Author + information
- Received March 30, 2006
- Revision received July 5, 2006
- Accepted July 6, 2006
- Published online December 19, 2006.
- John J. Mahmarian, MD, FACC⁎,1,⁎ (, )
- Leslee J. Shaw, PhD†,2,
- Neil G. Filipchuk, MD‡,3,
- Habib A. Dakik, MD§,
- Sherif S. Iskander, MD∥,4,
- Terrence D. Ruddy, MD¶,
- Milena J. Henzlova, MD#,4,
- Felix Keng, MD⁎⁎,
- Adel Allam, MD††,
- Lemuel A. Moyé, MD, PhD‡‡,
- Craig M. Pratt, MD, FACC⁎,
- INSPIRE Investigators
- ↵⁎Reprint requests and correspondence:
Dr. John J. Mahmarian, Department of Cardiology, Methodist DeBakey Heart Center, 6550 Fannin Street, SM-677, Houston, Texas 77030.
Objectives The purpose of this study was to determine whether gated adenosine Tc-99m sestamibi single-photon emission computed tomography (ADSPECT) could accurately define risk and thereby guide therapeutic decision making in stable survivors of acute myocardial infarction (AMI).
Background Controversy continues as to the role of noninvasive stress imaging in stratifying risk early after AMI.
Methods The INSPIRE (Adenosine Sestamibi Post-Infarction Evaluation) trial is a prospective multicenter trial which enrolled 728 clinically stable survivors of AMI who had gated ADSPECT within 10 days of hospital admission and subsequent 1-year follow-up. Event rates were assessed within prospectively defined INSPIRE risk groups based on the adenosine-induced left ventricular perfusion defect size, extent of ischemia, and ejection fraction.
Results Total cardiac events/death and reinfarction significantly increased within each INSPIRE risk group from low (5.4%, 1.8%), to intermediate (14%, 9.2%), to high (18.6%, 11.6%) (p < 0.01). Event rates at 1 year were lowest in patients with the smallest perfusion defects but progressively increased when defect size exceeded 20% (p < 0.0001). The perfusion results significantly improved risk stratification beyond that provided by clinical and ejection fraction variables. The low-risk INSPIRE group, comprising one-third of all enrolled patients, had a shorter hospital stay with lower associated costs compared with the higher-risk groups (p < 0.001).
Conclusions Gated ADSPECT performed early after AMI can accurately identify a sizeable low-risk group who have a <2% death and reinfarction rate at 1 year. Identifying these low-risk patients for early hospital discharge may improve utilization of health care resources at considerable cost savings.
There is no consensus as to the optimal strategy for evaluating risk in stable survivors of acute myocardial infarction (AMI). Ideally, this would be a noninvasive method which could be performed safely early after AMI so as to accurately identify low-risk patients and thereby facilitate early hospital discharge. Recent studies in acute coronary syndromes (ACS) propose that a routine invasive approach improves outcome over a conservative strategy of exercise treadmill testing (ETT) followed by selective coronary angiography in patients who demonstrate significant myocardial ischemia (1,2). However, preliminary data in patients after AMI suggest that myocardial perfusion imaging, when combined with a pharmacologic vasodilator such as adenosine, is an attractive noninvasive method for safely and reliably stratifying risk early after admission (3–5). The INSPIRE (Adenosine Sestamibi Post-Infarction Evaluation) is a prospective multinational clinical trial designed to establish the role of quantitative adenosine technetium-99m sestamibi single-photon emission computed tomography (ADSPECT) as the initial primary risk assessment tool for guiding therapeutic decision making in stable survivors of AMI.
The study cohort consisted of 728 stabilized patients ≥18 years of age who had either Q- or non–Q-wave AMI and were prospectively enrolled at 16 sites over 36 months (December 1999 through December 2002) with a subsequent 1-year follow-up (6). Study entry criteria are shown in Table 1.
The research protocol was approved by each site’s review board, and all patients gave written informed consent. The INSPIRE study design is published (6). The ADSPECT was performed within 10 days of hospital admission using a standard nitrate-enhanced rest-adenosine stress protocol. The nitrate-enhanced rest study was done to maximize detection of ischemia (7). Imaging was performed significantly earlier at U.S. than at non-U.S. sites and irrespective of initial clinical risk defined by the Thrombolysis In Myocardial Infarction (TIMI) risk score (Fig. 1).
Image data were sent electronically to the Nuclear Core Laboratory the day of acquisition for quantification of total and ischemic perfusion defect size (PDS) so as to direct patient management (6). The core laboratory also generated standard semiquantitative summed stress score (SSS), summed rest score (SRS), and summed difference score (SDS) (8). A strong correlation was seen between SSS and total PDS (r = 0.92; p < 0.001; PDS = 3.34 × SSS) and between SDS and ischemic PDS (r = 0.91; p < 0.0001; ischemic PDS = 3.54 × SDS).
INSPIRE risk groups
Patient risk and subsequent therapeutic decision making were prospectively defined by specific ADSPECT variables (4,5) (Fig. 2).Patients with a small (<20%) PDS were classified as low risk and most had a left ventricular ejection fraction (LVEF) of ≥35% (96%) and an ischemic PDS of <10% (97%). This group was discouraged from having coronary angiography unless they developed clinical instability (6). Infarct-free survival at 1 year was anticipated to be >95%, so ADSPECT results were used to expedite hospital discharge (4,5).
Patients with large (≥20%) but primarily nonischemic (<10%) defects were defined as intermediate risk and had catheterization and coronary revascularization (CR) at investigator discretion (5,6). It was anticipated that most would not have CR, because CR does not improve survival over medical therapy in patients without ischemic (9) or viable (10) myocardium.
Patients with large total (≥20%) and ischemic (≥10%) defects and an LVEF of <35% were defined as high risk and encouraged to have coronary angiography, whereas those with an LVEF ≥35% were randomized to receive either intensive medical therapy or CR (6). The results from the randomized arm of the INSPIRE study are reported elsewhere (11). Additional therapeutic considerations are listed in Table 2.
Prospectively defined events were cardiac death, reinfarction, and readmission for ACS or severe congestive heart failure (6). Procedure-related AMIs were identified based on standard clinical criteria rather than by routine surveillance with cardiac enzymes. Events were adjudicated by a blinded end point committee. The Data and Safety Monitoring Board met at prospectively defined intervals to monitor study conduct and event rates within the assigned groups.
The general statistical and cost analysis plans are published (6). Baseline clinical and ADSPECT differences between INSPIRE risk groups were determined using nonparametric methods with chi-squared analysis used to compare all discrete variables except TIMI risk score, where a Spearman correlation was used. Comparisons of continuous measures across INSPIRE risk groups were performed using analysis of variance techniques. Baseline clinical and scintigraphic data are presented as frequency (%) or mean ± standard deviation.
The primary end point was time to first cardiac event with a secondary end point of time to cardiac death/reinfarction (6). The trial had >80% power with a 2-sided type I error of 0.05 to detect differences between INSPIRE risk groups. Risk-adjusted multiple variable Cox proportional hazards modeling estimated the effect of baseline variables on primary and secondary end points. In each case, the proportional hazards assumption was met. The TIMI risk score for ST-segment elevation AMI was selected for comparison with ADSPECT, because it is a well-accepted clinical instrument used to predict outcome in patients who receive thrombolytics (12) and in more heterogeneous AMI populations (13). The TIMI risk score for ACS was not used, because it was derived primarily from patients without AMI (14).
Survival curves compared differences in overall and infarct-free survival among TIMI and INSPIRE risk groups. Other baseline variables assessed by univariate analysis are those in Table 3(6). Survival curves were also generated based on total and ischemic LVPDS, LVEF, and LV volumes. Unadjusted relative risk ratios were calculated for scintigraphic variables followed by risk-adjusted estimates of cardiac events controlling for the TIMI risk score and LVEF. A final multivariable model included PDS, LVEF, and TIMI risk score. Chi-squared statistic compared a TIMI risk model to one of TIMI risk and LVEF; TIMI risk, LVEF and total PDS; and TIMI risk and all 3 variables.
From the Cox model, predicted 1-year event rates were calculated and compared using analysis of variance (ANOVA) or general linear modeling (GLM). Total and ischemic PDS and LVEF were plotted in relation to predicted event rates. From a Cox model including total and ischemic PDS, event rates from 10% to 75% were calculated for specific total and ischemic PDS values. A final analysis compared predicted event rates from low to high INSPIRE risk groups based on gender, age, and infarct location.
Six U.S. sites participated in the cost substudy, and each patient allowed access to their hospital bills. Blinded summary charge reports for the index hospitalization were obtained from each hospital’s finance office and sent to the coordinating center. Total charges were adjusted, using a top-down approach, by multiplying total hospital charges by the hospital-specific cost-charge ratio. This estimated hospital cost was then discounted at a 5% annual rate and inflation-corrected by the U.S. medical service sector estimate of the consumer price index. Cost estimates were available for 205 INSPIRE patients and compared across INSPIRE risk groups using ANOVA or GLM. Nonparametric statistics were used owing to the nonlinearity of cost data, but in every case this analysis was similar to the ANOVA or GLM results.
Total hospital length of stay and time spent in intensive care units were available for all patients. These surrogates for resource intensity and infarct severity were compared across INSPIRE risk groups using GLM techniques adjusting for baseline TIMI risk score.
Baseline patient characteristics and differences across INSPIRE risk groups
Mean patient age for the entire INSPIRE cohort was 63 ± 12 years, 31% were women, and 17% had prior AMI (Table 3). The AMI location was anterior in 32% of patients, and 48% developed Q waves. Mean LVPDS was 26% (with 10% ischemia), and mean LVEF 46 ± 14%. Ischemic PDS was comparably small in low (2.8 ± 3%) and intermediate (3.7 ± 3%) INSPIRE risk groups, but significantly increased in high-risk (23.7 ± 12%) and randomized (21.3 ± 9%) populations (Table 3). Patients were evenly distributed into low (36% vs. 32%), intermediate (28% vs. 30%), high (10% vs. 9%), and randomized (26% vs. 29%) INSPIRE risk groups across U.S. and non-U.S. sites, respectively (p = NS).
Over a mean follow-up of 11 ± 3 months, 95 events occurred in 88 patients (12.1%) as follows: cardiac death, sudden (n = 10) or nonsudden (n = 8); nonfatal reinfarction (n = 34); and readmission for ACS (n = 24) or congestive heart failure (n = 12). Patient follow-up was complete at 1 year in 98%. No adenosine- or other procedure-related events occurred.
Coronary angiography, CR, and events
Few low-risk patients had initial or late invasive procedures (Table 4).In the intermediate-risk group, coronary angiography and CR were performed in 31% and 17% of patients by 1 month, which minimally increased by 1 year. Most high-risk patients had coronary angiography and subsequent CR, generally early after AMI. Coronary revascularization significantly lowered event rates in only the high-risk group (Table 4).
Outcome based on TIMI risk score and clinical variables
Cardiac event rates significantly increased with increasing TIMI risk score from low (8%) to intermediate (12.5%) to high (22%) (p = 0.042). A similar trend was observed when analysis was limited to cardiac death/reinfarction (p = 0.07). Univariate risk predictors for death/reinfarction were age (relative risk [RR] 1.03/year, 95% confidence interval [CI] 1.01 to 1.06; p = 0.01) and peak troponin I (RR 1.01/U increase, 95% CI 1.0 to 1.01; p < 0.0001), and for all events they were diabetes mellitus (RR 1.77, 95% CI 1.17 to 2.68; p = 0.006) and peak troponin I (RR = 1.01; 1.0 to 1.01/U increase; p = 0.009).
Outcome based on INSPIRE risk group
Overall cardiac (p = 0.007) (Fig. 3A)and death/reinfarction (p = 0.009) (Fig. 3B) event rates significantly increased across INSPIRE risk groups from low (5.4% and 1.8%) to intermediate (14.0% and 9.2%) to high (18.6 and 11.6%), respectively. These differences were observed despite a high initial rate of CR (50%) in the high-risk group. After adjustment for the TIMI risk score, INSPIRE risk categories still predicted total (p = 0.01) and death/reinfarction (p = 0.02) events.
Outcome based on total and ischemic PDS
Total and ischemic LV PDS predicted patient outcome both before (Table 5A).and after (Table 5B) adjustment for TIMI risk score and LVEF (Table 5C). Each 10% absolute increment in total or ischemic PDS increased the unadjusted relative risk for any event by 37% and 64%, respectively. The extent of scintigraphic scar also predicted outcome (Tables 5A and 5B), but not after adjustment for LVEF (Table 5C). The complementary prognostic information obtained from total and ischemic PDS are depicted in Figure 4.
Outcome based on LVEF and cardiac volumes
Relative risk for any event or death/reinfarction significantly increased as LVEF decreased (Table 5A) and irrespective of TIMI risk score (Table 5B). An LVEF ≥50% predicted a low 1-year event rate (5%), which increased to 27% in the 3% of patients with an LVEF <20%.
Left ventricular end-diastolic and end-systolic volumes also predicted events (Tables 5A and 5B) but not after adjustment for LVEF (Table 5C). This is explained by the strong inverse relationship between LVEF and end-diastolic (r = 0.76; p < 0.001) and end-systolic (r = 0.91; p < 0.001) volumes.
Incremental prognostic value of ADSPECT variables beyond TIMI risk score
The INSPIRE risk assessment significantly increased the baseline TIMI risk score global chi-square statistic from 4.7 to 16.5 (total events) and from 4.2 to 15.7 (death/reinfarction) (both p < 0.001).
The addition of total and ischemic PDS to the combined TIMI risk score and LVEF significantly improved the model for predicting all events, but only total PDS added to the risk model for predicting death/reinfarction (Fig. 5).The global chi-square also significantly increased when PDS was added first to the TIMI risk score before LVEF (p < 0.001).
Multivariate risk predictors
By stepwise logistic regression, total PDS emerged as the most important independent risk predictor (p < 0.0001), followed by ischemic PDS (p < 0.005) and LVEF (p = 0.05) (Table 6).The only significant predictor of death/reinfarction was total PDS.
Outcome based on the INSPIRE trial risk among patient subgroups
The value of ADSPECT was assessed in several clinically relevant and prospectively defined patient subgroups (6) (Fig. 6).Irrespective of patient age, gender, or AMI location, the INSPIRE risk criteria stratified patients in a similar fashion as observed in the overall population. Event rates were consistently low in all patient subgroups that were identified as low risk by ADSPECT.
Cost analysis and length of hospital stay
Low-risk INSPIRE patients spent fewer days in the coronary care unit, were discharged home earlier, and had a better 1-year outcome than those in the higher-risk groups (p < 0.0001) (Table 7).This translated into significantly lower hospital costs in the low-risk group compared with the intermediate- and high-risk INSPIRE groups, where invasive diagnostic and interventional therapeutic procedures were more frequently used.
The INSPIRE study is the largest prospective trial yet completed to use gated ADSPECT for the purpose of stratifying risk and thereby guiding therapeutic decision making in clinically stable patients early after AMI. The INSPIRE cohort alone doubles the current published observational experience with pharmacologic SPECT imaging in the post-AMI arena and allowed for risk stratification within a patient population demographically similar to that seen in other recent trials (13,15,16). The INSPIRE results establish that the commonly measured scintigraphic variables of total and ischemic PDS significantly improve the precision for assessing risk beyond that provided by the TIMI risk score alone or when combined with LVEF. Gated ADSPECT also identified a large group of patients with small and/or minimally ischemic defects, where an initial conservative approach of medical therapy is a reasonable alternative to an invasive evaluation.
Study design features
The refinements in risk stratification we report are attributable to several unique study design features incorporated in the INSPIRE trial: 1) the imaging protocol used a nitrate-enhanced rest followed by an ADSPECT stress study to optimize image quality and allow accurate quantification of myocardial ischemia and LVEF; 2) online quantitative SPECT analysis through a central core laboratory provided uniform image interpretation; 3) domestic and international investigators were enlisted to ensure enrollment of a large and diverse patient population; 4) the protocol encouraged adherence to specific and currently accepted medical and interventional therapeutic strategies; 5) patient risk was prospectively defined by ADSPECT within days of admission and then verified by a 1-year outcome analysis; and 6) the value of ADSPECT for predicting outcome was analyzed in relation to the TIMI risk score.
ADSPECT for assessing risk and guiding therapeutics
The cornerstone of risk stratification in stable AMI survivors is to rapidly identify high-risk patients who might benefit from CR and low-risk patients for whom medical therapy and early hospital discharge is appropriate. The INSPIRE study achieved this goal by using perfusion imaging to stratify a heterogeneous post-AMI population into several distinct risk categories.
The most clinically relevant primary observation is that ADSPECT can accurately identify a large low-risk population irrespective of age, gender, site of infarction, or clinical risk. This low-risk group, which represented one-third of all enrolled patients, had small (<20%) perfusion defects, preserved LV function, and minimal (<10%) residual ischemia. Previous studies in patients with chronic coronary artery disease or prior AMI have suggested that CR does not improve outcome beyond medical therapy when there is minimal (<10%) (9) or no (17) provokable ischemia. In support of this, our low-risk group had the lowest rate of CR, the shortest hospital stay, the lowest hospital-related costs, and yet the lowest (1.8%) cumulative death and reinfarction rate at 1 year. In most centers these low-risk patients would typically have undergone coronary angiography followed by percutaneous coronary intervention of any residual coronary stenosis. However, the INSPIRE trial demonstrates that such patients can be medically managed with a quite acceptable low overall event rate and, compared with invasive treatment, at a cost savings approaching 65% (18).
The prospectively defined intermediate-risk group represented an additional 29% of patients who would unlikely benefit from an initial invasive approach and for whom hospital costs could be reduced up to 25% (18). Although this group was anticipated to have a higher event rate than the low-risk group based on their larger total PDS and more severe LV dysfunction (4,5), they had only minimal residual ischemic myocardium to revascularize. It is not surprising that CR did not alter event rates in this group.
The prospectively defined high-risk INSPIRE group had the longest hospital stay, the highest rate of CR, and the highest hospital-related costs but represented only 9% of patients. Patients in this group had the highest 1-year overall event rate (18%), particularly when not revascularized. Other trials also suggest that patients with extensive ischemic viable myocardium and a low LVEF benefit most from CR (10). The cost analysis from the INSPIRE trial highlights the substantial increase in hospital expenses attributable to coronary interventional procedures. Gated ADSPECT can specifically identify high-risk ischemic patients so as to properly allocate health care resources.
Previous smaller or retrospective observational studies suggest that low- and high-risk groups can be identified using exercise (19–21) or pharmacologic stressor agents (3–5) when combined with perfusion scintigraphy. The Veterans Affairs Non–Q-Wave Infarction Strategies in Hospital trial successfully used planar imaging to risk-stratify patients following non–Q-wave AMI (15,22). The Invasive Versus Conservative Treatment in Unstable Coronary Syndromes trial recently showed that patients with non–Q-wave AMI have a similar outcome whether initially evaluated with an invasive or a conservative approach (23).
In contrast, the FRISC-II (Fragmin and Fast Revascularization During Instability in CAD) (2) and TACTICS (Treat Angina With Aggrastat and Determine Cost of Therapy With Invasive or Conservative Strategy) (1) studies reported superiority of a routine invasive approach over a noninvasive strategy of ETT and selective angiography in patients with ACS. Both are particularly germane to the INSPIRE trial in that they have strongly influenced current practice guidelines; and, within the cardiology community, have expanded the existing widespread use of an invasive approach to now include patients with AMI. These trials chose ETT as the noninvasive testing standard—a suboptimal technique for detecting ischemia (20,21) or predicting risk compared with perfusion imaging (3,20). Furthermore, in the TACTICS study ∼40% of all cardiac events occurred in the conservative strategy before the index stress test. The clinical benefit of an early invasive approach in the TACTICS study was based solely on event rate differences within this early critical time period; patient outcome in both groups was comparable beyond 30 days. The TIMI experience in 20,101 patients emphasizes that reinfarction after thrombolytic therapy occurs generally within 3 days of admission (16). Because events recur early after AMI, a “watchful waiting” approach is not clinically acceptable but generally unavoidable when ETT is selected as the stressor modality. By using adenosine in the INSPIRE trial, 54% of patients were imaged safely within 2 days of admission, facilitating very early identification and treatment of high-risk patients.
We did not track the percentage of eligible patients who were enrolled in the INSPIRE trial. However, based on previous trials, up to 70% of AMI patients are clinically stable and would be good candidates for initial evaluation with ADSPECT (24,25). The TIMI risk score for ST-segment elevation AMI may not be entirely applicable to our population, because not all had ST-segment elevation even though all patients had AMI. Finally, although routine surveillance with cardiac enzymes was not performed after invasive procedures to identify subclinical AMI, it is unlikely that ADSPECT would have predicted such procedure-related events.
Gated ADSPECT is a safe and accurate initial method for identifying: 1) a very large low-risk group after AMI who can be targeted for medical therapy and early hospital discharge; 2) patients with predominantly scar where initial medical treatment is appropriate; and 3) those with extensive ischemia where intensive medical and/or interventional therapy is warranted. This is a cost-efficient paradigm already embraced in clinical practice for evaluating risk and guiding therapeutic decision making in patients with suspected or known CAD (26). The findings from the INSPIRE trial provide justification for recommending that ADSPECT be incorporated within current practice guidelines as an appropriate initial strategy for risk-stratifying stable survivors of ST-segment elevation AMI who do not undergo primary PCI and the emerging larger group of patients with non–ST-segment elevation AMI.
The authors are indebted to Drs. David R. Holmes, Robert P. Giugliano, and Enrique F. Schisterman for their participation on the Data Safety and Monitoring Board; to Drs. Raymond Taillefer, Anthony Fung, Richard Steingart, Michael Freeman, Vincent Robinson, Charles Stone, and Dante Graves for patient recruitment; to Gerald H. Olszewski, Bradley K. Pounds, and Ray Russo for ensuring study completion; to the study coordinators and nuclear technologists; and to Janice S. Preslar and Maria E. Frias for their support.
↵1 Dr. Mahmarian is on the Advisory Board for CV Therapeutics and Astellas Pharma US.
↵2 Dr. Shaw has received research grants for the END and COURAGE trials
↵3 Dr. Filipchuk received research grants for the DIAD, CCORD, and AMISCAN studies
↵4 Drs. Iskander and Henzlova are on the Speakers’ Bureau for Bristol-Myers Squibb.
Funding for INSPIRE was provided by Bristol-Myers Squibb Medical Imaging, Astellas Pharma US, and Schering Plough Research Institute.
- Abbreviations and Acronyms
- adenosine technetium-99m sestamibi single-photon emission computed tomography
- acute coronary syndromes
- acute myocardial infarction
- analysis of variance
- confidence interval
- coronary revascularization
- exercise treadmill testing
- general linear modeling
- left ventricular ejection fraction
- perfusion defect size
- relative risk
- summed difference score
- summed rest score
- summed stress score
- Thrombolysis In Myocardial Infarction
- Received March 30, 2006.
- Revision received July 5, 2006.
- Accepted July 6, 2006.
- American College of Cardiology Foundation
- Brown K.A.,
- Heller G.V.,
- Landin R.S.,
- et al.
- Mahmarian J.J.,
- Mahmarian A.C.,
- Marks G.F.,
- Pratt C.M.,
- Verani M.S.
- Mahmarian J.J.,
- Shaw L.J.,
- Olszewski G.H.,
- et al.
- Sciagra R.,
- Bisi G.,
- Santoro G.M.,
- et al.
- Hachamovitch R.,
- Hayes S.W.,
- Friedman J.D.,
- Cohen I.,
- Berman D.S.
- Allman K.C.,
- Shaw L.J.,
- Hachamovitch R.,
- Udelson J.E.
- Mahmarian J.J.,
- Dakik H.A.,
- Filipchuk N.G.,
- et al.
- Morrow D.A.,
- Antman E.M.,
- Charlesworth A.,
- et al.
- Gibson C.M.,
- Karha J.,
- Murphy S.A.,
- et al.
- Ellis S.G.,
- Mooney M.R.,
- George B.S.,
- et al.
- Gibson R.S.,
- Beller G.A.,
- Gheorghiade M.,
- et al.
- Dakik H.A.,
- Mahmarian J.J.,
- Kimball K.T.,
- Koutelou M.G.,
- Medrano R.,
- Verani M.S.
- Barnett P.G.,
- Chen S.,
- Boden W.E.,
- et al.
- Committee to Revise the 1999 Guidelines for the Management of Patients With Acute Myocardial Infarction
- Braunwald E.,
- Antman E.M.,
- Beasley J.W.,
- et al.
- Shaw L.J.,
- Hachamovitch R.,
- Berman D.S.,
- et al.