Dermoscopic and reflectance confocal microscopy features of cutaneous squamous cell carcinoma

Squamous cell carcinoma (SCC) of the skin is a highly prevalent neoplasm. The management and the prognosis of this tumour are dependent on its invasiveness and its grade of differentiation.


Introduction
Squamous cell carcinoma (SCC) of the skin is responsible for 20% of skin malignancies. [1][2][3] Recent guidelines suggest that the management of high-risk SCC should be as early and aggressive as possible, with surgery considered the optimal choice. 4,5 Tumour aggressiveness and risk of recurrence depend on many factors, in detail: patient's immune efficiency, body site, tumour size, invasion into the subcutaneous tissue, perineural involvement and the grade of histopathological differentiation. [6][7][8] Poor differentiation is an independent risk factor for recurrence, metastasis and disease-specific death. [4][5][6][7][8] In contrast, well-differentiated SCC is associated with a 5-year recurrence-free survival rate of 83%. [4][5][6][7][8] Surgical biopsy and histologic examination are the gold standard for the diagnosis of SCC. 2 However, there is mounting evidence that dermoscopy and reflectance confocal microscopy (RCM) are useful tools for the bedside diagnosis of AK and SCC, with high sensitivity and specificity values. Dermoscopic criteria have been described for SCC, including keratin, scale, blood spots, white circles, white structureless zones and perivascular white halos. 9,10 In previous studies, keratin and white circles reached a diagnostic sensitivity and specificity for SCC of 79% and 87%, respectively. In addition, keratin, white circles, structureless whitish areas and scales with central distribution were shown to be associated with well-or moderately differentiated tumours. In contrast, poorly differentiated SCC revealed a predominantly red colour, resulting from the presence of bleeding and/or dense vascularity, in the absence of scale/keratin or other white-coloured criteria. 9 Reflectance confocal microscopy represents an add-on tool for the non-invasive diagnosis and management of SCC. Recent studies demonstrated that RCM image analysis performed by trained (expert) readers achieved sensitivity values ranging from 80.0 to 93.34% and specificity values ranging from 88.34 to 98.6%. 11-13 RCM grade of honeycomb atypia was highly correlated with the histopathological assessment of keratinocyte atypia in AK. Ulrich et al. found that the most common RCM findings of Bowen's disease (BD) were the disruption of the stratum corneum, an atypical honeycomb pattern in the epidermis, S-shaped blood vessels in the centre of the dermal papillae and 2 types of characteristic targetoid cells. 14 Regarding invasive SCC, large and comprehensive RCM studies are still lacking. Small population-sized studies identified that main RCM pattern of invasive SCC is a disarranged or atypical honeycomb pattern in the epidermis, round nucleated bright cells in the suprabasal epidermis and looping blood vessels in dermal papillae. 15,16 The aim of our study was to define the frequencies of the main RCM criteria for the diagnosis of invasive and in situ SCC and their correlations with the histologic grade of differentiation.

Materials and method
Squamous cell carcinoma cases were retrospectively collected in two centres in Italy (University of Modena and Reggio Emilia, and at the Arcispedale Santa Maria Nuova in Reggio Emilia). Ethics committee approval was waived because the study affected neither the routine diagnostic nor therapeutic management of these cases. Inclusion criteria were a definite histopathological diagnosis of SCC, including subtype classification, the availability of clinical, dermoscopic and RCM images of the tumour, and the availability of histopathological slides. Dermoscopy, RCM and histologic examination were performed as standards of care in our centres. The pathologic examination was conducted following the routine procedures: all lesions were defined as in situ SCC (intraepidermal carcinoma or Bowen's disease) or invasive SCC, based on the presence of an invasive component of the tumour, and invasive SCCs were classified into 'poorly differentiated', 'moderately differentiated' and 'well differentiated', based on the histologic grading of the tumour. Actinic keratosis was excluded from the study. Cases in which RCM images could not be evaluated because of poor image quality or the presence of extensive ulceration or hyperkeratosis were excluded. Dermoscopic images were captured by means of DermlitePhoto equipment (3Gen, Dana Point, CA) at 10-fold magnification. RCM images were acquired by means of a Vivascope 1500 or Vivascope 3000 (Caliber ID, Rochester, NY), which uses an 830-nm laser beam with a maximum power of 20 mW. Instrument and acquisition procedures have been described elsewhere. 17,18 Patient demographics and tumour characteristics were recorded, and two independent investigators (E.M. and M.M.) evaluated all clinical, dermoscopic and RCM images. Both were blinded to the clinical and histopathological diagnosis. If the two investigators failed to reach a consensus, a third investigator was involved (C.L.). Dermoscopic and RCM variables were selected on the basis of previously published data on SCC and our preliminary observations (Table 1).

Statistical analysis
The analysis was conducted in order to assess whether any dermoscopic or RCM criteria were associated to the different SCC types. The statistical analysis comprised descriptive statistics and Pearson's chi-squared test to analyse the different subgroups. The compared categories were as follows: 1) invasive vs. in situ SCC and 2) well, moderately and poorly differentiated invasive SCC. All statistical calculations were made with SPSS 17.0 (IBM, Armonk, NY, USA).

Results
143 SCCs from 143 patients were retrieved from the databases of the two academic centres; 121 had a complete set of dermoscopic and RCM images (mean age 78.79 years old), including 77 men and 45 women. The remaining 22 SCCs that were excluded from the study (22/143, 15.3%) either lacked some dermoscopic or RCM images or showed abundant hyperkeratosis or ulceration (>85% surface area) that hampered the quality of the images and their evaluation.
Upon RCM, invasive SCCs were significantly characterized by the presence of erosion/ulceration, architectural disarrangement, speckled nucleated cells in the dermis and absence of hyperkeratosis. Regarding the vascular features, the presence of buttonhole vessels was significantly associated with in situ SCC, while irregularly dilated vessels were associated with invasive SCC (Fig. 1). The assessment of poorly differentiated and other invasive SCC revealed that the RCM images of poorly differentiated SCC were characterized by architectural disarrangement, while well-or moderately differentiated SCC showed a more preserved architecture and the presence of speckled nucleated cells in the epidermis (Fig. 1).
From a dermoscopic point of view, in situ SCCs were characterized by the presence of white structureless areas and dotted or glomerular vessels. Polymorphic vessels were associated with invasive SCC (Fig 2). The comparison of the dermoscopic images of poorly differentiated and other invasive SCC showed that the predominance of red areas and the presence of erosion/ulceration were associated with poor differentiation. Conversely, the presence of white areas was associated with well-differentiated or moderately differentiated tumours (Fig. 2). Roundish to polygonal cells with speckled appearance and a dark nucleus within the epidermis. Their size is slightly larger than to the one of the surrounding keratinocytes. They are larger than the usual size of lymphocytes and have a polygonal shape that differentiate them from dendritic cells.

Speckled nucleated cells in the dermis
Roundish to polygonal cells with speckled appearance and a dark nucleus within the dermis. They are larger than the usual size of lymphocytes and have a polygonal shape that differentiate them from plump bright cells.

Targetoid cells
Large cell with a bright centre and a dark peripheral halo or a dark centre and a bright rim surrounded by a dark halo.

Discussion
In the current study, we focused on the dermoscopic and RCM features of squamous cell neoplasia, in order to correlate the presence of specific descriptors with invasiveness and histologic grade of differentiation. In our analysis, invasive SCC were characterized by erosion/ulceration, architectural disarrangement, speckled nucleated cells in the dermis and irregularly dilated vessels upon RCM examination. These features morphologically reproduce the deregulated growth of the tumour that develops necrotic areas on the epidermal surface and invades beyond the dermo-epidermal junction with abundant neo-angiogenetic phenomena. We introduced the term 'speckled nucleated cells' to define roundish to polygonal cells with speckled appearance and a dark nucleus. Their size is slightly larger than to the one of the surrounding keratinocytes and they can be differentiated from inflammatory because they are larger than the usual size of lymphocytes and have a polygonal shape that is different from the one of dendritic cells or plump bright cells. 19,20 On the other hand, in situ SCCs were characterized by a rich hyperkeratotic component and by the presence of buttonhole vessels inside dermal papillae. Architectural disarrangement was not as marked as in the invasive form, and there was no sign of invasion beyond the dermal epidermal junction. Our dermoscopic findings are consistent with previous studies reporting on the dermoscopic criteria of SCC. 9,10 In our group of patients, in situ SCCs were characterized by the presence of white structureless areas and dotted or glomerular vessels. Instead, polymorphic vessels were associated with invasive SCC. The second key point of our study was the identification of RCM or dermoscopic descriptors specific of poorly differentiated tumours. Among invasive SCCs, the recognition of poor differentiated neoplasms is of great clinical and therapeutic relevance. 4,21,22 It was demonstrated that this feature is associated with poor prognosis, with higher relapse and nodal involvement rates. [4][5][6][7][8] Even though, given the rarity of this type of neoplasm, only a small number of poorly differentiated invasive SCC were included in the study, we observed that poorly differentiated SCC are characterized by a massive architectural disarrangement and by the absence of round nucleated cells in the epidermis. These features reflected a more chaotic growth and a complete loss of the features associated with the regular epidermal differentiation. Furthermore, the identification of severe atypia of the honeycomb pattern that was assessed based on previous RCM studies 23 was associated neither with differences between invasive and in situ SCC, nor with the histologic grade of the invasive tumours. This is probably due to the fact that differently from AKs, the honeycomb atypia is often severe in all the SCC, either in situ, invasive, well, moderately or poorly differentiated. Dermoscopically, as previously reported by Lallas and colleagues, 9 we confirm that there is a predominance of red colour and of erosion/ulceration in the poorly differentiated tumour, and of white structureless areas in well-or moderately differentiated tumours SCC.

Conclusions
Clinical, dermoscopic and RCM information should be integrated in order to achieve the optimal therapeutic management for the patient. The clinical information should include tumour size, precise body site location, concurrent scars or chronic inflammation, presence of previous SCC, treatment failure and immunosuppression. [5][6][7][8]24 Based on our findings, dermoscopy and RCM imaging can allow a more accurate pre-surgical assessment of SCC. Further studies are needed in order to define the best surgical margins or therapeutic approaches in relation to the presence of the different RCM patterns.