Liver transplantation for hepatocellular carcinoma: need for a new patient selection strategy

Gd-EOB-DTPA-enhanced MRI radiomic features for predicting histological grade of hepatocellular carcinoma

BACKGROUND

Prediction models for the histological grade of hepatocellular carcinoma (HCC) remain unsatisfactory. The purpose of this study is to develop preoperative models to predict histological grade of HCC based on gadolinium-ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) radiomics. And to compare the performance between artificial neural network (ANN) and logistic regression model.

METHODS

A total of 122 HCCs were randomly assigned to the training set (n=85) and the test set (n=37). There were 242 radiomic features extracted from volumetric of interest (VOI) of arterial and hepatobiliary phases images. The radiomic features and clinical parameters [gender, age, alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), alanine aminotransferase (ALT), aspartate transaminase (AST)] were selected by permutation test and decision tree. ANN of arterial phase (ANN-AP), logistic regression model of arterial phase (LR-AP), ANN of hepatobiliary phase (ANN-HBP), logistic regression mode of hepatobiliary phase (LR-HBP), ANN of combined arterial and hepatobiliary phases (ANN-AP + HBP), and logistic regression model of combined arterial and hepatobiliary phases (LR-AP + HBP) were built to predict HCC histological grade. Those prediction models were assessed and compared.

RESULTS

ANN-AP and LR-AP were composed by AST and radiomic features based on arterial phase. ANN-HBP and LR-HBP were composed by AFP and radiomic features based on hepatobiliary phase. ANN-AP + HBP and LR-AP + HBP were composed by AST and radiomic features based on arterial and hepatobiliary phases. The prediction models could distinguish between high-grade tumors [Edmondson-Steiner (E-S) grade III and IV] and low-grade tumors (E-S grade I and II) in both training set and test set. In the test set, the AUCs of ANN-AP, LR-AP, ANN-HBP, LR-HBP, ANN-AP + HBP and LR-AP + HBP were 0.889, 0.777, 0.941, 0.819, 0.944 and 0.792 respectively. The ANN-HBP was significantly superior to LR-HBP (P=0.001). And the ANN-AP + HBP was significantly superior to LR-AP + HBP (P=0.007).

CONCLUSIONS

Prediction models consisting of clinical parameters and Gd-EOB-DTPA-enhanced MRI radiomic features (based on arterial phase, hepatobiliary phase, and combined arterial and hepatobiliary phases) could distinguish between high-grade HCCs and low-grade HCCs. And the ANN was superior to logistic regression model in predicting histological grade of HCC.

Preoperative prediction of hepatocellular carcinoma tumour grade and micro-vascular invasion by means of artificial neural network: a pilot study

BACKGROUND & AIMS

Hepatocellular carcinoma (HCC) prognosis strongly depends upon nuclear grade and the presence of microscopic vascular invasion (MVI). The aim of this study was to develop an artificial neural network (ANN) that is able to predict tumour grade and MVI on the basis of non-invasive variables.

METHODS

Clinical, radiological, and histological data from 250 cirrhotic patients resected (n=200) or transplanted (n=50) for HCC were analyzed. ANN and logistic regression models were built on a training group of 175 randomly chosen patients and tested on the remaining testing group of 75. Receiver operating characteristics curve (ROC) and k-statistics were used to analyze model accuracy in the prediction of the final histological assessment of tumour grade (G1-G2 vs. G3-G4) and MVI (absent vs. present).

RESULTS

Pathologic examination showed G3-G4 in 69.6% of cases and MVI in 74.4%. Preoperative serum alpha-fetoprotein (AFP), tumour number, size, and volume were related to tumour grade and MVI (p<0.05) and were used for ANN building, whereas, tumour number did not enter into the logistic models. In the training group, ANN area under ROC curves (AUC) for tumour grade and MVI prediction were 0.94 and 0.92, both higher (p<0.001) than those of logistic models (0.85 for both). In the testing group, ANN correctly identified 93.3% of tumour grades (k=0.81) and 91% of MVI (k=0.73). Logistic models correctly identified 81% of tumour grades (k=0.55) and 85% of MVI (k=0.57).

CONCLUSION

ANN identifies HCC tumour grades and MVI on the basis of preoperative variables more accurately than the conventional linear model and should be used for tailoring clinical management.