Introduction

Basal cell carcinoma (BCC) is the most common skin tumor nowadays, with approximately 2.8 million new cases a year in the USA, with tendential increasing [1]. In Europe, a recent Dutch study confirmed the growing tendency with a lifetime BCC risk of 1 in 5 (21 %) men and 1 in 6 (18 %) women [2]. Different histological subtypes of BCC have been described: nodular, morpheaform, infiltrative, superficial, ulcerated, baso-squamous carcinoma, and combinations of these subtypes. In advanced cases, removal approach is the first choice, with Mohs micrographic surgery as the gold standard [3]. However, for early and superficial BCCs, topical imiquimod or photodynamic therapy offer good therapeutic alternatives [4].

An early and correct identification of BCCs enables correct treatment and avoids unnecessary excisions; in case of diagnostic uncertainty, however, a pre- and sometimes even post-treatment biopsy is necessary in order to ensure the correct diagnosis.

The role of dermoscopy in the diagnosis of BCCs and prediction of their histological subtype has been widely demonstrated in the literature. Beginning with Menzies’ model, the combination of easily recognizable diagnostic features (absence of pigment network, leaf-like structures, spoke-wheel areas, large blue-gray ovoid nests, blue-gray globules, arborizing teleangiectasias, and ulceration) allowed a sensitivity of 93 % and a specificity of 89 % in the differential diagnosis with melanoma [5]. Various other studies reported promising results, which analyzed the features of pigmented and superficial basal cell carcinomas [5–12].

In the daily clinical routine, however, some cases raise the suspicion of BCC but remain unclear, with non-yelding dermoscopy. Especially in facial lesions of uncertain nature, the effective treatment might be delayed until progression to more invasive stages. Non-invasive tools, such as reflectance confocal microscopy (RCM) have the potential of ensuring earlier diagnosis and treatment.

Not only dermoscopy, but also RCM has shown its utility in the non-invasive diagnosis of BCC. In particular, RCM showed diagnostic sensitivity of 97 % and specificity of 93 % [13–17] and provided good correlations with the histological subtypes [18, 19].

The aim of our study was to establish the role of RCM in the diagnosis of BCCs in cases of difficult to diagnose facial BCCs, with indistinct clinical and dermoscopical features.

Materials and methods

We retrospectively examined 31 BCCs that had been screened in the Dermoscopy and Imaging outpatient clinic of the Department of Dermatology and Allergology of the Ludwig-Maximilian University of Munich between January 2014 and September 2015. We then selected 27 histologically proven BCCs which shared the absence of striking clinical features based on naked eye examination; we separately reviewed in a blinded manner the clinical, dermoscopic, and confocal microscopy features and evaluated the lesions meeting the common diagnostic criteria for BCCs.

Only lesions with complete macroscopic, dermatoscopic, and confocal documentation; histological diagnosis; and follow-up were included in the study. The selected patients were 12 males and 15 females with a mean age of 67 years. The tumors were located on the nose (63 %), on the cheeks 15 % while the remaining ones arose on other facial sites. Eleven (41 %) BCCs developed within a scar.

The dermoscopic images were captured with Fotofinder (FotoFinderSystems GmbH, Germany), at ×10 field magnification. RCM images were acquired using Vivascope 1500 (Mavig, Germany), a non-invasive diagnostic tool based on an 830-nm diode laser and a ×30 objective lens with a numerical aperture of 0.9. Its maximum laser power is 40 mW, which causes no tissue damage nor pain. Its penetration depth is up to 250–300 μm, its axial resolution 3–5 μm and its lateral resolution 1 μm. Three different layers corresponding to the epidermis, dermo-epidermal junction, and superficial dermis were examined for each lesion.

Clinical features were nodular structure, pearl-shiny borders, pigmentation, ulceration, raised borders, scaling, crusting, teleangiectasias, pink color, and fibrous-scarring appearance.

Dermoscopic features were selected according to the literature: arborizing vessels, leaf-like structures, blue-gray ovoid nests, multiple blue-gray dots, spoke-wheel structures, concentric structures, fine telangiectasias, micro-ulceration/multiple small erosions, shiny white to red structureless areas, and yellow stripes [7, 10, 12, 19]. Confocal parameters were defined based on already published data: ulceration, streaming of the epidermis, tumor islands, cord-like structures, dark silhouettes, peripheral palisading, peripheral clefting, onion-like structures, stromal reaction, presence of inflammatory (bright) particles and cells, dendritic cells, and increased vascularization (dilated vessels) [5, 15, 18–23].

Since the examination had no implications in the diagnostic process nor in the choice of therapeutic strategies, the approval of the ethics committee was not required.

Results

Of the examined tumors, the majority was of the nodular histological subtype (10); other histological subtypes were: superficial (7), morpheaform (7), infiltrative (1), and ulcerated (2). The mixed types were classified based on the most prevalent component.

All lesions were clinically unclear, with pink color (15) and fibrous-scarring appearance (10) as the most common features; other characteristics were nodular structure (7), pigmentation (2), ulceration (2), and scaling (1). In most cases, telangiectasias were present but not distinguishable from the background ones. The typical raised pearly-shine borders were not visible in any of the lesions. Dermoscopy was non-diagnostic and showed in most cases: unspecific teleangiectasias (20), arborizing vessels (11), and micro-erosions (14); other features were: blue-gray dots (6), blue-gray ovoid nests (1), and shiny white to red structureless areas (10); the typical leaf-like structures, spoke-wheel areas, concentric structures, and yellow stripes could not be detected.

In contrast, confocal microscopy revealed the presence of typical basal cell carcinoma criteria: peripheral palisading of the nuclei (24), streaming of the epidermis (22), dark silhouettes (19), stromal reaction (19), clefting (19), inflammatory particles (19), tumor islands (18), cord-like structures (9), ulceration (8), dendritic structures (7), and canalicular vessels (17).

The most common clinical features were pink color (55 %) and fibrous-scarring appearance (37 %). In videodermoscopy, the BCCs examined showed teleangiectasias (74 %), micro-ulcerations (52 %), and arborizing vessels (41 %). Confocal microscopy revealed peripheral palisading in 89 % of the cases, streaming of the epidermis (81 %), dark silhouettes (70 %), tumor islands (67 %), and stromal reaction, clefting, and inflammatory particles (70 %) (Table 1) (Figs. 1 and 2).

Table 1 Prevalence of clinical, dermoscopic, and confocal microscopy features expressed as relative (%) and absolute frequency (n)
Full size table
Fig. 1
figure 1

From left to right: clinical, dermoscopical, confocal, and histological detail of our lesions. a 73-year-old male with nodular BCC, note well-circumscribed tumor islands with clefting and palisading. b 75-year-old female with superficial BCC. c 36-year-old female with nodular BCC and infiltrative growth pattern, note tumor islands surrounded by inflammatory infiltrate. d 73-year-old male with recurrence of superficial BCC within a scar

Full size image
Fig. 2
figure 2

From left to right: clinical, dermoscopical, confocal, and histological detail of lesions. a 78-year-old male with superficial BCC. Note dark silhouettes surrounded by collagen in RCM. b 67-year-old male with nodular BCC. Note the tumor island in RCM corresponding to the basaloid island surrounded by stromal reaction in histology. c 75-year-old female with recurrence of superficial BCC on her nose. Note typical cord-like structures in RCM. d 49-year-old female with recurrence of nodular basal cell carcinoma on her nose

Full size image

Discussion

The clinical eye has been the only tool in the diagnosis of skin cancer for many years. There are few reports on the efficacy of naked eye examination in the diagnosis of BCC, which confirm, however, that the clinician’s ability to make a proper diagnosis is strictly dependent on his clinical experience and tumor stage. The positive predictive value for clinically diagnosing BCC ranged from 43 to 80 % when using main clinical criteria such as pearly borders, teleangiectasias, and pigmentation; in most cases, however, the patients were preselected among those referred to the study centers with suspicion of BCC [24–28]. In the absence of these clinical criteria, however, the diagnostic confidence was very low.

Since the beginning of the dermoscopy era in the 1980s, huge progress has been made in the early diagnosis of skin cancer, and diagnostic accuracy has reached a significative improvement, both in melanoma and non-melanoma skin cancer [29–33]. The role of dermoscopy in the early diagnosis of BCCs and prediction of their histological subtype has been validated in numerous studies. Sensitivity and specificity of dermoscopy in differentiating BCCs from other skin tumors are quite high, ranging from 95–87 to 96–97 %, respectively [5, 10, 34]. Furthermore, distinct dermoscopic criteria for superficial and other subtypes of BCC have been described, identifying leaf-like structures and short fine superficial telangiectasias in the absence of arborizing vessels, large blue-gray ovoid nests, or ulceration as predictors of superficial BCC with a sensitivity of 81.9 % and a specificity of 81.8 %. Ovoid nests, arborizing vessels, and large ulcerations were found in nodular and infiltrative BCCs; the latter were identified by shiny white-red structureless areas and small arborizing vessels with reduced branching tendency [7, 9, 34, 35].

The literature has, however, has shown significant variability in the prevalence of each specific dermoscopic feature for the diagnosis of BCC, with remarkable differences among studies, 28–81 % for arborizing vessels, 20–84 % for blue-gray ovoid nests, and 27–42 % for ulceration [5–12]. Moreover, while the diagnostic accuracy of pigmented BCCs has been widely tested, the findings concerning the diagnosis of non-pigmented BCCs are less specific, although arborizing vessels and micro-ulcerations are considered the most predictive features [10, 36].

In the daily clinical practice, however, the analysis of clinical and dermoscopic criteria might not be sufficient to ensure the diagnosis of BCC in lesions where the abovementioned parameters are lacking. In high-risk patients, when recurrence is suspected or when a lesion is enlarging, a skin biopsy should be taken, with the risk of overtreatment or low compliance.

In those cases, a significant help in improving the diagnostic accuracy of both clinical examination and dermoscopy is provided by RCM. The device, an infrared light microscope based on a diode laser of 830 nm, allows the visualization of skin structures with a cellular resolution and thus performing a kind of in vivo biopsy [37]. The use of in vivo RCM has been widely tested in the diagnosis and monitoring of many skin pathologies, including treatment monitoring of BCC under oral hedgehog inhibitors [38], with a large number of studies examining melanoma and non-melanoma skin cancer. In the last years, an ex vivo confocal device has been developed and shown excellent correlations with histology and utility in the therapeutic management of basal cell carcinomas [39–41].

The diagnostic criteria for the confocal diagnosis of BCC have been firstly described by Gonzales et al., and their diagnostic accuracy has been subsequently validated by numerous studies [13–16, 20, 42–44]. In particular, Nori and colleagues [20] reported a specificity of 95.7 % and a sensitivity of 82.9 % using a set of five criteria (presence of elongated monomorphic basaloid nuclei, polarization of these nuclei along the same axis of orientation, prominent inflammatory infiltrate, increased vasculature, and pleomorphism of the overlying epidermis indicative of actinic damage) applied to 152 skin lesions, while Guitera and colleagues [43] obtained 100 % sensitivity and 88.5 % specificity on a case series of 710 lesions which also included malignant melanoma. The group developed a diagnostic algorithm for the differential diagnosis of the two skin tumors, with polarization of the cells in the epidermis, telangiectasias, convoluted vessels, basaloid islands, dark silhouettes, and clefting as the positive features for the diagnosis of BCC. In 2011, Ulrich and colleagues described the presence of peritumoral clefting as an additional criterion that correlates in vivo with the peritumoral deposition of mucin [44].

As for dermoscopy, specific RCM criteria related to the histological subtype have been described. An accurate prediction of the histological subtype is useful in clinical practice, since it helps in therapeutic planning. Longo and colleagues identified cord-like structures, together with clefting and peripheral palisading as the most common features in superficial BCCs, with dark silhouettes outlined by bright coarse collagen fibers often present in non-pigmented superficial BCCs and infiltrative BCCs. Nodular BCCs show bright tumor islands, peripheral palisading, clefting, and enlarged blood vessels with sometimes ulceration [19].

In our study, at least four confocal criteria for the diagnosis of BCC were found in every lesion, ensuring a diagnostic confidence able to guide the patient in the subsequent management. Confocal microscopy enabled the in vivo diagnosis although the typical clinical BCC parameters were absent, and the dermatoscopic features not specific enough to demonstrate or exclude malignancy. In particular, RCM showed high diagnostic accuracy in case of lesions arising on pre-existing scars with suspicion of recurrence, with non-informative clinical and dermatoscopic features. The RCM features in these clinically indistinctive BCCs were also those described in the literature as the most specific for the diagnosis of BCC and part of the diagnostic algorithms, so that the use of RCM had an instant impact on the further management of the lesions. Additionally, in some cases, although dermoscopical examination suggested superficial BCCs of the case series would have been classified as of the superficial types, which would lead to inappropriate treatment such as topical or photodynamic therapy, RCM parameters showed features of nodular and infiltrative types and guided the clinical decision-making to surgical excision. In our hands, RCM achieved a high-enough diagnostic confidence to directly address the patient to Mohs surgery, without the necessity of performing a skin biopsy. Thus, the method is rapid, cost effective, and non-invasive for the patient.

In conclusion, in facial papues with unclear clinical features and non-diagnostic dermoscopy, confocal microscopy aids in the correct diagnosis and helps to avoid unnecessary biopsy and follow-up of the so-called invisible BCCs.