Electromechanical Properties of Perfusion/Metabolism Mismatch: Comparison of Nonfluoroscopic Electroanatomic Mapping with ¹⁸F-FDG PET

Study Population

Twenty-one consecutive patients (16 men, 5 women; mean age ± SD, 61 ± 11 y) with angiographically proven significant coronary artery disease and stable angina pectoris were included in the study. Exclusion criteria were recent myocardial infarction (within 3 wk before the endocardial mapping and ¹⁸F-FDG PET procedures), unstable angina, a LV ejection fraction < 30%, a LV wall thickness < 10 mm in any part of the left ventricle (to avoid an increased risk of perforating the thinned myocardium by an endocardial catheter), severe valvular heart disease, and severe peripheral atherosclerotic disease. Thirteen patients had experienced a previous myocardial infarction (3 ± 2 y before the study). Coronary angiography revealed multivessel disease in 13 patients. The LV ejection fraction measured by ventriculography (area length method) was 49% ± 17%.

Patients who were stratified to undergo NOGA mapping were studied by cardiac ¹⁸F-FDG PET of glucose utilization and ²⁰¹Tl myocardial resting perfusion imaging 8 ± 2 and 4 ± 2 d, respectively, before NOGA mapping. All procedures were performed according to institutional guidelines. Informed consent was obtained from all patients.

²⁰¹Tl SPECT Perfusion Imaging

To evaluate myocardial perfusion, 100 MBq ²⁰¹Tl were injected intravenously under resting conditions. Ten minutes later, SPECT image acquisition was performed using a dual-detector γ-camera with a noncircular clockwise orbit (Vertex MCD; ADAC Laboratories). A total of 32 projections (16 per head) was obtained by step-and-shoot acquisition over 180°, performing a simultaneous transmission scan with ¹⁵³Gd to obtain attenuation-corrected images. Transaxial images were reconstructed using fully automated software for attenuation and scatter correction (Vantage, ADAC Laboratories; ExSPECT, Emory University).

¹⁸F-FDG PET of Glucose Utilization

Patients were studied after fasting for at least 6 h. If fasting glucose levels were within the normal range, a 75-g oral glucose load was given to improve myocardial glucose uptake. Forty-five to 60 min later, the glucose level was determined, and 2–6 international units of insulin were injected intravenously if the glucose levels exceeded 140 mg/dL. This procedure was repeated until the glucose level was <140 mg/dL, at which time 400–500 MBq ¹⁸F-FDG were injected. Thirty to 90 min after tracer injection, image acquisition was performed on a high-resolution, full-ring PET camera (Advance; General Electric). A 1-min scout scan for correct positioning of the heart was followed by a 10-min transmission scan for attenuation correction (using a ⁶⁸Ge source) and a 30-min static emission scan for myocardial ¹⁸F-FDG uptake. Image reconstruction was performed to generate 35 transaxial images.

Analysis of SPECT and PET Images

Transaxial ²⁰¹Tl SPECT and ¹⁸F-FDG PET images were transferred to an image-processing workstation (Onyx; SGI) for image analysis. For both SPECT and PET images, transaxial images were reoriented into short-axis and long-axis images, and a polar map analysis was performed (Munich Heart software) to derive segmental tracer uptake in 12 segments (4 anatomic regions: septal, anterior, lateral, and posterior; 3 short-axis planes: apical, midventricular, and basal) in accordance with the results available from the NOGA mapping system.

To define segments without perfusion abnormality as the segments with “normal” glucose uptake (21), average segmental ²⁰¹Tl and ¹⁸F-FDG uptake values were normalized to the respective regions exhibiting the maximal ²⁰¹Tl uptake. Myocardial segments were classified into 4 groups according to viability criteria published in previous reports (17,22–26) differentiating viable myocardium (segments with normal perfusion or maintained glucose metabolism) from nonviable myocardium (hypoperfused segments with severely reduced glucose metabolism) as follows:

•Normal: Segments with ²⁰¹Tl uptake > 70% were considered as normally perfused.

All segments with reduced ²⁰¹Tl uptake (≤70%) were considered as hypoperfused and classified as:

• Perfusion/metabolism mismatch: hypoperfused myocardium with relatively increased glucose metabolism (²⁰¹Tl uptake ≤ 70%; ¹⁸F-FDG uptake > 70%),

• Perfusion/metabolism match: hypoperfused myocardium with concordantly reduced glucose metabolism (²⁰¹Tl uptake ≤ 70%; ¹⁸F-FDG uptake > 50% and ≤ 70%),

• Nonviable: hypoperfused myocardium without evidence of residual glucose metabolism (²⁰¹Tl uptake ≤ 70%; ¹⁸F-FDG uptake ≤ 50%).

NOGA Mapping

After diagnostic coronary angiography and contrast ventriculography, electroanatomic mapping was performed. Under fluoroscopic guidance, a 7-French mapping catheter with a deflectable tip was advanced into the LV cavity. Electroanatomic mapping was performed as previously described (11,27,28) to measure endocardial unipolar voltage (UpV) and local linear shortening (LLS). Polar map analysis was used to generate segmental values of the LV endocardial voltage and LLS in 12 myocardial segments indicated earlier. Thus, a direct segmental comparison of the maps for endocardial voltage and LLS with myocardial perfusion and glucose utilization was achieved (Figs. 1 and 2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Short-axis views of ¹⁸F-FDG PET (A) and ²⁰¹Tl SPECT scintigraphy (B) of 50-y-old patient with significant stenosis of circumflex artery with moderately decreased global LV function. Perfusion/metabolism mismatch is present in midlateral wall with reduced ²⁰¹Tl uptake and increased ¹⁸F-FDG uptake. Electrical activity (unipolar voltage) was nearly normal in all segments (C), whereas LLS showed decreased values in midlateral wall (D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Polar map analysis of ¹⁸F-FDG PET metabolic imaging (A) and ²⁰¹Tl perfusion imaging (B) and NOGA mapping (C, unipolar voltage; D, LLS) of patient with severe stenosis of diagonal branch of left anterior descending artery with perfusion/metabolism mismatch in basal and middle anterior and anterolateral walls with unipolar voltage values within normal range and reduced LLS values.

During NOGA mapping an average of 95 ± 28 endocardial points per patient was investigated. After filtering endocardialpoints with inadequate voltage signals due to unstable location of the catheter or due to rhythm disturbances, the average number of points drawing the LV silhouette was 78 ± 22. Since the septal and posterolateral basal segments in the polar map analysis of the endocardial mapping in most cases exhibit low voltage values of the mitral valve apparatus and the base of the heart (12), these basal segments (n = 63) were excluded from further analysis. In 17 segments, no definite interpretation could be made because of too few (<3) sampled data points. Thus, 172 (68%) of the 252 myocardial segments were available for data analysis.

Statistics

The mean ± SD is given for continuous data. Categoric variables are described as frequency distribution. ANOVA was used for data comparison between groups; regression analysis was used to compare continuous data. Receiver-operating-characteristic (ROC) analysis was used to derive the optimal threshold level (highest diagnostic accuracy) for UpV to differentiate PET viable myocardium and PET nonviable myocardium. P values < 0.05 indicated statistical significance. Statistical analyses were performed with the Statview software package (Statview 5.0.1; SAS Institute) and the CLABROC and LABROC computer software (29).

Endocardial Voltage

Endocardial voltage was 10.8 ± 4.6 mV in normal segments (n = 88), 9.3 ± 3.4 mV in hypoperfused segments with a perfusion/metabolism mismatch (n = 29), 6.9 ± 3.1 mV in hypoperfused segments with a perfusion/metabolism match (n = 40), and 4.1 ± 1.1 mV in nonviable segments (n = 15). Endocardial voltage was not significantly different between normal segments and hypoperfused segments with a perfusion/metabolism mismatch (P = not significant [NS]) but was significantly lower in hypoperfused segments with a perfusion/metabolism match and in nonviable segments (P < 0.0001; Table 1).

Relation Between Myocardial Perfusion, Glucose Metabolism, and Endocardial Voltage

Correlation between myocardial perfusion and endocardial voltage was moderate for all myocardial segments (r = 0.41, r² = 0.17, SEE = 4.1, P < 0.0001; Fig. 3A) and decreased slightly when only hypoperfused segments were included in the analysis (r = 0.38, r² = 0.14, SEE = 3.2, P < 0.001; Fig. 3C). There was no significant correlation between myocardial perfusion and endocardial voltage in hypoperfused segments with a perfusion/metabolism mismatch (r = 0.2, r² = 0.04, SEE = 3.4, P = NS; Fig. 3E).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Linear correlation between UpV by NOGA mapping and normalized ²⁰¹Tl and ¹⁸F-FDG uptake in all myocardial segments (A, ²⁰¹Tl: P < 0.0001, r = 0.41, r² = 0.17, SEE = 4.1; B, ¹⁸F-FDG: P < 0.0001, r = 0.45, r² = 0.20, SEE = 4.0), in hypoperfused segments (C, ²⁰¹Tl: P < 0.001, r = 0.38, r² = 0.14, SEE = 3.2; D, ¹⁸F-FDG: P < 0.0001, r = 0.6, r² = 0.36, SEE = 2.8), and in segments with perfusion/metabolism mismatch (E, ²⁰¹Tl: P = NS, r = 0.2, r² = 0.04, SEE = 3.4; F, ¹⁸F-FDG: P < 0.05, r = 0.45, r² = 0.21, SEE = 3.1).

In contrast, correlation between glucose metabolism and endocardial voltage was moderate when all myocardial segments were included in the analysis (r = 0.45, r² = 0.20, SEE = 4.0, P < 0.0001; Fig. 3B) and increased when only hypoperfused segments were considered (r = 0.6, r² = 0.36, SEE = 2.8, P < 0.0001; Fig. 3D). In hypoperfused segments with a perfusion/metabolism mismatch, glucose metabolism was significantly related to endocardial voltage (r = 0.45, r² = 0.21, SEE = 3.1, P < 0.05; Fig. 3F).

LLS

LLS was 9.5% ± 7.1% in normal segments, 9.6% ± 7.2% in hypoperfused segments with a perfusion/metabolism mismatch, 9.4% ± 8.3% in hypoperfused segments with a perfusion/metabolism match, and 5.4% ± 5.0% in nonviable segments. There was no significant difference between normal segments, hypoperfused segments with a perfusion/metabolism mismatch, and hypoperfused segments with a perfusion/metabolism match (Table 1).

ROC Analysis

The threshold level for endocardial voltage to identify PET viable segments with the highest achievable accuracy by ROC analysis was 5.2 mV. This threshold level was associated with a sensitivity and specificity of 85% for the detection of PET viable tissue by the mapping of endocardial voltage (Fig. 4A). ROC analysis of endocardial voltage and LLS to differentiate between PET viable segments (¹⁸F-FDG uptake > 50%) and PET nonviable segments (¹⁸F-FDG uptake ≤ 50%) showed an area under the ROC curve of 0.9 ± 0.03 and 0.69 ± 0.06, respectively (Fig. 4B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

(A) Determination of threshold value of unipolar endocardial voltage for identification of PET viable myocardium with sensitivity and specificity of 85% based on threshold of 5.2 mV. (B) ROC curves of UpV and LLS for differentiation between PET viable and PET nonviable myocardium with area under the curve of 0.9 ± 0.03 and 0.69 ± 0.06, respectively.

For the 4 anatomic regions, ROC analysis of endocardial voltage revealed an area under the curve of 0.91 ± 0.06 for the septal region, 0.91 ± 0.04 for the anterior region, 0.83 ± 0.07 for the lateral region, and 0.87 ± 0.09 for the inferior region.

Endocardial Voltage

Measurements of voltage obtained in normal myocardium (10.8 ± 4.6 mV) and in scarred segments (4.1 ± 1.1 mV) in this study are in good agreement with results of previous studies, which demonstrated that NOGA mapping can be used to identify viable myocardium (13–20) (Table 2).

Hypoperfused myocardial segments that exhibit a maintained glucose metabolism in the absence of adequate oxygen supply showed values for endocardial voltage that were similar to values derived in segments with a normal perfusion. This finding suggests that “electrical viability” is closely related to the metabolic state of the myocardial cell and can be retained even when perfusion is reduced.

A moderate correlation was found between myocardial ¹⁸F-FDG uptake and endocardial voltage (r = 0.45) when all segments were included in the analysis. The exclusion of myocardial segments with normal perfusion resulted in an improved correlation between ¹⁸F-FDG uptake and endocardial voltage (r = 0.6). In contrast, the correlation between endocardial voltage and myocardial perfusion decreased when only hypoperfused segments were investigated and became insignificant when only hypoperfused segments with a perfusion/metabolism mismatch were included. These findings support the concept of a close concordance between maintained or increased myocardial glucose metabolism and electrical activity in hypoperfused myocardium.

The present study confirms that resting perfusion imaging by ²⁰¹Tl is unable to differentiate hypoperfused segments with enhanced or preserved metabolism and preserved electrical activity from hypoperfused segments with a concordant reduction in metabolism and electrical activity.

In this study, the sensitivity and specificity of endocardial voltage for assessment of viability were 85% and, thus, comparable to data obtained by other techniques. The threshold value of endocardial voltage to differentiate viable and nonviable myocardium obtained by ROC analysis in this study was 5.2 mV and differed slightly from our previously published threshold value of 6.4 mV, which was obtained on the basis of ²⁰¹Tl late resting uptake (13). This lower threshold may partly be attributed to a lower accuracy of viability detection by ²⁰¹Tl and to differences in the study population.

LLS

Measurements of LLS were similar between normal segments, segments with a perfusion/metabolism mismatch, and segments with a perfusion/metabolism match. Accordingly, LLS cannot differentiate between normal and hypoperfused myocardium with maintained viability, as demonstrated by ROC analysis. The sensitivity and specificity of LLS for detection of viable and nonviable tissues were low, as demonstrated in previous studies (13,14,16,19) (Table 2). Therefore, our study confirms the limited value of LLS for the assessment of myocardial viability.

Limitations

This study included patients with an ejection fraction > 30%, as initially NOGA mapping with a 7-French catheter was considered to carry an increased risk of myocardial perforation in myocardium with a thinned wall. Therefore, whether the results also would apply to patients with a severely reduced LV function (<30% ejection fraction) remains to be determined.

Different criteria to define myocardial viability by ²⁰¹Tl or ¹⁸F-FDG imaging have been proposed in previous studies (6,22,23,25,32,33). In agreement with the study of Koch et al. (17), who compared NOGA mapping with SPECT and PET, in the present study a normalized ²⁰¹Tl tracer uptake > 70% was used for definition of normal perfusion and a normalized ¹⁸F-FDG tracer uptake ≤ 50% was used for definition of nonviable myocardium. Unlike an earlier study that used late ²⁰¹Tl imaging to delineate myocardial viability (13), resting ²⁰¹Tl SPECT in the present study was performed immediately after tracer injection to delineate resting myocardial perfusion.

The commercial software for NOGA analysis generates data displays in a polar map format. Therefore, segmental polar map analysis was used to achieve a direct semiquantitative comparison of results from NOGA mapping and scintigraphic imaging. However, differences may have occurred, as the 2 methods use different principles of signal processing: NOGA is based on a discrete point-per-point registration of values by a catheter, whereas scintigraphic images obtained by a tomographic acquisition represent a continuous imaging map. Moreover, small misalignments between SPECT and PET data cannot be excluded, as both studies were performed on different days and with different cameras. However, because all scintigraphic studies were processed with the same image analysis software, significant errors due to misalignment of segments are unlikely.

In NOGA mapping, the basal portion of the septal and posterolateral segments of the left ventricle are difficult to evaluate. Moreover, nondiagnostic endocardial signals due to an unstable location of the catheter or due to arrhythmias had to be excluded from analysis. Thus, NOGA mapping frequently does not cover the entire endocardial surface. Furthermore, segments with a perfusion/metabolism mismatch are not common in clinical practice, as reflected by the low number of 29 mismatch segments (17% of all measurable segments). Also, small areas of a perfusion/metabolism mismatch can be missed by segmental polar map analysis, particularly when overlapping between segments occurs.

Finally, even though the accuracy of PET viability imaging to predict functional recovery is well established (8,34), functional recovery as an endpoint of viability was not investigated independently in the present study.