Quantitative detection of myocardial ischaemia by stress echocardiography; a comparison with SPECT

The echocardiographic equipment used was a Sonos 5500 (Philips, Andover, Massachusetts, USA) with S3 probe and RTP using the power modulation angio-mode. Patients were examined in a left lateral recumbent position. The second-generation contrast agent Sonovue^® was infused in the left decubital vein using an infusion pump dedicated for this purpose (VueJect^® Esaote, Genova, Italy; Bracco™, Milan, Italy), which automatically rotates the syringe to prevent sedimentation. The infusion rate of Sonovue was set between 1.0 and 1.3 ml/min [27]. Adenosine and echo contrast were infused in the same peripheral venous catheter, using a separate infusion pump through a three-way tap. Adenosine was given at an infusion rate of 100 μg/kg/min during one minute, after which the infusion rate was increased to 140 μg/kg/min.

All 67 patients underwent RTP imaging (mechanical index = 0.1) during infusion of echo contrast, at rest and after a minimum of one minute of hyperaemia during adenosine stress (at 140 μg/kg/min). Image acquisition was started after at least one minute of Sonovue infusion. RTP image loops containing 8–10 heartbeats were collected from the parasternal long-axis and apical four- and two-chamber views, respectively. At the beginning of each loop a destruction impulse of 10 high mechanical index frames (mechanical index = 1.5) were given to destroy all contrast micro bubbles in the myocardium [28].

During RTP the angio-mode gain was set between 60 and 70%, depending on what was suitable for the individual patient as judged by a visual on-line assessment, and 2D greyscale gain was set at zero. Focus was set close to the base of the left ventricle. All images were stored digitally for later off-line analysis.

Qontrast^® was used to produce parametric images of contrast replenishment values for the RTP loops collected from both rest and stress. Two points were manually placed in the left ventricular cavity of the perfusion images. The first point was placed in the centre of the cavity where the apex "half-circle" ends, i.e. approximately two thirds from the base of ventricle, where it was always inside the cavity (never in the myocardium) during the complete loop. The point was placed in a cavity area that was fully opacified directly after the destruction impulse in the beginning of the loop, as well as throughout the entire RTP-loop, since this formed the basis of the maximum image contrast intensity reference-point. Any isolated frames not fulfilling these criteria were excluded from analysis. The second point was placed at the base of the ventricle, enabling the software to automatically outline the complete left ventricle, including both cavity and myocardium, with dotted "M-mode" lines crossing perpendicular through the myocardial wall (Figure 1).

The first frame was selected to be the one directly after destruction impulse frames.

Automatic perfusion analysis was then started. Qontrast^® uses an advanced image processing technique that recognizes the coherence of the dynamic image sequence in a space-time domain. The technique enables tracking of the myocardial pixel movement throughout the cardiac cycle and the entire RTP-loop. This increases the accuracy of the perfusion evaluation compared to triggered imaging, due to the higher number of quantifiable frames.

Three parametric images were then automatically generated from the perfusion analysis, displaying either the peak signal intensity (A), myocardial blood flow velocity (β) or myocardial blood flow (Axβ). These were generated for each pixel, from the replenishment curve of each pixel, according to the replenishment curve A = A(1-e^-βt) [29]. These parametric images were generated from RTP images in apical four- and two-chamber and parasternal long-axis view, at rest and stress, respectively.

To acquire quantitative values of A, β and Axβ, region of interests were manually traced both at rest and stress, corresponding to the distribution territories of the three main coronary arteries; left anterior descending (LAD), left circumflex (LCx) and right coronary artery (RCA) (Figure 2). Since earlier studies indicated that β is the most sensitive quantitative parameter [24, 30], special care was taken that the tracing would align correctly in the parametric β image, avoiding red areas. These either correspond to contrast in the left ventricular cavity, are due to perfusion artefacts originating from main coronary arteries, or are caused by mathematically generated high β-values due to very low A-values, which predominantly occur in rest images where the A-values are low for physiological reasons.

Comparing A, β and Axβ values at rest and stress (hyperaemia), the corresponding reserve values (A-r; β-r and Axβ-r) were derived by dividing the stress value with the matching rest value, thus resembling invasive measurement of coronary flow reserve. Accordingly, this generated three A-r, β-r, and Axβ-r values from the LAD territory, originating from the three different echocardiographic views (four-chamber, two-chamber, and long-axis views), two reserve values each from the LCx territory (four-chamber and long-axis views) and two from the RCA territory (four- and two-chamber views). The lowest A-r, β-r and Axβ-r value from any of the views corresponding to a coronary territory was selected for the ischaemia comparison with SPECT, since the lowest reserve value should originate from the most ischaemic or the least perfused territory.

Two separate visual image interpretation were performed with off-line analysis of myocardial perfusion and wall motion at RTP-ASE, using the EnConcert Image Diagnosis Application (Philips, Andover, Massachusetts, USA). The first visual analysis was a combined analysis of perfusion and wall motion. The second visual analysis, performed separately, was an analysis excluding wall motion to estimate the value of sole perfusion analysis. Visual and quantitative perfusion analyses were done on separate occasions, blinded to one another and blinded to the result of SPECT. Each segment of the left ventricular myocardium was attributed to the same coronary vessel territory at all analyses. Myocardial ischaemia was visually evaluated comparing rest and stress images, using both perfusion and wall motion analysis in a complementary manner. A visually detected perfusion defect during stress was used as the principal marker of ischaemia. Thus, a myocardial segment was considered ischaemic if perfusion was impaired in the stress images, compared to the rest images [8]. Perfusion defects were analyzed during the first four beats after the destruction impulse at rest and after two beats at peak stress. Wall motion was used in addition to reveal perfusion defect artefacts at rest and to evaluate segments with suspected perfusion artefacts at stress. Since wall motion should not be normal if a segment has a true perfusion defect at rest, a perfusion defect at rest was considered to be an artefact when wall motion was normal in that segment. A perfusion defect at peak stress was considered to be an artefact if there was a suspicion of a perfusion artefact, such as lateral or anterior shadowing from ribs or lungs, or basal segments shadowed by contrast. In such segments, the ischaemic evaluation was based on wall motion analysis alone. If wall motion decreased at stress compared to rest images, the segment was considered ischaemic. Since perfusion can be decreased without a decrease in wall motion at ASE, the use of solitary wall motion analysis in segments with perfusion artefacts might decrease the sensitivity with regard to ischaemia. However, this complementary use of wall motion analysis increases the number of interpretable segments without negatively affecting specificity [12, 18]. If there was disagreement between readers with regard to one segment, this particular segment was re-assessed in a joint reading until consensus was reached.

The rest and stress studies were performed using a 2-day protocol, starting with injection of 600 MBq ^99mTc-tetrofosmin at stress. Stress was simultaneous with the RTP-ASE. Normal findings at stress were not followed by a rest study [31, 32]. Pathological stress studies were followed by a rest study with injection of 800 MBq ^99mTc-tetrofosmin. Patients who had cardiac medications, which could interfere with the stress test, were informed to have their medication interrupted prior to the stress test. The decision whether to interrupt the drug administration was at the discretion of the referring physician. A five-minute adenosine infusion protocol was used. Starting the infusion with 100 μg/ml/min of adenosine for 1 minute, the dose was then increased to 140 μg/ml/min for two minutes before injecting ^99mTc-tetrofosmin. Infusion of adenosine was continued for 2 min after the injection of ^99mTc-tetrofosmin. The scintigraphic data were acquired one hour after the end of the stress test, using continuous SPECT over 180 degree elliptical rotation from the 45 degree right anterior oblique position, with a dual-head gamma camera (Siemens AG Medical Solutions, Erlangen, Germany). Low energy high-resolution collimator and a zoom factor of 1.0 were used. We obtained 64 projections in a 128 × 128 matrix, with an acquisition time of 20 s per projection. Tomographic reconstruction and calculation of short axis slice images were performed using Siemens software. A two-dimensional Butterworth pre-reconstruction filter was used with critical frequency of 0.35, order 5. For each patient, the same sets of short axis slices were then processed with an automatic software package (4D-MSPECT) on a Siemens e.soft workstation. The software package defined apex and base and generated, coronal, longitudinal, sagital tomographic slices as well as polar maps with schematic map of the territories of the main coronary arteries used for scoring. Radiotracer uptake of the vascular segments were scored visually and stress images were compared with rest images regarding ischaemia and infarct. The specialist in nuclear medicine who performed the scoring was blinded to the results of the RTP analysis.