Total Cavopulmonary Connection Flow With Functional Left Pulmonary Artery Stenosis
Angioplasty and Fenestration In Vitro
Background— In our multicenter study of the total cavopulmonary connection (TCPC), a cohort of patients with long-segment left pulmonary artery (LPA) stenosis was observed (35%). The clinically recognized detrimental effects of LPA stenosis motivated a computational fluid dynamic simulation study within 3-dimensional patient-specific and idealized TCPC pathways. The goal of this study was to quantify and evaluate the hemodynamic impact of LPA stenosis and to judge interventional strategies aimed at treating it.
Methods and Results— Simulations were conducted at equal vascular lung resistance, modeling both discrete stenosis (DS) and diffuse long-segment hypoplasia with varying degrees of obstruction (0% to 80%). Models having fenestrations of 2 to 6 mm and atrium pressures of 4 to 14 mm Hg were explored. A patient-specific, extracardiac TCPC with 85% DS was studied in its original configuration and after virtual surgery that dilated the LPA to 0% stenosis in the computer medium. Performance indices improved exponentially (R2>0.99) with decreasing obstruction. Diffuse long-segment hypoplasia was ≈50% more severe with regard to lung perfusion and cardiac energy loss than DS. Virtual angioplasty performed on the 3-dimensional Fontan anatomy exhibiting an 85% DS stenosis produced a 61% increase in left lung perfusion and a 50% decrease in cardiac energy dissipation. After 4-mm fenestration, TCPC baffle pressure dropped by ≈10% and left lung perfusion decreased by ≈8% compared with the 80% DS case.
Conclusions— DS <60% and diffuse long-segment hypoplasia <40% could be considered tolerable because both resulted in only a 12% decrease in left lung perfusion. In contrast to angioplasty, a fenestration (right-to-left shunt) reduced TCPC pressure at the cost of decreased left and right lung perfusion. These results suggest that pre-Fontan computational fluid dynamic simulation may be valuable for determining both the hemodynamic significance of LPA stenosis and the potential benefits of intervention.
Received December 20, 2004; revision received July 14, 2005; accepted August 26, 2005.
Palliative surgical treatments1,2 of the single ventricle heart defect result in diverse, complicated, and clinically critical morphologies. Among other factors, early deaths and poor outcomes are associated with high pulmonary vascular resistance and hypoplastic or distorted pulmonary arteries.3 All stage 1 reconstructions and palliative initial shunt operations have the potential to obstruct branch pulmonary arteries, having both immediate effects on pulmonary function and long-term effects in the form of asymmetrical arterial growth and remodeling patterns.4,5 Later stages of the total cavopulmonary connection (TCPC) surgery further complicate blood flow dynamics, affecting cardiac loading, lung perfusion, and hepatic blood distribution.6–8 Stenotic lesions of the pulmonary arteries can contribute to significant downstream hypoplasia of pulmonary arteries, ventilation/perfusion mismatch, impaired postsurgery revascularization, thromboembolism, limited exercise capacity, increased atrial level shunting, atrial/ventricular arrhythmias, and protein-losing enteropathy, all important concerns during and after surgery and for the quality of postoperative life.4,9,10
Surgical patch augmentation of the left pulmonary artery (LPA),11,12 balloon angioplasty, and fenestration13 are common interventions used to overcome elevated Fontan baffle pressures and poor left lung perfusion, which are associated with stenotic pulmonary arteries. In this study, to compare the hemodynamic outcomes associated with these options, LPA augmentation and fenestration were performed virtually in a computer medium on idealized models and on a selected patient-specific extracardiac TCPC.
Fenestration has been demonstrated to yield lower systemic venous pressures and improved ventricle filling, cardiac output, and overall oxygen delivery.14 Although some institutions advocate routine fenestration,9 others argue that it should be used more selectively,15 balancing the potential benefits against the risks and costs of an additional intervention for closure. A mathematical model, as presented here, that quantifies left and right lung perfusion, TCPC baffle pressure, the effects of different fenestration sizes, and atrial pressures would complement previous studies9,14,15 and contribute to clinical understanding and interventional plans.
The importance of geometric features for long-term TCPC efficiency has been demonstrated through a series of numerical and experimental in vitro studies.16–19 Recent anatomic fluid dynamic studies20,21 of the TCPC have tried to identify the relative contribution of different features and the configuration of the central “+”-shaped connection site and the connecting inferior (IVC) and superior vena cava (SVC) and left and right pulmonary arteries (RPA) to the global flow field. Each of these components has characteristic morphological variations and an equal-order-of-magnitude significance to the overall hydrodynamic performance and pressure drop. The influence of different IVC configurations for a given Glenn stage anatomy is studied via computational fluid dynamic (CFD), and the benefits of specific anastomoses are demonstrated.22 This article quantifies fluid dynamics effects of component-level morphological variations of the LPA outflow branch in idealized models and in patient-specific anatomies focused on stenotic LPA morphologies.
Cohort of TCPC With LPA Stenosis
A multicenter Fontan patient MRI database has been assembled to study the anatomic elements of the TCPC. Informed consent was obtained, and all associated studies were approved by the Internal Review boards of the Children’s Hospital of Philadelphia, Children’s Healthcare of Atlanta, University of North Carolina, and Georgia Institute of Technology. On examination of the MRI data and corresponding 3-dimensional (3D) TCPC reconstructions, a cohort (13 of 37 cases, 35%) exhibited LPA stenosis. Further analysis revealed that 10 of 13 patients with LPA stenosis (77%) carried the diagnosis of hypoplastic left heart syndrome in which the aorta is augmented with a patch of the larger main PA. The enlarged neo-aorta is observed to cause superior and anterior compression on the LPA, which is largely immobilized because of the left mainstream bronchus. A 3D anatomic reconstruction of a typical TCPC having discrete LPA stenosis is presented in Figure 1a.
Our database revealed 2 types of stenoses: Discrete stenosis (DS) or diffuse long-segment hypoplasia (DF). In DS (Figure 1a), the lesion is asymmetric about the vessel axis before distally returning to the proximal vessel diameter. A DF is characterized by a gradual taper, as shown in Figure 1b, in which the diameter does not return to its original caliber distal to the neo-aortic reconstruction. DF can also be associated with the mismatch between the native vessel and the proximal TCPC, producing a level of taper that can lessen as the native vessel grows if there is sufficient blood flow.
The TCPC models were analyzed with CFD techniques, providing detailed 3D flow fields, lung perfusions, pressure drops, and hydrodynamic power losses. The analysis requires creation of an idealized model geometry designed within a computer-aided design package or by reconstructing it from anatomic MRI patient data. The next step is grid generation in which the vessel volume is divided into thousands of smaller volumes or computational elements (grids or meshes). At each element, the governing conservation equations of mass and momentum for blood fluid flow are solved. Although there are no assumptions involved in the mathematical framework of blood flow, a credible numerical solution of the governing equations requires diligent boundary condition specification, global and local mesh verification studies, and in vitro experimental validation. Further references are provided in an earlier review.20 Specific technical details of CFD validation and verification studies are provided in the Appendix (found in the online-only Data Supplement at http://circ.ahjournals.org/cgi/content/full/CIRCULATIONAHA. 104.530931/DC1).
Idealized TCPC With DS and DF
In this study, idealized TCPC models with parametric dimensions are used to explore the effects of varying specific parameters such as degree of stenosis in the context of a well-understood idealized model. This process is analogous to altering a single variable (ie, scaled up or down in size) in a multivariable framework to investigate its impact on system performance.
The 1-diameter offset idealized TCPC model without stenosis, previously studied by Ensley et al,18 was reproduced as the reference model (Figure 2a) with a reduction in cross-sectional area (CSA) of 0%. CFD calculations23 and in vitro particle image velocimetry experiments24 on this reference configuration demonstrate low hydrodynamic power loss characteristics resulting from the stable buffer vortex located at the offset region.18 Larger, surgically impractical offset distances have higher loss indices and can result in undesirable hepatic blood distribution. In this model, all venae cavae and pulmonary arteries were held constant at 13.335 mm, a dimension retained by Ensley et al18 based on the chest MRI of an 8-year-old Fontan patient.18
Idealized models with various degrees of stenosis (0% to 80%) were created by introducing a smoothly varying obstruction into the functional LPA of the 1-dimensional offset reference model (Figure 2a). The 2 basic, anatomically observed, eccentric stenosis morphologies with convergent-divergent (DS) and convergent-only (DF) area variation were realized. The stenosis region, located 2.5 diameters distal from the SVC axis, had 1.5 and 2.0 diameters of convergent and divergent segments, respectively.
TCPC with Fenestration
The 1-dimensional offset idealized model having 80% stenosis in CSA (mean Fontan baffle pressure, 16 mm Hg) was modified to include a fenestrated communication between the pulmonary venous (common) atrium and the TCPC baffle (Figure 2b). Simulations were performed for 2- and 4-mm fenestrations, corresponding to the standard coronary bypass punches, and for a 6-mm opening, representing the upper-limit fenestration that may be used with an adjustable snare device.25 Five pulmonary venous atrium pressure values (4, 6.5, 9, 12, and 13.9 mm Hg) were imposed at the outlet of the fenestration to span the relevant range of transpulmonary pressure gradients. The fenestration was located on the IVC 2 vessel diameters proximal from the PA axis.
Patient-Specific Anatomical TCPC
Reconstruction Methods and Geometry
Previous reconstruction of 3D patient-specific morphologies from MRI has enabled detailed and realistic flow analyses.21 The anatomic models are obtained from patient MRI data by enhancing the out-of-plane image resolution with an adaptive control grid interpolation technique26 and then segmenting the vessels of interest. The patient-specific TCPC fluid dynamic analysis methodology, which uses in vitro computational and experimental models, is described elsewhere.20,27
A typical extracardiac TCPC anatomy (Fontan Anatomy/Flow Database ID CHOP013) with a severe DS (85% CSA) was selected from our database28 for analysis (Figure 1a). It featured hydraulic vessel diameters of 7.7, 8.8, 14.0, and 19.1 mm for the LPA, RPA, SVC, and IVC, respectively, and a smooth IVC baffle with uniformly constant diameter. Stenosis, where the hydraulic vessel diameter decreased to 3.0 mm, was located ≈15 mm (distal) from the IVC anastomoses.
The 3D computational models allow modification, which mimics surgical alteration seen in vivo. Thus, the results of virtual surgery can be evaluated computationally. This approach was applied to an 85% LPA stenosis with virtual dilatation to 0%. No other part of the original connection geometry shown in Figure 1a was altered virtually.
Hydrodynamic Power Loss
The cardiac workload consumed in the TCPC to overcome flow resistance can be calculated (Equation A-1 in the Appendix). This power loss index represents both the pressure drop and global TCPC flow quality and has been established as a valid parameter for comparing the overall performance of different TCPC designs and surgical morphologies.16–24 Furthermore, noninvasive clinical acquisition of this parameter in vivo is prospectively possible with cardiac phase-contrast MRI.23
Equal Vascular Lung Resistance Physiological Operating Point
Characteristics of TCPC models were obtained over a range of operating conditions by specifying varied LPA and RPA flow splits through a series of CFD simulations. Coupling these results with lumped parameter models of the pulmonary circulation enables comparison of different TCPC models or surgical alternatives in the same patient with known cardiac output, pulmonary vascular resistances, and IVC/SVC flow splits. Here, a vascular lung resistance of 1.8 Woods units was assumed for both lungs. This specific hemodynamic state is called the equal vascular lung resistance physiological operating point (EVLR) of the given TCPC.21 Uneven vascular lung resistances may also be analyzed by specifying different values for each lung.
Discrete and Diffuse Stenosis
Figure 3 shows TCPC models with various degrees of DS and DF. Characteristic hydrodynamic power loss variations are plotted as a function of LPA flow split. For each model, the EVLR point is indicated. The EVLR operation point was observed to shift toward more balanced pulmonary flow splits as stenosis decreased, resulting in improved lung perfusion. Left lung perfusion increased and power loss decreased with decreasing stenosis severity (R2>0.99). DF, a more severe condition than DS, resulted higher power losses and lower left lung perfusion, as shown in Figure 3. Good agreement between the computational results and in vitro experiments18 for the zero-diameter offset reference model was also demonstrated.
Sample velocity magnitude distributions and flow streamlines along the midcoronal sections of the selected idealized TCPC models are plotted in Figure 3 at the EVLR conditions. For stenosis levels >40% CSA, disturbed flow patterns and recirculation regions were observed in the DS models. Branch TCPC pressures are tabulated with reference to the LPA pressure level in Table 1. The pressure values demonstrated a trend similar to that of the power losses. All pressure values were found to increase significantly with increasing degree of stenosis (0 to 5.2 mm Hg with reference to LPA outlet). TCPC pressures were 43±16% higher in the DF stenosis models compared with the DS models.
Improvements with Fenestration
For the 80% CSA idealized model (DS-4), the 4-mm fenestration decreased the TCPC baffle pressure by 8% to 14%, depending on the pulmonary venous (common) average atrial pressure. As illustrated in Table 2, elevated pulmonary venous atrium pressures diminished the effects of the fenestration as a result of a lower transfenestration gradient. As expected, larger fenestrations lead to improved pressures in the TCPC baffle. On the other hand, the shunt through the 4-mm fenestration worsened lung perfusion by 8% to 17% on the left and 11% to 15% on the right. Values are summarized in Table 2 for other fenestration sizes and pulmonary venous atrium pressures calculated at the EVLR operation points.
In Figure 4, the cardiac power loss characteristics of selected fenestration models are plotted as a function of LPA flow split. Despite the LPA/RPA flow split differences, lower pulmonary venous atrium pressures and larger fenestrations were found to increase the power loss and systemic-to-atrial shunt flow rate.
Patient-Specific Anatomical Models
Pressure distribution along the TCPC pathway and flow streamlines are plotted in Figure 5 for the original extracardiac TCPC anatomy and the modified model without stenosis. Figure 5 shows that streamlines originating from the same locations in both models followed different paths. For the stenosis model, both IVC and SVC flows were preferentially directed to the right lung, whereas equal distribution was attained after stenosis dilatation. Swirling flow stream patterns, a common characteristic of the TCPC, were observed proximal to both branches, preferentially at the LPA.
Hemodynamic Improvements With Virtual Angioplasty
The hydrodynamic power loss characteristics were different for the 2 anatomies, as shown in Figure 6. The model with 85% CSA had very high energy loss, especially with increased LPA flow split. After LPA dilatation, the power loss curve was more symmetric with respect to the left/right flow split, causing the EVLR operating point to be much closer to an equal flow split. The virtual angioplasty operation resulted in notable performance improvements in left lung perfusion and hydrodynamic energy loss as 61% and 50%, respectively. The actual calculated values are labeled in Figure 5.
Virtual LPA angioplasty offers a method to evaluate surgical options based on patient MRI. Morphing a patient anatomy to mimic changes made during surgery allows postoperative anatomy to be realized computationally without in vivo execution. Flow conditions within the anatomy can then be evaluated with the methods described here. The promise of virtual surgery and simulation is further supported by the results of flow studies in which a marked decrease in cardiac power loss over all pulmonary flow splits and increased lung perfusion was observed, as illustrated in Figure 5. In addition to providing a 3D sense of the intended procedure, this quantification of surgical benefits can offer information useful for evaluating different intervention strategies such as angioplasty and/or fenestration.
Large variations in patient anatomies make it difficult to characterize the TCPC as 1 generalized model. Our comparative patient-specific hemodynamic study is supported by an extensive set of simulations in idealized models, ≈500 individual CFD runs, which demonstrated the relative effects of dimensional variations on fluid dynamic performance. For any LPA stenosis or otherwise inefficient TCPC, the pressure in the baffle is seen to increase. The pressure loss in a DS would recover to a certain extent, whereas the pressure loss does not recover in DF, causing higher pressures within the baffle. Furthermore, a DS has the potential to cause significant distal flow disturbance and separation.
Power loss and perfusion results showed similarities for the idealized and anatomic 80% CSA DS models. This resulted from balanced effects of pulmonary branch size and cardiac output in these models (nominal branch PA diameter of the anatomic model is almost half of the idealized model, whereas the total cardiac output is ≈2 times higher). Our experience has shown that anatomic morphology and flow should be used when available for exact quantitative assessments. Idealized models are usually designed to study a single feature of the anatomy such as stenosis, fenestration, caval offset, or PA curvature; in the actual patient-specific morphologies, the results of all geometrical features are observed in a combined way. However, the idealized model studies are essential for understanding, and they provide a comparative “know-how” or “lookup table” when advanced CFD techniques are not readily available.
These results enabled a comparison between the effect of zero-offset and stenosis severity. We have studied a zero-offset model in our earlier work, which showed that a 1-dimensional offset, which is the reference model used here, reduces the hydrodynamic loss by ≈50%. Therefore, in the context of an idealized model at 4L/min cardiac output, flow resistance created as a result of a DS stenosis of 60% CSA is equivalent to having a zero offset in the TCPC. Both models yield approximately the same power losses but do not result in the same lung perfusion because the zero-offset model should yield a 50/50 EVLR and stenosis skews the EVLR toward the nonstenotic branch.
Fenestrations at the IVC simultaneously reduced the LPA and RPA flow. This is similar to operating the TCPC at a lower cardiac output and reduces the power loss in the TCPC as long as the transfenestration pressure gradient is low. Higher gradients cause increased losses because of very high flow through the fenestration. Flow through the fenestration can be unsteady, depending on the pressure variation in the atrium and baffle flow changes over the cardiac cycle. Although flow pulsatility (unsteady effects) was not studied here, simulations with a number of quasi-steady pressure levels as used here can picture these effects grossly. Variations in the fenestration location may warrant further investigation, even though only a slight effect on the hydrodynamics is anticipated as long as the location remains on the IVC branch. Benefits of the reduced TCPC baffle pressure must be judged with regard to the corresponding reduction in lung perfusion. Based on the fluid dynamic factors alone, a cautious use of fenestrations or smaller diameters should be considered for patients with severely stenosed arteries and already poor left lung perfusion indices.
In current practice, the decision to intervene with LPA angioplasty rarely poses a question for DS; however, DF is more challenging. If amenable to intervention, highly stenosed morphologies are generally always addressed regardless of fenestration. The question of whether it is always necessary remains: Does the systematic intervention in these cases always lead to beneficial fluid dynamics and improved outcomes? The tools presented here offer a framework for elucidating this point. Our results showed that when applied individually, fenestration and angioplasty decrease the systemic venous pressure postoperatively by ≈3 mm Hg (≈20%) at moderate cardiac outputs. Angioplasty or stent placement involves risk and cost for the patient. Development of the proposed methodology to a dependable clinical level could prospectively identify cases in which intervention may not be necessary and the corresponding risk and cost are unwarranted.
In some cases, pediatric surgical decisions are based on “availability” or “necessity.” For example, fenestrations are usually created through the use of a standard adult coronary artery punch of fixed diameter. This arbitrary size is used without consideration of fluid dynamics but in practice generally works well. However, the benefits of using varied punch sizes are unexplored. Without quantitative studies, it is unknown whether a 5-mm punch would result in better hemodynamics and clinical outcome than a 2-mm punch. Furthermore, given the current clinical understanding, it is unknown to what degree the left and right lungs will be perfused for a given TCPC anatomy and fenestration size or how large a systemic venous pressure drop will be created. The quantitative methods demonstrated here provide hemodynamic founding for answering these difficult questions. This information should be especially valuable for patients with preoperative instability in whom prolonged bypass time may be detrimental and operating time crucial. Verification of this approach, to quantify and demonstrate dynamic stability, is feasible and likely to be beneficial for the patient.
A large number of cases need to be explored to establish the efficacy of this approach and to identify areas of study and problem. This is a long-term strategy that needs to be developed and explored. Preintervention noninvasive and virtual techniques place within reach the ability to avoid pitfalls and achieve optimal surgical outcomes.
This study quantified the possible improvements in TCPC hemodynamics brought forth by performing a virtual angioplasty operation on a stenotic LPA. An actual anatomic extracardiac TCPC with severe stenosis was selected to illustrate the applicability of the CFD technique. MRI anatomical reconstructions and the preceding MRI data acquisition methodology are integrated to the fluid dynamics analysis, which could enable preinterventional outcome assessment for improved patient care. The anatomic case study is supported with parametric computational simulations in idealized models with 2 common stenosis types. It was found that for the same LPA diameter, DF is a more problematic condition than DS. Results for the anatomic TCPC before and after the virtual angioplasty operation should be referred to when available, and the idealized models should serve as comparative data when the detailed patient specific simulations are not available.
Fenestration and angioplasty were found to be equally effective in reducing high TCPC baffle pressures resulting from increased degrees of LPA stenosis. For all fenestrations studied, however, both the left and right lung perfusions were reduced as a result of the systemic venous-to-common atrial shunt. This correlates well with the clinical observation that for patients with less-than-ideal pulmonary artery anatomies, fenestration lowers Fontan baffle pressure to less morbid levels while resulting in lower systemic arterial oxygen saturation.
This study presents the first in a long line of detailed investigations. Advanced analysis techniques could potentially be useful in the clinical evaluation of complex patient hemodynamics and anatomies. Once more experience is gained with this type of analysis and with advances in hardware and software development in the years to come, these methodologies have the potential to be clinically acceptable and useful in surgical and clinical patient management. Most importantly, assessment of the predictive value of CFD simulation in clinical practice requires prospective testing in Fontan patients before and after intervention.
This work was supported by grant HL67622 from the National Heart, Lung and Blood Institute.
The online-only Data Supplement can be found with this article at http://circ.ahjournals.org/cgi/content/full/CIRCULATIONAHA.104.530931/DC1.
Presented in part at the American Heart Association 2004 Scientific Sessions, New Orleans, La, November 7–10, 2004, and published in abstract form (Circulation. 2004;110[suppl III]:III-737).
McGuirk SP, Winlaw DS, Langley SM, Stumper OP, de Giovanni JV, Wright JG, Brawn WJ, Barron DJ. The impact of ventricular morphology on midterm outcome following completion total cavopulmonary connection. Eur J Cardiothorac Surg. 2003; 24: 37–46.
Goff DA, Blume ED, Gauvreau K, Mayer JE, Lock JE, Jenkins KJ. Clinical outcome of fenestrated Fontan patients after closure: the first 10 years. Circulation. 2000; 102: 2094–2099.
Agnoletti G, Boudjemline Y, Bonnet D, Sidi D, Vouhe P. Surgical reconstruction of occluded pulmonary arteries in patients with congenital heart defect: effects on pulmonary artery growth. Circulation. 2004; 109: 2314–2318.
Hjortdal VE, Emmertsen K, Stenbøg E, Frund T, Schmidt MR, Kromann O, Sorensen K, Pedersen EM. Effects of exercise and respiration on blood flow in total cavopulmonary connection. Circulation. 2003; 108: 1227–1231.
Cochrane AD, Brizard CP, Penny DJ, Johansson S, Comas JV, Malm T, Karl TR. Management of the univentricular connection: are we improving? Eur J Cardiothorac Surg. 1997; 12: 107–115.
Srivastava D, Preminger T, Lock JE, Mandell V, Keane JF, Mayer JE Jr, Kozakewich H, Spevak PJ. Hepatic venous blood and the development of pulmonary arteriovenous malformations in congenital heart disease. Circulation. 1995; 92: 1217–1222.
Sharma S, Goudy S, Walker P, Panchal S, Ensley A, Kanter K, Tam V, Fyfe D, Yoganathan A. In vitro flow experiments for determination of optimal geometry of total cavopulmonary connection for surgical repair of children with functional single ventricle. J. Am. Coll. Cardiol. 1996; 27: 1264–1269.
Ensley AE, Ramuzat A, Healy TM, Chatzimavroudis GP, Lucas C, Sharma S, Pettigrew R, Yoganathan AP. Fluid mechanic assessment of the total cavopulmonary connection using magnetic resonance phase velocity mapping and digital particle image velocimetry. Ann Biomed Eng. 2000; 28: 1172–1183.
Haas GS, Hess H, Black M, Onnasch J, Mohr FW, van Son JA. Extracardiac conduit Fontan procedure: early and intermediate results. Eur J Cardiothorac Surg. 2000; 17: 648–654.
The Fontan project. Available at: http://fontan.bme.gatech.edu. Accessed on September 15, 2005.