Background:  Simple, non-invasive markers of hepatic fibrosis are needed to assess the risks/benefit of HCV treatment in coinfected patients. The objective was to validate 2 previously described models and to develop a novel index to predict significant fibrosis.

Methods:  We analyzed correlates of significant fibrosis among 218 patients in the Johns Hopkins HIV clinic who had a liver biopsy and complete data available. Histology was scored by a single pathologist according to Ishak fibrosis stage (F0-6). Analyses were designed to differentiate patients with no or minimal fibrosis (≤F2) from those with significant fibrosis (≥F3). Univariate associations were examined using χ²-tests and Mann–Whitney tests; a predictive model was constructed by multivariate modeling of the independent variables. The areas under the receiver operating characteristic (AUROC) curve was calculated for this and others models, AST to platelet ratio index (APRI = AST/ULN)*100/platelet count) and FIB-4 (= age [years]* AST /(platelet count)*(ALT)½.)

Results:  The median age, 49 years; male, 67%; black, 83%; injecting drug users (IDU), 76%; clinically diagnosed alcohol abuse (past or active), 40%; median CD4 cell count, 345/mm³; median HIV RNA level, 309 copies/mL. Of 218 subjects, 55 (25%) had ≥F3. Patients with ≥F3 were more likely to abuse alcohol (54% >35%), have higher ALT, AST and total bilirubin levels, and lower platelet count and albumin. Significant fibrosis was not associated with age, sex, hyperglycemia, antiretroviral use, CD4 cell count, or HIV RNA level. In multivariate analysis, ≥F3 was independently associated: AST >1.25 x ULN (OR 4.2, 95%CI 1.8 to 10.0), platelet count <150,000/mm³ (3.7, 1.6 to 8.6), albumin <3.5 g/dL (2.3, 1.0 to 5.2) and alcohol abuse (3.5, 1.7 to 7.6). Regression model: risk score = –3.06 + 1.43 (if AST >1.25 x ULN, otherwise 0) + 1.3 (if platelet count <150,000; otherwise 0) + 0.81 (if albumin <3.5; otherwise 0) + 1.26 (if alcohol abuse; otherwise 0). The AUROC curve:  Johns Hopkins Fibrosis Index, 0.79; APRI, 0.76; FIB-4, 0.74. 

Conclusions:   In this urban, HIV clinic, categorical assessment of routine laboratory (albumin, AST and platelet count) and clinical data (alcohol abuse) accurately predicted significant fibrosis in HCV-infected adults. The utility of Johns Hopkins Fibrosis Index and other non-invasive models should be evaluated in other clinical settings. If validated, such indices may have a role in identifying patients with significant liver disease in settings in which access to liver biopsy is limited.