Prakash et al Congenital Heart Disease

A new diagnostic algorithm for assessment of patients with single

ventricle before a Fontan operation
Ashwin Prakash, MD, Muhammad A. Khan, MD, Rose Hardy, BA, Alejandro J. Torres, MD,
Jonathan M. Chen, MD, and Welton M. Gersony, MD

Objectives: Cardiac catheterization has a low diagnostic yield before a Fontan operation, and magnetic resonance
imaging and computed tomography are reliable alternatives to invasive angiography. A new diagnostic algorithm
to avoid cardiac catheterization in ‘‘low-risk’’ subjects before a Fontan operation is proposed.
Methods: The proposed algorithm would identify ‘‘high-risk’’ subjects on the basis of risk factors on medical
history, echocardiography, and noninvasive angiography. The efficacy of this algorithm in screening for subjects
deemed to be inoperable after catheterization was evaluated retrospectively in 151 children. For this analysis,
results of conventional angiography (assumed to be equivalent to noninvasive angiography) were used.
Results: According to the algorithm, 95 (63%) of 151 subjects had no risk factors (‘‘low risk’’) whereas 56
(37%) of 151 had 1 risk factor or more (‘‘high risk’’). Nine (6%) of 151 subjects were found to be inoperable
after catheterization and all 9 were correctly classified as high risk by the algorithm. In the 135 of 151 subjects
who underwent a Fontan operation, the algorithm predicted an adverse postoperative outcome with a sensitivity of
51% and specificity of 78%. However, this prediction was not improved by including elevated pulmonary artery
pressure or ventricular filling pressure as additional risk factors.
Conclusions: The proposed algorithm effectively screened for subjects who were deemed unsuitable for a Fontan
procedure. In addition, omitting preoperative invasive hemodynamic assessment did not impair prediction of
adverse postoperative outcomes. Prospective evaluation of such a noninvasive diagnostic strategy before the Fon-
tan operation is warranted.

angiography.2-6 Cardiac catheterization continues to be used

Earn CME credits at routinely for children who are referred for a Fontan opera-

CHD
http://cme.ctsnetjournals.org tion to perform hemodynamic assessment for the possibility
of elevated pulmonary artery pressure (PAP) or increased
ventricular filling pressure (VFP), which can only be identi-
The Fontan operation is the final surgical stage for most fied by invasive testing. However, in the modern era of
patients with a functionally single ventricle, resulting in timely staged palliation, it is unusual for patients with func-
a marked change in cardiovascular physiology.1 Since tionally single ventricle to be subjected to unrestrictive pul-
inception, detailed hemodynamic assessment and angio- monary blood flow and have significant pulmonary vascular
graphic evaluation of the extracardiac thoracic vasculature disease with resultant pulmonary hypertension. Ventricular
by cardiac catheterization before surgery has been consid- diastolic dysfunction is a risk factor for patients undergoing
ered to be mandatory. However, for most other types of con- the Fontan operation, but it may be difficult to identify,
genital heart disease, the use of preoperative catheterization especially with load-dependent parameters such as VFP.7
has become infrequent. In part, this is due to the advent of It is rare for a patient to be deemed unsuitable for a Fontan
newer noninvasive diagnostic techniques such as echocardi- operation as a result of routine invasive hemodynamic test-
ography, magnetic resonance imaging (MRI), and computed ing in the absence of known risk factors.8 Thus, given the
tomography (CT) as reliable replacements for invasive x-ray low yield of catheterization, it may be possible to limit
hemodynamic testing to only high-risk subjects who have
From the Division of Pediatric Cardiology, Columbia University College of Physi- been identified by clinical criteria and noninvasive tech-
cians and Surgeons, New York, NY. niques. However, it is not known whether using such an
Presented in part at the Annual Scientific Sessions of the American College of Cardi-
ology, 2007. approach would allow the identification of all subjects
Received for publication Oct 21, 2008; revisions received Feb 9, 2009; accepted for who are inoperable. It is also not known whether the omis-
publication March 9, 2009; available ahead of print June 1, 2009. sion of routine invasive hemodynamic testing in low-risk
Address for reprints: Ashwin Prakash, MD, Department of Cardiology, Children’s
Hospital, 300 Longwood Ave, Boston, MA 02115 (E-mail: ashwin.prakash@ patients would impair the ability to predict adverse postop-
cardio.chboston.org). erative outcomes.
J Thorac Cardiovasc Surg 2009;138:917-23 As a first step toward answering these questions, in the
0022-5223/$36.00
Copyright Ó 2009 by The American Association for Thoracic Surgery present study, we developed a diagnostic algorithm for
doi:10.1016/j.jtcvs.2009.03.022 assessment of patients before a Fontan operation. In this

The Journal of Thoracic and Cardiovascular Surgery c Volume 138, Number 4 917
 Congenital Heart Disease Prakash et al

Abbreviations and Acronyms

CPB ¼ cardiopulmonary bypass
CT ¼ computed tomography
ICU ¼ intensive care unit
MRI ¼ magnetic resonance imaging
NPV ¼ negative predictive value
PAP ¼ pulmonary artery pressure
PPV ¼ positive predictive value
VFP ¼ ventricular filling pressure
FIGURE 1. Proposed diagnostic algorithm: Characteristics on history,
echocardiography, or noninvasive angiography are used to identify high-
algorithm, invasive hemodynamic assessment is limited to risk subjects. Low-risk subjects undergo the Fontan operation without pre-
only high-risk subjects who have been identified by screen- ceding catheterization, whereas high-risk subjects undergo catheterization
ing for the presence of clinical risk factors and after assess- before the operation for hemodynamic testing.
ment by noninvasive imaging. We hypothesized that (1) this
new diagnostic algorithm can successfully identify subjects mined by the subject’s cardiologist, who had access to all available data, in-
cluding results of cardiac catheterization; (3) PAP and VFP at cardiac
who are inoperable and (2) omission of invasive hemody-
catheterization; (4) interventional procedures performed at catheterization;
namic testing does not impair the ability to predict adverse and (5) for subjects who subsequently underwent a Fontan operation, details
postoperative outcomes in low-risk patients. We retrospec- of the operation (type of Fontan, creation of a fenestration, duration of car-
tively tested the efficacy of this algorithm in a series of diopulmonary bypass), postoperative course (duration of intensive care unit
patients who had undergone routine cardiac catheterization [ICU] stay, and duration of pleural drainage), and data on survival to 1 year
after the operation. An adverse postoperative outcome was defined as the
to determine suitability for a Fontan operation.
occurrence of one or more of the following events: (1) death within 1
year of the operation; (2) prolonged pleural drainage (>15 days); (3) pro-
METHODS longed ICU stay (>10 days); and (4) cardiac catheterization within a 3-
Proposed Diagnostic Algorithm month period after the operation.
To consider the possibility of eliminating routine preoperative cardiac
catheterization, one must develop an alternative diagnostic strategy that Statistical Analyses
CHD

will provide data for assessing suitability for the operation and for predicting
Statistical analyses were performed with commercially available statisti-
an adverse postoperative outcome. We developed such a diagnostic algo- cal software (STATA SE 9.0; Stata Corp, College Station, Tex). To evaluate
rithm (Figure 1) for assessment before a Fontan operation on the basis of the efficacy of the proposed diagnostic algorithm in identifying subjects who
factors that have been shown to be associated with an increased risk of being are inoperable, we calculated sensitivity, specificity, positive predictive value
inoperable or having adverse outcomes after a Fontan operation.9-15 In this
(PPV), negative predictive value (NPV), likelihood ratios, and post-test prob-
algorithm, data from history, echocardiography, and noninvasive angiogra- ability as previously described.17-19 We also generated 95% confidence in-
phy (using MRI or CT) are used (Table 1). Subjects with 1 or more risk fac- tervals (CI) for sensitivity, specificity, PPV, and NPV. To evaluate the
tors are classified as ‘‘high risk’’ and undergo conventional cardiac proposed diagnostic algorithm’s efficacy in predicting adverse postoperative
catheterization, whereas subjects without any risk factors are classified as
outcomes, we similarly calculated sensitivity, specificity, PPV, NPV, likeli-
‘‘low risk’’ and undergo a Fontan operation without cardiac catheterization. hood ratios, and post-test probabilities. To test the hypothesis that the omis-
Angiographic evaluation of the pulmonary vasculature and systemic venous sion of invasive hemodynamic testing does not impair the ability to predict
anatomy is important because abnormalities in these areas are not reliably adverse postoperative outcomes, we repeated the above analyses after includ-
detected by echocardiography. Hence, several angiographic criteria are in-
ing the presence of elevated PAP (>15 mm Hg) and/or VFP (>10 mm Hg) as
cluded in our algorithm. Owing to the retrospective nature of this study, additional risk factors in the proposed algorithm. Furthermore, to identify
noninvasive angiographic data from MRI or CT were not available. Hence, predictors of an adverse postoperative outcome as previously defined, we
data from x-ray angiography were used to identify angiographic risk factors.
used multivariable logistic regression. Variables included for analysis were
For the purpose of this study, it was assumed that MRI or CT imaging is age at operation, PAP, VFP, ventricular dysfunction, atrioventricular valve
equivalent to x-ray angiography, as shown previously by numerous inves- regurgitation, creation of a fenestration, presence of heterotaxy or a defined
tigators.2-6,16 genetic syndrome, cardiopulmonary bypass (CPB) time, and the type of Fon-
tan operation (lateral tunnel or extracardiac conduit). In addition, further
Evaluation of Algorithm analyses were performed to identify predictors of the following individual
To test its diagnostic efficacy, we applied the diagnostic algorithm retro- outcomes: duration of ICU stay, duration of pleural drainage, and death
spectively to 151 consecutive children with a functionally single ventricle within 1 year after operation. Data for duration of ICU stay and duration
who underwent cardiac catheterization after the age of 1 year at the Morgan of pleural drainage were skewed and underwent logarithmic transformation
Stanley Children’s Hospital of New York Presbyterian between 2000 and before analysis to satisfy normalcy requirements. Multivariable linear regres-
2005, presumably to assess suitability for a Fontan operation. The Institu- sion was used to identify the predictors of log-transformed duration of ICU
tional Review Board of Columbia University Medical Center gave permis- stay and log-transformed duration of pleural drainage, and multivariable lo-
sion for the analysis of existing clinical data. Data recorded for each subject gistic regression was used to identify predictors of death within 1 year after
included the following: (1) presence of risk factors (Table 1) on history, operation. For all multivariable modeling using either logistic or linear re-
echocardiography, and x-ray angiography performed at the time of referral gression, a ‘‘backward elimination’’ method was used for selection of vari-
for preoperative evaluation; (2) suitability for the Fontan operation as deter- ables and P<.05 was required for inclusion of a variable in the final model.

918 The Journal of Thoracic and Cardiovascular Surgery c October 2009

Prakash et al Congenital Heart Disease

TABLE 1. Characteristics of high-risk subjects TABLE 2. Subject characteristics

History Diagnosis N
Chronic respiratory or lung disorder (chest x-ray abnormality lasting>3
Hypoplastic left heart syndrome 53 (35%)
months/need for supplemental oxygen > 3 months)
Unbalanced atrioventricular canal defect 26 (17%)
Heterotaxy
Double-inlet left ventricle 18 (12%)
Genetic syndrome
Tricuspid atresia 17 (11%)
Echocardiography
Pulmonary atresia with intact ventricular septum 12 (8%)
 Moderate atrioventricular valve regurgitation (qualitative grading)
Others 25 (17%)
 Moderate ventricular dysfunction (qualitative grading)
 Moderate aortic insufficiency (qualitative grading)
Pulmonary vein stenosis (mean Doppler gradient > 3 mm Hg)
Angiography
mal PAP and/or VFP. Several subjects had multiple risk fac-
 Moderate pulmonary artery branch stenosis or discontinuity tors. No inoperable subject had an elevated PAP and/or VFP
(qualitative grading) without an associated risk factor on history, echocardiogra-
Pulmonary vein stenosis (qualitative assessment) phy, or angiography.
Systemic venous anomaly (excluding left superior vena cava)
Subjects Who Underwent a Fontan Operation
RESULTS At the time of data analysis, 135 of the 142 operable sub-
A total of 151 consecutive children (63% male) who un- jects had undergone the Fontan operation at a median age of
derwent cardiac catheterization for assessment before a Fon- 3.2 years (interquartile range 2.5–3.9 years).
tan operation between 2000 and 2005 at Children’s Hospital Preoperative characteristics. Of the 135 subjects who un-
of New York Presbyterian were included in this study. Sub- derwent a Fontan operation, 95 were in the low-risk group
ject diagnoses are shown in Table 2. Hypoplastic left heart and 40 were in the high-risk group. PAP and VFP were sim-
syndrome was the most common diagnosis, in approxi- ilar in the low-risk (11  3 and 9  2 mm Hg) and high-risk
mately one third of the subjects. A majority (147/151, groups (12  3 and 9  3 mm Hg; P > .5 for both). In the
97%) of subjects had undergone a superior cavopulmonary low-risk group, 24 (27%) children had an elevated PAP
anastomosis with either a bidirectional Glenn (n ¼ 117) or (>15 mm Hg, n ¼ 7) and/or VFP (>10 mm Hg, n ¼ 23),
hemi-Fontan (n ¼ 30) operation. According to the proposed whereas in the high-risk group, 22 (55%; P ¼ .0009 using
algorithm, 95 (63%) subjects had no risk factors (low-risk a 2-sided Student t test) had an elevated PAP (n ¼ 5) and/

CHD
group) and 56 (37%) had 1 or more risk factors (high-risk or VFP (n ¼ 22). However, no patient undergoing a Fontan
group). operation had a PAP greater than 23 mm Hg or a VFP greater
than 14 mm Hg.
Identification of Inoperable Subjects by the Proposed Operative details and early postoperative outcomes.
Algorithm (Figure 2) The Fontan operation was performed with a lateral tunnel
A total of 9 subjects were deemed inoperable for the Fon- in 99 (74%) and an extracardiac conduit in 35 (26%) sub-
tan operation by their cardiologist after undergoing cardiac jects. A fenestration was performed in 81 (60%) subjects.
catheterization. The proposed algorithm performed well in Within the first month after the operation, there were 2
screening for inoperability, and all of the 9 inoperable sub- deaths and 1 patient required heart transplantation. Median
jects were correctly classified in the high-risk group. How- duration of stay in the ICU was 6 days (interquartile range
ever, 47 eligible subjects were also classified as high risk. 4–8 days). Median duration of pleural drainage was 9 days
In identifying inoperable subjects, the algorithm’s sensitiv- (interquartile range 7–12 days). An adverse postoperative
ity (100%, 95% CI 66–100) and NPV (100%, 95% CI outcome as previously defined occurred in 26% (95% CI
97–100) were high whereas the specificity (67%, 95% CI 19–34%) of subjects.
59–75) and PPV (16%, 95% CI 8–28) were low. For a sub- Prediction of postoperative outcomes using the algo-
ject classified as high risk by the algorithm, the likelihood ra- rithm with and without additional hemodynamic data.
tio was 3.03 and the post-test probability of being inoperable As seen in Table 4, the proposed algorithm had modest success
was 19% (increased from a pre-test probability of 6%). On in predicting adverse postoperative outcomes. Compared with
the other hand, for a subject classified as low risk by the al- the entire study population, presence in the high-risk group in-
gorithm, the likelihood ratio was 0.015 (assuming a sensitiv- creased the probability of an adverse outcome whereas pres-
ity of 99% instead of the actual 100% to allow calculation) ence in the low-risk group decreased the probability of an
and post-test probability of being inoperable was 0.1% (de- adverse outcome. The occurrence of an adverse postoperative
creased from a pre-test probability of 6%). The characteris- outcome was significantly associated with a high-risk status
tics of the 9 subjects who were inoperable are described in (c2 ¼ 10.8; P ¼ .001; odds ratio 3.75, 95% CI 1.53–9.16).
Table 3. All inoperable subjects had risk factors on history, However, prediction of an adverse postoperative outcome
echocardiography, or angiography and 6 of the 9 had abnor- was not improved by including an elevated PAP and/or VFP

The Journal of Thoracic and Cardiovascular Surgery c Volume 138, Number 4 919
 Congenital Heart Disease Prakash et al

(P ¼ .002, coefficient 0.74, 95% CI 0.08–1.4). Longer du-

ration of pleural drainage (after logarithmic transformation)
was independently associated with a longer duration of CPB
(P ¼ .002, coefficient ¼ 0.003, 95% CI 0.001–0.006 for 1-
minute increase in duration of CPB), heterotaxy syndrome
(P ¼ .04, coefficient ¼ 0.36, 95% CI 0.01–0.7), and the ab-
sence of a fenestration (P ¼ .008, coefficient ¼0.29, 95%
CI 0.49 to 0.08). Notably, there were no significant asso-
ciations between adverse postoperative outcomes and preop-
erative hemodynamic measurements (PAP or VFP).
FIGURE 2. Predicting operative eligibility: Results of retrospective test- Survival to 1 year. Five subjects died within 1 year after
ing of the algorithm on 151 children presenting for evaluation before a Fon- the operation, all of whom were classified in the high-risk
tan operation. All 9 ineligible (inoperable) subjects were correctly classified group according to the algorithm. The characteristics of
as high risk. All low-risk subjects were deemed eligible. these subjects and the clinical events leading to death in
each case are shown in Table 5. Four of the 5 had an unbal-
as an additional risk factor, and with this modified algorithm, anced atrioventricular canal defect. All 5 subjects had prom-
an adverse outcome was not significantly associated with inent and often multiple noninvasive risk factors. Three of
high-risk or low-risk status (c2 ¼ 3.3; P ¼ .07). the 5 subjects had heterotaxy, and 3 had severe atrioventric-
Predictors of an adverse outcome. On multivariable mod- ular valve regurgitation. Only 1 of the 5 had abnormal pre-
eling, an adverse postoperative outcome, as previously de- operative hemodynamics. The preoperative PAP was
fined, was independently associated with a defined genetic within normal limits in all 5 children and the preoperative
syndrome (P ¼ .02, odds ratio 17.7, 95% CI 1.5–208), het- VFP was elevated in 1 of the 5 children. On multivariable
erotaxy syndrome (P ¼ .02, odds ratio 4.7, 95% CI 1.3– modeling, the presence of greater than moderate atrioven-
17.1), a longer duration of CPB (P ¼ .008, odds ratio tricular valve regurgitation (P ¼ .003, odds ratio 46.3,
1.01, 95% CI 1.003–1.02 for 1-minute increase in duration 95% CI 3.6–593) and heterotaxy syndrome (P ¼ .049,
of CPB), and with the absence of a fenestration (P ¼ .01, odds ratio 17.3, 95% CI 1.01–172) were independently as-
odds ratio 0.29, 95% CI 0.11–0.74). On further analysis to sociated with risk of mortality within 1 year after the Fontan
identify predictors of individual adverse outcomes, a longer operation. PAP and VFP were not associated with risk of
CHD

ICU stay (after logarithmic transformation) was associated mortality within 1 year.
with a longer duration of CPB (P < .0001, coefficient ¼
0.006, 95% CI 0.003–0.009 for 1-minute increase in dura- Interventional Procedures Performed
tion of CPB), heterotaxy syndrome (P ¼ .01, coefficient ¼ Eighty-five transcatheter interventions were performed in
0.5, 95% CI 0.11–0.88), and a defined genetic syndrome 73 (48%) of 135 subjects who subsequently underwent the

TABLE 3. Subjects deemed inoperable for a Fontan operation

PAP VFP Angiographic
No Diagnosis (mm Hg) (mm Hg) Echo risk factors risk factors Clinical risk factors
1 Unbalanced AVCD 20 12 Severe ventricular dysfunction, — —
prosthetic AV valve stenosis
2 HLHS 19 18 — — Chronic lung disease, Noonan
syndrome
3 Unbalanced AVCD 25 18 Severe ventricular dysfunction, Left pulmonary Chronic lung disease,
moderate AVVR vein stenosis tracheostomy
4 Mitral atresia, DORV 27 21 Severe AVVR, decreased Hypoplastic RPA Chronic lung disease,
ventricular function tracheostomy
5 Mitral atresia, DORV, scimitar 24 13 — Scimitar vein Right lung hypoplasia
syndrome, TAPVC
6 HLHS 10 11 — Discontinuous LPA —
7 HLHS 14 8 — LPA occluded by thrombus —
8 HLHS 9 8 — Occluded SVC Chronic lung disease
9 HLHS 10 9 — Severe stenosis of head —
and neck vessels
AV, Atrioventricular; AVCD, atrioventricular canal defect; AVVR, atrioventricular valve regurgitation; DORV, double-outlet right ventricle; Echo, echocardiographic; HLHS,
hypoplastic left heart syndrome; LPA, left pulmonary artery; PAP, pulmonary artery pressure; RPA, right pulmonary artery; SVC, superior vena cava; TAPVC, totally anomalous
pulmonary venous connection; VFP, ventricular filling pressure.

920 The Journal of Thoracic and Cardiovascular Surgery c October 2009

Prakash et al Congenital Heart Disease

TABLE 4. Prediction of adverse postoperative outcomes using the a screening tool to identify subjects who are inoperable. In
algorithm with and without hemodynamic data addition, adverse postoperative outcomes, including death
Proposed Algorithm including within 1 year after the operation, were not associated with
algorithm PAP, VFP preoperative hemodynamic parameters; hence if such an al-
Sensitivity (95% CI) 51% (34%–69%) 62% (45%–79%) gorithm were used to screen patients, the omission of routine
Specificity (95% CI) 78% (68%–86%) 55% (45%–65%) invasive hemodynamic measurements would not impair the
PPV (95% CI) 45% (29%–62%) 29% (19%–40%) ability to predict adverse early postoperative outcomes.
NPV (95% CI) 82% (73%–89%) 81% (70%–89%)
High-risk status
Identifying Inoperable Subjects
Likelihood ratio 2.32 1.37
Pre-test probability 26% 26%
In identifying subjects who are inoperable, the proposed
Post-test probability 45% 25.4% algorithm is highly sensitive (100%) but has a low specific-
Low-risk status ity (67%). To maintain a high level of sensitivity, we chose
Likelihood ratio 0.63 0.69 to use a low threshold for classifying a subject as high risk
Pre-test probability 26% 26% (1 risk factor), and this likely resulted in a low specificity.
Post-test probability 18% 20% However, this low specificity is acceptable because this al-
CI, Confidence interval; NPV, negative predictive value; PAP, pulmonary artery pres- gorithm is being proposed only as a screening tool to identify
sure; PPV, positive predictive value; VFP, ventricular filling pressure.
high-risk patients who would then undergo invasive testing.
Notably, the inoperable subjects had relatively prominent
Fontan operation. A majority of these interventions included and often multiple risk factors and all were classified as
embolization of aortopulmonary collateral arteries (n ¼ 64, high risk according to the proposed algorithm. Furthermore,
75%). Other procedures performed included embolization inoperable status was rare (6%), likely resulting from a con-
of venovenous collaterals (n ¼ 15), stenting of pulmonary sistent policy of staged palliation and the rarity of pulmonary
artery stenosis (n ¼ 3), angioplasty for recurrent coarctation vascular disease. If the proposed algorithm were used to
of the aorta (n ¼ 2), and embolization of a right ventricular screen subjects who require cardiac catheterization, all the
sinusoid (n ¼ 1). Interventions were performed in 23 (58%) inoperable subjects would be correctly classified as high
of 40 subjects in the high-risk group and in 50 (56%) of 90 risk and would undergo cardiac catheterization. It should
subjects in the low-risk group. No relationship was seen be- be noted that these results are based on the assumption
tween the performance of an intervention and an adverse that MRI or CT imaging is equivalent to conventional

CHD
postoperative outcome. x-ray angiography.

DISCUSSION Predicting Adverse Postoperative Outcomes

In this study, we proposed and evaluated a diagnostic al- In addition to assessing eligibility, prediction of adverse
gorithm that does not use routine cardiac catheterization for postoperative outcomes is another goal of preoperative test-
low-risk subjects before the Fontan operation. In a retrospec- ing. In this study, we found that the proposed diagnostic al-
tive evaluation, the proposed algorithm was successful as gorithm allowed modest prediction of adverse postoperative

TABLE 5. Mortality within 1 year

Preoperative hemodynamics (mm Hg)
Timing of
Age at Nonhemodynamic death (days Clinical events related
No. Fontan (y) Diagnosis risk factors PAP VFP after Fontan) to mortality
1 1.7 Unbalanced Trisomy 21, 12 10 89 Prolonged pleural effusion, thrombus
AVCD severe AVVR in Fontan pathway, sepsis
2 3.8 Unbalanced Heterotaxy 12 9 90 Prolonged pleural and pericardial
AVCD effusion, thrombus in Fontan
pathway
3 7 Unbalanced Heterotaxy 13 9 25 Low cardiac output, renal failure, JET
AVCD
4 2 Unbalanced Heterotaxy, 10 7 22 Low cardiac output, respiratory
AVCD severe AVVR failure.
5 2 HLHS Severe AVVR, 9 5 80 Prolonged pleural effusion
occluded LPA
AVCD, Atrioventricular canal defect; AVVR, atrioventricular valve regurgitation; HLHS, hypoplastic left heart syndrome; JET, junctional ectopic tachycardia; LPA, left pulmonary
artery; PAP, pulmonary artery pressure; VFP, ventricular filling pressure.

The Journal of Thoracic and Cardiovascular Surgery c Volume 138, Number 4 921
 Congenital Heart Disease Prakash et al

events but, more importantly, the addition of invasive hemo- by several investigators in children with complex congenital
dynamic parameters to this algorithm did not improve this heart disease, including those with a functionally single ven-
prediction. Hence, if the proposed algorithm was used to tricle.2-6,16 A recently published randomized trial found that
screen subjects who require cardiac catheterization, the cardiac MRI is a safe, effective, and less costly alternative to
omission of routine invasive hemodynamic testing in the routine catheterization in the evaluation of selected patients
low-risk subjects would not impair the ability to predict before a bidirectional Glenn operation.16 Third, follow-up
adverse postoperative outcomes. Previous studies have beyond 1 year after the operation was not available for this
demonstrated that invasive hemodynamic parameters can study and hence long-term outcomes were not evaluated. Fi-
be used to predict postoperative outcomes.9-12 However, in nally, the effect of interventional procedures such as the coil-
our study, we did not find a significant relationship between ing of aortopulmonary collateral arteries in low-risk patients,
preoperative invasive hemodynamic parameters and postop- while controversial, was not assessed in this study. Although
erative outcomes. In fact, we found that adverse early post- the long-term benefit of these interventions in low-risk sub-
operative outcomes were associated with nonhemodynamic jects has not been established, it is not known whether elim-
factors such as the presence of heterotaxy or genetic syn- inating cardiac catheterization and interventional procedures
dromes and the duration of CPB. Hemodynamic parameters would affect outcome.
also were not predictive of mortality within 1 year. The only
predictors of mortality were the presence of greater than CONCLUSION
moderate atrioventricular valve regurgitation and heterotaxy In summary, we have shown that a noninvasive diagnostic
syndrome. The lack of association between preoperative algorithm such as the one proposed here could potentially be
hemodynamics and adverse postoperative outcomes may used as a screening tool to identify subjects who are inoper-
be related to a consistent policy of timely staged palliation able for a Fontan procedure. Second, our results indicate that
in the current surgical era and the resultant rarity of signifi- preoperative hemodynamic parameters are not associated
cantly elevated PAP or VFP. with early postoperative outcomes. Adverse postoperative
outcomes were associated with risk factors identified on
Interventional Procedures medical history, such as heterotaxy and genetic syndromes.
In addition to assessment of eligibility and prediction of Hence, the omission of routine cardiac catheterization in
postoperative outcomes, cardiac catheterization also pro- low-risk subjects would not impair identification of inoper-
vides an opportunity for interventional procedures before able subjects or those at risk for adverse early postoperative
CHD

the Fontan operation. In a significant proportion of our study outcomes. The use of the proposed algorithm could help
population (48%), an intervention was performed during avoid the risk of cardiac catheterization and reduce the
cardiac catheterization, which most frequently (75%) con- cost of evaluation in low-risk subjects while allowing the
sisted of embolization of aortopulmonary collateral vessels. opportunity for invasive hemodynamic testing and/or inter-
The clinical utility of routine embolization of small aortopul- ventional procedures in high-risk subjects. Owing to the
monary collateral vessels is not known. Data both support- retrospective nature of this study and because MRI or CT
ing and criticizing this practice have been presented by imaging was not performed in our subjects, the results
various investigators; hence, this practice is of unproven should be considered preliminary until further prospective
benefit.20-25 With the proposed algorithm, subjects with validation of such a strategy is performed. It should be noted
significant aortopulmonary collaterals would be identifiable that high-quality MRI or CT imaging is necessary if such
using noninvasive angiography and subsequent coil embo- a strategy is used. Finally, further investigation is warranted
lization could be performed if deemed necessary. Further to identify other predictors of adverse outcomes after the
investigation is needed to determine whether embolization Fontan operation and to evaluate whether routine emboliza-
of small aortopulmonary collateral vessels improves clinical tion of small aortopulmonary collateral arteries improves
outcomes. outcomes.

Study Limitations References

This study had several important limitations. First, the 1. Fontan F, Baudet E. Surgical repair of tricuspid atresia. Thorax. 1971;26:240-8.
2. Greil GF, Powell AJ, Gildein HP, Geva T. Gadolinium-enhanced three-dimen-
study is limited by its retrospective design. Hence, although sional magnetic resonance angiography of pulmonary and systemic venous anom-
the algorithm would omit cardiac catheterization for low- alies. J Am Coll Cardiol. 2002;39:335-41.
risk subjects, all the patients in this study underwent cardiac 3. Geva T, Greil GF, Marshall AC, Landzberg M, Powell AJ. Gadolinium-enhanced
3-dimensional magnetic resonance angiography of pulmonary blood supply in pa-
catheterization. Second, although data obtained with nonin- tients with complex pulmonary stenosis or atresia: comparison with x-ray angiog-
vasive angiographic techniques such as MRI or CT are raphy. Circulation. 2002;106:473-8.
a component of the proposed diagnostic algorithm, the accu- 4. Lee T, Tsai IC, Fu YC, Jan SL, Wang CC, Chang Y, et al. Using multidetector-
row CT in neonates with complex congenital heart disease to replace diagnostic
racy of MRI or CT was not directly compared with x-ray an- cardiac catheterization for anatomical investigation: initial experiences in techni-
giography, although this has been demonstrated previously cal and clinical feasibility. Pediatr Radiol. 2006;36:1273-82.

922 The Journal of Thoracic and Cardiovascular Surgery c October 2009

Prakash et al Congenital Heart Disease

5. Greil GF, Schoebinger M, Kuettner A, Schaefer JF, Dammann F, Claussen CD, drome: 30 years of experience in 142 patients. J Am Coll Cardiol. 2006;48:
et al. Imaging of aortopulmonary collateral arteries with high-resolution multide- 2301-5.
tector CT. Pediatr Radiol. 2006;36:502-9. 15. Kim SJ, Kim WH, Lim HG, Lee CH, Lee JY. Improving results of the Fontan pro-
6. Prakash A, Torres AJ, Printz BF, Prince MR, Nielsen JC. Usefulness of magnetic cedure in patients with heterotaxy syndrome. Ann Thorac Surg. 2006;82:1245-51.
resonance angiography in the evaluation of complex congenital heart disease in 16. Brown DW, Gauvreau K, Powell AJ, Lang P, Colan SD, Del Nido PJ, et al.
newborns and infants. Am J Cardiol. 2007;100:715-21. Cardiac magnetic resonance versus routine cardiac catheterization before bidirec-
7. Garofalo CA, Cabreriza SE, Quinn TA, Weinberg AD, Printz BF, Hsu DT, et al. tional Glenn anastomosis in infants with functional single ventricle: a prospective
Ventricular diastolic stiffness predicts perioperative morbidity and duration of randomized trial. Circulation. 2007;116:2718-25.
pleural effusions after the Fontan operation. Circulation. 2006;114(1 Suppl). 17. Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios. BMJ. 2004;329:168-9.
I56-61. 18. Akobeng AK. Understanding diagnostic tests 1: sensitivity, specificity and predic-
8. Ro PS, Rychik J, Cohen MS, Mahle WT, Rome JJ. Diagnostic assessment before tive values. Acta Paediatr. 2007;96:338-41.
Fontan operation in patients with bidirectional cavopulmonary anastomosis: are 19. Akobeng AK. Understanding diagnostic tests 2: likelihood ratios, pre- and post-
noninvasive methods sufficient? J Am Coll Cardiol. 2004;44:184-7. test probabilities and their use in clinical practice. Acta Paediatr. 2007;96:487-91.
9. Gaynor JW, Bridges ND, Cohen MI, Mahle WT, Decampli WM, Steven JM, et al. 20. Bradley SM. Management of aortopulmonary collateral arteries in Fontan pa-
Predictors of outcome after the Fontan operation: is hypoplastic left heart syn- tients: routine occlusion is not warranted. Semin Thorac Cardiovasc Surg Pediatr
drome still a risk factor? J Thorac Cardiovasc Surg. 2002;123:237-45. Card Surg Annu. 2002;5:55-67.
10. Cetta F, Feldt RH, O’Leary PW, Mair DD, Warnes CA, Driscoll DJ, et al. Im- 21. Bradley SM, McCall MM, Sistino JJ, Radtke WA. Aortopulmonary collateral
proved early morbidity and mortality after Fontan operation: the Mayo Clinic ex- flow in the Fontan patient: does it matter? Ann Thorac Surg. 2001;72:408-15.
perience, 1987 to 1992. J Am Coll Cardiol. 1996;28:480-6. 22. Ichikawa H, Yagihara T, Kishimoto H, Isobe F, Yamamoto F, Nishigaki K, et al.
11. Annecchino FP, Brunelli F, Borghi A, Abbruzzese P, Merlo M, Parenzan L. Fon- Extent of aortopulmonary collateral blood flow as a risk factor for Fontan opera-
tan repair for tricuspid atresia: experience with 50 consecutive patients. Ann tions. Ann Thorac Surg. 1995;59:433-7.
Thorac Surg. 1988;45:430-6. 23. Kanter KR, Vincent RN. Management of aortopulmonary collateral arteries in
12. Culbertson CB, George BL, Day RW, Laks H, Williams RG. Factors influencing Fontan patients: occlusion improves clinical outcome. Semin Thorac Cardiovasc
survival of patients with heterotaxy syndrome undergoing the Fontan procedure. Surg Pediatr Card Surg Annu. 2002;5:48-54.
J Am Coll Cardiol. 1992;20:678-84. 24. Spicer RL, Uzark KC, Moore JW, Mainwaring RD, Lamberti JJ. Aortopulmonary
13. Atz AM, Cohen MS, Sleeper LA, McCrindle BW, Lu M, Prakash A, et al. Func- collateral vessels and prolonged pleural effusions after modified Fontan proce-
tional state of patients with heterotaxy syndrome following the Fontan operation. dures. Am Heart J. 1996;131:1164-8.
Cardiol Young. 2007;17(Suppl. 2):44-53. 25. Triedman JK, Bridges ND, Mayer JE Jr, Lock JE. Prevalence and risk factors for
14. Bartz PJ, Driscoll DJ, Dearani JA, Puga FJ, Danielson GK, O’Leary PW, et al. aortopulmonary collateral vessels after Fontan and bidirectional Glenn proce-
Early and late results of the modified Fontan operation for heterotaxy syn- dures. J Am Coll Cardiol. 1993;22:207-15.

CHD

The Journal of Thoracic and Cardiovascular Surgery c Volume 138, Number 4 923