Subclinical Changes in Cardiovascular Function During the Neonatal Period in Infants Born to Mothers With Diabetes

Abstract

OBJECTIVE

The link between maternal diabetes and congenital heart defects is well-established; however, few studies have investigated subclinical cardiovascular dysfunction at birth, which may serve as an early indicator of future risk.

METHODS

We analyzed cardiovascular metrics, including heart rate and blood pressure during the postnatal hospital stay, in 1927 maternal-infant dyads at Nebraska Medicine (2012–2023), including 226 with gestational diabetes mellitus (GDM), 27 with type 1 DM (T1DM), 67 with type 2 DM (T2DM), and 1607 with no DM. Relationships between maternal diabetes subtype and neonatal cardiovascular metrics were examined using treatment effects modeling. Correlations between maternal glycemic burden, as assessed by average HbA1c and 1-hour oral glucose tolerance test (OGTT) during pregnancy, and newborn cardiovascular metrics were also evaluated.

RESULTS

Compared with no DM, T1DM was associated with higher newborn systolic blood pressure (SBP) (+4.6 mm Hg, p = 0.002), pulse pressure (PP) (+3.1 mm Hg, p = 0.006), and SBP variability (+1.2 mm Hg, p = 0.048). T2DM was associated with higher newborn SBP (+2.9 mm Hg, p = 0.03) and PP (+2.0 mm Hg, p = 0.02) and lower SBP variability (−1.3 mm Hg, p = 0.049). There were sigsnificant positive correlations between maternal HbA1c and newborn heart rate (p = 0.009, R² = 0.27), mean arterial pressure (p = 0.004, R² = 0.20), SBP (p < 0.001, R² = 0.22), diastolic BP (p = 0.03, R² = 0.17), and PP (p = 0.001, R² = 0.17). No correlations were observed between maternal 1-hour OGTT and newborn cardiovascular metrics.

CONCLUSIONS

These findings indicate early subclinical changes in cardiovascular physiology in offspring of women with DM. The differential impacts of diabetes subtypes on newborn BP and strong link to maternal glycemic burden suggest severity and timing of maternal diabetes influence neonatal cardiovascular physiology.

Introduction

Diabetes mellitus (DM) is one of the most pressing global health issues, with an attributed 3.4 million deaths annually worldwide.¹ The global prevalence of DM, including type 1 (T1DM), type 2 (T2DM), and gestational (GDM), increased from 108 million (4.7% of the global population) in 1980 to 589 million (11.1% of the global population) in 2025 and is projected to increase to 853 million by 2050.^1,2

The prevalence of DM during pregnancy is also rising rapidly.³ Maternal DM affects an estimated 23 million pregnancies worldwide, impacting 19.7% of all live births.¹ Offspring of women with DM have worse immediate and long-term health outcomes compared with those born from healthy pregnancies.^4,5 In particular, T1DM and T2DM during pregnancy have been found to confer risk of adverse neonatal outcomes such as macrosomia, hyperbilirubinemia, and hyperinsulinemic hypoglycemia.^6,7 During adulthood, offspring of women with DM are at increased risk of developing cardiovascular diseases (CVD), including pulmonary hypertension and heart failure; however, the etiology of these risks remains unclear and is not fully explained by genetics or socioeconomic status.^8–14

Risk factors for CVD have been noted during childhood and early adolescence in offspring of women with DM.^15–19 Several studies have demonstrated cardiac and vascular disruptions in this population, including increased aortic stiffness, left ventricular hypertrophy, and impaired myocardial function during the perinatal period.^13,17,20,21 Prior research has demonstrated disruptions at a cellular level, with reports of human umbilical vein endothelial cell dysfunction, a key model of the fetal vascular endothelium. These data suggest that intrauterine programming of cardiovascular dysfunction may contribute to cardiovascular outcomes in these offspring, reflecting the tenets of the developmental origins of health and disease theory.^22,23 The pro-inflammatory and pro-fibrotic gestational microenvironment during pregnancies with maternal DM can limit fetal growth and development, and may contribute to alterations in fetal vascular beds, influencing the cardiovascular physiology of neonates.²⁴ However, it remains unclear whether these changes contribute to system-wide disruptions, measurable in routine clinical exams, during the neonatal period, as seen among offspring of women with DM during childhood and adolescence.

To address this critical gap in knowledge, we conducted a retrospective cohort study examining clinically accessible measures of neonatal cardiovascular function in offspring born to women with and without DM during pregnancy. We sought to determine if subclinical alterations in cardiovascular function may be detected during the neonatal period, potentially linking intrauterine programming to future CVD risk.

Materials and Methods

Data Acquisition.

This retrospective cohort study analyzed clinical data from patients who opted in to research data collection at an academic medical center in the Midwestern United States between 2012 and 2023. Data were obtained from the Epic electronic health record (EHR) by the University of Nebraska Medical Center Research, Education, Administration, and Development of Biomedical Informatics (READi) Core. In accordance with the state of Nebraska age of majority laws, only maternal patients ≥20 years of age at the time of delivery and their newborns were included in this study. Data from newborns who were wards of the state were excluded from this study. Data from newborns diagnosed with cardiac or renal abnormalities or those delivered extremely preterm (<28 weeks gestation) were excluded to preserve focus on subclinical disruptions of cardiovascular function. Maternal diagnoses of GDM, T1DM, T2DM, and hypertensive disorders of pregnancy, including gestational hypertension, preeclampsia, eclampsia, and pre-existing chronic hypertension, were provided by clinicians and retrieved from the EHR. Maternal age and pre-pregnancy body mass index (BMI), up to 2 years prior to delivery of the newborn, were derived from the EHR. All maternal hemoglobin A1c (HbA1c) and 1-hour oral glucose tolerance test (1-hour OGTT) measurements during pregnancy were retrieved from the EHR and averaged. Our primary outcome measures newborn heart rate and blood pressure (BP) were compiled from the postnatal hospital stay. For each newborn, all available heart rate, systolic BP (SBP), and diastolic BP (DBP) measurements recorded in the EHR during hospitalization were used to calculate the individuals’ averages, mean arterial pressure (MAP), and SDs. These SDs represent variability of cardiovascular metrics at an individual level rather than population level. Considering the physiological changes in newborn heart rate and BP during the postnatal period, only averages and SD from newborns with ≥5 recorded heart rate or BP measurements were used for their respective analyses in order to limit the influence of sparse data and outliers to better capture true individual estimates.

Summary Statistics.

The distribution of data was evaluated before statistical testing. Continuous variables were evaluated for skewness and kurtosis to assess normality. Summary statistics were compiled for maternal and newborn demographics and stratified by maternal DM classification. Kruskal-Wallis tests were used to compare key demographic metrics between maternal DM classifications based on the non-parametric data distribution of continuous variables. χ² tests were used for categorical variables. Percentages, averages, SD, and statistical comparisons were computed based on non-missing data for each variable. Due to differences in data availability, sample sizes vary by analysis and are outlined in the study flow diagram (Figure 1). Analytic sample demographics are provided for the BP analyses in Supplemental Tables S1 and S2.

Figure 1.

Study flow diagram. Clinical data were obtained for maternal-infant dyads cared for at Nebraska Medicine between 2012 and 2023. After applying inclusion and exclusion criteria and filtering for missing covariate data, 1927 dyads were included in the study population.

Treatment Effects Estimation.

Inverse probability weighted regression adjustment (IPWRA) was used to balance the baseline covariates associated with risk for GDM and T2DM (maternal BMI and advanced maternal age).²⁵ A quasi-experimental approach was employed to estimate the average treatment effect on the treated for maternal DM on newborn cardiovascular metrics, including heart rate, heart rate variability, MAP, SBP, SBP variability, DBP, DBP variability, and PP. Outcome models were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal BMI, and advanced maternal age (≥35 years).²⁶ The use of IPWRA provided double-robust estimators, yielding valid estimates even if either the treatment or outcome model was mis-specified.

Special consideration for the unique pathophysiology of T1DM and lack of available predictors precluded the use of IPWRA. Instead, propensity score matching based on nearest-neighbor matching (NNM) considering maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal BMI, and advanced maternal age as covariates was applied.²⁷ The matching process ensured that patients in no DM and T1DM groups had similar covariate characteristics.²⁸ The resulting matched observations were used to estimate the average treatment effect on the treated for maternal DM on newborn cardiovascular metrics, including heart rate, SD of heart rate, MAP, SBP, SBP variability, DBP, DBP variability, and PP. The balance criteria for variables in both IPWRA and NNM included a standardized mean difference not exceeding ±0.1 and a ratio of variances between 0.9 to 1.1. The threshold of significance was set at p = 0.05. Analyses were performed using STATA 18.5 (StataCorp LLC, College Station, TX). Treatment effects analyses were visualized using GraphPad Prism 10 (GraphPad Software, Boston, MA).

Regression Analyses.

Associations between maternal HbA1c or 1-hour OGTT and newborn heart rate, SD of heart rate, MAP, SBP, SBP variability, DBP, DBP variability, and PP were assessed using linear regression. Logistic and linear regressions were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal BMI, and advanced maternal age. The threshold of significance was set at p < 0.05. Analyses and visualization were performed using STATA 18.5, and analysis files are included in the data supplement.

Results

Demographics.

A total study population of 1927 maternal-infant dyads was achieved for the analyses detailed in Figure 1. Of the 1927 maternal-infant dyads analyzed in this study, 1607 (83.3%) had no DM, 226 (11.7%) had GDM, 27 (1.4%) had T1DM, and 67 (3.5%) had T2DM (Table 1). The demographic characteristics of participants were broadly representative of the patient population served by Nebraska Medicine (Omaha, NE), where this study was conducted. Maternal age was significantly different between groups (p = 0.0001), with median ages of 31 (no DM), 32 (GDM), 29 (T1DM), and 33 years (T2DM). Pre-pregnancy BMI differed significantly (p = 0.0001), with a median BMI of 26.7 (no DM), 29.2 (GDM), 24.6 (T1DM), and 34.3 kg/m² (T2DM). With the exception of T1DM, all BMI medians fell within the overweight (25–29.9 kg/m²) and obese (>30 kg/m²) categories, consistent with BMI distributions among women in Nebraska aged 20 to 39 years, which comprised the bulk of our cohort.²⁹ The distribution of race differed significantly between groups (p < 0.001). Hypertensive disorders of pregnancy, including gestational hypertension, preeclampsia, eclampsia, and chronic hypertension during pregnancy, were more prevalent among individuals with DM (p < 0.001). Neonatal outcomes also differed significantly (Table 2). Neonatal intensive care unit (NICU) admission occurred in 18.5% of infants born to mothers with no DM compared with 27.9% in GDM, 55.6% in T1DM, and 53.7% in T2DM groups (p < 0.001). Offspring of mothers with DM were delivered at earlier gestational ages, with a median of 39.1 weeks infants born to mothers with no DM compared with 39.0 in GDM, 37.1 in T1DM, and 37.3 in T2DM (p = 0.0001).

Table 1.

Maternal Demographics*

	No DM (N = 1607)	GDM (N = 226)	T1DM (N = 27)	T2DM (N = 67)	p value
Age, yr	31 (27–34)	32 (29–36)	29 (26–32)	33 (29–36)	0.0001
BMI, kg/m2	26.7 (23.1–32.5)	29.2 (25.0–36.1)	24.6 (22.3–26.4)	34.3 (25.1–41.3)	0.0001
C-section	422 (26.3%)	73 (32.3%)	17 (63.0%)	40 (59.7%)	0.001
Hypertensive disorders of pregnancy					0.001
Gestational hypertension	164 (10.2%)	30 (13.3%)	5 (18.5%)	2 (3.0%)	—
Preeclampsia	188 (11.7%)	27 (12.0%)	7 (25.9%)	19 (28.4%)	—
Eclampsia	1 (0.1%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	—
Chronic hypertension	82 (5.1%)	25 (11.1%)	3 (11.1%)	19 (28.4%)	—
Race					0.001
American Indian/Alaska Native	7 (0.4%)	8 (3.5%)	0 (0.0%)	0 (0.0%)	—
Asian	23 (1.4%)	13 (5.8%)	0 (0.0%)	1 (1.5%)	—
Black/African American	226 (14.1%)	18 (8.0%)	2 (7.4%)	11 (16.4%)	—
Chose not to disclose	4 (0.3%)	1 (0.4%)	0 (0.0%)	0 (0.0%)	—
Multiracial	29 (1.8%)	3 (1.3%)	0 (0.0%)	2 (3.0%)	—
Native Hawaiian/Pacific Islander	2 (0.1%)	0 (0.0%)	0 (0.0%)	2 (3.0%)	—
Other	100 (6.2%)	8 (3.5%)	0 (0.0%)	9 (13.4%)	—
White/Caucasian	1216 (75.7%)	175 (77.4%)	25 (92.6%)	42 (62.7%)	—
Ethnicity					0.743
Not Hispanic/Latino	1476 (91.2%)	210 (92.9%)	27 (100.0%)	59 (88.1%)	—
Hispanic/Latino	128 (7.8%)	15 (6.6%)	0 (0.0%)	8 (11.9%)	—
Chose not to disclose	2 (0.1%)	1 (0.4%)	0 (0.0%)	0 (0.0%)	—
Unknown	1 (0.1%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	—
OGTT	n = 1360	n = 203	n = 1	n = 17
1-hr OGTT	113 (97–130)	152 (139–170)	411	170 (129–187)	0.0001
HbA1c	n = 316	n = 90	n = 27	n = 59
HbA1c	5.2 (5.0–5.4)	5.3 (5.0–5.6)	6.6 (6.0–8.3)	6.1 (5.6–6.8)	0.0001

BMI, body mass index; C-section, cesarean section; DM, diabetes mellitus; GDM, gestational diabetes mellitus; HbA1c, hemoglobin A1c; OGTT, oral glucose tolerance test; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus

^* Data presented as no. (%) for categorical variables and median (IQR) for continuous variables. Demographic and clinical characteristics for mothers of newborns included in the overall cohort are shown, stratified by diabetes mellitus classifications. Statistical comparisons between groups were performed using Mann-Whitney U tests for continuous variables and χ² tests for categorical variables.

Table 2.

Newborn Demographics*

	No DM (N = 1607)	GDM (N = 226)	T1DM (N = 27)	T2DM (N = 67)	p value
Gestational age, wk	39.1 (38.1–39.9)	39.0 (37.4–39.4)	37.1 (35.1–38.9)	37.3 (35.4–39.1)	0.0001
Birth weight, kg	3.3 (3.0–3.7)	3.4 (3.0–3.7)	3.8 (3.2–4.2)	3.2 (2.6–3.7)	0.0143
Birth length, cm	50 (48–52)	51 (48–52)	52 (50–53)	50 (48–51)	0.0764
Apgar 1	8 (7–8)	8 (7–8)	7 (6–8)	8 (7–8)	0.0192
Apgar 5	9 (9–9)	9 (9–9)	8 (7–9)	9 (8–9)	0.0001
NICU admission	297 (18.5%)	63 (27.9%)	15 (55.6%)	36 (53.7%)	0.001
Tobacco exposure in utero	96 (6.0%)	9 (4.0%)	2 (7.4%)	6 (9.0%)	0.430
Sex					0.535
Female	48.2% (775)	47.8% (108)	37.0% (10)	53.7% (36)	—
Male	51.8% (832)	52.2% (118)	63.0% (17)	46.3% (31)	—
Heart rate metrics, bpm
Heart rate	138 (134–142)	138 (134–142)	141 (136–150)	141 (134–148)	0.0025
Heart rate variability	13 (11–16)	13 (11–16)	14 (11–19)	15 (12–18)	0.0102
BP metrics, mm Hg	n = 244	n = 49	n = 15	n = 36
MAP	52 (49–55)	53 (49–55)	55 (52–58)	54 (50–57)	0.0212
SBP	71 (67–75)	71 (65–75)	76 (72–80)	72 (69–79)	0.0020
DBP	42 (40–45)	43 (40–46)	44 (42–48)	43 (40–46)	0.1506
PP	28 (26–31)	27 (26–29)	31 (29–34)	31 (27–34)	0.0043
SBP variability	11 (9–12)	10 (8–12)	11 (10–12)	10 (8–12)	0.5633
DBP variability	9 (8–11)	10 (8–12)	11 (10–13)	10 (8–11)	0.0928

BP, blood pressure; bpm, beats per minute; DBP, diastolic blood pressure; DM, diabetes mellitus; GDM, gestational diabetes mellitus; MAP, mean arterial pressure; NICU, neonatal intensive care unit; PP, pulse pressure; SBP, systolic blood pressure; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus

^* Data presented as no. (%) for categorical variables and median (IQR) for continuous variables. Demographic and clinical characteristics for newborns in the overall cohort are shown, stratified by DM classifications. Statistical comparisons between groups were performed using Mann-Whitney U tests for continuous variables and χ² tests for categorical variables.

Newborn Heart Rate.

In unadjusted analyses using Kruskal-Wallis tests, newborn heart rate and heart rate variability were significantly different between DM groups (Table 2). After adjustment using IPWRA, no differences in newborn heart rate or heart rate variability were detected between no DM, GDM, T1DM, or T2DM groups (Figure 2). All covariates included in the IPWRA models demonstrated adequate balance, with weighted standardized differences within ±0.1; however, the standardized difference and variance ratio for NNM of obese status in T1DM analyses were −0.1 and 0.8, respectively. Maternal HbA1c was positively correlated with newborn heart rate (p = 0.009, R² = 0.27) but had no correlation with newborn heart rate variability (Figure 3). Maternal 1-hour OGTT was not correlated with newborn heart rate or heart rate variability (Figure 4).

Figure 2.

Impact of maternal diabetes on measures of newborn heart rate. The forest plot displays β coefficients (points) and 95% CIs (horizontal lines) for the association between DM and newborn heart rate (HR) and HR variability expressed in beats per minute (bpm). Estimates were obtained using inverse probability weighted regression adjustment adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index and advanced maternal age. Positive β coefficients indicate higher values in the diabetes mellitus (DM) groups relative to no DM. Statistical significance was defined as p < 0.05 as indicated by 95% CIs that do not cross the null value (β = 0; dashed line).

Figure 3.

Impact of maternal hemoglobin A1c (HbA1c) on measures of newborn heart rate (HR). Scatter plots with overlayed linear regression 95% CI margins for the association between maternal HbA1c and newborn (A) HR and (B) HR variability expressed in beats per minute (bpm). Regressions were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index, and advanced maternal age. Statistical significance was defined as p < 0.05.

Figure 4.

Impact of maternal 1-hr oral glucose tolerance test (OGTT) on measures of newborn heart rate (HR). Scatter plots with overlayed linear regression 95% CI margins for the association between maternal 1-hr OGTT and newborn (A) HR and (B) HR variability expressed in beats per minute (bpm). Regressions were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index, and advanced maternal age. Statistical significance was defined as p < 0.05.

Newborn Blood Pressure.

In unadjusted analyses using Kruskal-Wallis tests, there were significant differences in MAP, SBP, PP, and DBP variability between DM groups (Table 2). IPWRA analyses revealed that both T1DM and T2DM were associated with higher newborn SBP (+4.6 mm Hg, p = 0.002; +2.9 mm Hg, p = 0.03) and PP (+3.1 mm Hg, p = 0.006; +2.0 mm Hg, p = 0.02) compared with no DM (Figure 5). SBP variability was significantly higher in T1DM (+1.2 mm Hg, p = 0.048) and lower in T2DM (−1.3 mm Hg, p = 0.049) compared with no DM. No differences were detected between newborn MAP, DBP, or DBP variability among DM groups compared with no DM. GDM was not associated with any significant differences in BP measurements compared with no DM. All covariates included in the IPWRA models demonstrated adequate balance, with weighted standardized differences within ±0.1, although the standardized difference and variance ratio for NNM of overweight, obese, and hypertensive disorder of pregnancy status in T1DM analyses were 0.2, −0.2, 0.3 and 1.2, 0.7, 0.8, respectively. Maternal HbA1c positively correlated with newborn MAP (p = 0.004), SBP (p < 0.001), DBP (p = 0.03), and PP (p = 0.001) (Figure 6). The strongest correlation was observed between maternal HbA1c and newborn SBP (R² = 0.22). No correlations were observed between maternal HbA1c and newborn SBP variability or DBP variability. Maternal 1-hour OGTT was not correlated with any newborn BP measurement (Figure 7).

Figure 5.

Impact of maternal diabetes on measures of newborn blood pressure. The forest plot displays β coefficients (points) and 95% CIs (horizontal lines) for the association between DM and newborn mean arterial pressure (MAP), systolic blood pressure (SBP), diastolic blood pressure (DBP), pulse pressure (PP), and SBP and DBP variability expressed in mm Hg. Estimates were obtained using inverse probability weighted regression adjustment adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index, and advanced maternal age. Positive β coefficients indicate higher values in the diabetes mellitus (DM) groups relative to no DM. Statistical significance was defined as p < 0.05 as indicated by 95% CIs that do not cross the null value (β = 0; dashed line).

Figure 6.

Impact of maternal hemoglobin A1c (HbA1c) on measures of newborn blood pressure. Scatter plots with overlayed linear regression 95% CI margins for the association between maternal HbA1c and newborn (A) mean arterial pressure, (B) systolic blood pressure (SBP), (C) diastolic blood pressure (DBP), (D) pulse pressure, and (E) SBP and (F) DBP variability expressed in mm Hg. Regressions were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index, and advanced maternal age. Statistical significance was defined as p < 0.05.

Figure 7.

Impact of maternal 1-hr oral glucose tolerance test (OGTT) on measures of newborn blood pressure. Scatter plots with overlayed linear regression 95% CI margins for the association between maternal 1-hr OGTT and newborn (A) mean arterial pressure, (B) systolic blood pressure (SBP), (C) diastolic blood pressure (DBP), (D) pulse pressure, and (E) SBP and (F) DBP variability expressed in mm Hg. Regressions were adjusted for maternal hypertensive disorders of pregnancy, delivery mode, newborn sex, gestational age, maternal body mass index, and advanced maternal age. Statistical significance was defined as p < 0.05.

Discussion

To our knowledge, this is the first study to define and compare the specific associations of maternal DM subtypes with the cardiovascular function of offspring during the neonatal period. Our analyses indicate statistically significant elevations in SBP and PP in offspring born from mothers with T1DM and T2DM. Moreover, there were statistically significant positive correlations between maternal HbA1c and newborn heart rate, MAP, SBP, DBP, and PP. Contrary to our hypothesis, there were no notable associations between GDM and newborn cardiovascular measurements. This may be explainable by the heterogeneity of disease burden among mothers with GDM. Most frequently, GDM is diagnosed between the 24th and 28th week of pregnancy, but onset may occur outside of this window.³⁰ Furthermore, differences in testing procedures among providers, including decisions to apply GDM diagnoses after a >180 mg/dL 1-hour OGTT or escalate to a 3-hour OGTT, may influence the degree of glycemic burden represented within the GDM group. The OGTT values used to diagnose GDM represent only a snapshot of glucose control at the time of assessment, differing from HbA1c, which represents chronic hyperglycemic burden. This may also explain why 1-hour OGTT results were not correlated with newborn cardiovascular measurement. Our results suggest that the level and timing of chronic hyperglycemic burden in pregnancy may be a key determinant of the impacts on cardiovascular function in offspring of women with DM. These results complement and build upon other studies that have noted differences in BP measurements among an aggregate of offspring of women with DM during the early neonatal period.³¹ Our results also reinforce the adverse neonatal outcomes associated with this population observed in other studies. We similarly found higher rates of NICU admission and lower gestational age. While birth length was not shown to be affected by maternal DM, newborn weight was generally higher, with offspring from mothers with T1DM being the heaviest. This phenomenon may be explained by Pedersen’s hypothesis that maternal hyperglycemia leads to fetal hyperglycemia and hyperinsulinemia, thus leading to increased fetal adipose tissue due to the anabolic properties of insulin.³²

While we did not directly examine the effects of maternal glycemic control on newborn cardiovascular function, the observed correlations between maternal HbA1c and newborn HR and BP suggest that glycemic control may play an important role in maintaining newborn cardiovascular homeostasis. Lifestyle modification remains the first-line approach for managing hyperglycemia during pregnancy; however, marked insulin resistance may necessitate pharmacologic intervention. Injectable insulin is commonly used for glycemic control during pregnancy and distinguishes the 2 central classes of gestational diabetes: GDM-A1, managed by dietary adjustments, and GDM-A2, requiring pharmacologic therapy, most often exogenous insulin. Insulin remains the preferred pharmacologic agent during pregnancy given the associations between metformin use and preterm birth and glyburide use and NICU admission. Repeat OGTT is most frequently used to assess glycemic control and determine needs for pharmacologic intervention in GDM, however, our data demonstrate that OGTT is a potentially less reliable metric for assessing impacts on newborn cardiovascular function. Unlike T1DM and T2DM, current standards of care do not routinely include HbA1c measurement following GDM diagnosis, although incorporating HbA1c assessment may help to identify pregnancies at risk for adverse cardiovascular adaptions in offspring.

The underlying mechanisms of cardiovascular changes among offspring of women with DM remains unclear; however, studies of adult DM may shed light on the pathophysiology of BP elevation. In adults, chronic hyperglycemic burden is known to induce vascular stiffening, mediated by cross-linking of collagen fibrils and loss of elastin content in large elastic arteries, like the aorta.^33,34 These changes can dampen the Windkessel effect, responsible for sustaining optimal BP during both systole and diastole.³⁵ When vessels remodel in this way, there are significant elevations in SBP and PP, as we have seen in offspring from mothers with T1DM and T2DM.³³ Similarly, at the tissue level, studies have characterized changes in placental and vascular structure in pregnancies with DM, including increased placental collagen, disorganized vascular collagens, and decreased umbilical cord elastin content.^36–38 These changes may be driven in part by endothelial dysfunction, which has been demonstrated in offspring of women with DM human umbilical vein endothelial cell dysfunction, a key model of fetal endothelial cells.²¹ Endothelial cells serve as critical regulators of vascular hemodynamics, act as the frontline for circulating insults, and function as progenitors for various cardiovascular tissues during development.^21,39 The microenvironment in fetal circulation during a pregnancy with DM, which is comprised of high glucose, high insulin, pro-inflammatory cytokines, and pro-fibrotic factors, can drive endothelial dysfunction and contribute to vascular remodeling.²¹ Studies have shown that disruption of endothelial function can result in vascular stiffening and autonomic dysfunction, which may be responsible for the changes we observed in SBP variability in offspring from mothers with T1DM and T2DM.^40–42 This is reinforced by other studies demonstrating autonomic dysfunction in offspring from mothers with T1DM.⁴³ The exact mechanisms of endothelial dysfunction in these offspring are still under investigation, but there are emerging hypotheses, including the endothelial to mesenchymal transition, which has been linked to CVD progression in adult DM.^44–46 The phenotypic shift from healthy endothelial cells to athero-prone mesenchymal-like cells via endothelial to mesenchymal transition has been posited as a potential mediator for cardiovascular disease risk in offspring of women with DM per the developmental origins of health and disease theory.^21,47

The key strengths of this study include a large, well-characterized cohort, extensive timeframe of data collection, and application of rigorous analytical approaches. The data presented were obtained from all available EHRs retrievable through our institutional health informatics core, encompassing an 11-year period and over 1900 maternal-infant dyads. In particular, the study benefits from intentional filtering for high quality data and the use of a doubly robust, inverse probability-weighted regression and NNM approaches to reduce bias and strengthen casual inference. However, several important limitations warrant consideration. This analysis was conducted at a single site, and the sample size for maternal HbA1c and newborn BP analyses was relatively small, with only 132 total observations. While several statistically significant differences and correlations were detected, the possibility of type II error must be acknowledged, particularly given the absence of post-hoc power analyses. This limitation is most relevant to the T1DM subgroup (n = 27), where we observed consistent trends toward higher HR, HR variability, and DBP variability that did not reach statistical significance and may reflect insufficient power rather than true absence of effect. It is also noteworthy that most newborns included in the BP analyses were admitted to the NICU at some point during their hospital stay, as this is where these measurements are most routinely obtained (Supplemental Table S1). Although infants with identifiable congenital cardiac or renal diseases were excluded from analyses, the study population may not fully represent a healthy newborn cohort. Additionally, since the exact timing of newborn heart rate and BP measurements were not known, there is potential for skewing toward higher values during longer hospital stays due to physiologic ontogenic development of the cardiovascular system. This is a particular limitation for the heart rate analyses, where NICU admissions, which generally result in longer hospital stays, were higher among offspring from mothers with DM than those from mothers without DM. Future research would benefit from the inclusion of data from additional medical centers and larger sample sizes to help validate these findings. Additionally, applying causal mediation analysis to decompose the total effect of maternal diabetes on newborn cardiovascular function, distinguishing direct effects from indirect effects operating through mediators such as birth weight, would help identify the underlying pathways driving these associations. Furthermore, longitudinal studies focused on endothelial function and vascular health in offspring of women with DM could provide critical insights into the mechanisms driving increased cardiovascular risk and identify opportunities for early detection or intervention.

Conclusions

Offspring of mothers with T1DM and T2DM during pregnancy have higher SBP and PP during the neonatal period compared with those born from mothers without DM. Maternal HbA1c is positively correlated with newborn heart rate, SBP, DBP, and PP. Together, these data demonstrate that the impacts of maternal DM on offspring cardiovascular function can be observed during the neonatal period and suggest that they may be dependent on DM severity and timing of onset. We speculate the mechanisms of cardiovascular function changes may be attributable to newborn endothelial dysfunction and vascular remodeling, which is known to occur in offspring of mothers with DM. Our findings align with the developmental origins of health and disease theory and may link the known increased risk of cardiovascular diseases in this population to the intrauterine programming of cardiovascular dysfunction.

ABBREVIATIONS

BMI

body mass index;

CVD

cardiovascular disease;

DBP

diastolic blood pressure;

EHR

electronic health record;

GDM

gestational diabetes mellitus;

HbA1c

hemoglobin A1c;

HDP

hypertensive disorders of pregnancy;

IPWRA

inverse probability weighted regression adjustment;

MAP

mean arterial pressure;

NICU

neonatal intensive care unit;

NNM

nearest neighbor matching;

OGTT

oral glucose tolerance test;

PP

pulse pressure;

READi

Research, Education, Administration, and Development of Biomedical Informatics;

SBP

systolic blood pressure;

T1DM

type 1 diabetes mellitus;

T2DM

type 2 diabetes mellitus