—  SPECIALTY CONFERENCE  —

Dermatopathology

Cases 3-6 - Molecular Cytogenetics as a Diagnostic Tool for Typing Melanocytic Tumors

Boris Bastian
UCSF Comprehensive Cancer Center
San Francisco, California


Click on each slide thumbnail image for an enlarged view
Clinical Histories:
Case 3: 47 year old man with a pigmented lesion on his right sole. CGH shows multiple aberration including amplifications of chromosomes 11q13 and 22q13.


Case 3 - Figure A - Acral skin biopsy, low power, showing a confluent nodular growth of atypical cells in the dermis with pagetoid spread through the overlying epidermis to the cornified layer. There is an asymmetric shoulder showing a lentiginous melanocytic proliferation.
Case 3 - Figure B - There is a confluent growth of atypical melanocytes along the dermoepidermal junction with pronounced pagetoid single cell spread to the cornified layer.


Case 3 - Figure C - At the lesional shoulder, a lentiginous array of fully transformed neoplastic melanocytes is present in the basal and immediate superbasilar spinous layers. The cells are hyperchromatic, polygonal and contracted and manifest fully transformed malignant cytologic features.
Case 3 - Figure D - The vertical growth phase component is predominantly spindled in character. There are admixed squamous pearls which represent down growths of squamous epithelium as a pseudoepitheliomatous reaction to the neoplasm. Cells show mitotic activity and fully transformed malignant features.

Case 4 : 5 year old girl with a lesion on the left flank that was removed and recurred one year later. CGH did not reveal any chromosomal aberrations.


Case 4 - Figure A - Low power examination shows a polypoid nodule with a symmetrical profile. There is an inverted apical growth of melanocytes along adnexal structures toward the base of the biopsy extending into the subcutaneous tissue.
Case 4 - Figure B - Along the dermoepidermal junction are large nests of melanocytes showing vertically oriented spindled shaped cells in fasicles with a "raining down" architecture.
Case 4 - Figure C - High power shows atypical spindled shaped melanocytes in dermal based nodules. Rare mitotic figures are apparent.

Case 5 : 33 year old man with a lesion of the left ear that was removed and recurred after several years. CGH showed multiple chromosomal aberrations.


Case 5 - Figure A - A low power micrograph shows confluent nodules of cells irregularly dispersed in the dermis associated with a dense sclerosing reaction.
Case 5 - Figure B - The melanocytic nests within the dense stromal collagen table show hyperchromatic and irregular nuclear contours.
Case 5 - Figure C - A high power micrograph shows nuclear hyperchromasia, endonuclear cytoplasmic inclusions and fully transformed cytologic features of malignancy.

Case 6 : 45 year old man with a pigmented lesion on his back. CGH showed multiple chromosomal aberrations, but no amplifications.


Case 6 - Figure A - A low power micrograph shows a dermoepidermal nevomelanocytic proliferation with an asymmetric architecture and a lateral junctional shoulder.
Case 6 - Figure B - (intermediate power): Along the junctional shoulder melanocytes show a pagetoid pattern of migration to the upper spinous layers consistent with a radial growth phase melanoma.


Case 6 - Figure C - There is a vertical growth phase nodule in the dermis which is surmounted by a radial growth phase component.
Case 6 - Figure D - (high power): The vertical growth phase component shows fully transformed malignant nuclear features with frequent and atypical mitoses.

Introduction
Histopathology is the gold standard for diagnosis of pigmented lesions of the skin. In spite of a great deal of morphologic variability, the majority of cases can be classified reliably with current pathological criteria. However, there is a significant subset of cases that are so ambiguous that no consensus can be reached even among expert pathologists(3-9).  The effect of the ambiguity on standard clinical practice is illustrated in a recent study from The Netherlands. An expert panel reviewed 1069 consecutive melanocytic lesions that had been submitted for review by clinical pathologists in order to identify the most common diagnostic problems. In 14% (22/158) of the cases that had been initially classified as invasive melanoma the panel considered the lesions as benign, and in 16.6% (85/513) the panel considered malignant what had been diagnosed as benign(10)  . Together with lymphoma, melanoma leads the list of pathology malpractice claims(11)  . Currently there is no method to definitively resolve these ambiguities. A clinical test that could enhance established diagnostic procedures would be of significant clinical benefit. Such a test would need to be applicable to routinely fixed tissue, because diagnostic problems typically arise after the specimen has already been processed. Several routes are conceivable that could lead to the identification of markers that could be detected by such a test. Among them, screening approaches permitting an analysis of thousands of markers are the most powerful route. However, most of these approaches rely on RNA as a source, which currently cannot be extracted in sufficient amounts and quality from formalin-fixed tissue. By contrast, DNA can be readily extracted from formalin-fixed tissue in quantity and quality adequate for a variety of analyses. We have relied on genomic analyses in order to identify potential diagnostic markers for melanocytic tumors.

Figure 1. Comparative Genomic Hybridization. Left: Total genomic DNAs are isolated from a 'test' and a 'reference' cell population, labeled with different fluorochromes, and hybridized to normal metaphase chromosomes. Cot-1 DNA is used to suppress hybridization of repetitive sequences. The resulting ratio of the fluorescence intensities of the two fluorochromes at a location on a chromosome is approximately proportional to the ratio of the copy numbers of the corresponding DNA sequences in the test and reference genomes. Right. A similar hybridization to an array of mapped clones permits measurement of copy number with resolution determined by the length of the clones and/or their map spacing.

Comparative genomic hybridization (CGH)
As originally described(13),  CGH detects and maps DNA sequence copy number variation throughout the entire genome onto a cytogenetic map supplied by metaphase chromosomes (Figure 1, left). Recently, an implementation of CGH has been described in which the metaphase chromosomes are replaced by arrays of genomic bacterial artificial chromosomes (BAC) clones (Figure 1, right)(14).  Relative copy number can then be measured at loci specified by the BAC clones by hybridization of fluorescently labeled test and reference DNAs as in conventional CGH. The use of metaphase chromosomes as the hybridization target has previously limited the resolution of CGH to 10-20 Mb, prohibited resolution of closely spaced aberrations, and only allowed linkage of CGH results to genomic information and resources with cytogenetic accuracy. In array CGH, the resolution is determined by the genomic spacing and/ or length of the target clones, and the positions of the clones are accurately known on the Human DNA sequences because each clone contains a sequence tag. Array CGH allows accurate quantification of DNA copy number variations over a wide dynamic range, including reliable detection of single copy deletions and duplications(15).  Array CGH provides substantially improved resolution and sensitivity compared to conventional CGH in the analysis of tumor genomes.

Figure 2: Chromosomal localization of DNA-sequence copy number changes in 32 randomly retrieved melanomas detected by CGH. Lines to the right of the chromosome ideograms represent gains, lines to the left represent losses. Thick lines indicate amplifications.

DNA copy number changes in primary cutaneous melanoma
We have performed conventional CGH on several hundred formalin fixed, paraffin embedded primary melanocytic lesions, most of which are melanomas. In an initial survey we screened 32 randomly selected primary cutaneous melanoma for chromosomal alterations using CGH(16)  Figure 2. Most of these were superficial spreading melanomas (SSM), the most common type in a light skinned population. This was the first comprehensive analysis of chromosomal gains and losses in primary melanoma. The most frequent aberrations were losses of chromosome 9 (81% of the tumors), most commonly affecting the p-arm. Further common losses occurred on chromosomes 10 (63%), 6q (28%), and 8p (22%). Gains in copy number involved chromosomes 7 (50%), 8q (34%), 6p (28%), 1q (25%), 20 (13%), 17 (13%), and 2 (13%)(16).  Most of these aberrations involved large chromosomal regions. One Acral Lentiginous Melanoma (ALM) was included in this group and it had high-level amplifications of several small chromosomal regions. This motivated a larger investigation of the genomic aberrations in 15 ALM. All the ALMs had at least one (mean 2.0) gene amplification, significantly more than a control set of 15 SSMs of comparable tumor thickness, in which only 2/15 (13%) had one amplification each (p<0.0001)(17)  . At least 15 different genomic regions were amplified in ALM. These involved small portions of chromosomal arms, sometimes including known oncogenes implicated in melanoma. The most frequently amplified regions in ALMs occurred at 11q13 (47%), 22q11-13 (40%), and 5p15 (20%). The region at 11q13 contains the cyclin D1 gene, that was shown to be overexpressed in all cases with the amplification and in about 20% of melanomas without the amplification(18)  . Inhibition of cyclin D1 expression with antisense RNA in cell lines that amplified or overexpressed cyclin D1 lead to massive apoptosis and tumor shrinkage in mouse xenograft models(18)  .

ALM exhibits several clinical and epidemiological features that distinguish it from the SSMs. For example, the incidence of ALM is approximately equal across all racial groups, and it develops on palmar, plantar and subungual skin, sites that have little exposure to sunlight and are protected from ultraviolet radiation by a thick stratum corneum. However, its classification as a distinct type of melanoma has been controversial. Our findings of the different character of the genetic events that are involved in the ALM and SSM coupled with its clinical features support the classification of ALM as a distinct melanoma subtype. More recent studies on over 40 melanomas from acral skin showed that the amplifier phenotype of ALM is independent of the histological growth pattern. Also tumors without a lentiginous growth pattern such as nodular or superficial spreading types showed this unique genetic characteristic as long as there were located on the non hair-bearing skin on acral sites (Bastian et al., manuscript in preparation). Glabrous melanoma may be a more accurate term to describe this entity.

DNA copy number changes in Spitz nevi
Most frequently problems of diagnostic ambiguity arise with Spitz nevus. Sophie Spitz initially described this condition as "juvenile melanoma" and regarded it as a subset of childhood melanoma that followed a benign course(19)  . It is now regarded as a form of melanocytic nevus. Spitz nevi account for about 1% of surgically removed melanocytic nevi(20)  and most frequently, but not exclusively, occur in children. Spitz nevi like melanoma can display a composition of melanocytes with abundant cytoplasm and large nuclei that contain macro-nucleoli. Mitotic figures can be numerous in either condition (21,22) . This overlap can make it impossible to make a diagnosis, resulting in well-documented lack of consensus, over- and misdiagnosis(8,23,24)  . A previous study used a panel of expert pathologists to review 102 lesions that were diagnosed as melanomas that occurred in patients under 17 years of age. They found that only 60 cases were correctly classified as melanoma and the majority of the remainder was reclassified as Spitz nevi by the expert panel(24)  .

CGH analysis of Spitz nevi showed that different from melanomas, which tend to have multiple chromosomal aberrations, Spitz nevi are usually normal by CGH analysis(25).  However, a few show a copy number increase of chromosome 11p, with occasional other aberrations. When used FISH with probes for chromosome 11p to screen a large number of Spitz nevi we found copy number increases of chromosome 11p of at least three fold in 12/102 cases (11.8%)(26).  In order to find out which gene drives the selection forces that lead to the accumulation of extra copies of chromosome 11p we focused on the HRAS oncogene, which maps to chromosome 11p. HRAS showed frequent oncogenic mutations in cases with copy number increase (8/12 or 67%), contrasting with rare HRAS mutations in cases with normal 11p copy numbers (1/21 or 5%, p<0.0001). Finding a mutated oncogene in a benign lesion was somewhat surprising. However, ras mutations are also frequently found in hamartomatous polyps of the colon, which only rarely progress to cancer. In contrast, dysplastic colon polyps, which have a high risk of progression, initially acquire mutations of APC, that are followed by ras mutations later(27).  Recently a high frequency of mutations in the BRAF oncogene have been described in other types of melanocytic nevi(28).  The BRAF gene is frequently mutated in melanoma and operates immediately down-stream in the MAP-kinase pathway(29).  Activation of the MAP-kinase pathway alone may not suffice for malignant transformation of melanocytes, as for example Ras genes need cooperating oncogenes or inactivation of p53 or p16 in order to transform normal cells(30,31)  These alterations may be missing in Spitz nevi with ras mutations and senescence induced by exhausted telomeres and a permanent G1-arrest.may prevent these lesions from further progression. Although, tumors with 11p copy number increases tended to be larger, predominantly intradermal, had marked desmoplasia and characteristic cytological features, they were clearly distinct from melanoma. Proliferation rates in the majority of these cases were low to absent. Follow-up of up to five years did not show any evidence of progression. Also, we did not find the gain of the whole arm of chromosome 11p seen in Spitz nevi in any of over 150 melanomas. In case that the Spitz nevi with the 11p gain would have a tendency to progress to melanoma, one would expect that at least some of the melanomas that we have analyzed would carry the 11p gain. Interestingly, one Spitz nevus of our initial study recurred several times at the excision site. We have studied 10 further cases of Spitz nevi that recurred at the excision site and found the isolated 11p gain in 3 cases(32).  These observations suggest that Spitz nevi with the 11p gain may have an increased risk of recurrence.

Conclusion
The use of comparative genomic hybridization and FISH on primary tumors has shown that the pattern of genomic aberrations differs significantly between melanoma and benign nevi and among different types of melanoma. The vast majority of primary melanomas have multiple chromosomal aberrations, whereas the vast majority of nevi do not have any. The few benign nevi that do have aberrations typically have a very restricted set, which does not occur in melanoma. These findings hold the promise to be useful in the classification of melanocytic tumors that are ambiguous by current histopathology. We have now begun to evaluate the potential of whole genome scanning using array CGH to predict outcome in melanocytic tumors that are histopathologically ambiguous. If we identify predictive markers for this group of cases, these may be potentially be developed into a test for clinical practice.

References

  1. D. S. Rigel, J. A. Carucci, CA Cancer J Clin 50, 215-236 (2000).
  2. C. Del Mar et al., Aust J Public Health 18, 221-3 (1994).
  3. R. Corona et al., J Clin Oncol 14, 1218-23 (1996).
  4. R. Jackson, Int J Dermatol 36, 497-8 (1997).
  5. S. L. Wakely, J. A. Baxendine-Jones, P. J. Gallagher, M. Mullee, R. Pickering, Am J Surg Pathol 22, 77-82 (1998).
  6. M. G. Cook, J Pathol 182, 247-9 (1997).
  7. M. Piepkorn, P. B. Odland, Arch Dermatol 133, 1393-6 (1997).
  8. W. Kempf, A. C. Haeffner, B. Mueller, R. G. Panizzon, G. Burg, Am J Dermatopathol 20, 478-82 (1998).
  9. E. R. Farmer, R. Gonin, M. P. Hanna, Hum Pathol 27, 528-31 (1996).
  10. K. C. Veenhuizen et al., J Pathol 182, 266-72 (1997).
  11. D. B. Troxel, J. D. Sabella, Am J Surg Pathol 18, 821-31 (1994).
  12. J. Goldes, S. Holmes, M. Satz, J. Cich, L. Dehner, Pediatr Dermatol 1, 295-8 (1984).
  13. A. Kallioniemi et al., Science 258, 818-21 (1992).
  14. D. Pinkel et al., Nat Genet 20, 207-11 (1998).
  15. D. G. Albertson et al., Nat Genet 25, 144-6 (2000).
  16. B. C. Bastian, P. E. LeBoit, H. Hamm, E. B. Bröcker, D. Pinkel, Cancer Res 58, 2170-5 (1998).
  17. B. C. Bastian et al., Cancer Res 60, 1968-73 (2000).
  18. E. R. Sauter et al., Cancer Res 62, 3200-6. (2002).
  19. S. Spitz, Am J Pathol 24, 591-609 (1948).
  20. E. M. Casso, C. M. Grin-Jorgensen, J. M. Grant-Kels, J Am Acad Dermatol 27, 901-13 (1992).
  21. M. Piepkorn, J Am Acad Dermatol 32, 248-54 (1995).
  22. D. Weedon, Skin Pathology (Churchill Livingstone, Edinburgh, 1997).
  23. R. L. Barnhill et al., Hum Pathol 30, 513-20 (1999).
  24. A. Spatz et al., Int J Cancer 68, 317-24 (1996).
  25. B. C. Bastian, U. Wesselmann, D. Pinkel, P. E. LeBoit, J Invest Dermatol 113, 1065-9 (1999).
  26. B. C. Bastian, P. E. LeBoit, D. Pinkel, Am. J. Pathol. 157, 967-972 (2000).
  27. J. Jen et al., Cancer Res 54, 5523-6 (1994).
  28. P. M. Pollock et al., Nat Genet 25, 25 (2002).
  29. H. Davies et al., Nature 417, 949-54. (2002).
  30. M. Barbacid, Annu Rev Biochem 56, 779-827 (1987).
  31. R. A. Weinberg, Cancer Res 49, 3713-21 (1989).
  32. J. D. Harvell, B. C. Bastian, P. E. LeBoit, Am J Surg Pathol 26, 654-661 (2002).