Role of genomic analysis in the classification of well differentiated hepatocellular lesions

Abstract

Background

The distinction of focal nodular hyperplasia (FNH) and hepatocellular adenoma (HCA) from well-differentiated hepatocellular carcinoma (WD-HCC) in noncirrhotic liver can be challenging. High-grade dysplastic nodule (HGDN) in cirrhosis can have overlapping features with WD-HCC. In some cases, HCA diagnosis is evident but glutamine synthetase (GS) staining is indeterminate for β-catenin activation, which does not allow reliable risk assessment. This study examines the role of genomic analysis in better categorization of WD hepatocellular lesions (WDHL).

Design

Genomic analysis using capture-based NGS assay was done in 23 WDHLs that could not be definitely classified based on morphology, reticulin stain and IHC, and were designated as ‘atypical hepatocellular neoplasms’ (AHNs). GS staining was classified as diffuse homogeneous (moderate to strong staining in >90 % of tumor cells), diffuse heterogeneous (50–90 %), not diffuse (<50 %) and borderline (not clear if more or less than 50 %).

Results

The genomic profile provided additional information for the diagnosis and/or risk assessment enabling a benign diagnosis in 15/23 cases (66 %) and HCC in 4/23 cases (17 %), while the diagnosis remained as atypical in the remaining 4 cases. Of the 4 cases with final HCC diagnosis, findings were suspicious but not diagnostic based on morphology/IHC; additional changes like TERT promoter mutation (n = 2), AXIN mutation (n = 1), CDKN2A loss (n = 2) and copy number alterations (n = 3) helped to support HCC. Of the 15 cases with a final benign diagnosis, the status of β-catenin activation was unclear based on GS stain in 8 cases, 2 of which showed CTNNB1 exon 7 mutation, 1 showed CTNNB1 exon 8 mutation, while genomic changes in 5 cases did not show any evidence of Wnt activation. FNH-like features were seen in 2 cases, but the genomic changes excluded FNH (CTNNB1 and ARID1A mutation). The final diagnosis was unchanged from the initial diagnosis of AHN in 4/23 cases (17 %) as the molecular findings did not favor HCC.

Conclusion

Genomic changes were helpful in characterization of WDHLs, supporting HCC in 17 % of cases and clarifying β-catenin activation status in all 7 cases with borderline GS staining. Genomic changes are not specific but can provide diagnostic clues in selected challenging cases that cannot be classified on morphology and IHC. Given the significant treatment implications of distinguishing between HCC and benign/premalignant entities, routine use of genomic analysis in diagnostically challenging settings should be considered.