Thallium-201 myocardial scintigraphy in coronary triple-vessel disease: An attempt to increase sensitivity using quantitative methods

Eur J Nucl Med (1987) 13:269 273 European N u c l e a r Journal of Medicine © Springer-Vex'lag 1987

Thallium-201 myocardial scintigraphy in coronary triple-vessel disease: An attempt to increase sensitivity using quantitative methods M. Singer 1, J. Miiller-Brand 1, M.E. Pfisterer 2, and R. Fridrieh 1 1 Department of Nuclear Medicine and 2 Department of Cardiology, University Hospital, CH-4031 Basel, Switzerland

Abstract. In order to increase the sensitivity of thallium-201 exercise scintigraphy in patients with triple vessel coronary disease (TVD), we first examined retrospectively myocardial scintigrams of 179 patients with TVD, as documented subsequently by cardiac catheterization. Ischemia had been diagnosed visually in 141 (79%), scar without ischemia in 25 (14%) and no apparent perfusion defect in 13 (7%) cases. The subset of TVD patients without scintigraphic ischemia (i.e. those with scar or no perfusion defect) was then compared to a control group with normal coronary angiography using four quantitative criteria: (1) in preset-count analog images, a quotient of the exposure times rest image/ stress image; (2) in preset-time digital images, a quotient of counts/pixel in stress image/rest image using two different myocardial regions of interest (ROI); (3) a similar quotient using paracardial lung ROIs of three different sizes; (4) the absolute values of stress lung uptake. Quotients (1) and (2) were expected to be lower in TVD patients than in normal controls due to exercise-induced global ischemia, quotient (3) and value (4) were expected to be higher due to exercise-induced left ventricular dysfunction with increased lung uptake of thallium-201. All results showed a tendency to confirm these hypotheses; significant differences (P<0.05) between patients and controls were obtained in all lung quotients and in 3 of 12 myocardial quotients. No significant differences were observed in the exposure time quotients of preset-count images and in the stress lung uptake. Due to overlapping values, it was not possible to fix normal and pathological ranges of any quotient. It is concluded that the increased ratio of stress/rest lung uptake reflecting exercise-induced pulmonary venous conges- tion seems to indicate TVD in apparently normal thallium- 201 scintigrams in a large number of patients.

Key words: Thallium 201 - Myocardial scintigraphy - Coronary disease

Exercise thallium-201 myocardial scintigraphy is a common diagnostic method for noninvasive assessment of coronary artery disease (CAD) (Bodenheimer et al. 1980; Hamilton 1979; H6r and Kanemoto 1981). Although it is recognized that myocardial scintigraphy has a higher sensitivity and specificity than conventional exercise ECG (Bodenheimer et al. 1980; H6r and Kanemoto 1981 ; Pfisterer and Mfiller-

Offprint requests to. J. Mi~ller-Brand

Brand 1981 ; Pfisterer 1982; Ritchie et al. 1978), its accuracy is still not sufficient to eliminate the necessity for cardiac catheterization in a considerable number of cases.

Accurate diagnosis of CAD is essential in patients with high-risk coronary patterns like triple-vessel disease (TVD) or left main stenosis, who should be evaluated for surgery. Although the frequency of abnormal thallium scintigrams increases with the number of vessels involved (Bodenheimer et al. 1980; H6r and Kanemoto 1981), false negative results still occur, especially in absence of relative perfusion differences due to similar-grade stenoses in all vessels, resulting in global ischemia (H6r and Kanemoto 1981; Pfisterer 1982). The reported sensitivity of visual ischemia detection in TVD patients varies between 70% and 100% (Dash et al. 1979; H6r and Kanemoto 1981; Leppo et al. 1979), but the scintigraphic identification of a TVD pattern has a much lower sensitivity and appears problematic with visual interpretation alone (Bodenheimer et al. 1980; Dash et al. 1979; Iskandrian et al. 1983; Maddahi et al. 1986).

Numerous techniques have been developed for quantitative assessment of planar thallium scintigrams (Boden- heimer et al. 1980; Buell et al. 1979). Computerized quantification allows image enhancement, background subtraction, calculation of global and regional uptake and washout as well as generation of circumferential profiles that can be compared to a normal range (Burow et al. 1979; Garcia et al. 1981 ; Maddahi et al. 1981 ; Van Train et al. 1986; Watson et al. 1981). Quantitative assessment increased sensitivity in patients with multivessel disease or moderate- degree stenoses (Berger et al. 198l; Maddahi et al. 1981, 1986). Additional quantitative criteria like stress lung thallium uptake (Boucher et al. 1983; Canhasi et al. 1985), de- layed redistribution (Beller et al. 1985) and global slow washout (Gewirtz et al. 1983), have been introduced.

However, quantification does not eliminate the need for visual judgment by an experienced observer (Beller 1984) and many of the computer programs are too time-consum- ing to be used in all patients in a busy routine setting, It was therefore the aim of the present retrospective analysis to evaluate simple quantitative criteria to detect ,,global lschemia" in apparently normal thallium scintigrams of TVD patients.

Patients and methods

Patients

For this study, the scintigrams of 179 consecutive patients studied between May 1984 and December 1985 were ana-

270

Table 1. Times of exposure (quotients rest image/stress image)

Projection Patients Controls Statistics (n = 25) (n = 17)

LAO 45 ° 1.65_+0.35 1.86_+0.46 n.s. LAO 75 ° 1.58_+0.30 1.78_+0.36 n.s. A.P. 1.60±0.16 1.75___0.46 n.s.

Mean 4- SD

lysed. In all patients, TVD was subsequently confirmed by coronary angiography ( > 50% stenoses in all 3 vessels). At visual interpretation, ischemia was found in 141/179 patients (79%), scar without ischemia in 25 (14%) whereas no perfusion defect was detected in 13 (7%) of patients.

The 38 patients without visually detected ischemia (i.e., with a ,,scar without ischemia" finding or no perfusion defect) were considered for quantitative comparison to a control group. Of these 38 subjects, preset count analog images and digital images were available in 25 and 20 patients respectively.

The control group consisted of 24 subjects with myocardial scintigrams and coronary angiograms effected for sus- picion of CAD. 15 of these subjects (subset 1) had anglo- graphically normal coronary arteries, whereas 9 subjects (subset 2) had insignificant coronary atheromatosis (7) or mitral valve prolapse (2). Digital images were available in all 24 subjects, analog images in 17.

Methods

Study protocol. A standard upright bicycle exercise test was performed in three-minute steps up to the appearance of symptoms or 90% of the maximum predicted heart rate. 2 mCi T1-201-chloride were injected intravenously one minute before termination of exercise. Thallium imaging was carried out with an Anger-type scintillation camera and an all-purpose parallel hole collimator.

Scintigraphic images were made immediately after exercise and 4 h later in LAO 45 °, LAO 75 ° and AP projections. The data were recorded in two ways: (1) as preset-count analog images (3000 counts over the hottest square centime- ter of the myocardium), registering the time of exposure in both stress and rest images; (2) as digital images with time preset to 5 rain, stored in a 64 x 64 pixel matrix. In order to record twice the same spot in analog images, the hottest spot of the stress image was indicated on a drawing and the camera was repositioned accordingly for the rest image.

Analysis of data. Analog images were interpreted visually by at least two experienced observers and classified as "ischemia", " sca r" o r " no pathological finding".

In preset-count analog images, the times of exposure of stress and rest images were compared resulting in a quotient as shown in Table 1.

Digital images were evaluated (1) for total myocardial activity and (2) for lung activity, using LAO 45 ° and AP projections. Myocardial activity (Fig. 1) was measured in elliptical (ME) and horseshoe-shaped (MH) ROIs. Back- ground activity (Fig. 2) was measured adjacent to the myocardium (B1, B2) and in the right paracardial lung field in AP projection (B3). Lung activity (Fig. 3) was measured using left paracardial ROIs of 3 different sizes (L1-L3).

Ler,

)Later<at

I ,~i k ero- ! Inferoapika[ lateral

B2

Se!c inferoa[

3 ApikaI lnferoopikal Lu L=, ~

A.R LAO/+5 ° Fig. 1. Regions of interest of the myocardium. ME = elliptical; MH = horseshoe-shaped

Fig. 2. Regions of interest for background activity (B1-B3)

Fig. 3. Regions of interest of the lung (L1-L3)

Counts per pixel activities were determined for each ROI. The myocardial ROIs were evaluated with and without background subtraction, subtracting either the corre- sponding horseshoe-shaped background field (BI, B2) in the same projection or the B3 field in AP projection. The stress activity was divided by the rest activity, resulting in 12 quotients referring to the myocardium (Table 2) and 6 quotients referring to the lung (Table 3). The absolute values of stress lung uptake were evaluated separately in Table 4. In Tables ~ 4 , both the whole control group (normal coronary arteries and minor abnormalities) and, separately, the control subset 1 with normal coronary arteries were used for comparison.

Table 2. Myocardial ROIs (quotients stress image/rest image)

Projection Quotient Patients (n20) Controls (n : 24) Controls (n = 15)

LAO 45 ° ME 1.51_+0.21 1.62_+0.21 n.s. 1.59_+0.23 n.s. MH 1.49+0.18 1.63-+0.21 P<0.05 1.59_+0.23 n.s. (ME-B1) 1.99-+0.83 2.15+0.55 n.s. 2.11_+0.64 n.s. (MH-B1) 1.86_+0.55 2.12-+0.48 n.s. 2.07_+0.57 n.s. (ME-B3) 1.87_+0.48 2.03_+0.38 n.s. 2.01_+0.43 n.s. (MH-B3) 1.83_+ 0.42 2.03_+ 0.37 n.s. 2.01_+ 0.44 n.s.

A.P. ME 1.44_+0.19 1.54_+0.23 n.s. 1.50_+0.25 n.s. MH 1.44_+0.21 1.55_+0.23 n.s. 1.51_+0.25 n.s. (ME-B2) 1.84_+0.44 2.14_+0.66 n.s. 2.08_+0.74 n.s. (MH-B2) 1.86_+0.44 2.16_+0.62 n.s. 2.13_+0.70 n.s. (ME-B3) 1.69_+0.32 1.93_+0.41 P<0.05 1.91_+0.47 n.s. (MH-B3) 1.70_+0.33 1.95_+0.40 P<0.05 1,93_+0.45 n.s.

271

Mean _+ SD For symbols, refer to Figures i and 2

Table 3. Lung ROIs (quotients stress image/rest linage)

Projection Quotient Patients (n = 20) Controls (n = 24) Controls (n = 15)

LAO45 ° LI 1.52_+0.17 1.41_+0.19 n.s. 1.40___0.16 P<0.05 L2 1.55_+0.20 1.43_+0.18 P<0.05 1.40_+0.13 P<0.05 L3 1.49_+0.20 1.37_+0.19 P<0.05 1.33__+0.14 P<0.05

A.P. L1 1.45_+0.30 1.32_+0.15 n.s. 1.26_+0.13 P<0.05 L2 1.34_+0.31 1.21 _+0.15 n.s. 1.16___0.13 P<0.05 L3 1.27_+0.26 1.17_+0.15 n.s. 1.10__+0.13 P<0.05

Mean-+ SD For symbols, refer to Figure 3

Statistical methods. Data are presented as mean values + / - s tandard deviat ion (SD). Student ' s t-test was appl ied to compare the mean values of pat ients and controls.

Results

1. Times of exposure

The quotients are summarized in Table 1. The mean values of the pat ients ' group (1.65, 1.58, 1.60) are generally lower than those of the control group (1.86, 1.78, 1.75), but no statistical differences could be found.

2. Myocardial ROIs

Quotients are tabula ted in Table 2. All values of the pat ients are lower than those of the controls.

When compar ing the patients with the whole control group ( n = 24), significant differences were found (1) in the L A O 45 ° projection, M H field (without background subtract ion) and (2) in the A P projection, ME and M H fields, after subtract ion of the B3 field.

The compar ison of the same patients with the subset 1 o f controls (n = 15) did not show statistical differences.

3. Lung ROIs

The resulting quotients are shown in Table 3. All values of the pat ients are higher than those of the controls.

When compar ing the pat ients with the whole control

Table 4. Lung ROIs (stress lung uptake)

Projection Quo- Patients Controls Controls tient (n = 20) (n = 24) (n = 15)

LAO 45 ° Lt 2 7 3 _ + 5 2 253_+65 n.s, 259_+66 n.s. L2 279___57 253_+69 n.s. 261_+72 n.s. L3 263_+ 54 235_+61 n.s. 243_+64 n.s.

A.P. Lt 2 6 0 _ + 5 3 229_+54 n.s. 231_+58 n.s. L2 238___50 208_+48 n.s. 211___52 n.s. L3 222-+43 198-+44 n.s. 198-+47 n.s.

Counts/pixel, mean-+ SD For symbols, refer to Fig. 3

group (n=24) , 2 of 6 quotients (L2 and L3 in LAO 45 ° projection) yield significant differences.

However, when the patients were compared only with the subset 1 of controls ( n = 15), all six quotients (ROIs L1-L3 in L A O 45 ° and AP projections) discriminated sig- nificantly between patients and controls. The mean values of the 15 subjects in subset 1 are generally lower than those o f the whole control group and have a smaller variabili ty.

The absolute values of stress lung uptake are tabulated in Table 4. Al though the uptake is generally higher in patients than in both subsets o f controls, no statistical differences can be found.

Discussion

The aim of this s tudy was to improve the sensitivity of thallium-201 exercise myocardia l scint igraphy in patients

272

with global myocardial ischemia by simple quantitative methods. In 38 of 179 patients with TVD documented by coronary angiography, no ischemia was found at visual evaluation. These patients were compared to a control group with normal (subset 1) or insignificantly altered (subset 2) coronary arteries. The exposure time quotients in preset-count analog images and the stress lung activity did not yield significant differences, whereas stress/rest activity quotients in myocardial and lung ROIs were partly able to discriminate between patients and controls (3 of 12 myocardial and all 6 lung quotients). The influence of using the whole control group or its subset 1 is discussed below.

Visual interpretation

A 93% frequency of pathological scintigrams (ischemia or scar) was observed in 179 patients with TVD. On the other hand, thallium-201 exercise imaging failed to demonstrate an ischemia in 38 patients or 21%, inspire or marked mor- phological abnormalities. These false-negative results un- derline that the sensitivity of visual interpretation is insufficient.

Times of exposure

Inspite of the general tendency to confirm the hypothesis that low quotients indicate inadequate perfusion increase under stress and/or slow myocardial washout, no statistical differences were found. The imaging technique (preset counts, counting the hottest myocardial area) may have contributed to the insufficient discrimination of patients and controls, since the hottest spot is the best perfused one and even in diffuse TVD, slight regional differences are seen. Moreover, it is difficult to identify twice the same reference spot of the myocardium.

Myocardial ROIs

A retrospective statistical discrimination of patients and controls was possible in 3 of 12 parameters. Designing horseshoe-shaped instead of elliptical ROIs resulted in slightly better discrimination (LAO 45 ° projection). As for background subtraction, the horseshoe-shaped areas (B1, B2) yielded too high variability of the quotients, probably because of liver interference in the apical portion of the ROI. In contrast, the use of the lung background area (B3) appeared to be beneficial, especially in AP projection where discrimination of both myocardial ROIs was improved.

The comparison of the patients with only the subset 1 of the control group (n = 15) did not show significant differences, due to the smaller number of patients. In sum- mary, this approach to myocardial washout quantification failed to yield a substantial diagnostic improvement. We attribute this to the variability of data (discussed below).

Lung ROIs

The analysis of stress lung thallium uptake (Table 4) showed only tendencies towards higher values in TVD patients. Discrimination was improved by dividing the stress uptake by the rest values (Table 3). This stress/rest ratio corrects for patient-dependent factors like fat tissue thick- ness that is a source of variability when considering absolute values. Significant differences were obtained in 2 of 6 pa-

rameters when the whole control group was considered. However, much better discrimination was observed in com- paring the patients only with the subset 1 of controls, i.e. subjects with entirely normal coronary arteries. The size of the lung ROI (L1, L2 or L3) seemed to be irrelevant.

It is remarkable that the subjects with entirely normal vessels had lower lung quotients than the whole control group, including subjects with insignificant coronary atheromatosis or mitral valve prolapse. On the other hand, the myocardial quotients of both subsets were homogeneous. An obvious interpretation of this finding is that the control group was homogeneous in terms of myocardial perfusion, but inhomogeneous with regard to left ventricular function. This is consistent with another study indicating that lung thallium uptake is a sensitive parameter of left ventricular dysfunction and may be elevated already in patients with moderate stenoses (Boucher et al. 1983).

Inspire of the good statistical discrimination, it was not possible to fix normal and pathological ranges of the lung quotients. Other authors (Levy et al. 1983) calculated the pulmonary thallium washout rate in CAD patients. Wash- out rates superior to 40% were considered pathological, resulting in a sensitivity and specificity of 64% and 75% respectively. The same calculation applied to the present study yielded only a 40% sensitivity, due to a larger overlap of data.

Variability

Many factors may have contributed to the considerable variability of the present data. Motion blur and low tracer energy are inherent in thallium myocardial imaging. Results are worsened when patients do not reach sufficient work- load due to exhaustion, ECG abnormalities or severe angina (Brown et al. 1984; Kaul et al. 1986). Substantial variability arises from imperfect camera reposition (Makler et al. 1985). Other phenomena like reverse redistribution (Silberstein and De Vries 1985) are not yet fully understood.

Conclusions

Recent data suggest that more sophisticated quantitative methods are needed for better scintigraphic detection and identification of multivessel CAD. Circumferential profile analysis on a multicenter basis (Van Train et al. 1986) and quantitative single photon emission tomography (Garcia et al. 1985) seem encouraging.

Although the present study did not offer a simple and sensitive quantitative parameter of multivessel coronary disease, it showed that an elevated stress/rest lung thallium uptake ratio may be a valuable supplementary indicator of CAD.

References

Beller GA (1984) Quantitative thallium-201 scintigraphy. Int J Car- diol 5 : 234-239

Beller GA, Gibson RS, Watson DD (1985) Radionuclide methods of identifying patients who may require coronary artery bypass surgery. Circulation 72: 9-22

Berger BC, Watson DD, Taylor GJ, Craddok GR, Martin RP, Teats CD (1981) Quantitative thallium-201 exercise scintigra- play for detection of coronary artery disease. J Nucl Med 22:585-593

Bodenheimer MM, Banka VS, Helfant RH (1980) Nuclear cardiol-

273

ogy. II. The role of myocardial perfuslon imaging using thallium-201 in diagnosis of coronary heart disease. Am J Cardiol 45 : 674-684

Boucher CA, Zir LM, Beller GA, Okada RD, Mekusick KA, Strauss H, Pohost G (1980) Increased lung uptake of thallium- 201 during exercise myocardial imaging: clinical, hemodynamic and angiographic implications in patients with coronary artery disease. J Am Cardiol 46:18%196

Brown KA, Boucher CA, Okada RD, Strauss H (1984) Quantifica- tion of pulmonary thallium-201 activity after upright exercise in normal subjects: importance of peak heart rate and propran- olol usage in defining normal values. Am J Card 53:1678-1682

Buell U, Kleinhans E, Seiderer M, Strauer BE (1979) Quantitative assessment of thallium-201 images Cardiovasc Radio[ 2:183-193

Burow RD, Pond M, Schafer AW, Becker L (1979) "Circumferen- tial profiles": a new method for computer analysis of thallium- 201 myocardial perfusion images. J Nucl Med 20:771 777

CanhasI B, Die M, Botvanick EH, Lancer P, Shaffman N (1985) Interaction of "supplementary" scintigraphic indicators of ischemia and stress electrocardiography in the diagnosis of mul- tlvessel coronary disease. J Am Colt Cardiol 6:581-588

Dash H, Massie BM, Botvinick EH, Brundage BH (1979) The noninvasive identification of left main and three-vessel coronary artery disease by myocardial stress perfusion scintigraphy and treadmill exercise electrocardiography. Circulation 60:276-283

Garcia EV, Maddahi J, Berman DS, Waxman A (1981) Space/time quantitation of thallium-201 myocardial scintigraphy. J Nucl Med 22:309-317

Garcia EV, Van Train KF, Maddahi J, Brisent F (1985) Quantifi- cation of rotational thallium-201 myocardial tomography. J Nucl Med 26:17-26

Gewirtz H, Paladino W, Sullivan M, Most AS (1983) Value and limitations of myocardial thallium washout rate in the noninvasive diagnosis of patients with triple-vessel coronary artery disease. Am Heart J 106:681-690

Hamilton GW (1979) Myocardial imaging with thallium-201 : the controversy over its clinical usefulness in ischemic heart disease. J Nucl Med 20:1201-1205

H6r G, Kanemoto N (1981) T1-201-myocardial scintigraphy: cur- rent status in coronary artery disease, results of sensitivity/specificity in 3092 patients and clinical recommendations. Nucl Med 20:136-147

Iskandrian AS, Hakki AH, Segal BL (1983} Assessment of the myocardial perfusion pattern in patients with multivessel coronary artery disease. Am Heart J 106:1089-1096

Kaul S, Chesler DA, Pohost GM Strauss HW (1986) Influence of peak exercise heart rate on normal thallium-201 myocardial clearance. J Nucl Med 27:26-30

Leppo J, Yipintsoi T, Blankstein R, Bonton R, Freeman L (1979) Thallium-201 myocardial scintlgraphy in patients with triple- vessel disease and ischemic exercise stress tests. Circulation 59:714-719

Levy R, Rozanski A, Berman DS, Garcia E, Swan HJ (1983) Anal- ysis of the degree of pulmonary thallium washout after exercise in patients with coronary artery disease. J Am Coll Cardiol 2:719-728

Liu P, Kiess M, Okada RD, Strauss H, Pohost G (1985) Increased thalhum lung uptake after exercise in isolated left anterior de- scending coronary artery disease. Am J Cardiol 55:1469-1473

Maddahi J, Garcia EV, Berman DS, Waxman A, Swan HJ (1981) Improved nomnvasive assessment of coronary artery disease by quantitative analysis of regional stress myocardial distribution and washout of thallium-201. Circulation 64:924-932

Maddahi J, Abdulla A, Garcia EV, Swan H J, Berman BS (1986) Noninvasive identification of left main and triple-vessel coronary disease: improved accuracy using quantitative analysis of regional myocardial stress distribution and washout of thallium-201. J Am Coll Cardlol 7:53-60

Makler PT, McCarthy DM, Velchik MG, Alavi A (1985) Sources of variability in quantitative thallium analysis. Invest Radiol 20 : 803-807

Pfisterer ME, Mfilter-Brand J (1981) Indikationen zur Thallium- Szintigraphle-Studie anhand yon 259 koronarographisch kon- trollierten F/illen. Angiokardlologie 3:297-304

Pfisterer ME (1982) Nuklearmedizinische Herzdiagnostik. Ed.: Springer, Berlin, Heidelberg, New York

Ritchie JL, Zaret BL, Strauss HW, Pitt B, Bermann HS (1978) Myocardial imaging with thallium-201 : a multicenter study in patients with angina pectoris or acute myocardial infarction. Am J Cardiol 42:345 352

Silberstein EB, De Vries DF (1985) Reverse redistribution phenom- enon in thallium-201 stress tests: angiographic correlation and clinical significance. J Nucl Med 26:707-710

Van Train KF, Berman DS, Garcia EV, Berger H J, Maddahi J (1986) Quantitative anaIysis of stress thallium-201 myocardial scintigrams : a multicenter trial. J Nucl Med 27 : 17-25

Watson DD, Campbell NP, Read EK, Gibson RS, Teats CD (1981) Spatial and temporal quantitation of plane thallium myocardial images. J Nucl Med 22:577-584

Received November 7, 1986 / May 5, 1987

Documents

Thallium-201 myocardial scintigraphy in coronary triple-vessel disease: An attempt to increase sensitivity using quantitative methods