Decreased Perfusion in the Lateral Wall of the Left Ventricle in PET/CT Studies with ¹³N-Ammonia: Evaluation in Healthy Adults

Participants and Setting

The research protocol was approved by the Institutional Review Board of the sponsoring institutions. The participants were part of the “Wounded Spirits, Ailing Hearts” study of the effects of posttraumatic stress disorder in American Indians. As part of an extensive evaluation, the participants underwent a physician-administered assessment that included asking about congenital heart disease, angina, coronary artery bypass surgery, percutaneous coronary angioplasty, and hospitalization for coronary heart disease. A positive answer to any question disqualified a participant from inclusion in this study. In addition, the blood pressure, heart rate, symptoms, and electrocardiograph during pharmacologic stress for PET/CT with ¹³N-ammonia were reviewed for any findings of coronary artery disease. Any indication of coronary artery disease disqualified the participant.

PET/CT Myocardial Perfusion Imaging

After a 4-h fast and placement of an intravenous line, the participant was placed in an inline PET/CT scanner with a 4-slice spiral CT component (DLS PET/CT; GE Healthcare). CT images of the lower chest were obtained in approximately 17 s during quiet breathing followed by the injection of 740 MBq (20 mCi) of ¹³N-ammonia intravenously. Four and a half minutes later, a 15-min electrocardiograph-gated set of static tomograms was obtained for evaluation of relative myocardial perfusion. The participant was allowed to get up at the end of the rest acquisition. At 45 min, the participant was again placed in the PET/CT scanner and a second CT scan of the lower chest was obtained. The participant was then stressed pharmacologically with a 45- to 60-mg infusion of dipyridamole intravenously over 4 min. Two minutes after the end of the infusion, 740 MBq (20 mCi) of ¹³N-ammonia were injected. Four and a half minutes later, a 15-min electrocardiograph-gated set of stress tomograms was obtained.

Image Analysis

The images were analyzed on a Xeleris workstation (GE Healthcare). In the mid portion of the transaxial images of the left ventricle, a vertically oriented rectangular region of interest was manually placed over the lateral wall and the average and maximum counts per pixel were recorded. The region of interest was then translated horizontally to overlie the septum, and again the average and maximum counts per pixel were recorded.

Then, using the original PET, CT, and fused images obtained transaxial to the body, the maximum extent to which the left ventricle in the PET image extended into the lung in the corresponding CT image was measured. In addition, the number of slices between the top of the right hemidiaphragm in the PET images and CT images was recorded as an indication of misalignment in the cephalocaudad direction.

Statistical Analysis

We assessed the correlation among the ratio of lateral wall counts relative to septal counts, the amount of misalignment at the left heart border, and the amount of misalignment at the top of the right hemidiaphragm using the Pearson correlation coefficient. Descriptive statistics for continuous variables included mean, SD, and range; percentages were computed for categoric variables.

Correlations were computed among counts in the lateral wall of the heart relative to the septum, degree of misalignment measured at the lateral heart border, and degree of misalignment at the top of the right hemidiaphragm while participants were at rest and under pharmacologic stress during imaging. A negative correlation indicates that as misalignment increases, the counts in the lateral wall relative to the septum decrease. For completeness, we present results for both average and maximum counts despite a strong correlation between the measures (r = 0.69 at rest, r = 0.74 at stress).

Paired t tests were used to compare the ratio of lateral wall counts relative to septal counts at rest and at stress. We also used paired t tests to compare the amount of misalignment at rest and at stress. In participants without misalignment at the left heart border, we used 1-sample t tests to examine if the ratio of lateral wall counts relative to septal counts differed from 1.0. A significant result indicates decreased counts in the lateral wall relative to the septum.

Because measures for the amount of misalignment did not follow a normal distribution, we also computed nonparametric equivalents to Pearson correlation (Spearman rank) and the paired t test (Wilcoxon signed ranks). For consistency, we present results from the parametric tests because statistical inference did not differ between the parametric and nonparametric tests.

Analyses were completed using Stata/SE 9.2 (StataCorp LP) for Windows (Microsoft). All statistical comparisons were 2-sided, and we used a P value of 0.05 as the threshold for significance.

Participants

All 32 participants had no history or current evidence of heart disease and were included in analyses. Most of the participants were female (81%), and the average age was 44 ± 9 y (range, 28–61 y).

Data Analysis

The amount of misalignment at the left heart border ranged from 0.0 to 11.3 mm at rest and 0.0 to 18.0 mm at stress. For the left heart border, 41% of participants at rest and 62% of participants at stress had misalignment (>0 mm). The amount of misalignment at the top of the right hemidiaphragm ranged from 0 to 2 slices at rest and 0 to 4 slices at stress. For the top of the right hemidiaphragm, 28% of participants at rest and 53% of participants at stress had misalignment (>0 slices).

Correlations among the ratio of lateral wall counts relative to septal counts, misalignment at the left heart border, and misalignment at the top of the right hemidiaphragm are shown in Table 1. At rest, we found significant negative correlations between lateral wall counts relative to the septum and misalignment at the left heart border for both average (r = −0.45) and maximum (r = −0.52) counts. The same association at stress was significant for maximum (r = −0.43), but not average (r = −0.26), counts. The correlation at rest between the amount of misalignment at the left heart border and the amount of misalignment at the top of the right hemidiaphragm was positive (r = 0.31) but not significant (P = 0.09). However, the same correlation at stress was significant (r = 0.45, P = 0.01).

Correlations between the lateral wall counts relative to the septum and the amount of misalignment at the top of the right hemidiaphragm were smaller than analogous associations at the left heart border and not significant.

Differences between rest and stress in the relative counts in the lateral wall, compared with the septum, the degree of misalignment at the left heart border, and the degree of misalignment at the top of the right hemidiaphragm, are presented in Table 2. Participants had a significantly lower lateral wall count relative to the septum at stress than at rest for both average (P < 0.01) and maximum (P = 0.02) counts, greater misalignment at the left heart border at stress than at rest (P = 0.02), and greater misalignment at the top of the right hemidiaphragm at stress than at rest (P < 0.01).

Finally, we examined the counts in the lateral wall relative to the septum in participants with no misalignment at the left heart border (Table 3). Among the 19 participants with no misalignment at rest, the average and maximum counts were lower in the lateral wall than in the septum (P = 0.03 and P = 0.01, respectively). Among the 12 participants with no misalignment at stress, both the average and the maximum counts were lower in the lateral wall than in the septum (P < 0.01 for both).