The signature of HBV-related liver disease in peripheral blood mononuclear cell DNA methylation

Clinical Epigenetics

Table 2 Multivariate logistic regression model to predict LC

Variables		B	OR	95% CI of OR	P
Intercept		11.52	1.01 e + 05		0.0326
cg05650055 (%)		− 0.19	0.83	0.68–0.92	0.0386
cg17149911 (%)		− 0.21	0.81	0.64–0.92	0.0471
AST (U/L)	≤ 15 or > 40(male) ≤ 13 or > 35(female)	Reference	Reference
	15–40 (male)	− 2.82	0.059	0.003–0.538	0.05
	13–35 (female)	− 2.82	0.059	0.003–0.538	0.05

B regression coefficient, OR odds ratio in favor of having LC
The LC prediction model was conducted for training and internal testing with 48 liver disease patients and its accuracy was examined using a confusion matrix analysis. The AUROC of CpG sites (cg17149911 + cg05650055) model and LC model to predict LC was 0.8542 (95% CI, 0.722–0.939), 0.917 (95% CI, 0.80–0.977), respectively in Fig. 4d and Supplementary table 5.

ISSN: 1868-7083