J Neuropathol Exp Neurol
Copyright Ó 2006 by the American Association of Neuropathologists, Inc.
Vol. 65, No. 10
October 2006
pp. 935Y944
REVIEW ARTICLE
Immunohistochemical Analysis of Metastatic Neoplasms of the
Central Nervous System
Mark W. Becher, MD, Ty W. Abel, MD, PhD, Reid C. Thompson, MD,
Kyle D. Weaver, MD, and Larry E. Davis, MD
Abstract
Metastatic neoplasms to the central nervous system are often
encountered in the practice of surgical neuropathology. It is not
uncommon for patients with systemic malignancies to present to
medical attention because of symptoms from a brain metastasis and
for the tissue samples procured from these lesions to represent the
first tissue available to study a malignancy from an unknown
primary. In general surgical pathology, the evaluation of a
metastatic neoplasm of unknown primary is a very complicated
process, requiring knowledge of numerous different tumor types,
reagents, and staining patterns. The past few years, however, have
seen a remarkable refinement in the immunohistochemical tools at
our disposal that now empower neuropathologists to take an active
role in defining the relatively limited subset of neoplasms that
commonly metastasize to the central nervous system. This
information can direct imaging studies to find the primary tumor
in a patient with an unknown primary, clarify the likely primary site
of origin in patients who have small tumors in multiple sites
without an obvious primary lesion, or establish lesions as late
metastases of remote malignancies. Furthermore, specific treatments can begin and additional invasive procedures may be
prevented if the neuropathologic evaluation of metastatic neoplasms provides information beyond the traditional diagnosis of
Bmetastatic neoplasm.[ In this review, differential cytokeratins,
adjuvant markers, and organ-specific antibodies are described and
the immunohistochemical signatures of metastatic neoplasms that
are commonly seen by neuropathologists are discussed.
Key Words: Brain metastases, Intermediate filaments, Tumor
markers, Unknown primary.
HISTORICAL PERSPECTIVE AND
INTRODUCTION
Traditionally, many neuropathologists’ strategy with
these specimens has been to first determine that they are not
high-grade glial neoplasms and then render a simple
From the Departments of Pathology (MWB, TWA) and Neurosurgery
(RCT, KDW), Vanderbilt University School of Medicine, Nashville,
Tennessee; and the Department of Neurology (LED), the New Mexico
VA Health Care System, Albuquerque, New Mexico.
Send correspondence and reprint requests to: Mark W. Becher, MD,
Neuropathology Division, Department of Pathology, Vanderbilt University School of Medicine, Medical Center North C-3314, Nashville,
TN 37232-2561; E-mail: [email protected]
diagnosis of Bmetastatic neoplasm[ or Bmetastatic carcinoma.[ This strategy likely developed over time for several
reasons. For many years, the diagnosis of carcinoma was
based on hematoxylin and eosin-stained sections and routine
stains such as mucicarmine may have suggested adenocarcinoma, but no further definition was possible. With immunohistochemistry in the 1980s, we were able to demonstrate
cytokeratin immunoreactivity in metastatic epithelial neoplasms but were unable to define the possible site of origin
for these tumors. The introduction of markers such as CEA
and B72.3, which were heralded as revolutionary markers
for gastrointestinal tumors and breast malignancies, added to
this frustration because they were eventually found to not be
organ-specific. At the same time, there was an explosion in
the number of classification schema, descriptions of subtle
variants of systemic neoplasms, and their corresponding
treatments. These many factors have imparted an air of
confusion over which markers are relevant to apply to brain
metastases that we encounter in surgical neuropathology.
This review presents readily available new markers, defines
strategies for evaluating central nervous system (CNS)
metastases, and discusses the immunohistochemical signatures of systemic malignancies that commonly spread to
the brain.
CLINICOPATHOLOGIC OVERVIEW
More than one million three hundred thousand new
cases of invasive cancer develop in Americans each year (1).
Metastatic neoplasms to the CNS are common complications
of systemic cancer (2). Although the true incidence is
unknown, it is estimated that up to 170,000 new cases of
brain metastasis are diagnosed in the United States each year
(3, 4). CNS metastases have been reported to occur in 20%
to 45% of patients with systemic cancer (5). As treatment
methods continue to improve for systemic cancers, patients
are living longer and thus may be at increased risk for
metastatic disease.
Signs and symptoms of brain metastasis can appear as
the first presentation of a systemic malignancy, present
simultaneously with symptoms of the primary tumor, or
develop as a late manifestation many years after diagnosis
and treatment of the primary cancer (5). Patients with
unknown primary tumors represent 3% to 10% of all new
patients with cancer (6, 7). The presenting signs and
symptoms of cerebral metastases depend on the level of
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
935
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Becher et al
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
increased intracranial pressure and the location of the lesion.
Cerebral dysfunction may result from destruction or replacement of brain tissue, mass effect of the tumor and surrounding
edema, irritation of adjacent cortical neurons, and cerebral
herniation (3). Some patients progress slowly over several
weeks with a headache or decline in cognition, whereas others
present more acutely with a severe headache, seizures, or
focal neurologic signs such as weakness or gait instability (4).
Radiographically, multiple ring enhancing parenchymal
lesions with central necrosis and surrounding edema or
FLAIR signal have a high likelihood to be metastatic in
origin, although high-grade gliomas can be multifocal. Single
lesions with similar radiographic features are more problematic because they cannot effectively be distinguished from
high-grade glioma, lymphoma, abscess, and even a large
demyelinating plaque based solely on imaging.
Most brain metastases arise from hematogenous spread
and thus the distribution of CNS metastases reflects the relative CNS blood flow with 80% of metastases in the cerebrum,
15% involving the cerebellum, and 5% in the brainstem and
deep structures (5). The majority of metastases are found
in the parietal lobe, likely as a result of the confluence of
the 3 major cerebral arteries followed by frontal and occipital
lobes (5, 8). Most cortical metastases begin at the grayY
white matter junction and in areas of borderline vascular supply.
Neurosurgical decisions regarding management of the
intracranial lesions relate to the number, size, and location of
the lesions and the role of neurosurgery in the treatment of
brain metastases continues to evolve (9, 10). Patients with
multiple metastatic lesions are typically treated with whole
brain radiation therapy, not neurosurgical resection. Those
with single, surgically accessible lesions are often treated
with surgical resection before radiation therapy. Stereotactic
radiosurgery may be also considered with single metastases
presumed to be from tumors that are classically resistant to
whole brain radiation therapy or patients who have local
control of their primary tumor and only isolated distant
metastases (oligometastasis) (11). Occasionally, patients
with multiple lesions will have a single large mass that is
symptomatic or life-threatening and requires emergent
neurosurgical intervention. Surgical resection of these large
metastatic tumors before radiation therapy will alleviate
mass effect and may also mitigate postirradiation edema and
contribute to resolution of neurologic symptoms. Studies by
Patchell and colleagues in 1990 shed new light on the role
for surgical therapy for patients with brain metastases by
demonstrating that the addition of surgical resection to
whole brain radiation therapy improved median survival
(40 weeks vs 15 weeks; p G 0.01) as well as functional independence (12). An additional indication for surgical intervention in the management of patients with intracranial
lesions (including patients with and without a known primary)
is to establish a diagnosis. Up to 10% of patients with known
systemic malignancies and a CNS lesion that is radiographically suspicious for metastasis will be found to have an intracranial process unrelated to the primary lesion, most frequently
high-grade gliomas (12). An important guiding principle is
to establish the diagnosis in the least invasive way possible,
although low-yield noninvasive tests likely contribute
936
substantially to delaying the diagnosis in patients with
suspected cancer (13). If a large lung mass is amenable to
biopsy, then that procedure would be preferable to a neurosurgical procedure on a neurologically asymptomatic patient
with a presumed brain metastasis. However, for example,
some patients with intrathoracic disease have contraindications for pulmonary biopsies and neurosurgical biopsy may be
preferable. Recent improvements in neurosurgical techniques,
including image-guided stereotactic localization, have considerably lowered intracranial surgical morbidity.
Thus, the neuropathologist is often the first pathologist to
have a tissue sample from a patient’s systemic malignancy.
Making a tissue diagnosis that suggests the primary tumor is
important for several reasons. First, it guides the clinical
workup to locate the primary systemic tumor. Second, the
histologic diagnosis aids in ruling in metastatic tumors that are
potentially curable such as germ cell tumors or lymphoma as
well as tumors that have a good response to therapy such as
breast, prostate, ovarian, and small cell lung carcinomas. Third,
many systemic neoplasms are treated with specific therapeutic
agents, individualized treatment regimens, and national protocols. Finally, the long-term prognosis of a patient with brain
metastases depends on several factors that include the type of
tumor, the number and size of metastases, extent of the primary
tumor, presence of secondary metastases in the body, Karnofsky
Performance Status score, level of neurocognitive functioning,
and patient age (3). Knowledge of the tumor type therefore
helps the oncologist develop appropriate strategies to maximize
the quality of life and life expectancy.
The spectrum of clinical scenarios that neuropathologists encounter with these specimens varies over a wide
range. Sometimes, only a brief clinical history of a poorly
defined lung nodule or abnormality on preoperative chest
x-ray is provided that leads the clinical team to the assumption
that the brain lesion may be metastatic lung carcinoma.
Occasionally, we have a patient with a diagnosis of carcinoma from a limited cytology fine needle aspirate from some
site that is presumed to be a primary neoplasm. Because
diagnostic immunohistochemistry sometimes cannot be performed on the limited amount of material in fine needle
aspirates, unless either multiple slides are prepared or a cell
block is generated, this might in fact be a metastatic tumor and
not the primary. For example, a fine needle aspirate of a small
pulmonary nodule interpreted as nonsmall cell carcinoma is
presumed to be a primary lung neoplasm but could be a metastatic lesion from a different primary site such as the colon.
Sometimes, the history of a remote tumor is uncovered days
after a neurosurgical procedure that would suggest a late
metastatic lesion from a previously known tumor.
Cost-effectiveness analyses of many anatomic pathology
tests, including immunohistochemistry, have not been extensively studied (14). For many medical tests, Bcost per increase
in patient life expectancy[ is a common matrix used to determine cost-effectiveness. If one assumes that all patients
with metastatic cancer have a universally poor prognosis and
thus a similar life expectancy, then the cost per increase in
patient life expectancy for any medical procedure in this cohort of patients would likely not be highly cost-effective. This
assumption is flawed because not all patients with metastatic
Ó 2006 American Association of Neuropathologists, Inc.
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
cancer have a universally negative outcome, nor is life
expectancy the most important variable in many cases in
which customized palliative treatments based on tumor type
would have a profoundly positive impact on quality of life. In
addition, increases in patient life expectancy are also likely
not the most appropriate measure of an anatomic pathology
test that imparts information rather then a tangible procedure
(14). Relative to the cost of other medical tests and
procedures in a patient with an unknown primary malignancy,
immunohistochemical tests are inexpensive, provide benefits
other than cost-effectiveness, and may avert more costly
procedures that otherwise would have been performed (14).
NEOPLASMS THAT METASTASIZE TO THE
CENTRAL NERVOUS SYSTEM
Central nervous system metastasis is a highly selective
nonrandom process consisting of a series of linked sequential events that involve molecular and genetic changes in the
tumor cells that impart metastatic potential, including
angiogenic properties, adhesive capacity, and CNS affinity
(15). Although there are wide variations in reported
incidences as a result of differing methods of case selection,
primary tumors in adults most likely to metastasize to the
brain include lung (18Y60%), breast (5Y21%), melanoma
(4Y16%), genitourinary (3Y10%), gastrointestinal (5Y12%),
and unknown primary site (2Y18%) (3, 4, 8, 16). CNS
involvement by a systemic leukemia/lymphoma could also
be included in this discussion. Choriocarcinoma, although an
uncommon malignancy, frequently spreads to the CNS (8).
The distribution of the primary sites of origin of brain
metastases does not simply reflect the frequency of neoplasms in the population. Some common malignancies,
Metastatic Neoplasms of the CNS
including prostate and pancreatic carcinoma, do not often
spread to the CNS. Lung primaries are one of the most
common neoplasms to initially present with widely metastatic tumors, including brain metastases, before recognition
of the primary tumor (17, 18). Other neoplasms tend to
develop CNS metastatic disease late in their course after the
primary and perhaps other metastatic lesions are known.
This is often the case with renal cell carcinoma. Another
major category of neoplasm that usually do not present with
metastatic tumors of unknown primaries is the sarcoma,
which usually develops bulky disease at the primary site
with local symptoms.
There is an extensive general surgical pathology
literature on the evaluation of metastatic neoplasms from
unknown primaries (17, 19Y21). The surgical pathology
workup of metastatic tumors of unknown primary requires
knowledge of a vast number of antibodies, diagnostic
staining patterns, crossreactivities of antibodies, different
tumor types, and complex algorithms. Attempting to apply
many of these algorithms to the neuropathologic evaluation
of brain metastases is difficult because not all neoplasms
metastasize to the CNS. In addition, surgical pathologists
often have to resolve additional issues other than the breadth
of possible tumor types. For example, one might need to
differentiate mesothelioma from other similar-appearing
intrathoracic neoplasms or define whether a pulmonary
lesion is primary or secondary. This level of complexity is
not relevant to neuropathologists. Thus, in practical terms,
neuropathologists do not need to be fully versed in the
immunohistochemical signature of every possible type of
metastatic neoplasm (22Y24). Rather, the immunohistochemical algorithm for common neoplasms that metastasize to the
CNS is less daunting and fairly straightforward (Table 1;
TABLE 1. Practical Algorithm for the Immunohistochemical Evaluation of Common Central Nervous System (CNS)
Metastatic Neoplasms
CNS metastatic neoplasm core first-round antibodies for epithelial tumors or Adenocarcinoma Cam 5.2, CK7, CK20, and TTF-1
If Cam5.2-positive
And TTF-1-positive
Positive CK7 and TTF-1 with negative CK20 suggests Nonsmall Cell Lung Carcinoma
Positive TTF-1 with negative CK7 and CK20
Add CD56 to rule in Small Cell Lung Carcinoma
And TTF-1-negative
Negative CK7 and positive CK20: Add CDX2 to rule in Colorectal Carcinoma
Negative CK7 and CK20
Add CD10, Vimentin, and RCCMa to rule in Renal Cell Carcinoma
Add CK5/CK6 to rule in Squamous Cell Carcinoma
Positive CK7 and negative CK20
Add GCFDP-15, ER, CA125 to differentiate Breast Carcinoma versus Endometrial Carcinoma
If Cam5.2-negative
And TTF-1-positive with negative CK7 and CK20
Add CD56 to rule in Small Cell Lung Carcinoma
And TTF-1-negative and negative CK7 and CK20
Add GFAP, melan-A, CD10, Vimentin, Hematopoietic markers (CD3 & CD20) to differentiate Epithelioid GBM, Melanoma, Renal Cell
Carcinoma, and Lymphoma
CNS metastatic neoplasm core first round antibodies for poorly differentiated neoplasms:
GFAP, synaptophysin, CAM5.2, CK7, CK20, melan-A, CD20 and CD3
Ó 2006 American Association of Neuropathologists, Inc.
937
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Becher et al
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
Fig. 1). If the initial workup as proposed below provides a
confusing or inconsistent immunohistochemical pattern, then
it is possible that the metastatic neoplasm is less well
differentiated than one with a classic immunohistochemical
signature (Table 2) or from a uncommon site of origin (8) and
would require further study and likely consultation with a
surgical pathologist. In addition, web-based resources are
available such as BImmunoQuery[ (www.ipox.org) that
allows one to search by antibody or neoplasm and gives upto-date statistics to help evaluate uncommon lesions or
atypical immunohistochemical reactivity patterns.
DIFFERENTIAL CYTOKERATINS
Intermediate filaments are found in all human cells and
can be divided into 6 categories: vimentin, desmin, lamins,
neurofilament, glial filament acidic protein (GFAP), and
cytokeratins. Vimentin is considered the Bprimordial[
intermediate filament (25), is found in nearly all fetal cells
in development, labels normal endothelial structures in many
cell types and tumors, and is classically used as a barometer
of tissue antigenicity relative to fixation and viability in
surgical pathology. Desmin is a muscle marker; lamins are
nuclear envelope proteins. Vimentin, desmin, and the lamins
have limited use in the context of CNS metastases. Neurofilament and GFAP are well known to the neuropathology
community. GFAP labels normal, reactive, and neoplastic
astrocytes, normal ependymal cells and neoplastic ependymal cell processes, and retinal Muller glial cells. Antibodies
to GFAP label Schwann cells in the peripheral nervous
system, Kupffer cells in the liver, chondrocytes, and myoepithelial cells (25). Some peripheral nerve sheath tumors
(25), chondroid tumors (25), hepatocellular carcinomas (26),
and tumors with myoepithelial differentiation may be GFAPimmunopositive (27, 28).
Cytokeratin antibodies have been used for many years
to label the intermediate filaments that define epithelial cell
populations. We now have numerous cytokeratin antibodies
that recognize cytokeratins of different molecular weights
and Bcocktails,[ or mixtures, of these antibodies. With these
reagents, we began to have the ability to define epithelium
from various organs as a result of their differential
FIGURE 1. Immunohistochemical relationship diagram for common central nervous system metastatic neoplasms.
938
Ó 2006 American Association of Neuropathologists, Inc.
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
Metastatic Neoplasms of the CNS
TABLE 2. Immunohistochemical Signatures of Common Central Nervous System Metastatic Neoplasms
GCDFP-15
CA125
CDX2
(nuclear)
Vimentin
CD10,
RCCMa
(nuclear)
+
+
j
+
j
j
j
j
j
j
j
j
+/j
j
j
+
j
+
j
j
j
j
j
j
+
j
j
j
+
j
j
j
j
j
j
j
j
+
j
+
j
j
j
j
j
j
j
j
+
j
j
+
j
j
j
j
+
j
j
j
+
+
j
j
j
j
j
j
+
j
j
j
+/j
j
j
j
j
j
j
j
j
j
+
+
+
j
+
j
j
j
j
j
j
+
j
j
+
+/j
+/j
j
j
j
j
j
j
+/j
j
j
CAM 5.2
Lung nonsmall
cell carcinoma
Lung small cell
carcinoma
Squamous cell
carcinoma
Melanoma
Breast
carcinoma
Endometrial
carcinoma
Renal cell
carcinoma
Colorectal
carcinoma
Gastric/
gastroesophageal
carcinoma
CK7 CK20
TTF-1
(nuclear)
Melan-A,
HMB-45,
CK5/CK6 CD56
S100
+, positive; j, negative; +/j, can be positive or negative.
expression of cytokeratins and the nomenclature of BCK#[
was developed to sort the cytokeratins by molecular weight
(29). Over the past several years, cytokeratin immunohistochemistry has been one of the most positive advances of
surgical pathology yet is understandable and useful for the
evaluation of CNS metastatic tumors (22, 30).
High-molecular-weight cytokeratins, typically identified with an antibody cocktail of CK5 and CK6 (BCK5/
CK6[), are found in squamous epithelia and almost all
squamous cell carcinomas (17). Low-molecular-weight cytokeratins are found in nearly all epithelia except squamous
epithelium. CAM5.2 is a low-molecular-weight cytokeratin
that recognizes simple nonstratified, ductal, and pseudostratified epithelium and does not crossreact with astrocyte
intermediate filaments (17, 30). In the brain, CAM5.2 labels
choroid plexus epithelium and some pituitary adenomas
(31). Cytokeratin AE1 or the cocktail of AE1/AE3, which is
commonly used in surgical pathology, recognizes lowmolecular-weight cytokeratins, but also labels normal astrocytes, reactive astrocytes, and neoplastic astrocytes (30).
Thus, in the brain, CAM5.2 is more useful to establish
epithelial origin in the differentiation of poorly differentiated
metastases from high-grade gliomas than is AE1/AE3.
CK7 and CK20 are the most useful of the newer
differential cytokeratins because they have different and
complementary reactivity in epithelia from many different
organs. CK7 is not found in normal colon, liver, prostate, or
squamous epithelium and is present in many simple, pseudostratified, and ductal epithelia. Specifically, CK7 labels lung
adenocarcinomas, lung nonsmall cell carcinomas, neuroendocrine neoplasms (although not small cell carcinoma of lung),
and carcinomas of breast, ovary, pancreas, and endometrium.
CK7 will also variably stain squamous cell carcinomas of the
lung. CK20 is found in normal and neoplastic colorectal
Ó 2006 American Association of Neuropathologists, Inc.
epithelium and a low percentage of breast carcinomas.
Presently, the immunoreactivity of CK7 and CK20 are often
referred to in the same sentence so that a nomenclature has
evolved such that a neoplasm is said to be BCK7-positive (or
negative)/CK20-negative (or positive).[ This is very helpful in
distinguishing 2 neoplasms that neuropathologists see as brain
masses, lung, and colon carcinoma. Thus, lung nonsmall cell
carcinomas are BCK7+/CK20j[ and colon carcinomas are
BCK7j/CK20+[ (Fig. 2).
ADJUVANT EPITHELIAL MARKERS
In addition to the differential cytokeratins, adjuvant
epithelial markers that are not organ-specific are sometimes
helpful in evaluating metastatic neoplasms of unknown
primary, especially those from epithelia that lack CK7 and
CK20 or are less well differentiated. BCarcinoembryonic
antigen[ (CEA) is a well-known glycoprotein cell membrane
protein family of endodermal differentiation. Polyclonal and
monoclonal antibodies are presently available. CEA labels
adenocarcinomas of the lung, colon, stomach, pancreas, and
bladder as well as endocervical and breast carcinomas. CEA
does not label mesothelioma, serous tumors of the ovary or
carcinomas of the prostate, kidney, adrenal, or endometrium.
BEpithelial membrane antigen[ (EMA) is a reagent well
known to neuropathologists as a result of its cell membrane
pattern of immunoreactivity of many meningiomas, focal
staining of some choroid plexus neoplasms, and the intracytoplasmic microlumina and luminal staining of ependymal
rosettes in ependymomas (32). EMA is not limited to visceral epithelial expression, however, and labels some soft
tissue neoplasms. Currently, both CEA and EMA have a role
as supplemental epithelial antigens in the workup of a metastatic tumor of unknown primary but are not likely included
939
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Becher et al
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
FIGURE 2. Typical immunohistochemical findings in metastatic neoplasms commonly encountered in surgical neuropathology.
(A) Lung adenocarcinoma (nonsmall cell carcinoma) is CK7+/CK20j and nuclei are TTF-1+/CDX2j. (B) Colon adenocarcinoma
is CK7j/CK20+ with CDX+ nuclei. (C) Breast carcinoma is CK7+/ER+/GCDFP-15+. (D) Melanoma is S100 protein+/HMB-45+/
melan-A+; reactive astrocytes express immunoreactivity with S100 protein (arrow). (E) Renal cell carcinoma is strongly CD10+/
RCCMa+. (F) Small cell carcinoma of the lung is CAM5.2+/CD56+/TTF-1+. (G) Endometrial carcinoma is CK7+/CA125+.
(H) Squamous cell carcinoma is CK5/CK6+/TTF-1j.
in the first round of immunohistochemistry for CNS metastases. Their use may eventually decline as organ-specific
940
markers become more widely used and additional markers
are discovered.
Ó 2006 American Association of Neuropathologists, Inc.
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
ORGAN-SPECIFIC MARKERS
Antigens that are organ-specific or limited to a
pathologically relevant biochemical family have greatly
enhanced our ability to evaluate metastatic neoplasms. In
addition to organ-specific markers, there are several antibodies that are not highly specific but are very helpful when
used in conjunction with other markers. The 2 most wellknown organ-specific markers are thyroglobulin for thyroid
and prostate-specific antigen (PSA) for the prostate.
Unfortunately, for the evaluation of metastatic neoplasms
to the CNS, these 2 markers have very little use because
these neoplasms rarely metastasize to the brain. Prostate
carcinoma is well known to metastasize to the dura,
however, and many of these tumors, depending on their
degree of differentiation, will label with antibodies to PSA.
PSA is also known to label some primary and metastatic
melanomas (17).
Relatively recently, 2 additional well-characterized
organ-specific markers, TTF-1 and CDX2, have come into
widespread use. These are transcription factors that are selectively expressed in embryogenesis and mature tissues that were
discovered through molecular biologic studies and applied to
tumor pathology. Thyroid transcription factor (TTF-1), also
known as Nkx2.1 or thyroid-specific enhancer-binding protein
(T/EBP), is a member of the Nkx homeodomain-containing
transcription factor family and is expressed primarily in normal
thyroid and lung cells (33). Relevant to the discussion of CNS
metastases, TTF-1 labels the nuclei of the majority of
carcinomas of lung origin (Fig. 2), including nonsmall cell
carcinoma, adenocarcinoma, small cell carcinoma, and neuroendocrine carcinoma (17, 34, 35). Lung squamous cell carcinomas are typically TTF-1j. TTF-1 is also expressed in the
diencephalon of the rat brain in embryogenesis, specifically in
the developing neurohypophysis (36), ependymal/subependymal
cells of the third ventricle (37), hypothalamus, and subfornical
organ (38). Knockout mice without the TTF-1/Nkx2.1 gene
have embryonic lethality but lack thyroid and pituitary glands
and have underdeveloped ventromedial and dorsomedial
nuclei of the hypothalamus, fused third ventricular walls,
and absent arcuate nuclei (39). TTF-1 does not label normal
brain (40), normal pituitary gland (41), pituitary adenomas
(34, 41), glioblastomas (42), or other primary brain tumors
(astrocytomas, oligodendrogliomas, medulloblastomas, ependymomas, and gangliogliomas) (40, 43) but has been reported
to label the nuclei of 2 ependymomas from the third ventricle
(43), perhaps through site-specific expression. Metastatic neoplasms, especially adenocarcinomas, in the brain that lack the
architecture of a thyroid neoplasm and are CK7+/CK20j/
TTF-1+ are essentially diagnostic of nonsmall cell carcinoma
of lung origin (Fig. 2).
CDX2 is a caudal-type homeobox gene that encodes
an intestine-specific homeodomain transcription factor that
is expressed in intestinal epithelium and has been found to
label primary and metastatic colorectal carcinomas (44, 45).
Although the mRNA is specific to intestinal epithelium,
immunohistochemical studies have shown CDX2 immunoreactivity in gastric and gastroesophageal junction carcinomas that may metastasize to the brain, and mucinous
Ó 2006 American Association of Neuropathologists, Inc.
Metastatic Neoplasms of the CNS
ovarian carcinoma, urinary bladder adenocarcinoma, pancreatic adenocarcinoma, and intestinal carcinoid that generally do not metastasize to the brain (44, 46). Gastric
adenocarcinomas and gastroesophageal junction adenocarcinomas, although immunoreactive with CDX2, usually label
with both CK7 and CK20, in contrast to colorectal carcinomas, which typically do not label with CK7 (19). Thus,
metastatic adenocarcinomas in the brain that are CK7j/
CK20+/CDX2+ have a high likelihood of being colorectal in
origin (Fig. 2).
Renal epithelial neoplasms are a complex set of
entities with varying cytologic and cytokeratin immunoreactivity profiles (47). Subclassification of these neoplasms
is not usually necessary in the evaluation of metastatic
neoplasms of the CNS. In general, renal cell carcinoma is
CK7j/CK20j and CAM5.2 or pancytokeratin, EMA, and
vimentin-immunoreactive (Fig. 2) (48). Many, but not all,
renal cell carcinomas are also immunoreactive with the
adjuvant markers CD10 (49, 50) and a nuclear marker called
Brenal cell carcinoma marker[ (RCCMa) (49, 50).
Metastatic breast carcinoma can sometimes be distinguished from other metastatic adenocarcinomas by the
cellular and architecture characteristics of typically bland,
nonmucinous breast carcinoma in a patient with known breast
carcinoma. The evaluation of metastatic breast carcinoma is
more difficult when no clinical history is known or the CNS
lesion is a late manifestation in a patient with a remote history
of breast carcinoma. Breast carcinomas are CK7+/CK20j
and TTF-1j/CDX2j (Fig. 2) and thus can generally be
distinguished from lung and gastrointestinal tumors (19, 51).
Adjuvant markers can be used to help confirm breast origin,
including gross cystic disease fluid protein 15 (GCDFP-15),
which is believed to have 99% specificity and up to 50%
sensitivity (17, 51). Estrogen receptor (ER) may have
nuclear immunoreactivity along with a positive GCDFP-15
in metastatic breast carcinoma, although ER is not a sensitive or specific test to detect breast carcinoma alone and can
be found in many lung carcinomas (17, 51).
Female genital tract neoplasms are not common sites of
origin for metastatic neoplasms to the CNS, but endometrial
carcinomas are occasionally encountered. Endometrial carcinomas are CK7+/CK20j/TTF-1j/CDX2j (Fig. 2) and
many label with the adjuvant markers ER and CA125 (19).
MELANOCYTIC MARKERS
Malignant melanoma frequently spreads to the CNS
along a continuum of the disease from metastatic neoplasms
that are the first presentation of the illness to widespread
metastatic neoplasms in a patient with a known primary and
clinically important neurologic symptoms to late or even
very remote manifestations in patients that no longer carry
the remote diagnostic data in their medical record. It is not
unusual for small primary melanomas to go undetected on
their skin by patients and the diagnosis be made on
presentation for a brain metastasis. Metastatic melanoma
stains readily with antibodies to vimentin and S100 protein
(so named as a result of its solubility in saturated or 100%
941
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Becher et al
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
ammonium sulfate solution), although in the evaluation of
CNS metastases, S100 protein has limited use because of its
widespread immunoreactivity with neurons, reactive astrocytes (Fig. 2), gliomas, neuropil, and Schwann cells (52).
Fortunately, we have 2 additional markers, HMB-45 and
melan-A, members of the gp100/pmel 17 glycoprotein group
that are primarily restricted to cells of melanocytic lineage
(53). The original HMB-45 reagent was reportedly highly
specific for melanocytic differentiation, although currently
the commercially available HMB-45 hybridoma products
have an overall sensitivity of approximately 60% and label a
variety of other tumors (53). Melan-A, also called MART-1
for Bmelanoma antigen recognized by T cells-1,[ mRNA is
believed to only be translated in melanocytic cells (54) and
strongly labels most melanomas (Fig. 2). The desmoplastic
subtype of melanomas that rarely metastasize to the brain
may label with S100 protein but do not label well with
HMB-45 or melan-A. Gliomas lack both HMB-45 and
melan-A immunoreactivity (55), although a small number
of melanomas express GFAP (56). Microphthalmia transcription factor protein (MTFP) is a newly discovered organspecific transcription factor expressed in melanocytes during
embryogenesis that labels normal and neoplastic melanocyte
nuclei that will likely become more widely used in the future
(57). In practice, many surgical pathologists use combinations of S100 protein and HMB-45 or melan-A to optimize
their individual sensitivities. For CNS metastatic melanomas, it is likely sufficient to use only melan-A to define
a metastatic melanoma.
A PRACTICAL STRATEGY FOR EVALUATING
CENTRAL NERVOUS SYSTEM
METASTATIC TUMORS
In practical terms, the cellular and architectural features
of the initial hematoxylin and eosin-stained sections will
determine the starting point of the first round of immunohistochemical studies. If there is an issue of epithelioid glioblastoma versus metastatic adenocarcinoma, the first round
would include GFAP. If the initial hematoxylin and eosinstained sections suggest a poorly differentiated neoplasm
without architectural definition to suggest phenotype, the
strategy is described subsequently. If the lesion is clearly
epithelial in nature with discreet cell borders, acinar configurations, or mucin/goblet cells, the core first round of antibodies for common CNS metastases might include CAM5.2,
CK7, CK20, and TTF-1 (Table 1). CAM5.2 label will limit
the possibilities to many of the common tumors that metastasize to the CNS, including nonsmall cell lung carcinoma,
small cell lung carcinoma, colorectal carcinoma, and breast
carcinoma. TTF-1 immunoreactivity will limit it to a
pulmonary process assuming thyroid has been ruled out based
on clinical history and hematoxylin and eosin architectural
features. CK7 immunopositivity then suggests nonsmall cell
lung carcinoma, primarily adenocarcinoma. TTF-1+/CK7 and
CK20j tumors are likely small cell lung carcinoma with the
appropriate architectural features and can be defined with the
addition of CD56. CAM5.2+/TTF-1j tumors can be sorted
out based on the differential cytokeratins and organ-specific
942
markers in the second round of studies. CK7j/CK20+
adenocarcinomas that are TTF-1j have a high likelihood of
being colorectal in origin and this can be confirmed by adding
CDX2. Gastric carcinomas or gastroesophageal carcinomas
also stain with CDX2 but are more likely to phenotype as
CK7+ and CK20+ and may contain distinctive signet ring
cell configurations. Many pulmonary squamous cell carcinomas are TTF-1j and are likely to be CK7j and CK20j.
Most squamous cell carcinomas, including lung, will label
with the cocktail CK5/CK6 (Fig. 2). Breast and endometrial
carcinoma are CK7+/TTF-1j tumors that can be differentiated with GCFDP-15, ER, and CA125. Adenocarcinomas
or moderate to poorly differentiated neoplasms that type out
as CAM5.2j/TTF-1j/CK7j/CK20j may be renal cell
carcinoma, melanoma, or lymphoma and can often be sorted
out with the second round of studies. Adjuvant epithelial
markers, EMA and CEA, can be used with CAM5.2-negative
tumors to refocus the evaluation on less well-differentiated
epithelial tumors. Renal cell carcinoma can have variable
pancytokeratin immunoreactivity, so should be pursued if the
differential cytokeratins CK7 and CK20 are both negative
regardless of the CAM5.2 immunoreactivity with the addition
of CD10, vimentin, and RCCMa. The vast majority of
melanomas will label with melan-A. Lymphomas will label
with a range of mature and immature hematopoietic markers as
described subsequently.
POORLY DIFFERENTIATED NEOPLASMS
Poorly differentiated neoplasms, both primary and
metastatic, are encountered in surgical neuropathology. The
strategy in general surgical pathology is to first attempt to place
the lesion into one of the major categories of neoplasms:
carcinoma, germ cell tumor, sarcoma, lymphoma, or melanoma (17). The cellular features and architecture on hematoxylin and eosin-stained sections may define these lesions
somewhat, and by definition, rule out many recognizable
neoplasms. For example, a poorly differentiated germ cell
tumor would not be a seminoma but rather would be an
embryonal carcinoma so that the immunohistochemical
strategies should be tailored to the possible entities. The age
of patient further defines the spectrum of possibilities. In
neuropathology, this list would expand to include poorly
differentiated primary brain tumors (primitive neuroectodermal tumors [PNETs], medulloblastomas, and other neuroblastic tumors) and would at the same time shrink down the
list of possibilities that a general surgical pathologist is
attempting to phenotype because soft tissue tumors and
muscle differentiation tumors are not likely to metastasize to
the brain. Although a general surgical pathologist faced with a
poorly differentiated neoplasm somewhere in the body might
do a pancytokeratin such as AE1/3 for carcinoma, S100 for
melanoma and soft tissue tumors, placental-like alkaline
phosphatase (PLAP) for germ cell tumors, CD45 as a general
hematopoietic marker, myo-D1 and desmin for muscle
differentiation, and CD99 for small round blue cell tumors
(Ewing_s/PNET), the neuropathologist’s strategy is likely to
be different. Neuropathologists would include markers for
PNET/neuroblastic tumors (synaptophysin and GFAP). Yet,
Ó 2006 American Association of Neuropathologists, Inc.
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
as described previously, one pancytokeratin in the first round
of immunohistochemistry followed by differential cytokeratins the next day if the epithelial marker is positive may not
be as effective a strategy in the evaluation of poorly differentiated tumors in the brain as using CAM5.2 with CK7 and
CK20 in the first round, which have a high likelihood of
defining common epithelial metastases to the brain without
the astrocyte crossreactivity of AE1/3. Given the reactivity of
S100 protein with neuroectodermal elements that constitute
the CNS, we might substitute melan-A for S100 protein for
melanocytic tumors. PLAP will label poorly differentiated
embryonal germ cell tumors that are rarely encountered in the
brain, but germ cell tumor and muscle differentiation markers
may not be in the first round of immunohistochemistry for a
brain mass. CD45, in our experience, fails to label some of
the lymphomas that are encountered in the brain because,
although it is a general hematopoietic marker, it is a fairly
mature marker and we tend to do a B cell marker (CD20) and
T cell marker (CD3) in the first round. In summary, we do
GFAP, synaptophysin, CAM5.2, CK7, CK20, melan-A,
CD20, and CD3 as our first round of studies for poorly
differentiated neoplasms in the brain (Table 1). After ruling
out a primary brain tumor, if an epithelial neoplasm is indicated with these stains, we would then do organ-specific
markers (TTF-1 and/or CDX2 and/or renal cell markers) as
indicated by the differential cytokeratins. If no specific
phenotype is suggested by the first round of studies, then
adjuvant markers (EMA and CEA), less mature hematopoietic markers (CD79a for immature B cells), and markers
for less common malignancies would be used.
ACKNOWLEDGMENTS
The authors thank Drs. Joyce E. Johnson and Marcia
L. Wills for their critical evaluation of the manuscript and
Ms. Faith A. Allen for her excellent assistance.
REFERENCES
1. Jemal A, Tiwari RC, Murray T, et al. Cancer statistics 2004. CA Cancer
J Clin 2004;54:8Y29
2. Vecht CJ. Clinical management of brain metastasis. J Neurol 1998;245:
127Y31
3. Posner JB. Neurologic Complications of Cancer. Philadelphia: FA
Davis, 1995
4. Lassman AB, DeAngelis LM. Brain metastases. Neurol Clin 2003;21:
1Y23
5. Polyzoidis KS, Miliaras G, Pavlidis N. Brain metastasis of unknown
primary: A diagnostic and therapeutic dilemma. Cancer Treat Rev
2005;31:247Y55
6. Van de Wouw AJ, Janssen-Heijnen ML, Coebergh JW, et al. Epidemiology of unknown primary tumours; Incidence and population-based
survival of 1285 patients in Southeast Netherlands, 1984Y1992. Eur J
Cancer 2002;38:409Y13
7. Pavlidis N, Briasoulis E, Hainsworth J, et al. Diagnostic and therapeutic
management of cancer of an unknown primary. Eur J Cancer 2003;39:
1990Y2005
8. Burger PC, Scheithauer BW, Vogel FS. Surgical Pathology of the
Nervous System and Its Coverings. Philadelphia: Churchill Livingstone,
2002
9. Sills AK. Current treatment approaches to surgery for brain metastases.
Neurosurgery 2005;57:S4:24Y32
Ó 2006 American Association of Neuropathologists, Inc.
Metastatic Neoplasms of the CNS
10. Modha A, Shepard SR, Gutin PH. Surgery of brain metastasesYis there
still a place for it? J Neurooncol 2005;75:21Y29
11. Sheehan J, Niranjan A, Flickinger JC, et al. The expanding role of
neurosurgeons in the management of brain metastases. Surg Neurol
2004;62:32Y40
12. Patchell RA, Tibbs PA, Walsh JW, et al. A randomized trial of surgery
in the treatment of single metastases to the brain. N Engl J Med 1990;
322:494Y500
13. Farag SS, Green MD, Morstyn G, et al. Delay by internists in obtaining
diagnostic biopsies in patients with suspected cancer. Ann Intern Med
1992;116:473Y78
14. Raab SS. Cost-effectiveness of immunohistochemistry. In: Dabbs DJ,
ed. Diagnostic Immunohistochemistry. Philadelphia: Churchill Livingstone, 2002:45Y56
15. Nathoo N, Chahlavi A, Barnett GH, et al. Pathobiology of brain
metastases. J Clin Pathol 2005;58:237Y42
16. Barnholtz-Sloan JS, Sloan AE, Davis FG, et al. Incidence proportions of
brain metastases in patients diagnosed (1973 to 2001) in the
Metropolitan Detroit Cancer Surveillance System. J Clin Oncol 2004;
22:2865Y72
17. Dabbs DJ. Immunohistology of metastatic carcinoma of unknown
primary. In: Dabbs DJ, ed. Diagnostic Immunohistochemistry. Philadelphia: Churchill Livingstone, 2006:180Y226
18. Quan AL, Videtic GM, Suh JH. Brain metastases in small cell lung
cancer. Oncology (Williston Park) 2004;18:961Y72
19. Dennis JL, Hvidsten TR, Wit EC, et al. Markers of adenocarcinoma
characteristic of the site of origin: Development of a diagnostic
algorithm. Clin Cancer Res 2005;11:3766Y72
20. Brown RW, Campagna LB, Dunn JK, et al. Immunohistochemical
identification of tumor markers in metastatic adenocarcinoma. A
diagnostic adjunct in the determination of primary site. Am J Clin
Pathol 1997;107:12Y19
21. DeYoung BR, Wick MR. Immunohistologic evaluation of metastatic
carcinomas of unknown origin. In: Santa Cruz DJ, ed. Seminars in
Diagnostic Pathology: Immunohistochemistry in Tumor Diagnosis.
Philadelphia: WB Saunders, 2000:184Y93
22. Perry A, Parisi JE, Kurtin PJ. Metastatic adenocarcinoma to the brain:
An immunohistochemical approach. Hum Pathol 1997;28:938Y43
23. Drlicek M, Bodenteich A, Urbanits S, et al. Immunohistochemical panel
of antibodies in the diagnosis of brain metastases of the unknown
primary. Pathol Res Pract 2004;200:727Y34
24. Giordana MT, Cordera S, Boghi A. Cerebral metastases as first symptom
of cancer: A clinico-pathologic study. J Neurooncol 2000;50:265Y73
25. Cerilli LA, Wick MR. Immunohistology of soft tissue and osseous
neoplasms. In: Dabbs DJ, ed. Diagnostic Immunohistochemistry.
Philadelphia: Churchill Livingstone, 2006:65Y120
26. Okamura N, Yoshida M, Shibuya A, et al. Cellular and stromal
characteristics in the scirrhous hepatocellular carcinoma: Comparison
with hepatocellular carcinomas and intrahepatic cholangiocarcinomas.
Pathol Int 2005;55:724Y31
27. Chen PC, Chen CK, Nicastri AD, et al. Myoepithelial carcinoma of the
breast with distant metastasis and accompanied by adenomyoepitheliomas. Histopathology 1994;24:543Y48
28. Alos L, Cardesa A, Bombi JA, et al. Myoepithelial tumors of salivary
glands: A clinicopathologic, immunohistochemical, ultrastructural, and
flow-cytometric study. Semin Diagn Pathol 1996;13:138Y47
29. Moll R, Franke WW, Schiller DL, et al. The catalog of human
cytokeratins: Patterns of expression in normal epithelia, tumors and
cultured cells. Cell 1982;31:11Y24
30. Oh D, Prayson RA. Evaluation of epithelial and keratin markers in
glioblastoma multiforme: An immunohistochemical study. Arch Pathol
Lab Med 1999;123:917Y20
31. Coons SW, Estrada SI, Gamez R, et al. Cytokeratin CK 7 and CK 20
expression in pituitary adenomas. Endocr Pathol 2005;16:201Y10
32. Kawano N, Ohba Y, Nagashima K. Eosinophilic inclusions in
ependymoma represent microlumina: A light and electron microscopic
study. Acta Neuropathol (Berl) 2000;99:214Y18
33. Yatabe Y, Mitsudomi T, Takahashi T. TTF-1 expression in pulmonary
adenocarcinomas. Am J Surg Pathol 2002;26:767Y73
34. Oliveira AM, Tazelaar HD, Myers JL, et al. Thyroid transcription factor-1
distinguishes metastatic pulmonary from well-differentiated neuroendocrine tumors of other sites. Am J Surg Pathol 2001;25:815Y19
943
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Becher et al
J Neuropathol Exp Neurol Volume 65, Number 10, October 2006
35. Srodon M, Westra WH. Immunohistochemical staining for thyroid
transcription factor-1: A helpful aid in discerning primary site of tumor
origin in patients with brain metastases. Hum Pathol 2002;33:642Y45
36. Lazzaro D, Price M, de Felice M, et al. The transcription factor TTF-1
is expressed at the onset of thyroid and lung morphogenesis and in
restricted regions of the foetal brain. Development 1991;113:1093Y104
37. Lee BJ, Cho GJ, Norgren RB Jr, et al. TTF-1, a homeodomain gene
required for diencephalic morphogenesis, is postnatally expressed in the
neuroendocrine brain in a developmentally regulated and cell-specific
fashion. Mol Cell Neurosci 2001;17:107Y26
38. Son YJ, Hur MK, Ryu BJ, et al. TTF-1, a homeodomain-containing
transcription factor, participates in the control of body fluid homeostasis
by regulating angiotensinogen gene transcription in the rat subfornical
organ. J Biol Chem 2003;278:27043Y52
39. Kimura S, Hara Y, Pineau T, et al. The T/ebp null mouse:
Thyroid-specific enhancer-binding protein is essential for the organogenesis of the thyroid, lung, ventral forebrain, and pituitary. Genes Dev
1996;10:60Y69
40. Bohinski RJ, Bejarano PA, Balko G, et al. Determination of lung as
the primary site of cerebral metastatic adenocarcinomas using monoclonal antibody to thyroid transcription factor-1. J Neurooncol 1998;40:
227Y31
41. Mantovani G, Corbetta S, Romoli R, et al. Absence of thyroid
transcription factor-1 expression in human parathyroid and pituitary
glands. Mol Cell Endocrinol 2001;182:13Y17
42. Prok AL, Prayson RA. Thyroid transcription factor-1 staining is useful
in identifying brain metastases of pulmonary origin. Ann Diagn Pathol
2006;10:67Y71
43. Zamecnik J, Chanova M, Kodet R. Expression of thyroid transcription
factor 1 in primary brain tumours. J Clin Pathol 2004;57:1111Y13
44. Werling RW, Yaziji H, Bacchi CE, et al. CDX2, a highly sensitive
and specific marker of adenocarcinomas of intestinal origin: An
immunohistochemical survey of 476 primary and metastatic carcinomas.
Am J Surg Pathol 2003;27:303Y10
45. Witek ME, Nielsen K, Walters R, et al. The putative tumor suppressor
Cdx2 is overexpressed by human colorectal adenocarcinomas. Clin
Cancer Res 2005;11:8549Y56
46. Moskaluk CA, Zhang H, Powell SM, et al. Cdx2 protein expression in
944
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
normal and malignant human tissues: An immunohistochemical survey
using tissue microarrays. Mod Pathol 2003;16:913Y19
Skinnider BF, Folpe AL, Hennigar RA, et al. Distribution of
cytokeratins and vimentin in adult renal neoplasms and normal renal
tissue: Potential utility of a cytokeratin antibody panel in the differential
diagnosis of renal tumors. Am J Surg Pathol 2005;29:747Y54
McGregor DK, Khurana KK, Cao C, et al. Diagnosing primary and
metastatic renal cell carcinoma: The use of the monoclonal antibody
Frenal cell carcinoma marker.’ Am J Surg Pathol 2001;25:1485Y92
Avery AK, Beckstead J, Renshaw AA, et al. Use of antibodies to RCC
and CD10 in the differential diagnosis of renal neoplasms. Am J Surg
Pathol 2000;24:203Y10
Pan CC, Chen PC, Ho DM. The diagnostic utility of MOC31, BerEP4,
RCC marker and CD10 in the classification of renal cell carcinoma and
renal oncocytoma: An immunohistochemical analysis of 328 cases.
Histopathology 2004;45:452Y59
Dabbs DJ. Immunohistology of the breast. In: Dabbs DJ, ed. Diagnostic
Immunohistochemistry. Philadelphia: Churchill Livingstone, 2006:
669Y745
Ludwin SK, Kosek JC, Eng LF. The topographical distribution of S-100
and GFA proteins in the adult rat brain: An immunohistochemical study
using horseradish peroxidase-labelled antibodies. J Comp Neurol 1976;
165:197Y207
Wick MR. Immunohistology of melanocytic neoplasms. In: Dabbs DJ,
ed. Diagnostic Immunohistochemistry. Philadelphia: Churchill Livingstone, 2006:162Y79
Kawakami Y, Eliyahu S, Delgado CH, et al. Cloning of the gene coding
for a shared human melanoma antigen recognized by autologous T cells
infiltrating into tumor. Proc Natl Acad Sci U S A 1994;91:3515Y19
Zimmer C, Gottschalk J, Goebel S, et al. Melanoma-associated antigens
in tumours of the nervous system: An immunohistochemical study with
the monoclonal antibody HMB-45. Virchows Arch A Pathol Anat
Histopathol 1992;420:121Y26
Banerjee SS, Harris M. Morphological and immunophenotypic variations in malignant melanoma. Histopathology 2000;36:387Y402
Takeda K, Yasumoto K, Takada R, et al. Induction of melanocytespecific microphthalmia-associated transcription factor by Wnt-3a.
J Biol Chem 2000;275:14013Y16
Ó 2006 American Association of Neuropathologists, Inc.
Copyright @ 2006 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.