Piedbois Pascal, Miller Croswell Jennifer
AstraZeneca, France.
Stat Methods Med Res. 2008 Oct;17(5):519-27. doi: 10.1177/0962280207081865. Epub 2008 Feb 19.
Surrogate endpoints in oncology research and practice have garnered increasing attention over the past two decades. This activity has largely been driven by the promise surrogate endpoints appear to hold: the potential to get new therapies to seriously ill patients more rapidly. However, uncertainties abound. Even agreeing upon a definition of a "valid" surrogate endpoint has not been a straightforward exercise; this article begins by highlighting differences in how this term has been previously captured and applied, as well as laying out the basic criteria essential for its application in advanced colorectal cancer. Ideally, these elements include (but are not limited to) ease of measurement, rapid indication of treatment effect, and, most importantly, reliable and consistent prediction of the true impact of a treatment on the ultimate outcome of interest: overall survival. The strengths and weaknesses of current potential surrogate endpoints in advanced colorectal cancer, including performance status, carcinoembryonic antigen plasma level, overall response rate, time to progression, and disease-free survival, are each considered in turn. Finally, limitations of surrogate endpoints in the clinical setting, including challenges in extrapolation to new therapies, and the incomplete provision of information about potential adverse effects, are discussed. Work remains to be done between physicians and statisticians to bridge the gap between that which is statistically demonstrable and that which will be clinically useful.The term ;surrogate endpoint' was virtually unknown by most oncologists 15 years ago. A search in PubMed [http://www.ncbi.nlm.nih.gov] based on the words ;surrogate and cancer' shows that more than 2000 papers were published in medical journals in the last 20 years, with a dramatic increase of interest in the last five years. Interestingly, the same trend is observed when the words ;surrogate and heart' are entered into PubMed, suggesting that the issue of surrogate endpoints goes beyond the field of oncology, although the frequency of discussion varies (Figure 1; note different y-axis scales for oncology and cardiology).The goal of the present paper is to discuss the main issues surrounding surrogate endpoints from a clinician's point of view, using as an example surrogate endpoints of overall survival (OS) in advanced colorectal cancer (ACC).
在过去二十年中，肿瘤学研究与实践中的替代终点受到了越来越多的关注。这一活动很大程度上是由替代终点似乎所具有的前景驱动的：即有可能更快地将新疗法提供给重症患者。然而，不确定性比比皆是。甚至就“有效”替代终点的定义达成一致也并非易事；本文首先强调了该术语此前在定义和应用方面的差异，并阐述了其在晚期结直肠癌中应用所需的基本标准。理想情况下，这些要素包括（但不限于）易于测量、能快速显示治疗效果，以及最重要的是，能可靠且一致地预测治疗对最终感兴趣结局（总生存期）的真实影响。本文依次考量了晚期结直肠癌当前潜在替代终点的优缺点，包括体能状态、癌胚抗原血浆水平、总缓解率、疾病进展时间和无病生存期。最后，讨论了替代终点在临床环境中的局限性，包括外推至新疗法时面临的挑战，以及关于潜在不良反应的信息提供不完整的问题。医生和统计学家之间仍需开展工作，以弥合统计学上可证明的内容与临床有用内容之间的差距。15年前，大多数肿瘤学家对“替代终点”一词几乎闻所未闻。在PubMed[http://www.ncbi.nlm.nih.gov]上以“替代与癌症”为关键词进行搜索，结果显示在过去20年里医学期刊上发表了2000多篇论文，且在过去五年中关注度急剧上升。有趣的是，当在PubMed中输入“替代与心脏”时也观察到了相同的趋势，这表明替代终点问题超出了肿瘤学领域，尽管讨论频率有所不同（图1；注意肿瘤学和心脏病学的y轴刻度不同）。本文的目的是从临床医生的角度讨论围绕替代终点的主要问题，以晚期结直肠癌（ACC）的总生存期（OS）替代终点为例。
