Histogram peritumoral enhanced features on MRI arterial phase with extracellular contrast agent can improve prediction of microvascular invasion of hepatocellular carcinoma

Abstract

Background: Preoperative microvascular invasion (MVI) prediction plays an important role in therapeutic decision-making of hepatocellular carcinoma (HCC). This study aimed to investigate the value of histogram based on the arterial phase (AP) of magnetic resonance imaging (MRI) with extracellular contrast agent compared with radiological features for predicting MVI of solitary HCC. Methods: In total, 113 patients with pathologically proven solitary HCC were retrospectively enrolled who received surgical resection and underwent preoperative abdominal MRI. The patients were divided into the ≤3 cm [small HCC (sHCC)] cohort and the >3 cm cohort. Based on pathological analysis of surgical specimens, the patients were classified into MVI negative (MVI−) and MVI positive (MVI+) groups. Peritumoral and intratumoral histogram features [mean, median, standard deviation (Std), coefficient of variation (CV), skewness, kurtosis] were acquired on AP subtraction images and radiological features [size, capsule, corona enhancement, corona enhancement thickness (CET), CET group]. Receiver operating characteristic (ROC) curve was constructed to assess predictive capability. Subgroup analysis of patients with a visible corona enhancement based on the CET cut-off value was performed.Results: None of the features extracted from the intratumor area were significantly different between the MVI+ and MVI− groups in both cohorts. Histogram defined peritumoral (peri-) mean, median, kurtosis, and radiological features including CET and CET group were associated with MVI in sHCCs. Peri-mean, median, Std and radiological features including incomplete capsule, CET, and CET group were associated with MVI in HCC >3 cm. In multivariate logistic regression analysis, the CET group and peri-mean were independent predictors for HCC >3 cm with an area under the curve (AUC) of 0.741. Peri-mean was an independent predictor for sHCC (AUC =0.798). Subgroup analysis of the corona enhancement using 8 mm as a cut-off value showed 100% sensitivity and negative predictive value (NPV).Conclusions: Peritumoral AP enhanced degree on MRI showed an encouraging predictive performance for preoperative prediction of MVI, especially in sHCCs. CET ≤8 mm could be used as a negative predictive marker for MVI.

Performance of multi-regional radiomics features and clinical-radiological variables in the prognostic analysis and risk stratification of single hepatocellular carcinoma

“…Many previous studies on peritumoral characteristics of HCC were based on a single sequence [16,31]. For instance, Wang et al [31] accurately predicted the status of MVI using peritumoral histogram features based on AP images.…”

“…Many previous studies on peritumoral characteristics of HCC were based on a single sequence [16,31]. For instance, Wang et al [31] accurately predicted the status of MVI using peritumoral histogram features based on AP images. Recently, Yu et al [17] developed a hepatobiliary phase image-based radiomics model to predict the vessels encapsulating tumor clusters in HCC, and the peritumoral radiomics model outperformed the intratumoral radiomics model alone.…”

The Role of Radiomics and AI Technologies in the Segmentation, Detection, and Management of Hepatocellular Carcinoma

“…Their combined model showed good performance (AUC = 0.926) in predicting MVI when using data derived from all phases [ 75 ]. Another study [ 76 ] reported encouraging results from analysis of the histogram of the peritumoral region in post-contrast arterial phase MRI images to predict MVI in patients with single HCC lesions. On the contrary, Dai et al [ 77 ] concluded that a CNN model created using radiomic features derived from the hepatobiliary phase of enhanced MRI images using a gradient boosting decision tree (GBDT) classifier had the highest predictive accuracy regarding MVI.…”

Intratumoral and peritumoral radiomics model based on abdominal ultrasound for predicting Ki-67 expression in patients with hepatocellular cancer

“…By using radiomics to analyze peritumoral tissue, it is possible to predict Ki-67 expression in tumors ( 17 ). And in HCC, peritumoral tissue was thought to be associated with MVI and invasiveness ( 18 , 19 ).However, few studies have evaluated the relationship between peritumoral tissues and Ki-67 expression in HCC patients.…”