Author + information
- Received August 4, 1999
- Revision received September 8, 2000
- Accepted October 12, 2000
- Published online February 1, 2001.
- Prem Soman, MD, PhD, MRCP∗,
- Raymond Taillefer, MD, FRCP(C)†,
- E.Gordon DePuey, MD, FACC‡,
- James E Udelson, MD, FACC§ and
- Avijit Lahiri, MBBS, MSc, MRCP, FESC, FACC∗,*
- ↵*Reprint requests and correspondence: Dr. Avijit Lahiri, Department of Cardiovascular Medicine, Northwick Park Hospital, Harrow, United Kingdom, HA1 3UJ
We prospectively compared dipyridamole single-photon emission computed tomography (SPECT) imaging with Tc-99m sestamibi and Tc-99m tetrofosmin for the detection of reversible perfusion defects in patients with mild-to-moderate coronary artery disease.
Tc-99m tetrofosmin has a lower first-pass myocardial extraction fraction compared to Tc-99m sestamibi and thus could underestimate mild perfusion defects.
Eighty-one patients with 50% to 90% stenosis in one or two major epicardial vessels without previous myocardial infarction, and seven with <5% probability of coronary artery disease underwent dipyridamole SPECT imaging with both agents. The SPECT data were analyzed quantitatively.
Tc-99m sestamibi detected reversible perfusion defects in a greater number of segments (total 363 and 285, p < 0.001, and mean ± SD, 2.2 ± 3.0 and 1.8 ± 2.5 per patient, p = 0.008, for Tc-99m sestamibi and Tc-99m tetrofosmin, respectively), demonstrated a larger extent of perfusion defect (mean ± SD, 15.8% ± 12.3% and 12.0% ± 11.4%, p < 0.03, for Tc-99m sestamibi and Tc-99m tetrofosmin, respectively) and more often correctly identified patients with disease in more than one coronary artery (p = 0.02). There was better defect contrast with Tc-99m sestamibi (defect/normal wall count ratios were 0.60 ± 0.15 vs. 0.73 ± 0.14 for Tc-99m sestamibi and Tc99m tetrofosmin, respectively, p = 0.01, for reversible defects seen in identical segments with both agents; and 0.73 ± 0.16 vs 0.79 ± 0.17, respectively, p <0.01, for reversible defects detected with either agent alone). There was no significant difference in diagnostic sensitivity or image quality.
These differences between two commonly used tracers may have significant diagnostic and prognostic implications.
Tc-99m sestamibi and Tc-99m tetrofosmin have similar physical properties that offer significant advantages over thallium-201 for gamma camera imaging (1–3). However, the myocardial extraction fraction of Tc-99m tetrofosmin is lower than that of Tc-99m sestamibi, and there is a relatively earlier “roll off” in the linear relationship between tracer uptake and myocardial blood flow for Tc-99m tetrofosmin (4), which could result in underestimation of perfusion abnormalities (5). Although two recent comparative studies of these two agents suggested diagnostic equivalence (6,7), they have never been compared specifically using vasodilator stress testing in patients with mild coronary disease, conditions in which the effects of a low myocardial extraction fraction are most likely to manifest (4,5,8,9).
Therefore, this study was designed to compare the diagnostic accuracy of Tc-99m sestamibi and Tc-99m tetrofosmin single-photon emission computed tomography (SPECT) for the detection of reversible perfusion abnormalities induced by dipyridamole stress in patients with mild-to-moderate coronary artery disease (CAD).
Patients with mild-to-moderate stable CAD, defined as 50% to 90% stenosis in one or two major epicardial coronary arteries on recent (<2 months) qualitative coronary arteriography, who consented to participate, were prospectively enrolled. In addition, patients with a low likelihood (<5% pretest probability) (10)of CAD were also included to eliminate reporting bias. Exclusion criteria included severe coronary disease (left main stem disease, >90% stenosis in any artery, three-vessel disease), previous myocardial infarction, significant valvular disease, left bundle branch block and clinically unstable CAD. Standard exclusion criteria for dipyridamole stress were applied (11).
The protocol was approved by the Institutional Review Boards of both institutions. After informed consent, all patients underwent dipyridamole stress SPECT imaging with Tc-99m sestamibi and Tc-99m tetrofosmin in random order within two weeks of each other. A two-day (U.K.) or same-day (Canada) rest-stress protocol was used (12), with individual patients subjected to the same protocol for imaging with both agents. Perfusion imaging was performed within two months of coronary arteriography.
The standard high-dose dipyridamole protocol (0.56 mg/kg, followed by 0.28mg/kg in the absence of adverse effects) was used (13). In any given patient the same dose of dipyridamole was used for imaging with both agents, and the tracer was injected 2 to 3 min after the completion of the infusion.
For the two-day stress rest protocol, 16 mCi of tracer was injected on each occasion. In the one-day protocol rest images were initially performed with 10 mCi of tracer followed 4 h later by the stress study with 30 mCi of tracer. SPECT imaging was performed 60 min after tracer injection with a large field-of-view, dual-headed gamma camera equipped with a high-resolution collimator using methods previously described (14).
The SPECT images were analyzed by two independent experts (EGD, JEU) blinded to the patient’s clinical details and tracer identity. The Tc-99m sestamibi and Tc-99m tetrofosmin scans of any given patient were read randomly. The left ventricle was divided into 17 segments on three short axes and one vertical long axis slices, and each segment was graded according to perfusion using a semiquantitative four-point system (1 = normal, 2 = mildly reduced, 3 = severely reduced, 4 = absent), and classified as normal, reversible (new or worsening stress defect) or fixed (identical stress and rest defects). The overall image quality was assigned one of four grades (excellent, good, fair, poor) based on count density, myocardial and left ventricular cavity definition and adjacent interfering visceral activity.
The extent of defect was quantified using CEqual software (15)as a percentage of the entire myocardium involved. The defect/normal wall activity ratio was used to calculate the severity of the defect and was determined in two groups of patients: those in whom reversible perfusion defects in identical segments were detected by both Tc-99m sestamibi and Tc-99m tetrofosmin, and those in whom reversible defects were detected by one agent only. For this analysis, we summed four short axis slices, taking care to choose identical slices from Tc-99m sestamibi and Tc-99m tetrofosmin images of any given patient. Regions of interest were then placed manually on the area of defect and the area of the myocardium with the maximum tracer uptake, with the operator blinded to the tracer identity. By determining the count densities within the defect and normal regions of interest for each defect, a mean defect/normal count density ratio was computed. The average of three values was taken.
Sensitivity and specificity were determined using standard formulae and compared using the McNemar test. The concordance between tetrofosmin and sestamibi for classifying scans (overall and segmental) into normal, reversible and fixed categories and for overall image quality was determined with kappa statistics. The difference in the average number of reversible segments detected per patient was compared using the paired ttest. Comparison of the two agents for diagnostic categorization was performed using the Stuart–Maxwell test (16)of marginal homogeneity (paired comparison of the row totals with the column totals, to see whether the two agents classify the same proportion of patients into the three categories) and Bowker’s test for symmetry (17)(a comparison of the symmetry of individual cells about the table diagonal), and image quality was compared using the Wilcoxon signed rank test. All data are expressed as mean ± standard deviation (SD). A p-value of <0.05 was considered significant.
Eighty-eight patients (63 men, 25 women) were included in the study. The mean age was 59.7 ± 12 years and mean weight 76.2 ± 15.4 kg. Fifty-eight and 30 patients were recruited at the U.K. and Canadian sites, respectively; 40 had the Tc-99m tetrofosmin scan performed first.
Eighty-one patients had 50% to 90% stenosis in one or two major epicardial coronary arteries; 52 had single and 29 two-vessel disease. One patient had normal coronary arteries and six patients were not subjected to arteriography because of <5% pretest likelihood of coronary disease.
Hemodynamics during stress testing
There were no significant differences in hemodynamic parameters during stress testing with Tc-99m sestamibi and Tc-99m tetrofosmin (p = NS for rest and peak heart rate and blood pressure).
Detection of reversible perfusion defects
Tc-99m sestamibi demonstrated reversible perfusion defects in a significantly greater number of myocardial segments (363 and 285 reversible segments by Tc-99m sestamibi and Tc-99m tetrofosmin, respectively, p < 0.0001, Table 1). Additional analysis of the U.K. subgroup showed that the average number of reversible segments detected per patient was 2.2 ± 3.0 and 1.8 ± 2.5 for Tc-99m sestamibi and Tc-99m tetrofosmin, respectively (p = 0.008, 95% CI for the difference 0.12–0.78). However, even though Tc-99m sestamibi demonstrated reversible perfusion defects in 50 patients compared with 45 by tetrofosmin, this difference did not achieve statistical significance.
Extent and severity of perfusion defects
The extent of perfusion abnormality demonstrated by Tc-99m sestamibi was also significantly greater compared with Tc-99m tetrofosmin (15.8 ± 12.3% and 12.0 ± 11.4%, respectively, p < 0.01). In patients in whom both Tc-99m sestamibi and Tc-99m tetrofosmin studies showed reversible perfusion defects in identical segments, the defect/normal wall ratio was 0.60 ± 0.15 for Tc-99m sestamibi and 0.73 ± 0.14 for Tc-99m tetrofosmin (p = 0.01). In patients where reversible perfusion defects were not detected concordantly by both agents, the defect/normal wall ratio was 0.73 ± 0.16 for Tc-99m sestamibi and 0.79 ± 0.17 for Tc-99m tetrofosmin (p < 0.01, Table 2), indicating that Tc-99m sestamibi demonstrated perfusion defects of greater severity. Figures 1 and 2⇓⇓show SPECT images from two patients with CAD in whom Tc-99m sestamibi demonstrated greater defect extent, severity and reversibility.
Concordance for categorization of defects
The overall concordance between Tc-99m sestamibi and Tc-99m tetrofosmin for categorizing normal, fixed and reversible scans was 78% (69/88, kappa = 0.60). Among patients with angiographic coronary disease, ten scans were positive by Tc-99m sestamibi (nine reversible, one fixed) and normal by Tc-99m tetrofosmin. In contrast, six scans were abnormal by Tc-99m tetrofosmin (five reversible, one fixed) and normal by Tc-99m sestamibi; of these, only four patients had angiographic coronary disease and two scans were false positive (p = NS).
On a segmental basis, the concordance for similar categorization was 86% (1,291/1,496, kappa = 0.67). Of 119 segments classified as normal by Tc-99m tetrofosmin, Tc-99m sestamibi demonstrated reversible and fixed defects in 101 and 18 segments, respectively. In contrast, there were only 54 segments classified as normal by Tc-99m sestamibi where Tc-99m tetrofosmin detected abnormalities (49 reversible, five fixed; p < 0.001 by both Stuart-Maxwell and Bowker’s tests, Table 2).
Prediction of two-vessel disease
In the 29 patients with two-vessel disease, Tc-99m sestamibi and Tc-99m tetrofosmin identified perfusion defects in two vascular territories in 14 and seven patients, respectively (p = 0.02, Fig. 3). Patients with three-vessel disease were excluded from the study.
Detection of CAD
The sensitivity of the two agents for the detection of mild-to-moderate CAD was not significantly different from each other (63% [51/81] for Tc-99m sestamibi and 58% [47/81] for Tc-99m tetrofosmin, p = 0.09). In the seven patients with <5% probability of CAD or with normal coronary arteries, the specificity of both agents was 57.1%.
There was overall agreement of 67% (59/88) for classification of images based on quality into excellent, good, fair and poor categories, p = NS (Table 3).
The results of this study indicate that in patients with mild- to-moderate CAD, dipyridamole SPECT imaging with Tc-99m sestamibi demonstrates greater perfusion defect extent, severity and reversibility than does Tc-99m tetrofosmin, and more accurately predicts disease in more than one vessel. The low sensitivity of both agents demonstrated in this study in comparison to previously published results (18–21)was expected as a result of the deliberate selection of patients with only mild-to-moderate coronary disease without previous myocardial infarction. The lack of a significant difference in the diagnostic sensitivity of the two agents despite a trend in favor of Tc-99m sestamibi is most likely a reflection of inadequate sample size. On the basis of current results it is estimated that 135 and 541 patients would be required to demonstrate a 10% and 5% difference in diagnostic sensitivity, respectively (at 5% level of significance and 80% power).
The parallel relationship between myocardial blood flow and the myocardial uptake of extractable flow tracers is limited by the roll-off phenomenon, a plateau of tracer uptake when myocardial blood flow is progressively increased (4,9,22). The precise flow rate at which this phenomenon occurs is determined by the first-pass myocardial extraction of the tracer (4,9). Hence, the myocardial uptake of Tc-99m teboroxime (Emax0.63 to 0.81) (23), thallium-201 (Emax0.59 to 0.73) (24), and Tc-99 sestamibi (Emax0.31 to 0.73) (25)plateau at progressively lower blood flow rates of 5 ml/g/min, 3 ml/g/min and 2 ml/g/min, respectively (4,26). Tc-99m tetrofosmin has a lower extraction fraction (Emax0.15 to 0.33) and may plateau when myocardial blood flow increases approximately 1.5-fold (5). The plateau effect limits the sensitivity of a tracer to detect CAD and reversible perfusion defects by reducing defect contrast (4,9). It may affect imaging with vasodilator stress more than dynamic exercise (5,9), and the detection of mild rather than severe coronary disease (8).
Accordingly, previous animal and clinical studies have demonstrated greater underestimation of coronary flow heterogeneity (27)and stress perfusion defect size and extent by both Tc-99m sestamibi and Tc-99m tetrofosmin compared with thallium-201 (28–32). Two of these studies (29,30)reported a similar defect extent on the rest Tc-99m sestamibi and redistribution thallium-201 images, suggesting that the smaller defect size on stress Tc-99m sestamibi imaging may have been due to its lower myocardial extraction during exercise induced vasodilation.
Two small intraindividual comparative studies of Tc-99m sestamibi and Tc-99m tetrofosmin showed diagnostic equivalence. However, Flamen et al. (6)did not report on coronary anatomy, whereas Acampa and colleagues (7)used dynamic exercise stress and patients with severe disease (of 25 patients, 15 and 14 had previous myocardial infarction and multivessel disease, respectively). Therefore, differences occurring as a result of low tracer extraction may not have manifested.
Implications for risk stratification
Although there is now a large volume of data involving several thousand patients establishing the prognostic utility of Tc-99m sestamibi myocardial perfusion imaging (14, 33–36), in the light of the current results this should not be empirically extrapolated to Tc-99m tetrofosmin imaging when vasodilator stress is used.
Limitations of the study
We were limited by the lack of a quantification program validated for both agents. Although the CEqual software to quantify perfusion defects has been validated for Tc-99m sestamibi but not for Tc-99m tetrofosmin, at least two previous studies have suggested that the choice of the quantitative program is unlikely to influence the imaging results to any significant degree (37,38).
Dipyridamole SPECT imaging with Tc-99m sestamibi demonstrates greater myocardial perfusion defect reversibility, extent and severity than Tc-99m tetrofosmin in patients with mild-to-moderate CAD, and is more likely to correctly identify patients with disease in more than one coronary artery. These findings may have significant diagnostic and prognostic implications.
We gratefully acknowledge the technical assistance of Usha Raval, HND; Uday Bhonsle, MSc; Carole Benjamin, CNMT; and Andre Gagnon, CNMT.
☆ This study was supported by a grant from the DuPont Pharmaceutical Company and the Michael Tabor Cardiac Research Fund.
- coronary artery disease
- single-photon emission computed tomography
- Received August 4, 1999.
- Revision received September 8, 2000.
- Accepted October 12, 2000.
- American College of Cardiology
- Leppo J.A.,
- Meerdink D.J.
- Glover D.K.,
- Ruiz M.,
- Yang J.Y.,
- Smith W.H.,
- Watson D.D.,
- Beller G.A.
- Flamen P.,
- Bossuyt A.,
- Franken P.R.
- Shanoudy H.,
- Raggi P.,
- Beller G.A.,
- et al.
- Watson DD, Glover DK. Overview of kinetics and modeling. In: Zaret BL, Beller GA, eds. Nuclear Cardiology: State of the Art and Future Directions. Mosby, 1999:3–12.
- Soman P.,
- Parsons A.,
- Lahiri N.,
- Lahiri A.
- Kang X.,
- Berman D.S.,
- Van Train K.F.,
- et al.
- Stuart A.
- Kahn J.K.,
- McGhie I.,
- Akers M.S.,
- et al.
- Zaret B.L.,
- Rigo P.,
- Wackers F.J.,
- et al.
- Beller G.A.
- Gray W.A.,
- Gewirtz H.
- Nielsen A.P.,
- Morris K.G.,
- Murdock R.,
- Bruno F.P.,
- Cobb F.R.
- Okada R.,
- Glover D.,
- Gaffney T.,
- Williams S.
- Dahlberg ST, Leppo JA. Tracer kinetics in the isolated heart model. In: Zaret BL, Beller GA, eds. Nuclear Cardiology: State of the Art and Future Directions. Mosby, 1999:27–36.
- Glover D.K.,
- Ruiz M.,
- Edwards N.C.,
- et al.
- Tamaki N.,
- Takahashi N.,
- Kawamoto M.,
- et al.
- Narahara K.A.,
- Villanueva-Myer J.,
- Thompson C.J.,
- Brizendine M.,
- Mena L.
- Maublant J.C.,
- Marcaggi X.,
- Lusson J.R.,
- et al.
- Nakajima K.,
- Taki J.,
- Shuke N.,
- Bunko H.,
- Takata S.,
- Hisada K.
- Brown K.A.,
- Altland E.,
- Rowen M.
- Berman D.S.,
- Hachamovitch R.,
- Kiat H.,
- et al.