The Morphologic Distinctions Between Hepatocellular Carcinoma
and Metastatic Carcinomas in Fine Needle Aspiration Cytology
Kim R. Geisinger
Wake Forest University
The liver is one of the major targets for deep organ fine needle aspiration biopsies. The
majority of these are performed to diagnose malignant neoplasms involving the liver. Hepatocellular
carcinoma (hepatoma) accounts for only 2% to 3% of all cancers involving the liver in the United States.
Still, hepatoma is the most frequent primary malignancy of this organ. Furthermore, hepatoma is much
more frequent in some parts of the world such as subSaharan Africa and Southeast Asia. Ninety-five
percent of all hepatic malignancies represent metastases. Common primary sites include adenocarcinomas
of the stomach, pancreas, and large bowel. These organs, as well as others, have direct portal drainage
and thus, the liver is a common site for metastatic disease.
Both hepatoma and metastatic carcinomas are associated with a poor prognosis. Yet, the
distinction between the two has clinical importance from both therapeutic and prognostic viewpoints. The
prognosis is extremely dismal in this country for individuals with a hepatoma, whereas a variety of
therapeutic interventions have improved the long term outlook for individuals with metastatic carcinoma,
especially those with a solitary metastasis in the right lobe.
As a general introduction, the aspiration cytology of hepatoma correlates to an extent
with the histologic grade of this neoplasm.  Moderately differentiated hepatomas are usually
easy to diagnose in aspiration smears. On the other hand, it may be very difficult at times to
distinguish benign hepatocytes from a well differentiated hepatoma.
At the other end of
the spectrum are poorly differentiated hepatocellular carcinomas. Here the problem is distinguishing it
from a metastasis.
Another aspect of the distinction between hepatoma and metastases is related to
morphologic variants of this primary malignancy. The latter include the small cell type, the clear cell
type, and pleomorphic hepatocellular carcinoma.
Well to moderately differentiated hepatocellular carcinomas are typically composed of
tumor cells with polygonal contours and sharply defined cellular borders.
In most cells,
the cytoplasm is opaque and granular in quality and scant to moderate in volume. A proportion of these
malignant cells may manifest cytoplasmic inclusions, including fat vacuoles and lipofuscin.
Characteristically, the tumor cells possess one or more centrally positioned nuclei which may appear
relatively perfectly round. The tumor cells are said to be more hyperchromatic and to possess more
prominent nucleoli than benign hepatocytes. A consistent attribute of hepatoma is an increase in the
nuclear-to-cytoplasmic ratios of the vast majority of the malignant cells (compared to benign
hepatocytes). Another important morphologic feature is the association of endothelial cells with the
malignant hepatocytes. Trabeculae can be defined as aggregates of neoplastic hepatocytes which are at
least partially surrounded by endothelial cells. In addition, sinusoidal capillaries are frequently seen
separating masses of hepatoma cells. Interestingly, the endothelial cells in hepatoma may be
phenotypically (and possibly genetically) different from those in benign liver.  The smear
background frequently contains large hepatocytic nuclei, at times with distinct nucleoli, and without any
visible attached cytoplasm.
In many instances, the distinction between hepatoma and a metastasis is straightforward on
the basis of routine microscopy. In an attempt to enhance our ability to make this distinction, Bottles
et al evaluated a series of hepatomas and metastases in aspiration biopsies by logistic regression
analysis.  Three features emerged as significant for hepatoma: polygonally shaped tumor
cells with centrally positioned nuclei, tumor cells separated by sinusoidal capillaries, and cytoplasmic
bile; the last feature was 100% specific but relatively insensitive. This series included metastatic
carcinomas, lymphomas, sarcomas, and melanomas.
More recently, Salamào evaluated 52 hepatomas and 56 metastatic adenocarcinomas for 11
cytomorphologic features; this data was also evaluated by regression analysis.  Two features
emerged as significant for hepatoma: polygonally shaped tumor cells with centrally positioned nuclei and
sinusoidal capillaries separating the neoplastic hepatocytes. In addition, one feature was significant
for metastatic adenocarcinoma, namely, the presence of lumen formation.
The Salamào study also investigated the variability in identification of specific
cytomorphologic features by the individual five pathologists.  For example, the overall
prevalence for the presence of malignant cells with polygonal contours and round, central nuclei was 68%
of the specimens. However, the range was from 56% to 74% among the five individuals. Similarly, the
overall prevalence for the presence of sinusoidal capillaries was 42%, with a range of 26% to 51%.
Although quite uncommon, primary adenocarcinoma of the liver (cholangiocarcinoma) needs to
be distinguished from both hepatoma and metastatic adenocarcinoma.  The former is generally
easy in that cholangiocarcinomas are usually well differentiated histologically, frequently showing well
formed glands with central lumens. The difficulty arises in distinguishing cholangiocarcinoma from
metastatic adenocarcinoma in smears. We are unaware of any single cytomorphologic attribute which allows
one to make this distinction clearly. Furthermore, immunocytochemistry is generally not of value (except
for distinguishing it from colonic carcinomas which are usually negative for cytokeratin 7). In contrast
to colonic carcinomas, the smear background in cholangiocarcinomas is usually clean in that it lacks a
tumor diathesis. In addition, smear cellularity may be relatively low, compared to many metastases.
One of the most common morphological variants of hepatocellular carcinoma is the
fibrolamellar form. This type of carcinoma should not be difficult to distinguish from conventional
hepatocellular carcinoma. The clinical history and the aspiration cytomorphology is quite suggestive.
Fortunately, many of the other morphologic variants of hepatocellular carcinoma are quite
uncommon. One of these is the small cell variant, which I believe represents a high grade hepatocellular
carcinoma. I have now personally examined a small subset of these neoplasms. In the vast majority,
intercellular cohesion is greatly reduced. Accordingly, most of the specimen consists of individually
dispersed neoplastic cells. The latter have solitary round nuclei and rather high nuclear-to-cytoplasmic
ratios. Thus, it presents as a small cell malignancy. Most of the neoplastic cells will not
demonstrate, by routine light microscopy, differentiation as hepatocellular carcinoma. In these cells,
the nuclei are often peripherally located in the scanty cytoplasm. I have now witnessed a couple of
examples in which cohesion is much better preserved. In any case, one prudent step is to examine the
entire specimen carefully for a small proportion of tumor cells which demonstrate more distinct
hepatocellular differentiation. This usually manifests as isolated malignant cells with greater volumes
of granular, dense cytoplasm and centrally positioned nuclei.
The differential diagnosis of small cell hepatocytic carcinoma may fall into two different
categories, based largely on patient age. In adults, one needs to consider primarily metastatic
neoplasms, including small cell carcinoma of bronchogenic origin, islet cell neoplasms, lymphoreticular
tumors, and other relatively infrequent neoplasms.
For small cell carcinoma of the lung, the clinical history is important. Morphologic
attributes which may assist include neoplastic cells which possess exceedingly high
nuclear-to-cytoplasmic ratios, nuclear molding, a lack of nucleoli, and a distinctly granular (salt and
pepper) chromatin pattern. In addition, immunocytochemistry may be exceedingly useful. In this setting,
the finding of a positive reaction for TTF-1 can be extremely useful.
The differential diagnosis also includes pancreatic endocrine neoplasms (islet cell
tumors). This may be more difficult, based purely on morphology, than small cell carcinoma of the lung.
This is because the neoplastic cells possess greater volumes of cytoplasm and often have an eccentric
nucleus which is perfectly round. A more distinctive plasmacytoid appearance of the tumor cells may
suggest a well differentiated neuroendocrine carcinoma. Endothelial cell associations, in addition,
would not be expected with these pancreatic primaries. Of course, immunocytochemistry pointing to
neuroendocrine differentiation would be most appreciated.
A neoplasm which closely resembles pancreatic endocrine neoplasms based purely on
cytomorphology is the primary granulosa cell tumor of the ovary. One distinctive feature is the finding
of well developed longitudinal nuclear grooves, which are typically more evident on the Papanicolaou
stained material. In addition, Call-Exner bodies may be observed in aspiration smears. Once again,
immunocytochemistry could prove most useful. One also needs to consider the possibility of nonHodgkin's
lymphomas and leukemias involving the liver. These may present as individually dispersed cells with high
nuclear-to-cytoplasmic ratios and small but distinct nucleoli, causing confusion with the small cell
variant of hepatoma. However, true intercellular cohesion will be lacking, nuclear contours may be more
irregular (better seen on the Papanicolaou stained specimens), and lymphoglandular bodies will be present
in the smear background. In our experience, most of these represent large cell lymphomas of the B-cell
type. A significant proportion of these are first discovered as a mass within the liver.
In the pediatric population, lymphoreticular malignancies also are important in hepatic
aspirates. The same cytomorphologic attributes should allow one to distinguish these neoplasms from
hepatoma. In addition, immunophenotyping by immunocytochemistry or flow cytometry can be conclusive.
One also needs to consider metastatic neuroblastoma. Several distinct morphologic
features may be helpful in distinguishing this from the small cell variant of hepatoma. In
neuroblastoma, the solitary malignant nuclei typically have distinctly granular, hyperchromatic
chromatin. Nucleoli may or may not be evident. These cells possess exceedingly high
nuclear-to-cytoplasmic ratios. Other helpful clues include identifying larger ganglion-like neoplastic
cells. The latter possess polygonal contours, moderate to abundant (and at times distinctly granular)
cytoplasm, and one or more peripherally located nuclei. The latter typically have a round contour,
vesicular chromatin, and prominent nucleoli. Finally, the smear background may contain neuropil. This
is better appreciated with the Diff-Quik stain, where it presents as brilliantly eosinophilic, fibrillary
material. Finally, one must consider the rare hepatoblastoma. On a cell for cell basis, this may be
difficult to distinguish from the small cell variant of hepatoma. Fortunately, this differential
diagnosis is exceedingly uncommon. In hepatoblastoma, the neoplastic cells, overall, are smaller than
with hepatoma. In addition, in general, cohesion is better preserved with hepatoblastoma.
Another morphologic variant of hepatoma is the pleomorphic type. This characteristically
consists of absolutely bizarre-appearing tumor giant cells with one or more nuclei intermingled with
(usually more numerous) smaller neoplastic hepatocytes. The malignant giant cells may or may not
manifest some degree of hepatocellular differentiation.
The differential diagnosis includes the pleomorphic or sarcomatoid forms of carcinoma, as
may arise in organs such as the lung, thyroid, pancreas, and kidney. With these metastases,
intercellular cohesion is almost completely lacking, there is no evidence of hepatocytic differentiation,
and neutrophils are a frequent and prominent component of the smears. The latter may include
cytophagocytosis of the segmented leukocytes by the malignant tumor cells. One would also have to
consider the possibility of a metastasis of a pleomorphic soft tissue sarcoma.  In the latter
situation, obviously, the clinical history is paramount.
One other morphologic, thankfully rare, variant of hepatocytic carcinoma is the clear cell
type. These typically resemble, for the most part, the conventional hepatocytic carcinoma except that
the cytoplasm is not dense and granular. Rather, it appears optically clear. Other cancers with
moderate to abundant clear cell cytoplasm obviously fall into this differential diagnosis. In this
setting, the architectural arrangement of the cells (including endothelial associations),
immunocytochemistry, and a lack of a primary neoplasm elsewhere are crucial.
In a significant proportion of cases, the distinction between metastatic adenocarcinoma
and hepatoma can be made based purely on convention light microscopy with direct smears. At times, the
architectural arrangement seen in cell blocks is also useful. Perhaps more importantly, cell blocks may
be used for immunocytochemistry. Immunocytochemistry is the most widespread ancillary diagnostic
procedure in this setting. A number of antibodies have been utilized in the distinction of metastatic
carcinoma and hepatoma. One of the most widely utilized is polyclonal carcinoembryonic antigen (CEA).
Polyclonal CEA characteristically produces a canalicular decoration with moderate levels of sensitivity
and high specificity. More recently, immunostaining with CD10 resulted in results that parallel those of
polyclonal CEA. Various cytokeratins may be useful in this distinction. Most hepatomas express low
molecular weight cytokeratin but are nonreactive with the high molecular weight forms. If a tumor is
positive for cytokeratin 20 and negative for cytokeratin 7, this is strong support for a colorectal or
appendiceal primary. A number of individuals have touted the use of alpha fetoprotein (AFP) due to its
high level of diagnosistic specificity. However, in my experience as well as that of many others, it
possesses excessively low diagnostic sensitivity. Markers of endothelium such as CD34 or CD31 may prove
useful in highlighting endothelial cells wrapping about neoplastic hepatocytes. Perhaps most useful is
the antibody HepPar-1. HepPar-1 is a monoclonal antibody apparently directed against an antigen specific
for the mitochondria in hepatocytes. HepPar-1 staining thus is quite specific and also possesses a high
level of diagnostic sensitivity.
The use of immunocytochemistry in conjunction with the clinical picture, laboratory
values, and the aspiration cytomorphology generally allows one to reliably and readily make the
distinction between primary hepatocellular carcinoma and metastatic carcinomas.
- Geisinger KR, et al. Liver. In Modern Cytopathology, Churchill Livingston, 2004, New York, pp. 505-542.
- Wee A, Sampatanukul P. Fine Needle Aspiration Cytology of the Liver. Diagnostic Algorithms. A Southeast Asian Perspective. Year Book Publishers, 2004, Bangkok, Thailand.
- Singh HK, et al. Fine-needle aspiration morphology of clear-cell hepatocellular carcinoma. Diagn Cytopathol 1997;17:206.
- Hughes JH, et al. The role of fine needle aspiration cytology in the evaluation of clear cell tumors. Cancer Cytopathol 1999;87:380-389.
- Chen X et al. Novel endothelial cell markers in hepatocellular carcinoma. Mod Pathol 2004;17:1198.
- Bottles K et al. A step-wise logistic regression analysis of hepatocellular carcinoma. An aspiration biopsy study. Cancer 1988;62:558.
- Salamào DR et al. Reproducibility of proposed cytologic criteria to discriminate hepatocellular carcinoma from metastatic adenocarcinoma in fine needle aspiration. Act Cytol 2001;45:857.
- Geisinger KR, Abdul-Karim FW. Fine Needle Aspiration Biopsies. In: Weiss SW, Goldblum JR. Enzinger and Weiss's Soft Tissue Tumors, 4th ed. Mosby, 2001, St. Louis, pp. 147-158.
- Minervini MI, et al. Utilization of hepatocyte-specific antibody in the immunocytochemical evaluation of liver tumors. Modern Pathol 1997;10:686.
- Lampe LW, Folpe AL. The diagnostic value of hepatocyte paraffin antibody 1 in differentiating hepatocellular neoplasms from nonhepatic tumors: a review. Adv Anat Pathol 2003;10:39.