Screening for coronary artery disease in patients with diabetes: A Bayesian strategy of clinical risk evaluation and exercise echocardiography

Screening for coronary artery disease in patients with diabetes: A Bayesian strategy of clinical risk evaluation and exercise echocardiography Dhrubo J. Rakhit, MD,a Melodie Downey, BSN,a Leanne Jeffries, BS,a Stuart Moir, MD,a Johannes B. Prins, MD, PhD,b and Thomas H. Marwick, MD, PhDa Brisbane, Australia

Objective Screening for coronary artery disease is constrained by its low prevalence in unselected patients. We compared the ability of clinical scores to identify a high-risk group with diabetes mellitus and investigated a Bayesian strategy by combination with exercise echocardiography (ExE). Methods

The Framingham risk score (FRS), a score based on the American Diabetes Association (ADA) screening guidelines, the United Kingdom Prospective Diabetes Study (UKPDS) risk engine, and a disease-specific diabetic cardiac risk score (DCRS) were calculated in 199 asymptomatic patients with type 2 diabetes mellitus undergoing ExE. The frequency of abnormal ExE and the proportion of these with coronary stenoses were sought in groups designated as high risk on the basis of optimal cutoffs for each score. All patients were followed up for 1 year.

Results

High risk was identified in fewer patients with the DCRS (27%) than FRS (38%, P = .02), ADA (41%, P = .004), and UKPDS (43%, P = .001). Exercise echocardiography was positive in 27 (14%); 11 of 23 proceeding to angiography showed significant stenoses. Areas under the receiver operator characteristic curves for prediction of a positive ExE were similar for DCRS, UKPDS, and FRS but less for ADA ( P = .04). Positive ExE was uncommon in low-risk patients (8%-11%) and most were false positives (58%-80%). Cardiovascular events (n = 9) were more likely in the high-risk compared with the low-risk UKPDS (9% vs 2%, P = .03) and DCRS (12% vs 2%, P = .01).

Conclusion Combination of the UKPDS or DCRS with ExE may optimize detection of coronary artery disease and cardiac events in asymptomatic patients, while minimizing the numbers of ExE and false-positive rate. (Am Heart J 2005;150:1074- 80.)

In patients with type 2 diabetes mellitus (DM), the risk of coronary artery disease (CAD) is increased 2- to 6-fold,1,2 and prevalence as high as 55% has been reported.3 Moreover, patients with CAD and DM have more extensive disease and increased risk of cardiac death.4 Silent ischemia is common,5 suggesting that some asymptomatic patients with CAD may first present with a cardiac event. Even asymptomatic patients with DM and multiple cardiovascular risk factors have an annual mortality of 3%.6 These data support efforts to identify and intervene on CAD in the preclinical phase.

From the aDepartment of Medicine, and bCentre for Diabetes and Endocrine Research, University of Queensland, Brisbane, Australia. Supported in part by a Center of Clinical Research Excellence Grant, National Health and Medical Research Council, Canberra, Australia. Submitted November 2, 2004; accepted January 14, 2005. Reprint requests: Thomas H. Marwick, MBBS, PhD, Department of Medicine, Princess Alexandra Hospital, University of Queensland, Ipswich Road, Brisbane, Q4102, Australia. E-mail: [email protected] 0002-8703/$ - see front matter n 2005, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2005.01.029

However, a recent review has highlighted difficulties in the detection of CAD in patients with DM, particularly those who are asymptomatic.7 The principal problem with screening is a Bayesian one, because the predictive value of a positive test is constrained by the low overall disease probability in unselected asymptomatic individuals. For example, a test with 80% sensitivity and specificity will have a positive predictive value of 50% if the disease prevalence is 20%, compared with 80% if the disease prevalence is 50% (ie, intermediate risk).8 A process whereby the patients at lowest risk could be removed from the screening program would minimize the chances of obtaining a false-positive test. Therefore, we sought to define a screening strategy in patients with DM using 4 clinical risk models—a score based on recommendations from the American Diabetes Association (ADA),9 the Framingham risk score (FRS),10 the United Kingdom Prospective Diabetes Study (UKPDS) risk engine11, and a recently described diabetic cardiac risk score12 (DCRS), before proceeding to noninvasive tests for vascular structure and function13 and exercise echocardiography (ExE)—which has been shown to predict mortality in patients with DM.14

American Heart Journal Volume 150, Number 5

Rakhit et al 1075

Table I. Diabetic cardiac risk score Variable Age (each decade over 50) Male sex Family history of premature CAD Current smoker LDL cholesterol z160 mg/dL (4.1 mmol/L) Triglyceride z451 mg/dL (5.1 mmol/L) HDL cholesterol z45 mg/dL (1.2 mmol/L) BP N140/90 Peripheral vascular disease Duration of NIDDM (each 5 y) Need for insulin GFR b90 mL/min Abnormal resting ECG

Score +3 +8 +3 +12 +6 +6 4 +8 +12 +2 +4 +5 +5

LDL, Low-density lipoprotein; HDL, high-density lipoprotein; BP, blood pressure; NIDDM, non–insulin-dependent diabetes mellitus; GFR, glomerular filtration rate.

Methods Study design We studied the results of coronary screening strategies in 199 asymptomatic patients with type 2 DM, recruited from the community and a hospital outpatient clinic for entry into a new lifestyle modification program. Clinical evaluation, ExE, and tests of vascular structure and function were performed at baseline, before entry into the program. Baseline demographic data, cardiovascular risk factors, and medications were documented, and a 12-lead electrocardiogram (ECG) was reviewed. Fasting serum was collected for biochemical analysis. Four sets of clinical scores were applied in each individual and noninvasive tests for vascular structure and function (brachial artery reactivity and intima-media thickness [IMT]) and ExE were performed in all. Coronary angiography was sought in those with positive ExE. All patients gave written, informed consent, and the study was approved by the Human Ethics Committee of Princess Alexandra Hospital.

Scoring systems American Diabetes Association score. This score was based on ADA guidelines,9 determined by the presence of DM with z2 cardiac risk factors (glomerular filtration rate b90 mL/min, active smoking, low-density lipoprotein cholesterol z160 mg/dL [4.1 mmol/L] or high-density lipoprotein cholesterol V35 mg/dL [b0.9 mmmol/L], hypertension [N140/90 mm Hg], family history of premature CAD). Low risk was identified by ADA score of b3 (ie, DM with no or one additional risk factor). Framingham risk score. The FRS was derived from clinical and lipid variables.10 UK Prospective Diabetes Study risk engine. An equation based on findings from the UKPDS11 was used to estimate the coronary heart disease risk. The factors used in this model include age at diagnosis, sex, race, smoking status, HbA1c, systolic blood pressure, total cholesterol/high-density lipoprotein ratio.

Table II. Clinical characteristics of patients as mean (SD) or percentage (number) at entry into study Age (y) Male (%) Body mass index (kg/m2) Duration of diabetes (y) Peripheral vascular disease (%) Abnormal ECG (%) Risk factors Positive family history of premature cardiovascular disease (%) Current smoker (%) Hypertension (%) Hyperlipidemia (%) High-sensitive CRP N3 mg/L (%) High-sensitive CRP (mg/L) Lipid profile Triglyceride (mg/dL) High-density lipoprotein (mg/dL) Low-density lipoprotein (mg/dL) Renal function Creatinine (mmol/dL) Glomerular filtration rate (mL/min) GFR b90 mL/min (%) Medications h-Blocker (%) ACE inhibitor (%) Angiotensin II antagonists (%) Calcium-channel blocker (%) Diuretics (%) Statin (%) Nitrates (%) Platelet inhibitors (%) Insulin therapy (%) Sulfonylurea (%) Metformin (%)

57.0 F 10.2 110 (55) 31.4 F 6.0 10.5 F 8.8 17 (9) 63 (32) 33 (17) 21(11) 51 (26) 21 (11) 117 (63) 5.7 F 6.2 184.2 F 185.1 50.1 F 15.4 104.2 F 34.7 0.09 F 0.05 71.4 F 38.6 148 (74) 12 (6) 82 (41) 32 (16) 29 (15) 4 (2) 82 (41) 2 (1) 23 (12) 73 (37) 70 (35) 121 (61)

Diabetic cardiac risk score. This score was modified from the recent report by the group of Machecourt et al,12 based on the outcome of patients undergoing myocardial perfusion imaging. The score for each variable is summarized in Table I and includes demographic (age, sex, risk factors), clinical (blood pressure, duration of diabetes, insulin dependence, peripheral vascular disease), biochemical (lipids, glomerular filtration rate), and ECG variables. A low-risk patient was defined by a DCRS V25. Biochemical markers. Standard biochemistry was performed, as well as an assay using a Beckman Coulter IMMAGE high-sensitivity c-reactive protein (hs-CRP) test.15 Tests of vascular structure and function Brachial artery reactivity. Studies were performed in the fasting state, with vasoactive medications withheld for 24 hours before the study. Longitudinal images of 6 to 8 cm of the brachial artery were taken above the elbow, at rest, after 4.5 minutes occlusion, after a further 15 minutes of rest, and 3 minutes after administration of 400 Ag of sublingual nitroglycerin.16 Vessel diameter was measured using automated edge detection, and the percentage diameter change with

American Heart Journal November 2005

1076 Rakhit et al

Figure 1

Findings of ROC analysis for scoring models for the prediction of positive ExE. The optimal cutoffs to define high- and low-risk individuals are illustrated, as well as the sensitivity and specificity for predicting an abnormal ExE.

cuff occlusion and nitrate was recorded as flow-dependent and -independent dilation, respectively.

Intima-media thickness Longitudinal B-mode ultrasonography of the left and right common carotid arteries was performed at the level of the carotid bifurcation in the anterior, lateral, and posterior planes.17 Offline analysis of magnified, frozen end-diastolic images was performed by an experienced observer with automatic edge-detection software (HDILab version 1.83H,

Philips Medical Systems, Andover, Mass). The far wall IMT was identified as the region between the lumen-intima interface and the media-adventitia interface; localized plaque was excluded.

Exercise echocardiography Patients underwent a maximal treadmill test using a Bruce or modified protocol, selected on the basis of anticipated functional capacity. Patients were prepared for standard 12-lead ECG monitoring; resting blood pressure and echocardiographic images (5 views) were digitally acquired as

American Heart Journal Volume 150, Number 5

Rakhit et al 1077

Table III. Findings of ExE, vascular imaging, and angiography in patients classified as low or high risk based on 4 clinical risk scores and clinical parameters

Table III. (continued) 4. DCRS (high risk N25) High risk (n = 54)

1. Framingham score (high risk N1.5% per year) High risk (n = 76) Abnormal ExE Mean IMT (mm) % Change FMD % Change NMD CAD/angiograms Event/total

15 (20%) 0.72 F 0.12 4.9 F 3.8 12.8 F 6.2 8/11 5/73 (7%)

Low risk (n = 123) 12 (10%) 0.66 F 0.11 5.0 F 9.2 14.6 F 9.3 3/12 4/116 (4%)

P .05 b.0001 .97 .15 .02 .23

Abnormal ExE Mean IMT (mm) % Change FMD % Change NMD CAD/angiograms Event/total

Abnormal ExE Mean IMT (mm) % Change FMD % Change NMD CAD/angiograms Event/total

14 (17%) 0.69 F 0.13 4.7 F 4.3 12.9 F 7.0 6/11 5/79 (6%)

Low risk (n = 118) 13 (11%) 0.68 F 0.11 5.1 F 9.2 14.6 F 9.0 5/12 4/110 (3%)

P .21 .44 .72 .15 .54 .30

3. UKPDS (high risk N1.6% per year) High risk (n = 85) Abnormal ExE Mean IMT (mm) % Change FMD % Change NMD CAD/angiograms Event/total

16 (19%) 0.73 F 0.12 4.7 F 4.4 13.1 F 7.1 9/13 7/81 (9%)

Low risk (n = 114) 11 (10%) 0.65 F 0.10 5.1 F 9.3 14.5 F 9.0 2/10 2/108 (2%)

P .06 b.0001 .73 .26 .02 .03

cine-loops consisting of 1 cardiac cycle. Blood pressure, heart rate, ECG changes, and symptoms were monitored at each stage. Standard end points were used to terminate the test and a diagnostic test was defined as reaching z85% maximal predicted heart rate. Digital images were obtained immediately after stress, stored on magneto-optical disk and analyzed offline by at least 2 trained observers18 blinded to the clinical data and risk scores. A positive test was defined by a resting or inducible wall motion abnormality; differences between the observers were resolved by consensus.

Coronary angiography Coronary angiography was recommended in all patients with positive ExE, although some patients or their referring physician declined. Stenosis severity was measured by an independent angiographer using quantitative coronary angiography (QCA; Philips Medical Imaging, Best, The Netherlands). Significant CAD was defined as a z50% decrease in luminal diameter in z1 major epicardial arteries or their major branches.

Data analysis All data are expressed as mean value F SD or frequency (%), unless otherwise stated. The optimal cutoff to assign high or

11 (8%) 0.67 F 0.11 5.0 F 8.6 14.3 F 8.8 2/10 3/137 (2%)

P b.0001 .04 .89 .26 .02 .01

5. IMT (high risk N0.64 mm)

2. ADA (DM + 2 risk factors) High risk (n = 81)

16 (30%) 0.72 F 0.14 4.8 F 3.6 12.8 F 6.7 9/13 6/52 (12%)

Low risk (n = 145)

Abnormal ExE % Change FMD % Change NMD CAD/angiograms No event

High risk (n = 107)

Low risk (n = 90)

P

16 5.3 F 4.7 12.9 F 7.5 7/12 95 (94%)

11 4.7 F 10.1 15.4 F 8.8 6/11 84 (97%)

.58 .59 .03 .86 .33

High risk n = 54

Low risk (n = 145)

P

17 0.67 F 0.12 5.4 F 10.0 14.1 F 8.5 7/15 87 (96%)

9 0.70 F 0.12 4.7 F 4.7 13.9 F 8.3 4/7 83 (94%)

.08 .18 .55 .88 .65 .48

6. hs-CRP (high risk N4 mg/L)

Abnormal ExE Mean IMT (mm) % Change FMD % Change NMD CAD/angiograms No event

FMD, Flow-mediated dilatation; NMD, nitrate-mediated dilatation.

low risk was obtained by receiver operator characteristic (ROC) analysis of the prediction of ExE findings using each score model. Groups were compared using the 2-tailed independent t test for continuous variables and the m2 test or Fischer exact test for noncontinuous variables, as appropriate. All statistical analyses were performed using SPSS for Windows 11.0 (SPSS Inc, Chicago, Ill).

Results Patient characteristics The clinical characteristics of the patients are summarized in Table II. Many of these patients were on medical treatment for one or more risk factors. High risk was identified in 54 (27%) with the DCRS, fewer than with ADA (81; 41%, P = .004), the Framingham score (76; 38%, P = .02), and the UKPDS (85; 43%, P = .001). Stress echocardiography In the 199 ExE studies, the average exercise capacity was 7.7 F 2.9 metabolic equivalents (METS) and 26 (13%) of patients had a rate pressure product b21 000. The hemodynamic response in most patients was normal;

American Heart Journal November 2005

1078 Rakhit et al

82 (41%) attained b85% predicted maximum heart rate. Five (2.5%) patients developed chest pain and 16 (8%) developed ST-segment changes. Wall motion was completely normal in 172 (86%) and 27 (14%) showed abnormal wall motion either at rest and after exercise (n = 5) or at after exercise only (n = 22). Of the 27 patients with abnormal ExE, 23 proceeded to angiography and significant (N50%) stenoses were identified in 11. Among the 12 false-positive ExE, 3 categories of false positives were identified: those with absolutely no CAD by angiography (n = 5), those with branch vessel disease (n = 2), and those with mild CAD (30%-50% diameter, n = 5).

Receiver operator characteristic analysis Figure 1 shows ROC analysis for prediction of positive ExE, including determination of the best cutoffs to define high and low risks. Comparison of areas under the curves showed no differences between DCRS, UKPDS, and FRS. However, the ADA score had a significantly smaller ROC area than DCRS ( P = .04) and UKPDS ( P = .04). Finally, DCRS and UKPDS had a larger ROC area than IMT ( P = .04) and CRP ( P = .002). Using DCRS, a score of 25 had a sensitivity of 60% and specificity of 77%. For the ADA, a score of z3 had a sensitivity of 52% and specificity of 62%; for the Framingham score, a risk of 1.5% per year had a sensitivity of 60% and specificity of 58%; and for the UKPDS, a cutoff of 1.6% per year gave a sensitivity of 60% and specificity of 60%. Findings of ExE and angiography in clinical risk groups The frequency of positive ExE is summarized in Table III. Positive ExE was slightly more common in the high-risk DCRS (30%) than high-risk ADA (17%, P = .09), high-risk Framingham score (20%, P = .19), or high-risk UKPDS (19%, P = .14). Thus, a positive ExE in combination with a high-risk DCRS, Framingham score, or UKPDS showed a high positive predictive value and contrasts with the same ExE result with a high-risk ADA score, suggesting that the prevalence of CAD was more homogeneous in the high and low ADA groups. In the low-risk groups, positive ExE was present in 11 by DCRS (8%; 20% with CAD), 13 by ADA (11%; 42% with CAD), 12 by Framingham score (10%; 25% with CAD), and 11 by UKPDS (10%; 20% with CAD). Vascular parameters Intima-media thickness was significantly higher in high-risk groups in the DCRS, UKPDS, and Framingham groups, but again was more homogeneous when the groups were defined by the ADA (Table III). Unfortunately, false-positive ExE results could not be predicted on the basis of IMT findings; mean IMT was not

significantly different between those with CAD versus those without CAD (0.69 vs 0.72 mm). Receiver operator characteristic analysis obtained an optimal cutoff of 0.64 mm to assign high risk (Figure 1) but this was not useful as a cutpoint. There were no differences in flow-mediated dilatation or nitrate-mediated dilatation between high- and low-risk groups using any of the clinical risk models.

Biochemical parameters Using a hs-CRP cutoff of 3mg/L,19 62% of patients were high-risk and 87% exceeded a cutoff of 1.5 mg/L. There was no relationship between hs-CRP findings and any of the scoring systems. Receiver operator characteristic analysis obtained a cutoff value of 4 mg/L to assign high risk (Figure 1), but this was not useful for screening. Follow-up Patients were followed up for a mean of 1.1 years. No patient died during the course of the study and 9 cardiovascular events were recorded (2 percutaneous interventions [PCIs], 6 coronary bypass grafting, and 1 nonfatal myocardial infarction). In the high-risk groups, 12% of patients had cardiovascular events using the DCRS, compared with 6% in the ADA profile, 7% in the Framingham profile, and 9% with UKPDS. Event rates were similar (2% - 4%) in the low-risk groups for the 4 scoring systems. High-risk patients had significantly higher event rates compared with low-risk patients using the DCRS (12% vs 2%, P = .01) and UKPDS (9% vs 2%, P = .03).

Discussion The results of this study suggest that screening for CAD in asymptomatic patients with DM can be optimized by use of a clinical score to select for further testing. In particular, use of an optimal DCRS cutoff of 25 (which is the same cutoff used previously12) assigned high risk to b30% of patients, which in a screening program would prevent unnecessary testing in the lowest-risk individuals, who contribute most of the falsepositive results. The ADA score, which is derived from current screening guidelines, was the least effective.

Screening for CAD in diabetes mellitus There is no uniform policy regarding CAD screening in patients with DM, reflecting the statistical challenges of screening patients at low risk and the decisions that should be made from the resulting data. Indeed, some authors argue that screening would make no useful contribution to management.20 As about 20% of asymptomatic type 2 male subjects with diabetes have significant CAD,21,22 the prevalence of CAD in asymptomatic subjects with DM may be insufficient to avoid the pitfalls of screening patients at low risk. Various strategies have been tried to focus

American Heart Journal Volume 150, Number 5

screening on the highest-risk patients. Janand-Delenne21 has recommended screening of male patients with diabetes N10 years or if one additional risk factor is present. Apart from screening higher-risk patients defined by a resting ECG suggestive of ischemia or infarction, or established occlusive arterial disease, the ADA guidelines9 recommend screening in those with a sedentary lifestyle, aged N35 years, with plans to begin a vigorous exercise program, and those with z2 risk factors in addition to diabetes. The Framingham risk profile is another means of identifying a higher-risk group, although assessment in other populations has suggested an overestimate of absolute coronary risk.23 In these patients with increased likelihood of disease, the optimal screening test is unclear. The Framingham profile has been combined with CRP24 or coronary artery calcium scoring in a nondiabetic population.25 Myocardial perfusion scintigraphy appears to be effective, but its cost and availability may be a limiting feature. The use of vascular imaging or ExE as screening tests in such a strategy has not been defined. Exercise echocardiography is more specific but has a lower sensitivity than perfusion scintigraphy for the detection of CAD26,27—this may actually be of some benefit in its application to a screening population. Its prognostic value in patients with known or suspected CAD28 is comparable with perfusion imaging.29 In patients with diabetes, ExE results have prognostic value in those with known or suspected CAD.30,31 In unselected asymptomatic individuals with diabetes or without diabetes, ExE does not add prognostic information32 and is not recommended in those with a low pretest probability of CAD,33 but its use in asymptomatic diabetes is undefined.

Clinical implications The finding of a positive ExE carries implications for both medical management and revascularization. As a positive test signifies the possibility of underlying CAD, an aggressive risk factor management approach should be adopted. The role of revascularization in asymptomatic patients with diabetes and CAD is unclear. These patients tend to have more diffuse CAD than their nondiabetic counterparts and often have poor target vessels, which may contribute to worse outcomes after revascularization. In randomized studies comparing revascularization either by coronary bypass surgery or by PCI, patients with DM who underwent coronary bypass surgery had a survival benefit over patients who had PCI.34,35 In addition, patients with diabetes have higher restenosis rates after PCI.36,37 Indeed, even after an invasive strategy, DM remains an independent predictor of death and myocardial infarction, suggesting that strict risk factor modification is important to reduce event rates.38 The availability of drug-eluting stents will likely have an impact on decision making in this situation.

Rakhit et al 1079

Limitations of the study The study sample is small and this may have constraints in statistical power. The lack of a reference test in the patients with negative ExE might engender concerns about verification bias, but 2 points are pertinent in this regard. First, the selective use of angiography is unavoidable in a screening population, whose risk is too low to justify an angiogram in this situation. Indeed, angiography was not deemed to be clinically justified in all patients with a positive ExE. We sought to overcome this limitation by short-term follow-up, which showed a benign course in those with a negative ExE. Second, we sought to compare the efficacy of various screening strategies rather than to define the accuracy of ExE in this population. It seems improbable that verification bias would influence the results of one screening tool more than others. The false-positive rate of ExE in this study likely reflects the behavior of the test in a low-risk group. However, diffuse atherosclerosis may have lead to underestimation of stenosis severity because of disease in the reference segment. Indeed, in subjects without DM, an abnormal ExE in the setting of binsignificant Q CAD may predict future events.30 In this respect, angiography is an imperfect reference standard in this population, and the use of intravascular ultrasound would be desirable, although it was not possible in this study. Conclusion Combination of ExE with the DCRS—and less effectively, the UKPDS—may optimize detection of CAD or future cardiac events in asymptomatic patients with DM, while minimizing the numbers of required ExE and minimizing false-positive ExE. This approach identifies those at highest risk of subsequent cardiovascular events, in whom adverse outcomes might be avoided by aggressive medical therapy and possibly revascularization, although the prognostic benefit of such a strategy remains undefined.

References 1. Pan WH, Cedres LB, Liu K, et al. Relationship of clinical diabetes and asymptomatic hyperglycemia to risk of coronary heart disease mortality in men and women. Am J Epidemiol 1986;123:504 - 16. 2. Stamler J, Vaccaro O, Neaton JD, et al. Diabetes, other risk factors, and 12-yr cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care 1993;16:434 - 44. 3. Kannel WB, McGee DL. Diabetes and glucose tolerance as risk factors for cardiovascular disease: the Framingham study. Diabetes Care 1979;2:120 - 6. 4. Kleinman JC, Donahue RP, Harris MI, et al. Mortality among diabetics in a national sample. Am J Epidemiol 1988;128:389 - 401. 5. Rutter MK, Wahid ST, McComb JM, et al. Significance of silent ischemia and microalbuminuria in predicting coronary events in asymptomatic patients with type 2 diabetes. J Am Coll Cardiol 2002;40:56 - 61.

1080 Rakhit et al

6. Kannel WB, McGee DL. Diabetes and cardiovascular disease. The Framingham study. JAMA 1979;241:2035 - 8. 7. Kamalesh M, Feigenbaum H, Sawada S. Challenge of identifying patients with diabetes mellitus who are at low risk for coronary events by use of cardiac stress imaging. Am Heart J 2004;147: 561 - 3. 8. Hennessy TG, Codd MB, Kane G, et al. Dobutamine stress echocardiography in the detection of coronary artery disease: importance of the pretest likelihood of disease. Am Heart J 1997;134: 685 - 92. 9. Consensus development conference on the diagnosis of coronary heart disease in people with diabetes: 10-11 February 1998, Miami, Florida. American Diabetes Association. Diabetes Care 1998;21: 1551 - 9. 10. Wilson PW, D’Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation 1998;97: 1837 - 47. 11. Stevens RJ, Kothari V, Adler AI, et al. The UKPDS risk engine: a model for the risk of coronary heart disease in type II diabetes (UKPDS 56). Clin Sci (Lond) 2001;101:671 - 9. 12. Machecourt J, Vanzetto G, Ormezzano O, et al. Diabetes, the silent killer: a therapeutic program in moderate-to-high risk diabetic patients decreases the occurrence of future cardiovascular events. Circulation 2003;108:332. 13. Haluska BA, Fathi R, Jeffriess L, et al. Noninvasive tests for arterial structure, function, and compliance: do they identify risk or diagnose disease? J Am Soc Echocardiogr 2004;17:195 - 202. 14. Marwick TH, Case C, Sawada S, et al. Use of stress echocardiography to predict mortality in patients with diabetes and known or suspected coronary artery disease. Diabetes Care 2002;25:1042 - 8. 15. Ridker PM, Cushman M, Stampfer MJ, et al. Inflammation, aspirin, and the risk of cardiovascular disease in apparently healthy men. N Engl J Med 1997;336:973 - 9. 16. Fathi R, Marwick TH. Noninvasive tests of vascular function and structure: why and how to perform them. Am Heart J 2001;141: 694 - 703. 17. Kennedy R, Case C, Fathi R, et al. Does renal failure cause an atherosclerotic milieu in patients with end-stage renal disease? Am J Med 2001;110:198 - 204. 18. Armstrong WF, Pellikka PA, Ryan T, et al. Stress echocardiography: recommendations for performance and interpretation of stress echocardiography. Stress Echocardiography Task Force of the Nomenclature and Standards Committee of the American Society of Echocardiography. J Am Soc Echocardiogr 1998;11: 97 - 104. 19. Pearson TA, Mensah GA, Alexander RW, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003;107:499 - 511. 20. Sayer JW, Timmis AD. Investigation of coronary artery disease in diabetes: is screening of asymptomatic patients necessary? Heart 1997;78:525 - 6. 21. Janand-Delenne B, Savin B, Habib G, et al. Silent myocardial ischemia in patients with diabetes: who to screen. Diabetes Care 1999;22:1396 - 400.

American Heart Journal November 2005

22. Wackers FJ, Young LH, Inzucchi SE, et al. Detection of silent myocardial ischemia in asymptomatic diabetic subjects: the DIAD study. Diabetes Care 2004;27:1954 - 61. 23. Brindle P, Emberson J, Lampe F, et al. Predictive accuracy of the Framingham coronary risk score in British men: prospective cohort study. BMJ 2003;327:1267. 24. Koenig W, Lowel H, Baumert J, et al. C-reactive protein modulates risk prediction based on the Framingham score: implications for future risk assessment: results from a large cohort study in southern Germany. Circulation 2004;109:1349 - 53. 25. Greenland P, LaBree L, Azen SP, et al. Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA 2004;291:210 - 5. 26. Geleijnse ML, Elhendy A. Can stress echocardiography compete with perfusion scintigraphy in the detection of coronary artery disease and cardiac risk assessment? Eur J Echocardiogr 2000;1: 12 - 21. 27. Schinkel AF, Bax JJ, Geleijnse ML, et al. Noninvasive evaluation of ischaemic heart disease: myocardial perfusion imaging or stress echocardiography? Eur Heart J 2003;24:789 - 800. 28. Marwick TH, Mehta R, Arheart K, et al. Use of exercise echocardiography for prognostic evaluation of patients with known or suspected coronary artery disease. J Am Coll Cardiol 1997;30: 83 - 90. 29. Olmos LI, Dakik H, Gordon R, et al. Long-term prognostic value of exercise echocardiography compared with exercise 201Tl, ECG, and clinical variables in patients evaluated for coronary artery disease. Circulation 1998;98:2679 - 86. 30. Elhendy A, Arruda AM, Mahoney DW, et al. Prognostic stratification of diabetic patients by exercise echocardiography. J Am Coll Cardiol 2001;37:1551 - 7. 31. Bigi R, Desideri A, Cortigiani L, et al. Stress echocardiography for risk stratification of diabetic patients with known or suspected coronary artery disease. Diabetes Care 2001;24:1596 - 601. 32. Marwick TH, Case C, Short L, et al. Prediction of mortality in patients without angina: use of an exercise score and exercise echocardiography. Eur Heart J 2003;24:1223 - 30. 33. Elhendy A, Shub C, McCully RB, et al. Exercise echocardiography for the prognostic stratification of patients with low pretest probability of coronary artery disease. Am J Med 2001;111:18 - 23. 34. The BARI Investigators. Seven-year outcome in the Bypass Angioplasty Revascularization Investigation (BARI) by treatment and diabetic status. J Am Coll Cardiol 2000;35:1122 - 9. 35. King III SB, Kosinski AS, Guyton RA, et al. Eight-year mortality in the Emory Angioplasty versus Surgery Trial (EAST). J Am Coll Cardiol 2000;35:1116 - 21. 36. Schofer J, Schluter M, Rau T, et al. Influence of treatment modality on angiographic outcome after coronary stenting in diabetic patients: a controlled study. J Am Coll Cardiol 2000;35:1554 - 9. 37. Abizaid A, Kornowski R, Mintz GS, et al. The influence of diabetes mellitus on acute and late clinical outcomes following coronary stent implantation. J Am Coll Cardiol 1998;32:584 - 9. 38. Norhammar A, Malmberg K, Diderholm E, et al. Diabetes mellitus: the major risk factor in unstable coronary artery disease even after consideration of the extent of coronary artery disease and benefits of revascularization. J Am Coll Cardiol 2004;43:585 - 91.