Network analysis of HBV‑ and HCV‑induced hepatocellular carcinoma based on Random Forest and Monte Carlo cross‑validation.

Abstract

Hepatocellular carcinoma (HCC) is one of the leading causes of cancer‑associated mortality worldwide. Hepatitis B virus (HBV) and hepatitis C virus (HCV) are two common risk factors for HCC. The majority of patients with HCC present at an advanced stage and are refractory to therapy. It is important to identify a method for efficient diagnosis at early stage. In the present study gene expression profile data, generated from microarray data, were pretreated according to the annotation files. The genes were mapped to pathways of Ingenuity Pathways Analysis. Dysregulated pathways and dysregulated pathway pairs were identified and constructed into individual networks, and a main network was constructed from individual networks with several edges. Random Forest (RF) classification was introduced to calculate the area under the curve (AUC) value of this network. Subsequently, 50 runs of Monte Carlo cross‑validation were used to screen the optimal main network. The results indicated that a total of 4,929 genes were identified in the pathways and gene expression profile. By combining dysregulated pathways with Z<0.05 and dysregulated pathway pairs with Z<0.2, individual networks were constructed. The optimal main network with the highest AUC value was identified. In the HCV group, the network was identified with an AUC value of 0.98, including 41 pairs of pathways, and in the HBV group, the network was identified with an AUC value of 0.94, including eight pairs of pathways. In addition, four pairs were identified in both groups. In conclusion, the optimal networks of HCV and HBV groups were identified with the highest AUC values. The use of these networks is expected to assist in diagnosing patients effectively at an early stage.