能谱CT结合影像纹理分析技术无创评价肺癌瘤内乏氧微环境状态的实验研究

Tumor hypoxia increases malignant phenotype, which is an important factor to chemoradiation resistance and poor prognosis. But so far there is no effective lung cancer in vivo hypoxia detection technology. Multi-parameter imaging of spectral CT with unique energy resolution and the resolution of the physical and chemical properties, can effectively show the tissue microenvironment. Our previous experimental results suggest that spectral CT iodine-based substances imaging and related parameters can reflect the lung tumor vascularization and heterogeneity, and image texture analysis technology can quantify subtle differences in image texture features, reflecting tumor hypoxia microenvironment difficultly to be observed visually. But quantitative analysis and spatial distribution of tumor hypoxia should be further explored. To this end, we intend to create a rabbit VX2 lung xenograft model, using the spectral CT imaging and texture features to obtain the parameters of tumor hypoxia. Therefore, this study aims to use image texture analysis based on spetral CT image to compare the texture features with immunohistochemical markers, to evaluate the the accuracy and relevance of this texture analysis method, and to establish radiological predictive model which reflect the degree of tumor hypoxia. This study from this new viewpoint of hypoxia imaging of spectral CT to lay a foundation for revealing tumor hypoxia and the variation of the microenvironment, and to provide adequate scientific basis for tumor individualized treatment.

肺癌乏氧是肿瘤恶性表型增加、对放化疗抗拒和患者预后不良的重要因素。但是到目前为止还没有有效的肺癌活体乏氧检测技术。能谱CT实现了CT的多参数功能成像, 可有效显示组织微环境。我们的预实验结果提示能谱CT功能成像及相关参数能够反映肺癌瘤内血管化及异质性,而图像纹理分析技术则可以将细微的纹理差异通过不同图像纹理特征量化,反映肉眼难以观察的瘤内乏氧微环境，但对肿瘤乏氧的定量分析及空间分布规律有待进一步探讨。为此，我们拟建立兔VX2肺移植瘤模型，获得肿瘤血管化及乏氧参数，并与免疫组织化学乏氧标志物进行对照研究，明确能谱CT成像结合图像纹理分析技术表征肿瘤乏氧的准确性及相关性，建立反映肿瘤乏氧程度的影像学预测模型。本研究将从能谱CT乏氧成像这个新视点为揭示肿瘤乏氧及微环境的变化规律奠定基础，为临床上开展肿瘤个体化治疗提供充分的科学依据。

肺癌是世界范围内肿瘤相关死亡的首要原因之一。肿瘤中的乏氧微环境是肺癌患者放疗抵抗的重要原因，能谱CT成像及相关参数能够反映肺癌瘤内血管化及异质性,而图像纹理分析则可以将不同图像纹理特征量化,反映瘤内乏氧微环境的差异。本研究建立兔VX2肺移植瘤模型，进行能谱CT成像以及纹理分析处理获得相关定量参数，并与免疫组织化学染色结果(肿瘤乏氧探针染色、HIF-1α表达水平及MVD计数)进行对照研究。研究结果发现能谱CT参数强乏氧组肿瘤静脉期标准化碘浓度（NIC）及静脉期能谱曲线斜率（λHU）显著高于弱乏氧组参数（P<0.05）。强乏氧组肿瘤较弱乏氧组肿瘤有更高的HIF-1α表达及MVD计数( P<0.05 )。Spearman相关性分析结果显示，能谱CT参数静脉期NIC及静脉期λHU与肿瘤乏氧探针染色、HIF-1α表达水平及MVD计数存在显著正相关性(r=0.609, r=0.608, r=0.432, r=0.472, r=0.457, r=0.529, P<0.05)。ROC分析结果显示，静脉期λHU具有较好的预测肿瘤乏氧状态的诊断效能，AUC为0.883，敏感性和特异性分别为85.7%和100%。纹理分析结果，65Kev最佳单能量图像分析显示纹理分析参数强乏氧组肿瘤动脉期峰度低于弱乏氧组肿瘤（P=0.046），强乏氧组肿瘤静脉期熵高于弱乏氧组肿瘤（P=0.012）。ROC分析结果显示，静脉期熵具有较好的预测肿瘤乏氧情况的诊断效能，AUC为0.789，敏感性和特异性分别为92.9%和63.6%。结合能谱CT参数及纹理分析，静脉期NIC结合静脉期熵较好的预测肿瘤乏氧情况的诊断效能，AUC为0.948，敏感性和特异性分别为92.9%和90.9%。能谱CT定量参数结合纹理分析在预测肿瘤乏氧状态方面具有较高的价值。