A Simulation Modeling Study to Support Personalized Breast Cancer Prevention and Early Detection in High-Risk Women

摘要： 在这个前所未有的发现和“大数据”时代，个性化护理是复杂的。拟议的研究重点 每年有超过10万名美国女性面临现实世界的选择，她们的风险高于平均水平。 由于乳腺密度和遗传易感性等风险因素而患上乳腺癌。妇女在高 有患上乳癌的风险的妇女有资格接受各种乳癌预防和早期检测方案。 目前的临床指南建议，这些妇女提供降低风险的药物， 除了每年的乳房X光检查之外，还可以使用磁共振成像（MRI）进行补充成像。中的每 这些选择的利弊各不相同，取决于个人的风险因素。年度 乳房X光检查和MRI可以早期发现肿瘤，从而实现早期诊断和提高生存率，但 与乳腺密度相关的假阳性相关的危害。降低风险的药物可降低 患乳腺癌的几率降低了近一半，但这些药物会根据年龄诱发更年期症状， 在一小部分女性中，增加了患子宫内膜癌或其他疾病的风险。最终将其识别为 妇女对干预措施的选择可能取决于她如何权衡这些不同因素的利弊。 选项和结果，考虑到个人风险。为了解决这些复杂性，过去的研究集中在以下两个方面： 单一风险因素，风险预测工具与选定的因素，或单独的筛查策略。 我们建议使用现存的癌症干预和监测建模网络（CISNET）模拟 模型，以综合临床风险因素的数据，以及早期发现与筛查和初级 预防与降低风险的药物，以提供个性化的数据，这将有助于确定妇女谁是 更有可能受益于各种干预措施或干预措施的组合，危害最小。的 目标1：a）提供关于好处的数据（例如，避免乳腺癌;早期发现和改善 生存）和危害（降低风险药物的副作用;筛查的假阳性） 降低风险的药物和筛查策略的个体化5年乳腺癌的风险 癌症、乳腺密度和不同体验的偏好（效用权重）;以及B）行为敏感性 分析以估计模型输入或假设中的不确定性对结果的影响。目标2：探索 将PRS添加到5年风险估计中以进一步个性化获益平衡信息的影响， 各种降低风险的药物和筛查策略的危害。目标3：对关键线人进行访谈 与卫生保健提供者一起指导未来使用模型结果来支持共同决策。 这项研究的结果将为指导高危女性的个性化护理提供新的数据。今后 研究，这些数据可以集成到一个对话援助，以促进共享决策期间 临床遭遇这项研究有助于国家癌症研究所的使命，以支持在 癌症预防和控制，并利用“大数据”将研究转化为临床实践。

Abstract: Personalized care is complex in this unprecedented era of discovery and ‘big data’. The proposed study focuses on the real-world choices facing over 100,000 US women each year who are at higher than average risk of developing breast cancer due to risk factors such as breast density and genetic predisposition. Women at high risk of developing breast cancer are eligible for various breast cancer prevention and early detection options. Current clinical guidelines recommend that these women are offered risk reducing medication, and supplemental imaging with magnetic resonance imaging (MRI) in addition to annual mammography. Each of these choices has a different profile of benefits and harms that will depend on individual risk factors. Annual mammography and MRI can detect tumors early, leading to early diagnosis and improved survival, but have harms related to false positives linked to breast density. Risk-reducing medications reduce the likelihood of developing breast cancer by nearly half, but these medications can induce menopausal symptoms based on age, and in a small percent of women, increase the risk of endometrial cancer or other conditions. Ultimately, a woman’s choice of intervention may depend on how she will weigh harms against benefits for these different options and outcomes given individual risk. To address these complexities, past studies have focused on either on single risk factors, risk prediction tools with selected factors, or screening strategies alone. We propose to use an extant Cancer Intervention and Surveillance Modeling Network (CISNET) simulation model to synthesize data on clinical risk factors and the impact of early detection with screening and primary prevention with risk-reducing medication to provide personalized data that will help identify women who are more likely to benefit from various interventions or combinations of interventions with the least harms. The aims are to: Aim 1: a) Provide data on the benefits (e.g. avoiding breast cancer; early detection and improved survival) and harms (side effects of risk-reducing drugs; false positives with screening) of various combinations of risk reducing medication and screening strategies personalized by individual 5-year risk of developing breast cancer, breast density, and preferences (utility weights) for different experiences; and b) Conduct sensitivity analysis to estimate the effects of uncertainty in model inputs or assumptions on results. Aim 2: Explore the impact of adding PRS to 5-year risk estimates to further personalize information on the balance of benefits and harms of various risk-reducing medication and screening strategies. Aim 3: Conduct key informant interviews with health care providers to guide the future use of model results to support shared decision making. The results of this study will provide novel data to guide personalized care for high-risk women. In future research, these data could be integrated into a conversation aid to facilitate shared decision making during clinical encounters. This study contributes to the National Cancer Institute’s mission to support advances in cancer prevention and control, and use ‘big data’ to enable the translation of research into clinical practice.