Received:11 August 2017Accepted:25 October 2017Published online:15 November 2017
AbstractInterferon-alpha (IFN-α) therapy of chronic hepatitis B (CHB) patients is constrained by limited response and side effects. We described a panel of circulating microRNAs (miRNAs) which could potentially predict outcome of IFN-α therapy. Here, we report development of a simplified scoring model for personalized treatment of CHB patients. 112 CHB patients receiving IFN-α treatment were randomly divided into a training (n = 75) or a validation group (n = 37). The expression of 15 candidate miRNAs was detected in training group with 5 miRNAs exhibiting significantly different levels (p < 0.0001) between early virological response (EVR) and non-early virological response (N-EVR). These 5 miRNAs were further tested in validation phase. Refinement analyses of results from training phase established a model composed of miR-210, miR-22 and alanine aminotransferase (ALT), with area under ROC curve (AUC) of 0.874 and 0.816 in training and validation groups, respectively. In addition, this model showed prognostic value for sustained virological response (SVR) (AUC = 0.821). Collectively, this simplified scoring model composed of miR-210, miR-22 and ALT can reproducibly predict the EVR and SVR of IFN-α therapy in CHB patients. The model should help to forecast the outcome of IFN-α treatment prior to therapy decision involving nucleoside analogs or IFNs.