- Email: [email protected]
Intercenter variability in outcome for patients treated with direct coronary angioplasty during acute myocardial infarction
Intercenter variability in outcome for patients treated with direct coronary angioplasty during acute myocardial infarction Timothy F. Christian, MD, FACC,a James H. O’Keefe, MD, FACC,b Marcus A. DeWood, MD, FACC,c Michael G. Spain, MD, FACC,d Cindy L. Grines, MD, FACC,b Peter B. Berger, MD, FACC,a and Raymond J. Gibbons, MD, FACCa Rochester, Minn.; Kansas City, Mo.; Royal Oak, Mich.; Spokane, Wash.; and Tulsa, Okla.
Background Direct coronary angioplasty is an effective therapy for acute myocardial infarction, but its success may be dependent on both ready availability and operator skill. The purpose of this study was to investigate the impact of the center performing direct coronary angioplasty for acute myocardial infarction while controlling for parameters known to affect outcome.
Methods and Results The study group consisted of 99 patients with ST elevation who were treated with direct angioplasty in four high-volume centers. Patients were injected with technetium-99m sestamibi intravenously and then taken to the cardiac catheterization laboratory. Antegrade flow was graded before and after direct coronary angioplasty. Single photon emission computed tomography was performed 1 to 6 hours after injection to measure myocardium at risk and residual blood flow to the jeopardized zone using previously published quantitative methods. A repeat sestamibi injection and tomographic acquisition were performed at hospital discharge to measure actual infarct size. There were no significant differences by center for baseline clinical characteristics, mean myocardium at risk (29% to 37% left ventricle [LV]), time to reperfusion (3.1 to 4.1 hours), residual blood flow, infarct location, or antegrade flow. Despite these similarities, there were differences in outcome measures by center. Mean infarct size was as follows: center 1, 15%; center 2, 12%; center 3, 10%, center 4, 23% (all LV; p = 0.11). Mean left ventricular ejection fraction at discharge also demonstrated significant differences: center 1, 0.57; center 2, 0.47; center 3, 0.53; center 4, 0.47 (p = 0.002). The prevalence of Thrombolysis in Myocardial Infarction grade 3 flow after angioplasty significantly differed by center: center 1, 92%; center 2, 94%; center 3, 87%; center 4, 71%; (p = 0.01). There was a low mortality rate for all four centers ranging from 0% to 6%. After adjustment for myocardium at risk, residual blood flow, and time to reperfusion, the primary outcome of the center where the angioplasty was performed was an independent determinant of both infarct size and left ventricular ejection fraction.
Conclusion The success of direct coronary angioplasty in reducing infarct size and preserving left ventricular function depends on the center performing the procedure. Direct measurement of the effectiveness of this reperfusion modality in community practice is required to assess the impact of this effect. (Am Heart J 1998;135:310-17.)
Direct coronary angioplasty is an effective method of reperfusion for acute myocardial infarction. Randomized clinical trials have shown that direct coronary angioplasty in selected centers is at least as effective as thrombolytic therapy if performed within 6 From the aDivision of Cardiovascular Diseases and Internal Medicine, Mayo Clinic and Mayo Foundation; the bDepartment of Cardiovascular Diseases, St. Luke’s Hospital, The Mid America Heart Institute, and the Department of Cardiovascular Diseases, William Beaumont Hospital; the cDepartment of Cardiovascular Diseases, Deaconess Medical Center, Spokane Heart Research Foundation; and the dDepartment of Cardiovascular Diseases, St. Francis Hospital, Tulsa Cardiology. Supported by a grant from Burroughs Wellcome Co., Research Triangle Park, N.C. Submitted Aug. 9, 1996; accepted July 24, 1997. Reprint requests: Timothy F. Christian, MD, Mayo Clinic, 200 First St. SW, West 16B, Rochester, MN 55905. E-mail: [email protected] Copyright © 1998 by Mosby, Inc. 0002/8703/98/$5.00 + 0 4/1/87274
hours of the onset of chest pain.1-3 Potential advantages of direct angioplasty include greater 90-minute patency rates and a more likely restoration of brisk antegrade flow compared with thrombolytic therapy.4 However, one potential disadvantage compared with thrombolytic therapy is the dependence on operator skill. The randomized clinical trials performed to date have been in selected centers highly proficient in coronary angioplasty.1-3 Intercenter variability of results has not been assessed. The purpose of this study was to investigate the variability in outcome among centers performing direct coronary angioplasty for acute myocardial infarction. The primary response variable was the impact of the center performing the procedure on infarct size after adjustment for other important variables known to
American Heart Journal Volume 135, Number 2, Part 1
Christian et al.
Table I. Clinical characteristics by center
No. patients Age (yr) Female Diabetes Vascular disease Smoking history Prior Current Hypercholesterolemia Hypertension Anterior location of MI Prior MI Prior CHF Culprit vessel = vein graft
Center 1
Center 2
Center 3
Center 4
p Value
37 66 ± 11 32% 27% 8%
33 59 ± 12 21% 21% 9%
15 62 ± 7 27% 0% 13%
14 59 ± 14 21% 14% 0%
— 0.09 >0.2 0.15 >0.2
27% 24% 43% 49% 30% 19% 0% 3%
36% 42% 42% 36% 24% 33% 9% 3%
27% 53% 40% 40% 27% 13% 0% 0%
29% 50% 36% 21% 36% 14% 0% 7%
0.14 >0.2 >0.2 >0.2 >0.2 0.10 >0.2
MI, Myocardial infarction; CHF, congestive heart failure.
influence infarct size—myocardium at risk, collateral flow, and time to coronary artery reperfusion.5,6 Such an approach is clinically feasible using sequential tomographic perfusion imaging with technetium-99mm sestamibi, which has been extensively used in prior studies of acute infarction.2,7-9
Methods The study group was selected from patients enrolled in a prospective randomized study of the use of poloxamer-188, a copolymer surfactant, as ancillary therapy with direct angioplasty for acute myocardial infarction (Table I). Patients were randomly assigned in a 2:1 ratio, drug-to-placebo, stratified by center. As previously reported, there was no demonstrable benefit of poloxamer-188 on myocardial salvage measured by acute and discharge 99mTc sestamibi imaging compared with placebo.10 Patients were eligible for this study if they met the following inclusion criteria: (1) enrollment in the randomized trial of poloxamer-188; (2) presentation within 6 hours of the onset of chest pain; (3) electrocardiographic ST-segment elevation of 2 mm in two contiguous anterior leads (V1 through V4) or 1 mm in two inferior leads (II, III, aVF). The following were reasons for exclusion from the randomized trial: (1) left bundle branch block on the resting electrocardiogram, (2) women of child-bearing potential without a negative serum pregnancy test, (3) prior use of thrombolytic therapy during the course of the myocardial infarction; (4) serum creatinine level ≥3 mg/dl. A total of 150 patients were enrolled in the randomized trial. In addition to the randomized trial criteria, the following exclusions were specific to this study: (1) incomplete acute 99mTc sestamibi imaging or clinical data (16 patients) and (2) performance of direct angioplasty at a center enrolling fewer than 10 patients into the trial (35 patients,
five centers). Radionuclide measurements for these 35 excluded patients are provided in footnotes to Tables II and III. Angiographic core laboratory analysis was not available for these patients; therefore, 99 patients formed the study group. The four centers that enrolled 10 patients or more meeting the inclusion criteria and none of the exclusion criteria included The Spokane Heart Research Institute and Deaconess Hospital, Spokane, Wash.; William Beaumont Hospital, Royal Oak, Mich.; Cardiology of Tulsa and St. Francis Hospital, Tulsa, Okla.; Mid America Heart Institute and St. Luke’s Hospital, Kansas City, Mo. All four centers meet the American College of Cardiology volume guidelines for centers performing direct angioplasty.11 The Mayo Clinic functioned as the core laboratory for the perfusion images but did not contribute any patients to the randomized trial. It should be noted that one center (center 1) had a competing randomized protocol that may have influenced the enrollment of patients into this trial.
Radionuclide data The methods of 99mTc sestamibi imaging in acute myocardial infarction have been previously described in detail.9,12 In brief, all patients were injected with 20 to 30 mCi of 99mTc sestamibi after giving informed consent for enrollment into the randomized trial of poloxamer-188 with direct coronary angioplasty. Tomographic imaging was performed 1 to 6 hours after injection depending on the duration of the coronary angioplasty procedure with standard tomographic imaging techniques but with different cameras and computer software at the individual centers.9 Because redistribution despite reperfusion therapy is minimal,13,14 images acquired after reperfusion still reflect myocardial blood flow before direct coronary angioplasty.14,15 This property has allowed the assessment of myocardium at risk without delaying reperfusion therapy. A second injection of 99mTc sestamibi and tomographic acquisition were performed 5 to 8 days later to assess final infarct size in 93 of 99 patients.
311
312
American Heart Journal February 1998
Christian et al.
Table II. Radionuclide and angiographic measurements
Myocardium at risk (% LV) Radionuclide residual flow (nadir) TIMI 0 or 1 flow before PTCA (% patients) Time to first balloon inflation (hr)
Center 1 (n = 37)
Center 2 (n = 33)
Center 3 (n = 15)
Center 4 (n = 14)
29 ± 19 0.26 ± 0.20 68% 3.9 ± 1.4
30 ± 20 0.25 ± 0.18 58% 4.1 ± 1.5
30 ± 19 0.31 ± 0.17 67% 3.4 ± 1.1
37 ± 25 0.24 ± 0.20 64% 3.1 ± 1.5
p Value >0.20 >0.20 >0.20 0.17
PTCA, Percutaneous transluminal coronary angioplasty. Combined data from the remaining five centers (35 patients) not qualifying for the study (enrollment of fewer than 10 patients meeting the inclusion criteria per center) were myocardium at risk 40% ± 22% LV and infarct size 16% ± 17% LV. LVEF was 0.48 ± 0.13.
Table III. Univariate outcome measures
Infarct size (% LV) Mean ± SD Range LVEF Mean ± SD Range TIMI 3 flow after PTCA
Center 1
Center 2
Center 3
Center 4
p Value
15% ± 16% 0%-62%
12% ± 13% 0%-51%
10% ± 10% 0%-27%
23% ± 23% 0%-67%
0.11
0.56 ± 0.11 0.37-0.78 92%
0.47 ± 0.08 0.27-0.64 94%
0.53 ± 0.11 0.37-0.73 87%
0.47 ± 0.17 0.22-0.79 71%
0.002 0.01
Combined data from the remaining five centers (35 patients) not qualifying for the study (enrollment of fewer than 10 patients meeting the inclusion criteria per center) were myocardium at risk 40% ± 22% LV and infarct size 16% ± 17% LV. LVEF was 0.48 ± 0.13.
The left ventricular perfusion defect size was quantified at the core laboratory (Mayo Clinic) with a previously described method that uses a threshold of 60% of maximal counts.9,12,16 This threshold has been validated in phantom studies,16 has provided close correlation with other surrogate measures of infarct size,7,17-20 and has been independently shown to provide the best separation of myocardial segments with improved contractility after revascularization from those unchanged after revascularization.21 The acute perfusion defect reflects myocardium at risk and the final defect reflects infarct size. A noninvasive measure of residual blood flow to the infarct zone was obtained for each patient from the acute sestamibi tomographic defect.9 This measure has previously been shown to be highly associated with acute angiographic collaterals in both single9 and multicenter22 studies of myocardial infarction and has been shown to be an independent predictor of infarct size in addition to myocardium at risk and duration of coronary occlusion. It has also been shown to correlate significantly with collateral blood flow measured by radio-labeled microspheres in an animal study of coronary occlusion.23 This measure, termed the nadir, is calculated as the lowest ratio of minimum to maximum counts per short-axis slice from the acute circumferential count profile curves (Fig. 1). All four centers were validated for quality control and measurement reproducibility by using a cardiac phantom. Defects of known size were inserted into the walls of the phantom
left ventricle (LV) and imaged. Measured defect size was compared with actual defect size over a range of defect sizes. Fig. 2 shows the results for the four centers in this study. These results are part of a larger study by O’Connor et al.24 analyzing all 14 centers involved in the evaluation of poloxamer-188.
Radionuclide ventriculography Left ventricular ejection fraction (LVEF) was determined in 96 of the 99 patients in the study group with gated equilibrium radionuclide ventriculography. Red cells were labeled with 99mTc pyrophosphate with a modified in vivo method,25 and LVEF was determined with commercial software in clinical use at each center. No central laboratory review or processing was performed. These studies were acquired 1 day after the second sestamibi study at discharge. Both infarct size and ejection fraction measures were available for 90 of 99 patients.
Coronary angiography All patients underwent acute coronary angiography as part of a treatment strategy for direct coronary angioplasty.26 Antegrade flow was scored with the Thrombolysis in Myocardial Infarction (TIMI) grading system by a single blinded observer (P.B.B.) not involved with any of the centers enrolling patients before or after intervention. TIMI 0 and 1 grades were combined because their outcomes have been shown to be nearly identical.27
American Heart Journal Volume 135, Number 2, Part 1
Figure 1
Christian et al.
Figure 2
Noninvasive measurement of collateral flow and determination of defect extent. Circumferential count profile curve from shortaxis slice demonstrating acute inferior perfusion defect. Extent of defect is defined by number of pixels with <60% of maximal counts. Residual flow (collateral and antegrade) to jeopardized zone is estimated by severity of perfusion defect. Severity (nadir) is determined by lowest ratio of minimum to maximum counts (B/A).
True versus tomographic defect size from cardiac phantom for four participating centers. Defects of known size are placed within walls of phantom and imaged at individual centers. There is very close correlation for each center despite different cameras and computer systems.
Statistics
Baseline radionuclide and angiographic measures
Data are presented as mean ± SD. Analysis of variance was used to compare continuous variables by angioplasty center. Noncontinuous variables were compared with chi-square contingency table analysis or Fischer’s exact test. Multivariable (analysis of covariance) analysis was used to determine whether the primary response variable of the center performing the procedure was an independent determinant of infarct size after adjustment for other variables known to affect the dependent variable. A similar (secondary) multivariable analysis was performed with LVEF as the dependent variable. A separate univariate analysis of infarct size and LVEF by center was performed with imputed values for the four patients who died before acquisition of these measures. The worst value reported among the four centers was assigned to these patients. All analyses were performed with Statview 4.1 (Abacus software) or Supernova 1.0 (Abacus software) commercially available statistical software programs.
The mean value for myocardium at risk for the overall study group was 30% ± 20% of the LV and was not significantly different among centers (Table II). The mean nadir for the study group was 0.26 ± 0.19, indicating that the average severity of the acute perfusion defect was 26% of maximal counts in the most severely hypoperfused short-axis tomographic slice. This acute measure of residual blood flow was not different among the four centers. The degree of persistent antegrade flow assessed by coronary angiography before direct coronary angioplasty was similar; TIMI grade 0 or 1 flow was present in 68%, 58%, 67%, and 64% of patients in centers 1 through 4, respectively.
Results Baseline clinical variables Table I shows clinical characteristics of the study group by center. There were no significant differences in these variables. Table II describes elapsed time from chest pain onset to first balloon inflation. The mean times, which ranged from 3.1 (center 4) to 4.1 hours (center 2), were not significantly different among centers.
Univariate radionuclide and angiographic outcome measures Although there were no clinical, radionuclide, or angiographic differences among centers, there were differences in outcome variables by center (Figs. 3 through 5, Table II). There was a trend for infarct size to differ among centers (Fig. 3, Table III). Centers 1 through 3 had mean infarct size measures ranging from 10% ± 10% LV to 15% ± 16% LV, and center 4 had a mean infarct size of 23% ± 23% LV. By analysis of variance, this trend did not reach statistical significance (p = 0.11). LVEF was significantly different by center (Fig. 4, Table III), with mean values ranging from 0.47 ± 0.14 (center 4) to 0.56 ± 0.10 (center 1) (p = 0.002).
313
314
American Heart Journal February 1998
Christian et al.
Figure 3
Mean and 95% confidence intervals of end point of infarct size by discharge 99mTc sestamibi perfusion imaging is presented by center as percent of LV. There is a trend toward difference in outcome by center that does not reach statistical significance (p = 0.11).
Figure 4
Figure 5
Antegrade flow after direct coronary angioplasty graded with TIMI criteria. There is significant difference by center following same pattern as infarct size.
Mortality rate There were four patients who died during the course of hospitalization. These patients did not have infarct size or LVEF measures and could not be included in the above analysis. Mortality rate as a percent of total patients enrolled (whether or not they were excluded from this substudy) by center was center 1, 2%; center 2, 5%; center 3, 0%; and center 4, 6%. Thus mortality rate was not significantly different by center. Including these patients in the univariate analysis by imputing the worst value for infarct size (67% LV) and LVEF (0.22) did not change the univariate analysis for these measures: infarct size p = 0.11 and LVEF p = 0.002.
Mean and 95% confidence intervals of LVEF determined by radionuclide ventriculography by center. This end point was not analyzed in a core laboratory. There is significant difference in outcome by center for this variable (p = 0.002).
There were 11 patients with TIMI flow <3 after angioplasty (10%), four with TIMI 0 or 1 flow, and seven with TIMI 2 flow. Fig. 5 shows the distribution of TIMI flow after angioplasty. Restoration of TIMI 3 flow after direct coronary angioplasty (90% of patients) ranged from 87% to 94% of patients for centers 1 through 3 but was only 71% for center 4 (Fig. 5). The distribution of TIMI flow after angioplasty was significantly different by center (p = 0.01).
Multivariable analysis of outcome Although there were no baseline differences for clinical, radionuclide, or angiographic variables by center on univariate analysis, it is possible that trends existed unfavorable to a center that, taken cumulatively, could profoundly affect the outcome measures. Because the elapsed time to reperfusion, magnitude of residual (collateral plus antegrade) flow, and extent of myocardium are known to be independently at risk associated with infarct size in models of reperfusion,5,6 these variables were forced into the model. The primary response variable of the center performing the procedure and the clinical and angiographic variables from Tables I and II were then tested for an independently significant asso-
American Heart Journal Volume 135, Number 2, Part 1
ciation with infarct size. Only the center where the procedure was performed provided additional independent information (Table IV). A similar significant association with outcome was seen in this multivariable design when LVEF was the dependent variable (Table IV).
Discussion Myocardium at risk and collateral flow are important determinants of infarct size.5 Prior studies have demonstrated that failure to account for all these variables can lead to erroneous conclusions regarding treatment efficacy.5 Restoration of flow in animal studies entails a simple release of a mechanical constrictor and does not account for the variability of success in restoring flow in clinical practice. Although direct coronary angioplasty is highly effective in restoring antegrade flow in acute myocardial infarction, there is much more potential for variability than in animal models of reperfusion. This study examined whether the variability inherent in direct coronary angioplasty at different centers had significant impact on final infarct size after accounting for the three other measurable variables (myocardium at risk, collateral flow, and time to reperfusion) known to be important. Overall, the results were impressive, with an overall 3% mortality rate. Myocardium at risk was 30% of the LV with a final infarct size of 14% of the LV, representing salvage of more than half the myocardium at risk on average. Mean ejection fraction was normal. All three measures of outcome compare very favorably to previously reported values of other trials.1-3 However, there were differences in outcome by center. Despite similar measures of myocardium at risk, collateral flow, and presumed duration of coronary occlusion, patients undergoing direct angioplasty at center 4 tended to have a poorer outcome as defined by this study. The differences in LVEF and antegrade flow after angioplasty were statistically significant. After controlling for factors known to determine infarct size (myocardium at risk, residual blood flow, time to reperfusion), the center where the procedure was performed was an independently significant factor in determining infarct size. The lower prevalence of TIMI 3 flow for this center after angioplasty suggests a possible mechanism for this difference. The impact on outcome of antegrade flow after reperfusion has been described in two large studies of myocardial infarction. Vogt et al.28 demonstrated that mortality rate was nearly three times greater in patients with TIMI 2 compared with patients with TIMI 3 flow and was similar to patients with TIMI 0 or 1 flow. This
Christian et al.
Table IV. Impact of the center performing PTCA on infarct size and LVEF after controlling for known determinants of outcome Variables Infarct size (n = 93) Myocardium at risk Residual blood flow Time to first inflation Center performing PTCA LVEF (n = 96) Myocardium at risk Residual blood flow Time to first inflation Center performing PTCA
p Value
0.02 0.0007 0.17 0.01 0.02 0.06 >0.20 0.0006
PTCA, Percutaneous transluminal coronary angioplasty.
finding may be caused by differences in resultant ventricular function after thrombolysis. The GUSTO investigators found that LVEF was significantly greater and regional wall motion abnormalities less extensive in patients with TIMI 3 flow compared with those with TIMI ≤2 flow.27 The results of this study, in which the lower the prevalence of TIMI 3 flow occurred in the center with the largest mean infarct size, suggest that these differences in mortality rate and left ventricular function may be caused by differences in myocardial salvage by antegrade flow after angioplasty. The differences in left ventricular systolic function were consistent with the measures of infarct size with the exception of center 2, where the mean LVEF was lower than centers 1 and 3 despite nearly identical values for infarct size. Fig. 4 demonstrates a narrow dispersion of values for center 2; 90% of patients had an ejection fraction <0.55 despite 25% of patients having normal perfusion images. This finding likely represents a lower range of normal values of this center and underscores the need for core laboratories to process and measure end points, which was not the case for LVEF in this trial. The range in values for center 4 was much broader. The discrepant outcome by center is in agreement with prior reports analyzing surgical outcome in the Coronary Artery Surgery Study.29,30 Mortality rate by site in 15 centers was analyzed and found to vary from 0.3% to 6.4%; a more than 20-fold range.29 The wide variability in outcome by center was preserved when baseline risk variables were adjusted.30 Variability by center in major complication rates has been described for elective coronary angioplasty as a function of patient volume.31 Consequently, it is not unexpected that there was vari-
315
316
American Heart Journal February 1998
Christian et al.
ability in outcome in this study of a therapy that is technically demanding.
Limitations The small number of patients available for analysis in two of the centers is a limitation of this study. There were consistent trends in the same direction for the three major outcome variables of this study: infarct size, ejection fraction, and antegrade flow after angioplasty, with only the latter two achieving statistical significance. Consequently, a type II error may be present for the univariate analyses. However, the multivariate analysis results are compelling. Had larger numbers of patients been enrolled, the trend in infarct size may have become significant by univariate analysis. These findings do suggest, however, that intercenter differences in outcome for direct angioplasty potentially exist. It should be noted that the use of acute imaging to define myocardium at risk adds considerable statistical power when analyzing outcomes in acute myocardial infarction treated with reperfusion therapy. Approximately 70% of the variability in outcome in infarct size can be accounted for by this measure alone.5 We have previously demonstrated that incorporation of myocardium at risk, collateral flow, and time to reperfusion accounts for 69% of the variability in infarct size in the clinical setting.9 Consequently, fewer patients are required to detect a significant impact on infarct size of a particular variable, such as the center where the reperfusion strategy occurs.32 This factor is in sharp distinction to trials that cannot account for baseline differences in these key variables but rely on mortality and reinfarction rates. Gibbons et al.2 demonstrated no significant difference in myocardial salvage between two reperfusion strategies (thrombolysis and direct angioplasty) with adequate statistical power with fewer than 50 patients per treatment arm. It should also be noted that the number of patients in two of the centers enrolling fewer patients was similar to the mean number of patients per center (n = 16) in a large published trial of direct angioplasty.1 Consequently, the limited number of patients in two of the centers in this study should be viewed within the context of the power of the technique. One center (center 1) had competing trials of direct angioplasty and thrombolytic therapy that may have biased enrollment at that center toward a sicker population in this trial. Despite this bias, the outcome was exceptionally good. None of the other three centers had competing trials. This study was conducted in four
tertiary referral centers with high patient volumes. How the inclusion of community-based centers would have impacted on the variability is unknown, but this issue clearly is key in examining the national use of direct coronary angioplasty as routine therapy for acute myocardial infarction. This study only examined short-term outcome variables. Consequently no conclusions regarding long-term outcome can be made from this study. The findings of this study imply that there may be a difference between the efficacy of direct coronary angioplasty and its effectiveness when applied on a broad scale. The efficacy (the performance of a test procedure under ideal conditions) of direct coronary angioplasty has been demonstrated in previous studies.1-3 The effectiveness (the use of a test or procedure for the individual patient in a routine setting) of direct coronary angioplasty remains to be determined. Our results should not be interpreted as evidence that direct coronary angioplasty is useful only in certain institutions. Rather, the findings presented here should be considered as suggestive evidence that significant differences in outcome may exist with this technique of reperfusion. Although mean values for outcome were excellent in the largest multicenter trial of direct coronary angioplasty,1 it is logical to presume that there were consistent outliers in both directions. This intercenter variability in outcome should be considered when comparing direct angioplasty to less skill-dependent modalities of reperfusion, such as intravenous thrombolysis.
References 1. Grines CL, Browne KF, Marco J, et al. A comparison of immediate angioplasty with thrombolytic therapy for acute myocardial infarction. N Engl J Med 1993;328:673-9. 2. Gibbons RJ, Verani MS, Behrenbeck T, et al. Feasibility of tomographic Tc-99m-hexakis-2-methoxy-2-methylpropyl-isonitrile imaging for the assessment of myocardial area at risk and the effect of acute treatment in myocardial infarction. Circulation 1989;80:1277-86. 3. Zijlstra F, deBoer MJ, Hoorntje JCA, Reiffers S, Reiber JHC, Suryapranata H. A comparison of immediate coronary angioplasty with intravenous streptokinase in acute myocardial infarction. N Engl J Med 1993;328:680-4. 4. Berger PB, Bell MR, Holmes DR, Gersh BJ, Hopfinspirger MR, Gibbons RJ. Time to reperfusion with direct coronary angioplasty and thrombolytic therapy in acute myocardial infarction. Am J Cardiol 1994;73:231-6. 5. Reimer KA, Jennings RB, Cobb FR, et al. Animal models for protecting ischemic myocardium: results of the NHLBI cooperative study. Comparison of unconscious and conscious dog models. Circ Res 1985;56:651-65. 6. Reimer KA, Jennings RB. The wavefront phenomenon of myocardial ischemic cell death. II. Transmural progression of necrosis within the
American Heart Journal Volume 135, Number 2, Part 1
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
framework of ischemic bed size (myocardium at risk) and collateral flow. Lab Invest 1979;40:633-44. Behrenbeck T, Pellikka PA, Huber KC, Bresnahan JF, Gersh BJ, Gibbons RJ. Primary angioplasty in myocardial infarction: assessment of improved myocardial perfusion with technetium-99m-isonitrile. J Am Coll Cardiol 1991;17:365-72. Pellikka PA, Behrenbeck T, Verani MS, Mahmarian JJ, Wackers FJTh, Gibbons RJ. Serial changes in myocardial perfusion using tomographic technetium-99m-hexakis-2-methoxy-2-methylpropyl-isonitrile imaging following reperfusion therapy of myocardial infarction. J Nucl Med 1990;31:1269-75. Christian TF, Schwartz RS, Gibbons RJ. Determinants of infarct size in reperfusion therapy for acute myocardial infarction. Circulation 1992; 86:81-90. Grines CL, O’Keefe JH, DeWood MA, et al. A prospective, randomized, placebo-controlled trial of RheothRx (Poloxamer 188, NF) to limit infarct size in humans [abstract]. Circulation 1993;80:I-292. Ryan TJ, Klocke FJ, Reynolds WA, ACP/ACC/AHA Task Force on Clinical Privileges in Cardiology. Clinical competence in percutaneous transluminal coronary angioplasty. Circulation 1990; 81:2041-6. Christian TF, Gibbons RJ, Gersh BJ, Gibbons RJ. Effect of infarct location on myocardial salvage assessed by technetium-99m-isonitrile. J Am Coll Cardiol 1991;17:1303-8. Okada RD, Glover D, Gaffney T, William S. Myocardial kinetics of technetium-99m-hexakis-2-methoxy-2-isonitrile. Circulation 1988; 77:491-8. DeCoster PM, Wijns W, Cauwe F, Robert A, Beckers C, Melin JA. Area at risk determination by technetium-99m-hexakis-2-methoxyisobutyl-isonitrile in experimental reperfused myocardial infarction. Circulation 1990;82:2151-62. Sinusas AJ, Trautman KA, Bergin JD, et al. Quantification of “area at risk” during coronary occlusion and degree of myocardial salvage after reperfusion with technetium-99m-methoxyisobutyl-isonitrile. Circulation 1990;82:1424-37. O’Connor MK, Hammell TC, Gibbons RJ. In vitro validation of a simple tomographic technique for estimation of percent myocardium “at risk” following administration of Tc99m isonitrile. Eur J Nucl Med 1990;17:69-76. Christian TF, Behrenbeck T, Pellikka, PA, Huber KC, Chesebro JH, Gibbons RJ. Mismatch of left ventricular function and infarct size demonstrated by technetium-99m-isonitrile imaging after reperfusion therapy for acute myocardial infarction: identification of myocardial stunning and hyperkinesia. J Am Coll Cardiol 1990;16:1632-8. Christian TF, Behrenbeck T, Gersh BJ, Gibbons RJ. Relation of left ventricular volume and function over one year after acute myocardial infarction to infarct size determined by technetium-99m sestamibi. Am J Cardiol 1991;68:21-6. Chareonthaitawee P, Christian TF, Hirose K, Gibbons RJ, Rumberger
Christian et al.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29.
30.
31.
32.
JA. Relation of initial infarct size to extent of left ventricular remodeling in the year after acute myocardial infarction. J Am Coll Cardiol 1995; 25:567-73. Christian TF, O’Connor MK, Hopfenspirger MR, Gibbons RJ. Comparison of reinjection thallium-201 and resting Tc-99m-sestamibi tomographic images for the quantificaiton of infarct size following acute myocardial infarction. J Nuc Card 1994;1:17-28. Udelson JE, Coleman PS, Metherall J, et al. Predicting recovery of severe regional ventricular dysfunction; comparison of resting scintigraphy with T1-201 and Tc-99m sestamibi. Circulation 1994;89:2552-61. Chareonthaitawee P, Christian TF, Berger PB, Higano ST, Gibbons RJ. Noninvasive prediction of collateral flow: a multicenter validation study. Am Heart J 1997;134:639-46. Christian TF, O’Connor MK, Gibbons RF, Ritman EL. Collateral flow determination during coronary occlusion with Tc-99m sestamibi: validation in two closed chest animal models. J Nucl Med 1997. In press. O’Connor MK, Gibbons RJ, Juni JE, O’Keefe J, Ali A. Quantitative myocardial SPECT for infarct sizing: feasibility of a multi-center trial evaluated using a cardiac phantom. J Nucl Med 1995;36:1130-6. Callahan RJ, Froelich HW, McKusick KA, Leppo J, Strauss HW. A modified method for the in vivo labelling of red blood cells with Tc99m: Concise communication. J Nucl Med 1982;23:315-8. The Thrombolysis in Myocardial Infarction (TIMI) Trial. N Engl J Med 1985;312:932-6. The GUSTO Angiographic Investigators. The effects of tissue plasminogen activator, streptokinase, or both on coronary artery patency ventricular function, and survival after acute myocardial infarction. N Engl J Med 1993;329:1615-22. Vogt A, vonEssen R, Tebbe U, Feurer W, Appel KF, Neuhaus KL. Impact of early perfusion status of the infarct related artery on short term mortality after thrombolysis for acute myocardial infarction: retrospective analyses of four German multicenter studies. J Am Coll Cardiol 1993;21:1391-5. Kennedy JN, Kaiser GC, Fisher LD, et al. Clinical and angiographic predictors of operative mortality from the Collaborative Study in Coronary Artery Surgery (CASS). Circulation 1981;63:793-802. Kennedy JW, Kaiser GC, Fisher LD, et al. Multivariate discriminant analysis of the clinical and angiographic predictors of operative mortality from the Collaborative Study in Coronary Artery Surgery (CASS). J Thor Cardiovasc Surg 1980;80:876-87. Kimmel SE, Berlin JA, Laskey WK. The relationship between coronary angioplasty procedure volume and major complications. JAMA 1995; 274:1137-42. Gibbons RJ, Christian TF, Hopfenspirger M, Hodge DO, Bailey KR. Myocardium at risk and infarct size after thrombolytic therapy for acute myocardial infarction: implications for the design of randomized trials of acute intervention. J Am Coll Cardiol 1994; 24:616-23.
317
Background Direct coronary angioplasty is an effective therapy for acute myocardial infarction, but its success may be dependent on both ready availability and operator skill. The purpose of this study was to investigate the impact of the center performing direct coronary angioplasty for acute myocardial infarction while controlling for parameters known to affect outcome.
Methods and Results The study group consisted of 99 patients with ST elevation who were treated with direct angioplasty in four high-volume centers. Patients were injected with technetium-99m sestamibi intravenously and then taken to the cardiac catheterization laboratory. Antegrade flow was graded before and after direct coronary angioplasty. Single photon emission computed tomography was performed 1 to 6 hours after injection to measure myocardium at risk and residual blood flow to the jeopardized zone using previously published quantitative methods. A repeat sestamibi injection and tomographic acquisition were performed at hospital discharge to measure actual infarct size. There were no significant differences by center for baseline clinical characteristics, mean myocardium at risk (29% to 37% left ventricle [LV]), time to reperfusion (3.1 to 4.1 hours), residual blood flow, infarct location, or antegrade flow. Despite these similarities, there were differences in outcome measures by center. Mean infarct size was as follows: center 1, 15%; center 2, 12%; center 3, 10%, center 4, 23% (all LV; p = 0.11). Mean left ventricular ejection fraction at discharge also demonstrated significant differences: center 1, 0.57; center 2, 0.47; center 3, 0.53; center 4, 0.47 (p = 0.002). The prevalence of Thrombolysis in Myocardial Infarction grade 3 flow after angioplasty significantly differed by center: center 1, 92%; center 2, 94%; center 3, 87%; center 4, 71%; (p = 0.01). There was a low mortality rate for all four centers ranging from 0% to 6%. After adjustment for myocardium at risk, residual blood flow, and time to reperfusion, the primary outcome of the center where the angioplasty was performed was an independent determinant of both infarct size and left ventricular ejection fraction.
Conclusion The success of direct coronary angioplasty in reducing infarct size and preserving left ventricular function depends on the center performing the procedure. Direct measurement of the effectiveness of this reperfusion modality in community practice is required to assess the impact of this effect. (Am Heart J 1998;135:310-17.)
Direct coronary angioplasty is an effective method of reperfusion for acute myocardial infarction. Randomized clinical trials have shown that direct coronary angioplasty in selected centers is at least as effective as thrombolytic therapy if performed within 6 From the aDivision of Cardiovascular Diseases and Internal Medicine, Mayo Clinic and Mayo Foundation; the bDepartment of Cardiovascular Diseases, St. Luke’s Hospital, The Mid America Heart Institute, and the Department of Cardiovascular Diseases, William Beaumont Hospital; the cDepartment of Cardiovascular Diseases, Deaconess Medical Center, Spokane Heart Research Foundation; and the dDepartment of Cardiovascular Diseases, St. Francis Hospital, Tulsa Cardiology. Supported by a grant from Burroughs Wellcome Co., Research Triangle Park, N.C. Submitted Aug. 9, 1996; accepted July 24, 1997. Reprint requests: Timothy F. Christian, MD, Mayo Clinic, 200 First St. SW, West 16B, Rochester, MN 55905. E-mail: [email protected] Copyright © 1998 by Mosby, Inc. 0002/8703/98/$5.00 + 0 4/1/87274
hours of the onset of chest pain.1-3 Potential advantages of direct angioplasty include greater 90-minute patency rates and a more likely restoration of brisk antegrade flow compared with thrombolytic therapy.4 However, one potential disadvantage compared with thrombolytic therapy is the dependence on operator skill. The randomized clinical trials performed to date have been in selected centers highly proficient in coronary angioplasty.1-3 Intercenter variability of results has not been assessed. The purpose of this study was to investigate the variability in outcome among centers performing direct coronary angioplasty for acute myocardial infarction. The primary response variable was the impact of the center performing the procedure on infarct size after adjustment for other important variables known to
American Heart Journal Volume 135, Number 2, Part 1
Christian et al.
Table I. Clinical characteristics by center
No. patients Age (yr) Female Diabetes Vascular disease Smoking history Prior Current Hypercholesterolemia Hypertension Anterior location of MI Prior MI Prior CHF Culprit vessel = vein graft
Center 1
Center 2
Center 3
Center 4
p Value
37 66 ± 11 32% 27% 8%
33 59 ± 12 21% 21% 9%
15 62 ± 7 27% 0% 13%
14 59 ± 14 21% 14% 0%
— 0.09 >0.2 0.15 >0.2
27% 24% 43% 49% 30% 19% 0% 3%
36% 42% 42% 36% 24% 33% 9% 3%
27% 53% 40% 40% 27% 13% 0% 0%
29% 50% 36% 21% 36% 14% 0% 7%
0.14 >0.2 >0.2 >0.2 >0.2 0.10 >0.2
MI, Myocardial infarction; CHF, congestive heart failure.
influence infarct size—myocardium at risk, collateral flow, and time to coronary artery reperfusion.5,6 Such an approach is clinically feasible using sequential tomographic perfusion imaging with technetium-99mm sestamibi, which has been extensively used in prior studies of acute infarction.2,7-9
Methods The study group was selected from patients enrolled in a prospective randomized study of the use of poloxamer-188, a copolymer surfactant, as ancillary therapy with direct angioplasty for acute myocardial infarction (Table I). Patients were randomly assigned in a 2:1 ratio, drug-to-placebo, stratified by center. As previously reported, there was no demonstrable benefit of poloxamer-188 on myocardial salvage measured by acute and discharge 99mTc sestamibi imaging compared with placebo.10 Patients were eligible for this study if they met the following inclusion criteria: (1) enrollment in the randomized trial of poloxamer-188; (2) presentation within 6 hours of the onset of chest pain; (3) electrocardiographic ST-segment elevation of 2 mm in two contiguous anterior leads (V1 through V4) or 1 mm in two inferior leads (II, III, aVF). The following were reasons for exclusion from the randomized trial: (1) left bundle branch block on the resting electrocardiogram, (2) women of child-bearing potential without a negative serum pregnancy test, (3) prior use of thrombolytic therapy during the course of the myocardial infarction; (4) serum creatinine level ≥3 mg/dl. A total of 150 patients were enrolled in the randomized trial. In addition to the randomized trial criteria, the following exclusions were specific to this study: (1) incomplete acute 99mTc sestamibi imaging or clinical data (16 patients) and (2) performance of direct angioplasty at a center enrolling fewer than 10 patients into the trial (35 patients,
five centers). Radionuclide measurements for these 35 excluded patients are provided in footnotes to Tables II and III. Angiographic core laboratory analysis was not available for these patients; therefore, 99 patients formed the study group. The four centers that enrolled 10 patients or more meeting the inclusion criteria and none of the exclusion criteria included The Spokane Heart Research Institute and Deaconess Hospital, Spokane, Wash.; William Beaumont Hospital, Royal Oak, Mich.; Cardiology of Tulsa and St. Francis Hospital, Tulsa, Okla.; Mid America Heart Institute and St. Luke’s Hospital, Kansas City, Mo. All four centers meet the American College of Cardiology volume guidelines for centers performing direct angioplasty.11 The Mayo Clinic functioned as the core laboratory for the perfusion images but did not contribute any patients to the randomized trial. It should be noted that one center (center 1) had a competing randomized protocol that may have influenced the enrollment of patients into this trial.
Radionuclide data The methods of 99mTc sestamibi imaging in acute myocardial infarction have been previously described in detail.9,12 In brief, all patients were injected with 20 to 30 mCi of 99mTc sestamibi after giving informed consent for enrollment into the randomized trial of poloxamer-188 with direct coronary angioplasty. Tomographic imaging was performed 1 to 6 hours after injection depending on the duration of the coronary angioplasty procedure with standard tomographic imaging techniques but with different cameras and computer software at the individual centers.9 Because redistribution despite reperfusion therapy is minimal,13,14 images acquired after reperfusion still reflect myocardial blood flow before direct coronary angioplasty.14,15 This property has allowed the assessment of myocardium at risk without delaying reperfusion therapy. A second injection of 99mTc sestamibi and tomographic acquisition were performed 5 to 8 days later to assess final infarct size in 93 of 99 patients.
311
312
American Heart Journal February 1998
Christian et al.
Table II. Radionuclide and angiographic measurements
Myocardium at risk (% LV) Radionuclide residual flow (nadir) TIMI 0 or 1 flow before PTCA (% patients) Time to first balloon inflation (hr)
Center 1 (n = 37)
Center 2 (n = 33)
Center 3 (n = 15)
Center 4 (n = 14)
29 ± 19 0.26 ± 0.20 68% 3.9 ± 1.4
30 ± 20 0.25 ± 0.18 58% 4.1 ± 1.5
30 ± 19 0.31 ± 0.17 67% 3.4 ± 1.1
37 ± 25 0.24 ± 0.20 64% 3.1 ± 1.5
p Value >0.20 >0.20 >0.20 0.17
PTCA, Percutaneous transluminal coronary angioplasty. Combined data from the remaining five centers (35 patients) not qualifying for the study (enrollment of fewer than 10 patients meeting the inclusion criteria per center) were myocardium at risk 40% ± 22% LV and infarct size 16% ± 17% LV. LVEF was 0.48 ± 0.13.
Table III. Univariate outcome measures
Infarct size (% LV) Mean ± SD Range LVEF Mean ± SD Range TIMI 3 flow after PTCA
Center 1
Center 2
Center 3
Center 4
p Value
15% ± 16% 0%-62%
12% ± 13% 0%-51%
10% ± 10% 0%-27%
23% ± 23% 0%-67%
0.11
0.56 ± 0.11 0.37-0.78 92%
0.47 ± 0.08 0.27-0.64 94%
0.53 ± 0.11 0.37-0.73 87%
0.47 ± 0.17 0.22-0.79 71%
0.002 0.01
Combined data from the remaining five centers (35 patients) not qualifying for the study (enrollment of fewer than 10 patients meeting the inclusion criteria per center) were myocardium at risk 40% ± 22% LV and infarct size 16% ± 17% LV. LVEF was 0.48 ± 0.13.
The left ventricular perfusion defect size was quantified at the core laboratory (Mayo Clinic) with a previously described method that uses a threshold of 60% of maximal counts.9,12,16 This threshold has been validated in phantom studies,16 has provided close correlation with other surrogate measures of infarct size,7,17-20 and has been independently shown to provide the best separation of myocardial segments with improved contractility after revascularization from those unchanged after revascularization.21 The acute perfusion defect reflects myocardium at risk and the final defect reflects infarct size. A noninvasive measure of residual blood flow to the infarct zone was obtained for each patient from the acute sestamibi tomographic defect.9 This measure has previously been shown to be highly associated with acute angiographic collaterals in both single9 and multicenter22 studies of myocardial infarction and has been shown to be an independent predictor of infarct size in addition to myocardium at risk and duration of coronary occlusion. It has also been shown to correlate significantly with collateral blood flow measured by radio-labeled microspheres in an animal study of coronary occlusion.23 This measure, termed the nadir, is calculated as the lowest ratio of minimum to maximum counts per short-axis slice from the acute circumferential count profile curves (Fig. 1). All four centers were validated for quality control and measurement reproducibility by using a cardiac phantom. Defects of known size were inserted into the walls of the phantom
left ventricle (LV) and imaged. Measured defect size was compared with actual defect size over a range of defect sizes. Fig. 2 shows the results for the four centers in this study. These results are part of a larger study by O’Connor et al.24 analyzing all 14 centers involved in the evaluation of poloxamer-188.
Radionuclide ventriculography Left ventricular ejection fraction (LVEF) was determined in 96 of the 99 patients in the study group with gated equilibrium radionuclide ventriculography. Red cells were labeled with 99mTc pyrophosphate with a modified in vivo method,25 and LVEF was determined with commercial software in clinical use at each center. No central laboratory review or processing was performed. These studies were acquired 1 day after the second sestamibi study at discharge. Both infarct size and ejection fraction measures were available for 90 of 99 patients.
Coronary angiography All patients underwent acute coronary angiography as part of a treatment strategy for direct coronary angioplasty.26 Antegrade flow was scored with the Thrombolysis in Myocardial Infarction (TIMI) grading system by a single blinded observer (P.B.B.) not involved with any of the centers enrolling patients before or after intervention. TIMI 0 and 1 grades were combined because their outcomes have been shown to be nearly identical.27
American Heart Journal Volume 135, Number 2, Part 1
Figure 1
Christian et al.
Figure 2
Noninvasive measurement of collateral flow and determination of defect extent. Circumferential count profile curve from shortaxis slice demonstrating acute inferior perfusion defect. Extent of defect is defined by number of pixels with <60% of maximal counts. Residual flow (collateral and antegrade) to jeopardized zone is estimated by severity of perfusion defect. Severity (nadir) is determined by lowest ratio of minimum to maximum counts (B/A).
True versus tomographic defect size from cardiac phantom for four participating centers. Defects of known size are placed within walls of phantom and imaged at individual centers. There is very close correlation for each center despite different cameras and computer systems.
Statistics
Baseline radionuclide and angiographic measures
Data are presented as mean ± SD. Analysis of variance was used to compare continuous variables by angioplasty center. Noncontinuous variables were compared with chi-square contingency table analysis or Fischer’s exact test. Multivariable (analysis of covariance) analysis was used to determine whether the primary response variable of the center performing the procedure was an independent determinant of infarct size after adjustment for other variables known to affect the dependent variable. A similar (secondary) multivariable analysis was performed with LVEF as the dependent variable. A separate univariate analysis of infarct size and LVEF by center was performed with imputed values for the four patients who died before acquisition of these measures. The worst value reported among the four centers was assigned to these patients. All analyses were performed with Statview 4.1 (Abacus software) or Supernova 1.0 (Abacus software) commercially available statistical software programs.
The mean value for myocardium at risk for the overall study group was 30% ± 20% of the LV and was not significantly different among centers (Table II). The mean nadir for the study group was 0.26 ± 0.19, indicating that the average severity of the acute perfusion defect was 26% of maximal counts in the most severely hypoperfused short-axis tomographic slice. This acute measure of residual blood flow was not different among the four centers. The degree of persistent antegrade flow assessed by coronary angiography before direct coronary angioplasty was similar; TIMI grade 0 or 1 flow was present in 68%, 58%, 67%, and 64% of patients in centers 1 through 4, respectively.
Results Baseline clinical variables Table I shows clinical characteristics of the study group by center. There were no significant differences in these variables. Table II describes elapsed time from chest pain onset to first balloon inflation. The mean times, which ranged from 3.1 (center 4) to 4.1 hours (center 2), were not significantly different among centers.
Univariate radionuclide and angiographic outcome measures Although there were no clinical, radionuclide, or angiographic differences among centers, there were differences in outcome variables by center (Figs. 3 through 5, Table II). There was a trend for infarct size to differ among centers (Fig. 3, Table III). Centers 1 through 3 had mean infarct size measures ranging from 10% ± 10% LV to 15% ± 16% LV, and center 4 had a mean infarct size of 23% ± 23% LV. By analysis of variance, this trend did not reach statistical significance (p = 0.11). LVEF was significantly different by center (Fig. 4, Table III), with mean values ranging from 0.47 ± 0.14 (center 4) to 0.56 ± 0.10 (center 1) (p = 0.002).
313
314
American Heart Journal February 1998
Christian et al.
Figure 3
Mean and 95% confidence intervals of end point of infarct size by discharge 99mTc sestamibi perfusion imaging is presented by center as percent of LV. There is a trend toward difference in outcome by center that does not reach statistical significance (p = 0.11).
Figure 4
Figure 5
Antegrade flow after direct coronary angioplasty graded with TIMI criteria. There is significant difference by center following same pattern as infarct size.
Mortality rate There were four patients who died during the course of hospitalization. These patients did not have infarct size or LVEF measures and could not be included in the above analysis. Mortality rate as a percent of total patients enrolled (whether or not they were excluded from this substudy) by center was center 1, 2%; center 2, 5%; center 3, 0%; and center 4, 6%. Thus mortality rate was not significantly different by center. Including these patients in the univariate analysis by imputing the worst value for infarct size (67% LV) and LVEF (0.22) did not change the univariate analysis for these measures: infarct size p = 0.11 and LVEF p = 0.002.
Mean and 95% confidence intervals of LVEF determined by radionuclide ventriculography by center. This end point was not analyzed in a core laboratory. There is significant difference in outcome by center for this variable (p = 0.002).
There were 11 patients with TIMI flow <3 after angioplasty (10%), four with TIMI 0 or 1 flow, and seven with TIMI 2 flow. Fig. 5 shows the distribution of TIMI flow after angioplasty. Restoration of TIMI 3 flow after direct coronary angioplasty (90% of patients) ranged from 87% to 94% of patients for centers 1 through 3 but was only 71% for center 4 (Fig. 5). The distribution of TIMI flow after angioplasty was significantly different by center (p = 0.01).
Multivariable analysis of outcome Although there were no baseline differences for clinical, radionuclide, or angiographic variables by center on univariate analysis, it is possible that trends existed unfavorable to a center that, taken cumulatively, could profoundly affect the outcome measures. Because the elapsed time to reperfusion, magnitude of residual (collateral plus antegrade) flow, and extent of myocardium are known to be independently at risk associated with infarct size in models of reperfusion,5,6 these variables were forced into the model. The primary response variable of the center performing the procedure and the clinical and angiographic variables from Tables I and II were then tested for an independently significant asso-
American Heart Journal Volume 135, Number 2, Part 1
ciation with infarct size. Only the center where the procedure was performed provided additional independent information (Table IV). A similar significant association with outcome was seen in this multivariable design when LVEF was the dependent variable (Table IV).
Discussion Myocardium at risk and collateral flow are important determinants of infarct size.5 Prior studies have demonstrated that failure to account for all these variables can lead to erroneous conclusions regarding treatment efficacy.5 Restoration of flow in animal studies entails a simple release of a mechanical constrictor and does not account for the variability of success in restoring flow in clinical practice. Although direct coronary angioplasty is highly effective in restoring antegrade flow in acute myocardial infarction, there is much more potential for variability than in animal models of reperfusion. This study examined whether the variability inherent in direct coronary angioplasty at different centers had significant impact on final infarct size after accounting for the three other measurable variables (myocardium at risk, collateral flow, and time to reperfusion) known to be important. Overall, the results were impressive, with an overall 3% mortality rate. Myocardium at risk was 30% of the LV with a final infarct size of 14% of the LV, representing salvage of more than half the myocardium at risk on average. Mean ejection fraction was normal. All three measures of outcome compare very favorably to previously reported values of other trials.1-3 However, there were differences in outcome by center. Despite similar measures of myocardium at risk, collateral flow, and presumed duration of coronary occlusion, patients undergoing direct angioplasty at center 4 tended to have a poorer outcome as defined by this study. The differences in LVEF and antegrade flow after angioplasty were statistically significant. After controlling for factors known to determine infarct size (myocardium at risk, residual blood flow, time to reperfusion), the center where the procedure was performed was an independently significant factor in determining infarct size. The lower prevalence of TIMI 3 flow for this center after angioplasty suggests a possible mechanism for this difference. The impact on outcome of antegrade flow after reperfusion has been described in two large studies of myocardial infarction. Vogt et al.28 demonstrated that mortality rate was nearly three times greater in patients with TIMI 2 compared with patients with TIMI 3 flow and was similar to patients with TIMI 0 or 1 flow. This
Christian et al.
Table IV. Impact of the center performing PTCA on infarct size and LVEF after controlling for known determinants of outcome Variables Infarct size (n = 93) Myocardium at risk Residual blood flow Time to first inflation Center performing PTCA LVEF (n = 96) Myocardium at risk Residual blood flow Time to first inflation Center performing PTCA
p Value
0.02 0.0007 0.17 0.01 0.02 0.06 >0.20 0.0006
PTCA, Percutaneous transluminal coronary angioplasty.
finding may be caused by differences in resultant ventricular function after thrombolysis. The GUSTO investigators found that LVEF was significantly greater and regional wall motion abnormalities less extensive in patients with TIMI 3 flow compared with those with TIMI ≤2 flow.27 The results of this study, in which the lower the prevalence of TIMI 3 flow occurred in the center with the largest mean infarct size, suggest that these differences in mortality rate and left ventricular function may be caused by differences in myocardial salvage by antegrade flow after angioplasty. The differences in left ventricular systolic function were consistent with the measures of infarct size with the exception of center 2, where the mean LVEF was lower than centers 1 and 3 despite nearly identical values for infarct size. Fig. 4 demonstrates a narrow dispersion of values for center 2; 90% of patients had an ejection fraction <0.55 despite 25% of patients having normal perfusion images. This finding likely represents a lower range of normal values of this center and underscores the need for core laboratories to process and measure end points, which was not the case for LVEF in this trial. The range in values for center 4 was much broader. The discrepant outcome by center is in agreement with prior reports analyzing surgical outcome in the Coronary Artery Surgery Study.29,30 Mortality rate by site in 15 centers was analyzed and found to vary from 0.3% to 6.4%; a more than 20-fold range.29 The wide variability in outcome by center was preserved when baseline risk variables were adjusted.30 Variability by center in major complication rates has been described for elective coronary angioplasty as a function of patient volume.31 Consequently, it is not unexpected that there was vari-
315
316
American Heart Journal February 1998
Christian et al.
ability in outcome in this study of a therapy that is technically demanding.
Limitations The small number of patients available for analysis in two of the centers is a limitation of this study. There were consistent trends in the same direction for the three major outcome variables of this study: infarct size, ejection fraction, and antegrade flow after angioplasty, with only the latter two achieving statistical significance. Consequently, a type II error may be present for the univariate analyses. However, the multivariate analysis results are compelling. Had larger numbers of patients been enrolled, the trend in infarct size may have become significant by univariate analysis. These findings do suggest, however, that intercenter differences in outcome for direct angioplasty potentially exist. It should be noted that the use of acute imaging to define myocardium at risk adds considerable statistical power when analyzing outcomes in acute myocardial infarction treated with reperfusion therapy. Approximately 70% of the variability in outcome in infarct size can be accounted for by this measure alone.5 We have previously demonstrated that incorporation of myocardium at risk, collateral flow, and time to reperfusion accounts for 69% of the variability in infarct size in the clinical setting.9 Consequently, fewer patients are required to detect a significant impact on infarct size of a particular variable, such as the center where the reperfusion strategy occurs.32 This factor is in sharp distinction to trials that cannot account for baseline differences in these key variables but rely on mortality and reinfarction rates. Gibbons et al.2 demonstrated no significant difference in myocardial salvage between two reperfusion strategies (thrombolysis and direct angioplasty) with adequate statistical power with fewer than 50 patients per treatment arm. It should also be noted that the number of patients in two of the centers enrolling fewer patients was similar to the mean number of patients per center (n = 16) in a large published trial of direct angioplasty.1 Consequently, the limited number of patients in two of the centers in this study should be viewed within the context of the power of the technique. One center (center 1) had competing trials of direct angioplasty and thrombolytic therapy that may have biased enrollment at that center toward a sicker population in this trial. Despite this bias, the outcome was exceptionally good. None of the other three centers had competing trials. This study was conducted in four
tertiary referral centers with high patient volumes. How the inclusion of community-based centers would have impacted on the variability is unknown, but this issue clearly is key in examining the national use of direct coronary angioplasty as routine therapy for acute myocardial infarction. This study only examined short-term outcome variables. Consequently no conclusions regarding long-term outcome can be made from this study. The findings of this study imply that there may be a difference between the efficacy of direct coronary angioplasty and its effectiveness when applied on a broad scale. The efficacy (the performance of a test procedure under ideal conditions) of direct coronary angioplasty has been demonstrated in previous studies.1-3 The effectiveness (the use of a test or procedure for the individual patient in a routine setting) of direct coronary angioplasty remains to be determined. Our results should not be interpreted as evidence that direct coronary angioplasty is useful only in certain institutions. Rather, the findings presented here should be considered as suggestive evidence that significant differences in outcome may exist with this technique of reperfusion. Although mean values for outcome were excellent in the largest multicenter trial of direct coronary angioplasty,1 it is logical to presume that there were consistent outliers in both directions. This intercenter variability in outcome should be considered when comparing direct angioplasty to less skill-dependent modalities of reperfusion, such as intravenous thrombolysis.
References 1. Grines CL, Browne KF, Marco J, et al. A comparison of immediate angioplasty with thrombolytic therapy for acute myocardial infarction. N Engl J Med 1993;328:673-9. 2. Gibbons RJ, Verani MS, Behrenbeck T, et al. Feasibility of tomographic Tc-99m-hexakis-2-methoxy-2-methylpropyl-isonitrile imaging for the assessment of myocardial area at risk and the effect of acute treatment in myocardial infarction. Circulation 1989;80:1277-86. 3. Zijlstra F, deBoer MJ, Hoorntje JCA, Reiffers S, Reiber JHC, Suryapranata H. A comparison of immediate coronary angioplasty with intravenous streptokinase in acute myocardial infarction. N Engl J Med 1993;328:680-4. 4. Berger PB, Bell MR, Holmes DR, Gersh BJ, Hopfinspirger MR, Gibbons RJ. Time to reperfusion with direct coronary angioplasty and thrombolytic therapy in acute myocardial infarction. Am J Cardiol 1994;73:231-6. 5. Reimer KA, Jennings RB, Cobb FR, et al. Animal models for protecting ischemic myocardium: results of the NHLBI cooperative study. Comparison of unconscious and conscious dog models. Circ Res 1985;56:651-65. 6. Reimer KA, Jennings RB. The wavefront phenomenon of myocardial ischemic cell death. II. Transmural progression of necrosis within the
American Heart Journal Volume 135, Number 2, Part 1
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
framework of ischemic bed size (myocardium at risk) and collateral flow. Lab Invest 1979;40:633-44. Behrenbeck T, Pellikka PA, Huber KC, Bresnahan JF, Gersh BJ, Gibbons RJ. Primary angioplasty in myocardial infarction: assessment of improved myocardial perfusion with technetium-99m-isonitrile. J Am Coll Cardiol 1991;17:365-72. Pellikka PA, Behrenbeck T, Verani MS, Mahmarian JJ, Wackers FJTh, Gibbons RJ. Serial changes in myocardial perfusion using tomographic technetium-99m-hexakis-2-methoxy-2-methylpropyl-isonitrile imaging following reperfusion therapy of myocardial infarction. J Nucl Med 1990;31:1269-75. Christian TF, Schwartz RS, Gibbons RJ. Determinants of infarct size in reperfusion therapy for acute myocardial infarction. Circulation 1992; 86:81-90. Grines CL, O’Keefe JH, DeWood MA, et al. A prospective, randomized, placebo-controlled trial of RheothRx (Poloxamer 188, NF) to limit infarct size in humans [abstract]. Circulation 1993;80:I-292. Ryan TJ, Klocke FJ, Reynolds WA, ACP/ACC/AHA Task Force on Clinical Privileges in Cardiology. Clinical competence in percutaneous transluminal coronary angioplasty. Circulation 1990; 81:2041-6. Christian TF, Gibbons RJ, Gersh BJ, Gibbons RJ. Effect of infarct location on myocardial salvage assessed by technetium-99m-isonitrile. J Am Coll Cardiol 1991;17:1303-8. Okada RD, Glover D, Gaffney T, William S. Myocardial kinetics of technetium-99m-hexakis-2-methoxy-2-isonitrile. Circulation 1988; 77:491-8. DeCoster PM, Wijns W, Cauwe F, Robert A, Beckers C, Melin JA. Area at risk determination by technetium-99m-hexakis-2-methoxyisobutyl-isonitrile in experimental reperfused myocardial infarction. Circulation 1990;82:2151-62. Sinusas AJ, Trautman KA, Bergin JD, et al. Quantification of “area at risk” during coronary occlusion and degree of myocardial salvage after reperfusion with technetium-99m-methoxyisobutyl-isonitrile. Circulation 1990;82:1424-37. O’Connor MK, Hammell TC, Gibbons RJ. In vitro validation of a simple tomographic technique for estimation of percent myocardium “at risk” following administration of Tc99m isonitrile. Eur J Nucl Med 1990;17:69-76. Christian TF, Behrenbeck T, Pellikka, PA, Huber KC, Chesebro JH, Gibbons RJ. Mismatch of left ventricular function and infarct size demonstrated by technetium-99m-isonitrile imaging after reperfusion therapy for acute myocardial infarction: identification of myocardial stunning and hyperkinesia. J Am Coll Cardiol 1990;16:1632-8. Christian TF, Behrenbeck T, Gersh BJ, Gibbons RJ. Relation of left ventricular volume and function over one year after acute myocardial infarction to infarct size determined by technetium-99m sestamibi. Am J Cardiol 1991;68:21-6. Chareonthaitawee P, Christian TF, Hirose K, Gibbons RJ, Rumberger
Christian et al.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29.
30.
31.
32.
JA. Relation of initial infarct size to extent of left ventricular remodeling in the year after acute myocardial infarction. J Am Coll Cardiol 1995; 25:567-73. Christian TF, O’Connor MK, Hopfenspirger MR, Gibbons RJ. Comparison of reinjection thallium-201 and resting Tc-99m-sestamibi tomographic images for the quantificaiton of infarct size following acute myocardial infarction. J Nuc Card 1994;1:17-28. Udelson JE, Coleman PS, Metherall J, et al. Predicting recovery of severe regional ventricular dysfunction; comparison of resting scintigraphy with T1-201 and Tc-99m sestamibi. Circulation 1994;89:2552-61. Chareonthaitawee P, Christian TF, Berger PB, Higano ST, Gibbons RJ. Noninvasive prediction of collateral flow: a multicenter validation study. Am Heart J 1997;134:639-46. Christian TF, O’Connor MK, Gibbons RF, Ritman EL. Collateral flow determination during coronary occlusion with Tc-99m sestamibi: validation in two closed chest animal models. J Nucl Med 1997. In press. O’Connor MK, Gibbons RJ, Juni JE, O’Keefe J, Ali A. Quantitative myocardial SPECT for infarct sizing: feasibility of a multi-center trial evaluated using a cardiac phantom. J Nucl Med 1995;36:1130-6. Callahan RJ, Froelich HW, McKusick KA, Leppo J, Strauss HW. A modified method for the in vivo labelling of red blood cells with Tc99m: Concise communication. J Nucl Med 1982;23:315-8. The Thrombolysis in Myocardial Infarction (TIMI) Trial. N Engl J Med 1985;312:932-6. The GUSTO Angiographic Investigators. The effects of tissue plasminogen activator, streptokinase, or both on coronary artery patency ventricular function, and survival after acute myocardial infarction. N Engl J Med 1993;329:1615-22. Vogt A, vonEssen R, Tebbe U, Feurer W, Appel KF, Neuhaus KL. Impact of early perfusion status of the infarct related artery on short term mortality after thrombolysis for acute myocardial infarction: retrospective analyses of four German multicenter studies. J Am Coll Cardiol 1993;21:1391-5. Kennedy JN, Kaiser GC, Fisher LD, et al. Clinical and angiographic predictors of operative mortality from the Collaborative Study in Coronary Artery Surgery (CASS). Circulation 1981;63:793-802. Kennedy JW, Kaiser GC, Fisher LD, et al. Multivariate discriminant analysis of the clinical and angiographic predictors of operative mortality from the Collaborative Study in Coronary Artery Surgery (CASS). J Thor Cardiovasc Surg 1980;80:876-87. Kimmel SE, Berlin JA, Laskey WK. The relationship between coronary angioplasty procedure volume and major complications. JAMA 1995; 274:1137-42. Gibbons RJ, Christian TF, Hopfenspirger M, Hodge DO, Bailey KR. Myocardium at risk and infarct size after thrombolytic therapy for acute myocardial infarction: implications for the design of randomized trials of acute intervention. J Am Coll Cardiol 1994; 24:616-23.
317