Comprehensive reference ranges for cardiovascular magnetic resonance: time to move on from single centre data?

Abstract

Cardiovascular magnetic resonance (CMR) provides gold standard, and often unique, measurements of cardiovascular structure, function and tissue character. Fundamental to such capabilities are clearly defined normal ranges. This study aimed to (1) Determine normal ranges for an extensive set of CMR measurements, and the inter-scan reproducibility of these measurements; (2) Determine the impact of common variations in practice, and; (3) Systematically evaluate the findings in the context of published reference ranges. One hundred and 22 healthy adults, including a minimum of 10 males and 10 females per age decile, underwent assessment including CMR (3 T, Siemens). Twenty participants returned for a second CMR. Image analysis was performed using cvi42 by experienced observers. Age- and sex-specific reference ranges, in tabular and normogram formats, and their interscan reproducibility, are provided for left ventricular mass, wall thickness, volumes and ejection fraction; right ventricular volumes and ejection fraction; longitudinal, radial and circumferential LV strains; atrial area, volume and strains; native T1, T2, T2*, aortic distensibility and pulse wave velocity. Measurement reproducibility improved when baseline scans were used for reference, e.g., for basal slice selection. Myocardial T1 was the most reproducible of all CMR measurements. Common variations in practice resulted in significant measurement differences e.g., indexed left atrial volume was larger (47.3 vs 40.3 ml/m², P < 0.0001), and its measurement less variable, when measured from atrial short-axis cine stacks compared to biplanar measurement from 4- to 2-chamber cines. Studies using similar methods to define normal ranges demonstrate clinically-relevant differences in the normal ranges produced. A comprehensive set of age and sex specific CMR reference ranges are provided, along with inter-scan reproducibility and the impact of common variations in practice. Single centre studies, whilst meticulous in design and delivery, result in clinically-relevant variations in normal ranges. We advocate that larger cohorts, including diverse ethnicities, such as the Healthy Hearts Consortium, may be a better approach to defining normal ranges for common CMR measurements.