Cardiovascular mortality remains the leading cause of death in Brazil, with atherosclerosis and ischemic heart disease playing prominent roles.¹ The high prevalence and mortality rates related to cardiovascular disease in Brazil can be largely attributed to poorly controlled risk factors. Controlling the main cardiovascular risk factors has great impact on reducing cardiovascular death.²

Many highly prevalent diseases with high mortality and morbidity rates are often diagnosed only after the first cardiovascular event, which can cause serious sequelae or even death. Considering this scenario, it is essential to establish cardiovascular assessment mechanisms that enable the identification and proper management of patients with cardiovascular disease.

The clinical presentation of patients with ischemic heart disease is quite variable, ranging from acute syndromes to silent ischemia.³ Symptoms include a late event in the ischemic cascade, and the patient already presents cardiac repercussions and increased risk of events long before their onset.⁴ Nuclear medicine is a method with greater sensitivity and power for the earlier detection of myocardial ischemia although with lower specificity.⁴

The active search for disease in the asymptomatic patient begins with a detailed history and physical examination.³ When a patient is asymptomatic, the assessment of myocardial ischemia proceeds in the form of screening, with the definition of cardiovascular risk occurring through the collection of clinical data and completion of basic complementary tests.³

In this context, myocardial perfusion scintigraphy (MPC) is for a method of investigating myocardial ischemia in some subgroups of asymptomatic patients. Some current applications of the test come from the appropriate use criteria (AUC) for myocardial perfusion scintigraphy (MPS) published in 2005 by the American College of Cardiology Foundation and the American Society of Nuclear Cardiology⁵ and updated in 2009.⁶ This document was an attempt to improve clinical outcomes with the efficient and fair allocation of healthcare resources based on the best available scientific evidence with the aim of improving the cost-effectiveness of medical care without restricting clinical judgment.⁵ These criteria were validated in 2013⁷ and have been incorporated since then into the latest ischemic heart disease, cardiovascular risk assessment, and nuclear medicine guidelines.

This study aimed to critically analyze the use of these criteria for asymptomatic patients at high cardiovascular risk undergoing MPS.

This cross-sectional, retrospective, and observational study involved the analysis of patients’ medical records. The study was developed at the Nuclear Medicine Department of Dante Pazzanese Institute of Cardiology. The project was registered at Plataforma Brasil and approved by the local ethics committee.

All patients who underwent MPS between December 6, 2017 and July 16, 2018 were selected for medical record analysis.

A total of 2,999 medical records of patients undergoing MPS to investigate ischemia were selected; 2,245 of them were available for analysis at the time of referral. The study included all patients undergoing MPS to investigate ischemia who had no symptoms of myocardial ischemia at the time of referral and the test, i.e., patients not reporting typical or atypical precordial pain, discomfort, or ischemic equivalent.⁸

All patients undergoing pharmacological stress testing were included. Patients with ineffective exercise test results were excluded unless the test result was considered positive for ischemia or presented transient low uptake in the imaging stage. A total of 490 patients met the inclusion criteria and were included in the final statistical analysis.

MPS was performed following the one- or two-day-protocol according to the service routine with or without the suspension of anti-ischemic medications according to the clinician and the standardized recommendations of the nuclear medicine department. The classical known and standardized criteria for myocardial ischemic response to exertion were adopted.⁹

The tests were performed in Millennium VG scintillation chambers (GE Medical Systems, Milwaukee, WI, USA), with two scintillation detectors angled at 90° with high-resolution low-energy parallel-hole collimators. Processing was performed using Xeleris™ workstations and Cedars-Sinai software, with sections in the major and minor vertical plane axes and in the major horizontal plane axis. Rest and stress stage sections were paired for the visual and qualitative analysis of radiopharmaceutical concentrations in the 17 segments of the myocardium and synchronized with the electrocardiography for the analysis of cardiac wall contraction, systolic and diastolic volume indices, and left ventricular ejection fraction.

According to well-established guidelines,¹⁰ the MPS was considered normal when the concentration of the marker was homogeneous in both phases (stress and rest). If the low uptake was reversible after rest, the test result was considered suggestive of ischemia. If the low uptake was fixed in both phases, the test result was considered suggestive of fibrosis. Finally, in case of associated fixed and reversible low uptakes, the test result was considered suggestive of fibrosis and ischemia.

Quantitative variables are described as mean, standard deviation, and minimum and maximum values. Categorical variables are described as frequencies and percentages. Fisher’s exact test or the chi-square test was used to assess the association between two categorical variables. Logistic regression models were adjusted for the univariate and multivariate analyses of factors associated with the outcomes of interest (fixed low uptake, transient low uptake, and changed test for at least one of fixed low uptake, transient low uptake, or transient ischemic dilation). The variables for the multivariate models were selected using two approaches: the univariate analysis results and the clinical importance of the variables, followed by stepwise backward regression with a probability of 0.10 for the output of the variables. The Wald test was used to assess the significance of each variable in the models. The estimated measure of association was the odds ratio with 95% confidence interval. P values < 0.05 indicated statistical significance. The data were analyzed using Stata/SE software version 14.1 (StataCorpLP, USA).

Overall, 22.27% of the total number of patients undergoing MPS to assess myocardial ischemia were asymptomatic. The clinical and epidemiological characteristics of the studied population are shown in Table 1.

The rate of tests considered appropriate by the AUC was 61.4%; that of tests considered uncertain was 28.8%; and that of tests considered inappropriate was only 9.8%. The predominant type of stress test requested was pharmacological with dipyridamole (61.4%). Exercise stress testing was performed in 38.4% of cases, all with treadmill exercise test and a predominance of modified Bruce and Bruce protocols.¹¹

Analysis of the imaging stage of the test showed a fixed low uptake in 43.5% of cases and a transient low uptake in 16.1%. Changed test results were defined as those with transient low uptake and/or fixed low uptake and/or transient ischemic dilation. The univariate analysis is shown in Table 2.

In the multivariate analysis ( Tables 3 and 4), the two models were adjusted according to the selection of the independent variables to be included. One of the models considered the results of the univariate analysis and the clinical relevance of the factors. The other, exclusively technical, considered the stepwise backward model, which makes the selection according to a pre-established criterion regarding statistical significance for the inclusion or exclusion of variables.Figures 1 and 2 present the arrangement of these data in a more visual and graphic format.

Analysis of the association between changed test results and degree of recommendation according to AUC showed a higher prevalence of changed test results in patients with appropriate and uncertain indications, especially in the group with uncertain indications. A similar result occurred when uncertain and appropriate indications were grouped ( Table 5).

The AUC criteria found in the study population and used to classify the indications as appropriate, inappropriate, or uncertain are presented in Table 6.

The AUC were not applied to the entire population of asymptomatic patients at the service; rather, it was applied respectively to only patients for whom an assessment of ischemia with MPS was previously requested. This characteristic makes the present study an assessment of test requests rather than an overall assessment of its use since it does not have the power to assess the adequacy to the AUC of this segment of an asymptomatic population of patients who did not undergo the test but may have been indicated for it.

Considering that greater care is needed for indications for MPS for asymptomatic patients and that there are already defined criteria for this indication with different degrees of recommendation according to the AUC, there was doubt as to whether these criteria would be adequate for assessing a population at such high cardiovascular risk.

The studied population was marked by high cardiovascular risk due to several factors. First, it was composed of patients already screened at other health services and referred due to their greater cardiac complexity, high prevalence of multiple risk factors, and difficultly controlling these risk factors for the most diverse reasons. These characteristics are evident in the results, with 87.3% of the patients in the study being at high cardiovascular risk, more than half already having known coronary artery disease, less than 30% having a low-density lipoprotein (LDL) level within the recommended target for patients at high cardiovascular risk, and almost half having an outpatient blood pressure above the recommended range.¹² Also, this population differed from the one assessed in the study of the validation of the use of the AUC published in 2013.⁷

The rate of inappropriate indications (9.8%) was quite low. Besides the expertise of a reference cardiology teaching center, a likely explanation for this result is that the high cardiovascular risk characteristic of this population is a factor alone that increases the degree of the recommendation in several of the AUC for asymptomatic patients. Another possible contribution to this result may be the usual difficulty in scheduling appointments and tests with ideal deadlines in the Brazilian Unified Health System given the extreme demand, leading to longer than desirable intervals between appointments. Under these circumstances, the test is performed later than it probably would be if the public services were not as saturated, which increases the probability of a properly performed test since, for many criteria, there is a minimum interval for the test not to be considered an inappropriate indication.

Another point to be carefully observed is that, although the ischemia assessment is the reason for the MPS, the division of criteria into appropriate, uncertain, and inappropriate does not necessarily mean that changed test results are expected in appropriate indications versus normal test results in inappropriate indications. This classification also consider the test’s ability to provide prognostic information and the usefulness of its result in the clinical context, and not just the exam result alone.

Sometimes it was difficult to find a criterion whose specifications met the patient’s exact profile. At other times, more than one criterion fit the patient’s profile, raising questions about which one to use. Therefore, doubtful patients were classified according to the criterion that was closest to their clinical situation or, in those with more than one possible criterion, the one referring to the condition of greatest clinical importance in the investigator’s opinion was chosen. There is a suggested hierarchy for the AUC; however, it remained difficult to choose the most appropriate criterion in situations with more complex conditions. The degree of recommendation changed a few times when there was doubt; when it did, it generally oscillated between uncertain and appropriate indications, which are degrees of recommendations with good acceptability in the definition of conduct as well as in patient management.

The ischemia assessment revealed that 16.1% of the target population of this study had ischemia, i.e., a rate similar to the one of another study conducted at the same service but that also included patients with atypical symptoms.¹³ This study was expected to present a lower rate of ischemia since it excluded patients with atypical symptoms, but this did not occur. This could mean that, in such a high-risk population, the prevalence of ischemia would remain similar regardless of the inclusion of patients with atypical symptoms.

Another Brazilian study¹⁴ addressed a specific situation contemplated in some of the AUC, specifically the prognostic value and clinical use of MPS in asymptomatic patients after percutaneous coronary intervention (PCI). MPS was suggested to play an important role in asymptomatic patients after they underwent angioplasty regardless of the time the test was performed after the procedure, i.e., the two-year milestone recommended of the guidelines could not separate patients with more severe outcomes. However, that study reported the important role of suspected incomplete revascularization in the indication of most of these tests. Considering the AUC published in 2009, which has specific criteria for suspected incomplete revascularization in addition to the criteria related to the time to perform the test after PCI, patients with suspected incomplete revascularization should not be evaluated with other asymptomatic patients in the post-PCI setting, as the role of MPS in suspected incomplete revascularization is well established.

The correlation between tests performed with appropriate or uncertain degree of recommendation and changed test results showed the good ability of the AUC to predict changed results.

The univariate analysis showed a correlation between classic risk factors and perfusion changes. The association between the use of anti-ischemic drugs and a changed test result is probably the result of a more aggressive clinical treatment in more severe patients. The same reasoning can be used to analyze that lower, not higher, LDL levels are correlated with a higher rate of changed test results.

In the multivariate analysis, when the variables involved were analyzed together, a personal history of CAD was the factor most associated with changed test results. MPC with pharmacological stress maintained a statistically significant association in the multivariate analysis as well, a result that has been demonstrated in other studies.^15-17 The possible explanation that has been suggested is that patients undergoing pharmacological stress testing would have more comorbidities and higher cardiovascular risk than those undergoing exercise stress testing, therefore increasing the association between changed test results and worse clinical outcomes.¹⁸

If, on the one hand, the study has limitations of being cross-sectional, observational, retrospective, and subject to the degree of quality of the notes taken in medical records; on the other hand, it approaches a realistic view of the current functioning of this service. Calculating cardiovascular risk with the calculator proposed by the Dyslipidemia Division of the Brazilian Society of Cardiology¹⁹ differs from what is proposed in the AUC, which used the ATP III score;²⁰ however, it is currently the most used commonly used form in this service and the one that best reflects how a patient’s cardiovascular risk has been calculated.

In addition, for the purpose of grouping into criteria, some patients were allocated to criteria that do not exactly represent their clinical situation, but this limitation was already described in the AUC publication.⁶ The authors did not exclude or create a category for unclassified patients who did not exactly fit into any of the criteria since this tool should be used to guide rather than restrict medical care. Thus, the results must be analyzed from the perspective of the power of such criteria to guide the search process for the correct use of the test. Additionally, the classification of indications as adequate or inappropriate is the result of an individual interpretation of the AUC by the investigator.

With the purpose of critically analyzing the role of the AUC in asymptomatic patients within a population, this study demonstrated that their retrospective application to guide the referral for MPS could predict that the adequate indication of the exam (appropriate and uncertain indications) is correlated with changed test results, especially in the subgroup of patients classified as having uncertain indications. Perhaps, in a population with a high cardiovascular risk as the one in this study, some indications considered uncertain might actually be appropriate.

Finally, considering these results, further studies are needed to identify whether certain clinical situations classified as uncertain in the AUC would be subject to changes in the degree of recommendation when the test is used in an asymptomatic population with high cardiovascular risk.