RESUMO
OBJECTIVE: Noninvasive tests that can be used in place of liver biopsy to diagnose fibrosis have major limitations. They either leave a significant proportion of patients without a definitive diagnosis or produce inaccurate results. Moreover, the performance of these tests is lower in HIV/hepatitis C virus (HCV) coinfection. Against this background, we examined the utility of serum matrix metalloproteinase 2 (MMP-2) and tissue inhibitor of metalloproteinase 1 (TIMP-1) measurements in combination with routine clinical data to predict fibrosis in HIV/HCV-coinfected patients. METHODS: Patients with a liver biopsy who had not received anti-HCV therapy were included in the study. A model including variables independently associated with fibrosis was constructed. Diagnostic accuracy was determined by measuring the area under the receiver operating characteristic curve (AUROC). Positive (PPV) and negative (NPV) predictive values were calculated. RESULTS: Ninety patients were included in the study. Aspartate aminotransferase (AST), platelet count and MMP-2 were predictors of significant fibrosis (F≥2) and cirrhosis (F4). A score constructed using these variables yielded an AUROC of 0.76 for F≥2 and 0.88 for F4. Score cut-offs detected (value ≥3.5) and excluded (value ≤1.5) F≥2 with a PPV of 87% and an NPV of 88%. Thirty-one patients (34%) were correctly diagnosed using these cut-offs, with four (13%) incorrect classifications. Cirrhosis was excluded with a certainty of 98% and diagnosed with a probability of 83%. Two (17%) of 12 patients were misclassified as having cirrhosis. The AST to platelet count index and MMP-2 levels were sequentially applied to detect F≥2. Forty-one patients (46%) were identified with this approach, with six (15%) misclassifications. CONCLUSION: MMP-2 levels can be used in combination with AST and platelet count to aid the diagnosis of liver fibrosis in HIV/HCV-coinfected patients.