乙型肝炎病毒相关性肝细胞癌预测模型的验证

Validation study of prediction models of hepatitis B virus-related hepatocellular carcinoma

目的验证已建立的乙型肝炎病毒相关性肝细胞癌预测模型（REACH—B评分模型及CU—HCC评分模型）。方法收集2004年10月1日至2014年5月1日初次人院、HBsAg阳性超过半年的患者，分为病例组（肝癌组）和对照组（非肝癌组），回顾性收集3年前及5年前相关指标，进行评分评估。据资料不同采用t检验、秩和检验、石。检验或受试者工作特征曲线进行统计学分析。结果共627例患者纳入3年乙型肝炎病毒相关性肝细胞癌预测REACH—B评分模型验证，其中病例组151例，对照组476例。REACH—B评分模型预测3年肝癌发生的曲线下面积为0．78，敏感度为73．00％，特异度为78．700／0。男性并ALT≥45U／L患者应用REACH-B评分模型对3年肝细胞癌发生的预测价值曲线下面积分别为0．89，敏感度为87．09％，特异度为83．86％。共159例患者纳入5年乙型肝炎病毒相关性肝细胞癌预测REACH—B评分模型及CU—HCC评分模型验证，其中病例组65例，对照组94例。REACH—B评分模型预测5年肝癌发生的曲线下面积为0．79，敏感度为73．60％，特异度为75．43％。CU—HCC评分模型预测5年肝癌发生的曲线下面积为0．76，敏感度为78．40％，特异度为77．40％。结论REACH—B评分模型和CU-HCC评分模型均可适用本地区人群;男性并ALT≥45U／L患者应用REACH—B评分模型对3年乙型肝炎病毒相关性肝细胞癌发生有较高的预测价值。

Objective To validate two previously published models （REACH-B score and CU-HCC score） for predicting the risk of developing hepatitis B virus （HBV）-related hepatocellular carcinoma （HCC）. Methods In-patients of the Liver Center of First Affiliated Hospital Fujian Medical University who tested positive for hepatitis B surface antigen （HBsAg; 〉 6 months） and were admitted for treatment between October 1, 2004 and May 1, 2014 were enrolled for study. The 627 study participants were grouped according to presence of HCC （151 in the HCC case group, and 476 in the non-HCC control group）. Relevant clinical data from 3 and 5 years prior to the current hospital admission were collected retrospectively and assessed using the REACH-B and CU-HCC scoring systems. A subset of the study participants （65 HCC cases, and 94 non-HCC controls） was used for the verification analysis of prediction for 5-year risk of HBV-related HCC. T-test, rank sum test, chi- square test and the receiver operating characteristic curve were used for statistical analyses. Results For the REACH-B score, prediction of 3-year risk of developing HCC had an area under the curve （AUC） of 0.78, a sensitivity of 73.00% and a specificity of 78.70%. In male patients with alanine aminotransferase （ALT） ≥ 45 U/L, the REACH-B score prediction of 3-year risk of developing HCC had an AUC of 0.89, a sensitivity of 87.09% and a specificity of 83.86%. The REACH-B score prediction of 5-year risk of HCC had an AUC of 0.79, a sensitivity of 73.60% and a specificity of 75.53%; the CU-HCC score prediction of 5-year risk of HCC had an AUC of 0.76, a sensitivity of 78.40% and a specificity of 77.40%. Conclusion Both the REACH-B and CU- HCC scoring systems can be used for HCC prediction among patients at the First Affiliated Hospital Fujian Medical University. For male patients with ALT ≥ 45 U/L, the REACH-B score may be a more sensitive predictor for 3-year risk of developing HBV-related HCC.