Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Circ Cardiovasc Imaging. Author manuscript; available in PMC 2010 September 1.
Published in final edited form as:
PMCID: PMC2789422

Noninvasive Quantification of Systemic To Pulmonary Collateral Flow: A Major Source of Inefficiency in Patients with Superior Cavopulmonary Connections



Systemic to pulmonary collateral flow (SPCF) is common in single ventricle patients (pts) with superior cavopulmonary connections (BDG). Because no validated method to quantify SPCF exists, neither its hemodynamic burden nor clinical impact can be systematically evaluated. We hypothesize that (1) the difference in total ascending aortic (Ao) and caval flow (SVC+IVC) and (2) the difference between pulmonary vein and artery flow (PV − PA) provide two independent estimators of SPCF.

Methods and Results

We measured Ao, SVC, IVC, right (RPA) and left (LPA) PA, left (LPV), and right (RPV) PV flows in 17 BDG pts during routine cardiac magnetic resonance imaging studies using through-plane phase contrast velocity mapping. Two independent measures of SPCF were obtained: (1) Ao − (SVC + IVC). (2) (LPV−LPA) + (RPV−RPA). Values were normalized to body surface area (BSA), Ao, and PV and comparisons made using linear regression and Bland-Altman analysis. SPCF ranged from 0.2–1.4 L/min for (1) and 0.2–1.6 L/min for (2) for an average indexed SPCF of 0.5–2.8 L/min/m2 : 11–53% (mean 37%) of Ao and 19–77% (mean 54%) of PV. The mean difference between (1) and (2) was 0.01 L/min (p=0.40, 2 S.D. range −0.45–0.47 L/min).


We present a noninvasive method for the SPCF quantification in pts with BDG. It should provide an important clinical tool in managing these patients. Furthermore, we show that SPCF is a significant hemodynamic burden in many patients with BDG physiology. Future investigations will allow objective study of the impact of collateral flow on outcome.

Keywords: single ventricle, collateral circulation, magnetic resonance imaging, blood flow, superior cavopulmonary connection


It has long been recognized that single ventricle patients with cavopulmonary anastomoses are susceptible to developing systemic to pulmonary collateral flow. 1, 2 There has been much investigation and speculation into the etiology of these collaterals. Cyanosis, pleural effusion, and decreased pulmonary blood flow have all been implicated in the development of collateral flow to the lungs.2, 3

Controversy exists over the prevalence of collateral flow and more importantly, the significance of these collaterals. Some investigations have shown that the presence of significant systemic to pulmonary collateral flow is a risk factor for pleural effusion, poor outcomes, and heart failure.1, 3, 4 However, other investigators have failed to show a difference in outcome based on amount of collateral flow.5 Some studies have demonstrated increased prevalence of collaterals in superior cavopulmonary connections compared to total cavopulmonary anastomoses.2 Collateral flow is potentially a significant source of power loss.and other research has suggested that collateral flow results in additional power loss in the Fontan pathway by transferring kinetic energy to the distal pulmonary vasculature and causing competitive flow losses.6

One of the major obstacles to investigating the importance of systemic to pulmonary collateral flow in single ventricle physiology has been the inability to accurately quantify this flow. Classically, these collaterals have been identified and then graded qualitatively by angiography.2 However, the validity of the grading systems has never been verified. When a decision is made to coil collaterals felt to be hemodynamically significant, there is no good method of assessing the effect of the procedure.

One potential method to quantify collateral flow involves the use of magnetic resonance imaging (MRI), utilizing through-plane phase contrast velocity mapping (PC-MRI), a technique which has been validated in multiple in vitro and clinical investigations710 We recently published flow data obtained from PC-MRI velocity mapping on 105 Fontan patients in which a significant difference was noted between the aortic flow and total caval flow. It was hypothesized that this difference was primarily due to systemic to pulmonary collateral flow.11

As a method of validating the difference between measured aortic and caval flow as an estimator of collateral flow, we propose a second estimator(the difference between measured pulmonary vein flow and pulmonary artery flow), providing two independent estimators of collateral flow.

The primary goal of this study is to validate the described non-invasive method to quantify systemic to pulmonary collateral flow. We hypothesize that collateral flow can be accurately estimated noninvasively in patients with superior cavopulmonary anastomoses by cardiac MRI utilizing phase-contrast velocity mapping. Furthermore, we hypothesize systemic to pulmonary collateral flow is a significant hemodynamic burden in many patients with superior cavopulmonary anastomoses.



We retrospectively investigated 17 consecutive patients who underwent superior cavopulmonary connections at the Children’s Hospital of Philadelphia and subsequently underwent a cardiac MRI from April to September of 2008. Patients were between 0.7 and 3.4 years of age (mean 2.1 years). There were 8 patients with hypoplastic left heart syndrome, 5 patients with tricuspid atresia, 3 patients with heterotaxy, and one with double inlet left ventricle. Ten patients had dominant right ventricles, 6 had dominant left ventricles, and one patient has two good sized ventricles. There were 13 patients with superior cavopulmonary anastomoses (3 with bilateral SVC’s, 2 with bilateral cavopulmonary connections and one with a small left SVC still connected to the left atrium) and 4 patients with hemi-Fontans.


In order to establish a control group, we retrospectively reviewed 13 two-ventricle patients (6 patients with arch anomalies and no prior surgery, as well as 7 post-operative 2-ventricle repair patients with no known residual shunts) who had complete pulmonary vein flow measurements. Seven of these patients also had vena caval velocity maps. These patients ranged in age from 1.7 years to 29 years.


All patients underwent cardiac magnetic resonance imaging consisting of balanced steady-state free precession (bright blood) and half-Fourier single-shot turbo spin-echo (dark blood) axial stacks, balanced steady-state free precession cine imaging for anatomy and to quantify ventricular function, gadolinium angiography, and through-plane PC-MRI cines as part of their routine clinical management. Multiplanar reformatting was used to set the position and angle of the imaging plane for anatomical and PC-MRI cines. Retrospectively gated, through-plane PC-MRI cines were performed in the aorta (native and/or neo-aorta), superior and inferior vena cava (SVC and IVC), right and left pulmonary arteries (RPA and LPA), and right and left pulmonary veins (RPV and LPV). We have previously described the typical parameters and positions we use to obtain the velocity mapping sequences.11 Figure 1 demonstrates the typical positions that were used to obtain the velocity maps. The aortic flow was measured at the sinuses. When two outflows were present (aorta and neo-aorta), they were measured separately and added for the total aortic flow. Care was taken to obtain the right pulmonary flow close to the SVC-RPA anastomosis so as to include the right upper lobe branch, which is often very close. The pulmonary veins were measured using an encoding velocity of 50 to 100 centimeters per second. When possible (based on whether all the pulmonary veins on one side formed a common vein of sufficient length before entering the atrium), all the pulmonary veins on one side were measured in one velocity map. This was more common for the left pulmonary veins, which often form a single confluence before joining the left atrium. When necessary, the upper and lower pulmonary veins, and occasionally a middle pulmonary vein, were measured separately.

Figure 1Figure 1
Top: Schematic showing the location of the phase contrast velocity used to calculate collateral flow. The yellow bars represent the location of the velocity maps. Note that often the left pulmonary veins could be obtained together as shown, but other ...

Analysis and Statistics

Velocity mapping sequences were previously analyzed using Argus flow analysis software on a Leonardo workstation (Siemens’, Inc.) to obtain the aortic, SVC, IVC, RPA, LPA, RPV and LPV flows (Q). These flow values were obtained from the patient’s MRI report. The collateral flow was then calculated for each patient by the two different methods of collateral flow (Qcoll-syst and Qcoll-pulm) from equations 1 and 2:



where Qcoll-syst and Qcoll-pulm represent the estimated collateral flow by comparing supply and return of the systemic and pulmonary systems respectively.

The collateral flow was normalized to aortic flow to determine the percent of cardiac output, to body surface area (BSA) to obtain an indexed flow, and to total pulmonary vein flow (QRPV + QLPV) to determine the percent of pulmonary flow from collateral flow. In addition, the total venous return to the heart was calculated (QIVC + QRPV + QLPV) to compare with the aortic flow as an indicator of internal consistency.

To establish inter and intra-observer variability, all the flow measurements of 4 patients were repeated by the same observer (28 flow measurements in all), and all the measurements of 4 different patients were repeated by a different observer (additional 28 flow measurements). The percent difference between the two measurements were calculated for each of the 7 measured flows. In addition, the mean and standard deviation of the difference between the two observations of the same collateral flow estimator was calculated. Reliability coefficients were calculated using single measure intraclass correlation for the intra- and inter-observer agreement for Qcoll-pulm and Qcoll-syst.

Ventricular volumes, including the absolute and indexed end-diastolic volume, end-systolic volume, and ejection fraction were also obtained from the clinical report. Estimated indexed collateral flow was compared to indexed end diastolic ventricular volumes by linear regression.

There were 7 patients who had cardiac catheterizations with aortic angiography adequate for collateral grading by angiography as described by Spicer and colleagues.4 Collateral grading by angiography was compared to the quantification of collateral flow by Spearman’s rank correlation coefficient for non-parametric parameters.

The two methods of calculating collateral flow were compared using linear regression, Bland-Altman analysis, intraclass correlation, and paired Student t-test. Control and SCPC collateral flow parameters were compared using unpaired Student t-tests. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written. The study was approved by the institutional review board.


Table 1 summarizes the 17 patients and the relevant data. Estimated collateral flow ranged from 0.2 to 1.4 L/min for Qcoll-syst and 0.2 to 1.6 L/min for Qcoll-pulm, which corresponds to an indexed collateral flow of 0.5 to 2.9 L/min/m2, with a mean of 1.8 L/min/m2. Collateral flow accounted for 11% to 54% of the aortic flow, with a mean of 37%. The estimated percent contribution of collateral flow to total pulmonary blood flow ranged from 19 to 77%, with a mean of 54%. The measured pulmonary to systemic flow ratio (Qp/Qs), when corrected for collateral blood flow, was on average 1.1, with a range of 0.6 to 1.7.

Table 1
Summary of Patients and Collateral Flow

There was a strong correlation between the two methods of estimating collateral flow (Figure 2) with a correlation coefficient of 0.80 and a regression slope of 0.78. The intraclass correlation coefficient was 0.81 (p<0.001) indicated good agreement. Looking at the Bland-Altman analysis in Figure 2, there was little bias, with a mean difference between the two methods of only 0.01 L/min, which was not statistically significant (p=0.40). The standard deviation of the difference between the two methods was 0.23 L/min, for a 2 S.D. range of −0.45 to 0.47 L/min

Figure 2Figure 2
Top: Qcoll-pulm vs. Qcoll-syst demonstrating excellent correlation between the two methods of estimating systemic to pulmonary collateral flow. Bottom: Bland-Altman plot of the difference between the two systemic to pulmonary collateral flow estimators ...

Venous return to the heart demonstrated excellent correlation with the measured aortic flow, with a correlation coefficient of 0.88 and a regression slope of 1.06 (Figure 3). It also showed little bias, with a mean difference between the two measures of only 0.01 L/min and a standard deviation of 0.27 L/min (Figure 3), for a 2 S.D. range of −0.53 to 0.55.

Figure 3Figure 3
Venous return to the heart (measured total pulmonary vein plus IVC flow) vs. measured aortic output for SCPC pts, demonstrating excellent agreement. Bottom: Bland-Altman plot for SCPC pts of the difference between the measured venous return to the heart ...

Comparison to controls

Table 2 summarizes the comparisons between the control group and the study population. In the control group, collateral flow was on average 0.2 L/min/m2, which is an order of magnitude less than the SCPC group and statistically significant. The measured collateral flow average 5% of both aortic and pulmonary blood flow, both significantly less than for the SCPC group. There is also excellent correlation between the pulmonary venous return and the aortic flow (Figure 3). Because the control group is retrospective, there is a significant difference between the age and BSA, making it less than ideal. However, there is significant overlap in the ages and collateral flow was not a function of age in this group, making this less of a concern.

Table 2
Summary of SCPC patients vs. controls. Controls were unoperated patients with arch anomalies and no intracardiac shutns, and post-operative 2-ventricle repair patients with no residual intracardiac shunts. Qcoll is indexed average measured collateral ...

Relationship between collateral flow and ventricular volumes

When using the entire cohort, there was no significant correlation between collateral flow and ventricular volumes. However, two patients, one with severe tricuspid regurgitation and another with moderate RV dysfunction, had very dilated ventricles for other reasons. When these two were excluded, indexed end-diastolic volume correlated significantly with indexed collateral flow (Figure 4) (r=0.59, p=0.02).

Figure 4
Indexed end-diastolic volume of the ventricle compared to the amount of indexed collateral flow, demonstrates significant correlation. Note that two outliers were deleted who had other reasons for significant RV dilation (severe tricuspid regurgitation, ...

There was no relationship between collateral flow and SCPC type, underlying type of heart disease or ventricular morphology. Collateral flow was slightly greater to the left lung on average (56% vs. 44%), but the difference was not significant (p = 0.25). Indexed collateral flow increased with time from the SCPC operation: the correlation was weak but significant (Figure 5).

Figure 5
With one outlier deleted who had a large amount of collateral flow soon after surgery, mean collateral flow (½ (Qcoll-syst + Qcoll-pulm)) indexed to BSA vs. time from SCPC surgery demonstrates a weak but significant positive correlation, suggesting ...

Angiographic grading of collateral flow did not correlate with collateral flow indexed to BSA, Ao flow, or pulmonary blood flow.

Inter- and Intra-observer variability

Intra-observer variability for all the flow measurements combined was 4% with a standard deviation of 4%. The inter-observer variability was 7% with a standard deviation of 7%. Both intra- and inter-observer differences between both measures of collateral flow were less than 0.1 L/min on average, with a standard deviation of 0.06 L/min. The intraclass correlation coefficient for intra-observer agreement was 0.986 for Qcoll-syst (p=0.0001) and 0.960 for Qcoll-pulm (p=0.001). The coefficient of inter-observer agreement was 0.836 for Qcoll-syst (p=0.05) and 0.986 for Qcoll-pulm (p=0.001).


We present a non-invasive method of quantifying systemic to pulmonary collateral flow in patients with cavopulmonary connections that provides two independent measures of collateral flow. There was excellent agreement between the two methods studied, providing internal validation for this noninvasive estimator.

In the 17 patients studied, the collateral flow averaged 54% of the total pulmonary blood flow and 37% of the cardiac output. In addition, in patients without other significant reasons for ventricular dilation, indexed collateral flow correlated with indexed end-diastolic volume. This suggests that collateral flow is a significant hemodynamic burden in this population and deserves further study. It may also explain previous findings by Fogel and colleagues that, despite the dogma that the superior cavopulmonary connection results in significant volume unloading, measured ventricular volumes were not significantly different after the cavopulmonary connection.12 The measured Qp/Qs mean of 1.1 in our study is in significant contrast to data obtained from oximetric evaluation at cardiac catheterization where typical measurements of Qp/Qs range from 0.4 to 0.6.13, 14 This suggests that cardiac catheterization significantly underestimates the actual pulmonary blood flow. This is not unexpected as the collateral connections enter the pulmonary tree distally precluding accurate measurement of mixed pulmonary arterial saturation. While indexed collateral flow demonstrated no correlation with angiographic grading of collateral flow, the limited number of patients with angiography prevents drawing conclusions.

While the current population studied has superior cavopulmonary connections, the methodology should be valid for patients with total cavopulmonary connections as well. This data suggests that collateral flow is in general quite significant in patients with superior cavopulmonary connections. Our previously reported study of Fontan flows suggest that collateral flow is less, but still significant in Fontan physiology.11 Further studies are required to quantify collateral flow in the Fontan population using the techniques described in this study.

Previous Studies

There have been few studies focused on quantifying systemic to pulmonary collateral flow. In 2001, Bradley and colleagues measured collateral flow directly in 32 patients while on bypass and about to undergo a Fontan operation.5 Collateral flow was present in all patients and ranged from 9 to 49% of pump flow. These values are similar to the range of 11 to 53 percent seen in our study. Systemic and pulmonary vascular resistances have significant effects on collateral flow. Given that the flows were measured on bypass, the measured collateral flows may well not reflect the baseline physiologic state. Additionally, this method has the obvious limitation of being invasive and thus not a practical means of tracking collateral flow over time.

Another notable study was performed by Inuzauka and colleagues15 and quantified collateral flow in 10 patients using a combination of cardiac catheterization oximetry data and perfusion scintillography. Scintillography data from a lower extremity injection quantified the ratio of collateral to systemic flow. Then oximetry data obtained from catheterization was used to calculate absolute collateral flow and total pulmonary blood flow. They obtained a mean indexed collateral flow of 1.75 L/min/m2, quite similar to our 1.8 L/min/m2. The estimated mean systemic flow for the patients was 3.4 L/min/m2 and the mean pulmonary flow was 3.0 L/min/m2. This corresponds to collateral flow being on average 34% of cardiac output and 58% of pulmonary blood flow, which agrees very well with our current study. The obvious disadvantage to this method is that it requires an invasive procedure with ionizing radiation in order to obtain the data. It is also fairly cumbersome and assumes complete mixing in the IVC, which is often not a valid assumption. Finally, it involves two separate procedures that are performed under different conditions and thus may limit the accuracy of individual measurements.

Finally, Grosse-Wortmann and colleagues recently described a method similar to the one presented here to quantify collateral flow in patients with both bidirectional cavopulmonary connections and Fontans.16 An important difference was the use of descending aortic flow as a surrogate to IVC flow. They reported similar values for collateral flow in SCPC patients, including average collateral flow using the pulmonary estimator of 1.4 L/min/m2, averaging 46% of total pulmonary blood flow and 36% of aortic flow and a mean Qp/Qs was 0.93 However, it should be noted that while our study demonstrated no significant bias between the two estimators, they reported a significant bias between the systemic and pulmonary collateral estimators, with the systemic estimator on average 0.64 L/min/m2 less than the pulmonary estimator. In addition, the intraclass correlation coefficient was lower at 0.73, compared to 0.81 in our study. These differences are likely accounted for by two important methodological differences: 1) their use of the descending aortic flow as a surrogate for IVC flow and 2) the measurement of aortic flow distal to the aortopulmonary anastomosis in patients with aortic reconstructions. We have demonstrated that the IVC flow can be reliably measured directly, and it will not be subject to inaccuracies caused by decompressing vessels or collateral vessels arising below the level of descending aortic measurement. The measurement of aortic flow distal to the aortopulmonary anastomosis of a reconstructed arch is not desirable secondary to the potential for disturbed flow from flow collisions between the native and neo-aortic flows and swirling in the often dilated reconstructed arch. We therefore propose that the proper way to measure aortic flow in patients with reconstructed arches is to measure the flows just above the level of the semilunar valves. When there are two semilunar valves which contribute to total aortic flow, they should be measured separately and summed to provide the total aortic flow. The study also reported difficulty in accurately measuring pulmonary vein flows in patients with complicated pulmonary vein anatomy, and excluded 7 of 36 patients from their analysis for this reason. We did not encounter this same difficulty, despite three patients with heterotaxy and two with abnormal pulmonary venous connections. No patients in our study undergoing MRI during the study period were excluded secondary to abnormalities of pulmonary veins. Finally, the prior study reported increased collateral flow to the side of a prior BT shunt, a finding that was not reproduced in our study. Contrary to the reported finding by Grosse-Wortmann and colleagues of a trend between younger age and higher collateral flow, we found the opposite, with a significant correlation between elapsed time from SCPC surgery and indexed collateral flow.


While we were unable to find any difference in collateral flow based on differences in anatomy or surgical type, the study is underpowered to find such differences. The described method could not be compared to a gold standard as is ideal when testing a new measurement because no gold standard exists. However, the fact that this method incorporates two independent measures of collateral flow largely overcomes this limitation and is a significant advantage both in validating the technique and applying it clinically. The control population is not age-matched to the study group, which is a limiting factor. However, the age range did overlap significantly (youngest control 1.6 years with three under 3 years of age), and there was no relationship between age and collateral flow in the control population, partially alleviating this concern. Because there was angiographic data on only a limited number of patients, we are unable to draw conclusions regarding the validity of angiographic grading of collaterals.


We have presented a series of patients who underwent non-invasive quantification of systemic to pulmonary collateral flow using cardiac magnetic resonance phase contrast velocity mapping. The excellent agreement between the two methods provides compelling evidence that this is an accurate method for quantifying collateral flow. Furthermore, this collateral flow represents a substantial hemodynamic burden to patients with SCPC (on average 37% of the cardiac output and 54% of the pulmonary blood flow).

We present a methodology similar to that recently reported by Grosse-Wortmann and colleagues, with important differences that result in improved accuracy and lower bias. Notably, our measurements do not appear subject to the occasional gross underestimation of collateral flow by the systemic collateral estimator noted in the other study. We propose that the IVC flow should be measured directly rather than indirectly by descending aorta flow, and that the aortic flow should be measured just above the semilunar valves in patients with Norwood-type arch reconstructions, rather than distal to the aortic to pulmonary anastomosis. When there is significant antegrade flow across two semilunar valves, they should be measured separately and summed together.

This study cannot address the important questions of why these collaterals develop, which patients are most susceptible to collateral formation, and perhaps most importantly what the long term impact of this significant collateral flow is. However, it does provide a non-invasive tool that should allow us to answer these important questions in future investigations. By monitoring these patients prospectively, we may be able to determine who develops collaterals, what the impact of these collaterals is on both short-term and long-term Fontan outcomes, and what happens to the collateral flow after Fontan completion.


Funding Sources

K.K.W. was supported in part by NIH K23 Grant HL089647 from the National Heart, Lung and Blood Institute.





1. McElhinney DB, Reddy VM, Tworetzky W, Petrossian E, Hanley FL, Moore P. Incidence and implications of systemic to pulmonary collaterals after bidirectional cavopulmonary anastomosis. The Annals of Thoracic Surgery. 2000;69:1222–8. [PubMed]
2. Triedman JK, Bridges ND, Mayer JE, Jr, Lock JE. Prevalence and risk factors for aortopulmonary collateral vessels after Fontan and bidirectional Glenn procedures. J Am Coll Cardiol. 1993;22:207–15. [PubMed]
3. Kanter KR, Vincent RN, Raviele AA. Importance of acquired systemic-to-pulmonary collaterals in the Fontan operation. The Annals of Thoracic Surgery. 1999;68:969–74. [PubMed]
4. Spicer RL, Uzark KC, Moore JW, Mainwaring RD, Lamberti JJ. Aortopulmonary collateral vessels and prolonged pleural effusions after modified Fontan procedures. Am Heart J. 1996;131:1164–8. [PubMed]
5. Bradley SM, McCall MM, Sistino JJ, Radtke WAK. Aortopulmonary collateral flow in the Fontan patient: does it matter? The Annals of Thoracic Surgery. 2001;72:408–15. [PubMed]
6. Ascuitto RJ, Ross-Ascuitto NT. Systematic-to-Pulmonary Collaterals: A Source of Flow Energy Loss in Fontan Physiology. Pediatric Cardiology. 2004;25:472–81. [PubMed]
7. Beerbaum P, Korperich H, Barth P, Esdorn H, Gieseke J, Meyer H. Noninvasive Quantification of Left-to-Right Shunt in Pediatric Patients : Phase-Contrast Cine Magnetic Resonance Imaging Compared With Invasive Oximetry. Circulation. 2001;103:2476–82. [PubMed]
8. Debatin JF, Leung DA, Wildermuth S, Botnar R, Felblinger J, McKinnon GC. Flow quantitation with echo-planar phase-contrast velocity mapping: in vitro and in vivo evaluation. J Magn Reson Imaging. 1995;5:656–62. [PubMed]
9. Greil G, Geva T, Maier SE, Powell AJ. Effect of acquisition parameters on the accuracy of velocity encoded cine magnetic resonance imaging blood flow measurements. J Magn Reson Imaging. 2002;15:47–54. [PubMed]
10. Powell AJ, Maier SE, Chung T, Geva T. Phase-Velocity Cine Magnetic Resonance Imaging Measurement of Pulsatile Blood Flow in Children and Young Adults: In Vitro and In Vivo Validation. Pediatric Cardiology. 2000;21:104–10. [PubMed]
11. Whitehead KK, Sundareswaran KS, Parks WJ, Harris MA, Yoganathan AP, Fogel MA. Blood Flow Distribution in a Large Series of Fontan Patients: A Cardiac Magnetic Resonance Velocity Mapping Study. JTCS. 2009 In press. [PMC free article] [PubMed]
12. Fogel MA, Weinberg PM, Chin AJ, Fellows KE, Hoffman EA. Late ventricular geometry and performance changes of functional single ventricle throughout staged fontan reconstruction assessed by magnetic resonance imaging. J Am Coll Cardiol. 1996;28:212–21. [PubMed]
13. Hoskote A, Li J, Hickey C, Erickson S, Van Arsdell G, Stephens D, Holtby H, Bohn D, Adatia I. The effects of carbon dioxide on oxygenation and systemic, cerebral, and pulmonary vascular hemodynamics after the bidirectional superior cavopulmonary anastomosis. J Am Coll Cardiol. 2004;44:1501–9. [PubMed]
14. Salim MA, Case CL, Sade RM, Watson DC, Alpert BS, DiSessa TG. Pulmonary/systemic flow ratio in children after cavopulmonary anastomosis. J Am Coll Cardiol. 1995;25:735–8. [PubMed]
15. Inuzuka R, Aotsuka H, Nakajima H, Yamazawa H, Sugamoto K, Tatebe S, Aoki M, Fujiwara T. Quantification of collateral aortopulmonary flow in patients subsequent to construction of bidirectional cavopulmonary shunts. Cardiol Young. 2008:1–9. [PubMed]
16. Grosse-Wortmann L, Al-Otay A, Yoo SJ. Aortopulmonary Collaterals After Bidirectional Cavopulmonary Connection or Fontan Completion: Quantification With MRI. Circ Cardiovasc Imaging. 2009;2:219–25. [PubMed]