基于MRI的机器学习模型和脂代谢重编程相关的分子标签用于术前预测肝癌微血管侵犯的研究

肝细胞癌的增殖、侵袭和迁移需要大量的脂肪酸来提供能量并合成新的细胞膜，脂代谢重编程是肝细胞癌侵袭和转移的关键因素。在前期工作中，申请者在转录组水平发现ACACA、ACSL1、SCD1、ACADSB、FADS1基因表达水平在肝癌有微血管侵犯病人样本中显著改变；免疫组化验证表明，有MVI组的SCD1阳性率显著高于无MVI组；申请者已验证了基于肝癌T2WI序列的机器学习模型可以较好的预测SCD1的表达和MVI。藉此，本项目将进一步利用肝癌临床样本、肝癌细胞株和裸鼠肝癌模型，验证ACACA、ACSL1、SCD1、ACADSB、FADS1等基因的表达水平，通过机器学习建立肝癌脂质代谢分子签名与磁共振影像融合预测模型，并在肝癌微血管侵犯病人样本中加以验证，确定脂代谢与肝癌MVI的关系，明确融合脂代谢分子签名和影像学特征的机器学习模型对肝癌微血管侵犯的预测价值。

The proliferation, invasion and migration of HCC need a lot of fatty acids to provide energy and synthesize new cell membrane. The reprogramming of lipid metabolism is the key factor of invasion and metastasis of HCC. In the preliminary work, the expression level of ACACA, ACSL1, SCD1, ACADSB and FADS1 genes in the samples of patients with microvascular invasion of HCC was significantly changed at the transcriptome level. Immunohistochemistry showed that the positive rate of SCD1 in MVI group was significantly higher than that in MVI group. The applicant has verified that the machine learning model based on T2WI sequence of hepatocellular carcinoma can better predict the expression of SCD1 and the results of MVI. Based on this, the project will further verify the expression level of ACACA, ACSL1, SCD1, ACADSB and FADS1 by using clinical samples of hepatocellular carcinoma, hepatocellular carcinoma cell lines and nude mice hepatocellular carcinoma models, establish a fusion prediction model of hepatocellular carcinoma lipid metabolism molecular signature and MRI image through machine learning, and verify it in the samples of HCC with MVI to determine the relationship between lipid metabolism and MVI. The purpose of this study is to determine the predictive value of a machine learning model that integrates the molecular signatures of lipid metabolism and imaging features for MVI of hepatocellular carcinoma.