Author + information
- Received August 5, 2009
- Revision received November 18, 2009
- Accepted December 14, 2009
- Published online May 1, 2010.
- Claas P. Naehle, MD⁎,
- Michael Kaestner, MD†,
- Andreas Müller, MD⁎,
- Winfried W. Willinek, MD⁎,
- Juergen Gieseke, PhD‡,
- Hans H. Schild, MD⁎ and
- Daniel Thomas, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Daniel Thomas, Department of Radiology, University of Bonn, Sigmund-Freud-Strasse 25, 53105 Bonn, Germany
Magnetic resonance angiography (MRA) is an established noninvasive imaging modality for detection and evaluation of vascular pathologies in children with congenital heart disease. Standard first-pass (FP)–MRA uses a 3-dimensional MRA sequence with an extracellular contrast agent, in which spatial resolution is limited by breath-hold duration, and image quality (IQ) is limited by motion artifacts. The purpose of this study was to compare the diagnostic confidence, IQ, and image artifacts of standard FP-MRA to a high-resolution, motion compensated steady-state (SS)–MRA of the thoracic vasculature in children and adolescents with congenital heart disease using a blood-pool contrast agent (gadofosveset trisodium). SS-MRA of the thoracic vasculature (technically successful in 90% of patients) offers superior diagnostic confidence and IQ compared with FP-MRA and shows fewer motion-related image artifacts. In addition, SS-MRA revealed findings missed by FP-MRA. Therefore, SS-MRA may prove specifically beneficial for imaging of thoracic vessels that are small and/or subject to motion.
Magnetic resonance angiography (MRA) has been established as a noninvasive alternative to conventional angiography for detection and evaluation of vascular pathologies in children and adolescents with congenital and acquired heart disease (1,2). The main advantages of MRA over conventional radiographic angiography and echocardiography for visualization of the thoracic vasculature are lack of radiation, noninvasiveness, a relatively large field of view, and independence of an acoustic window (1,2). To date, standard MRA for assessment of arterial and venous anomalies in children and adolescents is performed using a 3-dimensional (3D) single- or multiphase first-pass (FP)–MRA sequence with an extracellular contrast agent without cardiac gating during 1 (or more) breath-hold(s). Using this approach, spatial resolution is limited by the maximum breath-hold duration. In addition, standard contrast-enhanced 3D MRA is usually neither electrocardiogram (ECG)-gated nor does it use respiratory motion compensation, and thus it is prone to motion artifacts.
Recently, gadofosveset trisodium (Vasovist, Schering AG, Berlin, Germany [at this time named MS-325]), a blood-pool contrast agent, has been introduced, allowing for prolonged imaging up to 60 min after administration with potential improvement of spatial resolution compared with FP-MRA (3). Even though the superiority of high-resolution steady-state (SS)–MRA over standard-resolution FP-MRA has been reported in peripheral MRA, the value of gadofosveset trisodium for MRA of the thoracic vasculature has not been evaluated yet.
Thus, the aim of this study was to evaluate an ECG-gated, high-resolution, free-breathing SS-MRA sequence and compare it to a standard FP-MRA using a blood-pool contrast agent in children and adolescents with congenital heart disease and acquired disease of the thoracic vasculature.
Materials and Methods
We retrospectively identified 57 patients with congenital heart disease or acquired disease of the thoracic vasculature who underwent MRA of the thoracic vasculature between August 2008 and January 2009. Among these were 22 patients (8 to 18 years) in whom an MRA was performed after intravenous injection of gadofosveset trisodium. The institutional review board of the relevant institution approved the study protocol.
Magnetic Resonance (MR) Imaging
Contrast-enhanced MRA was performed after injection of gadofosveset trisodium using a commercial power injector (Spectris, Medrad, Volkach, Germany). The contrast agent was injected at a dose of 0.03 mmol/kg bodyweight and at a rate of 1 to 1.5 ml/s (depending on patient size), followed by a 20-ml saline flush at 1.5 ml/s.
All studies were performed on a clinical whole-body 1.5-T scanner (Achieva, Philips Medical Systems, Best, the Netherlands) equipped with 30-mT/m maximum field gradients and a 150-T/m/s slew rate, using a cardiac phased-array coil. After acquisition of scout images, functional steady-state free-precession cine images were acquired, covering the heart in the short axis, horizontal and vertical long axes, and the left ventricular outflow tract and, where appropriate, the right ventricular outflow tract. Flow measurements were performed in the ascending aorta and pulmonary trunk. MRA was performed at the end of the entire scanning procedure, where the FP-MRA was followed by the SS-MRA. For the multiphase FP-MRA, a standard T1-weighted 3D gradient-echo-sequence with a centric ordered k-space acquisition scheme was used. Using this approach, the scan duration for 1 dynamic scan is approximately 12 to 18 s. A total of 3 dynamics was acquired during 2 breath-holds. For the SS-MRA, a T1-weighted 3D inversion recovery sequence with ECG-triggering and navigator-based respiratory motion compensation was used. SS-MRA was acquired 2 to 3 min after FP-MRA. A similar pulse sequence has previously been described for contrast-enhanced imaging of the coronary arteries and shown to deliver excellent contrast-to-noise ratio for vascular structures (4). The inversion time was set to create the highest contrast between the thoracic vessels and the surrounding tissue, and was chosen using the Look-Locker approach. Detailed sequence parameters are given in Table 1.
Analysis of MR images with regard to image quality, image artifacts, and diagnostic content was performed using a commercially available, vendor-specific computer and software platform (View Forum 5.22, Philips, Best, the Netherlands).
Visual assessment of cardiac magnetic resonance (cmr) images
Both types of MRA sequences were analyzed separately by 2 experienced readers who were blinded to the results of the other reader as well as clinical data. For image analysis, the reviewers had access to both maximum intensity projections as well as the source data.
Image quality. Image quality was rated using maximum intensity projections of both types of MRA sequences using a 5-point grading scale. Pre-defined vessels (main pulmonary artery, proximal pulmonary arteries, lobar pulmonary artery branches, second-generation pulmonary artery branches, pulmonary veins, superior vena cava, aortic bulbus, ascending aorta, aortic arch including aortic isthmus, thoracic descending aorta, coronary ostia, coronary sinus) were graded separately for image quality with respect to border definition (blurring of vessel contours): 1 = excellent image quality (good delineation of the vessel border), 2 = good image quality (very little blurring of the vessel border), 3 = fair image quality (some vessel blurring), 4 = poor image quality (severe blurring of the vessel border), 5 = nondiagnostic image quality (vessel border unidentifiable).
Image artifacts. The CMR images were graded for presence of overall artifacts and their interference with diagnostic content using a 4-point grading scale: 1 = no artifacts present, 2 = mild artifacts without interference with diagnostic content, 3 = moderate artifacts degrading diagnostic content, 4 = severe artifacts resulting in nondiagnostic images.
Confidence of diagnosis. The CMR images were finally graded for overall diagnostic confidence diagnosis using a 3-point grading scale: 1 = high certainty of diagnosis, 2 = sufficient confidence of diagnosis, 3 = low certainty of diagnosis.
To determine vessel sharpness, an intensity profile along a user-specified segment along representative vessels (ascending aorta, left pulmonary artery left, upper pulmonary vein) was created with a public domain image processing software (ImageJ, U.S. National Institutes of Health, Bethesda, Maryland). To ensure data consistency, all vessel sharpness measurements were performed by Reviewer #1. The distance between 20% and 80% of maximum intensity was measured for each side of the profile and then averaged. The reciprocal of the averaged distance was taken as sharpness. Using this approach, a greater value for vessel sharpness was consistent with a better vessel definition.
Additional clinical information
Both types of MRA sequences were analyzed for additional clinical information. Additional clinical information was defined as any finding that could exclusively be obtained from either type of MRA sequence. In addition, it was verified that this information could not be obtained from any other type of CMR sequence (such as scout images, cine images, etc.) performed in the specific patient.
Data analysis was performed by using commercially available software (Analyse-it for Microsoft Excel version 2.12, Analyse-it Software, London, United Kingdom). All data are given as mean ± SD. Qualitative scores for image quality and quantitative measurements of vessel diameters and vessel sharpness were compared using the Wilcoxon signed-rank test as a normal distribution could not be assumed. Spearman correlation between both reviewers was calculated for inter-rater agreement on image quality grading. The level of significance was set to 0.05 for all tests.
Of the 22 patients identified following the inclusion criteria, 2 (9.1%) had to be excluded from further analysis, because poor performance of the respiratory navigator did not allow them to complete the SS-MRA scan. The remaining 20 patients constituted the final study population (mean age: 12.4 years, range 8 to 18 years; 15 men, 5 women). For further patient characteristics, see Table 2. In 1 of 20 patients (5%), the SS-MRA had to be repeated, because in the initial SS-MRA, an inappropriate inversion time was chosen. The mean inversion time leading to appropriate nulling of the surrounding soft tissue was 231 ± 24 ms (range 195 ms to 310 ms). Imaging duration for the SS-MRA was 5:53 ± 0:30 min (range 4:59 to 7:02 min).
For all vessels evaluated (except the lobar pulmonary artery branches graded by Reviewer #1 [p = 0.1465]), the SS-MRA showed significantly higher image quality grading (p < 0.05) for both reviewers (Table 3). This improvement in image quality was most apparent for the main pulmonary artery (mean improvement: 1.05 ± 0.82), the pulmonary veins (mean improvement: 1.18 ± 0.71), the aortic bulbus (mean improvement: 1.53 ± 0.75), the coronary ostia (mean improvement: 1.98 ± 1.01), and the coronary sinus (mean improvement: 1.43 ± 0.88), where the mean improvement in image quality translates into an improvement on the grading scale >1 step. Interobserver agreement was higher for the SS-MRA than the FP-MRA for all vessels except the lobar pulmonary artery branches (Table 3).
Overall image artifacts
Both reviewers judged that the FP-MRA showed significantly more overall image artifacts than the SS-MRA did (Reviewer #1—FP-MRA: 2.75 ± 0.43 vs. SS-MRA: 1.80 ± 0.75, p = 0.0001; Reviewer #2—FP-MRA: 2.70 ± 0.46 vs. SS-MRA: 1.75 ± 0.70, p < 0.0001). The mean reduction in overall image artifacts of 0.95 ± 0.69 translates into a 1-step improvement in the 4-point grading scale.
The SS-MRA yielded a significantly (p < 0.0001) higher overall diagnostic confidence for both reviewers (Reviewer #1—FP-MRA: 2.00 ± 0.45 vs. SS-MRA: 1.30 ± 0.46, p < 0.0001; Reviewer #2—FP-MRA: 2.00 ± 0.45 vs. SS-MRA: 1.30 ± 0.46, p < 0.0001). The mean difference in diagnostic confidence between the SS-MRA and the FP-MRA was 0.68 ± 0.54.
Compared to the FP-MRA, the SS-MRA yielded a significantly (p < 0.0001) higher vessel sharpness for the ascending aorta (FP-MRA: 0.38 ± 0.09 vs. SS-MRA: 0.53 ± 0.13, p < 0.0001), the left pulmonary artery (FP-MRA: 0.33 ± 0.05 vs. SS-MRA: 0.63 ± 0.11, p < 0.0001), and the left upper pulmonary vein (FP-MRA: 0.36 ± 0.09 vs. SS-MRA: 0.58 ± 0.11, p < 0.0001).
Additional clinical information
In 4 of 20 (20%) CMR examinations, the SS-MRA showed additional clinical information, which could be exclusively identified in the SS-MRA, and could not be depicted in the FP-MRA (Table 2). In none of the 4 patients (0%) could the additional clinical information be depicted in any other CMR sequence performed. These findings included a partial anomalous venous return in 2 patients and in 1 patient the coronary sinus and the coronary ostia, respectively. No additional clinical information was found for the FP-MRA versus the SS-MRA.
Congenital anomalies of the thoracic vascular system are an important cause of morbidity and mortality in infants and children (5), and although the perioperative and long-term morbidity and mortality have decreased, correct diagnosis is crucial for planning of surgery and the life long follow-up of these patients (1,5). In the past decade, MRA has become an important imaging modality for evaluation of pathologies of the thoracic vasculature. Especially in children and adolescents, the lack of radiation and the noninvasiveness is a major advantage of CMR angiography over catheterization and computed tomography in this patient population.
To date, standard MRA of the thoracic vessels is performed using contrast-enhanced FP 3D gradient-echo-sequences with extracellular contrast agents. Because of the properties of the extracellular contrast agents, imaging is restricted to the first pass, maximum second pass, of the contrast agent. Thus, data acquisition is limited to the breath-hold duration, which consequently limits the spatial resolution achievable. In addition, using standard, commercially available software packages, respiratory compensation techniques, and ECG-gating cannot be combined with FP-MRA. However, especially in children, the quality of the acquired image data is often compromised by respiratory motion artifacts due to the inability to follow verbal commands. Consequently, image quality and diagnostic confidence are often limited in FP-MRA of children and adolescents. Here, the use of a blood-pool agent has several advantages. Imaging time is not restricted to the first pass, but can be performed during the blood-pool–phase of up to 60 min after contrast agent injection (3). Therefore, motion artifacts can be reduced by combining navigator-respiratory compensation and ECG-gating, and the maximum spatial resolution achievable is no longer limited by the first pass of the contrast agent. It should be noted that navigator-based respiratory motion compensation is dependent on a good performance of the respiratory navigator. In our study, it was sometimes challenging to obtain a reliable navigator signal, especially in smaller children with only small liver-lung-interface and little respiratory excursion of the diaphragm. However, this was only the case in 2 of 22 (9.1%) patients.
In this study, the high-resolution SS-MRA showed higher image quality for all vessels rated and showed less overall image artifacts, resulting in a higher diagnostic confidence than FP-MRA. This may, in part, be ascribed to the fact that the high-resolution approach yielded significantly better vessel sharpness, thus improving vessel definition. Vessel sharpness itself was not only improved due to the higher spatial resolution, but also due to motion compensation through navigator-respiratory compensation in comparison to the breath-hold, nongated FP-MRA, and through ECG-gating. The acquisition window was specifically chosen to allow for motion artifact–free imaging of the pulmonary artery and pulmonary veins. A shorter acquisition window would have been desirable to improve assessment of the coronary ostia and proximal coronary arteries, but it was felt that an acquisition window of 120 to 150 ms was the best compromise between temporal resolution and scan duration. To minimize vessel blurring, especially of the pulmonary vessels, and thus to increase vessel sharpness, image acquisition was limited to end-expiration, and an appropriate length of the acquisition window was chosen to fit data acquisition into mid-diastole. This optimization of sequence parameters of the SS-MRA increased image quality especially in those vessels, which are mainly subjected to cardiac and respiratory motion (e.g., aortic bulbus, coronary ostia and sinus, pulmonary veins).
Although the SS-MRA offers all the advantages of a near isotropic 3D sequence, such as the possibility of multiplanar reformatting in any desired plane (Fig. 1), no information on flow dynamics can be obtained from the SS-MRA (Fig. 2). However, because dynamic information is important for assessment of intra- or extracardiac shunts in congenital heart disease as well as for the assessment of lung perfusion, the intention of this study is not to advocate the use of a single high-resolution sequence, but instead to demonstrate the added value of a combined imaging approach. Using the approach with a single injection of gadofosveset trisodium, both the dynamic information from the FP-MRA and the superior image quality of the SS-MRA can be used for image reading and interpretation. Whereas some authors have emphasized the value of ultrafast time-resolved dynamic MRA (contrast-enhanced robust-timing angiography–keyhole, time-resolved echo-shared angiographic technique, time-resolved imaging of contrast kinetics) for visualization of pulmonary vessels, because arteriovenous overlay is reduced (1), arteriovenous overlay was not felt to be an issue for vessel measurements in this study. The use of ultrafast time-resolved dynamic MRA and high-resolution SS-MRA warrants further investigation. However, at the time of this study, the required software package was not available on the CMR system used for this study.
A potential drawback of the high-resolution SS-MRA sequence is the susceptibility to arrhythmias. Also, image acquisition can become very long (>10 min) in patients with a very irregular breathing pattern, even when a motion-adaptive gating algorithm is used with navigator-respiratory gating. In fact, in our patient group poor navigator performance did not allow patients to finish the SS-MRA in 2 cases. However, in patients who are unable to follow verbal commands or to hold their breath (Fig. 2), the ability to use navigator-respiratory gating becomes clinically beneficial.
Currently, thoracic MRA in children and adolescents is most frequently used for diagnosis and follow-up of congenital heart disease. The accurate delineation of fast-moving vascular structures such as the coronary ostia, major aortopulmonary collaterals, or partial anomalous venous return poses a challenge for standard FP-MRA. In the present study, both reviewers rated the coronary sinus as nondiagnostic in 6 cases for the FP-MRA versus no case for the SS-MRA, and the coronary ostia as nondiagnostic in 18 cases (FP-MRA) versus 1 case (SS-MRA). In respect thereof, this noticeable improvement of image quality of the SS-MRA over the FP-MRA may constitute a clinically relevant advantage.
Additional clinical information
Multiple imaging modalities, such as chest radiography, echocardiography, conventional angiography, computed tomography, and MR are typically used to diagnose vascular anomalies and for planning of subsequent treatment. In comparison, CMR has been shown to demonstrate the abnormalities of pulmonary veins more accurately than cardiac angiography and echocardiography as it provides rapid and comprehensive 3D anatomic definition of the pulmonary veins (2). In up to 29% of patients, MRA has been demonstrated to clarify uncertain echocardiography findings; relative to X-ray angiography, 3D contrast-enhanced MRA was found to provide clinically new and important information in 46% of the patients (5). In this context, the superiority of 3D FP-MRA over conventional imaging techniques may even be exceeded by the SS-MRA, which showed additional clinically relevant findings in 4 patients (4 of 20, 20%) that could solely be seen in the SS-MRA and could not be detected in the FP-MRA (Table 2). It should be noted that in these 4 patients, the SS-MRA was rated an image quality that was in average 1.25 higher (range 0.5 to 2.5) than for the FP-MRA, indicating that the SS-MRA may prove especially helpful in case of an impaired image quality in the FP-MRA and serve as a clinically helpful add-on sequence. Two of these patients are especially noteworthy, as the planning of an invasive/surgical procedure primarily relied on CMR: 1) An 18-year-old patient with total situs inversus and repair of transposition of the great arteries was referred to identify a possible venous access to the coronary sinus before a planned implantation of a biventricular pacemaker system. In this patient, only the SS-MRA allowed for reliable identification of a possible transvenous access to coronary sinus for implantation of a right/left ventricular pacing lead (Fig. 3). This finding was later confirmed by cardiac catheterization. 2) A 12-year-old patient with repair of a common trunk was primarily referred to our department for quantification of a progressive insufficiency of the aortic valve. In this patient, the initial diagnosis of a quadricuspid aortic valve was made, necessitating recurrent surgery for valve repair/replacement. For planning of the procedure, the exact locations of the coronary ostia were important for the cardiac surgeons, and the coronary ostia could only be identified in the SS-MRA, but not in the FP-MRA (Fig. 4).
First, the study is limited by its retrospective design (e.g., possible patient selection bias). Second, the weight-adjusted dosage of gadofosveset trisodium can lead to injection of low volumes of contrast media, and thus may influence overall vessel delineation in the FP-MRA. However, in this retrospective analysis, the comparison of 2 different contrast agents was not possible and would be considered unethical in children. Third, 2 different MRA techniques are compared against each other without comparison against a gold standard technique (catheter angiography).
In summary, we demonstrated that a high-resolution, cardiac- and respiratory-gated MRA sequence during the steady-state phase of gadofosveset trisodium delivers superior image quality and improves vessel delineation compared with a standard dynamic MRA sequence during the first pass of the contrast agent. This improvement in image quality and sharpness allowed for exclusive identification of clinically relevant anatomic structures in the SS-MRA, which could not be detected in the FP-MRA.
A possible disadvantage of this imaging approach is the lack of dynamic information that cannot be derived from the high-resolution SS-MRA. Therefore, the combination of a FP and SS imaging protocol using a blood-pool contrast agent is a promising clinical approach for imaging of the thoracic vasculature, which warrants further investigation in the future. The proposed approach for imaging of the thoracic vasculature may prove specifically useful in children with congenital heart disease, where vessel conspicuity and delineation are limited by spatial resolution and respiratory motion artifacts.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- magnetic resonance angiography
- Received August 5, 2009.
- Revision received November 18, 2009.
- Accepted December 14, 2009.
- American College of Cardiology Foundation