Dermoscopic characteristics of nodular squamous cell carcinoma and keratoacanthoma

Background: Nodular squamous cell carcinoma (SCC) and keratoacanthoma (KA) may mimic a variety of other benign and malignant non-pigmented nodules. Objectives: To analyze the dermoscopic characteristics of nodular SCC and KA. Patients/Methods: Retrospective analysis of 50 nodular SCCs and 8 KAs collected from a tertiary dermatology referral center and a private dermatology practice in Melbourne, Australia, between 1 September 2009 and 1 October 2012. All lesions were nodules; defined as firm, elevated, round, palpable tumors with a diameter of 5 mm or more. Clinical and dermoscopic images were evaluated by two examiners in consensus. Results: Signs of keratinization were frequently observed and included keratin crust/scale (90% of SCCs, 100% of KAs), central keratin mass (32% of SCCs, 88% of KAs), white structureless areas (66% of SCCs, 50% of KAs), white circles (32% of SCCs, 38% of KAs) and white keratin pearls (14% of SCCs, 12% of KAs). Hemorrhage was present in 72% of SCCs and 88% of KAs and preferentially occurred centrally and in areas of keratinization. For nodular SCCs and KAs, we observed glomerular vessels (42% and 25% respectively), linear irregular vessels (36% and 25% respectively), atypical vessels (30% and 38% respectively) and hairpin vessels (30% and 25% respectively). Conclusions: Hemorrhage, keratinization and vascular features (glomerular, hairpin and linear irregular morphologies) are useful in diagnosing both nodular SCC and KA. Further research on the comparative dermoscopic characteristics of a range of amelanotic nodules is important in order to improve diagnosis of these clinically challenging tumors.


Introduction
Squamous cell carcinoma (SCC) is the second most common cutaneous malignancy after basal cell carcinoma (BCC), with an increasing incidence worldwide [1,2]. It may present in a variety of morphologies, including as a keratinizing nodule, which may be clinically indistinguishable from keratoacanthoma (KA). Although traditionally diagnosed clinically,

Results
Fifty cases of nodular SCC and 8 cases of KA were collected.
Keratin crust/scale was observed in the vast majority of cases (90% of SCCs and 100% of KAs) ( Table 1).
A central keratin mass was present in 32% of nodular SCCs and 88% of KAs. White structures were common, in decreasing frequency we observed; white structureless areas (66% of SCCs and 50% of KAs), white circles (32% of SCCs and 38% of KAs) and white keratin pearls (14% of SCCs and 12% of KAs). We observed that the keratin pearls were often clustered, forming a mosaic pattern of indented round foci of keratin. Hemorrhage was observed in 72% of SCCs and 88% of KAs and tended to be present centrally in areas of keratinization. Collarette surrounded 12% of SCCs and 25% of KAs.
The vascular pattern was polymorphic in 50% of nodular SCCs and 38% of KAs. Vessels were not seen in 4% of SCCs. Glomerular vessels were the most common vessel type and were observed in 42% of SCCs and 25% of KAs. For nodular SCCs and KAs, we also commonly observed linear irregular vessels (36% and 25% respectively), atypical vessels (30% and 38% respectively) and hairpin vessels (30% and 25% respectively); 71% of hairpin vessels were positioned peripherally with a radial arrangement. nodular SCC and KA may mimic a variety of other benign and malignant nodules. In the context of a new or growing non-pigmented nodule, the differential diagnosis may include nodular BCC, hypertrophic intraepidermal carcinoma, atypical fibroxanthoma, Merkel cell carcinoma and nodular or desmoplastic melanoma. The differing prognostic and therapeutic implications of each of these diagnoses make their distinction important. However, misdiagnosis is common, and a recent study found that misdiagnosed nodular melanoma was mistaken for nodular SCC in 38% of cases [3].
Dermoscopy is an important in-vivo, non-invasive diagnostic technique that permits visualization of morphological features not visible with the naked eye. It greatly enhances the diagnostic accuracy for pigmented skin lesions [4][5][6].
As non-pigmented nodules are a commonly encountered clinical diagnostic dilemma, this study sets out to describe the dermoscopic features of nodular SCC and KA. All lesions were excised and histopathology was reviewed to confirm the diagnosis. The diagnosis of KA was based on histopathological architecture and pattern of cell differentiation [19]. All lesions were nodules; defined as firm, elevated, round, palpable tumors with a diameter of 5 mm or more [20].

Materials and methods
Digital dermoscopic images were captured with a dermatoscope (DermLite DL3 dermatoscope, Heine, Herrsching, Germany) mounted on a digital camera (Cyber-shot DSC-W290, Sony Corporation, Tokyo, Japan). Tenfold magnification was used. Alcohol gel was used for immersion and precautions were taken to reduce compression artifact.
Lesions were excluded if the image quality was unsatisfactory.
Clinical and dermoscopic images were reviewed retrospectively. An initial meeting was held where a sample of 15 cases was scored and the literature reviewed to establish a tiple colors were present in all lesions, with 41% having 5 or 6 colors. Pigmented structures or blue-gray veil were not seen.
For nodular SCCs and KAs, the primary dermoscopic color was pink (62% and 75% respectively), white/yellow (32% and 25% respectively) and red (6% of SCCs). Mul-  cell carcinomas [24]. Given that collarette is not uncommonly seen in rapidly growing nodules, it is probably not a useful distinguishing feature.  White lines 6% 0% n/a n/a n/a n/a Collarette 12% 25% n/a n/a n/a n/a

Prior presentation
This study was presented in part at the 8th World Congress of Melanoma, Hamburg, Germany, July 17-20, 2013.
KAs developed a more polymorphic vascular pattern with an increased frequency of hairpin and linear irregular vessels.
When faced with a pink or red nodule, one of the most important diagnostic decisions to make is the distinction between a nodular SCC and a clinically non-pigmented nodular melanoma. Pigmented structures and blue-gray veil are important positive dermoscopic features that strongly favor the diagnosis of hypomelanotic nodular melanoma and these were both negative dermoscopic features in our series of nodular SCCs and KAs [25,26]. However, the diagnosis of truly amelanotic nodular melanoma becomes particularly challenging because it also lacks dermoscopic features of pigmentation.
Although none of the cases in our series contained pigmented structures, pigmented SCC is a recognized entity. The clinical and dermoscopic diagnosis of this rare tumor have been described in several case reports and the presence of keratin scale might be useful clue for the diagnosis [27][28][29][30][31][32].
Another common differential diagnosis of nodular SCC is nodular BCC. In our series, cases did not display the characteristic arborizing vessels associated with BCC [33,34].
Conversely, BCC rarely contains the glomerular, hairpin or linear irregular vessels, which were frequently observed in nodular SCC and KA.
Seborrheic keratosis may also be confused with nodular SCC. Dermoscopically, keratin and hairpin vessels are common to both, however, hairpin vessels have been found to be more predictive of seborrheic keratosis [33]. Seborrheic keratosis may also be pigmented and have milia-like cysts and comedo-like openings which aid diagnosis [35].
Merkel cell carcinoma is a rare tumor that may also mimic nodular SCC. Both are non-pigmented nodules that commonly contain linear irregular vessels [36,37]. However, other vascular features may help discriminate, with hairpin vessels favoring the diagnosis of SCC and arborizing vessels favoring the diagnosis of Merkel cell carcinoma. Hyperkeratosis is typically absent from Merkel cell carcinomas, which also tend to have a shinier cherry red appearance [38].
There were several limitations to this study. Firstly, the two examiners of images were not blinded to the diagnosis of SCC or KA. Secondly, we did not include nodules other than SCC and KA and the study was not designed to test the sensitivity nor specificity of dermoscopic criteria in differentiating nodular SCC and KA from other nodular lesions. Finally, the study was not designed to determine if dermoscopy alters the naked eye diagnosis of a nodular SCC or KA. For example, in Figure 1, lesions A-F appear as clinically keratinizing nodules, where the major differential diagnosis on naked eye examination would be SCC or KA and it is unlikely that dermoscopy would alter diagnosis in these cases. On the other hand, lesions G-I lack clinical clues of keratinization and it is here that dermoscopy may become useful in diagnosis. However,