Improved quantification of aortic regurgitation with direct regurgitant jet measurement by four-dimensional flow cardiovascular magnetic resonance in complex congenital heart disease

Abstract

Background

Due to the presence of complex flow states and significant jet eccentricity in patients with congenital heart disease (CHD), accurate quantification of aortic regurgitation (AR) using standard echocardiographic or conventional cardiovascular magnetic resonance (CMR) imaging measures remains challenging. Four-dimensional flow (4DF) CMR permits transvalvular flow quantification under non-laminar flow states, although it has not been well validated for AR quantification in CHD.

Methods

In 186 patients with moderate or complex CHD, we evaluated the agreement between different methods of AR quantification by 4DF CMR when compared to volumetry. Regurgitant flow volumes were measured (1) conventionally on time-resolved, velocity-encoded 4DF sequences at the aortic annulus, sinotubular junction (STJ), and ascending aorta (AAo), and via (2) direct regurgitant jet quantification 5 mm proximal to the vena contracta.

Results

Moderate overall agreement in AR quantification was observed between study methods (ρ = 0.58–0.73). Compared with conventional flow quantification at the annulus, STJ, and AAo, direct regurgitant jet measurements showed improved correlation with volumetry (ρ = 0.76), especially in patients with significant aortic dilation (r = 0.95–0.97). In this latter group, regurgitant flow quantification at all other aortic levels resulted in AR severity classifications that were nearly a full grade lower (mean aortic regurgitant fraction difference: 7–12% ± 10–12%; p<0.001).

Conclusion

4DF CMR permits AR quantification in complex CHD with comparable accuracy to volumetry. Under non-laminar or complex flow states, as observed with significant aortic dilation, direct regurgitant jet measurements may be preferable to regurgitant flow quantification at all other aortic levels.

Graphical abstract

1. Introduction

In patients with congenital heart disease (CHD), chronic aortic regurgitation (AR) can develop as a consequence of structural aortic valve disease, congenital aortopathies, or sequelae from preceding surgical correction or palliation [1]. If left untreated, hemodynamically significant AR can result in adverse left ventricular (LV) remodeling, with ultimate development of irreversible LV systolic dysfunction and related increased cardiovascular morbidity and mortality [2]. Therefore, in the presence of chronic AR, current consensus guidelines recommend serial quantification of AR severity, LV chamber dimensions, and LV systolic function as the predominant basis for informing surgical management [3]. Transthoracic echocardiography (TTE) is the preferred initial and primary imaging modality for assessment of AR severity and has the most robustly studied thresholds for surgical intervention [3]. However, accurate classification of AR severity using semi-quantitative TTE parameters can be challenging in the presence of limited acoustic windows, eccentric and/or multiple regurgitant jets, or complex outflow tract geometries [4].

Given these limitations, cardiovascular magnetic resonance (CMR) imaging has become increasingly utilized in the assessment of valvular regurgitation and provides an accurate and reproducible means of transvalvular flow quantification [5]. Although two-dimensional phase contrast (2DPC) CMR remains the standard for AR quantification in degenerative forms of valve disease, accuracy in flow measurements fundamentally relies on appropriate positioning of the imaging plane and laminar flow conditions [4], [6]. Thus, the uniform applicability of 2DPC CMR for AR quantification under complex flow states, as commonly observed in CHD, has been challenged [6], [7]. Four-dimensional flow (4DF) CMR has emerged as a similarly accurate and reliable means of transvalvular flow quantification and additionally permits measurement of regurgitant volumes (RVol) under non-laminar flow states via retrospective correction of valvular through-plane motion and flow angulation [6]. However, automated commercial software for valve tracking is not universally available, and manual flow quantification can be time and labor intensive. Even irrespective of these methodologic challenges, the optimal method for AR quantification by 4DF CMR remains largely unknown, with discrepancies persisting with regards to the ideal aortic measurement plane and influence of common anatomic variations on the accuracy of flow quantification [7], [8], [9], [10]. Use of 4DF CMR for AR quantification in CHD additionally has not been well validated, with prior studies being limited by small sample sizes and exclusion of complex forms of CHD [11], [12], [13].

Therefore, the aims of this study were to (1) compare aortic regurgitant fraction (RF) quantification via (a) volumetry, (b) direct regurgitant jet measurement, and (c) conventional regurgitant flow measurements in standard aortic imaging planes on 4DF CMR in a large cohort of patients with complex CHD, including many with conotruncal defects and associated aortopathy, and (2) determine the potential influence of common anatomic variations (aortic dilation, valve morphology, jet orientation) on the accuracy of flow quantification via each method.

2. Methods

2.1 Patient selection and study population

A single-center, retrospective study was performed on 186 patients with complex CHD and at least mild AR (aortic RF ≥5%) who underwent CMR examination for standard clinical indications at Stanford University. Eligible patients were identified via query of the institutional picture archiving and communication system (PACS) database from January 1, 2010 to December 31, 2023. Manual chart review was performed to confirm the diagnosis of moderate or complex CHD, in accordance with the 2018 ACC/AHA anatomic classification guidelines [14]. Patients with (1) significant mitral regurgitation (RF ≥5%), (2) intracardiac shunts, or (3) univentricular circulations, were excluded, as these states may introduce further error in volumetric-based measurements. Additionally, with univentricular circulations, accurate quantification of the net pulmonary blood flow is frequently hindered by the presence of aortopulmonary and systemic-to-pulmonary venous collaterals. This study was approved by the Stanford University Institutional Review Board.

2.2 Image acquisition

All examinations were performed on a 1.5 Tesla (T) or 3T MRI system (GE Healthcare, Milwaukee, Wisconsin; Optima 450 W and MR750) in accordance with standard CMR protocols for patients with CHD [15]. To permit comprehensive anatomic assessment, including characterization of aortic valve morphology and ventricular functional quantification, retrospectively electrocardiogram (ECG)-gated steady state free-precession (SSFP) imaging sequences were acquired in the standard orientations (short-axis multiplane, four-chamber, three-chamber, two-chamber, and outflow tract views). All images were acquired with an end-expiratory breath hold to minimize cardiac motion attributable to the respiratory cycle. Multiplane short-axis sequences (14–18 slices across the heart mass, 8 mm slices; 20 phases per cardiac cycle; flip angle 50–60°, skip “0”) were used for planimetric assessment of ventricular chamber size and function. Prior to 4DF sequence acquisition, contrast-enhanced magnetic resonance angiography (MRA) was performed to permit three-dimensional quantification of aortic dimensions. As per previously published institutional protocols [16], ferumoxytol (Feraheme, AMAG Pharmaceuticals, Waltham, Massachusetts) was preferentially utilized for scans performed at Lucile Packard Children’s Hospital in an effort to improve the velocity-to-noise ratio [17]. Conversely, CMR examinations performed at Stanford Healthcare predominantly utilized gadolinium-based contrast agents (MultiHance, Bracco Diagnostics, Inc, Monroe Township, New Jersey), given the lower associated cost and more frequent acquisition of delayed CMR sequences. These differences were based on currently established CMR imaging protocols at each institution.

4DF acquisitions were obtained on retrospectively ECG-gated, free-breathing sequences following intravenous contrast administration. Typical scan parameters were as follows: (1) 1.5T: echo time (TE)/repetition time (TR), 1.5–2.1 ms/3.6–5.1 ms; flip angle, 15°; temporal resolution, 28.6–83.6 ms; bandwidth, 62–125 Hz/pixel; spatial resolution, 0.8–1.5 × 0.8–1.8 × 0.9–3 mm³; velocity encoding (VENC), 250–450 m/s and (2) 3T: TE/TR, 1.7–2.5 ms/3.7–4.8 ms; flip angle, 15°; temporal resolution, 18.3–58.6 ms; bandwidth, 63–100 Hz/pixel; spatial resolution, 0.7–1.2 × 0.7–1.6 × 1.2–2.8 mm³; and VENC, 250–450 m/s. The total scan time for 4DF sequences ranged 8–15 min depending on the heart rate (HR) at the time of image acquisition.

2.3 Image processing and analysis

All CMR studies were analyzed and interpreted in a blinded fashion by a single researcher with expertise in congenital CMR (B.S.C). For assessment of interobserver variability, the aortic RF was quantified at the sinotubular junction (STJ) and by direct jet measurement in 18 (10% of study cohort) randomly sampled patients by a second investigator (D.E.C.). Planimetric analyses were performed using QMassMR (Medis Suite MR; Leiden, the Netherlands). The LV end-diastolic (LVEDV) and end-systolic (LVESV) volumes were quantified via manual tracing of ventricular endocardial borders on multiplane short-axis sequences and subsequently indexed to body surface area to permit normative interpretation [18]. Corresponding long-axis images were cross-referenced to ensure accurate delineation of the atrioventricular and semilunar valve planes [19]. When the morphologic right ventricle (RV) remained in the systemic position (n=2), RV volumes were measured with inclusion of RV trabeculations in the calculated blood volume [20]. Ventricular stroke volume (SV; EDV-ESV) and ejection fraction (EF; [(EDV-ESV)/EDV] × 100%) were then calculated. Aortic dimensions were measured at standard segmental locations on doubly obliqued, multiplanar reformatted images derived from contrast-enhanced MRA sequences. Aortic root diameters were measured from inner-edge to inner-edge, with the maximal sinus-to-sinus dimension reported [21]. For purposes of data analysis, aortic dilation was defined as a maximal aortic root or ascending aorta (AAo) diameter ≥4.0 cm in adults, or two standard deviations above published normative values in pediatric patients [21], [22].

4DF post-processing analyses were performed using a cloud-based software application (Arterys; San Francisco, California). For purposes of flow quantification, double-oblique measurements were obtained at standard region of interests (ROI), with manual tracing of contours performed during each phase of the cardiac cycle. Aortic FF and RVol were quantified at the aortic annulus, STJ, and AAo, per standard recommendations (Fig. 2) [16], [23]. Given prior evidence of improved reliability in valvular flow quantification of eccentric regurgitant jets with correction for annular through-plane motion [24], annular flow measurements were performed with and without valve tracking. Valve tracking was performed via manual adjustment of the ROI to the level of the annulus during each phase of the cardiac cycle [24].

Fig. 2.

Representative imaging planes used for quantification of aortic flow volumes. Forward and regurgitant flow quantification was performed at the aortic annulus (a, b), sinotubular junction (c, d), and ascending aorta (e, f). Direct regurgitant jet measurements were performed 5 mm beneath the annular plane (g, h) with orthogonal angulation

For direct aortic regurgitant jet quantification, the ROI was placed at the center of each regurgitant jet approximately 5 mm proximal to the regurgitant orifice and angulated perpendicular to the jet direction (Fig. 2). When aliasing was present, the measurement plane was repositioned to a region of more coherent flow proximally within the LV outflow tract. When multiple regurgitant jets were present, separate measurements were obtained for each regurgitant jet, with the combined RVol reported (Fig. 3). In the absence of any established “gold-standard” for 4DF-based AR quantification, volumetry was utilized as the reference standard for intermethod comparisons. Use of CMR for volumetric-based assessment in patients with complex CHD is well validated [25], and biventricular volumes, as measured by planimetry, have been shown to have excellent agreement with 4DF derived aortic and pulmonary flow measurements [12], [25]. To permit inclusion of patients with concomitant right-sided regurgitant valvular disease, the net pulmonary blood flow (NPV; FF-RVol) in the bilateral, proximal branch pulmonary arteries was used as a surrogate for right ventricular SV.

Fig. 3.

Direct regurgitant flow quantification in the presence of multiple, eccentric regurgitant jets. Two separate regurgitant jets are visualized in the coronal (a) and axial (b) planes. Regurgitant flow volumes for each regurgitant jet were separately quantified via measurements performed 5 mm beneath the annular plane and angulated perpendicular to the jet direction (c, d)

We quantified the aortic RF (%) using the following methods:

• (1)Conventional: Aortic FF (mL/beat) and RVol (mL/beat) were measured on time-resolved, velocity-encoded 4DF sequences at the annulus (± valve tracking), STJ, and AAo; aortic RF (%) = RVol/FF × 100.
• (2)Direct Jet: RVol was quantified on direct regurgitant jet measurement. Aortic FF measurements from study method (1) were utilized; aortic RF (%) = RVol/FF × 100.
• (3)Volumetry: aortic RF (%) = [(LV SV (planimetry) – NPV)/LV SV] × 100.

AR severity was classified as mild (aortic RF = 5–15%), moderate (aortic RF = 16–30%), or severe (aortic RF >30%) in accordance with published reference norms [26]. The specified cut-off values were chosen given their close approximation with semi-quantitative TTE parameters [4] and superior discriminatory ability in predicting adverse related cardiovascular outcomes [27], [28].

2.4 Statistical analysis

Statistical analysis was performed using SPSS (version 29.0; IBM, Armonk, New York). The Shapiro-Wilk test was used to assess for normality. Categorical variables are presented as frequencies with related percentages. Normally distributed continuous variables are described as means ± standard deviation (SD), and alternatively distributed continuous variables as medians with corresponding interquartile ranges (IQR). Comparisons between groups were performed using the Wilcoxon signed-rank test, paired-t test, and Mann-Whitney U test for continuous variables, and χ² test for categorical variables. Agreement between study methods was assessed through calculation of Pearson’s (r) and Spearman’s (ρ) correlation coefficients for parametric and non-parametric variables, respectively. Interobserver agreement was analyzed using intraclass correlation coefficients (ICC) generated from a two-way mixed model with absolute agreement. Correlation analyses were graphically depicted on linear regression scatter plots. The strength of the corresponding linear relationship was classified as: weak (0.3–0.5), moderate (0.5–0.7), or strong (>0.7). Variance between measurements was reported as median (IQR) or mean differences (± SD), when appropriate. For normally distributed variables, variance was graphically depicted using Bland-Altman plots, with corresponding 95% confidence intervals (± SD × 1.96). Statistical significance was defined as a p-value ≤0.05 (two-tailed) for all analyses.

3. Results

3.1 Characteristics of the study population

A total of 186 patients comprised the study cohort (Fig. 1). Demographic and clinical characteristics of the study population are summarized in Table 1. Within the final study cohort, CMR examinations were performed between 2017 and 2023, with no significant MRI scanner technologic modifications performed during the study period. CMR examinations were predominantly performed on 1.5T magnets (83%, n = 155) and with use of gadolinium-based contrast agents (72%, n = 134). However, there was no significant difference in AR severity classification based upon the MRI contrast agent utilized (p = 0.6) or relative magnet strength (p = 0.9). The median HR at the time of image acquisition was 68 bpm (IQR 59, 76), with no frequent ectopy or sustained arrhythmias observed during the acquisition of 4DF sequences.

Fig. 1.

Consort diagram outlining patient selection criteria. CTD connective tissue disease, dTGA d-transposition of the great arteries, DORV double outlet right ventricle, ccTGA congenitally corrected transposition of the great arteries

Table 1.

Baseline patient characteristics.

Variable	All Patients (n = 186)
Male	120 (65%)
Body surface area (m2)	1.8 (1.6, 2.0)
Age (years)	23.5 (16.3, 80.0)
Native anatomy
Conotruncal lesions Bicuspid aortic valve Shone’s complex Congenital aortic stenosis Connective tissue disease/aortopathy Congenitally corrected transposition Ebstein anomaly Other	69 (37%) 49 (26%) 17 (9%) 12 (6%) 11 (6%) 7 (4%) 1 (1%) 20 (11%)
Aortic valve morphology
Trileaflet Bicuspid Unicuspid Quadricuspid Prior cardiac surgery Aortic valve repair Aortic root surgery Eccentric jet	99 (53%) 81 (44%) 4 (2%) 2 (1%) 152 (82%) 31 (17%) 38 (20%) 125 (67%)
Holodiastolic flow reversal
Descending thoracic Abdominal aorta Aortic root (mm) Ascending aorta (mm)	68 (37%) 39 (21%) 36 (32, 41) 32 (27, 39)
LV parameters
LV EDVi (mL/m2) LV ESVi (mL/m2) LV ejection fraction (%)	106 (86, 123) 48 (39, 58) 55 (51, 57)
Field strength
1.5 Tesla 3 Tesla	155 (83%) 31 (17%)
Contrast
Ferumoxytol Gadolinium	49 (26%) 134 (72%)

EDVi indexed end-diastolic volume, ESVi indexed end-systolic volume, LV left ventricular

Data are represented as medians with interquartile ranges for continuous

variables and n (%) for categorical variables

The majority of patients were male (65%, 120/186), with a median age of 24 years (IQR 16, 80). A diverse spectrum of complex CHD diagnoses were represented (Fig. 1), including conotruncal lesions (n = 69), multi-level left-sided obstructive disease (n = 17), and connective tissue disorders (n = 11). As expected, most patients had undergone prior cardiac surgery (n = 152), with 33% (n = 62) having undergone aortic valve and/or aortic root interventions prior to CMR examination. Over half of the study population had trileaflet aortic valves (n = 99), with similar rates of aortic dilation observed in those with trileaflet (60%, 59/99) and bicuspid (63%, 51/81) valve morphologies (p = 0.8). Mixed aortic valve disease was present in 24% of patients (n = 46), with moderate-to-severe aortic stenosis (mean transaortic Doppler gradient ≥20 mmHg) present in nearly half (n = 22) of these patients.

A solitary regurgitant jet was present in the majority of patients (95%, n = 178), with eccentric jets observed in over half of the study cohort (66%, n = 125). Aliasing within the regurgitant jet was present in approximately 20% of patients (n = 35), and more commonly observed in those with mixed aortic valve disease (AOR = 3.9, 95%CI = 1.4–10.5; p = 0.008). Within the entire cohort, AR severity was classified as mild in 34% (n = 63), moderate in 39% (n = 73), and severe in 27% (n = 50) of patients. There was no significant association between AR severity classification and CHD lesion subtype (p = 0.9), aortic valve morphology (p = 0.9), or preceding aortic surgical intervention (p = 0.06).

3.2 Aortic RF quantification: comparison between study methods & interobserver reliability

The extent of agreement in aortic RF quantification between study methods is summarized in Table 2 and graphically depicted on linear regression analyses in Fig. 4. Overall, there was moderate intermethod agreement between each analyzed 4DF AR quantification technique and volumetry (ρ = 0.53–0.76). However, intermethod concordance was notably improved with direct regurgitant jet quantification, as judged by the higher observed correlation coefficients (ρ = 0.76), smaller median differences (1–2%), and narrower limits of agreement in aortic RF measurements (p>0.1). Overall, when compared to direct jet measurement (aortic RF = 19–21%) and volumetry (aortic RF = 21%), the median aortic RF values obtained by flow quantification in conventional aortic imaging planes were significantly lower (aortic RF = 8–15%, p<0.001). As the aortic FF measurements remained constant across 4DF study methods, this difference was predominantly accounted for by the relative underestimation of aortic RVol with flow quantification distal to the regurgitant jet orifice (Supplementary Table 1, p<0.001). Conversely, regurgitant flow quantification by direct jet assessment demonstrated good overall agreement with volumetry (ρ = 0.78; median RVol: 17.7 mL/beat v. 19.4 mL/beat; p = 0.81).

Table 2.

Correlation of aortic RF quantification between study methods.

Method*	Spearman (ρ)a	Median Difference (%)	p-value
Conventional 4DF v. Volumetry
Aortic Valve (S) Aortic Valve (T) STJ AAo	0.55 0.53 0.64 0.64	9.1 (2.4, 16.1) 9.7 (2.7, 17.1) 4.5 (−1.9, 10.8) 7.1 (0.6, 13.2)	<0.001 <0.001 <0.001 <0.001
Direct Jet v. Volumetry
Aortic Valve	0.76	0.5 (−5.4, 5.6)	0.65
STJ	0.76	−1.4 (−7.6, 3.8)	0.33
AAo	0.76	−0.8 (−7.7, 3.8)	0.32

RF regurgitant fraction, 4DF four-dimensional flow, (S) static annular plane, (T) valve tracking, STJ sinotubular junction, AAo ascending aorta

^*The median difference in aortic RF measurements are reported as percentages, with the corresponding interquartile range. Median aortic RF difference is reported with the 2nd listed variable as the reference point. Conventional aortic RF values were calculated via measurement of forward and regurgitant flow volumes in the specified aortic plane. Direct aortic RF values were calculated via direct measurement of the regurgitant jet volume, with separate quantification of forward flow in the specified aortic plane.

^ap-values for Spearman’s correlation analyses were all significant (p<0.001).

Fig. 4.

Comparison of aortic RF values obtained with regurgitant flow quantification in (1) conventional 4DF CMR aortic imaging planes (a–d) and on (2) direct regurgitant jet measurement (e–g). Comparisons were made using volumetry as the reference standard. There was moderate intermethod agreement in aortic RF quantification between study methods (ρ = 0.53–0.76). However, aortic RF values obtained via direct regurgitant jet measurement were more closely correlated with volumetry (ρ = 0.76) RF regurgitant fraction

Interobserver reliability was excellent across all 4DF study methods, with an ICC of 0.99 for aortic FF, aortic RVol, and direct jet measurements (p<0.001). In this subset of analyzed patients, there was no difference in ultimate AR severity classification between observers.

3.3 AR severity classification: comparison between study methods

The median aortic RF values and corresponding AR severity classification obtained by each study method are depicted in Table 3. When compared to volumetry, AR severity was differentially classified in 36–51% of patients (n=67, 95) using aortic RVol obtained in conventional aortic imaging planes, with a lower median aortic RF documented in 31–48% of these patients (n=21, 46). In this latter group, the median difference between aortic RF values was 17% (IQR 13, 24; p<0.001), resulting in an AR severity classification that was a full grade lower. Conversely, when compared to volumetry, a lower median aortic RF was obtained in only 18% of patients (n=10) with direct regurgitant jet quantification.

Table 3.

Aortic regurgitation severity grading by 4DF study methods and volumetry.

Method (n = 186)	Aortic RF (%)	Mild (0–15%)	Moderate (16–30%)	Severe (>30%)	p-valuea
SV Difference	21.0 (11.8, 29.4)	63 (34%)	80 (43%)	43 (23%)	N/A
Conventional 4DF*
Aortic Valve (S) Aortic Valve (T) STJ AAo	8.0 (3.0, 17.0) 8.0 (3.0, 17.0) 15.0 (5.0, 27.0) 10.0 (3.0, 23.0)	133 (72%) 136 (73%) 96 (52%) 112 (60%)	36 (19%) 37 (20%) 58 (31%) 51 (27%)	17 (9%) 13 (7%) 32 (17%) 23 (12%)	<0.001 <0.001 0.003 <0.001
Direct Jetb
Aortic Valve STJ AAo	19.0 (12.0, 29.0) 21.0 (13.0, 31.0) 21.0 (13.0, 31.0)	68 (37%) 62 (33%) 63 (34%)	81 (44%) 74 (40%) 72 (39%)	37 (19%) 50 (27%) 51 (27%)	0.72 0.68 0.58

RF regurgitant fraction, 4DF four-dimensional flow, SV stroke volume, (S) static annular plane, (T) annular tracking,

STJ sinotubular junction, AAo ascending aorta

Data are represented as medians with interquartile ranges for continuous variables and n (%) for categorical variables

^*Aortic RF was calculated via measurement of forward and regurgitant flow volumes in the specified aortic plane.

^ap-value is reported using volumetry as the reference standard.

^bDirect jet aortic RF was calculated via direct measurement of the regurgitant jet volume. Aortic forward flow was measured separately in the specified aortic plane.

3.4 Impact of slice location on aortic flow measurements

Agreement between aortic FF measurements obtained at each aortic level and RVol quantification between study methods is summarized in Supplemental Table 1 and graphically depicted in Fig. 5. No statistically significant differences were noted in aortic flow measurements obtained at the aortic annulus, irrespective of correction for annular valve plane motion (p = 1.0). Good overall agreement was observed between FF measurements at each aortic level (ρ = 0.93–0.95), with the median differences between aortic slices comprising only 1–5% of the total FF rates. LV SV additionally appeared to be an accurate surrogate for aortic FF (ρ = 0.86–0.89), with the strongest correlation noted with FF measurements obtained at the aortic annulus (median difference: 1.9 mL/beat, IQR: −5.2, 10.8; p = 0.5). However, when compared to volumetry, absolute FF measurements obtained at the STJ and AAo were lower in upwards of 75% (140/186) of patients (median difference: 8.7–9.4 mL/beat; p < 0.001), including in 79% (90/114) with significant aortic dilation (p < 0.001).

Fig. 5.

Aortic forward (a) and regurgitant (b) flow quantification via each study method. No significant differences were observed in forward flow measurements obtained at each aortic level, which similarly demonstrated good overall correlation with the left ventricular SV, as assessed on planimetry. When compared to flow quantification in conventional aortic imaging planes, RVol were significantly higher with direct regurgitant jet measurement and volumetric-based assessment. AV (S) aortic valve (static plane), AV (T) aortic valve (annular tracking), STJ sinotubular junction, AAo ascending aorta, LV left ventricular, SV stroke volume, RVol regurgitant volume

Aortic RVol were similarly closely correlated at each aortic level (ρ = 0.70–0.89). However, the absolute RVol quantified between 4DF study methods was notably discrepant (direct jet RVol: 17.7 mL/beat v. conventional RVol: 6.5–11.4 mL/beat; p<0.001). This difference was most pronounced at the aortic annulus, where RVol were nearly 40% lower using conventional 4DF quantification techniques (direct RVol: 17.7 mL/beat v. annular RVol: 6.5 mL/beat; p<0.001).

3.5 Aortic RF quantification: influence of valve morphology, jet characteristics, and aortic dilation

Given the marked anatomic heterogeneity observed in complex CHD, subgroup analyses were performed to evaluate for any differential aortic RF quantification between 4DF study methods in patients with bicuspid aortic valves, dilated aortas, or eccentric regurgitant jets (Table 4). As observed in the broader study cohort, aortic RF quantification via direct jet measurements were more closely correlated with volumetry in each subgroup (r = 0.70–97 v. r = 0.65–0.71). However, an even stronger correlation was observed in those with significant aortic dilation (r = 0.95–0.97 v. r = 0.69), where direct regurgitant jet measurements resulted in significantly more precise aortic RF quantification (median difference: 0–2% ± 3–4% v. 7% ± 11%, p<0.001) (Fig. 6). This finding was similarly attributable to the relative underestimation of aortic RVol with quantification more distally in the dilated aorta (STJ RVol: 12.3 mL/beat, AAo RVol: 8.8 mL/beat v. direct RVol: 19.1 mL/beat; p<0.001). In fact, in nearly all (90%, 103/114) patients with significant aortic dilation (≥4.0 cm), standard regurgitant flow quantification at the STJ or AAo resulted in a full grade lower AR severity classification.

Table 4.

Subgroup correlation of aortic RF quantification by 4DF study methods v. volumetry.

Method	Pearson (r)*	Mean Difference (%)	p-value
Aortic Dilation
STJ v. volumetry Direct (STJ) v. volumetry AAo v. volumetry Direct (AAo) v. volumetry	0.69 0.95 0.69 0.97	7.0 ± 11.0 2.0 ± 4.0 7.0 ± 11.0 0.0 ± 3.0	<0.001 0.19 <0.001 0.78
Eccentric Jets
STJ v. volumetry Direct (STJ) v. volumetry AAo v. Volumetry Direct (AAo) v. volumetry	0.65 0.70 0.65 0.72	3.0 ± 12.0 −2.0 ± 10.0 7.0 ± 11.0 −1.0 ± 10.0	<0.001 0.08 <0.001 0.08
Bicuspid Aortic Valve
STJ v. volumetry Direct (STJ) v. volumetry AAo v. volumetry Direct (AAo) v. volumetry	0.71 0.78 0.67 0.79	4.0 ± 11.0 −2.0 ± 9.0 7.0 ± 11.0 −2.0 ± 9.0	0.002 0.04 <0.001 0.02

RF regurgitant fraction, 4DF four-dimensional flow, STJ sinotubular junction, AAo ascending aorta

Mean aortic RF differences were calculated comparing each 4DF study method and volumetry. Mean differences are reported as percentages, with the corresponding standard deviation and p-value depicted. When direct regurgitant jet measurements were used for regurgitant flow quantification, the aortic plane utilized for forward flow measurements is displayed in parentheses

^*p-values for Pearson’s correlation analyses were all significant (p<0.001).

Fig. 6.

Bland-Altman plots comparing aortic RF quantification between 4DF study methods and volumetry in patients with aortic dilation (≥4.0 cm or Z-score ≥ +2). The mean aortic RF (solid line) and difference between aortic RF measurements (dashed lines) on 4DF CMR versus volumetry are depicted. Direct jet aortic RF values (r = 0.95–0.97 v. r = 0.69) were more closely correlated with volumetry in patients with significant aortic dilation. RF regurgitant fraction

4. Discussion

To the best of our knowledge, this study presents the largest analysis of AR quantification by 4DF CMR in patients with moderate or complex CHD. The principal findings of this study are (1) direct regurgitant jet measurement shows improved agreement with volumetry in AR quantification when compared to conventional 4DF CMR techniques; (2) quantification of aortic regurgitant flow distal to the regurgitant jet orifice results in significant underestimation of AR severity; and (3) agreement between 4DF CMR AR quantification techniques is particularly reduced under non-laminar or complex flow states. Although the accuracy of AR quantification by 2DPC and 4DF CMR is well documented in the literature, prior studies have largely been limited to evaluation of patients with degenerative or bicuspid aortic valve disease [7], [9], [29]. The present study included a broad pathophysiologic spectrum of congenital AR, including those with congenital aortic valve dysplasia, congenital aortopathies (conotruncal defects, connective tissue disorders), and prior surgical aortic valve and/or root interventions. For the CHD imaging specialist, this mimics the breadth of patients encountered clinically.

No prior studies have analyzed the utility of AR quantification via direct regurgitant jet measurement on 4DF CMR. However, direct regurgitant jet measurement has several theoretical advantages when compared to 2DPC CMR and conventional 4DF CMR AR quantification techniques. With the latter techniques, aortic imaging planes are prescribed to optimize assessment of aortic forward flow, often at the expense of quantification of regurgitant flow further distal from the regurgitant jet orifice. As expected, measurements obtained in the more distal aorta result in significant underestimation of regurgitant flow volumes [23]. Alternatively, with direct jet measurement, the regurgitant flow itself is separately quantified, thereby enabling maximal forward and regurgitant flow quantification [23].

Direct jet measurement additionally ensures maximal regurgitant flow volume quantification via improved regurgitant flow tracking. With flow tracking, measurement planes can be manually adjusted throughout the cardiac cycle to ensure optimal alignment with the regurgitant jet, and therefore often yield more accurate and higher regurgitant flow estimates [30]. In the setting of turbulent flow or significant jet eccentricity, as commonly encountered in patients with complex CHD or mixed aortic valve disease, flow tracking additionally allows for improved regurgitant jet streamline visualization and analysis [31]. With improved characterization of regurgitant flow patterns, the measurement plane can be repositioned to a region of more coherent flow proximal to the regurgitant orifice, thus minimizing intravoxel phase dispersion and signal loss via “visually-targeted” quantification [32]. Use of flow tracking has previously been shown to be an accurate and reproducible means of transvalvular flow quantification in the setting of atrioventricular valve (AV) regurgitation [30], [33], [34], although this is the first study to demonstrate similar applicability in AR.

Additionally, our findings suggest that flow tracking may provide a superior means of regurgitant flow quantification under non-laminar or complex flow states when compared with valve tracking. Valve tracking is a well validated and reliable means of flow correction for annular through-plane motion and has been shown to improve agreement in transvalvular flow quantification between different CMR modalities [24], [34]. However, in the present analysis, regurgitant flow volumes were not significantly changed with use of valve tracking, and still relatively underestimated when compared to values obtained with direct jet quantification. While valve tracking may be more applicable to the quantification of AV valve regurgitation, where multidirectional annular motion is more commonly observed [35], [36], prior studies in patients with mitral regurgitation have shown a similar underestimation of regurgitant flow volumes with use of valve tracking when compared to direct regurgitant jet quantification [30].

Thoracic aortic aneurysmal disease represents another altered hemodynamic flow state that may impact accurate aortic RF quantification in patients with complex CHD. In the present cohort, the underestimation of aortic forward and regurgitant flow volumes was even more pronounced in patients with aortic dilation. In the presence of significant aortic ectasia, flow dynamics in the distal ascending thoracic aorta are fundamentally altered, with accentuated flow vortices and supraphysiologic flow helicity more commonly observed [37]. This is the proposed mechanism by which “vortex rings” are created, ultimately resulting in lower distal net flow measurements due to partial-voluming with bidirectional flow [38].

Current consensus guidelines recommend aortic flow quantification at the STJ [15], although the ideal aortic plane for flow quantification in complex CHD has not been well established. As in the present analysis, prior studies including patients with CHD or other complex flow states have similarly shown improved correlation between aortic FF measurements and volumetry when measurements are obtained at the aortic annulus [8], [12], [29], [31]. However, significant discrepancies still persist, with some studies showing improved agreement with more distal flow quantification [4], [39]. Although lower FF measurements in distal aortic planes may be partially explained by coronary blood flow losses, discrepant flow measurements in complex CHD can additionally be explained by the presence of complex multidirectional flow and the corresponding partial-voluming incurred with significant aortic ectasia [31]. Therefore, given the significant anatomic heterogeneity observed in complex CHD, defining a standard plane for aortic flow quantification may not be feasible. Rather, given the improved regurgitant jet streamline visualization achievable with 4DF CMR, selection of an individualized “optimal” measurement plane may be preferable.

5. Limitations

While our study has several strengths, there are important limitations to be acknowledged. This study was performed at a comprehensive CHD care center with expertise in 4DF CMR image acquisition and post-processing analysis. Complex post-processing analyses are time intensive and require specialized training to ensure the accuracy and reproducibility of measurements. Therefore, use of 4DF CMR for routine AR quantification may not be feasible at all centers. However, there is mounting evidence supporting the accuracy of semiautomated 4DF analytical techniques, which may permit its more routine and widespread application [11], [24], [30], [36]. Additionally, two different CMR systems were used for image acquisition, with different magnet strengths, slight variations in acquisition parameters (i.e. HR, VENC), and different contrast agents utilized. However, as intermethod comparisons were performed in a paired fashion, this was unlikely to have significantly confounded our results. Although no significant differences in AR severity classification were observed based upon the type of MRI contrast agent utilized, further investigation is warranted to determine the comparable accuracy of various MRI contrast agents on AR quantification by 4DF CMR. Lastly, in the absence of any “gold-standard” for CMR-based AR quantification, volumetry was used as the reference standard for comparative analyses. As volumetric-based measurements were quantified on multiplanar SSFP sequences, which employed a standard end-expiratory breath hold, this could have resulted in relatively higher LV SV measurements when compared with aortic FF measurements obtained on free-breathing 4DF sequences. However, this would not account for the discrepancy in regurgitant flow measurements observed between 4DF study methods.

6. Conclusion

In conclusion, our study offers evidence supporting use of 4DF CMR for quantification of AR in complex CHD. As with 2DPC CMR, regurgitant flow quantification distal to the regurgitant jet orifice may result in significant underestimation of AR severity, especially in patients with significant aortic dilation. Direct regurgitant jet measurement may be preferable under these complex flow states and demonstrates comparable accuracy in aortic RF quantification to volumetry. Further investigation is required to determine if direct aortic regurgitant jet quantification by 4DF CMR provides incremental risk stratification over standard TTE-based parameters in predicting adverse LV remodeling and corresponding requirement for surgical intervention in congenital AR.

Funding

The authors have no financial disclosures or relationships to industry.

Author contributions

Makoto Takei: Formal analysis. Daniel E. Clark: Writing – review & editing, Methodology, Conceptualization. Brynn Connor: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Shiraz A. Maskatia: Writing – review & editing, Methodology, Conceptualization.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. . supplementary material

Supplementary material

.