Computed Tomography Angiography Images of Coronary Artery Stenosis Provide a Better Prediction of Risk Than Traditional Risk Factors in Asymptomatic Individuals With Type 2 Diabetes: A Long-term Study of Clinical Outcomes

Study Population

From January 2006 to December 2010, 933 asymptomatic patients with T2DM who were >30 years of age and had undergone CCTA were prospectively enrolled in the CRONOS-ADM (Coronary CT Angiography Evaluation for Clinical Outcomes in Asymptomatic Patients with Type 2 Diabetes Mellitus) registry to evaluate CCTA findings and the coronary artery calcium score (CACS) in the divisions of endocrinology and cardiology at two major cardiac centers in Korea (Seoul St. Mary’s Hospital, Seoul, Republic of Korea; and St. Vincent’s Hospital, Suwon, Republic of Korea). This registry included demographic characteristics; clinical, laboratory, CACS, and CCTA findings; and 7-year long-term clinical outcomes. A flowchart of the study population is depicted in Supplementary Fig. 1. A total of 390 patients refused to participate. A total of 67 patients were excluded for the following reasons: patients with type 1 diabetes, angina, or angina-equivalent symptoms according to the Rose questionnaire (12); patients receiving antiangina medications and patients who had a history of myocardial infarction (MI); and patients with previous coronary revascularization (either by percutaneous coronary intervention or bypass). Patients with an estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m² were also excluded. The diagnosis of T2DM was made using the 2010 criteria of the American Diabetes Association of fasting glucose ≥126 mg/dL, glycated hemoglobin (HbA_1c) ≥6.5% or ≥48 mmol/mol, and/or postchallenge glucose (glucose at 2 h after a 75-g oral glucose load) ≥200 mg/dL (13). A more detailed description of this study population is presented in our recent articles (14,15). No corporation was involved in the design, performance, or analysis of the study. This prospective observational study was approved by the institutional review board of our institution and performed in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (16). All patients provided written informed consent.

Scanning Protocol and Image Reconstruction

CCTA was performed using either a 64-slice multidetector computed tomography (MDCT) scanner (Light Speed VCT 64; GE Healthcare, Milwaukee, WI) or a dual-source computed tomography (DSCT) scanner (Somatom Definition; Siemens Healthcare, Forchheim, Germany). For the 64-slice MDCT scanning protocol, we used a slice collimation of 64 × 0.625 mm, a gantry rotation time of 350 ms, a pitch of 0.2, a tube voltage of 100–120 kV (depending on the patient’s BMI), and a tube current of 600 mA. For the DSCT scan protocol, slice collimation was performed at 2 × 32 × 0.6 mm using a z-flying focal spot, a gantry rotation time of 330 ms, a pitch of 0.2–0.5, a tube voltage of 100–120 kVp (depending on the patient’s BMI), and a reference tube current of 320 mA with an electrocardiogram-controlled tube current modulation. In each patient, 80 mL of the iodinated contrast agent iopromide (Ultravist 370; Schering AG, Berlin, Germany) or iomeprol (Iomeron 350; Bracco, Milano, Italy) was injected at a flow rate of 5 mL/s along with 50 mL of contrast mixture (15% contrast medium and 85% saline solution) at the same rate, which was controlled by bolus tracking in the ascending aorta (signal attenuation threshold 120 Hounsfield units [HU]). The scan delay was 7 s. In the absence of contraindications, each patient with a heart rate >80 bpm received an intravenous β-blocker, indenol or esmolol (Jeil Brevibloc; Jeil Pharma Co., Ltd., Seoul, Republic of Korea), 1 h before the scan, and a 0.3 mg sublingual dose of nitroglycerin was administered immediately before the scan. The estimated radiation dose ranged from 5 to 14 mSv. Images were reconstructed immediately after completing the scan and transferred to a computer workstation (MDCT: advantage Windows 4.3, GE Healthcare; DSCT: Syngo Multimodality Workplace, version 2008, Siemens Healthcare) for postprocessing, including axial and multiplanar reformatting, maximum intensity projection, and short-axis and cross-sectional views.

CCTA Analysis

All scans were analyzed by two radiologists with experience in interpreting several thousand CCTA scans. In accordance with the guidelines of the Society of Cardiovascular Computed Tomography, coronary segments were visually scored for the presence of coronary plaques using a 16-segment coronary artery model in an intent-to-diagnose manner (17). Segments were included in the analysis if the diameter was >1.5 mm. The severity of luminal diameter stenosis was scored as none (0% luminal stenosis), nonobstructive (plaques with a lumen narrowing <50%), obstructive (plaques with maximum stenosis ≥50%), or significantly obstructive (plaques with maximum stenosis ≥70%). CAD was diagnosed based on the maximum intraluminal stenosis in any of the segments of the major epicardial coronary arteries at the ≥50% stenosis threshold. Obstructive CAD in the branches of arteries was considered as corresponding to the major epicardial coronary arteries. The number of diseased vessels was assigned as one, two, three, or left main (LM) coronary artery vessels. The extent and severity of the CAD burden were measured using several CCTA scores (18,19). The CACS, atheroma burden obstructive score (ABOS), segment involvement score (SIS), and segment stenosis score (SSS) were included. CACS was assessed using dedicated software (Smartscore version 3.5, GE Healthcare; or Aquarius 3D Workstation, TeraRecon, San Mateo, CA). Coronary artery calcium levels were identified as the density exceeding the threshold of 130 HU (20). An overall Agatston score was recorded for each patient. The ABOS was defined as the number of plaques with >50% stenosis in the entire coronary artery tree. The SIS was calculated as the total number of coronary artery segments that exhibited plaques, regardless of the degree of luminal stenosis within each segment (minimum = 0; maximum = 16). The SSS was used to measure the overall extent of coronary artery plaques. Each individual coronary segment was graded as having no to severe plaques (i.e., scores from 0 to 3) based on the extent of obstruction of the luminal diameter of the coronary artery. The extent scores of all 16 individual segments were summed to yield a total score that could range from 0 to 48 (19). Plaques without any calcium were defined as noncalcified plaques (NCPs), and others were defined as mixed or calcified plaque (CPs), depending on consistency (using attenuation grade, i.e., mixed plaque ≤130 HU or CP >130 HU with an area ≥1.0 mm², respectively) (20,21).

Study End Point and Follow-up

The primary end point was a composite of all-cause mortality, nonfatal MI, and late coronary revascularization (≥90 days). Late coronary revascularization was considered to be ischemia driven if the diameter of the stenotic lesion was ≥50%, according to a quantitative analysis of the electrocardiographic changes observed at rest or a positive functional study in the distribution of the target lesion, or ≥70% with recurrent symptoms at ≥90 days after stent implantation. All clinical outcomes of interest were confirmed by source documents and were centrally adjudicated by a clinical events committee at the Cardiovascular Center of Seoul St. Mary’s Hospital, which consisted of an independent group of clinicians whose members were unaware of the patient’s status. Information related to censored survival data (death or survival) and cause of death (cardiac or noncardiac death) was obtained from the Korean Office of Statistics with a unique personal identification number to validate the complete follow-up data.

Statistical Analysis

Baseline and biochemical characteristics were summarized as the means ± SD for continuous variables and as absolute numbers and percentages for discrete variables. Differences in continuous variables between groups were evaluated using unpaired t tests or Mann-Whitney rank sum tests. Differences in discrete variables between groups were analyzed using χ² or Fisher exact tests. Cumulative event rates of CACS, CCTA-diagnosed obstructive CAD, ABOS, SIS, and SSS were calculated using a Kaplan-Meier estimator and compared using the log-rank statistic. A Cox proportional hazards model was used to calculate hazard ratios (HRs) with 95% CIs to describe the relationships between the various measures of CCTA-diagnosed CAD and composites of MACE, including all-cause mortality, nonfatal MI, and late revascularization. Each measure of CCTA-diagnosed CAD was adjusted for age, male sex, hypertension, hyperlipidemia, smoking, eGFR, and HbA_1c and was analyzed using SPSS 20.0 (SPSS-PC Inc., Chicago, IL). Statistical significance was accepted at P < 0.05 using a two-tailed test.

The discriminative ability of the models was assessed using Harrell’s C-index, which is analogous to the area under the receiver operator characteristic curve and was applied to composite MACE data The IDI was used to express the absolute improvement in the mean discrimination slope and the probability of discrimination between the base model (model A) and the new models (models B to G), whereas the relative IDI illustrated the difference in the separation of MACE and nonevents over the separation in the base model. We selected the conventional risk model based on the risk factors included in the 10-year atherosclerotic cardiovascular disease risk algorithm, which was published in the 2013 American College of Cardiology/American Heart Association guidelines and referenced in the Framingham study (22–24). The risk factors were matched with the variables adjusted in the Cox regression model. Larger positive IDI and relative IDI values indicate greater improvements in model discrimination. Improvements in subject risk reclassification were further assessed using NRI and were applied to the MACE data. We computed the two versions of the NRI: a category-based and a category-free NRI. The category-based NRI followed typical categories used for determining MACE risk (determined after 1 year as <5%, 5% to <10%, and ≥20%). We used χ² values and model Akaike information criterion (AIC) values as the parameters of model fit. The χ² test, AIC, C-index, IDI, and NRI were analyzed using SAS version 9.4 (SAS Institute, Cary, NC).