肿瘤基因组学的最新研究揭示了肝细胞癌（HCC）发病的核心驱动因子。我们的研究旨在探究MRI特征是否可以作为非侵入性标志物，用于预测HCC常见的遗传亚型。对42位患者的43个病理证实的HCC进行了基因测序，共测了447个与癌症相关的基因。此外，对这些患者进行了增强MRI随后进行活检或切除。回顾性评估了MRI特征，包括肿瘤大小，浸润性肿瘤边缘，扩散限制，动脉期强化，非周边性冲洗，增强包膜，周围肿瘤上皮细胞增生，静脉内肿瘤，肿块中的脂肪和血液制品，肝硬化和肿瘤异质性。使用Fisher确切检验将遗传亚型与影像特征相关联。评估使用相关MRI特征进行基因亚型的预测表现和读者间的一致性。 最常见的两种基因突变是TP53（13/43，30％）和CTNNB1（17/43，40％）。具有TP53突变的肿瘤更常在MRI上表现出浸润性肿瘤边缘（p = 0.01）；读者间协议几乎完美（kappa = 0.95）。CTNNB1突变与MRI上的周围肿瘤上皮细胞增生有关（p = 0.04），读者间协议为（kappa = 0.74）。MRI上的浸润性肿瘤边缘与TP53突变的相关性的准确性、灵敏度和特异性分别为74.4％，61.5％和80.0％。周围肿瘤上皮细胞增生与CTNNB1突变的相关性的准确性、灵敏度和特异性分别为69.8％，47.0％和84.6％。 在HCC中，MRI上的浸润性肿瘤边缘与TP53突变相关，周围肿瘤上皮细胞增生与CTNNB1突变相关。这些MRI特征的缺失可能是相应HCC遗传亚型的潜在负面预测因子，对预后和治疗反应产生影响。 ©2023作者，独家授权Springer Science+Business Media，LLC，Springer Nature的一部分。

Recent studies in cancer genomics have revealed core drivers for hepatocellular carcinoma (HCC) pathogenesis. We aim to study whether MRI features can serve as non-invasive markers for the prediction of common genetic subtypes of HCC.Sequencing of 447 cancer-implicated genes was performed on 43 pathology proven HCC from 42 patients, who underwent contrast-enhanced MRI followed by biopsy or resection. MRI features were retrospectively evaluated including tumor size, infiltrative tumor margin, diffusion restriction, arterial phase hyperenhancement, non-peripheral washout, enhancing capsule, peritumoral enhancement, tumor in vein, fat in mass, blood products in mass, cirrhosis and tumor heterogeneity. Fisher's exact test was used to correlate genetic subtypes with imaging features. Prediction performance using correlated MRI features for genetic subtype and inter-reader agreement were assessed.The two most prevalent genetic mutations were TP53 (13/43, 30%) and CTNNB1 (17/43, 40%). Tumors with TP53 mutation more often demonstrated an infiltrative tumor margin on MRI (p = 0.01); inter-reader agreement was almost perfect (kappa = 0.95). The CTNNB1 mutation was associated with peritumoral enhancement on MRI (p = 0.04), inter-reader agreement was substantial (kappa = 0.74). The MRI feature of an infiltrative tumor margin correlated with the TP53 mutation with accuracy, sensitivity, and specificity of 74.4%, 61.5% and 80.0%, respectively. Peritumoral enhancement correlated with the CTNNB1 mutation with accuracy, sensitivity, and specificity of 69.8%, 47.0% and 84.6%, respectively.An infiltrative tumor margin on MRI correlated with TP53 mutation and peritumoral enhancement correlated with CTNNB1 mutation in HCC. Absence of these MRI features are potential negative predictors of the respective HCC genetic subtypes that have implications for prognosis and treatment response.© 2023. The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.