Constructing a nomogram model for predicting liver cirrhosis based on serological indexes in patients with chronic hepatitis B
10.3760/cma.j.cn115455-20221103-00948
- VernacularTitle:基于血清学相关指标构建预测慢性乙型肝炎患者发生肝硬化的列线图模型
- Author:
Bin LUO
1
;
Ruifen ZHOU
;
Jianguang ZHU
Author Information
1. 咸宁市中心医院 湖北科技学院附属第一医院检验科，咸宁 437100
- Keywords:
Hepatitis B;
Liver cirrhosis;
Serology;
Nomograms
- From:
Chinese Journal of Postgraduates of Medicine
2023;46(9):791-798
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To analyze the influence of serological indexes on the liver cirrhosis (LC) in patients with chronic hepatitis B, and to construct a nomogram model.Methods:The clinical data of 220 patients with chronic hepatitis B in Xianning Central Hospital from January 2019 to December 2021 were retrospectively analyzed. Among them, 42 patients developed LC (LC group), and 178 cases did not develop LC (non-LC group). The patient′s fasting peripheral venous blood was taken in the morning. The platelet, red blood cell, white blood cell, fasting blood glucose, alanine aminotransferase (ALT), aspartate aminotransferase (AST), triacylglycerol (TG), total cholesterol (TC), total bilirubin (TBiL), albumin, globulin, alkaline phosphatase (ALP), γ-glutamyltransferase (GGT), prothrombin time (PT), thrombin time (TT), D-dimer (D-D), alpha-fetoprotein (AFP) and C-reactive protein (CRP) were detected. Receiver operating characteristic (ROC) curve was used to analyze the efficacy of each index in predicting LC in patients with chronic hepatitis B. Multivariate Logistic regression was used to analyze the independent risk factors for LC in patients with chronic hepatitis B. The R software "rms" package was used to construct a nomogram model to predict the LC in patients with chronic hepatitis B, the correction curve was used to internally verify the prediction model, and the decision curve evaluated the efficacy of the prediction model.Results:The TBiL, ALP, GGT, PT, TT, D-D, AFP and CRP in LC group were significantly higher than those in non-LC group: (50.57 ± 5.61) μmol/L vs. (46.69 ± 3.92) μmol/L, (105.23 ± 30.60) U/L vs. (75.14 ± 26.45) U/L, (68.73 ± 19.47) U/L vs. (50.39 ± 14.21) U/L, (13.88 ± 1.98) s vs. (13.01 ± 2.10) s, (18.88 ± 2.56) s vs. (15.98 ± 2.43) s, (2.62 ± 1.04) mg/L vs. (1.34 ± 0.63) mg/L, (4.19 ± 1.95) μg/L vs. (2.66 ± 1.21) μg/L and (8.54 ± 1.22) mg/L vs. (7.47 ± 0.79) mg/L, the platelet, ALT, AST and albumin were significantly lower than those in the non-LC group: (129.63 ± 32.66) × 10 9/L vs. (183.53 ± 56.31) ×10 9/L, (131.27 ± 22.19) U/L vs. (157.57 ± 38.67) U/L, (112.76 ± 19.57) U/L vs. (125.16 ± 21.84) U/L and (29.79 ± 6.17) g/L vs. (33.52 ± 5.89) g/L, and there were statistical differences ( P<0.01 or <0.05); there were no statistical differences in red blood cell, white blood cell, fasting blood glucose, TG, TC and globulin between the two groups ( P>0.05). ROC curve analysis result showed that the area under the curve (AUC) of AFP, platelet, ALT, AST, ALP, GGT, TBiL, albumin, D-D, CRP, PT and TT for predicting LC in patients with chronic hepatitis B were 0.731, 0.798, 0.723, 0.676, 0.766, 0.762, 0.710, 0.673, 0.856, 0.759, 0.603 and 0.786, and the optimal cut-off values were 4.64 μg/L, 162.56 × 10 9/L, 155.67 U/L, 122.37 U/L, 95.17 U/L, 68.96 U/L, 49.95 μmol/L, 28.8 g/L, 1.64 mg/L, 8.55 mg/L, 12 s and 18 s. Multivariate Logistic regression analysis result showed that AFP (>4.64 μg/L), platelet (≤162.56 × 10 9/L), ALP (>95.17 U/L), GGT (>68.96 U/L), D-D (>1.64 mg/L) and TT (>18 s) were independent risk factors for LC in patients with chronic hepatitis B ( OR = 1.278, 1.428, 1.488, 1.356, 1.513 and 1.369; 95% CI 1.109 to 1.369, 1.269 to 1.623, 1.217 to 1.894, 1.127 to 1.669, 1.342 to 1.878 and 1.169 to 1.583; P<0.05 or <0.01). The AFP, platelet, ALP, GGT, D-D and TT were used as predictors to construct a nomogram model for predicting the LC in patients with chronic hepatitis B. The correction curve of the nomogram model to predict the LC in patients with chronic hepatitis B was close to the ideal curve (C-index was 0.739, 95% CI 0.615 to 0.876); the decision curve analysis result showed that the prediction model had higher clinical net benefit when the risk threshold > 0.26 than a single index, and that it had significantly additional clinical net benefit. Conclusions:The AFP, platelets, ALP, GGT, D-D and TT are independent risk factors for LC in patients with chronic hepatitis B, and the nomogram model constructed based on these factors could provide important guidance for the prevention and treatment of LC in patients with chronic hepatitis B.
