Division of Medical Oncology & Hematology, Department of Medicine, Princess Margaret Cancer Centre and the University of Toronto, 610 University Avenue, Toronto, Ontario M5G 2M9, Canada.
Division of Medical Oncology & Hematology, Department of Medicine, Princess Margaret Cancer Centre and the University of Toronto, 610 University Avenue, Toronto, Ontario M5G 2M9, Canada; University of New South Wales, Sydney, NSW 2052, Australia.
Cancer Treat Rev. 2021 Mar;94:102167. doi: 10.1016/j.ctrv.2021.102167. Epub 2021 Feb 16.
The Fragility Indexquantifies the reliability of positive trials by estimating the number of events, which would change statistically significant results to non-significant results.
We identified randomized trials supporting drug approvals by the US FDA between 2009 and 2019 in lung, breast, prostate, and colon cancers and in melanoma. We reconstructed survival tablesand calculated the number of events, which would result in a non-significant result for the primary endpoint. The FI was then compared to the number of patients in each trial who withdrew consent or were lost to follow-up. Regression analyses were used to explore associations between RCT characteristics and FI and trials in which FI was lower or equal to number of participants who withdrew consent or were lost to follow-up.
Among 81 RCTs, the median FI was 28. The median number of patients who withdrew consent or were lost to follow up was 27. FI was equal or lower than the number of patients who withdrew consent or were lost to follow-up in 47 trials (58%). There was a modest increase in FI over time (p = 0.02). Trials with overall survival as the primary endpoint (p = 0.006) and those in the palliative setting (p < 0.001) had lower FI. There was no association with trial sample size or duration of follow-up.
Statistical significance of RCTs in common solid tumours can be reversed often with a small number of additional events. Post-approval RCTs or real-world data analyses should be performed to ensure results of registration trials are robust.
Fragility Index 通过估计事件数量来量化阳性试验的可靠性，这些事件数量会使统计学上显著的结果变为非显著结果。
我们确定了 2009 年至 2019 年间美国食品和药物管理局（FDA）批准的肺癌、乳腺癌、前列腺癌、结肠癌和黑色素瘤药物的随机试验。我们重建了生存表，并计算了导致主要终点非显著结果所需的事件数量。然后，将 FI 与每个试验中退出同意或失访的患者数量进行比较。回归分析用于探索 RCT 特征与 FI 之间的关联，以及 FI 低于或等于退出同意或失访的患者数量的试验。
在 81 项 RCT 中，FI 的中位数为 28。退出同意或失访的患者中位数为 27。47 项试验（58%）的 FI 等于或低于退出同意或失访的患者数量。FI 随时间呈适度增加（p=0.02）。以总生存期为主要终点的试验（p=0.006）和姑息治疗环境中的试验（p<0.001）FI 较低。FI 与试验样本量或随访时间无相关性。
常见实体瘤 RCT 的统计学意义通常可以通过少量额外事件逆转。应进行上市后 RCT 或真实世界数据分析，以确保注册试验的结果可靠。
