Zhang Lijun, Ko Chia-Wen, Tang Shenghui, Sridhara Rajeshwari
1 Division of Biometrics V, Office of Biostatistics, Office of Translational Sciences, Center of Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD, USA.
Ther Innov Regul Sci. 2013 Jan;47(1):95-100. doi: 10.1177/0092861512459180.
This study evaluated the circumstances under which the observed progression-free survival (PFS) benefit may translate into an overall survival (OS) benefit.
A simulation study, based on PFS and OS joint model decomposition, was conducted to evaluate the impact of crossover rates, survival post progression (SPP) lengths, and magnitudes of difference in median PFS on OS. Under different simulation scenarios, the degree of impact was investigated based on the probability of observing a significant OS benefit given an observed PFS benefit.
Using simulation parameters defined based on historical NDA datasets, the probability of detecting an OS benefit given the observed PFS benefit depends largely on crossover rate and SPP length (ie, SPP median times). Compared to no crossover, a crossover rate of ≤ 50% decreases the probability of detecting an OS benefit by at most 15% regardless of the SPP length. A crossover rate of > 50% decreases the probability of detecting an OS benefit much further, and the extent of decrease is proportional to the length of SPP.
Crossover rate and SPP length are important factors affecting the benefit translating from PFS to OS. This simulation study identified a threshold, of 50%, for crossover rate that is likely to confound OS effect. With a greater than 50% crossover rate, longer SPP further decreases the probability of translating a PFS benefit to OS.
本研究评估了观察到的无进展生存期（PFS）获益转化为总生存期（OS）获益的情况。
基于PFS和OS联合模型分解进行了一项模拟研究，以评估交叉率、进展后生存期（SPP）长度以及中位PFS差异幅度对OS的影响。在不同模拟场景下，根据在观察到PFS获益的情况下观察到显著OS获益的概率来研究影响程度。
使用基于历史新药申请（NDA）数据集定义的模拟参数，在观察到PFS获益的情况下检测到OS获益的概率很大程度上取决于交叉率和SPP长度（即SPP中位时间）。与无交叉相比，无论SPP长度如何，交叉率≤50%最多可将检测到OS获益的概率降低15%。交叉率>50%会进一步降低检测到OS获益的概率，且降低程度与SPP长度成正比。
交叉率和SPP长度是影响从PFS转化为OS获益的重要因素。本模拟研究确定了交叉率50%这一可能混淆OS效应的阈值。交叉率大于50%时，更长的SPP会进一步降低将PFS获益转化为OS的概率。
