Author + information
- James K. Min, MD,
- Rory Hachamovitch, MD, MSc,
- Alan Rozanski, MD,
- Leslee J. Shaw, PhD,
- Daniel S. Berman, MD and
- Raymond Gibbons, MD⁎ ()
- ↵⁎Address correspondence:
Dr. Raymond J. Gibbons, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905
Section Editor: Christopher M. Kramer, MD
imaging is presently in the crosshairs of policy makers in the U.S. government as the health care reform debate rages on. The era of proving cost effectiveness, not just efficacy, is upon us imagers whether we like it or not. Comparative effectiveness is the new buzzword in the imaging community. The debate in this issue of iJACC is about the utility of this standard to which newer imaging approaches are presently being held. Min et al. argue that this is a difficult standard, as assessing outcomes based on a diagnostic test is fraught with difficulty. They point out the advantages and limitations of single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) as it grew into its present place in the diagnostic armamentarium. They suggest that computed tomography coronary angiography (CTCA) has similar prognostic ability, at least in the intermediate term (long-term studies are ongoing), at a lower cost. They recommend comparative effectiveness studies of CTCA and SPECT MPI.
Raymond Gibbons points out the problem of spiraling imaging costs. He sets the bar high for CTCA, namely defining its additive value to existing testing before encouraging widespread adoption. He too underscores the excellent prognostic abilities of nuclear imaging. He points out limitations in the existing literature regarding CTCA, including the relatively short follow-up periods for prognostic studies due to the relative youth of the technology. Like Min et al., he stresses the importance of clinical trials to compare CTCA to existing imaging technologies in the evaluation of patients with suspected coronary artery disease (CAD). The emphasis of comparative effectiveness by the National Institutes of Health and the American Heart Association (AHA), among others, will hopefully bear fruit in this regard. We at iJACC hope that you enjoy the vigor of this debate in the imaging community. Both groups come to the same conclusion: comparative studies are sorely needed.
Assessing the Value of Cardiac Imaging: A New Era?
James K. Min, MD, Rory Hachamovitch, MD, MSc, Alan Rozanski, MD, Leslee J. Shaw, PhD, Daniel S. Berman, MD
CAD is the primary cause of mortality, accounting for one-third of deaths in the U.S. alone. Accurate diagnostic and prognostic assessment has been the mainstay of evaluating patients at risk for these events, primarily through functional imaging tests such as myocardial perfusion SPECT (MPS). More recently, newer anatomic imaging modalities, such as coronary computed tomographic angiography (CTA), have shown promise for evaluation of patients with suspected or known CAD.
Nevertheless, the enthusiasm for adopting such technologies has been tempered by a medical climate that increasingly emphasizes cost containment. Spending on imaging by the Centers for Medicare and Medicaid Services (CMS) more than doubled to $14 billion from 2000 to 2006 at ∼13% per year—almost twice that for other physician services, with cardiac imaging representing a prominent component of growth (1). In light of such observations, many have proposed a new benchmark for valuing cardiac testing, namely, demonstration that testing leads to improved outcomes at a reasonable cost. This concept of a “value-based health care system” has been implicitly adopted by CMS in a recent proposed National Coverage Determination (NCD) of coronary CTA, which called for the satisfaction of 3 criteria—addressing diagnostic accuracy; resource consumption and costs; and clinical outcomes—for demonstration of the benefit of coronary CTA. A more contemporary adjunct to the proposed NCD has been advocated by the Medicare Payment Advisory Commission and culminated in the Comparative Effectiveness Research Act, which suggested that newer technologies be subject to direct measure against pre-existing ones for demonstration of comparative value, that is, both clinical and cost effectiveness (2).
These criteria transcend those required historically for noninvasive cardiac testing, which have relied upon measures of diagnostic accuracy and prognostic risk stratification. As these recent shifts in perspective portend a new standard by which new imaging technologies may be considered, they warrant closer examination.
Potential approaches for assessing how diagnostic tests affect outcomes
The concept of relating a diagnostic test to future clinical outcomes is not straightforward. By its inherent properties as a diagnostic rather than therapeutic intervention, an imaging test merely supports clinical decisionmaking and cannot by itself directly improve outcomes. Imagining headlines proclaiming “Chest X-Rays Cure Pneumonia!” or “Nuclear Scans Prevent Heart Attacks!” underscores the obvious disconnect between supposed causal relationship between a diagnostic test and clinical outcome, unless the test findings can be linked to physician and patient behavior.
Instead, benefits from imaging are more easily measured by indirect means (Table 1). These measures fall into 5 general categories, including the impact of testing on: 1) patient lifestyle or medication compliance; 2) physician practice with scientific evidence; 3) improvement in quality of life; 4) reduction in downstream resource utilization; and 5) reduction in health care costs. Collectively, these factors may reduce rates of adverse events. Many of these criteria of benefit are not simple to assess since they are necessarily dependent on external factors, such as patient motivation, social support, and “medical literacy;” quality and quantity of medical support services; physician quality; and availability of medical reimbursement. These numerous factors create a complex interplay that can dilute the potential impact of any given testing modality and challenge the quantification of the indirect measures that are listed in Table 1.
This interchange between test, patient, and physician has caused some to question whether diagnostic tests can or even should be held to the same standards as therapeutic interventions. At present, no clear guidelines exist for appraisal of new diagnostic technologies under a value-based health care system. In fact, the most concrete method of assessing new diagnostic technologies is defined by the Food and Drug Administration (FDA) approval process—responsible for the authorization of new technologies for clinical use—which is driven largely towards reliability of manufacturer claims and patient safety, rather than improvement of patient-centered outcomes. Thus, a critical gap exists such that the design of a clinical study to assess the value of a diagnostic test may not be able to simultaneously satisfy the requirements set forth by the FDA and advocates of “value-based” imaging.
It is within this context that we examine the emerging evidence for coronary CTA—which serves as an example of a newly introduced diagnostic technology with great promise—to determine what further evidence, if any, is required for illustration of its utility to advocates of the value-based imaging paradigm. We begin with an examination of the evidence accrued for SPECT MPI—a mature technology for which widespread credence has developed with regards to its value—and use the SPECT MPI evidence base as a template upon which to guide future coronary CTA studies.
Lessons learned from SPECT MPI imaging
Since its introduction over 30 years ago, SPECT MPI has become a mainstay for diagnostic and prognostic assessment of patients with suspected and known CAD. Numerous studies support both of these applications.
A plethora of studies involving a wide range of patient populations have documented the diagnostic efficacy of SPECT MPI (3). Early diagnostic studies concerning SPECT MPI were performed with thallium-201; however, later studies confirmed the diagnostic efficacy of subsequently developed radiotracers such as Tc-99m sestamibi and tetrofosmin. One important lesson derived from these studies is that diagnostic performance estimates are profoundly altered when tests are used to guide patient management, with systematic overestimation of diagnostic sensitivity and underestimation of diagnostic specificity. Notably, few studies have been performed with SPECT MPI in which patients underwent the procedure for research purposes alone prior to invasive coronary angiography (ICA)—an approach that reduces referral bias and has become the standard approach used in assessing the diagnostic performance of more recently introduced modalities.
SPECT MPI also demonstrates robust prognostic value, with incident CAD risk increasing exponentially with extent and severity of myocardial perfusion abnormality. Based on this relationship, semiquantitative segmental scoring systems have been devised to mirror gradations of risk. The prognostic efficacy of SPECT MPI is widely applicable in numerous patient groups, including men and women, symptomatic and asymptomatic cohorts, diabetics, the elderly, following revascularization, following myocardial infarction, and in pre-operative patients. The accrued evidence in these diverse patient groups permits well-established practice guidelines and appropriateness criteria that incorporate SPECT MPI in many clinical decision-making scenarios (3).
In principle, risk stratification is based on the ability to discriminate patient risk on the basis of a test result, and in this regard, a normal SPECT MPI is strongly associated with low risk, whereas gradations of risk can be identified with progressive levels of SPECT MPI abnormality. Pooled data from published series of patients with normal SPECT MPI undergoing a >2-year follow-up identified a cardiac death or nonfatal myocardial infarction rate much lower than 1% per year, corresponding to well-accepted definitions of “low risk” (3). As might be expected, the risk associated with any level of SPECT MPI abnormality varies widely with baseline patient characteristics, including patient age, diabetic state, inability to exercise (i.e., pharmacologic SPECT MPI), abnormal rest electrocardiogram (ECG), prior CAD, and presence of dyspnea. In each of these patient subtypes, the findings of SPECT MPI add incremental value beyond the clinical state.
SPECT MPI and benefit from revascularization procedures
Although abnormal SPECT MPI prognosticates heightened risk of death and myocardial infarction, it does not necessarily follow that treatment of these individual patients with abnormal tests results in reduction of events. As the preponderance of CAD responsible for myocardial infarction or sudden cardiac death is due to nonobstructive coronary artery plaques for which SPECT MPI would expected to be normal, the challenge to demonstrate that SPECT MPI leads to improved patient outcomes—consistent with a value-based health care system—has lead to a subtle but important shift in research direction in the field. These more contemporary efforts have focused on whether the SPECT MPI findings can assist in the determination of therapeutic strategies with salutary effects on clinical outcome.
In a large single-center registry of 10,627 patients without prior CAD, SPECT MPI was examined for its ability to determine image-based thresholds that can guide appropriate treatment strategies (Fig. 1) (4). In matched cohorts—adjusted for differences between medically treated and revascularized patients and post-test referral patterns—patients with extensive ischemia by SPECT MPI exhibited a clear survival benefit when treated with revascularization (Fig. 1). By contrast, among those with no ischemia, cardiac death rates were higher with revascularization than with medical therapy. The cut point at which ischemia tipped the balance towards revascularization appeared to be ∼10% ischemic myocardium.
Another proof-of-principle, but nevertheless landmark study addressed how SPECT testing may shape clinical outcomes in the nuclear substudy of the COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) trial (5). In this study of 314 patients in whom both pre-randomization and 6- to 18-month post-randomization SPECT MPI was performed, patients assigned to percutaneous coronary intervention and optimal medical therapy (OMT) demonstrated greater ischemia reduction when compared with patients receiving OMT alone (33% vs. 19%, p = 0.0004). Importantly, the frequency of adverse events was higher in those with greater post-treatment ischemia. Although this study was not designed to prove that ischemia reduction directly reduces adverse CAD events, it nevertheless provides supportive evidence that imaging of ischemia does affect patient outcomes through guiding decisions for revascularization.
Although prognostic SPECT MPI data were primarily derived from well-performed large-scale registries, there is nevertheless a dearth of evidence from SPECT MPI from primary analyses of large randomized trials. Furthermore, as SPECT-MPI evolved in an era that preceded today's highly cost-conscious environment, few studies have evaluated the impact of SPECT MPI on measures related to cost. Additionally, there are clear limitations of SPECT-MPI: although it is clear that patients with normal SPECT-MPI generally have an excellent prognosis, problems in guiding individual patient management frequently arise after abnormal SPECT-MPI. Further, if the patient has a high pre-test likelihood of CAD, a normal SPECT study does not rule out obstructive CAD. Particularly bothersome in this regard are reports of underestimation of left main coronary artery or triple-vessel CAD by SPECT. Problems also arise in the substantial proportion of patients with equivocal or mildly abnormal SPECT studies. Although some early data suggested that medical management would be cost effective in this group as a whole, more recent data have shown that the hard cardiac event rates after testing are not low in many subgroups of these patients. Given these considerations, clinicians are often uncertain as to appropriate patient management following SPECT MPI.
Introduction of coronary CTA: a new algorithm of evaluation?
Coronary CTA has emerged in a contemporary era, where focus on evidence development for cardiac imaging transcends that which has been developed for older technologies, and instead proposes as necessary studies on patient-centered outcomes and cost effectiveness prior to widespread use. Indeed, critics of coronary CTA have suggested that studies are needed to demonstrate that use of coronary CTA provides additive value above and beyond that which is currently available. Although the effect on patient outcomes has not yet been shown with randomized trials, several studies have assessed the clinical benefit of coronary CTA.
Despite rapid improvements in CT technology, studies evaluating the diagnostic capability of coronary CTA have kept pace, with newer studies evaluating the specific performance of coronary CTA with 64-detector rows. Recent multicenter prospective trials have also evaluated the diagnostic performance of 64-detector row coronary CTA, confirming its high diagnostic efficacy for CAD detection in varying patient cohorts (6). Of note, the diagnostic receiver-operator characteristic curve area for detecting obstructive CAD, comparing coronary CTA with ICA, have shown clearly higher diagnostic accuracy than with any other noninvasive test.
It should be noted that the diagnostic accuracy studies with coronary CTA are less limited by referral biases than those that have affected SPECT MPI results, since the decision to perform ICA on these patients was not governed by the coronary CTA result. Nonetheless, biases still remain as selected patients were already identified as needing ICA and thus, had a relatively high pre-test likelihood of CAD. Critics have noted that the generalizability of coronary CTA study results to patients with low or intermediate pre-test likelihood of CAD—where the test is more commonly applied—is limited, although ethical conduct precludes subjecting such individuals to unnecessary invasive procedures. Notably, the ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) trial, the only study to date performed in low-intermediate CAD prevalence cohort, reported high sensitivity and specificity of coronary CTA. The ACCURACY trial, however, also reveals both the strengths and inherent limitations of testing with even a highly accurate modality in a relatively low-prevalence population (6). The negative predictive value (NPV) was extremely high (99%)—essentially ruling out the presence of CAD; however, as predicted by Bayes' theorem, the positive predictive value (PPV) was low (35% to 45%). The implications of the negative coronary CTA are clear: the patient with a normal study—free of coronary atherosclerosis—can now be treated with high confidence as not having CAD.
Given the recent introduction of 64-row coronary CTA, long-term survival analyses are not yet available. However, the predictive value of coronary CTA to assess intermediate-term risk has now been studied. In a recent study of 1,127 patients undergoing coronary CTA (7), gradations of risk for mortality were evident by increasing coronary artery plaque severity, location and distribution (Fig. 2). These data are the first, to our knowledge, to demonstrate that coronary CTA is accurate for prognosticating mortality risk, although limited by the small number of events recorded. The corollary to these findings is that a “normal” coronary CTA, as evidenced by no evident coronary artery plaque, conferred an exceptionally high rate of survival of 99.7% in the 15-month follow-up (Fig. 2) (7). Additional evidence using electron beam CT suggests that the “warranty period” of a normal coronary CTA extends to at least 7 years (8).
A recent study provided the first comparison of coronary CTA and SPECT MPI to predict incident all-cause mortality (9). In a prospective evaluation of 541 patients undergoing both coronary CTA and SPECT MPI followed for 1.8 years, the presence of CAD at the 50% stenosis threshold by coronary CTA demonstrated equivalent predictive value for all-cause mortality and nonfatal myocardial infarction, as compared with a summed stress score (SSS) ≥4 by SPECT MPI. Interestingly, the prognostic value of coronary CTA was independent and synergistic to SSS ≥4 by SPECT MPI.
Coronary CTA, resource utilization, and cost effectiveness
To assess how coronary CTA might influence resource utilization and costs, 1,938 patients undergoing coronary CTA were compared with 7,752 matched patients undergoing SPECT MPI (10). Downstream total and CAD costs for individuals undergoing coronary CTA were 27% (p < 0.001) and 33% (p < 0.001) lower compared with those undergoing SPECT MPI. These differences translated to a cost savings for coronary CTA, beyond the baseline test cost, of $445 (p < 0.01).
Importantly, no differences existed between coronary CTA and SPECT MPI individuals for rates of revascularization. Furthermore, individuals undergoing coronary CTA or SPECT MPI as an initial diagnostic test did not differ in rates of CAD hospitalization (4.2% vs. 4.1%), myocardial infarction (0.4% vs. 0.6%), or new-onset angina (3.0% vs. 3.5%). As the specific test results were not available, direct comparison of the relative cost effectiveness of these tests cannot be made. Nevertheless, to date, the totality of studies evaluating the cost efficiency of coronary CTA in individuals without known CAD indicates that it is cost saving compared with SPECT MPI, without differences in adverse CAD outcomes.
Coronary CTA has been introduced during a time of spiraling health care costs and reliance on noninvasive imaging tests in cardiology. Thus, increased vigilance for unnecessary testing is now desired, and the result appears to be the “raising of the bar” for new technologies such as coronary CTA. As mentioned before, certain medical policy advocates now suggest that coronary CTA, as well as other imaging modalities, not only be subject to traditional measures of diagnostic and prognostic efficacy, but also provide clear benefit of patient-centered outcomes and cost effectiveness. Data from large-scale registries will generate constructive hypotheses, but randomized clinical trials will be necessary to adequately answer the questions being posed. In addition, direct comparative effectiveness trials are necessary to evaluate new versus existing imaging tests (e.g., coronary CTA vs. SPECT MPI) and to study the ability of the information derived from imaging to guide patient management (e.g., SPECT MPI-determined referral for revascularization).
Although initial coronary CTA studies have begun to address these issues, full examination of benefit-oriented criteria is complex and more expensive to study compared with traditional assessments of diagnostic performance and risk stratification. Accordingly, funding entities now calling for such study should also be forthcoming with support of clinical trials that will be required to establish this new type of evidence.
The Health Care Crisis
Raymond Gibbons, MD
In his speech to congress, on February 24, 2009, President Obama said:“…we must also address the crushing cost of health care. …Health care reform cannot wait, it must not wait, and it will not wait another year” (11).
The President's words eloquently summarize the growing realization that the long-standing need for national health care reform is more urgent every day. Although many politicians, and commentaries, focus on the problem of the uninsured, the problem of escalating costs is probably more important. As shown in Figure 3, the annual growth in both public (federal and state) and private health care spending has consistently exceeded the annual growth in gross domestic product (GDP) over the past 30 years. The cumulative increase in private health care spending is dramatically reflected in health insurance premiums (Fig. 4), which are now 320% of what they were in 1991. During the same time, the Consumer Price Index has grown at less than half this rate. At least part of the increase in private health insurance premiums can be attributed to cost shifting, as Medicare reimbursement has fallen further and further behind the cumulative effects of inflation. Although the decrease in physician reimbursement that is mandated by the Sustainable Growth Rate (SGR) formula has not occurred since 2002, the modest increases that have occurred since then have been far less than inflation, widening the gap between Medicare physician reimbursement and inflation.
The crisis in health care costs will soon get much worse, with the retirement of the baby boom generation beginning in 2011. The combination of this aging of the population and the annual growth in health care spending will take an enormous toll on the federal budget (Fig. 5). By 2047, federal spending on Medicare and Medicaid will exceed 20% of GDP. In the history of our country, total federal revenue has never exceeded 20.9% of GDP. Thus, by 2047, there will be no federal revenue available for Social Security, interest on the federal debt, aid to education, defense, disaster aid, or anything else.
All of this “bad news” can be summarized in a single statement: “The current system is not sustainable” (12,13).
Imaging, particularly high-technology imaging, is part of the problem. The use of CT, magnetic resonance (MR), and positron emission tomography (PET) increased from 42 procedures per 1,000 Medicare beneficiaries in 1995 to 163 procedures per 1,000 Medicare beneficiaries in 2005 (Fig. 6), a nearly 4-fold increase. This compounded annual increase of 16% per year predated the use of CTCA. Other cardiac imaging procedures have also increased at a surprising rate. Between 1993 and 2001, stress imaging (stress SPECT imaging and stress echocardiography) increased in Medicare patients at an annual rate of 6% per year, which was far in excess of the increase in cardiac catheterization, revascularization, or acute myocardial infarction (14).
Congress is increasingly concerned about the growth of imaging. Medicare reimbursement for the technical component of high-technology imaging services was reduced in early 2006. In 2008, Congress mandated that the Secretary of Health and Human Services develop an imaging appropriateness demonstration project by 2010 and that all imaging laboratories performing services under Medicare be accredited by 2012. The National Priorities Partnership, which includes multiple federal agencies, multiple consumer groups, the AARP, the U.S. Chamber of Commerce, and the National Quality Forum (which includes both the American College of Cardiology [ACC] and the American Heart Association [AHA]) recently announced new goals, one of which is to “eliminate waste while insuring the delivery of appropriate care” (15). Among the 9 areas targeted for a 50% reduction are unwarranted diagnostic procedures, including CTCA.
During the past decade, the cardiovascular community has often embraced broader application of new technologies as part of an “increase the volume/grow the business” mentality, in response to declining rates of Medicare reimbursements. The cardiovascular community quickly embraced the use of pulsed Doppler and tissue Doppler echo parameters to assess the response to cardiac resynchronization therapy. Despite numerous reports that various parameters were useful, a carefully conducted, international, multicenter trial found that most of these parameters could not be reproducibly measured, and that the single parameter that was reproducible, the left ventricular pre-ejection interval, was not predictive of the response to resynchronization therapy (16).
Our health care system cannot afford to repeat this mistake with CTCA. As responsible stewards of our nation's health care resources, we must demand definitive evidence that CT coronary angiography has added value, either in improved patient outcomes, or reduced health care costs, or in some combination of the 2, before we adopt it on a widespread basis.
Existing approach to the noninvasive assessment of CAD
There is a well-established approach to the assessment of patients with chest pain and known or suspected CAD, which is reflected in existing clinical practice guidelines from both the ACC/AHA (17) and the European Society of Cardiology (ESC) (18). The first step in the evaluation of such patients is a careful history and physical examination to provide an estimate of the pre-test likelihood of CAD. The next step is often noninvasive stress testing with an exercise ECG, stress SPECT MPI, or stress echocardiography. These stress tests provide objective evidence of ischemia to determine whether the patient has functionally significant CAD that likely explains his or her symptoms. The test results can be placed into 3 broad categories:
1. When there is no evidence of ischemia, the patient can be reassured that CAD is not a likely cause of their symptoms, and alternative diagnoses can be pursued.
2. When there is evidence of ischemia, but the test results are not high risk, the patient can be managed medically.
3. When the test results show evidence of ischemia and a high risk of subsequent cardiac events, the patient should be referred for coronary angiography.
This strategy relies on an enormous published evidence base over many decades supporting the use of both the exercise ECG and stress imaging. The oldest portion of the evidence base relates to the diagnosis of CAD, and is well summarized in multiple clinical practice guidelines (Online Table 2). The evidence supporting the prognostic value of both the exercise ECG and stress imaging studies is equally robust, and probably even more important. Dozens of studies in tens of thousands of patients have consistently shown that patients with known or suspected CAD who do not have demonstrable stress-induced ischemia have very low cardiac event rates over the next 3 to 7 years (Online Tables 3 to 5) (19). Revascularization will likely not reduce the rate of subsequent death or myocardial infarction in such patients. If their symptoms prove refractory to medical therapy, they may require coronary angiography, but the goal of revascularization at that time will usually be to reduce symptoms rather than to improve hard outcomes.
In contrast, patients with severe stress-induced ischemia have far higher rates of subsequent cardiac events and therefore warrant early coronary angiography to determine whether they are candidates for revascularization to improve their outcomes. This strategy recognizes that there will often be differences between the functional significance of a coronary lesion and its anatomic appearance. The mere presence of an anatomic lesion is not sufficient to justify revascularization, as clearly reflected in existing clinical practice guidelines. The ESC Percutaneous Coronary Intervention Guidelines (20) have a class I recommendation for percutaneous intervention in stable coronary disease if there is “objective evidence of a large area of ischemia.” The ACC/AHA/Society of Coronary Angiography and Intervention Percutaneous Coronary Intervention Guidelines (21) have a class IIa recommendation for patients who are asymptomatic or mildly symptomatic when the vessels “subtend a moderate to large area of viable myocardium or…[are] associated with moderate or severe degree of ischemia on noninvasive testing.” Thus, objective demonstration of stress-induced ischemia is a well-established principle reflected in existing clinical practice guidelines for the management of patients with chronic CAD.
CT angiography compared with invasive angiography
In comparison, the published evidence base supporting CTCA is promising but limited. Multiple studies have shown that CTCA using the latest 64-slice scanners can produce results that compare favorably with ICA. A recent systematic review and meta-analysis (22) analyzed data from 28 studies in 1,286 patients, and reported a pooled sensitivity of 99% and a pooled specificity of 89%. However, these studies have, by necessity, focused on patients referred for ICA. Such patients are unlikely to reflect the full spectrum of patients with chest pain and known or suspected CAD. In general, they are more likely to have a high pre-test likelihood of CAD, and prior noninvasive tests showing ischemia.
Since publication of this meta-analysis, 2 significant additional studies have been reported. Budoff et al. (23) focused on patients with a low-to-intermediate risk of CAD in an important effort to broaden the evidence base. However, this study only included 32 patients with significant CAD by the usual 70% stenosis definition, a disease prevalence of 13.9% that is more “low” than “intermediate.” This disease prevalence is surprisingly low for patients with a mean age of 58 years and symptoms of typical and atypical angina, raising some concern about the generalizability of the results. Moreover, given the limited number of patients with CAD in the study, the lower 90% confidence limits for sensitivity and specificity were 76% and 79%, respectively. The study, therefore, did not achieve its stated goal of excluding values <80% for both sensitivity and specificity.
Miller et al. (24) reported results on 291 patients with calcium scores of 600 or less. The overall prevalence of significant CAD was considerably higher at 56%, using a stenosis of greater than 50% according to quantitative ICA. Although CT angiography did identify the presence and severity of obstructive CAD, the PPV of 91% and NPV of 83% were somewhat disappointing, and the authors appropriately concluded that “CT angiography cannot replace conventional coronary angiography at present.”
The results of Miller et al. (24) were consistent with several previous studies in the literature that found a limited association between stenosis severity by CTCA and stenosis severity by ICA. In 2005, Leber et al. (25) reported a correlation coefficient of only 0.54, and showed that many stenoses judged to be 20% to 70% by one approach were 0% by the other approach. More recently, Meijboom et al. (26) also reported a limited association (r = 0.53), although they did not find examples of such extreme discordance.
CT angiography compared with noninvasive stress testing
The more important question for practicing physicians is not whether CTCA will replace ICA, but whether it will replace noninvasive stress testing with or without imaging. What is the evidence comparing 64-slice CT coronary angiography with the approaches commonly used in clinical practice in the assessment of patients with known or suspected CAD: the exercise ECG, stress echocardiography, and stress SPECT MPI?
I am not aware of a single published study comparing the exercise ECG with 64-slice CTCA. It is possible that I have missed such studies, or that there are comparative data incorporated in the results section of other reports.
Similarly, I am not aware of a single published study comparing the results of stress echocardiography with 64-slice CT coronary angiography.
Finally, only a few studies have compared 64-slice CT coronary angiography with SPECT MPI. Figure 7 shows 5 representative studies (27–31) that have used a standardized approach to SPECT and include sufficient detail to permit calculation of PPV and NPV. Note that a total of only 385 patients are included. Only 1 of these studies is from the U.S. CTCA does have a high NPV (i.e., the absence of obstructive disease by CT angiography is generally associated with a negative SPECT perfusion image). However, the NPV is not 100%. This could reflect false positive perfusion images, or the presence of true stress-induced ischemia in the absence of obstructive disease, a well-established phenomenon (32,33). The PPV in these studies ranged from 45% to 67%. This may reflect the assessment of stenoses of intermediate severity, which might or might not be physiologically significant. However, several of these studies have reported that severe stenoses by CT angiography were not associated with detectable ischemia in a considerable percentage of patients.
The spectrum of patients included in these limited studies is probably not representative of the broad population of patients with stable chest pain, as 2 of the 4 studies included patients with known CAD, and a majority of the patients in 3 of the studies were referred for invasive angiography.
As indicated earlier, the evidence base supporting the prognostic value of noninvasive stress testing is at least as important as those studies examining the diagnostic value of the exercise ECG and stress imaging. What data are currently available regarding the prognostic value of CTCA in this regard? The available data have been well summarized elsewhere (34). At this time, the only prognostic studies with 64-slice CT angiography have limited follow-up, as the technology only recently became available. The only study with a mean follow-up of ≥5 years, a reasonable time frame for chronic CAD, utilized electron beam coronary angiography in a large cohort, but the symptomatic status of the patients was not reported (35).
Despite the health care crisis, and the limited evidence base in support of CTCA, some have continued to argue for more widespread application of this promising technology in the belief that the “evidence will certainly come.” Even if one is willing to accept this argument, which I am not, there remain some important unresolved questions that must be answered before widespread clinical application.
How will we evaluate and manage patients with discordant results by CTCA and stress imaging? Some patients will undergo anatomic assessment using CT coronary angiography and functional assessment using stress imaging. The results will not always be consistent. As already mentioned, some patients with “moderate” stenosis by CT coronary angiography will have functional evidence of ischemia; some will not. How will we manage patients without functional ischemia: what follow-up tests will be performed, and at what intervals? Will clinicians feel compelled to perform ICA in such patients to “clear the air”? Van Werkhoven et al. (36) have reported that 25% of patients with normal stress SPECT images will have obstructive CAD on CTCA; of these, 5% will have high-risk CAD. What will we tell these patients? How will they be managed, both initially and over time?
How will we manage patients with “mild CAD” by CTCA? This finding might improve patient compliance with lifestyle change and risk factor modification to prevent both progression of CAD and future cardiac events. However, it will also increase anxiety in a subset of patients, who realize for the first time that they have atherosclerosis. Will such patients demand or expect routine follow-up studies by CTCA to make certain that their disease has not progressed? To contemplate using CTCA for this purpose, we would need a lot more information about its reproducibility for the assessment of mild-to-moderate stenoses. Without this information, it will be impossible to judge whether a change over time really represents a change in the patient's lesion.
Will CT coronary angiography reduce the subsequent rate of ICA? This is often assumed to be the case. There are two published observational studies (37,38) suggesting that the use of CTCA may help to clarify “equivocal” or “intermediate-risk” SPECT MPI image results and thereby avoid ICA. However, given the poor association between the severity of stenosis by CTCA and ICA, will there be patients with “mild” disease by CTCA who are then referred to ICA? As I have detailed elsewhere (19), I have already seen an example in another medical center of just such a patient, who then underwent immediate intervention without any documentation of ischemia or symptoms. A carefully conducted regional study from Canada with contemporary controls (39) demonstrated that the rate of normal invasive coronary angiograms decreased after introduction of CTCA into their practice, but the overall rate of ICA increased.
Finally, there is the question of population radiation burden. In experienced hands, prospective gating can greatly decrease the radiation exposure involved in CTCA to a range that is similar to SPECT MPI. However, without these measures, the radiation exposure and risk will be considerably higher, particularly in young women. If CTCA becomes “the standard” for evaluating episodes of chest pain, how frequently will it be repeated in young women with atypical symptoms? Despite convincing published evidence that the “warranty period” for a normal exercise SPECT study in a patient without known CAD is greater than 5 years in young patients without diabetes, preliminary data from our laboratory (40) shows that 91% of the “routine” follow-up SPECT studies conducted in such patients are done before expiration of the warranty period. When will clinicians order follow-up CTCA in the absence of evidence about the “warranty period” of CT angiography in symptomatic patients?
Need for future studies
CTCA is a very promising technology that merits and demands future study. Shah et al. (41) have described the possible parameters for a carefully designed prospective study of CTCA for the diagnosis of CAD. I have personal knowledge of 2 other proposals for similar studies that were submitted to the National Heart, Lung, and Blood Institute this year. The design of a study of CTCA in patients with peripheral vascular disease has also been published (42). These are the kind of prospective, multicenter, “real-world” studies that need to be done before CTCA is adopted on a widespread basis. Congress has recognized the need for increased federal funding for comparative effectiveness research. The economic stimulus package signed into law included 1.1 billion dollars in incremental funding to the Agency for Healthcare Research and Quality, the National Institutes of Health, and the Secretary of Health and Human Services to help support such research. I personally hope that some of this incremental funding helps to support future studies of CTCA. The future of this promising technology, as well as the growing crisis in health care in our country, demand it. We as evidence-based physicians should await the results of these studies before embracing CTCA on a widespread basis. It is not yet ready for prime time.
Dr. Min serves on the speaker's bureau and medical advisory board and receives research support from General Electric Company. Drs. Berman and Shaw receive research support from General Electric. Dr. Gibbons has a research grant from King Pharmaceuticals for development of an adenosine agonist.
For additional Tables and References please see the online appendix.
Clinical Benefits of Noninvasive Testing: Coronary Computed Tomography Angiography as a Test Case
- American College of Cardiology Foundation
- Baucus M.
- ↵(2008) Baucus-Conrad Proposal Can Improve Quality, Lower Costs Throughout American Health Care System (U.S. Senate Committee on Finance, Washington, DC) August 2008 http://finance.senate.gov/press/Bpress/2008press/prb080108.pdf. Accessed February 10, 2010.
- Klocke F.J.,
- Baird M.G.,
- Lorell B.H.,
- et al.
- Hachamovitch R.,
- Hayes S.W.,
- Friedman J.D.,
- Cohen I.,
- Berman D.S.
- Shaw L.J.,
- Berman D.S.,
- Maron D.J.,
- et al.
- Budoff M.J.,
- Dowe D.,
- Jollis J.G.,
- et al.
- Min J.K.,
- Shaw L.J.,
- Devereux R.B.,
- et al.
- Ostrom M.P.,
- Gopal A.,
- Ahmadi N.,
- et al.
- van Werkhoven J.M.,
- Schuijf J.D.,
- Gaemperli O.,
- et al.
- Min J.K.,
- Shaw L.J.,
- Berman D.S.,
- Gilmore A.,
- Kang N.