The main cause of long-term morbidity and mortality after the arterial switch operation for the transposition of great arteries is a complication of the transferred coronary arteries.Reference Wernovsky, Mayer and Jonas1 These lesions are usually symptom free and several examinations have been made for the assessment of these patients at high risk for myocardial ischaemia.Reference Mussatto and Wernovsky2
Selective coronary angiography is the current gold standard for detecting anatomic coronary stenosis.Reference Legendre, Losay and Touchot-Kone3–Reference Cohen and Wernovsky5 However, it is an invasive, radiant examination and does not provide functional information about myocardial perfusion in the regions irrigated by the stenotic coronary artery.
On the other hand, positron emission tomography is the gold standard for evaluating myocardial viability and myocardial perfusion abnormalities,Reference Swada, Allman and Muzik6, Reference Rickers, Sasse and Buchert7 and allows regional quantification of absolute coronary flow reserve,Reference Ibrahim, Nekolla and Schreiber8, Reference Dayanikli, Grambow and Muzic9 but its availability is limited to a few centres and has poor spatial resolution.
Myocardial perfusion magnetic resonance imaging represents an alternative tool for evaluating regional blood flow. It is a non-invasive, non-radiant examination and allows extensive functional and anatomic information in less than 1 hour.Reference Lauerma10–Reference Crean and Merchant13 Despite the widespread use of this technique for ischaemia detection in the adult population, the experience in the paediatric patient is still limited.
The purpose of this paper is to report a single centre experience with myocardial perfusion magnetic resonance imaging for detecting ischaemia after arterial switch operation for the transposition of the great arteries.
Patients and methods
Using a prospectively maintained cardiovascular surgery database, 60 patients aged between 13 and 16 years with arterial switch operation for the transposition of great arteries currently being followed at our outpatient clinic, were invited from April, 2007 to April, 2008 to participate in this study after getting approval from the institutional research ethics board. Age limit was 13 years because the collaboration required for a 45–60-minute examination could not be guaranteed in younger children. At the time of the study, no patient was more than 16 years of age.
Patients with contraindications (nine) for cardiac magnetic resonance imaging and adenosine stress testing were excluded: pacemaker implantation (one), psychomotor delay (three), claustrophobia (one), advanced cardiac block (three), or severe bronchospasm (one). A total of 28 patients gave written informed consent and represent our study population.
The authors reviewed all operative and post-operative hospital complications and long-term follow-up events noted in the hospital records in detail. Coronary artery patterns were determined from the operative reports. We used the Yacoub and Radley-Smith classification for the description of the coronary pattern (see Table 1).
Table 1 Baseline characteristics.

ASO, arterial switch operation; CAO, aortic coarctation; F, female; M, male; TGA, transposition of great arteries; VSD, ventricular septal defect
Coronary artery pattern (Yacoub and Radley-Smith classification): Type A: left coronary artery originates from the left sinus and the right coronary artery from the right sinus; Type B: single coronary artery; Type C: two para-commissural ostia with or without intramural course; Type D: right coronary artery and circumflex originate from the right ostium, left descendent anterior alone from the left ostium; Type E: right coronary artery and left descendent anterior originate from the left ostium, circumflex alone from the right ostium
Patient functional evaluation included clinical examination, resting 12-lead electrocardiogram and echocardiogram for the evaluation of the ejection fraction. Coronary angiography or myocardial scintigraphy was not routinely performed.
Adenosine-related adverse events were graded as severe or life threatening if they resulted in substantial haemodynamic compromise requiring hospitalisation, moderate when requiring treatment but not hospitalisation and mild if minor transient symptoms occurred but with no need for medical treatment.
Magnetic resonance imaging protocol
All patients were instructed to refrain from smoking, drinking tea, coffee, or chocolate, and taking beta-blockers for at least 24 hours before the magnetic resonance study.
The patients were examined in the supine position with a 1.5-tesla magnetic resonance scanner (Symphony Vision, Siemens, Erlangen, Germany), which was equipped with quantum gradients of 30 millitesla per metre and used a four-element phased array radiofrequency coil. After the patients were positioned on the scanning table, intravenous access was established via the antecubital vein in both arms. Electrocardiogram monitoring leads and a brachial blood pressure cuff were applied. A single-lead electrocardiogram was continuously monitored on the magnetic resonance imaging console. Systolic and diastolic blood pressure were recorded using an automatic device at baseline and every 3 minutes throughout the procedure.
After a series of scout and axial steady-state free precession views, we acquired two-chamber, three-chamber, four-chamber and parallel short-axis cine views – 6–8-millimetre thick sections, 2-millimetre gaps – that included the entire left ventricle from the atrioventricular valves to the apex of the heart. Functional evaluation of the left ventricle was performed using a breath-hold steady-state free precession in the short-axis oblique plane. The imaging parameters were repetition time and echo time of 35.14 milliseconds and 3.2 milliseconds, respectively; flip angle at 55 degrees; 120 × 192 matrix, spatial resolution of 2.3 × 1.9, 6–8 millimetres of slice width, integrated Parallel Acquisition Techniques with a reduction factor of 2. Functional analysis was performed using standard software (Argus, Leonardo, Siemens Medical Solutions, United States of America, Inc.). For perfusion measurements at rest, three short-axis sections (basal, centre, and apical) were determined by choosing one section in the centre of the left ventricle and two sections at a distance of 20 millimetres from this reference point towards the base and the apex, respectively. For perfusion imaging, a single-shot, saturation-recovery, gradient-recalled echo sequence with steady-state free precession – repetition time and echo time of 203.8 milliseconds and 1.08 milliseconds, respectively; and flip angle at 50 degrees – with one saturation prepulse per slice before data readout – prepulse delay: 10 milliseconds, which yields an effective inversion time of 140 milliseconds; field of view: 35 × 30 centimetres; matrix: 192 × 108; resulting pixel size: 2.3 × 1.7millimetres; and section-thickness: 8 millimetres. Magnetic resonance stress perfusion scanning during adenosine infusion (Adenocor, Sanofi-Synthelabo, Maassluis, The Netherlands) was at a dose of 140 micrograms per kilogram per minute, total duration of adenosine infusion is 5 minutes. After 3.5 minutes of adenosine infusion, perfusion imaging is performed during and after the injection of a dose of 0.05 millimol per kilogram of gadobenate dimeglumine (Multihance®; Bracco-Altana, Germany) in an antecubital vein, flushed with saline at a flow rate of 3 millilitres per second using a power injector; three images per heartbeat are acquired during a breath hold in expiration for 60 consecutive heartbeats. After a 10-minute waiting period for equilibration of the contrast agent within the myocardium, a new perfusion scan was repeated at “rest” with injection of 0.05 millimol per kilogram of gadobenate dimeglumine but without adenosine, with the same parameters as described earlier.
Next, we acquired a contrast-enhanced three-dimensional magnetic resonance angiography using T1-weighted fast-gradient echo with multiple phases with 0.1 millimol gadobenate dimeglumine per kilogram. After a further 10-minute period, delayed contrast-enhanced imaging for the evaluation of cardiac viability was performed with an inversion recovery Flash two-dimensional sequence (breath hold, repetition time/echo time: 450 milliseconds/1.08, matrix: 120 × 192, flip angle: 50°, spatial resolution: 2.1 × 1.9, slice thickness: 80 millimetres), covering both ventricles in long- and short-axis and four-chamber views, after adjusting inversion time to obtain optimal suppression of normal myocardium signal with an inversion time of approximately 200–300 milliseconds.
During these two waiting periods, we acquired a new parallel short-axis cine steady-state free precession and a series of oblique coronal or oblique sagittal steady-state free precession images following right ventricular tract, ascending aorta, right and left pulmonary arteries and phase contrast magnetic resonance imaging – through plane sequences in the ascending aorta, pulmonary artery and right and left pulmonary arteries.
During the examination, a cardiologist, an anaesthetist, and a radiologist were present in the magnetic resonance suite to monitor the condition of the patient and to visually evaluate the images. Total duration of examination was 60 minutes.
Data analysis
Cine wall motion from baseline and stress perfusion images, perfusion and late gadolinium enhancement scans were interpreted independently by two different observers. Perfusion defects were determined by subjective visualisation. Perfusion defects were defined as focal regions of the myocardium, which had diminished and/or delayed contrast enhancement compared with normal myocardium. For visual assessment of perfusion deficits, adenosine stress, and rest scans were magnified and displayed simultaneously. If an area shows reduced signal intensity at stress but not at rest, it was considered pathological. Criteria for time perfusion deficits versus artefacts were sub-endocardial or trans-mural signal reduction, to see signal reduction in several frames and absence of breathing motion artefacts. Myocardial scars appear white in contrast to the black appearance of the normal myocardium at the delayed enhancement images. We did not perform quantitative methods to evaluate the myocardial delayed enhancement because we did not have any case of measurable infarction.
Results
Baseline characteristics of the 28 patients are listed in Table 1. Cardiac imaging information is shown in Table 2. No patient experienced peri-operative complications potentially related to myocardial ischaemia. All patients were symptom free and with no cardiac medications at the end of follow-up. All of them had normal sinus rhythm with no ST-T changes or abnormal Q waves on the resting electrocardiogram. No signs of coronary ischaemia were documented in the electrocardiogram during adenosine infusion, although the tracing could not be analysed during magnetic pulses due to signal artefacts. Left ventricular dimensions and shortening were within the normal range in all the patients. As stated previously, coronary angiogram and myocardial perfusion scintigraphy were not performed as routine examinations in the follow-up assessment of the patients. Nonetheless, five patients underwent late cardiac catheterisation as an interventional procedure to treat neopulmonary stenoses or as a preoperative evaluation. In the two cases in which selective coronary angiography was performed, no coronary stenoses were documented.
Table 2 Imaging test and adenosine tolerability.

A, abnormal; CAO, aortic coarctation; LV, left ventricle; MR, magnetic resonance; N, normal; NP, not performed; TGA, transposition of great arteries; VSD, ventricular septal defect
On the other hand, 11 patients were evaluated by nuclear myocardial perfusion imaging. Two patients had abnormal myocardial scintigraphy perfusion defects at the apical level consistent with the attenuation artefact.
Four patients underwent late re-operation. Indications for repeat surgery were neoaortic dilatation and severe aortic insufficiency (N = 1), re-coarctation (N = 2), and pulmonary stenosis (N = 1). No patient had re-intervention for coronary problems.
Magnetic resonance imaging tolerability
All examinations were well tolerated without major complications. One patient needed aminophyline administration for moderate bronchospasm crises after adenosine injection, which was graded as an adenosine-related moderate adverse event. Ten additional patients (36%) experienced mild adverse events related to adenosine injection such as rubor, chest discomfort, or cough.
Myocardial perfusion and late enhancement
Two first-pass myocardial perfusion magnetic resonance studies were excluded from analysis for technical reasons. After the analysis of 156 slices from the remaining 26 patients – three rest and three stress perfusion slices for each patient – no perfusion defects were detected.
Myocardial delayed enhancement was successfully performed in all 28 patients. In five patients, a septal, basal, ventricular wall sub-endocardial late enhancement was identified. All these patients had an underlying diagnosis of transposition of great arteries with large ventricular septal defect, which required patch closure at the time of arterial switch operation, and the magnetic resonance imaging findings were consistent with the patch tissue. No other late-enhancement magnetic resonance abnormalities were identified.
The two patients who had abnormal myocardial scintigraphy, perfusion defect at the apex, presented normal first-pass myocardial perfusion magnetic resonance imaging at this level and no late gadolinium enhancement.
Discussion
Late mortality in an arterial switch operation for the transposition of great arteries is known to be mainly related to myocardial ischaemia,Reference Tsuda, Imakita and Yagihara14–Reference Brown, Park and Turrentine18 and indeed, the most frequent indication for late re-operation in this group of patients.Reference Angeli, Raisky, Bonnet, Sidi and Vouhé19 The growth of the coronary arterial anastomosesReference Brutel de la Riviere, Quaegebeur and Brutel de la Riviere20, Reference Bartolani, Bianca, Patane and Mignosa21 and their patency remains a significant issue of long-term concern. Several authors have documented a prevalence of late coronary arteries abnormalities of 3–7.6%,Reference Legendre, Losay and Touchot-Kone3, Reference Tanel, Wernovsky and Landzberg4, Reference Von Bernuth22–Reference Bonnet, Bonhoeffer and Piechaud24 although the oldest survivors of the arterial switch operation in such series are just entering their third decade of life and this prevalence may increase as this population ages. Nowadays, coronary angiography is the main tool to detect silent coronary stenosis.Reference Bonhoeffer, Bonnet and Piechaud25 It is difficult, however, to judge the risks in the asymptomatic patient with obstructed coronary arteries. It is therefore important to evaluate myocardial perfusion, at rest and under stress, and viability before performing any further intervention.Reference Mavroudis, Backer and Muster26 Whether revascularisation should be carried out in the absence of evident myocardial ischaemia remains to be determined.Reference Bonnet, Bonhoeffer and Piechaud24, Reference Raisky, Bergoend and Agnoletti27
The methods that are currently available to evaluate myocardial perfusion and viability have been studied in detail in adults with acute and chronic coronary artery syndromes.Reference Kim, Fieno and Parrish28–Reference Giang, Nanz and Coulden32 Although the diagnostic accuracy and predictive profiles of these imaging techniques have been characterised in adults, their diagnostic utility in patients with arterial switch operation for the transposition of great arteries remains unknown and provides conflicting results.Reference Bonnet, Bonhoeffer and Piechaud24, Reference Acar, Maunoury and Bonnet33–Reference Weidiling, Wernovsky and Colan35 Studies using positron emission tomography have shown that the coronary arterial flow reserve may be decreased after the arterial switch operation, even in the absence of ischaemic symptoms.Reference Hauser, Bengel and Kuhn36–Reference Gagliardi, Adorisio, Crea, Versacci, Donato and Sanders39
The evaluation of myocardial perfusion and viability by magnetic resonance imaging offers several potential advantages. Studies in adults have shown that the superior spatial resolution of magnetic resonance imaging over nuclear techniques allows the detection of smaller defects and differentiation of sub-endocardial from trans-mural infarction.Reference Schwitter, Nanz and Kneifel40, Reference Kuhl, Beek and van der Weerdt41 Another advantage of magnetic resonance imaging is the ability to provide comprehensive anatomical and functional evaluation in a single examination; flattening of the pulmonary trunk, pulmonary artery branch stenosis, enlargement of aortic root and annulus, and aortic insufficiency are common anatomical residual lesions of arterial switch operation for the transposition of great arteries that can be thoroughly assessed by cardiac magnetic resonance imaging. However, its applicability to the paediatric population is yet to be established.
To the best of our knowledge, this is the first reported experience of the use of myocardial perfusion magnetic resonance imaging for myocardial ischaemia detection in the paediatric population with an arterial switch operation for the transposition of the great arteries. The follow-up protocol for patients with arterial switch operation for the transposition of great arteries in our institution consists of clinical, electrocardiogram and echocardiographic assessment on a yearly basis. Examinations devoted to the detection of coronary ischaemia have been traditionally reserved for symptomatic patients. As our population ages, however, the need for the early detection of asymptomatic ischaemia has become a reality. The low invasive nature of cardiac magnetic resonance and the fact that it is a non-radiant examination that provides thorough anatomical assessment makes it an appealing technique for the routine evaluation of our patients. Should a positive ischaemic test appear, cardiac catheterisation for evaluation of further intervention is indicated.
Our results are encouraging in terms of tolerability. Although minor symptoms related to adenosine administration were relatively common, those were transient and well tolerated. Only one patient experienced a moderate adverse event requiring medical treatment and no major complications occurred. On the other hand, the results of myocardial perfusion and late gadolinium enhancement were in good agreement with those obtained by clinical assessment, echocardiographic evaluation and electrocardiogram. No signs of myocardial ischaemia or necrosis were documented, even in the three cases of unfavourable coronary patterns C and E – patients 3, 23, and 27 – which have been proposed by different authors as being at high risk for poor outcome.Reference Legendre, Losay and Touchot-Kone3, Reference Bonhoeffer, Bonnet and Piechaud25, Reference Pasquali, Hasselblad, Li, Kong and Sanders43–Reference Mayer, Sanders, Jonas, Castañeda and Wernovsky45 In the smaller group of patients for whom myocardial scintigraphy was available (N = 11), there were two cases of discordant results. These gammagraphic defects involved only a small segmental area and may reflect arteriolar or capillary events occurring at the time of the operation, inhomogeneous myocardial protection, as proposed by some authors.Reference Weidiling, Wernovsky and Colan35 However, given their location, the most plausible explanation for these perfusion defects is attenuation artefact, which results from the non-uniform reduction of photon activity caused by the attenuation of the soft tissue.Reference Hendel, Berman and Cullom42
Finally, the lack of perfusion defects in our series compared with others may reflect the younger age of our population. Serial examinations will be required to detect perfusion defects as our population ages. On the other hand, regardless of the coronary pattern, no peroperative ischaemic events, which has been recognised as a risk factor for the development of coronary ischaemia in the long term,Reference Legendre, Losay and Touchot-Kone3, Reference Bonhoeffer, Bonnet and Piechaud25, Reference Pasquali, Hasselblad, Li, Kong and Sanders43–Reference Mayer, Sanders, Jonas, Castañeda and Wernovsky45 were documented in our patients. The different accuracy of the various perfusion techniques can be invoked which, in any case, may favour cardiac magnetic resonance imaging with a sensitivity of 91% and specificity of 94% when compared with positron emission tomography as a gold standard for coronary functional assessment.Reference Schwitter, Nanz and Kneifel40
Study limitations
Our study protocol did not include coronary angiography as part of the protocol and, therefore, no correlation with the current gold standard technique for coronary anatomy can be obtained. However, as stated above, the clinical significance of coronary stenosis found in the angiography remains controversial. In order to treat lesions with a functional impact, not just angiographic narrowing, in our current clinical practice coronary angiography is performed in the presence of symptoms or documented ischaemia in the stress test; cardiac magnetic resonance imaging with adenosine is used in this case. The long-term outcome of such a clinical approach remains to be seen.
Conclusion
The results of our preliminary experience show that magnetic resonance imaging evaluation of myocardial perfusion and viability is feasible in paediatric patients long after the arterial switch operation. Asymptomatic paediatric patients with a previous arterial switch operation for the transposition of the great arteries and normal resting electrocardiogram do not have perfusion defects on myocardial perfusion magnetic resonance imaging. We need further studies including coronary angiography correlation.
Acknowledgements
We express special thanks to F. Roses and V. Pineda. Without their help, this study could not be carried out.