In this study, a total of 259 individuals were classified into three cohorts (their clinical and demographic characteristics are presented in Supplementary Tables 1–3). The participants were enrolled from among the outpatients and inpatients of the cardiology and endocrinology departments of the People’s Hospital of Zhengzhou University from July 2015 to June 2016. Subjects with any of the following characteristics were excluded: (i) malignancies, (ii) liver and kidney dysfunction, (iii) any other clinically systemic acute or chronic inflammatory disease(s), (iv) autoimmune disease, (v) untreated hypertension, and (vi) any endocrine disease other than T2DM.
The process of this study is shown in Fig. 1 Each study subject was tested using the oral glucose tolerance test (OGTT) to determine whether they were healthy or had pre-diabetes or T2DM. Six control individuals and six T2DM patients were selected to donate venous blood samples, and total RNA was extracted for microarray analysis. The screened circRNAs were then validated in an independent cohort (control group, n = 20; pre-diabetes group, n = 20; and T2DM group, n = 20). After receiver operating characteristic (ROC) curve analysis, the circRNA with the best diagnostic value was selected as the biomarker, and its diagnostic value was validated in another independent cohort (control group, n = 60; pre-diabetes group, n = 63; and T2DM group, n = 64).
The definitions of pre-diabetes and T2DM and collection of whole blood samples
In this study, pre-diabetes and T2DM were diagnosed according to the 1998 Standards of the World Health Organization (WHO) . Thus, patients meeting either of the following criteria could be diagnosed as having T2DM: (i) fasting plasma glucose (FPG) ≥ 125 mg/dL (7.0 mmol/L), where fasting is defined as no caloric intake for at least 8 h or (ii) two-hour post-load plasma glucose ≥200 mg/dL (11.1 mmol/L) during an OGTT.
Additionally, patients meeting either of the following standards were diagnosed as having pre-diabetes: (i) FPG ≥110 mg/dL (6.1 mmol/L) and <125 mg/dL (7.0 mmol/L); or (ii) two-hour post-load plasma glucose ≥140 mg/dL (7.8 mmol/L) and <200 mg/dL (11.1 mmol/L) during an OGTT.
Blood sample collection was performed as follows: After overnight fasting, 2 mL of blood was collected from the median cubital vein of each patient before breakfast and then stored in ethylenediaminetetraacetic acid (EDTA) anticoagulant vacutainers. The total RNA was then extracted as soon as possible.
RNA extraction and Q-PCR
A fast total RNA extraction kit (Biotech, Beijing, China) was used to extract total RNA from 1 mL of whole blood according to the manufacturer’s instructions. RNA was then dissolved in RNase-free water. The yield and purity were measured by a NanoDrop 2000 instrument (Thermo Scientific, Waltham, MA, USA). The integrity of the RNA was determined by 1% formaldehyde denaturing gel electrophoresis. A PrimeScript RT Reagent Kit (Takara Bio, Nojihigashi, Kusatsu, Japan) was used for the production of complementary DNA (cDNA) by reverse transcription, according to the manufacturer’s instructions. Q-PCR was performed using SYBR-Green Premix Ex Taq (Takara Bio, Nojihigashi, Kusatsu, Japan) and monitored by an ABI PRISM 7500 Sequence Detection System (Applied Biosystems, Life Technologies, Waltham, MA, USA). The relative expression levels of circRNAs were determined via Q-PCR. The sequences of the primers used in the Q-PCR assay are shown in Supplementary Table 4.
CircRNA microarray analysis
The RNAs of the peripheral blood of six control subjects and six T2DM patients were extracted for microarray analysis. The purity and concentration of the RNA were determined by a NanoDrop ND-1000 instrument (Thermo Scientific, Waltham, MA, USA). The integrity of the RNA was evaluated using a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). The extracted RNAs were digested, dephosphorylated, denatured, amplified and labeled with Cy3-dCTP according to the manufacturer’s specifications. The purified RNAs were hybridized to a microarray (Agilent human circRNA Array V2.0) containing 170,340 human circRNA probes. The microarray data of the circRNAs were then analyzed using GeneSpring software V13.0 (Agilent Technologies, Santa Clara, CA, USA). The thresholds were as follows: fold change, ≥2 or ≤−2; P < 0.05 according to the t test.
Variables with different distributions were expressed as means ± standard deviations, medians (quartiles) or percentages when they fit. In the scatterplot of circRNA expression, the horizontal lines represent the median values. The Chi-square test was used for categorical variables, whereas the Kolmogorov–Smirnov and Shapiro–Wilk tests were performed to check data normality for continuous variables, followed by the test for homogeneity of variances. The clinical and demographic indicators were checked for significant differences by one-way analysis of variance (ANOVA), if the continuous variables were consistent with the normal distribution and homogeneity of variance; if not, the Kruskal–Wallis H test was used. The clinical diagnostic value of a given circRNA was verified by ROC curve analysis, and when the AUC was equal to 0.5, the circRNA was defined as having no diagnostic value. Furthermore, logistic regression analysis was performed to obtain an odds ratio (OR) when the relative expression of circRNAs was expanded by ten times. P < 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS 22.0 (SPSS Inc., Chicago, IL, USA).