore, future research should further enlarge the sample size, conduct multi-center study, and validate the effects of statin use on EPCs. Moreover, a comprehensive analysis on the safety of statins during pregnancy will be carried out, so as to provide new guidance for FGR assessment and treatment. Simultaneously, deeper exploration is needed to investigate the impact of FGR on older PE patients complicated with FGR.

Conclusion

In conclusion, this study pioneers an integrative biomarker model that synergizes cellular and hemodynamic metrics to predict PE-related FGR with high precision. By illuminating the interplay between endothelial repair mechanisms and placental perfusion, our findings redefine risk stratification paradigms and pave the way for targeted therapeutic strategies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

QYH is the guarantors of integrity of the entire study and contributed to the definition of intellectual content, manuscript review; YSL contributed to the study concepts, data acquisition, statistical analysis, manuscript preparation; YSL, XFZ contributed to the study design, literature research; FHZ, XFZ contributed to the data analysis, clinical studies, manuscript editing; All authors read and approved the final manuscript.

Funding

The authors received no funding from an external source.

Availability of data and materials

All data generated or analyzed during this study are included in this article. Further enquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was carried out complying with the ethical principles of the World Medical Association Declaration of Helsinki, relevant specifications and regulations for clinical research, as well as the guidelines from the Enhancing the QUAlity and Transparency Of Health Research (EQUATOR) network. The study protocol received approval from the Ethics Committee of The Affiliated Hospital of Xuzhou Medical University (Ethics approval number: XYFY2023-KL093-02).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.