A retrospective study was conducted involving pregnant women who received prenatal examinations and gave birth in the Department of Obstetrics, Maternity and Child Health Care of Guangxi Zhuang Autonomous Region, from January 1, 2016, to September 30, 2019. This study was designed and reported in strict accordance with the strengthening the reporting of observational studies in epidemiology (STROBE) checklist. Utilizing the hospital’s electronic medical record system, two participant groups were established: the non-diabetic macrosomia group (n = 170), which included pregnant women whose newborns had a birth weight $>$4000 g with no abnormal blood glucose levels during pregnancy; and the normal group (n = 186), which served as controls with newborns of normal weight, matched for gestational age. During the third stage of labor, 10 mL of umbilical venous blood samples were collected from 16 pregnant women (8 in the non-diabetic macrosomia group and 8 in the normal group). To minimize the interference of confounding factors, the participants were strictly matched for age and BMI (with a difference of 1) in addition to having comparable gestational age. All subjects underwent an oral glucose tolerance test (OGTT) at 24–28 weeks of gestation in our hospital. According to the diabetes diagnostic criteria recommended by the American Diabetes Association (ADA) guidelines, pregnant women with diabetes (including pre-pregnancy diabetes and gestational diabetes) were excluded from the study. Furthermore, pregnant women with any diseases that could affect neonatal metabolism and growth and development were excluded, such as metabolic disorders (e.g., thyroid dysfunction during pregnancy and Cushing’s syndrome), hypertensive disorders in pregnancy, and autoimmune diseases. Baseline data, laboratory test data, delivery data, and neonatal clinical data of all subjects were collected. The study protocol was approved by the Medical Ethics Committee of Guangxi Medical University (approval number: 2019-SB-067), and written informed consent was obtained from the participants.

During the third stage of labor, 10 mL of umbilical venous blood was collected from non-diabetic pregnant women and transferred into dipotassium ethylenediaminetetraacetic acid (K₂EDTA) vacuum blood collection tubes (BH1359, Bioroyee, Beijing, China), each. All blood samples were centrifuged at 1000 $\times$g at 4 °C for 15 min. Subsequently, at least 5 mL of plasma was isolated from each sample, aliquoted into multiple 1.5 mL Eppendorf tubes (0030125215, Eppendorf, Hamburg, Germany), and immediately stored at –80 °C for subsequent experimental analysis.

To eliminate cellular debris and large vesicles, 2 mL of plasma was first centrifuged at 2000 $\times$g for 10 min at 4 °C, followed by a second centrifugation at 10,000 $\times$g for 60 min. The resulting supernatant was then ultracentrifuged at 120,000 $\times$g for 2 h at 4 °C to enrich exosomes. The obtained pellet was gently rinsed with phosphate-buffered saline (PBS; 6-2554, Tianjingsha Gene Technology Co., Ltd., Beijing, China) and ultracentrifuged again at 120,000 $\times$g for 2 h at 4 °C to purify exosomes. Finally, the supernatant was carefully removed, and the exosome pellet was gently resuspended in 200 µL of PBS to obtain the exosome suspension, which was subsequently used for downstream experimental analyses.

The microstructure of the exosome was observed using a JEM-1200EX transmission electron microscope (TEM) (JEOL Ltd., Tokyo, Japan). The observation was carried out using an accelerating voltage ranging from 40–120 kV. Image data were acquired and exported. The particle size distribution and concentration of exosomes were measured via nanoparticle tracking analysis (NTA) (Nanosight, Wiltshire, UK). Each sample was analyzed in triplicate, and the mean particle size, modal particle size, and particle size distribution were calculated using NTA software (v3.2, Malvern Panalytical Ltd., Malvern, UK).

PBS (100 µL) was added to the resuspended exosomes, which were then incubated with 10 µL Fluorescein Isothiocyanate (FITC)-conjugated anti-CD63 antibody (557288, BD Biosciences, San Jose, CA, USA) or FITC-conjugated anti-CD81 antibody (551108, BD Biosciences, San Jose, CA, USA). Flow cytometry analysis was performed using an Accuri C6 flow cytometer (AccuriC6, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) according to the manufacturer’s instructions.

Plasma samples (3 mL) were extracted from umbilical venous blood using the Qiagen exoRNeasy/Plasma Midi Kit (77044, Qiagen, Hilden, North Rhine-Westphalia, Germany) following the manufacturer’s instructions. RNA sequencing was conducted by LC-Bio Technologies Co., Ltd. (Hangzhou, Zhejiang, China) on the Illumina HiSeq 2500 platform. Total RNA extracted from the samples was utilized to construct strand-specific whole-transcriptome libraries with the SMARTer® Stranded Total RNA-Seq Kit v2 - Pico Input Mammalian (Takara Bio USA, San Jose, CA, USA). After verifying the sequencing libraries met the required standards through quality control, paired-end 150-bp (PE150) sequencing was performed. Raw sequencing data were initially processed using Cutadapt (v3.4, Marcel Martin, Stockholm, Sweden, https://cutadapt.readthedocs.io/) to remove adapter sequences introduced during library preparation and filter out low-quality reads, obtaining high-quality valid data. These clean reads were then aligned to the human reference genome GRCh38 using Hisat2 (v2.2.1, Johns Hopkins University, Baltimore, MD, USA, https://daehwankimlab.github.io/hisat2/). Subsequently, transcript-level expression profiles for mRNA and lncRNA were generated based on the corresponding genome annotation file (general transfer format, GTF).

The Batch effect was eliminated using the R package limma (https://bioconductor.org/packages/limma/). Co-expression networks of mRNA and lncRNA were constructed with the WGCNA package (v1.47, https://cran.r-project.org/web/packages/WGCNA/) in R (v4.4.3, Foundation for Statistical Computing, Vienna, Austria). The WGCNA analysis workflow strictly followed the methodology described by Huang et al. [24], which encompasses critical steps for the selection of soft-thresholding parameters, module identification, and screening of key genes. Specifically, the pickSoftThreshold function was employed to determine the appropriate soft threshold parameter ($\beta$) for approximating the scale-free network topology. The selected $\beta$ was required to ensure that the scale-free topology index R² was $\ge$0.8 with a relatively high average connectivity. Based on this soft threshold, a weighted adjacency matrix was constructed, and modules were subsequently identified. Module eigengenes (MEs) were calculated for each module, and their correlations with clinical features were evaluated to identify gene modules significantly associated with clinical phenotypes. Gene modules showing significant correlations with clinical features (r $\ge$ 0.3, p 0.05) were selected for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis using the clusterProfiler R package. Hub genes were determined by calculating Gene Significance (GS) and Module Membership (MM). GS reflects the correlation between a gene’s expression level and a specific clinical trait or phenotype, while MM quantifies the importance of genes within a module.

Statistical analyses were performed using SPSS software (v25.0, IBM Corp., Armonk, NY, USA). The Shapiro-Wilk test was employed to evaluate the normality of continuous variables. Continuous variables with a normal distribution were presented as mean $\pm$ standard deviation (SD), and between-group comparisons were analyzed using the independent-samples t-test. Those deviating from a normal distribution were presented as median and interquartile range (IQR; 25th–75th percentiles, Q1–Q3), with group comparisons conducted via the Mann-Whitney U test. Categorical variables were described as counts and percentages [n (%)], and between-group comparisons were performed using the Chi-square test or Fisher’s Exact test. Graphs were generated using R software and Adobe Illustrator (v28.0, Adobe Inc., Mountain View, CA, USA). All statistical tests were two-sided, and p 0.05 was considered statistically significant.

Maternal age (p 0.001), diastolic blood pressure (DBP) (p = 0.012), primiparity (p = 0.008), and mode of delivery (p 0.001) had significant impacts on the occurrence of macrosomia (Table 1). Compared to the control group, the pre-pregnancy BMI (22.91 $\pm$ 3.57 vs. 20.86 $\pm$ 2.94, p 0.001) and overall weight gain (15.36 $\pm$ 4.05 vs. 13.70 $\pm$ 3.10, p 0.001) in non-diabetic mothers with macrosomic infants were higher. Weight gain was also significantly higher during the second [9.00 (7.50–11.00) vs. 8.50 (7.00–10.00), p = 0.006] and last trimesters [6.00 (4.00–7.80) vs. 5.35 (4.00–6.50), p = 0.009]. Notably, significant differences were observed in maternal blood lipid profiles [including high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides (TG)] and alkaline phosphatase (ALP) levels in the last trimester between the two groups (all p 0.05). Additionally, significant differences in clinical characteristics related to the placenta were identified between the macrosomia and control groups; the macrosomia group also had a higher proportion of male infants (Table 2).

This study utilized logistic regression analysis to identify factors independently associated with macrosomia in non-diabetic mothers. The chosen method of delivery was excluded from the analysis as it was based on fetal weight estimation. Although DBP showed a statistically significant difference in univariate analysis (p 0.05), its values remained within the range of normal physiological fluctuations. Moreover, there was a lack of sufficient evidence to support its role as an independent risk factor for macrosomia in non-diabetic pregnant women. Therefore, DBP was excluded from the multivariate regression analysis conducted in this study. After adjusting for the confounding effect of maternal age [25], the results showed that non-diabetic pregnant women with a higher pre-pregnancy BMI (odds ratio [OR] = 1.235, 95% confidence interval [CI]: 1.124–1.358, p 0.001), excessive overall weight gain (OR = 1.330, 95% CI: 1.177–1.503, p 0.001) and weight gain during the second (OR = 1.218, 95% CI: 1.004–1.477, p = 0.046) and third trimesters (OR = 1.189, 95% CI: 1.008–1.402, p = 0.039), higher LDL levels (OR = 1.827, 95% CI: 1.298–2.572, p = 0.001), and elevated ALP levels in the third trimester (OR = 1.007, 95% CI: 1.003–1.012, p = 0.002) were more likely to deliver macrosomic infants (Table 3). However, primiparas (OR = 0.524, 95% CI: 0.287–0.954, p = 0.035) and elevated HDL levels (OR = 0.180, 95% CI: 0.080–0.407, p 0.001) in non-diabetic mothers had a lower risk of macrosomic births. No multicollinearity was observed among these predictors.

Exosomes were successfully extracted from the umbilical venous blood. First, we analyzed the size and shape of the exosomes isolated from plasma (Fig. 1A). Next, we recorded the median size of the vesicles, the majority of which were between 60 nm and 100 nm (Fig. 1B). Flow cytometry analysis confirmed that the isolated vesicles were exosomes, as indicated by the specific biomarkers CD63 and CD81 (Fig. 1C).

Fig. 1.

Characterization of isolated exosomes. A representative TEM image of isolated exosomes. (A) Exosomes ranged in size from 20–80 nm (scale bar: 200 nm). (B) Exosomes isolated from the plasma derived from the cord blood of pregnant women with macrosomic neonates were evaluated by NanoSight. The plot shows a broad size distribution. (C) Flow cytometry of exosomes expressing CD63 and CD81. TEM, transmission electron microscope; FSC-A, Forward Scatter-Area; FL1-A, FLuorescence 1-Area.

We eliminated batch effects in two batches of the sequenced mRNA and lncRNA data using the R limma package. The datasets, containing 19,349 mRNAs and 9625 lncRNAs, were selected for analysis after removing the missing values of gene expression from the raw data, and the WGCNA package was filtered. Firstly, Soft threshold power values were selected in the mRNA and lncRNA groups. When $\beta$ = 20, the R² scale was set at 0.9360 to obtain a higher average connectivity degree in the mRNA group (Fig. 2A). Conversely, when $\beta$ = 30, the R² scale was set at 0.8012 to obtain a higher average connectivity degree in the lncRNA group (Fig. 2B). Therefore, the $\beta$ value determined distinct gene co-expression modules in mRNA and lncRNA of the exosome. The cluster dendrogram of all selected genes was clustered with the adjacency matrix.

Fig. 2.

Analysis of network topology of various soft-thresholding powers. mRNA (A) and lncRNA (B) power value filter. The left panel shows the scale-free fit index (y-axis) as a function of the soft-thresholding power (x-axis). The right panel displays the mean connectivity (degree, y-axis) as a function of the soft-thresholding power (x-axis). mRNA, messenger RNAs; lncRNA, long non-coding RNAs.

This study performed correlation analysis between the eigengene patterns within the co-expression modules and the clinical phenotype dataset (including continuous variables such as maternal age, pre-pregnancy BMI, overall weight gain, weight gain during first/second/last trimesters, and neonatal weight) (Fig. 3A,B). Key modules were identified by calculating the correlation coefficient “r” and p-value between module eigengenes and clinical traits. The mRNA module-trait relationship heatmap (Fig. 3A) revealed that lightyellow module was significantly positively correlated with pre-pregnancy BMI (“r” = 0.56, p = 0.020), green module was significantly positively correlated with weight gain during first trimester (“r” = 0.57, p = 0.020), and royalblue module exhibited a significant positive correlation with overall weight gain (“r” = 0.65, p = 0.006). The lncRNA module-trait relationship heatmap (Fig. 3B) showed that the darkgrey module was significantly correlated with weight gain during the first trimester (“r” = 0.60, p = 0.009). Based on the “r” and p values, the mRNA royalblue module and lncRNA darkgrey module, which have the highest correlation with clinical phenotypes, were selected as key modules for further investigation into their association with macrosomia occurrence in non-diabetic pregnancies.

Fig. 3.

Module-trait associations for mRNA (A) and lncRNA (B). Each row corresponds to a module eigengene, and the columns correspond to clinical indexes. Red indicates a positive correlation, and blue indicates a negative correlation. Each cell contains the corresponding correlation and p-value.

The analysis of KEGG pathways revealed various important signaling pathways corresponding to the clinical traits, including metabolic pathways such as the oxytocin signaling pathway, aldosterone synthesis and secretion, and cortisol synthesis and secretion (Fig. 4).

Fig. 4.

KEGG enrichment. KEGG analysis revealed that the enriched pathways were the oxytocin signaling pathway, aldosterone synthesis and secretion, and cortisol synthesis and secretion. KEGG, Kyoto Encyclopedia of Genes and Genomes.

Hub gene interaction networks were constructed for the mRNA royalblue module and the lncRNA darkgrey module, respectively (Fig. 5). Scatter plots depicting GS versus MM were also generated (Fig. 6). The criteria for screening hub genes in this study were set as $|$MM$|$ $\ge$0.88 and $|$GS$|$ $\ge$0.35. Through hub gene screening, core genes such as transmembrane protein 175 (TMEM175), phosphoglycerate kinase 1 (PGK1), and methionine adenosyltransferase 2B (MAT2B) in the mRNA royalblue module and HIF1A antisense RNA 2 (HIF1A-AS2) in the lncRNA darkgrey module were found to be closely associated with neonatal weight gain. Their MM and GS values were significantly higher than those of other genes (Table 4). By calculating relative expression levels, it was found that compared to normal-weight newborns, the expression of TMEM175 and HIF1A-AS2 was down-regulated, while the expression of PGK1 and MAT2B was up-regulated in non-diabetes-related macrosomia cases (Fig. 7). These findings suggested that these genes may play important regulatory roles in the development of macrosomia in non-diabetic pregnancies. They potentially influence fetal growth and development through complex network regulatory mechanisms, thereby promoting the occurrence of macrosomia.

Fig. 5.

The core gene interaction network of mRNA royalblue (A) and lncRNA darkgrey (B) modules.

Fig. 6.

Scatter plots of GS for neonatal weight and MM in mRNA royalblue (A) and lncRNA darkgrey (B) modules. There was a significant correlation between GS and MM in this module. MM, module membership; GS, gene significance.

Fig. 7.

The expression levels of TMEM175 (A), PGK1 (B), MAT2B (C) in the mRNA royalblue module and HIF1A-AS2 (D) in the lncRNA darkgrey module in umbilical cord blood.

Our study found that excessive gestational weight gain was a high-risk factor for macrosomia, aligning with the findings of earlier studies [26, 27]. A meta-analysis indicated that excessive gestational weight gain was an independent risk factor for macrosomia compared with moderate weight gain [28]. A retrospective study also found that accelerated weight gain rate in obese pregnant women before 24 weeks of gestation was associated with an increased risk of macrosomia [29]. Our study also confirmed that the pre-pregnancy BMI of the macrosomia group was significantly higher than that of the control group, verifying the protective effect of lower pre-pregnancy BMI against macrosomia [30]. Research has reported that multiparous women were more likely to give birth to macrosomic infants [31], which was also observed in our study. Our study also found that the cesarean section rate is significantly higher in non-diabetic macrosomic infants, further validating that fetal weight is a key factor in the decision-making process for delivery methods [32]. Omaña-Guzmán et al. [33] indicated that elevated DBP during pregnancy may be negatively correlated with fetal growth. Although this study found that maternal DBP was lower in the macrosomia group with a statistically significant difference, the DBP values remained within the normal physiological range. Therefore, the specific impact of blood pressure changes on the occurrence of macrosomia in non-diabetic pregnant women requires further validation through larger-scale clinical studies and mechanistic investigations. Additionally, the intrauterine metabolic environment shaped by maternal glucose and lipid profiles may also affect fetal growth [34]. Our research revealed that lower levels of maternal HDL and higher levels of LDL during the last trimester were significantly associated with an increased risk of macrosomia in non-diabetic pregnancies, indicating the importance of maternal lipid metabolism [35]. This is consistent with the results presented by Li et al. [36], who also reported that elevated LDL in the last trimester increased the risk of macrosomia. Peng et al. [34] also found that increased HDL levels in the last trimester were associated with lower neonatal weight. However, Wang et al. [37] observed no association between HDL-C levels and neonatal birth size. These differences may be attributed to regional variations in study populations, lifestyle factors, or underlying genetic background, warranting further exploration of the complex relationship between maternal lipid metabolism and fetal growth. Notably, although a previous study reported an association between elevated ALP levels and large for gestational age (LGA) infants in non-diabetic mothers [38], our findings showed no significant correlation between ALP levels in the first trimester and macrosomia in non-diabetic mothers. However, elevated ALP levels in the last trimester were associated with macrosomia, potentially as a result of the effects of placental and skeletal isoenzymes [39]. These conflicting results may show that the relationship between ALP levels and adverse effects on fetal outcomes can be complex and remain controversial. A study has found that pregnant women with extremely high levels of serum ALP levels gave birth to normal-weight infants [38]. This may imply that ALP alone does not affect macrosomia in infants, and it should be used together with other indicators to dictate clinical management [40]. These findings highlight the significance of managing pre-pregnancy weight, monitoring gestational weight gain, and regulating lipid levels for preventing macrosomia in non-diabetic mothers. In particular, lipid level regulation can be accomplished through a balanced diet rich in unsaturated fats and dietary fiber, along with the moderate physical activity advised by healthcare providers, to mitigate the effects of maternal obesity on the child [41]. Moreover, neonatal gender seems to be related to macrosomia in non-diabetic mothers. Placental samples showed higher methylation levels in males compared to females [42]. Studies indicated a higher proportion of male neonates among macrosomic births in non-diabetic pregnancies [1, 4, 43], consistent with our findings. Nevertheless, the specific regulatory mechanisms involved require further investigation.

Fetal macrosomia is more prevalent in mothers with pre-gestational diabetes and GDM compared to non-diabetic mothers; thus, mothers with GDM carry a higher risk. This increased risk is primarily due to the effects of maternal hyperglycemia, which leads to excess glucose transfer to the fetus and subsequent fetal hyperinsulinemia and fat storage [44, 45]. In non-diabetic mothers, fetal macrosomia is a complex condition resulting from the interplay of multiple factors such a dyslipidemia and gestational weight gain [35, 45, 46]. Moreover, an increase in maternal glucose metabolism in non-diabetic mothers can contribute to macrosomia [47]. A significant number of patients with GDM may also suffer from obesity, a factor associated with fetal macrosomia. The combination of maternal obesity and GDM appears to have a synergistic effect, increasing the risk even further [48]. However, obesity is an independent risk factor, and research indicates that a significant portion of infants with macrosomia are born to women who are obese without diabetes [49]. This suggests that BMI is a major contributor to macrosomia in both diabetic and non-diabetic mothers.

Currently, the application of exosomes in the research of the mechanism of pregnancy complications remains quite limited, and related explorations have not been extensively conducted [16]. Notably, the feasibility of exosomal RNA as a non-invasive biomarker also deserves attention. Although the initial equipment investment for isolation techniques (e.g., ultracentrifugation) is relatively high, the popularization of commercial kits has significantly reduced the cost of single detection. Moreover, exosomes are widely present in body fluids such as blood, urine, and milk, making sample collection convenient and non-invasive, which is suitable for large-scale population screening [50]. Additionally, combined with multi-omics analysis (such as small RNA sequencing), its specificity and sensitivity can be improved, providing a multi-dimensional solution for the early diagnosis and assessment of pregnancy complications [51].

In our study, RNA sequencing of umbilical venous blood exosomes was used to construct a co-expression network of mRNA and lncRNA. By analyzing continuous variables such as maternal age, pre-pregnancy BMI, and gestational weight gain, we found that the mRNA royalblue module and lncRNA darkgrey module were highly correlated with gestational weight gain. These modules may participate in physiological processes such as adipocyte differentiation and energy metabolism balance through cooperative regulation of gene expression [52]. As key mediators of intercellular communication, RNA molecules carried by exosomes may affect gene expression patterns in the placenta and maternal metabolic organs via blood circulation [53]. Furthermore, KEGG enrichment analysis revealed associations between macrosomia-related genes and the oxytocin signaling pathway, aldosterone synthesis and secretion, and cortisol synthesis and secretion pathways. As a study has shown, the oxytocin pathway may be involved in weight regulation [54]. Aldosterone is found to be elevated in obese individuals, and it is positively associated with BMI [55, 56], while cortisol promoted weight gain by enhancing fat storage and slowing down metabolism [57]. Therefore, we speculate that these three pathways may increase the risk of macrosomia by influencing maternal weight changes during pregnancy. However, the specific role of exosome RNA remains unclear, and future studies using animal models are needed to verify the causal relationship between exosomes in weight regulation during pregnancy and the occurrence of macrosomia, providing a theoretical basis for early intervention in pregnancy complications.

WGCNA analysis identified TMEM175, PGK1, and MAT2B as hub genes in the mRNA royalblue module, and HIF1A-AS2 as the hub gene in the lncRNA darkgrey module. Moreover, all these genes were closely related to gestational weight gain. As a key channel protein regulating the lysosome-mitochondria metabolic axis, TMEM175 mutations can lead to abnormal lysosomal acidification, which in turn impacts the lysosomal-mediated autophagy process and nutrient-sensing pathways (e.g., the mechanistic target of rapamycin (mTOR) pathway). Consequently, this can lead to cellular metabolic disorders [58, 59]. The mTOR pathway has been proven to play a significant role in regulating early stages of embryonic development [60], while TMEM175 is also capable of modulating mitochondrial function and facilitating mitophagy [59]. Lin et al. [61] observed reduced placental mtDNA copy numbers in macrosomic infants born to non-diabetic mothers, indicating a possible link between mitochondrial dysfunction and macrosomia, though the precise mechanisms remain unclear. Additionally, fetal growth mainly depends on the availability of sufficient nutrients and their transportation in the placenta, such as glucose. Glucose has been proven to be a key component promoting cell proliferation [62]. Research has demonstrated that regulating maternal glucose levels is essential for the growth and development of offspring. Furthermore, PGK1, a vital enzyme in the glycolytic pathway, influences glucose utilization [63]. The concentration of maternal glucose affects placental glucose transport, which provides energy for normal fetal intrauterine development. Therefore, we infer that PGK1 may regulate placental nutrient transport via glucose metabolism, and its dysregulation may increase glucose transport, leading to fetal overgrowth and macrosomia. Furthermore, fetal growth and development also require an adequate and balanced supply of amino acids for physiological processes such as protein synthesis. Methionine, an essential sulfur-containing amino acid, is critical for fetal growth and development [64]. Research has shown that methionine adenosyltransferase 2A (MAT2A) is involved in methionine metabolism and can catalyze the synthesis of S-adenosylmethionine (SAM) (a precursor of cysteine) [65]. Rubini et al. [66] confirmed that cysteine can affect fetal development. Furthermore, studies have reported that overexpression of MAT2A can promote lipid accumulation and significantly upregulate the expression levels of adipogenic marker genes such as peroxisome proliferator-activated receptor γ (PPAR$\gamma$), sterol regulatory element-binding protein-1c (SREBP-1c), and adipocyte protein 2 (aP2) [67]. PPAR$\gamma$, which regulates placental development, may affect macrosomia by modulating placental lipid transport [68]. Research has indicated that in the case of lncRNAs, the upregulation of HIF1A-AS2 can significantly promote trophoblast proliferation and regulate placental angiogenesis [69]. These two factors are fundamental to placental development and the pathogenesis of macrosomia in infants born to non-diabetic mothers. This may affect the formation and development of the placenta, enhance the efficiency of maternal nutrient transport to the fetus, and lead to excessive weight gain by a fetus exposed to a high-nutrition environment over extended periods of time [70]. However, no studies to date have directly explored the roles of TMEM175, PGK1, MAT2A, and HIF1A-AS2 in macrosomia in non-diabetic pregnancies. Their specific regulatory mechanisms and long-term effects on fetal development require further investigation.

This study constructed a differential transcriptome network based on systems biology to elucidate associations between differentially expressed RNAs and phenotypes (e.g., gestational weight gain), providing a basis for developing preventive strategies. However, there are several limitations. Firstly, its retrospective design inherently introduces selection bias. The inclusion criteria and patient characteristics may not reflect the broader population, and the availability and completeness of historical data may introduce additional confounding factors, thereby affecting the generalizability of the findings. Secondly, the single-center setting and relatively small sample size limit the robustness and generalizability of findings. Thirdly, there is a lack of comprehensive functional verification and a thorough assessment of clinical practicality. Although associations with exosome RNA were observed, the specific regulatory mechanisms (such as molecular pathways and interaction networks) remain unclear. Moreover, the clinical significance of exosomal RNA as a non-invasive screening biomarker, encompassing indicators such as sensitivity, specificity, and predictive value, requires further exploration. Future research should employ multi-center designs and expand sample sizes to verify result reliability. In vitro and in vivo functional experiments are needed to clarify the precise regulatory mechanisms of exosomal RNA. Additionally, well-designed prospective clinical trials should be conducted to systematically evaluate their diagnostic accuracy, prognostic value, and clinical applicability, thereby providing robust evidence for potential clinical application.