[18F]Fluorodeoxyglucose Single Photon Emission Computed Tomography
Can It Replace PET and Thallium SPECT for the Assessment of Myocardial Viability?
Background—New high-energy collimators for single photon emission computed tomography (SPECT) cameras have made imaging of positron-emitting tracers, such as [18F]fluorodeoxyglucose (18FDG), possible. We examined differences between SPECT and PET technologies and between 18FDG and thallium tracers to determine whether 18FDG SPECT could be adopted for assessment of myocardial viability.
Methods and Results—Twenty-eight patients with chronic coronary artery disease (mean left ventricular ejection fraction [LVEF]=33±15% at rest) underwent 18FDG SPECT, 18FDG PET, and thallium SPECT studies. Receiver operating characteristic curves showed overall good concordance between SPECT and PET technologies and thallium and 18FDG tracers for assessing viability regardless of the level of 18FDG PET cutoff used (40% to 60%). However, in the subgroup of patients with LVEF≤25%, at 60% 18FDG PET threshold value, thallium tended to underestimate myocardial viability. In a subgroup of regions with severe asynergy, there were considerably more thallium/18FDG discordances in the inferior wall than elsewhere (73% versus 27%, P<.001), supporting attenuation of thallium as a potential explanation for the discordant observations. When uptake of 18FDG by SPECT and PET was compared in 137 segments exhibiting severely irreversible thallium defects (scarred by thallium), 59 (43%) were viable by 18FDG PET, of which 52 (88%) were also viable by 18FDG SPECT. However, of the 78 segments confirmed to be nonviable by 18FDG PET, 57 (73%) were nonviable by 18FDG SPECT (P<.001).
Conclusions—Although 18FDG SPECT significantly increases the sensitivity for detection of viable myocardium in tissue declared nonviable by thallium (to 88% of the sensitivity achievable by PET), it will occasionally (27% of the time) result in falsely identifying as viable tissue that has been identified as nonviable by both PET and thallium.
In many patients with chronic coronary artery disease, impaired left ventricular function at rest arises, in part, from regions of ischemic or hibernating myocardium rather than scarred myocardium.1 2 Such asynergic but viable myocardial regions can be identified with radionuclide imaging techniques such as 18FDG metabolic imaging or thallium scintigraphy.3 Preserved or enhanced 18FDG uptake in asynergic myocardial regions identifies viable myocardium, which has been shown to predict not only improved regional and global function after revascularization but also improved survival compared with patients treated with medical therapy alone.4 5 6 The clinical application of 18FDG has been hampered by the limited availability and high cost of PET and cyclotron technology. Recently, because of the relatively long physical half-life of 18F (110 minutes), off-site production of 18FDG and subsequent transport to satellite nuclear cardiology laboratories has been proposed. This, combined with the advent of high-energy gamma camera collimators for SPECT, has made possible the use of 18FDG SPECT for detection of myocardial viability.7
Recent studies have shown that when SPECT flow tracers, such as 99mTc-sestamibi8 or early resting thallium,9 are used in combination with 18FDG SPECT, detection of myocardial viability is similar to that with PET. In routine clinical studies, stress-redistribution-reinjection thallium studies provide assessment of the presence and extent of coronary artery disease as well as myocardial viability.10 11 However, because of its low photon energy, soft-tissue attenuation of thallium may yield technically suboptimal images. The question arises, therefore, as to whether 18FDG SPECT could be used alone, without a perfusion agent, to replace thallium for routine clinical assessment of myocardial viability.
To address this question, we examined regional differences in 18FDG uptake as measured by SPECT and PET as well as the differences between thallium and 18FDG tracers. For the detection of viability, we used ROC analysis, thereby minimizing the difficulty of selecting arbitrary threshold values of viability for 18FDG SPECT or thallium SPECT. To correlate regional function with perfusion and/or metabolism, we applied gated tomographic radionuclide angiography, which allowed direct comparison of tomographic regional function with the assessment of myocardial viability by 18FDG SPECT, thallium SPECT, and 18FDG PET. The results of viability assessment are most relevant in patients with severe global left ventricular dysfunction and in regions with severe resting wall motion abnormalities. We therefore examined the ability of 18FDG SPECT to detect viable myocardium in all regions in such patients, especially in regions determined to be scarred by thallium.
We prospectively studied 28 patients (25 men and 3 women, 36 to 78 years old; mean, 62±12 years) with chronic coronary artery disease undergoing evaluation for myocardial viability. Eighteen patients had documented prior myocardial infarction, 11 had symptoms of angina, and 7 had heart failure symptoms. LVEF assessed by radionuclide angiography ranged from 8% to 61% (mean, 33±15%) at rest. All patients had either impaired left ventricular systolic function at rest or a regional wall motion abnormality. No patient with recent (<1 month) acute myocardial infarction was included in the study. Coronary angiography demonstrated significant stenosis (≥50% reduction in luminal diameter) of all three major epicardial coronary arteries in 17 patients, of two coronary arteries in 9 patients, and of one coronary artery in 2 patients. This study was approved by the Institutional Clinical Research Subpanel of the National Heart, Lung, and Blood Institute, and all patients gave informed consent.
Three-dimensional PET studies were performed on a General Electric whole-body camera (35 contiguous transaxial slices 4.25 mm apart; in-plane and axial reconstructed resolution of ≈6.5 mm). All patients were studied after an overnight fast. Approximately 1 hour before the injection of 18FDG, an oral dose of ≈25 to 50 g of glucose was administered. After the patient was positioned, an 8-minute attenuation correction scan was performed with 2 rotating 68Ge/68Ga line sources. Immediately thereafter, 5 mCi of 18FDG was injected, and data acquisition was begun. Transaxial tomographic images of 18FDG uptake were created by summing the data acquired during the 30- to 45-minute interval beginning 30 minutes after injection. Of the 28 patients studied, 6 had diabetes mellitus and 5 received insulin during the imaging procedure. Only patients with technically adequate PET studies were included in the 18FDG SPECT protocol.
Patients underwent 18FDG SPECT studies after completion of PET studies. Imaging was performed with a dual-headed (180° apart) SPECT camera (Vertex Genesys, ADAC Laboratories) equipped with a special high-energy collimator (resolution, 12 mm at 10 cm and 16 mm at 15 cm). Data were acquired in 120 steps over 360° (45 seconds per step) and reconstructed into transaxial slices (3.13 mm apart, 3.13-mm pixel size, reconstructed in-plane and axial resolution of ≈15 to 20 mm as measured with a line source in an elliptical chest phantom with lung inserts). Seventeen patients underwent SPECT imaging immediately after PET acquisition (mean time from injection to start of SPECT scan, 85±17 minutes). Because acquisition of 18FDG SPECT images so long after the completion of 18FDG PET studies raises the potential of unintentional bias against 18FDG SPECT because of isotope decay, a second injection of 10 mCi of 18FDG was given at rest in 11 patients immediately after PET imaging was completed, and SPECT images were acquired ≈40 minutes thereafter.
All patients underwent stress-redistribution-reinjection 201Tl SPECT, as previously described,10 within a mean of 2.7±4.8 weeks from 18FDG SPECT and PET studies. At peak stress (exercise or pharmacological), 3 mCi of thallium was injected intravenously. Thallium images were obtained with a three-headed rotating gamma camera equipped with a high-energy, medium-resolution, high-sensitivity collimator centered on the 68-keV photo peak with a 20% window. All images were reconstructed from projection data acquired over a 360° elliptical orbit around the patient’s thorax at 3° increments for 40 seconds each.
In each of these studies (thallium, 18FDG SPECT, and 18FDG PET), the transaxial images were resliced into a series of short-axis images, as previously described.12 These short-axis images were then sliced longitudinally, starting at 22.5° from midline, in 45° increments to yield four long-axis slices per patient, encompassing the entire left ventricle, as shown in Fig 1⇓. All three studies were resampled so as to have the same pixel dimension and matrix size: 2 mm per pixel in a 128×128 matrix. An operator-defined region of interest was drawn around the left ventricular myocardium of each of the radial long-axis slices and subdivided into 11 segments per slice (44 segments in all), and segmental 18FDG and thallium activities were then computed within each segment. The myocardial segment with the maximum mean counts per pixel on the thallium stress study was used as the normal reference segment for that patient. The same segment in the redistribution and reinjection thallium studies and 18FDG SPECT and 18FDG PET studies was identified and used as the reference segment for those studies. The activity of 18FDG or thallium in all other myocardial segments was then expressed as a percentage of the activity measured in the reference segment of the corresponding 18FDG SPECT, 18FDG PET, and thallium redistribution or reinjection image. The data were also analyzed by use of an alternate normalization scheme, in which the segment with the highest mean counts per pixel for thallium redistribution, thallium reinjection, 18FDG PET, and 18FDG SPECT was used as the reference segment for that individual study.
Myocardial segments were grouped on the basis of severity of reduction in thallium activity: normal (>85% of peak activity), mildly to moderately reduced (51% to 85% of peak), and severely reduced (≤50% of peak) activity. On the basis of previous reproducibility measurements in our laboratory,13 a region with reduced activity on thallium stress was considered reversibly ischemic if the normalized thallium activity on the subsequent redistribution or reinjection images for that region increased by ≥10%. Similarly, a region with <50% activity on thallium stress was considered severely irreversible if the normalized thallium activity on the subsequent redistribution and reinjection images did not increase by 10% or if the activity on both redistribution and reinjection images remained <50% regardless of any possible increase in thallium activity.
Because misregistration of small segments might account for variation between studies performed on the same patient, the reproducibility of segmental uptake in our laboratory was assessed by having two observers repeat these measurements on a cardiac phantom and on a patient. The interobserver variability for segmental tracer uptake was determined to be <5%.14 In addition to segmental analysis, we also grouped the 44 segments into 5 large regions per patient (anterior, apical, septal, lateral, and inferior) to allow 18FDG and thallium uptake in these large regions to be compared with regional wall motion.
Planar and SPECT Gated Radionuclide Angiography
Planar gated equilibrium radionuclide angiography was performed with a conventional Anger camera. Red cells were labeled with ≈25 mCi of [99mTc]pertechnetate. Three standard planar views (anterior, best septal left anterior oblique, and left lateral) were acquired at rest. LVEF was computed from the left anterior oblique view.
A subgroup of 18 patients also underwent SPECT gated radionuclide angiography to better separate cardiac structures and determine regional wall motion. Blood pool SPECT images were obtained with an ADAC (two 90° heads) scanner with a high-energy, high-sensitivity collimator, a 38×38-cm field of view, and a 128×128 matrix. The images were acquired in 60 azimuths over a 180° arc from 45° right anterior oblique to 45° left posterior oblique, with 60 seconds per step. For each patient, four gated long-axis slices were generated in a manner identical to that for thallium and 18FDG images, and regional wall motion was assessed by two observers in the same five myocardial regions per patient (anterior, apical, septal, lateral, and inferior) as normal, mildly hypokinetic, severely hypokinetic, akinetic, or dyskinetic. The grade assigned to a given region was the most severe regional asynergy within that region from all tomographic slices and views.
Data are presented as mean±SD. Group comparisons between 18FDG PET, 18FDG SPECT, and thallium uptake and differences between 18FDG and thallium with respect to regional wall motion were analyzed by either paired t test or χ2 analysis. The ability of 18FDG SPECT and thallium SPECT to determine viability was analyzed by use of ROC curves,15 ie, plots of true-positive rates versus false-positive rates. Each ROC curve was produced by dividing the total range of segmental values (ie, segment maximum to segment minimum) for the study in question (thallium or 18FDG SPECT) into 100 equal intervals and computing the sensitivity and (1−specificity) at each point. Comparison of ROC curves was performed by computing z scores from a comparison of the areas under the curves as in Reference 1515 .
Analysis of Regional 18FDG Uptake: SPECT Versus PET
To examine whether there are differences in regional 18FDG uptake between SPECT and PET technologies, a total of 977 segments were analyzed in 28 patients. There was a good overall agreement between 18FDG SPECT and 18FDG PET (r=.81, slope=0.79, SEE=0.004), with a mean difference in normalized 18FDG uptake assessed by the two techniques of 1.2% and a mean magnitude difference (ie, mean of the absolute values of the differences) of 9.6% (Fig 2⇓). When a 50% 18FDG threshold (for both PET and SPECT) was used to differentiate viable from nonviable myocardium, 18FDG SPECT provided information concordant with that of 18FDG PET in 920 of 977 segments (94%). When data from the 11 patients in whom SPECT images were acquired after a second injection of 18FDG were compared with data from the 17 patients who did not have 18FDG reinjection, the concordance between 18FDG SPECT and PET in the two patient groups was similar (95% versus 94%, P=NS).
ROC curves were used to assess the ability of 18FDG SPECT to determine viability, with 18FDG PET used as the gold standard. At 50% threshold value for 18FDG PET, the area under the ROC curve for 18FDG SPECT was 0.95±0.02. This excellent concordance between 18FDG SPECT and PET persisted when 40% and 60% 18FDG threshold values were applied with areas under the ROC curve of 0.96±0.02 and 0.94±0.02, respectively (Fig 3⇓).
18FDG and Thallium Uptake
To examine whether there are differences between 18FDG and thallium for detection of myocardial viability, a 50% threshold value was used for the two tracers to differentiate viable from nonviable myocardium. Stress-redistribution-reinjection thallium SPECT provided information concordant with that for 18FDG SPECT in 883 of 977 segments (90%) and in 896 segments (92%) with 18FDG PET. Furthermore, when the data were analyzed in the 777 segments demonstrating abnormal thallium uptake during stress, there was concordance regarding myocardial viability in 684 segments (88%) with 18FDG SPECT and 696 segments (90%) with 18FDG PET. Among the discordant segments, 21 of 85 segments that were nonviable by 18FDG SPECT and 22 of 100 segments that were nonviable by 18FDG PET were judged viable by thallium reinjection.
ROC curves were used to assess the ability of thallium to predict viability as defined by 18FDG PET. At 50% threshold value for 18FDG PET, the area under the ROC curve for thallium reinjection was 0.95±0.02 for all 977 segments. This excellent concordance between thallium and 18FDG PET persisted when data were analyzed in the 777 abnormal segments with the area under the ROC curve of 0.93±0.02. When different 18FDG PET threshold values were applied, at 40%, the area under the ROC curve for thallium was 0.95±0.02, and at 60%, the area under the ROC curve for thallium was 0.90±0.02. There was no statistically significant difference between the areas under the ROC curve for thallium and for 18FDG SPECT in abnormal segments, regardless of the 18FDG PET threshold value that was applied (40%, 50%, or 60%). A representative example of a patient demonstrating concordance between 18FDG SPECT, 18FDG PET, and thallium stress-redistribution-reinjection is shown in Fig 4⇓.
Relation to Severity of Left Ventricular Dysfunction
Because assessment of viability is of particular concern in patients with left ventricular dysfunction, the data were analyzed in two groups: group 1, representing 16 patients with normal or mildly to moderately impaired left ventricular function (LVEF>25%; mean, 44±8%), and group 2, representing 12 patients with severely impaired left ventricular function (LVEF≤25%; mean, 19±5%). The relation between 18FDG SPECT and 18FDG PET was similar in the two groups of patients. At 50% threshold value for 18FDG PET, the area under the ROC curve for 18FDG SPECT among group 1 patients was 0.97±0.02 and in group 2 patients was 0.94±0.02 (P=NS). Similarly, there was no statistically significant difference for the areas under the ROC curve between thallium and 18FDG SPECT in group 1 compared with group 2 patients when 40% and 50% 18FDG threshold values were applied. However, the areas under the ROC curves were significantly different between thallium (0.89±0.02) and 18FDG SPECT (0.95±0.02, P<.02) in group 2 patients when a 60% 18FDG PET threshold value was applied (Fig 5⇓).
18FDG Uptake in Severely Irreversible Thallium Defects
The ROC analysis above allowed assessment of overall ability to detect viable regions. Of more specific concern was the detection of potentially viable regions by 18FDG in territories previously determined to be nonviable by thallium.
Data Normalized to Peak Stress Thallium
Of the 777 abnormal segments, 432 defects (56%) were irreversible on redistribution-reinjection images: 295 with mildly to moderately reduced and 137 with severely reduced thallium activity. In these 137 segments, there was considerable discordance between thallium and 18FDG uptake. 18FDG PET verified 78 (57%) of these segments to be metabolically inactive and nonviable, of which 57 (73%) were also nonviable by 18FDG SPECT (Fig 6⇓). However, in the remaining 59 segments (43%), 18FDG PET identified these segments to be metabolically active and viable, of which 52 (88%) were also viable by 18FDG SPECT (P<.001). Of the 59 discordant segments (between thallium and 18FDG PET), 38 (64%) were located in the inferior region. Mean thallium uptake in segments that were judged to have severely irreversible defects but preserved 18FDG uptake by PET and SPECT was significantly higher (0.42±0.04 on stress) than in those with abnormally reduced 18FDG uptake (0.24±0.10, P<.001).
Among the 137 segments with severely reduced thallium uptake, 55 were located in the inferior territory and 85 were located in regions in which SPECT wall motion data were available. In the 55 inferior segments, there was significantly more discordance in subjects with severely impaired left ventricular function (LVEF≤25%), 27 of 30 segments (90%) compared with those with LVEF>25% (11 of 25 segments, 44%), P<.001. A representative example of a patient demonstrating discordance between 18FDG SPECT, 18FDG PET, and thallium stress-redistribution-reinjection is shown in Fig 7⇓. Of the 85 segments for which wall motion data were available, thallium and 18FDG PET uptake were concordant in 53 segments (62%) and discordant in 32. Of the 53 thallium/18FDG concordant segments, 49 (92%) were located in severely asynergic regions compared with 22 of 32 thallium/18FDG discordant segments (69%) (P<.02). Moreover, only 13 of 49 segments (27%) with severe asynergy and concordance between thallium and 18FDG uptake were located in the inferior region compared with 16 of 22 segments (73%) with discordant thallium and 18FDG uptake (P<.001).
Data Normalized to the Segment With Highest Peak Counts per Pixel for Each Individual Study
Because there could potentially be other approaches to normalizing the data, we repeated our data analysis, normalizing each study to its own highest segment. That is, the segment with the highest mean counts per pixel for thallium redistribution, thallium reinjection, 18FDG PET, and 18FDG SPECT was used as the reference segment for the corresponding individual study. Among 159 segments with severely irreversible thallium defects identified by this alternative normalization scheme, 74 segments (47%) were identified to be metabolically active and viable by 18FDG PET, of which 63 (85%) were also viable by 18FDG SPECT (P<.001). Of the 74 discordant segments (between thallium and 18FDG PET), 43 (58%) were located in the inferior region, whereas among 85 segments with concordant information, only 21 (25%) were located in the inferior region (P<.001). Among all 64 segments located in the inferior territory, significantly more discordance occurred in subjects with severely impaired LVEF (≤25%), 31 of 36 segments (86%) compared with those with LVEF>25%, 12 of 28 segments (42%), P<.001. Hence, the results appear to be unaffected by the normalization scheme applied.
18FDG and Thallium Uptake in Asynergic Regions Assessed by SPECT Gated Radionuclide Angiography
Because assessment of myocardial viability and decisions regarding revascularization are of clinical concern predominantly in asynergic regions, we directed our attention to the five myocardial regions in which regional wall motion could be assessed (apical, anterior, septal, inferior, and lateral, representing the three major coronary vascular territories per patient). Among the 18 patients who underwent SPECT gated radionuclide angiography, a total of 90 myocardial regions were analyzed. Regional wall motion was normal in 40, mildly hypokinetic in 9, severely hypokinetic in 13, akinetic in 23, and dyskinetic in 5.
Among the 41 severely asynergic regions (severely hypokinetic, akinetic, or dyskinetic), 39 (95%) had associated ECG Q waves, a history of prior myocardial infarction in the same vascular territory, or a critically stenosed coronary artery supplying the region. Myocardial viability was present in 23 of these 41 severely asynergic regions (56%) by thallium, 26 (63%) by 18FDG PET, and 28 (68%) by 18FDG SPECT (P=NS). 18FDG SPECT and PET provided concordant information regarding myocardial viability in 33 of 41 severely asynergic regions (80%) and discordance in 8 regions. Thallium SPECT provided concordant information with 18FDG PET in 32 of 41 asynergic regions (78%) and in 30 regions with 18FDG SPECT (73%) (Fig 8⇓). In contrast, among the 49 normal or mildly hypokinetic regions, 45 (92%) had preserved thallium uptake, 46 (94%) had preserved 18FDG uptake by SPECT, and 46 (94%) by PET.
ROC curves were used to assess the ability of thallium and 18FDG SPECT to predict viability as defined by 18FDG PET in the 41 severely asynergic regions. At 50% threshold value for 18FDG PET, the area under the ROC curve for 18FDG SPECT was 0.91±0.05 and for thallium reinjection, 0.89±0.06 (P=NS). The concordance between thallium, 18FDG SPECT, and 18FDG PET persisted when different 18FDG PET threshold values were applied (40% or 60%).
In patients with chronic coronary artery disease, we examined differences between SPECT and PET technologies and 18FDG and thallium tracers for differentiating viable from nonviable myocardium. Because in asynergic regions there may be a gradation of myocardial tissue from normal, hypoperfused but viable, to nonviable myocardium within the same region, the use of a single arbitrary cutoff value for thallium or SPECT 18FDG uptake would oversimplify this complex relationship. For example, an asynergic region with 60% 18FDG uptake may represent (1) severely hypoperfused but viable (hibernating) myocardium, (2) mixed scarred and normal (nonischemic) myocardium, or (3) mixed scarred and viable (hypoperfused) myocardium. Because the precise threshold with 18FDG SPECT to predict recovery of function after revascularization is not known, we used ROC analysis to compare the techniques and the tracers in their ability to detect viable tissue. Our findings suggest that the overall concordance between the two technologies (SPECT and PET) and the two tracers (18FDG and thallium) was excellent regardless of the level of 18FDG PET threshold value applied (40%, 50%, or 60%).
Comparison of Technologies: SPECT Versus PET
PET, with its high spatial resolution, high-count-density images, and the possibility for attenuation correction, allows accurate assessment of regional uptake using 18FDG (a glucose analogue) irrespective of weight or body habitus of patients. The clinical feasibility of imaging positron emitters such as 18FDG with high-energy collimator planar16 17 18 or SPECT19 techniques has been reported previously. In fact, imaging of 18F radiotracer with standard nuclear medicine equipment dates back to 1962, when attempts were made to perform bone scans by use of 18F.20
The poorer spatial resolution of SPECT, the lower sensitivity, and the lack of attenuation correction may cause some discordance between SPECT and PET images of the same tracer. Despite these differences, relative myocardial metabolic estimates of 18FDG SPECT are comparable overall to those with PET.7 9 In previous studies by Bax and coworkers,9 21 flow measurements by thallium SPECT were combined with measurement of metabolism by 18FDG. When thallium uptake and SPECT 18FDG were taken together to differentiate viable from nonviable myocardium in regions with impaired contraction, they found an agreement between PET and SPECT of 76%.9 In a subsequent study,21 the authors acquired thallium “reinjection” images 4 hours after the infusion of high-dose dobutamine and eliminated the 3- to 4-hour redistribution images. Because apparent washout of thallium may occur between redistribution and reinjection studies, reliance on reinjection images alone might underestimate defect reversibility and hence viability.10 22 Therefore, the findings of the above study21 cannot be compared with our results. In patients with coronary artery disease, when 18FDG SPECT and PET images were compared with thallium redistribution acquired 3 hours after the tracer was injected at rest, ≈20% of regions with severe resting thallium defects were metabolically viable by 18FDG imaging using either SPECT or PET.7 In the latter study, however, regional contractile function was not reported.
Our results suggest that using 40% to 60% 18FDG PET threshold values, 18FDG SPECT provides information comparable to that by 18FDG PET regarding detection of viability; concordance was found in patients with severely impaired global left ventricular function (LVEF≤25%) and in patients with severely asynergic regions. Overall concordance, however, is influenced by the number of normal segments. The more clinically relevant question is how well 18FDG SPECT predicts viability in segments judged nonviable by thallium (severely irreversible defects). In segments with severely irreversible thallium defects (scarred by thallium), 18FDG PET identified 43% of these segments to be metabolically active and viable, of which 88% were also viable by 18FDG SPECT. Among the segments that were judged to be nonviable by both thallium and 18FDG PET, 73% were also nonviable by 18FDG SPECT (P<.001). Thus, 18FDG SPECT provides incremental information regarding viability, similar to 18FDG PET, in segments judged nonviable by thallium. Despite the good overall concordance between 18FDG SPECT and PET, there were noticeable discordances, especially in regions of severe left ventricular dysfunction and regional asynergy and in regions with severely reduced 18FDG uptake on PET (27% of which appear to have more activity on SPECT). Some of these differences may be due to attenuation, others to the large counting statistical fluctuations in the data, especially with SPECT. However, the clinical significance of such discordances is unknown, especially in light of the relatively few subjects studied with severe left ventricular dysfunction.
In our study, the concordance between 18FDG SPECT and 18FDG PET was 95% among the patients studied after a second injection of 18FDG at rest and 94% among the patients in whom 18FDG was not reinjected. A possible explanation for the lack of observed difference between the two patient groups may relate to improvement in contrast between the myocardium and blood pool on delayed 18FDG images.23 Thus, the time delay to 18FDG SPECT imaging (performed after PET without a second 18FDG injection) may have partially offset the reduced sensitivity of SPECT and the loss of activity due to decay. In addition, the improved image quality inherent in late images may also have contributed to the good concordance between 18FDG SPECT and 18FDG PET, despite the poorer spatial resolution of 18FDG SPECT.
Comparison of Tracers: Thallium and 18FDG
Myocardial uptake of thallium reflects cell membrane integrity,24 whereas 18FDG uptake reflects the overall rate of transmembrane exchange and phosphorylation of glucose.25 26 Our data suggest that when 50% thallium threshold value was used, thallium provided concordant viability information with 18FDG SPECT in 90% of segments. Analysis of data in a dichotomous fashion is critically dependent on the exact threshold value chosen. ROC curve analysis,15 in particular comparison of areas under the ROC curves, has been shown to overcome this problem. When ROC analysis was applied to thallium and 18FDG data in all segments such that normal, reversible, mildly to moderately irreversible, and severely irreversible thallium defects were grouped together, regardless of the level of 18FDG PET cutoff used (40% to 60%), there was no statistical difference in ROC area between 18FDG SPECT and thallium SPECT. However, in segments with severely irreversible thallium defects (scarred by thallium), 43% of these segments were identified to be metabolically active by 18FDG PET, of which 88% were also viable by 18FDG SPECT. Of the discordant segments between thallium and 18FDG PET, 64% were located in the inferior segment. Thus, there was a tendency for thallium SPECT to underestimate the extent of myocardial viability, especially in the inferior region, whereas 18FDG SPECT was less affected by such regional variation. Furthermore, a significantly larger number of inferior segment discordances occurred in subjects with LVEF≤25% compared with those with LVEF>25% (P<.001). These findings are supported by a previous report from Altehoefer and colleagues,27 in which resting blood flow measurements by thallium were compared with 18FDG uptake by PET. In their study, although there was a linear relationship between thallium and 18FDG uptake in the three major coronary artery vascular territories, the correlation was lower in the right coronary artery vascular territory (r=.52). Furthermore, in segments with severely reduced thallium activity, there was 44% discordance between thallium and 18FDG (scarred by thallium but viable by 18FDG PET) in the right coronary artery vascular territory.
In our study, because SPECT was the common technology used for acquiring both 18FDG and thallium images, at least some of the observed differences in the inferior region between thallium SPECT and 18FDG SPECT are probably explained by the attenuation differences between the two tracers. For example, only 27% of segments with severe asynergy and concordance between thallium and 18FDG uptake were located in the inferior region, compared with 73% of segments with discordant thallium/18FDG segments (P<.001). Because the inferior wall is one of the most severely attenuated regions in the heart, this finding supports the greater attenuation of thallium as a potential explanation for the discordant observations. Attenuation effects might also be expected to cause discordance between thallium and 18FDG in women because of breast attenuation, but because there were only three women in our study, we were unable to examine this possible effect. Because SPECT does not correct for soft-tissue attenuation, there is regional heterogeneity of lower limits of normal for thallium, which ranges from ≈70% in the inferior region to 90% in the lateral wall.10 Assessment of myocardial viability is a clinical concern in patients with severely impaired left ventricular dysfunction. This group of subjects often have dilated hearts, accentuating the effects of attenuation, and our results support this concept. It remains unknown, however, how much of the observed discordance between thallium and 18FDG SPECT is related to the differences in physiological behavior of the two tracers and how much is due to the differences in their imaging properties (ie, attenuation). If the latter, then implementation of attenuation correction for SPECT might lead to greater concordance between thallium and 18FDG.
Tomographic Assessment of Regional Contractile Function
An important feature of our study is that we used SPECT radionuclide angiography for more accurate anatomic alignment of tomographic perfusion, metabolism, and regional contractile function. However, we did not have the opportunity to assess tomographic regional contractile function after revascularization. SPECT radionuclide imaging has shown improved segmental resolution, separation of overlapping structures, and localization of individual diseased coronary arteries compared with planar imaging.28 29 In the subset of patients who underwent SPECT gated radionuclide angiography, the agreement between 18FDG SPECT and PET for identifying myocardial viability in severely asynergic regions was 80%; thallium provided information concordant with that from 18FDG SPECT in 73% of these regions. In normal or mildly hypokinetic regions, thallium and 18FDG SPECT provided concordant information in 94% of the regions.
These data suggest that in patients with chronic coronary artery disease, there is overall good concordance between SPECT and PET technologies and thallium and 18FDG tracers for differentiating viable from nonviable myocardium. Although 18FDG SPECT significantly increases the sensitivity for detection of viable myocardium in tissue declared nonviable by thallium (to 88% of the sensitivity achievable by PET), it will occasionally (27% of the time) result in falsely identifying as viable tissue that has been identified as nonviable by both PET and thallium.
Selected Abbreviations and Acronyms
|LVEF||=||left ventricular ejection fraction|
|PET||=||positron emission tomography|
|ROC||=||receiver operating characteristic|
|SPECT||=||single photon emission computed tomography|
- Received August 25, 1997.
- Revision received October 29, 1997.
- Accepted November 1, 1997.
- Copyright © 1998 by American Heart Association
Rahimtoola SH. Coronary bypass surgery for chronic angina–1981: a perspective. Circulation. 1982;65:225–241.
Dilsizian V, Bonow RO. Current diagnostic techniques of assessing myocardial viability in patients with hibernating and stunned myocardium. Circulation. 1993;87:1–20.
Di Carli MF, Davidson M, Little R, Khanna S, Mody FV, Brunken RC, Czernin J, Rokhsar S, Stevenson LW, Laks H, Hawkins R, Schelbert HR, Phelps ME, Maddaahi J. Value of metabolic imaging with positron emission tomography for evaluating prognosis in patients with coronary artery disease and left ventricular dysfunction. Am J Cardiol. 1994;73:527–533.
Lee KS, Marwick TH, Cook SA, Go RT, Fix JS, James KB, Sapp SK, MacIntyre WJ, Thomas JD. Prognosis of patients with left ventricular dysfunction, with and without viable myocardium after myocardial infarction: relative efficacy of medical therapy and revascularization. Circulation. 1994;90:2687–2694.
Burt RW, Perkins OW, Oppenheim BE, Schauwecker DS, Stein L, Wellman HN, Witt RW. Direct comparison of fluorine-18-FDG SPECT, fluorine-18-FDG PET and rest thallium-201 SPECT for detection of myocardial viability. J Nucl Med. 1995;36:176–179.
Sandler MP, Videlefsky S, Delbeke D, Patton JA, Meyerowitz C, Martin WH, Ohana I. Evaluation of myocardial ischemia using a rest metabolism/stress perfusion protocol with fluorine-18-fluorodeoxyglucose/technetium-99m-MIBI and dual-isotope simultaneous-acquisition single-photon emission computed tomography. J Am Coll Cardiol. 1995;26:870–876.
Bax JJ, Visser FC, Blanksma PK, Veening MA, Tan ES, Willemsen ATM, van Lingen A, Teule GJJ, Vaalburg W, Lie KI, Visser CA. Comparison of myocardial uptake of fluorine-18-fluorodeoxyglucose imaged with PET and SPECT in dyssynergic myocardium. J Nucl Med. 1996;37:1631–1636.
Petretta M, Cuocolo A, Bonaduce D, Nicolai E, Cardei S, Berardino S, Ianniciello A, Apicella C, Bianchi V, Salvatore M. Incremental prognostic value of thallium reinjection after stress-redistribution imaging in patients with previous myocardial infarction and left ventricular dysfunction. J Nucl Med. 1997;38:195–200.
Perrone-Filardi P, Bacharach SL, Dilsizian V, Maurea S, Frank JA, Bonow RO. Regional left ventricular wall thickening: relation to regional uptake of 18fluorodeoxyglucose and 201Tl in patients with chronic coronary artery disease and left ventricular dysfunction. Circulation. 1992;86:1125–1137.
Dilsizian V, Freedman NMT, Bacharach SL, Perrone-Filardi P, Bonow RO. Regional thallium uptake in irreversible defects: magnitude of change in thallium activity after reinjection distinguishes viable from nonviable myocardium. Circulation. 1992;85:627–634.
Jousse F, Srinivasan G, Carson JM, Dilsizian V, Bacharach SL. Effect of shortaxis angle selection on quantitation of myocardial uptake. J Nucl Med. 1996;37:160. Abstract.
Hoflin F, Ledermann H, Noelpp U, Weinreich R, Rosler H. Routine 18F-2-deoxy-2-fluoro-d-glucose (18F-FDG) myocardial tomography using a normal large field of view gamma camera. Angiology. 1989;140:1058–1064.
Huitink JM, Visser FC, van Lingen A, Groeneveld JG, Bax JJ, van Leeuwen GR, Visser GM, Van Loon MJA, Teule GJJ, Visser CA. Feasibility of planar fluorine-18-FDG imaging after recent myocardial infarction to assess myocardial viability. J Nucl Med. 1995;36:975–981.
Bax JJ, Visser FC, van Lingen A, Huitink JM, Kamp O, van Leeuwen GR, Visser GWM, Teule GJJ, Visser CA. Feasibility of assessing regional myocardial uptake of 18F-fluorodeoxyglucose using single photon emission computed tomography. Eur Heart J. 1993;14:1675–1682.
Blau M, Nagler W, Bender MA. Fluorine-18: a new isotope for bone scanning. J Nucl Med. 1962;3:332–334.
Bax JJ, Cornel JH, Visser FC, Fioretti PM, van Lingen A, Reis AEM, Boersma E, Teulle GJJ, Visser CA. Prediction of recovery of myocardial dysfunction after revascularization: comparison of fluorine-18 fluorodeoxyglucose/thallium-201 SPECT, thallium-201 stress-reinjection SPECT and dobutamine echocardiography. J Am Coll Cardiol. 1996;28:558–564.
Dilsizian V, Bonow RO. Differential uptake and apparent thallium-201 ‘washout’ after thallium reinjection: options regarding early redistribution imaging before reinjection or late redistribution imaging after reinjection. Circulation. 1992;85:1032–1038.
Pohost GM, Zir LM, Moore RH, McKusick KA, Guiney TE, Beller GA. Differentiation of transiently ischemic from infarcted myocardium by serial imaging after a single dose of thallium-201. Circulation. 1977;55:294–302.
Wheeler TJ. Translocation of glucose transporters in response to anoxia in heart. J Biol Chem. 1988;263:19447–19454.
Young LH, Renfu Y, Russell R, Hu X, Caplan M, Ren J, Shulman GI, Sinusas AJ. Low-flow ischemia leads to translocation of canine heart GLUT-4 and GLUT-1 glucose transporters to the sarcolemma in vivo. Circulation. 1997;95:415–422.
Altehoefer C, vom Dahl J, Buell U, Uebis R, Kleinhans E, Hanrath P. Comparison of thallium-201 single-photon emission tomography after rest injection and fluorodeoxyglucose positron emission tomography for assessment of myocardial viability in patients with chronic coronary artery disease. Eur J Nucl Med. 1994;21:37–45.
Corbett JR, Jansen DE, Lewis SE, Gabliani GI, Nicod P, Filipchuk NG, Redish GA, Akers MS, Wolf CL, Rellas JS, Parkey RW, Willerson JT. Tomographic gated blood pool radionuclide ventriculography: analysis of wall motion and left ventricular volumes in patients with coronary artery disease. J Am Coll Cardiol. 1985;6:349–358.
Metcalfe MJ, Norton MY, Jennings K, Walton S. Improved detection of abnormal left ventricular wall motion using tomographic radionuclide ventriculography compared to planar radionuclide and single plane contrast ventriculography. Br J Radiol. 1993;66:986–993.