Caval Contribution to Flow in the Branch Pulmonary Arteries of Fontan Patients With a Novel Application of Magnetic Resonance Presaturation Pulse
Background—A complete understanding of fluid mechanics in Fontan physiology includes knowledge of the caval contributions to right (RPA) and left (LPA) pulmonary arterial blood flow, total systemic venous return, and relative blood flow to each lung.
Methods and Results—Ten Fontan patients underwent cine MRI. Three cine scans of the pulmonary arteries were performed: (1) no presaturation pulse, (2) a presaturation pulse labeling inferior vena cava (IVC) blood (signal void), and (3) a presaturation pulse labeling superior vena cava (SVC) blood. The relative signal decrease is proportional to the amount of blood originating from the labeled vena cava. This method was validated in a phantom. Whereas 60±6% of SVC blood flowed into the RPA, 67±12% of IVC blood flowed toward the LPA. Of the blood in the LPA and RPA, 48±14% and 31±17%, respectively, came from the IVC. IVC blood contributed 40±16% to total systemic venous return. The distributions of blood to each lung were nearly equal (RPA/LPA blood=0.94±11).
Conclusions—In Fontan patients with total cavopulmonary connection, SVC blood is directed toward the RPA and IVC blood is directed toward the LPA. Although the right lung volume is larger than the left, an equal amount of blood flow went to both lungs. LPA blood is composed of equal amounts of IVC and SVC blood because IVC contribution to total systemic venous return is smaller than that of the SVC. This technique and these findings can help to evaluate design changes of the systemic venous pathway to improve Fontan hemodynamics.
The fluid mechanics of pulmonary blood flow in Fontan physiology1 has been studied with increasing sophistication.1 2 3 4 5 6 The geometry of the systemic venous pathway has been implicated in various flow phenomena2 3 4 5 that may lead to energy losses.4 Much of this literature is predicated on the knowledge of the relative caval contributions to pulmonary flow and systemic venous return and the amount of blood flow to each lung. Furthermore, because a recent study suggests that caval contribution to total systemic venous return changes with age,7 it is likely that the various components of pulmonary blood flow also change, which has implications for systemic oxygen transport and the geometry of the systemic venous pathway. Until recently, quantification of these data in vivo has been nearly impossible.
Cine MRI, in which blood is signal-intense, can quantify this information in vivo.8 A presaturation pulse placed on blood entering the image plane produces a signal void.8 We hypothesized that when a presaturation pulse is placed on either vena cava of Fontan patients, the decrease in signal intensity in the pulmonary arteries is proportional to its contribution to each branch pulmonary artery. We used this concept to quantify, in a flow phantom in vitro and in Fontan patients in vivo, the contributions of both inferior (IVC) and superior (SVC) vena cava blood to right (RPA) and left (LPA) pulmonary artery flow, the contribution of each vena cava to total systemic venous return, and the relative amount of blood flow to both lungs.
A flow phantom was constructed to simulate the 4-way junction of the systemic venous pathway by connecting 4 pieces of flexible Tygon plastic tubing by use of a plastic 4-way “t”-connector (Figure 1⇓). Fluid inflow was directed into 2 opposing arms from 2 elevated fluid reservoirs (designated “superior” and “inferior”). Two other arms (designated “right” and “left”) carried outflow to 2 separate graduated cylinders placed below the 4-way “t”-connector. The “t”-connector had inputs and outputs perpendicular to each other. The tube inner diameter was 0.8 cm, the approximate diameter of the systemic venous pathway in Fontan patients. Clamps placed on each arm regulated each arm flow independently.
Flow rates were adjusted so that the mean velocity would be 30 to 40 cm/s, the approximate velocity in the systemic venous pathway of Fontan patients.6 One input reservoir contained water and the other contained 0.9% sodium chloride as a marker. Both reservoirs contained aqueous cupric sulfate to approximate the T1 relaxation properties of blood.
Measurement of the chloride concentration in both input and output reservoirs and measurement of flow rate in each output allowed calculation of the percentage of flow to each output reservoir. Chloride concentration was measured by a Ciba chloride analyzer.
The flow rate in the output limbs was measured with the graduated cylinder and stopwatch by determining the volume of fluid entering the graduated cylinder in 1 minute.
We prospectively studied 10 patients with functional single ventricles who underwent Fontan reconstruction at The Children’s Hospital of Philadelphia between June 1, 1995, and March 31, 1996. Patients were 1.8±0.3 years old and were studied 1.6±0.4 weeks after Fontan completion. Nine patients had hypoplastic left heart syndrome, and 1 had critical subaortic stenosis with multiple ventricular septal defects and aortic hypoplasia. All patients underwent a lateral wall tunnel total cavopulmonary connection with side-to-side anastomosis of the right SVC to the RPA with patch augmentation of that junction. Patients with insignificant flow across the fenestrations in their baffle were chosen, with average aortic saturation of 91.4±2.9%. For them to enter the study, they could have no discrete pulmonary artery stenosis (as delineated by MRI); the diameters of the LPA and RPA were within 20% of each other and >50% of that of the descending aortic diameter. Patients with bilateral SVC–to–pulmonary artery connections were excluded. Patients were stable enough to undergo a 1-hour MRI scan under sedation. Informed consent was obtained from all participants. No patient had arrhythmias that precluded imaging in the scanner.
Magnetic Resonance Imaging
All patients were sedated and tolerated this without incident.
1. Parameters we use to obtain spin-echo images of the heart have been described previously.8a We initially obtained spin-echo transverse images of the heart (Figure 2A⇑, top left) to identify the site of the branch pulmonary artery connection to the SVC and the lateral wall tunnel baffle.
2. With the transverse images used to precisely localize the center of the systemic venous pathway (by multiplanar reconstruction), sagittal spin-echo images were acquired spanning the region of the heart (Figure 2A⇑, right, and 2B, left) to localize the long axis of the systemic venous pathway. This was used to evaluate the superoinferior dimension of the branch pulmonary arteries for subsequent cine imaging and to determine placement of the presaturation band.
3. Gradient echo imaging (cine): Prospective ECG-gated, gradient-echo axial images were obtained at the level of the branch pulmonary artery anastomosis to the SVC (Figure 2A⇑, bottom left, and Figure 2B⇑, right. The position of the plane of the cine series is shown in Figure 2B⇑, left). Eight to 10 images were obtained, spanning the entire cardiac cycle. The saturation pulse takes 1.0 ms to apply and is applied 4.7 ms before the center of the excitation pulse used in the acquisition sequence. The TR ranged from 40 to 60 ms (depending on the R-R interval), the TE was 7.3 ms, field of view was 180 to 280 mm, matrix was 128×256 pixels (which was interpolated to 256×256), and number of excitations was 3. Slice thickness was determined by the superoinferior width of the branch pulmonary artery and ranged from 6 to 10 mm.
Three sets of gradient-echo images of the branch pulmonary arteries at the same level were obtained (Figure 2B⇑, right): (1) no presaturation pulse (Figure 2B⇑, top); (2) a presaturation band placed inferior to the image plane (Figure 2B⇑, middle) to destroy spins from IVC blood and produce a signal void. The band thickness extended from the IVC–right atrial junction to just below the imaging plane and ranged from 3 to 8 cm (Figure 2B⇑, middle left); (3) and a presaturation band placed superior to the image plane (Figure 2B⇑, bottom) to destroy the spins from SVC blood. The band thickness was equal to the previous presaturation band.
The leading edge of the saturation band was ≈3 mm from the imaging plane.
The entire study was obtained in 1 hour. Respiratory gating was not performed, because respiration plays a minor role.6
Images were downloaded onto a Sun SPARC station 10 (Sun Microsystems), and VIDA (volumetric image and display analysis)9 was used to manipulate the images.
Calculation of the contribution of fluid from the sodium chloride input reservoir to each output limb of the phantom used the formula (left limb is used as an example) where [Cl]L and [Cl]R are the chloride concentrations in the left and right output reservoirs and QL and QR are the flow rates to the left and right output reservoirs, respectively.
The mean blood signal intensity in each pulmonary artery just distal to the anastomosis site was measured in a rectangle 3 mm in length and the width of the branch pulmonary artery (Figure 2A⇑, bottom left). This was performed with the same rectangular shape and position used for all 3 cine image sets. The width of the branch pulmonary artery was also measured at that point. The data for each individual patient represent the integrated flow over the entire cardiac cycle.
The derivation of flow calculations to follow and exact formulas are given in the Appendix.
The IVC contribution to branch pulmonary artery flow is a function of both the decrease of total signal intensity in the branch pulmonary artery from the first to the second cine set (saturation of IVC blood producing a signal void) and the cube of the radius of the branch pulmonary artery. The SVC contribution to flow in each branch pulmonary artery is calculated similarly.
The percentage of blood in the LPA and RPA originating from the IVC is a function of the decrease in total signal intensity from the first to the second cine set (saturation of IVC blood producing a signal void) and the first to the third cine set (saturation of SVC blood producing a signal void) in the individual pulmonary artery. The percentages of blood in the LPA and RPA originating from the SVC were similar to that from the IVC, except that the numerators were different.
The amount of caval contribution to the total systemic venous return is a function of the sum of the decrease in total signal intensity from the first to the second cine set and the first to the third cine set in both branch pulmonary arteries.
The distribution of blood flow to each lung was the weighted average of percent caval contribution to the systemic venous return and the caval contribution to each pulmonary artery.
Comparisons between 2 means were made by the paired 2-way Student’s t test and the Wilcoxon rank sum test. All measurements are mean±SD. Intraobserver variability was determined by replicate measures using the coefficient of variation. A single trained observer performed all image analysis steps. Significance was taken at P<0.05. This analysis was performed on a PC using Excel 5.0 and JMP version 3.1.4 (SAS Institute Inc).
Validation in a Phantom
The left output limb mean velocity was 32 cm/s, and that of the right output limb was 40 cm/s. Sodium chloride measurement of percent contribution of the sodium chloride input limb to flow in each output limb was within 10% of MRI measurements (Figure 1⇑ and Table⇓).
Caval Contribution to Pulmonary Arterial Flow
Figure 2B⇑ displays the raw image data. Note how similar and uniform the signal is in both RPA and LPA in the top right panel without a presaturation pulse. Compare this with the decreased signal intensity in the LPA in the middle panel when the IVC blood is saturated. Also compare this with the decreased signal intensity in the RPA in the bottom panel when the SVC blood is saturated.
The first and second sets of bars in Figure 3⇓ display the percent IVC and SVC contributions, respectively, to LPA and RPA flow. The IVC contributed significantly more blood to the LPA than the RPA, with an LPA/RPA flow ratio of 2.5±1.5 (Figure 2B⇑, middle right). Significantly more blood in the SVC, however, was directed toward the RPA than the LPA, with an LPA/RPA flow ratio of 0.68±0.19 (Figure 2B⇑, bottom right). The LPA/RPA flow ratio for the IVC and SVC differed significantly from each other (P=0.009).
The third and fourth sets of bars in Figure 3⇑ display the amount of blood in the LPA and RPA originating from the IVC and SVC. Although the LPA blood was split nearly equally from flow originating in the IVC and SVC, RPA blood had a predominance of flow originating in the SVC.
Caval Contribution to Systemic Venous Return and Relative Pulmonary Blood Flow
Figure 2B⇑ again shows the raw image data. Note the generalized decrease in signal intensity in the pulmonary arteries in the bottom right panel relative to the middle right panel, consistent with the SVC contributing more flow to systemic venous return.
The first set of bars in Figure 4⇓ displays the amount of caval contribution to total systemic venous return. The IVC contributed significantly less blood to the total systemic venous return than the SVC. The second set of bars in Figure 4⇓ displays the amount of blood each branch pulmonary artery receives. The LPA and RPA received nearly equal amounts of blood, with an LPA/RPA ratio of 1.14±0.48.
The coefficient of variation for intraobserver variability was 6%.
The fluid mechanics of Fontan physiology has received much attention,2 3 4 5 6 10 11 12 13 14 15 16 17 18 19 and ways to optimize this physiology have been explored. This included work into the basic driving force of blood flow into the pulmonary arteries in the absence of a “pulmonary” pumping chamber,6 10 11 12 13 14 15 16 17 18 19 optimization of the geometry of the systemic venous pathway,2 3 4 5 visualization of the velocity profiles,6 and computational fluid dynamics using finite-element analysis.2 This body of literature is predicated to various degrees on the knowledge of caval contributions to the branch pulmonary arteries, caval contributions to total systemic venous return, and relative blood flow to each lung. This underlies the need to obtain these values in vivo.
It has classically been thought that IVC blood carries a majority of systemic venous return.2 5 Recent suggested modifications to systemic venous pathway geometry take advantage of this by attempting to direct more IVC flow to the larger right lung and SVC flow to the smaller left lung.20 Salim et al7 recently demonstrated in normal children that caval contributions to systemic venous return vary with age, from a low in the IVC of 45% at ≈2.2 years old (at 1.8 years old in our study, the IVC contribution was 40%) to a high of 65%, the adult value, at 6.6 years old. In the face of changing caval flow and a fixed systemic venous pathway geometry, the dynamics of systemic venous flow is likely to change with age. Indeed, IVC flow directed toward the LPA makes sense for a 2-year-old, but this same patient at almost 7 years old will wind up with a vast majority of total systemic venous return directed toward the smaller lung.
The purpose of this study was to delineate the components of pulmonary blood flow in vivo by use of MRI and a novel application of the presaturation pulse.
The presaturation pulse acts as a marker of blood from the vessel it is applied to by producing a signal void.8 By placing a presaturation “band” on a vessel and imaging “downstream,” one can observe where the blood from that vessel goes by the “negative washout.” This study used the concept of quantifying this flow by measuring the decrease in signal intensity, validated it in a phantom, and measured the various flow components in vivo. We used images triggered to the ECG because we have demonstrated that the vast majority of flow in the systemic venous pathway is due to cardiac effects.6
Our observation that a substantial amount of SVC blood preferentially streams to the RPA and a substantial amount of IVC blood preferentially streams to the LPA is consistent with anecdotal evidence by angiocardiography.2
The way blood preferentially streams from the vena cava is a function of the geometry of the systemic venous pathway, the amount of flow each vena cava carries, and the amount of flow each lung can accommodate. The geometry of the systemic venous pathway in our study group differs from the geometry studied by others in vitro.3 4 5 The systemic venous pathways in our study patients were created in 2 stages by use of the “hemi-Fontan” procedure: a side-to-side anastomosis of the SVC to the RPA, with patch augmentation anterior to the SVC-RPA connection.
The distance between the axial centers of the SVC and the intra-atrial baffle is called the “offset.” Although in our patients, there is no lateral offset of the SVC and IVC to each other,4 there is an offset anteroposterior to the RPA and LPA from the SVC and IVC. Sharma et al4 noted that a zero offset of the SVC and IVC carries a 100% power loss; however, they did not study a side-to-side anastomosis or an anteroposterior offset.
LPA blood originated from both venae cavae equally. This is because the IVC (lower flow) contributes a majority of its blood to the LPA (40% of systemic venous return×66% of IVC blood is directed to the LPA=0.26), whereas the SVC (larger flow) contributes a small amount of blood to the LPA (60% of systemic venous return×40% of the blood is directed to the LPA=0.24).
A majority of RPA blood originated from the SVC. This is because the SVC (higher flow) contributes a majority of its blood to the RPA (60% of systemic venous return×60% of the blood is directed to the RPA=0.36), whereas the IVC (lower flow) contributes a small amount of blood to the RPA (40% of systemic venous return×33% of the blood is directed to the RPA=0.12).
As mentioned, the IVC has classically been thought to carry a majority of systemic venous return.2 5 Some groups have modified the systemic venous pathway geometry to take advantage of this20 by channeling IVC blood into the RPA (larger lung) and SVC blood into the LPA (smaller lung). Our data demonstrated that without this modification, the reverse is true. However, the IVC flow in our age group is the smaller of the two venae cavae (40%), consistent with the recent work by Salim et al7 in healthy children (45%). To some degree, however, the distribution of caval flow to each pulmonary artery is a function of the amount of flow in each vena cava. How the geometry of the systemic venous pathway will distribute blood to each pulmonary artery when the Fontan patient gets older is unknown. The strategy of directing IVC flow into the RPA makes long-term sense, because the adult value is reached by the age of 6.6 years.
The amount of flow each vena cava contributes to each pulmonary artery also depends on how much blood flow each lung can accommodate. We calculated that equal amounts of blood went to both lungs even though the right lung is larger and should hold more blood.2 4 5 Indeed, that is the rationale behind channeling the vena cava with higher flow to the right lung.20 Tamir et al21 noted that of the 5 single-ventricle patients studied by lung perfusion scans, 2 had decreased perfusion to the right lung and 3 had normal perfusion to both lungs. del Torso et al22 noted that 7 of 19 Fontan patients had decreased perfusion to the right lung as well. None of these studies had patients with lateral wall tunnel total cavopulmonary connections that would influence the distribution of blood flow because of streaming effects and the more laminar flow profile.6
The geometry of the lateral wall tunnel total cavopulmonary connection may direct blood preferentially to the LPA, “filling up” the smaller left lung and leaving the right lung “underfilled”; this may be the reason why both lungs receive equal perfusion. If true, altering the geometry of the systemic venous pathway may optimize perfusion to both lungs and thereby optimize systemic oxygenation. Other causes for equal perfusion to both lungs are the patient’s age (affects caval contribution to the systemic venous return), thromboembolic phenomenon, or autonomic nervous system alterations.
Our studies were performed early after Fontan reconstruction, before hospital discharge. It is unknown how the flow components would change later in the patient’s course.
Because MRI measures fixed planes in space, through-plane motion may cause “partial volume” of some of the signal intensity. Because the pulmonary arteries are minimally pulsatile in this physiology, this would not change our results significantly.
Similarly, “partial-volume effects” can occur when blood vessels are imaged because the signal intensity at the edges averages the signal of the blood and the surrounding soft tissue. This problem was avoided by measuring the decrease in signal intensity from images with a presaturation pulse and those without a presaturation pulse. The difference between the signal intensities is solely from blood.
The saturation pulse takes 1.0 ms to apply and is applied 4.7 ms before the center of the excitation pulse used in the acquisition sequence. During that period, some recovery of spins would have occurred. This effect is not appreciable because of the small amount of time available for the spins to recover. It would also not change the relative percentages because spin recovery would occur in both pulmonary arteries.
The right-to-left flow from systemic to pulmonary venous pathways across the baffle fenestration would, in theory, alter the exact numbers for relative caval contribution to systemic venous return only. To minimize this, we studied patients with insignificant flow across this (aortic saturation, 91.4±2.9%).
It should be noted that this study was performed in sedated, supine patients. The findings may be altered with increased patient activity.
Finally, this technique is not applicable to patients with turbulent flow in the pulmonary arteries because turbulence would create a signal void and alter the calculations. This would probably apply only to patients with moderate to severe bilateral pulmonary artery obstruction.
This study has validated the use of a presaturation pulse to quantify the relative contributions of caval flow to pulmonary artery flow and total systemic venous return. It may be used to evaluate flow components in the various systemic venous pathways created surgically.
In Fontan patients, SVC blood is directed more toward the RPA and IVC blood is directed more toward the LPA. Nevertheless, LPA blood is composed of equal amounts of IVC and SVC blood, and the majority of RPA blood is from the SVC. In addition, the IVC accounts for only 40% of the total systemic venous return, and equal amounts of blood flow to each lung. These findings may help in designing a better Fontan procedure and change the way the systemic venous pathway geometry is created.
Derivation of Blood Flow Calculations
The assumptions (where k, k′, etc, are constants) are that (1) the imaging slice thickness equals the pulmonary artery diameter; (2) flow in the systemic venous pathway is laminar6 ; (3) the signal intensity is proportional to the volume of tissue; and (4) after caval blood saturation, the decrease in signal intensity, δI, is proportional to the blood volume (δVol) from the cava: The remaining signal is not equal to the signal from the unsaturated vena cava because nonblood tissue may be present in each pixel.
δVol enters an artery in time δt and fills the vessel for a distance (vm)(δt), where vm is mean velocity. Saturated blood is the sum of all δVol in that artery, which is the integral of δI in each pixel over area A extending the pulmonary artery width and the length of (vm)(δt). Area A is where R is the arterial radius. Thus, Substituting the product of the average signal drop (〈δI〉) in area A by area A (Figure 1⇑, Equation 3) over time δt into Equation 4, we obtain Velocity determination is not necessary for the final calculations (measuring proportional flow, not absolute flow). Assuming that the pressure gradient (dP/dx) and viscosity along both arteries are equal (no pulmonary artery stenosis or hypoplasia), Poiseuille’s law states that vm along each artery is proportional to the square of the radius: vm ∝ R2. Substituting in Equation 5, Because of periodic flow, we integrate Equation 6 over the cycle (heartbeat) to obtain where k‴=(k″)(δt). Writing D for ∑ 〈δI〉 (the sum of all average signal intensity drops over the cardiac cycle), we obtain Use the notation RL, RR=LPA and RPA radii, respectively; DIL, DSL=average intensity drop (D) in the LPA when the IVC and SVC are saturated, respectively; DIR, DSR=D in the RPA when the IVC and SVC are saturated, respectively; VIL, VSL=volume of blood (V) entering the LPA when the IVC and SVC are saturated, respectively; and VIR, VSR=V entering the RPA when the IVC and SVC are saturated, respectively.
1. Caval contribution to each pulmonary artery: Using the IVC and LPA as an example, the fraction of blood from the IVC that enters the LPA is Other ratios of caval blood entering the pulmonary arteries are written similarly.
2. The percentage of blood in each pulmonary artery from each vena cava: Using the IVC and LPA as an example, the fraction of blood in the LPA that originated from the IVC is Other ratios of pulmonary artery blood that originated from each vena cava are written similarly.
3. Caval contribution to total systemic venous return: Using the IVC as an example, the fraction of total systemic venous return that that is contributed from the IVC is The SVC contribution to total systemic venous return is written similarly.
4. Relative blood flow to each lung: Use the following notation: IVR, SVR=% IVC and % SVC contribution to total systemic venous return, respectively; IL, SL=% IVC and % SVC contribution to LPA flow, respectively; and IR, SR=% IVC and % SVC contribution to RPA flow, respectively (all values calculated from the previous equations). Using the LPA as an example, the fraction of total systemic venous return that enters the LPA is (IVR)(IL)+(SVR)(SL).
The % for RPA flow is written similarly, with IR and SR instead of IL and SL.
The authors acknowledge the help of Henry Drott, MD, in the phantom validation.
- Received February 19, 1998.
- Revision received November 9, 1998.
- Accepted November 18, 1998.
- Copyright © 1999 by American Heart Association
Fontan F, Baudet E. Surgical repair of tricuspid atresia. Thorax. 1971;26:240–248.
de Laval MR, Dubini G, Migliavacca F, Jalali H, Camporini G, Reddington A, Pietrabissa R. Use of computational fluid dynamics in the design of surgical procedures: application to the study of competitive flows in cavopulmonary connections. J Thorac Cardiovasc Surg. 1996;111:502–513.
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.
Lardo AC, Webber SA, del Nido PJ, Myers JL, Boyle GJ, Lee ET, Siewers RD, Cape EG. Does the right lung receive preferential blood flow in Fontan repair? Comparison of total cavopulmonary and atriopulmonary connections: an in vitro flow study. Circulation. 1995;92(suppl I):I-123. Abstract.
Salim MA, DiSessa TG, Arheart KL, Alpert BS. Contribution of superior vena caval flow to total cardiac output in children: a Doppler echocardiographic study. Circulation. 1995;92:1860–1865.
Horowitz AL. MRI Physics for Radiologists. New York, NY: Springer-Verlag; 1995:103–144.
Penny DJ, Redington AN. Doppler echocardiographic evaluation of pulmonary blood flow after the Fontan operation: the role of the lungs. Br Heart J. 1991;66:372–374.
Qureshi SA, Richheimer R, McKay R, Arnold R. Doppler echocardiographic evaluation of pulmonary artery flow after modified Fontan procedure: importance of atrial contraction. Br Heart J. 1990;64:272–276.
Nakazawa M, Nakanishi T, Okuda H, Satomi G, Nakae S, Imai Y, Takao A. Dynamics of right heart flow in patients after Fontan procedure. Circulation. 1984;69:306–312.
DiSessa TG, Child JS, Perloff JK, Wu L, Williams RG, Laks H, Friedman WF. Systemic venous and pulmonary arterial flow patterns after Fontan’s procedure for tricuspid atresia or single ventricle. Circulation. 1984;70:898–902.
Redington AN, Penny DJ, Shinebourne EA. Pulmonary blood flow after total cavopulmonary shunt. Br Heart J. 1991;65:213–217.
Laks H, Ardehali A, Grant PW, Permut L, Aharon A, Kuhn M, Isabel-Jones J, Galindo A. Modification of the Fontan procedure: superior vena cava to left pulmonary artery connection and inferior vena cava to right pulmonary artery connection with adjustable atrial septal defect. Circulation. 1995;91:2943–2947.