Freidlin Boris, Korn Edward L, Hunsberger Sally, Gray Robert, Saxman Scott, Zujewski Jo Anne
Biometric Research Branch and the Clinical Investigations Branch, Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD 20892, USA.
J Clin Oncol. 2007 May 20;25(15):2122-6. doi: 10.1200/JCO.2006.09.6198.
Progression-free survival is an attractive end point for clinical trials when an overall survival end point may be confounded by additional treatments administered after progression. When a trial is performed in an unblinded manner, however, there is the potential for bias between the treatment arms because of the subjective aspects of the progression end point. We discuss the magnitude of this potential bias and suggest methods for lessening it. We propose the carrying forward of any progression information to two designated time points for the statistical analysis for trials that are not blinded. This proposal, possibly combined with central review of progression scans for these two time points, essentially eliminates any bias, with little risk of major efficiency loss compared with using the reported progression times.
当总生存终点可能因疾病进展后给予的额外治疗而混淆时，无进展生存期是临床试验中一个有吸引力的终点。然而，当以非盲法进行试验时，由于进展终点的主观因素，各治疗组之间存在偏倚的可能性。我们讨论了这种潜在偏倚的程度，并提出了减轻偏倚的方法。对于非盲法试验，我们建议将任何进展信息结转至两个指定时间点进行统计分析。该建议可能与对这两个时间点的进展扫描进行中心审查相结合，基本上消除了任何偏倚，与使用报告的进展时间相比，主要效率损失的风险很小。
