基于磁共振影像基因组学的高级别浆液性卵巢癌分子分型和预后预测研究

High-grade serous ovarian carcinoma (HGSOC) is frequently diagnosed at an advanced stage, and has the characteristics of high recurrence rates and poor prognosis. The molecular subtypes of HGSOC were identified to have strongly associated with prognosis. Molecular characterization of HGSOC based on tumor samples are invasive and expensive procedures. A non-invasively and accurately method for predicting the molecular subtype of HGSOC may be helpful for making individualized treatment plan and improving the prognosis. Our previous studies have shown that MRI played an important role in characterization, therapeutic evaluation, and prognostic prediction. However, the associations between MRI features and HGSOC molecular subtypes are still unclear. Our projects will use the method of radiogenomics to extract the radiomics features from multiple sequences and multiple parameters MRI maps based on primary lesion, and combine with the regional distribution features of metastases. We also integrate the prognostically relevant genetic biomarkers with MRI radiomics features, to build a model of biological interpretability to accurately predict the molecular subtypes and prognosis of HGSOC, and to further clarify the association between molecular subtype and prognosis. This project is expected to provide an effectively and non-invasively quantitative MRI biomarkers, to predict the molecular subtypes and prognosis of HGSOC, and to build robust prediction model with great meaning on theory and in practice value.

高级别浆液性卵巢癌（HGSOC）就诊时常为进展期，具有易复发和预后差的特点。其分子分型和预后密切相关。基于肿瘤组织样本检测是一种有创且成本高昂的方法，寻找无创性精准预测肿瘤分子分型的方法，将有助于临床制定个体化诊疗方案和改善患者预后。我们前期研究工作表明MRI在卵巢癌的诊断、疗效评价和预后预测中发挥着重要作用。但MRI特征和HGSOC分子分型间的关系仍不明确。本项目拟采用影像基因组学方法，提取基于原发灶的多序列和多参数图MRI组学特征，结合转移灶的区域分布特征，融合预后相关的基因标志物，构建具有生物学可解释性的精准预测HGSOC分子分型和预后的模型，并阐明分子亚型与预后间的关系。本项目有望为预测HGSOC分子分型和预后提供有效且无创性的定量MRI标志物，并建立稳定的预测模型，具有重要的理论意义和应用价值。

卵巢癌是女性生殖系统第三常见恶性肿瘤，死亡率却居妇科恶性肿瘤之首。超过70%的卵巢癌为高级别浆液性卵巢癌（HGSOC）亚型，确诊时常为晚期。尽管在卵巢癌手术技术和全程管理模式方面已取得巨大进步，但晚期HGSOC的5年整体生存率仍约40%。有证据表明HGSOC分子亚型、减瘤结局及复发风险与患者预后密切相关。然而，临床实践中无创预测分子亚型、手术结局及复发风险均是非常具有挑战性的，亟需可靠的精准预测模型。近年来，影像组学作为一种新的图像分析方法可为解决这些临床难题提供一种新思路。因此，本项目旨在构建术前鉴定HGSOC分子亚型、预测手术结局和复发风险的MRI影像组学模型。本项目主要研究内容包括以下几个方面：1）晚期HGSOC肿瘤的MR图像分割、图像标准化及影像组学特征提取和筛选；2）临床放射学标志物的获得与筛选；3）融合影像组学特征和临床信息构建预测模型，进行模型的验证和优化。本项目取得的主要成果包括以下：1）构建融合临床和影像组学特征的列线图能够获得满意的预测效能，验证集的AUC可达0.803，为无创预测晚期HGSOC术后疾病残留提供有前景的方式；此外，本项目进一步构建了融合HGSOC腹盆腔转移灶评分的影像组学列线图，AUC高达0.900；项目团队也构建了无创预测晚期HGSOC肠系膜粟粒样转移的影像组学列线图；2）融合MRI影像组学和临床特征的融合模型获得最高的AUC，可达0.86，有助于检出具有高复发风险的HGSOC患者；3）结合三项影像学独立预测特征（缺乏孤立的卵巢肿块、较低的平均ADC值和较低的ADC值中位数）的列线图AUC高达0.816，为临床无创预测间质转化型HGSOC提供了新方法；4）融合MRI形态学特征、DWI的腹膜癌指数及两个临床因素所构建的融合模型的AUC高达0.863，为无创预测减瘤结局提供了有价值的新范式。本项目基于影像组学和DW-MRI研究，在European Radiology, American Journal of Roentgenology, Journal of Magnetic Resonance Imaging等高水平期刊发表SCI论文6篇。申请国家发明专利1项。总之，本团队在影像组学和DW-MRI方面的工作，为晚期HGSOC患者的评估和管理提供了新的思路和新方法，为个体化治疗提供了重要参考。