Author + information
- Published online February 4, 2019.
- Stephen G. Ellis, MD∗ ()
- ↵∗Address for correspondence:
Dr. Stephen G. Ellis, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, Ohio 44195.
Silent ischemia is not uncommon, but gathering evidence about its prevalence is challenging, as these patients often lack findings that trigger investigation for ischemia. However, ischemia has been found on 2.6% of SPECT scans from asymptomatic subjects with calcium scores of 11 to 100 (about the 50th percentile for a 65-year-old) and in 46% of asymptomatic subjects with scores >400, so it’s prevalence is certainly not rare (1,2).
Once detected, how should we best treat silent ischemia? In contrast to the situation with other better studied scenarios in cardiology, physicians lack high-quality data to inform thinking about how best to manage these patients. In this issue of JACC: Cardiovascular Interventions, Czarnecki et al. (3) present a relatively large, inclusive, and comprehensively analyzed nonrandomized study of patients with silent ischemia from Ontario, Canada, and report: 1) wide variation in the use of coronary revascularization across centers; and 2) an apparent strong benefit of revascularization (19% reduction in death and 42% reduction in myocardial infarction [MI] after a median 4.6 years of follow-up).
The first finding is expected, given the absence of definitive data on how to manage such patients and numerous reports of regional differences in procedural utilization (4). The second finding is perhaps more surprising, as cardiac therapies seldom lead to such a large reduction in risk for MI.
To investigate the plausibility of the latter finding, one must ask what high-quality data we have in similar patients, and what confounders might have been missed in this analysis (think of why you as a cardiologist would refrain from revascularizing a >70% stenosis in someone, albeit asymptomatic, you decided warranted catheterization: artery too small, arteries too diffusely narrowed, unapproachable chronic total occlusion, patient otherwise has a poor prognosis, or you do not believe revascularization is useful without symptoms).
Lessons from Outcomes in Similar Patients
To address the issue of similar patients, one might suppose that revascularization would have generally similar results in patients with silent ischemia and stable angina. Patients with stable ischemic symptoms and their physicians do often hope that percutaneous coronary intervention (PCI) of obstructive lesions will reduce the risk for MI and make them live longer. This latter hypothesis has been formally tested in the setting of silent ischemia only in the minority subset of patients with recent MI. The very small SWISSI II trial, reported a “perhaps too good to be true” 70% to 75% reduction in death and MI at 10 years for those patients randomized to revascularization (5).
However, many high-quality randomized clinical trials of PCI versus optimal medical therapy alone in patients with stable angina were performed in the pre–fractional flow reserve era, and at best, meta-analyses show a small reduction in long-term MI and mortality with PCI. The benefits are attenuated if one considers the risk (albeit controversial) for periprocedural MI. Several meta-analyses have studied this question using slightly different methodologies and have reached largely similar results. Using data from 8 randomized trials and 7,729 patients, Stergiopoulos et al. (6) found a 2% reduction in death and a 12% excess risk for MI with revascularization. About the same time, Pursnani et al. (7) studied 7,182 patients from 12 randomized trials and found a nonsignificant 15% reduction in death and a 7% reduction in MI. Importantly, Bangalore et al. (8) found a significant 24% reduction in spontaneous MI at the expense of a 4 times increase in index hospital MI with PCI plus optimal medical therapy versus optimal medical therapy alone.
When analyses are limited to studies requiring objective evidence of ischemia (not the case with Czarnecki et al. ), results with revascularization appear more favorable, with a significant 48% reduction in MI with PCI found in a 2015 meta-analysis (9). More recently, using fractional flow reserve to better evaluate the obstructive nature of specific lesions and cull out less prognostically important lesions from PCI, the FAME 2 (Fractional Flow Reserve Versus Angiography for Multivessel Evaluation) investigators found a 34% reduction in MI (p = 0.05), but no difference in mortality at 5 years (10).
Although most randomized trials are limited because of concerns about whether they can be generalized, as sicker more complex patients with multiple comorbidities are excluded, the concern with even well-conducted multivariate-corrected registries is that of unmeasured confounders. Why did the physician choose to treat a patient in a given fashion?
When we think about risk factors for death and MI that might not have been analyzed by Czarnecki et al. (3), where should we start, and how much might they affect the results? Studies of causes of death in persons 45 years and older in the United States, such as that from the Centers for Disease Control and Prevention, find heart disease, cancer, and “other” causes to be at the top, with 25% to 30% attributable deaths each, followed by chronic lung diseases, stroke, Alzheimer’s disease, and chronic liver disease at about 5% each (11). Age matters too. Data on cause of death in patients with coronary artery disease are harder to find. Analysis of our post-PCI database, focusing on about 8,000 patients without acute coronary syndromes treated since 2000, found the most powerful risk factor for all cause death to be low baseline hemoglobin, followed by advanced age, chronic obstructive pulmonary disease, low left ventricular ejection fraction, heart failure class, number of diseased vessels, and diabetes in that order. Unfortunately, we do not track other intuitive risk factors well known to treating physicians, such as advanced malignancy (not all cancers increase short-term risk for death), end-stage renal disease, critical limb ischemia, and frailty. Risk factors for MI include, of course, extent of disease, control and treatment of risk factors, exercise, socioeconomic status, and the stress of high- or intermediate-risk surgery (12,13). Importantly, most cardiovascular databases do not capture key details about noncardiac risk factors (e.g., the differences for patients with creatinine clearance of 15 vs. 59 [both chronic kidney disease], FEV1 of 1.9 vs. 0.5 [both chronic obstructive pulmonary disease], or basal cell vs. metastatic lung cancer [both malignancy]), all of which might well change treatment and most certainly change prognosis. Summarizing, there are many noncardiac causes of death for which we are not well prepared to adjust for risk factors, which may affect the results of studies such as that of Czarnecki et al. (3).
When we further consider the issue of confounders, we run across several reports of paradoxical undertreatment of sicker patients. For example, Cohen et al. (14), reporting from the very large CRUSADE (Can Rapid Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines) registry, found patients not undergoing early catheterization after non–ST-segment elevation MI to be older and have many more serious comorbidities than those who were offered invasive evaluation (14). Under use may be exaggerated under the stresses of public reporting of outcomes (15).
How to Weigh the Importance of Periprocedural “Enzyme Bumps”?
Before deciding how much credence to give the results of this study, and how we will manage these patients until better data become available, we also must decide how to define and diagnose periprocedural MI for both PCI and coronary artery bypass grafting. Physicians from different subspecialties cannot seem to agree on this. Depending on the definition, perirevascularization MI rates can range from 3% to 15% or more, and deciding which cutoff to use can totally alter the answer to the question “Which treatment is associated with lower risk for death and MI?” Interventionalists, of course, have “solved” this problem, in large part because of concerns raised by public reporting, by just no longer measuring post-PCI cardiac enzymes. In this trial, the definitions of both periprocedural and late MI are not provided.
Comparing Registry and RCT Outcomes
Last, it is well to recall results from better studied questions in which there are both registry and randomized trial data, to study the effect of confounding. A classic is that of Kirtane et al. (16), who found a 22% reduction in death with drug-eluting stents versus bare-metal stents (p < 0.001) in 182,000 patients from nonrandomized registries and absolutely no difference in mortality between the 2 types of stents in 9,470 patients studied in randomized clinical trials (p = 0.72). Sicker patients received bare-metal stents. Might sicker patients in this study have gotten the less aggressive therapy?
With all of this in mind, I would argue that the investigators of registry data attempting to ascertain a benefit of an intervention be circumspect with their conclusions (“these observational data suggest that there may be” rather than “a clinical benefit was observed”).
How to treat these patients? Until we have better data, perhaps with “a lot” of ischemia they should be revascularized, and with “a little ischemia” they should not. For me, it is worth the effort to try to enroll such patients in meaningful trials so that their successors can be treated with more confidence that we know what is best. The ISCHEMIA (International Study of Comparative Health Effectiveness With Medical and Invasive Approaches) trial (albeit already somewhat controversial) will have 20% to 25% of patients with silent ischemia (S. Bangalore, personal communication). I cannot wait until it is presented in 2019.
↵∗ Editorials published in JACC: Cardiovascular Interventions reflect the views of the authors and do not necessarily represent the views of JACC: Cardiovascular Interventions or the American College of Cardiology.
Dr. Ellis serves as a consultant for Abbott Vascular, Boston Scientific, and Medtronic.
- 2019 American College of Cardiology Foundation
- He Z.X.,
- Hedrick T.D.,
- Pratt C.M.,
- et al.
- Czarnecki A.,
- Qiu F.,
- Elbaz-Greener G.,
- et al.
- Pinho-Gomes A.C.,
- Azevedo L.,
- Ahn J.M.,
- et al.
- Pursnani S.,
- Korley F.,
- Gopaul R.,
- et al.
- Bangalore S.,
- Pursnani S.,
- Kumar S.,
- Bagos P.G.
- Gada H.,
- Kirtane A.J.,
- Kereiakes D.J.,
- et al.
- Fearon W.F.,
- Nishi T.,
- De Bruyne B.,
- et al.
- Cohen M.G.,
- Filby S.J.,
- Roe M.T.,
- et al.
- Kirtane A.J.,
- Gupta A.,
- Iyengar S.,
- et al.