Clin Mol Hepatol.  2023 Jul;29(3):733-746. 10.3350/cmh.2023.0034.

Classification of microvascular invasion of hepatocellular carcinoma: correlation with prognosis and magnetic resonance imaging

• Hwang YJ
• Bae JS
• Lee Y
• Hur BY
• Lee DH
• Kim H

• KMID: 2543991
• DOI: http://doi.org/10.3350/cmh.2023.0034

Abstract

Background/Aims
The microvascular invasion (MVI) of hepatocellular carcinoma (HCC) involves a wide histological spectrum, and it is unclear whether the degree of MVI correlates with patient prognosis or imaging findings. Here, we evaluate the prognostic value of MVI classification and analyze the radiologic features predictive of MVI.
Methods
Using a retrospective cohort of 506 patients with resected solitary HCCs, the histological and imaging features of MVI were reviewed and correlated with clinical data.
Results
MVI-positive HCCs invading ≥5 vessels or those with ≥50 invaded tumor cells were significantly associated with decreased overall survival (OS). The 5-year OS, recurrence-free survival (RFS), and beyond Milan criteria RFS rates were significantly poorer in patients with severe MVI compared with those with mild or no MVI. Severe MVI was a significant independent predictive factor for OS (odds ratio [OR], 2.962; p<0.001), RFS (OR, 1.638; p=0.002), and beyond Milan criteria RFS (OR, 2.797; p<0.001) on multivariable analysis. On MRI, non-smooth tumor margins (OR, 2.224; p=0.023) and satellite nodules (OR, 3.264; p<0.001) were independently associated with the severe-MVI group on multivariable analysis. Both non-smooth tumor margins and satellite nodules were associated with worse 5-year OS, RFS, and beyond Milan criteria RFS.
Conclusions
Histologic risk classification of MVI according to the number of invaded microvessels and invading carcinoma cells was a valuable predictor of prognosis in HCC patients. Non-smooth tumor margin and satellite nodules were significantly associated with severe MVI and poor prognosis.

Keyword

Hepatocellular carcinoma; Microvascular invasion; Prognosis; Magnetic resonance imaging