多模态超声组学精准预测直肠癌新辅助化疗疗效的实验研究

Colorectal cancer (CRC) has become the fifth most common cause of morbidity and mortality in China, and it is still rising. Because preoperative neoadjuvant chemotherapy is not effective for all patients, and its efficacy cannot be predicted early with the treatment of blindness, we propose a method to accurately predict the efficacy of chemotherapy in rectal cancer by multi-modal Ultrasomics. On the basis of previous studies on CEUS and elastography about rectal cancer, integrating multimodal ultrasound (2D, SWE, CDFI and CEUS), converting the digital image information into high throughput data to extract the quantitative characteristics of tumor biology (morphology, tumor heterogeneity, hardness and blood perfusion), data mining and establish the method of Ultrasomics. We used a new type of liquid gas conversion type nano vesicles we have successfully prepared to load miR-150, combined ultrasound cavitation to establish an experimental model of rectal cancer with the chemotherapy sensitivity can be precisely controlled, analyse its correlation with Ultrasomics, and build an individual nomogram prediction model to achieve the accurate prediction of efficacy of chemotherapy in rectal cancer by Ultrasomics. It is expected to establish a non-invasive, multi function and accurate medical imaging method and individualized evaluation model, and contribute to the early diagnosis and early intervention of the active treatment for rectal cancer.

结直肠癌已成为我国第五大高发病及高致死性肿瘤，且仍在持续上升。针对术前新辅助化疗疗效无法早期预测、具有治疗盲目性等难题，本项目在既往对直肠癌超声造影和弹性成像的研究基础上（发表于Radiology），联合已制备成功的液气转换型纳米囊泡，提出利用多模态超声组学（Ultrasomics）精准预测直肠癌化疗疗效的设想。通过整合多模态超声（二维、弹性、彩色多普勒和超声造影定量分析）图像，将数字图像信息转换为高通量大数据提取多维度肿瘤生物学定量特征（形态学、肿瘤异质性、硬度学和血流灌注学），进行数据挖掘建立Ultrasomics分析方法；利用液气转换型纳米囊泡负载miR-150，联合超声空化建立化疗敏感性可精准调控的直肠癌模型，建立个体化nomogram预测模型，实现Ultrasomics对直肠癌化疗敏感性的精准预测，从而有望建立一种无创性、多功能的精准影像医学新方法及个体化评估模式。

结直肠癌已成为我国第五大高发病及高致死性肿瘤，且仍在持续上升。针对术前新辅助化疗疗效无法早期预测、具有治疗盲目性等难题，本项目制备了负载miR-150的液气转换型阳离子纳米囊泡，并将囊泡联合超声空化用于调控直肠癌化疗敏感性，结果显示与空白组相比，超声空化释放miRNA-150-PFP@TNDs的化疗敏感组（186mm3）的肿瘤体积小于空白组即不敏感组（1308mm3），且生长速度慢于空白不敏感组（P<0.05），然后行多模态超声检查，利用超声组学软件获取多维度肿瘤生物学特征参数，建立超声组学的分析方法，建立了Ultrasomics预测直肠癌化疗疗效的人工神经网络（ANN）模型，利用单隐藏层的ANN网络，输入15个超声组学特征，包含28个神经元，1个输出层即超声空化释放miR-150 mimics-PFP@TNDs的化疗敏感组。结果显示该模型的预测敏感性、特异性为70.5%, 78.1%，建立了多模态超声组学对直肠癌化疗敏感性的预测模型。

内容获取失败，请点击重试