Study population
This prospective study protocol was approved by the Medical Ethics Committee of the Leiden University Medical Center (P14.095), and informed consent was signed by both parents/guardians of all subjects. Children with chromosomal disorder were excluded to preserve homogeneity of the population. Thirty-two patients after surgical CoA repair participated in the study and underwent a cardiovascular magnetic resonance imaging (MRI) examination. These patients were also included in a previous study with the aim to investigate the cardiac autonomic nervous system activity, cardiac function, and their relationship in children after CoA repair [21]. Eventually, eleven of them were excluded due to practical and emotional problems (e.g., patient movement and endurance) during the MRI. The included 21 patients were aged 13.7 ± 2.6 years (mean ± standard deviation), including 12 patients with bicuspid aortic valve (BAV) and 9 patients with tricuspid aortic valve (TAV). Mean age at CoA correction was 1.0 ± 1.8 years, performed by end-to-end anastomosis in 16, extended end-to-end anastomosis in 2, and subclavian flap in 1 patient. Only one patient, as result of a recurrent obstruction, underwent a reoperation using an autologous pericardial patch. None of the patients had a clinical indication for reintervention at the time of the MRI examination. The included patients were scanned between September 2015 and May 2016, and time between reconstruction and MRI was 12.6 ± 3.0 years.
The presence of an aortic reobstruction was determined based on the maximal flow velocity in the descending aorta, measured by a suprasternal transthoracic Doppler echocardiogram (VIVID 9, GE Healthcare, Norway) by a single observer (IN), supervised by an experienced clinician (AH) in all patients. The acquired images were stored and analysed offline using the EchoPAC software version 113 (General Electric Healthcare, Horten, Norway). Based on this analysis, the presence of a reobstruction was determined, defined as a maximal flow velocity larger than 2.5 m/s [22]. Using this criterion, the patient group was divided into two groups: twelve without and nine with recurrent obstruction.
MRI acquisition
The image acquisition consisted of two MRI through-plane phase-contrast MRI sequences to determine aortic PWV and one four-dimensional (4D) flow MRI sequence. MRI for all patients were performed on a 3-T scanner (Ingenia, Philips Healthcare, Best, The Netherlands) using a combination of both a FlexCoverage posterior coil in the table and a dStream Torso anterior coil, together providing up to 32-coil elements for signal reception. Concomitant gradient correction and local phase correction were performed from standard available scanner software.
The PWV was determined from high-temporal through-plane phase-contrast MRI using free breathing with retrospective electrocardiographic gating, for both the proximal aorta (ascending aorta plus aortic arch) and the descending aorta. This was accomplished by measuring the flow velocity through two planes positioned perpendicular to aortic centreline: the first plane intersecting both the ascending and thoracic descending aorta and the second plane intersecting the abdominal descending aorta, defined as proximal PWV and diaphragmatic PWV, respectively. The proximal PWV MRI sequence parameters were as follows: velocity encoding of 200–300 cm/s in feet-head direction, acquired temporal resolution 8.4 ms, reconstructed temporal resolution 4.1 ms (171 ± 24 phases), echo time 2.3 ms, repetition time 4.2 ms, flip angle 20°, field of view 350 × 350 × 8 mm, and acquired spatial resolution 2.8 × 2.8 × 8.0 mm. Acquisition time and heart rate were on average 75 ± 10 s and 83 ± 12 beats per min, respectively. Diaphragmatic PWV MRI sequence parameters were as follows: velocity encoding of 150–250 cm/s in feet-head direction, acquired temporal resolution 8.6 ms, reconstructed temporal resolution 4.1 ms (167 ± 23 phases), echo time 2.4 ms, repetition time 4.3 ms, flip angle 20°, field of view 350 × 350 × 8 mm, and acquired spatial resolution 2.8 × 2.8 × 8.0 mm. Acquisition time and heart rate were on average 78 ± 12 s and 84 ± 11 beats per min, respectively.
The aortic 4D flow MRI sequence used a hemidiaphragm respiratory navigator, a retrospective ECG gating, and a standard non-symmetrical four-point velocity encoding. Sequence parameters were as follows: velocity encoding of 200–350 cm/s in four directions, acquired temporal resolution 34.4 ms, reconstructed temporal resolution 29.2 ms (26 ± 4 phases), echo time 2.4 ms, repetition time 4.3 ms, flip angle 10°, field of view 350 × 350 × 52.5–72.5 mm, acquired spatial resolution 2.5 × 2.5 × 2.5 mm, segmentation factor 2, and sensitivity encoding factor 2 in anterior-posterior direction. Acquisition time was on average 4.9 ± 0.7 min excluding the respiratory compensation. Due to the acceptance window of the respiratory navigator, the actual acquisition time in the scanner approximately doubled.
Image analysis
The image analysis consisted of two parts to determine aortic PWV and the WSS. In order to obtain the PWV, the acquired proximal PWV and diaphragmatic PWV images firstly were analysed using the in-house developed software MASS (LUMC). This software was used to perform velocity mapping and to measure the length of both aortic segments on a multislice survey of the aorta. Lastly, these quantifications were imported into an in-house developed MATLAB-based application to determine the PWV of the proximal and descending aorta, using the foot-to-foot method (Fig. 1). In all subjects, the PWV image analysis was performed by a single observer (IN) with over 3-year experience in cardiovascular MRI, supervised by an experienced researcher (JW) with over 20 years’ experience in cardiovascular MRI. Additionally, the PWV ratio was also derived from these values, defined as the descending aorta PWV divided by the proximal aortic PWV. This PWV quantification method was previously validated and described in more detail by Grotenhuis et al. [23].
From 4D flow MRI, the WSS was determined using CAAS MR Solutions v5.0 (Pie Medical Imaging, Maastricht, The Netherlands), assuming a constant blood viscosity of 4 mPa s. This software was used to compute the WSS over five time phases and three consecutive aortic segments (Fig. 1): the aortic root plus the ascending aorta, the aortic arch, and the descending aorta (respectively; from the aortic valve to the brachiocephalic artery, from the brachiocephalic artery up and including the left subclavian artery, and from the subclavian artery to the abdominal descending aorta at the level of measurement of the diaphragmatic PWV). This was accomplished by firstly segmenting the aorta on a combined weighted magnitude and velocity image for all five available time phases, incorporating only the aorta and excluding the main branches (e.g., the subclavian and carotid arteries). Secondly, the anatomical segmentation planes were manually placed and imported perpendicular to the aortic wall. From proximally to distally on the aorta, these planes were positioned at the aortic valve, proximally against the brachiocephalic artery, distally against the subclavian artery, and 10 cm caudal below the diaphragm. Thirdly, for the five available time phases and each anatomical segment, the maximal WSS was exported from CAAS. Lastly, these maxima over the five time phases were used to determine the peak WSS for each anatomical segment over all five time phases. The WSS image analysis was performed by a single observer (IN) in all patients. The applied method to determine the WSS in the five systolic time phases was previously described and validated on the reproducibility by van der Palen et al. [24].
Statistical analysis
The statistical analysis was performed using the SPSS v23 software (IBM, Chicago, IL, USA). Differences between groups were compared using the independent sample t test or Mann-Whitney U test, respectively used for parametric scale data or non-parametric scale and ordinal data. Correlations between variables within groups were evaluated using the Pearson (rP) and Spearman rank (rS) correlations, respectively used for parametric scale data or non-parametric scale and ordinal data. The Levene test was used to verify the equality of variance and Shapiro-Wilk test to verify the normality of the data. The absolute correlation coefficient (rP or rs) was classified as follows: 0.30 < |r| < 0.50, weak; 0.50 < |r| < 0.70, moderate; 0.70 < |r| < 0.85, good; and |r| > 0.85, strong. All statistical tests were two-tailed, and a p value of less than 0.05 was considered significant. Data will be presented as mean values with standard deviations.