The Mapo study is a community-based cohort study that was initiated to investigate the prevalence, correlation, and progression of subclinical CVD. The study cohort comprises residents from Mapo-gu, a metropolitan district within Seoul, Korea, who visited a public health center for a routine health check-up. In the Mapo cohort study, men aged ≥45 years and women aged ≥55 years were enrolled and screened for subclinical atherosclerosis using CACS.

In the present study, we enrolled 456 individuals (285 men, 66.5±5.8 years) who were registered in Mapo-gu cohort between January 2011 and March 2013. Subjects were divided into two groups according to CACS. Selection was determined using an arbitrary cut-off CACS of 100 AU. We selected 228 individuals with a CACS of more than 100 AU and 228 age- and gender-matched individuals with a CACS of less than 100 AU. The Institutional Review Board at Yonsei University College of Medicine approved the study (4-2014-0754), and all participants gave informed consent.

All individuals underwent a baseline evaluation including an initial standardized questionnaire. Height, weight, body mass index (BMI), anthropometric data, and BP were measured. Total, low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose levels were measured from blood samples obtained after a 12-hour fast. hsCRP was measured as previously described.11 Smoking was defined as current or former smoker. Dyslipidemia was defined as any of the following: high total cholesterol (>200 mg/dL), high LDL-cholesterol (>160 mg/dL), low HDL-cholesterol (<40 mg/dL for men, <50 mg/dL for women), and/or high triglycerides (>150 mg/dL), or current use of lipid-lowering therapy. Participants were considered to have diabetes mellitus (DM) if they reported a history of DM and/or were receiving anti-diabetic treatment or had a fasting plasma glucose level of at least 126 mg/dL. Hypertension was defined as a self-reported history of hypertension, a history of antihypertensive medication use, and/or a BP of 140/90 mm Hg or more at the time of the visit.

All examinations were performed using a 320-row CT system (Aquilion ONE; Toshiba Medical Systems, Tokyo, Japan) with patients in the supine position, and were acquired during a single breath hold, which allows for image reconstruction at a single cardiac phase. Dual scanograms were used for planning the examination and determining the anatomical range to be covered. A non-enhanced prospective electrocardiogram-gated scan was performed to measure CACS with the following parameters: rotation time of 275 ms, slice width of 3.0 mm, tube voltage of 120 kV, and tube current using the Sure Exposure 3D® (Tochigi, Japan) with an automatic exposure control system. CACS was assessed on a postprocessing workstation (Vitrea fX, version 6.4; Vital Images, Minnetonka, MN, USA) using calcium score analysis software (VScore; Vital Images). Coronary scans were used to identify each focus of calcification with a minimal density of 130 Hounsfield units and a minimal voxel size of 1.03 mm³ along the course of the coronary artery.12 Agatston calcium scores were calculated to quantify the extent of coronary artery calcification.13

sST2 concentrations were measured in blood samples (serum) collected after an 12-hour fast, and immediately centrifuged for storage at -80℃. sST2 concentrations were measured using a high-sensitivity sandwich immunoassay (Presage ST2; Critical Diagnostics, San Diego, CA, USA). The performance characteristics of this assay have been recently described.14 The assay has a lower detection limit of 2 ng/mL, and meets acceptable analytical performance criteria, exhibiting an average coefficient of variation of 2.85% for intra-assay and 6.74% for inter-assay variation in our study.

Continuous variables are summarized as a mean±standard deviation, and categorical variables are summarized as a percentage of the group total. Variables not in the normal distribution were log transformed for analysis. Continuous variables were compared using independent t-tests or Mann-Whitney U test for non-normally distributed variables. Pearson's correlation analysis or Spearman test for non-normally distributed variables was used to determine simple correlation between continuous variables. A multivariate logistic regression analysis was conducted using all variables significantly correlated with CACS. The net reclassification improvement (NRI) and integrated discrimination index (IDI) for sST2 over hsCRP (and vice versa) was calculated, and the area under the receiver operating characteristic (ROC) curves (AUC) of sST2 and hsCRP were compared using STATA version 13.0 (Stata Corp LP, College Station, TX, USA). A two-tailed p value <0.05 was considered statistically significant. All other statistical analyses were performed with SPSS version 21.0 software (SPSS/IBM Corp., Chicago, IL, USA).

Baseline characteristics of study participants are summarized in Table 1. The mean age, BMI, and systolic BP (SBP) were 66±6 years old, 25.2±2.9 kg/m², and 124.2±14.9 mm Hg, respectively. Serum levels of sST2 ranged from 10.7 to 197.6 (median 27.9, interquartile range 22.2–35.4) ng/mL. Mean CACS was 237.2±465.0 AU, and there were 99 (21.7%) subjects with a CACS ≥300 AU. Patients with a CACS ≥300 AU were considered as high risk. There was a significant difference in hsCRP (1.27±1.95 mg/L vs. 2.90±9.00 mg/L for high risk patients, p=0.002) (Fig. 1A). No difference was observed in sST2 in this same high risk population (29.7±15.2 ng/mL vs. 32.6±12.2 ng/mL for high risk CACS, p=0.080) (Fig. 1B) between high risk and non-high risk CACS group. Women had a lower waist circumference, SBP, smoking history, fasting glucose, gamma glutamyltransferase (GGT), and uric acid levels. Women had higher dyslipidemia, statin use, total cholesterol, and HDL-cholesterol level. We observed a significant gender difference in serum sST2 levels (33.2±16.8 ng/mL vs. 25.4±8.2 ng/mL, for men vs. women, p<0.001). However, no difference in CACS and hsCRP were observed between genders.

Data regarding the association between sST2 or hsCRP levels and the clinical and laboratory parameters are summarized in Table 2. The LDL-cholesterol, alanine transaminase (ALT), creatinine, GGT, uric acid, white blood cell count (WBC), and CACS correlated with both sST2 and hsCRP levels. Total-cholesterol, glucose, aspartate transaminase (AST), blood urea nitrogen (BUN), serum creatinine, phosphate and hemoglobin correlated with only sST2 level, whereas only waist conference, HDL-cholesterol, and CACS correlated with only hsCRP level. Serum log sST2 level significantly correlated with log hsCRP level (r=0.128, p=0.006).

According to current AHA/ACC guidelines, we defined patients with a CACS >300 AU to be at high risk. In univariate regression analysis, both log sST2 [odd ratio (OR) 5.900, 95% confidence interval (CI) 1.471–23.668, p=0.012] and log hsCRP (OR 2.812, 95% CI 1.600–4.942, p<0.001) were significantly related to high CACS group. We compared the risk-stratifying power of sST2 and hsCRP in high risk patients using three statistical methods. First, using multivariate logistic regression, we compared the association of sST2/hsCRP with being classified as high risk (Table 3). After adjusting for other risk factors such as age, gender and current smoking status, only log hsCRP was significantly associated with high risk CACS (OR 2.450, 95% CI 1.1336–4.492, p=0.004). When we inserted sST2/hsCRP into the adjusted model as categorical variables (lower vs. higher tertile), only hsCRP (OR 1.858, 95% CI 1.051–3.284, p=0.033) correlated with a high risk CACS, in line with the multivariate regression analysis. Second, we compared AUC in ROC for high risk CACS. In this analysis, we observed no significant differences between sST2 (AUC 0.596, 95% CI 0.520–0.671) and hsCRP (AUC 0.580, 95% CI 0.514–0.647, p for difference=0.64). Third, to determine if any additive prognostic value for predicting high risk CACS was gained by combining sST2 with hsCRP, we compared the NRI for sST2 over hsCRP (Table 4). The NRI for sST2 over hsCRP was not statistically significant (continuous NRI 0.212, 95% CI -0.255–0.453, 20% & 25% risk cut-off NRI 0.0790, p=0.140, IDI 0.002, p=0.269). Meanwhile, however, the NRI for hsCRP over sST2 was significantly different (continuous NRI 0.238, 95% CI 0.001–0.474, 20% & 25% risk cut-off NRI 0.139, p=0.055, IDI 0.022, p=0.035). Collectively, these data indicated that hsCRP provides superior discrimination and risk reclassification for high risk CACS, compared to sST2. In addition, we chose a different cutoff value of CACS (100 or 200 AU) for defining high risk CACS, rather than 300 AU, although we could not find superiority for sST2 over hsCRP for predicting high risk CACS group (data not shown).

Given the observed gender differences in sST2 levels, we performed a subgroup analysis according to sex. In correlation analysis, in men, log sST2 level was associated with glucose (r=0.140, p=0.018), AST (r=0.209, p<0.001), ALT (r=0.185, p=0.002), GGT (r=0.193, p=0.001), WBC (r=0.232, p<0.001), and log hsCRP (r=0.154, p=0.009). These associations were not observed women, in whom log sST2 level was correlated with SBP (r=-0.164, p=0.032) and BUN (r=0.201, p=0.008). We did not find any statistical significance in comparing AUC, NRI, and IDI in either men or women (data not shown).