Introduction
Chronic aortic regurgitation is a cause of significant cardiac morbidity in children. Reference Ibrahim, Borrelli and Krupickova1 It is rarely a primary lesion and commonly occurs after aortic valvuloplasty for congenital aortic valve stenosis due to leaflet dysplasia. Progressive LV distension ensues secondary to the chronic volume loading and may lead to ventricular dysfunction that can be irreversible once a threshold volume is reached. Reference Selamet Tierney, Gal and Gauvreau2 Current adult guidelines for intervention in chronic aortic regurgitation are based on symptoms and two-dimensional echocardiography measurements of the left ventricle and function; however, these threshold dimensions do not apply to children. Reference Otto, Nishimura and Bonow3 In addition, children are often asymptomatic. MRI-based volume and function guidelines exist in the management of valvular regurgitation from chronic pulmonary regurgitation after tetralogy of Fallot repair, but there are no similar guidelines for aortic regurgitation despite the increasing use of MRI in the evaluation of aortic regurgitation. The limited data to guide surgical decision-making in children with chronic aortic regurgitation forms the premise of our study. We examined associations between MRI quantification of aortic regurgitant fraction, left ventricular volumetric function and echocardiographic measurements, and the subsequent need for surgical intervention in a paediatric and young adult population. The aim of this study was to examine cardiac MRI cut-off values for surgical intervention in children and young adults with chronic aortic regurgitation.
Materials and methods
A retrospective study of children and young adults with chronic aortic regurgitation who underwent at least one MRI was performed. Inclusion criteria included patients aged 1–21 years with aortic regurgitation for at least 1 year secondary to a bicuspid or dysplastic aortic valve. Exclusion criteria were more than mild aortic stenosis (velocity of > 2.2 m/sec or mean gradient of > 20 mmHg, on echocardiography or MRI), additional tricuspid, mitral, or pulmonary valvular regurgitation or known genetic conditions. The cohort consisted of two groups based on those whom aortic valve surgery was performed or not, within 1 year of the MRI. The timing and need for surgery were determined by the primary cardiologist based on a combination of symptomatology, clinical examination, and echocardiography findings. Exercise testing was available in 5/20 (25%) of the surgical group at the time of decision-making. MRI data was available in 17 out of 20 (85%) patients at the time of decision-making for surgical intervention with three patients (15%) undergoing MRI after the decision for surgery was made. Clinical variables collected included age, gender, height, weight, body surface area, heart rate, aortic valve morphology, aortic root z scores, and cardiac symptoms. A waiver of consent was obtained for this retrospective study which was approved by the University of Alberta Health Research Ethics board. All scans were performed on a Siemens Aera 1.5 T scanner (Siemens Medical Solutions, Erlangen, Germany) and analysed using CVI 42 software (Circle Cardiovascular Imaging, Calgary, Canada) to calculate left ventricular volumes, mass, ejection fraction, and aortic regurgitant volume and fraction. Sedation was used for children younger than 8 years. Electrocardiogram-gated balanced steady-state free precession imaging was used for cine imaging with 8–10 s inspiratory breath hold to minimise respiratory cardiac motion, whereas free breathing imaging was used for flow assessment and during sedation. Cine imaging data was divided into 25–30 phases using retrospective gating (temporal resolution 41–70 ms; echo time 1.40 to 1.54 ms; repetition time 2.80 to 3.08 ms; field of view 380 × 380 mm; flip angle 50 to 60°). Using semi-automatic contouring, the end-diastolic volume, end-systolic volume, stroke volume, ejection fraction, and mass were calculated and indexed to body surface area. Left ventricular dimensions were taken in end diastole and end systole from the three-chamber view. Phase-contrast velocity-encoded imaging through the aorta at the level of the right pulmonary artery (temporal resolution 25 to 55 ms; echo time 2.6 to 3.2 ms; repetition time 4.3 to 7.8 ms; field of view 320 × 320 mm; velocity window 2.5 to 4.0 m/s) was used to measure forward and reverse flow to calculate the regurgitant volume indexed to body surface area and the regurgitant fraction (%). All MRI data was collected from the report.
Statistical analysis
Continuous variables are described as mean ± standard deviation. Categorical variables are presented by frequency distribution and compared using Fisher’s exact test. MRI parameters were compared between the surgical and non-surgical groups using t-tests. Receiver operating characteristics analysis was performed to identify parameters with discriminatory ability towards the primary end point of surgery (area under the curve > 0.7). Cut-off values were determined using the Youden index.
Results
Out of a total of 40 patients (mean age 15 ± 5 years), 90% had a bicuspid aortic valve, 5% unicuspid, and 5% trileaflet. Balloon valvuloplasty had been performed in 11 (28%) of patients and surgical repair in 6 (15%). Additional procedures included coarctation of the aorta repair in five (12.5%) patients and ventricular septal defect closure in two patients. Symptoms were present in 28% of the whole group and included exertional dyspnea, decreased activity tolerance, dizziness, palpitations, or chest pain. The aortic root z scores of 2.3 (0.86–3.1) indicated that the aortas were not severely dilated. There were 20 patients who did not undergo surgery and 20 patients who underwent surgical intervention within 1 year of the MRI. Both groups were comparable for age, body size, symptoms, cardiac medications, and previous interventions (Table 1).
Demographic & clinical data comparison of the non-surgical and surgical groups
BSA, body surface area; VSD, ventricular septal defect.
* Symptoms included exertional dyspnea, decreased activity tolerance, dizziness, palpitations or chest pain.
The surgical group had significantly larger left ventricular end-diastolic volume (146 ± 42 versus 113 ± 37 ml/m2, p = 0.02); end-systolic volume (64 ± 22 versus 45 ± 20 ml/m2, p = 0.0009), and mass (100 ± 29 versus 79 ± 23 g/m2, p < 0.0001); lower ejection fraction (56 ± 6 versus 60 ± 7%, p = 0.03) and higher aortic regurgitant volumes (32 ± 16 versus 18 ± 10 ml/m2, p = 0.001) and regurgitant fraction (49 ± 13 versus 31 ± 13%, p = 0.02) compared to the non-surgical group (Table 2). There were no differences in mass:volume ratio or peak aortic velocity between the two groups. Receiver operating curves showed that ventricular volumes, ejection fraction, regurgitant volume and fraction demonstrated significant discriminatory power to distinguish between the surgical and non-surgical groups (Figure 1). The parameter with the highest discriminatory power was aortic regurgitant fraction with a sensitivity and specificity of 85% at a cut-off of 36% (Figure 2a). The second highest predictor was end-systolic volume with a cut-off value of 53 ml/m2 (sensitivity 85% and specificity 70%) to distinguish between the two groups (Figure 2c). This was followed by indexed aortic regurgitant volume, end-diastolic volume, and then ejection fraction (Figure 2b, d & e, Table 3). The two-dimensional left ventricular diameters showed moderate correlations with ventricular volumes (end-diastolic, r = 0.5 and end-systolic, r = 0.57).
Receiver operating curves for MRI parameters in predicting need for surgery. EDVi = end-diastolic volume indexed; EF = ejection fraction; ESVi = end-systolic volume indexed; LV = left ventricle; RF = regurgitant fraction.
Graphs displaying the differences in MRI parameters between patients who had no surgery versus those who underwent surgery. Data depicts the mean value and spread of the results with the dashed black line representing the cut-off values. EDVi = end-diastolic volume indexed; EF = ejection fraction; ESVi = end-systolic volume indexed; LV = left ventricle.
Cardiac MRI data comparisons between the non-surgical and surgical groups
EDVi = end-diastolic volume indexed; EF = ejection fraction; ESVi = end-systolic volume indexed; LV = left ventricle; RF = regurgitant fraction.
Sensitivity and specificity of cut-off values for surgical repair of aortic valve regurgitation
EDVi = end-diastolic volume indexed; EF = ejection fraction; ESVi = end-systolic volume indexed; LV = left ventricle; iRVol = indexed regurgitant volume; RF = regurgitant fraction.
Discussion
In a group of children and young adults with chronic aortic regurgitation, we identified MRI parameters that differentiated between those who did and did not undergo surgical intervention. Aortic regurgitant fraction demonstrated the highest discriminatory power followed by left ventricular end-systolic volume, aortic regurgitant volume, end-diastolic volume, and ejection fraction. The lack of comprehensive guidelines for the management of chronic aortic regurgitation in children is well recognised, highlighting the need for prognostic studies to inform clinical decision-making in this population. Reference Ibrahim, Borrelli and Krupickova1,Reference Gentles, French, Zeng, Milsom, Finucane and Wilson4
While current guidelines do recommend MRI in the evaluation of aortic regurgitation, class 1 indications for surgical intervention are based on symptoms and ejection fraction, while class 2 indications are based on left ventricular ejection fraction and end-systolic dimension. Reference Otto, Nishimura and Bonow3 We showed a moderate correlation between left ventricular linear dimension and volumes, and it is known that linear dimensions are limited by measurement error and can vary with the pattern of left ventricular enlargement. Reference Dujardin, Enriquez-Sarano, Rossi, Bailey and Seward5 MRI volume parameters have shown superior ability to discriminate outcome risk than left ventricular linear dimensions. Reference Malahfji, Crudo and Kaolawanich6 Comparison of the 2020 guidelines, 7 years from the 2014 guidelines, show little changes in the imaging criteria for surgical intervention which remain as two-dimensional measurements despite the growth and development of quantitative MRI during that time period. Reference Otto, Nishimura and Bonow3,Reference Nishimura, Otto and Bonow7 Sannino and Meucci in their editorial comment stated that these guidelines were supported by old and relatively small studies. Reference Sannino and Meucci8 At the same time, they acknowledged the expanded growth of MRI applications suggesting that left ventricular volumes may be a better parameter to guide decisions for aortic valve intervention. Reference Anand, Yang and Luis9,Reference Yang, Michelena and Scott10 The American College of Cardiology has identified an evidence gap in the relationship between left ventricular volumes and outcomes in aortic regurgitation, and data on the prognostic value of MRI is limited to small numbers. Reference Sannino and Meucci8
MRI has shown superior ability over echocardiography to predict the need for valve surgery due to its ability to quantify left ventricular volumes and aortic regurgitant volume and fraction. Reference Harris, Krieger and Kim11 Harris et al found that an aortic regurgitant volume > 50 ml on MRI was 100% sensitive in identifying patients who required aortic valve surgery; however, the number of patients meeting their primary end point was small. Reference Harris, Krieger and Kim11 One of the larger studies (n = 458) examined adults with aortic regurgitation who met the primary outcome of decrease in ejection fraction < 50%, development of symptoms, guideline indications for surgery, or death. They reported left ventricular volume cut-offs of end-systolic volume 43 ml/m2, end-diastolic volume 109 ml/m2, and end-systolic dimension 2 cm/m2 as predictive of their primary outcome. Reference Malahfji, Crudo and Kaolawanich6
The threshold values from adult studies are not applicable to younger children and infants, especially considering the growth potential at a younger age. This indicates a need for more studies in the paediatric group. Even if volumes or dimensions are indexed to body surface area, the relationship is non-linear and heteroscedastic in children which limits the direct application of indexed parameters from adults to children. Reference Gentles, French, Zeng, Milsom, Finucane and Wilson4 While regurgitant fractions may be more applicable across all age groups, the reported surgical threshold values vary throughout the literature. Some of the largest studies have reported predictive values as cut-off for aortic regurgitation fraction ranging from 33% Reference Myerson, d’Arcy and Mohiaddin12 , 35% Reference Vejpongsa, Xu, Quinones, Shah and Zoghbi13 , or 43%. Reference Malahfji, Crudo and Kaolawanich6 Our cut-off value of 36% in a younger population fits within these reported values. The range of regurgitant fraction thresholds emphasises the need for more precise definition of cut-off values that can be incorporated into the guidelines for aortic intervention.
There are variable reports on the optimal location for aortic flow measurements in bicuspid aortic valves due to the complex flow patterns downstream. An adult study suggested that flow should be measured at the aortic valve or outflow tract as measurement in the ascending aorta can underestimate the forward flow in bicuspid aortic valves. Reference Muzzarelli, Monney, O’Brien, Faletra, Moccetti, Vogt and Schwitter14 However, using four-dimensional flow comparison, another study found the best agreement with two-dimensional flow was at the level of the pulmonary artery, as we have performed. Reference Polacin, Geiger, Burkhardt, Callaghan, Valsangiacomo and Kellenberger15 While these were adult studies, we do not have similar studies in a purely paediatric population. The fact that none of our patients had severe aortic stenosis or significantly dilated aortic roots which are contributors to abnormal flow currents likely minimises these errors. Our cut-off values are therefore only applicable if the same methods are used.
The few studies in children that have examined predictive parameters after surgical aortic valve replacement in aortic regurgitation are based on echocardiography data. These found that preoperatively decreased ejection fraction and larger left ventricular dimensions or volumes were predictive of ventricular dysfunction after aortic valve surgery. Reference Selamet Tierney, Gal and Gauvreau2,Reference Gentles, French, Zeng, Milsom, Finucane and Wilson4,Reference Cox, Walton, Bartz, Tweddell, Frommelt and Earing16,Reference Lowenthal, Tacy, Behzadian and Punn17 The higher discriminatory power of end-systolic over end-diastolic volume in our study is in line with these and adult studies showing that end-systolic parameters were more predictive than end-diastolic parameters. Reference Harris, Krieger and Kim11,Reference Cox, Walton, Bartz, Tweddell, Frommelt and Earing16 This is likely due to end-systolic parameters more reflective of ventricular function and afterload. Reference Gentles, French, Zeng, Milsom, Finucane and Wilson4 The finding of higher post-operative end-systolic volume z scores as an independent risk factor for post-operative left ventricular dysfunction indicates that a combination of volumetric and functional assessment is required. Reference Gentles, French, Zeng, Milsom, Finucane and Wilson4 It is known that echocardiography tends to underestimate left ventricular volumes compared to MRI which is a limitation of echocardiography-based indices. Furthermore, the discrepancies in left ventricular measurement between echocardiography and MRI are worse in children with aortic regurgitation, stressing the need for accurate MRI volume measurements to guide clinical management. Reference Barros, Udine, Spurney, Olivieri and Loke18
The surgical options for aortic valve replacement are limited in young infants. Aortic valve repairs have limited longevity, and valve replacements require repeat intervention as the child grows. Thus, clear indications for the timing of surgery need to be developed in order to preserve ventricular function and reduce the number of interventions throughout their lifetime. While there was a significant difference between the left ventricular ejection fraction in both groups, the mean values and majority of the patients had values considered within the normal range, suggesting that ejection fraction is preserved until later in the course of the disease and should not be used as the only indication for surgical intervention. Since children have to live longer with their disease than adults, cut-off values using volume or regurgitant parameters to guide the timing of intervention are necessary in order to preserve left ventricular function before it declines.
Limitations
Our study only describes the difference in MRI parameters between surgical and non-surgical groups; but due to the retrospective nature, it is not able to make inferences about criteria for intervention. Decision-making for surgical intervention although based on symptoms and echocardiography data at our institution was likely also influenced by MRI data when available in 85% of patients. While it is likely that the MRI data may have played a confounding role, our decisions were based on clinical data as we had no MRI threshold values to guide these decisions. In addition, 15% of patients did not have MRI data at the time of decision-making as MRI is not a routine part of the decision-making process at our institution, which only slightly diminishes its influence as a potential confounder. We also do not have follow-up MRI data post-surgical intervention to determine if these cut-offs are predictive but aim to study this in the future.
Conclusions
We have shown in a group of children with chronic aortic regurgitation that aortic regurgitant fraction had the highest discriminatory power to distinguish patients who required surgical intervention with a cut-off value of 36%. Additional parameters, in order of discriminatory power, were left ventricular end-systolic volume 53 ml/m2, indexed aortic regurgitant volume 21 ml/m2, left ventricular end-diastolic volume 120 ml/m2, and then ejection fraction 59%. There is a lack of MRI studies with outcome data and cut-off values in children. Our study is one of the first to describe the differences in MRI parameters between surgical and non-surgical groups. Longitudinal studies are required to determine if these parameters are predictive and assess left ventricular remodelling after aortic valve surgery. However, they may serve as a useful guide in the younger population and serve as a basis for future prospective studies.
Acknowledgements
The authors wish to thank our dedicated MRI technicians who performed these studies: Wendy Chu, Rebecca Gray, Melissa Grzeszczak, Kelley Justice, Kam Ma and Justine Muller.
Financial support
The authors received no financial support for the research, authorship, and/or publication of this article.
Competing interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article
Ethical standard
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the University of Alberta Health Research Ethics Board (no. Pro0087295) on March 29, 2019, with the need for written informed consent waived.
Open access
