ABSTRACT

This systematic review and meta-analysis of conventional enhanced magnetic resonance imaging (MRI) were conducted to evaluate the diagnostic performance of imaging features of microvascular invasion (MVI) prediction in hepatocellular carcinoma (HCC).

Relevant studies on diagnosing MVI in HCC by MRI were searched in the MEDLINE, PUBMED, EMBASE, Cochrane library, and Web of Science databases. The pooled mean sensitivity and specificity were calculated using a random effects model. The corresponding positive likelihood ratio (PLR), negative likelihood ratio (NLR), and pooled diagnostic odds ratio (DOR) were calculated. The summary receiver operating characteristic (SROC) curve was used to summarize the overall diagnostic accuracy. Diagnostic performance was evaluated by determining the area under the curve (AUC). Regression analysis by subgroup and sensitivity analysis were used to explore potential sources of heterogeneity.

A total of 19 studies comprising 1920 HCC patients with 2033 tumors were ultimately enrolled. For the signs of the presence of peritumoral enhancement in the arterial phase, peritumoral hypointensity in the hepatobiliary phase, irregular non-smooth margin, and rim-like enhancement in the arterial phase, the pooled sensitivity values, the pooled specificity values, the pooled PLR values, the pooled NLR values, the pooled DOR values, and the values of the AUC of SROC curves were determined.

The conventional MRI features for predicting MVI showed poor diagnostic performance in HCC. Only signs of the presence of peritumoral enhancement in the arterial phase showed a moderate diagnostic accuracy.