Validação e comparação de testes laboratoriais simples como preditores de fibrose hepática em portadores de hepatite C crônica

Abstract

Background: hepatitis C virus (HCV) chronic infection is a relevant cause of progressive liver disease, leading to cirrhosis worldwide. Liver biopsy is the gold standard to assess hepatic fibrosis but non-invasive fibrosis markers are needed. The aspartate aminotransferase to platelet ratio index (APRI) have demonstrated good accuracy to predict significant fibrosis and cirrhosis. The Gotemborg University Cirrhosis Index (GUCI) and the Lok´s model may predict cirrhosis but lack specific points to predict significant fibrosis and external validation. We sought to validate these models in patients with HCV and to investigate new points, including predictive of significant fibrosis with GUCI and Lok models.Methods: 248 patients with liver biopsy (Ishak´s score) were included. Sensitivity(Se), specificity(Sp), positive(PPV) and negative(NPV) predictive values and likelihood ratios(LR) were calculated. Analysis of receiver operating characteristic (ROC) curves provided the new points. Areas under ROC curves (AUC) were calculated. Results: 140(56%) patients were male, mean age was 51 years. Fibrosis stages were F0- F2 in 79(32%), F3-F4 in 67(27%) and F5-F6 in 102(41%) patients. Using original points for prediction of significant fibrosis, Se and Sp to APRI were 78% and 97%, respectively. PPV and LR were 99% and 30. Se, Sp, PPV and NPV for cirrhosis were: 63%, 90%, 77%, 82% (APRI), 79%, 67% 63%, 83% (GUCI), 96%, 46%, 67% and 91% (Lok), respectively. AUC for APRI, GUCI and Lok models were 0.83, 0.83 and 0.81 to predict significant fibrosis and 0.82, 0.82, and 0.81 to predict cirrhosis, respectively. Novel cut off points to predict significant fibrosis were: APRI: <0.45/>1.2; GUCI: <0.45/>1.4 and Lok: <0.2/>0.6; and to predict cirrhosis were: APRI: <0.42/>2.6, GUCI: <0.46/>2.0 and Lok: <0.3/>0.76. Total number of classified patients ranged from 50% to 60%. These new points provided the following results: for prediction of significantfibrosis: APRI: Se, 88%; Sp, 89%; PPV, 96% and NPV: 72%; GUCI: Se, 88%; Sp, 89%; PPV, 96% and NPV, 72%; Lok, Se, 89%; Sp, 83%; PPV, 94% and NPV, 71%. For prediction of cirrhosis: APRI: Se, 90%, Sp, 86%; PPV, 79 and NPV, 94%; GUCI: Se, 97%, Sp, 78%; PPV, 81% and NPV, 96%; Lok: Se, 84%; Sp: 89%; PPV: 82% and NPV, 90%. The AUC for significant fibrosis were for APRI, GUCI e Lok: 0,83; 0,83 e 0,81; respectively and for prediction of cirrhosis: 0,82; 0,82 e 0,81; respectively.Conclusions: the validation set confirmed the performance for APRI and Lok´s model previously reported. GUCI did not perform as well as in the original report for evaluating cirrhosis. The novel cut-off points support the good performance potential of the three models for evaluation of fibrosis stage, mostly to confirm significant fibrosis and to exclude cirrhosis. They also suggest that GUCI and Lok´s model may be used not only for cirrhosis but also for significant liver fibrosis. These models might be used as surrogate markers to select patients for HCV treatment.