Purpose We aimed to develop and validate a radiomics model for differentiating hepatocellular carcinoma (HCC) from focal nodular hyperplasia (FNH) in non-cirrhotic livers using Gd-DTPA contrast-enhanced magnetic resonance imaging (MRI). Methods We retrospectively enrolled 149 HCC and 75 FNH patients treated between May 2015 and May 2019 at our center. Patients were randomly allocated to a training (n=156) and validation set (n=68). In total, 2260 radiomics features were extracted from the arterial phase and portal venous phase of Gd-DTPA contrast-enhanced MRI. Using Max-Relevance and Min-Redundancy, random forest, least absolute shrinkage, and selection operator algorithm for dimensionality reduction, multivariable logistic regression was used to build the radiomics model. A clinical model and combined model were also established. The diagnostic performance of the models was compared. Results Eight radiomics features were chosen for the radiomics model, and four clinical factors (age, sex, HbsAg, and enhancement pattern) were chosen for the clinical model. A combined model was built using the factors from the previous models. The classification accuracy of the combined model differentiated HCC from FNH in both the training and validation sets (0.956 and 0.941, respectively). The area under the receiver operating characteristic curve of the combined model was significantly better than that of the clinical model for both the training (0.984 vs. 0.937, p=0.002) and validation (0.972 vs. 0.903, p=0.032) sets. Conclusions The combined model provided a non-invasive quantitative method for differentiating HCC from FNH in non-cirrhotic liver with high accuracy. Our model may assist clinicians in the clinical decision-making process.

Background: Hepatocellular carcinoma (HCC) is one of the most common aggressive malignancies with increasing incidence worldwide. The oncogenic roles of transcription factors (TFs) were increasingly recognized in various cancers. This study aimed to develop a predicting signature based on TFs for the prognosis and treatment of HCC.Methods: Differentially expressed TFs were screened from data in the TCGA-LIHC and ICGC-LIRI-JP cohorts. Univariate and multivariate Cox regression analyses were applied to establish a TF-based prognostic signature. The receiver operating characteristic (ROC) curve was used to assess the predictive efficacy of the signature. Subsequently, correlations of the risk model with clinical features and treatment response in HCC were also analyzed. The TF target genes underwent Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses, followed by protein-protein-interaction (PPI) analysis.Results: A total of 25 differentially expressed TFs were screened, 16 of which were related to the prognosis of HCC in the TCGA-LIHC cohort. A 2-TF risk signature, comprising high mobility group AT-hook protein 1 (HMGA1) and MAF BZIP transcription factor G (MAFG), was constructed and validated to negatively related to the overall survival (OS) of HCC. The ROC curve showed good predictive efficiencies of the risk score regarding 1-year, 2-year and 3-year OS (mostly AUC >0.60). Additionally, the risk score independently predicted OS for HCC patients both in the training cohort of TCGA-LIHC dataset (HR = 2.498, p = 0.007) and in the testing cohort of ICGC-LIRI-JP dataset (HR = 5.411, p < 0.001). The risk score was also positively correlated to progressive characteristics regarding tumor grade, TNM stage and tumor invasion. Patients with a high-risk score were more resistant to transarterial chemoembolization (TACE) treatment and agents of lapatinib and erlotinib, but sensitive to chemotherapeutics. Further enrichment and PPI analyses demonstrated that the 2-TF signature distinguished tumors into 2 clusters with proliferative and metabolic features, with the hub genes belonging to the former cluster.Conclusion: Our study identified a 2-TF prognostic signature that indicated tumor heterogeneity with different clinical features and treatment preference, which help optimal therapeutic strategy and improved survival for HCC patients.

Purpose:This study aimed to develop and validate a radiomics model for differentiating between hepatocellular carcinoma (HCC) and focal nodular hyperplasia (FNH) in non-cirrhotic livers using Gd-DTPA contrast-enhanced magnetic resonance imaging (MRI).Methods:We retrospectively enrolled 149 HCC patients and 75 FNH patients seen between May 2015 and May 2019 at our center and randomly allocated patients to a training set (n = 156) and a validation set (n = 68). A total of 2,260 radiomics features were extracted from the arterial phase and portal venous phase of Gd-DTPA contrast-enhanced MRI. Using Max-Relevance and Min-Redundancy, random forests, and the least absolute shrinkage and selection operator algorithm for dimensionality reduction, multivariable logistic regression was used to build the radiomics model. A clinical model and combined model were also established. The diagnostic performance of the three models was compared. Results:Eight radiomics features were chosen to build a radiomics model, and four clinical factors (age, sex, HbsAg, and enhancement pattern) were chosen to build the clinical model. When evaluating the performance of three models, the clinical model that included clinical data and visual MRI findings achieved excellent performance in the training set (AUC, 0.937; 95% CI, 0.887–0.970) and the validation set (AUC, 0.903; 95% CI, 0.807–0.962), and there was no significant difference between the radiomics model and the clinical model. The AUC of the combined model was significantly better than that of the clinical model for both the training (0.984 vs. 0.937, p = 0.002) and validation (0.972 vs. 0.903, p = 0.032) sets.Conclusions:The combined model based on clinical and radiomics features can well distinguish HCC from FNH in non-cirrhotic liver. Our model may assist clinicians in the clinical decision-making process.

Background Hepatectomy is currently the most effective modality for the treatment of intrahepatic cholangiocarcinoma (ICC). The status of the lymph nodes directly affects the choice of surgical method and the formulation of postoperative treatment plans. Therefore, a preoperative judgment of lymph node status is of great significance for patients diagnosed with this condition. Previous prediction models mostly adopted logistic regression modeling, and few relevant studies applied random forests in the prediction of ICC lymph node metastasis (LNM). Methods A total of 149 ICC patients who met clinical conditions were enrolled in the training group. Taking into account preoperative clinical data and imaging features, 21 indicators were included for analysis and modeling. Logistic regression was used to filter variables through multivariate analysis, and random forest regression was used to rank the importance of these variables through the use of algorithms. The model’s prediction accuracy was assessed by the concordance index (C-index) and calibration curve and validated with external data. Result Multivariate analysis shows that Carcinoembryonic antigen (CEA), Carbohydrate antigen19-9 (CA19-9), and lymphadenopathy on imaging are independent risk factors for lymph node metastasis. The random forest algorithm identifies the top four risk factors as CEA, CA19-9, and lymphadenopathy on imaging and Aspartate Transaminase (AST). The predictive power of random forest is significantly better than the nomogram established by logistic regression in both the validation group and the training group (Area Under Curve reached 0.758 in the validation group). Conclusions We constructed a random forest model for predicting lymph node metastasis that, compared with the traditional nomogram, has higher prediction accuracy and simultaneously plays an auxiliary role in imaging examinations.