J Oral Pathol Med (2009) 38: 150–159

ª 2009 John Wiley & Sons A/S Æ All rights reserved doi: 10.1111/j.1600-0714.2008.00725.x

www.blackwellmunksgaard.com/jopm

Granular cell tumor of the oral cavity: updated

immunohistochemical profile
Marilena Vered1, William M. Carpenter2, Amos Buchner1
1
Department of Oral Pathology and Oral Medicine, The Maurice and Gabriela Goldschleger School of Dental Medicine, Tel Aviv
University, Tel Aviv, Israel; 2Department of Pathology and Medicine, University of the Pacific, Arthur A. Dugoni School of Dentistry,
San Francisco, CA, USA

BACKGROUND: Granular cell tumor (GCT) is a benign Introduction

lesion that occurs at different body sites with prepon-
Granular cell tumor (GCT) is a benign lesion charac-
derance to the oral cavity. It is generally believed to be of
terized by the accumulation of plump cells with abun-
schwann cell/neural cell origin. We used a large panel of
dant granular cytoplasm. The lesions tend to be poorly
both traditional and recently developed antibodies in an
circumscribed and extend into adjacent soft tissues.
attempt to trace the origin of GCTs on the basis of their
GCTs may arise in the skin, soft tissues, breast and
immunoprofile.
lungs, but over 50% involve the head and neck, with the
METHODS: The patients’ demographic data and the
tongue being the most common single site (1). GCT was
cytological and architectural features of the lesions were
originally described by Abrikosoff in 1926 and believed
analyzed in a large series of oral GCTs (n = 68). Forty-
to be of muscular differentiation (2), a theory that was
two lesions were also submitted to a panel of immuno-
subsequently abandoned. A wide variety of cell types
histochemical stains with antibodies against S-100, CD-68
have been proposed as the cells of origin, including
(KP-1 and PG-M1), vimentin, calretinin, NKI ⁄ C3, PGP9.5,
histiocytes, fibroblasts, myoblasts, neural sheath cells,
p75/NGFR and inhibin-a.
neuroendocrine cells, and undifferentiated mesenchymal
RESULTS: The tongue was the most common location of
cells (3–5). The histogenesis of GCTs has remained
oral GCTs (81%). The granular cells demonstrated a wide
enigmatic in spite of a vast number of immunohisto-
array of cytological features in terms of cell shape and
chemical and ultrastructural studies. A neural origin or
position of the nucleus. In addition, the lesions showed
differentiation, in particular of the schwann cell type, is
different architectural patterns, including infiltration’
currently in favor. This is based on the close anatomical
with satellite nodules. Interestingly, no recurrences were
relationship of GCTs to peripheral nerve fibers, on the
reported, even in lesions that were not completely
ultrastructural demonstration of myelin figures and
excised. Granular cells were usually found to be strongly
axon-like structures, and on its immunohistochemical
and diffusely positive for p75, vimentin, calretinin and
reactivity with S-100 protein, neuron-specific enolase
NKI/C3, inhibin-a, PGP9.5, and S-100.
and myelin proteins (6–8). GCTs from various sites of
CONCLUSIONS: Immunoreactivity of the granular cells
the body were recently found to be positive to a number
to a broad panel of antibodies that characterize different
of new markers of alleged neural differentiation as well
tissues does not confirm any particular cell type for the
as to several non-neural markers, such as inhibin-a (9).
histogenetic origin of GCTs. Furthermore, GCTs could
The aim of this study was to expand the present
be regarded as lesions that reflect a local metabolic or
knowledge on GCTs involving the oral mucosa by
reactive change rather than a true neoplasm.
reassessing the morphological features of a large series
J Oral Pathol Med (2009) 38: 150–159
of cases and delineating the immunohistochemical
profile of the lesional cells in light of recent observations
Keywords: granular cell tumor; immunoprofile; origin
from other sites of the body that revealed reactivity to
various new immunohistochemical markers.

Correspondence: Dr Marilena Vered, Department of Oral Pathology Materials and methods

and Oral Medicine, School of Dental Medicine, Tel Aviv University,
Tel Aviv 69978, Israel. Tel: +972 3 640 9305, Fax: +972 3 640 9250,
E-mail: mvered@post.tau.ac.il
Accepted for publication September 13, 2008
Case selection
The files of the Pacific Oral and Maxillofacial Pathology
Laboratory of Arthur A. Dugoni School of Dentistry,
 Immunoprofile of oral granular cell tumor
Vered et al.

University of the Pacific at San Francisco, CA, USA,
served as the source material for this study. Files were
systematically searched for all cases of oral GCTs
during a 20-year period (1987–2006). A total of 68 cases
were retrieved and information on the patient’s age,
gender and lesion location was obtained from the biopsy
submission forms. There were paraffin blocks from
various oral sites for 42 of these cases and so detailed
histopathological and immunohistochemical studies
could be carried out on them.
Morphological assessment
The following morphological parameters were assessed
on the hematoxylin and eosin-stained slides.
• Architecture of the lining of the oral epithelium –
normal width and regular rete ridges, acanthosis,
pseudoepitheliomatous hyperplasia (PSE), and
florid PSE.
• Lesions’ relationship to the lining of the oral epithe-
lium – direct contact with it or separated from it by a
band of connective tissue.
• Lesions’ architectural pattern – circumscribed nod-
ule, poorly circumscribed, infiltrative. In the infil-
trative pattern, the granular cells appear to infiltrate
or invade’ between the muscle fibers to a consider-
able distance, and satellite nodules that are remote
from the periphery of the main lesion can be
observed.
• Surgical margin status – either free of tumor or
involved by tumor.
Immunohistochemical stains
A panel of nine commercially available primary anti-
bodies was used (Table 1). Immunoreactivity was
detected using a SuperPicTure polymer kit (Zymed-
Invitrogen, Carlsbad, CA, USA) according to the
manufacturer’s recommendations. For negative con-
trols, the slides were treated with the same procedures,
including antigen retrieval, except for the application of
the primary antibodies.
151
Immunohistochemical stain assessment
Immunoreactivity was semi-quantitatively scored acc-
ording to the method used by Laskin et al. (10): 0 = no
detectable immunoreactivity, 1 = 1–10% of the tumor
cells are reactive, 2 = 11–25% of the tumor cells are
reactive, 3 = 26–50% of the tumor cells are reactive,
and 4 = >50% of the tumor cells are reactive. In
addition, immunoreactivity was graded according to the
staining intensity, as performed by Williams and
Williams (5) and Fine and Li (11): )’ = negative,
+’ = weak; ++’ = moderate, and +++’ =
strong.
Results
Clinical findings (n = 68)
The clinical findings reported in this study are based on
68 cases of oral GCT. The patients ranged in age from 7
to 82 years (mean 38.3 years and median 36 years).
There were 46 (68%) females and 22 (32%) males,
yielding a female-to-male ratio of 2.1:1. The tongue was
the most common oral location (55 cases, 80.8%). Of
those lingual lesions, 33 (48.5%) were located on the
dorsal surface, 19 (27.9%) on the lateral border and
three (4.4%) on the ventral surface. Of the remaining
tumors, four (5.9%) involved the lower lip, three (4.4%)
the upper lip, three (4.4%) the buccal mucosa, two
(2.9%) the soft palate and one (1.5%) the floor of the
mouth.
Morphological findings (n = 42)
The GCT cells were observed as being round, oval,
polygonal, or slightly elongated. The cell borders tended
to be indistinct, often giving the impression of a
syncytium. The nuclei ranged from small, dark and
hyperchromatic to large with a vesicular chromatin
pattern, and their location was in a peripheral or central
position. The cytoplasm was eosinophilic with fine-to-
coarse granules. Smaller cells containing granules were
occasionally found to be interspersed between the large
granular cells (interstitial cells).

Table 1 Source of antibodies and laboratory procedures

Antigen retrieval
Manufacturer Type of antibody procedure Concentration Control tissue
S-100 Dako A/S, Glostrup, Denmark Rabbit polyclonal Protease K 0.1% 1:500 Malignant melanoma
CD-68 DakoCytomation, Mouse, clone KP-1 Protease K 0.1% 1:100 Lymph node
Carpenteria, CA, USA
CD-68 Dako A ⁄ S, Glostrup, Denmark Mouse, clone PG-M1 Protease K 0.1% 1:100 Lymph node
Vimentin Dako A ⁄ S, Glostrup, Denmark Mouse, V-9 Citrate buffer pH = 6 1:500 Malignant melanoma
Calretinin Zymed, San Francisco, Rabbit polyclonal, Pressure cook, Nuclear Ready to use Mesothelioma
CA, USA ready to use Decloaker,a pH 9.5
NKI ⁄ C3 Biogenex, San Ramon, Mouse, clone AM077 – Ready to use Malignant melanoma
CA, USA
PGP9.5 Diagnostic BioSystems, Rabbit polyclonal Pressure cook, Nuclear 1:60 Adenocarcinoma of
Pleasanton, CA, USA Decloaker pH 9.5 pancreas
p75 ⁄ NGFR Diagnostic BioSystems, Mouse, clone ME 20.4 Protease K 0.1% 1:50 Malignant melanoma
Pleasanton, CA, USA
Inhibin-a ABD Serotec, Mouse, clone R1 Pressure cook, Nuclear 1:50 Normal human testis
Oxfordshire, UK Decloaker pH 9.5
a
Biocare Medical, Concord, CA, USA

J Oral Pathol Med

 Immunoprofile of oral granular cell tumor
Vered et al.
152
The cells were arranged in sheets, ribbons, and islands
separated by fibrous tissue septa. The granular cells
appeared to involve or replace the muscular tissue. They
grew along the muscle fibers and seem to merge with
them. There appeared to be a transition from normal
skeletal muscle fibers to granular cells. The granular
cells appeared to be streaming off or metaplastically
arising from the muscle fibers. The granular cells
enveloped small nerve bundles in only nine (21.4%)
cases, and replacement of nerves by tumor cells could be
observed in five of them.
We analyzed the morphology of the lining epithelium
in 40 cases (two lesions were devoid of epithelial lining).
Normal architecture was seen in 18 (45%) cases,
acanthosis in 12 (30%), PSE in four (10%) and florid
PSE, mimicking squamous cell carcinoma, in the
remaining six (15%) cases (Fig. 1). There were eight
cases of PSE and florid PSE in the tongue, one in the
soft palate and one in the lower lip mucosa. An
A scribed and one was infiltrative. No recurrence was
B immunohistochemical stains (Fig. 4).
Figure 1 (A) Florid pseudoepitheliomatous hyperplasia (PSE) asso-
ciated with granular cell tumor (hematoxylin and eosin, original
magnification ·40). (B) An area from A at a higher magnification that
highlights the apparent similarity between PSE and well differentiated
squamous cell carcinoma (hematoxylin and eosin, original magnifica-
tion ·100).
J Oral Pathol Med
assessment of the relationship of the lesions to the
lining epithelium revealed that 28 (70%) lesions were in
direct contact with the epithelium: 10 of them were
associated with PSE and florid PSE, 10 with acanthosis
and eight with normal appearing epithelium. In the
remaining 12 (30%) lesions, a band of connective tissue
was interposed between the lining epithelium and the
lesions: the lining epithelium had normal architecture in
10 lesions and it was occasionally acanthotic in two.
There was a direct contact between the lesion and the
lining epithelium in all the cases with PSE and florid
PSE.
Twenty-four (57%) lesions were poorly circum-
scribed, 14 (33.3%) had an infiltrative pattern, and
only four (9.5%) appeared as a circumscribed nodule.
The lesions involved the surgical margins in 23 (55%)
cases, and the surgical margins were clean in the
remaining 19 (45%). Among the cases with involved
margins, one lesion showed a circumscribed architec-
tural pattern, nine were poorly circumscribed and 13
were infiltrative. Among lesions with clean margins,
three were circumscribed, 15 were poorly circum-
reported for any of these cases, irrespective of the
margin status.
Immunohistochemical findings (n = 42)
The results obtained with the nine immunohistochem-
ical markers we used in this study are summarized in
Table 2. Vimentin, S-100, calretinin, NKI ⁄ C3, p75, and
PGP9.5 and inhibin-a produced a strong and diffuse
staining of the granular cells (Fig. 2A–G). The immu-
noreactivity of the granular cells for the various
antibodies highlighted the architectural pattern of the
lesions, either as a circumscribed nodule or as an
extensive infiltrative lesion with deposits of tumor cells
remote from the main mass (Fig. 3). Furthermore,
transition between the muscle fibers and the granular
cells could be easily identified with the aid of the
Most of the lesional cells had a staining score of 4 in
p75 (98%), vimentin (93%), calretinin and NKI/C3
(90%), inhibin-a (86%), PGP9.5 (81%) and S-100
(79%). In contrast, only 26% cases demonstrated a
staining score of 4 with KP-1, while a majority (71%) of
them were immunonegative. Interestingly, the interstitial
cells among the larger granular cells were KP-1 positive
in 46% cases. None of the lesions reached a staining
score of 4 when PG-M1 was used: 36% had a staining
score of 2 (11–25% stained cells) and 45% had a
staining score of 1 (1–10% stained cells), with 19% of
the lesions being immunonegative for PG-M1.
Characterization of the selected antibodies and their
expression in GCTs from various body sites and from
oral mucosa
For the purpose of comparing the immunohistochemical
findings of the present study with those of other studies
in the literature, we considered only articles published in
the English-language literature from 1990 to the present.
In the following brief overview, we first discuss the most
 Immunoprofile of oral granular cell tumor
Vered et al.

153
common immunostains that have been traditionally

Total (%)

(71.4)
Score 0

(4.8)

(7.1)
(19)
used for identification of GCTs (i.e. S-100 and CD-68)
followed by a description of the stains that have been

30
0

8
0
2
0
0
0
3
recently added to widen the immunohistochemical
profile of GCTs.

Total (%)

19 (45.2)
S-100
This protein belongs to a family of small acidic EF-hand
0
0

0
0
0
0
0
0
Score 1 (1–10%)

calcium-binding proteins initially discovered in brain

extracts (12). Although S-100 is considered one of the
+
0
0
0
0
0
0
0
0
0
most characteristic markers for melanomas, other S-100
positive tumor types enter into the differential diagnosis,
++
0
0
4
0
0
0
0
0
0
including selected histiocytic proliferations, poorly dif-
ferentiated carcinomas, peripheral nerve sheath tumors
and tumors that contain myoepithelial cells. S-100
+++

15
0
0

0
0
0
0
0
0
immunoreactivity of GCTs is the most widely used
marker for identification of the granular cells, which
constantly show a strong and diffuse staining (9, 11, 13–
Total (%)

16). This is also true for oral lesions (5, 17). However, it
15 (35.7)
1 (2.4)

1 (2.4)

is interesting to note that a new distinct clinical subtype

Staining pattern (score and intensity) for the various immunohistochemical stains in the 42 lesions of granular cell tumors

of GCTs, which do not show S-100 immunoreactivity

0
0

0
0
0
0
Score 2 (11–25%)

was identified less than two decades ago (18). One such
Interstitial cells were strongly immunoreactive (+++) in 15 cases, moderately (++) in four and weakly (+) in one case.

case from the oral mucosa has been reported in the

+
0
0
1
0
0
0
0
0
1

literature (19). In our current study of oral GCTs, all 42

cases were found to be diffusely and strongly positive for
++

S-100.
0
0
0
1
0
0
0
0
0
Staining intensity

CD-68
+++

This is a heavily glycosylated membrane protein, closely

14
0
0

0
0
0
0
0
0

related to the family of lysosomal-associated, mucin-

like membrane proteins (20, 21). A number of mono-
Total (%)

clonal antibodies are available against the CD-68

(2.4)

(4.8)
(4.8)
(9.5)

(2.4)
(4.8)
(21)

(19)

glycoprotein, including KP-1 and PG-M1. Positive

immunoreaction to these antibodies may be found in
9
1
0
2
2
4
8
1
2
Score 3 (26–50%)

cells of macrophage lineage. CD-68 was occasionally

identified in cells of fibroblastic (20) or schwannian (22)
+
0
0
0
0
0
0
1
0
0

lineages. It was also reported that the PG-M1 antibody

was dependent upon tissue pre-treatment procedures
++

and that its signal was less than that of KP-1 under
7
1
0
2
2
4
1
1
2

similar conditions (20). GCTs from different body

Immunostain intensity: +, weak; ++, moderate; +++, strong.

locations (including oral mucosa) were usually KP-1

+++

positive (15, 16, 23), with the exception of one study of

2
0
0
0
0
0
6
0
0

non-oral lesions (24). PG-M1 was investigated in only

one small series of non-oral lesions and it was found to
Total (%)

be weakly positive (4). In our current study, about 28%

(79)
(26)

(93)
(90)
(90)
(81)
(98)
(86)

of the cases were diffusely KP-1 positive, almost 80%

33
11

39
38
38
34
41
36
0

were PG-M1 positive, but the staining did not exceed

Score 4 (>50%)

25% of the lesional cells.

+
1
2
0
0
1
0
0
0
0

Vimentin
This is considered to be the primordial’ member of the
++
17

21
10
16

11
5
0

7
2

intermediate filament family (12). It is present in most

fetal cells and is ubiquitously expressed by mesenchymal
tissues in adult tissues. Therefore, it is not considered to
+++

be cell-type specific. Several studies have shown variable

15

18
27
22
27
39
25
4
0

staining intensity of GCT granular cells to vimentin

(4, 14, 25, 26). There were two cases of GCT from the
p75 ⁄ NGFR

oral mucosa reported in only one study and they were

Calretinin

Inhibin-a
Vimentin

NKI ⁄ C3

found to be vimentin negative (17). All the cases we now

Score of

PGP9.5
PG-M1
Table 2

stained

KP-1a
S-100

describe were vimentin positive, and almost all of them

cells

(93%) stained strongly and diffusely.

a

J Oral Pathol Med

 Immunoprofile of oral granular cell tumor
Vered et al.

154 A B

C D

E F

Figure 2 Lesional cells in granular cell tumors yield a positive immunoreactivity for p75 (A), vimentin (B), calretinin (C), NKI ⁄ C3 (D), inhibin-a
(E), PGP9.5 (F) and S-100 (G) (AEC method, original magnification ·200).

J Oral Pathol Med


A Calretinin
B cells of melanocytic origin (12). The corresponding
Figure 3 (A) Granular cell tumor demonstrates an infiltrative pattern
consistent with satellite nodules (AEC method, anti-NKI ⁄ C3 anti-
body, original magnification ·40). (B) Granular cell tumor demon-
strates an infiltrative pattern consistent with strands of granular cells
widely extending among the deep layers of the tongue muscle tissue
(AEC method, anti-NKI ⁄ C3 antibody, original magnification ·20).
Figure 4 Transition between striated muscle fibers and granular cells
highlighted by inhibin-a (AEC method, original magnification ·1000,
oil immersion).
Immunoprofile of oral granular cell tumor
Vered et al.
155
This was initially isolated from central nervous system
tissues. Calretinin belongs to the same family of EF-
hand proteins as the S-100 proteins (12). It is primarily a
marker of benign mesothelial cells. In addition, it has
been identified in mast cells, schwannomas, adrenocor-
tical tumors and in sex cord-stromal tumors of the
ovary. Calretinin was examined in two studies involving
GCTs (10, 11), and it was found to be positive in as high
as 93% of the tumors. Four cases of oral GCTs were
included in one study (11), and three of them were
weakly-to-moderately positive for calretinin. In con-
trast, 95% of the cases were strongly and diffusely
calretinin positive in our present study.
NKI/C3
gp100’ comprises a group of glycoproteins restricted to
nucleic acid sequence gives rise to gp100 and gp10 and
to PMel 17 by alternative splicing. Both gp100 and PMel
17 are localized to the inner membranes of pre-melano-
somes. NKI/C3 is one of the antibodies raised against
gp100/PMel 17. In general, these antibodies have
variable specificity and sensitivity for melanocytes, nevi
and melanomas, but there is evidence that they can label
non-melanocytic tumors, such as those composed of
epithelioid perivascular cells. On the other hand, mel-
anomas with a more fibroblastic/schwannian phenotype
are barely stained because of the loss of the pre-
melanosomes. NKI ⁄ C3 immunoreaction in non-oral
GCTs was examined in one small series (7) and found
to be positive in all cases. Expression of NKI/C3 in
GCTs of oral soft tissues has not been yet investigated.
All 42 cases in the present series were diffusely NKI/C3
positive.
PGP9.5
The protein gene product 9.5 (PGP9.5), also known as
ubiquitin carboxyl-terminal hydrolase (UCH-L1), is a
protein that had been initially isolated from whole brain
extract (27). It was originally felt that PGP9.5 was
strictly confined to neurons and neuroendocrine cells,
but it has since been documented in a vast array of cell
types, both normal and neoplastic, of both epithelial
and mesenchymal lineages. It is now believed that its
cellular expression reflects metabolically active regions,
irrespective of the cell origin. Expression of PGP9.5 in
GCTs was investigated in several studies (7, 15, 28) and
was found to be positive in all cases, including small
series of oral lesions (7, 15, 28). All 42 of our current
cases were strongly and diffusely positive for PGP9.5.
p75/NGFR
Also known as low-affinity nerve growth factor recep-
tor, this is a member of the tumor necrosis factor
receptor family (29). It may modulate binding of nerve
growth factor to the functionally high-affinity receptor
tyrosine kinase-A. Immunoreactivity for p75 was
reported in a large group of both fetal and adult tissues
as well as in neural and non-neural tumors. Normal cell
types that were found to be positive for p75 included
J Oral Pathol Med
 Immunoprofile of oral granular cell tumor
Vered et al.
156
adult pericytes, perivascular fibroblasts, basal cells of
several types of epithelia, perineural cells and dendritic
reticulum cells. Several types of mesenchymal, non-
neural tumors (e.g. dermato-fibrosarcoma protuberans,
rhabdomyosarcoma) together with neural tumors (e.g.
schwann cell tumors, malignant peripheral nerve sheath
tumors) were also found to be consistently positive for
p75. As such, it would appear that p75 is not a specific
marker for nerve sheath tumors. p75 immunoreactivity
was examined in a few studies on GCTs (10, 28, 29), in
which only two cases of the oral mucosa were included
(28): all cases were found to be p75 positive. All 42 cases
involving the oral mucosa in the current study were
found to be strongly and diffusely positive for p75.
Inhibin-a
This is a glycoprotein hormone that participates in the
regulation of the pituitary-gonadal feedback system
(12). Inhibin-a is a sensitive and relatively specific
marker of sex cord-stromal tumors of the ovary.
Adrenal cortical neoplasms also commonly express
inhibin-a. Inhibin-a was found to be negative in
melanomas, lymphomas and metastatic carcinomas.
Expression of inhibin-a in GCTs was investigated in a
series of studies (9, 11, 13, 15, 30) that also included a
few cases from the oral cavity (11, 15, 30). No sound
rationale for examining this marker in GCTs was
provided by Murakata et al. (9) who were the first to
introduce inhibin-a as part of the immunohistochemical
profile of this tumor. With the exception of one study by
Parfitt et al. (13), GCTs in general and those from the
oral mucosa in particular were found to be inhibin-a
positive in all the other studies. Thirty-eight of our 42
cases (90%) were diffusely positive for inhibin-a.
Discussion
The clinical findings of our study are generally in
agreement with those reported in the literature (3, 31,
32). Oral GCTs in most studies show a predilection for
females (twice as common than males), and the age range
of the patients is from childhood to elderly, with a mean
appearance in the fourth decade and 67–81% of the
lesions occurring in the tongue (3, 31). Reports on the
histomorphologic features of GCTs are inconsistent.
The findings of our study revealed that the granular cells
may be round, oval, polygonal or spindle shaped, that the
nuclei may be dark or vesicular and that their location is
not in a consistent position in the cell. Other authors have
failed to describe the variability of these features. Some
authors, for example, report that the nuclei of GCTs are
dark and hyperchromatic (1, 32, 33), while others report
that they are pale and vesicular (34–36). Some authors
claim that the nuclei are located in a central position
(1, 37), while others describe them in an acentric position
(35). We observed that the cell borders are usually
indistinct and that they sometimes give the impression of
syncytium, an observation supported by some other
investigators (1, 33, 37), but not by all (38, 39).
Oral GCTs reportedly exhibit PSE in about 50% of
the cases (32). It has been suggested that stimulation of
J Oral Pathol Med
basal cell proliferation occurs through an interaction
between the granular cells and the neighboring epithelial
cells (36). In our study, varying degrees of PSE were
present in 25% of the cases, and 10% of them exhibited
florid PSE mimicking squamous cell carcinoma. Other
authors reported that the presence of PSE ranged from
8.5% (31), to 31% (40), to 45% (3), and to 67% (36).
The difference probably lies in the definition of PSE and,
especially, in how PSE was differentiated from moder-
ate-to-severe acanthosis.
The association of nerve fibers with GCTs and
especially oral GCTs is of interest. In most of our study
cases, there was a close association between the granular
cells and the musculature, and merging of granular cells
with muscle fibers was prominent. In contrast, clustering
of granular cells around nerve fibers could be observed
in only nine (21%) cases and replacement of nerves by
tumor cells could be identified in five of them. Stewart
et al. (3) reported that neural investment of granular
cells could be identified in only 8% of their cases.
A large body of evidence at the morphological,
ultrastructural and immunohistochemical levels has
been accumulating during the past years in support of
the theory that the cell of origin for GCTs is neural with
a schwannian’ differentiation (1, 32, 33, 35, 41).
However, more recent findings have cast doubt on the
neural origin of these tumors (3, 6, 24).
Cytoplasmic granularity, such as that seen in GCT, is
not a feature unique to this type of lesion. It has been
observed to various extents in cells from a vast array of
conditions, encompassing benign and malignant tumors
as well as reactive lesions (6, 18, 25, 42–44). Further-
more, granular cells were experimentally induced fol-
lowing exposure to cyclophosphamide (6). In the oral
cavity, granular cells have been associated with lesions
other than GCTs, including ameloblastoma, ameloblas-
tic fibroma, odontogenic fibroma, odontogenic cysts,
congenital epulis of the newborn and oral lichen planus
(44). In addition, reactive granular cells that occur in
areas of tissue injury and gingival lesions resulting from
inflammatory hyperplasia may exhibit secondary gran-
ular cell change (45, 46).
Lesions, in which part of the cells exhibit granular
cells, have been found to be of neural, smooth muscle,
striated muscle, endothelial, primitive mesenchymal,
histiocytic and epithelial differentiation, further empha-
sizing the ambiguity of the granular cells (18). When
granular cells of different cell lineages were ultrastruc-
turally analyzed, they generally revealed lysosomal
granules and cytoplasmic filaments, similar or identical
to those observed in conventional GCTs (3, 42, 47). The
current belief that GCTs are of schwann cell origin has
been questioned in light of the observation that myelin
and myelin-forming cells are extremely rare in the
neurohypophysis and, in spite of this, tumors with
similar morphological and ultrastructural features of
typical GCT are found at this site (26). Collectively,
these observations may indicate that, on morphologic
grounds, cytoplasmic granularity is not powerful
enough to consider a GCT as being a distinct pathologic
entity. Analogically, other cytoplasmic changes, such as

those recognized in clear, oncocytic, rhabdoid and signet
cells which are not necessarily related to a specific cell of
origin, are seen in both benign and malignant condi-
tions, and usually do not constitute a separate entity (6,
18).
The immunohistochemical profile of GCTs has
undergone extensive analysis. They have been tradition-
ally known to be positive for the S-100 protein, which
has been used to serve as sound evidence for their
schwannian/neural origin. CD-68, a marker of lyso-
somes, mostly associated with macrophages, is also
usually positive in GCTs, and its presence is explained
by the assumption that schwann cells acquire lysosomes
during phagocytosis of myelin, a phenomenon which is
known to occur, for example, in peripheral nerves
showing Wallerian degeneration as well as in traumatic
neuromas (4, 23). With the considerable progression and
interpretation of immunohistochemical stains over the
last two decades, together with the improvement of
technical laboratory procedures, it became clearer that
positive immunoreaction to S-100 and CD-68 cannot
fully support a neural origin of the granular cells in
GCTs. The variety of cell types positive for theses
proteins is wider than was originally thought. For
example, S-100 reaction was identified in cells of non-
neural lineages, such as macrophages (48), normal
skeletal muscle cells and rhabdomyoma (3) and granular
cells in ameloblastoma (31, 49). Furthermore, in the
present study, the frequency of S-100 staining was
relatively low and lagged behind staining with p75,
NKI ⁄ C3 and PGP9.5. These latter immunostains are
not tissue-specific but are rather expressed in a very
large variety of adult tissues and their neoplastic
counterparts (as we noted in the section in which we
show the characterization of the selected antibodies and
their expression in GCTs). Adding to this, the positive
immunoreaction to inhibin-a, the granular cells of
GCTs takes on an immunohistochemical profile that
cannot be attributed to any defined cell lineage. Indeed,
this unusual and complex immunoprofile may be better
explained in terms of a stress-induced degenerative
process or a metabolic disorder that leads to loss of the
cellular proteins and organelles that once used to be the
footprints of the cell of origin and to the evolvement of
aberrant and uncharacteristic proteins (3, 6, 24, 43).
One of the most interesting observations in this study
was the transition of striated muscle cells to granular
cells, as highlighted by the inhibin-a staining results.
This reflects the modification that striated muscle cells
undergo and the process which gives them a granular
morphology and an altered immunohistochemical phe-
notype that is completely unrelated to that of normal
skeletal muscle. A similar evolution of granular cells
from striated muscle fibers in cases of GCTs was
reported by Stewart et al. (3) at both light and electron
microscopic levels. Moreover, those authors also found
similar modifications in granular cells that seemed to
evolve from peripheral nerve fibers. This observation
further strengthens the assumption that the appearance
of granular cells is not limited to one cell type, but is
rather site-dependent, and that the cell of origin may
Immunoprofile of oral granular cell tumor
Vered et al.
157
differ according to its accessibility at any site (3, 6, 15,
24, 26). As many of the GCTs occur in the tongue,
which is rich in striated muscle fibers, it is reasonable to
assume that these fibers could be the source for at least
some of the GCTs, thus supporting the initial theory on
the myogenic origin of these lesions. However, the
tongue is rich in nerve fibers encased by schwann cells
that could also be the source for GCTs that cannot be
morphologically distinguished from those of muscle cell
origin following an event of metabolic distress (3). In
contrast, in the subcutaneous tissues, which are devoid
of striated muscle fibers, other mesenchymal tissues, e.g.
nerve fibers, smooth muscles and endothelial cells, could
give rise to GCTs, given that they undergo granular cell
changes induced by metabolic stress.
Our large series of GCTs displayed three main
architectural patterns in which some of the lesions
appeared as small, well-circumscribed nodules, while
others were larger and poorly circumscribed, and yet
others had what appeared to be an impressive infiltrative
pattern with remote satellite nodules. On the basis of a
metabolic disorder theory, it can be assumed that the
former could represent a benign mesenchymal tumor
that underwent an extensive or complete granular cell
change. The remaining patterns may be compatible with
a diffuse process of metabolically induced cytoplasmic
granular change in the mesenchymal cells that are
naturally present. This assumption can be supported by
the fact that most of the lesions with an infiltrative
pattern are excised with positive margins [55% in the
present series and 71% in other series (40)] and yet there
are almost no recurrences, whereas in frank tumors,
recurrent lesions would be expected, especially in light of
such an apparently infiltrative behavior. Thus, including
the word tumor’ (meaning neoplasm) in labeling the
lesion appears to be inaccurate, as probably, at least in
the majority of the lesions, the process is rather
metabolic or reactive in nature and not neoplastic.
In summary, analysis of a large series of oral GCTs
enabled us to further expand our knowledge and insight
into these lesions. Morphologically, oral GCTs demon-
strated a wide variety of features and architectural
patterns, but they all still exhibited a benign behavior
and did not recur irrespective of the status of the
margins. Immunohistochemically, reactivity of the gran-
ular cells to a broad panel of antibodies that character-
ize different tissues does not confirm any particular cell
type for the histogenetic origin of GCTs. Furthermore,
GCTs could be regarded as lesions that reflect a local
metabolic or reactive change rather than a true neo-
plasm.
References
1. Speight P. Granular cell tumor. In: Barnes L, Eveson JW,
Reichart P, Sidransky D, eds. World Health Organization
Classification of Tumours. Pathology and genetics. Head
and neck tumours. Lyon: IARC Press, 2005; 185–6.
2. Abrikossoff A. Ueber Myome, ausgehend von der quer-
gestreifen willkurlichen Muskulatur. Virchows Arch A
Pathol Pathol Anat 1926; 260: 215–33.
J Oral Pathol Med
 Immunoprofile of oral granular cell tumor
Vered et al.
158
3. Stewart CM, Watson RE, Eversole LR. Oral granular cell
tumors: a clinicopathologic and immunocytochemical
study. Oral Surg Oral Med Oral Pathol 1988; 65: 427–35.
4. Kaiserling E, Ruck P, Xiao J-C. Congenital epulis and
granular cell tumor. A histologic and immunohistochem-
ical study. Oral Surg Oral Med Oral Pathol Oral Radiol
Endod 1995; 80: 687–97.
5. Williams HK, Williams DM. Oral granular cell tumours: a
histological and immunocytochemical study. J Oral Pathol
Med 1997; 26: 164–9.
6. Mentzel T, Wadden C, Fletcher CD. Granular cell change
in smooth muscle tumours of skin and soft tissue.
Histopathology 1994; 24: 223–31.
7. Mahalingam M, Lo-Piccolo D, Byers HR. Expression of
PGP 9.5 in granular sheath tumors: an immunohisto-
chemical study of six cases. J Cutan Pathol 2001; 28: 282–
6.
8. Miettinen M, Lehtonen E, Lehtola H, Ekblom P, Lehto V-
P, Virtanen I. Histogenesis of granular cell tumor – an
immunohistochemical and ultrastructural study. J Pathol
1984; 142: 221–9.
9. Murakata LA, Ishak KG. Expression of inhibin-[alpha] by
granular cell tumors of the gallbladder and extrahepatic
bile duct. Am J Surg Pathol 2001; 25: 1200–3.
10. Laskin WB, Fetsch JF, Davis CJ, Sesterhenn IA. Granular
cell tumor of the penis: clinicopathologic evaluation of 9
cases. Hum Pathol 2005; 36: 291–8.
11. Fine SW, Li M. Expression of calretinin and the alpha-
subunit of inhibin in granular cell tumors. Am J Clin
Pathol 2003; 119: 259–64.
12. Dabbs DJ. Diagnostic immunohistochemistry. China:
Churchill Livingstone, Elsevier, 2006.
13. Parfitt JR, Mc-Lean CA, Joseph MG, Streutker CJ,
Al-Hadad S, Driman DK. Granular cell tumours of the
gastrointestinal tract: expression of nestin and clinico-
pathological evaluation of 11 patients. Histopathology
2006; 48: 424–30.
14. Mazur MT, Shultz JJ, Myers JL. Granular cell tumor.
Immunohistochemical analysis of 21 benign tumors and
one malignant tumor. Arch Pathol Lab Med 1990; 114:
692–6.
15. Le BH, Boyer PJ, Lewis JE, Kapadia SB. Granular cell
tumor. Immunohistochemical assessment of inhibin-a,
protein gene product 9.5, S100 protein, CD68, and Ki-67
proliferative index with clinical correlation. Arch Pathol
Lab Med 2004; 128: 771–5.
16. Chrysomali E, Papanicolaou SI, Dekker NP, Regezi JA.
Benign neural tumors of the oral cavity. A comparative
immunohistochemical study. Oral Surg Oral Med Oral
Pathol Oral Radiol Endod 1997; 84: 381–90.
17. Junquera LM, De Vicente JC, Vega JA, Losa JL, Albertos
JM, Lopez-Arranz JS. Granular-cell tumors: an immuno-
histochemical study. Br J Oral Maxillofac Surg 1997; 35:
180–4.
18. Le-Boit PE, Barr RJ, Burall S, Metcalf JS, Benedict TS,
Wick MR. Primitive polypoid granular-cell tumor and
other cutaneous granular-cell neoplasms of apparent
nonneural origin. Am J Surg Pathol 1991; 15: 48–58.
19. Basile JR, Woo S-B. Polypoid S-100 negative granular cell
tumor of the oral cavity: a case report and review of the
literature. Oral Surg Oral Med Oral Pathol Oral Radiol
Endod 2003; 96: 70–6.
20. Kunisch E, Fuhrmann R, Roth A, Winter R, Lungerhau-
sen W, Kinne RW. Macrophage specificity of three anti-
CD68 monoclonal antibodies (KP1, EBM11, and PGM1)
widely used for immunohistochemistry and flow cyto-
metry. Ann Rheum Dis 2004; 63: 774–84.
J Oral Pathol Med
21. Dabbs DJ. Diagnostic immunohistochemistry. New York:
Churchill Livingstone, 2002; 323.
22. Filie AC, Lage JM, Azumi N. Immunoreactivity of S100
protein, alpha-1-antitrypsin, and CD68 in adult and con-
genital granular cell tumors. Mod Pathol 1996; 9: 888–92.
23. Kurtin PJ, Bonin DM. Immunohistochemical demonstra-
tion of the lysosome-associated glycoprotein CD68 (KP-1)
in granular cell tumors and Schwannomas. Hum Pathol
1994; 25: 1172–8.
24. Rickert CH, Kuchelmeister K, Gullotta F. Morphological
and immunohistochemical characterization of granular
cells in non-hypophyseal tumours of the central nervous
system. Histopathology 1997; 30: 464–71.
25. Deavers M, Guinee D, Koss MN, Travis WD. Granular
cell tumors of the lung. Clinicopathologic study of 20
cases. Am J Surg Pathol, 1995; 19: 627–35.
26. Nishioka H, Ii K, Llena JF, Hirano A. Immunohisto-
chemical study of granular cell tumors of the neurohy-
pophysis. Virchows Archiv B Cell Pathol 1991; 60: 413–7.
27. Campbell LK, Thomas JR, Lamps LW, Smoller BR,
Folpe AL. Protein gene product 9.5 (PGP 9.5) is not a
specific marker of neural and nerve sheath tumors: an
immunohistochemical study of 95 mesenchymal neo-
plasms. Mod Pathol 2003; 16: 963–9.
28. Hoshi N, Tsu-Ura Y, Watanabe K, Suzuki T, Kasukawa
R, Suzuki T. Expression of immunoreactivities to 75 kDa
nerve growth factor receptor, trk gene product and
phosphotyrosine granular cell tumors. Pathol Int 1995;
45: 748–56.
29. Fanburg-Smith JC, Miettinen M. Low-affinity nerve
growth factor receptor (p75) in dermatofibroma protub-
erans and other nonneural tumors: a study of 1,150 tumors
and fetal and adult normal tissues. Hum Pathol 2001; 32:
976–83.
30. Zamecnik M, Michal M, Mukensnabl P. Reactivity of
granular cell tumors for inhibin and other markers of sex
cord and steroid cell differentiation. Am J Surg Pathol
2003; 27: 413–4.
31. Mirchandiani R, Sciubba JJ, Mir R. Granular cell lesions
of the jaws and oral cavity: a clinicopathologig, immuno-
histochemical, and ultrastructural study. J Oral Maxillo-
fac Surg 1989; 47: 1248–55.
32. Regezi JA, Sciubba JJ, Jordan RCK. Oral pathology.
Clinical pathologic correlations. China: WB Saunders and
Elsevier, 2008; 168–70.
33. Kilpatrick SE, Renner JB, Creager A. Diagnostic muscu-
loskeletal surgical pathology. Clinicoradiologic and cyto-
logic correlations. China: Saunders, 2004; 86–8.
34. Neville BW, Damm DD, Allen CM, Bouquot JE. Oral and
maxillofacial pathology. Philadelphia: WB Saunders,
2002; 465–6.
35. Gnepp DG. Diagnostic surgical pathology of the head and
neck. Philadelphia: WB Saunders, 2001; 198–9.
36. Chrysomali E, Nikitatis NG, Tosios K, Sauk JJ, Papa-
nicolaou SI. Immunohistochemical evaluation of cell
proliferation antigen Ki-67 and apoptosis-related proteins
Bcl-2 and caspase-3 in oral granular cell tumor. Oral
Surg Oral Med Oral Pathol Oral Radiol Endod 2003; 96:
566–72.
37. Mc-Kee PH. Pathology of the skin. London: Mosby-
Wolfe, 1996; 16.41–2.
38. Cawson RA, Binnie WH, Speight P, Barrett AW, Wright
JM. Lucas’s pathology of tumors of the oral tissues.
London: Churchill Livingstone, 1998; 337–8.
39. Sapp JP, Eversole LR, Wysocki GP. Contemporary oral
and maxillofacial pathology. St Louis: Mosby, 2007; 309–
11.

40. Worsaae N, Schwartz O, Pindborg JJ. Follow-up study
of 14 granular cell tumors. Int J Oral Surg 1979; 8: 133–
9.
41. Weiss SW, Goldblum JR. Enzinger and Weiss’s soft tissue
tumors. St Louis: Mosby, 2001; 1178–85.
42. Lazar AJ, Fletcher CD. Primitive nonneural granular cell
tumors of skin. Clinicopathologic analysis of 13 cases.
Am J Surg Pathol 2005; 29: 927–34.
43. Chaudhry IH, Calonje E. Dermal non-neural granular cell
tumor (so-called primitive polypoid granular cell tumor): a
distinct entity further delineated in a clinicopathological
study of 11 cases. Histopathology 2005; 47: 179–85.
44. van der Meij EH, van der Waal I. Granular cells in oral
lichen planus. Oral Dis 2001; 7: 116–8.
45. Alguacil-Garcia A, Ahing S. Reactive granular cell lesion
in gingival fibrous hyperplasia: immunohistochemical
observation. Oral Surg Oral Med Oral Pathol 1988; 66:
697–700.
46. Odell EW, Morgan PR. Biopsy pathology of the oral
tissues. London: Chapman & Hall Medical, 1998; 128–31.
Immunoprofile of oral granular cell tumor
Vered et al.
159
47. Regezi JA, Zarbo RJ, Courtney RM, Crissman JD.
Immunoreactivity of granular cell lesions of skin, mucosa,
and jaw. Cancer 1989; 64: 1455–60.
48. Chan AC, Chan JK, Cheung MM, Kapadia SB. Haemato-
plymphoid tumours. In: Barnes L, Eveson JW, Reichart P,
Sidransky D, eds. World Health Organization classification
of tumours. Pathology and genetics. Head and neck
tumours. Lyon: IARC Press, 2005; 63–4.
49. Siar CH, Ng KH. Unusual granular cell odontogenic
tumor. Report of two undescribed cases with features of
granular cell ameloblastoma and plexiform granular cell
odontogenic tumor. J Nihon Univ Sch Dent 1993; 35: 134–8.
Acknowledgements
The study was supported by the Ed and Herb Stein Chair in Oral
Pathology, Tel Aviv University. The authors would like to thank
Mrs Hanna Vered for technical assistance and Ms Esther Eshkol for
editorial assistance.
J Oral Pathol Med