Specific Dermoscopy Patterns and Amplifications of the Cyclin D1 Gene to Define Histopathologically Unrecognizable Early Lesions of Acral Melanoma In Situ

Objective  To define early lesions of acral melanoma in situ that cannot be recognized histopathologically.

Design  A retrospective review of the clinical, dermoscopic, and histopathological findings.

Setting  University department of dermatology.

Patients  Thirty-three patients with melanocytic lesions on acral volar skin that were clinically suspected of being early melanomas.

Main Outcome Measures  Fluorescent in situ hybridization studies to detect the cyclin D1 gene amplification in proliferating melanocytes, which is a characteristic genetic aberration recently found in acral melanoma.

Results  Seventeen of 33 lesions were histopathologically diagnosed as either melanoma in situ (8 lesions) or benign melanocytic nevi (9 lesions). Amplification of the cyclin D1 gene was observed in 2 (25%) of the 8 melanomas in situ. None of the 9 nevi showed the amplification. The remaining 16 lesions were, however, difficult to classify histopathologically because most of them showed only a slight increase of nonatypical melanocytes in the basal cell layer of the epidermis. On dermoscopic examination, 9 of these 16 lesions exhibited the parallel ridge pattern that has been reported to be highly specific to melanoma in situ, and 4 (44%) of them had amplifications of the cyclin D1 gene. Amplifications were not found in any of the remaining 7 lesions that showed dermoscopic patterns suggestive of melanocytic nevi.

Conclusions  Cyclin D1 gene amplification detected by fluorescent in situ hybridization identified a very early progression phase of acral melanoma that precedes histopathologically defined melanoma in situ. The present study also indicates the specificity of the parallel ridge pattern on dermoscopy to detect melanomas on acral volar skin at such a very early developmental phase.

To improve the prognosis of patients with malignant melanoma, the early detection and correct diagnosis of lesions at the in situ phase are mandatory. In nonwhite populations, such as the Japanese, the diagnosis of pigmented skin lesions on acral sites, especially on the palms and soles, is important because nearly half of all cutaneous melanomas affect these sites.¹ In the past 2 decades, our group has characterized the clinical and histopathological features of malignant melanomas affecting the soles of the feet and has proposed several clinical and histopathological guidelines for the early detection of plantar melanomas.^2-4 More recently, we and others^5-7 introduced dermoscopic evaluation in the diagnosis of the pigmented lesions on acral volar skin and found that the recognition of several dermoscopic patterns is immensely helpful in differentiating melanomas from benign melanocytic nevi. These guidelines and criteria are quite effective in detecting early melanomas affecting acral volar skin. In the course of these investigations, however, we confronted diagnostic problems with several pigmented lesions on the palms and soles. These lesions showed clinical features suggestive of melanoma in situ, such as late onset, a large size (maximum diameter exceeding 7 mm), and dermoscopic findings of the parallel ridge pattern (PRP).⁸ However, histopathological examinations of these cases revealed only a slight increase of melanocytes, with no (or minimal) cytological atypia in the basal cell layer of the epidermis. Similar cases were reported⁹ previously in Japanese populations and were diagnosed as atypical melanosis of the foot. Although these lesions may represent a very early stage of melanoma in situ, currently there is no evidence for this.

Recently, Bastian et al^10-12 performed comparative genomic hybridization analyses in cutaneous melanomas and found that, in contrast to other types of cutaneous melanomas, acral melanomas had a high frequency of focused gene amplifications occurring at chromosome regions 11q13, 22q11–q13, 5p15, and 12q14. Furthermore, their investigation of the invasive and noninvasive portions of acral melanomas and acral melanomas in situ using fluorescence in situ hybridization (FISH) revealed that these amplifications occurred early in the radial growth phase of the acral melanomas before the formation of the invasive portion. Intriguingly, the amplifications were detected even in isolated melanocytes in the epidermis beyond the histopathologically recognizable extent of intraepidermal melanoma lesions, suggesting that such focused gene amplifications occur in a very early developmental phase of acral melanoma when the histopathological atypia is not obvious.¹¹ The most frequently amplified region in acral melanomas was 11q13, which contains the cyclin D1 gene (CCND1). A subsequent study¹³ revealed that the amplification of CCND1 was found in 44% of acral melanomas.

We reasoned that, based on these recent molecular cytogenetic findings, in addition to the clinical, histological, and dermoscopic findings, the identification of CCND1 amplification by FISH could be of help in classifying pigmented lesions on the palms and soles, especially in the diagnosis of early lesions of acral melanoma in situ. In the study described herein, we performed FISH analyses to detect CCND1 amplifications in 33 melanocytic lesions on acral volar skin that were clinically or dermoscopically suspected to be early melanoma.

From 1997 to 2004, following Saida’s⁴ guideline that encouraged the detection of early melanoma on the soles, we excised a total of 30 pigmented lesions on the acral volar skin of patients at the Department of Dermatology, Shinshu University Hospital, Matsumoto, Japan. The guideline recommends surgical excision of acral melanocytic lesions for histopathological evaluation when a lesion has 1 of the following clinical or dermoscopic features: (1) a maximum diameter exceeding 7 mm, (2) a highly irregular shape and/or color, or (3) a PRP revealed on dermoscopic examination. We obtained written informed consent from each patient. We included in the study 3 referral cases in which lesions suspected to be early melanomas had been excised at other hospitals. We excluded from the study cases of invasive acral melanomas, which are easily diagnosed clinically, as well as cases of congenital melanocytic nevi. Dermoscopic examinations were performed as described previously.⁶ The excised specimens were fixed in 10% buffered formalin and embedded in paraffin. All the pathology slides were evaluated using the standard criteria,¹⁴ and the difficult cases were reviewed by an expert dermatopathologist (T.S.).

Amplification of the CCND1 was examined in paraffin-embedded tissue sections using a dual-color FISH technique as described elsewhere.¹⁵ The dual-color probe mixture consisted of spectrum orange LSI CCND1 probe (band region, 11q13) and spectrum green CEP11 (Central Enumeration Probe for chromosome 11; Vysis Inc, Downers Grove, Ill) (band region, 11p11.1-q11; locus D11Z1).

The FISH signals were scored on a fluorescent microscope (Carl Zeiss, Oberkochen, Germany) equipped with a triple band-pass filter. Three-color images were captured by using a digital imaging analysis system (ISIS, version 5.0; Carl Zeiss). Copy numbers of CEP11 (green) and CCND1 (red) signals, respectively, were counted for at least 50 nonoverlapping proliferating melanocytes in the epidermis. Melanocytes were easily identified under a fluorescent microscope as cells sitting in the basal cell layer of the epidermis with cytoplasmic retraction. Because the FISH analyses were performed using paraffin sections, nuclear truncation by sectioning, as well as nonspecific staining, could potentially confound the scoring. Thus, we also counted the target and reference signals for more than 50 keratinocytes as controls of hybridization efficiency and specificity. The mean ± SD of the target–reference signal ratio in keratinocytes in 33 cases was 1.02 ± 0.06, indicating the feasibility of the scoring method. The lesions were considered to have CCND1 amplification if they fulfilled 2 criteria^16,17: (1) the ratio of the total number of CCND1 probe signals to the total number of centromere signals in all scored melanocytes was greater than 1.5, and (2) the mean copy number of CCND1 in melanocytes was significantly greater than that in keratinocytes. Statistical comparison was made with the unpaired t test, and P<.01 was regarded as significant. In addition, the percentage of melanocytes showing more than 2.5 times more test-probe signals than reference signals was also recorded for each specimen.

A total of 33 pigmented lesions on acral volar skin, excised under the suspicion of being early melanomas according to the clinical guideline proposed by Saida,⁴ were first classified into 3 groups based on the histopathological features (Figure 1 and Table). Eight lesions (group 1) showed the typical histopathological features of melanoma in situ (Figure 2), and 9 lesions (group 3) showed the features of benign melanocytic nevi (5 junctional nevi, 2 compound nevi, and 2 dermal nevi). In these 17 cases, the dermoscopic findings were mostly concordant with the histopathological diagnoses. All 8 cases designated as melanoma in situ showed either the PRP or irregular diffuse pigmentation, both of which are highly specific to melanoma in situ on acral volar skin.^6,8 Nine nevi showed benign dermoscopic patterns such as a parallel furrow pattern, a fibrillar or filamentous pattern, or a latticelike pattern.^5,7

However, the remaining 16 lesions (group 2) were difficult to classify histopathologically; 13 showed only a slight increase of nonatypical melanocytes in the basal cell layer of the epidermis (Table, cases 9- 21; Figure 3). These lesions could be a very early stage of melanoma in situ, but they could not be distinguished histopathologically from benign lentigo simplex. In the other 3 cases (Table, cases 22-24), small nests of melanocytes were found within the epidermis, associated with a marked proliferation of individual melanocytes in the basal cell layer, as well as in the lower portion of the spinous layer. These cases might have been either melanoma in situ or acral-lentiginous junctional nevi,¹⁸ but histopathological differentiation was also difficult. These 16 lesions with uncertain histopathological classifications were further subdivided into 2 groups based on the dermoscopic findings; 9 lesions (group 2A) showed PRP, and the remaining 7 lesions (group 2B) showed patterns suggestive of melanocytic nevi such as a parallel furrow pattern, a fibrillar or filamentous pattern, or a latticelike pattern (Table and Figure 1).

Amplification of CCND1 was detected with FISH in 2 (25%) of 8 lesions in group 1 (Figure 2) and 4 (44%) of 9 lesions in group 2A (Figure 3). No lesions in groups 2B or 3 showed CCND1 amplification. In 2 cases in group 1 (cases 1 and 2) with amplifications, about half of the melanocytes had more than 2.5 times more CCND1 probe signals than reference signals (Table and Figure 2). In case 9 in group 2A, nearly all melanocytes showed amplifications. In cases 10, 11, and 12 in group 2A, about 25% to 37% of melanocytes showed a CCND1 copy number increase. The findings in case 10 were most intriguing because the maximum diameter of a light brown macule on the sole of a 60-year-old woman was only 6 mm. However, the dermoscopic examination revealed the typical PRP. Although the histopathological findings showed only a slight increase of nonatypical melanocytes in the basal layer of the epidermis, amplification of CCND1 was observed with FISH in the epidermal melanocytes (Figure 3).

In this study, 33 melanocytic lesions on acral volar skin clinically or dermoscopically suspected of being early melanoma were histopathologically reviewed. Although half of the lesions were unambiguously classified as either melanoma in situ or benign melanocytic proliferations such as melanocytic nevus and lentigo simplex (groups 1 and 3), the remaining lesions (group 2) posed problems in histopathological diagnosis. Despite their atypical clinical features, such as a large size and a variegated color and/or irregular shape, most of these lesions showed minimal histopathological changes that did not provide evidence of malignancy. These lesions were essentially the same as those reported previously by Nogita et al.⁹ They distinguished these lesions from acral melanoma in situ by the lack of histopathological evidence of malignancy and designated them as “atypical melanosis of the foot” with undetermined biological behavior. However, we identified amplification of CCND1, a genetic hallmark of acral melanoma,¹¹ in proliferating melanocytes in a subset of such lesions, which strongly suggests that such atypical pigmented lesions on acral volar skin represent a very early phase of acral melanoma preceding histopathogically recognizable melanoma in situ. Cho et al¹⁹ actually observed a Korean patient with atypical melanosis of the foot; the melanosis increased its size and later showed histopathological features of acral melanoma in situ.

In our view, nearly all acral melanomas are not derived from preexisting benign nevi but begin with a clonal expansion of genetically altered melanocytes within a focus in the basal layer of the epidermis.²⁰ The present study shows that such neoplastic melanocytes in the initial stage of acral melanoma already harbor CCND1 amplification. In some cases, proliferation of such melanocytes can be very subtle, without apparent nuclear atypia even though the lesions are clinically recognized as large pigmented macules. Our observations are comparable to a report by Bastian et al,¹¹ who also demonstrated individual basal melanocytes with gene amplifications in the histopathologically healthy-looking skin immediately adjacent to an acral melanoma. Bastian²¹ termed these genetically aberrant cells with no morphological evidence of malignancy field cells. Our results provide direct evidence that clonal expansion of such field cells with genetic alterations (that is, CCND1 amplifications) is an early evolving stage of acral melanoma in situ. Analogous to the clonal patches of keratinocytes with p53 mutations found in sun-damaged skin,²² the clonal melanocytic field cells may acquire more genetic alterations that lead to apparent proliferation of these cells in due course.

According to the dermoscopic findings of the presence or absence of the PRP, group 2 lesions were further classified into 2 subgroups. The PRP is characterized by accentuated bandlike pigmentation on the ridge of the skin markings that are arranged in parallel on glabrous skin.⁸ A very recent multicenter study⁶ shows that the PRP is extremely specific and sensitive in detecting melanoma in situ affecting acral volar skin. In this context, it is interesting that amplifications of CCND1 were detected only in the group 2A cases that revealed the PRP on dermoscopy. The result, therefore, seems to further support the specificity of the PRP to melanoma on acral volar skin. The histopathological background of the PRP is a preferential proliferation of melanocytes in the crista profunda intermedia, the epidermal rete ridges situated under the ridges of the skin markings.⁸ The reason why the melanocytic field cells with genetic aberrations such as CCND1 amplifications preferentially proliferate in the rete ridges of the crista profunda intermedia is unknown. It was reported¹³ that CCND1 amplification in tumor cells is always associated with overexpression of the CCND1 protein. Although CCND1 is a well-known growth promoter, it may also function as a survival factor for tumor cells.¹³ It is thus speculated that neoplastic melanocytes harboring an increased gene dosage of CCND1 may preferentially survive and proliferate in the microenvironment provided by epidermal keratinocytes in the rete ridges of the crista profunda intermedia.

The frequency of CCND1 amplification in group 2A was 4 (44%) of 9 lesions, which was equal to the reported frequency of 44% in acral melanomas.^11,13 It was reported^11,12 that all acral melanomas appeared to have at least 1 gene amplification and that, in addition to 11q13, which contains the CCND1, chromosome regions including 4q12, 5p12, and 12q14 were also found to be amplified. These regions contain known oncogenes, such as platelet-derived growth factor receptor-α, human telomerase reverse transcriptase, and cyclin-dependent kinase 4. Therefore, it would be interesting to determine whether the amplifications of these oncogenes (other than CCND1) exist in group 2A cases that lacked CCND1 amplification. This would further verify the specificity and usefulness of the PRP in the diagnosis of early melanoma on acral volar skin. Because the histopathological findings did not provide convincing evidence of malignancy in the lesions classified as group 2A cases, dermoscopic observations might be superior to histopathological methods in detecting early lesions of melanoma.

Very recently, Soyer et al²³ pointed out the limitations of histopathological methods in the diagnosis of melanoma and emphasized the importance of a combined approach of histopathological and dermoscopic evaluation. The present investigation also shows that histopathological methods may not be the gold standard in the classification of melanocytic lesions on the acral volar skin, particularly in the diagnosis of an early evolving phase of acral melanoma, and emphasizes the importance of dermoscopic evaluation in the diagnosis of pigmented skin lesions. Furthermore, this study also highlights the usefulness and importance of molecular techniques in the field of dermatopathology, as we and others have already shown in other settings.^24-26

Correspondence: Minoru Takata, MD, PhD, Department of Dermatology, Shinshu University School of Medicine, Asahi 3-1-1, Matsumoto 390-8621, Japan (mtakata@hsp.md.shinshu-u.ac.jp).

Accepted for Publication: May 17, 2005.

Author Contributions:Study concept and design: Takata. Acquisition of data: Yamaura, Miyazaki, and Saida. Analysis and interpretation of data: Yamaura and Takata. Drafting of the manuscript: Takata. Critical revision of the manuscript for important intellectual content: Saida. Statistical analysis: Yamaura. Obtained funding: Saida. Administrative, technical, and material support: Saida. Study supervision: Saida.

Financial Disclosure: None.

Funding and Support: This study was supported by a Grant-in-Aid for Cancer Research (15-10) from the Ministry of Health, Labor and Welfare of Japan, Tokyo.

Acknowledgment: We thank Kazuyuki Matsuda, Department of Laboratory Medicine at Shinshu University Hospital, Matsumoto, Japan, for his helpful advice in FISH analyses.