Department of Health Outcomes & Biomedical Informatics, College of Medicine, University of Florida, Gainesville, FL.
Cancer Informatics Share Resource, University of Florida Health Cancer Center, Gainesville, FL.
JCO Clin Cancer Inform. 2022 Jul;6:e2100195. doi: 10.1200/CCI.21.00195.
Using real-world data (RWD)-based trial simulation approach, we aim to simulate colorectal cancer (CRC) trials and examine both effectiveness and safety end points in different simulation scenarios.
We identified five phase III trials comparing new treatment regimens with an US Food and Drug Administration-approved first-line treatment in patients with metastatic CRC (ie, fluorouracil, leucovorin, and irinotecan) as the standard-of-care (SOC) control arm. Using Electronic Health Record-derived data from the OneFlorida network, we defined the study populations and outcome measures using the protocols from the original trials. Our design scenarios were (1) simulation of the SOC fluorouracil, leucovorin, and irinotecan arm and (2) comparative effectiveness research (CER) simulation of the control and experimental arms. For each scenario, we adjusted for random assignment, sampling, and dropout. We used overall survival (OS) and severe adverse events (SAEs) to measure effectiveness and safety.
We conducted CER simulations for two trials, and SOC simulations for three trials. The effect sizes of our simulated trials were stable across all simulation runs. Compared with the original trials, we observed longer OS and higher mean number of SAEs in both CER and SOC simulation. In the two CER simulations, hazard ratios associated with death from simulations were similar to that reported in the original trials. Consistent with the original trials, we found higher risk ratios of SAEs in the experiment arm, suggesting potentially higher toxicities from the new treatment regimen. We also observed similar SAE rates across all simulations compared with the original trials.
In this study, we simulated five CRC trials, and tested two simulation scenarios with several different configurations demonstrated that our simulations can robustly generate effectiveness and safety outcomes comparable with the original trials using real-world data.
采用基于真实世界数据（RWD）的试验模拟方法，我们旨在模拟结直肠癌（CRC）试验，并在不同模拟场景下检查有效性和安全性终点。
我们确定了五项比较新治疗方案与转移性 CRC 患者中美国食品和药物管理局批准的一线治疗（即氟尿嘧啶、亚叶酸钙和伊立替康）作为标准治疗（SOC）对照臂的 III 期试验。使用 OneFlorida 网络的电子健康记录数据，我们使用原始试验的方案定义了研究人群和结局测量。我们的设计方案是（1）SOC 氟尿嘧啶、亚叶酸钙和伊立替康臂的模拟，以及（2）对照和实验组的比较有效性研究（CER）模拟。对于每种情况，我们都调整了随机分组、抽样和脱落。我们使用总生存期（OS）和严重不良事件（SAE）来衡量有效性和安全性。
我们对两项试验进行了 CER 模拟，对三项试验进行了 SOC 模拟。模拟试验的效果大小在所有模拟运行中均保持稳定。与原始试验相比，我们在 CER 和 SOC 模拟中均观察到更长的 OS 和更高的平均 SAE 数。在两项 CER 模拟中，与原始试验相比，模拟死亡的风险比相似。与原始试验一致，我们发现实验组的 SAE 风险比更高，表明新治疗方案可能具有更高的毒性。与原始试验相比，我们还观察到所有模拟中的 SAE 率相似。
在这项研究中，我们模拟了五项 CRC 试验，并测试了两种模拟方案，具有几种不同的配置，证明我们的模拟可以使用真实世界数据稳健地产生与原始试验可比的有效性和安全性结果。
