Which quantitative perfusion estimation model is better at diagnosing myocardial ischaemia? A CE-MARC sub-study
© Biglands et al.; licensee BioMed Central Ltd. 2014
Published: 16 January 2014
There are multiple methods for quantifying myocardial blood flow from dynamic contrast enhanced MRI (DCE-MRI) cardiac perfusion data sets. Currently there is no documented evidence to suggest the superiority of any of these models for diagnosing myocardial ischaemia. The aim of this study was to compare the diagnostic performance of four such methods.
This was a retrospective sub-study using data from the CE-MARC trial (Greenwood et al., Lancet, 2012). A 50 patient sample of patients were selected such that the distribution of risk factors and disease status within the sample was representative of the full CE-MARC cohort. Quantitative myocardial blood flow (MBF) estimates were obtained from the MRI data using four previously proposed models, commonly used in the quantitative cardiac perfusion literature. These models were: Fermi-constrained deconvolution, model independent deconvolution, the uptake model and the one compartment model. Myocardial Perfusion Reserve (MPR) ratios were calculated from the ratio of stress to rest MBF estimates. The presence of myocardial ischaemia was assessed using the consensus diagnosis of invasive, quantitative X-ray angiography and myocardial Single Photon Computed Tomography (SPECT) imaging. This provided a unique gold-standard combining independent anatomical and functional diagnostic measures. Receiver Operator Characteristic (ROC) curves were generated for each perfusion model using 1) the MPR, and 2) the stress MBF as the diagnostic measure. A DeLong, DeLong, Clarke-Pearson comparison was used to test for statistically significant differences in the Area Under the Curve (AUC) values of the four models.
There is no evidence to show that any of the models are superior in the diagnosis of myocardial ischaemia. However, the one compartment model should be avoided when using MPR as the diagnostic measure.
Theis work was funded by an NIHR doctoral training fellowship.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.