Scintigraphic spectrum of a patient population with suspected arrhythmogenic right ventricular dysplasia

Abstract

Gated radionuclide ventriculography (RNV), combined with inter- and intraventricular dyssynchrony measurement by phase analysis, is able to evidence right and left ventricular mechanical cardiac disorders and may contribute to the diagnosis of arrhythmogenic right ventricular dysplasia (ARVD). Nevertheless, the patients referred for suspicion of ARVD on the basis of symptoms, electrical abnormalities or family history of sudden death, are very heterogeneous and the examination findings spread out from strictly normal to severely abnormal. In order to describe the patient population encountered in "real life" we propose to use an automatic clustering method based on RNV results in order to segment the overall population into subgroups with coherent scintigraphic data in each one. A series of 130 consecutive patients presenting with various criteria suggestive of ARVD has been studied over a 3-year period. Seven variables have been extracted from gated RNV: left and right ejection fractions, visual semi-quantitative assessment of left and right ventricular volumes, left and right phase standard deviations and inter-ventricular dyssynchrony (IVD) measured from the phase histograms. The Self Organizing Map (SOM) clustering method has been applied to these data with various numbers of variables (right ventricular values only or values from both ventricles) and an increasing number of classes from two to nine. Including left ventricular variables and IVD in the analysis results in significant changes in classification compared to right ventricular data alone. Clustering into nine classes seems to be the most pertinent one and separates patients into four groups of normal result or insignificant left, right or bilateral abnormalities, two groups of isolated right ventricular abnormalities of increasing severity and three groups of severe bilateral abnormalities, right predominant with and without IVD, and left predominant. Automatic clustering of patients on the basis of scintigraphic results helps to understand the signification of the large spectrum of results encountered in clinical practice for patients whose common characteristic is to present some abnormalities or risk factors leading to investigations in the context of suspicion of ARVD. Although the final diagnosis remains questionable in a large proportion of patients, the knowledge of the various profiles of gated blood pool phase analysis may help for stratification of patients at risk of ARVD.

Prognostic value of right ventricular ejection fraction in pulmonary arterial hypertension

“…Normal RVEF values have been previously described at >30% in our centre [18,19]. Physicians reading ERNA were blinded to patient characteristics at any time, and had access to baseline RVEF at the time of follow-up RVEF measurement.…”

Identification of clinically relevant phenotypes in patients with Ebstein anomaly

“…17 Cluster analysis has been used successfully to classify patients with congenital abnormalities into clinically relevant subgroups based on their phenotype, 18 including patient classification based on cardiac imaging. [19][20][21][22] Adequate sample size for cluster analysis was determined considering the number of variables, which may exponentially increase data dimensionality. For this purpose, an approach has been proposed in which meaningful classifications can be obtained using a sample size of 2 × , where x is the number of variables.…”