Prognostic Implications of Myocardial Perfusion Single-Photon Emission Computed Tomography in the Elderly
Background— The goal of this study was to assess the clinical value of stress myocardial perfusion scintigraphy (MPS) in elderly patients (≥75 years of age).
Methods and Results— We followed up 5200 elderly patients (41% exercise) after dual-isotope MPS over 2.8±1.7 years (362 cardiac deaths [CDs], 7.0%, 2.6%/y) and a subset with extended follow-up (684 patients for 6.2±2.9 years; 320 all-cause deaths). Survival modeling of CD revealed that both MPS-measured ischemia and fixed defect added incrementally to pre-MPS data in both adenosine and exercise stress patients. Modeling a subset with gated MPS (n=2472) revealed that ejection fraction and perfusion data added incrementally to each other, further enhancing risk stratification. Unadjusted, annualized post–normal MPS CD rate was 1.3% but <1% in patients with normal rest ECG, exercise stress, or age of 75 to 84 years and was 2.3% to 3.7% in patients ≥85 years of age or undergoing pharmacological stress. However, compared with age-matched US population CD rates (75 to 84 years of age, 1.5%; ≥85 years, 4.8%), normal MPS CD rates were approximately one-third lower than the baseline risk of US individuals (both P<0.05). Modeling of all-cause death in 684 patients with extended follow-up revealed that after risk adjustment, an interaction between early treatment and ischemia was present; increasing ischemia was associated with increasing survival with early revascularization, whereas in the setting of little or no ischemia, medical therapy had improved outcomes.
Conclusions— Stress MPS effectively stratifies CD risk in elderly patients and may identify optimal post-MPS therapy. CD rates after normal MPS are low in all subsets in relative terms compared with the age-matched US population.
Received December 5, 2007; accepted September 21, 2009.
Cardiovascular care for the elderly poses a significant healthcare challenge. Constituting only 6.1% of the population, they suffer two thirds of all cardiovascular deaths.1 One-year cardiac death (CD) frequency in the United States is 1.46% in 75 to 84 year olds, increasing to 4.78% in individuals ≥85 years of age.2 Because coronary artery disease (CAD) in the elderly frequently presents with silent ischemia, atypical symptoms, or nonspecific functional status deterioration, identifying “at-risk” elderly is often difficult.3 Hence, stress imaging may play a role in the strategy to identify at-risk elderly individuals.
Clinical Perspective on p 2206
Because the elderly are a heterogeneous population, understanding the performance characteristics of stress imaging in this group is challenging. Some elderly cohorts are at lower risk (eg, normal rest ECG, no prior CAD, ability to exercise adequately)4–7 than others (eg, known CAD, inability to exercise). Hence, cardiac event rates after normal single photon emission computed tomography (SPECT) myocardial perfusion scintigraphy (MPS) may no longer be uniformly “low” (<1%/y).4,8 To understand the implications of MPS in the elderly, it is important to define the ability of the test to stratify risk and to yield incremental value in multiple patient subsets. To date, prognostic studies in elderly populations are limited in size and power, particularly relative to pharmacological stress5–11 and the value of gated MPS.10 Our goal was to examine the role of MPS for prognostic evaluation in the elderly (≥75 years of age) as defined by the ability to yield incremental prognostic value, to achieve clinically relevant risk stratification, and to add value to patient care by identifying patients who may benefit from aggressive versus conservative strategies.
We examined 5887 consecutive elderly patients (≥75 years of age) who underwent rest thallium-201/stress technetium-99m sestamibi dual-isotope MPS with exercise or adenosine stress between 1991 and 1999. If patients had multiple studies, only their initial study was considered. Patients with known nonischemic cardiomyopathy (left ventricular [LV] systolic dysfunction without obstructive CAD) or valvular heart disease were excluded.
Of the initial population, 172 patients (2.9%) were lost to follow-up, yielding a study cohort of 5715 patients. Of these, a subset of 2472 patients underwent gated MPS. Patients undergoing early (≤60 days after MPS) revascularization were censored from survival analyses,12,13 leaving a study cohort of 5200 patients (2158 exercise, 3042 adenosine stress). To enhance our analysis of post-gated MPS prognosis and MPS-associated patient benefit, we also examined a subset of 684 elderly patients tested between 1991 and 1997 (12% of total cohort) who underwent gated MPS and had long-term follow-up.
Imaging and Stress Protocol
Patients received 3 to 4.5 mCi thallium-201 at rest, and MPS was initiated 10 minutes afterward. Stress exercise or adenosine protocols with or without low-level treadmill exercise were performed as previously described, with stress injection of technetium-99m sestamibi (25 to 40 mCi). Twelve-lead ECG was monitored during stress tests. Patients with significant resting defects usually underwent 24-hour redistribution thallium-201 MPS to assess reversibility. In patients studied after 1994, the poststress acquisition used 8-frame gated MPS, and LV ejection fraction (EF), end-systolic volume, and end-diastolic volume were assessed with an automatic program.14
Image Interpretation and Scintigraphic and Other Indexes
Summed stress and rest scores were obtained by adding the scores of 20 myocardial segments.12 The sum of the differences between these 2 scores defined the summed difference score. These indexes were converted to a percent myocardium with abnormal stress, ischemic, or fixed defects by dividing the summed scores by 80, the maximum potential score (4 points×20 segments), and multiplying by 100.15 Normal scans and mildly and moderately to severely abnormal scans were defined as percent myocardium abnormal <5%, 5% to 10%, and >10% myocardium.15 For patients undergoing exercise, the Duke treadmill score was calculated as previously described.16 Normal rest ECG was defined as the presence of no or limited abnormalities (sinus tachycardia, first-degree atrioventricular block, premature atrial contraction, premature ventricular contraction, mild intraventricular conduction delay [≤0.12 seconds], or sinus bradycardia).
All patients prospectively enrolled in Cedars-Sinai Medical Center SPECT MPS registry were contacted once at least 1 year after their index MPS via mailed questionnaire, followed by scripted and blinded telephone interviews with patients not initially responding to the mailing. Deaths were identified and confirmed through our hospital-based patient information system, California state records, and death certificates forwarded from outside sources. Death, cardiac or noncardiac, and myocardial infarction were verified by 2 physicians’ (blinded) independent reviews of admission reports, discharge summaries, and consultation/laboratory reports. Catheterizations and revascularizations were verified by hospital records. The Social Security Death Index use was limited to confirmation of patient deaths occurring before the follow-up date. CD was defined as death resulting from any cardiac cause (lethal arrhythmia, myocardial infarction, or sudden death). Patients not confirmed to be deceased and without follow-up information were considered lost to follow-up. The average length of follow-up was 2.8±1.7 years. The subset of 684 patients who underwent gated MPS (between 1991 and 1997) had a second follow-up for death using Social Security Death Index data (mean follow-up, 6.2±2.9 years). Age group–specific mortality rates and cause of death for the year 2005 were obtained from the National Center for Health Statistics control.2
Continuous variables are given as mean±SD and compared by ANOVA test. Categorical variables are presented as frequencies and compared by a χ2 test. Bonferroni corrections were made as appropriate. Statistical significance was defined at P<0.05.
National Center for Health Statistics–identified age group death frequencies (per 100 000 population per year) were expressed as annual rates and compared with annualized CD rates from our cohort. Because we excluded patients with cardiomyopathy or valve disease, National Center for Health Statistics–published heart disease death rates were analogous to CD rates in our cohort.
The association between covariates of interest and survival time free of CD was assessed with Cox proportional-hazards analysis (S-plus 2000 software, Insightful Corp, Seattle, Wash). Initially, univariate analysis was performed (variables in Tables 1 and 2⇓ were considered), and the threshold for variable entry into multivariable models was P<0.05 (P<0.10 for interaction term inclusion) and for variable removal was P>0.10. Model assumptions were tested (proportional hazards, linearity, and additivity) when appropriate.17
Multivariable modeling considered (1) clinical and historical variables alone, (2) addition of nonimaging stress test data, (3) addition of fixed defect information, and (4) addition of imaging-detected ischemia. To assess the value of gated MPS data, we reexamined step 4 in a subset of patients who underwent gated MPS studies and added gated data as a final modeling step (step 5). Finally, the role of ischemia in risk assessment was examined using survival analysis of all-cause death (ACD) in the 684-patient subset defined above. A propensity score was derived from logistic regression modeling of early revascularization referral and used to adjust subsequent ACD modeling.
Patients undergoing pharmacological stress were older, more commonly were women, and more frequently had markers of greater risk (prior myocardial infarction and revascularization, greater age, diabetes, hypertension, angina, dyspnea, atrial fibrillation, and abnormal rest ECG; Table 1). Of these patients, 58% were referred for CAD diagnosis.
Patients undergoing gated MPS were 81±4 years of age; 51% were male; and 33% had exercise stress (P<0.05 versus the overall population). Prior CAD was present in 44%, abnormal MPS in 42%, and low LVEF in 25%. Patients with low LVEF presented more commonly with dyspnea (24%; P<0.05 versus the overall population) and as commonly with angina (40%). In this cohort, 56% had no prior CAD and were referred for CAD diagnosis.
In comparison (Table 1), the long-term follow-up subgroup was generally midway between the 2 other groups. However, that subgroup had more frequent diabetes and anginal symptoms than exercise stress patients and less digoxin use than pharmacological stress patients.
Compared with patients undergoing exercise stress, patients undergoing pharmacological stress had more severe and extensive MPS defects (largely resulting from greater percent myocardium fixed), larger LV end-diastolic volume, and reduced LV function (Table 2). Long-term follow-up patients were characterized by larger amounts of ischemia than the other 2 patient groups. Otherwise, they were similar to pharmacological stress patients with respect to EF and percent myocardium fixed, but their end-diastolic volume was closer to that of exercise patients.
During follow-up, 362 CDs (7.0%, 2.6%/y) and 1012 ACDs (19.5%, 7.4%/y) occurred. Of these, 187 CDs (51.7%) occurred in patients with gated MPS data, and 115 (61%) were in patients with low LVEF.
Multiple clinical, historical, and stress test data were univariate predictors of CD (Table 3), most notably age, prior CAD, digoxin use, abnormal rest ECG, and dyspnea. Male sex and diabetes were also predictors in pharmacological stress patients.
The subset of 684 elderly patients experienced 320 ACDs over a mean follow-up of 6.3±2.9 years (47% death rate, 7.4%/y) and underwent 83 early revascularization procedures (<90 days after MPS; 12.1%). Predictors of ACD included inability to perform the exercise tolerance test, diabetes mellitus, abnormal rest ECG, dyspnea, and prior CAD.
All MPS variables were predictive of CD and ACD. Hazard ratios (HRs) tended to be greater for CD than ACD for male sex, diabetes, MPS perfusion variables, abnormal rest ECG, CAD history variables, and dyspnea.
Overall CD rates were greater in patients undergoing pharmacological stress with prior CAD and with abnormal rest ECG compared with their counterparts (Table 4). No sex-related mortality differences were noted. Importantly, patients with normal rest ECG were the lowest-risk subgroup.
Unadjusted CD Rates and MPS Results
Analysis of the relationship between MPS results and CD risk included examination of post–normal MPS risk of CD, post-MPS risk stratification, incremental prognostic value of MPS over pre-MPS data, the prognostic impact of gated SPECT, and in the subset of 684 patients, the ability of MPS to identify which patients may benefit from revascularization.
Risk of CD After a Normal MPS
The unadjusted, annualized post–normal MPS CD rate was 1.3% but lower with exercise compared with pharmacological stress (Figure 1) and in “younger” (75 to 84 years of age) versus “older” (≥85 years of age) patients (Table 5). Annualized cardiac mortality was low (≤1%/y) in patients with normal rest ECG, age <85 years, or exercise stress test. However, in many subsets, post–normal MPS CD rates exceeded 1%/y. In patients ≥85 years of age, CD rates were between 2%/y and 4%/y, except in patients with normal rest ECG.
Table 5 and Figure 2 compare the event rates in our cohort with those reported for age-matched individuals in the general US population.2 Compared with the age-matched group, the 75- to 84-year-old group of our cohort was at higher risk for both ACD and CD. The ≥85-year-old group of our cohort was at lower risk for ACD but higher risk for CD compared with the general population counterparts (n=378 777 and 703 169, respectively, in the 2 age groups). In both age groups, with respect to both end points, patients with normal results tended to be at lower risk and those with abnormal results tended to be at higher risk than the age-matched population. Hence, normal MPS in elderly populations after MPS may be said to be associated with a relatively lower risk (compared with the general population) but not necessarily low absolute risk.
Risk Stratification After MPS
Unadjusted annual CD rates demonstrated significant risk stratification, with relatively low CD rates after normal MPS and significant increases in annual CD rates with increasing scan abnormality after both pharmacological and exercise stress (P<0.001; Figure 1). CD risk was stratified by MPS results (Table 4) in most patient subgroups except patients with normal rest ECG, likely because of normal LV function and insufficient power in the subset of those ≥85 years of age (n=123).
Prognostic Impact of LV Function Data
LVEF was significantly lower in patients with CD compared with others (42±18% versus 58±15%; P<0.0001). In the gated MPS patient subgroup (Figure 3), patients with reduced EF had higher unadjusted CD rates compared with those with normal EF in all perfusion categories (all P<0.01). Perfusion results stratified the risks of cardiac events in the overall cohort and in patients with normal LV function (unadjusted P<0.05) but not in patients with reduced EF, possibly because of insufficient power.
Multivariable Modeling for Incremental Value
Survival modeling of CD revealed that nonimaging stress test data added incrementally over clinical and historical data for both exercise and pharmacological stress (Tables 6 and 7⇓). Furthermore, both percent myocardium fixed and percent myocardium ischemic further enhanced these models. The overall strength of association was relatively lower with the exercise stress model compared with the pharmacological stress model and lower for percent myocardium ischemic compared with percent myocardium fixed in both models.
The initial models of clinical and historical data included patient age, abnormal rest ECG, dyspnea, and prior CAD in both the exercise stress and pharmacological stress subgroups. The pharmacological stress subgroup also included hypercholesterolemia, body mass index, and digoxin use. The addition of stress test data (model 2) increased χ2 significantly in both subgroups, with a proportionally greater increase in the exercise compared with the pharmacological stress subgroups. Of note, the exercise group included the Duke treadmill score as a significant predictor.
The addition of percent myocardium fixed further increased χ2 in both subgroups, more so in the pharmacological stress group. In the exercise stress model, the addition of percent myocardium fixed was accompanied by an interaction of this variable with patient history of prior myocardial infarction. Finally, the addition of percent myocardium ischemic further increased model χ2, again more so in the pharmacological stress group. In the exercise stress model, the addition of percent myocardium ischemic was accompanied by an interaction of this variable with the presence of anginal symptoms. Of note, in the final model, the HRs of percent myocardium ischemic and percent myocardium fixed were 1.26 and 1.61, indicating that fixed defects had approximately twice the prognostic impact of ischemic defects for predicting CD.
Incremental Prognostic Value of Combined Perfusion-Function Studies
In the 2472-patient subset with gated MPS data, after adjustment for age, digoxin use, atrial fibrillation, and heart rate ratio with Cox proportional-hazards modeling, both LVEF and percent myocardium abnormal were independent predictors for CD after adjustment (both P<0.005, χ2; LVEF, 28; percent myocardium abnormal, 9). Patients with both normal perfusion and LVEF had the highest survival; patients with abnormal perfusion and lower LVEF had the lowest survival. Differences across survival curves were significant (P<0.001 across these categories; Figure 4). Compared with patients with normal perfusion and normal EF, patients with abnormal stress perfusion and normal EF had an HR of 1.88 (95% confidence interval, 1.18 to 3.00); patients with normal stress perfusion and abnormal EF had an HR of 3.17 (95% confidence interval, 1.80 to 5.58); and patients with abnormal stress perfusion and abnormal EF had an HR of 5.71 (95% confidence interval, 3.92 to 8.32) (P<0.01 between the group with both normal perfusion and EF and every other group). Perfusion results combined with LV function further stratified the risk of CD in elderly patients with reduced or normal EF.
Modeling of ACD and Identification of Survival Benefit
Early logistic regression modeling of the most predictive model of revascularization referral included percent myocardium ischemic, EF, LV enlargement, anginal symptoms, and β-blocker use (global χ2, 86; percent myocardium ischemic χ2, 53; c index, 0.84). After adjustment for potential confounders, as well as a propensity score based on this logistic model, Cox proportional-hazards modeling identified a significant interaction between the use of early revascularization and percent myocardium ischemic; early revascularization increased risk in the setting of little or no ischemia but decreased risk in the setting of increasing amounts of ischemia (Figure 5). No such interaction was found with percent myocardium fixed, EF, or LV volumes.
In this study, we evaluated the clinical value of MPS in elderly patients, examining post–normal MPS CD risk, risk stratification by MPS, incremental prognostic value of MPS over pre-MPS data, prognostic impact of gated SPECT data, and whether MPS adds value via identification of optimal treatment. Although post–normal MPS CD rates were <1%/y in certain patient subsets, they were 2%/y to 4%/y in multiple subsets. To understand these results in the context of elderly patients’ greater risk, we compared post–normal MPS risk with age-matched CD rates in the United States (Table 4 and Figure 2), finding that post–normal MPS CD rates were lower than age-matched CD rates. Thus, although normal MPS did not confer absolute low risk (eg, <1%/y), it was associated with low risk relative to the age-matched population. With respect to risk stratification, normal MPS had lower CD rates, and CD risk increased with worsening MPS abnormalities.
Multivariable survival analysis identified both reversible and fixed defects as highly associated with CD and adding incremental value over pre-MPS data, the latter more than the former. Importantly, clinical factors (eg, age, rest ECG, prior CAD, Duke treadmill score, inability to exercise) added prognostic value over MPS data in this model, identifying patient subsets at greater cardiac risk after any MPS result. In the gated MPS subgroup, both perfusion and gated LV function data added incrementally over pre-MPS data for CD prediction. Abnormal perfusion further stratified risk in patients with normal or abnormal EF, whereas gated EF results stratified risk in patients both with normal and abnormal perfusion.
Finally, survival modeling in the 684-patient subgroup with extended follow-up revealed enhanced survival with medical therapy in the setting of little or no ischemia but an increasing survival benefit associated with the use of revascularization in the setting of increasing MPS-measured ischemia. Conversely, revascularization use in patients without significant ischemia was associated with increased ACD risk (Figure 5). No such relationship was present with other MPS data. Hence, MPS-identified ischemia may yield improved patient value by identifying improved survival with therapy based on MPS results. Furthermore, although percent myocardium fixed was a superior predictor of risk, only percent myocardium ischemic was able to identify patients who may potentially benefit from revascularization compared with medical therapy.
Post–Normal MPS Risk in the Elderly: How Do We Define Low Risk?
To date, normal MPS identification of low-risk patients (<1%/y risk of CD or hard events) has been the underpinning of clinically effective and cost-effective risk stratification because additional testing in these patients is unlikely.12,17 Although post–normal MPS risk is often low, higher event rates in higher-risk patients have been reported (eg, pharmacological stress, diabetics, elderly, prior CAD). We previously reported the impact of clinical factors on post-MPS risk and its temporal profile; increasing numbers of clinical risk markers resulted in greater risk.4,8 Previous studies in lower-risk elderly patients (eg, interpretable rest ECG, exercise stress, no prior CAD) have reported “low” event rates after normal MPS.4–7,11 However, previous studies were insufficiently powered to examine higher-risk subgroups, and the present study demonstrates that normal MPS in high-risk elderly patients does not have a low absolute mortality rate.
Is this finding of higher-than-expected event rates a failure of the test (it does not “work” and thus should not be used), an issue with the patients tested, or a flawed paradigm (all normal stress imaging studies are low risk)? Although these questions have previously been addressed, we present here a potential approach to assessing thresholds defining low risk in an elderly population by means of comparing post-MPS risk with that of the general elderly population. This approach suggests that the test works, but our paradigm of post-MPS risk must incorporate the basic Bayesian principle of considering baseline patient risk in post-MPS risk estimation.
Prognostic Value of LV Function
Previous studies reported that poststress gated SPECT LVEF, an important prognostic marker of CD, added incrementally to perfusion data for risk prediction.18–20 In this study, we show that combining LVEF and perfusion data enhances risk stratification, with greater risk with both abnormal perfusion and EF and lower risk with both normal perfusion and EF. The presence of either abnormal EF or stress perfusion increases risk incrementally. Compared with patients with normal MPS perfusion and function, CD risk increased almost 2-fold (HR, 1.88) with the finding of abnormal perfusion and 3-fold (HR, 3.17) with the finding of abnormal EF. The presence of both increased risk almost 6-fold (HR, 5.71). This finding extends previous data in 294 elderly patients demonstrating that gated SPECT data provided incremental prognostic value over clinical information and fixed defect for predicting all-cause or CD.10
Potential Role of Stress MPS in the Elderly
In the present study, MPS satisfies all criteria for adding incremental prognostic value and successfully stratifying risk an elderly population. A normal MPS in an elderly patient identifies a level of risk below that anticipated for the age-matched general US population, hence reassuring the referring physician with respect to pursuing medical therapy. On the other hand, in elderly patients, risk not only rises in the setting of an abnormal MPS result but commonly rises to levels higher than those reported in other patient subsets. More important, our findings also suggest that an MPS-guided approach with selective use of revascularization procedures may result in improved patient outcomes and enhanced value.
Comparison With Previous Studies
The Mayo Clinic group reported that in elderly patients, Duke treadmill score failed to risk stratify patients and predict CD, whereas MPS was a superior approach in elderly patients without prior CAD whose rest ECG was interpretable for exercise tolerance test and who could adequately exercise.7 The present study extends their findings and others5,9 on the prognostic value of MPS in the elderly but does not support the strategy of direct referral of elderly patients to MPS because Duke treadmill score and other exercise tolerance test data added incrementally in patients able to exercise. It must be noted that previous studies reported post-MPS event rates and risk stratification in elderly patients as subset analyses of larger studies,4,15 also demonstrating similar results.
Our study extends these previous results to a considerably larger cohort and is unique in being the first to evaluate combined perfusion-function studies in an elderly cohort, finding that these measures contribute incrementally to each other for the purpose of enhancing risk stratification. To the best of our knowledge, the present study is the first to use the general US population risk as a comparator of patient risk status and the first to evaluate the ability of MPS to identify optimal post-MPS therapy.
Risk-Based Versus Benefit-Based Testing
The present study also highlights important issues related to how we assess testing. Our results extend previous studies indicating that measures of scar and LV dysfunction (percent myocardium fixed, LVEF) dominate measures of ischemia (percent myocardium ischemic) for the prediction of CD.18–20 Indeed, the latter yielded limited incremental value over the former. Nevertheless, only ischemia, not LVEF or scar, was able to predict which post-MPS clinical strategy would enhance patient survival. Hence, there is an uncoupling of predictors of risk (scar, LVEF) from predictors of post-MPS physician action and of potential survival benefit (ischemia). This suggests that our standard methodologies assessing prognostic value identify measures of risk that do not drive patient management and that clinical decision making is based on measures that have marginal independent prognostic value (ischemia) but are intuitively accepted by physicians as the basis of identifying treatment candidates.
These findings raise 2 questions. First, is our current paradigm of validating stress imaging by assessing its ability to stratify risk a meaningful and accurate approach? The uncoupling of predictors of risk from predictors of physician action and possible survival benefit suggests that our analytical approaches may result in misleading clinical recommendations. Thus, should validation of testing also incorporate a benefit- or value-based approach, along with a risk-based approach, to test evaluation? This approach has been previously recommended for other reasons.17
Although imperfect, multivariable modeling of observational data series, when carefully applied, has been shown to yield results similar to those of clinical trials in similar populations. Although patients in observational studies better represent those seen in practice and, unlike subjects in clinical trials, can account for changes in therapy over time, the impact of selection biases, spurious observations, and missing or unmeasured covariates cannot be ignored.
Semiquantitative MPS interpretation used a 20-segment rather than the currently recommended 17-segment model; however, when converted to percent myocardium, the 2 approaches have similar prognostic performances.21 Importantly, summed scores and percent myocardium express identical information based on the same interpretation. Gated MPS was performed routinely in our center only after 1994; hence, this information is limited to a subset of patients in our study.
In elderly patients undergoing either pharmacological or exercise stress, MPS yielded risk stratification, incremental value, and identification of which patients may benefit from a revascularization strategy. Even in the elderly, risk increases dramatically with increasing age and other clinical factors, thus marking these cohorts as heterogeneous and necessitating examination of multiple patient subsets to understand prognostic test performance.
Sources of Funding
This work was supported in part by grants from Bristol-Myers Squibb Medical Imaging, Inc, Billerica, Mass, and Astellas Pharma, Inc, Deerfield, Ill.
Dr Berman has grants from Lantheus Medical Imaging, Inc and is on the speakers’ bureau for Astellas Pharma, Inc. Drs Berman and Germano participate in royalties to Cedars-Sinai Medical Center for licensure of software used in this study. Dr Hachamovitch has grants from Astellas Pharma, Inc, GE Healthcare, Siemens Medical Solutions, and Bracco and consultants for Lantheus Medical Imaging. The other authors report no conflicts.
Heart and Stroke Statistical Summary, 2005. Dallas, Tex: American Heart Association; 2006.
Kung HC, Hoyert DL, Xu JQ, Murphy SL. Deaths: Final Data for 2005: National Vital Statistics Reports. Hyattsville, Md: National Center for Health Statistics; 2008; 56: 121.
Ambepitiya GB, Iyengar EN, Roberts ME. Review: silent exertional myocardial ischaemia and perception of angina in elderly people. Age Ageing. 1993; 22: 302–307.
Schinkel AF, Elhendy A, Biagini E, van Domburg RT, Valkema R, Rizello V, Pedone C, Simoons M, Bax JJ, Poldermans D. Prognostic stratification using dobutamine stress 99mTc-tetrofosmin myocardial perfusion SPECT in elderly patients unable to perform exercise testing. J Nucl Med. 2005; 46: 12–18.
Valeti US, Miller TD, Hodge DO, Gibbons RJ. Exercise single-photon emission computed tomography provides effective risk stratification of elderly men and elderly women. Circulation. 2005; 111: 1771–1776.
De Winter O, Velghe A, Van de Veire N, De Bondt P, De Buyzere M, Van De Wiele C, De Backer G, Gillebert T, Dierckx R, De Sutter J. Incremental prognostic value of combined perfusion and function assessment during myocardial gated SPECT in patients aged 75 years or older. J Nucl Cardiol. 2005; 12: 662–670.
Berman DS, Hachamovitch R, Shaw LJ, Germano G, Hayes S. Nuclear cardiology. In: Fuster AR V, King S, O'Rourke RA, Wellens HJJ, eds. Hurst’s The Heart. New York, NY: McGraw-Hill Co; 2004: 525–565.
Germano G, Kiat H, Kavanagh PB, Moriel M, Mazzanti M, Su HT, Van Train KF, Berman DS. Automatic quantification of ejection fraction from gated myocardial perfusion SPECT. J Nucl Med. 1995; 36: 2138–2147.
Hachamovitch R, Hayes SW, Friedman JD, Cohen I, Berman DS. Comparison of the short-term survival benefit associated with revascularization compared with medical therapy in patients with no prior coronary artery disease undergoing stress myocardial perfusion single photon emission computed tomography. Circulation. 2003; 107: 2900–2907.
Hachamovitch R, Di Carli MF. Contemporary reviews in cardiovascular medicine: methods and limitations of assessing new noninvasive tests, II: outcomes-based validation and reliability assessment of noninvasive testing. Circulation. 2008; 117: 2793–2801.
Hachamovitch R, Hayes S, Friedman JD, Cohen I, Berman DS. Relative role of inducible ischemia versus ejection fraction in the prediction of survival benefit with revascularization compared to medical therapy in patients with no prior revascularization undergoing stress myocardial perfusion SPECT. J Nucl Cardiol. 2006; 13: 768–778.
Sharir T, Germano G, Kavanagh PB, Lai S, Cohen I, Lewin HC, Friedman JD, Zellweger MJ, Berman DS. Incremental prognostic value of post-stress left ventricular ejection fraction and volume by gated myocardial perfusion single photon emission computed tomography. Circulation. 1999; 100: 1035–1042.
Berman DS, Abidov A, Kang X, Hayes SW, Friedman JD, Sciammarella MG, Cohen I, Gerlach J, Waechter PB, Germano G, Hachamovitch R. Prognostic validation of a 17-segment score derived from a 20-segment score for myocardial perfusion SPECT interpretation. J Nucl Cardiol. 2004; 11: 414–423.
We assessed the clinical value of stress myocardial perfusion scintigraphy (MPS) in 5200 elderly patients (≥75 years of age) followed up for 2.8±1.7 years (362 cardiac deaths, 7.0%, 2.6%/y) and in a subset of 684 patients with extended follow-up (6.2±2.9 years; 320 all-cause deaths). The unadjusted cardiac death rate was 1.3%/y after normal MPS but was <1% in patients with normal rest ECG, exercise stress, or age of 75 to 84 years and was 2.3% to 3.7% in patients ≥85 years of age or undergoing pharmacological stress. Importantly, compared with age-matched US population CD rates (75 to 84 years of age, 1.5%; ≥85 years of age, 4.8%), normal MPS cardiac death rates were approximately one-third lower than the baseline risk of US individuals (P<0.05 for both). Survival modeling of cardiac death revealed that MPS data yielded incremental value over pre-MPS data. The addition of gated MPS in a subset of 2472 patients showed that perfusion and ejection fraction data added incrementally to each other and enhanced risk stratification. Modeling of all-cause death in 684 patients with extended follow-up revealed that after risk adjustment, an interaction between early treatment and ischemia was present so that increasing ischemia was associated with increasing survival with early revascularization, whereas in the setting of little or no ischemia, medical therapy had improved outcomes. In an elderly population, stress MPS stratifies cardiac death risk in elderly patients and may identify optimal post-MPS therapy. Postnormal MPS cardiac death rates are low in all subsets in relative terms compared with the age-matched US population.