Impact of Diabetes Mellitus on Prediction of Clinical Outcome After Coronary Revascularization by ¹⁸F-FDG SPECT in Patients with Ischemic Left Ventricular Dysfunction

Patient Population

The study population consisted of 130 consecutive patients with ischemic LV dysfunction who were already scheduled for surgical revascularization. The decision to perform revascularization was based on clinical grounds (symptoms, presence or absence of ischemia, coronary anatomy). Patients with primary cardiomyopathy or concomitant significant valvular disease were not included. All patients presented with heart failure as the predominant symptom, whereas 98 patients (75%) had accompanying angina pectoris. All patients were screened for diabetes mellitus before they were included in the study. A total of 34 patients had diabetes mellitus, defined according to the American Diabetes Association criteria (7). Of these 34 patients, 12 had diabetes mellitus type 1 (primarily due to pancreatic islet β-cell destruction) and 22 had diabetes mellitus type 2 (insulin resistance with an insulin secretory defect) (7). The remaining 96 patients did not have diabetes.

Study Protocol

The patients were prospectively studied according to a structured protocol. On being entered into the study, all patients underwent radionuclide ventriculography to assess left ventricular ejection fraction (LVEF), resting 2-dimensional echocardiography to identify dysfunctional myocardial tissue, and dual-isotope simultaneous-acquisition (DISA) SPECT, including ^99mTc-tetrofosmin to evaluate perfusion and ¹⁸F-FDG imaging to assess glucose use. Nine to 12 mo after revascularization, radionuclide ventriculography and resting 2-dimensional echocardiography were repeated. The local ethics committee approved the protocol, and all patients gave informed consent.

Assessment of LVEF

LVEF was assessed by radionuclide ventriculography. A small-field-of-view γ-camera (Orbiter; Siemens) was used, oriented in a 45° left anterior oblique position with a 5°–10° caudal tilt. After injection of ^99mTc-pertechnate–labeled autologous erythrocytes (550 MBq), radionuclide ventriculography was performed with the patient at rest and supine. LVEF was calculated by standard methods (Odyssey VP; Picker). An improvement in LVEF by at least 5% was considered significant.

Assessment of Regional Contractile Dysfunction

For echocardiography, a Hewlett-Packard Sonos-5500 imaging system was used, equipped with a 1.8-MHz transducer using second-harmonic imaging to optimize visualization of the endocardial border. Cine loops in 4 standard views were digitized (parasternal long and short axes, apical 2- and 4-chamber views). Two experienced readers, unaware of any other data, reviewed the pre- and postrevascularization images off-line in random order. The left ventricle was divided according to the standard 16-segment model suggested by the American Society of Echocardiography (8). Regional wall motion and systolic wall thickening were scored using a 5-point grading scale: 1 = normal, 2 = mildly hypokinetic, 3 = severely hypokinetic, 4 = akinetic, and 5 = dyskinetic. Segments with a wall motion score greater than 2 were considered dysfunctional. Improvement of segmental wall motion score after revascularization by at least 1 grade was considered significant, with the exception of improvement from dyskinesia to akinesia after revascularization.

Assessment of Perfusion and Glucose Metabolism

For optimal ¹⁸F-FDG uptake in the heart, all patients were asked to remain on a low-fat diet for the 24 h before the study. On the day of the study, the patients were allowed to have a light breakfast. Patients with diabetes mellitus were instructed to continue taking their antidiabetic medication. In all patients, venous blood samples ware taken at baseline to determine the glucose plasma concentration. In patients with a plasma glucose level greater than 9 mmol/L at baseline, 6 units of insulin (or more if considered necessary) were administered subcutaneously. One hour after intravenous injection of ^99mTc-tetrofosmin (600 MBq), standard SPECT using low-energy collimators was performed to evaluate resting perfusion as described previously (9). DISA SPECT, to evaluate perfusion and myocardial glucose use, was performed on all patients after acipimox administration. Acipimox enhances myocardial ¹⁸F-FDG uptake by reducing the plasma level of free fatty acids (10–12). All patients received two 250-mg oral doses of acipimox (Byk). The first dose was given immediately before the start of standard ^99mTc-tetrofosmin SPECT; the second dose was given immediately after completion of that acquisition. This schedule was chosen to ensure sufficiently low levels of serum free fatty acids at the time of ¹⁸F-FDG injection and to keep those levels low for a sufficient time during the period of ¹⁸F-FDG accumulation. Thirty minutes after standard ^99mTc-tetrofosmin SPECT, the patients received a low-fat, carbohydrate-rich meal. This small meal further enhanced myocardial ¹⁸F-FDG uptake by stimulating endogenous insulin release. In all patients, venous blood sampling was repeated to determine the plasma glucose concentration. In patients with a plasma glucose level greater than 9 mmol/L, additional insulin was administered. Sixty minutes after the meal, and immediately after reassessment of plasma glucose concentration, ¹⁸F-FDG (185 MBq) was injected and, after an additional 45 min to allow cardiac ¹⁸F-FDG uptake, DISA SPECT was performed. Perfusion and metabolic imaging were performed with the patient at rest without stressors.

A triple-head γ-camera (Prism 3000XP; Philips) was used. The camera system was equipped with commercially available high-energy 511-keV collimators (13). The energies were centered on the 140-keV photon peak of ^99mTc-tetrofosmin with a 15% window and on the 511-keV photon peak of ¹⁸F-FDG with a 15% window. Data were acquired over 360° (120 sectors of 3°), with the patient supine. Total imaging time was 32 min. Data were stored in a 64 × 64, 16-bit matrix. From the raw scintigraphic data, 6-mm-thick (1 pixel) transaxial slices were reconstructed by filtered backprojection using a Butterworth filter (cutoff frequency, 0.17 cycle per pixel; order, 3.5). Attenuation correction was not applied. Further reconstruction yielded standard short- and long-axis projections perpendicular to the heart axis. The ^99mTc-tetrofosmin and ¹⁸F-FDG data were reconstructed simultaneously to obtain an exact alignment of the perfusion and metabolic images.

The perfusion and ¹⁸F-FDG short-axis slices were displayed in polar maps, which were normalized to the region with maximum perfusion activity (set at 100%); the polar maps were divided into 16 segments matching the echocardiographic segments (8,9). Two experienced observers evaluated the SPECT images. The segments were divided into 4 groups based on the tracer activities: 0 = normal tracer uptake (activity > 75%), 1 = mildly reduced tracer uptake (activity ≤ 75% and > 50%), 2 = moderately reduced tracer uptake (activity ≤ 50% and > 25%), and 3 = severely reduced or absent tracer uptake (activity ≤ 25%). Dysfunctional segments (identified by resting echocardiography) were subsequently evaluated for viability. Segments with normal perfusion (^99mTc-tetrofosmin score of 1) and segments with a perfusion defect (score ≥ 2) but relatively increased ¹⁸F-FDG uptake (¹⁸F-FDG score > perfusion score; mismatch pattern) were considered viable. Segments with a perfusion defect and concordantly reduced ¹⁸F-FDG uptake (match pattern) were considered nonviable.

Statistical Analysis

Continuous data were expressed as mean ± SD, and percentages were rounded. Statistical analysis was performed with the BMDP statistical software package (BMDP Statistical Software Inc.). Continuous variables were compared using the Student t test for unpaired samples. Differences between proportions were compared using the χ² test. A P value of <0.05 was considered statistically significant.

Patient Characteristics

Baseline characteristics were comparable between the patients with diabetes mellitus and those without diabetes mellitus (Table 1). However, in the group of patients with diabetes mellitus, relatively more women were present (26% vs. 10%, P < 0.05). All patients presented with heart failure symptoms and had impaired LV function because of chronic coronary artery disease. The majority (123 patients, or 95%) had a previous myocardial infarction. At baseline, mean plasma glucose level was 11 ± 3 mmol/L in patients with diabetes and 5 ± 0.8 mmol/L in patients without diabetes. After subcutaneous administration of insulin (according to the SPECT protocol) and immediately before ¹⁸F-FDG injection, the mean plasma glucose level was 8 ± 3 mmol/L in patients with diabetes. Figure 1 shows examples of images obtained with DISA ^99mTc-tetrofosmin/¹⁸F-FDG SPECT.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

Examples of images obtained with DISA ^99mTc-tetrofosmin/¹⁸F-FDG SPECT after oral administration of acipimox. Baseline ^99mTc-tetrofosmin (A) and ¹⁸F-FDG (B) vertical long-axis images of patient with diabetes mellitus show perfusion defect in inferior region, with relatively preserved ¹⁸F-FDG uptake (mismatch pattern) indicating myocardial viability. Baseline ^99mTc-tetrofosmin (C) and ¹⁸F-FDG (D) vertical long-axis images of patient without diabetes mellitus show large perfusion defect in apical region with absent ¹⁸F-FDG uptake (match pattern), which indicates absence of myocardial viability.

Presurgical Assessment of Myocardial Viability

Regional contractile function, as assessed by 2-dimensional echocardiography at rest, was normal in 868 segments (42%) and abnormal in 1,212 (58%). Of the 1,212 dysfunctional segments, 676 were severely hypokinetic, 521 akinetic, and 15 dyskinetic. Patients with and without diabetes mellitus had a similar number of dysfunctional segments per patient (Fig. 2). Also the number of severely hypokinetic, akinetic, and dyskinetic segments was similar in both patient subsets. Metabolic imaging with ¹⁸F-FDG SPECT showed that 610 of 1,212 (50%) dysfunctional segments were viable. Patients with and without diabetes mellitus had a comparable number of dysfunctional but viable segments per patient (Fig. 2). A substantial amount of dysfunctional but viable tissue (≥4 viable segments) was present in 14 (41%) patients with and in 52 (54%) patients without diabetes mellitus (not statistically significant).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Bar graph illustrating number of dysfunctional segments and number of viable segments, which were comparable in patients with and in patients without diabetes mellitus (neither is statistically significant). Values are expressed as mean ± SD. DM = diabetes mellitus; + = present; − = absent.

Functional Outcome After Revascularization

Functional outcome after coronary revascularization in patients with and patients without diabetes mellitus in relation to the presence or absence of substantial myocardial viability is demonstrated in Table 2. After coronary revascularization, angina pectoris symptoms (Canadian Cardiovascular Society score) improved significantly; symptoms also improved in patients without substantial myocardial viability (those with <4 viable segments). Heart failure symptoms (New York Heart Association class) improved only in patients with a substantial amount of dysfunctional but viable myocardium, irrespective of the presence or absence of diabetes mellitus. In patients without substantial myocardial viability, heart failure symptoms did not improve. Noninvasive assessment of contractile function 9–12 mo after coronary revascularization was available in 109 patients (29 with and 80 without diabetes mellitus). A total of 1,744 segments were analyzed before and after coronary revascularization, with 966 (55%) demonstrating abnormal wall motion. Improvement of wall motion after coronary revascularization occurred in 423 (84%) of 501 dysfunctional segments that were viable according to ¹⁸F-FDG SPECT. None of the dyskinetic segments improved in function. In the group of patients with a substantial amount of myocardial viability (≥4 viable segments), LVEF improved significantly after revascularization. Conversely, in the group without substantial myocardial viability, LVEF did not improve. Improvement of angina pectoris, heart failure symptoms, segmental wall motion, and LVEF was not related to the presence or absence of diabetes mellitus. The presence of a substantial amount of viable tissue (≥4 viable segments) had a comparable sensitivity and specificity for prediction of improved LVEF in patients with and without diabetes (Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Bar graph illustrating sensitivities and specificities of ¹⁸F-FDG SPECT for prediction of improvement of LVEF after coronary revascularization. Presence of substantial amount of viable tissue (≥4 viable segments) had good sensitivity and specificity for prediction of improvement in LVEF, which was comparable in patients with and without diabetes (neither is statistically significant). DM = diabetes mellitus; + = present; − = absent.

Methodologic Considerations

Traditionally, ¹⁸F-FDG imaging is used in combination with PET; however, the availability of PET for cardiac imaging is limited. With the introduction of dedicated collimators for 511-keV photons, it has become possible to perform ¹⁸F-FDG imaging using SPECT, allowing ¹⁸F-FDG imaging on a larger scale (23–25). To obtain good image quality, optimization of metabolic conditions before and during ¹⁸F-FDG imaging is extremely important (26). To optimize metabolic conditions during metabolic imaging, hyperinsulinemic, euglycemic clamping can be used, which yields an excellent image quality (26,27), but this time-consuming technique is suboptimal for routine use. In daily practice, ¹⁸F-FDG imaging after oral administration of a nicotinic acid derivative (acipimox) may be preferred; this approach results in effective lowering of free fatty acids, resulting in increased cardiac glucose (and ¹⁸F-FDG) uptake. Excellent image quality with this approach has been shown, even in patients with diabetes mellitus (9), although additional insulin was sometimes needed to further increase cardiac ¹⁸F-FDG uptake and improve image quality. This was confirmed in the current study, with adequate image quality in all patients although occasional administration of insulin was needed.

Study Limitations

A potential limitation is that coronary angiography was not routinely repeated after coronary revascularization. Graft occlusion may have prevented recovery of viable segments. This may have blunted the power of ¹⁸F-FDG SPECT to predict functional improvement after coronary revascularization. Second, the SPECT protocol did not include administration of nitrates before injection of ^99mTc-tetrofosmin. Nitrate administration may further enhance the detection of viable myocardium (28). Finally, in the group of patients with diabetes mellitus, relatively more women were present; this may have influenced the current results.