Author + information
- Received September 12, 2003
- Revision received May 14, 2004
- Accepted June 23, 2004
- Published online January 4, 2005.
- Navin Rajagopalan, MD*,
- Todd D. Miller, MD, FACC*,* (, )
- David O. Hodge, MS†,
- Robert L. Frye, MD, FACC* and
- Raymond J. Gibbons, MD, FACC*
- ↵*Reprint requests and correspondence:
Dr. Todd D. Miller, Mayo Clinic, Gonda 5, 200 First Street, SW, Rochester, Minnesota 55905
Objectives The purpose of this study was to identify which asymptomatic diabetic patients are candidates for screening single-photon emission computed tomography (SPECT) imaging and to examine angiographic findings and mortality in patients according to SPECT imaging categories.
Background Previously we reported a high percentage of abnormal and high-risk SPECT imaging scans in asymptomatic diabetic patients.
Methods We examined the associations between several clinical and laboratory variables and a high-risk stress SPECT imaging scan in 1,427 asymptomatic diabetic patients without known coronary artery disease (CAD). Results of coronary angiography and long-term outcome were also analyzed.
Results An abnormal stress SPECT imaging scan was present in 826 patients (58%) and a high-risk scan in 261 patients (18%). Multivariate analysis demonstrated that seven variables were independently associated with a high-risk scan (model chi-square = 107, p < 0.0001). The two most important variables were electrocardiogram (ECG) Q waves (adjusted chi-square = 38.3, p < 0.001) and peripheral arterial disease (PAD) (adjusted chi-square = 13.9, p < 0.001). Coronary angiography was performed in 127 (49%) high-risk SPECT imaging patients, 61% of whom had angiographic high-risk CAD. Annual mortality rates for patient subsets categorized by SPECT imaging scans were high-risk 5.9%, intermediate-risk 5.0%, and low-risk 3.6% (p < 0.001 for differences between groups).
Conclusions High-risk findings on stress SPECT imaging were present in 18% of asymptomatic diabetic patients without known CAD. Patients with high-risk scans had a high prevalence of severe CAD and a high annual mortality rate. ECG Q waves and/or evidence of PAD identified the most suitable candidates for screening.
Approximately 75% of diabetic patients die of coronary artery disease (CAD) (1). Coronary artery disease is more likely to be silent in diabetic patients (2,3). American Diabetes Association guidelines recommend screening for CAD in asymptomatic diabetic patients who have an abnormal resting electrocardiogram (ECG) indicative of myocardial infarction (MI) or ischemia, peripheral arterial disease (PAD), or two or more additional CAD risk factors (4). These recommendations are the result of expert opinion and are not evidence-based. Stress single-photon emission computed tomography (SPECT) imaging is accurate for diagnostic and prognostic purposes in general and in diabetic populations (5–9). In a previous study we compared the yield of stress SPECT imaging in asymptomatic and symptomatic diabetic and nondiabetic patients (10). An abnormal SPECT imaging scan was present in 59% of asymptomatic diabetic patients, including 20% with a “high-risk” scan. The purpose of this study was to define the coronary anatomy and prognosis of asymptomatic diabetic patients with a high-risk scan and to identify a subset of asymptomatic diabetic patients who would benefit most from screening SPECT imaging.
The study was approved by the Mayo Clinic Institutional Review Board. The nuclear cardiology database was used to identify the study population (10). Between January 1986 and December 2000, 67,828 stress SPECT imaging studies were performed. The following exclusion criteria were applied: 1) clinical history of documented MI (n = 19,511); 2) prior percutaneous coronary artery intervention or bypass surgery (n = 22,841); 3) left bundle branch block or paced rhythm (n = 4,406); 4) significant valvular heart disease (n = 5,679); 5) refusal to authorize use of medical record for research purposes as required by Minnesota law (n = 77); and 6) technically suboptimal scan (n = 2,408). These criteria were applied to exclude patients with known CAD or with entities associated with false-positive SPECT imaging studies. For patients who underwent multiple studies (n = 9,736), only the first study was analyzed. From this population, 1,738 patients with diabetes who were coded as asymptomatic were identified.
The charts of all 1,738 patients were subsequently reviewed to address several potential limitations of the database. The goal of the study was to identify asymptomatic diabetic patients without present or past evidence of CAD. One limitation of using the database was related to coding of symptom status only at the time of stress SPECT imaging. For instance, a patient who developed chest pain three months before stress testing and was started on anti-anginal medication with resolution of symptoms would be coded as asymptomatic. Prior performance of coronary angiography at the Mayo Clinic or another institution was not recorded in the database. Peripheral arterial disease, which is common in diabetic patients and is a marker of CAD (11), also was not recorded. Finally, modest changes in the definition of diabetes occurred during the study period. The charts were reviewed to eliminate patients with previous chest symptoms or known CAD by coronary angiography; to ensure that all patients met the current definition of diabetes (fasting glucose >126 mg/dl) (12); and to collect information on PAD (including carotid artery disease), type of diabetes, and indication for testing. This review excluded 224 patients with a past history of chest symptoms or CAD by angiography and 87 patients who did not meet the definition of diabetes. The final study group consisted of 1,427 patients.
Clinical variables were obtained via the nuclear laboratory database and chart review. Hypertension was defined as blood pressure >140/90 mm Hg or prior diagnosis of hypertension. Hyperlipidemia was defined as total cholesterol or triglycerides >90th percentile for age or use of lipid-lowering medication. Family history of premature CAD was defined as CAD in a first-degree relative <60 years of age. The presence of PAD was determined by symptoms, physical examination, and/or vascular studies. Laboratory variables were obtained by cross-linking with the Mayo Clinic laboratory medicine database. Variables closest in time to the selected patient's stress study were chosen. Lipid values were available in 1,315 patients (92%), creatinine in 1,396 patients (98%), and glycosylated hemoglobin in 1,314 patients (92%).
Stress SPECT imaging
These methods have been described previously (13,14). Stress protocols employed treadmill exercise (n = 739), intravenous adenosine 140 μg/kg/min for 6 min or dipyridamole 0.56 mg/kg over 4 min (n = 688), or intravenous dobutamine in 10 μg/kg/min increments every 3 min to a peak dose of 40 to 60 μg/kg/min (n = 51).
Thallium-201 SPECT imaging was performed using a one-day protocol. Three to four mCi of thallium-201 were injected during stress. Five to 10 min later, an anterior planar image was acquired for 5 min for assessment of cardiac size and pulmonary uptake, followed by tomographic imaging over a 180° arc using the “step-and-shoot” method. Delayed images were acquired three to four hours later. After January 1, 1990, patients were re-injected with 1 mCi of thallium-201. Technetium-99m sestamibi SPECT imaging was performed as a 1-day (8 to 10 mCi rest, 22 to 24 mCi stress) or 2-day (30 mCi rest, 15 mCi stress) protocol. Tomographic imaging was started 45 to 60 min after injection of sestamibi.
Stress and rest images were displayed in 3 planes (short-axis, horizontal long-axis, vertical long-axis) divided into 24 segments. Uptake in each segment was graded by consensus of two experienced observers using a five-point scale (0 = absent, 1 = severely diminished, 2 = moderately diminished, 3 = mildly diminished, 4 = normal). Mild fixed defects were considered normal because the large majority represents soft tissue attenuation. Moderate or severe fixed defects or defects with any reversibility were considered abnormal. Summed stress score (SSS) and summed rest score (SRS) were determined as the summation of perfusion grading in all 14 short-axis segments. A normal image score is 56 (14 × 4). The SSS risk categories were assigned according to Cedars-Sinai criteria adapted for our scoring system as low-risk (SSS ≥53), intermediate-risk (SSS 48 to 52), and high-risk (SSS ≤47) (15).
Results of coronary angiography performed at the Mayo Clinic within six months were collected. Angiograms were interpreted subjectively by two experienced angiographers and graded according to Coronary Artery Surgery Study criteria (luminal diameter narrowing ≥50% left main or ≥70% left anterior descending, left circumflex, right coronary artery, or their major branches considered significant) (16).
Mortality data were obtained from Mayo Clinic records and the Social Security Death Index. Survival was estimated using the Kaplan-Meier method. Survival curves were compared using the log-rank statistic. Mean duration of follow-up was 5.8 ± 3.5 years.
Associations between a high-risk SPECT scan and 25 clinical and laboratory variables were tested using logistic regression models. Multivariate modeling included the bootstrap approach. One thousand bootstrap samples were taken from the population of 1,427 patients. Multivariate models were constructed using logistic regression and a stepwise selection technique for each of these 1,000 samples. Variables that entered at least 65% of the models were selected for the final multivariate models. Test indication (preoperative or other) was not initially identified as a variable for these analyses but subsequently was included in post-hoc analyses. For all analyses, a p value <0.05 was considered statistically significant.
Clinical variables are described in Table 1. The study population consisted predominantly of middle-age obese males with type 2 diabetes and a high prevalence of other risk factors. Median duration of diabetes was 10 years. Prevalence of PAD was 31%. Indications for stress testing were: 50% pre-surgical evaluation, 39% screening purposes, 3% vague non-chest symptoms such as fatigue or abdominal pain, and 5% various miscellaneous indications such as arrhythmias or sildenafil use. In 3% of patients, an indication could not be determined. Peripheral arterial disease was present in 44% of patients referred for pre-operative testing versus 19% of patients for a different indication.
ECG and laboratory values
Pertinent laboratory studies are summarized in Table 2.Median creatinine was 1.2 mg/dl, with 29% of patients having creatinine >1.5 mg/dl. Mean total cholesterol was 209 ± 53 mg/dl, low-density lipoprotein (LDL) cholesterol 127 ± 46 mg/dl, and high-density lipoprotein cholesterol 44 ± 15 mg/dl. Nearly one-fifth of patients were on a statin. Mean glycosylated hemoglobin was 9.1% ± 2.6%.
Stress SPECT imaging results
An abnormal stress SPECT imaging scan was present in 826 patients (58%) and a high-risk scan in 261 patients (18%). Univariate analysis revealed that 10 variables were significantly associated with a high-risk scan (Table 3).In the multivariate analysis, ECG Q waves demonstrated the strongest association (Table 4).The other six independent variables in the final model (chi-square = 107, p < 0.001) were PAD, glycosylated hemoglobin, male gender, age, pharmacologic stress, and LDL cholesterol.
Analysis of patients with ECG Q waves or PAD
The 131 patients with ECG Q waves had a significantly lower SRS than the rest of the population (51 ± 7 vs. 54 ± 4; p < 0.001). Because Q waves were strongly associated with a high-risk scan, another multivariate model was constructed excluding patients with Q waves. Four variables were found to be significantly associated with a high-risk scan: male gender, pharmacologic stress, PAD, and glycosylated hemoglobin. This model (chi-square = 69.4) was weaker than the original model. The prevalence of PAD in patients without a pre-operative indication was 19%. Peripheral arterial disease was significantly associated with a high-risk scan in this subset of patients (univariate chi-square = 4.7, p = 0.03).
Influence of referral bias
Patients referred to tertiary care centers are often “sicker” than community-based patients (17). Olmsted County, Minnesota, residents (community-based population, n = 145) were compared to patients who resided outside Olmsted County (referral population, n = 1282). In referral patients, test indication was 51% pre-surgical evaluation and 39% screening. These percentages were 41% and 46%, respectively, in the community cohort. There were no differences between the community and referral populations in clinical characteristics (Table 5)or in percentages of patients with abnormal scans (65% vs. 57%, p = 0.08) or high-risk scans (22% vs. 18%, p = 0.22).
Table 6shows the yield of high-risk scans in specific subsets of patients. The American Diabetes Association guidelines recommend screening diabetic patients with an ECG indicative of ischemia or MI, evidence of PAD, or two or more risk factors for CAD (4). Compared to the entire study population, patients with an abnormal resting ECG or with PAD were more likely to have a high-risk scan, but those with two or more risk factors for CAD were not. The National Cholesterol Education Program guidelines recommend treating hyperlipidemia in diabetic patients as aggressively as in patients with established CAD (18). A high-risk scan was found in 20% of patients with LDL cholesterol >100 mg/dl, versus 14% of those with LDL cholesterol <100 mg/dl (p = 0.02). Test indication (pre-operative versus other) was not significantly associated with a high-risk scan (univariate chi-square = 0.2, p = NS). In patients without pre-operative test indication, Q waves, or PAD (a “less selected” subset, n = 517), 15% still had a high-risk scan.
Coronary angiography was performed within 6 months in 212 patients (Table 7). Among high-risk SPECT imaging patients, the majority (61%) had left main, three-vessel, and/or proximal left anterior descending CAD, and only seven had no significant CAD. Relatively few low-risk SPECT imaging patients (16%) had high-risk angiographic CAD.
By 10 years, 44% of the population had died. Survival curves according to SPECT imaging SSS risk categories are shown in Figure 1.Annual mortality rates were: high-risk 5.9%, intermediate-risk 5.0%, and low-risk 3.6% (p < 0.001 for differences between groups). Post-hoc analyses were performed to determine if a truly low-risk (annual mortality <1%) subset of patients could be identified. Annual mortality in patients without ECG Q waves or PAD and with a completely normal SPECT imaging scan (n = 443) was lower but was still 2.9%. Annual mortality was higher in patients with pre-operative versus other indication (5.9% vs. 2.7%, p < 0.001). Annual mortality for patients without pre-operative indication whose scan was normal (n = 298) was 1.9%. For patients without pre-operative indication, Q waves, or PAD, annual mortality was 1.6% for those with a normal scan (n = 237) and 3.4% for those with a high-risk scan (n = 79).
This study is the largest to date to analyze the prevalence of CAD by stress SPECT imaging in asymptomatic diabetic patients. The yield of abnormal and high-risk scans was considerable. A high-risk scan is an indication to proceed with coronary angiography (19). Sixty-one percent of patients with a high-risk scan who underwent coronary angiography had high-risk anatomy, consistent with other studies in broader patient populations (13,15,20). Patients predicted to be at high risk by SPECT imaging did in fact have a high annual mortality rate of 5.9%. Patients predicted to be at low risk did have a significantly lower but nonetheless substantial annual mortality of 3.6%. Patient selection bias appeared to influence these results. Post-hoc analysis identified a subset of patients (no pre-operative indication, no Q waves, no PAD) whose annual mortality was 1.6% if their SPECT images were normal. Other studies have reported a higher-than-expected cardiac event rate in diabetic patients with normal SPECT images (21).
Which patients to screen?
A major goal of this study was to identify which asymptomatic diabetic patients might benefit most from screening. The most important variables were ECG Q waves and PAD. High-risk scans were present in 43% of patients with Q waves, in 26% with Q waves and/or ST-T wave abnormalities, and in 28% with PAD. Previous studies have found a high prevalence of CAD in diabetic patients with PAD (11,22,23). These findings support the American Diabetic Association guidelines, which recommend that asymptomatic diabetic patients with ECG abnormalities or PAD should be considered for screening (4).
Comparison to prior studies
The exact prevalence of silent CAD in asymptomatic diabetic patients is difficult to determine. The largest population-based autopsy study of 149 diabetic patients without antemortem evidence of CAD reported that approximately 50% of decedents 30 to 64 years of age and 75% of decedents ≥65 years of age had high-grade coronary atherosclerosis (24). In a study using electron beam computed tomography, 48% had a calcium score consistent with angiographically significant CAD (25). A wide range of abnormal stress myocardial perfusion studies between 4% and 57% has been reported (23,26–36). Direct comparison of these studies to our study is difficult given different selection criteria and other methodological differences. The Detection of Ischemia in Asymptomatic Diabetics (DIAD) study is an ongoing prospective trial designed to examine this issue (37,38). The prevalence of abnormal SPECT image was 16%, considerably lower than in our study. Patients enrolled in the DIAD study versus those in our study likely represent different ends of the spectrum of the asymptomatic diabetic population. Patients selected for clinical trials are commonly “healthier” than the larger population of patients with the disease process (39). Conversely, our patients were referred for stress SPECT imaging for clinical reasons. By chart review, pre-operative assessment was the test indication in half of these patients. Asymptomatic diabetic patients undergoing pre-operative assessment are likely to be at higher risk of CAD compared to the general population of asymptomatic diabetic patients. Patients with ECG abnormalities suggestive of underlying CAD were excluded from the DIAD study. We included these patients in our study because an ECG is commonly acquired as part of the screening process. ECG Q waves were a strong predictor of a high-risk scan. Patients with Q waves also had a significantly lower SRS, a measure of the extent and severity of the perfusion defect on the resting images (i.e., myocardial scar).
The major limitation is patient selection bias, acknowledged above in the discussion of the DIAD study. Clinicians may have been selecting the “sickest” asymptomatic diabetic patients for stress SPECT imaging. Interestingly, there were no differences between the community-based and referral patients. Another limitation relates to retrospective chart review for some data acquisition, which is often incomplete and not as accurate as prospective collection. The study period spanned the years 1986 to 2000. Conceivably, the prevalence of severe CAD and mortality rates could be lower at the present time because of more widespread use of statins and angiotensin-converting enzyme inhibitors in diabetic patients. The SPECT images were interpreted by physicians who were aware of the patients' clinical data, which may have influenced the SPECT reports. Assessment of left ventricular function was not performed in a uniform manner and therefore was not included as part of this study. It is possible that some patients with normal SPECT images may have had left ventricular dysfunction, contributing to their higher-than-expected mortality rate.
Our study cannot definitively address the role of stress SPECT imaging for screening all asymptomatic diabetic patients. The percentage of abnormal scans was higher than in most other studies (23,26–38). Patients with high-risk scans did in fact have a high annual mortality rate approaching 6%. Advantages of our study include the large study population and the use of stress SPECT imaging in a broad clinical practice, including performing the test in many patients as part of pre-operative assessment (a “real-world” practice). The results suggest a role for screening stress SPECT imaging in selected asymptomatic diabetic patients, especially those with ECG abnormalities or evidence of PAD. Although no randomized trial has proven that asymptomatic diabetic patients with severe CAD have an improvement in survival if treated with revascularization, bypass surgery is the treatment of choice for asymptomatic patients with severe CAD according to the American College of Cardiology/American Heart Association guidelines on the basis of expert consensus (19). The findings from our study will need to be interpreted in conjunction with the results of the ongoing DIAD study to formulate more definitive recommendations concerning screening stress SPECT imaging in asymptomatic diabetic patients.
The authors thank Pam McCabe and Tamie Tiedemann for secretarial preparation of the manuscript and Tammy Hudson for assimilation of the mortality data.
- Abbreviations and acronyms
- coronary artery disease
- confidence interval
- Detection of Ischemia in Asymptomatic Diabetics
- low-density lipoprotein
- myocardial infarction
- peripheral arterial disease
- single-photon emission computed tomography
- summed rest score
- summed stress score
- Received September 12, 2003.
- Revision received May 14, 2004.
- Accepted June 23, 2004.
- American College of Cardiology Foundation
- Nesto R.W.
- American Diabetes Association
- Ritchie J.L.,
- Bateman T.M.,
- Bonow R.O.,
- et al.
- Giri S.,
- Shaw L.J.,
- Murthy D.R.,
- et al.
- Berman D.S.,
- Kang X.,
- Hayes S.W.,
- et al.
- The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Expert Panel of Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
- ↵Gibbons RJ, Abrams J, Chatterjee K, et al. ACC/AHA 2002 guideline update for the management of patients with chronic stable angina: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for the Management of Patients with Chronic Stable Angina). Available at: www.acc.org/clinical/guidelines/stable/stable.pdf. Accessed August 2003.
- Goraya T.Y.,
- Leibson C.L.,
- Palumbo P.J.,
- et al.
- Abenavoli T.,
- Rubler S.,
- Fisher V.J.,
- Axelrod H.I.,
- Zuckerman K.P.
- Koistinen M.J.,
- Huikuri H.V.,
- Pirttiaho H.,
- Linnaluoto M.K.,
- Takkunen J.T.
- Janand-Delenne B.,
- Savin B.,
- Habib G.,
- Bory M.,
- Vague P.,
- Lassmann-Vague V.
- DeLorenzo A.,
- Lima R.S.,
- Sequeira-Filho A.G.,
- Pantoja M.R.
- Wackers F.T.H.,
- Zaret B.L.
- Wackers FJT, Young LH, Inzucchi SE, et al., for the DIAD Investigators. Detection of silent myocardial ischemia in asymptomatic diabetic subjects. The DIAD Study. Diabetes Care 2004;27:1954–61.
- Califf R.M.,
- DeMets D.L.