DOI: 10.1111/exd.
12880
www.wileyonlinelibrary.com/journal/EXD
Keratoacanthoma: a distinct entity?
Tobias Gleich, Elena Chiticariu, Marcel Huber and Daniel Hohl
Service of Dermatology, University Hospital Center and University of Lausanne, Lausanne, Switzerland
Correspondence: Daniel Hohl, CHUV, Service de Dermatologie, 1011 Lausanne, Switzerland, Tel.: +41 21 3140353, Fax: +41-21-314 0382,
e-mail: daniel.hohl@chuv.ch
Abstract: Keratoacanthoma (KA) are common but exceptional
benign tumors, often appearing on sun-exposed areas of light
skinned people and showing spontaneous resolution. The goal of
this study was to review existing literature, to point out the
etiological complexity of KA biology and to answer the
controversial debate if or not KA is a distinct entity or a variant
of squamous cell carcinoma (SCC). Relying on recent results, we
highlight that KA is an individual lesion with a unique molecular
Introduction
The first description of the ‘crateriform ulcer of the face’ was
given by Jonathan Hutchinson in 1889 (1), and since then many
different synonyms like molluscum sebaceum, self healing-epithe-
lioma and keratocarcinoma have been used (2–4). In the 19400 s
Freudenthal proposed the name Keratoacanthoma (KA) to high-
light acanthosis and keratosis as major histological changes of the
tumor (4–7). KA however is a benign tumor, characterized by
rapid growth over 4–5 weeks followed by spontaneous regression
(5,8). KA occurs mainly as a solitary lesion on sun-exposed loca-
tions (9–11), but multiple tumors appear in several syndromes
such as familial self-healing epithelioma of Ferguson-Smith, Xero-
derma pigmentosum or Muir-Torre syndrome (12,13). Morpho-
logically similar to squamous cell carcinoma (SCC), a better
biological understanding of KA as defined as a self-healing benign
tumor is needed to develop reliable diagnostic tools. Since recent
findings showed that altered TGFb signalling is the main cause of
familial KA, a detailed view on this pathway should allow new
insights.
Etiology
Sun exposure and thus UV light was known for a long time to be
necessary for KA development (14) as suggested by preferential
localization on sun exposed skin of elderly patients with light
complexions, an increased appearance in summer and autumn,
and their frequent emergence in Xeroderma pigmentosum
(10,11,15–19). Solitary and multiple KA have also been observed
in PUVA-treated patients (20,21).
Similarly, KA emergence was reported after cutaneous X-ray or
Megavolt radiation therapy (13,22,23). Further etiological factors
are immune-compromised or immune-suppressed patients (24),
skin trauma when pretreated with chemicals (18), carbondioxide
ablation therapy, likely by induction of a wound healing response
(13,22,23), and most importantly, genetic predisposition in multi-
ple KA of Ferguson Smith type (25). Drugs as Imiquimod and
BRAF-inhibitors induce KA as a side effect (26–31). Contradictory
results were published on the implication of human papilloma
viruses (HPV) (5,32–34).
ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Experimental Dermatology, 2016, 25, 85–91
Review
signature caused by alterations in the TGFb signalling pathway.
These recent findings will help to understand the nature of KA
and to develop new reliable diagnostic tools, simplifying the
discrimination of the histologically similar KA and SCC.
Key words: keratoacanthoma – paradoxical activation – squamous cell
carcinoma – TGFb – tumor
Accepted for publication 12 October 2015
Classification
Several forms of KA exist, with solitary KA as the most common
variant. Multiple KA have been described by Ferguson Smith,
Grzybowski and Witten-Zak.
Solitary KA
1. The typical solitary KA is rapidly growing up to 2 cm,
appearing as a rose nodule with a stretched shiny epidermis
and a keratotic centre. After this mature phase with a dura-
tion around 6 months, the lesions becomes smaller until
complete regression (5).
2. Giant KA, usually grow larger than 2 cm, predominantly
appears on eyelids and nose (35–37). Unusual vertical
growth invading dermal and underlying tissue is observed
(38–40).
3. Subungual KA is a rare KA variant, appearing under nails,
show rarely spontaneous regression and may destroy the
terminal phalanx (10,41,42).
4. Mucosal KA another extremely rare KA variant, has also no
tendency to regress (43) and is observed mostly intraorally
(44). Since intraoral mucous membranes bear no follicles it
was suggested, that these KA may arise from ectopic seba-
ceous glands (45).
5. Keratoacanthoma centrifugum is a rare exoendophytic type,
characterized by multifollicular origin, progressive periph-
eral expansion (2–40 cm) and atrophic central healing (46–
48). Resolving of the tumor occurs spontaneously and invo-
lution is mostly complete within week 12 (5,49).
Multiple KA
1. Ferguson Smith type, known as multiple self-healing squa-
mous epithelioma (MSSE), is the most common form of
multiple KA. This autosomal dominant genodermatosis is
characterized by sudden appearance and rapid growth, slow
regression and periodical recurrence (25,50). The number
of tumors varies from a few to hundreds, mostly leaving
deep scars after involution. With a mean onset in young
adolescence, appearance of KA is much earlier than the typ-
ically reported in sporadic cases (10). In 2011, Goudie et al.
85
 Gleich et al.
(51) identified 11 monoallelic mutations in transforming
growth factor beta receptor 1 (Tgfbr1) revealing the cause
of the MSSE phenotype.
2. Generalized eruptive keratoacanthomas of Grzybowski
appear spontaneously in 6–8 months represented by hun-
dreds to thousands follicular papules (52,53). Skin involve-
ment may be severe with masked facies and ectropion.
There is neither preference in sex nor in a familial pattern
(52,54). Fusing tumor nodules can form larger lesions
rather involving the mucosa than palms and soles. The
tumors may be indistinguishable from solitary KA (52).
3. Multiple KA of Witten and Zak type arise in a range of
1–30 papules which stay either small or increase their size
enormously. The lesions are showing central healing and
the life cycle with resolution is completed after several
months. A familial predisposition is assumed as well as sun-
light as an etiological factor (55).
4. KA in Muir-Torre syndrome appear solitary or multiple
(10). Muir-Torre syndrome is an autosomal dominant geno-
dermatosis combining visceral malignancy, notably colon
cancer with sebaceous neoplasms (56). The syndrome is
caused by germline mutations of hMSH2 and hMLH1
genes, involved in the mismatch repair system. KA appear-
ing within this syndrome is explained by the common
pilosebaceous gland origin of sebaceous neoplasms and KA
(10).
5. KA have been observed in patients suffering xeroderma pig-
mentosum, an autosomal recessive disorder with 100% pen-
etrance, caused by defects in genes coding for DNA repair
enzymes (57). Compared to normal cells, cells of XP
patients fail to repair UV-induced DNA lesions, thereby
linking sun exposure to a hypermutable genotype (58). This
failure of the repair system causes development of mela-
noma, basal cell carcinoma, SCC and KA in early childhood
on sun exposed areas of the body (15).
6. Non-familial multiple persistent KA have been observed on
conjunctivae, palms and soles. In one special case, the
tumor continued to grow over a time period of 35 years
(10). Affleck (59) described this distinct variant as sporadic,
idiopathic without spontaneous resolution.
7. Reactive KA is induced by treatment of melanoma patients
with the BRAF kinase inhibitors Vemurafenib or Sorafenib
(29,30). These patients often develop multiple disseminated
solitary KA on sun-exposed and non-sun-exposed areas
(60). The iatrogenic tumor formation usually occurs
8 weeks after inhibitor therapy initiation (29), by paradoxi-
cal activation of the ERK MAPkinase pathway (61).
Histopathology
Three different stages are determined by histological examination
during evolution of KA: (1) a proliferative phase, (2) the resting
fully developed mature stage and (3) regression (10). The majority
of tumors arises over the skin surface (62) embedded in a well-
differentiated squamous cell epithelium (63,64). A keratin-filled
epidermal invagination, eventually reaching into the dermis,
emerges from an adjacent hair follicle (65) (13,66). Atypical squa-
mous cells with multiple mitotic figures have been observed in
dermal extensions (13).
86
Proliferation
The initial, rapidly proliferating immature KA appears as a skin-
collared papule with an acanthotic hyperproliferating epidermis
characterized by hypergranulosis with large keratohyaline granules
and hyperkeratosis with premature keratinization. Enlarged ker-
atinocytes are noticed within the stratum basale and spinosum
(4). Eosin staining highlights keratinized areas with a pale cyto-
plasm giving the impression of a ‘glassy’ appearance (13,66). Later
in proliferation, a central depression and inflammatory infiltrates
consisting primarily of neutrophils and less eosinophils are present
(4,10). Perineural invasion is rarely observed. Invasion beyond the
level of eccrine glands is an ominous sign since KA usually not
extend deeper than sweat glands (10,67).
Maturity
The typical prominent dome shaped symmetric nodule with a ker-
atin plug in its centre is the prerequisite for correct diagnosis. Epi-
dermal lips rise around both sides and cover partially the top of
the crater (Fig. 1). Thereby, both elongated thin tumor lips are
covered on the surface and towards the crater by two epidermal
sheets which are in parallel with the skin surface. The keratin filled
crater plonges into multilocular spaces of epithelial tumor lobules
in the periphery in a semi-star shaped fashion (68,69). These are
composed of well-differentiated squamous cells at the dermal
interface (9). Thereby the tumor does not appear to grow verti-
cally into the dermis like SCC but into all directions below the
skin surface. Atypical basal keratinocytes are decreased in number
as are keratinocytes with pale and glassy cytoplasm (13). However,
apoptotic keratinocytes and intraepithelial abscesses primarily
composed of neutrophils and less eosinophils become prominent
(9,10). Horn pearls arranged in concentric keratinocyte layers with
a central eosinophilic keratin plug, also observed in SCC, are com-
mon (4,10,13). Increased dermal inflammatory infiltrates com-
posed of eosinophils, neutrophils, plasma cells and histiocytes are
observed (4,9).
(a) (b)
(c)
Figure 1. Keratoacanthoma in a proliferative to mature stage (a), in a regressing
stage (b) and a horn pearl (c). (a) A multilocular crater caused by the follicular
origin of the lesion and epithelial lips arising on both sites of the lesion are well
recognizable as well as inflammatory infiltrates of the dermis accumulating at the
borders to the epidermis. (b) Regressing keratoacanthoma (KA) maintaining the
typical architecture like multilocular crater and epithelial lips but the tumor
decreased in size and height. Dermal inflammation is reduced but still present in
this stage. (c) Horn pearl with keratotic plug, including eosinophilic cells.
ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Experimental Dermatology, 2016, 25, 85–91
 Keratoacanthoma

Regression the conclusion that KA is a benign neoplasm (8,51,85). Subungual

The third stage occurs relatively slowly. Most of the lesions invo- and mucosal KA appear under the global name of KA and arise
lute completely by week 12 (5). Involution is initiated by disap- on non-follicle bearing parts of the body. These borderline tumors
pearance of mitotic activity at the border of the lesion (4) and are histologically different from KA and SCC and show regression
loss of glassy, enlarged squamous cells. KA retain the crateriform only in a few cases (Table 3). Even if subungual KA show rather
architecture but decrease in height, followed by a reduction in progressive growth than spontaneous involution, cellular atypia
horizontal size (70) (Fig. 1). Around the tumor on the dermoepi- and the mitotic index are reduced compared to SCC (88,89) and
dermal interface a lichenoid inflammation reaction may be seen. several authors consider it to be benign (41,90,91).
The regression is likely based on immunological mechanisms
exhibited among others by activated CD4+ T lymphocytes (71).
Mitotic fibroblasts beneath the lesion forcing fibrosis and a dermal
Table 2. Table of differences of keratoacanthoma (KA) and squamous cell
infiltrate with multinucleated histiocytes may be observed (10). carcinoma (SCC)
Recent finding suggested that increased differentiation and loss of
KA SCC
proliferating epithelial cells together with inhibition of the Wnt-b-
catenin pathway are key steps in KA regression (72). Similarly, Biological differences
Rapid growth till 1–2 cm Slowly growing
SOX2 deletion in SCC led to decreased proliferation and increased Involution No involution
apoptosis and consequently to tumor regression in mice (73). Exoendophytic Predominantly endophytic
Histologpathologic differences
After complete resolution, the tumor leaves a fibrous scar without Epithelial lips –
adnexal structures (74). – Stromal desmoplasia
Discrimination: KA versus SCC Intraepithelial abscesses in the lesion Rarely observed
Rare ulceration Common
Classification controversy Distinct edge between Indistinct
Clinical and morphological distinction between KA and SCC may tumor and stroma
– Anaplasia
cause difficulties in certain cases (Tables 1, 2 and 4). Also the Intraepithelial abscesses with No association between
question whether KA is a distinct entity (75–77) or a variant of acantholytic cells eosinophils and acantholytic cells
Flask shaped –
SCC (43,78) was discussed controversially until molecular identifi- Epithelial collarette Rarely observed
cation of TGFbR1 mutations in Ferguson Smith syndrome and Symmetric Asymmetric
– Melanocytes present
genetic studies clarified that at least familial KA have bona fide an Absence of stromal desmoplasia Present
own pathogenetic cause (51,77,79,80). Indeed, the mutation rate
of TGFBR1 for supposed MSSE is about 85–90% (B. Lane, per-
sonal communication). A recent study analysing a large number
Table 3. Table of histopathological differences between solitary
of SCC by exome and targeted sequencing reported a comparable keratoacanthoma (KA) and its variants
frequency of ‘gatekeeper’ mutations in NOTCH1/2 (82%) while
Differences to
TGFBR1 mutations were much less frequent (81). Further studies corresponding Similarities to
are needed to confirm that this is a consistent fundamental dis- Variants Differences from squamous cell corresponding
of KA solitary KA carcinoma (SCC) SCC
tinction between all KA and SCC. Furthermore, the topic if KA is
essentially benign is lively debated and was forced by occasionally Ferguson • Deep
Smith extension
malignant behaviour of KA (82). If KA are truly able to form • Not
metastasis is still a matter of debate, in contrast with SCC which crateriform
metastasize in up to 5% (83,84). As reported, these metastasizing • No glassy
keratinocytes
KA could have been misdiagnosed as KA but are true SCC
Grzybowski • No differences
(5,8,69,76,78,85–87). In fact 10% as KA diagnosed tumors are
malignant and progressing SCC (9,69). Hallmarks of malignant Giant • Dimension
tumors such as, prominent mitoses or cytological atypia, are not Mucosal • No involution • Rare ulceration • No regression
observed in KA. Spontaneous regression and missing features • Rarely • Mitotic activity
crateriform • Low metastatic
characterizing a malignant phenotype, prompted us and others to • No fibrosis potential

KA • Morphology
centrifugum • Size
Table 1. Table of common features of keratoacanthoma (KA) and squamous marginatum
cell carcinoma (SCC) Subungual • No involution • Age of • Painful nodular
• Dyskeratotic appearance lesion on distal
Common biological features cells • KA evolves phalanx
Lateral growth predominant • No fibrosis rapidly, • Endo-exophytic
Metastatic potential • Little or no gathering • Radiometrically
Local tissue destruction (if untreated) atypia a tumor-like very similar
Similar etiologic factors (UV exposure) configuration
Common histopatological features • KA can regress
Intraepithelial polymorphonuclear abscesses • No increase
Parakeratosis of Ki-67 or
Dyskeratosis p53 expression
Cup shaped
Atypical squamous cells
Glassy keratinocytes Comparison of the rare subungual and mucosal KA with its SCC counterparts.

ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Experimental Dermatology, 2016, 25, 85–91 87
 Gleich et al.

In sum, the recently found TGFbR1 driver mutations in familial Proteins functioning in cellular adhesion processes
KA, Comparative Genomic Hybridization studies, micro array Differences in VCAM1 and ICAM1 expression in KA compared to
experiments and missing clinical features of a malignant pheno- SCC were not sufficient for a reliable marker (98). Significantly
type in KA are in line with our view that KA and SCC are two increased expression of stromelysin (ST3) was found in SCC and
distinct lesions (51,77,79,80). Unfortunately, there is still a notable even more in metastatic SCC allowing to predict their biological
lack of a reliable molecular marker, allowing for the unequivocal behaviour (99).
distinction of KA from SCC in the laboratory. Cell surface receptors
Histological criteria EGFR is a receptor tyrosine kinase affecting epithelial and ker-
The requirement to distinguish KA from SCC is complete excision atinocyte differentiation, proliferation and tumorgenesis (100).
of the lesion reaching to the subcutis including the crater, both Canonical EGFR signalling occurs through mitogen activated
lips and the surrounding of uninvolved epidermis (10). KA are kinase pathway (MAPK), and thus through Ras and Raf proteins.
exoendophytic whereas the majority of cutaneous SCC show inte- Comparative genetic hybridization identified different chromoso-
rior proliferation (13). For reliable diagnosis, dermatolopatholo- mal aberration patterns in KA compared with SCC providing evi-
gists have to include several criteria. Architectural attributes are dence that KA and SCC are two distinct entities (80).
essential since both tumors possess nearly identical cytological fea- Consequently, EGFR copy numbers were altered more often in
tures (Tables 1–3) (10). SCC but amplification was not significant, and no increased EGFR
Cribier et al. examined 14 criteria on 262 previously diagnosed protein expression could be observed (101–103). However, this is
samples and defined five significant criteria: (1) epithelial lips, controversial, as increased expression of EGFR was detected in the
(2) sharp outline between tumor and stroma favouring KA- (3), majority of primary and metastatic SCC but not in KA (100,104).
and ulceration, (4) abundant mitotic cells, (5) pleomorphism/ This may indicate that EGFR expression in skin carcinogenesis is
anaplasia SCC-favouring. These features allowed a consensual rather facilitating progression (104).
diagnosis in 88% of cases, although they did not increase the As in many tumors, alteration of Ras and Raf proteins are also
sensitivity in difficult cases (92). Other studies claimed inflamma- important for SCC and KA formation. Both can develop as sec-
tion or pattern of keratinization as reliable distinctive features ondary tumors upon melanoma treatment with BRAF inhibitors
(Table 4) (10,75). (29,105).
Cytological markers As shown recently, TGFb signalling is implicated in KA forma-
Cytokines tion. Eleven monoallelic mutations in the Tgfbr1 gene were speci-
PCR screening showed elevated levels of interleukin 10 (IL10) fied in Ferguson Smith disease (51). The role of TGFb signalling
and a decrease of granulocyte macrophage colony-stimulating in KA and SCC formation will be discussed in detail in this
factor (GM-CSF) in KA compared with SCC. IL10 usually report.
exhibits anti-tumoral but also immunosuppressive functions Cell cycle and apoptosis regulators
(93,94) thereby, inhibiting eosinophile GM-CSF and Th1 Ki-67 expression varies in KA since proliferative KA display higher
cytokine production (95). Low GM-CSF serum concentration immunostaining frequency than mature KA. Staining of p53 was
leading to an immunosuppressive state probably enables rapid strong and mainly localized to expanding tumor islands in KA
tumor growth on one hand and allows involution on the other and a diffuse nuclear staining throughout the whole lesion in
hand (96). SCC. Expression of p53 in KA and SCC is not sufficiently discrim-
Some studies reported diffuse distribution of transforming inatory to use it as a diagnostic tool but may be supporting the
growth factor-a (TGFa) in the majority of KA. Forty per cent of differential diagnosis of subungual KA versus subungual SCC
SCC exhibit a focal enrichment of TGFa whereas 90% of KA exhi- (88,89,106,107). Cyclin A and B exhibited predominantly basal
bit no staining in this area (97). and parabasal staining in KA, whereas it was diffuse in SCC (108).
In sum, these and many other targets have been tested in the
last decades, but no reliable marker was identified exhibiting suffi-
Table 4. Table of features to reliable distinguish keratoacanthoma (KA) and cient sensitivity and specificity.
squamous cell carcinoma (SCC) described in publications
Genetics
KA SCC Array CGH
Histopathology Array comparative genomic hybridization (array CGH) allows the
Cribier et al.: Cribier et al.: discrimination of KA and SCC in 85% of the cases. Recurrent
• Epithelial lips • Ulceration
• Sharp outline between • Numerous mitosis
aberrations on chromosome 7, 8 and 10 have been the best pre-
tumor and stroma • Pleomorphism/anaplasia dictors for SCC, suggesting a role in prevention of apoptosis or
Schwartz et al.: Schwartz et al.: activation of proliferation and infiltration for genes located in
• Intratumoral abscesses with • Intratumoral abscesses contain these areas. In contrast chromosomes 17, 19, 20 and X seemed to
granulocytes and acantholytic cells few or no inflammatory cells
be aberrant mainly in KA, indicating crucial genes for KA devel-
Weedon et al.: Weedon et al.: opment. These and other differences in the aberration pattern
• Pattern of keratinization • Pattern of keratinization
between the two tumors were proven to be significant, leading to
Genetics the assumption that KA and SCC are two distinct entities (79). A
Li et al.: Li et al.:
• Aberrations on chromosome: • Aberrations on chromosome: recent study investigating the relationship between KA and SCC
17, 19, 20, X 7, 8, 10 by DNA microarray technique concludes that KA and SCC are

ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
88 Experimental Dermatology, 2016, 25, 85–91

distinct, further defining KA as a distinct regressing neoplasm
(77).
Molecular pathways contributing to SCC and KA
formation
TGFb signalling in KA and SCC
The implication of TGFb signaling in KA formation was demon-
strated recently by Goudie et al. They ascertain a clear correlation
between loss of function mutations in the Tgbr1 gene and a MSSE
phenotype. With high-throughput genomic sequencing they iden-
tified the disease locus on chromosome 9 and by exome sequenc-
ing they localized in three unrelated families several distinct
mutations in Tgfbr1. In total, 11 monoallelic mutations in 18 fam-
ilies have been described, occurring in the cytoplasmic kinase
domain or in the extracellular ligand-binding domain. In the cyto-
plasmatic region of the receptor nonsense or frameshift mutations
occurred, resulting in a null mutation. In the extracellular ligand-
binding domain frameshift, nonsense or missense mutations have
been demonstrated, affecting the binding of TGFb or the interac-
tion with TGFbR2. These mutations were shown to be null sup-
ported by the fact that the bases of three of four of the missense
mutations are highly conserved across vertebrates. These heterozy-
gous MSSE mutations exhibit a loss of function, which is covered
by the functional protein transcribed by the wildtype allele (51).
The paradoxical role of TGFb signalling could explain the rapid
evolution and involution of KA after ablation of TGFb mediated
signal transduction. Additionally, a discontinuous membranous
expression was observed in spontaneous KA but rarely in SCC. If
this was due to mutation in the receptor could not be answered
definitively (109).
Transforming growth factor beta receptor 1 and 2 (TGFbR1
and TGFbR2) are transmembrane serine/threonine kinases, trans-
ducing signals of the TGFb protein superfamily from the cell sur-
face to the cytoplasm. Downstream signalling occurs mostly
through SMAD-proteins but also affects Ras, PI3K/Akt- or
MAPK-pathways (110,111). TGFb signalling is found in a large
number of tissues where it controls differentiation and growth
(112) and alterations are implicated in malignant transformation
of keratinocytes and tumorgenesis (113,114).
Transgenic mice overexpressing TGFb, were protected against
an early development of tumors but promoted tumor metastasis
later in tumorgenesis (111). A widely accepted hypothesis is that
an effect favouring tumor development is mediated by paracrine
signalling, stimulating angiogenesis and inflammation at late stages
whereas autocrine signalling seems to function as tumor inhibitor
at least in early stages (115). Secondary effects as onset of inflam-
mation or angiogenesis upon TGFb overexpression seem to be
responsible for keratinocyte hyperproliferation of head and neck
epithelium in vivo (116).
TGFbR1 deletion leads in a small number of animals to sponta-
neous formation of HNSCC, thus indicating that loss of the recep-
tor is not an initiation event in tumor formation (117).
Concomitant loss of TGFbR1 and of PTEN provoked the forma-
tion of spontaneous tumors (118). Combining TGFbR2 ablation
with either overexpression of oncogenic K-ras or DMBA treatment
led rapidly to the formation of aggressive growing SCC (119,120).
Glick et al. (113) reported significantly augmented tumor forma-
tion after skin graft experiments of TGFb1 negative keratinocytes
expressing virally transduced oncogenic v-ras.
ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Experimental Dermatology, 2016, 25, 85–91
Keratoacanthoma
Haploinsufficient TGFb1+/ mice developed fewer benign
tumors with reduced incidence and size compared with TGFb1+/+
mice. Even if TGFb+/+ mice had a higher onset of benign tumors,
no increased numbers of SCC were observed (121).
The fact that KA and SCC appearance is linked to TGFb alter-
ation is not sufficient to conclude that they are one entity. Since
the molecular background of SCC and KA differs strongly it is
very probable that the outcome of TGFb signalling is also altered
(77). Consequently, stem cells within the same SCC tumor, exhibit
opposite phenotypes depending on if they respond to TGFb stim-
uli or not (122). Furthermore, rather than Tgfbr1 mutations,
recently identified NOTCH1/2 receptor mutations are considered
as early gatekeeper mutations in SCC development (81).
Par3 signalling in KA formation
The Par3 pathway is implicated in KA formation and affects
downstream targets of TGFbR1 such as PI3K/Akt and MAPK.
Specifically, Par3 is involved in apico-basal cell polarity and asym-
metric cell division. Loss of cell polarity appears to be a prerequi-
site in tumor formation and progression. In DMBA/TPA tumor
models, Par3 overexpression facilitates papilloma formation but
reduces KA development. Opposite effects were observed in the
same tumor model with Par3 ablation, what may be explained by
its different intracellular localizations (123). Whether staining of
Par3 may be helpful in histopathology needs to be investigated.
RAF inhibitors favours KA and SCC development
A specific category of KA appears in melanoma patients treated
with BRAF inhibitors (29,61,105). Ninety per cent of all BRAF
mutations are V600E changes, causing a higher BRAF activity
thereby activating MAPK signalling pathway independently of
RAS (124,125). A side effect of the mutation-specific BRAF V600E
drug Vemurafenib in melanoma patients is the appearance of SCC
or KA in 15–20% of the patients. Sixty per cent of these epidermal
lesions have a mutated RAS leading to BRAF inhibitor mediated
paradoxical MAPK pathway activation in BRAF wild-type cells
(29). Melanoma patients treated with Sorafenib also developed
SCC or KA lesions in up to 7% of the cases (30,31).
Interferon treatment
Imiquimod, a nucleoside analogue of the imidazoquinolone fam-
ily, is usually applied against actinic keratoses and basal cell carci-
noma. It leads to upregulation of pro-inflammatory cytokines due
to activation of the NFkB pathway. Recently, there is a growing
evidence of Imiquimod induced KA as a secondary effect of the
treatment (26,28).
Treatment of KA and SCC
Treatment of patients with diagnosed KA include excision (stan-
dard or Mohs), radiotherapy, systemic retinoids and intralesional
methotrexate (19,126). Mohs surgery should be used in the appli-
cable case (19). Application of a Imiquimod 5% cream led to fast
regression of KA (127). Regarding the patients interest, conserva-
tive handling or Mohs surgery are the treatments of choice.
Conclusion
The etiology of KA involves many different factors, of which some
are seemingly more evident than others giving rise to a broad
spectrum of KA variants (10,11,13–30,32,33,69,128). Genetic pre-
disposition or spontaneous formation adds another level to the
complex classification system of KA. Today we conclude that SCC
and the benign neoplasm KA are two distinct entities, prototypi-
cally evidenced by alterations in the TGFb pathway in Ferguson
89
 Gleich et al.

Smith Disease, chromosomal aberration differences and pro-
nounced alterations in gene expression (51,77,79,80). Since
TGFbR1 mutations were identified as driver mutations in Fergu-
son Smith Disease, it would be important to identify mutations
leading to spontaneous KA and other KA variants e.g. by identify-
ing more disease loci followed by high-throughput genomic
sequencing. Additionally, the identification of pathways implicated
in the spontaneous involution of KA would be of serious interest.
With regard to the development of new therapeutics it would be
interesting to investigate the immune response in KA during dif-
ferent stages, leading to new insights concerning presented tumor
antigens or identification of specific reactive T-cells. Up to now,
dermatopathologists have a difficult task to distinguish KA and
SCC on histopathological grounds. For this purpose a fully excised
lesion is helpful, but to discriminate both tumors in an early stage
remains challenging. Consequently, it is essential to find morpho-
logical, biological and/or molecular markers which are showing
reliable differences between these two lesions to prove that a given
KA is not a SCC. Taken together, KA is a rapidly growing, sponta-
neously regressing tumor. Recently published data lead us to
assume that KA is distinct from SCC and caused by early gate-
keeper alterations in the TGFb pathway.
Acknowledgements
This work has been supported by grants to M.H. and D.H. by the Swiss
National Science Foundation and the Dind Cottier Foundation.
Author contribution
T.G. wrote and conceptualized this review. E.C. and M.H. revised the
paper. D.H. conceptualized, revised and coordinated the work.
Conflict of interests
The authors have declared no conflicting interests.

References
1 Hutchinson J. Trans Pathol Soc London 1889:
40: 275–281.
2 Maccormac H, Scarff R. Br J Dermatol 1936:
48: 624–626.
3 Dunn J, Smith J F. Br J Dermatol 1934: 46:
519.
4 Kwittken J. Mt Sinai J Med 1975: 42: 127–
135.
5 Rook A, Whimster I. Br J Dermatol 1979: 100:
41–47.
6 Miedzinski F. Postepy Dermatol (Poznan) 1990:
7: 61–70.
7 Musso L, Proc R. Soc Med 1950: 43: 838.
8 Mandrell J C, Santa Cruz D. Semin Diagn
Pathol 2009: 26: 150–163.
9 Beham A, Regauer S, Soyer H P et al. Adv Anat
Pathol 1998: 5: 269–280.
10 Schwartz R A. J Am Acad Dermatol 1994: 30:
1–19; quiz 20–12.
11 Schwartz R A. Keratoacanthoma. In: Chu A C,
Edelson R L, eds. Malignant Tumors of the
Skin. London: Chapman & Hall, 1999: 96–118.
12 Karaa A, Khachemoune A. Int J Dermatol
2007: 46: 671–678.
13 Schwartz R A. J Surg Oncol 1979: 12: 305–
317.
14 Kingman J, Callen J P. Arch Dermatol 1984:
120: 736.
15 Dogliotti M, Caro I. Int J Dermatol 1976: 15:
524.
16 Dufresne R, Marrero G, Robinson-Bostom L. Br
J Dermatol 1997: 136: 227–229.
17 Chuang T-Y, Reizner G T, Elpern D J et al. Arch
Dermatol 1993: 129: 317.
18 Ghadially F, Barton B, Kerridge D. Cancer
1963: 16: 603–611.
19 Schwartz R A. Dermatol Surg 2004: 30: 326–
333; discussion 333.
20 Sina B, Adrian R M. J Am Acad Dermatol
1983: 9: 686–688.
21 Maddin W, Wood W S. J Cutan Pathol 1979:
6: 96–100.
22 Gewirtzman A, Meirson D H, Rabinovitz H.
Dermatol Surg 1999: 25: 666–668.
23 Shaw J C, Storrs F J, Everts E. J Am Acad Der-
matol 1990: 23: 1009–1011.
24 Wagdy Dessoukey M, Omar M F, Abdel-
Dayem H. J Am Acad Dermatol 1997: 37:
478–480.
25 Ferguson-Smith M, Wallace D, James Z et al.
Birth Defects Orig Artic Ser 1971: 7: 157.
26 Foxton G, Delaney T. Australas J Dermatol
2011: 52: 66–69.
27 Pini A M, Koch S, Scharer L et al. J Am Acad
Dermatol 2008: 59: S116–S117.
28 D’Addario S, Carrington P R. Acta Derm Vener-
eol 2006: 86: 366–367.
29 Sharma A, Shah S R, Illum H et al. Drugs
2012: 72: 2207–2222.
30 Arnault J P, Mateus C, Escudier B et al. Clin
Cancer Res 2012: 18: 263–272.
31 Di Marco V, De Vita F, Koskinas J et al. Ann
Oncol 2013: 24 (Suppl 2): ii30–ii37.
32 Zelickson A. Acta Derm Venereol 1962: 42: 23.
33 Payne D, Newman C, Tyring S. J Am Acad Der-
matol 1995: 33: 1047.
34 Ghadially R, Ghadially F N. Keratoacanthoma.
In: Fitzpatrick T B, Eisen A Z, Wolff K, eds. Der-
matology in General Medicine. New York:
McGraw-Hill, 1993: 845–855.
35 Leonard A L, Hanke C W. J Drugs Dermatol
2006: 5: 454.
36 De Visscher J, van der Wal J, Starink T M et al.
Oral Surg Oral Med Oral Pathol Oral Radiol
Endod 1996: 81: 193–196.
37 Webb A, Ghadially F. J Pathol Bacteriol 1966:
91: 505–509.
38 Lejman K, Starzycki Z. Br J Vener Dis 1977: 53:
65–67.
39 Rapaport J. Arch Dermatol 1975: 111: 73.
40 Saito M, Sasaki Y, Yamazaki N et al. Plast
Reconstr Surg 2003: 111: 1561–1562.
41 Cramer S, Am J. Clin Pathol 1981: 75: 425.
42 Patel M R, Desai S. J Hand Surg Am 1989: 14:
139–142.
43 Choonhakarn C, Ackerman A. Dermatopathol
Pract Concept 2001: 7: 7–16.
44 Janette A, Pecaro B, Lonergan M et al. J Oral
Maxillofac Surg 1996: 54: 1026–1030.
45 Svirsky J A, Freedman P D, Lumerman H. Oral
Surg Oral Med Oral Pathol 1977: 43: 116–122.
46 Belisario J C. Aust J Dermatol 1965: 8: 65–72.
47 V’Lckova-Laskoska M T, Laskoski D S. Clin Exp
Dermatol 2008: 33: 259–261.
48 Sung C S, Chuang F C, Ho J C et al. Dermato-
logica Sinica 2014: 32: 25–28.
49 Aydin F, Senturk N, Sabanciler E et al. Clin Exp
Dermatol 2007: 32: 683–686.
50 Smith J F. Br J Dermatol 1934: 46: 267–272.
51 Goudie D R, D’Alessandro M, Merriman B
et al. Nat Genet 2011: 43: 365–369.
52 Consigli J, Gonzalez M, Morsino R et al. Br J
Dermatol 2000: 142: 800–803.
53 Grzybowski M. Br J Dermatol 1950: 62: 310–
313.
54 Schwartz R A, Blaszczyk M, Jablonska S. Der-
matology 2002: 205: 348–352.
55 Witten V H, Zak F G. Cancer 1952: 5: 539–
550.
56 Ponti G, Losi L, Di Gregorio C et al. Cancer
2005: 103: 1018–1025.
57 Ponti G. Ponz de Leon M. Lancet Oncol 2005:
6: 980–987.
58 DiGiovanna J J, Kraemer K H. J Invest Dermatol
2012: 132: 785–796.
59 Affleck A G. Int J Dermatol 2007: 46: 1105.
60 Harvey N T, Millward M, Wood B A. Am J Der-
matopathol 2012: 34: 822–826.
61 Holderfield M, Deuker M M, McCormick F
et al. Nat Rev Cancer 2014: 14: 455–467.
62 Levy E J, Cahn M M, Shaffer B et al. J Am
Med Assoc 1954: 155: 562–564.
63 Brenn T, McKee P, Mihm M. Keratoacanthoma.
In: McKee P et al., ed. Pathology of Skin.
Philadelphia: Elsevier Mosby, 2005: 1221–
1227.
64 Fisher E R, McCoy M M, Wechsler H L. Cancer
1972: 29: 1387–1397.
65 Ghadially F N. Cancer 1961: 14: 801–816.
66 Wade T, Ackerman A. J Dermatol Surg Oncol
1978: 4: 498.
67 Cooper P H, Wolfe J T 3rd. Arch Dermatol
1988: 124: 1397–1401.
68 Ghadially F. J Pathol Bacteriol 1958: 75: 441–
453.
69 Yus E S, Simo n P, Requena L et al. Am J Der-
matopathol 2000: 22: 305–310.
70 Ko C J, McNiff J M, Bosenberg M et al. J Am
Acad Dermatol 2012: 67: 1008–1012.
71 Patel A, Halliday G, Cooke B et al. Br J Derma-
tol 1994: 131: 789–798.
72 Zito G, Saotome I, Liu Z et al. Nat Commun
2014: 5: 3543.
73 Boumahdi S, Driessens G, Lapouge G et al.
Nature 2014: 511: 246–250.
74 Blessing K, Nafussi A, Gordon P. Histopathol-
ogy 1994: 24: 381–384.
75 Weedon D. Curr Diagn Pathol 2003: 9: 259–265.
76 Weedon D, Malo J, Brooks D et al. ANZ J Surg
2010: 80: 129–130.
77 Ra S H, Su A, Li X et al. Mod Pathol 2015: 28:
799–806.
78 Hodak E, Jones R E, Ackerman A B. Am J Der-
matopathol 1993: 15: 332–342.
79 Li J, Wang K, Gao F et al. J Invest Dermatol
2012: 132: 2060–2066.
80 Clausen O P F, Aass H C D, Beigi M et al. J
Invest Dermatol 2006: 126: 2308–2315.
81 South A P, Purdie K J, Watt S A et al. NOTCH1
mutations occur early during cutaneous squa-
mous cell carcinogenesis. J Invest Dermatol
2014: 134: 2630–2638.
82 Schnur P L, Bozzo P. Plast Reconstr Surg 1978:
62: 258–262.
83 Joseph M G, Zulueta W P, Kennedy P J. ANZ J
Surg 1992: 62: 697–701.
84 Brantsch K D, Meisner C, Schonfisch B et al.
Lancet Oncol 2008: 9: 713–720.
85 Ko C J. Clin Dermatol 2010: 28: 254–261.
86 Jackson I T, Brown G E D. Br J Plast Surg 1970:
23: 373–379.
87 Ledo E. Int J Dermatol 1992: 31: 777–778.
88 Honma M, Kato N, Hashimoto M et al. Clin
Exp Dermatol 2011: 36: 57–62.
89 Gonzalez-Rodriguez A J, Gutierrez-Paredes E
M, Montesinos-Villaescusa E et al. Actas Der-
mosifiliogr 2012: 103: 549–551.

ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
90 Experimental Dermatology, 2016, 25, 85–91

90 Levy D W, Bonakdarpour A, Putong P B et al.
Skeletal Radiol 1985: 13: 287–290.
91 Oliwiecki S, Peachey R D, Bradfield J W et al.
Clin Exp Dermatol 1994: 19: 230–235.
92 Cribier B, Asch P-H, Grosshans E. Dermatology
1999: 199: 208–212.
93 Kundu N, Beaty T L, Jackson M J et al. J Natl
Cancer Inst 1996: 88: 536–541.
94 Lippitz B E. Lancet Oncol 2013: 14: e218–e228.
95 Borish L. J Allergy Clin Immunol 1998: 101:
293–297.
96 Lowes M, Bishop G, Cooke B et al. Br J Cancer
1999: 80: 1501.
97 Ho T, Horn T, Finzi E. Arch Dermatol 1991:
127: 1167.
98 Melendez N D, Smoller B R, Morgan M. Mod
Pathol 2003: 16: 8–13.
99 Asch P-H, Basset P, Roos M et al. Am J Der-
matopathol 1999: 21: 146–150.
100 Groves R W, Allen M H, MacDonald D M. J
Cutan Pathol 1992: 19: 66–72.
101 Jacobs M S, Persons D L, Fraga G R. J Cutan
Pathol 2013: 40: 447–454.
102 Reuter C, Morgan M, Eckardt A. Br J Cancer
2007: 96: 408–416.
103 Schlauder S M, Calder K B, Moody P et al. Am
J Dermatopathol 2007: 29: 559–563.
ª 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Experimental Dermatology, 2016, 25, 85–91
104 Shimizu T, Izumi H, Oga A et al. Dermatology
2001: 202: 203–206.
105 LaPresto L, Cranmer L, Morrison L et al. JAMA
Dermatol 2013: 149: 279–281.
106 Wrone-Smith T, Bergstrom J, Quevedo M E
et al. J Dermatol Sci 1999: 19: 53–67.
107 Connolly M, Narayan S, Oxley J et al. Clin Exp
Dermatol 2008: 33: 625–628.
108 Tran T A, Ross J S, Boehm J R et al. J Cutan
Pathol 1999: 26: 391–397.
109 Berger F, Geddert H, Faller G et al. Pathol Res
Pract 2014: 210: 596–602.
110 Massague J. Nat Rev Mol Cell Biol 2012: 13:
616–630.
111 Bierie B, Moses H L. Nat Rev Cancer 2006: 6:
506–520.
112 Massague J, Blain S W, Lo R S. Cell 2000: 103:
295–309.
113 Glick A, Popescu N, Alexander V et al. Proc
Natl Acad Sci U S A 1999: 96: 14949–14954.
114 Levy L, Hill C S. Cytokine Growth Factor Rev
2006: 17: 41–58.
115 De Wever O, Mareel M. J Pathol 2003: 200:
429–447.
116 Lu S L, Reh D, Li A G et al. Cancer Res 2004:
64: 4405–4410.
Keratoacanthoma
117 Bian Y, Terse A, Du J et al. Cancer Res 2009:
69: 5918–5926.
118 Bian Y, Hall B, Sun Z J et al. Oncogene 2012:
31: 3322–3332.
119 Lu S L, Herrington H, Reh D et al. Genes Dev
2006: 20: 1331–1342.
120 Malkoski S P, Haeger S M, Cleaver T G et al.
Clin Cancer Res 2012: 18: 2173–2183.
121 Perez-Lorenzo R, Markell L M, Hogan K A
et al. Carcinogenesis 2010: 31: 1116–1123.
122 Oshimori N, Oristian D, Fuchs E. Cell 2015:
160: 963–976.
123 Iden S, van Riel W E, Schafer R et al. Cancer
Cell 2012: 22: 389–403.
124 Pratilas C A, Xing F, Solit D B. Curr Top Micro-
biol Immunol 2012: 355: 83–98.
125 Joseph E W, Pratilas C A, Poulikakos P I et al.
Proc Natl Acad Sci U S A 2010: 107: 14903–
14908.
126 Annest N M, VanBeek M J, Arpey C J et al. J
Am Acad Dermatol 2007: 56: 989–993.
127 Di Lernia V, Ricci C, Albertini G. J Eur Acad
Dermatol Venereol 2004: 18: 626–629.
128 Barrett J C. Environ Health Perspect 1993: 100:
9–20.
91