Supplementary data: Considerations for emulations of randomized controlled trials using real-world data: learnings from an emulation of MONALEESA-2

<b>These are peer-reviewed supplementary materials for the article</b><b> </b><b>'</b><b>Considerations for emulations of </b><b>randomized controlled trials using </b><b>real-world data: learnings from an </b><b>emulation of MONALEESA-2</b><b>'</b><b> </b><b>published in the</b><b> </b><b><i>Journal of Comparative Effectiveness Research</i></b><b>.</b><b>Appendix Table 1:</b> Operationalization of trial inclusion criteria<b>Appendix Table 2:</b> Operationalization of trial exclusion criteria<b>Appendix Table 3:</b> Absolute standardized differences after IPTW and PS matching for the primary analysis of ribociclib plus letrozole vs. contemporaneous letrozole<b>Appendix Table 4:</b> Absolute standardized differences after IPTW and PS matching for the subgroup analysis of ribociclib plus letrozole vs. contemporaneous letrozole<b>Appendix Table 5:</b> Absolute standardized differences after IPTW and PS matching for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 1<b>Appendix Table 6:</b> Absolute standardized differences after IPTW and PS matching for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 2<b>Appendix Figure 1:</b> Weighted Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. contemporaneous letrozole<b>Appendix Figure 2:</b> Matched Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. contemporaneous letrozole<b>Appendix Figure 3:</b> Weighted Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 1<b>Appendix Figure 4:</b> Matched Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 1<b>Appendix Figure 5: </b>Weighted Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 2<b>Appendix Figure 6:</b> Matched Kaplan-Meier curve for the primary analysis of ribociclib plus letrozole vs. letrozole historical cohort 2<b>Aim:</b> The eligibility criteria of randomized controlled trials can exclude groups of patients indicated for the therapy post-approval and the trial populationmay not be generalizable to routine care.We attempted to emulate the MONALEESA-2 trial of ribociclib plus letrozole for women with advanced breast cancer using real-world data (RWD) to assess the impact of modifying entry criteria of the trial, which found a survival benefit associated with ribociclib (hazard ratio [HR] = 0.76). <b>Materials &amp; methods:</b> Post-menopausal women with recurrent or metastatic breast cancer were identified in a linked electronic health recordsclaims database. Treatment groups were ribociclib plus letrozole or letrozole without ribociclib (referred to as letrozole). Overall survival was compared between the two groups using Cox proportional hazards models with inverse probability of treatment weights and propensity score matching. <b>Results:</b> There were 132,406 patients who initiated ribociclib plus letrozole or letrozole from 13 March 2017 (ribociclib approval) to 30 September 2023. After applying trial entry criteria, the sample size was 3912 patients. Compared with real-world patients, trial participants tended to be younger (&gt;50% were &lt;65 years old compared with 38%) and more commonly had liver or lung metastases (&gt;50% vs &lt;15%). Among realworld patients, those treated with ribociclib plus letrozole had higher comorbidity scores (mean Elixhauser Comorbidity Index Score 15 vs 9) and were more likely to have metastatic disease burden than patients treated with letrozole (85% vs 45%). We were unable to emulate the trial findings; all HRs in this analysis were &gt;1. <b>Conclusion:</b> Real-world patients may differ from those participating in randomized controlled trials. Along with data source limitations, such as missing clinical information or incomplete capture of mortality, this can impact the ability to emulate trials in RWD. However, RWD is key for describing patients in routine care and to answer relevant questions following the approval of new therapies.

本内容为发表于《比较疗效研究期刊》（*Journal of Comparative Effectiveness Research*）的论文《使用真实世界数据模拟随机对照试验的思考：来自MONALEESA-2试验模拟的经验教训》的同行评议补充材料。

附录表1：试验纳入标准的操作化定义
附录表2：试验排除标准的操作化定义
附录表3：瑞波西利联合来曲唑 vs 同期单用来曲唑主要分析中，逆概率加权（Inverse Probability of Treatment Weighting, IPTW）与倾向得分匹配（Propensity Score, PS）后的绝对标准化差值
附录表4：瑞波西利联合来曲唑 vs 同期单用来曲唑亚组分析中，IPTW与PS匹配后的绝对标准化差值
附录表5：瑞波西利联合来曲唑 vs 来曲唑历史队列1主要分析中，IPTW与PS匹配后的绝对标准化差值
附录表6：瑞波西利联合来曲唑 vs 来曲唑历史队列2主要分析中，IPTW与PS匹配后的绝对标准化差值
附录图1：瑞波西利联合来曲唑 vs 同期单用来曲唑主要分析的加权Kaplan-Meier曲线
附录图2：瑞波西利联合来曲唑 vs 同期单用来曲唑主要分析的匹配Kaplan-Meier曲线
附录图3：瑞波西利联合来曲唑 vs 来曲唑历史队列1主要分析的加权Kaplan-Meier曲线
附录图4：瑞波西利联合来曲唑 vs 来曲唑历史队列1主要分析的匹配Kaplan-Meier曲线
附录图5：瑞波西利联合来曲唑 vs 来曲唑历史队列2主要分析的加权Kaplan-Meier曲线
附录图6：瑞波西利联合来曲唑 vs 来曲唑历史队列2主要分析的匹配Kaplan-Meier曲线

**研究目的**：随机对照试验的纳入标准往往会排除获批后适配该疗法的患者群体，试验人群亦无法外推至常规临床诊疗场景。本研究尝试利用真实世界数据（Real-World Data, RWD），针对晚期乳腺癌女性患者模拟MONALEESA-2试验（该试验证实瑞波西利可带来生存获益，风险比[Hazard Ratio, HR]=0.76），以评估调整试验入组标准所产生的影响。

**材料与方法**：在关联电子健康记录-理赔数据库中，筛选复发或转移性乳腺癌的绝经后女性患者。治疗分组为瑞波西利联合来曲唑组，与单用来曲唑组（下称来曲唑组）。采用纳入逆概率加权与倾向得分匹配的Cox比例风险模型，比较两组患者的总生存期。

**研究结果**：2017年3月13日（瑞波西利获批日期）至2023年9月30日期间，共计132406例患者启动瑞波西利联合来曲唑或来曲唑治疗。应用试验入组标准后，最终样本量为3912例。与真实世界患者相比，试验参与者更趋年轻（≥50%的受试者年龄<65岁，而真实世界人群中该比例仅为38%），且更常合并肝或肺转移（≥50% vs <15%）。在真实世界患者中，接受瑞波西利联合来曲唑治疗的患者合并症评分更高（Elixhauser合并症指数平均得分为15 vs 9），且相较接受来曲唑治疗的患者，更易出现较高的转移性疾病负荷（85% vs 45%）。本研究未能复现原试验结果：本次分析中所有风险比均大于1。

**结论**：真实世界患者群体可能与随机对照试验的参与者存在显著差异。加之数据来源本身存在局限性，例如临床信息缺失或死亡率记录不全，这些因素均可能制约利用真实世界数据模拟临床试验的可行性。然而，真实世界数据仍是描述常规诊疗场景下患者特征、解答新药获批后相关临床问题的核心工具。