Population characterization

The general characteristics were compared, and it revealed that gender, age, region, daily temperature, breathing, pulse, systolic blood pressure, and diastolic blood pressure exhibited no statistically significant difference between the HF group and control group. (P > 0.05, Table 1).

Table 1 Baseline characteristics of the study populationCpG methylation of DIO3-FA27 of HF

Using prism8.0 software, a violin plot comparing methylation levels of the promoter region of selenoprotein gene DIO3-FA27 in patients with HF and healthy control groups was fitted, indicating that a nonsignificant difference was observed in methylation fragments in the promoter region of DIO3-FA27 between patients with HF and healthy controls (P > 0.05, Fig. 2).

Fig. 2

Comparison of methylation levels in the DIO3-FA27 promoter region between patients with HF and healthy controls

Various degrees of CpG methylation of DIO3-FA27 in HF patients

The violin diagram results comparing methylation levels of DIO3-FA27 promoter region in patients with different HF degrees illustrated that in the heart function of patients with HF, the methylation fragments of the promoter region of the DIO3-FA27 promoter, DIO3-FA27_CpG_11.12 (P = 0.0382), and DIO3-FA27_CpG_23.24 (P = 0.0427) were statistically significant (P < 0.05, Fig. 3f–j).

Fig. 3

Comparison of methylation levels in the promoter region of DIO3-FA27 in patients with heart failure with different cardiac functions

Analysis of influencing factors of clinical biochemical indices, CpG methylation, and HF

Logistic regression was deployed to analyze the effects of various clinical biochemical indices on patients with HF, and the results were as follows:

Univariate logistic regression results manifested that triglyceride (TG), serum total cholesterol (TC), high-density lipoprotein (HDL), human Apolipoprotein A-I (ApoA-1), low-density lipoprotein (LDL), and ApoB were probably related to patients with HF. Meanwhile, prothrombin time activity (PTA), prothrombin time ratio (PTR), APTT, and fibrin degradation products (FDP) in coagulation indicators were probably correlated with HF. Among the renal function indices, cholinesterase (CHE) and urea were probably associated with patients with HF. Moreover, alkaline phosphatase (ALP) and indirect bilirubin (IBIL) in liver function indicators were probably related to patients with HF. Furthermore, mean corpuscular volume (MCV) and mean corpuscular hemoglobin concentration (MCH) were connected to patients with HF (P < 0.05). The effect of the methylation fragment of the promoter region of DIO3-FA27 on patients with HF represented a nonsignificant difference (P > 0.05, Fig. 4).

Fig. 4

Univariate Logistic regression for HF. (A Coagulation index and HF. B Blood lipid index and HF. C Blood routine indices and HF. D Renal function indices and HF. E Liver function indices and HF. F DIO3-FA27 promoter region methylation tablets and HF)

Therefore, the five items of blood lipids, coagulation, liver function, renal function, and blood routine in clinical biochemical indices, DIO3-FA27_CpG_11.12, and DIO3-FA27_CpG_23.24 may be related to HF.

In the analysis of differences between different cardiac function grades and biochemical indices, we can find that there are statistically significant differences in PTR and ApoA-1 between cardiac function I/II and III/IV. (P < 0.05, Table 2).

Table 2 Comparison of biochemical indices with different cardiac functions \((\pm}})\) 

The multivariate logistic regression results (Table 3) elucidated that APTT levels had significant differences (P < 0.05) in the Q3 (OR = 27.00, 95% CI 1.08–674.12) and Q4 groups (OR = 264.04, 95% CI 3.76–18548.42), which indicated that APTT was positively linked to patients with HF after the APTT level increased to Q3 group. Meanwhile, FDP levels were significantly different (P < 0.05) in the Q2 (OR = 0.00, 95% CI 0.00–0.09), Q3 (OR = 0.02, 95% CI 0.0–0.57), and Q4 groups (OR = 0.03, 95% CI 0.00–0.90), revealing that FDP level has a negative connection to patients with HF. A positive correlation existed between urea and patients with HF in the Q3 group (OR = 0.11, 95% CI 0.01–1.24). In the blood routine index, MCV was positively associated with patients with HF at the levels of Q3 and Q4 groups (OR = 0.07, 95% CI 0.00–0.84; OR = 0.03, 95% CI 0.00–0.37); HDL was positively linked to patients with HF in Q3 and Q4 groups (OR = 21.00, 95% CI 1.78–248.10; OR = 81.00, 95% CI 4.36–1504.46). In liver function indices, ALP was positively related to patients with HF at the Q4 group level (OR = 0.03, 95% CI 0.00–0.37).

Table 3 Multivariate logistic regression analysis of the influence of five biochemical markers on patients with HFDose–response relationship between HF differential CpG site methylation level and various clinical biochemical indices

Taking the 25th, 50th, and 75th percentiles of methylation levels at HF differential CpG sites as three nodes, a restrictive cubic spline model was established for eight clinical biochemical indices: blood pressure, blood lipid, coagulation, liver function, renal function, protein, electrolytes, and blood routine. However, it was found that the methylation levels of HF differential CpG sites DIO3-FA 27_CpG_11.12 and DIO3-FA 27_CpG_23.24 were only associated with coagulation, liver function, and kidney function. There exists a correlation and dose–response relationship between the four indices of blood routine.

HF differential CpG site methylation level in relation to dose–response of coagulation indices

Using restricted cubic spline regression, the dose–response curves of methylation levels at DIO3-FA27_CpG_11.12 and DIO3-FA 27_CpG_23.24 sites on coagulation indices were further fitted. The solid line in Fig. 5 represents the OR of the coagulation index. The results (Fig. 5) illustrate that a correlation exists between plasma PT, PTA, PTR, partially activated APTT, fibrinogen degradation product FDP, and DIO3-FA27_CpG_11.12 site methylation level where PT, PTA, PTR, and FDP have all “U-shaped” nonlinear correlations with DIO3-FA27_CpG_11.12. However, no dose–response relationship exists (Poverall > 0.05, Pnon-linnear < 0.05, Fig. 5A, B, C, and D).

Fig. 5

Relationship between HF differential CpG site methylation level and coagulation index

The APTT and DIO3-FA27_CpG_11.12 had an “inverted U-shaped” nonlinear association, but no dose–response relationship was observed (Poverall > 0.05, Pnon-linnear < 0.05, Fig. 5E).

HF differential CpG site methylation level in relation to dose response of liver function

The dose–response relationship between the methylation levels of DIO3-FA27_CpG_11.12 and DIO3-FA27_CpG_23.24 sites on liver function is observed, and the solid lines in Fig. 6 represent the OR values of each liver function index. The analysis results (Fig. 6) revealed that mAST and TBA were associated with DIO3-FA27_CpG_11.12, both exhibiting a “U-shaped” linear dose–response relationship (Poverall < 0.05, Pnon-linnear > 0.05), where mAST and DIO3-FA27_CpG_11.12 had a negative linear correlation. With the elevated methylation level at this site, the mAST OR value decreased (Fig. 6A).

Fig. 6

Dose–response relationship between HF differential CpG site methylation level and liver function indices

The TBA was positively and linearly correlated with the methylation level of DIO3-FA27_CpG_11.12 site. With the escalated methylation level of this site, the OR of TBA also increased (Fig. 6B).

HF differential CpG site methylation level corresponding to dose response of renal function

The dose–response relationship between the methylation levels of DIO3-FA27_CpG_11.12 and DIO3-FA27_CpG_23.24 sites on renal function (Fig. 7) indicated that CHE and CO2 were associated with DIO3-FA27_CpG_11.12 methylation levels, where CHE and the methylation level of this site demonstrated an “inverted U-shaped” nonlinear correlation, without dose–response relationship (Poverall > 0.05, Pnon-linnear < 0.05). Meanwhile, CO2 revealed a “U-shaped” nonlinear association with methylation levels at this site but without a dose–response relationship (Poverall > 0.05, Pnon-linnear < 0.05, Fig.7A, B).

Fig. 7

Dose–response relationship between HF differential CpG site methylation level and renal function indices

HF differential CpG site methylation level corresponding to dose response of blood routine indices

The dose–response relationship between methylation levels of DIO3-FA27_CpG_11.12 and DIO3-FA27_CpG_23.24 on routine 1 indices (Fig. 8) indicated that MONO%, RDW-SD, RDW-CV%, DIO3-FA27_CpG_11.12, and DIO3-FA27_CpG_23.24 exhibited “U-shaped” nonlinear associations with dose–response relationship (Poverall < 0.05, Pnon-linnear < 0.05). Therefore, with the heightened DIO3-FA27_CpG_11.12 and DIO3-27_CpG_23.24 methylation levels, the OR of MONO%, RDW-SD, and RDW-CV% gradually decreased and then gradually increased, (Fig.  8A–F).

Fig. 8

Relationship between HF differential CpG site methylation level and dose–response of blood routine 1 index

PDW was nonlinearly associated with the methylation level of DIO3-FA27_CpG_11.12 sites but without a dose–response relationship (Poverall > 0.05, Pnon-linnear < 0.05). Furthermore, PDW exhibited a “U-shaped” nonlinear association with the DIO3-FA27_CpG_23.24 methylation level, with a dose–response relationship (Poverall < 0.05, Pnon-linnear < 0.05). As the methylation level of this site elevates, the OR of PDW first gradually decreases and then gradually increases (Fig. 8G, H).

The MPV was nonlinearly associated with DIO3-FA27_CpG_11.12 as a “U-shaped,” with a dose–response relationship (Poverall < 0.05, Pnon-linnear < 0.05). With the elevated methylation level of this site, the OR of MPV first gradually decreases and then gradually increases (Fig.  8I).

The dose–response relationship between HF differential CpG sites DIO3-27_CpG_11.12 and DIO3-FA27_CpG_23.24 in routine 2 revealed the following results: RBC, PLT, HGB, NEUT, HCT, MCHC, MCV, and DIO3-FA27_CpG_23.24 showed a nonlinear association of “inverted U-shaped” and a dose–response relationship (Poverall < 0.05, Pnon-linnear < 0.05). Briefly, with the elevated methylation level of this site, the OR of RBC, PLT, HGB, NEUT, HCT, MCHC, and MCV gradually increased to a certain value and then gradually decreased (Fig. 8J-L, Fig. 9A, B, F, G).

Moreover, LYMPH, MCH, EO%, and DIO3-FA27_CpG_23.24 methylation levels all exhibited “U-shaped” nonlinear associations, with a dose–response relationship (Poverall < 0.05, Pnon-linnear < 0.05). The results indicated that with the raised methylation levels at this site, the OR of LYMPH, MCH, and EO% gradually decreased and then increased (Fig. 9C–E).

Fig. 9

Relationship between HF differential CpG site methylation level and dose–response of blood routine 2 index

According to the results of RCS, these four types of clinical biochemical indices: coagulation indices (PT, PTA, PTR, APTT, and FDP), liver function indicators (mAST and TBA), renal function indicators (CHE and CO2), blood routine indicators (MONO%, RDW-CV, MPV, HGB, PLT, RBC, PDW, RDW-SD, NEUT, EO%, MCV, MCH, MCHC, HCT, and LYMPH), and HF have a dose–response relationship. Results of RCS model with P＞0.05 for other clinical biochemical indices with methylation levels of the key CpGs are attached to Additional file 1.

ROC analysis for coagulation, liver function, renal function, and blood routine diagnosis of HF

According to the RCS results, the ROC curves of coagulation, liver function, kidney function, and blood routine indices were established, and the results showed that the AUC (95% CI) of the four biochemical index models for diagnosing HF was 0.93 (0.84–1.00), 0.50 (0.32–0.68), 0.73 (0.57–0.88), and 0.96 (0.90–1.00). The blood routine index combined with HF was the best among the four indicators (Table 4 and Fig. 10).

Table 4 ROC curve parameters of four biochemical markersFig. 10

ROC curves of four types of clinical biochemical indices for HF diagnosis (A ROC curve of coagulation index combined diagnosis of HF. B ROC curve of liver function indices combined to diagnose HF. C ROC curve of renal function indices combined for the diagnosis of HF. D ROC curve for combined diagnosis of HF by blood routine indices)