Evaluation of Coronary Artery Z-Score Models in a German Population-Based Study Cohort of 310 Patients

Objectives: Coronary artery aneurysms (CAA) pose a complication on the course of Kawasaki’s disease (KD). Diagnostic accuracy of CAA significantly influences diagnostic and therapeutic decisions for the patients. Seven different Z-score models, indicating the numerical measurement of the CA diameter’s relationship to the mean, have been derived from studies on North American and Asian healthy children.

First, those models have never been uniformly evaluated with data collected solely from a representative KD patient cohort. Furthermore, no recommendation exists as to which Z-score model is suitable to apply on European patients.

Methods: Echocardiographic reports performed during the acute phase of KD were recruited from either the active population-based German Pediatric Surveillance Study (ESPED) or the Ludwig-Maximilian-University of Munich. Those reports went through a systematical screening regarding information of the CA diameters. All patients evaluated fulfilled the AHA diagnostic criteria for KD. Seven Z-score models were applied on all the available CA diameters. The calculated scores were compared for normal distribution and stability concerning the diagnosis of CA ectasia (CAE, Z-score >2 and ≦ 2.5) and CAA (Z-score > 2.5).

Results: In 310 out of 408 echocardiographic reports, at least one CA diameter was reported. A precise breakdown: 82% left main coronary artery (LMCA, n = 255), 73% right coronary artery (RCA, n = 226), and 24% left anterior descending (LAD, n = 74). In all but five reports of those without documented CA dimensions, the arteries were judged by the treating physician as of a normal diameter. Between all Z-scores calculated by the seven models, percentage of achieved CAA varied considerably; for LMCA, 25–45%, and for RCA and LAD, 30–39%, respectively. CAE was less frequently diagnosed; therefore, showed less variance (LMCA 8–11%, RCA 2–6%, and LAD 1.5–6%). Among all CAs and all scores evaluated most concordant results were found between the Kobayashi and Dallaire Z-score models, however, at the same time diagnosing higher rates of CAA than the other models.

Conclusion: There are considerable discrepancies between the different Z-score models available in terms of determining CAA. Since the Z-score models proposed by Kobayashi et al and Dallaire et al obtained similar and higher rates of CAA, we would propose to use either one of these models. This way a considerable amount of underdiagnosis and delayed treatment initiation might be prevented.

No conflict of interest has been declared by the author(s).