Multifactoral breast cancer risk prediction accounting for ethnic and tumor diversity

摘要 乳腺癌风险评估工具在临床实践中被广泛用于指导筛查决策。 时机和方式、生活方式干预、基因检测、预防性治疗和降低风险的手术。 虽然在实践中使用了一些工具，但它们面临着各种挑战，包括：(1)适度 由于缺乏纳入一套全面风险因素的统一模式而导致的歧视能力； 无法产生亚型特定风险，特别是考虑到乳腺癌和/或侵袭性亚型 仅对激素受体阳性肿瘤有效的预防性内分泌治疗；(Iii)缺乏数据以 为不同种族人口建立模型；以及，(Iv)模型缺乏验证，特别是在医疗保健方面 可以在实践中广泛传播模型的环境。在这份提案中，我们将吸收和分析 从参与NCI队列联盟的研究到以下研究的大量和多样化的女性样本数据 开发一种全面的工具，以预测主要种族的总体和分类型的乳腺癌风险 在美国的团体。我们还建议在不同的临床环境中前瞻性地验证该模型，包括 风险分层筛查试验。在目标1中，我们将开发一个全面的模型来预测 包括家族史信息在内的多种族妇女的总体乳腺癌；多基因 风险评分；人体测量、生活方式和生殖因素；荷尔蒙生物标志物；以及 乳房X光摄影密度。在目标2中，我们将针对特定的乳腺癌亚型定制这些风险模型，尤其是 雌激素受体阴性和阳性癌症。在目标3中，我们将评估这些风险预测的有效性 综合医疗保健系统中的模型、乳房X光检查登记和持续的基于风险的 美国乳房X光摄影筛查试验。所产生的模型可以在不同的临床环境中使用 指导预防治疗或风险分层筛查计划，增加乳腺癌死亡人数 预防的同时最大限度地减少过度诊断和过度治疗。

Abstract Breast cancer risk assessment tools are widely used in clinical practice to guide decisions regarding screening timing and modality, life-style interventions, genetic testing, preventive therapy, and risk-reducing surgery. Although a number of tools are used in practice, they face various challenges including: (i) modest discriminatory ability due to lack of a unified model that incorporates a comprehensive set of risk-factors; (ii) inability to produce sub-type specific risk, especially considering aggressive subtypes of breast cancer and/or prophylactic endocrine therapy that is effective only for hormone receptor positive tumors; (iii) lack of data to build models for different ethnic populations; and, (iv) scant validation of models, especially in healthcare settings where models can be widely disseminated in practice. In this proposal, we will assimilate and analyze data on a large and diverse sample of women from studies participating in the NCI Cohort Consortium to develop a comprehensive tool that will predict breast cancer risk, overall and by sub-types, across major ethnic groups in the US. We further propose to prospectively validate the model in different clinical settings, including a risk-stratified screening trial. In Aim 1 we will develop a comprehensive model for predicting absolute risk of overall breast cancer for women from multiple ethnicities, incorporating information on family history; polygenic risk-scores (PRS); anthropometric, life-style and reproductive factors; hormonal biomarkers; and mammographic density. In Aim 2 we will tailor these risk models to specific breast cancer subtypes, notably estrogen receptor negative and positive cancers. In Aim 3 we will evaluate the validity of these risk prediction models in integrated health care systems, mammography registries, and an ongoing risk-based mammographic screening trial in the US. The resulting models could be used in diverse clinical settings to guide preventive therapy or risk-stratified screening programs, increasing the number of breast cancer deaths prevented while minimizing overdiagnosis and overtreatment.