In total, 9,269 consecutive South Korean individuals aged 20 years and older who had undergone self-referral CCTA evaluation as part of a general health examination in the Health Screening and Promotion Center at Asan Medical Center from January 2007 to December 2011. Among these, 7,129 individuals (76.9%) agreed to participate in this study. Possible risks associated with CCTA were explained and informed consent was obtained. Exclusion criteria include subjects with (1) a previous history of angina or myocardial infarction; (2) abnormal rest electrocardiographic results, i.e., pathological Q waves, ischemic ST segments or T wave changes, or left bundle-branch blocks; (3) insufficient medical records; (4) structural heart diseases; (5) a prior history of open heart surgery or percutaneous coronary intervention; (6) a previous cardiac procedure; or (7) renal insufficiency (creatinine >1.5 mg/dL). Finally, 6,434 subjects were enrolled (Supplementary Fig. 1). The study was approved by the local Institutional Review Board of the Asan Medical Center, Seoul, Korea (approval number: 2016-1068).

The demographic information was collected from a database maintained by the Health Screening and Promotion Center at the Asan Medical Center. Medical history including angina, myocardial infarction, stroke, structural heart disease, open heart surgery, percutaneous coronary intervention, previous cardiac procedures, diabetes mellitus, hypertension, or hyperlipidemia; a family history of CAD; and smoking status was taken from the responses in the systemized self-report questionnaire issued prior to the general health examination. A family history of CAD was defined as having a first-degree relative of any age on the self-report questionnaire [13]. Patients with diabetes was defined as (1) subjects with a self-reported history of diabetes and/or treatment with dietary modification, or use of anti-diabetic medication on the systemized questionnaire and (2) those with a fasting plasma glucose (FPG) ≥126 mg/dL or a glycosylated hemoglobin (HbA1c) level ≥6.5% [13,14]. By their diabetes control, participants with diabetes were classified as controlled diabetes (diabetes with HbA1c <7%) and uncontrolled diabetes (diabetes with HbA1c ≥7%), respectively [1]. Those without diabetes were categorized as normal group. Hypertension was defined as a blood pressure ≥140/90 mm Hg or a self-reported history of hypertension and/or use of anti-hypertensive medication. Hyperlipidemia was also defined as total cholesterol ≥240 mg/dL a self-reported history of hyperlipidemia and/or use of anti-hyperlipidemic treatment.

Height and weight were obtained while subjects wore light clothing without shoes. The body mass index (BMI) was calculated as weight in kilograms divided by the square of the height in meters. Obesity was defined as a BMI of ≥25.0 kg/m² using the cut-offs of BMI for Asian population. The waist circumference (cm) was measured midway between the costal margin and the iliac crest at the end of a normal expiration. The blood pressure was measured on the right arm after a rest of ≥5 minutes using an automatic manometer with an appropriate cuff size. Left ventricular ejection fraction was measured by echocardiography. After overnight fasting, early morning blood samples were drawn from the antecubital vein into vacuum tubes and subsequently analyzed in the central, certified laboratory of the Asan Medical Center. Measurements included the concentrations of FPG, uric acid, creatinine, high-sensitivity C-reactive protein (hs-CRP), and several lipid parameters. Fasting total cholesterol, high density lipoprotein cholesterol, low density lipoprotein cholesterol (LDL-C), triglyceride, uric acid and creatinine were measured by an enzymatic colorimetric method using a Toshiba 200FR Neo (Toshiba Medical System Co. Ltd., Tokyo, Japan). FPG were measured by an enzymatic colorimetric method using a Toshiba 200 FR autoanalyzer (Toshiba). Ion-exchange high-performance liquid chromatography (Bio-Rad Laboratories Inc., Hercules, CA, USA) was used to measure HbA1c levels. Serum hs-CRP level was measured according to a high-sensitivity assay by using a latex particle-enhanced immunoturbidometric assay (Roche Diagnostics, Mannheim, Germany). All enzyme activities were measured at 37℃ [15].

CCTA was conducted using either single-source 64-slice CT (LightSpeed VCT; GE, Milwaukee, WI, USA) or dual-source CT (Somatom Definition; Siemens, Erlangen, Germany). Subjects with no contraindication to β-adrenergic blocking agents and with an initial heart rate greater than 65 beats per minute received an oral dose of 2.5 mg bisoprolol (Concor; Merck, Darmstadt, Germany) 1 hour before the CT examination. CT scanning was performed in the prospective electrocardiogram (ECG)-triggering mode or the retrospective ECG-gating mode with ECG-based tube current modulation. Two puffs (2.5 mg) of isosorbidedinitrate (Isoket spray; Schwarz Pharma, Monheim, Germany) were sprayed into the patient's oral cavity before contrast injection. During CCTA acquisition, 60 to 80 mL of iodinated contrast (Iomeron 400; Bracco, Milan, Italy) was injected at 4 mL/second, followed by a 40 mL saline flush. A region of interest was placed in the ascending aorta, and image acquisition was automatically initiated once a selected threshold (100 HU) had been reached using bolus tracking. A standard scanning protocol was used, and the tube voltage and tube current-time product were adjusted according to the patient's body size as follows: 100 or 120 kVp tube voltage; 240 to 400 mAs per rotation (dual-source CT); and 400 to 800 mA (64-slice CT) tube current [12].

All CCTA scans were analyzed using a dedicated workstation (Advantage Workstation, GE; or Volume Wizard, Siemens) by experienced cardiovascular radiologists (D.H.Y., J.W.K., and T.H.L.). According to the guidelines of the Society of Cardiovascular Computed Tomography, a 16-segment coronary artery tree model was used [16]. A coronary artery calcium score was measured as described, with categorized by scores of 0, 1 to 10, 11 to 100, 101 to 400, and >400 [17]. Plaques were defined as structures >1 mm² within and/or adjacent to the vessel lumen, which could be clearly distinguished from the lumen and surrounding pericardial tissue. Plaques containing calcified tissue involving more than 50% of the plaque area (density >130 HU) were classified as calcified, plaques with <50% calcium were classified as mixed, and plaques without calcium were classified as non-calcified lesions [18]. The contrast-enhanced portion of the coronary lumen was semi-automatically traced at the site of maximal stenosis and compared with the mean value of the proximal and distal reference sites [19]. Stenosis ≥50% was defined as significant. The overall plaque burden was determined from coronary artery plaque scores calculated from modified Duke prognostic scores, segment stenosis scores, and segment involvement scores, as described [20]. In addition, high-risk CAD was defined as at least 2-vessel coronary disease with proximal left anterior descending (LAD) artery involvement, 3-vessel disease, or left main (LM) disease [21].

Follow-up clinical data were obtained by a review of medical records or telephone interviews using trained personnel through to the end of June 2017. A cardiac event was defined as a composite of all-cause death, myocardial infarction, unstable angina requiring hospitalization, or coronary revascularization. The diagnosis of myocardial infarction was based on the presence of new Q waves in at least two contiguous leads, or an elevation of creatine kinase or its myocardial band isoenzyme to at least three times the upper limit of the normal range at follow-up. Revascularization was performed if there was a stenosis of at least 50% of the diameter noted on invasive coronary angiography with a positive stress test result or if there was a stenosis of at least 70% seen on invasive coronary angiography [22].

Categorical variables are expressed as frequencies with percentages, and continuous variables as the mean and standard deviation. Between-group comparisons were performed by using the Pearson's chi-square test or Fisher's exact test for categorical variables, and by using the one-way analysis of variance or Kruskal-Wallis test for numerical variables, as appropriate. Logistic regression analyses were performed to evaluate the independent relationships between the diabetes control and subclinical coronary atherosclerosis on CCTA. For the multivariable analyses, we adjusted clinically and statistically important covariates such as age, sex, obesity, hypertension, hyperlipidemia, current smoking, a family history of CAD, and hs-CRP. Unadjusted and adjusted odds ratios with 95% confidence intervals for the logistic regression were calculated. Survival curves were assessed using the Kaplan-Meier method and compared using the log-rank test. All reported P values are two sided, and P values of <0.05 were considered statistically significant. Data manipulation and statistical analyses were conducted using the SPSS software version 18.0 (SPSS Inc., Chicago, IL, USA).

The mean age of study participants was 53.7±7.6 years and 4,694 (73.0%) were male. Of them, 5,319, 747, and 368 participants were categorized as normal, controlled diabetes, and uncontrolled diabetes, respectively. The baseline characteristics of the study population according to the diabetes control are listed in Table 1, Supplementary Table 1. In controlled diabetes, the mean FPG and HbA1c were 121.8±17.7 mg/dL and 6.2%±0.5%. On the other hand, in uncontrolled diabetes, the mean FPG and HbA1c were 158.6±40.8 mg/dL and 8.1%±1.2%.

CCTA findings according to the diabetes control are shown (Table 2, Supplementary Table 2). A total of 236 coronary segments (0.2%) were not interpretable due to artifacts. There was significant difference in CACS according to the diabetes control (P<0.001). The prevalence of any atherosclerotic, calcified, non-calcified, or mixed plaque increased with the diabetes control (P for all<0.001). Plaque burden scores such as segment involvement score, segment stenosis score, and modified Duke prognostic score also increased with the diabetes control (P for all<0.001). Of study participants, 494 (7.7%) had significant coronary arteries stenosis (≥50% diameter stenosis) in at least one coronary artery on CCTA. The prevalence of significant stenosis, multi-vessel disease, significant stenosis in the LM or proximal LAD artery, high-risk CAD significantly increased according to the diabetes control (P for all<0.001).

Controlled diabetic individuals had more calcified and mixed plaques than the normal individuals. However, there were no statistically significant differences in the adjusted odds ratios for any atherosclerotic and non-calcified plaque, significant stenosis, multi-vessel disease, significant stenosis in the LM or proximal LAD, and high-risk CAD between the normal and controlled diabetic individuals. On the other hand, uncontrolled diabetic individuals had significantly associated with any subclinical coronary atherosclerosis compared with normal individuals (P for all<0.05) (Table 3).

Supplementary Fig. 2 shows the study flow. During the follow-up period (median 5.4 years [interquartile range, 4.4 to 6.4 years]), a total of 209 cardiac events occurred in 193 patients: 67 all-cause deaths, five non-fatal myocardial infarctions, 15 acute coronary syndrome requiring hospitalization, and 122 coronary revascularizations (Table 4, Supplementary Table 3). Fig. 1 shows Kaplan-Meier survival curves according to the diabetes control. The 6-year cardiac event-free survival rates were 97.2%±0.2% in normal individuals, 96.4%±0.7% in controlled diabetic individuals, and 93.9%±1.3% in uncontrolled individuals (log-rank P<0.001). There was no significant difference in cardiac events between normal and controlled diabetic individuals (P=0.365). However, uncontrolled diabetes was associated with an increased risk of cardiac events compared with normal individuals (P<0.001) and controlled diabetic individuals (P=0.023).