Keratoacanthoma: Clinical and histopathologic

features of regression
Christine J. Ko, MD, Jennifer M. McNiff, MD, Marcus Bosenberg, MD, PhD, and Keith A. Choate, MD, PhD
New Haven, Connecticut
Background: The clinical and histopathologic features of regressing keratoacanthomas have not been
adequately described in the literature.
Objective: True keratoacanthomas (ie, squamous tumors with evidence of spontaneous resolution) were
studied clinically and histopathologically.
Methods: Nineteen crateriform tumors with a partial biopsy histopathologically compatible with keratoacanthoma were followed over time for correlation with biologic behavior (ie, regression). Tumors
displaying spontaneous resolution, arbitrarily defined as a decrease in size of at least 25%, were categorized
as keratoacanthomas.
Results: Seven regressing keratoacanthomas tended to show flattening before a decrease in diameter.
Histopathologically, there was variable epidermal hyperplasia with generally prominent hyperkeratosis,
retained crateriform architecture, and dermal fibrosis.
Limitations: This study has a small sample size.
Conclusions: Regressing keratoacanthomas show persistent crateriform architecture, clinically and
histopathologically. Lesions become flatter before decreasing in diameter, and keratinocytes appear banal
and lack glassy pink cytoplasm during regression. ( J Am Acad Dermatol 2012;67:1008-12.)
Key words: crateriform; keratoacanthoma; regression; squamous cell carcinoma; squamous lesion;
squamous proliferation.

INTRODUCTION
As it is currently well recognized that keratoacanthoma can be indistinguishable from squamous cell
carcinoma, clinically as well as histopathologically,1
it is increasingly difficult to recognize the regressing
keratoacanthoma. Several factors contribute to this,
including 1) many dermatopathologists and dermatologists consider so-called keratoacanthoma to be a
variant of squamous cell carcinoma2 as it is not
possible to predict biologic behavior (ie, regression)
with microscopic features alone; 2) case reports of
metastasizing keratoacanthoma2; and 3) the current tendency to excise keratoacanthomas rather
than to observe for regression.3
Although keratoacanthoma may simply be a selfe
resolving variant of squamous cell carcinoma, abrogating the term keratoacanthoma complicates the
From Yale University School of Medicine.
Funding sources: None.
Conflicts of interest: None declared.
Reprint requests: Christine J. Ko, MD, 333 Cedar St, PO Box 208059,
New Haven, CT 06520. E-mail: christine.ko@yale.edu.

1008

nomenclature of regressing lesions, which tend to

retain a crateriform architecture clinically but do not
resemble squamous cell carcinoma histopathologically.3,4 Furthermore, as keratoacanthoma-like lesions and squamous cell carcinoma have been
described to develop in patients treated with sorafenib5 and vemurafenib,6 it is especially relevant to
distinguish these lesions, when possible. In this
report, the term keratoacanthoma is used for
crateriform tumors displaying spontaneous regression (ie, tumor shrinkage of at least 25% at followup), and clinical and histopathologic findings of 7
such cases are described.

METHODS
Tumors that displayed rapid growth and crateriform architecture underwent partial biopsy (one
Published online April 23, 2012.
0190-9622/$36.00
2012 by the American Academy of Dermatology, Inc.
doi:10.1016/j.jaad.2012.02.041

J AM ACAD DERMATOL

Ko et al 1009

VOLUME 67, NUMBER 5

fourth of the tumor was removed in a wedge from 12

Clinically, the 6 tumors displaying regression had
to 3 oclock). The biopsy was examined by at least 2
flattened compared with initial presentation, and 5 of
experienced dermatopathologists for features of
6 had decreased in diameter as well. The 13 tumors
keratoacanthoma (eg, crateriform architecture, epithat were not smaller on follow-up were the same
thelial lip, glassy keratinocytes, sharp demarcation
size or larger. Only 3 of the 6 patients were aware that
of tumor from stroma, intraepidermal neutrophil/
their tumors were smaller or flatter compared with
eosinophil abscesses, eosinophils in the infiltrate,
initial presentation.
fibrosis, entrapped elastic fiOne additional tumor
bers, apoptotic keratinocytes,
(case 7) had a clinical history
CAPSULE SUMMARY
lack of prominent cytologic
of rapid growth and crateriatypia, absence of atypical
form architecture (Fig 2, A);
Keratoacanthoma is considered by many
mitoses, lack of infiltrative
the clinician highly susto be a variant of squamous cell
growth pattern, and perineupected a squamous cell carcarcinoma.
ral invasion).
cinoma. The partial biopsy
A true keratoacanthoma can be
If a crateriform architecspecimen showed a cystdefined as a crateriform squamous
ture and glassy keratinocytes
like cavity, a thin rim of eptumor that shows spontaneous
were present with at least 5
idermis lining the cyst, and
regression.
of the other features, the tumassive central hyperkeratomor was followed up over
sis (Fig 2, B). This tumor did
The clinical and histopathologic features
time. By 1 month of clinical
not
fit the histologic criteria
of regressing keratoacanthoma have not
follow-up, tumors were asfor
inclusion (the tumor
been adequately described in the
sessed for clinical regression,
lacked
glassy keratinocytes),
literature.
arbitrarily defined as tumor
but at 1-month follow-up
Regressing keratoacanthoma retains a
size at least 25% smaller comshowed complete regression
crateriform architecture clinically and
pared with the initial preof the remainder. Therefore
histopathologically, with central
sentation. Tumor size was
the initial biopsy of this tuhyperkeratosis rimmed by a banal
measured by length/width
mor was considered to repepidermis of variable thickness and
as well as by an estimate of
resent the regressing stage of
underlying fibroplasia.
tumor volume based on
keratoacanthoma.
baseline and follow-up cliniA common finding in all
cal presentation and photogregressing keratoacanthomas
raphy. All tumors, regardless of the presence of
was persistent crateriform architecture over time,
regression, were excised by 1 month after initial
clinically and microscopically. All 7 regressing tumors
presentation. Tumors with evidence of clinical relacked glassy keratinocytes. In 4 tumors, the epidermis
gression at follow-up were classified as keratoacrimming the crater appeared thin with flattening of the
anthoma. Tumors without evidence of clinical
rete ridge pattern, and the crater was filled with
regression at follow-up were classified arbitrarily as
hyperkeratosis. In 3 tumors, the epidermis was hypersquamous cell carcinoma.
plastic with thin rete descending from an acanthotic
epidermis (Fig 3). All 7 tumors showed dermal inflammation and fibrosis.
d

RESULTS
Twenty-two tumors presented with rapid clinical
growth and crateriform architecture (Fig 1, A). Three
tumors were eliminated as the partial biopsy did not
meet histopathologic criteria for keratoacanthoma.
Of the 19 remaining tumors with histopathologic
features on partial biopsy compatible with keratoacanthoma (Table I), 6 showed evidence of clinical
regression at 1-month follow-up (Table II) (Fig 1, B).
Comparing the initial biopsy specimens of the 6
tumors with regression against the 13 tumors without
regression at follow-up, tumors showing clinical
regression lacked atypical mitoses, showed dermal
fibrosis, and had intraepidermal neutrophilic/eosinophilic abscesses (see Table I).

DISCUSSION
Clinical7 and histopathologic features4 of regressing
keratoacanthoma have not been recently described in
the literature. Major textbooks of dermatopathology
mention regressing keratoacanthoma only briefly,8,9
and the life of a keratoacanthoma was best chronicled by Dr. Ackerman almost 20 years ago.10 To our
knowledge, only one prior study on keratoacanthoma correlated histopathologic findings with biologic behavior.3 In that study,3 the authors focused on
the histopathologic findings in the initial biopsies of
lesions that regressed (keratoacanthoma) versus
those that did not (squamous cell carcinoma). The
majority of regressing lesions were left untreated and

J AM ACAD DERMATOL

1010 Ko et al

NOVEMBER 2012

Fig 1. Case 2. A, Keratoacanthoma. At initial presentationcrateriform tumor with stretched,

pink epidermis. B, Keratoacanthoma. Three weeks latermarked flattening of the lesion.
C, Keratoacanthoma. Initial partial biopsythere are down-growing lobules of keratinocytes.
D, Keratoacanthoma. Three weeks laterfrozen section after debulking before Mohs surgery;
the tumor shows a thinned epidermis lining a cavity with marked hyperkeratosis.

Table I. Initial histopathologic features of 19 tumors clinically compatible with keratoacanthoma at initial
presentation*
Present in
Histologic findings

Crateriform
Epithelial lip
Sharp demarcation tumor/stroma
Glassy keratinocytes
Fibrosis
Neutrophil/eosinophil abscesses
Eosinophils
Entrapped elastic fibers
Apoptotic keratinocytes
Lack of atypical mitoses
Lack of cytologic atypia
Lack of infiltrative growth
Perineural invasion

Tumors ultimately showing

regression (6 total)

6 (100%)
6 (100%)
6 (100%)
6 (100%)
4 (except cases 5 and 6) (67%)
6 (100%)
6 (100%)
4 (except cases 1 and 6) (66%)
6 (100%)
6 (100%)
2 (except cases 2, 3, 5, and 6) (33%)
6 (100%)
0 (0%)

Tumors ultimately without

regression (13 total)

13
13
12
9
1
3
9
9
13
7
4
12
0

(100%)
(100%)
(93%)
(75%)
(8%)
(25%)
(75%)
(75%)
(100%)
(54%)
(33%)
(93%)
(0%)

*That is, rapid growth, crateriform architecture.

followed up to complete clinical resolution, without

additional histopathologic evaluation.3
The 7 tumors in our study were clinically and
histopathologically evaluated at two different time
points and displayed spontaneous regression. As the
tumors were excised at follow-up, we cannot know
with absolute certainty that the tumors would have

resolved completely (with the exception of case 7).

However, in a prior study on regressing keratoacanthomas, tumors showing evidence of regression
at 1-month follow-up continued on to complete
resolution.3
Clinically, our keratoacanthomas reached a stable
size and appeared to remain crateriform with a

J AM ACAD DERMATOL

Ko et al 1011

VOLUME 67, NUMBER 5

Table II. Clinical features of 7 regressing keratoacanthomas

Size (cm)
Case No.

1
2
3
4
5
6
7

Age (years)

Sex

Site

Duration before biopsy

Initial

Final

57
74
74
90
80
83
78

M
M
F
F
F
M
F

Forearm
Hand
Forearm
Nose
Upper arm
Upper back
Forearm

3 months
3 months
Unknown
1 month
1 month
6 months
Unknown

2.3 3 1.9
2.4 3 2
1.2 3 1.5
0.6 3 0.6
131
131
131

2.1 3 1.5
2.5 3 2.0
0.6 3 0.6
0.5 3 0.4
131
0.5 3 0.7
0

Time between visits

1
3
1
1
1
1
1

month
weeks
month
month
week
month
month

Fig 2. Case 7. A, Keratoacanthoma. At initial presentationcrateriform tumor. B, Keratoacanthoma. Initial partial biopsymassive central hyperkeratosis lined by a rim of thinned
epidermis.

Fig 3. Case 1. Keratoacanthoma. At 1-month follow-up,

excision of the remaining tumor showed hyperkeratosis in
a shallow cup-shaped lesion that was rimmed by epidermal hyperplasia with thin, down-growing rete.

decrease in height (flattening of the tumor) before a

decrease in diameter. This finding correlates well
with loss of both epidermal hyperplasia and glassy
(enlarged) keratinocytes. The stable crateriform
architecture also correlates well with the central
hyperkeratosis seen in our cases. Our finding of
variable epidermal hyperplasia has been described
previously for regressing keratoacanthomas.3,4
Of 19 tumors that appeared to have an initial
presentation compatible with keratoacanthoma on
clinical and histopathologic grounds, only 6 showed
evidence of regression by 1-month follow-up.

A limitation of our study is that it remains possible

that the 13 tumors that did not show evidence of
regression during our follow-up period may have
displayed regression at a later date. Certainly, it can
take months or longer for a keratoacanthoma to
resolve7; however, because of the constraints of our
study protocol, we were not able to follow the
tumors for longer periods. Our findings confirm
that on initial biopsy, it can be impossible to differentiate keratoacanthoma (ie, a tumor that will regress) from squamous cell carcinoma (ie, a tumor
that will not regress).3 However, our results support
that keratoacanthoma (ie, a tumor that will regress)
generally lacks atypical mitoses and has a greater
tendency to display fibrosis and intraepidermal abscesses of neutrophils/eosinophils compared with
squamous cell carcinoma (ie, non-regressing squamous tumors) (see Table I).
Although it can be argued that the partial biopsy of
these tumors induced regression, it is clear that rapid
regression was not induced in the majority of the
tumors that underwent biopsy for this study. This also
reflects clinical practice, as the majority of skin cancers,
if biopsied only partially, do not go on to regress
spontaneously. The mechanism of regression in keratoacanthoma remains unclear and bears further study.

1012 Ko et al

Our study suggests that the keratoacanthoma (ie,

a squamous tumor that spontaneously regresses) is a
distinct neoplasm with reproducible clinical and
histopathologic features in its regressing stage.
Clinically, lesions are crateriform and may have a
history of having decreased in height; it is important
to note that 3 of 6 patients in this study were not
aware that their lesions had shrunk in size, so
patient-derived history cannot necessarily be relied
upon to assess regression. Microscopically, regressing keratoacanthomas show a keratin-filled crater
rimmed by variable epidermal hyperplasia composed of banal (non-glassy, non-atypical) keratinocytes with dermal fibrosis and inflammation.
Although we did not encounter a keratoacanthoma
that became less sessile and more exophytic with
regression, this morphology has been described.10
Thus, our small sample size may not comprehensively include all patterns of regression.
While keratoacanthoma may very well be a selfresolving variant of squamous cell carcinoma, as
patients on sorafenib5 and vemurafenib6 are at risk
for both keratoacanthoma and squamous cell carcinoma, it would be useful to accurately predict tumor
course, if possible. As keratoacanthoma is sometimes
initially biopsied in the regressing stage (eg, case 7),
clinicians and dermatopathologists can more accurately make a diagnosis when aware of the features
of regressing keratoacanthoma. The diagnosis of
regressing keratoacanthoma should be entertained
for a clinically crateriform tumor with histopathologic features of central hyperkeratosis, a

J AM ACAD DERMATOL

NOVEMBER 2012

hyperplastic to thin rim of banal keratinocytes, and

underlying dermal fibrosis.
The authors thank Sumaira Aasi, Jennifer Choi, and
Leonard Milstone for their contribution of cases.
REFERENCES
1. Cribier B, Asch P, Grosshans E. Differentiating squamous cell
carcinoma from keratoacanthoma using histopathological
criteria. Is it possible? A study of 296 cases. Dermatology
1999;199:208-12.
2. Hodak E, Jones RE, Ackerman AB. Solitary keratoacanthoma is
a squamous-cell carcinoma: three examples with metastases.
Am J Dermatopathol 1993;15:332-42, [discussion 43-52].
3. Magalhaes RF, Cruvinel GT, Cintra GF, Cintra ML, Ismael AP, de
Moraes AM. Diagnosis and follow-up of keratoacanthoma-like
lesions: clinical-histologic study of 43 cases. J Cutan Med Surg
2008;12:163-73.
4. Blessing K, al Nafussi A, Gordon PM. The regressing keratoacanthoma. Histopathology 1994;24:381-4.
5. Arnault JP, Wechsler J, Escudier B, Spatz A, Tomasic G, Sibaud
V, et al. Keratoacanthomas and squamous cell carcinomas in
patients receiving sorafenib. J Clin Oncol 2009;27:59-61.
6. Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P,
Larkin J, et al. Improved survival with vemurafenib in melanoma
with BRAF V600E mutation. N Engl J Med 2011;364:2507-16.
7. Rook A, Whimster I. Keratoacanthomaa thirty year retrospect. Br J Dermatol 1979;100:41-7.
8. Weedon D. Chapter 31: Tumors of the epidermis [pp 667-708].
In: Weedons Skin Pathology. 3rd ed. China: Churchill Livingstone; 2010. p. 704.
9. Kirkham N. Chapter 29: Tumors and cysts of the epidermis [pp
791-849]. In: Elder DE, Elenitsas R, Johnson BL, Murphy GF,
Xu X, eds. Levers Histopathology of the skin; 10th ed.
Philadelphia: Lippincott Williams and Wilkins. 2009. p. 837.
10. Ackerman AB, Ragaz A. Keratoacanthoma. In: The Lives of
Lesions: Chronology in Dermatopathology. New York: Masson
Publishing USA, Inc.; 1984. pp. 102-9.