OUP user menu

Gastrointestinal Stromal Tumor Markers in Cutaneous Melanomas
Relationship to Prognostic Factors and Outcome

Raul S. Gonzalez MD, Grant Carlson MD, Andrew J. Page MD, Cynthia Cohen MD
DOI: http://dx.doi.org/10.1309/AJCP9KHD7DCHWLMO 74-80 First published online: 1 July 2011


Melanoma expresses c-kit, a gastrointestinal stromal tumor marker, but has not been extensively evaluated for protein kinase C θ (PKCθ) or DOG1, and these stains have not been correlated with prognostic factors. We immunostained 62 primary cutaneous and 15 metastatic melanomas for polyclonal c-kit (pc-kit), monoclonal c-kit (mc-kit), PKCθ, and DOG1 and correlated results with prognostic parameters and survival. Of the cutaneous melanomas, 34 (55%) stained for pc-kit, 30 (48%) for mc-kit, 11 (18%) for PKCθ, and 2 (3%) for DOG1. The Breslow depth was 1.00 mm or less in 21 (68%) of 31 pc-kit+ cutaneous melanomas compared with 7 (27%) of 26 pc-kit– melanomas (P = .002). The pc-kit+ melanomas had less nodal disease (1/31 [3%] vs 9/25 [36%]; P = .001) and local recurrence (1/33 [3%] vs 6/27 [22%]; P = .021) but no statistically significant difference in the rate of distant metastases (13/32 [41%] vs 14/27 [52%]; P = .388) or survival (10/34 [29%] vs 16/39 [41%]; P = .301). We found that pc-kit correlates better with prognostic parameters than does mc-kit.

Key Words:
  • Melanoma
  • Gastrointestinal stromal tumors
  • Immunopathology
  • c-kit
  • DOG1

Melanoma has been the subject of many immunohistochemical studies. In addition to markers such as S-100 protein and HMB-45, primary cutaneous melanoma has occasionally been found to stain positively for markers associated with other neoplasms.1 One such marker is CD117/c-kit, which is typically associated with gastrointestinal stromal tumors (GISTs).2,3 c-kit is a receptor tyrosine kinase that serves as a target for stem cell factor; this interaction is necessary for normal melanocyte function and proliferation.4 Studies indicate that approximately half of primary cutaneous melanomas demonstrate c-kit positivity.5,6

While c-kit is the prototypical marker for GIST, the marker is not 100% sensitive.7 In recent years, additional immunohistochemical markers have been found to be of diagnostic usefulness for c-kit– GISTs.8 Among these are protein kinase C θ (PKCθ) and DOG1.

PKCθ is a novel PKC isotype that was found to be expressed in GIST via transcriptional profiling. It has not been detected in numerous other mesenchymal neoplasms.9 This novel PKC isotype has previously been described as vital to mature T-cell activation and interleukin-2 gene induction.10 DOG1 is a protein of unknown function that is considered sensitive and specific for the diagnosis of GIST.11 It was discovered to be expressed in GIST through gene expression profiling.12 Despite the known expression of c-kit in primary cutaneous melanomas, no studies so far have investigated the staining properties of melanoma in regard to PKCθ9,1315 or DOG111,15 in more than a handful of cases.

Of the aforementioned GIST markers, only c-kit has been evaluated for prognostic significance in cutaneous melanomas.5,16 Staining patterns for PKCθ and DOG1 in melanoma have never been statistically compared with any of the aforementioned markers or with overall survival.

To evaluate whether immunohistochemical analysis for PKCθ and DOG1 may have a role in the diagnosis of cutaneous melanoma, our study aimed to evaluate the staining patterns of these markers in melanoma. Furthermore, given the lack of data regarding the prognostic significance of these markers, as well as for polyclonal c-kit (pc-kit) and monoclonal c-kit (mc-kit), expression of the 4 markers was compared with a number of prognostic variables and with outcome.

Materials and Methods

With Emory University Institutional Review Board (Atlanta, GA) approval, a review of a prospectively collected melanoma database identified 92 cases of melanoma in the archives of the Department of Pathology, Emory University that had H&E-stained slides, the corresponding blocks, and complete surgical pathology reports available. The reports included site of lesion, histologic type, Breslow depth, Clark level, and lymph node metastases. Clinical patient data contained in the database included adjuvant treatment and overall survival (average follow-up time, 66 months) and were available for 77 cases: 62 primary cutaneous melanomas and 15 metastatic melanomas from various sites. Additional information regarding local recurrence and distant organ metastases was also available. Each patient was included in the database only once; as a result, metastatic lesions from database patients with primary cutaneous melanomas were not available, nor were the primary cutaneous lesions from database patients with metastatic melanomas.

Tissue microarrays were constructed with two 1-mm formalin-fixed, paraffin-embedded cores from each case and from 2 benign melanocytic nevi. Five-micrometer sections from each microarray were deparaffinized and dehydrated. Immunostaining used standard heat-induced epitope retrieval, the DAKO EnVision Detection System, and the DAKO Autostainer (DAKO, Carpinteria, CA). Four antisera were used: pc-kit, mc-kit, PKCθ, and DOG1 Table 1. The primary antibody dilution, retrieval solution and pH, and detection system were optimized before the study was undertaken.

Antigen retrieval was in EDTA (pH 8.0) for the DOG1 antibody and in citrate buffer (pH 6.0) for the remaining antibodies, using an electric pressure cooker for 5 minutes at 120°C (15 psi), with cooling for 10 minutes before immunostaining.

All slides were loaded on the DAKO automated system, exposed to 3% hydrogen peroxide for 5 minutes, primary antibody for 30 minutes, labeled polymer horseradish peroxidase for 30 minutes, diaminobenzidine as chromogen for 5 minutes, and DAKO automation hematoxylin as counterstain for 5 minutes. These incubations were performed at room temperature; between incubations, sections were washed with tris(hydroxymethyl)aminomethane-buffered saline. Polysorbate-20 was included in this buffer. A known positive control of GIST was used with each run. Negative controls had primary antibody replaced by buffer.

View this table:
Table 1

The slides were viewed on an Olympus BX41 light microscope (Olympus, Hamburg, Germany) and assessed for presence of immunohistochemical staining and for the predominant growth pattern of the lesion (in situ, radial, or vertical). In clinical practice, and also for the purposes of this study, focal and/or weak but convincing staining was considered a positive result. Results were obtained semiquantitatively. Intensity of staining was graded from 0 to 3+ (0, no staining; 1+, focal and/or weak staining; 2+, moderate staining; 3+, strong staining). For the percentage of cells staining, cases were scored as negative (no cells immunopositive), 1+ (1%–9% of cells staining), 2+ (10%–50% of cells staining), and 3+ (>50% of cells staining).

The presence, intensity, cell percentage, and lesional distribution of staining for each of the 4 antibodies were statistically compared with each other and then with the following 6 parameters for the primary cutaneous melanomas: Breslow depth (≤1.00, 1.01–2.00, 2.01–4.00, and >4.00 mm), Clark level (1–5), lymph node metastasis, distant organ metastasis, local recurrence, and overall survival. The prognostic parameters were not statistically analyzed against one another because a higher Clark level, a higher Breslow depth, and the presence of nodal and distant metastasis have been established as harbingers of a worse prognosis.17 In the metastatic melanomas, the staining properties for each antibody were statistically compared with overall survival only.

Outcomes were examined with categorical variables as previously described. Statistics were analyzed with χ2 tests for each respective outcome. A P value of less than .05 was considered statistically significant. All statistics were calculated by using SPSS, version 17.0 (SPSS, Chicago, IL).


Table 2 summarizes clinical data for the patients. Of the 62 primary cutaneous melanomas, 34 (55%) stained for pc-kit Image 1A, 30 (48%) for mc-kit Image 1B, and 11 (18%) for PKCθ Image 1C; 2 (3%) of 60 cases stained for DOG1 Image 1D; of these, 1 demonstrated 2+ staining intensity in 20% of cells and the other expressed 3+ staining intensity in 5% of cells. The expression of pc-kit, mc-kit, and DOG1 was membranous and cytoplasmic; PKCθ expression was cytoplasmic. The stains pc-kit and mc-kit primarily stained the same lesions; 28 stained for both markers, 6 stained for pc-kit but not mc-kit, and 2 stained for mc-kit but not pc-kit. No histologic subtype of melanoma showed a statistically significant predilection for expressing any of the immunohistochemical markers.

Only 2 (13%) of the 15 metastatic melanomas expressed pc-kit, which is significantly lower than the 55% of primary melanomas that expressed pc-kit (P = .004). Similarly, none of the metastatic melanomas (0%) expressed mc-kit compared with 48% of the primary melanomas (P = .001). Of the 15 metastases, 3 (20%) stained for PKCθ, which is similar to the 18% staining frequency among primary melanomas (P = .839). None of the metastases (0%) expressed DOG1 compared with 3% of the primary melanomas (P = .474). Of the 2 nevi, 1 (50%) stained for pc-kit, and both were immunonegative for mc-kit, PKCθ, and DOG1.

The statistically significant data are presented in Table 3, Table 4, and Table 5. Of the 31 cutaneous melanomas that expressed pc-kit and had Breslow depth data available, 21 (68%) had a Breslow depth of 1.00 mm or less compared with 7 (27%) of 26 cutaneous melanomas that were negative for pc-kit; conversely, 2 (6%) of these 31 pc-kit+ melanomas had a Breslow depth of more than 4.00 mm, whereas 11 (42%) of these 26 pc-kit– melanomas had this Breslow depth (P = .002; Table 3). Of the 62 primary cutaneous melanomas, 5 did not have Breslow depth data available. In addition, of the 21 pc-kit+ primary melanomas with a Breslow depth of 1.00 mm or less, 17 (81%) showed 3+ staining intensity, with the remaining 4 showing 2+ intensity (P = .002; Table 3). More than half (7/13 [54%]) of the melanomas with a predominantly vertical growth phase that stained with pc-kit had a Breslow depth of 1.00 mm or less (P = .008; Table 3), and 19 (68%) of 28 pc-kit+ melanomas with more than 50% of cells staining had a low Breslow depth (P = .012; Table 3).

The primary melanomas that stained positively for pc-kit had less nodal disease (1/31 [3%] vs 9/25 [36%]; P = .001; Table 4) and less local recurrence (1/33 [3%] vs 6/27 [22%]; P = .021; Table 4). Six melanomas lacked nodal disease data, and 2 lacked local recurrence data.

No significant correlation was seen between pc-kit staining and distant metastatic disease: 13 (41%) of 32 pc-kit+ cases were associated with distant metastases compared with 14 (52%) of 27 pc-kit– cases (P = 0.388).

View this table:
Table 2

Similarly, there was no statistically significant correlation between pc-kit staining and overall survival: 10 (29%) of 34 patients with pc-kit+ melanoma (primary and metastatic) vs 16 (41%) of 39 patients with pc-kit– melanoma (P = .301).

Node status was the only parameter studied in which mc-kit staining showed the same significance as pc-kit staining in primary melanomas, as only 1 (4%) of 28 (P = .005) that stained with mc-kit demonstrated positive nodal disease (Table 5). Of 28 cases negative for mc-kit, 9 (32%) showed nodal disease. Local recurrence was observed in 1 (4%) of 28 mc-kit+ melanomas compared with 6 (19%) of 32 mc-kit– melanomas (P = .068; Table 5).

Image 1

Immunohistochemical staining of primary cutaneous melanoma. A. 3+ expression of polyclonal c-kit, 90% of cells, in situ growth phase (×400). B, 2+ expression of monoclonal c-kit, 90% of cells, vertical growth phase, with interspersed melanophages (×400). C, 3+ expression of protein kinase C θ, 90% of cells, in situ growth phase (×400). D, 2+ expression of DOG1, 20% of cells, vertical growth phase (×400).

View this table:
Table 3

No statistically significant relationships or patterns were demonstrated between staining parameters in the small numbers of melanomas expressing PKCθ and DOG1 and the chosen prognostic factors or outcome.


Staining with c-kit was observed in more than half (55%) of the primary cutaneous melanoma cases examined, which agrees with previous findings. Janku et al5 found pc-kit staining in 55.2% of cutaneous melanomas, and Guerriere-Kovach et al6 reported positive c-kit staining in more than 10% of cells in 48.8% of primary melanomas (including cutaneous and uveal lesions). c-kit is neither a sensitive nor a specific marker for melanoma and, therefore, is unlikely to be of help during the typical diagnostic workup of a neoplasm suggestive of melanoma. However, c-kit was found to be present more often in melanomas of lower Breslow depth that did not recur following surgical resection and that did not metastasize to lymph nodes (ie, better prognostic parameters). In keeping with this finding, the frequency of c-kit positivity in metastatic melanoma was significantly reduced compared with the frequency in primary cutaneous lesions.

Previous studies examining the role of c-kit in melanoma came to similar conclusions. Janku et al5 examined 261 patients with stage 1 and 2 melanoma and found no correlation between c-kit expression and Breslow depth, Clark level, or disease-free survival. The study did not include local recurrence, lymph node metastases, or distant metastases in its analysis. Potti et al,16 studying 202 cases, also found no correlation between c-kit positivity and survival; no prognostic variables were considered in the analysis. While these studies did not determine a correlation between c-kit staining and Breslow depth or nodal spread of disease, Potti et al16 noted that c-kit was found in more than half of the superficial spreading melanomas the study analyzed. They postulated that c-kit was therefore seen primarily in early disease, and metastatic melanomas may have down-regulated their expression of c-kit.

Other studies have also shown that metastatic melanomas often lose expression of c-kit, along with other proteins such as MITF and bcl-2.1820 These studies, however, did not differentiate between lymph node and organ metastases when discussing their findings. Indeed, despite the link between c-kit expression and nodal disease, our study found no correlation between c-kit staining and the likelihood of vascular metastasis to distant organs. This finding indicates that while c-kit expression in melanoma is less often associated with lymphatic spread, it has no relationship to hematogenous metastasis.

Our results also show that melanomas positive for c-kit are no more or less likely to lead to a patient’s death, a finding that confirms the findings of previous studies.5,6 Therefore, while retention of c-kit expression may indicate that a melanoma is more indolent clinically, it does not seem to indicate a better long-term prognosis.

View this table:
Table 4
View this table:
Table 5

Stains for mc-kit did not yield the same statistically significant findings as pc-kit. pc-kit stained more primary cutaneous melanomas than mc-kit did, making it a more sensitive marker. This same finding has been reported in GISTs stained for both types of c-kit antibody.21 Thus, staining for c-kit on cases of melanoma should be performed using the polyclonal antibody.

Mutations of c-kit are often found in certain melanoma subtypes, most notably acral and mucosal melanomas.22 While the rate of c-kit mutation in mucosal melanomas is approximately 15%,22,23 the frequency in primary cutaneous melanoma has been reported as 1.7%.24 Presence of a c-kit mutation does not necessarily correlate with immunohistochemical staining for c-kit antigen because roughly half of primary cutaneous melanomas stain positively for c-kit. In addition, c-kit– melanoma cell lines do not seem to possess a deletion or rearrangement of the c-kit gene.16 This inconsistency between c-kit staining and c-kit mutation, coupled with the low rate of c-kit mutations in primary cutaneous melanomas, may offer an explanation for why therapy targeted at mutant c-kit protein, such as imatinib, has so far not yielded consistent clinical benefit.25

Unlike in melanoma, the correlation between c-kit mutation and immunohistochemical findings in GISTs is well-characterized, as 98% of lesions with a gain-of-function c-kit mutation display at least moderate immunohistochemical positivity.26 GISTs lacking a c-kit mutation are often found to possess a mutually exclusive mutation in platelet-derived growth factor receptor α (PDGFRA).26,27 Overall, more than 90% of GISTs demonstrate a mutation in 1 of these 2 genes.28 While some melanomas have been shown to overexpress PDGFRA,29 a study of 26 cutaneous melanomas that overexpressed PDGFRA or lacked a c-kit mutation found that none of them demonstrated a PDGFRA mutation.30

Most studies looking at PKCθ expression in melanoma have included few cases but have reported generally consistent results. Blay et al9 analyzed 5 melanomas of unknown subtype for PKCθ staining and found them all to be negative; similarly, Motegi et al13 found that 2 melanomas of unknown subtype were negative for PKCθ. Duensing et al14 performed immunoblot analysis on 2 melanomas and determined weak, but not absent, PKCθ expression in both. However, Wong and Shelley-Fraser15 stained 5 melanomas of unknown subtype and determined that 3 of them (60%) demonstrated strong positivity for PKCθ.

To date, only 2 published studies have looked at DOG1 expression in melanomas. Espinosa et al11 examined 10 desmoplastic melanomas and found positive expression in 1 case (10%); no other subsets of melanoma were included in the study. The 5 melanomas studied by Wong and Shelley-Fraser15 did not demonstrate positive staining for DOG1.

Our study also found low rates of PKCθ and DOG1 positivity in cutaneous melanoma. PKCθ expression was observed in only 18% of primary cutaneous melanomas, and even fewer (3%) expressed DOG1. Furthermore, both DOG1-expressing melanomas showed staining in fewer than 25% of cells. PKCθ has already been shown to be typically negative in mesenchymal tumors such as leiomyoma and schwannoma.9,13 This study confirms that the majority of primary cutaneous melanomas are also typically negative.

DOG1 is considered a specific marker for GIST.11 This study expands the list of neoplasms that can be excluded with DOG1 positivity by including primary cutaneous melanoma. This may be of value in cases such as spindle-cell anorectal lesions, in which the differential diagnosis includes melanoma (S-100 protein+, c-kit+, and DOG1–) vs GIST (S-100 protein–, c-kit+, and DOG1+).31,32 While c-kit expression is seen in cutaneous melanomas and GISTs, the lesions differ biologically in many other respects, including the different rates of expression of PKCθ and DOG1 and the difference in frequency of mutations in genes such as c-kit and PDGFRA.


Upon completion of this activity you will be able to:

  • discuss the expected results of immunohistochemical staining for c-kit (polyclonal and monoclonal), protein kinase C θ, and DOG1 on primary cutaneous melanomas.

  • correlate the results of c-kit immunostaining in primary cutaneous melanomas with prognostic factors associated with the lesion.

The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.

The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.

Questions appear on p 153. Exam is located at www.ascp.org/ajcpcme.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
View Abstract