Author + information
- Received May 22, 2003
- Revision received July 10, 2003
- Accepted July 22, 2003
- Published online January 21, 2004.
- Gregory S. Thomas, MD, MPH, FACC*,* (, )
- Michael I. Miyamoto, MD, MS, FACC*,
- A.Peter Morello III, ScB*,
- Haresh Majmundar, CNMT*,
- Jennifer J. Thomas, BA*,
- Christine H. Sampson*,
- Rory Hachamovitch, MD, MSc, FACC† and
- Leslee J. Shaw, PhD‡
- ↵*Reprint requests and correspondence:
Dr. Gregory S. Thomas, Mission Internal Medical Group, 26732 Crown Valley Parkway, Suite 155, Mission Viejo, California 92691, USA.
Objectives The purpose of this study was to evaluate the prognostic value of community-based myocardial perfusion imaging (MPI) and to assess the incremental value of individual components of 99mTc-sestamibi single photon emission computed tomography (SPECT).
Background Although the most rapid growth of MPI has been in community outpatient laboratories, its prognostic value has not been validated in this setting.
Methods We prospectively followed 1,612 consecutive patients undergoing stress 99mTc-sestamibi SPECT in an outpatient community laboratory who experienced 71 hard events over 24 ± 7 months (0.2% lost to follow-up).
Results Patients whose scans were normal incurred an annualized event rate of 0.4%, compared with 2.3% for those with abnormal scans (p < 0.0001). Subset analysis demonstrated comparable risk stratification for women and men, diabetics, patients with normal resting ECGs, and those referred for pharmacologic and exercise stress. After adjusting for pre-test variables, multivariable Cox regression analysis found the most potent independent components of MPI to be, in order of importance, transient ischemic dilation, extent of reversibility, post-stress ejection fraction, extent and severity of the stress perfusion defect, and the overall test result (normal or abnormal). Each 1% decrement of ejection fraction predicted a 3% increase in risk (p = 0.0009). Post-MPI angiography and revascularization increased commensurate with the extent and severity of MPI result.
Conclusions The prognostic value of perfusion imaging is portable and transferable to the outpatient community setting, with multiple components of MPI providing incremental prognostic information.
The increase in the use of myocardial perfusion imaging (MPI) in the U.S. over the last decade has been striking. Myocardial perfusion imaging use increased almost fourfold between 1992 and 2001—a year in which 7.8 million studies were performed (1,2). In 2003, an estimated 1 in 15 Americans 40 years or older will undergo MPI. A substantial portion of this growth has been driven by a shift of performance of MPI from the hospital to the physician's office or related outpatient facility. Between 1992 and 2001, the percentage of all scans performed in the latter setting increased from 13% to 33%, totaling 2.6 million in 2001 (1,2). The total annual cost of these physicians' offices and related outpatient facilities scans currently exceeds $2.5 billion. In addition, of inpatient and outpatient perfusion imaging performed in a hospital location, 73% of studies are performed in hospitals with <400 beds, the vast majority of which would be expected to represent nonteaching, community hospitals (1). Although the prognostic value of MPI in academic hospital-based centers has been well documented (3–7), there is little evidence that similar prognostic information can be obtained in the community.
Concomitantly, the wide availability of technetium-based agents and expanded computer processing power has facilitated the increasing use of gated single photon emission computed tomography (SPECT). Between 1996 and 2001, the percentage of perfusion scans in which ejection fraction (EF) was obtained increased from 10% (Center for Medicare and Medicaid Services, 2002) to 88% (1). However, few studies have evaluated either the prognostic value of EF obtained by gated SPECT or other nonperfusion characteristics such as transient ischemic dilation (TID) or resting left ventricular (LV) size when assessed by these new 99mTc-labeled agents.
The aims of this study were, thus, threefold: 1) to evaluate the prognostic value of MPI in the community setting; 2) to determine the components of 99mTc-sestamibi MPI that best predict outcome; and 3) to evaluate the post-test utilization of cardiac catheterization and revascularization after MPI performed in this nonacademic setting.
The study population consisted of 1,782 consecutive patients undergoing 99mTc-sestamibi stress gated SPECT imaging between August 1997 and March 1999 in the outpatient nuclear cardiology laboratory of the 42-physician Mission Internal Medical Group in Mission Viejo, California. During this period, 1,612 patients underwent 1,792 studies. All patients provided written informed consent for MPI and long-term follow-up of their health status, including permission to contact their physicians and health care institutions to determine their health status and to present the results of their study and follow-up for scientific purposes.
Stress testing protocol
Exercise and pharmacologic testing was physician supervised and performed within the guidelines of the American College of Cardiology/American Heart Association (8). Imaging was carried out within the guidelines of the American Society of Nuclear Cardiology (9). The method of stress was exercise in 63% of patients, adenosine in 34%, and dobutamine in 3%. Exercise studies were symptom-limited, with a goal of ≥85% maximum predicted heart rate. Among patients undergoing adenosine stress, 66% underwent simultaneous low-level treadmill exercise during the 6-min adenosine infusion of 140 μ/kg/min (termed “Adenoex”), as previously described (10). Simultaneous treadmill exercise was not performed if the patient was judged by the supervising physician to be physically unable to walk for 6 min or if his/her resting electrocardiogram (ECG) demonstrated left bundle branch block or ventricular pacing.
Tomographic imaging was performed, as described previously (10), with a single-head Seimens ZLC 7500 Orbiter camera equipped with 75 photomultiplier tubes (Hoffman Estates, Illinois). A 180° arc was used with 32 projections, 40 s per projection for 99mTc-sestamibi and 45 s for 201Tl. Dual isotope imaging with resting 201Tl and stress 99mTc-sestamibi was performed if the patient was <240 pounds, and two-day 99mTc-sestamibi imaging study was used for patients of greater weight or challenging body habitus. The SPECT imaging was performed with eight-frame gating 15 to 60 min post stress. Images were displayed using an 80 × 80 pixel matrix with acquisition software authored by Lear (11)that included EF. Motion, attenuation, and scatter correction and quantitative perfusion or LV function algorithms were not available.
The technologist reviewed the raw data of each scan for patient motion, and the processed images of each stress scan for an inferior defect, to determine if a second, 8-min non-gated SPECT scan should be performed. If motion was judged to be ≥2 pixels, a repeat non-gated SPECT scan was performed. Additionally, if the technologist observed an inferior defect on the stress images, the patient underwent a non-gated SPECT scan in the prone position. In total, a second post-stress SPECT scan was performed in 27% of patients.
Each study was interpreted by one of two nuclear cardiologists certified by the Certification Board of Nuclear Cardiology (G.S.T., M.I.M.) and included an overall five-point global score of normal, probably normal, equivocal, probably abnormal, or abnormal. A scan was defined to be “abnormal” if either perfusion, LV function, or LV size was abnormal. Expected abnormal septal motion secondary to bundle branch block, ventricular pacing, or after cardiac surgery was regarded as normal. For the purposes of the prognostic component of this investigation and as prospectively specified, scans interpreted as normal, probably normal, or equivocal were grouped as normal. Those scored as abnormal or probably abnormal were grouped as abnormal. Only in evaluating the use of post-test coronary angiography and revascularization was the five-point scoring system retained. Left ventricular size at rest was scored semi-quantitatively on a four-point scale as normal or mildly, moderately, or severely enlarged. Similarly, TID was described on a four-point scale as absent, mild, moderate, or severe. Both resting LV size and TID were assessed by visual evaluation. Interpreters “read around” the expected difference in cavity size between the two isotopes when dual isotope imaging was performed. Representative examples of TID scoring as mild, moderate, and severe with dual isotope imaging are shown in Figure 1.
As would be expected in an office setting, clinical information was often available to the interpreting nuclear cardiologist. In 88% of cases, the interpreting cardiologist was not the referring cardiologist.
Only qualitative perfusion information was entered in the initial clinical MPI reports. Retrospectively, in order to quantitatively assess the extent and severity of the stress defect and the extent of reversibility, the two nuclear cardiologists performed a blinded review of the qualitative reports of all patients to provide a semi-quantitative score of the extent and severity at stress and extent of reversibility. The review was performed without knowledge of patient identity, age, or clinical information. The extent and severity of observed perfusion defects were scored on a four-point scale, based on their size and intensity described in the initial report (Table 1A). Defect extent and severity ranged from a score of 1 (small, mild-to-moderate intensity defect) to a score of 4 (large, severe intensity defect). Where multiple defects were observed, defect extent was determined by the sum of the size of all defects. A separate reversibility (or ischemia) score was similarly assigned (Table 1B). The reversibility score combined elements of defect size with the absolute change in intensity of tracer activity within the predominant defect between stress and resting conditions. All fixed or predominantly fixed defects were assigned a reversibility score of 0. Correspondingly, large defects with a moderate-to-severe change in intensity from stress to rest were assigned a reversibility score of 4.
To evaluate for the potential of the clinical information to bias the interpretation of the initial study and to evaluate the reproducibility of scan interpretation, we performed a blinded re-review of the images of a random sample of 10% of studies. Only patient size, gender, and type of test were available during the blinded re-review, performed jointly by the two nuclear cardiologists an average of four years after the initial interpretation. No studies in the re-review were excluded because of poor quality, or the presence of artifacts such as those caused by patient motion, or gut overlap. Concordance for the global score was 82% (133 of 162) with a kappa value of 0.68 (p < 0.0001). Corresponding kappa values for extent and severity and reversibility scores were both 0.4 (p < 0.0001).
Follow-up and ascertainment of end points
Patients were contacted an average of 24 ± 7 months after testing. A scripted phone interview was used to determine the clinical status of each patient including the occurrence of myocardial infarction (MI), coronary angiography, and revascularization. The interviewer was blinded to the results of MPI. If a patient could not be reached by phone, clinical status was obtained by a letter to the patient or through a review by an experienced RN or physician of the patient's office or hospital chart. In the case of death, information was obtained from family members, medical records, the Social Security Death Index, and death certificate, which was obtained for all patients who died.
Cause of death was ascertained by review of the medical record and death certificate by investigators blinded to the results of MPI. Myocardial infarctions were confirmed by death certificate or chart review requiring clinical symptoms and cardiac enzyme elevation, either CPK-MB or troponin, greater than the hospital upper limit of normal. Coronary angiography and revascularization were confirmed by chart review. To identify events that patients did not recall, admissions to the major local hospital were searched for all admissions by study patients over the study period. Finally, the databases of the catheterization laboratories of the two local hospitals performing angiography and revascularization were searched for study patients, as was the Seattle Systems Apollo database for the regional cardiac surgeons (Bellevue, Washington).
Of the 1,612 patients, follow-up was obtained in 99.81% (1,609 of 1,612 total patients). Mean follow-up of surviving patients was 24 ± 7 months, ranging from a minimum of 13 to a maximum of 36 months.
The primary end point for this analysis was cardiac death or nonfatal MI. Comparisons of categorical variables were completed using a chi-square statistic. Continuous variables were compared for the occurrence of cardiac events by one-way analysis of variance techniques. Univariable and multivariable Cox proportional hazards models were constructed to estimate cardiac death and MI. In particular, we assessed the relationship of myocardial perfusion and ventricular function data to the primary end point in unadjusted and risk-adjusted survival analysis. First-order interaction effects were tested for gender and the type of stress performed in the prognostic analyses. Subset analyses were performed using a stratified Cox proportional hazard model. For risk-adjustment, a clinical pre-test likelihood hazard score was derived using age, gender, resting normal or abnormal ECG, diabetes, hypertension, history of coronary artery disease (CAD), history of smoking, and family history of premature CAD. Survival curves were developed using an SPSS 11.0 statistical software package (Chicago, Illinois). For the survival analysis, patients undergoing coronary revascularization were included up until the date of their procedure and censored thereafter. Reproducibility was assessed by means of percent agreement and the Kappa statistic. Time to cardiac catheterization, percutaneous coronary intervention (PCI), and coronary artery bypass graft surgery (CABG) were assessed using Cox proportional hazard techniques.
Clinical characteristics and SPECT results
The average age of the patients was 65 ± 12 years, and 62% were men (Table 2). Dual isotope rest 201Tl and stress 99mTc-sestamibi imaging was performed in 85% of patients. Table 3compares the imaging results of the 171 patients who underwent revascularization within 60 days of MPI to those who did not. Imaging parameters were more abnormal in the early revascularization group for all characteristics other than resting LV size.
Overall two-year event-free survival
Over the follow-up period, 27 cardiac deaths and 62 MIs occurred, totaling 71 unique events. Unadjusted Cox death or MI-free survival at two years was 97.3%, or an annualized event rate of 1.35%.
Cox survival analysis
Elderly, male patients who were diabetic or past or current smokers had an increased rate of cardiac death or nonfatal MI (Table 4). Among patients with normal scans, event-free survival was 99.6% at one year and 99.2% at two years. All events were nonfatal MI. For those with abnormal scans, event-free survival at two years was 95.4% (p < 0.0001). Mean EF (± SD) was 58 ± 11% in those without events, compared with 51.3 ± 12% for those with events (p < 0.0001).
Event-free survival based on the extent and severity of the stress defect is shown in Figure 2A. Two-year survival for patients with extent and severity scores of 0, 1, 2 to 3, and 4 was 99.0%, 97.0%, 95.3%, and 91.7%, respectively (p < 0.0001). Similarly, event-free survival corresponding to the reversibility scores of 0 to 1, 2 to 3, and 4 was 98.6%, 96.0%, and 91.3%, respectively (p < 0.0001) (Fig. 2B).
Categorized by LV size at rest, patients with no, mild, moderate, and severe LV enlargement experienced unadjusted two-year survival rates of 97.0%, 96.7%, 92.0%, and 87.5%, respectively (p = 0.012). Categorized by TID, two-year event-free survival for those with none, mild, and moderate-to-severe TID was 97.0%, 93.8%, and 88.0%, respectively (p < 0.0001) (Fig. 3A). Including only those who underwent dual isotope imaging, those with none, mild, and moderate-to-severe TID had event-free survival rates of 99.0%, 95%, and 92.5%, respectively (p < 0.0001) (Fig. 3B). Transient ischemic dilation stratified patients who underwent either exercise stress or adenosine stress testing (p < 0.0001) (Figs. 3C and 3D). Event-free survival at two years was 99%, 96%, and 93.5% for exercising patients with none, mild, and moderate-to-severe TID, respectively. Similarly, for those undergoing adenosine stress, overall survival was 97%, 93%, and 89% for none, mild, and moderate-to-severe TID, respectively.
There was a significant interaction effect for the prognostic value of TID and the type of stress performed. That is, the relative risk of death or infarction in patients with TID was greatest when it occurred with adenosine testing with low-level exercise (relative risk = 2.23-fold, 95% confidence interval [CI] = 1.5 to 3.4, p < 0.0001), compared with the borderline increase in risk seen when it occurred with adenosine testing without exercise (relative risk = 1.9, 95% CI = 0.98 to 3.7, p = 0.055).
Event-free survival was directly related to post-stress EF (Fig. 4A), such that event-free survival decreased from 99.0% to 87.0% with a decreasing post-stress EF of 70% to 20% (p < 0.0001). Integrating function and perfusion, overall event-free survival for patients in those with an EF above or below 40%, with and without inducible ischemia, is shown in Figure 4B and 4C. For those whose EF was ≥40%, the two-year event-free survival rate worsened from 97.8% to 92.0% with increasing ischemia. For those patients with an EF <40%, survival worsened from 94.2% to 80.4% with increasing ischemia. Similarly, in a stratified analysis of patients with an EF ≥50%, event-free survival was 98.5% in those with a reversibility score of 0 to 1 compared with 93.7% in those with a reversibility score of 4 (p = 0.012). Among those with an EF <50%, event-free survival worsened from 95.5% in those with reversibility scores of 0 to 1 compared with 86% in those with a score of 4 (p = 0.012). When analyzing patients by post-stress EF deciles of 40% to 49% down to <20%, event-free survival ranged from 95.0% to 89.7% in those with ischemia (p < 0.0001) compared with 97.8% to 91% in those without ischemia (p < 0.0001).
Cox multivariable regression model
After adjusting for age, gender, resting ECG, prior known CAD, cigarette smoking, hypertension, hypercholesterolemia, diabetes, and family history of CAD, the most important independent prognostic variables, in order of importance, were: TID, extent of reversibility, post-stress EF, extent and severity of the stress perfusion defect, and the global score (Table 5). For TID, the relative risk increased 3.3-fold for patients with moderate-to-severe TID and 1.6-fold for patients with mild TID (p = 0.0002). For post-stress EF, each 1% decrement in EF represented a 3% increase in risk (p = 0.0009).
As detailed in Table 6, MPI predicted outcome in the following key subsets: women and men, diabetics, patients with known or suspected CAD, patients referred for chest pain, those with prior abnormal exercise ECGs, and those with normal or abnormal resting ECGs. Among diabetics with abnormal studies, women experienced a greater relative risk than men, 5.2 versus 2.7.
We also compared the event-free survival rates for those undergoing exercise and adenosine stress testing. For those undergoing exercise stress, normal and abnormal scans were associated with annualized event rates of 0.4% and 2.2%, respectively (p < 0.0001). For those undergoing adenosine stress without low-level exercise, annualized event rates were 0.9% and 3.3% for those with normal and abnormal scans, respectively (p < 0.0001). For those undergoing adenosine combined with low-level exercise (Adenoex), normal and abnormal studies were associated with annualized event rates of 1.1% and 4.7%, respectively (p < 0.0001). In fact, a test for an interaction effect was significant (chi-squared = 22, p < 0.0001), such that the overall relative risk increased 3.91-fold (95% CI = 2.3 to 6.7) for an abnormal scan in patients who underwent Adenoex, while the relative risk was only 3.2-fold higher (95% CI = 1.6 to 6.4) for an abnormal test result when adenosine was infused without exercise.
Post-test use of cardiac catheterization and coronary revascularization
During the study period, 499 patients (31%) underwent coronary angiography, 252 (16%) underwent PCI, and 105 (7%) underwent CABG (Table 7). Catheterization was rarely performed in patients with normal scans: 2.0% at 90 days and 3.0% at the end of two years (p < 0.0001). Catheterization increased with increasing ischemia, with 47.0% of those with reversibility scores of 4 undergoing angiography within 90 days and 67.0% within two years (p < 0.0001).
Coronary revascularization increased linearly with extent of ischemia. Among those patients with normal scans, PCI was performed in only 0.3% during the first 90 days and 1.0% by the end of two years (p < 0.0001), while CABG was performed in 0.4% and 0.7% over the same time intervals (p < 0.001). The highest rates of PCI (37% at two years) and CABG (14.0% at two years) were seen in those with the most ischemia, those with a reversibility score of 4. As seen with reversibility, catheterization, PCI, and CABG increased with increasing extent and severity of the stress defect and with decreasing post-stress EF (p < 0.0001) (Table 7).
It is well established that MPI performed with 201Tl (3), 99mTc-sestamibi (5,6), and 99mTc-tetrofosmin (7)provides important incremental prognostic information. Our study is unique in that MPI was performed in the community, in a physician's office setting, while each of the previous studies had been carried out in an academically based nuclear cardiology/medicine program, with all except that of O'Keefe et al. (12)being hospital based.
The 0.4% annual hard event rate in patients with normal MPI found in this study parallels the rates of <1% observed in previous studies (5,13–15). This event rate in patients with normal scans is lower than the rate for the general population of the U.S. (5,16). The event rate for patients with abnormal scans in our study, 2.3% per year, is lower than observed in previous investigations. Corresponding annual event rates in patients with abnormal scans were 3.5% to 7.1%, 5.4%, and 7.4% in studies by Hachamovitch et al. (13), Boyne et al. (14), and Iskander et al. (5), respectively. The difference is probably due to: 1) the inclusion of inpatients in the previous studies; 2) the increasing use of revascularization manifested by a 10.6% rate of revascularization within the 60 days of MPI in our contemporary study compared with 6.8% among patients in the earlier Hachamovitch et al. study (13); 3) the ongoing decrease in age-specific cardiac mortality occurring in the U.S. (17); and 4) the increasing use of MPI that may have resulted in lower risk individuals making up a greater percentage of patients.
After adjusting for clinical factors, the most powerful components of MPI, in decreasing level of prognostic significance, were: TID, defect reversibility, post-stress EF, extent and severity of the stress defect, and the global score.
The finding that TID adds powerful prognostic information to dual isotope imaging, though not previously demonstrated, is welcome given its long-established prognostic value with 201Tl imaging (18)and the more recent demonstration of its prediction of severe and extensive CAD with dual isotope imaging (19). Challenges in the evaluation of TID include the differing physical characteristics of 99mTc-sestamibi and 201Tl and the necessary delay in 99mTc-sestamibi imaging after stress. Despite these issues, we found that increasing degrees of TID predicted increasing risk, and this was consistent across both exercise and adenosine stress. As has been previously demonstrated (6), we also found that increasing degrees of defect reversibility and increasing extent and severity of the stress defect provided substantial incremental prognostic value.
The availability of 99mTc perfusion agents provides the opportunity to gate images and, thus, obtain a post-stress EF, leading several authors to predict that this would offer additional incremental prognostic value (6,20). This potential has only recently been evaluated. The work of Sharir et al. (21)demonstrated that a gated SPECT post-stress EF of <45% was an independent predictor of cardiac events. In a follow-up study of 2,686 patients, Sharir et al. (22)separately evaluated the predictors of cardiac death and nonfatal MI, finding that gated SPECT post-stress EF best predicted cardiac death, while ischemia best predicted nonfatal MI. Kroll et al. (23)studied 120 inpatients after MI who were treated medically and found that predischarge post-stress gated SPECT EF provided incremental prognostic benefit among patients with an EF of <40%. Our study confirms and extends these findings; in particular, a 3% increase in risk was observed for every 1% decrease in EF (p = 0.0009). In addition, we found that, despite adjustment for EF, ischemia continued to provide incremental prognostic information.
As typically occurs in the interpretation of nuclear cardiology images, some clinical information was available to the interpreting physicians for many of the patients in this study. To evaluate the potential influence of this information on scan interpretation, a re-read of 10% of the studies was performed with the interpreters blinded to clinical data. Agreement between the initial interpretation of the global score and the blinded re-read four years later was good, with 82% concordance (K = 0.68). This is comparable, though slightly less, than the concordance rates in the intraobserver reproducibility studies of Danias et al. (24), 83% to 88% (K = 0.54 to 0.73), and Golub et al. (25), 88% to 93% (K = 0.76 to 0.87). Although the impact of clinical information could have resulted in the small difference in reproducibility between the studies, other factors also are likely to have played a role. The images interpreted for the study of Danias et al. (24)were performed with the benefit of motion correction software, while such software was not available in the current study. In the study of Golub et al. (25), only studies of “good-to-excellent” quality were selected for inclusion. As well, Golub et al. (25)included only patients undergoing exercise stress, 16% of whom were women. In the current study, a random sample of all studies was interpreted in the blinded re-read, including those of suboptimal quality. A total of 37% were performed with pharmacologic stress, with its increased potential for significant overlap of gastrointestinal contents, and 38% were performed in women, in whom artifacts would be expected to be more frequent. Given the results of our blinded re-review, the impact of clinical data was probably small.
We observed, as previously described, that patients referred for adenosine stress are at higher risk for events than those referred for exercise stress and that MPI can effectively stratify these patients (26–28). In contrast with others, however, we had the opportunity to specifically evaluate the prognostic value of the addition of low-level treadmill exercise to adenosine. Analogous to simultaneous semi-supine bicycle exercise pioneered by Pennell et al. (29), the protocol has been found to be safe, to improve image quality, and to decrease hypotensive and arrhythmic side effects (10,29,30,31). More recently, Samady et al. (32)demonstrated an increase in the extent and severity of reversible defects when simultaneous exercise at the first two stages of the modified Bruce protocol was combined with the standard adenosine infusion of 140 μg/kg/min, compared with an adenosine infusion alone.
Annualized event rates for patients undergoing adenosine infusion without simultaneous exercise in our study were 0.9% and 3.3% for those with normal and abnormal perfusion scans, respectively (p < 0.0001), compared with 1.1% and 4.7% for those undergoing Adenoex (p < 0.0001). The difference in relative risks, 3.2 for adenosine alone and 3.91 for the combination, is significant (p < 0.001), raising the potential that the combination provides greater prognostic value than adenosine alone. An alternative explanation, however, is that the difference could have been caused by unaccounted for differences in the populations undergoing each test.
The presence of TID occurring with Adenoex also provided further prognostic value compared with adenosine alone, yielding relative risks of 2.23 for the combination and 1.9 for adenosine alone (p < 0.001). These findings add further evidence to the thesis that adenosine testing can be enhanced by the addition of low-level exercise.
Our work provides further evidence that diabetes itself is an incremental risk factor for cardiac events even after allowing for the results of perfusion and function imaging (28,33,34). However, as others have demonstrated, MPI allowed substantial stratification among diabetics with annualized event rates of 0.9% and 5.8% in patients with normal and abnormal scans, respectively (p < 0.0001). This stratification is concordant to that found by Kang et al. (33)who observed an annualized event rate in diabetics with normal and abnormal MPI of 1.2% and 4% to 8%, respectively, and that of Giri et al. (34).
Normal resting ECG
Myocardial perfusion imaging has typically been reserved for patients with abnormal resting or exercise ECGs or those requiring pharmacologic stress (8). Hachamovitch et al. (35)recently demonstrated that MPI provided incremental prognostic information in 3,058 consecutive patients with normal resting ECGs. They observed an annual hard event rate of 0.4% among patients with normal perfusion compared with 2.8% to 3.9% for those with abnormal perfusion. We confirmed these findings in a different cohort of patients with normal resting ECGs with annualized event rates of 0.3% and 1.7% for those with normal and abnormal scans, respectively (p < 0.0001).
Women, constituting 38% of our cohort with an average age of 66 ± 11 compared with 64 ± 12 years for men, experienced lower annualized event rates than men but were indeed stratified by perfusion imaging; women with normal and abnormal SPECT scans had event rates of 0.4% and 1.8%, respectively (p < 0.0001). The effectiveness of 99mTc-based MPI in women is consistent with a large earlier study of 1,394 women (36). The studies differed, however, in the substantially higher event rates experienced among women with abnormal SPECT scans in the previous study (11.5%). The higher event rate observed in their cohort may be explained by the inclusion of inpatients, the aforementioned difference in time periods relative to less frequent referral to revascularization (4.3% for women in the previous study vs. 7.8% in our study), overall improvements in cardiac care, and the increase in MPI usage.
Resource utilization after MPI
Although this study was conducted in a physician's office environment in which economic incentives for self-referral to coronary angiography would be expected to exist, clinicians effectively used MPI as a “gatekeeper” to catheterization. The rate of angiography for patients with normal studies, 2% at 90 days, is similar to previous works in the academic setting. This low rate held over two years with only an additional 1% of patients undergoing catheterization by that time. This is consistent with the only other study performed in what could be regarded as a private practice setting, that of Bateman et al. (37), in which the rate was similarly low: 3.5% over 8.9 months for patients with scans that were either normal, equivocal, or had a nonreversible defect. We found, as have others (4,28,38), that frequency of angiography also increased in proportion to the amount of reversibility. In addition, Shaw and others (4,28)have also demonstrated that referral to revascularization increased with increasing ischemia. We confirmed this finding but also found revascularization to increase proportionally to two other components of MPI found to add incremental prognostic risk in our multivariate analysis, namely the extent and severity of the stress defect and decreasing post-stress EF. Although these latter findings are not unexpected, they have not been previously demonstrated and are consistent for the utilization of both PCI and CABG, the frequency of each increasing linearly with increasing reversibility, increasing extent and severity of the stress defect, and with decreasing EF.
Health policy implications
Levin et al. (39)recently challenged the justification for both the increasing rate of referral of patients for MPI and the increasingly frequent add-on measurement of EF by gated SPECT. Although our study did not evaluate the appropriateness of referral to MPI, it found the post-test use of MPI to be appropriate. The low 2% rate of referral to catheterization within 90 days and the appropriately increasing rate of referral to PCI and CABG with increasing risk based on the results of MPI are reassuring. We also found that the use of post-stress EF added incremental prognostic information beyond that obtained by perfusion alone.
The availability and use of clinical information by the interpreting cardiologist could have affected the interpretation of the MPI exam and, thus, its prognostic significance. However, the high concordance and kappa value between the initial clinical global score interpretation and the blinded re-review four years later suggest that any impact this may have had was small. As well, the scoring of perfusion defect extent and severity and reversibility was performed retrospectively from the image reports rather than the images themselves. The use of a clinically validated scoring system in real time would have been preferred (9). Although the evaluation of TID was performed in real time, it was performed visually rather than quantitatively. The use of quantitative software for the evaluation of LV volumes post stress and at rest may have enhanced the assessment and reproducibility of TID. In addition, in the evaluation of TID in the multivariable analysis, both single and dual isotope studies were included in the evaluation of the relative risk of TID.
Because the electrographic response to exercise or pharmacologic stress was not prospectively catalogued, we were unable to evaluate the value of the exercise ECG. However, the resting ECG was abnormal in 65% of patients and 34% of patients underwent adenosine stress, it is therefore unlikely that this information would substantially impact the results.
The use of a brief, second non-gated post-stress scan in 27% of our patients is atypical of community-based laboratories. Although currently available motion correction software (40)generally obviates the need to rescan for patient motion, this does not provide further evaluation of inferior defects.
In sum, 99mTc-sestamibi-based MPI is transferable to community-based laboratories, and multiple components of perfusion imaging provide incremental prognostic information after multivariate adjustment for clinical factors. These components include TID, defect reversibility, post-stress EF, extent and severity of the stress defect, and the global score. The incremental value of MPI is consistent across all subgroups evaluated, including women, diabetics, patients with normal resting ECGs, and those referred for adenosine imaging, both with and without simultaneous exercise. Use of post-test coronary angiography, PCI, and CABG increased commensurate with increased prognostic risk as assessed by MPI.
☆ Partial grant support supplied by Bristol-Myers-Squibb Medical Imaging and Fujisawa Healthcare.
Presented, in part, at the 2002 Annual Scientific Session of the American College of Cardiology.
- coronary artery bypass graft
- coronary artery disease
- electrocardiogram/electrocardiographic/ electrocardiography
- ejection fraction
- left ventricular
- myocardial infarction
- myocardial perfusion imaging
- percutaneous coronary intervention
- single photon emission computed tomography
- transient ischemic dilation
- Received May 22, 2003.
- Revision received July 10, 2003.
- Accepted July 22, 2003.
- American College of Cardiology Foundation
- ↵Nuclear Medicine Market Summary Report 2001–02. Des Plaines, IL: IMV Medical Information Division
- Nuclear Medicine Market Summary Report 1991–92. Des Plaines, IL: Technology Marketing Group
- Brown K.A.
- Iskander S.,
- Iskandrian A.E.
- Beller G.A.,
- Zaret B.L.
- Gibbons R.J.,
- Balady G.J.,
- Beasley J.W.,
- et al.
- ↵American Society of Nuclear Cardiology. Imaging guidelines for nuclear cardiology procedures, part 2. J Nucl Cardiol 1999;6:G47–84
- ↵Lear J, Pratt J, Roberts D, et al. Gamma camera image acquisition, display and processing with the personal microcomputer. Radiology 1990;175:241–5
- Hachamovitch R.,
- Berman D.S.,
- Shaw L.J.,
- et al.
- United States Life Tables, 1999. National Vital Statistics Reports 2002;6:1–8
- ↵Braunwald E. Preface. In: Braunwald E, editor. Heart Disease: A Textbook of Cardiovascular Disease. 5th ed. Philadelphia, PA: W.B. Saunders Company, 1997:xvii
- Weiss A.T.,
- Berman D.S.,
- Lew A.S.,
- et al.
- Mazzanti M.,
- Germano G.,
- Kiat H.,
- et al.
- Sharir T.,
- Germano G.,
- Kang X.,
- et al.
- Sharir T.,
- Germano G.,
- Kavanagh P.B.,
- et al.
- Danias P.G.,
- Ahlberg A.,
- Travin M.I.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Berman D.S.,
- Kang X.,
- Hayes S.W.,
- et al.
- Pennell D.J.,
- Mavrogeni S.I.,
- Forbat S.M.,
- et al.
- Giri S.,
- Shaw L.J.,
- Murthy D.R.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Matsumoto N.,
- Berman D.S.,
- Kavanagh P.B.,
- et al.