Aims

Hepatic angiomyolipoma (AML) often shows epithelioid morphology with inconspicuous fat. Epithelioid component can mimic hepatocellular adenoma (HCA) or carcinoma (HCC). The aims of this study were to examine the expression of commonly used markers for HCA or HCC in hepatic AML and highlight pitfalls in diagnosis.

Methods and results

Resected hepatic AMLs (n = 16) were reviewed; reticulin stain, immunohistochemistry for glutamine synthetase (GS), β-catenin and liver fatty acid binding protein (LFABP) were performed along with Sanger sequencing of exon 3 of CTNNB1 and next-generation sequencing (NGS). Predominant epithelioid component (≥50%) was seen in 80% of cases. Foamy macrophage was present in 33% of cases. High-risk histological features were often present in tumours with benign outcome: marked atypia (19%), mitoses (20%) and necrosis (33%). GS staining (≥10% of tumour) was seen in epithelioid components in 13 (87%) cases, and was diffuse (>50% of tumour) in six (40%) cases. LFABP staining or nuclear β-catenin staining was not seen in any case. Sanger sequencing and NGS did not reveal CTNNB1 mutation in any tested case. NGS demonstrated TSC2 mutations in all five cases tested.

Conclusions

The predominance of epithelioid component resembling HCA or HCC is common in hepatic AML. Absence of LFABP and presence of fat can be mistaken for HNF1α-inactivated HCA. Diffuse GS staining can be mistaken for β-catenin-activated HCA or HCC. Diffuse GS expression is not related to CTNNB1 mutation. All tested cases showed TSC2 mutation, supporting this as the driving genetic event for hepatic AML.

The current clinical practice is based on the assumption of strong correlation between diffuse glutamine synthetase expression and β-catenin activation in hepatocellular adenoma and hepatocellular carcinoma. This high correlation is based on limited data and may represent an oversimplification as glutamine synthetase staining patterns show wide variability in clinical practice. Standardized criteria for interpreting diverse glutamine synthetase patterns, and the association between each pattern and β-catenin mutations is not clearly established. This study examines the correlation between glutamine synthetase staining patterns and β-catenin mutations in 15 typical hepatocellular adenomas, 5 atypical hepatocellular neoplasms and 60 hepatocellular carcinomas. Glutamine synthetase staining was classified into one of the three patterns: (a) diffuse homogeneous: moderate-to-strong cytoplasmic staining in >90% of lesional cells, without a map-like pattern, (b) diffuse heterogeneous: moderate-to-strong staining in 50-90% of lesional cells, without a map-like pattern, and (c) patchy: moderate-to-strong staining in <50% of lesional cells (often perivascular), or weak staining irrespective of the extent, and all other staining patterns (including negative cases). Sanger sequencing of CTNNB1 exon 3 was performed in all cases. Of hepatocellular tumors with diffuse glutamine synthetase staining (homogeneous or heterogeneous), an exon 3 β-catenin mutation was detected in 33% (2/6) of typical hepatocellular adenoma, 75% (3/4) of atypical hepatocellular neoplasm and 17% (8/47) of hepatocellular carcinomas. An exon 3 mutation was also observed in 15% (2/13) of hepatocellular carcinomas with patchy glutamine synthetase staining. The results show a modest correlation between diffuse glutamine synthetase immunostaining and exon 3 β-catenin mutations in hepatocellular adenoma and hepatocellular carcinoma with discrepancy rates >50% in both hepatocellular adenoma and hepatocellular carcinoma. The interpretation of β-catenin activation based on glutamine synthetase staining should be performed with caution, and the undetermined significance of various glutamine synthetase patterns should be highlighted in pathology reports.