Data mining analysis for the investigation of clinical features of chronic hepatitis C

我们用数据挖掘的方法分析了预测聚乙二醇化干扰素联合利巴韦林治疗慢性丙型肝炎疗效的因素。我们发现，在早期治疗期间，肝脏脂肪变性、低密度脂蛋白水平、高血糖水平和高GGT水平预示着病毒学应答不良。ISDR和Core70中的丙型肝炎病毒序列是独立于其他因素的应答预测因子。2009年，IL28B附近的SNP被发现与干扰素治疗的反应最密切相关。我们发现，虽然IL28B与病毒下降的概率有关，但ISDR是独立于IL28B的SVR的预测因子，因为ISDR与肝炎复发有关。在寻找与肝癌发生相关的因素中，低白蛋白、高FIB-4指数、肝脏脂肪变性和高血糖被确定为显著的预测因素。高血糖和肝脏脂肪变性与肝细胞癌的发生发展有关，与肝纤维化无关。这些代谢特征是否可以作为治疗干预的靶点，以降低肝癌的发病率，有待进一步研究。

We have analyzed factors that are predictive of response to peg-interferon plus ribavirin in chronic hepatitis C using data mining method. We found that hepatic steatosis, lower LDL-cholesterol level, higher glucose level and higher GGT level was predictive of poor virological response during the early treatment period. The HCV sequences in the ISDR and Core 70 was predictors of response independently of other factors. In 2009, a SNP near the IL28B was found to be most closely associated with the response to interferon therapy. We found that while IL28B was associated with the probability of viral decline, ISDR was a predictor of SVR independently of IL28B since ISDR was related to relapse of hepatitis. In the search for the factors associated with HCC development, lower albumin, higher FIB-4 index, hepatic steatosis and hyperglycemia was identified as a significant predictor. Hyperglycemia and hepatic steatosis was related to HCC development independent of liver fibrosis. Whether thee metabolic features could be the target for therapeutic intervention to reduce the incidence of HCC awaits further investigation.