Serum lncRNA H19/miR-675 /PPARα expression before middle gestation and their associations with macrosomia risk in singleton pregnancies without gestational diabetes mellitus: a preliminary study

Study population

This nested case-control study was conducted within a prospective cohort of pregnant women. Participants were women with singleton pregnancies between 13 and 20⁺⁶ weeks’ gestation as determined by ultrasound, who attended prenatal care at four hospitals in hospitals in Wenzhou, China, between July 2020 and April 2021. Cases were mothers who delivered a liveborn with macrosomia (birthweight ≥ 4,000 g). Controls were randomly selected from mothers within the same cohort who delivered a live, term newborn with normal birthweight (ranging from 2,500 g to 3,999 g). Controls were individually matched to each case based on both gestational age at blood collection (within ±1 week) and delivery date (within ±1 week).

Inclusion criteria for all participants were: term pregnancy (37 to 42 weeks), singleton gestation, delivery without assisted technology, and a liveborn without congenital malformations. Exclusion criteria included pre-pregnancy metabolic diseases or long-term medication use, congenital or genetic disorders, and maternal blood samples collected after 20 weeks’ gestation. All eligible women were informed about the study, and provided written informed consent. Ethical approval was obtained from the Wenzhou Medical University Ethics Committee (2018035).

The measurement of serum Lipid profile

The maternal serum concentrations of triglyceride (TG), total cholesterol (TC), high density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) were measured by automatic biochemical analyzer (Mindray BS480, Shenzhen, China) in the same hospital laboratory. The detection kits (oxidase methods for TG and TC, catalase removal method for HDL-C and surfactant method for LDL-C) were purchased from Mindray (Shenzhen, China).

Collection of survey information and delivery records

Participants completed a structured questionnaire that collected data on socio-demographic characteristics (maternal age, ethnicity, height and weight, and annual household income), lifestyle before 20 weeks’ gestation (weekly exercise frequency, nighttime sleep duration, and diet based on a simplified food-frequency questionnaire), and disease history (self-reported metabolic, genetic, congenital, mental disorders). Delivery information (neonatal sex, birthweight, survival status, and delivery mode) and maternal health records during pregnancy (gestational weight gain, and complications) were retrieved from hospital electronic medical records.

Maternal blood sample collection and preservation

Fasting venous blood samples (five ml) were collected by trained clinicians and left at room temperature (approximately 26 °C) for 20 mins. Samples were centrifuged at 3,000 rpm for 5 min, and serum aliquots were transported on dry ice to the laboratory, transferred into RNase-free cryotubes, and stored at −80 °C.

Quantitative real-time polymerase reaction

The total RNA was extracted from maternal serum using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and then reverse transcribed into cDNA using a reverse transcription kit (Takara, Tokyo, Japan). The RNA samples, isolated with 10 µL nuclease-free water, was quantified using Multiskan GO UV/Vis Microplate Spectrophotometer (Thermo Fisher Scientific, MA, USA). RNA samples with A260/A230 ratios between 1.8 and 2.0 and A260/A280 ratios greater than 2 were selected. Formaldehyde agarose gel was used to examine the integrity of all RNA samples (28S/18S ≈ 2.0, no visible smear). Total RNA of 600 ng was reverse transcribed for 15 min at 37 °C with PrimerScript RT Reagent Kit (TaKaRa Bio Inc., Dalian, China) according to the manufacturer’s instruction. The reaction was terminated by heating 85 °C for 5 s, and then were stored at −20 °C until next day. The cDNA levels were analyzed using a CFX96 Touch quantitative real-time polymerase chain reaction (qRT-PCR) detection system (Bio-Rad, Hercules, CA, USA) with SYBR green dye (Roche, Basel, Switzerland) according to the manufacturer’s guidelines. Each sample, containing two µL of RT product, was replicated by three times in a MicroAmp Fast Optical 96-well plate (Bio-Rad, Hercules, CA, USA) in a reaction volume (10 µL) of 2X Universal SYBR Green Fast qPCR Mix (Abclonal, Wuhan, China), and the real-time PCR was conducted with an initial denaturation of 3 min at 95 °C accompanied by 44 cycles of 5 s at 95 °C, 15 s at 53 °C, 15 s at 72 °C. Reverse transcription of miR-675 and miR-39-3p were conducted with stem-loop primers (Ribobio, Guangzhou, China). MiR-39-3p (Ribobio, Guangzhou, China) was used as the external reference for lncRNA H19 and miR-675, and standardized values were calculated. The primers of lncRNA H19 (141bp) was synthesized by Sangon (Shanghai, China). The primer sequences of lncRNA H19: Forward, 5′-GGACGTGACAAGCAGGACAT-3′; Reverse, 5′-CATAGTGCCGACTCCGTG-3′. The threshold cycles (Ct) were calculated using the CFX Maestro software. Reactions for each sample prepared technical duplicates and blank negative controls, and the relative expression was calculated by the equation 2-ΔΔCt. Variability was defined by the standard deviation (SD) of mean Cq values and low SD values corresponded to stable miR-675 or lncRNA H19. As this was a screening study, not all MIQE (Minimum Information for Publication of Quantitative Real-Time PCR Experiments) specifications were provided. All Real-time PCR experiments were compliant with the MIQE guidelines.

Enzyme-linked immunosorbent assay

The serum was thawed at 4 °C after being removed from the −80 °C refrigerator. The concentration of the PPARα protein in maternal serum was measured using enzyme-linked immunosorbent assay kits (Cusabio, Wuhan, China) according to the manufacturer’s instructions.

Statistical analysis

Analyses were conducted in R software version 4.1.3. Questionnaire data were double-checked and entered into EpiData 3.1. Multiple imputation was applied to covariates with <30% missing data. Normally distributed continuous variables were presented as mean ± standard deviation (SD) and compared using unpaired or paired t test. Skewed continuous variables were expressed as median (P25, P75), and compared using the Wilcoxon signed-rank test or Mann–Whitney U test. Categorical variables were summarized as counts (%), and compared using the McNemar test, chi square test, or Fisher’s exact test. A two-tailed P value < 0.05 indicated statistical significance.

Correlation analyses were performed to assess the relationships among maternal serum lipids, lncRNA H19, miR-675, and PPARα protein levels. Pearson’s correlation was used for variables that followed a bivariate normal distribution; otherwise, Spearman’s rank correlation was applied. Conditional multivariate logistic regression was used to estimate Odds Ratios (ORs) between lncRNA H19, miR-675, PPARα and macrosomia, along with the corresponding 95% confidential intervals (CIs). Restricted cubic spline (RCS) based on logistic regression was performed, and the expression value of the first quartile of lncRNA H19, the third quartile of miR-675, and the first quartile of PPARα at OR = 1 was treated as the references. Subgroup analyses were performed to evaluate the influences of three molecular indexes (lncRNA H19, miR-675, PPARα) on the occurrence of macrosomia in various subgroups, including overweight before pregnancy (yes/no), parity (primiparous/multiparous), and maternal age (</≥ 30 years). Sensitivity analysis was performed in pregnant women who delivered ≤ 40 weeks’ gestation, or who were without history of macrosomia, or whose blood samples collected ≤ 13 weeks’ gestation.

Receiver operating characteristics (ROC) curves were used to evaluate the predictive performance of each of three molecular indexes (lncRNA H19, miR-675, PPARα) and combined on the occurrence of macrosomia, quantified using area under curve (AUC). The net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to assess incremental predictive performance beyond the basic models.

Participant characteristics

Maternal and neonatal characteristics in the macrosomia group (n = 47) and matched control group are presented in Table 1, and the screening process is shown in Fig. S1. The corresponding prospective cohort of 898 participants summarized in Table S1, imputed data before and after processing are presented in Table S2. The overall prevalence of macrosomia was 6.24%. The mean birthweight of newborns with macrosomia was 4,103.43 ± 141.77 g, compared with 3,358.40 ± 288.94 g in the control group.

Expression levels of lncRNA H19, miR-675, PPARα protein level in maternal serum

There was no statistically significant difference in lncRNA H19 (Z = −0.344, P = 0.731) or miR-675 (Z = −1.376, P = 0.169) between the two groups (Figs. 1A–1B). Likewise, the serum protein expression level of PPARα did not differ significantly between groups (Z < 0, P = 0.999) (Fig. 1C).

Correlation analysis of serum lipids with serum lncRNA H19, miR-675, and PPARα protein level

Positive correlations were observed between serum PPARα protein levels and TC (r_s =  − 0.32, P = 0.03), and between serum PPARα protein level and LDL-C only in control group (Figs. 2B–2D). No associations were detected between PPARα and TG, PPARα and HDL-C, serum lipids and lncRNA H19, or serum lipids and miR-675 (Figs. 2E–2L).

Associations of serum lncRNA H19, miR-675, and PPARα protein level with macrosomia

The RCS analyses indicated no correlation between lncRNA H19, miR-675, or PPARα protein levels and the occurrence of macrosomia (Fig. S2). As shown in Table 2, maternal serum lncRNA H19 and miR-675 were not associated with the risk of macrosomia. Compared with the tertile 3 of PPARα protein levels, tertile 2 was associated with a lower risk of macrosomia (adjusted OR: 0.15; 95% CI [0.03–0.87]), after adjustment for gestational age, pre-pregnancy BMI, and maternal age.

Stratified analysis evaluated whether the associations between PPARα protein and macrosomia varied across prespecified subgroups (Fig. 3). The associations observed in most subgroups was consistent with the main findings of Model 3 (Table 2). No significant interactions were detected (Fig. 3). Notably, in pregnant women with pre-pregnancy BMI <24 kg/m², tertile 2 of maternal serum PPARα protein levels was associated with a 70% reduction in the risk of macrosomia compared with the tertile 3 (OR: 0.30, 95% CI [0.09–0.99]; P = 0.049).

Sensitivity analysis for PPARα protein levels were broadly consistent with the main findings (Table S3). When restricting the analysis to pregnancies with ≤ 40 weeks’ gestation, tertile 2 remained significantly associated with a reduced risk of macrosomia after covariate adjustment (Model 3: OR, 0.20, 95% CI [0.04–0.99]), whereas the corresponding main-model estimate was OR: 0.17 (95% CI [0.03–1.01]) (Table S3, Table 2). Similar results were observed among women without a history of macrosomia and among those with maternal blood samples collected ≤ 13 weeks. Consistent results were also noted after excluding gestational age >40 weeks (Table S3 and Table 2).

Incremental combination predictive power of serum lncRNA H19-miR-675- PPARα in macrosomia

The basic model included TC, TG, HDL, LDL, maternal age, ethnicity, parity and BMI before pregnancy, and the FBG model further included fasting blood glucose before 20 weeks’ gestation. Given the limited sample size, an exploratory assessment of predictive performance was conducted. Adding the three biomarkers to the basic clinical model did not significantly improve AUC (AUC: 0.799 vs. 0.760, P = 0.41) (Table 3 and Fig. S3). However, improvements in NRI and IDI—while requiring cautious interpretation due to sample size constraints—suggest that this biomarker combination merits further evaluation in larger cohorts as potential early indicators (Table 3).