文献速递消化道肿瘤与早癌肝病

接受阿替利珠单抗联合贝伐珠单抗治疗的肝细胞癌患者风险预测模型的开发与验证

Development and validation of a risk prediction model for patients with hepatocellular carcinoma receiving atezolizumab-bevacizumab.
PubMed 文献 · Hepatology · 2025-06-30

该研究基于韩国多中心接受一线阿替利珠单抗联合贝伐珠单抗(AB)的晚期肝细胞癌患者,构建并验证了CRAPT-M评分模型,用于预测总生存期。模型将患者分为低、中、高风险组,三组中位OS分别为22.4、12.9和6.7个月,预测准确性优于CRAFITY模型并在外部验证队列中表现稳健。

要点

  • CRAPT-M模型基于5个独立预后因素:CRP≥1.0 mg/dL、白蛋白<3.5 g/dL、PIVKA-II≥1500 mAU/mL、总胆红素≥1.0 mg/dL和大血管侵犯。
  • 低风险(评分≤4)、中风险(5-12)和高风险(≥13)组的中位OS分别为22.4、12.9和6.7个月(p<0.001)。
  • CRAPT-M在12、24、36个月的时间依赖AUC分别为0.785、0.737和0.742,预测准确性持续高于CRAFITY模型。

核心结论

  • CRAPT-M模型对接受AB治疗的晚期肝细胞癌患者表现出稳健的总生存期预测能力,为预后评估和临床决策提供了有价值的工具。

研究设计与样本

  • 多中心研究,纳入韩国接受一线阿替利珠单抗联合贝伐珠单抗治疗的晚期肝细胞癌患者,形成推导队列(n=456)和验证队列(n=205)。

主要发现

  • 多变量分析识别出5个独立预后因素:C反应蛋白≥1.0 mg/dL(HR 2.07;p<0.001)、白蛋白<3.5 g/dL(HR 1.60;p=0.002)、维生素K缺乏或拮抗剂-II诱导蛋白≥1500 mAU/mL(HR 1.60;p=0.002)、总胆红素≥1.0 mg/dL(HR 1.50;p=0.006)和大血管侵犯(HR 1.49;p=0.009)。
  • 低风险组(评分≤4)中位OS为22.4个月(95%CI 18.6-25.0),中风险组(5-12)为12.9个月(95%CI 8.7-14.8),高风险组(≥13)为6.7个月(95%CI 5.1-7.7),组间差异显著(p<0.001)。
  • CRAPT-M在12、24和36个月的时间依赖ROC曲线下面积分别为0.785、0.737和0.742,持续高于CRAFITY模型。
  • 模型在外部验证队列中表现出稳健的预测性能,校准良好,敏感性分析中判别能力一致。

关键数值与推荐

  • CRAPT-M评分将患者分为低风险(≤4分)、中风险(5-12分)和高风险(≥13分)。

临床怎么用

  • 作者认为CRAPT-M模型可作为接受AB治疗的晚期肝细胞癌患者预后评估和临床决策的有价值工具。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND AND AIMS: Atezolizumab plus bevacizumab (AB) has become the standard first-line treatment for advanced HCC. However, identifying reliable prognostic biomarkers remains a critical challenge. We aimed to develop a comprehensive scoring system to predict overall survival (OS) in advanced HCC patients receiving first-line AB. APPROACH AND RESULTS: We included patients with advanced HCC receiving first-line AB from multiple centers in Korea, forming a derivation cohort ( n =456) and a validation cohort ( n =205). Multivariable analysis identified 5 independent prognostic factors: C-reactive protein ≥1.0 mg/dL (HR 2.07; p <0.001), albumin <3.5 g/dL (HR 1.60; p =0.002), protein induced by vitamin K absence or antagonist-II ≥1500 mAU/mL (HR 1.60; p =0.002), total bilirubin ≥1.0 mg/dL (HR 1.50; p =0.006), and macrovascular invasion (HR 1.49; p =0.009). We developed the CRAPT-M model, named after these factors' initial letters. Patients were categorized into low (≤4), intermediate (5-12), and high (≥13) risk groups by CRAPT-M score. Median OS differed significantly: 22.4 (95% CI, 18.6-25.0), 12.9 (95% CI, 8.7-14.8), and 6.7 (95% CI, 5.1-7.7) months for low-risk, intermediate-risk, and high-risk groups, respectively ( p <0.001). Time-dependent area under the receiver operating characteristic for CRAPT-M demonstrated consistently higher predictive accuracy than the CRAFITY model, with values of 0.785, 0.737, and 0.742 at 12, 24, and 36 months, respectively. The model demonstrated robust predictive performance in the external validation cohort, with excellent calibration and consistent discrimination across sensitivity analyses. CONCLUSIONS: The CRAPT-M model demonstrated robust OS prediction, offering a valuable tool for prognosis estimation and clinical decision-making in advanced HCC patients receiving AB.

来源:Hepatology,查看原文。版权归原作者所有。

引用

  1. Nam H, Kim DY, Kim DY, et al. Development and validation of a risk prediction model for patients with hepatocellular carcinoma receiving atezolizumab-bevacizumab. Hepatology. 2025. PMID: 40587822. doi:10.1097/HEP.0000000000001444

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。