A Clinical and Biological Review of Keratoacanthoma

A. Tisack; A. Fotouhi; C. Fidai; B.J. Friedman; D. Ozog; J. Veenstra


The British Journal of Dermatology. 2021;185(3):487-498. 

In This Article

Clinical Behaviour of Cutaneous Squamous Cell Carcinoma and Keratoacanthoma

Epidemiology of Cutaneous Squamous Cell Carcinoma

cSCC is the most common type of cancer with metastatic potential and historically has accounted for approximately 20% of all cutaneous malignancies. However, recent studies indicate that cSCC is increasing in incidence and may constitute up to 50% of nonmelanoma skin cancer.[15–17] Malignant transformation of keratinocytes arises primarily in photodamaged skin of elderly patients with fair skin.[18] Major risk factors include repeated ultraviolet exposure, radiation, immunosuppression, chronic nonhealing wounds and human papillomavirus (HPV) infection.[19–22] Additionally, the advent of targeted molecular therapies and immune checkpoint inhibitors has led to increased incidence of cSCC stemming from dysregulated cell-regulatory pathways.[23–27]

Although the prognosis of cSCC is typically favourable, nodal metastasis rates range from 4% to 6%, but have been reported as high as 30% for high-risk locations such as the lip and ear.[28,29] The 5-year survival rate for metastatic cSCC may be as low as 34·4%, and approximately 1·5% of all cases are fatal.[30–32] Both the American Joint Committee on Cancer (AJCC) Staging Manual and the Brigham and Women's Hospital (BWH) Tumor Classification System have been utilized for cSCC staging, the latter of which may predict outcomes more accurately.[33]

Epidemiology of Keratoacanthoma

The incidence, rate of regression and persistence of KA remain poorly characterized. Reports estimate the incidence of KA to range between 100 and 150 cases per 100 000 individuals; however, this is likely a gross underestimation due to misclassification of these lesions as well-differentiated cSCC, under-reporting, or spontaneous regression before diagnosis.[34,35] Risk factors for KA development are similar to those of well-differentiated cSCC. However, uniquely, cutaneous trauma (i.e. surgery, laser resurfacing, radiation) appears to be an additional risk factor for local KA development, suggesting that dysregulated inflammatory responses may contribute towards its pathogenesis.[4,34,36–38] Neither the AJCC nor BWH incorporate any additional consideration for KA-type cSCC.

Diagnosis of Keratoacanthoma

Currently, diagnosis of KA is based upon three key facets:[38] (i) characteristic clinical presentation of a rapidly developing crateriform lesion over the course of weeks to months; (ii) triphasic evolution consisting of proliferation, stabilization and regression (for untreated lesions) and (iii) histopathology of an adequate specimen with intact architecture. Distinguishing a KA from well-differentiated cSCC relies on often subtle architectural and cytological differences.[39]

Morphology of Keratoacanthoma

Gross. Solitary KAs are the most common type, arising as minute papules that mature into dome- or bud-shaped, sharply circumscribed, umbilicated nodules with a central hyperkeratotic plug.[38] The process from origin to spontaneous resolution usually occurs over 4–6 months and they may heal with or without prominent scarring. There are several other variants of KAs and associated syndromes (Table 1).[40–68] Each of these is classified as KA due to morphology; however, based on their disparate behaviours they may harbour different genetic drivers.

Imaging. The keratinocytic origin of both KA and cSCC makes discrimination unreliable by dermoscopy.[69] KAs are often characterized by a central, yellow-brown, structureless keratin plug surrounded by elongated hairpin vessels, but this is not specific to KA.[70,71] One study of 32 cSCCs and 29 KAs found central keratin to be more common in KAs (51% vs. 30%),[69] whereas another study of 100 KAs and 410 cSCCs found 'branching vessels' more frequently in KAs (25% vs. 10·3%).[72] However, the low sensitivity and high variability of these features limit the clinical application of these findings. Reflectance confocal microscopy has also been investigated, but has similar issues to dermoscopy, with poor discrimination between KA and cSCC.[73,74]

Micro: Histopathological Architecture and Cytology. Differentiating KA from well-differentiated cSCC on routine histology can be difficult due to subtle distinguishing features. This is further complicated by the high frequency of biopsy samples that fail to include the complete tumour architecture.[3,5] Inadequate sampling of a KA is more likely to lead to a diagnosis of cSCC and potentially overtreatment. Best exemplifying the dermatopathologist's dilemma is the vast discrepancy in reported ratios of cSCC to KA, ranging from 2·5 : 1 to 139 : 1, among medical centres across Great Britain and Ireland.[3]

The histopathological features of KAs are phase dependent (Figure 1).[75,76] Early lesions consist of an exo-endophytic proliferation of pale squamous cells in lobules, some resembling distorted infundibular structures.[39] These start in contiguity with the adjacent epidermis and then progressively extend into the mid-to-deeper portions of the reticular dermis, with further extension beyond the sweat glands being unusual.[77] In later biopsies, the infundibular structures become more cystic and hyperkeratotic, coalescing to form a central keratin plug.

Figure 1.

Keratoacanthoma evolution: histopathological features.

Well-developed KAs are largely symmetrical, with most peripheral tumour islands demonstrating little infiltration beyond the confines of the central mass.[78] Typically, there is buttressing of the surrounding normal epidermis around the tumour.[79] The peripheral keratinocytes notably have enlarged, pink, glassy-appearing cytoplasm, a low nuclear-to-cytoplasmic ratio, and minimal nuclear atypia.[39,76,80] A mixed infiltrate of inflammatory cells is common. In some cases, neutrophils and eosinophils may be prominent, often extending into the epithelial islands and forming small microabscesses.[78]

Regressing lesions are characterized by a well-formed crater of keratin with thinning of the surrounding squamous epithelium, fewer overall squamous lobules, and progressive development of underlying dermal fibrosis.[76,78] The main histopathological features that are thought to exclude a KA and confirm a diagnosis of cSCC are the presence of asymmetry, extension beyond the sweat glands, signs of infiltration and associated desmoplasia, and the presence of more conspicuous nuclear atypia.[39,81] Often only one or two of these features are present (with the assessment of atypia being subjective), making the diagnosis especially difficult. Both murine models and human KAs have been utilized to characterize each of the phases by histopathology (Figure 1).[40,76,78]

Rare cases of metastasis purported to originate from KAs can be found in the literature; however, these exceptional scenarios can be challenged in a number of ways.[82–84] Firstly, cSCC can probably arise within a KA, and it is that component that would be likely to metastasize.[85] Secondly, some KAs that have metastasized may have truly been cSCCs with a distinct follicular pattern of differentiation.[86] Thirdly, a number of visceral carcinomas that can metastasize to the skin have the capability to masquerade as KA.[87,88] Depending on the adequacy of the original cutaneous biopsy, these former possibilities may or may not be detectable at the time of the initial diagnosis. The future development of a reliable, distinctive proteomic signature will help differentiate true KA from these other possibilities. Finally, some immunocompromised patients with large KAs have developed metastasis of unclear clinical significance.[84] This scenario undoubtedly calls for additional investigation regarding clinical outcome and whether or not further treatment that would be directed towards cSCC is necessary to prevent further morbidity or mortality. Interestingly, no cases of death from a definite KA have been reported.[82]

Furthermore, there is a significant prognostic difference between perineural invasion in KA and cSCC. While perineural invasion is a poor prognostic factor in cSCC when involving a nerve in the subcutis > 0·1 mm in calibre, no metastases or direct deaths are attributable to the presence of perineural invasion in KA.[89,90] One study of 18 patients with KAs of the head and neck with invasion of nerves (ranging from 0·04 mm to 0·22 mm) treated with excision had no recurrences or metastases after a follow-up period ranging from 3 to 12 years.[90] Similarly, in another study, four patients with perioral KAs exhibiting perineural invasion were treated with excision and had no metastasis after follow-up periods of 7–44 months. Based on a larger study of 3465 KAs, perineural invasion had a reported incidence of 0·2%, while estimates for cSCC range from 2% to 14%.[17,51,85,89–92]

Treatment of Keratoacanthoma

Since the discovery of KA, management has remained controversial. Although a suspected regressing KA could be monitored for several months, it is difficult to predict the maximum size before the lesion regresses or how it will ultimately heal.[93] The current standard treatment of KA is that of a well-differentiated cSCC – surgical excision with clear margins, although this may be excessive given the dubious metastatic propensity of KA.[94] While tumours on the trunk and extremities can often be successfully excised with relatively little surgical morbidity, patients with lesions developing on sensitive areas or those with numerous lesions are more susceptible to increased surgical morbidity and disfigurement, such as ectropion.[36,38] Unfortunately, the skin of the head and neck is one of the most common areas affected, due to repeated sun exposure.[95] This problem is amplified in the at-risk elderly population, who commonly have facial lesions, frequent comorbidities and limited physiological reserve.[95,96] Other successful, less invasive treatment options for KAs are summarized in Table 2.[93,97–106] Based on the current literature, we recommend treating histopathologically definitive KAs conservatively after careful consideration to location, patient risk factors and associated procedure risks.