Prognostic Value of Gated Myocardial Perfusion Imaging for Asymptomatic Patients With Type 2 Diabetes: The J-ACCESS 2 investigation

The J-ACCESS 2 prognostic registry is a prognostic cohort study of 513 patients from 50 institutions (8) who were registered between June 2004 and September 2005. Certified physicians specializing in diabetes at all institutions participated in the 3-year follow-up. The inclusion criteria included age ≥50 years with type 2 diabetes and either a maximal carotid artery intima-media thickness (max IMT) of ≥1.1 mm (10) or a urinary albumin level of ≥30 mg/g creatinine (11) or at least two of the following, abdominal obesity (BMI ≥25 kg/m² and abdominal circumference ≥85 cm for men and ≥90 cm for women), hypo–HDL cholesterolemia (HDL cholesterol level >40 mg/dl), hypertriglyceridemia (triglyceride level ≥150 mg/dl), and hypertension (blood pressure ≥130/85 mmHg). The abdominal circumference criteria conformed to Japanese guidelines. Exclusion criteria comprised MI, effort angina, and unstable angina. Patients with A1C ≥10% within 1 month before enrollment or evidence of nephropathy (serum creatinine measurement ≥1.5 mg/dl within 1 month before enrollment) were excluded. Also excluded were those with valvular disease, idiopathic cardiomyopathy, evidence of abnormalities on rest electrocardiography (ECG) such as atrial fibrillation, NYHA (New York Heart Association) class III or IV heart failure at the time of myocardial perfusion SPECT, and peripheral arterial disease (Fig. 1). Estimated glomerular filtration rates (eGFRs) were calculated according to the equation proposed by Cockcroft and Gault (12). On the basis of these criteria, we selected patients who were asymptomatic in terms of ischemic heart disease but who had potential event risks from diabetes complications.

The institutional review board of each hospital approved the study, which complied with the Ethical Guidelines for Epidemiological Research in Japan. Written informed consent was obtained from all participants before starting the study.

Stress-rest myocardial perfusion SPECT images were obtained using the 1-day technetium-99m–tetrofosmin protocol at 94% of the institutions and the 2-day protocol at the remainder. Exercise, dipyridamole, adenosine, and adenosine-triphosphate stress tests were performed in 71 15, 7, and 6% of the patients, respectively. The administered dose of technetium-99m–tetrofosmin was 200–400 MBq (average 331 MBq) for the first study and 700–800 MBq (748 MBq) for the second. The gated study proceeded on patients at rest. Standard SPECT image acquisition protocols used 64 × 64 matrices over 360° or 180° rotations.

Myocardial perfusion images of short-axis, vertical long-axis, and horizontal long-axis images were generated using prefiltering (mainly Butterworth) and ramp filters. All SPECT images were interpreted by experienced physicians using a 20-segment model, as in the J-ACCESS study (3). The results of the segmentation were comparable with estimations based on the 17-segment model (13). Each of the myocardial segments was visually scored using a 5-point system: 0, normal; 1, mildly reduced; 2, moderately reduced; 3, severely reduced; and 4, absent. Totals were calculated as summed stress, rest, and difference scores (SSS, SRS, and SDS) as validated by precedent prognostic studies (7). The maximum score was 80 points (20 segments × 4 points/segment). Risk-based grouping was based on two SSS severity categories: <9 (normal or mildly abnormal) and ≥9 (moderately or severely abnormal) as categorized in the J-ACCESS study (3).

Quantitative gated SPECT proceeded as described (3,14). SPECT slices were reconstructed using the standard software provided by the manufacturer. Gated SPECT images were analyzed at each institution using QGS software (Cedars-Sinai Medical Center, Los Angeles, CA) with automated processing. Inappropriate edge tracing was manually adjusted. Values for left ventricular ejection fraction (LVEF, %), end-diastolic volume (EDV, ml), and end-systolic volume (ESV, ml) were obtained.

Cardiovascular events were investigated at 1, 2, and 3 years after registration, and the present final prognostic analysis was based on 3-year follow-up data from 485 of the 513 patients because 7 who underwent revascularization within 30 days of registration, 7 who died because of noncardiac causes, and 14 who did not attend the hospital were excluded. The end point of the follow-up was hard events defined as sudden or cardiac death and acute coronary syndrome. Total events additionally included severe heart failure requiring hospitalization, percutaneous coronary intervention, coronary artery bypass grafting, de novo stable angina, unstable angina, nonsevere heart failure, transient ischemic attack of the brain, as well as stroke, and peripheral artery disease.

Continuous variables are expressed as means ± SD. We applied the Wilcoxon rank sum test to compare results from patients with and without cardiovascular events and applied the χ² test to categorical data. The independent variables in the univariate Cox proportional hazard model included age, sex, cardiac risk factors (chest pain, BMI, hypertension, dyslipidemia, diabetes, history of smoking, family history of coronary artery disease, and ECG abnormalities), summed perfusion defect scores, quantitative gated SPECT parameters of ejection fraction (EF), and volumes. Medications before the follow-up were also included. The dependent variable was the occurrence of total events. The threshold value for events was an EF of 45% and male and female subject ESV values of 60 and 40 ml, respectively (13). The relative hazard ratios (HRs) and 95% CIs were calculated. The multivariate Cox proportional model was applied using a forward stepwise method based on statistically significant independent variables excluding medication in the univariate Cox proportional hazards model. Two prognostic severity categories based on SSS values of <9 or ≥9 were included in the multivariate analysis. Statistical significance was defined as P < 0.05.

Among the criteria for patient registry, IMT ≥1.1 mm and urinary albumin ≥30 mg/g of creatinine were satisfied in 37 and 36% of the patients, respectively. The frequency of the other criteria was hypertension (blood pressure ≥130/85 mmHg or taking hypotensive drugs) in 81%, hypertriglyceridemia (triglyceride level ≥150 mg/dl, HDL cholesterol level <40 mg/dl, or taking hypolipidemic drugs) in 80%, and abdominal obesity in 21%. BMI was 24.8 ± 3.6 kg/m², and IMT was 1.6 ± 0.8 mm. With respect to other risk factors, 19% of patients were current smokers, and 7% had a family history of coronary artery disease. Rest ECG findings were atypical in 89 (21%) of 426 (88%) of the patients. Exercise ECG findings suggested myocardial ischemia in 50 (28%) of 180 (37% of the total) patients (Table 1).

Five patients died during the 3-year follow-up, three due to sudden death and two to cardiac death. 9 patients were diagnosed with nonfatal acute coronary syndrome, 3 developed heart failure requiring hospitalization, 10 were diagnosed with stable angina pectoris, and 3 were diagnosed with nonsevere heart failure. Cerebrovascular accidents and transient ischemic attacks occurred in 15 and 2 patients, respectively, and 6 patients were diagnosed with peripheral artery disease. Recurrent and initial percutaneous coronary intervention proceeded in one and six patients, respectively. Other cardiovascular accidents occurred in two patients.

Figure 2 shows Kaplan-Meier curves for event-free survival for cardiac events. At the end of follow-up, the event-free rates were 0.88 and 0.77 for the groups with low and high SSS, respectively (P = 0.009). The urinary excretion rate of albumin was higher in the high SSS group (P = 0.017). ECG revealed a higher incidence of segmental wall motion abnormalities (P = 0.013) and the myocardial perfusion defect scores of SRS and SDS were higher (both P < 0.0001) in the high SSS group. Both EDV and ESV were higher, whereas EF was lower in the high SSS group (all P < 0.0001; Table 1).

The following event summary is based on sudden or cardiac death and nonfatal cardiovascular events. Table 1 compares patients with and without events. Patients with events were older (P = 0.05) and had higher total cholesterol levels (P = 0.020). They also had a significantly lower eGFR (P = 0.014) and were frequently medicated with ticlopidine (P = 0.011) or insulin (P = 0.039). The results of nuclear studies did not significantly differ.

Univariate analysis showed that the significant variables for total events were age, current smoking, use of insulin or ticlopidine, high total cholesterol level, low eGFR, low LVEF, and SSS ≥9 (Table 2). Multivariate Cox regression analysis applied using these significant variables but excluding medication revealed that SSS ≥9 (HR 3.385 [95% CI 1.783–6.426]; P = 0.0002), eGFR (0.982 [0.971–0.992]; P = 0.0008), and current smoking (2.083 [1.194–3.632]; P = 0.0097) were independent predictors of cardiac events.

Coronary artery disease is a leading cause of death among Western and Japanese patients with diabetes. The prevalence of diabetes in the Japanese population is rapidly increasing as the lifestyle becomes more Westernized. Therefore, the cardiovascular event rate among asymptomatic diabetic Japanese patients is a matter of considerable concern.

The rate of cardiac deaths and cardiovascular events among patients with type 2 diabetes in the J-ACCESS 2 population during the 3-year follow-up was 13%. Scholte et al. (15) found a comparable rate of cardiovascular events in 6 (5%) of 120 patients during a follow-up period of 12 months, whereas the Detection of Ischemia in Asymptomatic Diabetics (DIAD)-2 study identified a 0.6% annual cardiovascular event rate among asymptomatic diabetic patients (16). The offsetting factors of healthier patients capable of exercise testing and more intensive therapy might have been associated with the rather low cardiac event rate in the DIAD-2 study. Asymptomatic patients were selected in the J-ACCESS 2 investigation from the viewpoint of coronary artery disease but with potential event risks included in the selection criteria. In addition, because the patients were indicated for myocardial perfusion imaging by a physician, patients with relatively higher risk might have been included.

The rate of cardiac deaths and cardiovascular events in the J-ACCESS 2 study was 2.7-fold higher than that of recent epidemiological surveys conducted by the Japanese Diabetes Society (17). The rate of cardiovascular events excluding noncardiac death and heart failure in the J-ACCESS 2 study of 56 events over a period of 3 years was also 2.3-fold higher than that (1.7% per annum) of the nonaspirin group in the more recent Japanese Primary Prevention of Atherosclerosis with Aspirin for Diabetes (JPAD) study of asymptomatic Japanese patients with type 2 diabetes (18). Furthermore, this high cardiovascular event rate was quite comparable with those of groups given intensive and standard treatment (22.1 and 22.4% over a period of 5 years, respectively) in the ADVANCE trial and with the findings of the ACCORD trial (19,20). The J-ACCESS study demonstrated remarkably low cardiovascular risk among Japanese, even among those with type 2 diabetes (3). Asymptomatic patients with type 2 diabetes were enrolled for the J-ACCESS 2 study, and we validated markers of high risk such as a max IMT of ≥1.1 mm (10), a urinary albumin level of >30 mg/g creatinine (11), and at least two of the four relevant conditions to ensure the registration of those with appropriate diagnostic yields. Considering the substantial difference in cardiovascular event rates between Western and Japanese diabetic patients, we consider that these conditions for enrolment into J-ACCESS 2 might have specifically selected diabetic patients at high risk for cardiovascular diseases.

The role of SPECT for detecting myocardial ischemia has been validated (5,6), and general consensus has also been reached regarding prognostic evaluation. Because the risk of cardiac death and MI increases with increasing degrees of scan abnormalities, the summed score is now widely accepted as described in the guidelines of the American Cardiology Society (7). The rate of sudden or cardiac deaths and cardiovascular events in the J-ACCESS 2 population with asymptomatic type 2 diabetes was 1.9-fold higher among individuals with high than with low SSS. The cardiac event rate (including hospitalization and sudden or cardiac death) was 1.9-fold higher among diabetic patients with than without myocardial ischemia evaluated by myocardial perfusion scintigraphy in the MERIDIAN (Multicentre Trial of Early Revascularisation In Patients with Diabetes Mellitus Type 2 and Mild Anginal Symptoms) trial (21).

To apply the multivariate model, the total numbers of patients and events were limited from the final analysis. We therefore used five variables with the smallest P values according to the univariate Cox proportional hazards model. Current smoking and a low eGFR, as well as a high SSS, were found to be independently associated with the increased rate of cardiovascular events in the present study. A large-scale community-based population study revealed an independent, graded association between a reduced eGFR and the risk of death and cardiovascular events (22). Irie et al. (23) also found in a 10-year prospective cohort study that individuals with a reduced eGFR had a 1.65- or 1.81-fold higher risk of cardiovascular diseases. Univariate Cox hazard analysis revealed a higher incidence of cardiovascular events among patients treated with insulin or ticlopidine at baseline. Further analyses on possible affection of baseline medication or additional medication on total events in the J-ACCESS 2 study might clarify the clinical meaning of these observations.

Each patient was scored at one of the 50 participating institutions because collecting and evaluating all images at a core center was unfeasible. However, variability of SSS and SRS scores among these institutions was acceptable, and the precision of the quantitative gated SPECT parameters regarding EF and volumes was excellent (14).

The frequency of a high SSS was lower in the J-ACCESS 2 (10%) than in the J-ACCESS study (34%). The incidence of major cardiac events including sudden or cardiac death, acute coronary syndrome, and severe heart failure requiring hospitalization during a 3-year follow-up in J-ACCESS 2 was 3.5% (17 of 485), which is a little lower than that of diabetic patients without prior MI in the J-ACCESS study (5.7%). Ongoing analysis of the treatment modality might support the decreased rate of major cardiac events in J-ACCESS 2.

In conclusion, the cardiovascular event rate was high and comparable among asymptomatic Japanese and Western diabetic patients. Abnormal myocardial perfusion images might be prognostic for a further twofold higher risk of cardiovascular disease among asymptomatic diabetic patients with conventional cardiovascular risks such as dyslipidemia, hypertension, and abdominal obesity, as well as high carotid atherosclerosis or microalbuminuria. A reduced eGFR and current smoking were other independent predictors for identifying high-risk groups among asymptomatic diabetic patients. The present data indicate that myocardial ischemia would be useful for risk stratification of cardiovascular events in asymptomatic diabetic patients.

The J-ACCESS 2 study was supported by a grant from the Japan Cardiovascular Research Foundation.

No potential conflicts of interest relevant to this article were reported.

Y.Y. and K.N. wrote the manuscript and analyzed the data. H.K. reviewed/edited the manuscript. T.I., A.K., R.K., K.S., and N.Y. contributed to the discussion. T.N. was the chief investigator who analyzed and summarized the data.

The authors acknowledge the work of all investigators involved the J-ACCESS 2 study (a complete list of the J-ACCESS study investigators is available in the online appendix at http://care.diabetesjournals.org/cgi/content/full/dc09-2370/DC1). The authors also thank N. Foster for preparing the manuscript and for editorial assistance, as well as Kenji Ichioka and the staff of the J-ACCESS investigation office for their logistical assistance with the multicenter study.