We examined the data of 372 subjects who underwent a health check-up from November 2011 to February 2013 at Severance Hospital, Yonsei University College of Medicine, Seoul, Korea. Subjects who underwent both laboratory testing of biomarkers for T2DM and CAP were included. The following exclusion criteria were applied.

1) Subjects with chronic liver disease of any etiology other than NAFLD [positivity for hepatitis B surface antigen (HBsAg; n=4) or hepatitis C antibody (anti-HCV; n=2)]

2) Subjects with potential risk for developing secondary hepatic steatosis [excessive (>20 g/day) alcohol consumption (n=20), medications (n=4; steroid, n=3; tamoxifen, n=1)]

3) Unreliable liver stiffness value (LSV) or LSV measurement failure (n=2)

After excluding 32 of 372 participants, a total of 340 subjects were included in the final analysis. This study was approved by the independent Institutional Review Board of Severance Hospital of Yonsei University College of Medicine and conformed to the ethical guidelines of the 1975 Helsinki declaration. All patients signed a written informed consent.

Anthropometric data, such as body mass index (BMI) and systolic and diastolic blood pressure, were measured. A questionnaire was given to all participants to investigate family history of diabetes mellitus, frequency and amount of alcohol consumption, and smoking status (never, past, current). T2DM was defined as fasting glucose ≥126 mg/dL, and/or 2-hour post-loading glucose ≥200 mg/dL, and/or hemoglobin A1c (HbA1c) ≥6.5%, or history of glucose-lowering medications. The prediabetes group included subjects with impaired fasting glycemia (fasting glucose 100 to 125 mg/dL) and/or impaired glucose tolerance [75 g oral glucose tolerance test (OGTT), 140 to 199 mg/dL], and/or HbA1c 5.7 to 6.4%.21 Hypertension was defined as taking anti-hypertensive medication or blood pressure higher than 140/90 mm Hg at initial examination. Metabolic syndrome was defined according to the criteria established by the National Cholesterol Education Program Adult Treatment Panel III, except for the determination of abdominal obesity by waist circumference. We used a waist circumference cutoff level based on the report by the International Diabetes Institute/Western Pacific World Health Organization/International Obesity Task Force. Metabolic syndrome was diagnosed if three of the five following features were satisfied: 1) waist circumference >90 cm in men and >80 cm in women; 2) triglycerides ≥150 mg/dL (1.7 mmol/L); 3) high-density lipoprotein (HDL) cholesterol <40 mg/dL (1.03 mmol/L) in men and <50 mg/dL (1.29 mmol/L) in women; 4) blood pressure ≥130/85 mm Hg; and 5) fasting plasma glucose ≥100 mg/dL (5.6 mmol/L).

The laboratory tests included serological tests for HBsAg, anti-HCV, aspartate aminotransferase, alanine aminotransferase (ALT), total bilirubin, gamma glutamyltranspeptidase (GGT), and lipid profiles (triglycerides and total HDL, and low-density lipoprotein cholesterol). Biomarkers related to T2DM included fasting glucose, fasting insulin, fasting C-peptide, HbA1c, glycoalbumin, homeostasis model assessment of insulin resistance of insulin resistance (HOMA-IR), homeostasis model assessment of beta cell function (HOMA-β), and high sensitive C-reactive protein (hs-CRP). HOMA-IR was calculated as fasting serum insulin (IU/dL)×fasting serum glucose (mg/dL)/22.5, and HOMA-β as [20×fasting serum insulin (IU/dL)]/[fasting serum glucose (mg/dL)-3.5].

All subjects underwent abdominal ultrasonography and transient elastography (FibroScan®, EchoSens, Paris, France). Abdominal ultrasonography was performed by two experienced hepatologists. Liver steatosis by ultrasonography was assessed as absent, mild, moderate, or severe on the basis of the liver-kidney difference in echo amplitude caused by abnormally intense echoes arising from the hepatic parenchyma due to liver fat, echo attenuation by fat into the deep portion of liver, and blurring of the blood vessel structures in the liver.22 Two hepatologists who performed ultrasonography independently reviewed all ultrasonographic findings. In cases of discrepancy, a final consensus opinion was adopted.

CAP and liver stiffness measurement were performed on the right lobe of the liver through the intercostal spaces with subjects lying in the dorsal decubitus position with the right arm in maximal abduction by one experienced technician who was blind to the subjects' clinical data. The principle of CAP calculation has been described previously.23 Briefly, CAP measures ultrasonic attenuations by the liver fat at 3.5 MHz using signals acquired by the FibroScan®, and is simultaneously calculated with LSV using the same signals. CAP value and LSV are expressed as dB/m and kPa, respectively. The median values of successful measurements were selected as the representative CAP values and LSV. The interquartile range (IQR) was defined as the interval containing 50% of valid measurements between the 25th and 75th percentiles. As an indicator of variability, the ratio of the IQR of CAP values and LSV to the median (IQR/Mcap and IQR/M, respectively) was calculated. In this study, only measurements with at least 10 valid shots and a success rate of at least 60% were considered reliable and were used for statistical analysis.24 A CAP value of 250 dB/m was selected as the cutoff for presence of steatosis and a CAP value of 300 dB/m was the cutoff for moderate to severe steatosis.20

Data are expressed as means±standard deviation, median (range), or n (%), as appropriate. Categorical data were compared using chi-square test and continuous data were compared with independent t-test. Differences of median CAP values according to diabetic group were analyzed with the Kruskal-Wallis test. Binary logistic regression analysis was used to identify the factors associated with T2DM. Correlations between CAP value and biomarkers related to T2DM were evaluated by Spearman correlation coefficients. A one-way analysis of variance was performed to compare HOMA-IR and fasting C-peptide according to CAP groups (CAP value <250 vs. CAP value 250–300 vs. CAP value ≥300). All statistical analyses were performed using SPSS 19.0 (SPSS Inc., Chicago, IL, USA) and two-sided p-values <0.05 were considered significant.

Among the 340 participants (191 men and 139 women), 66 (19.4%) subjects had T2DM. Among 274 (80.6%) subjects without T2DM, 202 (59.4%) had pre-diabetes and 72 (21.2%) showed normal glucose tolerance (NGT). The baseline characteristics of study subjects according to patients with and without T2DM are shown in Table 1. According to the glucose tolerance status, patients with T2DM rather than patients with pre-diabetes or NGT were more likely to be older and male (both p<0.05). These subjects also had higher BMI, waist circumference, and triglyceride level, as well as elevated biomarkers related to T2DM (fasting glucose, fasting insulin, fasting C-peptide, HbA1c, glycoalbumin, and HOMA-IR) (all p<0.05). CAP values and LSV were also significantly higher in patients with T2DM than those with pre-diabetes or NGT (CAP values, 264.5±56.1 dB/m vs. 246.9±47.4 dB/m vs. 227.3±47.5 dB/m, p<0.001; LSV, 5.5±3.5 kPa vs. 4.6±1.6 kPa vs. 4.4±1.7 kPa, p=0.003).

The proportion of subjects with presence of NAFLD (CAP value ≥250 dB/m) increased according to the glucose tolerance status [31.9% (n=23) in NGT; 47.0% (n=95) in pre-diabetes, 57.6% (n=38) in T2DM; NGT vs. pre-diabetes, p=0.027; NGT vs. T2DM, p=0.003; pre-diabetes vs. T2DM, p=0.157]. The proportion of subjects with moderate to severe NAFLD (CAP value ≥300 dB/m) also increased according to the glucose tolerance status (9.7% in NGT, 15.3% in pre-diabetes, 33.3% in T2DM; NGT vs. pre-diabetes, p=0.321; NGT vs. T2DM, p=0.001; prediabetes vs. T2DM, p=0.02) (Fig. 1A). In addition, the proportion of subjects with NAFLD diagnosed by ultrasonography differed according to the glucose tolerance status (27.8% in NGT, 35.6% in pre-diabetes, 54.5% in T2DM; NGT vs. pre-diabetes, p=0.425; NGT vs. T2DM, p=0.024; pre-diabetes vs. T2DM, p=0.057). The proportion of subjects with moderate to severe NAFLD diagnosed by ultrasonography also increased accord ing to the glucose tolerance status (6.9% in NGT, 8.4% in prediabetes, 22.7% in T2DM; NGT vs. pre-diabetes, p=0.406; NGT vs. T2DM, p=0.031; pre-diabetes vs. T2DM, p=0.044) (Fig. 1B).

Subjects with T2DM had significantly higher median CAP values than those with NGT (265 dB/m vs. 231 dB/m, p<0.001) or pre-diabetes (265 dB/m vs. 245 dB/m, p=0.003). The median CAP value was significantly higher in the pre-diabetes group than in the NGT group (245 dB/m vs. 231 dB/m, p=0.001) (Fig. 2A). In subjects with NAFLD (CAP value ≥250 dB/m), the median CAP values increased according to the glucose tolerance status: 259 dB/m in NGT, 278 dB/m in pre-diabetes, and 304 dB/m in the T2DM group (comparison between groups; T2DM vs. NGT, p=0.006; T2DM vs. pre-diabetes, p=0.026; prediabetes vs. NGT, p=0.077) (Fig. 2B). In subjects with NAFLD (diagnosed by ultrasonography), the median CAP values also increased according to the glucose tolerance status: 265 dB/m in NGT, 278 dB/m in pre-diabetes, and 302 dB/m in the T2DM group (comparison between groups; T2DM vs. NGT, p=0.042; T2DM vs. pre-diabetes, p=0.047; pre-diabetes vs. NGT, p=0.484) (Fig. 2C).

In univariate binary logistic regression analyses, T2DM was significantly associated with fasting glucose, fasting insulin, fasting C-peptide, HbA1c, glycoalbumin, and HOMA-IR (all p<0.05). In addition, among all variables described in Table 1, T2DM was associated with age, male gender, BMI, waist circumference, triglyceride, HDL-cholesterol, LSV, and CAP value (all p<0.05).

Multivariable logistic regression analysis was performed to control for other established risk factors for T2DM and to determine whether CAP value was independently associated with T2DM (Table 2). After adjusting for age, gender, BMI, waist circumference, triglyceride, HDL-cholesterol, and LSV, subjects with CAP value ≥300 dB/m were found to have a 2.8-fold higher risk of having T2DM than those with CAP value <250 dB/m [p=0.017; odds ratio (OR)=2.83; 95% confidence interval (CI), 1.21–6.64]. Subjects with a CAP value of 250–300 dB/m had a very similar risk of developing T2DM as those with CAP value <250 dB/m (p=0.374; OR=1.10; 95% CI, 0.59–2.16). As expected, age was also closely related to the risk of T2DM (p<0.001; OR=1.07; 95% CI, 1.04–1.11). However, gender, BMI, waist circumference, triglycerides, HDL-cholesterol, and LSV showed no significant associations with T2DM.

Biomarkers related to T2DM and other factors were measured in all subjects (n=340) to investigate their correlation with CAP value (Table 3). Among biomarkers related to T2DM, CAP value showed significant correlation; HOMA-IR (ρ=0.407, p<0.001) and fasting C-peptide (ρ=0.402, p<0.001). In addition, CAP value showed a correlation with fasting insulin (ρ=0.395, p<0.001), fasting glucose (ρ=0.379, p<0.001), HbA1c (ρ=0.345, p<0.001), hs-CRP (ρ=0.277, p=0.001), and HOMA-β (ρ=0.132, p<0.001). Among other factors, CAP value showed significant correlation with BMI (ρ=0.491, p<0.001), triglycerides (ρ=0.458, p<0.001), and waist circumference (ρ=0.403, p<0.001). Moreover, CAP showed a correlation with ALT (ρ=0.291, p<0.001), GGT (ρ=0.268, p<0.001), HDL-cholesterol (ρ=-0.218, p<0.001), and total bilirubin (ρ=-0.109, p=0.046). HOMA-IR was significantly higher in the group with CAP value ≥300 dB/m than in the groups with CAP value between 250–300 or CAP value <250 dB/m (HOMA-IR, 3.00±1.99 vs. 1.64±1.04 vs. 1.63±2.59, respectively; p<0.001) (Fig. 3A). Subjects with a CAP value >300 dB/m showed significantly higher fasting C-peptide than those with CAP value between 250–300 dB/m or <250 dB/m (fasting C-peptide; 2.71±0.95 ng/mL vs. 2.29±1.62 ng/mL vs. 1.97±0.88 ng/mL, respectively; p<0.001) (Fig. 3B).