摘要
激素受体阳性(hormone receptor positive,HR+)乳腺癌作为乳腺癌最主要的分子亚型(占比70%~80%),其显著的肿瘤异质性及肿瘤微环境(tumor microenvironment,TME)介导的治疗抵抗是制约新辅助治疗(neoadjuvant therapy,NAT)疗效提升与个体化诊疗实施的核心因素。当前临床依赖Ki-67等指标检测、穿刺活检等手段预测NAT疗效时,受时空异质性与指标波动性影响,难以精准评估治疗响应,亟需无创、高效的替代技术。动态对比增强磁共振成像(dynamic contrast-enhanced magnetic resonance imaging,DCE-MRI)凭借其功能性成像优势,可量化反映肿瘤血管生成、血管渗透性等TME关键特征,为无创解析TME生物学行为提供了重要窗口。而深度学习(deep learning,DL)技术通过自主挖掘DCE-MRI图像中超越人眼识别的深层时空特征,突破了传统影像组学的局限,为构建高精准度的TME表征与NAT疗效预测模型提供了新范式。本文系统梳理了HR+乳腺癌TME的异质性特征、DCE-MRI在TME功能评估中的技术优势、DL驱动的影像特征挖掘策略及多模态整合研究进展,重点阐述该交叉领域的关键技术瓶颈,并展望未来基于“影像-病理-分子”多组学融合的研究方向,旨在为HR+乳腺癌精准诊疗的临床转化提供理论参考与技术路径。
Hormone receptor-positive(HR+)breast cancer,the most prevalent molecular subtype(70%to 80%of cases),is characterized by significant tumor heterogeneity and treatment resistance mediated by the tumor microenvironment(TME).This heterogeneity and treatment resistance represent the core bottlenecks limiting the improvement of neoadjuvant therapy(NAT)efficacy and the implementation of individualized diagnosis and treatment.Currently,clinical prediction of NAT efficacy relies on methods such as needle biopsy and Ki-67 detection.However,affected by spatiotemporal heterogeneity and indicator fluctuation,these methods cannot accurately evaluate treatment response,creating an urgent need for superior non-invasive predictive tools.Dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)leverages its functional imaging advantages to quantitatively reflect key TME features,such as tumor angiogenesis and vascular permeability.This provides a crucial means for the non-invasive analysis of TME biological behaviors.Deep learning(DL)technology,by autonomously mining deep spatiotemporal features in DCE-MRI images that exceed human visual recognition,breaks through the limitations of traditional radiomics.It offers a new paradigm for constructing high-precision TME characterization and NAT efficacy prediction models.This article systematically reviews the heterogeneous characteristics of HR+breast cancer TME,the technical advantages of DCE-MRI in TME functional evaluation,DL-driven imaging feature mining strategies,and research progress in multimodal integration.It focuses on elaborating key technical bottlenecks in this interdisciplinary field.Additionally,it prospects the future research direction based on the integration of"imaging-pathology-molecular"multi-omics.The aim is to provide theoretical references and technical paths for the clinical transformation of precise diagnosis and treatment of HR+breast cancer.
作者
耿熙坪
孙艺瑶
张勇
赵丹
GENG Xiping;SUN Yiyao;ZHANG Yong;ZHAO Dan(Outpatient Department,Liaoning Cancer Hospital,Shenyang 110042,China;School of Intelligent Medicine,China Medical University,Shenyang 110122,China;Department of Pathology,Liaoning Cancer Hospital,Shenyang 110042,China;Department of Medical Imaging,Liaoning Cancer Hospital,Shenyang 110042,China)
出处
《磁共振成像》
北大核心
2026年第2期175-181,共7页
Chinese Journal of Magnetic Resonance Imaging
基金
辽宁省科技计划联合计划(自然科学基金-面上项目)(编号:2024-MSLH-272)
国家癌症中心攀登基金项目(编号:NCC201906B01)。
关键词
激素受体阳性乳腺癌
肿瘤微环境
新辅助治疗
动态对比增强磁共振成像
多组学
人工智能
深度学习
磁共振成像
hormone receptor positive breast cancer
tumor microenvironment
neoadjuvant therapy
dynamic contrast-enhanced magnetic resonance imaging
multi-omics
artificial intelligence
deep learning
magnetic resonance imaging
