Altered caval flow and wall shear stress in single ventricle circulation: insights from ferumoxytol-enhanced 4D flow MRI

Abstract

Aims

Patients with single ventricle (SV) congenital heart disease typically undergo surgical palliation that reroutes the superior/inferior vena cava (SVC/IVC) directly to the pulmonary arteries, which alters vascular flow dynamics. Four-dimensional flow (4DF) cardiovascular magnetic resonance (CMR), particularly with ferumoxytol contrast, enables comprehensive assessment of flow distribution and wall shear stress (WSS) in small paediatric vessels. This study aimed to evaluate the differences in caval flow and WSS between SV and BiV patients.

Methods and results

We retrospectively analysed patients who underwent clinically indicated ferumoxytol-enhanced (FE) 4DF CMR between 2019 and 2024. Patients with anomalous systemic or pulmonary venous return were excluded. 4DF data were evaluated on the cloud-based Tempus platform. Flow in the SVC, IVC, aorta, and descending aorta was quantified and indexed to the body surface area. The average WSS over time was measured. Among 387 patients (n = 126 for SV; n = 261 for BiV), SV patients had lower indexed SVC flow (1.11 ± 0.51 vs. 1.50 ± 0.69 L/min/m², P < 0.0001) and lower WSS (16.0 ± 6.4 vs. 22.7 ± 9.1 cPa, P < 0.0001) compared with BiV patients. The SVC/caval flow ratio declined with age and body surface area, remaining independently lower in the SV group after multivariable adjustment (P = 0.0005).

Conclusion

SV physiology is associated with lower SVC flow and WSS independent of age or body size, suggesting altered systemic venous return intrinsic to cavopulmonary circulation. 4DF CMR is a valuable tool for assessing and monitoring systemic venous physiology in congenital heart disease.

Graphical Abstract

Graphical Abstract.

Introduction

The cavopulmonary connection in patients with a single ventricle (SV) anatomy represents one of the most extensively studied flow environments in congenital heart disease. These surgical connections are used to palliate a variety of complex congenital cardiac defects; but while the bidirectional Glenn (BDG) and Fontan operations have significantly improved outcomes for patients with SV physiology, they also introduce fundamental alterations to vascular haemodynamics, including reductions in wall shear stress (WSS).^1,2 In a cavopulmonary circulation, the absence of a subpulmonary ventricle creates a passive venous pathway in which pulmonary blood flow depends on systemic venous circulation. Thus, detailed knowledge of caval flow dynamics is essential to understanding the physiological consequences of these anastomoses. Decades of flow quantification research is starting to yield an understanding of the haemodynamic characteristics of caval flow circulation over a patient’s lifespan.^3,4

Prior studies have primarily relied on echocardiography and 2D phase-contrast (2D-PC) cardiac magnetic resonance (CMR) to assess flow dynamics in congenital heart disease. SVC and inferior vena cava (IVC) flows have been described in Fontan patients, with a suggestion that the changes in caval flow distribution may be driven by somatic growth.⁵ Caval flow distribution, defined by the relative contribution of SVC and IVC flow to total systemic venous return, can influence Fontan haemodynamics, including hepatic flow distribution.³ However, caval flow proportions are also shaped by extracardiac and developmental factors, including age-related changes in somatic growth and cerebral blood flow.⁵ Although prior studies have characterized caval flow distribution in patients with Fontan physiology, most have relied on targeted 2D phase-contrast CMR acquisitions planned at specific vascular locations, which are not always performed for patients with biventricular (BiV) physiology. As a result, direct comparison of caval flow distribution and related flow-derived parameters between SV and BiV physiology using a comprehensive CMR flow approach remains limited.

With the recent advances and widespread adoption of four-dimensional flow (4DF) CMR, more detailed and quantitative evaluations of cardiovascular physiology have become possible. In particular, FE 4DF CMR allows for accurate flow quantification across multiple vascular territories in a single acquisition, with excellent reproducibility in measuring flow in smaller vessels when compared to traditional 2D-PC methods.⁶ This non-invasive modality also supports comprehensive assessment of WSS, the result of the frictional haemodynamic forces exerted by blood flow on the endothelial surface, offering new insights into vascular physiology and thrombosis risk.^7,8 The SV physiology predisposes patients to chronically elevated central venous pressure and venous congestion, which can contribute to extracardiac complications including hepatic congestion, lymphatic dysfunction, protein-losing enteropathy, plastic bronchitis, and thrombotic risk.⁹ These features suggest potential alterations to caval flow distribution and venous WSS in SV patients compared with BiV patients.

Thus, our central hypothesis was that SV patients exhibit altered caval flow distributions as well as changes to WSS compared with their BiV counterparts. Thus, in this study, we sought to compare caval flow splits and WSS parameters in patients with SV physiology vs. those with BiV physiology, leveraging a consistent FE 4DF MRI protocol we have performed over the past years.

Methods

Study design

This single-centre, retrospective, Institutional Review Board-approved study evaluated a series of clinically indicated CMR examinations acquired between 2019 and 2024 using an FE 4DF sequence. Studies were included if they provided diagnostic-quality acquisitions that yielded clinical flow measurements, as well as normal systemic/pulmonary venous return (i.e. absence of an interrupted IVC and no evidence of partial/total anomalous pulmonary venous return). Patients ≤18 years were selected for this examination. Exclusion criteria encompassed any dataset with anomalous venous anatomy or significant coil artefacts that prevented reliable assessment of the SVC, IVC, ascending aorta (AAo) or descending aorta (DA). Selected cases included patients with both SV and BiV physiology. The SV cohort was further categorized by ventricular anatomy into pre-Glenn, BDG, and Fontan groups for analysis, with additional exclusion of SV patients with antegrade pulmonary blood flow that could directly impact cavopulmonary flow, as well as those with abnormal systemic/pulmonary venous drainage that directly alter expected SVC/IVC drainage and confound assessment of caval flow distribution. BiV patients were defined as those with two functioning ventricles contributing to both systemic and pulmonary circuits, excluding those with abnormal systemic/pulmonary venous drainage. Isolated intracardiac shunts such as atrial/ventricular septal defects were not excluded as they do not directly reroute systemic venous return. The retrospective analysis had an approximate 2:1 sample size distribution of BiV and SV cohorts. Studies included patients who were either awake or anaesthetized, utilizing positive pressure ventilation by a certified paediatric cardiac anaesthesiologist. Conventional volumetric measurements from cine imaging included indexed end-diastolic volume (EDVi), ejection fraction (EF %), and cardiac index (CI).

4D flow sequence parameters

All 4DF acquisitions were performed using a validated protocol tailored to patient body surface area. An iPAT acceleration factor of 2 and 2–3 views per segment were applied, with the sequence acquired using a matrix size of 72 × 160. For neonates and small infants (BSA <0.5 m²), an isotropic resolution of 1.3 mm was used with a field of view of 200 mm, whereas for larger children (0.5 m² to 1.5 m²), a resolution of 1.8 and 2.8 mm for adult-sized patients (>1.5 m²). Echo times ranged from 2.37 to 2.55 ms, and temporal resolution spanned 39.3 to 82.1 ms. Velocity encoding (VENC) was optimized for vessel-specific flow dynamics, with settings ranging from 150 to 250 cm/s. This strategy was previously validated against 2D-PC, with an R² of 0.98 for FE studies.⁷ Typical FE studies had scan durations ranging from 7 to 10 min.

4D flow measurements of caval flows

All 4DF data were subsequently uploaded to the cloud-based Tempus platform for analysis, with measurements performed under the oversight of experienced readers. Two experienced CMR readers (YHL, RVV) independently reviewed all studies for consistency and measurement accuracy. To obtain flow measurements of vessels, planes were placed perpendicular to the vessels and velocity vectors were calculated.⁷ Regions of interest (ROIs) were manually drawn at the vessel-lumen interface measurements to obtain flows at the SVC, IVC, AAo, and DA. Flow measurements were obtained at standardized anatomical landmarks. A representative acquisition and the corresponding flow measurement process are illustrated in Figure 1.

Figure 1.

Volume-rendered 4D flow MRI demonstrating analysis planes and regions of interest in biventricular and single-ventricular patients. Flow/WSS through the caval veins (blue rings) and systemic arteries (red rings) quantified by placing measurement planes orthogonal to each vessel.

• In BiV patients or SV patients prior to bidirectional Glenn, the SVC was assessed at the midpoint between the innominate vein confluence and the SVC–right atrium (RA) junction. In SV patients with a bidirectional Glenn and Fontan physiology, the SVC was assessed at the midpoint between the innominate vein confluence and the superior cavopulmonary connection. For consistency in flow measurements, vessel visualization, and avoiding any potential interaction with diaphragm motion, the DA at the level of the diaphragm served as the main metric for lower body flow. The IVC flow was also measured at the confluence of the hepatic veins.

• AAo flow was quantified at the level of the branch pulmonary arteries.

All flow measurements were indexed to BSA to account for patient size differences. An estimated aortopulmonary (AP) shunt burden was calculated as (AAo flow—SVC flow—DA flow)/AAo flow, with values expressed as percentages of AAo flow. In addition, SVC flow values were expressed as a ratio relative to the total venous return (SVC/caval flow), which provided a normalized measure for comparison across the cohort.

Wall shear stress (WSS) measurements

Average WSS (cPa) was quantified using the Tempus platform, leveraging a previously validated algorithm process.¹⁰ This methodology processes the 4DF MRI data with a Sobel filter to detect edge gradients and delineate the vessel boundary without manual segmentation. After the manual drawing of ROIs at the vessel-lumen interface, the software automatically computed the average WSS values over time. Hard cut-off distances of 3 mm for the interior and 1 mm for the exterior of the drawn contour were applied to standardize the measurement process. The average WSS over time was selected for analysis.

Statistical analysis

All analyses were performed in GraphPad Prism and Python (statsmodels). Continuous data were assessed for normality, and WSS was log₁₀-transformed prior to analysis to satisfy model assumptions. To assess normality of the data, the Shapiro–Wilk test, histograms and Q–Q plots were assessed with GraphPad and Python. Between-group comparisons (BiV vs. SV) used Welch’s t-test or Mann–Whitney U test as appropriate. Relationships between variables were explored with Pearson correlation and linear regression (univariable and multivariable) to identify determinants of SVC/Caval flow ratio and log-transformed WSS. Multivariable linear regression models were constructed using candidate predictors selected a priori based on physiologic relevance and flow variables that were statistically significant in univariable analysis. Model assumptions were evaluated using residual diagnostics and normality testing, and multicollinearity was assessed using variance inflation factors. Statistical significance was defined as P < 0.05. As a sensitivity analysis, the multivariable SVC-to-caval flow ratio model was repeated using directly measured IVC flow rather than DA flow as a surrogate. Additional sensitivity analyses were also performed in response to the physiologic heterogeneity of the SV cohort (stratified by pre-Glenn, BDG, or Fontan) and differential use of general anaesthesia using Kruskal–Wallis and Mann–Whitney U testing.

Results

Demographics and baseline CMR metrics

A total of 387 patients, consisting of 261 patients with BiV circulation and 126 patients with SV circulation, were included in the analysis. In the BiV cohort, the median age was 12.0 years (IQR: 5.0–16.0), and 36.4% were female. General anaesthesia was used in 23.4% of cases. The most common diagnoses included repaired Tetralogy of Fallot/double outlet right ventricle (n = 87) and repaired D-transposition of the great arteries with arterial switch operation (n = 51). In the SV cohort, the median age was 8.9 years (IQR: 3.5–16.0), and 31.5% were female, with a higher rate of general anaesthesia used in 53.2% of cases. Eight (6%) were pre-Glenn; 50 (39%) had a BDG, and 66 (52%) had a Fontan. The most common diagnoses included hypoplastic left heart syndrome (n = 45) and tricuspid atresia (n = 29).

While age and BSA were not significantly different between the groups, the SV cohort trended towards a left-skewed distribution with younger/smaller patients clustering in early childhood, whereas the BiV group showed a broader distribution, across age and body size (Table 1). Systemic ventricular volumetric and functional indices differed between groups. Compared with BiV patients, the systemic ventricle in SV patients had a significantly larger EDVi (122.4 ± 53.0 vs. 97.9 ± 30.6 mL/m², P < 0.0001), lower EF (48.6 ± 9.8% vs. 55.0 ± 7.9%, P < 0.0001), and higher CI (4.7 ± 1.7 vs. 4.2 ± 1.3 L/min/m², P = 0.008).

Table 1.

Demographic and cardiac MRI-derived characteristics of bi-ventricular (BiV; n = 261) vs. single-ventricular (SV; n = 124) cohorts

	BiV (n = 261)		SV (n = 124)	P
Female	95 (36%)		39 (31%)	0.342
Age (years; median + IQR)	12 (5–16)		8.9 (3.5–16)	0.1614
Body Surface Area (m2; median + IQR)	1.36 (0.76–1.70)		0.99 (0.74–1.38)	0.088
Diagnosis	Repaired TOF/DORV (87) Repaired d-TGA with ASO (52) ASD/VSD/AVSD (25) Other (36) Coarctation (21) Bicuspid Aortic Valve (21) Vascular Ring/Normal (19)		HLHS (45) Tricuspid Atresia (29) DORV (21) uAVSD (13) Other (8) PA/IVS (5) DILV (3)
General anaesthesia (%)	61 (23%)		67 (54%)	<0.0001a
	RV	LV
EDVi (mL/m2)	97.9 ± 30.6	81.7 ± 22.2	122.4 ± 53.0	<0.0001a
EF (%)	55.0 ± 7.9	59.2 ± 6.7	48.6 ± 9.8	<0.0001a
CI (L/min/m2)	4.2 ± 1.3	3.8 ± 1.2	4.7 ± 1.7	<0.0001a

Categorical data are compared by χ² or Fisher’s exact test; continuous data compared by Mann–Whitney U test or Welch’s t-test, as appropriate.

ASD, atrial septal defect; ASO, arterial switch operation; AVSD, atrioventricular septal defect; BSA, body surface area; CI, Cardiac Index (L/min/m²); DILV, double-inlet left ventricle; DORV, double-outlet right ventricle; EDVi, end-diastolic volume indexed to BSA (mL/m²); EF, ejection fraction (%); HLHS, hypoplastic left heart syndrome; LV, left ventricle; PA/IVS, pulmonary atresia with intact ventricular septum; RV, right ventricle; SV, single ventricle; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; uAVSD, unbalanced atrioventricular septal defect.

Significant differences (P < 0.05) are highlighted in^a.

Caval flow split by age and body surface area

The mean indexed SVC flow was 1.373 ± 0.659 L/min/m², while AAo flow was 3.530 ± 1.154 L/min/m², DA flow was 1.684 ± 0.749 L/min/m², and IVC flow was 1.633 ± 0.589 L/min/m². The mean SVC/caval flow across the entire cohort was 0.450 ± 0.120. Visual inspection of histograms and normal Q–Q plots demonstrated that the SVC-to-caval flow ratio was skewed towards lower ratios (see Supplementary data online, Figure S1). There was a strong inverse relationship between age and the proportional SVC flow. Specifically, SVC/caval flow ratio was negatively correlated with both ages (r = −0.530, 95% CI: −0.598 to −0.454, P < 0.0001) and BSA (r = −0.585, 95% CI: −0.647 to −0.515, P < 0.0001).

Biventricular vs. single ventricle flow measurements

Caval and arterial flow characteristics differ significantly between SV and BiV circulations (Table 2). Indexed SVC flow was also significantly lower in the SV group (1.11 ± 0.51 L/min/m²) compared with BiV (1.50 ± 0.69 L/min/m², P < 0.0001), whereas the indexed SVC was not significantly different between the groups (P = 0.3187). Likewise, indexed IVC flow was lower with similar vessel size. Aortic flow indexed to BSA was similar between SV (3.51 ± 1.30 L/min/m²) and BiV patients (3.54 ± 1.08 L/min/m², P = 0.804). The net DA flow was lower in SV patients (1.37 ± 0.43 L/min/m²) compared with BiV (1.76 ± 0.61 L/min/m², P < 0.0001). The estimated AP shunt burden was higher in SV patients compared with BiV patients (25.8 ± 18.3% vs. 10.2 ± 12.2%, P < 0.0001).

Table 2.

Comparison of indexed vessel diameters flows and wall shear stress (WSS) between bi-ventricular (BiV; n = 261) and single-ventricular (SV; n = 124) cohorts

	BiV (n = 261)	SV (n = 124)	P-value
Vessel diameter (mm/m2)
SVC	14.30 ± 6.57	14.97 ± 5.88	0.3187
IVC	19.88 ± 9.15	21.36 ± 7.97	0.1048
AAo	22.22 ± 9.64	27.33 ± 9.76	<0.0001
DA	13.35 ± 4.95	13.74 ± 4.63	0.4478
Flow (L/min/m2)
SVC	1.50 ± 0.69	1.11 ± 0.51	<0.0001
IVC	1.84 ± 0.75	1.35 ± 0.62	<0.0001
AAo	3.54 ± 1.08	3.51 ± 1.30	0.804
DA	1.76 ± 0.61	1.37 ± 0.43	<0.0001
WSS (cPa)
SVC	22.72 ± 9.14	16.03 ± 6.41	<0.0001
IVC	22.28 ± 9.86	16.21 ± 6.74	<0.0001
AAo	25.79 ± 10.72	19.51 ± 8.63	<0.0001
DA	100.20 ± 33.89	89.27 ± 29.93	0.0015

Values are mean ± SD; P-values from Welch’s t-test. Significant differences (P < 0.05) are demonstrated.

AAo, ascending aorta; DA, descending aorta; IVC, inferior vena cava; SVC, superior vena cava.

In a multivariable linear regression model including age, BSA, circulation type (SV vs. BiV), and general anaesthesia, all variables were significantly associated with the SVC/caval flow ratio (Table 3, model R² = 0.377, P < 0.0001). Both age (P = 0.018) and BSA (P < 0.0001; Figure 2) were negatively associated with SVC/caval flow ratio. After adjusting for these variables, SV circulation remained independently associated with lower SVC/caval flow (P = 0.0005). General anaesthesia was associated with a modest but statistically significant increase in SVC/caval flow ratio (P = 0.0168). The findings were preserved in a sensitivity analysis using directly measured IVC flow instead of DA flow to calculate the SVC-to-caval flow ratio (see Supplementary data online, Table S4).

Table 3.

Multivariable linear regression for predictors of SVC-to-caval flow ratio

Variable	Estimate (β)	Standard Error	95% CI	P-value
Intercept	0.5781	0.0166	0.5455 to 0.6107	<0.0001
Age	−0.002	0.0009	−0.0037 to −0.0004	0.018
BSA	−0.0847	0.015	−0.1141 to −0.0552	<0.0001
SV vs. BiV (reference: BiV)	−0.0385	0.011	−0.0601 to −0.0168	0.0005
Use of general anaesthesia	0.032	0.0133	0.0058 to 0.0582	0.0168

All predictors were statistically significant. SV circulation remained independently associated with a lower SVC/caval ratio after adjusting for covariates. The model had an R² = 0.377. No multi-collinearity was detected (VIFs all <3.1), and residuals passed all tests for normality. Significant associations (P < 0.05) are demonstrated.

Figure 2.

Relationship between the normalized superior caval flow ratio (SVC/caval flow) and body surface area (BSA). In both cohorts, relative SVC flow declines as BSA increases. Single-ventricular patients (SV; blue squares and regression line) exhibit a lower SVC/caval flow ratio compared with biventricular patients (BiV; red circles and regression line).

Biventricular vs. single ventricle WSS measurements

WSS values differed significantly between SV and BiV patients across multiple vascular territories (Table 2). In the SVC, average WSS was significantly lower in SV patients (16.0 ± 6.4 cPa) compared with BiV patients (22.7 ± 9.1 cPa, P < 0.0001). SV patients also had lower WSS in the IVC, AAo, and DA when compared with BiV patients. The log-transformed wall shear stress was approximately normally distributed across most of its range (see Supplementary data online, Figure S1).

In a multivariable linear regression model with the log-transformed SVC WSS as the dependent variable, several variables were independently associated with SVC WSS (Table 4, R² = 0.503, P < 0.0001). Higher age (P = 0.0003), BSA (P = 0.0010, Figure 3), and descending aortic flow (P = 0.0035) were all negatively associated with log SVC WSS. Indexed SVC flow was a strong positive predictor of SVC WSS (β = 0.0600, P < 0.0001). General anaesthesia was not statistically significant in this model (P = 0.239). SV circulation remained independently associated with lower SVC WSS compared with BiV (P < 0.0001), even after adjusting for the covariates.

Table 4.

Multivariable linear regression for predictors of log-transformed SVC wall shear stress (WSS)

Variable	Estimate (β)	Standard Error	95% CI	P-value
Intercept	1.432	0.03514	1.362 to 1.501	<0.0001
Age (years)	−0.00405	0.001007	−0.0062 to −0.0019	0.0003
BSA (m2)	−0.06618	0.02003	−0.1056 to −0.0268	0.001
SV vs. BiV (reference: BiV)	−0.1497	0.01457	−0.1791 to −0.1204	<0.0001
SVC indexed flow (L/min/m2)	0.06003	0.01124	0.04195 to 0.09010	<0.0001
DA Flow (L/min/m2)	−0.03676	0.01252	−0.06139 to −0.01213	0.0035
Use of general anaesthesia	−0.02061	0.01748	−0.05497 to 0.01375	0.239

Age, body surface area (BSA), descending aortic flow, and single ventricle circulation were each independently associated with lower SVC WSS, while indexed SVC flow was positively associated. Use of general anaesthesia was not a significant predictor. The model had an R² = 0.503, with acceptable multicollinearity (all VIFs <3.4). Significant associations (P < 0.05) are demonstrated.

Figure 3.

Relationship between log₁₀–transformed wall shear stress (WSS) and body surface area (BSA). Both biventricular (BiV; red circles and regression line) and single-ventricular (SV; blue triangles and regression line) cohorts demonstrate a decline in WSS with increasing BSA. SV patients consistently exhibit lower WSS across the BSA range compared with BiV patients.

Sensitivity analyses within the SV cohort and potential influence of general anaesthesia

Given the physiologic heterogeneity of the SV cohort, additional subgroup analysis was performed by surgical stage (see Supplementary data online, Table S1). The SV cohort included eight pre-Glenn patients, 50 BDG patients, and 66 Fontan patients. SVC/caval flow ratio differed significantly across stages, with lower values in Fontan patients compared with earlier-stage SV physiology [Pre-Glenn: 0.51 (0.47–0.57), Glenn: 0.53 (0.42–0.61), Fontan: 0.37 (0.31–0.42), P < 0.0001]. Indexed SVC flow and SVC WSS also differed significantly across stages, with the lowest values observed in the Fontan subgroup.

To evaluate the potential influence of general anaesthesia, BiV-vs.-SV comparisons were repeated separately in patients studied without general anaesthesia and in those studied under general anaesthesia (see Supplementary data online, Tables S2 and S3). In both strata, SV patients continued to demonstrate a lower SVC/caval flow ratio, lower indexed SVC flow, and lower SVC WSS compared with BiV patients. Among patients not under general anaesthesia, the SVC/caval flow ratio was lower in SV compared with BiV patients [0.37 (0.31–0.43) vs. 0.41 (0.36–0.47), P = 0.0019]. A similar difference was observed among patients under general anaesthesia [0.49 (0.40–0.59) vs. 0.55 (0.48–0.62), P = 0.0083].

Associations of SVC/caval flow ratio and WSS within the single ventricle cohort

Within the SV cohort, both SVC/caval flow ratio and SVC WSS did not demonstrate significant associations with SV-specific haemodynamic parameters. Higher SVC/caval flow was moderately correlated with younger age (r = −0.495, P < 0.0001), lower body surface area (r = −0.554, P < 0.0001), and a modest association with CI was present (r = 0.305, P = 0.0008). SVC WSS followed a similar pattern, with inverse associations with age (r = −0.552, P < 0.0001) and BSA (r = −0.605, P < 0.0001), and CI (r = 0.460, P < 0.0001). A weak inverse association was observed between SVC WSS and descending aortic flow (r = −0.223, P = 0.013). After correcting for age/BSA, neither SVC/caval flow ratios nor SVC WSS had any significant relationships with AP shunt burden, SVEDVi or SVEF.

Discussion

The purpose of this study was to compare caval flow splits and WSS parameters in patients with SV physiology vs. those with BiV physiology. We used FE 4DF CMR to measure and compare flow parameters in the cavae among children and young adults with SV physiology at various stages of palliation with those exhibiting BiV physiology. The primary results of this study included (i) characterization of caval flow splits in a large congenital heart disease cohort as stratified by BSA, age, and BiV vs. SV status; (ii) patients with SV physiology exhibited relatively lower SVC flow and lower SVC WSS compared with patients with BiV physiology, independent of BSA or age.

Optimization of the Fontan pathway prioritizes achieving balanced hepatic flow distribution, which is essential to prevent complications such as pulmonary arteriovenous malformations. However, the dynamics of systemic venous return change significantly with age, introducing challenges in maintaining consistent hepatic flow.³ The distribution of systemic venous return in infants, children, and adults with normal BiV circulation (proportionally greater lower-body/peripheral growth compared with head and brain growth) has also been described by echocardiographic Doppler.⁶ Somatic growth leads to changes in the ratio of upper and lower body segments as well as alterations to regional blood flows.^5,11 Similarly, normative Doppler profiles of SVC, hepatic, and pulmonary venous flows in healthy children show clear age-related variations in SVC velocities.¹² Thus, the age-related decline in relative SVC contribution observed in our study confirms previously described developmental changes, rather than representing a novel physiological observation. CMR studies by Whitehead et al. and Ghosh et al. have similarly described that the change in flow distribution in Fontan patients occurs longitudinally, supporting the idea that somatic growth influences caval flow patterns in SV patients.^4,13 In a recent study including 414 Fontan patients, Govindarajan et al. demonstrated a non-linear, exponential decline in the SVC-to-IVC inflow ratio from early childhood through adolescence, with a transition from SVC-dominant to IVC-dominant flow.³ Therefore, age-related changes in venous inflow have practical implications for Fontan pathway design and longitudinal interpretation of flow measurements. When surgical planning aims to optimize hepatic flow distribution, the current SVC/IVC flow distribution should be interpreted in the context of expected somatic growth and the anticipated decrease in relative SVC contribution over time. Similarly, when SVC WSS is used to assess haemodynamic status, lower relative SVC flow should be interpreted with respect to age and BSA rather than assumed to reflect disease progression.

Importantly, SV physiology remained independently associated with a lower SVC/caval flow ratio after adjustment for age, BSA, and general anaesthesia, suggesting that cavopulmonary physiology has an independent contribution to altered caval flow distribution beyond somatic growth alone. Reductions in cavopulmonary circulation could be potentiated by positive pressure ventilation and general anaesthesia. However, the utilization of modern volatile anaesthetics is known to reduce cerebral metabolism while increasing cerebral blood flow; this remains consistent with the modest increase in SVC/caval flow ratio observed.¹⁴ Even after accounting for somatic growth, SV patients exhibited disproportionately lower SVC flow, supporting the conclusion that surgical physiology directly contributes to these differences. As described by Gewillig et al., the cavopulmonary circuit ultimately acts as a ‘dam’ with upstream congestion. This configuration not only reduces kinetic energy across the pulmonary vascular bed but also imposes a bottleneck that limits overall cardiac output.^15–17 Our findings reinforce this concept and are further supported by longitudinal data from Arrigoni et al., who demonstrated progressive declines in indexed cavopulmonary flows in Fontan patients despite preserved ventricular function.¹⁸ The overall reduction in caval flows may also reflect aortopulmonary collateral flow, although it does not have a direct effect on the SVC/caval flow ratio. The contribution of the present study is the application of a consistent FE 4DF CMR approach that allowed for direct examination of flow measurements and WSS in patients with SV or BiV physiology. Together, the results underscore the importance of understanding flow constraints in SV patients for long-term surveillance and anticipatory management of Fontan failure.

We demonstrated that WSS in the SVC was independently lower in SV patients. Venous systems typically have lower WSS and a differential response to WSS compared with the dynamic arterial blood flow.^19,20 In SV physiology, this reduction may reflect a predisposition to stasis, a core component of Virchow’s triad. Alongside increased coagulability and endothelial dysfunction, low WSS is thought to contribute to thrombosis risk.^21,22 Experimental studies have previously demonstrated that non-pulsatile pulmonary flow induces endothelial dysfunction and increased apoptosis of vascular smooth muscle cells.^23,24 Histopathological analysis of intra-acinar pulmonary vessels of Fontan patients who died also demonstrates eccentric intimal fibrosis associated with thrombosis.²⁵ Recent computational fluid dynamic modelling by Govindarajan et al. showed that WSS decreases with upsizing of the Fontan conduit, and oversizing a Fontan could potentially introduce non-uniform flow with regions of abnormally low WSS. These zones were associated with disturbed flow, suggesting that low WSS may be detrimental.²⁶ This concept is reinforced by the increased presence of thrombus in atrio-pulmonary Fontans and the understanding that thrombosis in Fontan patients is multifactorial, involving both anatomic and flow-related factors.²⁷ Lower WSS in the pulmonary arteries has also been implicated as a form of abnormal mechano-transduction that mediates the development of aortopulmonary collaterals.⁸ Our findings of lower SVC WSS in SV patients further highlight the impact of Fontan physiology on flow patterns and its potential role in predisposing to thrombotic complications. At the same time, the measurement of WSS is dependent on flow and vessel geometry; in SV patients, it may be hard to separate this effect without other parameters such as oscillatory shear stress, which also has a role in endothelial dysfunction.²⁸

Finally, reduced WSS in the Fontan vasculature may reflect a broader spectrum of abnormal flow physiology. In the aorta, prior work has shown that regions of low WSS, such as those in bicuspid aortic valve disease, predict progressive dilation and remodelling through impaired mechanotransduction, endothelial inflammation, and extracellular matrix turnover.^29–32 Turbulent-like flow patterns in the proximal neo-aorta after Norwood palliation can also contribute to adverse vascular remodelling.³³ Romanowicz et al. recently demonstrated that Fontan patients exhibit reduced arterial flow and oxygen delivery in select vascular territories, particularly the iliac and renal arteries, despite preserved oxygen content, with deficits linked to collateral losses and correlated with exercise capacity.³⁴ These studies, combined with our findings, underscore that diminished WSS is not an isolated venous phenomenon but part of a systemic pattern of altered haemodynamics that promotes stasis, endothelial dysfunction, and prothrombotic risk in the venous system, while simultaneously restricting end-organ perfusion in the arterial circulation, together highlighting the central role of shear in shaping long-term vascular health in SV patients.

Limitations

Our study is limited in that it was conducted as a single-centre, retrospective, cross-sectional analysis of a heterogeneous population, with variability in spatial resolution and flow sequence parameters across age. To mitigate this, we included only FE studies, which provide higher signal and improved measurement accuracy, particularly in small infants.⁷ It is important to note that, since normal healthy controls rarely require FE CMR, a direct age and sex-matched comparison could not be performed to isolate the SV pathophysiology against the normal healthy circulation. Although the sequence used in our analysis was free-breathing with respiratory navigation, prior real-time CMR examinations suggest minimal differences in net flow compared to breath-hold techniques, making it unlikely to significantly influence our results.³⁵ While age was not statistically different between groups, the SV cohort skewed younger, with a lower median and interquartile range. This skew could compound our interpretation of caval flow distribution. However, our findings were adjusted for age, BSA and use of general anaesthesia, supporting a physiologic basis for this difference. There is still, however, a risk of potential residual confounding variables.

The additional subgroup analysis reinforces that SV physiology should not be interpreted as a single uniform state, although these abnormalities are more pronounced in those with cavopulmonary connections (the Fontan and Glenn groups). Fontan patients demonstrated the lowest SVC/caval flow ratio and SVC WSS, consistent with the progressive haemodynamic constraints imposed by cavopulmonary circulation and somatic growth. However, the persistence of lower SVC flow and WSS in SV patients regardless of the use of general anaesthesia suggests that the primary differences cannot be solely explained by cohort heterogeneity. We also excluded studies with excessive motion artefacts or with stents or coils, which may limit generalizability. Future studies using 4DF MRI could leverage this dataset to estimate cerebral blood flow and further explore the interaction between pulmonary venous return, interventricular forces, and diastolic function, to better characterize how these factors influence caval flow distribution.

Conclusion

SVC flow and WSS are significantly lower in SV patients compared with those with BiV physiology, regardless of age or BSA. In addition, 4DF cardiac MRI remains a valuable tool to assess how caval flow distribution evolves with age, size, and somatic growth, offering insights into the cavopulmonary connection and long-term outcomes.

Supplementary data

Supplementary data are available at European Heart Journal - Imaging Methods and Practice online.

Author contributions

Vasupradha Suresh Kumar (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing—original draft, Writing—review & editing [lead]), Sarah Kollar (Conceptualization, Investigation, Methodology, Resources, Writing—review & editing [equal]), Ravi V. Vegulla (Conceptualization, Data curation, Investigation, Resources, Supervision, Writing—original draft, Writing—review & editing [equal]), Areen Almarkhan (Data curation, Writing—review & editing [supporting]), Andrew Waberski (Data curation, Supervision, Validation, Writing—review & editing [supporting]), and Yue-Hin Loke (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing [lead])

Consent

No informed consent was required as no identifiable patient information or images are included in this manuscript.

Funding

None declared.

Data availability

The data used in this article are available within the article and its online supplementary material.

Lead author biography

Vasupradha Suresh Kumar, MD. Vasupradha Suresh Kumar is an assistant professor and paediatric cardiologist at Hasbro Children’s/Brown University. She earned a medical degree from the University of Arkansas in Little Rock, Arkansas. Dr. Kumar completed a paediatrics residency at Hasbro Children’s/The Warren Alpert Medical School of Brown University and a paediatric cardiology fellowship at Children’s National Hospital in Washington, DC. Her research interests lie in preventative cardiology, imaging in congenital heart disease, and social determinants of health. Dr. Suresh Kumar is conversant in Tamil, Spanish, and English and she is a member of the American Academy of Pediatrics.