Author + information
- Received June 15, 2001
- Revision received September 17, 2001
- Accepted October 18, 2001
- Published online January 16, 2002.
- Gerald F Greil, MD∗,
- Andrew J Powell, MD, FACC∗,‡,
- Hans P Gildein, MD∗ and
- Tal Geva, MD, FACC∗,†,‡,§,* ()
- ↵*Reprint requests and correspondence:
Dr. Tal Geva, Department of Cardiology, Children’s Hospital, 300 Longwood Avenue, Boston, Massachusetts 02115, USA.
Objectives The goal of this study was to evaluate the diagnostic value of gadolinium (Gd)-enhanced three-dimensional (3D) magnetic resonance angiography (MRA) in patients with congenital and acquired anomalies of the pulmonary and systemic veins.
Background Gadolinium-enhanced 3D MRA is a fast magnetic resonance imaging technique that has shown great promise in the evaluation of large and medium-sized arteries. However, its application to venous anomalies has not been studied in detail.
Methods The study retrospectively analyzed all patients who underwent Gd-enhanced 3D MRA examination from January 1998 through January 2001, were diagnosed with anomalies of the pulmonary or systemic veins and had additional diagnostic data available for comparison with the MRA findings.
Results Sixty-one patients (age 1 day to 60 years) were included. Image acquisition was completed in 29 ± 6.9 s. Pulmonary venous anomalies were found in 37 patients, systemic venous anomalies in 17 patients and both pulmonary and systemic venous anomalies in 7 patients. Compared with available diagnostic information by other modalities, all known or suspected venous anomalies were imaged by 3D MRA. In three patients, catheterization did not detect anomalies of the pulmonary veins that were subsequently diagnosed by MRA. The 3D MRA diagnoses were followed by 10 interventional catheterization procedures and 15 operations. In 74% of patients, 3D MRA either diagnosed previously unsuspected venous anomalies (28%) or added new clinically important information (46%). The mechanism of pulmonary vein compression in eight patients was determined by MRA but not by other imaging modalities. Using a five-level grading system for MRA image quality (1 = nondiagnostic; 5 = excellent), the average grade was 4.6 ± 0.6, with a 0.28 ± 0.6 mean grade difference between two independent observers.
Conclusions Gadolinium-enhanced 3D MRA is capable of rapidly and accurately diagnosing a wide spectrum of pulmonary and systemic venous anomalies and is a useful noninvasive alternative to diagnostic catheterization.
Anomalies of the pulmonary and systemic veins vary widely in their anatomic spectrum, clinical presentation and outcome (1). Although some venous anomalies do not cause a significant hemodynamic burden, the majority of pulmonary venous anomalies and a substantial number of systemic venous anomalies have important clinical and surgical implications. Accurate delineation of the anomalous connection, drainage, course and relation to neighboring structures of abnormal veins is important for the diagnosis and planning of transcatheter and surgical therapies (2).
During the past two decades echocardiography has become the principal noninvasive diagnostic modality for venous anomalies, whereas cardiac catheterization continues to be the diagnostic reference standard (3–6). However, both modalities have important limitations. Echocardiography is dependent on acoustic windows, which tend to become progressively more restricted as body size increases and as a result of postoperative changes. Moreover, because ultrasound energy is attenuated by the presence of air in the trachea, bronchi and lungs, imaging of nearby vessels can be difficult (6). Cardiac catheterization is an expensive procedure that carries risks associated with its invasive nature, ionizing radiation exposure and use of iodinated contrast agents (7). Magnetic resonance imaging (MRI) overcomes most of these limitations. Standard MRI techniques, including spin and gradient echo sequences, have been shown to accurately delineate anomalies of the pulmonary and systemic veins (8–12). However, these MRI techniques require relatively long scan times for complete anatomical coverage.
Gadolinium (Gd)-enhanced three-dimensional (3D) magnetic resonance angiography (MRA) is a fast imaging technique that has shown great promise in the evaluation of the arterial tree (13–15). However, its application to venous anomalies has been demonstrated in only a few case reports (16–19). The present study, therefore, was designed to evaluate the diagnostic value of 3D MRA in a cohort of pediatric and adult patients with congenital and acquired anomalies of the pulmonary and systemic veins and to compare these findings with those from cardiac catheterization, echocardiography, surgery and, when available, autopsy.
All patients referred to the Cardiac MRI Program at the Children’s Hospital, in Boston, from January 1998 through January 2001 who fulfilled the following criteria were included in this study: 1) diagnosis of pulmonary or systemic venous anomaly by any imaging modality; 2) underwent Gd-enhanced 3D MRA; and 3) had an echocardiogram, cardiac catheterization, computed tomography (CT), surgical confirmation or autopsy. Review of the medical records and computer databases was approved by the Children’s Hospital Committee on Clinical Investigations.
Magnetic resonance imaging studies were performed with a commercially available 1.5 T scanner (either a Signa Horizon HighSpeed with operating systems 5.7/5.8 or Signa Horizon LX EchoSpeed with operating systems 8.1/8.3, GE Medical Systems, Milwaukee, Wisconsin). The following radiofrequency coils were used depending on patients’ size: body (n = 3), torso phased array (n = 43), cardiac phased array (n = 8), head (n = 9) and shoulder phased array (n = 4). General anesthesia was used in patients who were unable to cooperate. Using a fast spoiled gradient echo recall sequence (echo time 1.5 to 4 ms, repetition time 9 ms, flip angle 30°, matrix 256 × 128, one-signal average), localizing images were obtained in the coronal, axial and sagittal planes. A non-ECG-triggered 3D spoiled gradient echo pulse sequence was prescribed from an axial localizing image in either the coronal (n = 31) or sagittal (n = 36) planes, depending on the desired anatomical coverage and body habitus (20). The 3D MRA acquisition parameters and imaging times used in this study are summarized in Table 1.
Gadopentetate dimeglumine (0.2 mmol/kg; Magnevist, Berlex Laboratories, New Jersey) was injected via a peripheral intravenous cannula either by hand (in patients weighing <10 kg) or by a power injector (Medrad, Indianola, Pennsylvania) at a rate ranging from 1.5 to 2.5 cc/s. Patients were instructed to take several deep breaths before image acquisition. The time delay between start of contrast injection and data acquisition was determined by the “best estimate” method (21)and ranged from 5 to 6 s in infants to 16 to 17 s in adults. Two sequential 3D MRA acquisitions were performed 10 to 15 s apart during which patients were asked to hold their breath. In patients under anesthesia, ventilation was suspended during imaging. The MRA data were acquired during free breathing in three patients who were unable to hold their breath.
The source images of the MRA datasets were first examined on the MRI operating console before the patient left the scanner to ensure adequate image quality. The studies were then archived on optical disks and reviewed on a commercially available computer workstation (Advantage Windows version 2.0, GE Medical Systems, Milwaukee, Wisconsin) using a combination of the following techniques (22): 1) user-defined subvolume maximal intensity projections (MIP); 2) user-defined multiplanar reformatting (MPR); and 3) 3D shaded-surface displays. Using these viewing techniques, all pulmonary veins, the left and right innominate veins, the superior vena cava (SVC), coronary sinus and the inferior vena cava (IVC) cranial to the renal veins were examined in detail. The MRA findings and the morphology, course, connections and anatomic relations to neighboring structures were recorded. In addition, all MRA datasets were independently reviewed for image quality by two investigators (G. F. G., H. P. G.) using a five-level scoring system modified from Danias et al. (23): 1) poor-quality information—nondiagnostic; 2) structures visible but markedly blurred—diagnosis suspected but not established; 3) anatomy visible with moderate blurring—able to establish diagnosis; 4) minimal blurring—good quality diagnostic information with definite diagnosis; and 5) sharply defined borders—excellent quality diagnostic information.
The MRA diagnoses were compared with all other available diagnostic information for discrepancies. The diagnostic yield of the MRA findings was classified in each patient as: 1) providing a previously unsuspected diagnosis; 2) confirming a suspected diagnosis but adding new clinically relevant information; or 3) confirming a suspected diagnosis without adding new clinically relevant information.
Continuous data are reported as mean ± SD or median, as appropriate. The agreement between observers regarding image quality grading was assessed by the Wilcoxon signed-rank test. A p value <0.05 was considered significant.
Of the 703 cardiac MRI examinations performed during the study period, 61 patients (9%) fulfilled inclusion criteria. Six patients had follow-up scans for a total of 67 MRAs. The median age was 15 years (mean 20 ± 17 years, range 1 day to 60 years) and the median weight was 50 kg (mean 43 ± 28 kg, range 2.7 to 97 kg). There were 29 female patients. The examination was performed under general anesthesia in 24 patients (median age 0.8 years) and no sedation was required in the remaining 43 patients (median age 29 years). All MRA studies were technically successful without adverse events. The mean duration of each of the two MRA data acquisitions was 29 ± 6.9 s (range: 21 to 45 s). Image analysis was typically accomplished in ≤15 min per patient. Multiple user-defined subvolume MIP and MPR angiographic-like projections were constructed nearly instantaneously, and 3D shaded surface models were typically reconstructed in ≤5 min per dataset. In addition to 3D MRA, 55 patients had fast gradient cine MR sequences for dynamic imaging of blood flow and quantitative assessment of ventricular function, 41 patients had phase contrast sequences for flow quantification and 14 patients had either standard or fast spin echo with double inversion recovery sequences performed for imaging of vessel walls or airways.
Magnetic resonance angiography clearly demonstrated the origin, course, connections (either to the atria or to an extracardiac structure) and any stenosis or dilation of the pulmonary and systemic veins in all patients. The spatial relationship between venous structures and adjacent nonvascular structures such as the trachea, spine and chest wall were clearly delineated. Table 2summarizes the venous anomalies in the 61 study patients. Of the 20 patients with partially anomalous pulmonary venous connection (PAPVC), 6 patients had anomalous connection of the left upper pulmonary vein to the left innominate vein, 10 patients had anomalous connection of the right upper pulmonary vein to the SVC, 1 patient had two anomalously connecting pulmonary veins (left upper pulmonary vein to left innominate vein and right upper pulmonary vein to right SVC) and 3 patients had scimitar syndrome (Fig. 1). Of the four patients with totally anomalous pulmonary venous connection (TAPVC), one was a newborn with heterotaxy syndrome and mixed connections of the pulmonary veins to the SVC via an unusually tortuous venous channel that coursed through the right lung parenchyma as well as a stenotic connection to the IVC through the portal vein (Fig. 2). In the other patient with supracardiac TAPVC, the left upper pulmonary vein was not identified by echocardiography and MRA was performed to exclude mixed connections. Magnetic resonance angiography clearly demonstrated the connection of the left upper pulmonary vein to the ascending vertical vein at a higher than usual site. Among the 19 patients with pulmonary venous obstruction, 11 patients had intrinsic stenosis (Fig. 3) and 8 patients had extrinsic compression (Fig. 4). In 4 of these 19 patients, pulmonary venous stenosis or compression was associated with systemic venous anomalies. Figure 5demonstrates IVC obstruction in a patient after Baffes procedure (24)as an example of acquired systemic venous anomaly.
Diagnostic yield of MRA
Compared with diagnostic information available by other modalities—echocardiography (n = 29), catheterization (n = 28), surgery (n = 15), autopsy (n = 3) and CT (n = 1)—all of the confirmed venous anomalies were accurately diagnosed by Gd-enhanced 3D MRA. Previously unsuspected diagnoses of venous anomalies were found by MRA in 17 patients (28%). In another 28 patients (46%), the suspected diagnoses were confirmed and additional clinically important information was provided. An example in this category is the group of patients in whom compression of the pulmonary veins was suspected by cardiac catheterization (n = 8). In addition to confirming the diagnosis, MRA also delineated the mechanism of compression (dilated atria after Fontan operation with atriopulmonary anastomosis in three patients, compression between the atria and descending aorta in four patients and severe scoliosis in one patient), allowing for planning of transcatheter and surgical interventions. In the remaining 16 patients (26%), the referral diagnoses were confirmed without additional information. In three patients, cardiac catheterization did not diagnose anomalies of the pulmonary veins that were subsequently demonstrated by MRA. Partially anomalous pulmonary venous connection was diagnosed by 3D MRA in two patients (confirmed at surgery in one patient and by a follow-up catheterization in the other) and pulmonary vein stenosis in one patient (confirmed by a follow-up catheterization and a subsequent 3D MRA). The 3D MRA diagnoses were followed by 10 interventional catheterization procedures and 15 operations.
Image quality was independently evaluated by two investigators using the five-level scoring system defined earlier. The mean score was 4.6 ± 0.6 with 63% of the scans receiving a score of 5, 32% receiving a score of 4 and 5% receiving a score of 3 (all in patients who were unable to hold their breath). A close agreement was found between the two observers with a mean difference of 0.28 ± 0.6 (range −1.0 to 1.0) (p = 0.12).
The results of this study demonstrate that Gd-enhanced 3D MRA is a rapid imaging technique capable of diagnosing a wide spectrum of pulmonary and systemic venous anomalies in patients ranging in age from newborn to adult. When the MRA findings were compared to those at catheterization, surgery and autopsy, no venous anomaly was missed. In fact, among the 28 patients who underwent cardiac catheterization before MRA, two instances of PAPVC and one instance of pulmonary vein stenosis were found by MRA but were not diagnosed by catheterization. This observation is not surprising given that imaging of vascular structures by X-ray angiography requires a relatively nearby injection of contrast media. Consequently, unsuspected pulmonary or systemic venous anomalies may not be targeted for angiography. However, Gd-enhanced 3D MRA demonstrates all the major pulmonary and systemic veins and arteries in the thorax, upper abdomen and the base of the neck with a single injection into a peripheral vein. The MRA data also provided new clinically important information in 45 of the 61 patients (74%) and was followed by 10 interventional catheterization procedures and 15 surgeries related to the pulmonary and systemic venous anomalies.
Echocardiography has been the primary noninvasive imaging modality for evaluation of the pulmonary (1,3,5)and systemic venous systems (4,5). However, the sensitivity of transthoracic echocardiography in detecting venous anomalies is limited by its dependency on acoustic windows (6). Transesophageal (5)and contrast echocardiography (25)may increase the diagnostic sensitivity of cardiac ultrasound. However, transesophageal echocardiography is a relatively invasive procedure, its field of view is narrow and its ability to visualize vascular structures outside the mediastinum is very restricted.
The experience with this cohort indicates that one of the most important advantages of 3D MRA is the ability to process the images offline and to reformat angiographic-like projections in any arbitrary plane. This capability, provided by the 3D nature of the data, allows for evaluation of tortuous blood vessels and for understanding their spatial relationships with neighboring nonvascular structures. Offline image postprocessing proved especially useful for determining the mechanism of pulmonary vein compression and stenosis. For example, by tilting the reformatted image plane from a standard projection to oblique planes parallel or perpendicular to the vessel in question, the viewer can interactively navigate through the 3D volume data to determine whether the pulmonary veins are extrinsically compressed or intrinsically stenosed (Figs. 3 and 4). The 3D MRA data, which are available within seconds after the conclusion of the scan, can also be used for additional targeted MR imaging to further investigate the venous pathology. For example, fast-gradient cine MR may be performed to detect abnormal flow patterns and intracardiac anomalies (26), and phase contrast sequences to quantify flow (e.g., pulmonary-to-systemic flow ratio, differential pulmonary blood flow, etc.) (27). In this cohort, fast spin echo with double inversion recovery proved especially valuable for evaluation of vessel wall morphology in patients with pulmonary vein stenosis.
The time interval between contrast injection and image acquisition in this study was determined using the “best estimate” method (21). Alternative methods, including bolus tracking and a test bolus timing (21,28), aim to synchronize image acquisition with the onset of peak contrast concentration in the target vessel. The experience with this cohort indicates that initiation of image acquisition within 6 to 16 s from the onset of contrast injection and the acquisition of two sequential datasets, each lasting ∼20 to 40 s provided high-quality 3D MRA datasets. Absence of “pure” arterial or venous phases was not found to decrease the diagnostic quality of the images or to hinder the ability to demonstrate venous anomalies. Furthermore, many patients with congenital heart disease have shunt lesions that lead to early venous-arterial mixing of the contrast bolus.
Newer 3D MRA sequences using high-performance gradients may obviate the need for a timing strategy. Because a complete 3D MRA dataset can be acquired in 4 to 7 s (29), several sequential acquisitions can be obtained immediately following contrast injection. Subsequently, the best-quality dataset is selected for image analysis. Alternatively, the arrival of the contrast bolus to the target vessel can be imaged using a real-time magnetic resonance sequence, prompting initiation of a high-resolution 3D MRA image acquisition.
Although the majority of the patients in this cohort were able to hold their breath during image acquisition, a diagnostic-quality study was successfully obtained in the three patients in whom the sequence was performed during free breathing. In these patients, image quality decreased approximately one grade and was associated with some blurring of vessel margins.
The sensitivity of Gd-enhanced 3D MRA in diagnosing venous anomalies may not have been fully determined by this investigation because not all 703 patients who underwent MRI examinations had confirmatory tests such as catheterization, surgery or autopsy. In other words, among the 642 patients who were not included in this study, there may have been patients with venous anomalies that were not detected by MRA or any other modality. However, a large-scale study in which all patients undergo both catheterization and 3D MRA may not be practical or ethical. Furthermore, no instance of missed venous anomaly has come to clinical attention to date. Finally, this study has not defined the sensitivity of MRA in diagnosing small coronary sinus septal defects and tiny venous collaterals (<1 mm).
The results of this study indicate that Gd-enhanced 3D MRA is suitable for the diagnosis of congenital and acquired anomalies of the pulmonary and systemic veins in a wide range of patient ages and can be considered a noninvasive alternative to X-ray angiography. Furthermore, the 3D nature of the data allows one to employ multiple viewing and reconstruction techniques that enhance interpretation. In conjunction with information obtained by other sequences, MRI/MRA can provide a comprehensive evaluation of cardiovascular anatomy and physiology that facilitates planning for transcatheter and surgical therapies.
☆ Dr. Greil was supported by grant #1751/1 from the Deutsche Forschungsgemeinschaft (DFG), Bonn, Germany.
- computed tomography
- inferior vena cava
- maximal intensity projection
- multiplanar reformatting
- magnetic resonance angiography
- magnetic resonance imaging
- partially anomalous pulmonary venous connection
- superior vena cava
- totally anomalous pulmonary venous connection
- Received June 15, 2001.
- Revision received September 17, 2001.
- Accepted October 18, 2001.
- American College of Cardiology
- Geva T.,
- Van Praagh S.
- Santoro G.,
- Ballerini L.,
- Bialkowski J.,
- et al.
- van der Velde M.E.,
- Parness I.A.,
- Colan S.D.,
- et al.
- Choi J.Y.,
- Jung M.J.,
- Kim Y.H.,
- et al.
- Stumper O.,
- Vargas B.J.,
- Rijlaarsdam M.,
- et al.
- Geva T,
- Kreutzer J.
- Vitiello R.,
- McCrindle B.W.,
- Nykanen D.,
- et al.
- Snider A.R.,
- Silverman N.H.,
- Turley K.,
- et al.
- Ferrari V.A.,
- Scott C.H.,
- Holland G.A.,
- Axel L.,
- St. John Sutton M.
- Yucel E.K.,
- Anderson C.M.,
- Edelman R.R.,
- et al.
- Vick G.,
- Ravekes W.,
- Muthupillai R.,
- et al.