Editorial for “Traveling Volunteers – A Multi‐Vendor, Multi‐Center Study on Reproducibility and Comparability of 4D Flow Derived Aortic Hemodynamics in Cardiovascular Magnetic Resonance”

Because of numerous technical and software innovations, we have a vast armamentarium of tools for noninvasive and nonionized characterization of flow with magnetic resonance imaging (MRI). Early concepts included displacement imaging of blood/tissue, followed by time of flight feature tracking or tagging. Parallel imaging advances have increased the amount of information available and the image resolution while maintaining the same or shortened acquisition time. Breakthroughs in phase contrast and velocity encoding have enabled us to quantitatively analyze flow in two or three dimensions.1 Multiple velocity-encoded phase-contrast acquisitions offered analyses on very short time scales and in-between cardiac beats.2 Combination of these tools has allowed the three-dimensional (3D) analysis of flow in any tissue or direction, and the comprehensive assessment of complexed time resolved dynamics (4D).2 The use of higher magnetic fields (3–7 tesla) has increased the number of convoluted sets of data available for a wide range of technical, pathophysiologic, and clinical utilizations.2 Modern MRI machines allow convenient examination of any type of flow in nearly every dimension and direction regardless of the anatomy, potential barriers (i.e., bone structures), type of vessel (i.e., large or small arteries/veins), and tissue status (ruptured or shunted). Conventional MRI use computes analyses for velocity, pressure, volume, and others, similar to noninvasive Doppler ultrasound imaging or contrast angiography, the latter of which uses ionized radiation. Despite the reasonably expected benefits from analyses of rather large vessels and flow through cardiac valves, a completely new world was discovered. Inconspicuous alternations of flow detected by various advanced flow MRI sequences were revealed as pathophysiologic distinctions for different types of heart and pulmonary vessel diseases. Several groups of authors reported on changes and inefficient blood flow propagation in the ventricles of patients with ischemic heart disease and dilated cardiomyopathy.3, 4 Furthermore, alternations of flow were found in both ventricles of children with pulmonary artery hypertension (PAH), whilst a significant response was recorded in the right ventricle (RV) after inhaled nitric oxide challenge.5 Patients with PAH presented with increased RV dimensions, increased RV residual volumes, and decreased direct RV inflow and RV systolic function.5 François et al analyzed 4D MRI velocity mapping in patients with repaired tetralogy of Fallot vs. controls.6 In this study, patients with repaired tetralogy had increased diastolic over systolic flow in both caval veins into the RV, increased diastolic vortical inflow patterns in right atrium and RV, and increased vortical and helical flow features in pulmonary arteries compared to controls.6 The 4D characterization of flow is potentially promising as it was shown to be in correlation with functional grade.3, 4 Subtle alternations of flow to some degree precede the infrastructural geometric remodeling of various heart diseases such as ischemia or cardiomyopathies.7 These expansions of flow MRI studies are encouraging; however, it is important to point out that currently there is a general lack in standardization among various machines, software, centers, as well as the evidence-based merits on benefits and safety of clinical applications. In the present study, the authors offered an innovative approach using a multicentric, multivendor (three vendors) analysis of 15 patients, with scan–rescan control.8 The aim was to improve the knowledge of flow MRI clinical confounders, compare differences in MRI machines, and quantify the interobserver agreement of the method. This would be the first study systematically reporting on sources of differences in order to define clinical reproducibility of MRIs for clinical 4D hemodynamic analysis.8 Currently, there is an abundance of available protocols, pre- and post-processing software, center-dependent features, and magnetic fields for the MRIs, all of which may potentially perplex the results and clinical interpretation of findings. Furthermore, the optimized protocols and referral points of interest for studies on aorta and aortic valve analyses are still not unanimously defined.9 Hemodynamic data of interest includes peak velocity, forward volume, and average and maximal wall shear stress (WSS) through nine points of the thoracic aorta.8 Planes P3–P9 showed significant difference among vendors for forward flow (P < 0.01), maximal WSS (P < 0.05), average WSS (P < 0.05), and for peak velocity among scanners I vs. III (P < 0.05).8 Location of plane at the sinotubular junction demonstrated the higher intervendor stability. Scan–rescan reproducibility and intra- and inter-observer variability were good to excellent (intraclass correlation coefficient (ICC) 0.8) with the best agreement being seen for forward flow volume (ICC ≥ 0.98).8 This study is of premiere importance as it is the first study that measured and recognized differences among various MRI machines, all of which must be taken into account when making comparisons or designing timely relevant therapeutic interventions. Even though the results of the study are promising, they are still tentative. Further explorations of results would be valuable, if taken in stepwise manner through several head-to-head settings. There would be special interest around estimations of the accuracy of various vendors to the reference standard and clinical endpoints. Also, a prospective validation with inclusion of a greater number of cases, within greater velocity range (at least in the range of 1–6 m/second), which would sufficiently cover clinical spectrum of diseases like aortic stenosis would be beneficial.10