Fetal aortic intima-media thickness as an early predictor of preeclampsia in patients with pregestational diabetes

Introduction

Preeclampsia (PE) is a complication that usually appears in the second half of pregnancy. Although the exact pathomechanism remains unclear, it is characterised by impaired endothelial function and incomplete remodeling of spiral arteries, which are crucial for feto-maternal circulation [1]. In developed countries, PE affects 3-5% of first pregnancies. Hypertensive disorders of pregnancy, of which PE is an example, represent the leading cause of maternal mortality and morbidity [2, 3]. Based on the onset of the disease, it is possible to distinguish early-onset PE (< 34 weeks) and late-onset PE (≥ 34 weeks). Early-onset PE is associated with poorer neonatal outcomes and with a higher incidence of prematurity. However, it is late-onset PE that occurs more frequently and is less likely to be predicted [4, 5]. The results of the ASPRE study indicate that aspirin is effective in the prevention of pre-eclampsia in high-risk patients; however, it was found to be ineffective in terms of term PE [6, 7].
If pregnancy is complicated by pregestational diabetes, the risk of PE is 3 times higher than in the background population [8-10]. Additionally, in patients with long-term diabetes, proteinuria might also be present as a complication of diabetes not associated with placental or endothelial dysfunction. Hence, the question that frequently arises within this group of patients is whether the symptoms point to the development of PE or to the progression of proteinuria in the course of kidney disease [11, 12]. Moreover, diabetic patients are more likely to develop term PE; therefore, aspirin prophylaxis was found to be ineffective in this group of patients [11, 13].
The abovementioned lower detection rate of late-onset PE and the lack of effective prevention methods necessitate the development of new screening tools and diagnostic approaches to achieve a better understanding of its pathomechanism. Ultrasound evaluation provides information with regard to the potential fetal distress. Nevertheless, abnormalities in the Doppler examination usually occur as a manifestation of an already-existing imbalance between fetal demands and placental potential. With the improvement of ultrasonography in obstetrics, novel parameters may be assessed, and it is possible to evaluate their applicability in predicting adverse neonatal and maternal outcomes.
Cardiologists employ ultrasound measurement of intima-media thickness (IMT) in the carotid artery as a risk factor for cardiovascular disease (CVD) [14, 15]. However, recent studies additionally demonstrate the potential role of the fetal IMT as an early predictor of the subsequent placental insufficiency [16, 17], where the most accessible vessel for measuring IMT is the fetal abdominal aorta (aIMT) [18].
Our study aimed to assess aIMT performance as an early predictor of PE and fetal growth restriction (FGR) in patients with pregestational diabetes (PGDM), who are particularly prone to developing placental complications. Therefore, an effective, non-invasive method of predicting adverse placental function is essential for both the timely diagnosis and the treatment.

Material and methods

Subjects

A nested case-control study was conducted, involving 81 patients with PGDM. All women following delivery were classified either into the PE group (n = 18) or to the control group (n = 63), based on the pregnancy results obtained. Patients were recruited between December 2019 and December 2022. Patients were all Caucasian and received standard pregnancy care for diabetes, as recommended by the Polish Diabetes Association and Polish Gynaecological Society, targeting a fasting glucose level of 3.8-5.0 mmol/l, 1-h postprandial glucose below 7.0 mmol/l, and glycated haemoglobin (HbA1c) below 6.5% (45.5 mmol/mol) in the first trimester and below 6.0% (42 mmol/mol) in the second and third trimester [19]. All the women were administered with intensive insulin therapy, multiple daily insulin (MDI) injections, or continuous subcutaneous insulin infusions (CSII). According to the Polish recommendations, all patients in our group received 150 mg of aspirin daily from the 12^th-36^th week of gestation to prevent PE [19]. The therapeutic blood pressure target in women diagnosed with hypertensive disorders of pregnancy was blood pressure < 135/85 mm Hg [19].
Data regarding personal, obstetric, and familial history were obtained, including smoking, age, and parity during the first hospital appointment. Gestational age was adjusted to the crown-rump length measured before the 14^th week of gestation if necessary.
Following the delivery, information regarding the obstetrical outcome, complications, delivery, and neonatal data was supplemented in our database.
Exclusion criteria comprised multiple pregnancies, a history of gestational hypertension or PE, major fetal abnormalities, or aneuploidy.
In order to diagnose PE, the criteria provided by the International Society for Study of Hypertension in Pregnancy (ISSHP) [20] were employed, including adjustments for patients with diabetic kidney disease [11]: in patients who had not been previously diagnosed with chronic hypertension (n = 79) [20] – systolic blood pressure (BP) ≥ 140 mm Hg, or diastolic BP ≥ 90 mm Hg on 2 specific instants, which occurred for the first time after 20 weeks of gestation, and one of the following complications with the onset in the second half of pregnancy: (1) proteinuria (≥ 300 mg/24 h or >100% increase in proteinuria in proteinuric patients); (2) serum creatinine > 1 mg/dl (>90 μmol/l) or > 50% increase in serum creatinine within 7 days; (3) elevation of transaminase levels > 40 IU/l; (4) neurological complications (eclampsia, altered mental status, blindness, stroke, clonus, severe headache, persistent visual scotomata); (5) haematological complications (thrombocytopaenia <150 G/l, disseminated intravascular coagulation, haemolysis); and (6) uteroplacental dysfunction (FGR, proteinuric patients placental insufficiency) [21].
At least one of the abovementioned criteria had to be present for patients without previous chronic hypertension, except for uteroplacental dysfunction, according to ISSHP [20]. Fetal growth restriction was diagnosed according to the Delphi consensus criteria: estimated fetal weight or abdominal circumference below the third centile or absent end-diastolic flow in the umbilical artery (UmA) or EFW/AC below the tenth centile together with abnormalities in Doppler ultrasound evaluation [21].

Monitoring of laboratory and clinical measurements

The following laboratory tests were performed in each trimester in all the patients: HbA1c and concentration of serum triglycerides, as described in our previous paper [22]. Blood samples for all the routine analyses were collected following overnight fasting and immediately transported to the accredited university hospital laboratory with the ISO 9000 quality management certification.
In terms of the clinical characteristics, they included maternal age, height, pre-gestational weight, body weight before (≤ 7 days) delivery, weight gain, parity, smoking status, and type and duration of diabetes, which are presented in Table 1, and the data concerning the delivery (neonatal results, timing, cord blood gas analysis, placental weight) are presented in Table 2. First anthropometric measurements (height, weight, and body mass index [BMI]) and blood pressure measurements were performed at the onset of the study. Blood pressure during the first hospitalisation was measured 4 times daily, and the mean systolic and diastolic values were recorded in the study data.

Ultrasound examination

One skilled operator (DB) performed every ultrasound scan using an ultrasound machine equipped with 12L-RS linear probe 4-12MHz (Voluson S10 Expert, GE Medical Systems, Chicago, IL, USA). Routine ultrasound scans were performed during regular appointments in the study group at the 26^th-28^th, 32^nd-34^th, and around the 38th week of gestation.
All aIMT measurements were obtained in the fetal aorta in the coronal view between the renal and iliac arteries on a frozen image during the end-diastole, using cine-loop capability. The position of the fetus was first verified with a C2-9-RS curved array transducer (2.5-9.1 MHz), then fetal biometry was assessed, estimating fetal weight using the Hadlock 3 formula. Subsequently, Doppler velocimetry of the UmA, medial cerebral artery (MCA), and uterine artery on both arteries were evaluated. Ultrasound software automatically calculated pulsatility index (PI), PI in UmA, MCA, and the mean PI in uterine arteries. In the course of the study, insonation of the abovementioned arteries was performed with an angle as close to zero degrees as possible (maximum angle of 30 degrees) during fetal rest with at least 3 similar waveforms. PI in the UmA or mean PI in uterine arteries above the 95^th percentile for gestational age, as well as PI below the 5^th centile in the MCA for gestational age, were defined as abnormal. The examination was concluded with the assessment of the fetal aIMT using a 12L-RS transducer on a vertebral artery preset, and aIMT was defined as the distance between the lumen of the aorta and the internal hyperechogenic adventitia border of the vessel. Each measurement was triple checked in different artery areas in a single image, and the arithmetic mean was calculated. Intra-observer bias was eliminated by a single sonographer performing all the measurements, which were stored in our clinical database.
Written informed consent was obtained from each patient prior to enrolment, ultrasound evaluation, and blood sampling. The study was approved by the Bioethics Committee and conducted according to the Declaration of Helsinki (No. 291/19). The Bioethics Committee reviewed the study protocol and confirmed that the conducted research was not a clinical trial.

Statistical analysis

In order to determine the cut-off point in the fetal abdominal aIMT, which is most accurate in predicting PE, ROC curves were developed and the area under the curve (AUC) was determined with 95% CI, and its significance was evaluated using DeLong’s method. Because the measurement of aIMT was performed between the 26^th-28^th and/or 32^nd-34^th weeks of gestation, a ROC curve was developed for the measurements from the 26^th-28^th gestational weeks (38 patients). Thereafter, 2 further ROC curves were derived in 73 patients, considering subsequent measurements as independent and as dependent (cluster data, CI based on Bootstrap). On the basis of the ROC curves, an optimal cut-off point with the highest Youden Index was determined. Then, aIMT was compared using Student’s t-test. Finally, a group of patients with an increased aIMT was identified, with a new variable referred to as “aIMT ≥ cut-off” if aIMT was greater than or equal to the cut-off point in any measurement, where the cut-off point was a specific number.
Similarly, an attempt was made to determine a specific cut-off value of aIMT for predicting FGR.
Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy (ACC), and diagnostic odds ratio (DOR) for the variable “aIMT ≥ cut-off” was compared with the parameters assessed in a routine ultrasound Doppler examination, which may indicate early symptoms of placental insufficiency, such as:
1) Umbilical artery pulsatility index (UmA PI) or mean uterine artery pulsatility index (UtA PI) above the 95^th percentile for gestational week, or MCA pulsatility index (MCA PI) below the 5th percentile for gestational week;
2) Estimated fetal weight (EFW) below the 3^rd percentile, or EFW below the 10^th percentile with UmA PI, or UtA PI above the 95^th percentile, or MCA PI below the 5^th percentile for gestational week.
Additionally, patients who subsequently developed PE were compared with the control group, as well as with the subjects who did and did not meet the criteria of FGR. Student’s t-test or the Mann-Whitney test were employed to examine variables depending on how these variables fulfilled the test assumptions. To verify the normality of distribution, the Shapiro-Wilk test was used, whereas the Fisher-Snedecor test was applied for the equality of variances. The χ² test or Fischer’s exact test were employed to compare the binary variables when the Cochrane assumption for group size was not satisfied.
Moreover, logistic regression models were developed, initially with simple linear regression, then with multiple regression, to assess the prediction of PE by itself, as well as adjusted for other variables. Odds ratio (OR) with 95% CI and p-value for the Wald test were determined.
In each analysis, p < 0.05 was considered statistically significant. The analyses were performed using PQStat v1.8.4, Poland, and Bootstrap in Stata/IC 16.1, USA.

Results

In the studied population, 18 patients with PGDM developed PE, whereas 63 women without PE constituted the control group (Figures 1, 2). Five patients developed early-onset PE before the 34^th week of gestation. In the study group, the mean gestational age of diagnosing PE was 34.28 ±3.05 weeks. The investigated group of patients was homogenous; there were no significant differences in the maternal age, pre-gestational BMI, type and duration of diabetes, parity, or smoking status between the studied population and the control group. Interestingly, there were no significant differences in HbA1c levels in any of the trimesters between the patients who subsequently developed PE and the controls. Patients presenting higher triglyceride concentrations (p = 0.0101) in the third trimester were more likely to develop PE (Table 1).
The data concerning the delivery from patients with PE and the controls were then compared. As a result, it was observed that patients with PE delivered smaller newborns (p < 0.0001) who were supplied by smaller placentas (p < 0.0001) than the patients without PE. Furthermore, the gestational week of the delivery was lower in patients diagnosed with PE than in the control group (p = 0.0001). However, the placental weight/birth weight ratio was not significantly different between patients with PE and the controls (Table 2).
Subsequently, aIMT obtained were analysed in terms of representing a potential predictor of PE. It was found that mean aIMT in fetuses from pregnancies complicated by PE was 0.64 ±0.05 mm, compared to 0.49 ±0.01 mm in patients without PE. In 38 patients, the measurements were performed between the 26^th-28^th gestational weeks, which provided the basis for a ROC curve with AUC (95% CI) = 0.83 (0.58, 1), and a cut-off point for an increased risk of PE at aIMT ≥ 0.56 mm (Table 3, Figures 3, 4).
Because the measurements obtained between the 32^nd and 34^th gestational weeks were not significantly different from those from the 26^th-28^th gestational weeks (p = 0.828912), 2 additional ROC curves were determined, combining measurements from the 26^th-28^th weeks of gestation with the ones from the 32^nd-34^th weeks of gestation, which also increased the size of the studied group. In the first curve, the second measurement in the same patient was considered independent data – AUC (95% CI) = 0.705 (0.53, 0.88). In contrast, in the second curve, the second measurement in the same patient was treated as dependent (cluster) data – AUC (95% CI) = 0.69 (0.52, 0.87). The cut-off point for an increased risk of developing PE remained the same, at aIMT ≥ 0.56 mm, despite the addition of data from the measurements from the 32^nd-34^th gestational weeks in both ROC curves.
In further analyses, aIMT ≥ 0.56 mm was used as a cut-off point in predicting PE in our group of patients with pre-gestational diabetes mellitus.
Following the aforementioned results, the findings in aIMT measurements were compared with the abnormalities in the Doppler examination performed in the same patients during the same scan (Table 4).
Measurement of aIMT compared to the abnormalities in the Doppler examination of the fetus, or signs of FGR, showed the highest accuracy as a predictor of PE in patients with PGDM (Table 4).
Subsequently, in view of being a potential tool for predicting PE, the aIMT results were adjusted in terms of maternal parameters, which may affect the obtained results, such as maternal age, smoking status, pre-gestational BMI, type of diabetes, and parity. After adjusting, fetal abdominal aIMT was found to be relevant in predicting PE (p = 0.0078) with an adjusted OR of 6.31. Since the OR for developing PE in patients with PGDM is nearly 4 times higher if abnormalities in Doppler ultrasound are present OR (95% CI) = 3.97 (1.27, 12.43), the effectiveness of aIMT in predicting PE for maternal parameters and any Doppler abnormalities found was adjusted. After adjusting for maternal age, pre-gestational BMI, parity, smoking status, type of diabetes, and abnormalities in the Doppler examination, aIMT remained significantly higher (p = 0.0094) in patients who subsequently developed PE. However, the patients with aIMT ≥ 0.56 mm were 5.27 times more likely to develop PE regardless of the abovementioned parameters (Table 5).
In further analysis, FGR was used as an end-point, and patients were retrospectively classified into the FGR group and the controls, which were then compared (Table 6). It was observed that older women were more susceptible to FGR in our group of PGDM patients (p = 0.0149). Furthermore, newborns of mothers with type 2 diabetes and shorter duration of diabetes were also prone to growth restriction. Additionally, serum triglycerides in the first and second trimesters were significantly higher in patients with pregnancy complicated by FGR (Table 6).
Table 7 presents the obstetric outcome of pregnant patients affected by pre-gestational diabetes mellitus and FGR. The birth weight and placental weight were lower in pregnancies complicated by FGR irrespective of the placental weight/birth weight ratio (p = 0.0544) (Table 7).
However, assessing aIMT in predicting FGR was found to be ineffective. An ROC curve was developed to evaluate the performance of aIMT as a potential FGR predictor. Nevertheless, since the area under the curve was insignificant for predicting subsequent FGR, the aIMT cut-off point could not be determined (Table 8).
Finally, the authors decided to compare the patients presenting an increased fetal abdominal IMT with patients with aIMT below the cut-off point and to verify the differences between these groups (Table 9). According to our observations, patients with a higher pregestational BMI and a lower weight gain were prone to an increased fetal abdominal aorta intima-media thickness (Tables 9 and 10).

Discussion

In patients with pre-gestational diabetes, the risk of developing PE is 2- to 4-times higher as compared to the background population [8-10, 23]. In fact, the prevalence of PE in the studied group was 22.22%, and 5 out of 18 patients satisfying the PE criteria were diagnosed before the 34^th week of gestation.
An increased fetal aIMT constituted a strong predictor of developing subsequent PE in the studied population. Furthermore, even after adjusting for age, smoking status, type of diabetes, parity, and abnormalities in the Doppler assessment, increased aIMT remained a significant and independent PE risk factor.
Our results indicate that an elevated aIMT represents an early form of vascular adaptation to high peripheral resistance of an impaired placenta. According to other conducted studies, an increased aIMT correlates with peak systolic velocity and systolic-diastolic aorta diameter change [17]. The aIMT may also be a parameter that provides insight into the fetal circulation [24]. Moreover, it allows detection of an ongoing pathology in the fetal circulation prior to the clinical manifestations, which predominantly occur in the late third trimester [25].
Our assessment of aIMT as an early predictor of PE in a high-risk population seems promising. The thickening of intima-media in fetal arteries reflects the adaptations to an increased peripheral resistance. Thus, it is reasonable to speculate that an increased aIMT and a high-velocity flow in the UmA constitute indirect evidence of the hyperkinetic fetal circulation/elevated fetal blood pressure [26]. Fetuses exposed to the elevated blood pressure in the developing circulatory system show an impaired endothelial function, which impacts cardiovascular risk throughout their lives. In fact, follow-up studies involving children from pregnancies affected by PE demonstrated an increased cardiovascular risk in future life [27-30]. However, these patients did not appear to be affected by the risk of cancer or non-cardiovascular morbidity [27]. Bearing that in mind, it seems vital to consider PE as a CVD of the mother and the fetus because it impacts the cardiovascular risk for both. Consequently, it is essential to develop a preferably non-invasive parameter/measurement, which will allow to examine the condition of the fetal circulation, as well as the possible vascular complications. Interestingly, the thickness of the abdominal aorta intima-media was not gestational age-dependent between the 26^th and 34^th weeks of gestation (p = 0.83).
Furthermore, our observations support the data that aspirin prophylaxis may not be beneficial among patients with pregestational diabetes [11, 31]. This result might reflect the ratio of term PE in our group, in which prophylaxis was less effective than in the preterm PE subjects [6, 32]. In the analysed group, only 5 out of 18 patients developed PE before the 34^th week of gestation. Although late-onset PE is less likely to be predicted [6, 7], aIMT efficiently predicted PE in the studied group of patients, in which 72.2% of PE cases developed after the 34^th week of gestation. Additionally, aIMT seemed more effective in predicting late-onset PE than the Doppler evaluation, which is known to be less likely to be altered in term PE.
In view of the inadequate intrauterine environment, as well as the recommendations stating that pregnancies complicated by PE should be delivered promptly, patients with PE in our study delivered earlier (p = 0.0001) [20, 33]. As a result, neonatal birth and placental weights were lower among women with PE (p < 0.0001). However, the placental/fetal weight ratio was not different in the studied population compared with the control group (p = 0.61). In other studies, this ratio was altered in pregnancies complicated by hypertensive disorders or placental abruption [34, 35].
In terms of predicting FGR, aIMT did not appear to be statistically significant. Thus, it was not possible to determine a cut-off point to predict the subsequent restriction of fetal growth. Moreover, it may be assumed that this observation stemmed from the fact that patients with late-onset PE predominated, whereby FGR is uncommon [4, 32].
It is of note that patients with a shorter history of diabetes, particularly with type 2 diabetes, were prone to develop FGR. In this group of patients, diabesity (obesity with type 2 diabetes) may modify the l-arginine/nitric oxide and insulin/adenosine axis [36, 37]. Therefore, it may affect the fetoplacental endothelium, increasing the risk of placental complications. Moreover, prior to the pregnancy in patients with type 2 diabetes, the time necessary for the diagnosis is longer than in type 1 diabetes, due to the milder initial symptoms. Hence, the exposure to hyperglycaemia prior to initiating the treatment is longer, and frequently vascular complications are already present at the time of the diagnosis. According to the Bennett et al. data, 29% of fetuses from mothers with vascular complications of type 2 diabetes will be small for gestational age (below the 10^th centile) [38].
As expected, fetuses diagnosed with growth restriction (FGR) presented lower birth weight and placental weight (p < 0.0001). Nevertheless, in our study the fetal/placental birth weight ratio did not differ as compared to the controls, although in some studies such a correlation was observed [34, 35].
Chronic hypertension was diagnosed in one patient, who subsequently developed PE, as well as in one patient included in the control group. Considering the small number of patients with chronic hypertension, this group was not analysed separately.
Interestingly, patients with growth-restricted fetuses presented higher concentrations of serum triglycerides in the first and second trimesters, whereas in patients with PE, such elevated levels were observed in the third trimester. According to numerous studies, triglyceride concentrations correlate with the incidence of PE [8, 39], poor glycaemia, and weight gain control [40]. Furthermore, dyslipidaemia may impair trophoblast invasion, thus contributing to a cascade of pathophysiological events that lead to the development of PE [41, 42]. This thesis is supported by the fact that triglyceride accumulation in the endothelial cells has been associated with a decreased release of prostacyclin [43]. Similarly, an increase in triglycerides has also been linked with a significant shift in LDL particle sizes to subtypes of smaller diameter. The formation of a smaller variant of LDL has been shown to contribute to endothelial dysfunction in PE by means of the stimulation of thromboxane synthesis by endothelial cells and an increase in intracellular calcium in vascular smooth muscles [42-44].
In addition to the ultrasound measurements, certain biochemical parameters may be useful in identifying patients at risk of placental complications. For instance, the levels of natriuretic peptides tend to be higher in patients presenting hypertensive disorders of pregnancy. However, corin, which converts inactive pro-atrial natriuretic peptide into an active form, may constitute an efficient second-trimester predictor of the subsequent impaired placental function [22, 45].
Finally, patients presenting with an increased aIMT were compared with those in whom aIMT was not elevated (cut-off point aIMT ≥ 0.56 mm). The only difference between these patient groups was in pre-gestational BMI and weight gain in the course of pregnancy. According to other studies, higher maternal BMI is associated with the altered maternal haemodynamics, consequently increasing the risk of hypertensive disorders during pregnancy [5, 46, 47].
The results of our study indicate that fetal aIMT constitutes a potentially effective and feasible tool, which may be applicable when differentiating patients at a higher risk of PE. However, larger, prospective studies need to be conducted to validate our findings. Currently, the available data concerning aIMT derive from studies with small sample sizes.

Strengths and limitations

The strengths of the presented study include the size of the population with pregnancy complicated by PGDM and the lack of intra-observer bias, because one skilled ultrasonographer performed all the measurements. Additionally, the study design is also worth noting, whereby the sonographer did not know whether the patient would be classified into the studied population or the control group, because the subjects were assigned to the groups based on the obstetric outcome.
As for the weaknesses of the study, it was observed that patients with long-term pregestational diabetes usually develop subclinical vascular complications, which may be impossible to detect when enrolling in the study, although they may impact the results. While designing the study, we did not aim to assess aIMT in non-diabetic patients. We determined that comparing healthy pregnant women to diabetic patients regarding the risk of PE might introduce bias due to various factors that induce vascular changes, potentially leading to PE. In further studies comparing the predictive value of aIMT with other predictors of placental insufficiency such as the sFlt-1/PlGF ratio might be interesting. Because we collected patients serum, that might be a starting point for subsequent studies.

Conclusions

The thickening of the intima-media complex is a form of subclinical vascular adaptation to an increased placental resistance and can be observed before abnormalities in the routine Doppler scans. Our study suggests an increased aIMT and a high-velocity flow in the UmA represent indirect evidence of hyperkinetic fetal circulation/elevated fetal blood pressure. Recognising this, it becomes crucial to view PE as a cardiovascular issue affecting both the mother and the fetus. This condition has implications for the future cardiovascular health of both parties. Therefore, aIMT shows promise as a potential early predictor of PE, particularly in patients with pregestational diabetes.

Disclosures

The study was approved by the Bioethics Committee at Poznań University of Medical Sciences and was conducted according to the Declaration of Helsinki (No. 291/19).
The research received no external funding.
The authors declare no conflict of interest.

References