- Email: [email protected]
Long-Term Prognostic Value of Appropriate Myocardial Perfusion Imaging
Long-Term Prognostic Value of Appropriate Myocardial Perfusion Imaging Angela S. Koh, MBBS, MPHa,b,*, Weng Kit Lye, MScc, Shaw Yang Chia, BSca, Jennifer Salunat-Flores, MDa, Ling L. Sim, MSca, Felix Y.J. Keng, MBBSa,b, Ru San Tan, MBBSa,b, and Terrance S.J. Chua, MBBSa,b Appropriate use criteria (AUC) for single-photon emission computed tomography myocardial perfusion images (SPECT-MPIs) were developed to address the growth of cardiac imaging studies. Long-term prognostic value of AUC in SPECT-MPI has not been tested in existing cohorts. We sought to determine the long-term prognostic value of MPI classified as appropriate. AUC was evaluated in a prospectively designed cohort of patients who underwent clinically indicated MPI. MPI studies were classified based on 2009 AUC for SPECT-MPI. Data regarding downstream coronary angiography (cath), revascularization and all-cause mortality, cardiac death, and nonfatal myocardial infarction (MI) were collected from national registries. Among n [ 1,129 MPI scans that received an appropriate grading, 148 allcause deaths, 109 MIs, 58 cardiac deaths, 152 caths, 113 revascularization procedures occurred over a mean follow-up period of 5.4 – 1.2 years (0.9% cardiac death rate per year, 1.8% MI rate per year). Most of the scans were low-risk normal MPI scans (summed stress score £3; 74.1%). An abnormal scan was associated with higher rates of MI (19.5% vs 6.2%, hazard ratio 1.72, p [ 0.017) and cardiac death (13.4% vs 2.3%, hazard ratio 2.12, p [ 0.016). In conclusion, MPI scans classified as appropriate have long-term prognostic value, despite a high proportion of low-risk scans. This provides support for clinicians to consider the use of appropriate grading in addition to MPI scan results in patient management. Ó 2017 Elsevier Inc. All rights reserved. (Am J Cardiol 2017;119:1957e1962) The rise in utilization of myocardial perfusion examinations in the United States led to the initial development of imaging guidelines for nuclear cardiology by the American College of Cardiology Foundation, American Heart Association, and American Society of Nuclear Cardiology (ASNC) in 2009.1 Since then, using these guidelines, groups within and outside the Unites States have examined the appropriate use of single-photon emission computed tomography myocardial perfusion imaging (SPECT-MPI) known as appropriate use criteria (AUC), including assessments of downstream cardiac procedures such as coronary angiography, coronary revascularization, and short-term major adverse cardiac events, in relation to grading based on AUC.2e7 SPECT-MPI variables are well established as strong predictors of clinical outcomes with solid long-term data. As we increasingly examine the use of AUC in SPECT-MPI studies, more data containing longer term follow-up of AUC, particularly the value of an MPI classified as appropriate are important, particularly in an era of a National Heart Centre Singapore, Singapore, Singapore; bDukeNational University of Singapore Medical School, Singapore, Singapore; and cCentre for Quantitative Medicine, Duke-National University of Singapore Medical School, Singapore, Singapore. Manuscript received December 27, 2016; revised manuscript received and accepted March 10, 2017. Dr. Koh has received grants from National Medical Research Council of Singapore (NMRC/TA/0031/2015). See page 1962 for disclosure information. *Corresponding author: Tel: (þ65) 6704-8961; fax: (þ65) 6222-9258. E-mail address: [email protected] (A.S. Koh).
0002-9149/17/$ - see front matter Ó 2017 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2017.03.026
balancing risk-to-benefit ratio related to radiation exposure8 and clinical outcomes.9 In this regard, whether an appropriately classified SPECT-MPI has long-term prognostic value, over longer term follow-up beyond what has been published, remains unknown. In this study, we test the hypothesis that an appropriate SPECT-MPI scan, continues to prognosticate adverse cardiac events over the long term. Methods We have previously reported the AUC grading of all consecutive studies referred to our MPI laboratory for single-photon emission tomography (SPECT) between February 2009 and July 2009 according to the 2009 American College of Cardiology/American Heart Association AUC.7 Briefly, the patient’s medical history and indications for testing were recorded on the day of their MPI appointment. Based on AUC, MPI studies were then immediately classified into appropriate, inappropriate, uncertain, or unclassified groups. Imaging was performed with a stress-rest protocol using either technetium-99m tetrofosmin or sestamibi. Stress testing was performed by exercise or dipyridamole with rest and poststress-gated SPECT imaging at the discretion of the nuclear cardiologist. The study was approved by the local institutional review board. Electrocardiogram-gated MPI was performed using a stress first, rest second, protocol using dual-head gamma cameras (Philips CardioMD and Philips Vertex). Singlephoton emission computed tomography (SPECT) MPI was performed in accordance with guidelines published by the ASNC. Images were acquired in step-and-shoot mode over a www.ajconline.org
1958
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline clinical and imaging characteristics Variable
Appropriate (n¼1129)
Age (years SD) Men Women Weight (kg SD) Chinese Malay Indian Others Hypertension Hyperlipidemia Diabetes mellitus Smoker Family History of CAD Prior MPI Test Indications for MPI Detection of symptomatic CAD Detection of CAD or risk assessment without ischemic equivalent Risk assessment with prior test results and/or known chronic stable CAD Preoperative risk assessment Risk assessment within three months of an acute coronary syndrome Risk assessment post PCI or CABG Myocardial perfusion Normal (SSS¼0-3) Mildly abnormal (SSS¼4-8) Moderately abnormal (SSS¼9-13) Severely abnormal (SSS>13) Myocardial ischemia None (SDS¼1) Mild (SDS¼2-4) Moderate (SDS¼5-7) Severe (SDS>7) Left Ventricular Ejection Fraction 60% < 60%
61.6 10.6 683 (60.5%) 446 (39.5%) 67.6 14.2 826 (73.2%) 94 (8.3%) 165 (14.6%) 44 (3.9%) 817 (72.4%) 879 (77.9%) 363 (32.2%) 179 (15.9%) 180 (15.9%) 114 (10.1%) 620 (54.9%) 64 (5.7%) 178 (15.8%) 250 (22.1%) 15 (1.3%) 2 (0.2%) 837 138 77 77
(74.1%) (12.2%) (6.8%) (6.8%)
890 140 61 38
(78.8%) (12.4%) (5.4%) (3.4%)
754 (66.8%) 375 (33.2%)
CABG ¼ coronary artery bypass graft surgery; CAD ¼ coronary artery disease; MPI ¼ myocardial perfusion imaging; PCI ¼ percutaneous coronary intervention; SSS ¼ summed stress score; SDS ¼ summed difference score.
180 semicircular orbit with 64 stops. No attenuation correction was applied. After filtered backprojection, shortaxis, horizontal, and vertical long-axis tomographic slices were created. The images were normalized to local maximal myocardial activity. Reconstructed SPECT-MPI slices were interpreted semiquantitatively using a 20-segment model. Each segment was scored by consensus by 3 nuclear cardiologists (F.K., T.R.S., and T.C.) using a 5-point scoring system for radiotracer uptake (0 ¼ normal, 1 ¼ mild reduction of tracer uptake, 2 ¼ moderate reduction of uptake, 3 ¼ severe reduction of uptake, and 4 ¼ absence of uptake). The summed stress score (SSS), summed rest score, and summed difference score (SDS) were calculated in the usual manner. The SSS is a measure of both reversible and fixed defects. The SDS is a measure of exercise-induced ischemia. Categorizations of summed scores as normal,
Figure 1. Flow chart of study cohort.
mild, moderate, and large are listed in Table 1. Left ventricular ejection fraction (LVEF) was derived from poststress SPECT images using commercially available software (AutoQuant; Cedars-Sinai Medical Center, Los Angeles, California). Vital status was evaluated through record linkage with the National Registry of Births and Deaths to determine overall mortality. It is a statutory requirement that death must be registered within 24 hours of its occurrence in the whole country. Cardiac deaths were ascertained based on International Classification of Diseases (ICD), Ninth Revision, codes 410 to 414, as well as ICD 10 I21.9, I24.9, I25.1, and I25.9. For the computation of the incidence of acute myocardial infarction (AMI), all the episodes included were identified centrally through the Singapore Myocardial Infarction Registry and diagnosed as. (1) Definitive AMIedefinite ECGs, or symptoms (typical or atypical), together with probable ECG and abnormal biochemical markers suggestive of myocardial necrosis, or typical symptoms and abnormal biochemical markers with ischemic/noncodable/unavailable ECG. (2) Clinical AMIeelectrocardiographic changes suggestive of AMI but not supported with raised cardiac biochemical markers or typical symptoms, or at least 2 of the following criteria: clinical history of prolonged chest pain >20 minutes; raised biochemical markers of myocardial necrosis; serial electrocardiographic tracings showing ST-T changed from baseline or Q waves duration that are 0.03 seconds in 2 or more contiguous leads. Furthermore, ICD-9 Clinical Modification code of 410 was used to identify AMI cases in the data sources before 2012, whereas ICD-10 Australian Modification codes I21 and I22 were used for the cases diagnosed from 2012 (inclusive) onward. Patients who were nonresidents (n ¼ 80) and had repeated MPI (n ¼ 8) were excluded from the current analysis. Patients with complete information in outcomes and covariates were used in the analysis. Patients with AUC grading other than appropriate were excluded from this analysis. For this analysis, there were 1,129
Coronary Artery Disease/Long-term Prognostic Value of Appropriate MPI
1959
Table 2 Cardiac procedures based on presence of ischemia (SDS 2) Procedure Coronary angiography* Coronary revascularization†
Total procedures (n¼1129)
Ischemia (n¼239)
No ischemia (n¼890)
OR (95% CI)z
P-valuez
152 (13.5%) 113 (10.0%)
99 (41.4%) 78 (32.6%)
53 (6.0%) 35 (3.9%)
5.33 (2.98, 9.53) 11.01 (6.90, 17.56)
<0.001 <0.001
CI ¼ confidence interval; OR ¼ odds ratio; SDS ¼ summed difference score. * Coronary angiography procedures occurring within 1 year after the index cardiac catheterization. † Coronary revascularization (PCI/CABG) procedures occurring within 1 year after the index cardiac catheterization. z OR (95% CI) and p values are derived from logistics regression models between groups of ischemia and no ischemia (ref.), adjusted for clinical covariates (including age/gender/weight/hypertension/hyperlipidemia/diabetes mellitus/smoking/family history of coronary artery disease [CAD]/EF).
patients recorded at the end of the follow-up period (Figure 1). Outcome assessors were blinded to MPI findings and AUC classification. Coronary angiography and revascularization (percutaneous coronary intervention [PCI] and coronary artery bypass graft [CABG]) procedures performed within 1 year after the MPI were considered to be triggered by the MPI findings. The primary end point was composite end point of cardiac death or nonfatal MI. Secondary end point was all-cause mortality. Continuous variables were summarized as mean (SD) and 2-sample t-test was used to compare between 2 normally distributed variables. Categorical variables were summarized as count (percentage), and the chi-square test was used to compare dichotomous variables. A multivariable logistic regression model was used to estimate the odds ratio of ischemia to nonischemia groups with coronary angiography procedures and coronary revascularization (PCI/CABG) procedures. The model was adjusted for age, gender, weight, hypertension, hyperlipidemia, diabetes mellitus, smoking status, family history of coronary artery disease, and EF. Both univariate and multivariable Cox proportional hazard regression models were used to estimate hazard ratios of abnormal MPI to normal MPI with (1) allcause mortality, (2) myocardial infarction (MI), (3) all-cause mortality or MI, (4) cardiac mortality, and (5)cardiac mortality or MI as outcomes. The multivariable model was adjusted for the same set of clinical covariates above with including coronary revascularization. Two-sided p <0.05 was considered as statistically significant. SAS version 9.4 (SAS Institute, Cary, North Carolina) was used to perform the analysis. Results The study cohort consisted of 1,129 consecutive patients referred for clinically indicated SPECT-MPI study (Figure 1), followed up for mean 5.4 1.2 years. Baseline patient characteristics are summarized in Table 1. There were 60.5% men with a mean age of 61.6 10.6 years. Over 72.4% of the study cohort had hypertension, 77.9% had hyperlipidemia, whereas 32.2% had diabetes mellitus. Most of the MPI scans were normal (SSS 3; 74.1%), had lower proportions of myocardial ischemia (SDS 1 [78.8%], and with LVEF of 60% [66.8%]). Over the course of follow-up, there were 152 (13.5%) coronary angiography (cath) procedures with subsequent coronary revascularization in 113 (10.0%) (either PCI or CABG surgery) that occurred within 1 year of the MPI scan (Table 2). Patients with stress-induced ischemia (defined as
SDS 2) had significantly higher rates of cath (41.4% vs 6.0%, p <0.001) and revascularization rates (32.6% vs 3.9%, p <0.001) than those without ischemia. Patients with ischemia had significantly higher odds of cath and revascularization than those without ischemia (all p <0.001). There were 148 all-cause mortality deaths (13.1%), of which 58 were cardiac deaths (5.1%) and 109 (9.7%) nonfatal MIs. Patients with abnormal MPI had higher rates of all-cause mortality, cardiac death, and nonfatal MI than those with normal MPI (Table 3). In the presence of a normal MPI scan, we observed annualized events rate of 0.4% for cardiac death and 1.1% for MI. Risk-adjusted survival curves demonstrate a striking difference with respect to primary end point of cardiac death or nonfatal MI particularly between normal, moderately, and severely abnormal MPI scans (Figure 2). After adjustments for clinical covariates, event rates for primary end point of cardiac death or nonfatal MI, but not the secondary end point of all-cause mortality, were higher in patients with abnormal MPI than those patients with normal MPI (hazard ratio 1.81, 95% CI 1.23 to 2.66, p ¼ 0.003; Figure 3). Discussion Following on this quality assessment project that started in the year 2009 where we had previously used the 2009 AUC for nuclear cardiology by the American College of Cardiology Foundation, American Heart Association, and ASNC, to assess the appropriate use of myocardial perfusion scans in our center,7 including examination of shortterm clinical outcomes data among a subset of MPI studies,10 we provide new evidence in this analysis that demonstrates long-term prognostic value of AUC. In this prospective cohort of patients followed for at least 5 years, we observed sustained long-term value of an appropriate grading in prognosticating hard clinical outcomes of patients referred for clinically indicated MPI. We demonstrate that an appropriate grading continued to predict outcomes, taking into account clinical risk factors, and traditional MPI variables. This demonstrates the robust value of an appropriate grading within the AUC, observed for the first time in a tertiary level setting, surpassing previous data observed in more community-based settings and which were limited to shorter duration of follow-up.4 A major strength of our study is the prospective design of the study sample. We had deliberately designed this study as a quality and outcomes project right from the start in 2009. The collection of all study variables were prospective, in particular assessment of appropriate grading was performed
1960
The American Journal of Cardiology (www.ajconline.org)
Table 3 Clinical events based on degree of myocardial perfusion imaging abnormality (SSS 4) Event
Total Events
All-cause mortality Myocardial infarction (MI) All-cause mortality or MI Cardiac Death Cardiac Death or MI
148 109 218 58 144
(13.1%) (9.7%) (19.3%) (5.1%) (12.8%)
Abnormal MPI (n¼292) 71 57 103 39 79
Normal MPI (n¼837)
(24.3%) (19.5%) (35.3%) (13.4%) (27.1%)
77 52 115 19 65
(9.2%) (6.2%) (13.7%) (2.3%) (7.8%)
HR (95% CI)* 2.89 3.42 2.96 6.28 3.94
(2.10, (2.35, (2.26, (3.63, (2.84,
4.00) 4.98) 3.86) 10.88) 5.47)
p-value <.0001 <.0001 <.0001 <.0001 <.0001
Adjusted HR (95% CI) 1.27 1.72 1.39 2.12 1.81
(0.86, (1.10, (1.02, (1.15, (1.23,
1.87) 2.69) 1.91) 3.91) 2.66)
p-value 0.223 0.017 0.040 0.016 0.003
CI ¼ confidence interval; HR ¼ hazard ratio; MPI ¼ myocardial perfusion imaging; SSS ¼ summed stress score. * HR (95% CI) and p values were derived from Cox proportional hazards models between group of abnormal MPI and normal MPI, adjusted for clinical covariates (including age/gender/weight/hypertension/hyperlipidemia/diabetes mellitus/smoking/family history of coronary artery disease [CAD]/EF/ revascularization).
Figure 2. Outcomes based on myocardial perfusion imaging findings. Time-to-outcome Cox proportional hazards curves based on severity of perfusion abnormality (SSS). The hazard ratios (HRs) were calculated relative to the normal group and adjusted for clinical covariates.
by independent observers at the point of performing the MPI scan. This crucial step minimized potential misclassification biases otherwise obtained from retrospective chart review or computer logic. Owing to a robust national diseases registry where death notification was a mandatory requirement across the country by law, and a national MI registry where MI were centrally monitored, the study team accrued prospective clinical outcome data over the years, minimizing losses to follow-up.
Despite being a tertiary center, the accrual of a high event rate over time was a function of the long period of follow-up and less because the study sample was a high-risk population, commonly seen in tertiary settings. Although there was a rather high burden of coronary risk factors in the study sample, most of the MPI studies (74.1%) were normal, with the majority having no ischemia (78.8%) or normal EFs (66.8%). Such a sample would intuitively provide greater test to the AUC, by showing that these
Coronary Artery Disease/Long-term Prognostic Value of Appropriate MPI
1961
Figure 3. Outcomes based on myocardial perfusion abnormality. Time-to-outcome Cox proportional hazards curves of patients with abnormal versus normal MPI. Hazard ratios (HRs) were calculated and adjusted for clinical covariates.
appropriate MPI scans prognosticated clinical events despite being generally low-risk MPI scans. Longer term outcomes studies such as ours draw conclusions based on the actual sustaining value of the appropriate grade, powered by events over time, rather than by high rates of events driven in the shorter term by highly abnormal MPI scans. In addition, although we had defined a composite primary end point of cardiac death or nonfatal MI, each end point of cardiac death or nonfatal MI was by itself significantly powered to draw conclusions about their associations with appropriate grading. The scans with myocardial ischemia had understandably greater use of subsequent coronary angiography and revascularization, consistent with other studies that have looked at rates of downstream utilization in relation to AUC.3 Even in the presence of a normal MPI commonly regarded as low risk, we observed a considerable annualized cardiac death or nonfatal MI rates of 0.4% and 1.1%, respectively, in this cohort, where all MPI scans were classified as appropriate. For years, normal SPECTMPI has been used to provide a “warranty” for low cardiac risk, associated with low rates of events such as <0.5% risk of cardiac death or <0.3% of risk of MI per year.11 Since then, various patient cohorts have been identified
as having increased risk despite normal SPECT MPI scans, such as patients with diabetes mellitus,12 atrial fibrillation13 and patients who undergo pharmacological stress.14 Our results suggest that patients with normal MPI classified as appropriate, may benefit from closer clinical scrutiny. This is especially novel because hardly any prospective cohort have reported long-term event rates among normal MPI defined as SSS 3, particularly defined by an appropriate classification. Our novel data may reflect the intrinsic quality embedded within AUC, not encompassed within traditional SPECT MPI scan variables, and which therefore provides a fresh angle to support the use of AUC as a risk stratification tool in the assessment of quality imaging.9 Our observation further indicates the need for more data to report the prognostic value of an appropriate grading with normal MPI, among other patient cohorts. Finally, the prognostic value of SPECT is heavily dependent on the patient population being studied. We acknowledge that this study, although representative of a tertiary center within our country, may only apply to settings similar to ours and may not apply to other tertiary centers in other parts of the world. However, our study represents steps along the road to quality improvement, provides novel data
1962
The American Journal of Cardiology (www.ajconline.org)
toward quality imaging, which is the ultimate goal of the AUC.15 We acknowledge limitations in the study. First, our study is observational and therefore does not allow for causal inferences. Second, we attributed coronary angiography and revascularization procedures within 1 year from the index MPI scan as being triggered by the index MPI scan result. This duration may be viewed as being prolonged compared with other studies that use 180 days from MPI result, possibly diminishing the effect that cath was truly triggered by the MPI result. Sensitivity analyses adjusting for duration between MPI scan and procedures might provide better clarification, which was not done. However, our results emphasizing the long-term hard clinical outcomes are still the main message of the study. In addition, we did not include assessment of other adverse MPI variables such as transient ischemic dilatation; however, given the low rates of abnormal MPI scans in this cohort, such adverse features might not have made a difference to the conclusions. Finally, it is still possible that adjustments may not have accounted for unknown confounders, leading to the possibility of residual confounding. Acknowledgment: The authors are indebted to the patients and the healthcare staff of the nuclear laboratory for making this study possible over the years. Disclosures The authors have no conflicts of interest to disclose. 1. Hendel RC, Berman DS, Di Carli MF, Heidenreich PA, Henkin RE, Pellikka PA, Pohost GM, Williams KA. ACCF/ASNC/ACR/AHA/ ASE/SCCT/SCMR/SNM 2009 appropriate use criteria for cardiac radionuclide imaging: a report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, the American Society of Nuclear Cardiology, the American College of Radiology, the American Heart Association, the American Society of Echocardiography, the Society of Cardiovascular Computed Tomography, the Society for Cardiovascular Magnetic Resonance, and the Society of Nuclear Medicine. Circulation 2009;119:e561ee587. 2. Doukky R, Hayes K, Frogge N. Appropriate use criteria for SPECT myocardial perfusion imaging: are they appropriate for women? J Nucl Cardiol 2016;23:695e705. 3. Khawaja FJ, Jouni H, Miller TD, Hodge DO, Gibbons RJ. Downstream clinical implications of abnormal myocardial perfusion single-photon emission computed tomography based on appropriate use criteria. J Nucl Cardiol 2013;20:1041e1048.
4. Doukky R, Hayes K, Frogge N, Balakrishnan G, Dontaraju VS, Rangel MO, Golzar Y, Garcia-Sayan E, Hendel RC. Impact of appropriate use on the prognostic value of single-photon emission computed tomography myocardial perfusion imaging. Circulation 2013;128:1634e1643. 5. Gibbons RJ, Miller TD. Single-photon emission computed tomography appropriateness: does it matter for patient outcomes? Circulation 2013;128:1595e1597. 6. Soine LA, Cunningham SL, Motzer SA, Inoue LY, Caldwell JH. Application of appropriate use criteria for stress myocardial perfusion imaging at two academic medical centers: compliance and association with image findings. J Am Acad Nurse Pract 2012;24:200e208. 7. Koh AS, Flores JL, Keng FY, Tan RS, Chua TS. Evaluation of the American College of Cardiology Foundation/American Society of Nuclear Cardiology appropriateness criteria for SPECT myocardial perfusion imaging in an Asian tertiary cardiac center. J Nucl Cardiol 2011;18:324e330. 8. Doukky R, Frogge N, Appis A, Hayes K, Khoudary G, Fogg L, Williams KA Sr. Impact of appropriate use on the estimated radiation risk to men and women undergoing radionuclide myocardial perfusion imaging. J Nucl Med 2016;57:1251e1257. 9. Bhattacharyya S, Lloyd G. Improving appropriateness and quality in cardiovascular imaging: a review of the evidence. Circ Cardiovasc Imaging 2015;8:e003988. 10. Koh AS, Flores JL, Keng FY, Tan RS, Chua TS. Correlation between clinical outcomes and appropriateness grading for referral to myocardial perfusion imaging for preoperative evaluation prior to non-cardiac surgery. J Nucl Cardiol 2012;19:277e284. 11. Hachamovitch R, Berman DS, Shaw LJ, Kiat H, Cohen I, Cabico JA, Friedman J, Diamond GA. Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death: differential stratification for risk of cardiac death and myocardial infarction. Circulation 1998;97:535e543. 12. Berman DS, Kang X, Hayes SW, Friedman JD, Cohen I, Abidov A, Shaw LJ, Amanullah AM, Germano G, Hachamovitch R. Adenosine myocardial perfusion single-photon emission computed tomography in women compared with men. Impact of diabetes mellitus on incremental prognostic value and effect on patient management. J Am Coll Cardiol 2003;41:1125e1133. 13. Abidov A, Hachamovitch R, Rozanski A, Hayes SW, Santos MM, Sciammarella MG, Cohen I, Gerlach J, Friedman JD, Germano G, Berman DS. Prognostic implications of atrial fibrillation in patients undergoing myocardial perfusion single-photon emission computed tomography. J Am Coll Cardiol 2004;44:1062e1070. 14. Rozanski A, Gransar H, Hayes SW, Friedman JD, Hachamovitch R, Berman DS. Comparison of long-term mortality risk following normal exercise vs adenosine myocardial perfusion SPECT. J Nucl Cardiol 2010;17:999e1008. 15. Fazel R, Gerber TC, Balter S, Brenner DJ, Carr JJ, Cerqueira MD, Chen J, Einstein AJ, Krumholz HM, Mahesh M, McCollough CH, Min JK, Morin RL, Nallamothu BK, Nasir K, Redberg RF, Shaw LJ; American Heart Association Council on Quality of Care and Outcomes Research, Council on Clinical Cardiology, and Council on Cardiovascular Radiology and Intervention. Approaches to enhancing radiation safety in cardiovascular imaging: a scientific statement from the American Heart Association. Circulation 2014;130:1730e1748.
0002-9149/17/$ - see front matter Ó 2017 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2017.03.026
balancing risk-to-benefit ratio related to radiation exposure8 and clinical outcomes.9 In this regard, whether an appropriately classified SPECT-MPI has long-term prognostic value, over longer term follow-up beyond what has been published, remains unknown. In this study, we test the hypothesis that an appropriate SPECT-MPI scan, continues to prognosticate adverse cardiac events over the long term. Methods We have previously reported the AUC grading of all consecutive studies referred to our MPI laboratory for single-photon emission tomography (SPECT) between February 2009 and July 2009 according to the 2009 American College of Cardiology/American Heart Association AUC.7 Briefly, the patient’s medical history and indications for testing were recorded on the day of their MPI appointment. Based on AUC, MPI studies were then immediately classified into appropriate, inappropriate, uncertain, or unclassified groups. Imaging was performed with a stress-rest protocol using either technetium-99m tetrofosmin or sestamibi. Stress testing was performed by exercise or dipyridamole with rest and poststress-gated SPECT imaging at the discretion of the nuclear cardiologist. The study was approved by the local institutional review board. Electrocardiogram-gated MPI was performed using a stress first, rest second, protocol using dual-head gamma cameras (Philips CardioMD and Philips Vertex). Singlephoton emission computed tomography (SPECT) MPI was performed in accordance with guidelines published by the ASNC. Images were acquired in step-and-shoot mode over a www.ajconline.org
1958
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline clinical and imaging characteristics Variable
Appropriate (n¼1129)
Age (years SD) Men Women Weight (kg SD) Chinese Malay Indian Others Hypertension Hyperlipidemia Diabetes mellitus Smoker Family History of CAD Prior MPI Test Indications for MPI Detection of symptomatic CAD Detection of CAD or risk assessment without ischemic equivalent Risk assessment with prior test results and/or known chronic stable CAD Preoperative risk assessment Risk assessment within three months of an acute coronary syndrome Risk assessment post PCI or CABG Myocardial perfusion Normal (SSS¼0-3) Mildly abnormal (SSS¼4-8) Moderately abnormal (SSS¼9-13) Severely abnormal (SSS>13) Myocardial ischemia None (SDS¼1) Mild (SDS¼2-4) Moderate (SDS¼5-7) Severe (SDS>7) Left Ventricular Ejection Fraction 60% < 60%
61.6 10.6 683 (60.5%) 446 (39.5%) 67.6 14.2 826 (73.2%) 94 (8.3%) 165 (14.6%) 44 (3.9%) 817 (72.4%) 879 (77.9%) 363 (32.2%) 179 (15.9%) 180 (15.9%) 114 (10.1%) 620 (54.9%) 64 (5.7%) 178 (15.8%) 250 (22.1%) 15 (1.3%) 2 (0.2%) 837 138 77 77
(74.1%) (12.2%) (6.8%) (6.8%)
890 140 61 38
(78.8%) (12.4%) (5.4%) (3.4%)
754 (66.8%) 375 (33.2%)
CABG ¼ coronary artery bypass graft surgery; CAD ¼ coronary artery disease; MPI ¼ myocardial perfusion imaging; PCI ¼ percutaneous coronary intervention; SSS ¼ summed stress score; SDS ¼ summed difference score.
180 semicircular orbit with 64 stops. No attenuation correction was applied. After filtered backprojection, shortaxis, horizontal, and vertical long-axis tomographic slices were created. The images were normalized to local maximal myocardial activity. Reconstructed SPECT-MPI slices were interpreted semiquantitatively using a 20-segment model. Each segment was scored by consensus by 3 nuclear cardiologists (F.K., T.R.S., and T.C.) using a 5-point scoring system for radiotracer uptake (0 ¼ normal, 1 ¼ mild reduction of tracer uptake, 2 ¼ moderate reduction of uptake, 3 ¼ severe reduction of uptake, and 4 ¼ absence of uptake). The summed stress score (SSS), summed rest score, and summed difference score (SDS) were calculated in the usual manner. The SSS is a measure of both reversible and fixed defects. The SDS is a measure of exercise-induced ischemia. Categorizations of summed scores as normal,
Figure 1. Flow chart of study cohort.
mild, moderate, and large are listed in Table 1. Left ventricular ejection fraction (LVEF) was derived from poststress SPECT images using commercially available software (AutoQuant; Cedars-Sinai Medical Center, Los Angeles, California). Vital status was evaluated through record linkage with the National Registry of Births and Deaths to determine overall mortality. It is a statutory requirement that death must be registered within 24 hours of its occurrence in the whole country. Cardiac deaths were ascertained based on International Classification of Diseases (ICD), Ninth Revision, codes 410 to 414, as well as ICD 10 I21.9, I24.9, I25.1, and I25.9. For the computation of the incidence of acute myocardial infarction (AMI), all the episodes included were identified centrally through the Singapore Myocardial Infarction Registry and diagnosed as. (1) Definitive AMIedefinite ECGs, or symptoms (typical or atypical), together with probable ECG and abnormal biochemical markers suggestive of myocardial necrosis, or typical symptoms and abnormal biochemical markers with ischemic/noncodable/unavailable ECG. (2) Clinical AMIeelectrocardiographic changes suggestive of AMI but not supported with raised cardiac biochemical markers or typical symptoms, or at least 2 of the following criteria: clinical history of prolonged chest pain >20 minutes; raised biochemical markers of myocardial necrosis; serial electrocardiographic tracings showing ST-T changed from baseline or Q waves duration that are 0.03 seconds in 2 or more contiguous leads. Furthermore, ICD-9 Clinical Modification code of 410 was used to identify AMI cases in the data sources before 2012, whereas ICD-10 Australian Modification codes I21 and I22 were used for the cases diagnosed from 2012 (inclusive) onward. Patients who were nonresidents (n ¼ 80) and had repeated MPI (n ¼ 8) were excluded from the current analysis. Patients with complete information in outcomes and covariates were used in the analysis. Patients with AUC grading other than appropriate were excluded from this analysis. For this analysis, there were 1,129
Coronary Artery Disease/Long-term Prognostic Value of Appropriate MPI
1959
Table 2 Cardiac procedures based on presence of ischemia (SDS 2) Procedure Coronary angiography* Coronary revascularization†
Total procedures (n¼1129)
Ischemia (n¼239)
No ischemia (n¼890)
OR (95% CI)z
P-valuez
152 (13.5%) 113 (10.0%)
99 (41.4%) 78 (32.6%)
53 (6.0%) 35 (3.9%)
5.33 (2.98, 9.53) 11.01 (6.90, 17.56)
<0.001 <0.001
CI ¼ confidence interval; OR ¼ odds ratio; SDS ¼ summed difference score. * Coronary angiography procedures occurring within 1 year after the index cardiac catheterization. † Coronary revascularization (PCI/CABG) procedures occurring within 1 year after the index cardiac catheterization. z OR (95% CI) and p values are derived from logistics regression models between groups of ischemia and no ischemia (ref.), adjusted for clinical covariates (including age/gender/weight/hypertension/hyperlipidemia/diabetes mellitus/smoking/family history of coronary artery disease [CAD]/EF).
patients recorded at the end of the follow-up period (Figure 1). Outcome assessors were blinded to MPI findings and AUC classification. Coronary angiography and revascularization (percutaneous coronary intervention [PCI] and coronary artery bypass graft [CABG]) procedures performed within 1 year after the MPI were considered to be triggered by the MPI findings. The primary end point was composite end point of cardiac death or nonfatal MI. Secondary end point was all-cause mortality. Continuous variables were summarized as mean (SD) and 2-sample t-test was used to compare between 2 normally distributed variables. Categorical variables were summarized as count (percentage), and the chi-square test was used to compare dichotomous variables. A multivariable logistic regression model was used to estimate the odds ratio of ischemia to nonischemia groups with coronary angiography procedures and coronary revascularization (PCI/CABG) procedures. The model was adjusted for age, gender, weight, hypertension, hyperlipidemia, diabetes mellitus, smoking status, family history of coronary artery disease, and EF. Both univariate and multivariable Cox proportional hazard regression models were used to estimate hazard ratios of abnormal MPI to normal MPI with (1) allcause mortality, (2) myocardial infarction (MI), (3) all-cause mortality or MI, (4) cardiac mortality, and (5)cardiac mortality or MI as outcomes. The multivariable model was adjusted for the same set of clinical covariates above with including coronary revascularization. Two-sided p <0.05 was considered as statistically significant. SAS version 9.4 (SAS Institute, Cary, North Carolina) was used to perform the analysis. Results The study cohort consisted of 1,129 consecutive patients referred for clinically indicated SPECT-MPI study (Figure 1), followed up for mean 5.4 1.2 years. Baseline patient characteristics are summarized in Table 1. There were 60.5% men with a mean age of 61.6 10.6 years. Over 72.4% of the study cohort had hypertension, 77.9% had hyperlipidemia, whereas 32.2% had diabetes mellitus. Most of the MPI scans were normal (SSS 3; 74.1%), had lower proportions of myocardial ischemia (SDS 1 [78.8%], and with LVEF of 60% [66.8%]). Over the course of follow-up, there were 152 (13.5%) coronary angiography (cath) procedures with subsequent coronary revascularization in 113 (10.0%) (either PCI or CABG surgery) that occurred within 1 year of the MPI scan (Table 2). Patients with stress-induced ischemia (defined as
SDS 2) had significantly higher rates of cath (41.4% vs 6.0%, p <0.001) and revascularization rates (32.6% vs 3.9%, p <0.001) than those without ischemia. Patients with ischemia had significantly higher odds of cath and revascularization than those without ischemia (all p <0.001). There were 148 all-cause mortality deaths (13.1%), of which 58 were cardiac deaths (5.1%) and 109 (9.7%) nonfatal MIs. Patients with abnormal MPI had higher rates of all-cause mortality, cardiac death, and nonfatal MI than those with normal MPI (Table 3). In the presence of a normal MPI scan, we observed annualized events rate of 0.4% for cardiac death and 1.1% for MI. Risk-adjusted survival curves demonstrate a striking difference with respect to primary end point of cardiac death or nonfatal MI particularly between normal, moderately, and severely abnormal MPI scans (Figure 2). After adjustments for clinical covariates, event rates for primary end point of cardiac death or nonfatal MI, but not the secondary end point of all-cause mortality, were higher in patients with abnormal MPI than those patients with normal MPI (hazard ratio 1.81, 95% CI 1.23 to 2.66, p ¼ 0.003; Figure 3). Discussion Following on this quality assessment project that started in the year 2009 where we had previously used the 2009 AUC for nuclear cardiology by the American College of Cardiology Foundation, American Heart Association, and ASNC, to assess the appropriate use of myocardial perfusion scans in our center,7 including examination of shortterm clinical outcomes data among a subset of MPI studies,10 we provide new evidence in this analysis that demonstrates long-term prognostic value of AUC. In this prospective cohort of patients followed for at least 5 years, we observed sustained long-term value of an appropriate grading in prognosticating hard clinical outcomes of patients referred for clinically indicated MPI. We demonstrate that an appropriate grading continued to predict outcomes, taking into account clinical risk factors, and traditional MPI variables. This demonstrates the robust value of an appropriate grading within the AUC, observed for the first time in a tertiary level setting, surpassing previous data observed in more community-based settings and which were limited to shorter duration of follow-up.4 A major strength of our study is the prospective design of the study sample. We had deliberately designed this study as a quality and outcomes project right from the start in 2009. The collection of all study variables were prospective, in particular assessment of appropriate grading was performed
1960
The American Journal of Cardiology (www.ajconline.org)
Table 3 Clinical events based on degree of myocardial perfusion imaging abnormality (SSS 4) Event
Total Events
All-cause mortality Myocardial infarction (MI) All-cause mortality or MI Cardiac Death Cardiac Death or MI
148 109 218 58 144
(13.1%) (9.7%) (19.3%) (5.1%) (12.8%)
Abnormal MPI (n¼292) 71 57 103 39 79
Normal MPI (n¼837)
(24.3%) (19.5%) (35.3%) (13.4%) (27.1%)
77 52 115 19 65
(9.2%) (6.2%) (13.7%) (2.3%) (7.8%)
HR (95% CI)* 2.89 3.42 2.96 6.28 3.94
(2.10, (2.35, (2.26, (3.63, (2.84,
4.00) 4.98) 3.86) 10.88) 5.47)
p-value <.0001 <.0001 <.0001 <.0001 <.0001
Adjusted HR (95% CI) 1.27 1.72 1.39 2.12 1.81
(0.86, (1.10, (1.02, (1.15, (1.23,
1.87) 2.69) 1.91) 3.91) 2.66)
p-value 0.223 0.017 0.040 0.016 0.003
CI ¼ confidence interval; HR ¼ hazard ratio; MPI ¼ myocardial perfusion imaging; SSS ¼ summed stress score. * HR (95% CI) and p values were derived from Cox proportional hazards models between group of abnormal MPI and normal MPI, adjusted for clinical covariates (including age/gender/weight/hypertension/hyperlipidemia/diabetes mellitus/smoking/family history of coronary artery disease [CAD]/EF/ revascularization).
Figure 2. Outcomes based on myocardial perfusion imaging findings. Time-to-outcome Cox proportional hazards curves based on severity of perfusion abnormality (SSS). The hazard ratios (HRs) were calculated relative to the normal group and adjusted for clinical covariates.
by independent observers at the point of performing the MPI scan. This crucial step minimized potential misclassification biases otherwise obtained from retrospective chart review or computer logic. Owing to a robust national diseases registry where death notification was a mandatory requirement across the country by law, and a national MI registry where MI were centrally monitored, the study team accrued prospective clinical outcome data over the years, minimizing losses to follow-up.
Despite being a tertiary center, the accrual of a high event rate over time was a function of the long period of follow-up and less because the study sample was a high-risk population, commonly seen in tertiary settings. Although there was a rather high burden of coronary risk factors in the study sample, most of the MPI studies (74.1%) were normal, with the majority having no ischemia (78.8%) or normal EFs (66.8%). Such a sample would intuitively provide greater test to the AUC, by showing that these
Coronary Artery Disease/Long-term Prognostic Value of Appropriate MPI
1961
Figure 3. Outcomes based on myocardial perfusion abnormality. Time-to-outcome Cox proportional hazards curves of patients with abnormal versus normal MPI. Hazard ratios (HRs) were calculated and adjusted for clinical covariates.
appropriate MPI scans prognosticated clinical events despite being generally low-risk MPI scans. Longer term outcomes studies such as ours draw conclusions based on the actual sustaining value of the appropriate grade, powered by events over time, rather than by high rates of events driven in the shorter term by highly abnormal MPI scans. In addition, although we had defined a composite primary end point of cardiac death or nonfatal MI, each end point of cardiac death or nonfatal MI was by itself significantly powered to draw conclusions about their associations with appropriate grading. The scans with myocardial ischemia had understandably greater use of subsequent coronary angiography and revascularization, consistent with other studies that have looked at rates of downstream utilization in relation to AUC.3 Even in the presence of a normal MPI commonly regarded as low risk, we observed a considerable annualized cardiac death or nonfatal MI rates of 0.4% and 1.1%, respectively, in this cohort, where all MPI scans were classified as appropriate. For years, normal SPECTMPI has been used to provide a “warranty” for low cardiac risk, associated with low rates of events such as <0.5% risk of cardiac death or <0.3% of risk of MI per year.11 Since then, various patient cohorts have been identified
as having increased risk despite normal SPECT MPI scans, such as patients with diabetes mellitus,12 atrial fibrillation13 and patients who undergo pharmacological stress.14 Our results suggest that patients with normal MPI classified as appropriate, may benefit from closer clinical scrutiny. This is especially novel because hardly any prospective cohort have reported long-term event rates among normal MPI defined as SSS 3, particularly defined by an appropriate classification. Our novel data may reflect the intrinsic quality embedded within AUC, not encompassed within traditional SPECT MPI scan variables, and which therefore provides a fresh angle to support the use of AUC as a risk stratification tool in the assessment of quality imaging.9 Our observation further indicates the need for more data to report the prognostic value of an appropriate grading with normal MPI, among other patient cohorts. Finally, the prognostic value of SPECT is heavily dependent on the patient population being studied. We acknowledge that this study, although representative of a tertiary center within our country, may only apply to settings similar to ours and may not apply to other tertiary centers in other parts of the world. However, our study represents steps along the road to quality improvement, provides novel data
1962
The American Journal of Cardiology (www.ajconline.org)
toward quality imaging, which is the ultimate goal of the AUC.15 We acknowledge limitations in the study. First, our study is observational and therefore does not allow for causal inferences. Second, we attributed coronary angiography and revascularization procedures within 1 year from the index MPI scan as being triggered by the index MPI scan result. This duration may be viewed as being prolonged compared with other studies that use 180 days from MPI result, possibly diminishing the effect that cath was truly triggered by the MPI result. Sensitivity analyses adjusting for duration between MPI scan and procedures might provide better clarification, which was not done. However, our results emphasizing the long-term hard clinical outcomes are still the main message of the study. In addition, we did not include assessment of other adverse MPI variables such as transient ischemic dilatation; however, given the low rates of abnormal MPI scans in this cohort, such adverse features might not have made a difference to the conclusions. Finally, it is still possible that adjustments may not have accounted for unknown confounders, leading to the possibility of residual confounding. Acknowledgment: The authors are indebted to the patients and the healthcare staff of the nuclear laboratory for making this study possible over the years. Disclosures The authors have no conflicts of interest to disclose. 1. Hendel RC, Berman DS, Di Carli MF, Heidenreich PA, Henkin RE, Pellikka PA, Pohost GM, Williams KA. ACCF/ASNC/ACR/AHA/ ASE/SCCT/SCMR/SNM 2009 appropriate use criteria for cardiac radionuclide imaging: a report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, the American Society of Nuclear Cardiology, the American College of Radiology, the American Heart Association, the American Society of Echocardiography, the Society of Cardiovascular Computed Tomography, the Society for Cardiovascular Magnetic Resonance, and the Society of Nuclear Medicine. Circulation 2009;119:e561ee587. 2. Doukky R, Hayes K, Frogge N. Appropriate use criteria for SPECT myocardial perfusion imaging: are they appropriate for women? J Nucl Cardiol 2016;23:695e705. 3. Khawaja FJ, Jouni H, Miller TD, Hodge DO, Gibbons RJ. Downstream clinical implications of abnormal myocardial perfusion single-photon emission computed tomography based on appropriate use criteria. J Nucl Cardiol 2013;20:1041e1048.
4. Doukky R, Hayes K, Frogge N, Balakrishnan G, Dontaraju VS, Rangel MO, Golzar Y, Garcia-Sayan E, Hendel RC. Impact of appropriate use on the prognostic value of single-photon emission computed tomography myocardial perfusion imaging. Circulation 2013;128:1634e1643. 5. Gibbons RJ, Miller TD. Single-photon emission computed tomography appropriateness: does it matter for patient outcomes? Circulation 2013;128:1595e1597. 6. Soine LA, Cunningham SL, Motzer SA, Inoue LY, Caldwell JH. Application of appropriate use criteria for stress myocardial perfusion imaging at two academic medical centers: compliance and association with image findings. J Am Acad Nurse Pract 2012;24:200e208. 7. Koh AS, Flores JL, Keng FY, Tan RS, Chua TS. Evaluation of the American College of Cardiology Foundation/American Society of Nuclear Cardiology appropriateness criteria for SPECT myocardial perfusion imaging in an Asian tertiary cardiac center. J Nucl Cardiol 2011;18:324e330. 8. Doukky R, Frogge N, Appis A, Hayes K, Khoudary G, Fogg L, Williams KA Sr. Impact of appropriate use on the estimated radiation risk to men and women undergoing radionuclide myocardial perfusion imaging. J Nucl Med 2016;57:1251e1257. 9. Bhattacharyya S, Lloyd G. Improving appropriateness and quality in cardiovascular imaging: a review of the evidence. Circ Cardiovasc Imaging 2015;8:e003988. 10. Koh AS, Flores JL, Keng FY, Tan RS, Chua TS. Correlation between clinical outcomes and appropriateness grading for referral to myocardial perfusion imaging for preoperative evaluation prior to non-cardiac surgery. J Nucl Cardiol 2012;19:277e284. 11. Hachamovitch R, Berman DS, Shaw LJ, Kiat H, Cohen I, Cabico JA, Friedman J, Diamond GA. Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death: differential stratification for risk of cardiac death and myocardial infarction. Circulation 1998;97:535e543. 12. Berman DS, Kang X, Hayes SW, Friedman JD, Cohen I, Abidov A, Shaw LJ, Amanullah AM, Germano G, Hachamovitch R. Adenosine myocardial perfusion single-photon emission computed tomography in women compared with men. Impact of diabetes mellitus on incremental prognostic value and effect on patient management. J Am Coll Cardiol 2003;41:1125e1133. 13. Abidov A, Hachamovitch R, Rozanski A, Hayes SW, Santos MM, Sciammarella MG, Cohen I, Gerlach J, Friedman JD, Germano G, Berman DS. Prognostic implications of atrial fibrillation in patients undergoing myocardial perfusion single-photon emission computed tomography. J Am Coll Cardiol 2004;44:1062e1070. 14. Rozanski A, Gransar H, Hayes SW, Friedman JD, Hachamovitch R, Berman DS. Comparison of long-term mortality risk following normal exercise vs adenosine myocardial perfusion SPECT. J Nucl Cardiol 2010;17:999e1008. 15. Fazel R, Gerber TC, Balter S, Brenner DJ, Carr JJ, Cerqueira MD, Chen J, Einstein AJ, Krumholz HM, Mahesh M, McCollough CH, Min JK, Morin RL, Nallamothu BK, Nasir K, Redberg RF, Shaw LJ; American Heart Association Council on Quality of Care and Outcomes Research, Council on Clinical Cardiology, and Council on Cardiovascular Radiology and Intervention. Approaches to enhancing radiation safety in cardiovascular imaging: a scientific statement from the American Heart Association. Circulation 2014;130:1730e1748.