Evaluation anomalies
of complex congenital ventricular with magnetic resonance imaging
Complex ventricular anomalies are frequently associated with abnormalities of thoracic and abdominal situs, arterioventricular connection, and venous connection. The definition of all components of these anomalies is difficult to accomplish with imaging techniques. This study compared the effectiveness of electrocardiographic (ECG) gated spin-echo magnetic resonance imaging (MRI) with cardiac angiography for the evaluation of all components of central cardiovascular anatomy in patients with the clinical diagnosis of single or common ventricle or complete atrioventrlcular (AV) septal (canal) defect. MRI studies and angiograms of 29 patients were evaluated independently. A sequential approach was used to define cardiac anatomy assessing nine anatomic features in each patient. MRI provided 261 observations and angiography provided 209 observations. In the mutual 209 observations, only 17 discrepancies were found. Comparison of MRI and angiography in individual cases showed that MRI was as effective as angiography in the depiction of ventricular anomalies, including determination of morphology and evaluation of the size of the ventricles, the orientation of the ventricular septum relative to the AV valves, as well as the origins and spatial relationships of the great arteries. MRI was more informative for the determination of thoracic and abdominal situs and systemic and pulmonary venoatrial connections, but was not as effective for the evaluation of semilunar valves. Thus MRI provides complete evaluation of central cardiovascular anatomy and is effective in the anatomic assessment of most components of complex ventricular anomalies. (AM HEART J 1990;120:133.)
Barbara A. Kersting-Sommerhoff, MD, Lisa Diethelm, MD, Paul Stanger, MD, Renee Dery, MD, Stanley M. Higashino, MD, Sarah S. Higgins, MS, and Charles B.Higgins, MD. San Francisco and Oakland, C&if.
Complex ventricular anomalies have proved to be a diagnostic challenge for angiography as well as for noninvasive imaging techniques. Angiography is the standard imaging technique used for definitive diagnosis and preoperative assessment, but superimposition of cardiac structures can be problematic in these complex cases when a projectional imaging technique is employed. Complex ventricular anomalies are frequently associated with abnormalities of thoracic and abdominal situs, arterioventricular connection, and venous connection, so that definition of all components of these anomalies requires imaging of the entire central cardiovascular anatomy. Electrocardiographic (ECG) gated magnetic resonance imaging (MRI) has been used for the diagnosis of many congenital anomalies of the heart.‘-I2 The From the Departments of Radiology and Pediatrics (Cardiology) of the University of California Medical Center, San Francisco, and the Department of Cardiology in the Childrens Hospital Medical Center, Oakland. Received
for publication
Reprint University
requests: Charles of California,
4l1120435
Jan.
16, 1990;
accepted
Feb.
B. Higgins, MD, Department San Francisco, CA 94143.
26, 1990. of Radiology,
L 308,
ability to image in multiple orthogonal planes and the high natural contrast between rapidly flowing blood and internal cardiac structures are useful attributes of this technique. The effectiveness of MRI in comparison with angiography in the complete definition of complex cardiovascular anomalies has not been evaluated previously. The purpose of the current study was to assess the effectiveness of MRI in the diagnosis of complex congenital ventricular anomalies such as single ventricle and atrioventricular septal (canal) defects. In an attempt to determine the accuracy of MRI diagnoses, angiography was used as the “truth standard” and MRI diagnoses were correlated with angiographic findings in all patients. MRI and angiography were compared in their effectiveness for defining all components of central cardiovascular anatomy in these complex lesions. METHODS Twenty-nine patients (17 males and 12 females) with the clinical diagnosis of single or common ventricle or complete atrioventricular septal (canal) defect who had not under133
134
Kersting-Sommerhoff
Table
I. MRI findings
Patient No. (age, sex)
et al.
Thoracic
situ.5
American
Venoatrial connections
Atria1 septum
Ventricular morphology
loop
AV valve morphology
(23 yr, M) (7 yr, F) (30 yr, F) (16 yr, F) (9 yr, M)
Solitus Solitus Solitus Solitus Solitus
Normal Normal Normal Normal Normal
2’ ASD 2” ASD Intact Intact 2O ASD
Single Single Single Single Single
6. (12 yr, F) 7. (3 mo, F) 8. (17 yr, M)
Solitus Solitus Solitus
Normal Normal Normal
2 ventricles 2 ventricles 2 ventricles
D 0.4 D 6) D (L)
Atresia (right) AV canal AV canal
9. (10 yr, M)
Inversus
Bilateral SVC
Atresia (right)
Bilateral SVC
2 ventricles RV dominant Single LV
L CR)
Right isomerism Right isomerism Right isomerism Left isomerism Left isomerism
2’ ASD 1” ASD Common atrium Common atrium Common atrium lo ASD
D @A
2 ventricles
L (JJ
Common valve AV canal
2 ventricles
L (R)
AV canal
2 ventricles
D 03)
AV canal
2 ventricles LV dominant
L 6)
AV canal
2“ ASD
2 ventricles
D (RI
2 valves
1’ ASD
2 ventricles
L CR)
AV canal
Common atrium
2 ventricles, LV dominant
D (RI
AV canal
1. 2. 3. 4. 5.
10. (16 yr, M) 11. (11 yr, M) 12. (16 yr, M) 13. (20 yr, M) 14. (19 yr, M) 15. (9 yr, M)
Left isomerism
16. (5 yr, M)
Left isomerism
17. (16 yr, M)
Left isomerism
Bilateral
SVC
TAPVR
Common atrium
Normal IVC interruption PAPVR IVC interruption PAPVR IVC interruption PAPVR Bilateral SVC
atrium lo ASD
gone repair were studied with ECG gated MRI. The patients ranged in agefrom 3 months to 30 years. Six children were 3 years old or younger. Sedation was generally necessaryin children under 7 years of ageand was administered as90 to 100 mg/kg of chloral hydrate orally 30 minutes prior to the study or 5 to 6 mglkg of pentobarbital sodium intramuscularly (maximum dose100 mg). Imageswere acquired using a cryogenic 0.35 T superconducting magnet (Diasonics MT/S, Milpitas, Calif.). First echocardiographic imageswith an echo delay time (TE) of 28 to 30 msecwere obtained. In somepatients second echo (TE = 56 to 60 msec)were acquired additionally. Since ECG gating wasapplied in all patients, the repetition time (TR) equaledthe R-R interval and dependedon the patient’s heart rate. Tomograms in the transverse plane were obtained in all patients and additional imagesin the sagittal or coronal plane were acquired depending on the findings on the initial transverse run. Multiple contiguous slicesencompassingthe entire heart and extending from the aortic arch through the upper abdomenwere acquired in all subjects.Slice thickness was 10 mm in adults and 5 mm in most children.
LV LV LV LV LV
Ventricular (apex)
July 1990 Heart Journal
L L L L L
6) (L) (L) 6) UJ
Atresia (left) Atresia (left) 2 valves 2 valves 2 valves
Angiographic studieswereavailable for all patients; they were done at two different institutions. Each study included one or more injections into the dominant ventricular chamber and the angiogramswere recorded on 35 mm cineangiography film at 30 frames/set. Image analysis. MRI studiesand angiogramswereevaluated independently by two different teams of observers. The observerswere awareof the generaldiagnosisbut had no knowledge of anatomic details or of the results of the other imaging study. The teams operated as a consensus panel and analyzed the cardiac anatomy using a sequential approach.lO,l1 The morphology of nine components was defined: (1) visceroatrial situs, (2) systemicand pulmonary venoatrial connections, (3) atria1 septum, (4) ventricular morphology, (5) ventricular loop and direction of the cardiac apex, (6) atrioventricular (AV) valve morphology, (7) ventricular septum, (8) spatial relationship of the great arteries and arterioventricular connections,and (9) the status of the semilunar valves and ventricular outflow tracts. Image quality of the MRI studies was assessed using a system describedpreviously.3 “Excellent” quality images provided distinct delineation of internal cardiac structures
Volume Number
120 1
Ventricular septum
MRI of complex
Great arterial relationship
Semilunar valvular and
subvalvar stenosis
Sub PS Normal Sub PS Normal Normal
VSD Inflow VSD Inflow VSD
L-TGA L-TGA L-TGA L-TGA L-TGA PA D-DORV D concordant D concordant
VSD
L-DORV
Normal
BVF
A-DORV
Sub PS
Inflow VSD
L-DORV
Normal
Inflow VSD
L-DORV
Sub PS
Inflow VSD
D-DORV
Normal
Inflow VSD
L concordant
Normal
VSD
L concordant
PS Sub PS
Inflow VSD
L-DORV
Normal
Inflow VSD
L-DOLV
PS
BVF BVF BVF BVF BVF
PS Normal Sub PS
Sub PS
and myocardium without signal dropout from any regions of the atria1 or ventricular walls. In “diagnostic” images, occasionallossof signal or noisein the phaseencoding direction degraded overall quality, but cardiac structures were clearly delineated. In “barely diagnostic” images,the diagnosiscould still be made, but visualization of the cardiac structures wasobscuredby signallossor noise.“Nondiagnostic” images did not permit adequate diagnosis because cardiac chambers and great vessels were not discernible. Terminology and morphologic criteria. Some debate exists regarding the terminology for complex congenital ventricular anomalies.12-1g Since the aim of this study was not to classify these complex anomaliesbut to assessthe role of MRI in their evaluation, controversial terms were avoided and a nomenclature proposed previously by &anger et al.” wasused.The following terms were usedto describe the ventricular anomaliesin this study. “Single ventricle” hasboth or a commonAV valve(s) entering one ventricle. When this is a morphologically left ventricle, there commonly exists a rudimentary right ventricular chamber and the two are connected by a bulboventricular
ventricular anomalies I 35
foramen. When the dominant ventricle is of right ventricular morphology, there may only be a trabecular pouch or no secondventricular chamber.“Common ventricle” hasa rudimentary rim of septum separating two ventricular cavities with an AV valve entering the inflow portion of each, or a commonAV valve entering both cavities.15,l6 “Complete atrioventricular septal (canal) defect” consistsof a primum atria1septal defect, a commonAV valve, and a large inflow ventricular septal defect.21p 23The common valve may be more committed to one of the ventricles, with that chamber being the dominant one, implying that the other ventricle is smaller than normal. “AV valve atresia” is associatedwith an ipsilateral hypoplastic ventricle. The contralateral ventricle is normal or enlarged. The presenceor absenceof an infundibulum was the major criterion usedto identify ventricular morphology on MRI tomograms. The presenceof an infundibulum was considered to be characteristic of a right ventricle; when there wasno detectable musclebetween the AV valve and the adjacent semilunar valve, the chamberwasconsidered to be a morphologic left ventricle. The “bulboventricular loop,” according to Van Praagh et al.)5 wasdescribedas D-loop when the morphologic right ventricle was located to the right of the morphologic left ventricle; conversely, when the morphologic right ventricle waslocated to the left of the morphologic left ventricle, L-loop was present. “Ventricular septal defects” were classifiedaseither part of an AV septal (canal) defect, a bulboventricular foramen in single ventricle, or unspecified. Subclassification of “atria1 septal defects” included primum (AV canal), secundum, or sinus venosusdefect or common atrium, and wasaccording to criteria previously established.g Classification of “thoracic situs” was inferred from the main stem bronchial
anatomy
as well es from the relation-
ship of the central pulmonary arteries to the bronchi (epior hyparterial). Right isomerismis characterized by bilateral right-sidedness (eparterial bronchi); left isomerism refers to bilateral left-sidedness(hyparterial bronchi). The abdominal visceral situs was inferred from the location of stomach and liver, the presence or absence of spleen (asplenia, polysplenia), and the aortocaval relationship (ipsi- or contralateral). RESULTS
Sixteen of the MRI studies were rated “excellent” quality; 13 were of “diagnostic” quality. However, for two patients the initial study was considered nondiagnostic and was repeated, in one case with sedation. These second studies were rated to be “excellent” and “diagnostic,” respectively. The major factors affecting quality were somatic motion or respiratory degradation of the signal. The findings of MRI for each patient with respect to all nine features are presented in Table I. Since nine anatomic features were assessed on both MRI studies and angiograms in each patient, a total number of 261 mutual observations
July
136
Kersting-Sommerhoffet al.
Table
I. cont.
Patient No. (age, sex) 18.
(5
yr, F)
19. (1 yr, M)
Thoracic situs
American
Venoatrial connections
Atria1 septum
AV valve morphology
Normal
Intact
Single LV
L CL)
Bilateral SVC PAPVC Normal
Common atrium 2’ ASD
2 ventricles
L (R)
Single LV
D 6)
2 ventricles, LV dominant 2 ventricles, RV dominant 2 ventricles, LV dominant Single RV
I. (R)
valves, DILV Atresia (right)
D (L)
2
D (L)
2
D (R)
Atresia (right)
2 ventricles, LV dominant 2 ventricles
D CL)
Atresia (left)
L (RI
AV canal
2 ventricles RV dominant Single LV
D NJ
2 ventricles, LV dominant
D (L)
valves, DILV 2 valves, DIRV Atresia (right)
20.
(20
yr, F)
21.
(14
yr, F)
Inversus
Normal
22.
(7
yr, F)
Solitus
Normal
Common atrium Intact
23.
(I1 yr, F)
Solitus
Normal
Intact
24.
(11
25.
(2
yr, M)
Right isomerism Solitus
Bilateral SVC PAPVC Normal
Common atrium 2” ASD
26.
(3
yr, F)
27.
(28
Situs inversus Solitus
IVC interruption Normal
Common atrium 2’ ASD
28.
(7
mo, M)
Solitus
Normal
Intact
29.
(2
yr, F)
Solitus
Normal
2’
yr, M)
Ventricular loop (apexi
Solitus Right isomerism Solitus
yr, M)
Ventricular morphology
1990
Heart Journal
ASD
valves, DILV AV canal 2
2
valves, DIRV valves
2
D 6)
lo
ASD, Primum atrial septal defect; 2’ ASD, secundum atria! septal defect; A-DORV. double-outlet right ventricle with anteriopasteriorly related great arteries; AV, atrioventricular; BVF, Bulboventricular foramen; D concordant, normally related great arteries with concordant ventriculoarterial connection; D-DORV, double-outlet right ventricle with the pulmonary artery to the right of the aorta; DILV. double-inlet left ventricle; DIRV, double-inlet right ventricle; D Loop, normally related ventricles with the right ventricle to the right of the ventricular septum; IVC, inferior vena cava; (L), leftward directed cardiac apex; L Loop, inverted ventricular relationship with the left ventricle to the right of the ventricular septum; LV, Morphologic left ventricle; L-concordant, pulmonary artery to the left of the aorta but with concordant ventricular arterial connection; L-DOLV, double-outlet left ventricle with the pulmonary artery to the left of the aorta; L-DORV, double-outlet right ventricle with the pulmonary artery to the right of the aorta; L-TGA, transposition of great arteries with aorta arising from right and pulmonary artery from the left ventricle; pulmonary artery lies to the left of the aorta; PA, pulmonary atresia: PAPVR, partial anomalous pulmonary venous return; PS, pulmonic (valvar) stenosis; (R), rightward directed cardiac apex; RV, morphologic right ventricle: Sub AS, subaortic stenosis (outflow tract stenosis); Sub PS, subpulmonic stenosis (outflow tract stenosis); SVC, Superior vena cava; TA, tricuspid atresia; TAPVR, total anomalous pulmonary venous return; VSD, ventricular seutal defect: D-TGA, same as L-TGA except that pulmonary artery lies to the right of the aorta; PAPVC, partial anomalous pulmonary venous connection
and possible correlations were obtainable. However, 52 of these correlations could not be made due to insufficient information from angiography. A total of 209 correlations could be made with the current data (Fig. 1). In 17 instances MR findings differed from angiographic results. These results are presented in Fig. 2. Thoracic and abdominal situs. Thoracic and abdominal situs were defined in all 29 patients with MRI, while angiography provided insufficient information for a conclusive diagnosis in 10 cases and equivocal results in two. The two discrepancies occurred in patients with the angiographic diagnosis of polysplenia. In both patients, inferior vena caval interruption was detected by MRI, but since neither a spleen nor splenules were visualized, the diagnosis of “asplenia” was made. Systemic
and
pulmonary
venoatrial
connections.
Systemic and pulmonary venoatrial connections were displayed with MRI in all 29 patients. Systemic
venous anomalies were present in 10 patients and pulmonary venous anomalies were present in six; five patients had both types of anomalies. Venoatrial connections were defined with both techniques in 14 cases; among these the only discrepancy occurred in a patient with anomalous venous return that was interpreted as partially anomalous by MRI and as totally anomalous by angiography. MRI was interpreted as demonstrating the right pulmonary veins draining into the right side of the common atrium, while angiography, with a selective injection into the innominate vein, visualized a common pulmonary venous trunk draining into the right innominate vein. Ventricular
and
atrloventrlcular
valve
morphology.
Ventricular morphology was demonstrated in every patient with sufficient detail on MRI studies to permit evaluation of the ventricular loop, number, and type of AV valves, as well as their ventricular commitment and status, and the orientation of the
Volume Number
120 1
MRI of complex
ventricular
anomalies
137
Ventricular septum
Great arterial relationship
Semilunar ualuular and subvalvar stenosis
BVF
L-TGA
Normal Normal
VSD
L-TGA, PA D concordant PA L-DORV
VSD
L-DORV
PS
double-outlet right ventricle. In one of these, angiography diagnosed transposition of the great arteries, while in the other the additional diagnosis of pulmonary atresia was made by angiography. Semilunar valves. The majority of discrepancies occurred in the evaluation of ventricular outflow tracts and semilunar valves. In 16 cases, MRI diagnosed stenoses of the right ventricular outflow tract or the pulmonic valve correctly in comparison with angiography (Fig. 5). In five cases angiography demonstrated valvar or subvalvar stenoses (four pulmonic stenoses, one aortic stenosis) that were not identified with MRI imaging.
Intact
D concordant PA D-TGA PA
Normal
DISCUSSION
D concordant
Sub PS
L-TGA
PS
VSD
D-DORV
Normal
BVF
L-TGA
Sub AS
VSD
D concordant
PS
Inflow
VSD
BVF
VSD
VSD Inflow
VSD
Normal Sub PS
Normal
ventricular septum (Figs. 3 and 4). Ventricular morphology and ventricular loop were demonstrated with both techniques in 25 patients. The six discrepancies among these cases concerned the assignments of the type of dominant chamber in two patients and the type of ventricular loop in four cases. AV valve morphology was described consistently in 26 of the 27 patients with the two techniques. The single discrepancy occurred in one patient in whom the MRI diagnosis was right AV valve atresia and the angiographic diagnosis was that both valves were present. Echocardiography in this patient depicted only one AV valve. Atrial and ventricular septa. The depiction of the normal atrial septum (five cases) and of atrial septal defects (24 cases) was possible by MRI in 29 cases and by angiography in 19 cases. There were no discrepancies between MRI and angiography with regard to atrial septal defects. The ventricular septum and anomalies (27 cases) of the ventricular septum were assessed by MRI in 29 cases and by angiography in 27 cases without discrepant observations between the two techniques. Great vessel relationship. Number, spatial relationship, and ventricular connection of the great arteries were identified in 29 patients with MRI (Fig. 4), while angiography did not provide sufficient detail in four patients. The two discrepancies between techniques were found in patients with MRI characteristics of
The current study assessed the effectiveness of MRI imaging for the anatomic assessment of 29 patients with complex congenital ventricular anomalies. MRI identified in all patients the anatomic features required to define the complex morphology in a segmental approach. In an attempt to determine the accuracy of MRI findings, angiography, as the only universally available verifying test in patients with congenital heart disease, was employed to establish the true diagnosis. However, the role of angiography as a “gold standard” was compromised, since in several cases MRI provided more information than the respective angiographic study. In the current study, 52 correlations between angiography and MRI could not be made due to insufficient information from angiographic studies. This problem was caused in part by the retrospective nature of the current study. In some instances technical factors, such as the lack of specific contrast injections to obtain venograms, resulted in the inability to define venoatrial connections. Moreover, frequently angiography did not provide enough information on visceral position and orientation for the definition of complex visceroatrial situs. The current study shows that if only studies with complete diagnoses using both techniques are compared, MRI depicts most aspects of cardiovascular segmental anatomy in a detailed manner that is at least equivalent to that of angiography. This has also been suggested by previous studies.7> ’ A third technique, such as surgery or postmortem examination, would have been desirable to have a more rigorous standard for the accuracy of MRI diagnoses. In this study, however, many of the patients were not amenable to corrective surgery due to the complexity of their lesions. Echocardiographic findings were consulted in an attempt to clarify discrepancies between MRI and angiographic findings. In some of the cases this proved to be helpful, as noted in the Results section; in other cases, however, the
138
Kersting-Sommerhoff
et al.
American
Total
t?ARl
July 1990 Heart Journal
Angiography
Fig. 1. Block diagram comparing the total number of possible observations (column 1, n = 261) with the actual number of observationsthat could be made by MRI (column 2, n = 261) and angiography (column
3, n = 209).
tn Ti *G= a” 20 b ft 10
Agreement Discrepancy
2
Fig.
MRI
2. Block diagram comparingthe number of agreementsand disagreementsbetweenangiography and observationsfor all nine features that were evaluated.
true diagnosis could not be established conclusively. While a direct comparison of MRI and echocardiography would be interesting, this was not the intent of the current study. Presumably, such a comparison will be done in the future. However, the current study
indicates potential problems in using angiography as the arbiter between the results of noninvasive imaging techniques. The major advantage of MRI imaging in congenital heart disease is the tomographic mode. Cardiac
Volume Number
120 1
MRI of complex ventricular
anomalies
Fig. 3. A to C, Patient No. 21 with situs inversus, commonatrium, tricuspid atresia, and double-outlet right ventricle. The transverse images(A to C) depict the situs inversus with the apex and the spleen (S) on the right side, while the liver (L) is located on the left side of the body. K, Kidney. At the baseof the heart, aorta (A) and pulmonary artery (P) are located nearly side by side. 6 demonstratesthe dominant left ventricular chamber (LC) located inferiorly and the smallright ventricular chamber (RC) in an anterosuperior position. Aorta and pulmonary artery originate from the right ventricular outflow chamber (double-outlet right ventricle). CA, Common atrium.
Fig. 3. D and E, Coronal MR imagesof patient No. 21 (seelegend to Fig. 3 A to C). Curved arrow passes through the bulboventricular foramen and points to the site of subpulmonicstenosis.Abbreviations asin Fig. 3A to C.
139
July
140
Kersting-Sommerhoff
et al.
American
1990
Heart Journal
Fig. 4. Transverse (A) and coronal (B through D) imagesof patient No. 18 with single ventricle and Ltransposition of the great arteries. At the baseof the heart (A), the aorta (A) is located anterior and to the left of the pulmonary artery (P) that originates from the dominant chamber. The coronal imagesdepict the dominant left ventricular-type chamber (LC) and the origin of the pulmonary artery. The dominant chamber communicateswith the rudimentary right chamber (RC) that gives rise to the aorta through a bulboventricular foramen (curued arrow, C). RA, Right atrium.
anatomy is depicted without the superimposition of structures in other planes that occurs with a projectional technique such as angiography. Although axial angiography addresses this difficulty, it requires considerable skill on the part of the angiographic team to achieve optimal delineation of cardiac structures. Similarily, the quality of an echocardiographic study depends entirely on the experience and skill of the examiner. The tomographic nature of MRI, on the other hand, is inherent and requires no special maneuvering. MRI therefore offers the advantage of a more uniform technique when practiced by different physicians. While advantages have been proclaimed for the use of MR images in planes parallel to the cardiac long and short axes, no studies have actually systematically compared these planes with
the standard orthogonal planes. Significant advantages have never been proven. These cardiac axis planes are more time-consuming to obtain, and for technical reasons frequently have decreased image quality compared with the orthogonal planes. The results of the current study indicate that MRI can be as precise as angiography in the depiction of ventricular anomalies, including determination of morphology and evaluation of the size of the ventricles, the orientation of the ventricular septum relative to the AV valves, as well as in determining the origins and spatial relationships of the great arteries. MRI appears more informative for the determination of situs and systemic and pulmonary venoatrial connections, areas in which angiography may not provide sufficient information for a diagnosis without
Volume Number
120 1
MRI
of
complex ventricular anomalies
14 1
Fig. 5. Patient No. 29 with situs solitus and tricuspid atresia. Transverse images(A and B) showhigh signal intensity fatty tissue and myocardium extending from the atrioventricular groove into the expected location of the tricuspid valve (large arrow). The hypoplastic right ventricle (RV) as well as the hypertrophied left ventricle (LV) can be seenon both transverse and coronal (C and D) images.Note the infundibular pulmonic stenosis(arrow) in imageB and the dilated inferior vena cava (WC) entering the right atrium in image D. RVOT, Right ventricular outflow tract.
additional injections. Venoatrial venous connections can nearly always be determined from transverse MR images.g Transverse and coronal images permit diagnosis of mainstem bronchial anatomy as well as the relationship of the central pulmonary arteries to the bronchi (epi- or hyparterial). MRI appears to be at least as accurate in the diagnosis of the number and the size of AV valves, and of atrial and ventricular septal defects.3p ’ The major limitation of spin-echo MRI in the evaluation of congenital heart disease is the functional evaluation of semilunar valves and ventricular outflow tracts. The diagnosis of these lesions by spin-echo MRI is based largely on secondary signs such as the identification of thickened valves, ventricular hypertrophy, or annular narrowing and/or
dilated pulmonary arteries. At present, a combination of MRI and echocardiography in the clinical setting would address this difficulty and provide anatomic as well as functional information noninvasively. Furthermore, gradient refocused (tine) MR1241 25 and velocity-encoded tine MR126 will broaden MRI capabilities beyond the morphologic assessment of cardiovascular anomalies. REFERENCES
HigginsCB, Byrd BF, FarmerDW, OsakiL, SilvermanNH, Cheitlin MD. Magneticresonance imagingin patientswith congenitalheart &ease. Circulation1984;70:851-60. 2. FletcherBD, JacobsteinMD, NelsonAD, Riemenschneider TA, Alfidi RJ. Gatedmagneticresonance imagingof congenital cardiacmalformations. Radiology1984;150:137-40. 3. Didier D, HigginsCB, FisherMR, OsakiL, SilvermanNH, 1.
142 Kersting-Sommerhojf
4. 5. 6. I.
8.
9.
10. 11. 12.
13.
et al.
Cheitlin MD. Congenital heart disease: gated MR imaging in 72 patients. Radiology 1986;158:227-35. Mirowitz SA, Gutierrez FA, Canter CE, Vannier MW. Tetralogy of Fallot: MR findings. Radiology 1989;171:207-12. Link KM, Herrera MA, D’Souza VJ, Formanek AG. MR imaging of Ebstein’s anomaly: results in four cases. AJR 1988;150:363-7. Akins EW, Martin TD, Alexander JA, et al. MR imaging of double outlet right ventricle. AJR 1989;152:128-30. Kersting-Sommerhoff BA, Diethelm L, Teitel DF, et al. Magnetic resonance imaging of congenital heart disease: sensitivity and specificity using receiver operating characteristic curve analysis. AM HEART J 1989;188:155-63. Kersting-Sommerhoff BA, Sechtem UP, Higgins CB. Evaluation of pulmonary blood supply by nuclear magnetic resonance imaging in patients with pulmonary atresia. J Am Co11Cardiol 1988;11:166-71. Diethelm L, Dery R, Lipton MJ, Higgins CB. Sensitivity and specificity of magnetic resonance imaging in the identification and localization of atral level shunts. Radiology 1987;162: 181-6. Guit GL, Bluemm R, Rohmer J, et al. Levotransposition of the aorta: identification and segmental anatomy using MR imaging. Radiology 198&161:673-g. Van Praagh R. The importance of segmental situs in the diagnosis of congenital heart disease. Semin Roentgen01 1985; 2%254-71. Anderson RH, Macartney FJ, Tynan M, Becker AE, Freedom RM. Godman MJ. et al. Univentricular atrioventricular connection: the single ventricle trap unsprung. Pediatr Cardiol 1984;4:273-80. Anderson RH, Becker AE, Tynan M, Macartney FJ, Rigby ML, Wilkinson JL. The univentricular atrioventricular connection: getting to the root of a thorny problem. Am J Cardiol 1984:54:822-8.
American
July 1990 Heart Journal
14. Van Praagh R, David I, Van Praagh S. What is a ventricle: the single ventricle trap. Pediatr Cardiol 1984;2:79-84. 15. Van Praaeh R. Onelev PA. Swan HJC. Anatomic tvoes of single or co&mon ven%les in man. Morphologic and geometric aspects of 60 necropsied cases. Am J Cardiol 1964;14:367-86. 16. Macartney FJ, Partridge JB, Scott 0, Deverall PB. Common or single ventricle. An angiographic and hemodynamic study of 42 patients. Circulation 1976;53:543-54. 17. Lev M, Liberthson RR, Kirkpatrick JR, Eckner FAO, Arcilla RA. Single (primitive) ventricle. Circulation 1969;39:57’7-91. 18. Soto B, Pacific0 AD, Sciasco GD. Univentricular heart: an angiographic study. Am J Cardiol 1982;49:787-94. 19. Thies WR, Soto B, Diethelm E, Bargeron LM, Pacific0 AD. Angiographic anatomy of hearts with one ventricular chamber: the true single ventricle. Am J Cardiol 1985;55:1363-6. 20. Stanger P, Rudolph AM, Edwards JE. Cardiac malpositions. An overview baaed on study of sixty-five necropsy specimens. Circulation 1977;56:159-72. 21. Becker AE, Anderson RH. Pathology of congential heart disease. London: Butterworth & Co, Ltd, 1981:77-92. 22. Baron MG. Endocardial cushion defects. Radio1 Clin North Am 1968;3:343-60. 23. Jacobstein MD, Fletcher BD, et al. Evaluation of atrioventricular septal defects by magnetic resonance imaging. Am J Cardiol 1985;55:1158-61. 24. Higgins CB, Sechtem UP, Pflugfelder P. Cine MR: evaluation of cardiac ventricular function and valvular function. Int J Card Imaging 1988;3:21-8. 25. Sechtem U, Pflugfelder P, Cassidy MC, White RD, Cheitlin M, Schiller ND. Quantification of regurgitant volumes in patients with mitral or aortic regurgitation by tine MRI. Radiology 1988;167:425-30. 26. Pelt NJ, Shimakawa A, Glover GH. Phase contrast tine MRI. Society of Magnetic Resonance Imaging. 1989;l:lOl.
of complex congenital ventricular with magnetic resonance imaging
Complex ventricular anomalies are frequently associated with abnormalities of thoracic and abdominal situs, arterioventricular connection, and venous connection. The definition of all components of these anomalies is difficult to accomplish with imaging techniques. This study compared the effectiveness of electrocardiographic (ECG) gated spin-echo magnetic resonance imaging (MRI) with cardiac angiography for the evaluation of all components of central cardiovascular anatomy in patients with the clinical diagnosis of single or common ventricle or complete atrioventrlcular (AV) septal (canal) defect. MRI studies and angiograms of 29 patients were evaluated independently. A sequential approach was used to define cardiac anatomy assessing nine anatomic features in each patient. MRI provided 261 observations and angiography provided 209 observations. In the mutual 209 observations, only 17 discrepancies were found. Comparison of MRI and angiography in individual cases showed that MRI was as effective as angiography in the depiction of ventricular anomalies, including determination of morphology and evaluation of the size of the ventricles, the orientation of the ventricular septum relative to the AV valves, as well as the origins and spatial relationships of the great arteries. MRI was more informative for the determination of thoracic and abdominal situs and systemic and pulmonary venoatrial connections, but was not as effective for the evaluation of semilunar valves. Thus MRI provides complete evaluation of central cardiovascular anatomy and is effective in the anatomic assessment of most components of complex ventricular anomalies. (AM HEART J 1990;120:133.)
Barbara A. Kersting-Sommerhoff, MD, Lisa Diethelm, MD, Paul Stanger, MD, Renee Dery, MD, Stanley M. Higashino, MD, Sarah S. Higgins, MS, and Charles B.Higgins, MD. San Francisco and Oakland, C&if.
Complex ventricular anomalies have proved to be a diagnostic challenge for angiography as well as for noninvasive imaging techniques. Angiography is the standard imaging technique used for definitive diagnosis and preoperative assessment, but superimposition of cardiac structures can be problematic in these complex cases when a projectional imaging technique is employed. Complex ventricular anomalies are frequently associated with abnormalities of thoracic and abdominal situs, arterioventricular connection, and venous connection, so that definition of all components of these anomalies requires imaging of the entire central cardiovascular anatomy. Electrocardiographic (ECG) gated magnetic resonance imaging (MRI) has been used for the diagnosis of many congenital anomalies of the heart.‘-I2 The From the Departments of Radiology and Pediatrics (Cardiology) of the University of California Medical Center, San Francisco, and the Department of Cardiology in the Childrens Hospital Medical Center, Oakland. Received
for publication
Reprint University
requests: Charles of California,
4l1120435
Jan.
16, 1990;
accepted
Feb.
B. Higgins, MD, Department San Francisco, CA 94143.
26, 1990. of Radiology,
L 308,
ability to image in multiple orthogonal planes and the high natural contrast between rapidly flowing blood and internal cardiac structures are useful attributes of this technique. The effectiveness of MRI in comparison with angiography in the complete definition of complex cardiovascular anomalies has not been evaluated previously. The purpose of the current study was to assess the effectiveness of MRI in the diagnosis of complex congenital ventricular anomalies such as single ventricle and atrioventricular septal (canal) defects. In an attempt to determine the accuracy of MRI diagnoses, angiography was used as the “truth standard” and MRI diagnoses were correlated with angiographic findings in all patients. MRI and angiography were compared in their effectiveness for defining all components of central cardiovascular anatomy in these complex lesions. METHODS Twenty-nine patients (17 males and 12 females) with the clinical diagnosis of single or common ventricle or complete atrioventricular septal (canal) defect who had not under133
134
Kersting-Sommerhoff
Table
I. MRI findings
Patient No. (age, sex)
et al.
Thoracic
situ.5
American
Venoatrial connections
Atria1 septum
Ventricular morphology
loop
AV valve morphology
(23 yr, M) (7 yr, F) (30 yr, F) (16 yr, F) (9 yr, M)
Solitus Solitus Solitus Solitus Solitus
Normal Normal Normal Normal Normal
2’ ASD 2” ASD Intact Intact 2O ASD
Single Single Single Single Single
6. (12 yr, F) 7. (3 mo, F) 8. (17 yr, M)
Solitus Solitus Solitus
Normal Normal Normal
2 ventricles 2 ventricles 2 ventricles
D 0.4 D 6) D (L)
Atresia (right) AV canal AV canal
9. (10 yr, M)
Inversus
Bilateral SVC
Atresia (right)
Bilateral SVC
2 ventricles RV dominant Single LV
L CR)
Right isomerism Right isomerism Right isomerism Left isomerism Left isomerism
2’ ASD 1” ASD Common atrium Common atrium Common atrium lo ASD
D @A
2 ventricles
L (JJ
Common valve AV canal
2 ventricles
L (R)
AV canal
2 ventricles
D 03)
AV canal
2 ventricles LV dominant
L 6)
AV canal
2“ ASD
2 ventricles
D (RI
2 valves
1’ ASD
2 ventricles
L CR)
AV canal
Common atrium
2 ventricles, LV dominant
D (RI
AV canal
1. 2. 3. 4. 5.
10. (16 yr, M) 11. (11 yr, M) 12. (16 yr, M) 13. (20 yr, M) 14. (19 yr, M) 15. (9 yr, M)
Left isomerism
16. (5 yr, M)
Left isomerism
17. (16 yr, M)
Left isomerism
Bilateral
SVC
TAPVR
Common atrium
Normal IVC interruption PAPVR IVC interruption PAPVR IVC interruption PAPVR Bilateral SVC
atrium lo ASD
gone repair were studied with ECG gated MRI. The patients ranged in agefrom 3 months to 30 years. Six children were 3 years old or younger. Sedation was generally necessaryin children under 7 years of ageand was administered as90 to 100 mg/kg of chloral hydrate orally 30 minutes prior to the study or 5 to 6 mglkg of pentobarbital sodium intramuscularly (maximum dose100 mg). Imageswere acquired using a cryogenic 0.35 T superconducting magnet (Diasonics MT/S, Milpitas, Calif.). First echocardiographic imageswith an echo delay time (TE) of 28 to 30 msecwere obtained. In somepatients second echo (TE = 56 to 60 msec)were acquired additionally. Since ECG gating wasapplied in all patients, the repetition time (TR) equaledthe R-R interval and dependedon the patient’s heart rate. Tomograms in the transverse plane were obtained in all patients and additional imagesin the sagittal or coronal plane were acquired depending on the findings on the initial transverse run. Multiple contiguous slicesencompassingthe entire heart and extending from the aortic arch through the upper abdomenwere acquired in all subjects.Slice thickness was 10 mm in adults and 5 mm in most children.
LV LV LV LV LV
Ventricular (apex)
July 1990 Heart Journal
L L L L L
6) (L) (L) 6) UJ
Atresia (left) Atresia (left) 2 valves 2 valves 2 valves
Angiographic studieswereavailable for all patients; they were done at two different institutions. Each study included one or more injections into the dominant ventricular chamber and the angiogramswere recorded on 35 mm cineangiography film at 30 frames/set. Image analysis. MRI studiesand angiogramswereevaluated independently by two different teams of observers. The observerswere awareof the generaldiagnosisbut had no knowledge of anatomic details or of the results of the other imaging study. The teams operated as a consensus panel and analyzed the cardiac anatomy using a sequential approach.lO,l1 The morphology of nine components was defined: (1) visceroatrial situs, (2) systemicand pulmonary venoatrial connections, (3) atria1 septum, (4) ventricular morphology, (5) ventricular loop and direction of the cardiac apex, (6) atrioventricular (AV) valve morphology, (7) ventricular septum, (8) spatial relationship of the great arteries and arterioventricular connections,and (9) the status of the semilunar valves and ventricular outflow tracts. Image quality of the MRI studies was assessed using a system describedpreviously.3 “Excellent” quality images provided distinct delineation of internal cardiac structures
Volume Number
120 1
Ventricular septum
MRI of complex
Great arterial relationship
Semilunar valvular and
subvalvar stenosis
Sub PS Normal Sub PS Normal Normal
VSD Inflow VSD Inflow VSD
L-TGA L-TGA L-TGA L-TGA L-TGA PA D-DORV D concordant D concordant
VSD
L-DORV
Normal
BVF
A-DORV
Sub PS
Inflow VSD
L-DORV
Normal
Inflow VSD
L-DORV
Sub PS
Inflow VSD
D-DORV
Normal
Inflow VSD
L concordant
Normal
VSD
L concordant
PS Sub PS
Inflow VSD
L-DORV
Normal
Inflow VSD
L-DOLV
PS
BVF BVF BVF BVF BVF
PS Normal Sub PS
Sub PS
and myocardium without signal dropout from any regions of the atria1 or ventricular walls. In “diagnostic” images, occasionallossof signal or noisein the phaseencoding direction degraded overall quality, but cardiac structures were clearly delineated. In “barely diagnostic” images,the diagnosiscould still be made, but visualization of the cardiac structures wasobscuredby signallossor noise.“Nondiagnostic” images did not permit adequate diagnosis because cardiac chambers and great vessels were not discernible. Terminology and morphologic criteria. Some debate exists regarding the terminology for complex congenital ventricular anomalies.12-1g Since the aim of this study was not to classify these complex anomaliesbut to assessthe role of MRI in their evaluation, controversial terms were avoided and a nomenclature proposed previously by &anger et al.” wasused.The following terms were usedto describe the ventricular anomaliesin this study. “Single ventricle” hasboth or a commonAV valve(s) entering one ventricle. When this is a morphologically left ventricle, there commonly exists a rudimentary right ventricular chamber and the two are connected by a bulboventricular
ventricular anomalies I 35
foramen. When the dominant ventricle is of right ventricular morphology, there may only be a trabecular pouch or no secondventricular chamber.“Common ventricle” hasa rudimentary rim of septum separating two ventricular cavities with an AV valve entering the inflow portion of each, or a commonAV valve entering both cavities.15,l6 “Complete atrioventricular septal (canal) defect” consistsof a primum atria1septal defect, a commonAV valve, and a large inflow ventricular septal defect.21p 23The common valve may be more committed to one of the ventricles, with that chamber being the dominant one, implying that the other ventricle is smaller than normal. “AV valve atresia” is associatedwith an ipsilateral hypoplastic ventricle. The contralateral ventricle is normal or enlarged. The presenceor absenceof an infundibulum was the major criterion usedto identify ventricular morphology on MRI tomograms. The presenceof an infundibulum was considered to be characteristic of a right ventricle; when there wasno detectable musclebetween the AV valve and the adjacent semilunar valve, the chamberwasconsidered to be a morphologic left ventricle. The “bulboventricular loop,” according to Van Praagh et al.)5 wasdescribedas D-loop when the morphologic right ventricle was located to the right of the morphologic left ventricle; conversely, when the morphologic right ventricle waslocated to the left of the morphologic left ventricle, L-loop was present. “Ventricular septal defects” were classifiedaseither part of an AV septal (canal) defect, a bulboventricular foramen in single ventricle, or unspecified. Subclassification of “atria1 septal defects” included primum (AV canal), secundum, or sinus venosusdefect or common atrium, and wasaccording to criteria previously established.g Classification of “thoracic situs” was inferred from the main stem bronchial
anatomy
as well es from the relation-
ship of the central pulmonary arteries to the bronchi (epior hyparterial). Right isomerismis characterized by bilateral right-sidedness (eparterial bronchi); left isomerism refers to bilateral left-sidedness(hyparterial bronchi). The abdominal visceral situs was inferred from the location of stomach and liver, the presence or absence of spleen (asplenia, polysplenia), and the aortocaval relationship (ipsi- or contralateral). RESULTS
Sixteen of the MRI studies were rated “excellent” quality; 13 were of “diagnostic” quality. However, for two patients the initial study was considered nondiagnostic and was repeated, in one case with sedation. These second studies were rated to be “excellent” and “diagnostic,” respectively. The major factors affecting quality were somatic motion or respiratory degradation of the signal. The findings of MRI for each patient with respect to all nine features are presented in Table I. Since nine anatomic features were assessed on both MRI studies and angiograms in each patient, a total number of 261 mutual observations
July
136
Kersting-Sommerhoffet al.
Table
I. cont.
Patient No. (age, sex) 18.
(5
yr, F)
19. (1 yr, M)
Thoracic situs
American
Venoatrial connections
Atria1 septum
AV valve morphology
Normal
Intact
Single LV
L CL)
Bilateral SVC PAPVC Normal
Common atrium 2’ ASD
2 ventricles
L (R)
Single LV
D 6)
2 ventricles, LV dominant 2 ventricles, RV dominant 2 ventricles, LV dominant Single RV
I. (R)
valves, DILV Atresia (right)
D (L)
2
D (L)
2
D (R)
Atresia (right)
2 ventricles, LV dominant 2 ventricles
D CL)
Atresia (left)
L (RI
AV canal
2 ventricles RV dominant Single LV
D NJ
2 ventricles, LV dominant
D (L)
valves, DILV 2 valves, DIRV Atresia (right)
20.
(20
yr, F)
21.
(14
yr, F)
Inversus
Normal
22.
(7
yr, F)
Solitus
Normal
Common atrium Intact
23.
(I1 yr, F)
Solitus
Normal
Intact
24.
(11
25.
(2
yr, M)
Right isomerism Solitus
Bilateral SVC PAPVC Normal
Common atrium 2” ASD
26.
(3
yr, F)
27.
(28
Situs inversus Solitus
IVC interruption Normal
Common atrium 2’ ASD
28.
(7
mo, M)
Solitus
Normal
Intact
29.
(2
yr, F)
Solitus
Normal
2’
yr, M)
Ventricular loop (apexi
Solitus Right isomerism Solitus
yr, M)
Ventricular morphology
1990
Heart Journal
ASD
valves, DILV AV canal 2
2
valves, DIRV valves
2
D 6)
lo
ASD, Primum atrial septal defect; 2’ ASD, secundum atria! septal defect; A-DORV. double-outlet right ventricle with anteriopasteriorly related great arteries; AV, atrioventricular; BVF, Bulboventricular foramen; D concordant, normally related great arteries with concordant ventriculoarterial connection; D-DORV, double-outlet right ventricle with the pulmonary artery to the right of the aorta; DILV. double-inlet left ventricle; DIRV, double-inlet right ventricle; D Loop, normally related ventricles with the right ventricle to the right of the ventricular septum; IVC, inferior vena cava; (L), leftward directed cardiac apex; L Loop, inverted ventricular relationship with the left ventricle to the right of the ventricular septum; LV, Morphologic left ventricle; L-concordant, pulmonary artery to the left of the aorta but with concordant ventricular arterial connection; L-DOLV, double-outlet left ventricle with the pulmonary artery to the left of the aorta; L-DORV, double-outlet right ventricle with the pulmonary artery to the right of the aorta; L-TGA, transposition of great arteries with aorta arising from right and pulmonary artery from the left ventricle; pulmonary artery lies to the left of the aorta; PA, pulmonary atresia: PAPVR, partial anomalous pulmonary venous return; PS, pulmonic (valvar) stenosis; (R), rightward directed cardiac apex; RV, morphologic right ventricle: Sub AS, subaortic stenosis (outflow tract stenosis); Sub PS, subpulmonic stenosis (outflow tract stenosis); SVC, Superior vena cava; TA, tricuspid atresia; TAPVR, total anomalous pulmonary venous return; VSD, ventricular seutal defect: D-TGA, same as L-TGA except that pulmonary artery lies to the right of the aorta; PAPVC, partial anomalous pulmonary venous connection
and possible correlations were obtainable. However, 52 of these correlations could not be made due to insufficient information from angiography. A total of 209 correlations could be made with the current data (Fig. 1). In 17 instances MR findings differed from angiographic results. These results are presented in Fig. 2. Thoracic and abdominal situs. Thoracic and abdominal situs were defined in all 29 patients with MRI, while angiography provided insufficient information for a conclusive diagnosis in 10 cases and equivocal results in two. The two discrepancies occurred in patients with the angiographic diagnosis of polysplenia. In both patients, inferior vena caval interruption was detected by MRI, but since neither a spleen nor splenules were visualized, the diagnosis of “asplenia” was made. Systemic
and
pulmonary
venoatrial
connections.
Systemic and pulmonary venoatrial connections were displayed with MRI in all 29 patients. Systemic
venous anomalies were present in 10 patients and pulmonary venous anomalies were present in six; five patients had both types of anomalies. Venoatrial connections were defined with both techniques in 14 cases; among these the only discrepancy occurred in a patient with anomalous venous return that was interpreted as partially anomalous by MRI and as totally anomalous by angiography. MRI was interpreted as demonstrating the right pulmonary veins draining into the right side of the common atrium, while angiography, with a selective injection into the innominate vein, visualized a common pulmonary venous trunk draining into the right innominate vein. Ventricular
and
atrloventrlcular
valve
morphology.
Ventricular morphology was demonstrated in every patient with sufficient detail on MRI studies to permit evaluation of the ventricular loop, number, and type of AV valves, as well as their ventricular commitment and status, and the orientation of the
Volume Number
120 1
MRI of complex
ventricular
anomalies
137
Ventricular septum
Great arterial relationship
Semilunar ualuular and subvalvar stenosis
BVF
L-TGA
Normal Normal
VSD
L-TGA, PA D concordant PA L-DORV
VSD
L-DORV
PS
double-outlet right ventricle. In one of these, angiography diagnosed transposition of the great arteries, while in the other the additional diagnosis of pulmonary atresia was made by angiography. Semilunar valves. The majority of discrepancies occurred in the evaluation of ventricular outflow tracts and semilunar valves. In 16 cases, MRI diagnosed stenoses of the right ventricular outflow tract or the pulmonic valve correctly in comparison with angiography (Fig. 5). In five cases angiography demonstrated valvar or subvalvar stenoses (four pulmonic stenoses, one aortic stenosis) that were not identified with MRI imaging.
Intact
D concordant PA D-TGA PA
Normal
DISCUSSION
D concordant
Sub PS
L-TGA
PS
VSD
D-DORV
Normal
BVF
L-TGA
Sub AS
VSD
D concordant
PS
Inflow
VSD
BVF
VSD
VSD Inflow
VSD
Normal Sub PS
Normal
ventricular septum (Figs. 3 and 4). Ventricular morphology and ventricular loop were demonstrated with both techniques in 25 patients. The six discrepancies among these cases concerned the assignments of the type of dominant chamber in two patients and the type of ventricular loop in four cases. AV valve morphology was described consistently in 26 of the 27 patients with the two techniques. The single discrepancy occurred in one patient in whom the MRI diagnosis was right AV valve atresia and the angiographic diagnosis was that both valves were present. Echocardiography in this patient depicted only one AV valve. Atrial and ventricular septa. The depiction of the normal atrial septum (five cases) and of atrial septal defects (24 cases) was possible by MRI in 29 cases and by angiography in 19 cases. There were no discrepancies between MRI and angiography with regard to atrial septal defects. The ventricular septum and anomalies (27 cases) of the ventricular septum were assessed by MRI in 29 cases and by angiography in 27 cases without discrepant observations between the two techniques. Great vessel relationship. Number, spatial relationship, and ventricular connection of the great arteries were identified in 29 patients with MRI (Fig. 4), while angiography did not provide sufficient detail in four patients. The two discrepancies between techniques were found in patients with MRI characteristics of
The current study assessed the effectiveness of MRI imaging for the anatomic assessment of 29 patients with complex congenital ventricular anomalies. MRI identified in all patients the anatomic features required to define the complex morphology in a segmental approach. In an attempt to determine the accuracy of MRI findings, angiography, as the only universally available verifying test in patients with congenital heart disease, was employed to establish the true diagnosis. However, the role of angiography as a “gold standard” was compromised, since in several cases MRI provided more information than the respective angiographic study. In the current study, 52 correlations between angiography and MRI could not be made due to insufficient information from angiographic studies. This problem was caused in part by the retrospective nature of the current study. In some instances technical factors, such as the lack of specific contrast injections to obtain venograms, resulted in the inability to define venoatrial connections. Moreover, frequently angiography did not provide enough information on visceral position and orientation for the definition of complex visceroatrial situs. The current study shows that if only studies with complete diagnoses using both techniques are compared, MRI depicts most aspects of cardiovascular segmental anatomy in a detailed manner that is at least equivalent to that of angiography. This has also been suggested by previous studies.7> ’ A third technique, such as surgery or postmortem examination, would have been desirable to have a more rigorous standard for the accuracy of MRI diagnoses. In this study, however, many of the patients were not amenable to corrective surgery due to the complexity of their lesions. Echocardiographic findings were consulted in an attempt to clarify discrepancies between MRI and angiographic findings. In some of the cases this proved to be helpful, as noted in the Results section; in other cases, however, the
138
Kersting-Sommerhoff
et al.
American
Total
t?ARl
July 1990 Heart Journal
Angiography
Fig. 1. Block diagram comparing the total number of possible observations (column 1, n = 261) with the actual number of observationsthat could be made by MRI (column 2, n = 261) and angiography (column
3, n = 209).
tn Ti *G= a” 20 b ft 10
Agreement Discrepancy
2
Fig.
MRI
2. Block diagram comparingthe number of agreementsand disagreementsbetweenangiography and observationsfor all nine features that were evaluated.
true diagnosis could not be established conclusively. While a direct comparison of MRI and echocardiography would be interesting, this was not the intent of the current study. Presumably, such a comparison will be done in the future. However, the current study
indicates potential problems in using angiography as the arbiter between the results of noninvasive imaging techniques. The major advantage of MRI imaging in congenital heart disease is the tomographic mode. Cardiac
Volume Number
120 1
MRI of complex ventricular
anomalies
Fig. 3. A to C, Patient No. 21 with situs inversus, commonatrium, tricuspid atresia, and double-outlet right ventricle. The transverse images(A to C) depict the situs inversus with the apex and the spleen (S) on the right side, while the liver (L) is located on the left side of the body. K, Kidney. At the baseof the heart, aorta (A) and pulmonary artery (P) are located nearly side by side. 6 demonstratesthe dominant left ventricular chamber (LC) located inferiorly and the smallright ventricular chamber (RC) in an anterosuperior position. Aorta and pulmonary artery originate from the right ventricular outflow chamber (double-outlet right ventricle). CA, Common atrium.
Fig. 3. D and E, Coronal MR imagesof patient No. 21 (seelegend to Fig. 3 A to C). Curved arrow passes through the bulboventricular foramen and points to the site of subpulmonicstenosis.Abbreviations asin Fig. 3A to C.
139
July
140
Kersting-Sommerhoff
et al.
American
1990
Heart Journal
Fig. 4. Transverse (A) and coronal (B through D) imagesof patient No. 18 with single ventricle and Ltransposition of the great arteries. At the baseof the heart (A), the aorta (A) is located anterior and to the left of the pulmonary artery (P) that originates from the dominant chamber. The coronal imagesdepict the dominant left ventricular-type chamber (LC) and the origin of the pulmonary artery. The dominant chamber communicateswith the rudimentary right chamber (RC) that gives rise to the aorta through a bulboventricular foramen (curued arrow, C). RA, Right atrium.
anatomy is depicted without the superimposition of structures in other planes that occurs with a projectional technique such as angiography. Although axial angiography addresses this difficulty, it requires considerable skill on the part of the angiographic team to achieve optimal delineation of cardiac structures. Similarily, the quality of an echocardiographic study depends entirely on the experience and skill of the examiner. The tomographic nature of MRI, on the other hand, is inherent and requires no special maneuvering. MRI therefore offers the advantage of a more uniform technique when practiced by different physicians. While advantages have been proclaimed for the use of MR images in planes parallel to the cardiac long and short axes, no studies have actually systematically compared these planes with
the standard orthogonal planes. Significant advantages have never been proven. These cardiac axis planes are more time-consuming to obtain, and for technical reasons frequently have decreased image quality compared with the orthogonal planes. The results of the current study indicate that MRI can be as precise as angiography in the depiction of ventricular anomalies, including determination of morphology and evaluation of the size of the ventricles, the orientation of the ventricular septum relative to the AV valves, as well as in determining the origins and spatial relationships of the great arteries. MRI appears more informative for the determination of situs and systemic and pulmonary venoatrial connections, areas in which angiography may not provide sufficient information for a diagnosis without
Volume Number
120 1
MRI
of
complex ventricular anomalies
14 1
Fig. 5. Patient No. 29 with situs solitus and tricuspid atresia. Transverse images(A and B) showhigh signal intensity fatty tissue and myocardium extending from the atrioventricular groove into the expected location of the tricuspid valve (large arrow). The hypoplastic right ventricle (RV) as well as the hypertrophied left ventricle (LV) can be seenon both transverse and coronal (C and D) images.Note the infundibular pulmonic stenosis(arrow) in imageB and the dilated inferior vena cava (WC) entering the right atrium in image D. RVOT, Right ventricular outflow tract.
additional injections. Venoatrial venous connections can nearly always be determined from transverse MR images.g Transverse and coronal images permit diagnosis of mainstem bronchial anatomy as well as the relationship of the central pulmonary arteries to the bronchi (epi- or hyparterial). MRI appears to be at least as accurate in the diagnosis of the number and the size of AV valves, and of atrial and ventricular septal defects.3p ’ The major limitation of spin-echo MRI in the evaluation of congenital heart disease is the functional evaluation of semilunar valves and ventricular outflow tracts. The diagnosis of these lesions by spin-echo MRI is based largely on secondary signs such as the identification of thickened valves, ventricular hypertrophy, or annular narrowing and/or
dilated pulmonary arteries. At present, a combination of MRI and echocardiography in the clinical setting would address this difficulty and provide anatomic as well as functional information noninvasively. Furthermore, gradient refocused (tine) MR1241 25 and velocity-encoded tine MR126 will broaden MRI capabilities beyond the morphologic assessment of cardiovascular anomalies. REFERENCES
HigginsCB, Byrd BF, FarmerDW, OsakiL, SilvermanNH, Cheitlin MD. Magneticresonance imagingin patientswith congenitalheart &ease. Circulation1984;70:851-60. 2. FletcherBD, JacobsteinMD, NelsonAD, Riemenschneider TA, Alfidi RJ. Gatedmagneticresonance imagingof congenital cardiacmalformations. Radiology1984;150:137-40. 3. Didier D, HigginsCB, FisherMR, OsakiL, SilvermanNH, 1.
142 Kersting-Sommerhojf
4. 5. 6. I.
8.
9.
10. 11. 12.
13.
et al.
Cheitlin MD. Congenital heart disease: gated MR imaging in 72 patients. Radiology 1986;158:227-35. Mirowitz SA, Gutierrez FA, Canter CE, Vannier MW. Tetralogy of Fallot: MR findings. Radiology 1989;171:207-12. Link KM, Herrera MA, D’Souza VJ, Formanek AG. MR imaging of Ebstein’s anomaly: results in four cases. AJR 1988;150:363-7. Akins EW, Martin TD, Alexander JA, et al. MR imaging of double outlet right ventricle. AJR 1989;152:128-30. Kersting-Sommerhoff BA, Diethelm L, Teitel DF, et al. Magnetic resonance imaging of congenital heart disease: sensitivity and specificity using receiver operating characteristic curve analysis. AM HEART J 1989;188:155-63. Kersting-Sommerhoff BA, Sechtem UP, Higgins CB. Evaluation of pulmonary blood supply by nuclear magnetic resonance imaging in patients with pulmonary atresia. J Am Co11Cardiol 1988;11:166-71. Diethelm L, Dery R, Lipton MJ, Higgins CB. Sensitivity and specificity of magnetic resonance imaging in the identification and localization of atral level shunts. Radiology 1987;162: 181-6. Guit GL, Bluemm R, Rohmer J, et al. Levotransposition of the aorta: identification and segmental anatomy using MR imaging. Radiology 198&161:673-g. Van Praagh R. The importance of segmental situs in the diagnosis of congenital heart disease. Semin Roentgen01 1985; 2%254-71. Anderson RH, Macartney FJ, Tynan M, Becker AE, Freedom RM. Godman MJ. et al. Univentricular atrioventricular connection: the single ventricle trap unsprung. Pediatr Cardiol 1984;4:273-80. Anderson RH, Becker AE, Tynan M, Macartney FJ, Rigby ML, Wilkinson JL. The univentricular atrioventricular connection: getting to the root of a thorny problem. Am J Cardiol 1984:54:822-8.
American
July 1990 Heart Journal
14. Van Praagh R, David I, Van Praagh S. What is a ventricle: the single ventricle trap. Pediatr Cardiol 1984;2:79-84. 15. Van Praaeh R. Onelev PA. Swan HJC. Anatomic tvoes of single or co&mon ven%les in man. Morphologic and geometric aspects of 60 necropsied cases. Am J Cardiol 1964;14:367-86. 16. Macartney FJ, Partridge JB, Scott 0, Deverall PB. Common or single ventricle. An angiographic and hemodynamic study of 42 patients. Circulation 1976;53:543-54. 17. Lev M, Liberthson RR, Kirkpatrick JR, Eckner FAO, Arcilla RA. Single (primitive) ventricle. Circulation 1969;39:57’7-91. 18. Soto B, Pacific0 AD, Sciasco GD. Univentricular heart: an angiographic study. Am J Cardiol 1982;49:787-94. 19. Thies WR, Soto B, Diethelm E, Bargeron LM, Pacific0 AD. Angiographic anatomy of hearts with one ventricular chamber: the true single ventricle. Am J Cardiol 1985;55:1363-6. 20. Stanger P, Rudolph AM, Edwards JE. Cardiac malpositions. An overview baaed on study of sixty-five necropsy specimens. Circulation 1977;56:159-72. 21. Becker AE, Anderson RH. Pathology of congential heart disease. London: Butterworth & Co, Ltd, 1981:77-92. 22. Baron MG. Endocardial cushion defects. Radio1 Clin North Am 1968;3:343-60. 23. Jacobstein MD, Fletcher BD, et al. Evaluation of atrioventricular septal defects by magnetic resonance imaging. Am J Cardiol 1985;55:1158-61. 24. Higgins CB, Sechtem UP, Pflugfelder P. Cine MR: evaluation of cardiac ventricular function and valvular function. Int J Card Imaging 1988;3:21-8. 25. Sechtem U, Pflugfelder P, Cassidy MC, White RD, Cheitlin M, Schiller ND. Quantification of regurgitant volumes in patients with mitral or aortic regurgitation by tine MRI. Radiology 1988;167:425-30. 26. Pelt NJ, Shimakawa A, Glover GH. Phase contrast tine MRI. Society of Magnetic Resonance Imaging. 1989;l:lOl.
