Project 1

项目摘要--项目1 筛查乳房X光检查可降低晚期乳腺癌的发病率，从而降低乳腺癌死亡率。在被诊断为乳腺癌的常规筛查女性中，约有22%的人患有晚期癌症，黑人女性的晚期癌症发病率是白人女性的两倍多。我们已经将晚期癌症比率确定为关键的筛查结果，因为它最能区分谁是乳腺癌死亡风险最高的人。识别癌症风险较高的女性，并以她们为目标进行有效的筛查战略，可以降低乳腺癌死亡率。因此，项目1旨在推进一种新的基于风险的筛查范例，以确定女性的绝对晚期癌症风险，以便为筛查强度和补充成像决策提供信息。目的1)利用乳腺癌监测联盟在40-74岁女性和晚期乳腺癌筛查中的数据，我们将开发、评估和验证新的风险预测模型，以识别晚期乳腺癌的高危女性。我们将结合临床乳腺癌风险因素、自我报告的种族/民族、成像特征和人工智能输出，使用结合了现代数据自适应建模方法的新型离散时间生存模型来估计累积的6年晚期癌症风险，并检查按种族/民族进行的模型预测公平性。目的2)我们将在因果推断框架内进行统计学模拟随机试验，根据晚期癌症风险评估普通筛查人群中乳腺癌死亡率的年度筛查和两年一次的筛查。目的3)三个已建立的乳腺癌模拟模型将被用来评估人群水平筛查策略的长期益处(避免乳腺癌死亡和增加寿命)和危害(例如假阳性和活检)，这些筛查策略针对的是乳房X光检查频率(每年、两年一次)和基于癌症风险的补充MRI。这项建议将回答：1)哪些接受乳房X光检查的女性患晚期癌症的风险较高，需要两年一次的乳房X光检查的替代策略，哪些女性风险较低，可以较少进行筛查？2)与两年一次的乳房X光检查相比，哪些高级癌症风险水平需要进行类似的乳腺癌死亡率降低？3)与基于年龄的策略相比，使用高级癌症风险来告知筛查频率和方式的基于风险的筛查策略是否改善了结果？因此，该项目将提供证据，指导妇女和提供者根据晚期癌症风险制定公平的筛查战略，以最大限度地提高筛查效益，同时将危害降至最低。

PROJECT SUMMARY – Project 1 Screening mammography results in lower breast cancer mortality by reducing the incidence of advanced breast cancer. Advanced cancer occurs in about 22% of routinely screened women diagnosed with breast cancer, with advanced cancer rates more than 2-fold higher in Black vs. White women. We have identified advanced cancer rate as a key screening outcome because it best discriminates who has the highest risk of breast cancer death. Identifying women at increased advanced cancer risk and targeting them with effective screening strategies could reduce breast cancer mortality. Thus, Project 1 aims to advance a new risk-based screening paradigm that identifies women’s absolute advanced cancer risk to inform screening intensity and supplemental imaging decisions. Aim 1) Using Breast Cancer Surveillance Consortium data on screening exams from women aged 40-74 years, and advanced breast cancers, we will develop, evaluate, and validate new risk prediction models to identify women at high risk of advanced breast cancer. We will combine clinical breast cancer risk factors, self-reported race/ethnicity, imaging features, and Artificial Intelligence output to estimate cumulative 6-year advanced cancer risk using novel discrete time survival models incorporating modern data-adaptive modeling approaches for annual and biennial screeners and examine model predictive equity by race/ethnicity. Aim 2) We will statistically emulate a randomized trial within a causal inference framework to evaluate annual vs. biennial mammography screening on breast cancer mortality among the general screening population according to advanced cancer risk. Aim 3) Three established breast cancer simulation models will be used to evaluate the long-term benefits (breast cancer deaths averted, and life years gained) and harms (e.g., false positives and biopsies) of population-level screening strategies that target mammography frequency (annual, biennial) and supplemental MRI based on advanced cancer risk. This proposal will answer: 1) Which women undergoing screening mammography are at high risk of advanced cancer and require an alternative strategy to biennial mammography, and which women are at low risk and can be screened less often? 2) What advanced cancer risk levels require annual mammography vs. biennial mammography for similar reductions in breast cancer mortality? 3) Does a risk-based screening strategy using advanced cancer risk to inform screening frequency and modality improve outcomes compared to an agebased strategy? Therefore, this project will provide evidence to guide women and providers towards equitable screening strategies based on advanced cancer risk to maximize screening benefits while minimizing harms.