Proton nuclear magnetic resonance-based metabonomic models for non-invasive diagnosis of liver fibrosis in chronic hepatitis C: Optimizing the classification of intermediate fibrosis

Table 3 Linear discriminant analysis metabonomic models, APRI and FIB-4 performances to predict significant fibrosis, advanced fibrosis and cirrhosis in 69 chronic hepatitis C patients from Pernambuco/Brazil

	Biopsy		P value	Sensitivity		Specificity		LR+		LR-		A (%)
Significant fibrosis
Model (n = 69)	F2-F4	F0-F1		(%)	95%CI	(%)	95%CI		95%CI		95%CI
≥ F2	41	2	< 0.0011	97.6	87.4-99.9	92.6	75.7-99.1	13.2	3.5-50.1	0.03	0.004-0.2	95.7
< F2	1	25
APRI (n = 68)
> 0.5	36	13	0.0012	85.7	71.5-94.6	50	29.9-0.70	1.71	1.2-2.6	0.3	0.1-0.7	72
≤ 0.5	6	13
> 1.5	19	2	0.0011	45.2	29.9-61.3	92.3	74.9-99.0	5.9	1.5-23.2	0.6	0.4-0.8	63.2
≤ 1.5	23	24
Advanced fibrosis
Model (n = 69)	F3-F4	F0-F2		(%)	95%CI	(%)	95%CI		95%CI		95%CI
≥ F3	27	2	< 0.0011	96.4	81.7-99.1	95.1	83.5-99.4	19.8	5.1-76.5	0.04	0.005-0.3	95.7
< F3	1	39
FIB-4 (n = 68)
> 1.45	25	24	0.0121	89.3	71.8-97.7	40	24.9-56.7	1.5	1.1-2.0	0.3	0.1-0.8	60.3
≤ 1.45	3	16
> 3.25	21	3	< 0.0011	75	55.1-89.3	92.5	79.6-98.4	10	3.3-30.3	0.3	0.1-0.5	85.3
≤ 3.25	7	37
Cirrhosis
Model (n = 69)	F4	F0-F3		(%)	95%CI	(%)	95%CI		95%CI		95%CI
F4	18	1	< 0.0011	100	81.5-100	98	89.6-99.9	33.83	6.9-163.7	0.033	0.002-0.4	98.6
< F4	0	50
APRI (n = 68)
> 1.00	14	16	0.0021	77.8	52.4-93.6	68	53.3-80.5	2.4	1.5-3.9	0.3	0.1-0.8	70.6
≤ 1.00	4	34
> 2.00	9	9	0.0082	50	26.0-74.0	82	68.6-91.4	2.8	1.3-5.9	0.6	0.4-1.0	73.5
≤ 2.00	9	41

¹Fisher's exact test;

²χ² test;

³Estimated value. A: Accuracy; LR+: Positive likelihood ratio; LR-: Negative likelihood ratio; APRI: Aspartate aminotransferase to platelet ratio index; FIB-4: Fibrosis index based on four factors.