| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of Miami School of Medicine, Jackson
Memorial Hospital, WW279, 1611 N.W. 12th Ave., Miami, FL 33136.
2 Present address: Department of Radiology, Tel-Aviv Medical Center, 6 Weizman
St., Tel Aviv, Israel 64239.
3 Department of Pediatrics, University of Miami School of Medicine, Miami, FL
33136.
4 Present address: Department of Pediatric and Adolescent Medicine, Section of
Pediatric Cardiology, Mayo Clinic, Rochester, MN 55905.
5 Department of Pathology, University of Miami School of Medicine, Miami, FL
33136.
Received January 2, 2002;
accepted after revision August 27, 2002.
Address correspondence to J. E. Fishman.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Twenty-six consecutive pediatric patients underwent MR imaging with ECG-gated non–breath-hold spin-echo T1-weighted non–fat-suppressed and fat-suppressed sequences. The MR images were evaluated for thinning or fat signal in the right ventricular wall and for enlargement or increased trabeculation of the right ventricle or right ventricular outflow tract. Cine MR imaging was used to assess wall motion abnormalities. Cardiac biopsy was performed in 17 patients. Biopsy results and other clinical findings suggesting arrhythmogenic right ventricular cardiomyopathy were tabulated.
RESULTS. Two MR imaging studies were of poor quality as a result of arrhythmias, and one study was incomplete. In the 23 remaining patients, there were (mean ± SD) 1.5 ± 1.0 and 0.8 ± 1.0 findings of possible arrhythmogenic right ventricular cardiomyopathy in the non–fat-suppressed and the fat-suppressed sequences, respectively. Fat-compatible signal in the myocardium was detected in 16 (70%) of 23 non–fat-suppressed studies and in five (22%) of 23 fat-suppressed studies (p = 0.003). The non–fat-suppressed sequence had a higher sensitivity (75% vs 43%) and a lower specificity (38% vs 75%) for fatty infiltration than did the fat-suppressed sequence when correlated with the biopsies. The linear correlation between all MR findings and all clinical diagnostic criteria, including biopsy, was better for the combination of cine and both T1 sequences (r = 0.58) than for the non–fat-suppressed (r = 0.53) or fat-suppressed (r = 0.46) T1 sequences alone.
CONCLUSION. MR imaging showed moderate correlation with the clinical criteria in the diagnosis of arrhythmogenic right ventricular cardiomyopathy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Standardized diagnostic criteria for arrhythmogenic right ventricular cardiomyopathy have been proposed on the basis of the presence of major and minor criteria encompassing genetic, electrocardiographic, arrhythmic, morphofunctional, and histologic factors [5]. Imaging criteria have included depiction of the enlargment of the right ventricle with global or segmental wall motion abnormalities, in the absence of left ventricular abnormalities, as shown on echocardiography, angiography, radionuclide scintigraphy, or MR imaging [1, 5]. Of these, MR imaging is considered a promising technique for delineating right ventricular anatomy and wall function, and has the ability to characterize tissue, specifically by differentiating fat from muscle [6].
The pediatric population poses a special challenge to MR imaging in the diagnosis of arrhythmogenic right ventricular cardiomyopathy. This cardiomyopathy is a progressive disease occurring over many years; thus, diagnosis in children may require the detection of small amounts of fat in normally thin right ventricular walls [7]. Motion artifacts are often more severe in children than in adults because pediatric heart and respiratory rates are usually faster, and children often cannot maintain their position for long. Even if sedation is routinely administered, a short MR study using the high-est-yield sequences is to be preferred. The purpose of our study was to evaluate three MR imaging sequences for the assessment of arrhythmogenic right ventricular cardiomyopathy in the pediatric population. Sequences were compared for their technical quality and for their individual and combined correlation with biopsy and other clinical findings of possible arrhythmogenic right ventricular cardiomyopathy [5].
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MR studies were performed on a 1.5-T super-conducting magnet (Edge; Marconi, Cleveland, OH). The studies were performed on patients in the prone position using a surface coil or dedicated cardiac coil. Patients were not sedated except for two patients who were less than 7 years old. All studies were cardiac gated, with the leads placed on the patients' backs. For anatomic localization, we performed a preliminary study using a 7-mm slice thickness. Subsequently, high-resolution ECG-gated spin-echo T1-weighted non–breath-hold sequences were performed in the axial plane, first without and then with a spectral fat-suppression pulse. The same slice locations were used for both sequences. The imaging range encompassed the pulmonary artery, superiorly, to the apex of the heart, inferiorly. The following parameters were used: slice thickness, 5–6 mm; gap, 0.5–1 mm; matrix, 192 x 256; field of view, 32; TR, R-R interval; TE, 22; and number of signal averages, 4–6. A presaturation band was placed behind the right ventricle to avoid flow signal artifacts in the phase-encoding axis, which can produce factitious bright-signal foci in the free wall of the right ventricle. After the T1 sequences, the cine study was performed using a non–breath-hold velocity flow refocused spoiled gradient-echo sequence (cine MR imaging) obtained in the axial plane with a TE of 4.6, a flip angle of 50–60, and number of signal averages of 2.
All MR examinations were retrospectively evaluated by consensus between an attending thoracic radiologist and a fellow who were unaware of the clinical and biopsy findings. The technical quality of the images obtained with non–fat-suppressed and fat-suppressed sequences was compared using the following 3-point rating scale: 0, poor quality (artifacts prohibit assessment of wall thickness or fatty infiltration); 1, moderate quality (artifacts present but wall readily distinguishable from blood pool); and 2, high quality (artifact-free). Both non–fat-suppressed and fat-suppressed sequences were evaluated for the following four criteria that indicate possible arrhythmogenic right ventricular cardiomyopathy: abnormal signal compatible with or suggesting fatty infiltration of the walls of the right ventricle or the right ventricular outflow tract, thinning of the walls of the right ventricle or right ventricular outflow tract (consisting of wall thickness < 2 mm at any point in the cardiac cycle or a focal, abrupt decrease in wall thickness adjacent to a normal appearing wall), enlargement of the right ventricle or right ventricular outflow tract (i.e., larger than the aortic outflow tract), and qualitatively heavy right ventricular trabeculation. These criteria were selected on the basis of previous studies [8, 9, 10]. If signal intensity suggesting fat was seen, a record was made as to whether it was present on only one image or on multiple images within the walls of the right ventricle or right ventricular outflow tract. Combining the results from non–fat-suppressed and fat-suppressed sequences yielded up to eight findings of arrhythmogenic right ventricular cardiomyopathy. Cine gradient-echo MR images were evaluated for evidence of focal or diffusely abnormal wall motion. Thus, there were a maximum of nine MR imaging findings indicating possible arrhythmogenic right ventricular cardiomyopathy with the use of all three sequences. Fisher's exact test was used to compare the non–fat-suppressed and fat-suppressed findings.
To quantify how reliably the non–fat-suppressed and fat-suppressed
sequences identified the presence or absence of fatty infiltration, we
performed signal intensity (SI) pixel measurements in the single area of
highest (non–fat-suppressed sequence) or lowest (fat-suppressed
sequence) signal in the right ventricular wall for each patient. These areas
were considered to most likely represent fatty infiltration, if present.
Measurements were also performed in nor-mal-appearing right ventricular muscle
and epicardial fat (Fig. 1A,
1B). Normalization of SI values
was performed using the formula:
 |
|
|
Biopsies were interpreted by a single attending pediatric pathologist who was unaware of the clinical and MR findings. Biopsy results were categorized as those showing definitive fatty infiltration, those showing small or minimal areas of fat considered equivocal for the diagnosis of arrhythmogenic right ventricular cardiomyopathy, and those free of fat. Clinical criteria were determined according to the standardized diagnostic criteria proposed by the Task Force of the Working Group on Myocardial and Pericardial Disease of the European Society of Cardiology and of the Scientific Council on Cardiomyopathies of the International Society and Federation of Cardiology [5]. According to the task force, major criteria include familial disease confirmed at surgery or necropsy, epsilon waves or prolongation of the QRS wave complex in the right precordial leads of the electrocardiogram, severe dilatation, and reduction of right ventricular ejection fraction with no or only mild left ventricular involvement, localized right ventricular aneurysms, severe segmental dilatation of the right ventricle, and fibrofatty replacement of myocardium at endomyocardial biopsy. There are also nine minor criteria [5]. For the purpose of categorizing our patients with an overall clinical score for arrhythmogenic right ventricular cardiomyopathy, we scored the presence of a major criterion as two points and a minor criterion as one point. The task force guidelines require a score of four or more points to diagnose arrhythmogenic right ventricular cardiomyopathy. Image-derived results, such as right ventricular dilatation and reduction in ejection fraction, were taken from angiography and not MR imaging, so as not to compare MR imaging with itself.
Correlation between MR findings and the clinical diagnostic criteria score was only obtained when biopsy was performed because biopsy results are part of the standardized diagnostic criteria. Linear regression was performed to compare MR imaging results with clinical criteria. Statistical significance was defined as a p value of less than 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MR Findings and Correlation with Biopsy and Clinical Criteria
There was a mean of 1.5 ± 1.0 findings on the
non–fat-suppressed sequences indicating possible arrhythmogenic right
ventricular cardiomyopathy, and a mean of 0.8 ± 1.0 findings on the
fat-suppressed sequences (p > 0.05, not significant). The
difference was almost entirely due to varying rates of detection of fat. High
signal compatible with or suggesting fat in the right ventricular wall was
detected in 16 (70%) of 23 non–fat-suppressed studies, but low signal
suggesting fat was detected only in five (22%) of 23 fat-suppressed studies
(p = 0.003). Other findings were not significantly different between
non–fat-suppressed and fat-suppressed sequences: regional thinning of
the right ventricular wall was seen in 12 (52%) of 23 non–fat-suppressed
studies and in seven (30%) of 23 fat-suppressed studies (p = 0.13);
enlargement of the right ventricular outflow tract was seen in two patients
(9%) in both sequences; and heavy trabeculation in the right ventricle was
seen in one patient (4%) in both sequences. Focal or generalized hypokinesis
or bulges were seen on 11 (46%) of 24 cine studies (including the patient who
had only non–fat-suppressed imaging).
Cardiac biopsy was performed in 17 patients, but in one patient the non–fat-suppressed and fat-suppressed images were all of poor quality. Of the 16 remaining patients, two had biopsies with definitive fatty infiltration of the right ventricular myocardium, six with equivocal fatty infiltration, and eight with no evidence of fat. One patient who was biopsied underwent only non–fat-suppressed imaging. For the eight patients with biopsies containing some fat or fibrofatty changes, suspected fatty areas were seen in six (75%) of eight non–fat-suppressed sequences and three (43%) of seven fat-sup-pressed sequences. For eight patients without fat in the biopsy, suspected fatty areas were seen in five non–fat-suppressed sequences (63%) and in two fat-suppressed sequences (25%). In the non–fat-suppressed sequences, SI scores were 46.5% ± 27.5% when fat was confirmed at biopsy and 18.1% ± 19.4% when no fat was present at biopsy (p = 0.05). For the fat-suppressed sequence, the SI score was 17.7% ± 36.9% in possible areas of fat in patients with positive biopsy findings and 9.3% ± 72.3% for negative biopsy findings (p = 0.61, not significant). Table 1 shows the correlation between biopsy grade (positive, equivocal, or negative) and the number of non–fat-suppressed images on which fat was suggested. Two of five patients with fat signal on more than one MR image had definitive infiltration at biopsy, two had equivocal biopsies, and one had a negative finding at biopsy; none of the 11 patients with fat signal on one image (n = 6) or no images (n = 5) had definitive infiltration at biopsy (p = 0.08).
|
MR sensitivity, specificity, and positive and negative predictive values for each of the spin-echo sequences related to the biopsy results for fatty infiltration of the right ventricular wall are presented in Table 2. In this analysis, the findings of both the definitive and equivocal biopsies were considered positive. There is a reciprocal relationship between sensitivity and specificity for the non–fat-suppressed and fat-suppressed sequences, with non–fat-sup-pressed images showing better sensitivity and fat-suppressed images showing better specificity with respect to biopsy results. According to the clinical criteria, two patients had four positive findings of arrhythmogenic right ventricular cardiomyopathy and were subsequently diagnosed with this disease. These patients were the same two patients who had definitive biopsy results. Among patients who were biopsied, wall thinning was seen on non–fat-sup-pressed images in both patients diagnosed with arrhythmogenic right ventricular cardiomyopathy, as well as in six others (sensitivity, 100%; specificity, 57%); thinning was seen on fat-sup-pressed images in one patient with arrhythmogenic right ventricular cardiomyopathy. Neither patient with enlargement of the right ventricular outflow tract on MR imaging was diagnosed with arrhythmogenic right ventricular cardiomyopathy (sensitivity and specificity, 0%). The one patient with heavy right ventricular trabeculation was diagnosed with arrhythmogenic right ventricular cardiomyopathy. Both patients who were diagnosed with arrhythmogenic right ventricular cardiomyopathy had dysmotility on cine studies, as did three others (sensitivity, 100%; specificity, 79%). The correlation between all MR findings of all three sequences and the summed clinical diagnostic criteria (major and minor) for arrhythmogenic right ventricular cardiomyopathy is shown in Figure 2. These criteria, which include biopsy, revealed linear correlation coefficients with an MR imaging r value of 0.58 for the combination of non–fat-suppressed images, fat-suppressed images, and cine MR imaging; coefficients had an r value of 0.57 for the combination of non–fat-suppressed and cine images, an r value of 0.53 for the non–fat-suppressed images alone, and an r value of 0.46 for the fat-suppressed images alone.
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
MR image quality can be limited because of artifacts from respiratory motion, cardiac pulsation, and flowing blood. The inherent arrhythmic nature of arrhythmogenic right ventricular cardiomyopathy will further interfere with ECG gating and thereby reduce image quality. In using two non–breath-hold spin-echo T1-weighted sequences, we found that only slightly more than half of the images in our study had a high quality, although one sequence frequently had better quality than the other in individual cases. Despite the limitations of each individual sequence, there was a greater correlation between the number of positive criteria on MR imaging and clinical diagnostic criteria for arrhythmogenic right ventricular cardiomyopathy for the combination of all three sequences (including cine MR imaging) than for any single sequence. Although using the combination of the three sequences requires prolongation of the total imaging time, this method might contribute to the diagnostic ability of the study.
For the detection of fat in the right ventricular wall, the two spin-echo sequences performed differently. The non–fat-suppressed sequence showed relatively high sensitivity and low specificity in contrast to the relatively high specificity and low sensitivity of the fat-suppressed sequence. In our patients, foci of signal compatible with or suggesting fat were detected in 70% of the studies on the spin-echo T1 images and in 22% of the studies performed with fat suppression. Other studies have shown fat-detection rates on non–fat-suppressed spin-echo T1-weighted images of 22–100% [8, 9, 10, 12] (Table 3). There are many potential sources of error in the detection of right ventricular fat. ECG-gated spin-echo sequences have variable T1-weighting from patient to patient because the TR depends on the R-R interval [12]; fat will be less intense on longer TR T1-weighted sequences. Fat suppression may be inhomogeneous, particularly when using a surface coil, as in our study. Molinari et al. [10] stated that all hyperintensities interpreted as adipose replacement using non–fat-suppressed sequences were identified and suppressed in the images obtained with fat suppression. We find, however, that a retrospective analysis of images at the same location often reveals that fat-compatible signal seen on non–fat-suppressed imaging may not have apparent fat-compatible signal on fat-suppressed imaging, likely because of the lower contrast between fat and muscle on the fat-suppressed sequence (Fig. 3A, 3B). The lower contrast could account for the lower sensitivity of the fat-suppressed sequence as well as for the results of our SI measurements. Non–fat-suppressed sequences showed an SI difference between fat and non–fat containing pixels in patients with and without fat at biopsy that was marginally significant (p = 0.05), but differences on fat-suppressed images were not significant. Other fat-suppression methods, such as opposed-phase imaging, might be better suited to reveal arrhythmogenic right ventricular cardiomyopathy [13]. The normal presence of epicardial fat makes identification of true intramyocardial fat difficult because no structure with different MR signals separates these two tissue layers. Observing fat signal instead of muscle signal where the right ventricular wall should be found may lead an observer either to the diagnosis of right ventricular thinning, with only the epicardial fat being seen, or to the diagnosis of replacement of the right ventricular myocardium by fat (Fig. 4). The same problem was encountered by Schick et al. [14], who used chemical shift selective breath-hold cine imaging as an alternative technique for the evaluation of arrhythmogenic right ventricular cardiomyopathy. The researchers showed fat signal in the right ventricular wall on the fat selective images and reduced water signal on the water selective sequences. However, they often found difficulties in determining whether the origin of the detected fat was the infiltrated myocardium or the epicardium.
|
|
|
Several MR imaging criteria have been proposed for the diagnosis of arrhythmogenic right ventricular cardiomyopathy because encountering fatty tissue in myocardium does not appear to constitute sufficient grounds for the diagnosis [9]. The relatively low specificity (38%) of high-signal foci seen on non–fat-suppressed spin-echo T1-weighted images and the low sensitivity (43%) of low-signal foci on the fat-suppressed sequences found in our series support the need for more accurate diagnostic findings to establish a diagnosis of arrhythmogenic right ventricular cardiomyopathy on MR imaging. Additional criteria that have been used include thinning of the right ventricular wall, dilatation of the right ventricle or right ventricular outflow tract, prominent trabeculation within the right ventricle or the right ventricular outflow tract, and evidence of focal dyskinesia. As we have described, wall thinning and dysmotility were seen on the MR images in both our patients who were diagnosed with arrhythmogenic right ventricular cardiomyopathy, with specificities between 57% and 79%. The histopathology of arrhythmogenic right ventricular cardiomyopathy may vary depending on various pathologic factors, because two types of arrhythmogenic right ventricular cardiomyopathy have been described. The purely adipose form is characterized by partial to total replacement of the right ventricular wall by fatty tissue with normal or thickened myocardium and no fibrosis. The other form is fibrofatty, characterized by fibrous tissue and fat with myocyte atrophy and myocardial thinning in the common sites (triangle) of dysplasia, with secondary aneurysmal dilatation [15, 16, 17]. Depiction of fatty infiltration in this pathologic form may be difficult because fibrous tissue is generally intermingled with fat, and if these foci are small then there will be superimposition of the two MR signals within the voxel. Fibrosis, unless extreme, would be even more difficult to confidently identify on unenhanced MR imaging than fat. Morphometric analysis of endomyocardial biopsy sections in patients with arrhythmogenic right ventricular cardiomyopathy has revealed greater amounts of fibrous tissue in younger patients but a higher prevalence of fatty tissue in older patients [18]. A further complication includes a normal degree of fatty involution of the right ventricular wall in older individuals.
Few if any prospective studies define the diagnostic sensitivity and specificity of MR imaging in the evaluation of arrhythmogenic right ventricular cardiomyopathy, partially because of the low frequency of biopsy. The increasing use of MR imaging may lead to even fewer biopsies because visualization of a thin wall on MR imaging may prompt concerns for possible perforation of the right ventricular wall, precluding biopsy in three patients in our study. Even so, endomyocardial biopsy is limited by imperfect sensitivity (67% in one study [19]). Sampling error can occur because for reasons of safety, samples are usually taken from the septum and not the free right ventricular wall. The septum is much less likely than the free wall to be affected by the disease. In addition, fibrofatty replacement primarily affects the subepicardial layer rather than the subendocardial layer, which is the biopsy site [20]. In our series, six patients (35%) had equivocal biopsy findings, in which small amounts of adipose or fibrosis were seen but were insufficient for definitive diagnosis of arrhythmogenic right ventricular cardiomyopathy. Both our patients with positive (unequivocal) biopsy findings had fat signal on more than one MR image; the findings in the six patients with equivocal biopsies, however, were equally likely to show fat signal on several images, one image, or no images (Table 1).
Our study has several limitations. Only non–breath-hold spin-echo and cine MR sequences were used for two reasons. Black blood breath-hold sequences were not available on our machine until the end of this study; such sequences generally improve imaging of the heart wall [21]. Another advantage of these breath-hold, double inversion recovery sequences is that all images are obtained at the same point (diastole) in the cardiac cycle. However, many young children, especially those less than 10 years old, cannot reliably breath-hold long enough for breath-hold MR imaging. Another drawback of breath-hold sequences is the possibly unreliable slice-to-slice positioning, which could negatively impact detection of small foci of disease. These drawbacks of breath-hold sequences apply to cine sequences as well, most of which were also performed without breath-holding but with two signal averages. The use of navigator pulses, which are becoming more commonly available on advanced MR imaging units, may improve non–breath-hold images. Pediatric patients suspected to have arrhythmogenic right ventricular cardiomyopathy often have severe arrhythmias resulting in poor cardiac gating, which averaging will somewhat ameliorate. We have since added short-axis cine imaging to better depict the inferior wall and to improve assessment of the size of the right ventricle. The low number of patients who were definitively diagnosed with arrhythmogenic right ventricular cardiomyopathy and the number of patients with an equivocal biopsy result created difficulties in establishing values regarding the diagnostic accuracy of MR imaging. Our study is also limited in that only patients who underwent biopsies were included in the overall assessment of MR imaging results versus clinical criteria because biopsy is a major criterion. The use of MR imaging can lead to bias in selecting patients for biopsy, as we have described.
In summary, arrhythmogenic right ventricular cardiomyopathy remains a challenging diagnosis with multiple, imperfect criteria proposed for diagnosis. Approximately half of the non–breath-hold spin-echo T1-weighted images without and with fat suppression were of optimal quality. Often, one sequence provided more high-quality images than the other. The non–fat-sup-pressed sequence showed a larger difference in signal intensity (contrast) between fat and nonfatty tissue than did the fat-suppressed sequence, and the non–fat-suppressed sequence SI scores correlated more accurately with biopsy. Our SI formula might serve as a template for more quantitative models to assess the presence of fat. Wall thinning and dysmotility were present in both our patients who were definitively diagnosed with arrhythmogenic right ventricular cardiomyopathy. The combination of three sequences had a higher correlation with clinical criteria for arrhythmogenic right ventricular cardiomyopathy than individual sequences, although the correlation was only moderate. Despite its poor sensitivity for fat, we continue to perform fat-suppressed imaging for the slightly higher image quality obtained, which may reveal other morphologic abnormalities. For both pathologic and technical MR imaging reasons, the detection of small foci of apparent fat without other structural or contractile abnormalities, particularly if seen on only one image, is probably not sufficient for the diagnosis of arrhythmogenic right ventricular cardiomyopathy.
Acknowledgments
We thank Pradip Pattany for pulse sequence programming and assistance with
other MR imaging issues.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. Castillo, H. Tandri, E. R. Rodriguez, K. Nasir, J. Rutberg, H. Calkins, J. A. C. Lima, and D. A. Bluemke Arrhythmogenic Right Ventricular Dysplasia: Ex Vivo and in Vivo Fat Detection with Black-Blood MR Imaging Radiology, July 1, 2004; 232(1): 38 - 48. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
