基于倾向指数匹配法的肝癌患者生存分析

摘要

Objective Evaluate the effects of radiofrequency ablation (RFA) and surgical resection ( SR) to Hepatocellular Carcinoma (HCC) patients between the two groups after balancing the covariates by use of matching method based on propensity score. Methods Using cali-per matching method based on propensity score to match the two groups of HCC(Hepatocellular Carcinoma) patients. After matching a new sample with covariates balanced between the two groups will be obtained. And then use the new sample to make survival analysis. Results Age.child-pugh. family history and HBsAg become balance between the groups after caliper matching based on propensity score,which are imbalance before matching. Before matching, the COX regression analyzing shows us that the Chlid-pugh,family history with HCC,positive HbsAg,over highed AFP and GT are the dangerous factors effecting the mortality rate of HCC patients (HR ( Harvard Ratio) ＞ 1, P ＜ 0.05). But there are no significant differences of the mortality rates between the two groups ( P = 0. 202). After matching, the one-year survival rate of RFA and SR are 33. 24％ ,44. 37％ , respectively. The two-year survival rates are 12.75％ , 16. 70％ , respectively. The median survival time is 0. 66 year and 0. 84 year, respectively. There are significant differences of survival rate between the groups ( P = 0.032 ). Conclusion The therapeutic effects of SR are superior to RFA.%目的 用倾向指数匹配法均衡射频消融术( RFA)和外科切除术(SR)两组中期肝癌病人组间的协变量后,评价两种治疗方式的效果.方法 应用倾向指数卡钳匹配法对两组肝癌病人进行匹配,得到两组间各协变量均衡的样本,并用新样本做生存分析.结果 使用卡钳匹配法对肝癌病人两组间的协变量进行平衡,匹配前年龄、child分级、家族史、是否有乙肝在两组间的不均衡在匹配以后达到了均衡.匹配之前,Cox回归分析表明,Child分级高、有肝癌家族史、HbsAg阳性、过高的AFP及GT等是影响病人死亡率的危险因素(风险比HR＞1,P＜0.05),但是射频消融术和外科切除术两组间死亡率的差异无统计学意义(P =0.202);匹配之后,射频消融术组和外科切除术组的1年生存率分别为33.24％、44.37％,2年生存率分别为12.75％、16.70％,中位生存时间分别为0.66年和0.84年,两组间生存率的差异有统计学意义(P =0.032).结论 外科切除术的治疗效果优于射频消融术的治疗效果.