- Email: [email protected]
Reclassification of Cardiovascular Risk in Patients With Normal Myocardial Perfusion Imaging Using Heart Rate Response to Vasodilator Stress
Reclassification of Cardiovascular Risk in Patients With Normal Myocardial Perfusion Imaging Using Heart Rate Response to Vasodilator Stress Fahad M. Iqbal, MDa,*, Wael Al Jaroudi, MDb, Kumar Sanam, MDc, Aaron Sweeney, MDa, Jaekyeong Heo, MDc, Ami E. Iskandrian, MDc, and Fadi G. Hage, MDc,d Previous studies have shown that patients with normal vasodilator myocardial perfusion imaging (MPI) findings remain at a greater risk of future cardiac events than patients with normal exercise MPI findings. The aim was to assess improvement in risk classification provided by the heart rate response (HRR) in patients with normal vasodilator MPI findings when added to traditional risk stratification. We retrospectively studied 2,000 patients with normal regadenoson or adenosine MPI findings. Risk stratification was performed using Adult Treatment Panel III framework. Patients were stratified by HRR (percentage of increase from baseline) into tertiles specific to each vasodilator. All-cause mortality and cardiac death/nonfatal myocardial infarction (MI) £2 years from the index MPI were recorded. During follow-up, 11.8% patients died and 2.7% patients experienced cardiac death/nonfatal MI in the adenosine and regadenoson groups, respectively. The patients who died had a greater Framingham risk score (12 – 4 vs 11 – 4, p [ 0.009) and lower HRR (22 – 16 vs 32 – 21, p <0.0001). In an adjusted Cox model, the lowest tertile HRR was associated with an increased risk of mortality (hazard ratio 2.1) and cardiac death/nonfatal MI (hazard ratio 2.9; p <0.01). Patients in the highest HRR tertile, irrespective of the Adult Treatment Panel III category, were at low risk. When added to the Adult Treatment Panel III categories, the HRR resulted in net reclassification improvement in mortality of 18% and cardiac death/nonfatal MI of 22%. In conclusion, a blunted HRR to vasodilator stress was independently associated with an increased risk of cardiac events and overall mortality in patients with normal vasodilator MPI findings. The HRR correctly reclassified a substantial proportion of these patients in addition to the traditional risk classification models and identified patients with normal vasodilator MPI findings, who had a truly low risk of events. Ó 2013 Published by Elsevier Inc. (Am J Cardiol 2013;111:190e195) Traditional risk factors such as age, gender, blood pressure, smoking history, cholesterol, and diabetes mellitus have been shown to predict coronary heart disease risk and death in a number of studies.1e5 Global risk algorithms, such as the Framingham risk score, or its modification in the third report of the National Cholesterol Education Program’s Expert Panel on the Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel [ATP] III), encapsulate these risk factors to estimate the 10-year coronary heart disease event risk.3,6 Myocardial perfusion imaging (MPI) using adenosine or regadenoson is an established method for detecting coronary heart disease and for risk stratification of patients with exercise limitations.7e9 Compared to patients undergoing exercise testing, those referred for vasodilator-based MPI have a greater pretest a Tulane University Heart and Vascular Institute, Tulane University, New Orleans, Louisiana; bDivision of Cardiovascular Medicine, Section of Cardiac Imaging, Heart and Vascular Institute, Cleveland Clinic, Cleveland, Ohio; cDivision of Cardiovascular Disease, University of Alabama at Birmingham, and dDivision of Cardiology, Birmingham Veterans Affairs Medical Center, Birmingham, Alabama. Manuscript received August 14, 2012; revised manuscript received and accepted September 18, 2012. See page 195 for disclosure information. *Corresponding author: Tel: (504) 988-6139; fax: (504) 988-4237. E-mail address: fi[email protected] (F.M. Iqbal).
0002-9149/12/$ - see front matter Ó 2013 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.amjcard.2012.09.013
probability of coronary heart disease and, therefore, have a slightly greater risk of cardiac death and myocardial infarction (MI) despite normal MPI findings (1% to 2%/year for vasodilator stress vs <1% for exercise).10 Furthermore, the prognostic value of functional status provided by the exercise portion of the test is not available.11 The best method for risk stratification in this population is not well established. We have recently shown that the heart rate response (HRR) to adenosine and regadenoson is an independent predictor of outcome, provides incremental prognostic information to clinical and imaging variables, and helps in better risk stratification.7,12,13 The present study examined whether the addition of HRR to adenosine or regadenoson to traditional risk stratification in patients with normal MPI findings improves risk stratification. Methods The study population included 2,000 patients who underwent vasodilator MPI at the University of Alabama at Birmingham: 1,000 consecutive patients with normal adenosine MPI findings and another 1,000 consecutive patients with normal regadenoson MPI findings (after our center switched completely from using adenosine to using regadenoson).14 For the purposes of the present retrospective study, normal MPI findings were defined as a normal perfusion www.ajconline.org
Coronary Artery Disease/Vasodilator HRR Reclassification Table 1 Baseline characteristics of study population according to all-cause mortality Characteristic Age (yrs) Men Race White Black Other Hypertension Hyperlipidemia Diabetes mellitus Chronic kidney disease End-stage renal disease Congestive heart failure Peripheral vascular/cerebrovascular disease Previous myocardial infarction Previous coronary intervention Previous bypass surgery Framingham risk score Former smoker Current smoker In-patient stress test
Alive (n ¼ 1,764) Dead (n ¼ 236) 58.99 12.24* 778 (44.1%)† 1,108 626 30 1,397 883 654 459 452 115 300
(62.8%) (35.5%) (1.7%) (79.2%)* (50.1%)* (37.1%) (26.0%) (25.6%) (6.5%) (17.0%)
116 (6.6%) 214 (12.1%)† 115 (6.5%) 11.26 4.20† 454 (25.7%)† 316 (17.9%) 378 (21.4%)
62.71 11.52* 122 (51.7%)† 162 70 4 162 76 82 74 60 19 42
(68.6%) (29.7%) (1.7%) (68.6%)* (32.2%)* (34.7%) (31.4%) (25.4%) (8.1%) (17.8%)
17 (7.2%) 17 (7.2%)† 16 (6.8%) 11.97 3.70† 84 (35.6%)† 46 (19.5%) 51 (21.6%)
* p <0.001. † p <0.05.
Table 2 Medication use by event status for all-cause mortality Medication Aspirin Plavix b Blockers Angiotensin-converting enzyme inhibitors/receptor blockers Calcium channel blockers Other antihypertensives Thiazide diuretics Loop diuretics Potassium-sparing diuretics Statins Insulin Metformin Sulfonylureas Glitazones
Alive (n ¼ 1,764)
Dead (n ¼ 236)
704 204 839 850
(39.9%)* (11.6%)† (47.6%)† (48.2%)*
62 15 88 82
(26.3%)* (6.4%)† (37.3%)† (34.7%)*
518 288 406 392 146 732 295 190 187 84
(29.4%) (16.3%) (23.0%)† (22.2%) (8.3%)† (41.5%)* (16.7%) (10.8%)† (10.6%) (4.8%)†
58 36 34 64 35 56 41 13 18 4
(24.6%) (15.3%) (14.4%)† (27.1%) (14.8%)† (23.7%)* (17.4%) (5.5%)† (7.6%) (1.7%)†
191
Table 3 Hemodynamic parameters by event status for all-cause mortality Parameter
Alive (n ¼ 1,764)
Dead (n ¼ 236)
Heart rate at rest (beats/min) Systolic blood pressure at rest (mm Hg) Diastolic blood pressure at rest (mm Hg) Heart rate response Left ventricular ejection fraction (%)
71.37 13.00* 132.38 21.54*
75.79 15.15* 128.27 21.70*
75.07 11.26*
71.82 11.11*
31.91 20.79* 66.88 9.45
21.62 15.95* 67.25 9.80
* p <0.001.
Table 4 Laboratory results by event status for all-cause mortality Alive (n ¼ 1,207) Total cholesterol (mg/dl) High-density lipoprotein (mg/dl) Low-density lipoprotein (mg/dl) Triglycerides (mg/dl) Serum creatinine (mg/dl)z Estimated glomerular filtration rate (ml/min/1.73 m2)
177.47 46.01 103.62 157.25 1.18 59.22
49.99* 18.94 39.31† 116.32 0.72 36.38
Dead (n ¼ 111) 166.74 47.62 95.74 159.01 1.19 56.03
44.06* 34.05 38.67† 107.31 0.54 36.29
* p <0.001 for patients who had died versus those alive at 2 yrs after MPI within same category. † p <0.05 for patients who had died versus those alive at 2 yrs after MPI within same category. z Serum creatinine and estimated glomerular filtration rate values were averaged for patients without end-stage renal disease and not requiring dialysis.
* p <0.001. † p <0.05.
pattern and normal left ventricular ejection fraction (LVEF) 50%. A total of 3,000 patients undergoing MPI were screened, patients with abnormal MPI findings (n ¼ 617) or LVEF <50% (n ¼ 383) were excluded. Using this database, we have previously shown that patients with normal regadenoson MPI findings experienced outcomes similar to those with normal adenosine MPI findings.14 Patients selected for the present study underwent either adenosine or regadenoson MPI. Both agents were delivered using peripheral intravenous access. Adenosine was administered as a continuous infusion (140 mg/kg/min for 6 minutes), and regadenoson was administered as a single
Figure 1. Scatter plot demonstrating relation between Framingham risk score and HRR. Scatter plot has superimposed linear regression line and corresponding equation.
bolus of 0.4 mg, followed by a saline flush. Technetium99m-sestamibi determined by the patient’s weight was injected at 3 minutes into the infusion of adenosine and 10 to 20 seconds after the saline flush following regadenoson administration, as previously described.14 All the studies were done in the absence of accompanying exercise. The heart rate and blood pressure were measured at baseline and every 2 minutes until the completion of the study.
192
The American Journal of Cardiology (www.ajconline.org)
Table 5 Hazard ratios for heart rate response (HRR) with and without adjustment All-Cause Mortality HR (95% CI) Unadjusted Adjusted for baseline heart rate Adjusted for age and gender Adjusted for age, gender, baseline heart rate, and ATP III risk category Adjusted for age, gender, baseline heart rate, ATP III risk category, and LVEF
2.54 2.25 2.47 2.09 2.10
(1.97e3.29) (1.71e2.96) (1.91e3.19) (1.59e2.76) (1.59e2.77)
p Value <0.001 <0.001 <0.001 <0.001 <0.001
Cardiac death/MI HR (95% CI) 4.01 2.95 4.03 2.92 2.93
(2.14e7.51) (1.52e5.75) (2.15e7.55) (1.49e5.71) (1.50e5.72)
p Value <0.001 0.001 <0.001 0.002 0.002
CI ¼ confidence interval; HR ¼ hazard ratio.
Figure 2. Annualized event rates of all-cause mortality and cardiac death/ nonfatal MI categorized by HRR tertile.
Medications (including b blockers) and caffeinated beverages were withheld on the morning of the stress portion of the study. Gated single photon emission computed tomography images were acquired 1 hour after tracer injection using a dual-head detector gamma camera with a 20% energy window centered on the 140-keV gamma peak. Cameras were operated in an elliptical 180 acquisition orbit with 32 projections and 30 to 40 seconds per projection. Gating was done with 8 to 16 frames per RR cycle. Images were processed using filtered back projection. All MPI studies were interpreted visually, aided by automated quantitative analysis, by readers who were unaware of subsequent events and without attenuation or scatter correction. At rest images were obtained whenever uncertainty was present in the interpretation of the stress images, as previously described.15 The LVEF and end-diastolic and end-systolic volumes were measured from the stress gated images using the method previously described by Germano et al.16 The variables abstracted from the patients’ medical records included patient demographics (age, gender, race); co-morbidities, including hypertension, hyperlipidemia, diabetes mellitus, stroke, previous MI, previous cardiovascular interventions, such as percutaneous coronary intervention or coronary artery bypass grafting; history of tobacco use; medication usage at MPI, and laboratory results (serum creatinine and lipid panel). The estimated glomerular filtration rate was calculated according to the Modification of Diet in Renal Disease study formula.17 Patients were considered to have chronic
kidney disease if their estimated glomerular filtration rate was 15 to 60 ml/min/1.73 m2. Patients were considered to have end-stage renal disease if their estimated glomerular filtration rate was <15 mL/min/1.73 m2 or if they were taking renal replacement therapy. The HRR was calculated as the maximum percentage of change from baseline, as previously described.12,13,18 For each of the stress agents, patients were divided into 3 groups according to the HRR tertiles. The HRR tertiles for adenosine were defined as <16%, 16% to 32%, and >32%. The corresponding tertiles for regadenoson were <21%, 21% to 37%, and >37%. The primary outcome of interest was defined as all-cause mortality and was determined using the Social Security Death Index (assessed on July 31, 2011). The secondary outcome was a composite of cardiac death and nonfatal MI (as documented by the appropriate combination of symptoms, electrocardiographic findings, and enzyme changes).19 Cardiac death was defined as death from fatal arrhythmia, MI, or heart failure and was determined by reviewing the electronic medical records and death certificates. All events were censored at 2 years after the index MPI study. The interval to an event was defined as the duration from the baseline MPI study to death, cardiac death/nonfatal MI, or the end of 2 years of follow-up. All statistical analyses were performed using the Statistical Package for Social Sciences, version 18, for Windows (SPSS, Chicago, Illinois). Student’s t test and the MannWhitney U test were used to compare continuous variables and the chi-square test or Fischer’s exact test to compare categorical variables, as appropriate. The Framingham risk score for the 10-year risk of hard coronary heart disease events were calculated using age, gender, total cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, active treatment of hypertension, and smoking status.6 Patients were then stratified into risk groups according to the calculated Framingham risk score (low, <10%; intermediate, 10% to 20%; and high, >20%). Patients with preexisting diabetes mellitus or vascular disease at baseline were assigned to the high-risk group, irrespective of calculated Framingham risk score according to ATP III.6 Patients were additionally substratified within ATP III categories according to the HRR tertile. Cox proportional hazards models were then used to calculate the hazard ratios and corresponding 95% confidence intervals for the primary and secondary end points in relation to HRR tertiles. The
Coronary Artery Disease/Vasodilator HRR Reclassification
193
Figure 3. Reclassification of patients experiencing all-cause mortality using baseline ATP III risk category modified by HRR. High-risk (red), intermediate-risk (purple), and low-risk (blue) patients presented separately.
Figure 4. Reclassification of patients experiencing cardiac death/nonfatal MI (CDMI) using baseline ATP III risk category modified by HRR. High-risk (red), intermediate-risk (purple), and low-risk (blue) patients presented separately.
variables entered in the model were age, gender, ATP III category, and LVEF. Variables with co-linearity were entered into the model 1 at a time. Stepwise forward selection was used to create the final model. The incremental effect of adding HRR to the ATP III categorization for predicting outcomes was evaluated using the net reclassification improvement.20 After risk categorization according to the ATP III guidelines, patients in the lowest HRR tertile (i.e., blunted HRR) were reclassified up 1 risk category, to a maximum of high risk. Patients in the middle HRR tertile were not reclassified. Finally, patients in the highest HRR tertile were reclassified down 1 risk category, to a minimum of low risk. The goal was to determine whether reclassification would assign patients who developed events to a higher risk category and those who did not to a lower risk category. The statistical significance of the net reclassification improvement was assessed, as described by Pencina et al.20 The institutional review board for human research at the University of Alabama at Birmingham approved the present study.
Results The baseline characteristics (Table 1), medication use (Table 2), hemodynamic parameters (Table 3), and laboratory results (Table 4) of the patients are presented grouped by stress agent. The cohort consisted of 2,000 patients with a mean age of 59 12 years and an LVEF of 67 10%; 45% were men, 64% were white, 37% had diabetes mellitus, 24% had a history of cardiovascular disease, 26% end-stage renal disease, and 18% were active smokers. After 2 years of follow-up, 236 patients (11.8%) died. Patients who died were older, more likely to be men, former smokers, have a lower low-density lipoprotein level and a greater Framingham risk score, and were less likely to have hypertension, hyperlipidemia, or a history of percutaneous coronary intervention (Table 1). In addition, those who died were less likely to be receiving cardiovascular medications known to improve survival (aspirin, clopidogrel, b blockers, angiostensin-converting enzyme inhibitors/ angiostensin receptor blockers, and statins; Table 2) and had a greater heart rate at rest, lower systolic and diastolic blood
194
The American Journal of Cardiology (www.ajconline.org)
pressure at rest, and lower HRRs (Table 3). The secondary end point (cardiac death/nonfatal MI) occurred in 54 patients (2.7%). Scatter plot analyses showed that HRR was significantly associated with Framingham risk score but with very poor correlation (Figure 1). On Cox proportional hazard analysis, and after controlling for age, gender, LVEF, and ATP III category, a HRR in the lowest tertile was associated with an increased risk of the primary (overall mortality) and secondary (cardiac death/nonfatal MI) outcomes (hazard ratio 2.10, 95% confidence interval 1.59 to 2.77; and hazard ratio 2.93, 95% confidence interval 1.50 to 5.72, respectively; Table 5). Patients with normal perfusion on MPI and classified to be at low risk by ATP III experienced an annual mortality and cardiac death/nonfatal MI rate of 5.5% and 1.0%, respectively (n ¼ 635). Patients classified to be in the highest HRR tertile, irrespective of ATP III categorization, had an annual mortality and cardiac death/nonfatal MI rate of 3.1% and 0.5%, respectively (n ¼ 691; Figure 2). Alternatively, patients with a HRR in the highest tertile and classified in the low- or intermediate-risk categories by ATP III had an annual mortality and cardiac death/nonfatal MI rate of 2.9% and 0.1%, respectively (n ¼ 359). Adding HRR to the classic risk stratification with ATP III resulted in an net reclassification improvement for all-cause mortality of 18.2% (8.9% for deaths and 9.3% for survivors, p <0.0001; Figure 3) and cardiac death/nonfatal MI of 21.9% (7.6% for events and 14.3% for nonevents, p ¼ 0.022; Figure 4). Discussion The results of the present study have demonstrated that risk stratification with HRR is useful in patients with normal vasodilator MPI findings when added to a classic risk stratification model, such as the ATP III risk categories, or by itself. Patients with normal vasodilator MPI findings have a residual risk attributable to elevated pretest risk. In this population, even patients classified as low risk using the ATP III algorithm were at a relatively high risk of events during the subsequent 2 years (annual mortality rate of 5.5% and annual cardiac death/nonfatal MI rate of 1.0%). Patients with an HRR in the highest tertile had event rates that were lower, irrespective of their ATP III classification (annual mortality rate 3.1% and annual cardiac death/ nonfatal MI rate of 0.5%), not unlike those reported for normal exercise MPI findings.21 Furthermore, patients who were classified as at low or intermediate risk by ATP III and had a HRR in the highest tertile (18% of the population) were at very low risk of events in the 2 years after the test (annual mortality rate 2.9% and cardiac death/nonfatal MI rate 0.1%). A large body of evidence supports the prognostic use of MPI.8 Both perfusion abnormalities and a depressed LVEF are proven strong predictors of a poor outcome.22 Clinically, the value of normal MPI findings might be even more important than abnormal test findings, because it signals a benign outcome, despite the presence of factors associated with greater risk.14 Navare et al10 combined data from 24 studies (almost 15,000 patients) to show that the
cardiac event rate was significantly greater with normal and abnormal test results from pharmacologic stress MPI versus exercise stress MPI. Patients with normal pharmacologic MPI had an annualized event rate for cardiac death and myocardial infarction of 1.78% compared to 0.65% for normal exercise MPI (p <0.001). Rozanski et al21 followed up >6,000 patients who underwent exercise or adenosine MPI and had normal test results for a mean of 10 years. Annualized mortality was significantly greater in the adenosine cohort with (3.9% vs 1.6%, p <0.0001) and without (4.3% vs 1.1%, p <0.0001) propensity matching. The increased risk has been attributed to an increased pretest risk in the population referred for pharmacologic stress, rather than an inaccuracy of the test itself. This pretest risk is a reflection of the increased prevalence of poor prognostic indicators in patients referred to pharmacologic stress, as well as poor functional capacity that precludes exercise. Patients with normal vasodilator MPI findings, therefore, have a residual risk that is two- to fourfold greater than those with normal exercise MPI findings. Risk classification in patients with normal MPI findings (with both exercise and pharmacologic stress tests) has not been well studied. Previous reports12,23,24 found that HRR can risk stratify patients with normal vasodilator MPI findings. Our data have expanded on these earlier reports by showing that HRR provides incremental information to traditional risk stratification models. Although the traditional Framingham risk factors have been used in multiple diverse studies for risk stratification, it is important to note that patients with normal vasodilator MPI findings are unique in several respects. First, they have an indication for stress testing. Second, they have normal perfusion and LV function. Third, they have limited functional capacity. Our results have indicated that the patients in our cohorts with low ATP III risk continued to be at relatively high risk. In the study by Rozanski et al,21 propensity matching for these and other factors did not account for the increased risk in patients with vasodilator MPI findings. The change in heart rate in response to adenosine receptor agonists is mediated by the autonomic nervous system.25 Because cardiac autonomic dysfunction is independently associated with increased cardiovascular risk,26 these hemodynamic changes are expected to carry prognostic significance. A blunted HRR to adenosine,13 dipyridamole,24 or regadenoson12 has been associated with worse outcomes. This easily derived indicator provides prognostic value that is independent of traditional MPI data and other known prognostic indicators. In a recent study that followed up 1,156 patients who underwent regadenoson MPI for a mean of almost 2 years, patients with a normal HRR had a relatively low annualized total mortality despite the presence of risk factors.12 The present study demonstrated that the HRR can contribute to better risk stratification in patients with normal vasodilator MPI findings. The overall effect on discrimination with the addition of HRR to traditional risk factors, as encompassed by ATP III categorization, was significant and substantial (net reclassification improvement of 18% for overall mortality and 22% for cardiac death/nonfatal MI), irrespective of the vasodilator used in our reclassification analysis. These results suggest that including HRR in the
Coronary Artery Disease/Vasodilator HRR Reclassification
initial screening for vascular disease risk assessment will measurably improve the ability to discriminate future cases and noncases. Of particular interest is the identification of a subgroup that is truly at low risk for 2 years after the MPI study, patients with an HRR in the upper tertile had a risk that is not unlike that seen with normal exercise MPI findings. Furthermore, a smaller subset of patients with low to intermediate ATP III risk and high HRR had an exceptionally low risk. Our study had several limitations. The results were obtained from a single, tertiary care academic institution and might not be generalizable. Despite the large sample size, these results should be considered exploratory, because we used a retrospective study design with all the limitations that poses. We are aware that in a study that depends on a retrospective chart review, ensuring the completeness of follow-up for nonfatal events (our secondary outcome of cardiac death/ nonfatal MI) is problematic. It is for this reason we used allcause mortality as the primary outcome for the study because follow-up using the Social Security Death Index is virtually complete and the outcome unbiased. Finally, the cutoffs for risk were artificial. For example, our data used a lower limit cutoff of risk of 10%. A cutoff of 6% has recently been advocated as a better separator of low and intermediate risk.1,5,20 Changing the categories to <6%, 6% to 20%, and >20% did not alter the findings of our study. A similar argument can be made with respect to the categorization of the HRR. A more simple categorization we have recently used identifies those with a HRR of 40% as normal and those with a HRR <20% as blunted. Using this unified categorization, which is not dependent on the population studied or the stress agent used, the net reclassification improvement for all-cause mortality remained unchanged at 18%.
7. 8. 9. 10.
11.
12.
13. 14.
15. 16.
17. 18.
Disclosures Dr. Iskandrian is a consultant for Gilead Sciences and Astellas Pharma; and Dr. Hage is the recipient of an investigator initiated grant from Astellas Pharma. The remaining authors have no conflicts of interest to disclose. 1. Wilson PWF, Pencina M, Jacques P, Selhub J, D’Agostino R, O’Donnell CJ. C-reactive protein and reclassification of cardiovascular risk in the Framingham Heart Study. Circ Cardiovasc Qual Outcomes 2008;1:92e97. 2. Sivapalaratnam S, Boekholdt SM, Trip MD, Sandhu MS, Luben R, Kastelein JJP, Wareham NJ, Khaw KT. Family history of premature coronary heart disease and risk prediction in the EPIC-Norfolk prospective population study. Heart 2010;96:1985e1989. 3. Lee HM, Le H, Lee BT, Lopez VA, Wong ND. Forced vital capacity paired with Framingham Kfor prediction of all-cause mortality. Eur Respir J 2010;36:1002e1006. 4. Elias-Smale SE, Proença RV, Koller MT, Kavousi M, van Rooij FJA, Hunink MG, Steyerberg EW, Hofman A, Oudkerk M, Witteman JC. Coronary calcium score improves classification of coronary heart disease risk in the elderly: the Rotterdam study. J Am Coll Cardiol 2010;56:1407e1414. 5. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837e1847. 6. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third Report of the National
19.
20. 21.
22.
23.
24.
25. 26.
195
Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002;106:3143e3421. Hage FG, Iskandrian AE. Heart rate response during vasodilator stress myocardial perfusion imaging: mechanisms and implications. J Nucl Cardiol 2010;17:536e539. Iskandrian AE, Garcia EV. Nuclear Cardiac Imaging: Principles and Applications. New York: Oxford University Press, 2008. Iskandrian AE, Garcia EV. Atlas of Nuclear Cardiology: Imaging Companion to Braunwald’s Heart Disease. Philadelphia: Saunders Elsevier, 2011. Navare SM, Mather JF, Shaw LJ, Fowler MS, Heller GV. Comparison of risk stratification with pharmacologic and exercise stress myocardial perfusion imaging: a meta-analysis. J Nucl Cardiol 2004;11: 5531e5561. Shaw LJ, Peterson ED, Shaw LK, Kesler KL, DeLong ER, Harrell FE Jr, Muhlbaier LH, Mark DB. Use of a prognostic treadmill score in identifying diagnostic coronary disease subgroups. Circulation 1998;98:1622e1630. Hage FG, Dean P, Iqbal F, Heo J, Iskandrian AE. A blunted heart rate response to regadenoson is an independent prognostic indicator in patients undergoing myocardial perfusion imaging. J Nucl Cardiol 2011;18:1086e1094. Hage FG, Dean P, Bhatia V, Iqbal F, Heo J, Iskandrian AE. The prognostic value of the heart rate response to adenosine in relation to diabetes mellitus and chronic kidney disease. Am Heart J 2011;162:356e362. Iqbal FM, Hage FG, Ahmed A, Dean PJ, Raslan S, Heo J, Iskandrian AE. Comparison of the prognostic value of normal regadenoson to normal adenosine myocardial perfusion imaging using propensity score matching. JACC Cardiovasc Imaging 2012;5:1014e1021. Iskandrian AE. Stress-only myocardial perfusion imaging: a new paradigm. J Am Coll Cardiol 2010;55:231e233. Germano G, Kiat H, Kavanagh PB, Moriel M, Mazzanti M, Su HT, Van Train KF, Berman DS. Automatic quantification of ejection fraction from gated myocardial perfusion SPECT. J Nucl Med 1995;36: 2138e2147. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1eS266. Hage FG, Heo J, Franks B, Belardinelli L, Blackburn B, Wang W, Iskandrian AE. Differences in heart rate response to adenosine and regadenoson in patients with and without diabetes mellitus. Am Heart J 2009;157:771e776. Hachamovitch R, Hayes SW, Friedman JD, Cohen I, Berman DS. Stress myocardial perfusion single-photon emission computed tomography is clinically effective and cost effective in risk stratification of patients with a high likelihood of coronary artery disease (CAD) but no known CAD. J Am Coll Cardiol 2004;43:200e208. Pencina MJ, D’Agostino RB, D’Agostino RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157e172. 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. Shaw LJ, Hage FG, Berman DS, Hachamovitch R, Iskandrian AE. Prognosis in the era of comparative effectiveness research: where is nuclear cardiology now and where should it be? J Nucl Cardiol 2012;19:1026e1043. Bhatheja R, Francis GS, Pothier CE, Lauer MS. Heart rate response during dipyridamole stress as a predictor of mortality in patients with normal myocardial perfusion and normal electrocardiograms. Am J Cardiol 2005;95:1159e1164. Mathur S, Shah AR, Ahlberg AW, Katten DM, Heller GV. Blunted heart rate response as a predictor of cardiac death in patients undergoing vasodilator stress technetium-99m sestamibi gated SPECT myocardial perfusion imaging. J Nucl Cardiol 2010;17:617e624. Hage FG, Iskandrian AE. Cardiac autonomic denervation in diabetes mellitus. Circ Cardiovasc Imaging 2011;4:79e81. Maser RE, Mitchell BD, Vinik AI, Freeman R. The association between cardiovascular autonomic neuropathy and mortality in individuals with diabetes: a meta-analysis. Diabetes Care 2003;26:1895e1901.
0002-9149/12/$ - see front matter Ó 2013 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.amjcard.2012.09.013
probability of coronary heart disease and, therefore, have a slightly greater risk of cardiac death and myocardial infarction (MI) despite normal MPI findings (1% to 2%/year for vasodilator stress vs <1% for exercise).10 Furthermore, the prognostic value of functional status provided by the exercise portion of the test is not available.11 The best method for risk stratification in this population is not well established. We have recently shown that the heart rate response (HRR) to adenosine and regadenoson is an independent predictor of outcome, provides incremental prognostic information to clinical and imaging variables, and helps in better risk stratification.7,12,13 The present study examined whether the addition of HRR to adenosine or regadenoson to traditional risk stratification in patients with normal MPI findings improves risk stratification. Methods The study population included 2,000 patients who underwent vasodilator MPI at the University of Alabama at Birmingham: 1,000 consecutive patients with normal adenosine MPI findings and another 1,000 consecutive patients with normal regadenoson MPI findings (after our center switched completely from using adenosine to using regadenoson).14 For the purposes of the present retrospective study, normal MPI findings were defined as a normal perfusion www.ajconline.org
Coronary Artery Disease/Vasodilator HRR Reclassification Table 1 Baseline characteristics of study population according to all-cause mortality Characteristic Age (yrs) Men Race White Black Other Hypertension Hyperlipidemia Diabetes mellitus Chronic kidney disease End-stage renal disease Congestive heart failure Peripheral vascular/cerebrovascular disease Previous myocardial infarction Previous coronary intervention Previous bypass surgery Framingham risk score Former smoker Current smoker In-patient stress test
Alive (n ¼ 1,764) Dead (n ¼ 236) 58.99 12.24* 778 (44.1%)† 1,108 626 30 1,397 883 654 459 452 115 300
(62.8%) (35.5%) (1.7%) (79.2%)* (50.1%)* (37.1%) (26.0%) (25.6%) (6.5%) (17.0%)
116 (6.6%) 214 (12.1%)† 115 (6.5%) 11.26 4.20† 454 (25.7%)† 316 (17.9%) 378 (21.4%)
62.71 11.52* 122 (51.7%)† 162 70 4 162 76 82 74 60 19 42
(68.6%) (29.7%) (1.7%) (68.6%)* (32.2%)* (34.7%) (31.4%) (25.4%) (8.1%) (17.8%)
17 (7.2%) 17 (7.2%)† 16 (6.8%) 11.97 3.70† 84 (35.6%)† 46 (19.5%) 51 (21.6%)
* p <0.001. † p <0.05.
Table 2 Medication use by event status for all-cause mortality Medication Aspirin Plavix b Blockers Angiotensin-converting enzyme inhibitors/receptor blockers Calcium channel blockers Other antihypertensives Thiazide diuretics Loop diuretics Potassium-sparing diuretics Statins Insulin Metformin Sulfonylureas Glitazones
Alive (n ¼ 1,764)
Dead (n ¼ 236)
704 204 839 850
(39.9%)* (11.6%)† (47.6%)† (48.2%)*
62 15 88 82
(26.3%)* (6.4%)† (37.3%)† (34.7%)*
518 288 406 392 146 732 295 190 187 84
(29.4%) (16.3%) (23.0%)† (22.2%) (8.3%)† (41.5%)* (16.7%) (10.8%)† (10.6%) (4.8%)†
58 36 34 64 35 56 41 13 18 4
(24.6%) (15.3%) (14.4%)† (27.1%) (14.8%)† (23.7%)* (17.4%) (5.5%)† (7.6%) (1.7%)†
191
Table 3 Hemodynamic parameters by event status for all-cause mortality Parameter
Alive (n ¼ 1,764)
Dead (n ¼ 236)
Heart rate at rest (beats/min) Systolic blood pressure at rest (mm Hg) Diastolic blood pressure at rest (mm Hg) Heart rate response Left ventricular ejection fraction (%)
71.37 13.00* 132.38 21.54*
75.79 15.15* 128.27 21.70*
75.07 11.26*
71.82 11.11*
31.91 20.79* 66.88 9.45
21.62 15.95* 67.25 9.80
* p <0.001.
Table 4 Laboratory results by event status for all-cause mortality Alive (n ¼ 1,207) Total cholesterol (mg/dl) High-density lipoprotein (mg/dl) Low-density lipoprotein (mg/dl) Triglycerides (mg/dl) Serum creatinine (mg/dl)z Estimated glomerular filtration rate (ml/min/1.73 m2)
177.47 46.01 103.62 157.25 1.18 59.22
49.99* 18.94 39.31† 116.32 0.72 36.38
Dead (n ¼ 111) 166.74 47.62 95.74 159.01 1.19 56.03
44.06* 34.05 38.67† 107.31 0.54 36.29
* p <0.001 for patients who had died versus those alive at 2 yrs after MPI within same category. † p <0.05 for patients who had died versus those alive at 2 yrs after MPI within same category. z Serum creatinine and estimated glomerular filtration rate values were averaged for patients without end-stage renal disease and not requiring dialysis.
* p <0.001. † p <0.05.
pattern and normal left ventricular ejection fraction (LVEF) 50%. A total of 3,000 patients undergoing MPI were screened, patients with abnormal MPI findings (n ¼ 617) or LVEF <50% (n ¼ 383) were excluded. Using this database, we have previously shown that patients with normal regadenoson MPI findings experienced outcomes similar to those with normal adenosine MPI findings.14 Patients selected for the present study underwent either adenosine or regadenoson MPI. Both agents were delivered using peripheral intravenous access. Adenosine was administered as a continuous infusion (140 mg/kg/min for 6 minutes), and regadenoson was administered as a single
Figure 1. Scatter plot demonstrating relation between Framingham risk score and HRR. Scatter plot has superimposed linear regression line and corresponding equation.
bolus of 0.4 mg, followed by a saline flush. Technetium99m-sestamibi determined by the patient’s weight was injected at 3 minutes into the infusion of adenosine and 10 to 20 seconds after the saline flush following regadenoson administration, as previously described.14 All the studies were done in the absence of accompanying exercise. The heart rate and blood pressure were measured at baseline and every 2 minutes until the completion of the study.
192
The American Journal of Cardiology (www.ajconline.org)
Table 5 Hazard ratios for heart rate response (HRR) with and without adjustment All-Cause Mortality HR (95% CI) Unadjusted Adjusted for baseline heart rate Adjusted for age and gender Adjusted for age, gender, baseline heart rate, and ATP III risk category Adjusted for age, gender, baseline heart rate, ATP III risk category, and LVEF
2.54 2.25 2.47 2.09 2.10
(1.97e3.29) (1.71e2.96) (1.91e3.19) (1.59e2.76) (1.59e2.77)
p Value <0.001 <0.001 <0.001 <0.001 <0.001
Cardiac death/MI HR (95% CI) 4.01 2.95 4.03 2.92 2.93
(2.14e7.51) (1.52e5.75) (2.15e7.55) (1.49e5.71) (1.50e5.72)
p Value <0.001 0.001 <0.001 0.002 0.002
CI ¼ confidence interval; HR ¼ hazard ratio.
Figure 2. Annualized event rates of all-cause mortality and cardiac death/ nonfatal MI categorized by HRR tertile.
Medications (including b blockers) and caffeinated beverages were withheld on the morning of the stress portion of the study. Gated single photon emission computed tomography images were acquired 1 hour after tracer injection using a dual-head detector gamma camera with a 20% energy window centered on the 140-keV gamma peak. Cameras were operated in an elliptical 180 acquisition orbit with 32 projections and 30 to 40 seconds per projection. Gating was done with 8 to 16 frames per RR cycle. Images were processed using filtered back projection. All MPI studies were interpreted visually, aided by automated quantitative analysis, by readers who were unaware of subsequent events and without attenuation or scatter correction. At rest images were obtained whenever uncertainty was present in the interpretation of the stress images, as previously described.15 The LVEF and end-diastolic and end-systolic volumes were measured from the stress gated images using the method previously described by Germano et al.16 The variables abstracted from the patients’ medical records included patient demographics (age, gender, race); co-morbidities, including hypertension, hyperlipidemia, diabetes mellitus, stroke, previous MI, previous cardiovascular interventions, such as percutaneous coronary intervention or coronary artery bypass grafting; history of tobacco use; medication usage at MPI, and laboratory results (serum creatinine and lipid panel). The estimated glomerular filtration rate was calculated according to the Modification of Diet in Renal Disease study formula.17 Patients were considered to have chronic
kidney disease if their estimated glomerular filtration rate was 15 to 60 ml/min/1.73 m2. Patients were considered to have end-stage renal disease if their estimated glomerular filtration rate was <15 mL/min/1.73 m2 or if they were taking renal replacement therapy. The HRR was calculated as the maximum percentage of change from baseline, as previously described.12,13,18 For each of the stress agents, patients were divided into 3 groups according to the HRR tertiles. The HRR tertiles for adenosine were defined as <16%, 16% to 32%, and >32%. The corresponding tertiles for regadenoson were <21%, 21% to 37%, and >37%. The primary outcome of interest was defined as all-cause mortality and was determined using the Social Security Death Index (assessed on July 31, 2011). The secondary outcome was a composite of cardiac death and nonfatal MI (as documented by the appropriate combination of symptoms, electrocardiographic findings, and enzyme changes).19 Cardiac death was defined as death from fatal arrhythmia, MI, or heart failure and was determined by reviewing the electronic medical records and death certificates. All events were censored at 2 years after the index MPI study. The interval to an event was defined as the duration from the baseline MPI study to death, cardiac death/nonfatal MI, or the end of 2 years of follow-up. All statistical analyses were performed using the Statistical Package for Social Sciences, version 18, for Windows (SPSS, Chicago, Illinois). Student’s t test and the MannWhitney U test were used to compare continuous variables and the chi-square test or Fischer’s exact test to compare categorical variables, as appropriate. The Framingham risk score for the 10-year risk of hard coronary heart disease events were calculated using age, gender, total cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, active treatment of hypertension, and smoking status.6 Patients were then stratified into risk groups according to the calculated Framingham risk score (low, <10%; intermediate, 10% to 20%; and high, >20%). Patients with preexisting diabetes mellitus or vascular disease at baseline were assigned to the high-risk group, irrespective of calculated Framingham risk score according to ATP III.6 Patients were additionally substratified within ATP III categories according to the HRR tertile. Cox proportional hazards models were then used to calculate the hazard ratios and corresponding 95% confidence intervals for the primary and secondary end points in relation to HRR tertiles. The
Coronary Artery Disease/Vasodilator HRR Reclassification
193
Figure 3. Reclassification of patients experiencing all-cause mortality using baseline ATP III risk category modified by HRR. High-risk (red), intermediate-risk (purple), and low-risk (blue) patients presented separately.
Figure 4. Reclassification of patients experiencing cardiac death/nonfatal MI (CDMI) using baseline ATP III risk category modified by HRR. High-risk (red), intermediate-risk (purple), and low-risk (blue) patients presented separately.
variables entered in the model were age, gender, ATP III category, and LVEF. Variables with co-linearity were entered into the model 1 at a time. Stepwise forward selection was used to create the final model. The incremental effect of adding HRR to the ATP III categorization for predicting outcomes was evaluated using the net reclassification improvement.20 After risk categorization according to the ATP III guidelines, patients in the lowest HRR tertile (i.e., blunted HRR) were reclassified up 1 risk category, to a maximum of high risk. Patients in the middle HRR tertile were not reclassified. Finally, patients in the highest HRR tertile were reclassified down 1 risk category, to a minimum of low risk. The goal was to determine whether reclassification would assign patients who developed events to a higher risk category and those who did not to a lower risk category. The statistical significance of the net reclassification improvement was assessed, as described by Pencina et al.20 The institutional review board for human research at the University of Alabama at Birmingham approved the present study.
Results The baseline characteristics (Table 1), medication use (Table 2), hemodynamic parameters (Table 3), and laboratory results (Table 4) of the patients are presented grouped by stress agent. The cohort consisted of 2,000 patients with a mean age of 59 12 years and an LVEF of 67 10%; 45% were men, 64% were white, 37% had diabetes mellitus, 24% had a history of cardiovascular disease, 26% end-stage renal disease, and 18% were active smokers. After 2 years of follow-up, 236 patients (11.8%) died. Patients who died were older, more likely to be men, former smokers, have a lower low-density lipoprotein level and a greater Framingham risk score, and were less likely to have hypertension, hyperlipidemia, or a history of percutaneous coronary intervention (Table 1). In addition, those who died were less likely to be receiving cardiovascular medications known to improve survival (aspirin, clopidogrel, b blockers, angiostensin-converting enzyme inhibitors/ angiostensin receptor blockers, and statins; Table 2) and had a greater heart rate at rest, lower systolic and diastolic blood
194
The American Journal of Cardiology (www.ajconline.org)
pressure at rest, and lower HRRs (Table 3). The secondary end point (cardiac death/nonfatal MI) occurred in 54 patients (2.7%). Scatter plot analyses showed that HRR was significantly associated with Framingham risk score but with very poor correlation (Figure 1). On Cox proportional hazard analysis, and after controlling for age, gender, LVEF, and ATP III category, a HRR in the lowest tertile was associated with an increased risk of the primary (overall mortality) and secondary (cardiac death/nonfatal MI) outcomes (hazard ratio 2.10, 95% confidence interval 1.59 to 2.77; and hazard ratio 2.93, 95% confidence interval 1.50 to 5.72, respectively; Table 5). Patients with normal perfusion on MPI and classified to be at low risk by ATP III experienced an annual mortality and cardiac death/nonfatal MI rate of 5.5% and 1.0%, respectively (n ¼ 635). Patients classified to be in the highest HRR tertile, irrespective of ATP III categorization, had an annual mortality and cardiac death/nonfatal MI rate of 3.1% and 0.5%, respectively (n ¼ 691; Figure 2). Alternatively, patients with a HRR in the highest tertile and classified in the low- or intermediate-risk categories by ATP III had an annual mortality and cardiac death/nonfatal MI rate of 2.9% and 0.1%, respectively (n ¼ 359). Adding HRR to the classic risk stratification with ATP III resulted in an net reclassification improvement for all-cause mortality of 18.2% (8.9% for deaths and 9.3% for survivors, p <0.0001; Figure 3) and cardiac death/nonfatal MI of 21.9% (7.6% for events and 14.3% for nonevents, p ¼ 0.022; Figure 4). Discussion The results of the present study have demonstrated that risk stratification with HRR is useful in patients with normal vasodilator MPI findings when added to a classic risk stratification model, such as the ATP III risk categories, or by itself. Patients with normal vasodilator MPI findings have a residual risk attributable to elevated pretest risk. In this population, even patients classified as low risk using the ATP III algorithm were at a relatively high risk of events during the subsequent 2 years (annual mortality rate of 5.5% and annual cardiac death/nonfatal MI rate of 1.0%). Patients with an HRR in the highest tertile had event rates that were lower, irrespective of their ATP III classification (annual mortality rate 3.1% and annual cardiac death/ nonfatal MI rate of 0.5%), not unlike those reported for normal exercise MPI findings.21 Furthermore, patients who were classified as at low or intermediate risk by ATP III and had a HRR in the highest tertile (18% of the population) were at very low risk of events in the 2 years after the test (annual mortality rate 2.9% and cardiac death/nonfatal MI rate 0.1%). A large body of evidence supports the prognostic use of MPI.8 Both perfusion abnormalities and a depressed LVEF are proven strong predictors of a poor outcome.22 Clinically, the value of normal MPI findings might be even more important than abnormal test findings, because it signals a benign outcome, despite the presence of factors associated with greater risk.14 Navare et al10 combined data from 24 studies (almost 15,000 patients) to show that the
cardiac event rate was significantly greater with normal and abnormal test results from pharmacologic stress MPI versus exercise stress MPI. Patients with normal pharmacologic MPI had an annualized event rate for cardiac death and myocardial infarction of 1.78% compared to 0.65% for normal exercise MPI (p <0.001). Rozanski et al21 followed up >6,000 patients who underwent exercise or adenosine MPI and had normal test results for a mean of 10 years. Annualized mortality was significantly greater in the adenosine cohort with (3.9% vs 1.6%, p <0.0001) and without (4.3% vs 1.1%, p <0.0001) propensity matching. The increased risk has been attributed to an increased pretest risk in the population referred for pharmacologic stress, rather than an inaccuracy of the test itself. This pretest risk is a reflection of the increased prevalence of poor prognostic indicators in patients referred to pharmacologic stress, as well as poor functional capacity that precludes exercise. Patients with normal vasodilator MPI findings, therefore, have a residual risk that is two- to fourfold greater than those with normal exercise MPI findings. Risk classification in patients with normal MPI findings (with both exercise and pharmacologic stress tests) has not been well studied. Previous reports12,23,24 found that HRR can risk stratify patients with normal vasodilator MPI findings. Our data have expanded on these earlier reports by showing that HRR provides incremental information to traditional risk stratification models. Although the traditional Framingham risk factors have been used in multiple diverse studies for risk stratification, it is important to note that patients with normal vasodilator MPI findings are unique in several respects. First, they have an indication for stress testing. Second, they have normal perfusion and LV function. Third, they have limited functional capacity. Our results have indicated that the patients in our cohorts with low ATP III risk continued to be at relatively high risk. In the study by Rozanski et al,21 propensity matching for these and other factors did not account for the increased risk in patients with vasodilator MPI findings. The change in heart rate in response to adenosine receptor agonists is mediated by the autonomic nervous system.25 Because cardiac autonomic dysfunction is independently associated with increased cardiovascular risk,26 these hemodynamic changes are expected to carry prognostic significance. A blunted HRR to adenosine,13 dipyridamole,24 or regadenoson12 has been associated with worse outcomes. This easily derived indicator provides prognostic value that is independent of traditional MPI data and other known prognostic indicators. In a recent study that followed up 1,156 patients who underwent regadenoson MPI for a mean of almost 2 years, patients with a normal HRR had a relatively low annualized total mortality despite the presence of risk factors.12 The present study demonstrated that the HRR can contribute to better risk stratification in patients with normal vasodilator MPI findings. The overall effect on discrimination with the addition of HRR to traditional risk factors, as encompassed by ATP III categorization, was significant and substantial (net reclassification improvement of 18% for overall mortality and 22% for cardiac death/nonfatal MI), irrespective of the vasodilator used in our reclassification analysis. These results suggest that including HRR in the
Coronary Artery Disease/Vasodilator HRR Reclassification
initial screening for vascular disease risk assessment will measurably improve the ability to discriminate future cases and noncases. Of particular interest is the identification of a subgroup that is truly at low risk for 2 years after the MPI study, patients with an HRR in the upper tertile had a risk that is not unlike that seen with normal exercise MPI findings. Furthermore, a smaller subset of patients with low to intermediate ATP III risk and high HRR had an exceptionally low risk. Our study had several limitations. The results were obtained from a single, tertiary care academic institution and might not be generalizable. Despite the large sample size, these results should be considered exploratory, because we used a retrospective study design with all the limitations that poses. We are aware that in a study that depends on a retrospective chart review, ensuring the completeness of follow-up for nonfatal events (our secondary outcome of cardiac death/ nonfatal MI) is problematic. It is for this reason we used allcause mortality as the primary outcome for the study because follow-up using the Social Security Death Index is virtually complete and the outcome unbiased. Finally, the cutoffs for risk were artificial. For example, our data used a lower limit cutoff of risk of 10%. A cutoff of 6% has recently been advocated as a better separator of low and intermediate risk.1,5,20 Changing the categories to <6%, 6% to 20%, and >20% did not alter the findings of our study. A similar argument can be made with respect to the categorization of the HRR. A more simple categorization we have recently used identifies those with a HRR of 40% as normal and those with a HRR <20% as blunted. Using this unified categorization, which is not dependent on the population studied or the stress agent used, the net reclassification improvement for all-cause mortality remained unchanged at 18%.
7. 8. 9. 10.
11.
12.
13. 14.
15. 16.
17. 18.
Disclosures Dr. Iskandrian is a consultant for Gilead Sciences and Astellas Pharma; and Dr. Hage is the recipient of an investigator initiated grant from Astellas Pharma. The remaining authors have no conflicts of interest to disclose. 1. Wilson PWF, Pencina M, Jacques P, Selhub J, D’Agostino R, O’Donnell CJ. C-reactive protein and reclassification of cardiovascular risk in the Framingham Heart Study. Circ Cardiovasc Qual Outcomes 2008;1:92e97. 2. Sivapalaratnam S, Boekholdt SM, Trip MD, Sandhu MS, Luben R, Kastelein JJP, Wareham NJ, Khaw KT. Family history of premature coronary heart disease and risk prediction in the EPIC-Norfolk prospective population study. Heart 2010;96:1985e1989. 3. Lee HM, Le H, Lee BT, Lopez VA, Wong ND. Forced vital capacity paired with Framingham Kfor prediction of all-cause mortality. Eur Respir J 2010;36:1002e1006. 4. Elias-Smale SE, Proença RV, Koller MT, Kavousi M, van Rooij FJA, Hunink MG, Steyerberg EW, Hofman A, Oudkerk M, Witteman JC. Coronary calcium score improves classification of coronary heart disease risk in the elderly: the Rotterdam study. J Am Coll Cardiol 2010;56:1407e1414. 5. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97:1837e1847. 6. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third Report of the National
19.
20. 21.
22.
23.
24.
25. 26.
195
Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002;106:3143e3421. Hage FG, Iskandrian AE. Heart rate response during vasodilator stress myocardial perfusion imaging: mechanisms and implications. J Nucl Cardiol 2010;17:536e539. Iskandrian AE, Garcia EV. Nuclear Cardiac Imaging: Principles and Applications. New York: Oxford University Press, 2008. Iskandrian AE, Garcia EV. Atlas of Nuclear Cardiology: Imaging Companion to Braunwald’s Heart Disease. Philadelphia: Saunders Elsevier, 2011. Navare SM, Mather JF, Shaw LJ, Fowler MS, Heller GV. Comparison of risk stratification with pharmacologic and exercise stress myocardial perfusion imaging: a meta-analysis. J Nucl Cardiol 2004;11: 5531e5561. Shaw LJ, Peterson ED, Shaw LK, Kesler KL, DeLong ER, Harrell FE Jr, Muhlbaier LH, Mark DB. Use of a prognostic treadmill score in identifying diagnostic coronary disease subgroups. Circulation 1998;98:1622e1630. Hage FG, Dean P, Iqbal F, Heo J, Iskandrian AE. A blunted heart rate response to regadenoson is an independent prognostic indicator in patients undergoing myocardial perfusion imaging. J Nucl Cardiol 2011;18:1086e1094. Hage FG, Dean P, Bhatia V, Iqbal F, Heo J, Iskandrian AE. The prognostic value of the heart rate response to adenosine in relation to diabetes mellitus and chronic kidney disease. Am Heart J 2011;162:356e362. Iqbal FM, Hage FG, Ahmed A, Dean PJ, Raslan S, Heo J, Iskandrian AE. Comparison of the prognostic value of normal regadenoson to normal adenosine myocardial perfusion imaging using propensity score matching. JACC Cardiovasc Imaging 2012;5:1014e1021. Iskandrian AE. Stress-only myocardial perfusion imaging: a new paradigm. J Am Coll Cardiol 2010;55:231e233. Germano G, Kiat H, Kavanagh PB, Moriel M, Mazzanti M, Su HT, Van Train KF, Berman DS. Automatic quantification of ejection fraction from gated myocardial perfusion SPECT. J Nucl Med 1995;36: 2138e2147. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1eS266. Hage FG, Heo J, Franks B, Belardinelli L, Blackburn B, Wang W, Iskandrian AE. Differences in heart rate response to adenosine and regadenoson in patients with and without diabetes mellitus. Am Heart J 2009;157:771e776. Hachamovitch R, Hayes SW, Friedman JD, Cohen I, Berman DS. Stress myocardial perfusion single-photon emission computed tomography is clinically effective and cost effective in risk stratification of patients with a high likelihood of coronary artery disease (CAD) but no known CAD. J Am Coll Cardiol 2004;43:200e208. Pencina MJ, D’Agostino RB, D’Agostino RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157e172. 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. Shaw LJ, Hage FG, Berman DS, Hachamovitch R, Iskandrian AE. Prognosis in the era of comparative effectiveness research: where is nuclear cardiology now and where should it be? J Nucl Cardiol 2012;19:1026e1043. Bhatheja R, Francis GS, Pothier CE, Lauer MS. Heart rate response during dipyridamole stress as a predictor of mortality in patients with normal myocardial perfusion and normal electrocardiograms. Am J Cardiol 2005;95:1159e1164. Mathur S, Shah AR, Ahlberg AW, Katten DM, Heller GV. Blunted heart rate response as a predictor of cardiac death in patients undergoing vasodilator stress technetium-99m sestamibi gated SPECT myocardial perfusion imaging. J Nucl Cardiol 2010;17:617e624. Hage FG, Iskandrian AE. Cardiac autonomic denervation in diabetes mellitus. Circ Cardiovasc Imaging 2011;4:79e81. Maser RE, Mitchell BD, Vinik AI, Freeman R. The association between cardiovascular autonomic neuropathy and mortality in individuals with diabetes: a meta-analysis. Diabetes Care 2003;26:1895e1901.