The predominant cause of death in diabetes mellitus (DM) is coronary artery disease (CAD). Little is known about prevalence of silent ischemia in developing nations. We compared prevalence of silent ischemia in DM to a control group by exercise myocardial perfusion imaging (MPI) and electrocardiogram (ECG) in developing nations. The prospective multinational Ischemia Assessment with Exercise imaging in Asymptomatic Diabetes study recruited participants at 12 sites in Asia, Africa, and Latin America. DM participants were age- and gender-matched 2:1 to non-DM individuals with ≥1 CAD risk factor. Subjects underwent exercise tests that were interpreted in core labs in blinded fashion. The study included 392 DM and 205 control participants. Among participants with diagnostic ECGs, a similar proportion of DM and controls had ischemic ECG (15% vs 12%, p = 0.5). A significantly higher proportion of DM group had MPI abnormalities compared with controls (26% vs 14%, p <0.001). In participants with ischemia on MPI, only 17% had ischemic ECG, whereas in those without ischemia on MPI, 10% had ischemic ECG. In a multivariable model, DM was independently associated with abnormal MPI (odds ratio 2.1, 95% confidence interval 1.3–3.5, p = 0.004). Women were less likely to have ischemia by MPI than men (10% vs 30%, p <0.001) and concordance between ECG and MPI was much worse in women. In conclusion, in this large prospective study, asymptomatic DM participants had (1) more ischemia by exercise MPI than ECG, (2) more ischemia by MPI but not ECG than control group, and (3) ischemia by MPI was less in women than men.
According to the World Health Organization, 347 million people worldwide have diabetes mellitus (DM), with 80% living in low- and middle-income countries. The predominant cause of death in DM is coronary artery disease (CAD), and DM is commonly considered a CAD-risk equivalent. Testing for myocardial ischemia with electrocardiogram (ECG) and/or myocardial perfusion imaging (MPI) stress tests is widely used in DM to establish diagnosis of CAD and for risk stratification. These tests have similar diagnostic and prognostic capabilities in symptomatic DM as in the general population. Using stress tests to screen for CAD in asymptomatic patients with DM is controversial. Although myocardial ischemia is more likely to be silent in DM, little is known about prevalence of ischemia in asymptomatic patients, especially in developing nations. The most recent guidelines for assessment of risk in asymptomatic adults state that exercise ECG may be considered in intermediate-risk individuals and stress MPI in DM, mostly based on results of studies performed in developed countries. The Role of Myocardial Perfusion Imaging in Ischemia Assessment with Exercise imaging in Asymptomatic Diabetes (IAEA Diabetes) study was performed with support from International Atomic Energy Agency in developing countries to determine prevalence of ischemia by exercise ECG and MPI in patients with asymptomatic type 2 DM and compare results to a control group with other CAD risk factors. The trial was designed to compare prevalence of ischemia by exercise MPI to ECG, which is less expensive, more widely available, and the preferred modality according to current guidelines.
Methods
The primary hypotheses for this study were that (1) ischemia by MPI is more common in DM than control group, and (2) ischemia by MPI is more common than ischemia by ECG. Key secondary hypotheses were that (1) DM participants have lower exercise capacity than controls, (2) DM duration affects prevalence of myocardial ischemia, (3) DM participants have lower left ventricular ejection fraction (LVEF) than controls, and (4) prevalence of ischemia is higher in DM participants with additional risk factors.
The study population was prospectively recruited at 12 sites from Asia, Africa, and Latin America. DM participants were age- and gender-matched in a 2:1 ratio to non-DM individuals with ≥1 CAD risk factor. Inclusion criteria were (1) lack of symptoms of angina or angina-equivalent, (2) age >40 years, (3) normal or near normal resting ECG (abnormalities such as right bundle branch block, slight T wave abnormalities, left anterior hemi block, first degree AV Block were permitted), (4) deemed by physician to be able to exercise >3 minutes, (5) DM duration >5 years, and (6) presence ≥1 CAD risk factor in control group including hypertension (blood pressure >140/90 mm Hg or use of antihypertensive medications); dyslipidemia (LDL cholesterol >130 mg/dl or use of cholesterol-lowering medications); active tobacco consumption and family history of premature atherosclerosis defined as death, myocardial infarction (MI), or coronary revascularization before age 60 years in women and 50 years in men.
Exclusion criteria were (1) history of coronary revascularization, (2) abnormal coronary angiography with >50% stenosis, (3) history of MI, (4) stroke, (5) end-stage renal disease or serum creatinine >3 mg/dl, (6) symptomatic peripheral vascular disease or limb amputation, (7) severe associated disease that limits life expectancy (<1 year), (8) severe valvular disease or any known cardiomyopathy, (9) any of the following ECG abnormalities: old MI, left bundle branch block, paced rhythm, Wolf-Parkinson-White pattern or other pre-excitation syndromes, left ventricular hypertrophy with repolarization changes and atrial fibrillation, (10) previous positive stress test of any type, (11) body size precluding adequate quality MPI (>300 pounds), and (12) inability to sign a consent form or cooperate with study personnel. Data were collected using a central web-based database implemented and maintained at University of Ljubljana. Institutional review boards at each participating center approved the protocol. The study was conducted in compliance with the Declaration of Helsinki. All participants gave written informed consent.
On the day of stress test and before exercising, complete history and physical examination was performed. Blood and urine sample were obtained. Medications were withheld on that morning. Baseline 12-lead ECG was obtained and symptom-limited exercise test using either Bruce protocol or upright bicycle ergometer was performed. ECGs were recorded each minute of exercise and for first 3 minutes during recovery. ECGs were interpreted in blinded manner in a core lab at the University of Alabama at Birmingham. Ischemia was considered present with ≥1-mm ST elevation or depression of horizontal or downsloping variety or 1.5-mm ST depression of upsloping variety in 2 contiguous leads at 80 msec from the J point. The test was considered nondiagnostic if participant failed to reach 85% of age-adjusted maximal predicted heart rate.
Each participant had rest and exercise single-photon emission computed tomography (SPECT) MPI with technetium-99m methoxy-isobutyl-isonitrile using 1-day (8–9 mCi rest, 15–25 mCi stress) or 2-day (15–25mCi for both) protocols. All images were obtained 45 minutes after radiotracer injection using rotating gamma camera equipped with low-energy, high-resolution, parallel hole collimator with 20% symmetric energy window centered at 140 keV. Sixty-four projections (20–30 seconds per projection), 8 or 16 frames per R-R cycle, with a 64 × 64 or 128 × 128 matrix were obtained over a 180° orbit. Images were transferred to a core lab at Spedali Civili and University of Brescia, Italy. For image reconstruction, raw data were visually inspected in cine mode and, if needed, motion corrected by MoCo software. Both stress and rest SPECT images were reconstructed by Ordered Subsets Expectation Maximization (OSEM) iterative reconstruction. No attenuation or scatter correction was applied. A summed rest score (SRS), summed stress score (SSS), and summed difference score (SDS) were calculated based on visual aided by automated analysis according to accepted guidelines. Perfusion was considered to be abnormal if SSS was >3. LVEF was determined by automatic software of gated images. Readers were blind to clinical and ECG data and subjects’ status (DM vs control).
To have 90% power at significance level of 0.05, a sample size of 185 control and 370 DM participants was needed, given the assumption that prevalence of MPI abnormalities would be 15% in control and 30% in DM group. Allowing for dropouts and protocol violations, desired sample size was set at 200 control and 400 DM participants.
Numerical variables were summarized as means ± SD, and categorical variables were summarized as frequencies (percentages). The groups were compared using Mann-Whitney test for numerical variables, and chi-squared test with continuity correction for categorical variables. Multivariable logistic regression was used to investigate association between MPI ischemia and DM, after adjustment for age, gender, body mass index, CAD risk factors, maximum achieved workload, and use of statins or beta blockers; covariates were selected in advance, and recruiting center was included in the model as a random effect. We provide a graphical representation of variability in the estimated association between DM and MPI abnormalities across recruiting centers using gender- and age-adjusted logistic regression models with Firth’s correction. A similar model was estimated for DM participants with addition of DM duration and insulin treatment. Results for these models were presented as odds ratios (OR) with 95% confidence intervals (CI) and p values derived from the Wald test.
All reported p values were 2-sided. Statistical analyses were performed using R.
Results
The study population consisted of 597 participants (392 DM and 205 control subjects). The most common reasons for exclusion (n = 74) included an abnormal ECG (n = 27), missing or uninterpretable MPI (n = 19), pharmacologic stress (n = 12), age ≤40 years (n = 7), absence of risk factors in control group (n = 3), abnormal coronary angiography (n = 2), and MI (n = 1). Baseline characteristics are shown in Table 1 . CAD risk factors were more prevalent in the control group. DM participants achieved lower workload than control participants and had lower peak heart rate even after accounting for age. Hemodynamic changes are summarized in Table 2 .
| Characteristic | DM Group (n = 392) | Control Group (n = 205) | p Value |
|---|---|---|---|
| Age (yrs) | 60 ± 8 | 57 ± 9 | <0.001 |
| Male gender | 237 (61%) | 114 (56%) | 0.3 |
| Races/ethnicities | 0.004 | ||
| African | 24 (6%) | 7 (3%) | |
| Arab | 34 (9%) | 31 (15%) | |
| Asian | 27 (7%) | 12 (6%) | |
| Caucasian | 98 (25%) | 28 (14%) | |
| Indian | 62 (16%) | 37 (18%) | |
| Hispanic | 147 (38%) | 90 (44%) | |
| Hypertension | 287 (73%) | 140 (68%) | 0.2 |
| Dyslipidemia ∗ | 219 (56%) | 141 (69%) | 0.003 |
| Cigarette smoker | 0.01 | ||
| Current | 71 (18%) | 54 (26%) | |
| Former | 56 (14%) | 37 (18%) | |
| Never | 265 (68%) | 114 (56%) | |
| Family history of CAD | 120 (31%) | 93 (45%) | <0.001 |
| CAD risk factors † | <0.001 | ||
| 0 | 26 (7%) | 0 | |
| 1 | 114 (29%) | 47 (23%) | |
| 2 | 180 (46%) | 103 (50%) | |
| 3 | 65 (17%) | 45 (22%) | |
| 4 | 7 (2%) | 10 (5%) | |
| Body mass index (kg/m 2 ) | 28 ± 5 | 27 ± 4 | 0.1 |
| Blood glucose (mg/dl) | 150 ± 58 | 91 ± 13 | <0.001 |
| Hemoglobin A1c >6% || | 252 (68%) | 4 (3%) | <0.001 |
| Creatinine (mg/dl) | 0.96 ± 0.29 | 0.89 ± 0.22 | 0.002 |
| Urine albumin/creatinine (mg/g) ‡ | <0.001 | ||
| <30 | 227 (64%) | 151 (86%) | |
| >30 | 86 (24%) | 21 (12%) | |
| >300 | 39 (11%) | 3 (2%) | |
| Lipid profile (mmol/l, mg/dl) § | |||
| Total Cholesterol | 5.6 ± 2.0, 217 ± 77 | 5.8 ± 2.1, 224 ± 81 | 0.1 |
| LDL | 3.3 ± 1.3, 128 ± 50 | 3.5 ± 1.5, 135 ± 58 | 0.3 |
| HDL | 1.4 ± 0.7, 54 ± 27 | 1.3 ± 0.4, 50 ± 15 | 0.7 |
| Triglycerides | 2.2 ± 1.9, 195 ± 168 | 2.3 ± 1.4, 204 ± 124 | 0.03 |
| Aspirin ¶ | 126/382 (33%) | 61/199 (31%) | 0.6 |
| Beta-blocker ¶ | 123/387 (32%) | 61/203 (30%) | 0.7 |
| ACE-I/ARB ¶ | 214 (55%) | 92 (45%) | 0.03 |
| ACE-I | 184/385 (47%) | 74/202 (37%) | |
| ARB | 36/374 (10%) | 20/201 (10%) | |
| Calcium antagonist ¶ | 100/386 (26%) | 35/201 (17%) | 0.03 |
| Diuretic ¶ | 72/382 (19%) | 34/201 (17%) | 0.6 |
| Statin ¶ | 138/385 (36%) | 82/201 (41%) | 0.3 |
| Diabetes treatment | |||
| Diet only | 24 (6%) | 0 | |
| Oral hypoglycemic | 344 (88%) | 0 | |
| Insulin | 73 (19%) | 0 | |
| Duration of diabetes (yrs) | 11 ± 6 | 0 |
∗ LDL cholesterol >130 mg/dl or use of cholesterol-lowering medications.
† CAD risk factors include hypertension, dyslipidemia, current tobacco consumption, and family history of CAD.
‡ Data not available for 70 participants (40 DM and 30 control).
§ Data not available for 2 participants (DM) on LDL and for 67 (45 DM and 22 control) on triglyceride.
|| Data not available on 99 participants (18 DM and 81 control).
¶ Number of participants on which data are available is listed as the denominator.
| DM Group (n = 392) | Control Group (n = 205) | p Value | |
|---|---|---|---|
| Exercise type | 0.008 | ||
| Treadmill | 301 (77%) | 177 (86%) | |
| Bicycle | 91 (23%) | 28 (14%) | |
| Exercise duration (min) | 7.9 ± 3.6 | 9.1 ± 3.4 | <0.001 |
| Exercise workload (METs) | 8.0 ± 3.1 | 8.9 ± 2.7 | <0.001 |
| HR (beats/min) | |||
| Baseline | 83 ± 14 | 79 ± 13 | 0.005 |
| Peak | 148 ± 20 | 153 ± 18 | <0.001 |
| Exercise HR (% of maximally predicted for age) | 91 ± 10 | 94 ± 10 | <0.001 |
| Systolic BP (mm Hg) | |||
| Baseline | 135 ± 18 | 125 ± 32 | 0.002 |
| Peak | 176 ± 27 | 162 ± 44 | 0.004 |
| Diastolic BP (mm Hg) | |||
| Baseline | 83 ± 10 | 79 ± 20 | 0.2 |
| Peak | 93 ± 14 | 86 ± 22 | <0.001 |
| ECG ischemia ∗ | 0.06 | ||
| Positive | 47 (12%) | 22 (11%) | |
| Negative | 270 (70%) | 158 (79%) | |
| Nondiagnostic | 67 (17%) | 21 (10%) | |
| Myocardial perfusion | <0.001 | ||
| Normal (SSS ≤3) | 291 (74%) | 177 (86%) | |
| Abnormal (SSS >3) | 101 (26%) | 28 (14%) | |
| Perfusion defect size | 0.006 | ||
| No defect | 221 (56%) | 144 (70%) | |
| Small (1–2 segments) | 58 (15%) | 26 (13%) | |
| Intermediate (3–5 segments) | 76 (19%) | 25 (12%) | |
| Large (>5 segments) | 37 (9%) | 10 (5%) | |
| SSS | 3.0 ± 5.4 | 1.5 ± 3.5 | <0.001 |
| SRS | 0.6 ± 2.0 | 0.2 ± 1.3 | <0.001 |
| SDS | 2.3 ± 4.5 | 1.3 ± 3.2 | 0.005 |
| LVEF (%) | 65 ± 11 | 69 ± 11 | <0.001 |
| LVEF <50% | 7 (3%) | 22 (6%) | 0.32 |
| Transient ischemic dilation † | 21 (6%) | 11 (6%) | >0.99 |
∗ ECG not available for 12 participants (8 DM and 4 control).
† Data on transient ischemic dilation not available for 36 DM and 4 control participants.
ECG data were not available for 12 of 597 subjects (8 DM and 4 control). ECG was not diagnostic in response to exercise for 88 of 585 subjects (15%), and more DM participants had nondiagnostic ECGs than control participants (17% vs 10%, p = 0.03). Among participants with diagnostic ECGs similar proportion of DM and control participants had ischemic ECG changes (15% vs 12%, p = 0.5). Except for 1 participant in each group who experienced ST elevation with exertion, ischemic ECG represented ST depression. In participants with ischemic ECG, severe ischemia (≥2-mm ST shift) occurred more frequently in DM participants (56% vs 27%, p = 0.03).
SSS for each participant stratified by DM status and exercise ECG findings is shown in Figure 1 . A higher proportion of DM than control participants had abnormal perfusion (SSS >3) on SPECT imaging (26% vs 14%, p <0.001). Both SSS and SDS were higher in DM than control participants ( Table 2 ). In participants with SSS >3, SRS was not different between groups (2.0 ± 3.5 vs 1.2 ± 3.1, p = 0.08). In the control group, a similar proportion of participants had ischemia by ECG and MPI (14% vs 11%, p = 0.50). In the DM group, more participants had ischemia by MPI than by ECG (26 vs 15%, p <0.001).
More DM than control participants had perfusion abnormalities that were moderate or large ( Figure 1 , p = 0.002). These findings were unchanged when SDS was used instead of SSS. The DM group also had larger defects when assessed as the number of myocardial segments involved ( Table 2 ). Majority of participants with MPI ischemia had negative rather than nondiagnostic ECGs in both groups ( Figures 1 and 2 ).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
