General information
There were 151 pediatric patients underwent liver biopsy. and 35 cases were excluded. 116 patients (80 males and 36 females) were included in the study. The median age was 6 years old. there were 65 (56.03%) cases in the no significant fibrosis group and 51(43.97%) cases in the significant fibrosis group (Table 1).
Table 1
Characteristics of the Patients in training Set
Variable | Training Set (n = 116) |
Age(y) | 6(3–10) |
Male gender (n, %) | 80(68.96) |
Liver fibrosis stage† (n, %) | |
S0 | 34(29.31) |
S1 | 31(26.72) |
S2 | 32(27.59) |
S3 | 11(9.48) |
S4 | 8(6.90) |
† Liver fibrosis stage: Take the fibrosis score of liver biopsy as the gold standard |
Correlation analysis between variables and liver fibrosis stages
The spearman correlation analysis was used to test the correlation between variables and liver fibrosis stages (Table 2). The results showed that TBA, GGT, AST, Ratio of Direct bilirubin to Total bilirubin (D/T) and ALT were positively correlated with fibrosis stages, PA, CHE were negatively correlated with fibrosis stages. All the correlation coefficients of these variables were less than 0.5., so it was difficult to assess the fibrosis stages by s single indicator in children with CHB.
Table 2
Spearman's correlation between Variables and Liver fibrosis stage
Variable † | Training Set(n = 116) | R Value | P Value |
Age(year) | 6.00(3.00–10.00) | -0.129 | 0.167 |
D/T | 0.30(0.30–0.40) | 0.203 | 0.029* |
ALT (U/L) | 55.00(33.00-150.00) | 0.187 | 0.045* |
AST(U/L) | 57.50(35.50–127.00) | 0.286 | 0.002** |
AST/ALT | 1.00(0.70–1.30) | 0.095 | 0.312 |
GGT(U/L) | 19.00(15.00-39.65) | 0.329 | 0.000** |
ALP (U/L) | 249.00(220.00-296.00) | -0.002 | 0.980 |
Globulin (g/L) | 25.30(22.70–28.70) | 0.045 | 0.635 |
Albumin (g/L) | 43.40(41.50–45.30) | 0.061 | 0.514 |
Prealbumin (mg/L) | 163.90(141.85-190.45) | -0.285 | 0.002** |
Cholinesterase (U/L) | 8194.00(7348.50–9567.00) | -0.191 | 0.004** |
Total protein (g/L) | 68.55(64.65–71.90) | 0.034 | 0.713 |
Total bile acid (µmol/L) | 10.30(5.85–20.80) | 0.292 | 0.001** |
Total bilirubin (µmol/L) | 6.30(4.80–10.30) | 0.041 | 0.663 |
Indirect bilirubin (µmol/L) | 4.00(3.05–6.65) | -0.040 | 0.669 |
Direct bilirubin (µmol/L) | 2.20(1.50–3.20) | 0.138 | 0.139 |
APTT(s) | 35.00(32.55–37.45) | 0.156 | 0.094 |
INR(s) | 0.95(0.91-1.00) | -0.039 | 0.678 |
Thrombin time (s) | 11.60(10.90–12.40) | -0.051 | 0.584 |
Fibrinogen (g/L) | 2.83(2.59–3.28) | -0.165 | 0.077 |
Prothrombin time (s) | 11.20(10.60–11.80) | 0.007 | 0.942 |
MPV(fl) | 8.19(7.64–9.21) | 0.055 | 0.555 |
Platelet count (109/L) | 287.90(245.00-336.10) | -0.113 | 0.227 |
White blood cell (109/L) | 7.67(6.10–9.66) | -0.040 | 0.669 |
Thrombocytocrit (ml/L) | 0.24(0.20–0.28) | -0.078 | 0.405 |
Lg[HBVDNA(copies/ml)] | 7.36(6.72–8.01) | -0.054 | 0.567 |
†variables are presented as median(interquartile range (IQR)) |
Abbreviation: D/T, direct bilirubin to total bilirubin ratio; ALT, Alanine transaminase; AST, Aspartate transaminase; AST/ALT, aspartate transaminase to ratio; GGT, gamma-glutamyl transpeptidase; ALP, Alkaline phosphatase; APTT, activated partial thromboplastin time; INR, international normalized ratio; MPV, mean platelet volume; Lg, logarithm base 10;R Value, correlation coefficient. |
Difference analysis between the no significant fibrosis group and significant fibrosis group.
According to the liver biopsy, the patients were divided into two groups: no significant fibrosis group (S < 2 n = 65) and significant fibrosis group (S ≥ 2 n = 51). Difference analysis between the two groups had been shown in Table 3. The results suggested that PA(P = 0.002), ALT(P = 0.045), AST(P = 0.002), CHE(P = 0.004), GGT(P = 0.000) and TBA(P = 0.001) were independent predictors of significant fibrosis.
Table 3
The difference analysis between the two groups in training set
Variable † | No Significant Fibrosis n = 65 | Significant Fibrosis n = 51 | Statistics Value | P Value |
Age(year) | 3.00(6.00–10.00) | 2.00(6.00-9.50) | Z=-0.648§ | 0.517 |
D/T | 0.30(0.30–0.40) | 0.40(0.30–0.40) | Z=-1.639§ | 0.101 |
ALT (U/L) | 45.00(28.00-119.00) | 72.00(39.50–203.00) | Z=-2.128§ | 0.033* |
AST(U/L) | 45.00(30.00–79.00) | 76.00(42.50–173.00) | Z=-3.010§ | 0.003** |
AST/ALT | 1.00(0.70–1.30) | 1.00(0.75–1.44) | Z=-0.773§ | 0.440 |
GGT(U/L) | 17.00(14.00–21.00) | 29.00(17.00-47.50) | Z=-3.873§ | 0.000** |
ALP (U/L) | 243.00(223.00-295.00) | 252(208–295) | Z=-0.117§ | 0.907 |
Globulin (g/L) | 24.90(22.40–28.70) | 26.00(23.30–28.70) | T=-1.172‡ | 0.244 |
Albumin (g/L) | 43.40(41.50–45.30) | 43.40(41.60-45.35) | T=-0.031‡ | 0.975 |
Prealbumin (mg/L) | 175.00(146.60-202.10) | 152.60(126.95–178.70) | T = 3.338‡ | 0.001** |
Cholinesterase (U/L) | 8691.00(7519.00-9845.00) | 7981.00(7044.00-8632.50) | Z=-2.172§ | 0.030* |
Total protein (g/L) | 68.30(64.80–71.90) | 69.10(64.70–71.70) | T=-0.611‡ | 0.542 |
Total bile acid (µmol/L) | 8.40(5.10-15.73) | 15.50(6.65-26.00) | Z=-2.535§ | 0.012* |
Total bilirubin (µmol/L) | 6.50(5.00-8.70) | 5.70(4.80–8.70) | Z=-0.320§ | 0.749 |
Indirect bilirubin (µmol/L) | 4.50(3.30-6.00) | 3.80(2.90–7.30) | Z=-0.281§ | 0.779 |
Direct bilirubin(µmol/L) | 2.20(1.40–2.90) | 2.20(1.60–3.65) | Z=-1.303§ | 0.193 |
APTT(s) | 34.70(32.50–37.30) | 35.20(32.90–37.50) | T=-0.092‡ | 0.926 |
INR(s) | 0.95(0.91–1.02) | 0.94(0.90–0.99) | Z=-0.914§ | 0.361 |
Thrombin time (s) | 11.60(10.90–12.50) | 11.60(10.95–12.20) | T=-0.090‡ | 0.929 |
Fibrinogen (g/L) | 2.89(2.62–3.28) | 2.68(2.51–3.27) | Z=-1.196§ | 0.232 |
Prothrombin time (s) | 11.20(10.70–12.00) | 11.10(10.60–11.70) | Z=-0.818§ | 0.413 |
MPV(fl) | 8.13(7.50–8.97) | 8.21(7.92–9.34) | Z=-1.179§ | 0.238 |
Platelet count (109/L) | 288.70(250.80-343.20) | 282.60(240.70–330.00) | T = 0.763‡ | 0.447 |
White blood cell (109/L) | 7.64(6.25–10.13) | 8.05(5.80–9.35) | Z=-1.093§ | 0.274 |
Thrombocytocrit (ml/L) | 0.25(0.20–0.29) | 0.24(0.21–0.28) | T=-0.001‡ | 0.900 |
Lg[HBVDNA(copies/ml)] | 7.39(6.73–8.28) | 7.32(6.77–7.81) | Z=-1.046§ | 0.296 |
†Variables are presented as median(interquartile range (IQR)).‡Normal distribution data: Independent sample T-test. §Non-normal distribution data: Mann-Whitney U test. |
Abbreviation: D/T, direct bilirubin to total bilirubin ratio; ALT, Alanine transaminase; AST, Aspartate transaminase; AST/ALT, aspartate transaminase to ratio; GGT, gamma-glutamyl transpeptidase; ALP, Alkaline phosphatase; APTT, activated partial thromboplastin time; INR, international normalized ratio; MPV, mean platelet volume; Lg, logarithm base 10. |
The formula Establishment of noninvasive model and selection of cut-off values
After taking logarithm change of variables that with statistical significance in difference analysis and Spearman correlation analysis. Fitting multiple model was calculated by logistic regression analysis. And determine the final model by backward stepwise procedures: Y=-1.803LN (PA (mg/L)) + 0.769LN(GGT(U/L)) + 6.436 (AUROC 0.732, 95%CI 0.642–0.810). Because the formula was difficult to calculate in application, we used the mathematical relations to simplify it and got the G index (AUROC 0.733, 95%CI 0.643–0.811):
G Index = 600 × GGT(U/L)/(PA(mg/L))2
When the G index value ≤ 0.28, it could be consideredwithout no significant liver fibrosis (the sensitivity was 86.27% and the specificity was 38.46%). in all the 51 patients in significant fibrosis group, there were only 7 cases (13.73% six cases in S2, one in S3) with G index value ≤ 0.28. In all the 116 patients, the negative predictive value (NPV) was 78.13%, and diagnostic accuracy (DA) was 59.48% (69/116) (Table 4).
Table 4
Accuracy of G index and APRI in Predicting Significant Fibrosis
Model | Patients n = 116 n(%) | S0-1 n = 65 n(%) | S2-4 n = 51 n(%) | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | DA (%) |
G index | | | | | | | | |
≤ 0.28 | 32(28) | 25(38) | 7(14) | 86.27 | 38.46 | 52.38 | 78.13 | 59.48 |
> 0.28 | 84(72) | 40(62) | 44(86) | | | | | |
≤ 1.16 | 88(76) | 57(88) | 31(61) | | | | | |
> 1.16 | 28(24) | 8(12) | 20(39) | 39.22 | 87.69 | 71.43 | 64.77 | 66.38 |
APRI | | | | | | | | |
≤ 0.26 | 28(24) | 21(32) | 7(14) | 86.27 | 32.31 | 50.00 | 75.00 | 56.03 |
> 0.26 | 88(76) | 44(68) | 44(86) | | | | | |
≤ 0.9 | 90(78) | 57(88) | 33(65) | | | | | |
> 0.9 | 26(22) | 8(12) | 18(35) | 35.29 | 87.69 | 69.23 | 63.33 | 64.66 |
FIB-4 | | | | | | | | |
≤ 0.05 | 14(12) | 9(14) | 5(10) | 88.24 | 12.31 | 45.10 | 64.29 | 47.41 |
> 0.05 | 102(88) | 56(86) | 46(90) | | | | | |
≤ 0.4 | 104(90) | 60(92) | 44(86) | 17.65 | 95.38 | 58.33 | 57.69 | 57.76 |
> 0.4 | 12(10) | 5(8) | 7(14) | | | | | |
Abbreviations: PPV, positive predictive value; NPV, negative predictive value; DA, diagnose accuracy. |
When the G index value > 1.16, it could be considered that children have significant liver fibrosis, the specificity was 87.69% and the specificity was 39.22%. This specificity suggested that the majority of children without significant liver fibrosis had a G index value ≤ 1.16. Of the 65 children with CHB without significant liver fibrosis, only 8 (12.31%) had a G index > 1.16. And the positive predictive value was 71.43%. Of the 116 children with CHB who were enrolled, 20 of 28 children with G index > 1.16 had significant liver fibrosis. The DA was 66.38 (77/116), as shown in Table 4. When the values of G index outside 0.28 and 1.16, pediatric patients could reduce liver biopsy need in up to 52% of individuals, with an overall accuracy of 75%.
Comparison of the G index model to the two pre-existing non-invasive indexes model
Using MedCale19.07 to analyze AUROCs of models that including APRI, FIB-4 and G Index (Fig. 1). The AUROCs of G Index, APRI and FIB-4 were 0.733 (95% CI 0.643–0.811), 0.680 (95% CI 0.587–0.764) and 0.601 (95% CI 0.506–0.691) respectively. According to Delong test, there was a statistical difference between the G index and FIB-4 in children with CHB (P = 0.0365) (Table 5). Although there was no significant difference between the G Index and APRI (P > 0.05). The cut-off values of APRI (Table 4) in children were 0.26 and 0.9 which were different from adults (0.5 and 1.5) 6. As for FIB-4, its AUROC was significantly less than that of G index(P < 0.05). And the optimal cut-off values (0.05 and 0.4) in children (Table 4) were also different from adults (1.45 and 3.25) 7.
Table 5
Pairwise comparison of ROC curves
Model comparison | Difference Between Areas | Standard Error | P Value | 95% Confidence Interval |
G Index and APRI | 0.733 − 0.680 = 0.053 | 0.0329 | 0.1061 | -0.0113 to 0.118 |
G Index and FIB-4 | 0.733 − 0.601 = 0.132 | 0.0633 | 0.0365* | 0.00830 to 0.256 |
APRI and FIB-4 | 0.680 − 0.601 = 0.079 | 0.0626 | 0.2068 | -0.0437 to 0.202 |
Abbreviations: APRI, aminotransferase-to-platelet ratio index; FIB-4, fibrosis index based on 4 factors. |