PD-1/PD-L1抑制剂治疗非小细胞肺癌的成本-效用分析方法学系统评价

费正洋 张雪珂 王樱澄 陈平钰 芮明军 马爱霞

中圖分类号 R956 文献标志码 A 文章编号 1001-0408（2021）20-2499-10

DOI 10.6039/j.issn.1001-0408.2021.20.11

摘 要 目的：为提高我国细胞程序性死亡受体1（PD-1）/细胞程序性死亡配体1（PD-L1）抑制剂治疗非小细胞肺癌相关药物经济学研究的质量提供依据。方法：计算机检索Embase、PubMed、Medline、Cochrane图书馆、中国知网、万方数据、维普网等中英文数据库，搜集PD-1/PD-L1抑制剂治疗非小细胞肺癌的成本-效用研究，检索文献的发表时间为2016年1月-2021年1月。对纳入研究进行资料提取，采用卫生经济学评价报告标准共识清单对其进行质量评价后，从模型框架、模型参数、不确定性分析等方面入手对方法学相关数据进行总结和对比。结果与结论：最终纳入17项研究，研究整体质量较高，但在方法学上有着较大的差异。16项研究采用了基于三状态的马尔可夫模型或分区生存模型;研究时限最短为5年，最长为终身;模型周期最短为1周，最长为6周;8项研究采用了标准参数分布的方法进行参数拟合，7项研究同时结合了KM曲线或样条模型等其他参数估计方法;有11项研究进行了模型外推验证;所有研究均仅考虑了直接医疗成本，并以质量调整生命年为产出指标报告了对应的增量成本-效果比;16项研究同时进行了确定型敏感性分析和概率敏感性分析以提高模型结果的稳健性。建议未来我国的相关研究应保证报告格式的完整性，选择合适的阳性对照方案，根据可获得的数据形式选择经济学评价模型并建立合理的研究假设，采用Cholesky分解等方法探索参数拟合的不确定性，结合外部数据进行外推验证，并在缺乏头对头临床试验时采取适当的间接比较方法，以提高我国相关药物经济学研究质量。

关键词 细胞程序性死亡受体1/细胞程序性死亡配体1抑制剂;非小细胞肺癌;经济学评价模型;系统评价;方法学

Systematic Review of Cost-utility Analysis Methodology for PD-1/PD-L1 Inhibitors in the Treatment of Non-small Cell Lung Cancer

FEI Zhengyang1，ZHANG Xueke1，WANG Yingcheng1，CHEN Pingyu1，2，RUI Mingjun1，MA Aixia1，2（1. School of International Pharmaceutical Business， China Pharmaceutical University， Nanjing 211198， China; 2. Center for Pharmacoeconomics and Outcomes Research， China Pharmaceutical University， Nanjing 211198， China）

ABSTRACT   OBJECTIVE： To provide reference for improving the quality of programmed cell death protein 1 （PD-1）/programmed cell death 1 ligand （PD-L1） inhibitors in the treatment of non-small cell lung cancer related pharmacoeconomic studies in China. METHODS： Retrieved from Embase， PubMed， Medline， Cochrane Library， CNKI， Wanfang database， VIP and other Chinese and English database， cost-utility studies about PD-1/PD-L1 inhibitors in the treatment of non-small cell lung cancer published during Jan. 2016-Jan. 2021 were collected. The data of the included studies were extracted. After the quality of the included studies was evaluated by using the Consolidated Health Economic Evaluation Reporting Standards list， the relevant data were summarized and compared from the aspects of model framework， model parameters and uncertainty analysis. RESULTS & CONCLUSIONS： A total of 17 studies were finally included， the overall quality of them was high but the differences in methodology were great. Markov model or partition survival model based on three states was adopted for 16 studies. The time horizon ranged from 5 years to lifetime; the cycle length ranged from 1 week to 6 weeks. A total of 8 studies used the standard parameter distribution method for parameter fitting， and 7 studies additionally adopted other parameters estimation methods as KM curves or spline models. Eleven studies performed the validation of model extrapolation. All studies considered the direct medical costs and reported the incremental cost-effectiveness ratio using quality-adjusted life years as the health outcome. Sixteen studies conducted the deterministic sensitivity analysis and probabilistic sensitivity analysis to improve the stability of the model. It is suggested that studies should keep the integrity of the report format， choose the appropriate positive comparators， select the health economic model and construct reasonable assumptions according to the available data format， use Cholesky decomposition to explore the uncertainty of the parameter fitting， perform the validation of extrapolation combined with external data and use the appropriate indirect comparison in the absence of the head-to-head clinical trials to improve the quality  of related pharmacoeconomic studies in China.