Development of Magnetic Resonance Fingerprinting (MRF) to Assess Response to Neoadjuvant Chemotherapy in Breast Cancer

摘要 在女性中，乳腺癌是最常见的确诊癌症，也是癌症相关死亡的主要原因。 在世界范围内，2020年约有230万新病例和685,000人死亡。新辅助化疗 (NAC)通常用于乳腺肿瘤手术前缩小肿瘤大小。不幸的是，由于 对于乳腺肿瘤的遗传和表型异质性，并不是所有的患者对传统的NAC都有反应。 目前，只有大约22%的患者表现出病理完全应答(PCR)，而其余的患者没有出现 患者对化疗有部分反应(占全部患者的54%)或无反应。早期预测 肿瘤对化疗的反应识别无反应者可以1)减少不必要的副作用和 与无效治疗相关的成本，以及2)帮助医生更早地定制治疗计划，以达到更好的效果 治疗效果和提高存活率。目前监测肿瘤对化疗的反应是基于 通过体检测量肿瘤大小，这是主观的，很难量化，最重要的是， 与潜在的生物变化相比，是暂时延迟的。定量、可重复、客观的方法 这可以在肿瘤大小显著变化之前及早检测到肿瘤的生理变化 改善治疗结果和病人护理质量。然而，定量成像具有重要的意义 技术挑战，这在临床环境中很少进行。在这里，我们建议利用Magic 磁共振指纹(MRF)，一个革命性的定量磁共振新平台，由我们的 团队，开发新的成像生物标记物，用于早期评估乳房女性的治疗反应 癌症。我们的团队开发了一种乳房磁共振成像方法，可以同时生成定量的3D T1和T2 地图只需约6分钟，重复性极佳。我们还将我们的MRF方法扩展到同时 量化大脑的T1、T2和ADC图，没有图像失真。在这里，我们计划优化这一新的 专门用于乳腺癌妇女的弛豫/扩散MRF方法(目标1)。新奇深度学习 将开发方法以提供快速(5分钟)和高分辨率(1.2毫米各向同性)的采集 全乳房覆盖以及基于云计算的高效后处理管道(目标2)。 最后，我们将在两个患者队列中评估所开发的早期预测治疗反应的方法。 HER2阳性或三阴性乳腺癌患者(目标3)。在这个项目成功完成后， 开发的mrf技术将为早期预测治疗提供实用的定量乳房检查。 对NAC和其他治疗方法(激素治疗、基于抗体的靶向治疗等)的反应为 乳腺癌妇女，最终目标是减少符合条件的受试者和定制的无效治疗 获得最佳治疗结果的治疗方法。

Abstract In women, breast cancer is the most commonly diagnosed cancer and leading cause of cancer related deaths worldwide, with approximately 2.3 million new cases and 685,000 deaths in 2020. Neoadjuvant chemotherapy (NAC) is commonly applied to reduce the tumor size before surgery for breast neoplasms. Unfortunately, due to the genetic and phenotypic heterogeneity of breast tumors, not all patients respond to conventional NAC. Currently, only about 22% of patients show pathologic complete response (pCR), while the remaining non-pCR patients show either partial response (54% of all patients) or no response to chemotherapy. Early prediction of tumor response to chemotherapy to identify non-responders could 1) reduce unnecessary side effects and costs related to ineffective therapy, and 2) help physicians tailor the treatment plan earlier to achieve better therapeutic outcomes and improve survival. Monitoring tumor response to chemotherapy is currently based on tumor size measured by physical exam, which is subjective, difficult to quantify, and most importantly, temporally delayed compared to underlying biological changes. Quantitative, repeatable and objective methods that could provide an early detection of tumor physiological changes before size changes could significantly improve treatment outcome and the quality of patient care. However, quantitative imaging poses significant technical challenges, which is rarely performed in the clinical setting. Here, we propose to leverage Magnetic Resonance Fingerprinting (MRF), a revolutionary new platform for quantitative MR that was invented by our team, to develop new imaging biomarkers for early assessment of treatment response in women with breast cancer. Our team has developed a breast MRF method to simultaneously generate quantitative 3D T1 and T2 maps in ~6 minutes with excellent reproducibility. We have also expanded our MRF method to simultaneous quantify T1, T2 and ADC maps of the brain with no image distortion. Here, we plan on optimizing this new relaxometry / diffusion MRF method specifically for women with breast cancer (Aim 1). Novel deep learning methods will be developed to provide a fast (<5 minute) and high resolution (1.2 mm isotropic) acquisition for whole-breast coverage along with an efficient post-processing pipeline based on cloud computation (Aim 2). Finally, we will evaluate the developed method for early prediction of treatment response in two patient cohorts with either HER2-positive or triple negative breast cancers (Aim 3). Upon successful completion of this project, the developed MRF technique will provide a practical quantitative breast exam for early prediction of treatment response to NAC and other treatment methods (hormone therapy, antibody-based target therapy, etc.) for women with breast cancer, with the ultimate goal to reduce ineffective treatment in eligible subjects and tailor the treatment methods for optimum therapeutic outcomes.