Prospective validation of a noninvasive index for predicting liver fibrosis in hepatitis C virus-infected liver transplant recipients

Abstract

We previously developed a mathematical model, the Hospital Universitario La Fe (HULF) index, as an alternative to protocol liver biopsy (PLB) to estimate significant fibrosis (SF) in patients who underwent liver transplantation (LT) for liver damage caused by chronic HCV infection. In the present study, we sought to validate this noninvasive index. The commonly derived clinical and laboratory data for calculating the HULF index were prospectively collected over 2.7 years from patients undergoing LT and PLB. The sensitivity, specificity, positive and negative predictive values, and diagnostic capacity were evaluated with receiver operating characteristic curve analysis. Biopsy was performed 93 times in 86 LT patients. The prevalence of SF (F3-F4 on the Knodell scoring system) was 32%. The intraobserver and interobserver concordance was high (kappa = 0.94 and kappa = 0.75, respectively) in identifying SF in PLB. For low scores, the HULF index discarded an SF diagnosis with a sensitivity of 90% and a negative predictive value of 89%. The area under the receiver operating characteristic curve was 0.68. The precision of the HULF index did not improve with the incorporation of donor age and body mass index into the multivariate analysis. Applying the index would have prevented 24% of the biopsy procedures performed. In conclusion, the HULF index was prospectively validated with data commonly obtained in standard clinical practice. Because the index distinguishes a subgroup of HCV LT patients with a low probability of having SF, PLB would be avoided in those patients.