Author + information
- Received March 12, 2012
- Revision received June 19, 2012
- Accepted June 26, 2012
- Published online September 18, 2012.
- Mark A. Fogel, MD⁎,†,⁎ (, )
- Thomas W. Pawlowski, BA⁎,
- Kevin K. Whitehead, MD⁎,†,
- Matthew A. Harris, MD⁎,†,
- Marc S. Keller, MD†,
- Andrew C. Glatz, MD⁎,
- Winnie Zhu, PhD†,
- David Shore, BS⁎,
- Laura K. Diaz, MD‡ and
- Jonathan J. Rome, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Mark A. Fogel, The Children's Hospital of Philadelphia, Division of Cardiology, 34th Street and Civic Center Boulevard, Philadelphia, Pennsylvania 19104
Objectives This study investigated whether cardiac magnetic resonance (CMR) and echocardiography (echo) can replace catheterization (cath) for routine evaluation prior to Fontan and under what circumstances CMR and cath are used together.
Background Routine cath prior to Fontan has been utilized for years; noninvasive methods, however, may be sufficient.
Methods This study reviews clinical data in 119 consecutive patients investigating 3 groups: those who underwent CMR alone (MR; n = 41), cath alone (C; n = 41), or both cath and CMR (C+M; n = 37) prior to Fontan.
Results No clinically significant differences were noted in patient characteristics, hemodynamics, or clinical status prior to or after surgery between the C and MR groups. CMR added information in 82%. There were no discrepant findings between CMR and cath data in the C+M group. Diagnostic success was ≥95% in all groups. Of those undergoing Fontan completion, the C+M group had similar outcomes to C and MR; C and CMR were utilized in combination to assess aortopulmonary collaterals or the need for an intervention or evaluate its success. Echo could not delineate pulmonary arterial anatomy in 46% to 53% of patients. The C+M and C groups were exposed to 6.8 ± 4.1 mSv of radiation.
Conclusions Single ventricle patients not requiring an intervention can undergo successful Fontan completion with CMR and echo alone with similar short-term outcomes to C, which was used as a control, preventing an invasive test and exposure to radiation. CMR can add information in a significant number of patients. Cath and CMR are utilized together for interventions and assessment of aortopulmonary collaterals.
- bidirectional Glenn
- cardiac catheterization
- cardiac magnetic resonance
- single ventricle
Improved outcomes for functional single ventricle patients undergoing Fontan completion (1,2) are due in part to preoperative imaging that identifies multiple parameters correlated with postoperative outcomes (3–7). Routine evaluation has included echocardiography (echo) and cardiac catheterization (cath); however, both have shortcomings. Poor acoustic windows may limit visualization of cardiovascular structures (8) and the lack of quantitative single ventricle functional and flow assessment are limitations of echo. Cath is inherently an invasive, 2-dimensional modality associated with a small but significant morbidity and utilizes ionizing radiation that may increase cancer risk (9–11). In select groups of single ventricle patients, noninvasive imaging modalities may suffice (8,12–14). Ro et al. developed a set of criteria to determine the need for cath prior to Fontan based mainly on echo data, hemoglobin, and pulse oximetry (8). Their criteria were developed in order to identify all patients who died or required additional cath interventions. The criteria in that study had a negative predictive valve of 93%; the positive predictive value (52%) and sensitivity (25%), however, were poor, primarily due to unsuccessful imaging of the pulmonary arteries (PAs) resulting from poor acoustic windows. They suggested that cardiac magnetic resonance (CMR) may have helped.
CMR, successfully used for years to assess single ventricle patients (15), is an excellent noninvasive tool for assessing cardiovascular anatomy (16,17), ventricular function (18–20), and blood flow. CMR obtains data unavailable to other modalities, such as assessment of aortopulmonary collateral flow which imposes a volume load on the ventricle, even in the hemi-Fontan/bidirectional Glenn stage (21).
We hypothesized that in a group of routine pre-Fontan patients, CMR and echo alone provide sufficient pre-operative evaluation to allow successful Fontan completion. The purpose of this study is to test this hypothesis by comparing short-term outcomes and accuracy of diagnostic testing in 3 groups: patients undergoing CMR and echo alone (MR); cath and echo (C); and CMR, cath, and echo (C+M). The MR group was the test group while the C group was used as a control. We also sought to determine in what instances all 3 modalities are used in pre-operative evaluation (C+M group). Finally, we tested the hypothesis that the addition of CMR to echocardiographic imaging in the Ro criteria would successfully identify patients who may not require cardiac catheterization prior to Fontan completion. Preoperative demographics and hemodynamics were analyzed in each group to demonstrate comparability for operative risk.
A search of CMR and cath databases was performed for all single ventricle patients prior to Fontan completion between January 2005 and October 2009. Three groups were identified: 1) patients with C+M; 2) patients with only CMR; and 3) patients with only C. All patients who underwent CMR were enrolled while a set of consecutive C patients picked at random during the study period were identified equal in number to the CMR group for comparability. For C+M, studies were performed within a year of each other; no clinical changes were noted in between studies.
CMR was performed on a Siemens 1.5-T Sonata or Avanto MRI system (Siemens Medical Solutions, Malvern, Pennsylvania) with data analyzed using the standard Siemens' analysis package. The CMR protocol is listed in Table 1.
Routine pre-Fontan cath was performed as per protocol. Hemodynamic measurements (pressures and oxygen saturation) were obtained in the atria, ventricle(s), PAs, and aorta followed by angiography of the ventricle, aortic arch, and PAs. Interventions were performed as determined by the physician performing the catheterization.
A standard pediatric echo was performed as delineated in the guidelines of Lai et al. (22). Two-dimensional, Doppler, and color flow mapping of the relevant cardiovascular structures included evaluation of single ventricle anatomy and function, atrioventricular valve regurgitation (AVVR), PA and superior vena cava (SVC) to PA anastomosis, ventricular outflow tract, and aorta.
All CMR, cath, and echo data were reviewed. CMR data included ventricular volumes and mass, ventricular performance parameters, and anatomic data with quantitative measures of cardiovascular structures. Cath data included hemodynamic measurements and anatomic findings with quantitative measures of cardiovascular structures; calculated data (systemic and pulmonary blood flow and pulmonary vascular resistance [PVR]) were also noted. Anatomic findings for both CMR and cath included assessment of SVC to right pulmonary artery (RPA) anastomosis, PAs, pulmonary venous anatomy, ventricular outflow tract obstruction, aortic assessment including coarctation, presence of a left SVC, decompressing vessel, or other venous anomalies. Self-reported serious complications from CMR and cath (e.g., days in the intensive care unit [ICU]) were recorded. Echos were examined to determine if each provided a comprehensive evaluation, with a primary criterion of whether the PAs were visualized. Surgical and hospital reports were evaluated for type of fenestrated Fontan (lateral tunnel or extracardiac), bypass time, circulatory arrest time, surgical findings, complications, ICU days, and the length of hospital stay. Reviews of CMR, cath, surgical reports, and clinical outcome were all blinded to the other's test results. Ro criteria were applied to these patients, stating that any of the following findings indicated a need for cath: 1) left pulmonary artery (LPA) not visualized or stenotic; 2) significant AVVR; 3) significantly diminished ventricular contractility; 4) aortic coarctation; 5) restrictive atrial septal defect; or 6) evidence of a decompressing vessel. For this study, by CMR, a stenotic branch PA was defined as <25% of the total cross-sectional area of the branch PA (7); significant AVVR was defined as >20% regurgitant fraction (moderate regurgitation); diminished ventricular contractility was defined as ejection fraction <45%; aortic coarctation was defined as narrowing with a gradient; and restrictive atrial septal defect was defined as a small defect with turbulence noted. By echo, these were all assessed qualitatively.
Effective radiation dose
A total effective radiation dose (mSv) for each cath was estimated based on a conversion of the dose-area product (μGy·m2) using the Monte Carlo method. A straight frontal projection angle was assumed. Using data collected from our current system with available fluoroscopy time and measured dose-area product for each case, a series of best-fitting lines from linear regression was generated for a variety of different weight categories. Using the equation for the best-fitting line for the 5 to 15 kg weight category, we converted fluoroscopy time to dose-area product for the older cases, then converted the dose-area product to mSv dose.
Descriptive statistics were used, recorded as mean ± SD. Pairwise comparisons were made using an unpaired Student t test. An analysis of variance was used to compare all 3 groups for surgical and clinical outcomes. Diagnostic success rates were calculated for each group by comparing preoperative diagnoses with surgical findings. In C+M, diagnostic success rate between CMR and cath was also calculated. In C+M, only patients who had a completed Fontan within the study date range were included in the statistical analysis for outcomes. To compare categorical data (intervention in each cath group), Fisher exact and chi-square tests were used where applicable. A p value ≤0.05 was considered significant.
The study comprised 119 pre-Fontan patients. Of those, 78 consecutive patients were referred to CMR, 37 in C+M and 41 in MR. Of 246 patients evaluated by cath in this time frame, 41 consecutive patients were in C. All patients in the C and MR groups completed the Fontan, however, in the C+M group, 6 patients had not completed the Fontan (2 died unrelated to cath or CMR, 1 underwent heart transplantation, and 3 are awaiting completion). Those who did not undergo Fontan in C+M are not included in outcomes data (as, for example, length of stay after a Fontan would not apply) or hemodynamics (which is meant to demonstrate comparability prior to Fontan as it relates to quantitative outcome data) but are included for comparison between cath and CMR findings for accuracy of CMR (validating CMR vs. cath) as well as disposition of patients in C+M group (determining when CMR and cath can be utilized together in patient management). Both patients who died had chronic respiratory/lung disorders and the 1 who underwent heart transplantation had moderate ventricular dysfunction (end-diastolic volume of 187 cc/m2 and an ejection fraction of 33% by CMR). Patient demographics are described in Table 2. Patients fell into each group depending on whether the referring cardiologist felt they required an intervention at cath or had a practitioner bias toward either cath or CMR prior to Fontan (comparability between all 3 groups is discussed subsequently). The number of patients in each group are listed in Table 3.
PAs were visualized by echo in 51% of patients in the C+M group, 46% in the MR group, and 54% in the C group (p = 0.73).
The addition of CMR added significant clinical information in 64 (82%) of the patients who underwent that test (Fig. 1). Echo identified 7 patients with a normal LPA, later discovered by CMR to be hypoplastic/stenotic; in 8 others, the LPA, described as hypoplastic/stenotic by echo, was subsequently found to be normal on CMR. One patient described by echo with no AVVR had significant AVVR on CMR, whereas 1 patient had AVVR described as moderate by echo but quantitatively had minimal AVVR on CMR. For structures not visualized on echo, CMR identified 8 patients with LPA hypoplasia/stenosis, 9 patients with normal LPA, and 13 with some form of decompressing vessels. The addition of CMR enabled characterizing the LPA and decompressing vessels in 5 with incomplete/absent echo studies, quantification of collateral flow in 13, RPA stenosis or hypoplasia in 3, and delineation of pulmonary venous anatomy in 2. Five patients had 2 findings and 1 patient had 3 findings (70 findings in 64 patients). To summarize, the discrepancies between echo and CMR were largely extracardiac vascular anatomy (PAs, pulmonary veins, decompressing vessels and collaterals) with 2 differences in assessment of AVVR regurgitation.
Only 1 complication was noted in the MR group (endotracheal tube becoming obstructed because of secretions).
Disposition of C+M Patients
Of the 37 patients, 4 were referred for intervention at cath prior to CMR; 2 underwent an intervention and CMR was performed to evaluate the result. In the 2 who did not undergo intervention, cath was truncated and CMR was utilized to perform a full evaluation. Thirty-three underwent CMR prior to cath, and 11 eventually had an intervention. Of the remaining patients, 20 met Ro criteria for cath (12 by CMR and echo and 8 by CMR alone); 2 met Ro criteria to avoid cath by CMR alone but proceeded to cath despite the negative CMR results because of practitioner bias for cath prior to Fontan (Table 3, Fig. 2).
C Only Patients
Of the 41 patients, 6 (15%) underwent an intervention (2 identified by echo prior to cath, 3 not identified by echo, and 1 ablation). The other 35 patients underwent “routine” cath prior to Fontan because of practitioner bias for cath prior to Fontan.
In C+M, no complications were noted, while in C, 3 complications were present (2 patients with complete heart block (1 required chest compressions and atropine) and 1 with inadvertent entry into the right carotid artery). The number of days in the ICU after cath are shown in Table 4 (range 0 to 1 days in C and 0 to 2 days in C+M).
In the C+M and C groups, radiation dose estimate was 6.8 ± 4.1 mSv with a range of 2.1 to 27.9 mSv.
The Ro criteria and predictive values
Of the 78 CMRs performed, 47 Ro criteria were changed in 43 (55%) patients; 3 patients had multiple criteria changed. Of the 43 patients whose Ro criteria changed with CMR, 31 (72%) avoided catheterization, usually as a result of delineation of pulmonary arterial anatomy or decompressing vessels (in 30 of 31 patients). Twenty-three percent (n = 10) were identified with indications for catheterization that had not been identified by echo (usually the presence of a stenotic pulmonary artery). Two patients (5%) went to cath despite the negative Ro criteria because of physician referral bias for cath. In addition, CMR added information that did not change the Ro criteria in 21 (27%) additional patients (see Noninvasive Imaging subsection and Fig. 2). In summary, CMR either changed Ro criteria or added additional information in 64 (82%) of patients.
By utilizing the patients in the C+M group, the positive and negative predictive values of echo and CMR including the Ro criteria to determine findings or interventions at cath were each 100%. The C group alone was used to determine these parameters for the sole use of echo prior to cath utilizing the Ro criteria; the positive and negative predictive values were 25% and 85%, respectively.
Comparison Between Groups in Pre-Operative Evaluation
To assess pre-operative clinical status in each group and demonstrate comparability of risk for Fontan completion, cath data from C+M and C are shown in Table 4 and CMR data from C+M and MR are shown in Table 5. An example of CMR imaging in a pre-Fontan patient is shown in Figure 1. No differences were noted in qualitative ventricular shortening or AVVR on echo between all 3 groups.
In comparing hemodynamics in cath patients, no differences were noted between C+M and C in age at cath, body surface area, interventions, pressures in the SVC, PAs and the single ventricle, pulmonary and systemic vascular resistance, hemoglobin, systemic arterial O2 saturation, and ICU days. When the number of patients with measured RPA and LPA pressures were divided into groups with ≤15 or >16 mm Hg or divided into those with ≤12, 13 to 16, or >17 mm Hg, no significant differences between C+M and C were noted. Patients in C, however, were significantly closer to Fontan completion than the C+M group at the time of cath. Additionally, the C+M group had statistically lower systemic (3.6 vs. 4.1 l/min/m2) and pulmonary flows (2.3 vs. 2.6 l/min/m2) compared with C. In patients who underwent cath, PVR ranged from 1.54 ± 0.52 Wood units to 1.63 ± 0.61 Woods units and ventricular end-diastolic pressure ranged from 6.08 ± 4.32 mm Hg to 6.18 ± 3.70 mm Hg. These parameters were similar to the 3 patients in the C+M group who are awaiting Fontan completion (for scheduling reasons).
Similarly, no differences between C+M and MR groups were noted with regard to age at CMR, body surface area and measures of ventricular performance (ejection fraction, end-diastolic volume, and cardiac index). Similarly to the cath comparison, patients in MR were closer to Fontan completion than those in the C+M group.
Ten decompressing vessels/left SVC were noted in the C+M group while 4 each were noted in the C and MR groups (p = 0.052).
Short-term surgical outcome and hospital course are shown in Table 6 and demonstrates no significant differences in any clinical outcome parameters between MR, C, and C+M. All patients who underwent Fontan completion survived. All patients who had pleural effusions underwent hospital stays of >20 days.
Diagnostic success and comparison between cath and CMR
In the 37 C+M patients, no discrepant anatomic findings were noted between CMR and cath. No difference at surgery was noted between CMR or cath with the exception of LPA hypoplasia: 1 patient in C+M (96%), 2 in CMR only (95%), and 1 in cath only (98%) had the preoperative finding of LPA hypoplasia underneath the neoaortic arch and no PA arterioplasty was performed as is the surgeon's usual practice when LPA hypoplasia is identified (p = 0.96 between groups). If the surgeon did not perform a PA arterioplasty, the presumption is that he did not feel the LPA was hypoplastic.
Linear measurements were compared between CMR and cath in the C+M group in the SVC, PAs, and aorta. Table 7 and Figure 3 list the raw measurements and the Bland-Altman plot (all structures included), respectively. No significant differences between CMR and cath were noted with the raw data; the Bland-Altman plot demonstrated a narrow band between 2 standard deviations of approximately ± 1 mm. Interobserver variability ranged from 2.3% to 5.6% depending on the structure.
This study demonstrated that single ventricle patients not requiring an intervention can undergo successful Fontan with CMR and echo evaluation alone with similar short-term outcomes to patients who underwent C as a control. This study demonstrated comparability of preoperative demographics and hemodynamics in each group which allowed for comparison between preoperative evaluation strategies showing comparable preoperative risk. Differences in ventricular morphology were not statistically significant; recent data has demonstrated that morphology has little impact on short-term outcome (23,24). The positive and negative predictive values of echo and CMR to determine findings or interventions at cath were each 100%; when CMR is not used, positive and negative predictive values are lower, especially the negative predictive value (our data in the C group using echo alone is very similar to the original Ro criteria findings). Cath and CMR are usually used in combination to assess aortopulmonary collaterals or to assess the need for or evaluate the success of an intervention. CMR assessment of cardiovascular structures was accurate when compared with cath and surgical observation; the diagnostic success rate was ≥95% when compared to surgical observation with no quantitative difference between CMR and cath with regard to linear measurements or structures visualized. The addition of CMR on a routine basis changed the Ro criteria in 55% avoiding the need for cath and added additional information in another 27%. By dividing the study population into 3 different groups, this study covers all 3 combinations of CMR and cath as preoperative investigations prior to Fontan.
There are no prior studies investigating the use of routine CMR for preoperative evaluation prior to Fontan. With the percentage of complete echos in our study ranging from 46% to 53% across groups, echo alone appears insufficient. Cath is a common supplement for preoperative evaluation and provides the benefit of immediate intervention if required. CMR, however, provides an equally effective and safer alternative for diagnosis.
Previous studies have investigated noninvasive preoperative evaluations throughout the stages of single ventricle reconstruction. As mentioned, Ro et al. (8) developed a set of criteria to determine the need for cath prior to Fontan based mainly on echo data, hemoglobin and pulse oximetry (8). In that study, they suggested that CMR may have helped and indeed, applying the Ro criteria in our patient population changed the need for cath in 55%.
Two studies by Brown et al. (12,13) investigated the utility of cath prior to bidirectional Glenn/hemi-Fontan, 1 retrospective (12) and 1 prospective (13). In their retrospective analysis, they demonstrated that unless echo indicated a need, routine cath prior to surgery rarely provided new information or resulted in an intervention. Echo, however, was only 70% successful in completely characterizing pre-operative patients and 73% successful at PA assessment. The prospective study compared both CMR and cath prior to bidirectional Glenn/hemi-Fontan and found CMR was safer than cath and incurred less cost.
A proposed algorithm by Prakash et al. (25) suggested dividing single ventricle patients who were to undergo Fontan into a “high risk” and “low risk” group based on certain criteria including history (respiratory or lung disorder, heterotaxy or genetic syndromes), echocardiography (valve insufficiency, ventricular dysfunction, pulmonary vein stenosis), or angiography (branch PA stenosis, pulmonary vein stenosis, or systemic venous anomalies other than a left SVC). They stated that the algorithm could effectively screen patients unsuitable for Fontan and may be able to determine which patients needed invasive preoperative testing. The sensitivity and negative predictive values were 100%, however, the specificity and positive predictive values were low (67% and 16%, respectively). All of the 3 cases in our study that could not progress to Fontan because of death or heart transplantation met criteria for a “high risk” Fontan, unsuitable for that operation.
CMR is a safer, noninvasive test when compared with cath, which has complications such as risk of bleeding, ICU stays, prolonged hospitalization, and involves ionizing radiation with cancer risk (9–11,26). Our study showed an effective radiation dose of 6.8 mSv on average (range 2.1 to 27.9 mSv), falling into the range of other reported pediatric cath studies (27,28). In the current study, there was only 1 complication in all patients undergoing CMR (1%) and 3 complications in all patients undergoing cath (4%) with varying severity; endotracheal tube obstruction because of secretions in the CMR group versus complete heart block (1 with chest compressions and atropine) and inadvertent entry into the right carotid artery in the cath group.
This study is retrospective and is subject to all of the inherent biases that come with this type of investigation. In particular, the decision to perform either a cath or CMR prior to Fontan completion was the bias of the clinical practice of the referring physician.
Statistically, some of the comparisons made were underpowered, however, from a clinical standpoint, the differences between the groups were not significant (e.g., the difference cardiopulmonary bypass time and circulatory arrest time between C and MR groups on average, of 0.4 and 3 min, respectively, is not clinically significant enough to warrant detection).
In addition, immediate surgical results and short-term clinical outcomes were used to compare groups and no intermediate or long-term follow-up was conducted. Although immediate results are promising, a prospective, randomized clinical trial is warranted to compare CMR and cath prior to Fontan completion for diagnostic success rates as well as short- and long-term clinical outcomes.
Calculation of the average effective radiation dose was an estimate and may not represent precise deposition of the effective radiation dose.
Implications and Conclusions
Noninvasive imaging consisting of CMR and echo can completely characterize the relevant information needed prior to Fontan completion on a routine basis as well as need for cath in select patients. CMR was able to accurately assess cardiovascular anatomy and physiology compared with cath and surgical observation. Because little difference was noted between groups in preoperative hemodynamics or anatomy, intraoperative surgical complications or immediate clinical outcome, a strong case may be made for eliminating cath from routine preoperative evaluation. We recommend that cath should be utilized if: 1) noninvasive evaluation reveals need for potential interventions; 2) ventricular function/hemodynamics appear poor on noninvasive studies; 3) Ro criteria are met with CMR; 4) CMR is contraindicated; or 5) noninvasive imaging methods cannot obtain necessary information. Cath should not be required to assess PVR or ventricular end-diastolic pressure routinely, as our data demonstrated normal values. If poor ventricular shortening is identified or if there is a clinical/imaging suspicion of elevated PVR or ventricular end-diastolic pressure, the patient should be directed to cath. Immediate or long-term complications from cath should limit its use in routine settings. The authors recognize that our institutional bias is not to embolize aortopulmonary collaterals, dilate stenotic branch pulmonary arteries, or arch abnormalities prior to Fontan and this may change the decision-making process to perform CMR only at other institutions to C+M.
Dr. Fogel has received grants from Siemens, Edwards Lifesciences, and Kereos. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrioventricular valve regurgitation
- catheterization and echocardiography alone
- cardiac catheterization and cardiac magnetic resonance
- cardiac magnetic resonance
- intensive care unit
- left pulmonary artery
- cardiac magnetic resonance and echocardiography alone
- pulmonary artery
- pulmonary vascular resistance
- superior vena cava
- Received March 12, 2012.
- Revision received June 19, 2012.
- Accepted June 26, 2012.
- American College of Cardiology Foundation
- Garofalo C.A.,
- Cabreriza S.E.,
- Quinn T.A.,
- et al.
- Sano T.,
- Ogawa M.,
- Yabuuchi H.,
- et al.
- Ro P.S.,
- Rychik J.,
- Cohen M.S.,
- Mahle W.T.,
- Rome J.J.
- Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation, National Research Council
- Brown D.W.,
- Gauvreau K.,
- Powell A.J.,
- et al.
- Fogel M.A.
- Fogel M.A.,
- Weinberg P.M.,
- Chin A.J.,
- Fellows K.E.,
- Hoffman E.A.
- Kang I.S.,
- Redington A.N.,
- Benson L.N.,
- et al.
- Lorenz C.,
- Walker E.,
- Morgan V.
- Whitehead K.K.,
- Gillespie M.J.,
- Harris M.A.,
- Fogel M.A.,
- Rome J.J.
- Lai W.W.,
- Geva T.,
- Shirali G.S.,
- et al.,
- Task Force of the Pediatric Council of the American Society of Echocardiography, Pediatric Council of the American Society of Echocardiography
- Beels L.,
- Bacher K.,
- De Wolf D.,
- Werbrouck J.,
- Thierens H.
- Yakoumakis E.N.,
- Gialousis G.I.,
- Papadopoulou D.,
- et al.
- Bacher K.,
- Bogaert E.,
- Lapere R.,
- De Wolf D.,
- Thierens H.