Retrospective analysis of placental pathology results from pregnant women who received regular prenatal care and delivered at The First Affiliated Hospital of Harbin Medical University between September 2021 and February 2024. Based on placental pathology, 129 cases were classified into the MVM group with 57 cases and the non-MVM group with 72 cases. Maternal characteristics (age, gravidity, BMI, gestational age) and pregnancy-related complications and comorbidities (hypertensive disorders, thyroid dysfunction, gestational diabetes, placental implantation diseases) were analyzed. All placental samples were obtained with informed consent, and the study was approved by the local Ethics Committee (Ethics Approval No. 2024381), in compliance with the Declaration of Helsinki.

This study defined the following inclusion criteria for placental pathology results collection: 1) Placenta gross appearance without significant abnormalities; 2) Complete prenatal and delivery records of the mother; 3) Informed consent was obtained from both the mother and her family.

The exclusion criteria for placental pathology results collection were as follow: 1) presence of severe liver dysfunction, renal impairment, hematological diseases, congenital genetic disorders, or autoimmune disorders; 2) fetal developmental abnormalities; 3) multiple pregnancies, stillbirths, or fetal demise; 4) abnormal placental appearance; 5) incomplete clinical data.

Placental tissue from the central region was collected during cesarean delivery under sterile conditions. Each sample weighed approximately 100 mg and was rinsed with non-enzymatic physiological saline (Solarbio) to remove any residual blood. The tissue was then blotted dry and immersed in a preservation solution (Beyotime) at a ratio of 1:10 (tissue to solution). The samples were stored overnight at −4 °C and subsequently transferred to a −80 °C freezer for long-term preservation. Additionally, a separate portion of the placental tissue was fixed in 4% paraformaldehyde solution for 48 h, followed by routine paraffin embedding for subsequent immunohistochemical experiments.

Peripheral blood samples (3 mL) were collected from fasting mothers prior to the cesarean delivery. The blood was preserved in collection tubes containing separation gel and clot activator, then refrigerated at 4 °C overnight. The samples were centrifuged at 1,000 g for 15 min at 4 °C. The supernatant was collected, aliquoted, and stored at −80 °C for analysis to be conducted within 1 month.

The inclusion criteria for clinical samples were as follows: 1) gestational age of ≥34 weeks; 2) singleton live births; 3) absence of fetal developmental abnormalities; 4) healthy mothers without gestational comorbidities or complications. Participants were grouped by maternal age: 1) the advanced maternal age (AMA) group, comprising women aged between 35 and 50 years; 2) the normal control (NC) group, comprising women aged between 20 and 34 years. The exclusion criteria were as same as those in placental pathology results collection.

Prior to the experiment, samples were retrieved from the −80 °C freezer. A total of 25 μL of each sample was mixed with 225 μL of diluent in sterile EP tubes, resulting in a 10-fold dilution. After incubating at room temperature for 1h, required assay plates were removed from their aluminum foil packaging. Any remaining plates were resealed in self-sealing bags and stored at 4 °C. Standard, blank, sample, and control wells for the sample diluent were set up, with each condition prepared in triplicate. Following the procedure outlined in the ELISA kit manual (Mlbio), samples were added in the designated order. After adding the stop solution, optical density (OD) values were measured within 15 min at 450 nm using a microplate reader.

To assess oxidative stress levels in placental tissue, 100 mg of placental tissue was minced and then mixed with physiological saline at a volume ratio of 1:9. The tissue was then homogenized using a tissue grinder. The mixture was centrifuged at 3,000 rpm for 5 min to obtain the supernatant, which served as a 10% tissue homogenate. Protein concentration in the homogenate was quantified using the BCA Protein Assay Kit (Beyotime), following the manufacturer’s instructions. For malondialdehyde (MDA) quantification, samples were processed according to the MDA Assay Kit protocol (NJJCBIO). The samples were incubated in a boiling water for 40 min, then cooled to room temperature under flowing water and centrifuged at 100 rpm for 10 min. The supernatant was collected, and absorbance was measured at 530–540 nm using a microplate reader to calculate the MDA content. In addition, another aliquot of the 10% placental tissue homogenate was processed using the Superoxide Dismutase (SOD) Assay Kit (NJJCBIO). Samples were incubated at 37 °C for 40 min, followed by the addition of a color reagent, and absorbance was measured at 550 nm.

Tissue sections (4 μm) were prepared using an automatic microtome and placed in an incubator for 30 min to facilitate dehydration and rehydration. Antigen retrieval was performed via microwave for 10 min, followed by cooling of the slides to room temperature and washing three times with PBST. Nonspecific binding sites of sections were blocked with 10% goat serum (1 h) and incubated with the primary antibody overnight at 4 °C. After PBST washes, the secondary antibody was applied (1 h), followed by DAB staining for 5 min, and hematoxylin counterstaining for 30 s. Slides were examined using an optical microscope (×200), and SIRT3-positive cells in interstitial regions were counted in each field. Immunohistochemical (IHC) staining was evaluated by two pathologists using a scoring system: no staining (“none”) was assigned a score of 0 for both intensity and distribution; weak staining (“weak”) received a score of 1 for intensity and a distribution of 1%–25%; moderate staining (“moderate”) received a score of 2 for intensity and a distribution of 26%–50%; strong staining (“strong”) received a score of 3 for intensity and a distribution of 51%–75%; and very strong staining (“very strong”) was assigned a score of 4 for intensity with a distribution of 75%–100%. The cumulative score was calculated as the product of staining intensity and distribution.

Placental tissue (100 mg) was ground in liquid nitrogen, lysed in 1 mL RIPA buffer containing PMSF, and sonicated for 1 min. After incubation on ice for 30 min with mixing, the lysate was centrifuged (12,000 rpm, 25 min, 4 °C), and the supernatant was collected for protein quantification using a BCA protein assay kit. Proteins were mixed with a 5× loading buffer (4:1), denatured at 100 °C, aliquoted (200 μL), and stored at −20 °C. For SDS-PAGE, 8 μL of protein sample and 3 μL of molecular weight marker were loaded per lane and separated at 200 V for 25 min). Proteins were transferred onto a PVDF membrane (300 mA, 25 min, ice-cooled). The membrane was blocked (5% BSA, 1 h), incubated overnight at 4 °C with primary antibodies (anti-SIRT3 and anti-β-actin), and subsequently secondary antibodies at room temperature for 1 h. After washing with TBST, protein bands were visualized using enhanced chemiluminescence (ECL) and imaged on a Tanon-520 Gel Imaging System.

The primer sequences for human SIRT3 were as follows:

Forward (F): 5′-GAA​GTG​GAG​GCA​GCA​GTG​ACA​AG-3′,

Gene expression in placental tissues was assessed using qRT-PCR, with synthesized cDNA serving as a template. The reaction was conducted using gene-specific primers and a SYBR^® Green PCR kit (EZBioscience) on a StepOnePlus Real-Time PCR System (Bio-Rad Biosystems). The thermal cycling conditions included an initial denaturation at 95 °C for 15 min, followed by 40 amplification cycles of 94 °C for 15 s, 60 °C for 30 s, and 72 °C for 30 s. GAPDH was employed as an internal control for normalization. Relative mRNA expression levels were then calculated using the 2^−ΔΔCT method.

For the sequencing analysis, a total of 6 samples were randomly chosen from each group. Ensuring each sample integrity throughout the processing. RNA-seq libraries were prepared using the Illumina TruSeq™ RNA Sample Prep Kit, and sequenced on the Illumina Novaseq 6000 platform for high-throughput analysis. Gene expression levels in placental tissues were quantified using FPKM. Principal Component Analysis (PCA) was performed using the FactoMineR v2.6 package in R to assess sample clustering and identify potential outliers. Differential gene expression analysis between the AMA and NC samples was carried out using the DESeq2 v1.34 package. DEGs were defined as those with an adjusted P-value <0.05 and an absolute Fold Change value ≥2. To elucidate the biological function of DEGs, GO and KEGG pathway enrichment analyses were performed utilizing the clusterProfiler v4.2.0 package. Gene Set Enrichment Analysis (GSEA) was utilized to evaluate enrichment results with a normalized enrichment score (|NES|) > 1, a false discovery rate (FDR) q-value <25%, and an adjusted P-value <0.05, indicating statistical significance. Bayesian network analysis of enriched pathways was conducted using CBNplot v1.2.1 to infer regulatory interactions.

The WGCNA package was employed to contruct a gene co-expression network. Initially, sample clustering was performed based on clinical data, with an outlier threshold of 60 applied to exclude aberrant samples. The Pearson correlation matrix was then transformed into a weighted adjacency matrix through the power function f (X) = X^β, where the optimal soft thresholding power (β) was identified. A power of 5 was selected, and the adjacency matrix was converted into a topological overlap matrix (TOM). The TOM facilitated clustering based on topological overlap (TO) dissimilarity (1 - TOM). Genes were clustered based on TOM using average linkage hierarchical clustering. Gene modules were defined by grouping genes with similar expression patterns, enforcing a minimum module size of 30. Modules with a dissimilarity threshold below 0.25 were merged. Key hub genes within co-expressed modules were identified based on gene significance (GS > 0.2, P-value <0.05) and module membership (MM > 0.8, P-value <0.05). Modules exhibiting the highest correlation with gestational age were selected for further analysis. The genes within these significant modules were intersected with DEGs to filter overlapping candidates. GO and KEGG pathway enrichment analyses of key genes were conducted using the clusterProfiler package. Statistical significance was identified as an adjusted P-value <0.05. The filtered key genes were further analyzed using the STRING database for protein-protein interaction (PPI) network construction. Cytoscape v3.10.1 was used to compute topological parameters of the network. The cytoHubba plugin was utilized to identify core genes associated with gestational age, refining the selection of crucial nodes within the PPI network.

RNA-seq data (GSE133592) from trophoblast cells and clinical records of 32 pregnant women were obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo), categorized into young (10 cases), intermediate (16 cases), and AMA (6 cases) groups.

Data analysis was conducted using SPSS (IBM SPSS Statistics 26.0) and R (version 4.3.0) software. Continuous variables were presented as mean ± standard deviation (SD) or median with interquartile ranges (IQRs), depending on their distribution. Categorical variables were expressed as counts and percentages. For group comparisons, one-way analysis of variance (ANOVA) was applied for normally distributed variables with homogeneous variance. For continuous variables, Student’s t-test or the Mann–Whitney U test was applied, depending on the data distribution. Logistic regression analysis was employed for the statistical examination of outcomes related to factors and categorical variables. The Fisher exact test or chi-square test was used to compare categorical variables between groups, as appropriate. Statistical graphs were generated using Graphpad 9.5 and R 4.3.0 software.

The clinical characteristics of the study cohort are presented in Table 1. Comparing with the Non-MVM group, there were significant increases in age (p < 0.05), the proportion of AMA (p < 0.05), body mass index (BMI) (p < 0.05), and the incidence of pregnancy-induced hypertension (p < 0.05) in the MVM group. No statistically significant differences were observed in other indicators, including gestational diabetes, thyroid dysfunction, and placental implantation diseases. Logistic regression analysis was performed to assess the correlation between AMA, BMI, pregnancy-induced hypertension, and MVM, with detailed results in Table 2. The analysis identified AMA as an independent risk factor for MVM (OR = 3.022, 95% CI 1.337–6.832). Additionally, Elevated BMI (OR = 1.145, 95% CI 1.037–1.246) and pregnancy-induced hypertension (OR = 2.671, 95% CI 1.065–6.701) were identified strong independent associations with MVM. Based on retrospective analysis of placental pathology results, AMA was identified as an independent risk factor for maternal vascular malperfusion (MVM). To further investigate the underlying molecular mechanisms, this study performed transcriptome sequencing (RNA-seq) on placental tissues from AMA group (≥35 years) and normal control group (NC group, 20–34 years), aiming to delineate differential gene expression profiles between the cohorts.

Principal component analysis (PCA) demonstrated distinct inter-group separation in transcriptional profiles, whereas samples within each group exhibited cohesive clustering patterns (Figure 1A). A total of 731 significantly differentially expressed genes (DEGs) were identified from the AMA group compared to the NC group (|log2FC| > 1, adjusted p < 0.05), with 445 genes upregulation and 286 genes downregulation in the AMA group. DEGs were ranked by log2 (fold change) magnitude (Figure 1B), with the top most significantly 50 upregulated and downregulated genes visualized through heatmap analysis (Figure 1C).

Systematic functional profiling of DEGs was performed through bioinformatics framework, incorporating GO term enrichment, KEGG pathway mapping, and GSEA. GO enrichment analysis of DEGs across different maternal age groups was performed, subsequently identifying the top-ranked entries within the three standard GO categories: biological processes (BP), cellular components (CC), and molecular functions (MF) (Figure 2A). The results from the GO analysis indicated that DEGs are primarily enriched in processes including energy metabolism, oxidative stress, angiogenesis, and NAD(P)H metabolism. Among these processes, the pathways associated with oxidative stress and angiogenesis are identified as critical enrichment pathways, suggesting their involvement in placental development in pregnancies affected by AMA (Figures 2B–D). The circular plot was employed to visually emphasize key entries associated with the three major GO categories, while providing a comprehensive overview of the functional enrichments corresponding to the differentially expressed genes (DEGs) identified in this study (Figures 2E–G). The KEGG pathways analysis categorized biological metabolic pathways into six distinct categories: Metabolism, Genetic Information Processing, Environmental Information Processing, Cellular Processes, Organismal Systems, and Human Diseases. The top 20 representative KEGG enrichment results are illustrated in Figure 3A, with the regulatory relationships among pathways displayed in Figure 3B. The upregulated pathways include oxidative phosphorylation and reactive oxygen species (ROS) production, while the downregulated pathways encompass resistance to EGFR tyrosine kinase inhibitors, PI3K-AKT signaling, and the VEGF signaling pathway. During placental angiogenesis and metabolic processes, ROS signaling pathway and the VEGF signaling pathway have been shown to interact, as illustrated in Figure 3C. Additionally, GSEA demonstrated significant enrichment of DEGs in pathways related to oxidative phosphorylation, oxidative stress, and NAD(P)H metabolism. In contrast, downregulation was predominantly observed in pathways inflammasome formation and responses to VEGFA, as shown in Figure 3D.

To identify the most relevant gene modules associated with placental tissue samples of AMA, WGCNA was performed on the GSE133592 dataset for gene clustering (Figure 4A). A co-expression network was constructed based on gene expression profiles. Scale-free genes from this network were obtained using β (soft threshold parameter) = 5, with R ² = 0.85 through transformation of the weighted adjacency matrix of the TOM (Figure 4B). Hierarchical clustering dendrogram analysis of the co-expression network revealed 11 distinct co-expressed gene modules, as illustrated by color-coded branches in Figure 4C. The gray module contained unassigned genes, while the black module (51 genes) demonstrated the strongest positive correlation (correlation coefficient = 0.47) with AMA clinical phenotypes (Figure 4D). GO enrichment analysis of the black module revealed significant enrichment of age-associated biological processes, including hypoxia response, oxidative stress response, ketone body metabolism, and endothelial cell proliferation pathways (Figure 4E). PPI network analysis conducted via the STRING database (Figure 4F) identified top hub genes using CytoHubba plugin and Maximum Neighborhood Component (MNC). Ten high-scoring genes were prioritized, with six core hub genes (SIRT3, TLR6, AOX1, ARG1, CRYAB, and HGF) exhibiting the highest connectivity. Among these, SIRT3 demonstrated the maximal degree score, indicating its central regulatory role within the network (Figure 4G).

The baseline characteristics of the included individuals used in the placental transcriptome sequencing and verification experiments are detailed in Supplementary Table S1. There were no statistically significant differences in the basic information of the patients between the two groups, except for the age. The serum ELISA results indicated significant differences in the levels of placental growth factor (PLGF) and soluble fms-like tyrosine kinase-1 (sFlt-1) between the AMA group (n = 15) and NC group (n = 15). The PLGF level in the high-risk pregnancy group was lower than that in the NC group (p = 0.005), as shown in Figure 5A. Conversely, the sFlt-1 level in the AMA group was significantly higher compared to the NC group. Additionally, the PLGF/sFlt-1 ratio of the AMA group was higher than that of the NC group (p = 0.000). Correlation analyses revealed a positive correlation was observed between age and serum sFlt-1 levels (r = 0.5263, p = 0.0028). Furthermore, the result demonstrated strong inverse correlations between maternal age and both serum PLGF levels (r = −0.5952, p = 0.0005) and the PLGF/sFlt-1 ratio (r = −0.7503, p < 0.0001).

As shown in Figure 5B, by Immunohistochemical staining, PLGF expression in the placenta was decreased while sFlt-1 expression was increased in the AMA group (n = 15) compared to the NC group (n = 15), significantly (p < 0.005). Furthermore, HE staining revealed a reduction in the decidual vascular space in the AMA group (n = 30) (p < 0.005). The expression levels of PLGF and sFlt-1, as determined by Western blot, showed the same trend as immunohistochemical staining (Figure 5C). These results indicate that angiogenesis is abnormally impaired in placentas of AMA, with concomitant dysregulated remodeling of spiral arteries.

In previous results, SIRT3 as the predominant hub gene, is involved in regulating oxidative stress response within mitochondrial function. Compared to the NC group (n = 6), both protein and mRNA levels of SIRT3 in the placenta of AMA pregnancies were significantly decreased (p < 0.05, Figures 5D–F). Furthermore, the levels of MDA, a marker of oxidative stress, were found to be elevated in the placentas of AMA pregnancies (p < 0.05, Figure 5G), while the activity of SOD, an important antioxidant enzyme, was significantly lower compared to the NC group (n = 30) (p < 0.05). SIRT3 downregulation in the placenta of AMA pregnancies suggests a potential mechanism of oxidative stress damage and poor placental perfusion. The decreased antioxidant capacity due to low SIRT3 levels may contribute to oxidative stress-related placental injuries, which are associated with adverse pregnancy outcomes in AMA pregnancies.