This new FISH assay was first evaluated by Gerami et al10 on a cohort of 83 melanomas and 86 typical nevi, with a sensitivity of 86.7% and a specificity of 95.4%. Their results were confirmed by the study of Morey et al12 on 10 metastases of melanomas, 10 primary melanomas and 10 nevi showing a sensitivity of 90% and a specificity of 95%, and more recently by a new study by Gerami et al13 on 110 nevi and 123 melanoma (sensitivity 83% and specificity 94%). Using the same updated threshold value on a series of 20 non-ambiguous melanomas and 19 nevi, we observed quite similar sensitivity of 85% and specificity of 90%. In our study, two out of 19 nevi were FISH positive as well as 4/86 nevi in the study by Gerami and 1/10 nevi in Morey's study. Our two cases as well as Morey's nevus had only one positive criterion. No criterion other than histopathological review can be used to consider such cases as truly benign and therefore as ‘false-positive’ FISH data, as such lesions may have been cured by excision, so their spontaneous outcome is unknown. Such ‘false-positive’ cases emphasize the need to combine morphology and FISH data to achieve a final diagnosis, and underline the lack of interest in performing FISH for the diagnosis of typically benign or malignant cases. Concerning false-negative cases, our three FISH-negative melanomas were thin SSM (Breslow) between 1.2 and 2 mm. However, the impact of thickness on FISH results was not shown by the study by Gerami et al.10 A recent study proved the feasibility of this FISH test for in situ (very thin) melanocytic lesions in the so-called lentiginous junctional melanoma of the elderly.14 In this study, three out of 19 in situ melanomas were FISH negative. The main difficulty concerning melanocytic lesions is to define the gold standard for malignancy. It might be best to use histopathological diagnosis. Yet, diagnoses may differ between experts.4 It could also be to base judgment on metastatic evolution. Yet, well-excised melanomas may be cured and never metastasize. A particularly doubtful diagnosis for thin lesions has no impact on survival because complete excision leads to complete remission. On the other hand, the impact of a doubtful diagnosis in thick melanocytic lesions is great because treatment and follow-up are quite different. Therefore, our study shows that the histopathological diagnosis combined with FISH data could improve the accuracy of diagnosis.
In our study, the most frequent FISH-positive criterion was RREB1 gain (71%), as observed by others.10, 12, 13 However in Morey's study, CCDN1 gain was as frequent as RREB1 gain in all type of melanomas (70%),12 whereas we observed only 48% of CCDN1 gain in FISH-positive cases. CCDN1 gain has been found associated with ALM and chronically sun-damaged skin melanoma subtypes.13, 15 In our series, ALM cases shared the same chromosome aberration: CCDN1 gain and MYB deletion. This is in agreement with the genetic classification of melanomas based on sun exposure proposed by Bastian et al.11 To improve the diagnosis of acral melanoma or chronic sun-induced damage melanomas, the analysis of KIT mutations could provide ancillary data.16 Curtin et al16 found KIT mutations in 36% of acral melanomas and 28% of chronic sun-induced damage melanomas, but never in melanoma on skin without chronic sun-induced damage melanomas. The impact of FISH analysis of melanoma has to be determined not only for diagnostic purposes, but also for prognosis and treatment.
In ambiguous melanocytic tumors, such genetic analysis would primarily be very useful from a diagnostic point of view, as it would help pathologists to distinguish benign and malignant melanocytic lesions. Indeed, ambiguous lesions simulating a melanoma represent the most frequent cause of lawsuits for pathologists.9 Different genetic approaches have been used to facilitate this differential diagnosis. Koh et al17 proposed a DNA microarray-derived gene expression profiling study after microdissection on formalin-fixed and paraffin-embedded tissue sections of melanoma and nevus, including difficult melanocytic lesions. Thirty-six significant differentially expressed genes were identified. In comparison with nevi, melanomas expressed higher levels of genes promoting signal transduction, transcription and cell growth, but expression of L1CAM was reduced in melanomas relative to nevi. For this reason, gene signatures established using DNA microarray gene expression profiling could be useful for distinguishing melanomas from nevi as a supplement to standard histology. The reproducibility of such an approach has never been tested and its use in routine practice may not be as robust as FISH analysis.
Among ambiguous melanocytic tumors, the main group of problematic lesions is Spitz tumors (45/90 in our study) as frequently reported in the literature (eg, by Barnhill et al2). In our study, among 28 out of the 45 Spitz tumors diagnosed as ‘favor malignant’ (A+), 12 were FISH positive and six of them underwent metastasis. Conversely, among the 16 FISH-negative ‘favor malignant’ Spitz cases, five also displayed metastases showing that histopathological analysis remains the gold standard for establishing the diagnosis of malignancy. A broader spectrum of FISH probes and further genetic analyses need to be evaluated in such FISH-negative patients with metastatic outcome. Among the 14 ‘favor benign’ (A−) Spitz nevi, only one was FISH positive (CCDN1 gain) and was even recurrence free at 5-year follow-up. Two out of the three discordant Spitz cases were FISH positive without unfavorable course. Therefore, FISH positivity in ambiguous cases may strengthen the diagnosis of malignancy when at least one pathologist favors malignancy. In such ‘favor malignant’ Spitz cases, a FISH-negative result should not modify the expert interpretation. In ‘favor benign’ ambiguous cases, negative FISH data may lead to ‘conservative’ excision, as suggested by others.18 Using the same FISH melanoma kit, Gerami et al10 analyzed 27 ambiguous melanocytic tumors reviewed by two dermatopathologists also mostly represented by Spitz tumors (24/27). Six patients developed metastases and 21 remained free with at least a follow-up period of 5 years. FISH was positive in all six metastatic patients and in 6/21 non-metastatic patients. Among such Spitz tumors, a technical difficulty for interpretation is polyploidy. Eight of our nine polyploid cases corresponded to Spitz tumors (seven ‘favor malignant’ and one ‘favor benign’). Interestingly, Isaac et al19 reported four out of 41 Spitz nevi as polyploidy that were confirmed by X-chromosome probe. The disease-free survival times between FISH-positive and FISH-negative patients were significantly different (P=0.003), as we observed in our study with a trend for better prognosis for FISH-negative patients (P=0.07). This could be very useful in the future because prognosis of melanocytic lesions seems to be worse when the number of chromosomal anomalies increases. Therefore, FISH data associated with histopathological diagnosis could change the patient's management.
In our study, integration of FISH data along with histopathological diagnosis was found relevant for patient's management as the FISH result was found to improve the specificity of histopathological expert diagnosis alone (76% instead of 52%). On the other hand, the sensitivity of the FISH test alone was also improved (90 vs 43%). The most difficult issue for ambiguous melanocytic lesions is to determine what is the proof of malignancy between expert diagnosis or metastatic outcome. If expert diagnosis is taken to be the gold standard for ambiguous melanocytic lesions, the specificity of this FISH test is quite high (90.6%). However between experts, the diagnostic reproducibility of ambiguous melanocytic tumors is quite low.4 On the other hand, if outcome is considered the most important criterion, FISH was found to provide good specificity (80%). Even metastatic outcome is probably not so unfavorable in childhood and teenage Spitz tumors, as recently shown by Busam et al.20
Other applications of such a FISH melanoma kit have been reported. It could be useful for microstaging of melanoma, allowing the distinction between melanoma arising on nevus vs nevoid melanoma.21 It could also be useful if it really changes the treatment of patients, for example, with a thickness greater than 1 or 1.5 mm (according to the protocol), leading to interferon treatment or other adjuvant management such as sentinel node biopsy or adjuvant vaccine therapy. Other FISH applications have been reported to facilitate the differential diagnosis between (I) epithelioid blue nevus and cutaneous metastatic melanoma simulating blue nevus22 or (ii) nevoid melanomas and mitotically active nevi,23 (iii) conjunctival nevi from melanomas24 or (iv) intranodal nevus from metastatic melanoma.25
In conclusion, a single genetic test allowing the differential diagnosis between benign and malignant ambiguous melanocytic lesions does not yet exist. However, the practical value of the FISH test used in this study was that it reinforced the histopathological diagnosis of malignancy. In unequivocal melanomas it was highly sensitive. In ambiguous melanocytic tumors, especially in thick lesions, slides should first be submitted to a referring pathologist to improve the diagnosis and then a positive FISH test would strengthen the diagnosis of ‘favor malignancy’. However, FISH testing did not formally identify patients at risk of metastasis. In unequivocal benign melanocytic ambiguous lesions, caution should be taken if the FISH test is positive, whereas a negative test would be an ancillary argument for benignity in ‘favor benign’ cases.18 In our opinion, FISH positivity alone without histopathological diagnosis of malignancy should not lead to modifying the patient's treatment. Finally, in cases with discordance between pathologists, the value of a positive FISH test was high when a single pathologist favored malignancy. Therefore, ambiguous lesions with doubtful or discordant histopathological analyses should be tested by FISH. In this group, FISH-positive cases should be managed in the same way as malignant and ‘favor malignant’ cases. Further genetic testing should be conducted on FISH-negative ‘favor malignant’ cases.
Fluorescent in situ hybridization used to label chromosomes during the cell cycle at S phase has identified an apparent asynchrony of replication between the parental alleles of an imprinted gene (Kitsberg et al., 1993). The region showing asynchrony can extend beyond the cluster of imprinted genes to include nonimprinted genes. The asynchrony can go in both directions (maternal early or late) but, in contrast to the behavior of nonimprinted genes, there is no correlation between late replication and silencing (Knoll et al., 1994). In contrast to the fluorescent in situ hybridization technique, bromodeoxyuridine incorporation detects replication asynchrony only in highly expressing cells, and shows that the extent of asynchrony between imprinted parental alleles can be less than between loci along a single chromosome. It has been suggested that the asynchrony observed with both these techniques may indicate allelic differences in chromatin structure rather then DNA replication (Kawame et al., 1995; Bickmore and Carothers, 1995). In support of this, increased chromatin packaging on nontranscribed alleles for Snrpn and M6P-Igf2r has been described (Watanabe et al., 2000). In addition, several imprinted genes have now been shown to have differential chromatin organization over the maternal and paternal promoter of an imprinted gene (Greally et al., 1999; Hark and Tilghman, 1998; reviewed in Feil and Khosla, 1999). However, it is not yet known if chromatin plays a part in acquisition of the imprint, or in the maintenance of imprinted expression, or perhaps has no significance for imprinting.