Cardiac investigation for prognosis in coronary artery disease: where negative is positive

Despite improvements in health care, cardiovascular disease remains the leading cause of morbidity and mortality worldwide.1 New-onset suspected stable angina is a widespread clinical problem. Following Bayesian analysis, if a patient has a very low-pretest probability of coronary artery disease (CAD), the risk of a hard cardiac event (cardiac death or acute myocardial infarction) is very low and therefore no further test is required for prognosis. Conversely, patients with suspected stable angina and an intermediate to high-pretest probability of CAD should benefit from risk stratification by a non-invasive test that has the ability to predict a good outcome if the test is normal.2 This provides a sort of ‘warranty’ period for at least the medium term and helps to reassure both the patient and physician. An unequivocally normal result also precludes the need for further cardiac investigations and this can be both comforting and convenient for the patient. Furthermore, in an era of increasing cardiac investigation and spiralling health care costs, a reliably normal/negative initial test which can avoid further testing should benefit health care systems by delivering both efficient, high-quality care and at the same time limit costs. Presently, a plethora of non-invasive cardiac investigations are available to investigate stable chest pain for suspected or known coronary artery disease. These investigations may be divided into functional tests that detect reversible myocardial ischaemia or anatomical tests which delineate the coronary artery arteries. Functional tests involve stress testing with physiological exercise or pharmacological agents for the detection of myocardial ischaemia. The most widely used test continues to be the exercise electrocardiography (ExECG) which has a Class 1A indication for patients who can exercise and have no resting ECG abnormalities.2,3 However, imaging techniques which are widely available and commonly used are single-photon-emission tomography (SPECT) and stress echocardiography (SE). Newer techniques include cardiac magnetic resonance (CMR) imaging and positron-emission tomography (PET). Coronary computerized tomographic angiography (CCTA) is a non-invasive technique that assesses the coronary artery anatomy. All of these techniques are advocated for the assessment of suspected angina both by the European and the American guidelines.2,3 The meta-analysis conducted by Smulders et al.4 is timely, addressing the relative prognostic value of a negative test of the above contemporary cardiac imaging techniques, in symptomatic patients with stable chest pain. The analysis was well conducted and took into account the history of known CAD, prevalence of CAD and actual hard cardiac event rate in each of the studies selected for the analysis. True to the present clinical scenario, SPECT followed by SE formed the bulk of the study population, closely followed by CCTA. The number of CMR trials for the assessment of CAD either by wall motion or perfusion was relatively small comprising 25% and 10% of the SE and SPECT population, respectively; the PET population was considerably smaller. Unlike real-world circumstances, the number of patients who underwent ExECG was lower than that evaluated by SPECT, SE, or CCTA and this can be accounted for by publication bias. To the best of our knowledge, this was the largest and most robust meta-analysis on this subject and the conclusions were clear. If by functional testing there was no evidence of myocardial ischaemia or by non-invasive anatomic testing there was no obstructive coronary artery disease, then the event rates were comparable and very low. An initial apparent lower event rate of a normal CCTA over a normal SPECT/SE was accounted for after adjustment for the population event rate and prevalence of known CAD. This was also true when patients without known CAD and suspected angina were evaluated. Thus, according to the meta-analysis, a normal CCTA which excludes significant obstructive CAD directly does not confer any lower risk compared to a normal functional test which excludes reversible ischaemia. This was clearly highlighted by two large prospective randomized studies comparing CCTA with functional testing—PROMISE and SCOT-HEART. Both trials showed that strategies based on anatomic vs. functional testing have similar outcomes.5,6 However, it must be pointed out that the mean duration of follow-up of the population in the meta-analysis was intermediate at around 2 years, similar to the population in PROMISE and SCOT-HEART trials. It was also very interesting to note that CCTA was predominantly performed in a very low-risk group in whom the prevalent hard event rate was only 0.72%/year. A negative scan reduced it by 50%. In contrast, the prevalent event rate of the population evaluated by SPECT/SE was 4–5 times higher with reduction in event rate by 50% which is similar to CCTA when the functional test was negative for ischaemia. It is similarly interesting to note that ExEcg was performed in those with a 20% prevalence of previous AMI and possibly resting ECG changes. While good exercise tolerance predicts good outcomes, ECG changes may lead to unnecessary downstream tests. Another aspect to ponder on, though not shown in this study, is that the negative predictive value for cardiac events did not change significantly in those with an intermediate pretest probability of CAD, demonstrated in a meta-analysis comparing SE and SPECT.7 In a meta-analysis, it needs to be highlighted that when comparing two different studies, even if most confounding factors are adjusted for, there is always a possibility that certain peculiarities specific to a given population studied may not have been captured. Ideally, when two techniques are compared, it should be in the same patient population. The authors acknowledged the intrinsic limitation of this type of research methodology and analysis. Another caveat to consider in this meta-analysis is the definition of normality. For functional testing, the absence of reversible abnormalities was considered normal. It is known that fixed WMA/perfusion defect carries risk. This factor was not adjusted for in the analysis which may account for borderline statistical significance between CCTA and SE after other potential confounding factors were adjusted for. However, when patients with suspected CAD only are considered the difference between CCTA and SE were abolished probably because of low likelihood of prevalence of fixed abnormalities in this population. So, given that the outcome of a negative test is similar across all contemporary technologies what factors should determine the best initial test for the investigation of patients with suspected stable angina? Let us consider the largest group of patients investigated for CAD i.e. those with new-onset suspected stable angina and no previous history of CAD. Both PROMISE and SCOT-HEART demonstrated that event rates in such a population are low, approximating to 1.5%/year. This is likely because of healthier lifestyles, better control of modifiable cardiovascular risk factors, and earlier institution of preventative measures. Thus, to investigate such patients one needs a relatively safe test (ALARA principle—as low as reasonably achievable), one which is widely available, highly feasible, provides high-diagnostic tests (low-inconclusive tests) and has a low-initial cost. Both CCTA and nuclear scans are techniques that involve ionizing radiation; the latter is also relatively costly. CMR is not readily available and is costly. SE and Ex ECG are widely available, do not involve ionizing radiation and are relatively inexpensive. However, ExECG can only be performed in those able to exercise and is indeterminate in approximately 20% of patients giving rise to further downstream testing.6 On the contrary, SE with the advent of transpulmonary contrast agents is diagnostic in almost 99% of patients.8 In a recent randomized study, cost to diagnosis of CAD was cheaper with SE than ExECG in patients presenting with suspected stable angina.8 There is now evidence to suggest that SE with transpulmonary contrast agents could be the initial test of choice in such a population.9 A further advancement is the ability of ultrasound to assess atherosclerosis and myocardial ischaemia simultaneously using carotid ultrasound and SE, respectively. This combination provided incremental prognostic information in patients with suspected CAD.10 Reassuringly, this information by SE-CU is very similar to that provided by SPECT-CT.11 In the high-risk population, similar considerations should be applied to determine the most cost-effective and safe approach for diagnostic testing. Conflict of interest: R.S. has received speaker fees from Bracco (Milan, Italy), Lantheus Medical Imaging and Philips (Eindhoven, Holland). R.S.K.: None.