Author + information
- Received February 27, 2006
- Revision received July 12, 2006
- Accepted July 23, 2006
- Published online November 21, 2006.
- Srijita Sen-Chowdhry, MA, MBBS, MRCP⁎,†,⁎ ( )(, )
- Sanjay K. Prasad, MD, MRCP†,
- Petros Syrris, PhD⁎,
- Ricardo Wage, DCR(R)†,
- Deirdre Ward, MBBS, MRCPI⁎,
- Robert Merrifield, PhD‡,
- Gillian C. Smith, MSc†,
- David N. Firmin, PhD†,
- Dudley J. Pennell, MD, FACC† and
- William J. McKenna, MD, DSc, FACC⁎
- ↵⁎Reprint requests and correspondence:
Dr. Srijita Sen-Chowdhry or Professor William J. McKenna, Cardiology In The Young, The Heart Hospital, 16–18 Westmoreland Street, London W1G 8PH, United Kingdom.
Objectives We sought to assess the utility of cardiovascular magnetic resonance (CMR) in the evaluation of arrhythmogenic right ventricular cardiomyopathy (ARVC) in relation to diagnostic criteria and genotype.
Background Timely diagnosis of ARVC is difficult as clinical findings may be subtle and nonspecific in early disease. The role of CMR is controversial owing to the absence of a standardized protocol, insufficient experience with the modality, and inherent difficulties in imaging the right ventricle.
Methods Comprehensive CMR examination was performed in 232 patients undergoing evaluation for suspected ARVC. CMR outcomes were compared with: 1) prospective clinical diagnosis using Task Force guidelines, with and without the proposed modifications for familial ARVC; and 2) gene-carrier status in 35 individuals from genotyped families.
Results CMR studies were positive in all 64 patients who prospectively fulfilled Task Force criteria, resulting in 100% sensitivity. Specificity in relation to Task Force criteria was low (29%). Of the 119 apparent false positives detected by CMR, however, 63 fulfilled modified diagnostic criteria for familial ARVC and 7 were obligate gene carriers, suggesting that CMR frequently identifies individuals with early disease, in whom Task Force criteria are relatively insensitive. This was borne out by evaluation of genotyped individuals (26 gene-positive and 9 gene-negative), in whom CMR had a sensitivity of 96% and a specificity of 78%.
Conclusions CMR is a valuable component of the diagnostic workup for ARVC when performed with a dedicated protocol by specialists with experience in analysis of volumes, right ventricular wall motion, and delayed-enhancement imaging.
Clinical diagnosis of arrhythmogenic right ventricular cardiomyopathy (ARVC) is frequently confounded by the nonspecific nature of associated findings and the absence of a single, definitive diagnostic test. The 1994 Task Force (TF) diagnostic guidelines aimed to address these difficulties by differential weighting of structural, histologic, electrocardiographic, arrhythmic, and familial features into major and minor criteria, according to their specificity (1).
Preliminary elucidation of the genetic basis of ARVC over the past 5 years has had two-fold results. First, the isolation of disease-causing mutations in plakoglobin, desmoplakin, plakophilin-2, and desmoglein-2 has suggested a central role for impaired cell adhesion in the pathogenesis of ARVC (2,3). Second, the accompanying studies of genotype–phenotype associations have highlighted the limitations of the TF criteria, which were developed at a time when clinical experience with ARVC was dominated by symptomatic index cases with advanced disease expression (4,5). Consequently, the guidelines are highly specific but lack sensitivity for early disease. Modified criteria have been proposed to facilitate diagnosis of familial ARVC in first-degree relatives with incomplete disease expression (Table 1)(1,6).
Particularly challenging is detection of the so-called “concealed” phase of ARVC, during which clinical markers are subtle or absent but patients may nonetheless be at risk of sudden cardiac death (SCD) (1). Strategies to enable timely diagnosis include clinical implementation of genetic analysis and development of sensitive imaging techniques.
The role of cardiovascular magnetic resonance (CMR) in the diagnosis of ARVC remains unresolved. CMR originally attracted interest as a noninvasive means of tissue characterization owing to the ability of fast-spin echo sequences to differentiate fat from normal myocardium. Acquisition of gradient-echo ciné images further allows assessment of volumes, function, and regional wall motion abnormalities (RWMA). Late enhancement (LE) has been shown to correlate with scarring after myocardial infarction and fibrofatty changes in ARVC (7,8). Nevertheless, the subjectivity inherent in the interpretation of wall thinning, localized contraction abnormalities, and intramyocardial fat has generated considerable controversy (9–11). Other criticisms of CMR include the absence of a standardized protocol and studies systematically comparing it with other imaging modalities. Studies assessing the diagnostic value of CMR have hitherto been limited by small sample numbers and the need to employ the TF criteria as the only available gold standard (11–12).
We sought to assess the diagnostic accuracy of CMR in 232 patients undergoing evaluation for suspected ARVC using 3 different gold standards: 1) TF guidelines; 2) extended diagnostic criteria, incorporating the proposed modifications for familial ARVC and obligate gene carriers identified through pedigree analysis; and 3) gene-carrier status in the subset of individuals from genotyped families.
Ethical approval was obtained from local committees, and participating individuals gave informed consent. The Heart Hospital is a tertiary referral center with a well-established clinical ARVC service, run in conjunction with a genetic screening program. The study population comprised 232 patients attending either of 2 dedicated clinics between January 2003 and February 2005 for assessment of suspected ARVC. All underwent cardiac evaluation for ARVC, including clinical history, pedigree analysis, 12-lead electrocardiogram (ECG), signal-averaged ECG, conventional 2-dimensional echocardiography, exercise testing, and ambulatory ECG monitoring.
Prospective clinical diagnosis
Patients were assigned to one or more of the following diagnostic categories independently of the results of CMR assessment:
1. TF criteria: all patients fulfilling TF diagnostic guidelines for ARVC;
2. Extended diagnostic criteria, incorporating the following:
Over the 2-year recruitment and 1-year follow-up periods, disease-causing mutations were identified in several ARVC families from the broader Heart Hospital cohort and a number of individuals underwent predictive testing. As a result, the gene-carrier status of 35 participating individuals became known.
Cardiovascular magnetic resonance was performed at the National Heart and Lung Institute, Imperial College, on a 1.5-T Sonata scanner (Siemens Medical Solutions, Erlangen, Germany) using a comprehensive dedicated protocol (Table 2)(13–16).
Eight data collectors participated in image acquisition and/or analysis. Clinical and genetic data were not available to operators or readers at the CMR unit. Nevertheless, familiarity with clinical details was presumed for the main operator (S.S.C.) owing to her direct involvement in patient care at The Heart Hospital, and she was regarded as unblinded by default. The remaining data collectors were considered independent.
All CMR studies were reviewed by the main operator (S.S.C.) and at least 2 independent readers. The readers subjectively classified each scan into 1 of 4 categories: Diagnostic, in which sufficient hallmark features were recognized to support a diagnosis of ARVC on the basis of CMR alone; Strongly Suspicious, in which unequivocal abnormalities consistent with ARVC were identified, but were considered insufficient per se to establish a diagnosis without clinical correlation; Possible, in the presence of mild or nonspecific abnormalities, the significance of which was difficult to interpret; or Normal. In instances where one or more readers differed in classification of a scan, the concordant opinions were considered final, with more independent reviews sought if necessary.
For assessment of diagnostic accuracy, Diagnostic and Strongly Suspicious scans were considered positive, whereas scans classified as Possible or Normal were considered negative.
Twelve months after the completion of the study (January 2006), the main CMR operator (S.S.C.) reinterpreted 150 randomly selected CMR scans from the original cohort. Names and other identifying details were removed from the scans before viewing to ensure blinding. Her subsequent classification of the scans was compared with her original opinion during the course of the study as an indicator of intraobserver variability.
Concordance with the opinion of the main CMR operator was assessed for readers who had reviewed >60 scans as an indicator of interobserver variability.
Statistical calculations were performed using web-based computation facilities (courtesy of Professor Richard Lowry) (17). Linear weighted kappa concordance coefficient was calculated as an indicator of intraobserver and interobserver variability. The four designations (Diagnostic, Strongly Suspicious, Possible, and Normal) were considered ordinal categories. The default relative distance of 1 was increased to 3 between the Strongly Suspicious and Possible categories to reflect the positive/negative dichotomy.
Sensitivity, specificity, and positive and negative predictive values were assessed in relation to TF guidelines, extended diagnostic criteria, and genotype. The relative utility of key CMR features (ventricular volumes, RWMA, intramyocardial fat, and LE) was evaluated in genotyped patients. Receiver-operating characteristic (ROC) curve analysis was performed for ordinal and categorized continuous variables (18). The presence or absence of each abnormality was considered a dichotomous variable, and association with genotype was analyzed using the Fisher exact probability test.
Patient baseline characteristics
The study sample comprised 232 patients, ages 34 ± 16 years (range 11 to 76 years), with an approximately equal male-to-female ratio (0.95:1). Indications for referral were as follows: 34 (15%) had symptomatic arrhythmia of right ventricular (RV) origin; 6 (3%) had cardiac symptoms and electrocardiographic or echocardiographic features that raised suspicion of ARVC; 1 had recurrent syncope without documented arrhythmia; 1 was a cardiac arrest survivor with documented paroxysmal atrial fibrillation but otherwise normal cardiac assessment including coronary angiogram; the remainder had a family history of ARVC (n = 173 [75%]) or SCD with cause unascertained (n = 17 [7%]).
Based on clinical evaluation before CMR examination, 64 patients (28%) fulfilled TF diagnostic criteria for ARVC. Applying the proposed modifications for familial ARVC enabled clinical diagnosis in a further 63 patients (27%). Seven patients were identified as obligate gene carriers based on familial assessment and pedigree analysis, but did not satisfy original or modified diagnostic criteria. A total of 134 patients (58%) therefore fulfilled the extended diagnostic criteria as previously defined.
Scan duration was 40 to 80 min. The protocol was generally well tolerated. Twenty-seven children (ages <18) were not eligible for administration of gadolinium-DTPA because of age and weight stipulations. Five adults declined contrast injections because of needle phobia or claustrophobia and consequent reluctance to undergo the full-length protocol.
During reporting, the readers classified CMR studies as Diagnostic in the presence of moderate to severe RV dilation and/or impairment, aneurysms, and marked myocardial fatty replacement or LE. The Strongly Suspicious category included studies with mild to moderate ventricular dilation and/or impairment, prominent RWMA, and/or intramyocardial fat or LE. Scans were rated Possible in the presence of mild RWMA of uncertain significance and/or volumes at the upper limit of normal.
The intraobserver kappa for the main operator was 0.87 (range 0.81 to 0.93) (n = 150). Three of the other readers reviewed more than 60 scans. Interobserver kappa coefficients were 0.47 (range 0.31 to 0.63) (n = 60), 0.58 (range 0.46 to 0.70) (n = 110), and 0.84 (range 0.76 to 0.92) (n = 143).
The results of CMR examination in relation to clinical status are shown in Table 3.All 64 patients who prospectively fulfilled TF criteria had unequivocal abnormalities on CMR, resulting in 100% sensitivity. However, almost 70% of patients who did not satisfy TF criteria also had CMR studies Diagnostic or Strongly Suspicious of disease expression. The overall specificity of CMR in relation to TF criteria therefore was only 29%. Of the 134 patients fulfilling extended diagnostic criteria, all had Strongly Suspicious or Diagnostic changes on CMR; the sensitivity of CMR therefore remained 100%, and the specificity increased to 50%. In the 63 patients fulfilling modified criteria for familial ARVC, inclusion of CMR abnormalities would have enabled definitive TF diagnosis in 47 (75%).
Of the 35 patients with CMR studies evaluated as Possible, 19 (54%) had nondiagnostic clinical features such as incomplete right bundle branch block, supraventricular arrhythmia, and/or couplets and bigeminy on ambulatory ECG monitoring but <200 ventricular extrasystoles in 24 h.
The genotyped subset was composed of 15 men and 20 women, age 34 ± 16 years (range 11 to 72) from 15 families with distinct mutations in desmoplakin (n = 6), plakophilin-2 (n = 6), desmoglein-2 (n = 2), and another cell adhesion protein (n = 1) (Syrris P, McKenna WJ, unpublished data, March 2006). CMR evaluation was conducted before mutation screening in 30 individuals from this subset. Disease-causing mutations were identified in 11 of these patients using a candidate gene strategy; the other 19 received predictive testing after isolation of the causative gene in their families. In the 5 remaining individuals, CMR was performed after successful genotyping. Four were from a single family genotyped by linkage analysis. The fifth underwent CMR as part of his clinical workup after predictive testing demonstrated gene-carrier status.
Twenty-six of the patients in the genotyped subset were gene positive. Of these, 12 fulfilled TF guidelines and a further 8 satisfied modified diagnostic criteria. CMR studies were considered Diagnostic (n = 23) or Strongly Suspicious (n = 2) in a total of 25 (96%). The exception was a boy age 12 whose CMR study was ranked as Possible on the basis of mild localized dilation at the RV outflow tract and subtricuspid region. His other investigations were unremarkable.
Five gene-positive individuals who did not fulfill the extended diagnostic criteria were correctly identified by CMR before genotyping: 1) an asymptomatic 19-year-old boy with marked RV RWMA and prominent left ventricular (LV) LE on CMR. All other investigations were unremarkable; 2) a boy of 13, in whom biventricular dimensions were at the upper limit of normal on echocardiography. CMR volume assessment showed moderate biventricular dilation in comparison with age- and sex-matched controls, and RWMAs were observed in both ventricles on CMR; 3) an asymptomatic 47-year-old woman with incomplete right bundle branch block on 12-lead ECG and LV dimensions at the upper limit of normal on echocardiography. CMR showed biventricular dilation and RV RWMA; 4) a 42-year-old woman with syncope in whom clinical evaluation revealed inverted T-waves in V1through V6, late potentials on signal-averaged ECG, and >1,000 ventricular extrasystoles in 24 h. Conventional 2-dimensional echocardiography was unremarkable. Three minor criteria are insufficient for TF diagnosis and there was no family history of ARVC, precluding use of modified criteria. CMR demonstrated multiple RV aneurysms, prominent RV fatty replacement, and LV LE; and 5) a 48-year-old woman with a longstanding history of palpitation who had recently lost her daughter to ARVC. Her 12-lead ECG showed inverted T-waves in V1to V2, insufficient to fulfill modified diagnostic criteria. Visualization of the RV was unsatisfactory on conventional 2-dimensional echocardiography; however, CMR showed moderate biventricular dilation, RV RWMA, and minor LE at the inferolateral LV wall.
Among the 9 gene-negative patients, one 34-year-old woman satisfied TF criteria. Her mother had suffered SCD with confirmation of ARVC on postmortem examination. The patient herself had a longstanding history of palpitation. A 12-lead ECG and signal-averaged ECG were normal, but Holter monitoring on one occasion showed >400 ventricular extrasystoles including a salvo of nonsustained ventricular tachycardia. Two-dimensional echocardiography showed RV RWMA. Her CMR study was rated as Suspicious, as was that of her identical twin sister. A missense mutation in plakophilin (419C→T) was subsequently isolated in all other clinically affected members of the family.
Four CMR studies from gene-negative patients were classified as Possible because of abnormal trabeculation (n = 1), mild localized dilation at the RV subtricuspid region (n = 2), and mild LV dilation (n = 1). The remaining 3 gene-negative individuals had normal scans.
Comparison of diagnostic accuracy
The diagnostic accuracy of the TF guidelines, extended criteria, and CMR in the genotyped sample are compared in Table 4.Again, both Diagnostic and Strongly Suspicious studies were considered positive for evaluation of sensitivity and specificity. The area under the ROC curve for CMR was 0.98. Inclusion of scans rated Possible in the positive category would have increased the sensitivity of CMR from 95% to 100%, at the cost of reducing the specificity from 78% to 33%. Conversely, assigning positive status to the Diagnostic scans alone, excluding studies considered Strongly Suspicious, would have improved the specificity to 100% while reducing the sensitivity to 88%.
Incorporation of both clinical and CMR data would have enabled TF diagnosis in 20 of 26 gene-positive patients, improving the sensitivity of the diagnostic guidelines from 46% to 77% without any reduction in the 89% specificity.
Of 26 gene-positive patients, CMR volume analysis showed dilation and/or systolic impairment of the RV in 20 (77%) and the LV in 14 (54%). In 1 gene-negative patient, a boy age 12, LV volumes were mildly enlarged compared with age-matched and sex-matched controls. Abnormal trabeculation was reported in 50% of gene carriers, but also in 44% of gene-negative patients.
Eighteen gene-positive patients (69%) had RV aneurysms and/or segmental dilation and RWMA judged to be severe (Figs. 1A and 1B).Marked regional dilation and RWMA were not observed in gene-negative patients. In contrast, mild localized dilation and/or hypokinesia was present in all gene carriers and 56% of gene-negative patients.
Of the subgroup of 35 genotyped patients, delayed-enhancement imaging was performed in 28. Five children were not eligible according to the study protocol, and 2 adults declined. LV LE was detected in the 22 gene-positive patients who received gadolinium-DTPA in a subepicardial or midmyocardial distribution, or both. The extent of LE was minor in 5, and moderate to extensive in 18 (82%) (Fig. 1D). Minor LE was reported in 2 gene carriers and 2 gene-negative twin sisters (1 of whom satisfied TF criteria). RV LE was less easy to distinguish from intramyocardial fat, but was considered unequivocal in 13 gene carriers and no gene-negative individuals. Fatty replacement of the RV myocardium (Fig. 1C) was identified in 65% of gene-positive patients and the sole gene-negative patient who fulfilled TF criteria.
The diagnostic value of various CMR parameters is shown in Table 5,together with the area under the ROC curve, where applicable. The parameters showing the strongest association with gene-carrier status were abnormal RV volumes, severe segmental dilation and/or RV aneurysms, and LV LE.
The past decade has seen gradual evolution in clinical perception of CMR assessment for suspected ARVC. Initial enthusiasm for a noninvasive technique that is not restricted by acoustic windows was tempered by concerns regarding interobserver variability, lack of experience with the modality at most imaging centers, insufficient resolution for detection of wall thinning, overinterpretation of RWMA, and difficulties in differentiating normal epicardial fat from true myocardial adipose replacement (9,19). An additional cautionary note was the paucity of studies assessing its diagnostic accuracy, albeit in the absence of a true gold standard (10,11).
Since genetic analysis remains in its infancy in ARVC and is not yet widely available, the initial phase of our study mirrors the approach at most centers, where noninvasive clinical evaluation is the mainstay of diagnosis. Among 232 patients undergoing evaluation for ARVC, the sensitivity of CMR was 100% when applying TF criteria as a gold standard. The conspicuously high detection rate may be attributed, at least in part, to the limitations of the TF guidelines, which reflect the overt form of the disease that is easily recognized. The majority of study patients who fulfilled TF criteria had abnormalities on 2-dimensional echocardiography. In patients with structurally severe disease, all imaging modalities are likely to provide satisfactory evaluation, and the ability of CMR to do so is therefore not exceptional. In contrast, patients with milder phenotypes pose a significant diagnostic challenge.
White et al. (11) investigated the relative utility of CMR and RV angiography in the diagnosis of ARVC. Employing TF criteria as the gold standard, RV angiography had a specificity of 100%, whereas that of CMR was only 60%. The high false-positive rate was ascribed to over-reading of subtle RV RWMA detected by CMR. Although our study elicited a still lower specificity for CMR in comparison with TF guidelines (29%), subsequent analysis indicates an alternative explanation. Of the 119 apparent false positives identified by CMR, over 50% fulfilled the modified diagnostic criteria for familial ARVC and a further 7 were obligate gene carriers. The obligate gene carriers are of particular interest because other clinical investigations were within normal limits, whereas the CMR was unequivocally abnormal. Thus, our results suggest that CMR may be capable of detecting an early form of ARVC.
Investigation of the genotyped subset lends support to this hypothesis. The TF guidelines identified 46% of genetically affected individuals; the proposed modifications for the diagnosis of familial ARVC increased the sensitivity to 77% without any loss in specificity. The specificity of TF diagnosis in the genotyped sample was <100% because of the single gene-negative patient who fulfilled criteria, based on a family history of pathologically proven ARVC, nonsustained ventricular tachycardia on Holter monitoring, and RV RWMA on echocardiography. The 419C→T (S140F) missense mutation in plakophilin-2 isolated in her family has been reported in other cohorts and is thought to be pathogenic rather than a polymorphism (2,5). Reconciling the genetic data with the clinical profile is difficult, and involves ascribing the echocardiographic abnormalities to over-reading and the arrhythmia to an unrelated cause. Nevertheless, we cannot exclude the possibility of a second mutation in this family, and the genotype remains under investigation.
The high sensitivity of CMR among genotyped patients (96%) merits further discussion. CMR evaluation in our study differed from standard clinical protocols in several key aspects. First, acquisition of ciné images with high temporal resolution (<23 ms) facilitated accurate volume estimation and assessment of RWMA, at the cost of increasing the breath hold (14). Second, late-enhancement imaging was performed in the majority of patients. The feasibility of the longer research protocol in a clinical setting remains to be established. Third, study patients with frequent ventricular extrasystoles were placed on antiarrhythmic medication before scanning to reduce interference with ECG gating. Finally, the specificity of 78% raises the possibility that the readers had a relatively low threshold for classifying studies as Suspicious; that the area under the ROC curve was 0.98 nonetheless affirms the diagnostic value of CMR.
The negative predictive value of CMR in the genotyped sample was 88%. The sole false-negative result was a 12-year-old boy with mild RV RWMA of uncertain significance. The absence of more prominent morphologic changes may be a corollary of age-related penetrance in ARVC; clinical manifestations are uncommon before adolescence. Nevertheless, it should be emphasized that a normal CMR study does not exclude ARVC. Postmortem studies show that fibrofatty replacement is frequently microscopic rather than overt, and hence not discernible by imaging (1). Furthermore, ventricular arrhythmia has been reported in a Naxos patient without apparent histologic substrate (2). The limitations of any imaging modality in excluding ARVC are therefore underscored. Likewise, we do not advocate the use of CMR, or any single investigation, as a sole screening test for ARVC, although our results strongly support inclusion of CMR as part of a comprehensive noninvasive workup.
In their study of 2-dimensional echocardiography in probands with ARVC, Yoerger et al. (20) highlighted the importance of quantifying RV enlargement and dysfunction. Consistent with their findings was the notable diagnostic accuracy of RV volume analysis in our cross-comparison of key CMR parameters. Another strong feature was LV LE, occurring in a subepicardial or midwall pattern distinct from the subendocardial involvement observed in ischemic heart disease (7). Specific genotype–phenotype associations may, however, have skewed the results; of the 26 gene-positive individuals in our sample, 16 (62%) had desmoplakin mutations. Frequent and marked LV involvement has been reported in several families with defects in desmoplakin (2,4). Unequivocal RV LE was less commonly identified, a probable corollary of protocol design; inversion times were set to null the LV myocardium, and the inversion recovery sequence was not fat-suppressed. Abnormal trabeculation has been reported in over half of ARVC probands assessed by echocardiography and CMR (12,20). Although a similar proportion of gene carriers in our study showed this feature, it was also observed in 4 gene-negative individuals, resulting in a low specificity (56%).
CMR studies were frequently assigned a Possible rating in the presence of soft signs such as abnormal trabeculation, mild localized dilation, and subtle RWMA. Around half of these patients had nonspecific ECG abnormalities and supraventricular or low-grade ventricular arrhythmia, raising suspicion of early ARVC. Nevertheless, the results of genotype analysis indicate that, with one exception, patients in the Possible category were not affected. In the concealed phase of ARVC, morphologic abnormalities may be discernible only by highly sensitive imaging techniques, CMR and contrast-enhanced echocardiography being the main candidates. Operators in both modalities may be susceptible to over-reading because the spectrum of normal morphology and function within the thin-walled, trabeculated, pyramidal RV remains to be defined.
Reproducibility therefore remains a central issue in the assessment of ARVC by CMR. Intraobserver concordance was excellent for the main operator, suggesting that pattern recognition may be learned. Strengthening this premise is the high interobserver concordance (kappa = 0.84) shown by one of the readers, who had reported >150 ARVC scans alongside the main operator prior to involvement in the study. Concordance was lowest (although still moderate) for the reader with the least pre-study and intra-study experience, and limited background knowledge of ARVC. These preliminary findings suggest 3 key factors in improving reproducibility for an individual CMR reviewer: extensive experience with reporting ARVC scans, understanding of the clinical manifestations and pathogenesis of ARVC, and training alongside an expert reader to learn pattern recognition. At a broader level, CMR will realize its full potential as a diagnostic tool in ARVC only after systematic evaluation of a large genotyped population, who represent the ideal patients for comparing the changes of early disease with normal variants. The preliminary data yielded by our study should provide impetus for further work in this important area.
Key limitations include the case mix, which is not wholly representative of the patient population seen at other institutions. Our sample included a preponderance of relatives undergoing familial evaluation (82%), reflecting the special interests of our center; only 18% were referred for symptomatic profiling, the predominant reason for assessment elsewhere.
The main CMR operator was considered unblinded owing to her role as a clinician treating study patients. Familiarity with clinical details is a representation of reality, since referring physicians commonly provide a summary of prior investigations, but raises the possibility of systematic bias. This was reduced by involvement of at least 2 independent readers with each case. Furthermore, the high level of concordance achieved on blinded viewing 12 months later supports the validity of the original categorizations.
In spite of controversies surrounding its role in ARVC, CMR is a valuable diagnostic tool that complements echocardiography and substantially enhances the sensitivity of clinical diagnosis, particularly in early disease. Consistent with this was the 96% sensitivity and 78% specificity of CMR in the genotyped subset of patients, only 46% of whom satisfied TF criteria. Incorporation of CMR results would have enabled TF diagnosis in a further 30% of proven gene carriers, and in 75% of patients who prospectively satisfied modified criteria only. Since accuracy and reproducibility are strongly operator dependent, however, CMR evaluation of ARVC may be best conducted at specialist centers, using a dedicated comprehensive protocol, by experts with a clinical background in the field and extensive experience in analysis of volumes, RV RWMA, and delayed-enhancement imaging.
The authors would like to thank Sripurna Das, PhD, for her helpful comments on the manuscript. This work is dedicated to the memory of Asifa Quraishi, MBBS, MRCP.
Supported by the British Heart Foundation (Drs. Sen-Chowdhry, Syrris, Ward, and McKenna), the European Commission 5th Framework Program (ARVC/D project, QLG1-CT-2000-01091), and CORDA (Drs. Prasad and Pennell).
- Abbreviations and Acronyms
- arrhythmogenic right ventricular cardiomyopathy
- cardiovascular magnetic resonance
- late enhancement
- left ventricle/left ventricular
- receiver-operating characteristic
- right ventricle/right ventricular
- regional wall motion abnormalities
- sudden cardiac death
- Task Force
- Received February 27, 2006.
- Revision received July 12, 2006.
- Accepted July 23, 2006.
- American College of Cardiology Foundation
- Pilichou K.,
- Nava A.,
- Basso C.,
- et al.
- Bauce B.,
- Basso C.,
- Rampazzo A.,
- et al.
- Syrris P.,
- Ward D.,
- Asimaki A.,
- et al.
- Hamid M.S.,
- Norman M.,
- Quraishi A.,
- et al.
- Tandri H.,
- Saranathan M.,
- Rodriguez E.R.,
- et al.
- Maceira AM, Prasad SK, Pennell DJ. Reference right ventricular systolic and diastolic function normalized to age, gender and body surface area from steady state free precession cardiovascular magnetic resonance. Eur Heart J 2006. In press.
- ↵VassarStats. Available at: http://faculty.vassar.edu/lowry/vshome.html. Accessed July 10, 2006.
- ↵Eng J. ROC analysis: web-based calculator for ROC curves. Available at: http://www.rad.jhmi.edu/jeng/javarad/roc/JROCFITi.html. Accessed July 10, 2006.
- Yoerger D.M.,
- Marcus F.,
- Sherrill D.,
- et al.