Children between 0 and 18 years of age were referred for a clinically indicated CMR examination to assess RV function, PA regurgitation and pulmonary perfusion, from January 2019 through 2022, and were retrospectively included. The control subjects consisted of adolescents referred for a clinically indicated CMR for suspected arrhythmogenic right ventricular cardiomyopathy, with a normal clinical work up and CMR without pathological findings. Patient demographics were registered from medical records.

The examinations were performed on a 3T or 1.5T MRI scanner (3T Signa Architect or 1.5T Signa Artist, GE Medical systems, Waukesha, WI, USA) with a standardized clinical protocol in free breathing, including 4D flow (described in detail in Supplementary Material S1). During 2019–2020, all examinations (n = 19) were performed on 3T, due to insufficient quality of the 4D flow on the 1.5T scanner. During 2021–2022, n = 16 patients were examined on the 3T and n = 8 on the 1.5T scanner after software adaptations. Balanced steady-state free precession (bSSFP) sequences performed in free breathing were used for analysis of ventricular function. An ECG-gated, respiratory compensated and motion-corrected 4D PC CMR flow protocol was used for flow analysis. Velocity encoding (VENC) was optimized to avoid velocity aliasing and was chosen individually for each patient, based upon the highest velocity measured on echocardiography the same day or the day before the MRI examination, typically ranging from 125 to 350 cm/s. General anaesthesia without intubation or sedation was used when needed. The scan protocol lasted∼45 min, and the 4D flow sequence duration was 6–8 min.

Examinations were reviewed and postprocessing was done offline by one pediatric radiologist (A.N) using the software Cardio AI (Tempus Medical/Arterys, Redwood Shores, CA, USA), where ventricular function and flow data were analysed. Ventricular function was analysed by semiautomatic and manually adjusted segmentation on bSSFP cine images in the short axis plane, in accordance with present guidelines (16). Flow was reconstructed from the CMR 4D flow volume using double-oblique planes perpendicular to the dominant flow direction, in the ascending aorta and in the main pulmonary artery (Figures 1a–d). For mitigating the influence of measurement noise 5 closely spaced (distance: 1 mm) manually traced measurements were performed; in the mid pulmonary artery and in the ascending aorta just above the level of the sinotubular junction (Figure 1e). The WSS value is an average of WSS in all voxels around the circumference of the vessel. The lumen contours were manually adjusted at all time frames of the cardiac cycle to provide best fitting to the vessel contour. Both peak and mean WSS were registered. Peak WSS is the highest WSS value measured in that vessel and can be in either systole or diastole. Mean WSS, also referred to as time-averaged WSS (TAWSS) is an average of all WSS values across all 20 or 30 time frames of the cardiac cycle. Vessel diameter derived from CMR 4D flow images were indexed to BSA, using the DuBois formula (17). Peak vessel diameter was measured the same way as WSS, using the highest diameter measured from all time frames; typically the highest values are found in systole. Mean vessel diameter is an average of the diameters of all time frames of the cardiac cycle. Wall shear stress was calculated by the software using velocity data from the 4D flow and a presumed blood viscosity of 3.5 mPas (Figure 1f).

Continuous variables are presented as means ± standard deviation (SD) if normally distributed and otherwise as medians and ranges. Categorical variables are expressed as counts and percentages of the total. The data was tested for normal distribution, using the Kolmogorov—Smirnov and Shapiro–Wilk test. Categorical and continuous data were analysed as appropriate. The presented values of mean WSS and peak WSS were averaged over the five ROIs in both the Ao and PA and compared between groups using independent samples t-test. The correlations between parameters were assessed using Pearsons’ rank correlation test for continuous variables or Fisher's exact test used for categorical data and small samples. A p-value less than 0.05 was considered significant. All statistical analyses were performed in SPSS Statistics for Macintosh (IBM Corp. Released 2021. IBM SPSS Statistics for Macintosh, Version 28.0. Armonk, NY: IBM Corp)

Comparing the average of five measurements of peak and mean WSS in the PA, significant differences between rTOF patients and controls regarding both average peak WSS and average mean WSS, (average peak WSS 69 ± 13 cPa vs. 93 ± 29 cPa, p < 0.001 and average mean WSS 35 ± 13 cPa vs. 17 ± 5 cPa, p < 0.001) were revealed. There was also a wider distribution of WSS values in the rTOF group. Despite there being significant differences in aorta diameter/BSA, both peak and mean, between rTOF patients and controls, there was no significant difference in neither mean nor peak WSS in the ascending aorta (p = 0.14–0.76) (Figure 2).

When adjusting for age > or <10 years of age, no significant difference in WSS aorta or PA was revealed.

In recent years, WSS has been suggested to constitute a risk marker for vascular remodelling and later development of atherosclerotic disease and aneurysms in the systemic circulation (19, 20). The WSS is reported to regulate transcription in vascular remodelling (21), and high levels of WSS have predominantly been shown to aggravate manifest atherosclerosis (22). Which roles these pathways play in the pathogenesis of long-term complications in patients with CHD is still unknown.

WSS can be calculated in different ways; from echocardiography or from using computational fluid dynamics (CFD) from computed tomography or CMR (23, 24). CMR 4D flow yields information about the velocity gradients in blood adjacent to the vessel wall, and the shear force can be calculated by commercially available software. In many institutions, 4D flow is now a standard part of clinical CMR analysis for CHD. The litterature further highlights the potential of WSS, currently mostly applied in research, to become a clinically useful biomarker (25–27). However, determining WSS based on CMR 4D flow has disadvantages; there is a tendency towards underestimation because of limited spatial and temporal resolution (28). Also, in complex flows the range of velocities in each voxel can be wide and vary between individuals. In this study, where many of the patients were expected to display regurgitant flow, we used circumferential WSS which is more sensitive to helical or turbulent flow in the transverse direction of the vessel, rather than longitudinal WSS which is more sensitive to shear forces in the longitudinal axis, as would be expected in a stenosis.

Concerning the aorta, WSS has been studied in mainly bicuspid aortic valve and Marfan syndrome (29, 30), where it has been shown to predict progressive dilatation (31). One CMR 4D flow study by Schäfer et al. (32) has focused on the proximal thoracic aorta of adolescents with rTOF repaired in infancy and shows many similarities to our study, concerning patient age and study group composition. We found significant dilatation of the ascending aorta in the rTOF patients, whereas Schäfer et al. did not. On the other hand, they found elevated levels of WSS in the ascending aorta, whereas we did not. Despite difference in findings, our conclusions were similar; that patients that have undergone early repair of TOF are at risk of late aortic complications. In addition we found a decreasing indexed ascending aorta diameter with age. While the study by François et al. relies exclusively on echocardiography data and therefore cannot be fully translated to our work, it similarly demonstrates a successive normalization of indexed aortic diameters in rTOF patients who underwent corrective surgery in infancy, effectively halting progressive dilatation of the ascending aorta (33). Other rTOF studies have shown normalized aortic diameters in mid childhood (34, 35), which is contrary to our results; even though median age for repair in our study was 5 months, there persisted a significant difference in indexed ascending aorta diameter in patients older than 10 years. This is an important finding, and although the study has a cross-sectional design, it suggest that early repair does not necessarily protect against ascending aortic dilatation. Although different studies have shown various results, aortic diameter may possibly need surveillance in this patient group. Several studies have focused on WSS in the PA, mainly in patients with pulmonary arterial hypertension (36, 37). Only a few studies have evaluated WSS after TOF repair; these have shown that patients with repaired TOF exhibit elevated levels of WSS in the PA and its branches (38, 39). Rizk et al. examined adult patients and included 24 patients with rTOF and 11 controls, where only 17 of the patients hade undergone the TAP repair (14). They found circumferential WSS (both peak and mean) significantly higher in the PA of patients with rTOF than in controls. Hudani (39) also studied adults, 17 patients and 20 controls, and in concordance with our study found significantly elevated levels of circumferential WSS along the PA compared to controls. One pediatric rTOF study by Hu et al. (40) where 25 patients and 10 controls were younger (8.44 ± 4.52 and 8.2 ± 1.22 respectively) than in our study, also revealed increased levels of PA WSS. The above studies have shown only weak or no correalations between PA regurgitant fraction an PA WSS, which is in line with our results in adolescent patients.

Our cross sectional study included all patients that were examined at our institution during the relevant time period without excluding those with stenoses, competent valves and importantly; those who had already undergone pulmonary valve replacement. This is both a strength and a weakness in our study; patients with rTOF display different pathologic traits and highly variable disease progressions. This highlights the difficulties in displaying how PA regurgitation influences WSS in a clinical setting.

Moreover, WSS is sensitive to differences in flow that are not described in simplistic terms of velocity or regurgitation. WSS is affected by spatial and temporal resolution (especially important in a pediatric cohort), respiratory motion and field strength (41). We have used both 1.5T and 3T in clinical practice with the majority (35/43) examined on 3T, which may have slightly influenced our results (42). Inter-observer variability and reproducibility also limit how WSS can be implemented in clinical routine. A combination of all the above mentioned probably partly explain our findings. In some cases though there were trends that did not reach significance, for instance RVEF vs. PA WSS and PA regurgitant fraction vs. PA WSS. Using small patient cohorts always pose a risk of type 2-errors where true correlations goes unnoticed. We are aware of this drawback with our study and hope to be able to contribute to prospective research in this area. There are still no longitudinal studies showing how elevated WSS in the PA of rTOF patients can predict vascular remodelling and future need for additional interventions.

While WSS has been shown to influence the risk for vascular remodelling in the systemic circulation; in the PA the relationship remains to be elucidated. In patients with rTOF, RVOT and PA are often wide due to earlier surgery, e.g., TAP, as in most of the patients (n = 35/43) in our study. Interestingly, there were no significant differences in PA diameter in the rTOF patients older than 10 years and controls, which could be attributed to a larger proportion of patients having had reinterventions in the older cohort. In our study, there was no correlation between degree of dilatation and levels of WSS in the PA. Our patients were on average 11.6 years old, and we know that in rTOF, vessel dilatation progresses over a long period of time. Even though other research groups have shown correlations between WSS and regurgitation, our patients may have been too early in their disease progression to exhibit sufficient degrees of dilatation or regurgitation. In support of this theory, the patients did not display left ventricular enlargement or reduced RVEF. Against that reasoning stands that the average rTOF patient in our study had a right ventricle enlargement, indicating that the rTOF patients were on average on an advanced stage in their disease progression.

Further studies are required to clarify the relationship between elevated WSS in the pulmonary artery and dilatation/regurgitation, as well as between WSS and ascending aortic dilatation in rTOF patients. Moreover, the lack of studies correlating WSS with clinical endpoints must be addressed before WSS can be integrated into clinical management guidelines.

Our sample was small, and the rTOF group was also somewhat heterogenous concerning the method and timing of repair, although the majority of patients had undergone the TAP at some point, that could introduce bias to the results. The control group included slightly older individuals which can also introduce an uncertainty regarding vascular remodelling that could potentially be more advanced in an older age group. However, these are slowly developing changes and the few years of difference in age is probably not significant (43).