MRI Validated Echocardiographic Technique to Measure Total Cardiac Volume: A Tool for Donor Recipient Size Matching In Pediatric Heart Transplantation

Abstract

Background

Our aim is to develop and validate an accurate method for estimating total cardiac volume (TCV) using standard echocardiographic imaging that can be easily employed to aid in donor recipient size-matching in pediatric heart transplantation.

Methods

30 patients who underwent echocardiography and cardiac magnetic resonance imaging (cMRI) were identified. TCV was measured on cMRI. TCV was determined echocardiographically by two methods: a volume measurement using the modified-Simpson’s method on a 4-chamber view of the heart; and a calculated volume measurement which assumed a true-elliptical shape of the heart. These two methods where compared to the value obtained by cMRI using the concordance correlation coefficient (CCC).

Results

TCV by method 1 correlated well with cMRI (CCC = 0.98%, CI=0.97, 0.99). TCV by method 2 had a CCC = 0.90 (CI = 0.9464, 0.9716) when compared to cMRI. Left ventricular end-diastolic volume (LVEDV) also correlated as a predictor of TCV in patients with structurally normal hearts and could be described by the equation: TCV = 6.6 (LVEDV) + 12 (R² = 0.97).

Conclusion

Echocardiographic assessment of TCV for recipients and their potential donors is a simple process and can be prospectively applied as part of donor evaluation.

Introduction

The outcomes for cardiac transplantation have improved over time, and this therapy has become a common approach for patients with end stage heart disease. A fundamental challenge in heart transplantation, especially in pediatric cases, is determining the suitability of the size of a donor organ for a particular recipient. Currently, size matching is based on weight; however, this method is not a consistent predictor of “good match.” Although no uniform criteria for size matching in pediatric heart transplantation exist, guidelines based on body weight have been proposed. Generally, for infant recipients, the donor body weight should not be more than 300% of the body weight of the recipient. For recipients greater than 18 months, it has been suggested that the body weight of the donor should be no more than 25–50% greater than the recipient’s weight. The use of hearts from donors who weigh less than the recipient has been discouraged, though in older children and adolescents, hearts from smaller donors may be used as long as the weight is no less than 75% of the recipient’s weight¹. Liberalization of recipient criteria as well as expansion of the donor pool has been reported by some institutions to have increased the acceptance of undersized hearts, potentially leading to more postoperative complications². The lack of strict data derived guidelines for donor-recipient size matching may actually reflect the fact that weight itself is not an ideal surrogate marker for cardiac size. Some institutions, in fact, have reported using height rather than weight for donor-recipient matching, arguing that edema can affect weight, and that weight therefore could be a poor predictor of cardiac workload requirements³.

Size mismatch has been identified as a potential mortality risk in pediatric cardiac transplantation⁴. Fullerton et al reported that large size mismatches were well tolerated in the pediatric population⁵. However, Tamisier et al found in a review of their institution’s transplants that while oversized hearts were typically well tolerated; undersized hearts remained a significant cause of early mortality⁶. In adult heart transplant recipients with elevated pulmonary vascular resistance, Patel et al found that oversized hearts had better outcomes while undersized grafts did poorly⁷. While oversized grafts may often be well tolerated, the postoperative course may be more challenging due to a need for delayed sternal closure or a propensity for compressive atelectasis.

Routinely, all potential cardiac donors undergo echocardiographic evaluation prior to offer. A facile, easily reproducible method of determining total cardiac volume (TCV) by echocardiography would aid in donor-recipient size matching. We sought to develop and evaluate an echocardiographic measure of TCV that may aid in donor-recipient size matching in order to minimize complications related to size mismatch and potentially expand the acceptable donor pool for a given recipient.

Methods

Pediatric patients who had undergone both cardiac magnetic resonance imaging (cMRI) and echocardiography (Echo) within a 2 month period were identified retrospectively. Demographic data including age at time of study, body surface area (BSA), and diagnosis were recorded. cMRI was performed on a Siemens Symphony 1.5 Tesla magnet and was used as the reference standard for determination of TCV. Contiguous axial dark-blood HASTE images are routinely obtained from the diaphragm inferiorly through the great vessels superiorly as part of all cMRI studies in our institution. These static axial images obtained in diastole were used to estimate the TCV. The epicardial border was traced on all slices through the heart, and TCV was determined by summing the product of the cardiac area with known, fixed thickness for each slice (Figure 1).

Figure 1.

Total cardiac volume (TCV) determined by cMRI. Volumes for each slice were determined by multiplying the area by the known slice thickness. These volumes were summated from apex to base to obtain the total cardiac volume.

TCV was determined by Echo using two candidate techniques. Both techniques were performed using measurements made offline on a Siemens Syngo Dynamic Workstation. Method 1 employed a modified Simpson’s method applied to a standard 4-chamber apical view (Figure 2). Method 2 assumed an elliptical shape of the heart and volume was calculated using the equation: V = 4/3(π)(X)(Y)(Z), where V is volume; Y is the longest dimension along the major axis of the heart; X is the longest dimension perpendicular to Y; and Z is measured along the septum from an anterior-to-posterior direction in the parasternal short-axis view (Figure 3). Left ventricular end-diastolic volume (LVEDV) was also measured on those patients with structurally normal hearts using the modified Simpson’s method from the apical four chamber view. Echocardiographic estimates of TCV were then compared to the volumes obtained by cMRI using Bland-Altman plots and concordance correlation coefficients (CCC). In a Bland Altman the difference of two measures is plotted against the average (which is used to represent the true value). The mean of the difference is the bias of using one method rather than the other. The CCC measures the variation around the line of agreement. A value of ≥ 0.99 reflects almost perfect agreement and a value 0.95 is substantial agreement. Further each of the two calculated measures by two independent raters was compared using CCC. Linear regression of the cMRI volume as the dependent and LVEDV as the independent measure was also performed in patients with normal cardiac anatomy. This regression was employed to derive a line of best fit describing the relationship between LVEDV by echo and TCV by cMRI in patients with structurally normal hearts. To provide further validation, we performed TCV assessment using the preferred echo technique herein described for an additional 100 patients with normal hearts. A line of best fit was then derived that allows for calculation of TCV by echo based on LVEDV obtained during the same echo study. Study data were collected and managed using REDCap electronic data capture tools hosted at the Children’s Hospital of Wisconsin.⁸

Figure 2.

A - apical view of a child with tricuspid atresia with intact ventricular septum and absent pulmonary valve. B - Method 1 Echo estimation of total cardiac volume using the modified Simpson’s method.

Figure 3.

Method 2 – Illustration of the “X, Y, and Z planes.” The Y plane is the longest dimension along the major axis of the heart; X plane is the largest dimension perpendicular to the Y plane; Z plane is measured along the septum from an anterior-to-posterior direction in the parasternal short-axis view.

Results

30 patients who had undergone both cMRI and Echo within a 2 month period were identified. These included 10 patients with normal hearts, 11 patients with cardiomyopathies, and 9 patients with congenital heart disease. Table 1 shows study subject demographics (at echo) while table 2 lists the spectrum of congenital heart disease. No significant changes in BSA were found between the dates of the echocardiogram and cMRI.

Table 1.

Subject Demographics

	Normals	Cardiomyopathy	Congenital Heart Disease
Number	10	11	9
Gender	6F 4M	3F 7M	3F 6M
Age yrs: mean(range)	9.8 (0.7–21)	12.6 (0.03–17.8)	7.3 (0.08–14)
BSA: mean(range)	1.1 (0.3–1.9)	1.6 (0.2–21)	0.8 (0.2–1.4)

Table 2.

Frequency of diagnosis of congenital heart diseases

Frequency of Congenital Heart Defects
Hypoplastic Left Heart Syndrome	3
Tricuspid Atresia with Intact Ventricular Septum	1
Tetralogy of Fallot	2
Double Outlet Right Ventricle with Subaortic Ventriculoseptal defect	1
Total Anomalous Pulmonary Venous Return	1
Scimitar Syndrome	1
Total	9

TCV by method 1 correlated extremely well with volumes obtained by cMRI with a concordance correlation coefficient (CCC) of 0.98, 95% CI (0.97, 0.99). TCV by method 2 also agreed well with a CCC of 0.90, 95% CI (0.95, 0.97). Further the inter observer CCC was 0.98, 95% CI (0.96, 0.99) and 0.98, 95% CI (0.98, 1) respectively. When evaluating each patient group separately, there appeared to be better correlation among normal hearts, cardiomyopathies, and congenital heart disease using method 1, (CCC =0.99, 0.97, 0.97), (Figure 4). Method 2 also correlated fairly well for each group (CCC = 0.98, 0.89, 0.93)) (Figure 5). Bland-Altman Plots were constructed to explore agreement between measurements obtained by method 1 and method 2 with cMRI for each diagnosis (Figure 6A&B). Again, the plots reveal that volumes obtained by method 1 had greater agreement with volumes obtained by cMRI than method 2.

Figure 4.

Total cardiac volume determined by method 1 compared to volume obtained by cMRI for each diagnosis with line of agreement. Volumes by method 1 have excellent agreement with volumes by cMRI with CCC ≥0.97.

Figure 5.

Total cardiac volume determined by method 2 compared to volume obtained by cMRI for each diagnosis with line of agreement. Volumes by method 2 have good agreement with volumes by cMRI with CCC≥.89; however, the agreement with method 1 is superior.

Figure 6.

Figure 6A. Bland-Altman plots for Method 1 vs. cMRI in normal hearts. Dashed lines indicate 20% of the mean variance; the lower horizontal line indicates the bias (mean of the difference).

Figure 6B: Bland-Altman plots for Method 2 vs. cMRI in normal hearts. Dashed lines indicate 20% of the mean variance; the higher horizontal line indicates the bias (mean of the difference).

LVEDV by echocardiography was recorded and correlated with TCV by cMRI for all patients with structurally normal hearts. R²=0.95 (p<0.0001). Linear fitting found that total cardiac volume as determined by cMRI could be estimated from the LVEDV. This relationship was initially defined using the 10 normal patients included in the cMRI/echo cohort. To provide a more robust relationship, a line of best fit was identified using 100 determinations of TCV by echo and their respective LVEDV also obtained by echo at the same study. As shown previously, the relationship between TCV by echo and TCV by cMRI becomes less consistent at larger heart sizes. For this reason, we chose to identify the line of best fit weighted to optimize the predictive value for patients with TCV < 400 ml. The resulting relationship between LVEDV and TCV for structurally normal hearts can be described by the equation TCV = 6.6 (LVEDV) + 12 (Fig 7a). Despite the increased scatter at higher cardiac volumes, this equation agrees well with data derived directly from the cMRI volumes across the spectrum of heart sizes (R² = 0.94 p<.001) (Fig 7b).

Figure 7.

Figure 7A. Total cardiac volume by modified Simpson’s (Method 1) plotted against left ventricular end-diastolic volume (LVEDV) for hearts with TCV < 400 ml.

Figure 7B: Line of best fit weighted for hearts with TCV < 400 derived from echo data applied to the full spectrum of hearts studied by cMRI.

Discussion

Donor-recipient size matching can be difficult, particularly in pediatrics. Currently, size matching is weight based. However, pediatric transplant patients often have asymmetric or dilated hearts and their weight may vary due to disease related factors such as failure to thrive or tissue edema; thus weight may not be a reliable predictor of suitability. A reliable, facile and quick estimation of total cardiac size may be a better predictor of how well a donor heart will fit into a given recipient.

In this study, we sought to develop and evaluate an echocardiographic measure of total cardiac volume. Comparing these methods to volumes obtained by cMRI shows that accurate estimates of total cardiac volume can be obtained by echocardiographic measures. Application of a modified Simpson’s method (method 1) appears to be an excellent indicator of total cardiac volume. This method proved to be superior to our alternative measurement technique and this may be for several reasons. First, the actual heart shape is more analogous to the shape of an egg rather than a true ellipse. Second, method 1 involves only one physical measurement whereas method 2 depends on measurements in 3 different dimensions. This triples the chance of measurement error. Another advantage of the modified Simpson’s approach is that it is easily obtainable from the apical imaging plane, which is a standard echocardiographic view. Both methods, however, had less agreement at larger heart sizes, likely due to difficulty in acquiring clear echocardiographic imaging of the anterior right ventricular free wall. Therefore, this method is particularly well suited for the pediatric heart transplant population.

Importantly, normal hearts were also found to have a nearly linear relationship between LVEDV and TCV. Recognizing that for the most part, all potential cardiac donors have structurally normal hearts; the only donor data needed to determine TCV is a standard LVEDV which is a routine echocardiographic measurement. Hence, using the formula we obtained, donor TCV can be easily calculated at the time of offer and compared to the recipient TCV as part of the donor evaluation process.

Recently at our institution, we have prospectively applied this technique for all cardiac transplant recipients. Our current practice is to measure recipient TCV at the time of listing and then back calculate a donor LVEDV that would represent a size matched recipient. This allows for rapid screening at the time of a donor offer. Early in our experience, we had two interesting cases that may be particularly illustrative. The first case was a 4 month old girl with dilated cardiomyopathy requiring inotropic support. The patient’s weight at the time of transplant was 4 kilograms. A donor heart was accepted from a donor weighing 10 kilograms giving a donor-recipient weight ratio of 2.5. There was some concern that the donor heart would be prohibitively oversized; however our TCV determination showed the donor heart to be approximately 160ml while the recipient’s heart was 180ml. The patient underwent transplant with chest closure in the operating room without significant complications. A second patient with double outlet right ventricle, mitral atresia, and left ventricular hypoplasia underwent cardiac transplantation because he was not a candidate for single-ventricle palliation due to severe tricuspid insufficiency and severe heart failure. His weight at transplantation was 4.3kg and a donor heart was accepted from a child who weighed 10.4kg giving a donor-recipient weight ratio of 2.4. Using our echo technique, we determined the recipient’s cardiac volume to be 102ml. Donor heart TCV was estimated to be 200mL based on application of our best-fit equation to the LVEDV measured on donor echo. Transplant proceeded as the patient was critically ill and clinically declining. Although this patient ultimately exhibited end organ recovery and survived, the chest was unable to be closed for several weeks and the patient did have postoperative morbidity attributable to the oversized graft.

The concept of measuring donor and recipient TCV to facilitate decision making in pediatric cardiac transplantation is actually not an alternative to weight based matching but rather a new paradigm. The status quo of matching by weight includes a reliance on the dual assumptions that 1. donors are generally “normal” patients with an average relationship between weight and heart size and 2. recipient heart size to weight relationships can be visually estimated using chest radiography. Retrospective analysis of donor recipient size matching can only tell us about the strengths and limitations of donor-recipient weight ratios but cannot be informative as to the potential value of donor-recipient volumetric comparisons. Furthermore, there is no way to retrospectively assess the potential advantage of a volumetric approach; i.e. we cannot know whether TCV assessment could have allowed centers to accept a suitable organ declined due to unacceptable weight or prevented acceptance of a graft that proved to be a poor fit from a size perspective. Hopefully, with prospective evaluation and direct comparison to weight based decision making, we can determine the utility of this new approach to donor-recipient size matching in pediatric heart transplantation. Finally, with experience, the limits and implications of true size mismatching by volume can be determined.

There are some limitations to our study. First, the study population is relatively small. A second limitation is the assumption that cMRI is an accurate measure of total cardiac volume. We believe that cMRI is the best modality currently available to obtain volumetric data. Finally, the proposed linear equation has demonstrably higher concordance with both echo and cMRI derived TCV determinations at TCV < 400 ml. This may be due to the previously described challenge with capturing the complete cardiac image at larger heart sizes or it may be due to the relationship between LVEDV and TCV becoming non-linear with increasing size or age. Despite this limitation, the predictive value is highly concordant in the pediatric age range where donor recipient size matching may be most challenging.

In conclusion, we believe that TCV for donors and recipients can be accurately estimated from images obtained through routine echocardiographic examinations. This measurement may allow for improved donor-recipient size matching in pediatric heart transplant patients. TCV for the recipient can be easily estimated using standard echocardiographic views and can be compared to a donor TCV based on LVEDV which is measured and reported routinely as part of donor echocardiographic evaluations. These techniques have the potential to improve donor allocation and outcomes by helping transplant centers make more informed decisions about suitability of potential donor organs from a size matching perspective.