Association of circular RNAs and environmental risk factors with coronary heart disease

A frequency-matched case-control study was conducted in which case groups and controls must have the same gender and age ratio (±3 years). A research factor with a small OR was selected to calculate the appropriate sample size, according to the pre-experimental results and considered the following: α = 0.05, β = 0.10, P₀ = 0.260 and OR = 1.70. Calculations using PASS revealed N₁ = N₂ = 379 people. In this study, 1300 questionnaires, including 650 cases and 650 controls, were distributed, and all were collected between March 2016 and September 2017 from the First Affiliated Hospital of Fujian Medical University and the Affiliated Union Hospital of Fujian Medical University, China. After corresponding the age and gender, 585 CHD patients and 585 controls were enrolled in this study. CHD was defined using the following criteria: (1) significant stenosis (≥ 50%) of > 1 major coronary artery was confirmed by present cardiac catheterisation, (2) documented history of prior myocardial infarction (MI) and a prior coronary revascularisation procedure (percutaneous coronary intervention or coronary artery bypass graft), (3) patients in the stable stage after acute MI and patients with ST-segment elevation/depression on ECG. The subjects without medical history of cardiovascular diseases were selected as the non-CHD subjects. All subjects were long-term residents in Fujian, subjects with other types of heart disease, serious brain organic diseases, malignant tumours, hepatic or renal dysfunction, recent infections and endocrine system diseases were excluded from the study. This study protocol conformed to the ethical guidelines of the Declaration of Helsinki. The protocol was approved by the ethics committee of Fujian Medical University School. All selected patients and controls provided informed consent.

All interviewer who conducted the standard questionnaire were specially trained. The questionnaire included demographic characteristics (age, gender, marital status, educational level), lifestyle habits (smoking, alcohol drinking, diet, exercise), social psychological factors (, anxiety and depression levels), Physiological index (height, weight, waist circumference (WC)) and family history of cardiovascular disease. The questionnaire was listed in Additional file 1: Table S1. Individuals who have consumed more than 20 packs of cigarettes or one cigarette a day for at least 1 year were defined as smokers. Alcohol drinkers were defined according to literature [17]. Each exercise lasting for no less than 20 min will be recorded as an effective physical exercise, according to this standard, the weekly physical activity of all subjects were determined. Anxiety and depression were evaluated by self-rating depression scale (SDS) and self-rating anxiety scale (SAS) [18]. SAS scores were divided into < 50, 50-59, 60-69, and ≥ 70, which were used to indicate normal, mild anxiety, moderate anxiety, and severe anxiety, respectively. SDS was categorised as < 53, 53–62, 63–73 and ≥ 74, which were used to indicate normal, mild depression, moderate depression, and severe depression, respectively. Body mass index (BMI) = weight (kg) / [height (m)]² and categorised to three scales as follows: < 18.5 kg/m² underweight, 18.5–24.0 kg/m² normal and ≥ 24.0 kg/m² overweight and obese. A male waistline cutoff of 85 or higher means abdominal obesity, and a female waistline cutoff of 80 or more means abdominal obesity. According to gender, WC was classified into two categories: A male WC with a cutoff of 85 or higher implies abdominal obesity, whereas a female WC with a cut-off value of 80 or more suggests abdominal obesity.

The samples used for microarray analysis include 4–12 cases [19]. Therefore, five patients with similar age, disease and disease duration and with no other diseases and five controls with similar general conditions, age and sex was selected for microarray analysis. The distribution of demographic variables is listed in Additional file 2: Table S2. The RNAs of the peripheral blood samples from five CHD cases and five non-CHD subjects were extracted for microarray analysis. According to the manufacturer’s instructions, total RNA was extracted from 1 mL of whole blood sample by using a fast total RNA extraction kit (Bioteke, Beijing, China). RNA was then dissolved in RNase-free water. NanoDrop1000 instrument (NanoDrop, England) was used to measure the yield and purity of the RNA.. All total RNA was of high purity with OD260 / 280 in the range of 1.80-2.00. 1% formaldehyde denaturing gel electrophoresis was used to determine the integrity of the RNA. The extracted RNAs were digested, dephosphorylated, denatured, amplified and labelled with Cy3-dCTP according to the manufacturer’s specifications. The purified RNAs were hybridised to a microarray (Agilent human circRNA Array V2.0) containing 170,340 human circRNA probes. GeneSpring software V13.0 (Agilent Technologies, Santa Clara, CA, USA) was then used to analyze microarray data for circRNA. The thresholds were fold change (FC) of > 2 or < − 2 and p < 0.05 according to the t-test. The data were log-2 transformed and median-centred by genes through CLUSTER 3.0 software and then analysed using hierarchical clustering with average linkage.

Representative subjects were selected using stratified sampling from the first part of the study. The subjects in the case group were divided into two layers according to age (i.e. < 65 years old and ≥ 65 years old) and stratified by gender according to each age group. The sex ratio of all subjects in the case group were determined. According to the random number table, the corresponding number of male and female subjects was finally selected at each age level. The control group were also sampled using the same method. According to the pre-experimental results, the PASS software obtained the following result: N₁ = N₂ = 25, considering α = 0.05, β = 0.10, P₀ = 0.50 and OR = 1.8. Therefore, at least 25 samples were required in each group. Therefore, 30 cases and controls were finally determined for peripheral blood quantitative real-time polymerase chain reaction (qRT-PCR) validation, and 100 cases and controls were used for peripheral blood leukocyte qRT-PCR validation. A fast total RNA extraction kit (Bioteke, Beijing, China) was used to extract total RNA from peripheral blood in 30 CHD patients and 30 non-CHD controls. Total RNA from peripheral blood leukocytes in 100 CHD patients and 100 non-CHD controls was extracted by using TRIzol reagent (Invitrogen, USA). The examination of RNA concentration, purity, and integrity were as described above.

Reverse transcription of quantified RNA was performed using PrimeScript RT Reagent Kit (Takara Bio Inc., Shiga, Japan) according to the manufacturer’s instructions. qRT-PCR was used to measure the expression levels of circRNAs, Which was performed on the LightCycler 480 Real-Time PCR System (Roche, Switzerland) with the SYBR® Premix Ex Taq™ II kit (Takara Bio Inc., Shiga, Japan). The reaction conditions were listed as follows: Amplification curves were obtained by 45 cycles of 95 °C 30 s, 95 °C 5 s, 60 °C 34 s, whereas dissolution curves were obtained by one cycle of 95 °C 15 s, 60 °C 1 min, 95 °C 15 s. GAPDH was used as internal control. 2^-△△CT method was used to calculate the expression levels of circRNAs. The primer sequences are shown in Additional file 3: Table S3.

Data with normal distribution were presented as the mean value±SD and compared using two-tailed Student’s t-test. Skewed data were represented as median (25th–75th quartile) and compared using Mann–Whitney U test. Discrete variables were displayed as percentages, distribution differences were examined by the chi-square (χ²) test. The diagnostic cut-offs of circRNAs were obtained from the receiver operating characteristic (ROC) curve. Univariate analysis of meaningful variables was included in the multivariate analysis. Crossover analysis was used to assess the association between CHD and risk factors. A p-value < 0.05 (two-tailed) was considered significant. All statistical analyses were performed using SPSS 25.0 software.

The distribution of all subjects according to demographic variables is shown in Table 1. No significant difference was found in their general demographic characteristics, including age, gender, marital status and educational level (all p > 0.05), thereby indicating that the frequency matching was adequate (Table 1).

Univariate analysis on the associations between individual behavioural factors and risk of CHD showed that salty diet. Social psychological factor analysis showed that anxious people are highly prone to CHD. BMI (≥24.00) and abdominal obesity were significantly different in CHD and non-CHD subjects. However, light diets and physical exercise (one to two times per week) were protected factors for CHD. Unconditional logistic regression analysis was further used to evaluate the associations between the environmental factors (high-salt diet, light diet, active exercise, anxiety, BMI and abdominal obesity) and CHD. Significant increased risk effects for CHD were associated with anxiety {odds ratio (OR) = 2.34; 95% confidence interval (95% CI): 1.67–3.28} and being overweight (BMI ≥ 24.00) {OR = 1.47; 95% CI: 1.13–1.91}. Physical exercise (one to two times per week) {OR = 0.41; 95% CI: 0.23–0.75} was the protective factor for CHD (Table 2).

The heat map, volcano plot and scatter plot of microarray assay showed abnormal expression of circRNAs in the CHD cases (Fig. 1). The microarray study identified that 3423 circRNAs were differentially expressed in CHD and non-CHD subjects based on FC and p-value (FC > 2, p < 0.05). Four circRNAs were selected for qRT-PCR validation analysis to independently validate our results and determine the roles of circRNAs in CHD. Selection was based on the following: (1) hsa_circ_0008507, hsa_circ_0125589 and hsa_circ_0000284 are among the most abundant and have significantly differentially expression according to microarray analysis, (2) overexpression of hsa_circ_0001946 in cardiac myocyte cells promotes cell apoptosis [20] and silencing hsa_circ_0000284 can alleviate retinal vascular dysfunction [21]. In summary, the circRNAs associated with the occurrence and development of CHD include hsa_circ_0125589, hsa_circ_0008507, hsa_circ_0001946 and hsa_circ_0000284 (Table 3).

We selected 30 CHD cases and 30 non-CHD controls with corresponding age and gender as qRT-PCR subjects. The expression levels of hsa_circ_0001946 and hsa_circ_0000284 in CHD were significantly elevated compared with those in the non-CHD subjects (Fig. 2). No significant differences were found in the expression levels of hsa_circ_0125589 and hsa_circ_0008507. The expression level of hsa_circ_0001946 was consistent with the microarray assay, whereas that of hsa_circ_0000284 was contrary to the result of the microarray.

We selected 100 CHD cases and 100 non-CHD controls as qRT-PCR subjects, all of which corresponded in terms of age and gender. The expression levels of hsa_circ_0001946 and hsa_circ_0000284 in the peripheral blood leukocytes in CHD group were significantly elevated compared with those in the non-CHD subjects. No significant difference was found in the expression level of hsa_circ_0125589 in peripheral blood leukocytes between the case and control groups (p > 0.05) (Fig. 3). These results were consistent with the qRT-PCR validation of peripheral blood samples. However, the expression level of hsa_circ_0008507 was up-regulated, which was inconsistent with the qRT-PCR validation of peripheral blood samples. The expression levels of hsa_circ_0008507 and hsa_circ_0001946 were consistent with the microarray assay, whereas that of hsa_circ_0000284 was contrary to the result of the microarray assay.

Two logistic regression analysis models were further used to analyse the association of factors (age, gender, marital status and education level in Model 1 and age, gender, marital status, education level, anxiety, BMI and physical activity in Model 2) with circRNA. CHD was the dependent variable (0 = no, 1 = yes), and the related environmental factors and four verified circRNAs in peripheral blood leukocytes were the independent variables. The results suggested that the high expression levels of hsa_circ_0008507 {(OR = 1.30; 95% CI: 1.13–1.51 in Model 1), (OR = 1.29; 95% CI: 1.11–1.50 in Model 2)}, hsa_circ_0001946 {(OR = 1.25; 95% CI: 1.06–1.47 in Model 1), (OR = 1.20; 95% CI: 1.01–1.42 in Model 2)}, hsa_circ_0000284 {(OR = 1.88; 95% CI: 1.24–2.84 in Model 1), (OR = 2.05; 95% CI: 1.32–3.19 in Model 2)} were the independent risk factors of CHD in the two models (Table 4).

The ROC curve in 100 cases and 100 controls of peripheral blood leucocytes of hsa_circ_0008507 showed an AUC of 0.75 (95% CI: 0.68–0.82). The maximum Youden’s index was 0.46 with a sensitivity of 0.86 and a specificity of 0.60, corresponding to a diagnostic cut-off of 1.975. The ROC curve of hsa_circ_0001946 showed that the AUC was 0.71 (95% CI: 0.64–0.79). The maximum Youden’s index was 0.37 with a sensitivity of 0.85 and a specificity of 0.52, corresponding to a diagnostic cut-off of 1.00. The ROC curve of hsa_circ_0000284 showed that the AUC was 0.68 (95% CI: 0.61–0.76). The maximum Youden’s index was 0.37 with a sensitivity of 0.66 and a specificity of 0.71, corresponding to a diagnostic cut-off of 1.42 (Fig. 4).

The diagnostic cut-offs were used to define high and low expression groups of hsa_circ_0008507, hsa_circ_0001946 and hsa_circ_0000284. Stratification by demographic characteristics and main lifestyle factors revealed the statistically significant associations between hsa_circ_0008507, hsa_circ_0001946 and CHD risk in young and old individuals, males and females, smokers and non-smokers, abdominal obesity and non-abdominal obesity individuals and passive smokers and active smokers. The statistically significant associations between hsa_circ_0000284 and CHD risk were also observed in these comparison groups. In addition, the combined effect of hsa_circ_0008507 and smoking was observed. Compared with non-smoking individuals with low hsa_circ_0008507 expression, smokers with high hsa_circ_0008507 expression showed the highest magnitude of OR in CHD risk (OR 6.94, 95% CI 1.95–24.72). A statistically significant multiplicative interaction was found between hsa_circ_0008507 and smoking for CHD (OR multiplicative: 5.68; 95% CI: 1.08–29.84; p = 0.040) (Tables 5 and 6).