Department of Epidemiology and Biostatistics, School of Public Health, Texas A&M University, College Station, TX, United States of America.
Yale Cancer Center, Yale School of Medicine, New Haven, CT, United States of America.
Contemp Clin Trials. 2021 Nov;110:106589. doi: 10.1016/j.cct.2021.106589. Epub 2021 Oct 9.
Waterfall plots have been increasingly used to visualize tumor response measurements in Phase II clinical trials. Despite the growing popularity of waterfall plots, quantitative summaries and distribution features of the data indicating antitumor activities are typically not reported. Statistical summaries from the raw and augmented data may provide valuable information for understanding such features. This issue has not been discussed adequately in the literature or fully recognized within the oncology community. In this article, we propose to augment the data using a statistical distribution system. Summary statistics of the data set corresponding to waterfall plot can be calculated using the original sample of the tumor changes or the augmentation sample, which may lead to additional insights into the treatment effect. We demonstrate the proposed method in numerical studies and in a Phase II clinical trial investigating the efficacy of a treatment for ovarian carcinoma. We recommend the proposed statistical analyses for making inferences in addition to the waterfall plot visualization.
瀑布图已越来越多地用于可视化 II 期临床试验中的肿瘤反应测量。尽管瀑布图越来越受欢迎，但通常不会报告定量总结和数据的分布特征，这些数据可以表明抗肿瘤活性。原始数据和扩充数据的统计摘要可为了解这些特征提供有价值的信息。这个问题在文献中讨论得不够充分，肿瘤学界也没有充分认识到这一点。在本文中，我们建议使用统计分布系统来扩充数据。可以使用肿瘤变化的原始样本或扩充样本计算对应于瀑布图的数据集中的统计摘要，这可能会进一步了解治疗效果。我们在数值研究和一项针对卵巢癌治疗效果的 II 期临床试验中演示了所提出的方法。我们建议除了瀑布图可视化之外，还可以进行所提出的统计分析以进行推断。
