Multivariable Assessment of Coronary Artery Disease Using Cardiac CT Imaging

Background

Multidetector computed tomography (MDCT) is the most recent addition to the arsenal of cardiac imaging modalities. With its unparalleled spatial resolution and well established techniques for contrast enhancement using conventional iodine-based agents, it allows visualization of coronary arteries and is thus increasingly used as an alternative to invasive coronary angiography (ICA) (de Roos A. et al. 07,Deetjen et al. 07,Schroeder et al. 08). The diagnostic value of noninvasive coronary angiography (CTCA) has been established against conventional techniques used for the diagnosis and evaluation of coronary artery disease (CAD), including ICA (Budoff et al. 07,Leber et al. 05,Raff et al. 05,Rubinshtein et al. 07) and SPECT myocardial perfusion imaging (MPI) (Hacker et al. 07,Rubinshtein et al. 07,Schuijf et al. 06). Nevertheless, the physiological significance of intermediate grade stenosis detected by CTCA in individual patients is unknown and such patients are routinely referred for stress testing in order to define individual therapeutic strategy. It has been suggested that intramyocardial distribution of contrast during the arterial phase of enhancement may be related to myocardial perfusion (Cury et al. 07). Several studies have demonstrated hypo-enhanced areas corresponding to myocardial scar tissue in a small number of patients post myocardial infarction (MI) (Gerber et al. 06,Henneman et al. 06,Mahnken et al. 05,Nieman et al. 06,Nikolaou et al. 05), and in animal models of acute MI (George et al. 07,Gerber et al. 06,Hoffmann et al. 04,Lardo et al. 06). Our hypothesis is that perfusion information, which can be extracted from images acquired for CTCA without additional radiation exposure or contrast load, could be a useful addition to the MDCT evaluation of ischemic heart disease (IHD).

Accordingly, we recently completed a study designed to determine the value of MDCT assessment of resting myocardial perfusion in consecutive patients referred to CTCA. In this study, we developed and tested a new technique for quantitative assessment of myocardial perfusion based on analysis of MDCT images acquired for CTCA. The accuracy of resting MDCT perfusion was tested against ICA as well as MPI. Both protocols included a detailed investigation of the sources of inter-technique discordance.

Comparisons against ICA revealed that the majority of perfusion abnormalities detected on MDCT images at rest were associated with either prior MI, as previously reported (Gerber et al. 06,Henneman et al. 06,Mahnken et al. 05,Nieman et al. 06,Nikolaou et al. 05), or reduced blood supply secondary to significant stenosis. This previously unknown finding may have important clinical implications in the context of detection of myocardial ischemia. Although comparisons against resting MPI data showed high levels of agreement, we noted a large number of perfusion defects that were not confirmed by resting MPI. These apparent "false positive" findings were found to be either directly related to suboptimal image quality or were true positives when compared to stress MPI. This latter surprising finding may probably be explained by the effects of nitroglycerin used during MDCT imaging, as well as possible vasodilating effects of the iodine-based contrast media (Limbruno et al. 00), which may to some extent mimic those of vasodilator stress agents used during MPI, namely adenosine or dipyridamole.

The main conclusion of these recent studies was that future studies are needed to explore the full diagnostic potential of MDCT perfusion when used in combination with vasodilator stress.

Objectives

Accordingly, we are planning a new study in which MDCT imaging will be performed during vasodilator stress in consecutive patients referred for clinically indicated CTCA. Myocardial perfusion will be assessed using quantitative volumetric analysis of myocardial x-ray attenuation and compared to either ICA or MPI findings in a subgroup of patients who also undergo one of these tests.

Methods

We will prospectively study 120 consecutive patients referred to CTCA for the evaluation of CAD. MDCT imaging will be performed according to the standard clinical protocol, which will be modified to include the vasodilator stress agent Regadenoson (Astellas Pharmaceutical) recently approved by the FDA for clinical use. This selective A2A agonist will be administered according to the manufacturer's guidelines. Imaging will be performed during its peak effect.

Standard contraindications to CTCA will be observed, including known allergies to iodine, renal dysfunction (creatinine >1.4 mg/dL), inability to perform a 10 sec breath-hold. Images will be obtained using an MDCT scanner (256-channels, Philips) using retrospective ECG-gating. A nonionic iodinated contrast agent (Omnipaque-350, Amersham) will be injected into a right antecubital vein (80-120 ml depending on body weight, at 5-6 ml/sec), followed by a 20-50 ml chaser bolus (70% saline, 30% contrast, at 5 ml/sec). Image acquisition will be triggered by the appearance of contrast in the descending thoracic aorta, and performed during suspended respiration.

Additional set of images will be acquired 10 min later in order to visualize delayed contrast enhancement, which is used to estimate viability in hypoperfused myocardium. This set of images will be acquired without injection of contrast or Regadenoson. Prospective ECG-gating will be used to obtain a single phase of a cardiac cycle in order to minimize total radiation dose.

Regional MDCT perfusion measurements

Volumetric MDCT perfusion analysis will be performed using custom software from the same phase of the cardiac cycle used for CTCA (75% of RR interval in most patients). Semi-automated detection of the endo- and epicardial surfaces will be performed based on the level set approach, as described previously (Corsi et al. 05), and the myocardium will be divided into 16 segments (6 basal, 6 mid-ventricular, 4 apical) using standard segmentation. In each 3D myocardial ROI, mean x-ray attenuation will be measured and divided by the mean attenuation measured in the corresponding ROI in the control group of normal subjects. This normalization will compensate for inter-segmental heterogeneity in x-ray attenuation. The resultant value will then multiplied by the ratio between the mean of the highest three attenuation values measured in the control group and in the individual patient. This rescaling will compensate for differences in contrast levels between patients. The resultant value will be used as the MDCT myocardial perfusion index.

Objective detection of regional MDCT perfusion abnormalities

By definition, MDCT perfusion index (subendocardial and transmural) obtained in the control group approximately equal to 1 in all segments. The SD of this index averaged over the 16 segments, SD16, will be used to determine the threshold for automated detection of perfusion abnormalities, which will be defined as [1-SD16] for all segments. To this effect, in each patient, segments in which the perfusion index is below this threshold will be considered abnormal. A territory of an individual coronary artery will be considered abnormal when the perfusion index is abnormal in at least one segment. For the patient-by-patient analysis, abnormal perfusion will be diagnosed when at least one territory is abnormal.

Inter-technique comparisons

Coronary anatomy depicted on each patient's MDCT volume rendering of the heart will be used to determine the perfusion territory of each artery and its major branches, i.e. to assign each myocardial segment to the territory of a specific coronary artery. Inter-technique comparisons will be performed on a segment-by-segment, vascular territory and patient-by-patient basis. Inter-technique agreements will be assessed by counting concordances (true positive and true negative) as well as discordances (false positive and false negative) on a segment, vascular territory and patient basis. For every comparison, these counts will be used to calculate sensitivity, specificity, positive and negative predictive values (PPV, NPV) and overall accuracy.

Anticipated results

We anticipate that approximately 60% of the study patients will have either MPI or ICA (or both) data available as a reference for comparisons with MDCT. We anticipate that combining MDCT imaging with vasodilator stress will prove to be highly feasible and that perfusion abnormalities detected on MDCT images will correlate with the findings of stress MPI and/or ICA.

Significance

To our knowledge, this will be the first study to validate quantitative MDCT evaluation of myocardial perfusion imaging with vasodilator stress against MPI/ICA reference in consecutive patients referred for CTCA. Because the addition of stress perfusion information will allow elucidating the clinical significance of coronary lesions in the same test, such addition promises not only to improve the accuracy of cardiac CT in the diagnosis and evaluation of IHD, but is also likely to prove as a cost-effective, single-stop alternative to costly serial testing. We anticipate that the results of our study will support the use of this methodology in every patient referred for CTCA, similar to the routine use of vasodilator stress with MPI.