Atypical Fibroxanthoma: A Histological and
Immunohistochemical Review of 171 Cases
Trevor W. Beer, MBChB, MRCPath, FRCPA, Paul Drury, BSc (Med Sci),
and Peter J. Heenan, MBBS, FRCPath, FRCPA
Abstract: The clinical and histological features of 171 atypical
fibroxanthomas (AFX) from a single institution in Western Australia
are outlined. This area experiences high levels of solar radiation, and
all assessable biopsies showed solar elastosis. Patients were aged
between 41 and 97 years (median age 74), with 76% of tumors
occurring in men (male to female ratio approximately 3 to 1). Most
tumors were small, with a median diameter of 10 mm and a range of
4–35 mm. Only 5% exceeded 20 mm in diameter. Most AFX were
well-circumscribed dermal lesions, with limited invasion of subcutis
in a minority. Histological variants identified included keloidal (n =
8), clear cell (n = 3), and granular cell (n = 3), plaque like (n = 4), and
myxoid (n = 1). Bland cytological appearances (spindle cell non-
pleomorphic AFX) were noted in 5 tumors, with osteoclast-like giant
cells in 2. Features suggesting regression were present in 22 cases.
Two cases recurred locally, none metastasized. No tumors expressed
melanocytic or epithelial markers. Seventy-four percent of cases
expressed smooth muscle actin, typically strongly and diffusely. No
AFX stained with desmin. Only 1 of 50 cases was CD117 positive. In
conclusion, AFX may show awide range of histological appearances,
and a panel of immunohistochemical markers is essential to make the
correct diagnosis. Histological mimics, such as poorly differentiated
squamous cell carcinoma, must be carefully excluded. Specific
diagnosis is important because there seems to be a very low risk of
recurrence or metastasis despite the frequently alarming histology.
Key Words: Atypical fibroxanthoma, case series, immunohisto-
chemistry, literature review, misdiagnosis
(Am J Dermatopathol 2010;32:533–540)
Atypical fibroxanthoma (AFX) is a relatively rare skin
neoplasm of uncertain histogenesis.1,2The majority of affected
patients are elderly with evidence of chronic sun damage. The
clinical appearances are not specific and overlap with other
skin neoplasms. Histologically, the features are frequently
highly atypical and alarming, but clinical behavior is indolent,
with only occasional recurrences at the excision site. Meta-
stases have been reported but are exceptionally rare. Many
previous reports of metastasis are of disputed authenticity
because they may predate the use of immunohistochemistry
or provide insufficient data to substantiate the diagnosis.2The
microscopic diagnosis is one of exclusion, and AFX must be
differentiated from tumors that carry a more sinister prognosis.
Immunohistochemistry plays a vital role in this distinction.
Lesions to be excluded include poorly differentiated squamous
cell carcinoma, melanoma, and leiomyosarcoma. Although
many antibodies have been evaluated to assist in the diagnosis
of AFX, most lack specificity and have poor positive predic-
tive value. Diagnosis generally relies on a lack of staining
for melanocytic, epithelial markers and desmin, with variably
positive staining for nonspecific markers such as actin, CD68,
MATERIALS AND METHODS
The histology computer archive in this laboratory was
searched for all cases diagnosed as AFX by 3 dermatopathol-
ogists (that included T.W.B. and P.J.H.) during a 9½-year
period from January 2000 to mid-2009. The histological
features of every case were reviewed on hematoxylin-
eosin–stained sections, and all immunostains were rescruti-
nized. A variety of immunohistochemical stains had been
performed on the original neoplasms that always included
a selection of epithelial and melanocytic markers. Clinical data
were correlated with these findings, and follow-up information
obtained. Standard diagnostic criteria were used to verify the
diagnosis.1,2In equivocal cases, supplementary immunostains,
further sections, or additional clinical information were
sought to clarify the diagnosis. All immunostaining had been
performed using the avidin–biotin complex method with
appropriate positive and negative controls.
The computer search identified 183 cases reported as
AFX. After case reviews, 12 neoplasms were reclassified as
tumors other than AFX (details below), leaving 171 tumors in
169 patients (2 patients developed 2 tumors each).
The patients were aged between 41 and 97 years, with
a mean age of 75 (median 74). There were 128 males and 41
females (ie 76% male, giving a male to female ratio of appro-
ximately 3:1). A variety of neoplasms had been suggested
clinically, the most common being basal cell carcinoma,
although the possibility of AFX had been suggested in rare
cases (Fig. 1). In 1 patient, the tumor had arisen at the site of
From the Cutaneous Pathology, Western Australia, Australia.
Reprints: Dr. Trevor W. Beer, MBChB, MRCPath, FRCPA, Cutaneous
Pathology, 26 Leura Street, Nedlands, Western Australia 6012, Australia
Copyright ? 2010 by Lippincott Williams & Wilkins
Am J Dermatopathol?Volume 32, Number 6, August 2010www.amjdermatopathology.com|533
a wood splinter injury on the forearm several months before.
All patients were residents of Western Australia, mainly from
Perth or surrounding areas. These locations are subject to
prolonged periods of high intensity solar ultraviolet radiation.
Two patients developed 2 separate AFX during the study
period. There were no remarkable clinical or histological
features inthese cases to distinguish them from the other cases.
Only 2 patients experienced tumor recurrence at the excision
site. In one of these cases, the original tumor had been
incompletely excised, and in the second case, the original
tumor was present 0.5 mm from the deep excision margin. No
metastases were recorded, and no lymphovascular or peri-
neural invasion were identified in any AFX.
Thevast majority of AFX occurred on the head and neck
(155 cases or 91%) (Fig. 2). Thirty-three AFX were diagnosed
on the ears, suggesting that these sites may be preferentially
affected, given their relatively small surface area. All cases on
the ear occurred in males. There was no indication that lesions
on the limbs or trunk occurred in younger patients as has been
alluded to in some older literature. The median age for tumors
on the limbs was 79 and on the trunk 75, similar to the median
age of the overall population (74 years). Where laterality was
relevant, more caseswere reported on the right side of the body
compared with the left (67 versus 51), but this difference was
not statistically significant (x2test P = 0.141).
Cases Recategorized on Review
Twelve cases were considered on review of the clinical,
histological, and immunohistochemical features not to repre-
sent AFX (Table 1). In 3 cases, the findings were ambivalent,
meaning that a precise diagnosis could not be ascertained with
certainty—these cases were therefore excluded from the study.
Five tumours showed some unequivocal cytokeratin positivity,
at least focally. In several cases, this was demonstrated
more clearly by immunostaining with 34betaE12 which had
not been originally applied. Four of these tumors exhibited
FIGURE 1. The clinical appearance of an AFX on the forehead of
an 80-year-old man. The tumor forms a well-circumscribed
dome-shaped nodule with ulceration. Photograph courtesy of
Dr. C Quirk, Dermatology Specialist Group, Ardross, Western
FIGURE 2. Anatomical location of the 171 AFXs. Ninety-one
percent were present on the head and neck, and 99% were
seen on sun-damaged skin.
534|www.amjdermatopathology.comq 2010 Lippincott Williams & Wilkins
Beer et al
Am J Dermatopathol?Volume 32, Number 6, August 2010
surrounding or overlying solar keratosis, and one showed
adjacent in situ squamous cell carcinoma. These cases were
reclassified as poorly differentiated (sarcomatoid) squamous
cell carcinomas. Two of these carcinomas contained osteo-
clast-like giant cells and have been described in detail in
a previous publication.3The review also revealed a metastatic
melanoma that had originally been diagnosed as AFX on
a partial biopsy that showed negative S100, HMB45, and
Melan A staining. However, the full excision sample and
subsequent further similar skin nodules revealed foci of
strong S100, Melan A, and HMB45 positivity. The discrep-
ancy seemed to have arisen due to biopsy sampling error in
a metastatic melanoma showing only patchy melanocytic
marker staining. Three tumors were reclassified as malignant
fibrous histiocytoma (MFH) because of extensive, deep
invasion of subcutis or skeletal muscle, large areas of necrosis,
and repeated recurrences.
Solar elastosis was assessable in 161 (94%) of the
cases and was present in 160 (99%). It was graded subjectively
as absent in 1 case, mild in 3, moderate in 7, and severe in 150.
In 10 tumors, it could not be adequately measured due to
insufficient surrounding dermis for examination.
The median size of the neoplasms was 10 mm diameter
(range 4–35 mm), with only 9 (5%) exceeding 20 mm in
diameter. The vast majority were well circumscribed, although
a few showed irregular, poorly defined, or infiltrative margins.
Invasion of subcutis was seen in 45 (26%) cases and was
generally only focal and of limited extent. In 13 tumors, there
was insufficient deep tissue to adequately assess possible
invasion of subcutis or deeper structures. A number of lesions
showed focal, early extension into underlying fascia (2 cases,
both on the scalp), and skeletal muscle (1 case, located on the
Ulceration was present in 77 (45%) AFX and was
typically focal and of only superficial depth. Necrosis was
uncommon and seen only in a small area of 6 tumors (3.5%).
Cytologically, most tumors showed highly atypical cells with
nuclear enlargement, pleomorphism, hyperchromasia, and
frequent mitoses (Fig. 3). Most were of spindle cell appearance
(137 cases, 80%), and a fascicular arrangement reminiscent
of smooth muscle was sometimes observed. Atypical mitotic
figures were common. The cytoplasm of the malignant cells
was mostly scanty, lightly eosinophilic, and poorly defined.
In 7 cases, cells of epithelioid appearance dominated, which
showed ovoid cells with larger amounts of pale cytoplasm,
sometimes with clearly defined cytoplasmic borders. In
27 cases, there was a mixed spindle and epithelioid cell
morphology. A number of tumors showed incidental foci
of hemosiderosis associated with adjacent hemorrhage.
Keratinization or identifiable intercellular bridges were never
features. Away from areas of ulceration, inflammatory cells
were present in a minority of AFX and were typically sited
toward the periphery of the lesions. These cells were pre-
dominantly small lymphocytes and macrophages. Mast cells
were scattered through all of the tumors in variable number,
being clearly delineated in the 51 cases where tryptase staining
Many neoplasms exhibited bizarre anaplastic cells,
which were sometimes multinucleated. Osteoclast-like giant
cells were also noted in 4 tumors from the original search
of the computer archive. However, 2 of these turned out to be
poorly differentiated sarcomatoid squamous cell carcinomas
on detailed review that had been misdiagnosed as AFX. These
cases have been highlighted in a previous publication.3In
brief, osteoclast-like giant cells in the 2 AFX were distin-
guished from neoplastic cells by their smaller nuclei, mono-
morphism, and lack of atypia or mitotic activity. They were
CD68 positiveand actin negative and were interpreted as fused
aggregates of reactive macrophages.
Five AFX showed a relatively bland cytomorphology
(Fig. 4; Table 2). Although some cellular atypia was always
present, pleomorphism and hyperchromasiawere more limited
than in the classic cases, with fewer mitoses seen. These cases
TABLE 1. Review of the Clinical, Histological, and Immunohistochemical Findings in the Original 183 AFXs Led to Reclassification
of 12 Cases
Revised Diagnosis No. CasesComment
Squamous cell carcinoma
Original partial excision was negative for S100, HMB45, and Melan A
Cytokeratin staining was patchy but more clearly seen in several cases with subsequent 34betaE12 staining
Reclassified due to extensive or deep tissue invasion, extensive necrosis, and repeated recurrences
Clinical, histological, or immunohistochemical features were equivocal, so firm diagnosis was not possible
FIGURE 3. The classic histological appearances of an AFX. The
lesion shows highly pleomorphic cells with prominent nuclear
atypia and scattered multinucleated forms. Many mitotic
figures are present (hematoxylin and eosin stain).
q 2010 Lippincott Williams & Wilkins
Am J Dermatopathol?Volume 32, Number 6, August 2010Atypical Fibroxanthoma
were all of spindle cell type with 4 located on the head or face
and 1 on the shoulder. The median patient age was 82 years,
with 2 male and 3 female patients. One of the bland tumors
showed areas of low cellularity and fibrosis, suggesting
regression. The pattern of immunostaining was similar to
that seen in the more usual cytological patterns of AFX.
The review identified 4 cases (2%) of the rare plaque-like
variant of AFX,4in which spindle-shaped tumor cells invade
as a superficial band in the upper dermis (Fig. 5).
A single myxoid pattern AFX was seen (Fig. 6) with
alcian blue–positive stromal mucin demonstrable. Three clear
cell AFX were identified. These showed copious amounts of
clear colorless cytoplasm with an epithelioid or mixed cyto-
logical morphology. Three granular cell AFX were observed
which showed finely granular eosinophilic cytoplasm in cells
of epithelioid or mixed pattern—these have been defined in
a previous publication.5Coarse, glassy, eosinophilic, keloid-
like collagen fibers were present in occasional AFX (Fig. 7). In
total, 15 tumors showed some keloidal fibers, but in 8 (4.7%),
they were sufficiently prominent to render classification as
keloidal AFX variants. In 22 AFX, there seemed to be focal
loss of tumor with replacement by poorly cellular fibrous
tissue containing a light lymphohistiocytic infiltrate. This did
not show cellular atypia or conspicuous mitotic activity and
suggested tumor regression (Fig. 8). In 5 of these cases, there
were some associated keloidal collagen fibers.
The background epidermis showed areas of solar
keratosis formation in some patients, with in situ squamous
cell carcinoma noted in a few cases. This did not show direct
continuity with the AFX. Overlying, but morphologically and
immunohistochemically distinct in situ melanoma was seen
in 1 case. Two cases showed AFX in collision with a basal cell
carcinoma and 3 cases in collision with a morphologically and
immunohistochemically distinct squamous cell carcinoma.
Staining for S100 protein was negative in the tumor cells
in all 171 cases but did highlight junctional melanocytes and
FIGURE 4. Bland cytological features were seen in 5 of the AFXs
in this series. In this zone, the features could easily be
misinterpreted as a benign lesion such as dermatofibroma.
Other areas of the tumor showed features more characteristic
of AFX (hematoxylin and eosin stain).
TABLE 2. Uncommon Variants or Features of AFX Identified
in the Study
Histological Variant or PatternNo. Cases % of All Tumors
Bland cytomorphology (spindle cell
Intratumoral osteoclast-like giant cells
FIGURE 5. Plaque-like AFX. This is a rare form of tumor, which
grows as a superficial zone of atypical spindle-shaped cells
confined to the upper dermis (hematoxylin and eosin stain).
FIGURE 6. A single case of the rare myxoid pattern AFX was
identified in the review, with alcian blue positivity detected in
the stroma. This appearance invokes consideration of entities
that might not normally be included in the differential
diagnosis of AFX (hematoxylin and eosin stain).
536|www.amjdermatopathology.comq 2010 Lippincott Williams & Wilkins
Beer et al
Am J Dermatopathol?Volume 32, Number 6, August 2010
intratumoral and surrounding Langerhans cells (Table 3).
The nature of the latter was confirmed by their bland dendritic
morphology and in 52 of the cases, by colocalized CD1a
staining. In 4 tumors, Langerhans cells were very prominent,
as demonstrated by diffuse or focal accentuated S100 and
CD1a staining of dendritic cells. In most cases, however,
Langerhans cells were sparse and predominantly located
toward the edge of the neoplasms. They were typically in-
creased in areas of inflammation. No tumors showed staining
with Melan A or HMB45, which were performed in 80 and
51 of the cases, respectively.
CD68 decorated the neoplastic cells of 98 of the 106
cases examined (92%). No staining was seen in 7 tumors, and
in 1 case, the staining was equivocal. Vimentin was stainable
in 48 of the 49 cases examined (98%), with positivity in both
the spindle and epithelioid cell variants. The single negative
case had a mixed spindle and epithelioid cytological pattern.
Frequent expression of smooth muscle actin was identi-
fied in the tumors, with 64 of 87 cases stainable (74%). In most
of these positively stained tumors, staining was strong and
diffuse, although it was only focal in 11 cases. None of the 73
AFX stained for desmin were positive.Epithelial markers were
negative in all cases and included broad spectrum cytokeratins
MNF116 (156 cases), AE1 and 3 (154 cases); high molecular
weight cytokeratin 34bE12 (13 cases), EMA (4 cases); and
Cam 5.2 (5 cases). In a small number of AFX, care was
required to distinguish staining of reactive hyperplastic
epidermis or included distorted adnexal ductal structures
from tumor cell staining. Only 7 tumors were stained for
CD34, which was negative in every case.
CD117 was stainable in the neoplastic cells in 1 of 50
cases examined (2%), as detailed in a previous study.6This
positive case had no other features clinically, histologically, or
immunohistochemically to distinguish it from the negatively
stained tumors. CD117 did, however, reveal variable numbers
of mast cells in the tumors, the nature of which was confirmed
by positive staining for mast cell tryptase in all cases.
Ten AFX were stained for CD10, which was diffusely
and strongly positive in 9 cases. One case showed only limited,
patchy staining of tumor cells.
AFX is an uncommon dermal neoplasm that exhibits
a malignant histological appearance, but an indolent clinical
course.1The cell of origin is disputed, with results from
various electron microscopic and immunohistochemical
studies suggesting an undifferentiated mesenchymal cell that
can show features of histiocytic, fibroblastic, or myofibro-
blastic lineage.7–10One view is that AFX merely represents
a superficial variant of MFH. However, the concept of MFH
is disputed, with most examples considered to be anaplastic
forms of various sarcomas, carcinomas, melanoma, or even
lymphoma.11–17Features traditionally used to separate MFH
from AFX include deeper invasion, larger size, frequent
necrosis, and, sometimes, metastasis in MFH. A number of
investigators have sought specific markers to help determine if
AFX and MFH are analogous lesions, with varying results.
FIGURE 7. Keloid-type collagen fibers were occasionally iden-
tified in the AFXs. In 8 cases, they were prominent, indicating
the rare keloidal pattern AFX (hematoxylin and eosin stain).
FIGURE 8. In 22 AFXs, there were foci of poorly cellular fibrous
tissue without cytological atypia replacing part of the tumor,
suggesting regression. In this case, a well-defined border is
seen between an area of cellular tumor and an underlying zone
where the tumor has regressed (hematoxylin and eosin stain).
q 2010 Lippincott Williams & Wilkins
Am J Dermatopathol?Volume 32, Number 6, August 2010Atypical Fibroxanthoma
Evidence to support them being different entities includes
differential expression of H and K ras,18MIB1,19CD74,20and
ploidy differences.19,21Similarities have also been claimed,
with comparable AgNOR numbers,22p53 and bcl2 expres-
sions, and apoptotic and proliferation markers.23
This report represents the largest series of AFX reported
in the literature to date. All cases were reviewed histologically
with the diagnosis supported by a tailored panel of immuno-
histochemistry. A number of previous case series, including
the review of 140 cases by Fretzin and Helwig in 1973, were
performedbefore the routineapplication of immunostaining or
used only limited, unstipulated, or low specificity immuno-
histochemical reagents.24–27It is, therefore, not possible to be
certain of the diagnosis in many of the cases in these reports,
so their conclusions need to be treated with caution.
In this series, AFX showed a strong male predilection
(male to female ratio 3:1) and was mostly seen in older patients
(median age 74). The Western Australian population is
exposed to high levels of solar radiation and shows very high
rates of ultraviolet light–associated skin cancers such as basal
cell carcinoma and melanoma.28,29AFX seems to occur pre-
dominantly on sun-damaged skin, with solar elastosis present
in 99% of the patients in this study where dermis could be
adequately assessed. Reflecting this, the principle sites of oc-
currence are areas likely to be frequently exposed to sunlight.
There was a particular predilection for the head and neck
(91% of cases), especially the ears. All 33 cases diagnosed on
the ears were in men, suggesting that sun protection of the
ears by long hair in females may be of importance.
Prior trauma was noted at the site of 1 AFX in this study
(from a wood splinter on the forearm, 3 months before), but
physicians and patients were not specifically asked about
any history of trauma, so the frequency of this possible asso-
ciation is unknown. Evidence of a link to prior trauma has
been previously suggested in the literature.30Theoretically,
this prospect may be feasible, because the trauma repair
processes lead to a proliferation of the type of cells implicated
as being likely cells of origin of AFX.
Some early studies suggested that there was a subset of
AFX that occurred on the limbs of younger patients.24–27This
is now disputed, and it is likely that those tumors were other
entities such as atypical fibrous histiocytoma.31In the current
review, the limbs were uncommon sites for AFX, and the
few tumors that did occur at those sites did not show any
predilection for younger patients.
Multiple AFX were uncommon, with just 2 patients
developing separate metachronous tumors. AFX has a very
low risk of recurrence at the site of excision or metastasis, with
no metastases seen in this cohort, and only 2 local recurrences
(a 1% recurrence rate). In one of the recurrent cases, the
original tumor was known to have been incompletely excised,
and in the second case, the lesion was only 0.5 mm from the
deep excision margin. The 88 patients reviewed by Mirza and
Weedon4reportedly showed no true recurrences or metastases.
The 25-year experience of treating 91 patients at the Rochester
Mayo clinic reported recurrence in 2 cases treated with wide
local excision and no recurrences in patients treated by Mohs
surgery (n = 59) or by other means (n = 6).32However, one of
the recurrent tumors was noted to be incompletely excised
originally and the other 3.0 mm from the excision margins.
Several publications have suggested that Mohs surgery should
be performed to treat AFX.32–34However, because of the
minimal risk of recurrence, we do not believe that Mohs sur-
gery is necessary for the treatment of AFX. Nevertheless, we
would suggest that in rare instances, where tumors are incom-
pletely excised or only very marginally excised, reexcision
with a small rim of normal tissue should be considered.
Histologically, the majority of AFX were circumscribed
dermal neoplasms with minimal involvement of subcutis or
deeper tissues. Although ulceration was frequent (seen in 45%),
necrosis was rare and never more than microscopic in extent.
Awareness of the uncommon histological variants
observed in this study is important to avoid misdiagnosis.
Five tumors showed spindle-shaped cells of relatively bland
cytomorphology, with no clinical or immunohistochemical
features to distinguish these lesions from more usual forms of
AFX. This tumor variant has only rarely been described pre-
viously under the title ‘‘spindle cell nonpleomorphic AFX.’’35
As expected, the plaque-like pattern of AFX4was infrequent in
this series, being present in just 4 cases. Clear cell and granular
cell tumors were uncommon (3 cases of each), but may lead to
contemplation of alternative diagnoses to those traditionally
considered with AFX.36,37Similarly, the rare occurrence of
widespread myxoid change in AFX (seen in only 1 patient
in this series) demands consideration of a different spectrum
of lesions to those typically entertained. The myxoid pattern of
AFX has only recently been described in the literature in
a series of 4 cases.38
Osteoclast-like giant cells seem to be infrequent in AFX
(2 cases in this series) and must be distinguished from tumor
giant cells, which show larger, more atypical, and pleomorphic
nuclei. They were CD68 positive and seem to represent fused
macrophages. Isolated cases of osteoclast-like giant cells in
AFX have been reported over the years, but supportive evi-
dence that the tumors are truly examples of AFX is sometimes
limited.39–45Interestingly, 2 neoplasms originally retrieved
from our archive as examples of AFX with osteoclast-like
TABLE 3. Selected Immunohistochemical Findings in the
Series of 171 AFXs
AntibodiesNo. Cases StainedNumber Positive % Positive
AE1 and 3*
Smooth muscle actin
*Broad spectrum cytokeratins.
†High molecular weight cytokeratin.
538|www.amjdermatopathology.comq 2010 Lippincott Williams & Wilkins
Beer et al
Am J Dermatopathol?Volume 32, Number 6, August 2010
giant cells were subsequently proven to be squamous cell
carcinomas after further scrutiny and immunohistochemistry.
Keloidal change is an unusual finding in AFX, which was only
very recently recognised in the literature, in a series of 9
cases.46Fibrous tissue apparently replacing zones of 22 AFX
suggested areas of regression. This concept has received little
attention in the literature but could be an important feature,
possibly leading to tumor involution. Some of the tumors
with regression also demonstrated keloidal change. Recently,
regression was described in 7 AFX in abstract form47but was
alluded to earlier in a single case of highly sclerotic AFX.48
The concept of regression in AFX deserves further attention
and may partly explain the generally small size and indolent
behavior of these tumors, despite their alarming histological
Perineural or intraneural invasion appear to be rare
events in AFX, if indeed they occur at all. Two cases were
reported in 2006, but both of these tumors also showed
invasion of subcutis and deep fascia, suggesting that they may
in fact represent forms of MFH or other poorly differentiated
tumors.49No lymphovascular invasion was seen in our cohort.
Immunohistochemistry is considered to be essential for
the accurate identification of AFX.1Identical histological
appearances can be associated with other tumors that include
amelanotic melanoma, poorly differentiated squamous cell
carcinoma and leiomyosarcoma. Table 4 shows a suggested
immunohistochemical panel for the differentiation of AFX
from its mimics (Table 4). Although the neoplastic cells in
AFX were always negative for S100 and other melanocytic
markers, S100 positive Langerhans cells were often present in
the tumors, occasionally in large numbers. This has been little
recognised in the literature but can lead to misdiagnosis as
melanoma.50,51CD1a can be useful to identify the true nature
of the Langerhans cells in AFX.
Frequent expression of actin in this series may givesome
support to the notion that at least some AFX are derived from
myofibroblasts. AFX remains largely a diagnosis of exclusion,
and the majority of markers used to confirm the diagnosis lack
specificity. We would not accept expression of cytokeratin
as being compatible with a diagnosis of AFX. If specific
neoplastic cell–localized staining is identified, this should be
confirmed with additional epithelial markers, and such a lesion
should be classified as a carcinoma. Previous suggestions that
all AFX are forms of squamous cell carcinoma can no longer
be substantiated.52It is highly likely that many lesions
previously diagnosed as AFX have included a mixture of
different tumors, including AFX, carcinomas, and other
neoplasms. This again emphasises the critical need for a panel
of carefully performed and evaluated immunohistochemistry.
Use of the high molecular weight cytokeratin 34bE12 has
been suggested as a helpful addition to discern some poorly
differentiated squamous cell carcinomas.53Our experience in
this study and in day to day work supports this concept.
Similarly, p63 has been shown to be a sensitive marker for
distinguishing poorly differentiated squamous cell carcinomas
from AFX,54–56although in one study, 3 of 23 AFX showed
focal weak nuclear staining.57
The identification of CD117 staining in AFX has
been a source of confusion in a number of previous reports,
which suggested that AFX are CD117-positive tumors. This
phenomenon has been discussed in detail in a recent publi-
cation,6which indicated that the frequent presence of posi-
tively stained cells in AFX was due to included mast cell,
rather than neoplastic cell staining. CD117 was only positive
in the neoplastic cells of 1 of the 50 cases stained in this series.
AFXs are uncommon skin neoplasms that typically
occur on sun-damaged skin of elderly patients. They form
a definable pathological entity with fairly consistent clinical
and histological features but require a panel of immunohis-
tochemical stains for definitive diagnosis. Care must be taken
to exclude histological mimics, especially poorly differentiated
squamous cell carcinoma. Specific diagnosis is important
because there seems to be a very low risk of recurrence or
metastasis despite the frequently alarming pathological
appearances. Regression may occur and could be an important
reason for the indolent behavior of these elusive neoplasms.
TABLE 4. A Suggested Immunohistochemical Panel for the Diagnosis of AFX
Antibodies Expected ReactionComments
Broad spectrum cytokeratin
(e.g. MNF116, AE1 & 3)
Smooth muscle actin
Essential to exclude melanoma. Also highlights Langerhans cells, which may be prominent
Optional, if S100 is negative
Optional, if S100 is negative
NegativeBeware of included normal adnexae or hyperplastic epidermal downgrowths
Highlights some squamous cell carcinomas more prominently than broad spectrum cytokeratins
74% positive in this series
Useful to exclude leiomyosarcoma in actin positive cases
92% positive in this series
Most AFX are positive, but the specificity is low, making it of limited discriminatory value
q 2010 Lippincott Williams & Wilkins
Am J Dermatopathol?Volume 32, Number 6, August 2010 Atypical Fibroxanthoma
REFERENCES Download full-text
1. Beer TW, Lam M, Heenan PJ. Fibrous tissue tumors of the skin. In: Elder
DE. Lever’s Histopathology of the Skin. 10th ed. Philadelphia, PA:
Lippincott Williams and Wilkins; 2009:978–979.
2. Weiss SW, Goldblum JR. Atypical fibroxanthoma. In: Enzinger and
Weiss’s Soft Tissue Tumors. 5th ed. Mosby; 2008:403–406.
3. Beer TW. Squamous cell carcinoma with osteoclast-like giant cells: 2 cases
misdiagnosed as atypical fibroxanthoma. Am J Dermatopathol. In press.
4. Mirza B, Weedon D. Atypical fibroxanthoma: a clinicopathological study
of 89 cases. Australas J Dermatol. 2005;46:235–238.
5. Beer TW. Keloidal and granular cell change in a series of 171 atypical
fibroxanthomas. J Cutan Pathol. In press.
6. Beer TW, Haig D. CD117 is not a useful marker in atypical fibro-
xanthoma. Am J Dermatopathol. 2009;31:649–652.
7. Alguacil-Garcia A, Unni KK, Goellner JR, et al. Atypical fibroxanthoma
of the skin: an ultrastructural study of two cases. Cancer. 1977;40:1471–
8. Barr RJ, Wuerker RB, Graham JH. Ultrastructure of atypical fibrox-
anthoma. Cancer. 1977;40:736–743.
9. Longacre TA, Smoller BR, Rouse RU. Atypical fibroxanthoma: multiple
immunohistologic profiles. Am J Surg Pathol. 1993;17:1199–1209.
10. Ma CK, Zarbo RJ, Gown AM. Immunohistochemical characterization of
atypical fibroxanthoma and dermatofibrosarcoma protuberans. Am J Clin
11. Fletcher CDM. Pleomorphic malignant fibrous histiocytoma: fact or
fiction? A critical reappraisal based on 159 tumors diagnosed as
pleomorphic sarcoma. Am J Surg Pathol. 1992;16:213–228.
12. Fletcher CDM.The evolving classificationof softtissue tumours: an update
based on the new WHO classification. Histopathology. 2006;48:3–12.
13. Hollowood K, Fletcher CDM. Malignant fibrous histiocytoma: morpho-
logic pattern or pathologic entity? Semin Diagn Pathol. 1995;12:210–220.
14. Nakayama R, Nemoto T, Takahashi H, et al. Gene expression analysis of
soft tissue sarcomas: characterization and reclassification of malignant
fibrous histiocytoma. Mod Pathol. 2007;20:749–759.
15. Pezzi CM, Rawlings MS Jr, Esgro JJ, et al. Prognostic factors in 227
patients with malignant fibrous histiocytoma. Cancer. 1992;69:2098–2103.
16. Weiss SW. Malignant fibrous histiocytoma. Am J Surg Pathol. 1982;6:
17. Weiss SW, Enzinger FM. Malignant fibrous histiocytoma: an analysis of
200 cases. Cancer. 1978;41:2250–2266.
18. Sakamoto A, Oda Y, Itakura E, et al. H-, K-, and N-ras gene mutation in
atypical fibroxanthoma and malignant fibrous histiocytoma. Hum Pathol.
19. Oshiro Y, Fukuda T, Tsuneyoshi M. Atypical fibroxanthoma versus
benign and malignant fibrous histiocytoma. A comparative study of their
proliferative activity using MIB-1, DNA flow cytometry, and p53
immunostaining. Cancer. 1995;75:1128–1134.
20. Lazova R, Moynes R, May D, et al. LN-2 (CD74). A marker to distinguish
atypical fibroxanthoma from malignant fibrous histiocytoma. Cancer.
21. Worrell JT, Ansari MQ, Ansari SJ, et al. Atypical fibroxanthoma: DNA
ploidy analysis of 14 cases with possible histogenetic implications.
J Cutan Pathol. 1993;20:211–215.
22. Merot Y, Durgniat AC, Frenk E. Nucleolar organizer regions in
fibrohistiocytic tumors of the skin. J Cutan Pathol. 1990;17:122–126.
23. Westermann FN, Langlois NE, Simpson JG. Apoptosis in atypical
fibroxanthoma and pleomorphic malignant fibrous histiocytoma. Am J
24. Dahl I. Atypical fibroxanthoma of the skin. A clinico-pathological study
of 57 cases. Acta Pathol Microbiol Scand [A]. 1976;84:183–197.
25. Evans HL, Smith JL. Spindle cell squamous carcinomas and sarcoma-like
tumors of the skin: a comparative study of 38 cases. Cancer. 1980;45:
26. Fretzin DF, Helwig EB. Atypical fibroxanthoma of the skin. A
clinicopathologic study of 140 cases. Cancer. 1973;31:1541–1552.
27. Leong AS, Milios J. Atypical fibroxanthoma of the skin: a clinicopath-
ological and immunohistochemical study and a discussion of its
histogenesis. Histopathology. 1987;11:463–475.
28. English DR, Kricker A, Heenan PJ, et al. Incidence of non-melanocytic
skin cancer in Geraldton, Western Australia. Int J Cancer. 1997;73:
29. Threlfall TJ, Thompson JR. Cancer Incidence and Mortality in Western
Australia, 2005. Department of Health, Western Australia, Perth,
Australia; 2007. Statistical Series Number 81.
30. Starink TH, Hausman R, Van Delden L, et al. Atypical fibroxanthoma
of the skin. Presentation of 5 cases and a review of the literature. Br J
31. Weiss SW, Goldblum JR. Malignant fibrous histiocytoma. In: Enzinger
and Weiss’s Soft Tissue Tumors. 5th ed. Mosby; 2008:406–425.
32. Ang GC, Roenigk RK, Otley CC, et al. More than 2 decades of treating
atypical fibroxanthoma at Mayo clinic: what have we learned from 91
patients? Dermatol Surg. 2009;35:765–772.
33. Davis JL, Randle HW, Zalla MJ, et al. A comparison of Mohs
micrographic surgery and wide excision for the treatment of atypical
fibroxanthoma. Dermatol Surg. 1997;23:105–110.
34. Seavolt M, McCall M. Atypical fibroxanthoma: review of the literature
and summary of 13 patients treated with mohs micrographic surgery.
Dermatol Surg. 2006;32:435–441.
35. Calonje E, Wadden C, Wilson-Jones E, et al. Spindle-cell non-
pleomorphic atypical fibroxanthoma: analysis of a series and delineation
of a distinctive variant. Histopathology. 1993;22:247–254.
36. Kemmerling R, Dietze O, Mu ¨ller S, et al. Aspects of the differential diag-
nosis of clear-cell lesions of the skin in connection with the rare case of
a clear-cell atypical fibroxanthoma. Pathol Res Pract. 2009;205:365–370.
37. Wright NA, Thomas CG, Calame A, et al. Granular cell atypical
fibroxanthoma: case report and review of the literature. J Cutan Pathol.
38. Patton A, Page R, Googe P, et al. Myxoid atypical fibroxanthoma:
39. Ferrara N, Baldi G, Di Marino MP, et al. Atypical fibroxanthoma with
osteoclast-like multinucleated giant cells. In Vivo. 2000;14:105–107.
40. Khan ZM, Cockerell CJ. Atypical fibroxanthoma with osteoclast-like
multinucleated giant cells. Am J Dermatopathol. 1997;19:174–179.
41. Orlandi A, Bianchi L, Ferlosio A, et al. The origin of osteoclast-like giant
cells in atypical fibroxanthoma. Histopathology. 2003;42:407–410.
with osteoclast-like giant cells. Am J Surg Pathol. 1997;21:213–218.
43. Val-Bernal JF, Corral J, Ferna ´ndez F, et al. Atypical fibroxanthoma with
osteoclast-like giant cells. Acta Derm Venereol. 1994;74:467–470.
44. Wilson PR, Strutton GM, Stewart MR. Atypical fibroxanthoma: two
unusual variants. J Cutan Pathol. 1989;16:93–98.
45. Zelger BG, Soyer HP, Zelger B. Giant cell atypical fibroxanthoma: does it
really exist? Am J Dermatopathol. 1999;21:108–110.
46. Kim J,McNiff JM. Keloidal atypical fibroxanthoma: a case series. J Cutan
47. Stefanato C, Robson A, Calonje E. The histopathological spectrum of
atypical fibroxanthoma with regression. J Cutan Pathol. 2009;36:166.
48. Bruecks AK, Medlicott SA, Trotter MJ. Atypical fibroxanthoma with
prominent sclerosis. J Cutan Pathol. 2003;30:336–339.
49. Dettrick A, Strutton G. Atypical fibroxanthoma with perineural or intra-
neural invasion: report of two cases. J Cutan Pathol. 2006;33:318–322.
50. McKee PH, Calonje E, Granter SR. Fibrohistiocytic tumors. In: Pathology
of the Skin. 3rd ed. Philadelphia, PA: Elsevier Mosby; 2005:1757–1760.
51. Ricci A Jr, Cartun RW, Zakowski MF. Atypical fibroxanthoma. A study
of 14 cases emphasizing the presence of Langerhans’ histiocytes with
implications for differential diagnosis by antibody panels. Am J Surg
52. Zelger B, Soyer HP. Between Scylla and Charybdis: mythology in
dermatopathology. Dermatopathol Pract Concept. 2000;6:348–355.
53. Gray Y, Robidoux HJ, Farrell DS, et al. Squamous cell carcinoma detected
by high-molecular-weight cytokeratin immunostaining mimicking atyp-
ical fibroxanthoma. Arch Pathol Lab Med. 2001;125:799–802.
54. Dotto JE, Glusac EJ. p63 is a useful marker for cutaneous spindle cell
squamous cell carcinoma. J Cutan Pathol. 2006;33:413–417.
55. Gleason BC, Calder KB, Cibull TL, et al. Utility of p63 in the differential
diagnosis of atypical fibroxanthoma and spindle cell squamous cell
carcinoma. J Cutan Pathol. 2009;36:543–547.
56. Greene LA, Cooper K. Atypical fibroxanthoma: an immunohistochem-
istry update. Adv Anat Pathol. 2008:15;374.
57. Hall JM, Saenger JS, Fadare O. Diagnostic utility of P63 and CD10 in
distinguishing cutaneous spindle cell/sarcomatoid squamous cell carcino-
mas and atypical fibroxanthomas. Int J Clin Exp Pathol. 2008;1:524–530.
540|www.amjdermatopathology.comq 2010 Lippincott Williams & Wilkins
Beer et al
Am J Dermatopathol?Volume 32, Number 6, August 2010