TNF-α抑制剂作为类风湿关节炎首个生物制剂的头对头比较
该研究基于挪威NOR-DMARD研究数据,采用目标试验模拟和以药品成本为工具变量的方法,比较了英夫利西单抗、培塞利珠单抗、依那西普、戈利木单抗和阿达木单抗在未使用过生物制剂的类风湿关节炎患者中3个月缓解的疗效。工具变量分析提示英夫利西单抗缓解率高于培塞利珠单抗,依那西普优于戈利木单抗,但置信区间宽且多数比较无统计学显著性。调整观察性混杂因素的回归分析未显示实质性差异,结论需随机对照试验验证。
要点
- 工具变量分析提示TNFi疗效可能不等,英夫利西单抗优于培塞利珠单抗,依那西普优于戈利木单抗。
- 多数比较置信区间宽且无统计学显著性,结果不确定性高。
- 标准回归分析未显示实质性差异,提示需处理未观测混杂。
核心结论
- 该假设生成性研究提示TNFi作为RA首个生物制剂可能并非同等有效,但估计不确定性大,需随机对照试验验证。
研究设计与样本
- 基于挪威NOR-DMARD研究,纳入以TNFi作为首个生物制剂治疗的成年RA患者,采用目标试验模拟方法。
- 以国家年度招标中的药品价格作为工具变量,采用新型工具变量法估计3个月缓解的因果效应,并与调整基线变量的标准回归分析比较。
主要发现
- 工具变量分析提示英夫利西单抗相比培塞利珠单抗3个月缓解率更高。
- 工具变量分析提示依那西普优于戈利木单抗。
- 多数比较的置信区间宽且无统计学显著性,敏感性分析确认结果总体方向。
- 调整观察性混杂因素的回归分析未显示实质性差异。
临床怎么用
- 当前证据不足以支持在未使用过生物制剂的RA患者中优先选择某种TNFi,临床决策应个体化。
局限性
- 估计不确定性大,多数比较无统计学显著性,研究为假设生成性。
- 观察性设计可能存在未观测混杂,工具变量分析结果需随机对照试验验证。
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
OBJECTIVES: TNF-α inhibitors (TNFis) are effective biologic drugs for rheumatoid arthritis (RA), but their comparative efficacy is unclear due to limited direct comparisons. The objective of this work was to compare the efficacy of infliximab (INX), certolizumab pegol (CZP), etanercept (ETN), golimumab (GOM) and adalimumab (ADM) in achieving remission at 3 months in bio-naïve RA patients. METHODS: Using data from the NOR-DMARD study of adult RA patients in Norway starting a TNFi as their first biologic treatment, we employed a target trial emulation approach. TNFi selection is mainly determined by drug prices in annual national tenders, allowing us to use drug cost as an instrumental variable (IV). Applying a novel IV method, causal effects were estimated with remission at 3 months as the end point. Results were compared with standard regression analysis adjusted for baseline variables. RESULTS: The IV analysis suggested potential differences in efficacy among TNFis, with INX showing higher remission rates compared with CZP, and ETN performing better than GOM. However, confidence intervals were wide, and many comparisons were not statistically significant. Sensitivity analyses confirmed the overall direction of results. Regression analysis adjusting for observed confounders showed no substantial differences, underscoring the importance of addressing unobserved confounding effects. CONCLUSION: This hypothesis-generating study provides evidence that TNFis may not be equally effective. The uncertainty in the estimates highlights the need for a pragmatic randomized controlled trial to validate the findings. With decreasing costs of TNFis, such studies are increasingly feasible and critical to guide clinical decision-making.
引用
- Kazemi AA, Kristianslund EK, Sexton J, et al. A head-to-head comparison of TNF-α inhibitors as the first biologic treatment of rheumatoid arthritis. Rheumatology (Oxford). 2026. PMID: 41533706. doi:10.1093/rheumatology/keag007
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。