Clear Cell Variant of Fibrolamellar Carcinoma of the Liver

Fibrolamellar carcinoma of the liver is an uncommon but distinctive type of hepatocellular carcinoma (HCC) first described by Edmondson in 1956 and later popularized in the 1980s by Berman et al1 and Craig et al.2 It is characterized by trabeculae of polygonal oxyphilic cells separated by parallel lamellae of hyalinized collagen. Distinction from conventional HCC is of clinical significance because the prognosis is better due to higher resectability and more indolent growth.2 It typically occurs in noncirrhotic livers of younger patients. There is no association with hepatitis B virus, and the serum α-fetoprotein level is often not elevated.

In this report, we describe a case of fibrolamellar carcinoma with a prominent clear cell component, a feature not previously documented in the English-language literature.2,3 

A 59-year-old woman, known to be a hepatitis B virus carrier, was found to have a liver mass on routine checkup. Computed tomography showed a solitary nodule in the right lobe of the liver. The serum α-fetoprotein level was not elevated. Right hepatectomy was performed on July 1999. The patient was well, with no evidence of recurrence, in the latest follow-up 13 months after the operation.

Hematoxylin-eosin–stained slides were prepared from paraffin blocks according to standard protocols. Histochemical stains for periodic acid–Schiff (PAS), diastase-PAS, Alcian blue (pH 2.5), and trichrome were performed. Immunohistochemical staining with the labeled streptavidin-biotin-peroxidase detection system was performed using the Ventana automated immunostainer (Ventana, Tucson, Ariz). The antibodies HEP-PAR1 (hepatocyte-specific antigen, clone OCH1E5, 1:25, Dakopatts, Glostrup, Denmark), mitochondria Ab-1 (clone 113–1, 1:25, NeoMarkers, Freemont, Calif), and hepatitis B surface antigen (clone S1–210, 1:100, Signet, Dedham, Mass) were applied.

For electron microscopy, formalin-fixed tumor tissue was further fixed in 2.5% (wt/vol) glutaraldehyde. The tissue was subsequently postfixed in osmium tetroxide and embedded in epoxy resin. Ultrathin sections were double stained with uranyl acetate–lead citrate and examined with a Philips CM100 transmission electron microscope (Philips Electron Optics, Eindhoven, The Netherlands).

The tumor, measuring 50 × 40 × 25 mm, was a predominantly circumscribed, bosselated, and firm mass situated in the subcapsular area of the right lobe of the liver. It was light tan and interspersed with concentric whitish streaks on the cut surface (Figure 1). The surrounding liver tissue did not show cirrhosis.

Microscopic examination revealed a tumor with pushing borders composed of trabeculae and small islands of polygonal cells with focal formation of pseudoglandular structures (Figure 2, A). The tumor was traversed by parallel strands of hyalinized collagen sprinkled with small lymphocytes. The cytoarchitectural features were consistent with fibrolamellar carcinoma. The peculiar concentric streaks observed in the gross specimen were due to coalesced zones of fibrosis, with the collagen fibers being oriented radially (Figure 2, B). In focal areas where there was scanty fibrous stroma, sinusoids separated the trabeculae of tumor cells.

There were 2 populations of tumor cells, namely oncocytic cells and clear cells. The oncocytic cells had granular eosinophilic cytoplasm, and some contained pale bodies and cytoplasmic hyaline globules (Figure 2, A). They blended imperceptibly with the clear cells, which comprised 50% of the total cell population (Figure 2, C). Transitional forms between these 2 cell types were evident. The clear cells were larger than the oncocytic cells and had abundant clear cytoplasm that was not water clear but exhibited lightly eosinophilic flocculent or reticulated strands (Figure 2, D). In both cell types, the nuclei were central or eccentric, demonstrating moderate nuclear atypia and distinct nucleoli.

Except for a few small clusters, both the clear cells and oncocytic cells were negative for glycogen (PAS and diastase-PAS negative). A small proportion of clear cells showed weak staining for Alcian blue. Trichrome stain produced a densely granular appearance in the cytoplasm of the oncocytic cells and a sparsely granular to finely reticulated pattern in the cytoplasm of the clear cells. The hyaline globules in the oncocytic cells were highlighted by PAS, diastase-PAS, and trichrome stains.

Both the clear cells and oncocytic cells showed diffuse strong granular staining for the hepatocyte-specific antigen (HEP-PAR1). Antimitochondria antibody produced a granular cytoplasmic staining in the oncocytic cells, but a rarefied and weaker staining in the clear cells similar to the reticulated pattern demonstrated by trichrome (Figure 3). Hepatitis B surface antigen was present in the adjacent nonneoplastic hepatocytes, but not in the neoplastic cells.

The ultrastructural tissue preservation was suboptimal, but the oncocytic cells were evidently packed with mitochondria. Within the cytoplasm of the clear cells, there were large empty-looking vesicular structures that probably represented ballooned mitochondria, as evidenced by the presence of occasional cristae (Figure 4). Some normal-sized mitochondria were present between the vesicular structures. There were tumor cells that showed abundant normal mitochondria mixed with some dilated mitochondria, corresponding to the transitional form seen on light microscopy. Glycogen particles were sparse.

Cytoplasmic clearing can be observed in many tumor types because of accumulation of cytoplasmic vesicles, glycogen, lipopolysaccharides or mucopolysaccharides, mitochondrial swelling, or tissue-processing artifact.4 In this reported case, clear cells accounted for a significant component of the fibrolamellar carcinoma, a feature not previously described in the English-language literature.2,3 Initially, we considered the possibility of “acquired storage disease,” resulting from enzyme defects in the neoplastic cells (such as glycogen or mucopolysaccharides), as the cause for cytoplasmic clearing, but we found no histochemical support for this possibility. The presence of vesicles with occasional cristae on electron microscopy indicated that the clearing was due to ballooning of mitochondria. The positive staining in a reticulated pattern with trichrome and antimitochondria antibody further supported that the membranes of the vesicles were derived from mitochondria. In fact, small numbers of swollen mitochondria have been described in classic fibrolamellar carcinoma.5 Cystic or vesicular degeneration of mitochondria has similarly been reported to account for clear cell change in oncocytic thyroid neoplasms,6 uterine epithelioid leiomyoma,7 and uterine intravenous leiomyoma.8 In such cases, the clear cells still retain a finely reticulated quality, reminiscent of washed-out oncocytes.

Clear cells have been recognized to occur in conventional HCC,3 either focally or predominantly; the term clear cell HCC or clear cell carcinoma is sometimes applied for the latter circumstance.9 The clarity of the cytoplasm is most often due to accumulation of glycogen or, rarely, fat or water. Clear cell fibrolamellar carcinoma can potentially be misdiagnosed as clear cell HCC, especially in core biopsy specimens when the typical oncocytic cells are not included. However, the cytoplasm of the clear cell HCC is usually water clear, and the glycogen content can be demonstrated by positive PAS staining that is digestible with diastase. Furthermore, the characteristic lamellae of hyalinized collagen are lacking. It is important to distinguish between the 2 because fibrolamellar HCC is associated with a more favorable prognosis than clear cell HCC, which behaves no differently from conventional HCC.9 However, more studies are required to ascertain whether the clear cell variant of fibrolamellar HCC shares the same clinical course with the classic fibrolamellar HCC. Fibrolamellar carcinoma does not show association with hepatitis B virus. The hepatitis B positivity in the current case is probably incidental because this virus is prevalent in the Chinese population and there is no evidence of cirrhosis in the liver.10