If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for reprints: Ajit P. Yoganathan, PhD, Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, 387 Technology Cir, NW, Suite 200, Atlanta, GA 30313.
The use of Y-grafts for Fontan completion is hypothesized to offer more balanced hepatic flow distribution (HFD) and decreased energy losses. The purpose of this study was to evaluate the hemodynamic performance of Y-grafts over time using serial cardiac magnetic resonance data and to compare their performance with extracardiac Fontan connections.
Ten Fontan patients with commercially available Y-graft connections and serial postoperative cardiac magnetic resonance data were included in this study. Patient-specific computational fluid dynamics simulations were used to estimate HFD and energy losses. Y-graft performance was compared with 3 extracardiac conduit Fontan groups (n = 10 for each) whose follow-up times straddle the Y-graft time points.
Y-graft HFD became significantly more balanced over time (deviation from 50% decreased from 18% ± 14% to 8% ± 8%; P = .015). Total cavopulmonary connection resistance did not significantly change. Y-grafts at 3-year follow-up showed more balanced HFD than the extracardiac conduit groups at both the earlier and later follow-up times. Total cavopulmonary connection resistance was not significantly different between any Y-graft or extracardiac conduit group.
Y-grafts showed significantly more balanced HFD over a 3-year follow-up without an increase in total cavopulmonary connection resistance, and therefore may be a valuable option for Fontan completion. Additional follow-up data at longer follow-up times are still needed to thoroughly characterize the potential advantages of Y-graft use.
Y-graft hepatic flow distribution became significantly more balanced over an approximately 3-year follow-up with comparable TCPC resistance to extracardiac conduit options. Therefore, Y-grafts may be a valuable option for Fontan completion with inherent advantages in hepatic flow distribution.
In this procedure, the inferior vena cava (IVC) is anastomosed to the pulmonary arteries (PAs), typically via a conduit/graft, bypassing the subpulmonary ventricle. This reconfiguration separates oxygenated and deoxygenated blood, but creates a 1-pump circulatory system that requires passive blood flow through the pulmonary circulation.
In this physiology, the single ventricle must provide all cardiac power for the entire circulatory system in the presence of an added resistance due to the total cavopulmonary connection (TCPC).
Efforts to minimize energy loss in the TCPC are hypothesized to result in improved patient outcomes and may benefit the inherently overburdened single ventricle. Additionally, due to the direct connection of the IVC to the PAs, hepatic flow has a limited opportunity to mix with upper body flow, which can result in unbalanced hepatic flow distribution (HFD) to the left and right lungs, potentially leading to the formation of pulmonary arteriovenous malformations (PAVMs).
There is currently a limited understanding of the composition and production of this hepatic factor as well as the exact amount needed to preserve vascular structure. However, a balanced HFD distribution is an important factor to prevent PAVM progression. Both energy loss and HFD distribution have been the focus of numerous previous studies.
The Fontan procedure has evolved over time, progressing through the atriopulmonary, intra-atrial, and extracardiac options, with improvements in outcomes. More recently, the bifurcated Y-graft option has been selectively implemented and is proposed to alleviate the above-mentioned concerns by offering lower energy losses by avoiding caval flow collisions and encouraging more balanced HFD by directing hepatic flow through 2 Y-graft branches toward the left pulmonary artery (LPA) and right pulmonary artery (RPA).
However, these benefits have not yet been fully realized. In our previous study comparing the hemodynamic performance of commercially available Y-grafts with the more traditional lateral tunnel and extracardiac conduit (ECC) options, no definite advantages were seen in patients with Y-grafts.
Overall, HFD was not statistically different between graft types and TCPC resistance was higher in the Y-graft cohort. Furthermore, extremely unbalanced HFD was seen more often in patients with Y-grafts. A key limitation noted in that previous study was the difference in follow-up times between the Y-graft (<1 month) and lateral tunnel/ECC (∼9 months) patients. In a study by Yang and colleagues,
These authors illustrate that in the immediate postoperative setting the theoretical advantages of the Y-graft have not always been seen, but the anecdotal experience in our institution has been favorable. No larger series exist evaluating the midterm outcomes of patients receiving Y-graft Fontans to help advise future practice.
Motivated by these previous observations regarding this new group of Fontan patients, we sought to evaluate how the hemodynamic performance of Y-grafts changes over time. In this study, serial cardiac magnetic resonance (CMR) data and computational fluid dynamics (CFD) were used to investigate these changes. Finally, Y-graft performance was compared with ECC patients.
Ten Fontan patients with Y-graft connections were included in this study. Inclusion criteria included single ventricle patient with Y-graft Fontan connection; serial, postoperative CMR and phase-contrast CMR data; and sufficient image quality to accurately segment the TCPC and blood flow waveforms at each TCPC inlet/outlet throughout the cardiac cycle. All Y-graft patient data were received from Children's Healthcare of Atlanta (institutional review board No. H09279) and are numbered chronologically in order of surgery date. All Y-graft connections were implanted using commercially available aortoiliac polytetrafluoroethylene bifurcated grafts (Gore, Flagstaff, Ariz) without further modification. An ECC comparison group with identical follow-up times was not available. Therefore, to put the Y-graft results in context with ECC patients, several unique ECC groups were chosen (each n = 10) with follow-up times both less than and greater than the Y-graft cohort. Each ECC group included consecutive patients within a predefined follow-up window as specified in Table 1. All ECC patient data were received from Children's Hospital of Philadelphia (institutional review board No. H05236). A sensitivity analysis was performed to ensure these small comparison groups were representative of larger Fontan cohorts. Clinical data, including age, body surface area, gender, ventricle morphology, presence of fenestration and dates of surgery, and CMR studies were obtained for each patient.
Table 1Clinical and hemodynamic data for the extracardiac conduit (ECC) comparison groups
ECC group 1 (follow-up <3 y) (n = 10)
ECC group 2 (follow-up 3-6 y) (n = 10)
ECC group 3 (follow-up >10 y) (n = 10)
3.30 ± 0.95
6.90 ± 1.0
16.3 ± 3.33
0.63 ± 0.06
0.86 ± 0.18
1.48 ± 0.28
Follow-up time (y)
0.67 ± 0.55
5.0 ± 0.63
13.5 ± 1.78
Cardiac index (L/min/m2)
4.46 ± 1.16
3.66 ± 0.86
3.43 ± 0.58
IVC flow (L/min/m2)
1.08 ± 0.32
1.79 ± 0.79
1.72 ± 0.54
SVC flow (L/min/m2)
1.84 ± 0.81
1.43 ± 0.63
0.75 ± 0.26
LPA flow (L/min/m2)
1.27 ± 0.64
1.09 ± 0.42
1.07 ± 0.35
RPA flow (L/min/m2)
1.21 ± 0.51
1.78 ± 0.89
1.33 ± 0.43
3.06 ± 0.86
2.90 ± 0.67
2.46 ± 0.50
Collateral flow (L/min/m2)
1.94 ± 1.40
0.82 ± 0.98
0.96 ± 0.76
PFD deviation (percentage points)
13 ± 9
11 ± 9
9 ± 7
LPA stenosis (%)
35 ± 24
49 ± 19
32 ± 20
RPA stenosis (%)
Overall PA stenosis (%)
27 ± 17
40 ± 13
26 ± 15
TCPC resistance (WU)
0.29 ± 0.37
0.12 ± 0.07
0.21 ± 0.15
HFD deviation (percentage points)
24 ± 15
21 ± 18
12 ± 5
Values are presented as mean ± standard deviation for normally distributed variables, median (interquartile range) for nonparametric data, or as number. ECC, Extracardiac conduit; BSA, body surface area; HLHS, hypoplastic left heart syndrome; IVC, inferior vena cava SVC, superior vena cava; LPA, left pulmonary artery RPA, right pulmonary artery; Qs, systemic return; PFD, pulmonary flow distribution; PA, pulmonary artery; TCPC, total cavopulmonary connection; WU, Wood unit; HFD, hepatic flow distribution.
All CMRs were performed with a Siemens 1.5 T magnetic resonance imaging system (Siemens Medical Solutions, Malvern, Pa). Patients were scanned supine, head first in the scanner, with echocardiogram leads placed. After localizers were obtained, a stack of contiguous, static, diastolic steady state free precession images were obtained from the diaphragm to thoracic inlet to assess anatomy. Slice thickness was generally 1 to 3 mm and in plane resolution was 1 × 1 mm.
Retrospectively gated, through plane phase contrast magnetic resonance was used to assess flows in the vena cavae, branch pulmonary arteries, and across the aortic valve. Inferior vena cava flow was measured suprahepatic. Velocity encoding was generally 150 cm/s for the aorta and 60 cm/s for the other vessels. Slice thickness was generally 4 mm with in plane resolution of 1 × 1 mm. The number of phases was a function of the heart rate and ranged from 20 to 30.
Anatomic Reconstruction and Blood Flow Segmentation
Patient-specific anatomies were reconstructed from axial CMR images using methods previously developed.
Geomagic Studio (Geomagic Inc, Research Triangle Park, NC) was used to fit a surface around the reconstructed point cloud and export the surface for mesh generation. Patient-specific blood flow waveforms were segmented from phase contrast magnetic resonance images for all vessels of interest using previously validated methods.
The 3-dimensional anatomy data were imported into ANSYS workbench (Ansys Inc, Canonberg, Pa), where vessel extensions of length 10 × (vessel diameter, mm) were added to overcome entrance effects and establish an appropriate velocity profile. A polyhedral mesh of approximately Davg/30 mm elements was used to achieve mesh independent results, where Davg is the average vessel diameter.
All simulations were performed using ANSYS Fluent (Release 17.0), which is a finite volume pressure based Navier-Stokes solver. Blood was modeled as a single-phase Newtonian fluid (μ = 0.04 g/(cm·s), ρ = 1.06 g/cm3) and rigid walls were assumed. The appropriate patient-specific blood flow waveforms extracted from phase contrast magnetic resonance imaging were used as boundary conditions for each TCPC inlet and outlet. Twenty cardiac cycles were simulated to overcome transition effects and achieve period stability, using the final cycle for data analysis.
HFD and TCPC resistance were assessed using CFD. HFD was quantified by seeding massless particles at the IVC and calculating the total flux of particles leaving the LPA and right PA (RPA). HFD is defined as where θ is the total flux of particles throughout a cardiac cycle. HFD is discussed in terms of HFD deviation from 50%, and an even (50/50) split of hepatic flow to the LPA and RPA is assumed as ideal. This convention gives a 0 HFD deviation for balanced HFD cases, and an HFD deviation of 50 for extreme cases (all hepatic flow to either the LPA and RPA). Pulmonary flow distribution (PFD) is calculated as where Q is flow rate. Collateral flow was calculated as the difference between aortic and vena cava flow. TCPC resistance is calculated as where is the pressure drop across the TCPC (calculated via CFD) and Qs is the systemic venous flow. Flow rates are indexed to body surface area.
Percent stenosis was calculated for the PAs using the following equation:
where Amin, LPA and Amin, RPA are the minimum cross-sectional areas of the LPA and RPA, and Aavg, LPA and Aavg, RPA are the average cross-sectional areas of the LPA and RPA. A combined outlet stenosis value was used in place of unilateral PA stenosis measurements to give a more accurate representation of the total outlet obstruction to flow and allow a better comparison between patients. Additionally, IVCs and superior vena cavas showed negligible percent stenosis and therefore were not included in this calculation. All vessel areas for this metric were calculated using Vascular Modeling Toolkit version 1.0.1 (Orobix, Bergamo, Italy).
SPSS version 25, (IBM-SPSS Inc, Armonk, NY) was used for statistical analyses. The Shapiro-Wilk test was used to determine normality for each parameter. Pearsos and Spearman correlations were used to investigate bivariate correlations between TCPC resistance and HFD with clinical, hemodynamic, and anatomical parameters. Paired sample t tests or Wilcoxon signed-rank tests were used to test for significant changes over time in the Y-graft cohort for parametric and nonparametric data, respectively. Independent sample t tests and Wilcoxon sum rank tests were used to compare means between the Y-graft and ECC comparison groups for parametric and nonparametric data, respectively. Statistical significance was determined at the P = .05 level. Values are presented as average ± standard deviation for normally distributed variables and median (interquartile range [IQR]) for nonparametric data.
Clinical and hemodynamic data for all Y-grafts at both time points are shown in Table 2. Age increased from 3.26 ± 1.02 years to 5.65 ± 1.25 years across the 2 time points, with an average follow-up time of 14 days (IQR, 7-25 days) and 3.02 years (IQR, 2.13-3.31 years), respectively. The cohort included 4 men and the most common underlying anatomy was hypoplastic left heart syndrome, present in 3 patients. Clinically, these Y-graft patients are doing well, with no Fontan complications (eg, protein losing enteropathy or plastic bronchitis) and oxygen saturations in the mid to upper 90s. These patients are currently aged 7 to 10 years and none are being evaluated for transplantation.
Table 2Clinical and hemodynamic data for the Y-graft cohort
Values are presented as mean ± standard deviation for normally distributed variables, median (interquartile range) for nonparametric data, or as number. BSA, Body surface area; HLHS, hypoplastic left heart syndrome; IVC, inferior vena cava; SVC, superior vena cava; LPA, left pulmonary artery; RPA, right pulmonary artery; Qs, systemic return; PFD, pulmonary flow distribution; PA, pulmonary artery; TCPC, total cavopulmonary connection; WU, Wood units; HFD, hepatic flow distribution.
Indexed IVC flow increased from 0.57 L/min/m2 (IQR, 0.45-0.89 L/min/m2) to 1.03 ± 0.25 L/min/m2 (P = .005). Similarly, Qs increased from 1.80 ± 0.34 L/min/m2 to 2.40 ± 0.53 L/min/m2 (P = .020). Superior vena cava, LPA, RPA, collateral flow, and cardiac index showed no significant changes between time points. Pulmonary flow distribution became more balanced (PFD deviation from 50% decreased from 13% ± 11% to 9% ± 8%) although not statistically significant (P = .173). Finally, PA stenosis did not significantly change over time.
HFD significantly improved over time (paired sample t test, P = .015) (Table 2). HFD deviation decreased from 18% ± 14% to 8% ± 8%. Nine out of 10 patients had more balanced HFD at time 2 than time 1 (Figure 1). On average, there was a 10 ± 10 point improvement in HFD deviation. Individual HFD values and follow-up times are shown in Figure 1, A. Streamlines for all Y-grafts at both timepoints are shown in Figure 2. Overall, HFD was relatively balanced at time 2 with an HFD deviation of only 8% ± 8%.
TCPC resistance decreased from 0.30 ± 0.21 to 0.17 ± 0.07 wood units (WU), but was not statistically significant (paired sample t test, P = .117) (Table 2). TCPC resistance decreased for 8 out of 10 patients (Figure 3). Individual resistance values can be found in Figure 3, A.
Comparison With ECC Control Groups
Three ECC comparison groups were chosen using consecutive patients with <3-year follow-up, 3- to 6-year follow-up, and >10-year follow up (n = 10 for each group). These 3 groups were chosen to straddle the Y-graft follow-up times because a comparison group with identical, serial follow-up times was not available. The clinical and hemodynamic data for all 3 ECC comparison groups are shown in Table 1. Follow-up times were used as a selection criterion, and was therefore different for the 3 comparison groups: 0.67 ± 0.55 years, 5.0 ± 0.63 years, and 13.5 ± 1.78 years, respectively. All ECC patients in the 2 younger comparison groups are free from Fontan complications. Two patients in the oldest comparison group were reported to have mild protein losing enteropathy. None of these patients are being evaluated for transplantation.
Both Y-grafts and ECC patients showed more balanced HFD as patients aged (Figure 4). Y-grafts improved quicker than ECC. The Y-grafts at time 2 (36-month follow-up) were significantly more balanced than the ECC group with an earlier follow-up time (P = .009), and more balanced, although not statistically significant, than the 2 ECC groups with later follow-up times (P = .059 and P = .209, respectively) (Figure 4, A). TCPC resistance was not significantly different between any 2 groups, although slightly higher in the Y-graft patients at time 1 (Figure 5). Large variations in TCPC resistance were seen at the earliest timepoints in both Y-grafts and ECC patients, and these variations decreased in the later timepoints.
The concept of using bifurcated Y-grafts for Fontan completion was initially met with interest as an intuitive solution for achieving balanced HFD and avoiding caval flow collisions. However, early evaluations failed to confirm these potential advantages.
The operation is inherently more complex than more traditional Fontan completions, in that it involves 3 anastomotic sites rather than 2 and has caliber changes throughout the bifurcated baffle as flow moves from the main trunk into the Y-arms. Additionally, the geometry of a Y-shaped baffle lends itself to more potential for twisting and kinking than does a straight tube baffle. These limitations notwithstanding, we have not found the Y-graft Fontan to be inferior to the traditional external or later tunnel connections, simply that it was not superior in the metrics evaluated on early follow-up studies. We therefore believe its potential for streamlined flow and improved efficiency merits ongoing consideration and study (Video 1). Figure 6 highlights 1 representative patient and summarizes these findings. Although the patient numbers in this study are challenging from a statistical standpoint, it does represent by far the largest such cohort of Y-graft Fontans with midterm follow-up data.
Y-graft patients showed more balanced HFD as follow-up time increased. (Figure 4). The Y-grafts (at 36-month follow up) had a lower HFD deviation than any of the ECC groups.
In general, more balanced HFD over time (for either graft type) may be explained in part by two primary factors: more balanced overall PFD and inferior systemic venous return constituting a larger percentage of Qs as patients age. Though HFD is multifactorial, one key component is pulmonary flow distribution. In most cases, as PFD becomes more balanced, HFD becomes more balanced. In this study, both patient cohorts showed more balanced PFD as age increased (Tables 1and 2). Secondly, it is well known that the IVC receives a larger portion of Qs as patients age.
As this percentage increases, graft positioning, flow interactions, and the location of stenosis may have a diminished effect.
In our Y-graft cohort, the early time point was obtained within the first month of the operation in all cases. Although these data are valuable and represents the patients' postoperative hemodynamic profiles, there are transient factors, such as fluid shifts, pleural effusions, and tissue edema, that were likely still resolving in many of these patients at the time of CMR. The effect of these postoperative changes on their CMR and CFD results are not entirely clear, but it is possible that some of the maldistribution of hepatic factor in these patients was only transient, and once their systems had reached a new equilibrium a more balanced state existed. For the reasons above, as children age and IVC flow increases, it is expected that HFD will become more balanced regardless of graft type. Both of these factors would be expected to occur at similar rates for the Y-graft and ECC groups, so the fact that Y-grafts exhibited an accelerated improvement in HFD when compared with the ECC group (Figure 4, B) suggests that using a Y-graft may provide an advantage over ECC connections in terms of achieving a balanced HFD. Whether the early time point underrepresented how well balanced the HFD was in Y-graft patients or whether they had more substantial normalization over time, the end result appears to be a more favorable hemodynamic state for these patients.
Even with the built-in 50% stenosis inherent to the design of commercially available Y-grafts (the trunk diameter is double that of the individual arms, resulting in a 50% loss of cross sectional area as the blood moves through the graft), TCPC resistance was not significantly different between the Y-graft and ECC groups. Overall PA stenosis was higher in the ECC groups, but not to the extent of equaling the total (Y-graft + PA) stenosis seen in Y-graft patients. Therefore, avoiding caval flow collisions through Y-graft use likely plays a role in decreasing energy losses and lowering TCPC resistance. This finding encourages the use of a customized Y-graft design that would preserve cross-sectional area throughout the bifurcation.
By combining a stenosis-free graft design with the ability to avoid caval flow collisions, it is reasonable to think that Y-grafts could offer lower TCPC resistance than the standard ECC connections. It remains unknown how TCPC resistance will differ between Y-grafts and ECC connections at follow-up times beyond 36 months (our longest Y-graft follow-up time). With increasing cardiac output as patients age, avoiding caval flow collisions may become a progressively more important factor. However, investigating this hypothesis requires longer term follow-up data.
Future Y-Graft Use
Based on the current data, Y-graft Fontan connections are a viable surgical option. After the immediate post-operative state, Y-grafts offered more balanced HFD than ECC connections without sacrificing energy efficiency. Though longer follow up data on this unique cohort will be instructive, future Y-graft use may be a proactive step to balance HFD and reduce PAVM development. Additionally, as suggested in previous studies, the use of an area preserving Y-graft design may offer further hemodynamic benefits.
As with studies of patients with Fontan circulation, a small sample size limits the power of statistical analysis. However, even with our limited patient number a significant improvement in HFD was observed. Additionally, the lack of an ECC comparison group with identical follow-up times required the use of several unique comparison groups. Although this does not allow for a direct comparison with serial ECC data, choosing follow-up times that straddle the Y-grafts can provide meaningful comparisons. Furthermore, although a 3-year follow-up time for this novel group of Y-graft patients is the longest to date, these new data provide limited understanding of potential long-term consequences. In addition, a 50/50 flow split to the left and right lung will not be ideal for every individual patient. This ratio was used as a consistent approximation for all patients in this study. Finally, it is important to note that all Y-graft patients in this study used commercially available Y-grafts, and results may differ with more customized designs that preserve cross-sectional area.
Y-grafts showed significantly more balanced HFD as follow-up time increased. Previous Y-graft concerns, including increasing resistance without improving HFD, may have been driven by altered hemodynamic parameters in the immediate postoperative state that did not allowing for physiologic adaptation. Additional follow-up data at increased follow-up times are needed to thoroughly characterize the potential advantages of Y-graft use.
Conflict of Interest Statement
Authors have nothing to disclose with regard to commercial support.
Supported in part by American Heart Association Predoctoral Fellowship 17PRE33630117. ANSYS software was provided through an academic partnership between ANSYS Inc and the Cardiovascular Fluid Mechanics Lab at the Georgia Institute of Technology.
Long-term outcomes after the Fontan operation for single-ventricle lesions remain suboptimal, with many patients having complications related to resistance through the Fontan circuit or maldistribution of hepatic blood flow between the lungs. These mechanisms of Fontan failure have persisted through various iterations of Fontan anatomy that have been developed during the almost 5 decades since the first successful atriopulmonary connection in 1971. A recent modification to the Fontan operation is the use of a bifurcated Y-graft to complete the inferior cavopulmonary connection.
The Fontan procedure represents the last stage of surgical palliation for functional single-ventricle defects. As a staged operation in the current era, the Fontan procedure creates a pathway that directs inferior vena cava (IVC) blood flow into the lungs via the branch pulmonary arteries. Over time, a number of modifications of the Fontan procedure have been made, including an atriopulmonary connection, intra-atrial lateral tunnel, extracardiac conduit, and extracardiac Y-graft.