Retrospective Study
Copyright ©The Author(s) 2025.
World J Hepatol. Mar 27, 2025; 17(3): 104534
Published online Mar 27, 2025. doi: 10.4254/wjh.v17.i3.104534
Table 1 Characteristics of participants
Characteristics
Included patients (n = 103)
Demographic
Age (years), median (IQR)53 (38-61)
Female86 (83)
BMI, median (IQR)26 (21-28)
Baseline histological features (METAVIR)
Biopsy specimen length (mm), median (IQR)12 (11-15)
F010 (10)
F128 (27)
F221 (20)
F324 (23)
F420 (19)
Fibrosis ≥ F265 (63)
Fibrosis ≥ F344 (43)
Cirrhosis F420 (19)
A010 (10)
A130 (29)
A242 (41)
A321 (20)
Activity ≥ A263 (61)
Steatosis > 012 (12)
Biopsy diagnosis
AIH83 (81)
AIH and MASLD1 (1)
AIH and PBC19 (18)
NITs for significant fibrosis (≥ F2)
FibroTest > 0.4855 (53)
FIB-4 ≥ 1.366 (64)
VCTE ≥ 8 kPa61 (59)
NITs for advanced activity (≥ A2)
ActiTest > 0.5248 (47)
ALT ≥ 5063 (61)
Other serum components, median (IQR)
ALT (IU/L)67 (37-160)
AST (IU/L)52 (34-117)
GGT (IU/L)73 (31-170)
Total bilirubin, μmol/L16.7 (11-25)
Apolipoprotein-A1 (g/L)1.57 (1.25-1.80)
Haptoglobin (g/L)0.58 (0.56-1.30)
Alpha-2 macroglobulin (g/L)2.34 (1.93-2.88)
Platelets (g/L)201 (156-266)
Table 2 Test performance according to statistical methods by features
Features and test
Obuchowski weighted AUROC mean (standard error)
Standard binary AUROC mean (95%CI)
Fibrosis5 stages, 10 comparisonsF2, F3, F4 vs F0, F1F3, F4 vs F0, F1, F2F4 vs F0, F1, F2, F3
FibroTest0.919 (0.014)0.83 (0.73-0.90)0.83 (0.73-0.90)0.80 (0.68-0.88)
VCTE0.933 (0.014)0.86 (0.77-0.92)0.87 (0.78-0.93)0.87 (0.78-0.94)
FIB-40.882 (0.019)0.71 (0.60-0.80)0.75 (0.63-0.83)0.80 (0.67-0.89)
Activity4 stages, 8 comparisonsA2, A3 vs A0, A1--
ActiTest0.921 (0.016)0.86 (0.76-0.92)--
ALT0.905 (0.019)0.83 (0.73-0.90)--
Table 3 Diagnostic performance of biopsy, FibroTest, fibrosis-4 index, and vibration-controlled transient elastography for significant fibrosis (F2, F3, and F4) and advanced fibrosis (F3 and F4) compared with that of the gold standard model and Bayesian latent class model
Parameters and cutoffs
Biopsy assumed as perfect reference
Bayesian LCM
Gold standard model vs LCM
Number of patients103103Difference (Z-test P-value)
Prevalence significant fibrosis63.1 (53.0-72.2)66.2 (56.0-77.0)3.1% increase (0.67)
Biopsy stage METAVIR ≥ F2
Sensitivity10095.4 (83.1-100)4.6% decrease (NA)
Specificity100100 (100-100)NA (NA)
Positive predictive value100100 (100-100)NA (NA)
Negative predictive value10091.9 (67.3-99.9)8.1% decrease (NA)
FibroTest with cutoff of > 0.48
Sensitivity73.8 (61.2-83.6)73.1 (61.5-82.6)0.7% decrease (0.93)
Specificity81.6 (65.1-91.7)86.1 (70.6-98.4)4.5% increase (0.69)
Positive predictive value87.3 (74.9-94.3)91.1 (79.9-99.1)3.8% increase (0.65)
Negative predictive value64.6 (49.4-77.4)62.0 (45.7-75.4)2.6% decrease (0.82)
FIB-4 with cutoff of ≥ 1.3
Sensitivity69.2 (56.4-79.8)70.0 (58.6-80.2)0.8% increase (0.93)
Specificity44.7 (29.0-61.5)49.8 (40.8-66.1)5.1% increase (0.58)
Positive predictive value68.2 (55.4-78.8)73.5 (61.5-85.0)5.3% increase (0.55)
Negative predictive value45.9 (29.8-62.9)46.1 (31.2-61.9)0.2% increase (0.99)
VCTE with LSM cutoff of ≥ 8 kPa
Sensitivity81.5 (69.6-89.7)80.7 (69.9-89.2)0.8% decrease (0.92)
Specificity78.9 (62.2-89.9)83.6 (67.2-97.2)4.7% increase (0.69)
Positive predictive value86.9 (75.2-93.8)90.6 (79.5-98.7)3.7% increase (0.65)
Negative predictive value71.4 (55.2-83.8)68.8 (50.6-82.1)2.6% decrease (0.83)
Prevalence advanced fibrosis42.7 (33.1-52.8)50.2 (37.1-65.2)7.5% increase (0.37)
Biopsy stage METAVIR ≥ F3
Sensitivity10085.2 (64.6-99.8)14.8% decrease (NA)
Specificity100100 (100-100)NA (NA)
Positive predictive value100100 (100-100)NA (NA)
Negative predictive value10087.0 (63.7-99.9)13% decrease (NA)
FibroTest with cutoff of > 0.58
Sensitivity70.5 (54.6-82.8)66.1 (51.2-80.1)4.4% decrease (0.69)
Specificity83.1 (70.6-91.1)86.4 (74.2-96.3)3.3% increase (0.71)
Positive predictive value75.6 (59.4-87.1)83.3 (66.6-95.9)7.7% increase (0.52)
Negative predictive value79.0 (66.5-87.9)71.9 (53.2-85.2)7.1% decrease (0.54)
FIB-4 with cutoff of ≥ 2.67
Sensitivity38.6 (24.7-54.5)38.2 (25.0-52.3)0.4% decrease (0.97)
Specificity88.1 (76.5-94.7)91.7 (79.9-99.5)3.6% increase (0.67)
Positive predictive value70.8 (48.8-86.6)82.4 (56.5-99.2)11.6% increase (0.50)
Negative predictive value65.8 (54.2-75.9)59.8 (44.2-72.6)6% decrease (0.55)
VCTE with LSM cutoff of ≥ 9.5
Sensitivity84.1 (69.3-92.8)84.0 (72.2-92.9)0.1% decrease (0.94)
Specificity72.9 (59.5-83.3)81.7 (65.2-98.8)8.8% increase (0.42)
Positive predictive value69.8 (55.5-81.3)82.2 (62.4-99.0)12.4% increase (0.32)
Negative predictive value86.0 (72.6-93.7)83.5 (69.4-92.9)2.5% decrease (0.80)
Table 4 Comparison of the diagnostic performance of biopsy, ActiTest, and alanine aminotransferase for advanced activity (A2 and A3) with the gold standard model and Bayesian latent class model
Parameters and cutoffs
Biopsy assumed as perfect reference
Bayesian LCM
Gold standard model vs LCM
Number of patients103103Difference (Z-test P-value)
Prevalence advanced activity61.2 (51.0-70.5)47.8 (37.7-58.3)13.4% decrease (0.07)
Biopsy stage METAVIR ≥ A2
Sensitivity10095.8 (87.5-99.6)4.2% decrease (NA)
Specificity10070.9 (57.4-82.3)29.1% decrease (NA)
Positive predictive value10075.1 (62.7-85.6)24.9% decrease (NA)
Negative predictive value10094.8 (84.5-99.5)5.2% decrease (NA)
ActiTest with cutoff of > 0.52
Sensitivity73.0 (60.1-83.1)97.3 (85.8-100)24.3% increase (0.006)
Specificity95.0 (81.8-99.1)99.4 (93.4-100)4.4% increase (0.59)
Positive predictive value95.8 (84.6-99.3)99.3 (92.6-100)3.5% increase (0.60)
Negative predictive value69.1 (55.0-80.5)97.7 (86.6-100)28.6% increase (0.002)
ALT with cutoff of ≥ 50
Sensitivity81.0 (68.7-89.4)99.5 (95.0-100)18.5% increase (0.006)
Specificity70.0 (53.3-82.9)74.3 (61.6-85.7)4.2% increase (0.69)
Positive predictive value81.0 (68.7-89.4)78.0 (65.8-88.4)3.0% decrease (0.73)
Negative predictive value70.0 (53.3-82.9)99.4 (94.0-100)29.4% increase (0.006)