Receiver operating characteristics curve analysis of factors predictive of nonresponse to interferon therapy in patients with chronic hepatitis C.

OBJECTIVES: 1) to identify pretreatment variables predictive of nonresponse to interferon-alpha (IFN-alpha) in patients with chronic hepatitis C, and 2) to establish a prognostic index in these groups using receiver operating characteristics curve analysis. METHODS: 132 patients were treated with IFN-alpha at a dose of 3 megaunits three times a week for 3-12 months. The response was compared in patients with a complete response vs nonresponders, and patients with a sustained response vs nonresponders plus relapsers. Factors predictive of response were identified by analyzing clinical, biochemical, virological and histological variables. RESULTS: The sustained response rate was 12.8% at 24 months of follow-up. The pretreatment characteristics with a predictive value (PV) according to area under the ROC curve and 95% confidence interval > 0.5 were age, known duration of infection, history of transfusion, GGT, serum ferritin levels, viral load, genotype, and grade and stage of the histological lesion. The positive PV (the probability of predicting absence of response when the variable is present) was notably greater than the negative PV (mean: 94.9% vs 24.8%, respectively). In addition, when 4 and 6 variables were present, the positive PV was 100% and sensitivity was 60.2% and 22.1%, respectively. The predictive variables independently associated with an absence of response were genotypes 1, 4 and 5, GGT > 24 IU/l and grade of the histological lesion > 6. CONCLUSIONS: It was possible to predict the absence of both primary and posttreatment response with an acceptable degree of reliability.