++
The principles underlying the efficient use of stress nuclear techniques and the optimal use of the test results are discussed below. The most common applications are identifying inducible ischemia in patients with suspected disease, assessing the likelihood that a patient with known CAD has ischemia, evaluating patients prior to noncardiac interventions, and assessing the magnitude of ischemia for prognostic purposes. The section that follows explores these applications and discusses the evidence for the use of SPECT and PET-MPI in specific patient populations commonly encountered in clinical cardiology settings. Evidence for SPECT is emphasized below, because PET is covered in Chapter 19.
+++
Selecting the Appropriate Patients for Diagnosis of Obstructive CAD Using Stress Nuclear MPI
++
Central to appropriate patient selection for nuclear imaging and the interpretation of test results for purposes of establishing the diagnosis of obstructive CAD is the assessment of an individual patient’s pretest likelihood of CAD based on demographic, clinical, and historical information. Bayes’ theorem underlies the selection of SPECT- or PET-MPI for diagnostic testing (see Chap. 19). For a given test result, the post-test likelihood is a function of three variables: patient pretest likelihood of disease and the test sensitivity and specificity. The degree to which the test result alters the post-test likelihood is directly affected by the pretest likelihood of disease. For any test with less than 100% sensitivity and specificity, the greatest shift in post-test likelihood of disease occurs in patients with an intermediate pretest likelihood of CAD. As a benchmark, for a test with sensitivity and specificity of 90%, a patient with a pretest likelihood of 50%139 will have a post-test likelihood of 10% (low) with a normal test result and a 90% likelihood (high) with an abnormal test result. For purposes of assessing the post-test likelihood of CAD when using SPECT- or PET-MPI, the patient’s pretest likelihood takes into account all available information, including age, sex, symptoms, risk factors, the degree of coronary atherosclerosis if known (eg, from a CAC score), and the results of the non-nuclear stress testing components of the examination (eg, the duration of exercise and degree of ST-segment depression).
++
The pretest likelihood or risk assessment is conventionally estimated from published nomograms or by using available computerized programs. Recent data, however, suggest that the commonly used approaches to assessing the pretest likelihood of CAD, based on the work of Diamond and Forrester,139 may not be applicable in the types of patients currently being referred for noninvasive testing. From a contemporary series, Cheng and coworkers reported that the Diamond-Forrester criteria markedly overestimated pretest likelihood of CAD in patients referred for coronary computed tomography angiography (CCTA).140 In 8106 patients from the COronary CT Angiography EvaluatioN For Clinical Outcomes: An InteRnational Multicenter (CONFIRM) registry with nonanginal angina, atypical angina, or typical angina, the Diamond-Forrester pretest likelihood of angiographically significant CAD was 51%; however, the observed frequency of 50% or more stenosis on C CTA in these patients was 18%. This overestimation of CAD likelihood was confirmed in the recent PROMISE trial. In the patients in the coronary CT arm of the trial, the pretest likelihood of obstructive CAD was 53% while at least 50% stenosis on C CTA was observed in only 10.7%. As noted below, the overestimation of the pretest likelihood of obstructive CAD is likely a principal factor in the recently observed low frequency of abnormal SPECT-MPI studies.141
+++
Diagnostic Accuracy of SPECT and PET-MPI
++
The accuracy of diagnostic testing for CAD has been defined on the basis of sensitivity and specificity as compared to an anatomic stenosis standard of either a 50% or 70% diameter-narrowing determined by invasive coronary angiography. A large meta-analysis of 86 studies comprising 10,870 patients and examined the diagnostic accuracy of exercise or pharmacologic stress MPI for detecting significant CAD.142 Pooling studies published between January 2002 and October 2009, the authors found that SPECT performed without attenuation correction or gating (63 studies) had a sensitivity of 87% and a specificity of 70%. The addition of gating information increased specificity to 78% (19 studies), and the use of attenuation correction further increased specificity to 81% (12 studies).
++
Several meta-analyses have compared the sensitivity and specificity of SPECT versus PET-MPI techniques. McArdle and colleagues reported on a comparison of the two modalities limiting the SPECT studies to those that used attenuation correction and gating, thus allowing for comparison of contemporary techniques.143 Pooling 15 PET and 8 SPECT studies (1344 and 1755 patients), the authors reported a pooled sensitivity of 90% (confidence interval [CI], 0.88-0.92) and specificity of 88% (CI, 0.85-0.91) for PET and sensitivity of 85% (CI, 0.82-0.87) and specificity of 85% (CI, 0.82-0.87) for SPECT. The areas under the summary receiver-operating characteristic curves were 0.95 and 0.90 for PET and SPECT (P < .0001), respectively. This study was limited by interstudy heterogeneity and likely the differential referral patterns for each modality. Parker and coworkers performed a bivariate meta-analysis to compare the sensitivity and specificity of PET versus SPECT stress MPI using a 50% stenosis or more threshold to define obstructive CAD.144 The authors identified 117 studies, including 108 evaluating SPECT-MPI, 4 evaluating PET-MPI, and 5 evaluating both modalities. A significantly higher pooled mean sensitivity was found with PET (93% [95% confidence interval or CI, 88%-96%]) compared with SPECT (88%) [95% CI, 86%-90%] (P = .035). No significant difference in specificity was present between PET (81% [95% CI, 67%-90%]) and SPECT (76% [95% CI, 72%-79%]) (P = .39). In a multicenter European cohort of 475 symptomatic patients undergoing CCTA and nuclear MPI (Evaluation of Integrated CAD Imaging in Ischemic Heart Disease [EVINCI]) trial, the diagnostic accuracy of detection of obstructive CAD was higher for PET than for SPECT-MPI.145 For PET-MPI, the diagnostic accuracy, sensitivity, and specificity were 85%, 81%, and 89%, respectively. By comparison, for SPECT-MPI, the diagnostic accuracy, sensitivity, and specificity were 70%, 73%, and 67%, respectively. Given the physiological basis of these tests, a more relevant comparison would be based on FFR by invasive angiography as a gold standard. As part of a comparison of multiple modalities (SPECT, PET, stress echocardiography, CT, or magnetic resonance [MR]) to invasive FFR 37, studies were identified (n = 2048 patients).146 Using FFR as a gold standard, PET had a lower negative likelihood ratio compared to SPECT (PET 0.14 [95% CI, 0.02-0.87]; SPECT 0.39 [95% CI, 0.27-0.55]) with a greater positive likelihood ratio (PET 7.43 [95% CI, 5.03-10.99]; SPECT 3.76 [95% CI, 2.74-5.16]). PET, MR, and CT perfusion performed similarly in excluding abnormal FFR, whereas MR and PET had similar positive likelihood ratios that were greater than that of CT. SPECT and stress echocardiography were inferior to the three other modalities with respect to both metrics.146
+++
Referral or Partial-Verification Bias
++
A major limitation in assessing the diagnostic accuracy of nuclear MPI to detect CAD is that the decision to perform the gold standard test—catheter-based coronary angiography—after nuclear testing is strongly influenced by the MPI result, thereby biasing the population available for the analysis of test accuracy. This is referred to as past test referral bias or verification bias. The referral to invasive angiography is largely driven by the presence of ischemia, particularly the extent and severity of perfusion abnormalities, as well as anginal symptoms, ECG changes, and other clinical factors. This referral pattern results in an overestimation of test sensitivity and a reduction in test specificity, with the most dramatic change occurring in specificity.147 Importantly, only a small proportion of patients with normal or low-risk MPI findings proceed to invasive coronary angiography, resulting in more variability in specificity measures than in sensitivity. In the extreme case—catheterization in all abnormal MPI but none in normal MPI—test sensitivity and specificity would approach 100% and 0%, respectively. For example, Miller and colleagues reported that in a large series from the Mayo Clinic that the uncorrected SPECT specificity was less than 10% in men, because the referral for invasive angiography in patients having SPECT-MPI is strongly affected by the SPECT results.148 Adjustments for referral bias have been described.
++
The normalcy rate has been suggested as surrogate for assessing test specificity without requiring the angiographic standard; thus, it is not being affected by the marked post-test referral bias associated with assessment of specificity based on invasive coronary angiographic stenosis.147 The normalcy rate is defined as the percentage of patients with normal test results in a population with a low (< 10%) pretest risk of CAD. The 2003 American College of Cardiology (ACC)/AHA/ASNC guidelines reported a normalcy rate for SPECT-MPI of 91%. Even in obese patients, normalcy rates for SPECT-MPI of greater than 90% have been reported when attenuation correction or combined supine/prone imaging is used.22,149 This measure of “specificity” has limitations that the individuals being tested—generally those who would not be appropriate for testing—are much “healthier” than the patients. Potentially fewer artifacts might occur in these patients (eg, they might be less likely to move during the study), and they are likely to have higher quality images (eg, greater coronary blood flow leading to greater radiopharmaceutical uptake). Thus, normalcy rate is likely to overestimate true test specificity.
+++
Risk Assessment of the Patient with Known or Suspected CAD
+++
Principles of Risk Stratification
++
The first of the prognostic studies concerning exercise MPI was published by Brown and coworkers in 1983, based on a small cohort of only 100 patients.150 In 1986, Ladenheim and colleagues reported a landmark study involving the follow-up of 1689 diagnostic patients undergoing exercise 201Tl MPI.151 This approach of prognostic accuracy of stress MPI overcame many of the limitations of diagnostic accuracy that have been previously discussed, improving separation of low- to high-risk patient subsets. Both the number of perfusion defects and the severity of perfusion defects were shown to be related to adverse cardiac outcomes in an exponential fashion, with extent and severity of ischemia as independent predictors of outcome (Fig. 18–17). When PET-MPI is used, measurements of myocardial flow have been shown to add information over perfusion defects.
++
++
These early results led to a risk-based approach for the application of MPI to patients with known or suspected CAD with stable symptoms. In a risk-based approach, the focus is not on predicting the presence of CAD but on identifying patients at risk for specific, potentially preventable adverse events. Subsequent management then focuses on reducing the risk of these outcomes, whether cardiac death, nonfatal myocardial infarction (MI), or CAD progression. A major goal of management of stable ischemic heart disease (SIHD) patients following stress MPI is adherence and, as needed, intensification of guideline-directed medical therapy.152 For the high-risk, invasive diagnostic and therapeutic procedures are limited to patients who are most likely to benefit from them. This approach has been extensively evaluated using registry data for both SPECT and PET-MPI.
++
For the purposes of risk assessment, it has been proposed that low risk be defined as a less than 1% annual cardiac mortality and intermediate risk be defined by the range of 1% to 3% per year.153 Because the mortality risk for patients undergoing revascularization is 1% or more (see Chaps. 20 and 44), symptomatic patients with a less than 1% annual mortality risk would not appear to be candidates for revascularization to improve survival. The values of these risk thresholds vary according to the population being tested.
+++
Risk Stratification in Patients with Suspected or Known CAD
+++
Incremental Prognostic Value
++
The clinical value of SPECT- or PET-MPI for prognostic assessment of CAD results from the incremental or added prognostic information yielded overall data available prior to the test (clinical, historical, and stress data), as first demonstrated by Ladenheim and colleagues.154 The incremental prognostic value of SPECT- and PET-MPI has since been documented extensively, including a broad spectrum of patient subsets.
+++
Event Risk after a Normal Scan
++
Extensive literature exists documenting the risk of adverse events following a normal stress SPECT-MPI. A normal scan is generally associated with a less than 1% annual risk of cardiac death or MI, but in higher risk cohorts this risk can be about 2% per year or higher. A meta-analysis of the prognostic value of a normal stress perfusion scan (n = 29,788) revealed that the annual risk of MI or cardiac death after a normal stress MPI is 0.5% (95% CI, 0.3%-0.7%).155 This low event rate is critical in applying nuclear test information to risk stratification, because in the absence of significant or limiting symptoms, it implies that patients with normal perfusion scans can be managed conservatively. Such an approach requires careful follow-up for signs of clinical worsening and treatment of cardiac risk factors and related symptoms (see Chap. 32). Increasingly, many studies are using all-cause death as an end point, and the degree of comorbidity impacts the annual mortality rates. Thus, the all-cause mortality rates following a normal MPI are expectedly higher than that of cardiac-specific end points.156
++
The risk of cardiac events is consistently lower in patients with normal, than in those with abnormal SPECT-MPI, but there are multiple subsets of patients with a normal scan in which the absolute risk is not low. There is a strong temporal component as well in the assessment of risk. A study examining predictors of risk and its temporal characteristics in a series of 7376 patients with normal stress SPECT-MPI identified the following as markers of increased risk and shortened time to a hard event: use of pharmacologic stress, known CAD (Fig. 18–18A), diabetes mellitus (in particular, female diabetics), and advanced age.24 Hence, a dynamic temporal component of risk was present, and the existence of a warranty period for specific patient groups was defined (Fig. 18–18B). The warranty period of a normal scan was further examined by Acampa and colleagues, who documented the importance of diabetes and reduced LVEF (< 45%) in identifying those with a slightly higher CAD event risk and shorter warranty period.157 Increased risk after normal SPECT has also been reported in patients with dyspnea as the presenting symptom.158 The importance of exercise duration has also been reported. In 6069 patients with a normal SPECT-MPI study followed for a mean of 10.2 years, the all-cause mortality rate rose progressively as the exercise duration associated with the SPECT study declined. Patients who could not exercise for more than 3 minutes had a mortality that was comparable to that of patients requiring pharmacologic stress.59 In a subsequent study, of 12,232 patients with a normal exercise SPECT-MPI study followed for a mean of 11.2 years, hypertension, smoking, diabetes, and exercise capacity were significant predictors, beyond age, of long-term mortality (Fig. 18–19). Combining these predictors, the mortality ranged from 0.2% per year in patients exercising 9 minutes or more and with none of these CAD risk factors to 1.6% per year in patients exercising less than 6 minutes and having two or more of these risk factors.159 Thus, the lower risk and more functional the patient, the lower the expected annual event rates following a normal MPI. The converse is also true—that the annual event rates are increased in higher risk cohorts, including those of advanced age or with known cardiovascular disease, impaired LVEF, and/or functional impairment. By comparison, patients with normal SPECT-MPI who performed some exercise at the time of stress MPI (combined low-level exercise with pharmacologic stress) have been reported to have event rates that are intermediate, between those who underwent exercise SPECT-MPI and those who had pharmacologic SPECT-MPI without exercise.160
++
++
++
The contextual nature of risk requires some consideration of expected population mortality and cardiac event risk. As an example, from a study of elderly patients aged 75 to 84 years and 85 years or older, the annual cardiac mortality was 1.0% and 3.3%, respectively. Although these rates are not low, they are lower than the expected cardiac mortality for individuals of these ages in the US population (1.5% and 4.8%, respectively; P < .05 vs normal SPECT-MPI patients).161
++
The importance of stress-only SPECT protocols has been emphasized in the preceding portions of this chapter. The low risk associated with a normal rest/stress SPECT-MPI has been reported in patients studied with the stress-only protocols. In the largest study to date, Chang and coworkers reported all-cause mortality rates over a 4.5-year median follow-up in 16,854 consecutive patients with a normal gated stress SPECT (stress-only protocol in 8034 patients, both stress and rest imaging in 8820 patients),93 finding lower annualized unadjusted mortality after a stress-only protocol compared to a stress-rest protocol (2.6% vs 2.9%; P = .02). The very low risk associated with a stress-only protocol has been confirmed and extended by other authors.92,162,163
++
In some patients, extensive CAD may be missed because of balanced reduction of flow.164 The latter would lead to a severe underestimation of the extent of ischemia by SPECT-MPI.165 However, many of the patients with normal SPECT-MPI found subsequently to have high-risk anatomic lesions had abnormal ancillary markers on their SPECT study (eg, TID,129 a rest-to-poststress decrease in LVEF, increased lung tracer uptake, or small perfusion defect below threshold to be considered abnormal).165 A recent study of 580 patients with normal or minimal perfusion abnormalities on stress SPECT-MPI (SSS < 4), who underwent invasive angiography within 2 months after SPECT, reported that the finding of TID, abnormal EF response, or SSS of more than 0 were predictors of the small proportion (7.2%) who were found to have high-risk angiographic findings. Use of a CAC scan can be very helpful in identifying underlying atherosclerotic disease risk, as discussed later in the chapter. When the CAC score is high (400 Agatston units or higher), a normal stress MPI would not exclude significant CAD, as the score would indicate a high likelihood of obstructive CAD. Kaminek and colleagues reported that, in patients with known CAD, a combined approach of SPECT-MPI perfusion-function assessment and CAC improves the identification of multivessel CAD.166 In general, when nonperfusion abnormalities are present in patients with otherwise normal SPECT-MPI, they are by themselves insufficient to warrant proceeding to invasive coronary angiography. PET-MPI, through the assessments of absolute rest and stress myocardial blood flow and coronary flow reserve (CFR), can improve risk detection and a greater understanding of the burden of obstructive CAD, functionally limiting mild epicardial CAD, or coronary microvascular dysfunction.167
+++
Event Risk with a Borderline/Equivocal Scan
++
As noted in the interpretation section, in some patients semiquantitative visual scoring or quantitative analysis results in a study being neither completely normal (percent myocardium abnormal = 0) nor definitely abnormal (percent myocardium abnormal ≥ 5%). In these circumstances, the scan is interpreted as “borderline” or “equivocal.” The equivocal SPECT-MPI study is associated with an increased risk compared to the normal scan. In a study of 18,200 patients followed for a mean of 2.7 years for cardiac death, a significant increase in risk was found in patients with equivocal stress SPECT scans (P < .001) (Fig. 18–20).102 In a recent study of prognostic assessment using a CZT camera, significantly increased risk of cardiac events was noted in patients with hypoperfusion below the threshold to be considered abnormal.168 In a recent report of a subset of a large database studying patients with angiographically identified high-risk obstructive CAD following a borderline/equivocal SPECT-MPI study, the presence of a score greater than 0 (but lower than the abnormal threshold of ≥ 5%) was a variable associated with a higher than expected event rate.169 Thus, for purposes of risk assessment, the scan category of “equivocal” or “borderline” should not be interpreted as being completely normal. Nonetheless, the “equivocal” or “borderline” scan result should not by itself prompt referral to invasive coronary angiography, because a preponderance of patients in this scan category do not have significant CAD.
++
+++
Event Risk with Abnormal Scan
++
The relationship of varying extent and severity of perfusion abnormalities with cardiac outcomes has been reported in numerous, diverse patient subsets.103,153,170,171 Consistently, increasing perfusion scan abnormality is associated with an increasing risk of cardiac events, and this risk is further increased in patients with reduced LVEF. The severity of perfusion defects correlates with the degree of stenosis, and extent of perfusion defect correlates with the amount of myocardium subtended by the stenosed vessels.151 As described above, the percent myocardium abnormal, as estimated by 17-segment scoring or quantitative assessments such as TPD, combines the assessments of both extent and severity of perfusion abnormality, and the general categories of abnormality have been described.25,27,101 Illustration of how these parameters affect risk in various patient subsets is shown in Fig. 18–21. Annual cardiac event rates generally range from 0.7% to 6.7% for patients with normal, mild, moderate, and severely abnormal perfusion scans, with significant variability associated with each level of defect extent and severity; depending on the pretest likelihood of CAD.25 Findings similar to those observed with 99mTc-sestamibi have been described with 201Tl and 99mTc-tetrofosmin SPECT as well as in a multicenter PET registry.155,172
++
+++
Importance of Prescan Data
++
As noted, prescan information adds incrementally to SPECT data for predicting outcomes. For all groups of patients, risk assessment in an individual patient is improved by taking into account findings other than those of the scan. The presence of high-risk clinical or historical markers identifies a subset of patients at greater risk for any level of scan abnormality (see Fig. 18–21).25,101,173,174 Additionally, for any degree of perfusion defect, the risk of adverse events after SPECT-MPI is greater in patients with higher clinical risk, those with including atrial fibrillation,174 diabetes,175,176 advanced age,161 reduced LVEF, dyspnea,158,177 high CAC scores, known CAD, reduced functional capacity, and inability to exercise.
+++
Mildly Abnormal SPECT-MPI
++
The presence of a mildly abnormal scan (5-9% myocardium) usually implies the presence of CAD but a moderately elevated risk of cardiac event. However, the risk is higher in a variety of subgroups with significant comorbidities and presentations, as mentioned above. The need to take into account all risk predictors is of particular importance in this patient group. A highly important consideration is that with SPECT-MPI or PET-MPI (unless absolute blood flow is being assessed), detection of abnormality is spatially relative; that is, perfusion defects are detected by their comparison to other myocardial regions. Thus, SPECT or PET-MPI without absolute flow measurements may underestimate the extent of hemodynamically significant CAD. Although patients with mildly abnormal SPECT-MPI results have an elevation in cardiac event risk, generally optimal medical therapy approaches are the mainstay in this patient subset, with referral to invasive angiography limited to patients with refractory symptoms.
+++
Moderately to Severely Abnormal SPECT-MPI
++
As discussed earlier, this category of scan abnormality is associated with high levels of patient risk and even greater in patients with high-risk cardiovascular comorbidities, increased LV size/reduced LV function, extensive scar, or reduced LVEF. As discussed later, patients in this SPECT- or PET-MPI category with extensive ischemia are identified as those in whom knowledge of coronary anatomy may define clinical management. This patient subset benefits from revascularization, as opposed to guideline-directed medical management alone, as is discussed in detail later in this chapter.
+++
Added Value of LV Function and Size Using Gated SPECT or PET
++
Assessment of LV function on gated SPECT or PET-MPI provides incremental value over perfusion in assessing prognosis. Poststress LVEF and LV ESV provide incremental information over the perfusion defect assessment in the prediction of cardiac death.178 In a report of 6713 patients, using separate criteria for EF and ESV for men and women, Sharir and coworkers reported that perfusion (percent myocardium ischemic), function (LVEF), and LV volumes (ESV) provide incremental prognostic information regarding cardiac death and hard events122 (Fig. 18–22), with particularly high event rates noted among women with reduced LVEF and more than 10% myocardium ischemic. A subsequent study reported that SPECT-MPI determined ischemia and gated LVEF added incrementally to each other for the prediction of adverse events.179
++
++
Other important information that can be derived from gated SPECT-MPI and are related to risk has not been widely included in literature reporting the prognostic assessment (see Table 18–6). These include regional or global stunning. (31, 180) TID of the left ventricle124,125,129 and pulmonary uptake of radioactivity.134 Extensive reversibility of resting 201Tl perfusion defects (as determined by 24-hour 201Tl imaging after rest 201Tl/stress 99mTc-sestamibi SPECT-MPI) has been shown to be predictive of a higher mortality than would be predicted by rest or stress perfusion defect abnormalities alone.181 As noted below, with the most commonly performed PET-MPI approach—82Rb PET—unlike SPECT, imaging is performed during as opposed to after pharmacologic stress. Any fall in EF is considered abnormal and has been associated with critical coronary stenosis.182
++
As noted above, PET-MPI offers an important characteristic that is not provided by current SPECT-MPI—the quantitation of absolute rest and stress myocardial blood flow and CFR. These absolute flow measurements provide added value in risk assessment as discussed below. Furthermore, they provide an ability to recognize when there has been inadequate vasodilation stimulus caused by caffeine intake by the patient—detected as no increase in flow between rest and stress. Of importance, from a diagnostic perspective, a completely normal PET-MPI study, including normal CFR, has been reported to essentially exclude the presence of high-risk epicardial CAD.167 In addition, CFR and maximally myocardial blood flow provide assessments of overall myocardial blood flow and thus evaluation for focal epicardial coronary stenosis as well as diffuse epicardial disease and microvascular disease or microvascular dysfunction. This capability of PET flow measurements provides information that is complementary to the invasive assessment of FFR, allowing the identification of diffuse epicardial disease and microvascular dysfunction. These have their own prognostic and therapeutic implications beyond those associated with assessment of an individual epicardial coronary lesions.183,184
+++
Sex-Based Differences in the Prognostic Value of SPECT-MPI
++
Recent guidelines for cardiac imaging in women have been published by the AHA.185 In women, because of breast tissue artifact, false-positive SPECT-MPI examinations are most notable in the anterior and anterolateral segments of the heart and are more common with 201Tl than with the 99mTc agents.185 Improved accuracy has been reported with use of the 99mTc agents as well with combined acquisition of gated EF and wall motion imaging and prone imaging and the use of validated attenuation correction algorithms.185,186
++
Regarding prognosis, pooled data involving more than 7500 women noted annual rates of cardiac death or nonfatal MI of 0.4% for those with low-risk or normal SPECT-MPI.187 High-risk MPI findings elevated a woman’s risk by nearly 10-fold, with annual rates of major cardiac events of 6.3% for all women and 10.9% for diabetic subsets of women.187 With separate criteria for abnormality of ventricular function in women, there was similar prognostic content of combined perfusion and function information from gated SPECT in men and women.122
++
Coronary vascular dysfunction (often termed coronary microvascular disease [CMD] or dysfunction) has been proposed as a mechanism for mislabeled “false-positive” stress testing results in women, suggesting that some of these studies may represent true perfusion abnormalities without significant epicardial coronary stenosis. Evidence suggests that these SPECT- and PET-MPI perfusion findings may be associated with increased near-term risk of major cardiac events.188 Multiple reports suggest that prognostically important CAD states not involving epicardial, obstructive CAD occur more frequently in women than in men and that PET-MPI could be useful for detecting this process. A current active field of investigation is to more clearly define the types of CMD and the prevalence of mild epicardial CAD stenosis or high-risk atherosclerotic plaque features (eg, expansive or positive remodeling), in women which may also influence CFR findings.
+++
PET-MPI in Risk Assessment
++
For PET-MPI, the past decade has witnessed a surge in publications. Along with the assessment of incremental prognostic value, risk stratification, and test performance in various patient subgroups and the role of change in LVEF, the impact of CFR on patient outcomes has also been extensively reported. As recently summarized,189 a large number of studies have evaluated the prognostic value of stress PET-MPI.182,190,191,192,193,194,195 As described for SPECT-MPI, the risk of a normal stress PET is very low, with risk increasing as a function of the extent and severity of perfusion defects.182,190,191,193,194
++
A recent meta-analysis assessing the prognostic value of stress PET perfusion identified 20 eligible studies (7 prospective) included a median sample size of 551 patients (interquartile range [IQR] 232-1291) with follow-up ranging from 1.0 to 7.3 years for clinical outcomes.196 These included four studies using NH3, fifteen using Rb82, and one combining patients who underwent either. These studies were published between 1993 and 2013. The authors reported that similar to SPECT, a significant increase in risk of cardiac death was present in patients with at least 10% ischemic myocardium (summary hazard ratio [HR] 1.66, 95% CI 1.51-1.82) or scarred (fixed defect) myocardium (HR 1.83, CI 1.70-1.96). Furthermore, analyses revealed that, an SSS of 4 or more (6% or more) and a CFR less than 2 identified significant associations with major adverse cardiovascular events (HR, 95%, CI 2.30 [1.53-3.44] and 2.11 [1.33–3.36]), respectively].
+
A recent multicenter registry evaluated the prognostic value of stress PET-MPI in 7061 patients with known or suspected CAD who underwent a rest/stress 82Rb PET drawn from four sites.194 Risk-adjusted analyses revealed that over a median follow-up of 2.2 years, stress PET results added incremental value over preimaging data for the prediction of cardiac death. Compared to normal studies, mildly, moderately, and severely abnormal studies were associated with HRs of 2.3 (1.4-3.8), 4.2 (2.3-7.5), and 4.9 (2.5 to 9.6; all P < .001), respectively. A study comparing the prognostic value of stress PET by sex was also reported from a subset of 6037 patients in this multicenter effort.190 In this study, men had greater likelihood of CAD and more frequently had PET abnormalities. Men had a higher mortality during follow-up (5-year mortality: 6.0% vs 3.7%, P < .001; 115 vs 54 deaths). PET results added incremental value in both men and women and had numerically similar adjusted HRs (women 1.81 [1.54-2.14]; men 1.71 [1.52–1.94]). Hence, stress cardiac PET was prognostically valuable for both women and men. This registry has also been used to ascertain and compare the prognostic implications of PET-MPI in normal, overweight, and obese patients.197 Body mass index (BMI) was grouped in three patient categories: normal (< 25 kg/m2), overweight (25-29.9 kg/m2), or obese (≥ 30 kg/m2, with a mean BMI of 30.5 ± 7.4 kg/m2). PET results defined by a percent myocardium abnormal were categorized into normal (0%), mild (1%-9.9%), moderate (10%-19.9%), and severe (≥ 20%). In 6037 patients, a normal PET-MPI was associated with an excellent prognosis with very low annual cardiac death rates in normal (0.4%), overweight (0.4%), and obese (0.2%) patients. Furthermore, in all three BMI-defined categories, greater abnormality of PET-MPI results were associated with increased patient risk. Similarly, PET-MPI results added incremental prognostic value and enhanced risk stratification in all patient subgroups.
++
As noted above, one of the advantages of cardiac stress PET compared to stress SPECT is the acquisition of peak stress, rather than poststress image data. Hence, PET-MPI provides a rest LVEF, peak stress LVEF, and an LVEF reserve (stress LVEF–rest LVEF). The independent and incremental prognostic implications of rest and peak stress LVEF have been reported.182,192,194,198 The prognostic value of the LVEF reserve information was first investigated by Dorbala and colleagues182 in 985 consecutive patients who underwent gated rest/vasodilator stress 82Rb PET. Over a mean follow-up of 1.7 years, the unadjusted annualized rates of cardiac death or nonfatal MI were greater in patients with an LVEF reserve less than 0% than in those with an LVEF reserve greater than 0% (2.1% vs 5.3%, P < .001). Multivariable survival modeling revealed that after adjustment for clinical, historical, and resting EF data, the LVEF reserve added incremental prognostic value for the prediction of both the end point of cardiac death or nonfatal MI and for all-cause mortality (Fig. 18–23).
++
++
In the MPI literature, the terms coronary blood flow and coronary flow reserve, or CFR, have been used interchangeably with the terms myocardial blood flow and myocardial flow reserve. A distinct clinical advantage afforded by the use of PET-MPI over SPECT-MPI is the ability to capture accurate, validated, and reproducible measures of rest and stress absolute flow and flow reserve on a per-vascular territory and global LV basis, as discussed in Chapter 19. In the first large study examining the prognostic implications of impaired vasodilator function relative to indices of PET myocardial perfusion, Murthy and coworkers192 examined a cohort of 2783 consecutive patients referred for stress PET who were followed up for a median of 1.4 years (IQR 0.7-3.2 years), with cardiac death as the primary end point. After adjustment for multiple factors, including rest LVEF, SSS, and LVEF reserve, compared to the highest tertile of CFR, the lowest tertile had a HR of 5.6 (95% CI, 2.5-12.4) and the middle tertile had a HR of 3.4 (95% CI, 1.5-7.7) (Fig. 18–24).192 Similarly, Ziadi and colleagues195 evaluated the added prognostic information provided by CFR using Rb82 PET-MPI in 677 patients who were followed for adverse events (median follow-up ~ 1 year). In a Cox model adjusted for pre–PET-MPI data, CFR was an independent predictor of the end point of cardiac death or MI (HR 3.3; 95% CI, 1.1-9.5; P < .029). This relationship persisted after adjustment for stress perfusion data.
++
Murthy and colleagues extended this work in an investigation of sex-related differences in 405 men and 813 women.199 This cohort was identified on the basis of having no prior history of CAD and no evidence of CAD-associated perfusion defects on their PET studies. A threshold for abnormality of CFR was less than 2.0. In both men and women, a CFR less than 2.0 was a frequent finding (51% and 54%, equivalence P = .0002). Patients were followed over a median 1.3 years (IQR 0.5-2.3 years) for major adverse cardiac events (cardiac death, nonfatal MI, late revascularization, or hospitalization for heart failure). In both sexes, CFR was incrementally predictive of major adverse cardiac events (HR, 0.80 [95% CI, 0.75-0.86] per 10% increase in CFR; P < .0001). This finding shows an inverse relationship with higher CFR equating to lower risk and vice versa. Interestingly, the authors also reported that in a subgroup of 404 patients without evidence of CAC and with normal stress perfusion imaging in both sexes, a CFR of less than 2.0 was common (44% of men vs 48% of women). Future work is needed to identify its putative role as a therapeutic target.
++
An additional report focused on examining the prognostic significance of CFR among diabetics as compared to nondiabetics.200 In this report, PET-assessed CFR was performed in a total of 1172 diabetic and 1611 nondiabetic patients who were followed up for a median of 1.4 years (IQR 0.7-3.2 years) for occurrence of cardiac death. An impaired CFR was defined as that falling below the median value and was associated with an adjusted 3.2- and 4.9-fold increase in cardiac death risk in diabetics and nondiabetics (P < .0004). CFR findings had a profound impact on risk assessment among diabetic patients without prior CAD—those with a reduced CFR experienced risk equivalent to nondiabetic patients with CAD while those with preserved CFR had a level of risk similar to that of patients without CAD or without diabetes; both of which were similar to patients with normal PET-MPI and LVEF (Fig. 18–24).200 These data support an integral role for CFR in assessing risk. Even in patients without CAD and with normal LV function, after adjusting for multiple confounders, impaired CFR was associated with both the occurrence of a positive troponin and with major downstream adverse events.201
++
+++
Integrating Clinical Data with Nuclear MPI Results for Risk Assessment
++
A challenge facing clinicians attempting to apply nuclear MPI results to patient care is to distill all information reported after SPECT-MPI—clinical, historical, stress test, perfusion, and function data—for risk of adverse events for an individual patient or likelihood of obstructive CAD as discussed above. Accurate final estimates of risk must incorporate all relevant prognostic factors. Ideally, validated prognostic assessments can be developed integrating all available sources of information, including nuclear MPI results, and be incorporated into nuclear MPI reporting.
++
When assessing risk, incorporation of non-nuclear variables obtained at the time of testing is particularly difficult with pharmacologic stress because prognostically important variables used in exercise testing, such as exercise duration and exertional chest pain, are not components of pharmacologic stress testing protocols. To test the potential of a multivariable integration of findings into a single score, a score was developed in 5873 patients studied by adenosine stress who experienced 387 cardiac deaths on follow-up (6.6%).58 The authors derived a complex score taking into account all significant variables, including age, percent myocardium ischemic, percent myocardium fixed, diabetes, dyspnea as the presenting symptom, resting HR, peak HR, and ECG findings. Separate scores were calculated for therapeutic choices of medical therapy or revascularization (Fig. 18–25). The results of this study, although not in common clinical use, point the way to how test results may be analyzed in the future. In this regard, improvement in prediction of obstructive coronary disease and post-MPI revascularization has been recently demonstrated by deriving integrated scores by machine learning from imaging and clinical features.202,203 With the increased applications of machine learning methods and computer-assisted test analysis and reporting, application of these comprehensive assessments—which have the potential of systematically taking into account all available information—may become practical tools to improve the clinical interpretation and reporting of nuclear MPI studies.
++
+++
Estimating the True Prognostic Value of Nuclear MPI and Post-Test Referral Bias
++
Although there is compelling evidence that SPECT- and PET-MPI are effective in the prognostic stratification, current data on risk stratification by nuclear MPI may underestimate the strength of this modality because of a prognostic counterpart to the verification bias described earlier. Most prognostic analyses performed to date are comprised of patients undergoing follow-up medical management and without documentation of the adequacy of guideline-directed care. Moreover, many survival analyses also censor revascularized patients at the time of their invasive procedure with varying physician practices of prompt or delayed referral to coronary revascularization. The rationale for censoring of revascularized patients is the traditional viewpoint that surgical intervention improved outcome. This practice is less commonplace today because current randomized trial evidence does not support risk reduction with coronary revascularization as compared to medical therapy.152,204 Thus, there is a varied relationship between the test results and the referral to revascularization.205 The result is a potential underestimation of the prognostic value of SPECT-MPI; the patients at the highest risk, with the most abnormal test results, are removed from the population being studied.173,205,206
++
The reduction in observed event rates resulting in this referral bias in medically treated patients with severe amounts of ischemia on SPECT-MPI206 has been quantified. Thus, reported event rates in observational studies limited to medically treated patients may be high risk but underestimate the true risk in these patients; the patients with the greatest ischemic risk were referred for revascularization and censored from the prognostic studies. Moreover, the quality of medical management is quite disparate, and the relative adherence to SIHD guidelines is unknown.152 Statistical techniques may be applied in observational cohorts that may yield insight into the potential for stress MPI to define candidates and guide management of SIHD patients.25,58,205
+++
Nuclear MPI and Clinical Decision Making: Assessment of Cardiac Risk versus Potential Survival Benefit from Revascularization—Randomized Trials and Observational Findings in SIHD
++
The paradigm of the application of nuclear MPI to patient risk assessment conceptually yields a series of strata, representing the spectrum of risk and where along the continuum of risk the patient undergoing nuclear-MPI falls, after consideration of clinical, historical, stress test, and nuclear MPI results. Beyond risk stratification, optimal selection of patient treatment should be based on reasonable estimates of potential patient benefit with respect to a relevant and important end point with one treatment option versus an alternative. Moreover, the current evidence on the potential benefit of stress MPI should be considered in light of the randomized trial evidence and guideline-directed medical therapy approaches, which largely support equipoise on the use of surgical revascularization even for those with moderate to severe ischemia.152
+++
Identifying Optimal Post-MPI Patient Management
++
Evolving evidence indicates that beyond identifying patient risk, a principal role of nuclear MPI in a testing strategy is the identification of patients who may accrue a clinical benefit in terms of risk reduction with a specific therapeutic approach. Observational evidence is available from registry data and randomized trials regarding the impact of stress MPI for medical management decisions of patients with SIHD.207,208,209,210,211 With regard to the observational evidence on a survival benefit with coronary revascularization, in a study of 10,627 patients without prior MI or revascularization, and who underwent stress SPECT-MPI, a survival benefit for patients undergoing medical therapy versus revascularization was shown in the setting of no or mild ischemia, whereas patients undergoing revascularization had an increasing survival benefit over patients undergoing medical therapy when moderate to severe ischemia was present (> 10% of the total myocardium ischemic) (Fig. 18–26).25 Furthermore, the roles of stress perfusion and gated SPECT-MPI EF in identifying this observational benefit have also been compared.179 Although EF, percent myocardium ischemic, and the percent myocardium fixed are all predictors of cardiac death, the first is by far the best predictor of cardiac mortality. Conversely, only inducible ischemia identified patients who would benefit from revascularization compared with medical therapy (Fig. 18–27). With increasing amounts of ischemia, increasing survival benefit for revascularization over medical therapy was found, across all values of EF. A subsequent study by the same investigators examined this relationship in a large cohort of 13,969 patients, both with and without prior CAD, who were followed-up for an average of 8.7 years after MPI. As previously described, patients without prior CAD (n = 8791) had improved survival with revascularization in the setting of significant ischemia, as did patients with prior revascularization but no prior MI (n = 1542). However, as an overall group, patients with prior MI (n = 3216) did not show benefit with early revascularization as a function of the extent of ischemia (Fig. 18–28). However, patients with more than 10% to 15% of their myocardium ischemic, but with less than 10% of their myocardium with fixed defects, had a survival benefit with revascularization compared to medical therapy alone. These observational findings suggest that it was patients with extensive scar in whom ischemia did not identify a survival benefit.209,212
++
++
++
++
The value of MPI-assessed ischemia to identify patients who may benefit from revascularization has been assessed in other observational patient cohorts. In 5200 patients who were aged 75 years or older, increasing ischemia was associated with increasing survival with early revascularization. Of interest, the threshold of ischemia at which this survival benefit appeared to be present was higher (≥ 15%) than observed in the previously described.213 Sorajja and colleagues extended these results in a series of 826 asymptomatic diabetic patients without known CAD who had undergone stress SPECT-MPI. Improved post-MPI survival with revascularization during follow-up was limited to those patents with high-risk SPECT treated with coronary artery bypass grafts (CABGs).214 Finally, in a recent study of asymptomatic patients with history of prior revascularization, no benefit with revascularization was found, despite the presence of significant ischemia.215 This observational survival benefit was striking in higher risk patients (ie, elderly, requiring adenosine stress, women, and diabetics). Thus, in assessing treatment options in an individual patient, cardiac risk factors, comorbidities, coincidental anatomic findings, and EF all have to be considered along with ischemia to determine the potential advantages of a specific therapeutic strategy.
++
Observational studies, such as those discussed above, are primarily intended for hypothesis generation and have significant limitations. These include concerns regarding generalizability and dependency on risk adjustment techniques. Use of the latter does not obviate the impact of selection biases, unmeasured covariates, and other factors, on the results of the study.156 For instance, to what degree the intensity of medical therapy in these studies is equivalent to that used in clinical trials is uncertain. However, these studies are effectiveness studies—treatment received by patients was dictated by clinical practice. Hence, they may better represent patients seen in practice and changes in practice over time than in randomized trials.
++
These results from single-site observational series have been recently extended by the results of prospective randomized clinical trials of SIHD.204,216 Shaw and coworkers210 reported several important results from the nuclear substudy of the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial. First, in this subset of COURAGE patients, the addition of percutaneous coronary intervention (PCI) to optimal medical therapy (OMT) resulted in more effective reduction of ischemia and angina symptoms than OMT alone. The reduction in ischemia was documented by comparing pretreatment (baseline) MPI to repeat stress MPI 1 year later. Importantly, regardless of treatment assignment, the magnitude of residual ischemia determined by follow-up SPECT-MPI was proportional to the risk of death or MI, and patients with a greater than 5% reduction in ischemia, measured quantitatively by the TPD at stress minus the TPD at rest, had a reduced rate of death or MI (Fig. 18–29).210 Importantly, this substudy within the COURAGE trial was powered to examine an ischemia reduction and not for clinical events. Accordingly, this significant univariable analysis was not statistically significant in adjusted analyses.
++
++
An additional study from the COURAGE trial evaluated the results of site-interpreted MPI in 1381 patients randomized to OMT alone versus PCI and OMT who had baseline MPI studies.217 Using the results of interpretation of defect extent using a six-segment model, researchers found no difference in rates of death and MI in patients with three segments or more with ischemia versus those patients with less than three segments. Furthermore, patient risk did not increase with worsening extent of ischemia, suggesting a failure to risk stratify. Importantly, this report is unique in reporting the results of MPI interpretation and outcomes from numerous sites in the United States and Canada. The results suggest the possibility that differences between “expert” reads to a variety of readers with varying experience affect outcomes.
++
The most recent study from COURAGE examined 621 patients with baseline quantitative MPI data and quantitative anatomic data to determine the predictive value of these two measures on patient outcomes (end point of death, MI, and non–ST-segment elevation acute coronary syndromes [ACS]) in medically treated patients as well as to determine whether either of these factors could identify which patients would benefit from one therapeutic approach versus another.218 The authors found that, both with and without risk adjustment, coronary anatomic burden and LVEF were significantly associated with patient outcomes. Neither ischemia nor treatment assignment was predictive, although a significant interaction between ischemic burden and coronary artery anatomy was present. Most importantly, however, neither coronary anatomy nor ischemic burden was predictive of which treatment option would be predictive of improved survival.
++
Finally, the nuclear substudy from Bypass Angioplasty Revascularization Investigation 2 Diabetes (BARI 2D) included data from SPECT-MPI studies performed 1 year after enrollment in 1505 patients randomized to medical therapy versus medical therapy with revascularization.211 Thus, the ability to compare reductions in MPI abnormalities following intercurrent randomized treatment was not possible because of a lack of data collection in the prerandomization setting. However, comparing 1-year MPI results in patients who received revascularization plus medical therapy had smaller total defect size and lower percent myocardium ischemia when compared to those receiving medical therapy alone. At 1 year after randomization, 59% of revascularized patients had ischemia compared to 49% with OMT alone (P < .001). Furthermore, 4-year risk of death and MI increased with more severe and extensive MPI defects.211 As was the case in the COURAGE nuclear substudy, the relationship between treatment, MPI results, and outcomes was not statistically significant. Although there are statistical power issues in the randomized trial comparisons for clinical effectiveness, these data support the main SIHD trial findings, such as from COURAGE and BARI 2D, that revascularization reduces angina and ischemic findings as compared to medical therapy for patients with SIHD. However, no differences in long-term event reduction were reported for either trial when comparing PCI versus medical therapy.
++
There is little published data evaluating the ability of stress PET-MPI to observationally identify which patients may accrue a survival benefit with revascularization with medical therapy compared to OMT alone. In a study of 2223 patients followed for a median of 3.6 years after PET-MPI, Taqueti and colleagues examined the association of post-MPI treatment, the amount of ischemia present, and post-test patient outcomes.219 Cox proportional hazards modeling adjusting for multiple potential confounders, including a propensity score to correct for nonrandomized treatment allocation (P < .0001), revealed significant interactions between ischemia and the use of revascularization (ie, survival increased with the use of revascularization for patients with more extensive and severe ischemia) and between CFR and the number of anti-ischemic medications used (P = .01). As with SPECT, these findings are observational and hypothesis generating, thus further investigations are necessary.
+++
Stress MPI in the SIHD Clinical Practice Guidelines
++
Applications of nuclear MPI in SIHD are included in the recent ACCF/AHA clinical practice guidelines on SIHD152 as well as in the European Society of Cardiology (ESC) guidelines for management of stable CAD.220 They are also covered in the ACCF/AHA multimodality appropriate use criteria (AUC), as discussed below. The ACCF/AHA SIHD guideline assigned a class Ib level of evidence to exercise MPI for patients with an intermediate to high pretest risk who have an ECG for which the exercise response cannot be interpreted and a class Ia for those unable to exercise. An exercise stress MPI in this same subset with an intermediate to high risk in the presence of an interpretable ECG was assigned a class IIa level of evidence. More details on these indications are also provided in the AUC section and in the sections on specific patient cohorts. With regards to ischemia-guided management, the SIHD clinical practice guideline recommend initiation of guideline-directed medical therapy as the initial post-test treatment strategy. However, they also provide variable recommendations for revascularization, including consideration of patient preferences and clinical factors, as well as the high-risk stress test findings.152,220 In general, the index approach for most patients is OMT with invasive angiography limited to those failing to improve symptoms following treatment intensification. In the ESC guidelines, consideration of revascularization is limited to patients with more than 10% ischemic myocardium.220 However, as discussed above, there is no randomized trial evidence to demonstrate a therapeutic risk reduction with revascularization as compared to OMT in patients with moderate to severe ischemia (ie, > 10% ischemic myocardium). Current randomized trials fail to support such a threshold benefit and the lack of consistent recommendations across the guidelines reflects the variable evidence base. The ongoing International Study of Comparative Health Effectiveness With Medical and Invasive Approaches (ISCHEMIA) trial is examining this primary aim with enrollment of about 5000 patients with moderate to severe ischemia on stress imaging. Patients are being randomized to a routine invasive strategy with cardiac catheterization followed by revascularization when appropriate plus OMT or to a conservative strategy of OMT. In the latter group, catheterization and revascularization are reserved for those who fail OMT. Additional SIHD and other MPI trials using MPI are discussed in the special patient population sections below.
+++
Post-MPI Resource Utilization and Cost-Effectiveness Findings
++
A number of investigators have examined the relationship between SPECT-MPI and subsequent patient management. Overall, there is a clear relationship present in patients with normal scans; only a small proportion undergoes early post–SPECT-MPI cardiac catheterization, usually as a result of clinical symptomatology.100,161,208,209,221,222,223,224,225
++
As first shown by Hachamovitch and coworkers, the extent and severity of reversible defects shown by the SPECT-MPI result are the dominant factors driving subsequent resource utilization25 (Fig. 18–30). Importantly, even among the patients at greatest post-MPI risk—patients with high postexercise treadmill test likelihood of CAD and moderate to severe MPI abnormalities—only about half the patients were referred to catheterization. These findings, coupling MPI results to catheterization rates and potential underuse of catheterization, have been confirmed by this group and other authors.100,161,208,209,221,222,223,224,225 Regarding the cost-effectiveness of this approach, Shaw and colleagues,225 in a multicenter study of 11,249 patients, extended these results to show that a strategy of SPECT-MPI with selective subsequent catheterization produced a substantial reduction (31%-50%) in costs for all levels of pretest clinical risk compared with a direct catheterization approach (Fig. 18–31), with essentially identical outcomes as assessed by cardiac death and MI rates. Importantly, in the SPECT-MPI strategy, rates of revascularization, rates of cardiac catheterization after normal SPECT-MPI, and the frequency of normal coronary angiographic findings were significantly reduced.225
++
++
++
A recent prospective, multicenter study, the Study of Myocardial Perfusion and Coronary Anatomy Imaging Roles in Coronary Artery Disease (SPARC), examined referral rates to catheterization and use of medical therapy after clinically ordered SPECT, PET, and CCTA in patients from 40 sites.223 In a cohort of 1703 patients with no prior CAD and intermediate to high pretest likelihood of CAD, the authors found that referral rates to catheterization mimicked those in previous studies; after the most abnormal test results, 38% to 61% of patients were not referred to catheterization (Fig. 18–32). Additionally, about half these patients with significant abnormalities did not receive medical therapy. Although there were significant increases in the use of aspirin and beta-blockers post-testing, the change in use of lipid-lowering agents was not significant. Furthermore, only one in five patients were on all three of these medications, most were on one or two of them, and there was no increase in the proportion of patients with increased number of medications after testing (Fig. 18–33). The study did not examine the root cause of these results, but they do suggest that additional studies are needed and, perhaps, postnoninvasive utilization of resources may need to be considered as a quality metric.
++
++
++
The data are robust on the cost consequences that follow the above discussion on post-testing resource consumption testing patterns. Although the older nuclear MPI literature is replete with cost-effectiveness analysis, more contemporary evidence has also recently been reported. In a report including a large number of Medicare beneficiaries (n = 282,830), 180-day costs of stress MPI were compared to those of CCTA.226 An index testing strategy of CCTA was associated with higher rates of follow-up coronary angiography and revascularization procedures and, as a result, had higher total health care spending (P < .001). The costs for stress MPI were about $4200 lower over 6 months when compared to CCTA.
++
Similarly, in another report from the SPARC registry, 1703 patients with suspected CAD were enrolled and underwent CCTA (n = 590), stress PET-MPI (n = 548), or stress SPECT-MPI (n = 565).227 In this registry, stress PET-MPI patients had the highest risk, with 2-year mortality of 5.5%, followed by stress SPECT-MPI (1.7%), and lowest with CCTA (0.7%). The stress PET-MPI patients had the highest costs at $6647 at 2 years, with costs of $4909 for CCTA and $3695 for stress SPECT-MPI. In a decision analytic model, the incremental cost-effectiveness ratio for CCTA versus SPECT was $11,700 per life year saved. These results support that risk is a prominent factor impacting resource consumption patterns and costs following testing; that is, when comparing the cost-effectiveness of tests, risk adjustment is of paramount importance.
++
The Cost Effectiveness of Noninvasive Cardiac Testing (CeCAT) trial studied 898 patients referred for nonurgent coronary angiography who were randomized in a 1:1:1:1 basis to SPECT, cardiac magnetic resonance (CMR), stress echocardiography, or catheterization. The study was designed to study the noninvasive tests as a gatekeeper to invasive coronary angiography, evaluating mortality and costs.228 CMR was associated with a higher mortality as compared to the other modalities, and the mortality in the other groups were not significantly different from the mortality in the cath eterization group. Overall, the frequency of cath was eterization reduced by 20% to 25% in the noninvasive testing arms. Stress SPECT-MPI was economically superior to the other noninvasive tests and to cath eterization, with a more than 70% probability of being cost effective (in bootstrap simulations) when compared to stress CMR imaging and echocardiography. The 2-year costs for stress SPECT-MPI were lower than those for invasive coronary angiography. The costs for stress echocardiography were actually higher than for coronary angiography, which the authors attribute to the relatively high risk of the population being studied.
++
A synthesis of the cost data underscores the importance of pretest risk in assessing cost effectiveness. As expected, as risk increases, there are greater treatment intensity and greater costs of care. Cost-effectiveness of the various noninvasive and invasive approaches depends to a large extent on the risk of the patients being tested.
+++
Comparative Effectiveness
++
A recent large comparative effectiveness trial compared stress testing, which was predominantly MPI, to CCTA. Sponsored by the National Institutes of Health (NIH)–National Heart, Lung, and Blood Institute (NHLBI), the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) evaluated 10,003 symptomatic patients who were randomized to an initial strategy of CCTA versus functional testing (67% nuclear MPI, 23% stress echocardiography, and 10% exercise ECG).229 Although all enrollees were symptomatic, the majority (78%) had atypical symptoms with the result being a low risk cohort with a low rate of CAD events. The primary end point was death, ACS, or major procedural complication, which occurred in approximately 3% of randomized patients at 2 years of follow-up. The primary end point was similar by randomization strategy with an HR of 1.04 (95% CI, 0.83-1.29; P = .75). Similarly, the Scottish COmputed Tomography of the HEART Trial (SCOT-HEART) randomized 4146 patients with suspected CAD to CCTA or standard of care arms. In both arms, stress testing was usually performed.230 The results of this trial do not provide information regarding comparative effectiveness with MPI directly, because only about 10% of the patients in the standard of care arm underwent MPI.
++
To provide value, MPI must be applied in a population with at least intermediate risk or likelihood of CAD. Notably, the PROMISE trial provided evidence that the pretest likelihood of obstructive CAD and of risk is overestimated by current clinical algorithms. The event rate in the population was substantially lower than expected based on historical data. Whereas the mean pretest likelihood of obstructive CAD by the Diamond-Forrester classification was 53.3 ± 21.4%, the observed prevalence of 50% stenosis or more on CCTA in the CCTA arm was only 10.7% and the prevalence of abnormality on stress testing was only 11.7%. A change in true likelihood of CAD according to age, sex, and symptoms, since the time of the initial descriptions by Diamond and Forrester, may have occurred over time, accounting for these observations. In this regard, recent findings in 39,515 patients undergoing SPECT-MPI over an 18-year period have shown a dramatic fall in the frequency of abnormal SPECT-MPI procedures, despite a rise in the apparent pretest likelihood of CAD (Fig. 18–34). These observations suggest the need to use new algorithms for assessing pretest likelihood of CAD and of risk in patients considered for referral to noninvasive testing. To this end, the recently described CONFIRM risk score developed by Min and colleagues provides an easily applied approach, which has been validated in patients undergoing CCTA and SPECT-MPI studies.231
++
+++
AUC for Optimal Patient Selection
++
The ACC, in conjunction with multiple professional organizations, has developed AUC for applications of testing in cardiology. Such AUC play a complementary role along with clinical practice guidelines. The documents were developed to guide health care coverage decisions and are now mandated by many private payers in the United States. Use of the AUC or similar approaches will be required for Medicare coverage beginning in 2018.232 The initial AUC were published for cardiac radionuclide imaging, which were then updated.233,234 Recently, a multimodality AUC document was published and replaced prior published documents.235 Although the recent document includes multimodality criteria including exercise ECG, stress echocardiography, stress magnetic resonance imaging, CCTA, and stress nuclear MPI, we will highlight the criteria directly relevant to stress SPECT and PET-MPI. Of note, the AUC considers the assessments of SPECT and PET-MPI as the same.
++
In the multimodality AUC for detection and risk assessment in SIHD, three principal scenarios are discussed and rated based on AUC categories: suspected SIHD, known SIHD, and preoperative cardiac assessment.235 Each modality is rated individually, and the ratings of each procedure were accomplished by evaluating the evidence within the modality and not as a comparison between the modalities. The multimodality AUC categorizes indications for cardiac imaging as appropriate, maybe appropriate, or rarely appropriate. Moreover, the term maybe appropriate also indicates that some of the patients within a given indication will be candidates for MPI. Further, the term appropriate does not mandate testing but supports that a large proportion of patients within a given indication will be candidates for nuclear MPI. Importantly, the term rarely appropriate does not prohibit testing but implies that only a limited number of patients within a given indication will be candidates for nuclear MPI. For nuclear MPI, rarely appropriate indications largely include low-risk or asymptomatic patients. Several published reports have examined the proportion of appropriate indications based on patient referrals to an SPECT-MPI laboratory.234,236,237,238 From the largest series, Hendel and coworkers239 presented data on 5928 patients, with 71%, 15%, and 14% of the studies being categorized as appropriate, maybe appropriate, or rarely appropriate, respectively. The most common rarely appropriate indications were (1) asymptomatic low-risk patients, (2) asymptomatic patients less than 2 years postrevascularization, (3) evaluation of chest pain in low-probability patients with interpretable ECG who were able to exercise, (4) asymptomatic patients (or patients with stable symptoms) with known CAD less than 1 year after angiography or prior abnormal stress nuclear study, and (5) preoperative assessment prior to low-risk surgery. These five indications accounted for 12% of all studies and 92% of the studies that were considered inappropriate. By reducing utilization of inappropriate or rarely appropriate studies, considerable cost savings could be achieved. From a recent meta-analysis in 22 studies in 23,343 patients, the proportion of rarely appropriate testing was 15%.240 This analysis also indicated that rarely appropriate MPI had a 59% reduced odds of being abnormal (P < .0001) when compared to patients with appropriate indications for MPI.
++
Decision support tools have been developed for use of the AUC and can guide improvements in appropriate referral patterns and reduce unwarranted testing patterns. Lin and colleagues published findings following the introduction of an AUC decision support tool, noting an increase appropriate use (pre: 49% to post: 61%, P = .02) and marked reductions in inappropriate use (pre: 22% to post: 6%).241
++
Reports examining clinical outcomes based on AUC are limited.242,243 From one series of 1511 patients, the relative hazard for CAD death or MI was elevated nearly four-fold for patients with abnormal MPI when those tested with appropriate or maybe appropriate indication were compared to patients with a rarely appropriate indication (P = .006). A smaller study of 280 patients demonstrated that appropriate studies are more likely to result in abnormal results requiring subsequent revascularization compared to rarely appropriate and maybe appropriate stress studies.244 These studies provide evidence of greater risk stratification by testing when used in patients with an appropriate indication than in patients with rarely appropriate indications.