Left ventricular remodelling and vascular adaptation to pregnancy in women with type 1 diabetes

Abstract

Background

Evidence regarding cardiovascular adaptation to pregnancy in women with pregestational diabetes is limited. Our study aimed to describe left ventricular (LV) remodelling and vascular adaptation to pregnancy in women with type 1 diabetes.

Methods

In this prospective cohort study, three consecutive cardiac MRI scans were conducted on age-matched and BMI-matched pregnant women with pregestational type 1 diabetes and pregnant women without diabetes. The scans were performed at gestational weeks 15–20, 26–30 and 34–37 from November 2020 to April 2023. Data collection and analysis included LV imaging results, brachial blood pressure and stroke volume derived functional vascular parameters.

Results

The study included 24 women with pregestational type 1 diabetes and 39 controls. Compared with controls, women with type 1 diabetes had significantly reduced LV end-diastolic volume index: 64.2±11.1 mL/m² vs 77.6±13.9 mL/m² (p<0.001), reduced end-systolic volume index: 26.0±7.3 mL/m² vs 33.5±7.7 mL/m² (p=0.003), increased concentricity: 0.84±0.13 g/mL vs 0.68±0.10 g/mL (p<0.001), reduced stroke volume index: 38.2±7.2 mL/m² vs 44.1±8.4 mL/m² (p=0.008), reduced cardiac index: 3.37±0.55 L/min/m² vs 3.62±0.51 L/min/m² (p=0.046), reduced global longitudinal strain: −13.5±2.3% vs −15.2±2.1% (p=0.04), increased myocardial T1 values: 998±28 ms vs 983±25 ms (p=0.03), increased systolic blood pressure: 128.1±7.8 mmHg vs 117.4±11.1 mmHg (p=0.007), increased mean arterial pressure: 94.8±6.7 mmHg vs 88.1±9.7 mmHg (p=0.03), increased total peripheral vascular resistance: 28.9±5.3 mmHg·min·m²/L vs 24.7±4.0 mmHg·min·m²/L (p=0.001), and reduced total arterial compliance: 0.79±0.19 mL/m²/mmHg vs 1.02±0.21 mL/m²/mmHg (p<0.001).

Conclusion

Our study provides evidence of impaired LV remodelling and suboptimal vascular adaptation to pregnancy in women with type 1 diabetes when compared to women without diabetes. The clinical implications of these findings, particularly their association with the development of later cardiovascular disease, require further investigation.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Cardiovascular adaptation during pregnancy is well documented and can be considered a physiological stress test for the cardiovascular system.

• However, while diabetes is a recognised risk factor for cardiovascular disease, the specific nature of cardiovascular adaptations in pregnancies complicated by pregestational diabetes mellitus remains poorly understood.

WHAT THIS STUDY ADDS

• This study provides evidence of impaired left ventricular remodelling and inferior vascular adaptation to pregnancy in women with type 1 diabetes when compared with women without diabetes.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study not only provides new insights into the intersection of diabetes, pregnancy and cardiovascular health but also establishes a foundation for further research that could lead to an optimised clinical practice for women with type 1 diabetes at risk for heart failure.

Introduction

The maternal cardiovascular system undergoes significant changes during pregnancy to ensure sufficient supply of oxygen and nutrients to the placenta and growing fetus.¹ Cardiovascular adaptation to pregnancy is well described and may be regarded as a cardiovascular stress test due to a massive volume overload reaching a peak increase in plasma volume in the third trimester by 48%.² The maternal heart adapts to volume overload through left ventricular (LV) remodelling, which increases cardiac output by 31%.³ However, the nature of cardiovascular adaptation to pregnancy complicated by pregestational diabetes mellitus is still largely unknown.

Diabetes mellitus is associated with an increased lifetime risk of heart failure (HF) even when accounting for HF risk factors such as coronary heart disease, blood pressure, body weight, cholesterol levels and age.⁴ The pathophysiology of diabetes-related HF, known as diabetic cardiomyopathy, is not fully understood and often remains undiagnosed, as the initial stage with LV diastolic dysfunction occurs subclinically.5,7 These changes are thought to result from a combination of reduced glucose uptake, increased fatty acid use and the build-up of harmful by-products such as reactive oxygen species and advanced glycation end products. This contributes to inflammation, fibrosis and impaired myocardial function.⁵ As diabetic cardiomyopathy can be challenging to recognise, sensitive diagnostic tools are required.⁸ Furthermore, as the initial stage of diabetic cardiomyopathy may be reversible, timely intervention is crucial.^{6 9}

Considering pregnancy as a window of opportunity to uncover early stage cardiovascular disease, we hypothesise that pregnant women with type 1 diabetes exhibit subclinical signs of LV dysfunction and vascular maladaptation. Therefore, the present study aims to describe maternal LV remodelling and vascular adaptation during type 1 diabetes pregnancies using cardiac magnetic resonance imaging (CMR), the gold-standard non-invasive technique for assessing cardiac structure and function.¹⁰

Methods

Study population

This cohort study was conducted at Aalborg University Hospital from November 2020 to April 2023 as part of the fetal growth and placental function in pregestational diabetes cohort. All participants gave written and oral informed consent. Furthermore, the study was registered at ClinicalTrials.gov Identifier: NCT04801121.

Participants were included at their routine first trimester fetal ultrasound scan. Exclusion criteria were age <18 years, multiple pregnancies, fetal malformations, abnormal karyotype, gestational diabetes, maternal height from spine to chest >43 cm, severe claustrophobia, or any other contraindications to CMR and inability to read or understand Danish. Women with known cardiovascular disease were not excluded from the study. Women with type 1 diabetes were diagnosed before pregnancy according to the International Classification of Diseases.

The participants in this study were not engaged in the research design. However, the potential burden of the relatively time-intensive examinations was carefully considered. Specifically, for women who had pre-existing appointments for diabetes management with an obstetrician, the examinations were intentionally scheduled on the same day as their routine diabetes assessments to minimise the additional burden on the participants.

Anthropometry and biochemical marker assessment

Maternal characteristics were obtained from the patient records. At each examination, mean systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured in the supine position based on three measurements after at least 5 min rest. Mean arterial pressure was calculated by adding DBP and a third of pulse pressure. Furthermore, glycated haemoglobin (HbA1c) was analysed at each visit. An oral glucose tolerance test was made in controls at gestational age (GA) 28 weeks to rule out gestational diabetes. Gestational diabetes was defined as a 2-hour blood glucose value >9.0 mmoL/L, and controls diagnosed with gestational diabetes were excluded from the study.

CMR protocol

Maternal CMR was performed at three visits during pregnancy; GA 15–20 weeks, 26–30 weeks and 34–37 weeks denoted as first visit, second visit and third visit, respectively. All scans were performed on a 1.5 T scanner system (Optima MR450w, GE Healthcare, Milwaukee, Wisconsin, USA). The woman was placed in a left lateral position to avoid aortocaval compression. The CMR scans were ECG-gated and obtained during breath holds. No sedatives or contrast agents were used. After localising images, balanced steady-state free-precession cine images including short-axis stack, standard four-chamber, three-chamber and two-chamber views were obtained. Slices of 8 mm thickness with a 2-mm gap from the base through the apex of the heart were used.

The images were acquired with a typical flip angle of 45°, echo time 1.5 ms, repetition time 3.5 ms, matrix size 224 × 224, field of view 420 × 420 mm, 20 views per segment and 40 reconstructed phases. Native T1 mapping images were acquired in a midventricular short-axis slice using modified Look-Locker inversion recovery sequence with a 5–(3)−3 scheme, slice thickness 8 mm, flip angle 35°, field of view 360 × 360 mm.

CMR analyses

Cine images were analysed offline using the cvi42 V.5.14 software (Circle Cardiovascular Imaging, Canada). The analyses were performed by a trained research assistant (ÁWO) and supervised by an experienced cardiologist specialised in CMR (TZ). The images were anonymised and CMR analyses were therefore blinded to any clinical information, in particular, diabetes status and GA.

Endocardial and epicardial borders of LV were traced automatically and corrected manually as needed in short-axis views with four-chamber and two-chamber images used as references. End-diastolic phase and end-systolic phase were visually identified as the largest and smallest chamber areas on the midventricular slice, respectively. The LV cavity was obtained by tracing endocardial borders in each slice at end diastole and end systole. LV wall volume was obtained by tracing endocardial and epicardial borders at end diastole (figure 1). Measurements from each slice were summed using the method of disks. Myocardial mass was estimated by multiplying the myocardial wall volume at end diastole by the density of myocardium (1.05 g/mL). The trabeculae and papillary muscles were included in the blood volume. Basal slices were excluded when myocardium was present in less than 50% of the short-axis circumference. Global longitudinal strain (GLS) was calculated from balanced steady-state free-precession images in two-chamber and four-chamber views using a feature tracking algorithm. Native T1 mapping values were obtained from the midventricular septum in short-axis view by manually tracing the region of interest in the interventricular septum.

Figure 1. Illustration of endocardial and epicardial tracing of the left ventricle during diastole from base to apex in short-axis view. The first basal slice is excluded due to less than 50% of the myocardial presence in the circumference. Trabeculae and papillary muscles are included in the blood volume. Green = epicardium. Red = endocardium.

Cardiac remodelling was analysed with regard to concentricity defined as LV mass (LVM) to LV end-diastolic volume (LVEDV) ratio.¹¹ Concentric remodelling was defined as an increase in concentricity while eccentric remodelling was defined as a decrease in concentricity. Also, vascular parameters derived from LV stroke volume (LVSV) including total peripheral vascular resistance, total arterial compliance, effective arterial elastance and stroke work were calculated as follows:¹²

$$Totalarterialcompliance=\frac{LVSV}{Pulsepressure}$$

$$Effectivearterialelastance=\frac{MAP}{LVSV}$$

$${\displaystyle Strokework=MAP\times LVSV}$$

Given the possible implications of maternal habitus, LVM and LV volumes were indexed to body surface area yielded from the baseline height and weight.

Statistical analysis

Normally distributed variables were reported as mean and SD while categorical variables were reported as number and proportion. An independent sample t-test was used to test the differences in means of GA between groups at each visit. Multiple linear regression adjusted for tobacco smoking and GA at CMR scan was used to test the differences in means of continuous variables between groups at each visit. Fisher’s exact test was used to test the independent proportions between groups at each visit. Participants with missing data were excluded from the related analysis but contributed to the remaining analyses if data were available. For all analyses, values of p<0.05 were considered statistically significant. Statistical analyses were performed using R V.4.2.1. A power calculation was not performed prior to study initiation, as the sample size was determined based on feasibility within a fixed recruitment period.

Results

A total of 64 pregnant women were initially included. Pregnant women without diabetes were denoted as controls. One control was excluded due to a diagnosis of gestational diabetes at gestational week 28. This resulted in a total of 63 study subjects: 24 with pregestational type 1 diabetes and 39 controls providing a total of 161 CMR scans. In the type 1 diabetes group, 14 (58.3%) completed all three CMR scans. Among the controls, 30 (76.9%) completed all three scans. The number of scans per visit for the type 1 diabetes group was 24, 20 and 14 at the first, second and third visits, respectively. The number of scans per visit for the controls was 39, 34 and 30 at the first, second and third visits, respectively. Reasons for not completing all three CMR scans were preterm delivery and logistical challenges mainly due to CMR unavailability. A few CMR slices in the control group were excluded due to artefacts resulting in exclusion of one GLS analysis at the first visit. Likewise, a few T1 mapping analyses were excluded from the control group data at the first, second and third visits due to artefacts resulting in 2, 2 and 1 missing GLS analyses, respectively.

The clinical characteristics of the study population are shown in table 1. There were no significant differences in age or body mass index between groups; however, there was a higher proportion of tobacco smokers among women with type 1 diabetes compared with the control group (42% vs 15%, p=0.04).

Table 1. Clinical characteristics of the study population at baseline.

Characteristic	Type 1 diabetes (n=24)	Controls (n=39)	P value
Age, mean±SD, years	29.7±4.3	29.8±4.3	0.91
BMI, mean±SD, kg/m2	29.0±4.7	28.4±6.5	0.63
BSA, mean±SD, m2	1.87±0.17	1.88±0.17	0.81
HbA1c, mean±SD, mmoL/moL	51.3±10.0	30.5±3.3	<0.001
Diabetes duration, mean±SD, years	14.7±8.6	–	–
Multiparous, n (%)	13 (54)	21 (54)	0.99
Tobacco smoker, n (%)	10 (42)	6 (15)	0.04

Groups are compared by independent sample t-test (continuous data) and Fisher’s exact test (binary data).

BMI, body mass index; BSA, body surface area; HbA1c, glycated haemoglobin.

Table 2 shows indexed data for LV, blood pressures and functional vascular parameters in type 1 diabetes pregnancy compared with controls adjusted to tobacco smoking and GA.

Table 2. Comparison of cardiac and vascular parameters through pregnancy between women with type 1 diabetes and controls.

Parameters	First visit			Second visit			Third visit
	Type 1 diabetes (n=24)	Controls (n=39)	P value	Type 1 diabetes (n=20)	Controls (n=34)	P value	Type 1 diabetes (n=14)	Controls (n=30)	P value
GA (weeks)	17.0±1.2	17.1±1.1	0.73	28.4±0.9	28.5±0.6	0.53	34.7±0.6	35.4±0.9	0.006
HbA1c (mmoL/moL)	51.3±10.0	30.5±3.3	<0.001	47.1±11.2	29.5±3.3	<0.001	46.0±8.0	32.0±3.7	<0.001
Cardiac parameters
LVM (g/m2)	47.8±5.7	47.7±6.3	0.98	51.4±6.4	51.2±6.6	0.81	53.1±6.7	52.1±8.3	0.77
LVEDV (mL/m2)	77.8±10.4	83.0±12.8	0.16	68.6±10.4	80.1±13.6	0.02	64.2±11.1	77.6±13.9	<0.001
LVESV (mL/m2)	30.5±6.5	33.3±5.6	0.05	27.5±6.0	32.9±7.4	0.05	26.0±7.3	33.5±7.7	0.003
LVSV (mL/m2)	47.3±6.5	49.7±9.7	0.56	41.1±6.4	47.2±8.1	0.03	38.2±7.2	44.1±8.4	0.008
LVEF (%)	61.0±5.0	59.6±5.1	0.11	61.0±5.0	59.2±5.1	0.84	59.7±6.2	57.0±5.3	0.27
GLS (%)*	−17.8±1.6	−17.5±2.4	0.68	−15.7±2.5	−16.4±2.6	0.21	−13.5±2.3	−15.2±2.1	0.04
CI (L/min/m2)	3.60±0.46	3.62±0.43	0.63	3.71±0.52	3.76±0.55	0.87	3.37±0.55	3.62±0.51	0.046
HR (bpm)	76.8±9.7	74.7±13.3	0.61	91.0±11.2	81.3±14.0	0.04	89.1±12.1	84.1±14.2	0.26
Concentricity (g/mL)	0.62±0.07	0.58±0.88	0.16	0.76±0.10	0.65±0.11	0.01	0.84±0.13	0.68±0.10	<0.001
T1 value (ms)†	996±32	987±27	0.22	1005±33	980±26	0.004	998±28	983±25	0.03
Brachial blood pressures
SBP (mmHg)	120.0±11.3	118.0±10.4	0.4	123.5±9.1	114.0±9.7	<0.001	128.1±7.8	117.4±11.1	0.007
DBP (mmHg)	71.6±8.4	70.0±10.0	0.5	70.7±9.0	67.1±8.4	0.08	78.2±8.0	73.4±10.1	0.12
MAP (mmHg)	87.7±8.3	86.0±8.5	0.39	88.3±7.5	82.7±7.9	0.006	94.8±6.7	88.1±9.7	0.03
PP (mmHg)	48.4±9.7	48.0±11.9	0.81	52.8±10.6	46.9±8.7	0.049	49.9±8.9	44.0±8.3	0.09
Derived vascular functional parameters
TAC (mL/m2/mmHg)	1.01±0.22	1.07±0.26	0.54	0.81±0.18	1.03±0.20	0.001	0.79±0.19	1.02±0.21	<0.001
TPVR (mmHg·min·m2/L)	24.7±4.1	24.2±4.2	0.89	24.2±3.8	22.4±3.9	0.11	28.9±5.3	24.7±4.0	0.001
Ea (mmHg·m2/mL)	1.89±0.35	1.83±0.54	0.85	2.19±0.36	1.82±0.44	0.006	2.55±0.45	2.09±0.55	0.002
SW (mmHg·mL/m2)	4137±612	4228±698	0.95	3639±696	3892±701	0.54	3627±709	3872±796	0.14

^*n = 38, 34 and 30 at first, second and third visits, respectively, in the control group.

^†n = 37, 32 and 29 at first, second and third visits, respectively, in the control group.

CI, cardiac index; DBP, diastolic blood pressure; Ea, effective arterial elastance; GA, gestational age; GLS, global longitudinal strain; HbA1c, glycated haemoglobin; HR, heart rate; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVM, left ventricular mass; LVSV, left ventricular stroke volume; MAP, mean arterial pressure; PP, pulse pressure; SBP, systolic blood pressure; SW, stroke work; T1, myocardial T1 time; TAC, total arterial compliance; TPVR, total peripheral vascular resistance.

LV parameters

Figures2AH and 3A,B demonstrate differences in LV structural and functional parameters between groups. No significant differences between groups regarding LV parameters were observed at the first visit. In type 1 diabetes pregnancy, there was an increase in LVM through gestation. However, no significant difference between the groups was observed (figure 2G). Meanwhile, at the second and third visit, there was a significantly higher degree of concentric hypertrophy in type 1 diabetes pregnancy compared with controls: 0.84±0.13 g/mL vs 0.68±0.10 g/mL (p<0.001, third visit, figure 2H). The higher level of concentricity was a result of a significantly reduced LVEDV in type 1 diabetes pregnancy at the second and third visits when compared with controls: 64.2±11.1 mL/m² vs 77.6±13.9 mL/m² (p<0.001, third visit, figure 2A). Myocardial T1 value was significantly higher in type 1 diabetes pregnancy at the second and third visits when compared with controls: 998±28 ms vs 983±25 ms (p=0.03, third visit, figure 3B).

Figure 2. LV structural and functional parameters at the first, second and third visits conducted at 15–20 weeks, 26–30 weeks and 34–37 weeks of gestation, respectively. Groups are compared by multiple linear regression adjusted for tobacco smoking and gestational age at CMR. Data are indexed to body surface area and expressed as means and SD. CMR, cardiac magnetic resonance imaging; LV, left ventricular.

Figure 3. Global longitudinal strain and native T1 mapping at the first, second and third visits conducted at 15–20 weeks, 26–30 weeks and 34–37 weeks of gestation, respectively. Groups are compared by multiple linear regression adjusted for tobacco smoking and gestational age at CMR. Data are expressed as means and SD. CMR, cardiac magnetic resonance imaging.

LVSV was significantly lower in type 1 diabetes pregnancy than in controls at the second and third visits: 38.2±7.2 mL/m² vs 44.1±8.4 mL/m² (p=0.008, third visit, figure 2C). Accordingly, cardiac index was significantly reduced in type 1 diabetes pregnancy at the third visit: 3.37±0.55 L/min/m² vs 3.62±0.51 L/min/m² (p=0.046, figure 2E).

No difference was seen between groups regarding systolic function reflected in LV ejection fraction (LVEF). However, GLS was reduced in type 1 diabetes pregnancy compared with controls: −13.5±2.3% vs −15.2±2.1% (p=0.04, figure 3A) at the third visit.

Vascular parameters

Vascular parameters are demonstrated in figure 4A–H. No statistically significant difference was observed between groups with regard to vascular parameters at the first visit. Pulse pressure was significantly increased at the second visit: 52.8±10.6 mmHg vs 46.9±8.7 mmHg (p=0.049, figure 4D) and although not significant, likewise, there was a trend towards a higher pulse pressure in the third visit: 49.9±8.9 mmHg vs 44.0±8.3 mmHg (p=0.09). SBP was significantly increased in type 1 diabetes pregnancy compared with controls at the second and third visits: 128.1±7.8 mmHg vs 117.4±11.1 mmHg (p=0.007, third visit, figure 4B). Also, total arterial compliance was significantly reduced at the second and third visits: 0.79±0.19 mL/m²/mmHg vs 1.02±0.21 mL/m²/mmHg (p<0.001, third visit, figure 4E) while total peripheral vascular resistance was significantly increased at the third visit: 28.9±5.3 mmHg·min·m²/L vs 24.7±4.0 mmHg·min·m²/L (p=0.001, figure 4G). Likewise, effective arterial elastance was significantly increased at the second and third visits: 2.55±0.45 mmHg·m²/mL vs 2.09±0.55 mmHg·m²/mL (p=0.002, third visit, figure 4F). There was no statistically significant difference in stroke work between the groups.

Figure 4. Brachial blood pressures (BPs) and derived functional vascular parameters at the first, second and third visit conducted at 15–20, 26–30 and 34–37 weeks of gestation, respectively. Groups are compared by multiple linear regression adjusted for tobacco smoking and gestational age at CMR. Functional vascular parameters are indexed to body surface area. Data are expressed as means and SD. CI, cardiac index; CMR, cardiac magnetic resonance imaging; PP, pulse pressure; SVi, stroke volume index.

Correlation analysis

A Pearson correlation analysis was conducted to assess the association between T1 values and HbA1c, which showed no statistically significant correlation (r = –0.13, p=0.33).

Discussion

In this cohort study, we observed significant differences in LV remodelling and vascular adaptation between women with type 1 diabetes and controls during pregnancy. However, no significant differences in cardiovascular parameters were observed between groups in early pregnancy, suggesting that there was no difference in cardiovascular status between groups prior to pregnancy. In mid and late pregnancy, women with type 1 diabetes developed significantly reduced LV volumes when compared with controls. Yet, no significant difference was observed in LVM between groups. Accordingly, women with type 1 diabetes developed a significantly higher degree of LV concentricity. Moreover, we observed significant differences in blood pressures and functional vascular parameters in mid and late pregnancy, including pulse pressure, SBP, total arterial compliance, effective arterial elastance and total peripheral vascular resistance, suggestive of a reduced capacity to compensate for volume overload during pregnancy in women with type 1 diabetes.

LV remodelling

Prior research on cardiac remodelling in pregnant women with type 1 diabetes is limited. Airaksinen et al reported a reduced echocardiographic LV end-diastolic diameter in 17 pregnancies with type 1 diabetes compared with 11 non-diabetic pregnancies.¹³ Our observation of reduced LVEDV is in line with the diametric findings of Airaksinen et al. However, unlike our study, they reported a reduction in LVM in type 1 diabetes pregnancy.¹³ The discrepancy between these two studies regarding LVM may stem from differences in assessment methods. As noted by Ducas et al, echocardiography tends to underestimate LVM compared with CMR.¹⁰ This is further supported by Bellenger et al, who demonstrated that CMR also provides significantly higher reproducibility than echocardiography.¹⁴

LV concentricity was not addressed in the study by Airaksinen et al and, to our knowledge, has not been previously described in pregnancies with type 1 diabetes. Thus, the increased concentric hypertrophy during type 1 diabetes pregnancy is a novel finding. While the underlying pathophysiology remains unknown, the higher concentricity results from reduced LVEDV, which may point to decreased LV compliance with subsequent reduced filling, as seen in the early phase of diabetic cardiomyopathy.¹⁵

In this study, we found no differences in LVEF between the groups. LVEF is a standard cardiac parameter for evaluating LV systolic function, reflecting the overall contractile capacity of the LV. However, LVEF does not differentiate between the contractility of circumferential and longitudinal myocardial fibres. A reduction in the contractile function of the longitudinal fibres is captured by GLS, which may precede dysfunction in the circumferential fibres without affecting LVEF.¹⁶ Notably, we found reduced GLS in type 1 diabetes pregnancies. Furthermore, we observed increased T1 mapping values in the type 1 diabetes group compared with controls. A recent study by Seno et al highlighted a correlation between GLS and T1 mapping in predicting all-cause mortality or HF hospitalisation in patients without signs of coronary artery disease, independent of LVEF.¹⁷ Therefore, our findings of reduced GLS and elevated T1 values in type 1 diabetes pregnancies may suggest early indicators of diabetes-associated HF.

Vascular adaptation

Efficient cardiovascular regulation relies on well-functioning vasculature.¹⁸ In our study, we discovered several vascular parameters that significantly differed between type 1 diabetes pregnancies and controls. Specifically, we observed an increase in pulse pressure in type 1 diabetes pregnancies. Pulse pressure is an indicator of arterial elasticity, and elevated pulse pressure reflects arterial stiffness, leading to reduced arterial compliance. In normal pregnancies, high arterial compliance helps to accommodate increased plasma volume without a rise in blood pressure.¹⁹ However, in type 1 diabetes pregnancies, we found reduced total arterial compliance and increased pulse pressure, suggesting that women with type 1 diabetes have impaired compensation for volume overload during pregnancy. Additionally, we observed an increase in total peripheral vascular resistance in type 1 diabetes pregnancies. Reduced total arterial compliance and elevated total peripheral vascular resistance are commonly associated with arterial hypertension.^{18 20} However, we did not observe hypertension in this cohort. Nonetheless, women with type 1 diabetes had significantly higher SBP within the normal range when compared with controls during mid and late pregnancy, which may have contributed to the increased concentric hypertrophy in this group. In all regards, these findings suggest that pregnancy may unmask subclinical arterial dysfunction in women with type 1 diabetes.

Hyperglycaemia and T1 mapping

Poor glycaemic control is associated with a fourfold higher risk of HF in the non-pregnant type 1 diabetes population when compared with optimal glycaemic control, even when adjusted for other cardiovascular risk factors.²¹ Interestingly, most women with type 1 diabetes in the current study failed to meet the recommended glycaemic level.²² A systematic review by Salvador et al identified a significant correlation between HbA1c levels and myocardial fibrosis (MF) in patients with diabetes assessed by CMR techniques.²³ In this study, we used native T1 mapping to detect myocardial changes, including MF.^{24 25} MF is associated with reduced myocardial compliance, potentially impairing cardiac function.²³ However, we did not observe an association between HbA1c and native T1 values in our cohort, suggesting that the observed myocardial tissue alterations may not be directly related to glycaemic control in this study population.

While elevated native T1 may indicate MF, it can also reflect interstitial oedema or water retention, especially in the context of pregnancy. A previous study by Nii et al found no significant increase in native T1 during healthy pregnancy, supporting the idea that pregnancy alone may not drive fibrosis-like changes.²⁶

In our study, although the absolute native T1 values in both groups were generally within or near the expected normal range for healthy myocardium, we observed statistically significant differences between the type 1 diabetes and control groups. Furthermore, native T1 values may be influenced by technical factors, including motion artefacts. We mitigated this by excluding data sets with severe artefacts, but minor motion-induced variation cannot be entirely ruled out.

Given the small sample size, particularly in the third trimester subgroup, the findings should be interpreted with caution. The relative group differences suggest a possible myocardial tissue alteration in women with type 1 diabetes, but these may not necessarily reflect fibrosis in a pathological sense. Further research is needed to determine whether these observed cardiac changes return to baseline postpartum.

Overall, our observations in type 1 diabetes pregnancy align with previous literature on the early pathophysiology of diabetic cardiomyopathy. Initial stages of diabetic cardiomyopathy are characterised by MF, hypertrophy and diastolic dysfunction with preserved LVEF and reduced GLS, potentially progressing to systolic dysfunction and overt HF.¹⁵ Thus, our findings suggest that the cardiac alterations seen in pregnant women with type 1 diabetes reflect an early stage of diabetic cardiomyopathy.

Strengths and limitations

The primary strength of the study is the use of CMR, the gold standard for assessing cardiac mass and volumes.¹⁴ Another strength is the longitudinal study design, with three CMR scans performed in each woman at different stages of pregnancy. Moreover, to ensure objective and unbiased CMR analysis, the CMR reader was blinded to both diabetic status and GA at the time of CMR.

A limitation is the absence of prepregnancy and postpartum CMR data. Although the first scan at GA 15–20 weeks showed no significant differences between groups, suggesting similar cardiovascular status prior to pregnancy, we cannot confirm whether group differences existed before pregnancy. Furthermore, it remains uncertain whether observed changes in LV function and vascular parameters are reversible postpartum. Another limitation is the relatively small sample size, especially in the diabetic group at the third visit, though statistically significant differences were identified between groups. Women with known cardiovascular pathology were not excluded, which may introduce heterogeneity but reflects real-world clinical populations.

Conclusion

Maternal LV remodelling during type 1 diabetes pregnancy is marked by reduced LV volumes and function, while vascular adaptation is characterised by decreased arterial compliance and elevated arterial pressures compared with non-diabetic pregnancy. These findings suggest that pregnancy reveals subclinical cardiovascular dysfunction in women with type 1 diabetes. Future studies are warranted to investigate the possible link between these deviations and later development of cardiovascular disease in women with type 1 diabetes.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.