Expression of the antiapoptotic protein survivin has been demonstrated in some melanocytic lesions and is believed to be required for melanoma cell viability. However, its diagnostic value in differentiating melanomas from nevi has not yet been examined.
Tissue microarray blocks were constructed with paraffin-fixed tissue of 19 nevi, 18 dysplastic nevi, 24 malignant melanomas, and 31 metastatic melanomas. Sections were then reacted with three antisurvivin antibodies (two monoclonal and one polyclonal) assessing labeling intensity (absent or weak, and moderate to strong) as well as the percentage of cells labeled (< 25%, ≥ 25%).
Of the antibodies evaluated, the polyclonal one was found to be the most sensitive. Nuclear immunoreactivity for survivin (i.e., ≥ 25% of cells exhibiting and/or at least moderately intense staining) was seen in a subset of melanomas but not in nevi or dysplastic nevi (P < 0.05)
Survivin is a member of the inhibitor of apoptosis (IAP) family of proteins.1 The main function of survivin is to inhibit apoptosis, and it is thus overexpressed in many cancers.2–10 Survivin is expressed in both melanoma9, 10 and benign melanocytic neoplasms,9, 11–13 as well as in keratinocytic neoplasms.14 It has been shown to be required for the maintenance of melanoma cell viability.9 For these reasons, survivin in melanoma has been investigated both for its value as a prognostic marker10 and as a potentially important therapeutic target.15 However, any possible utility of the analysis of survivin expression in the histologic diagnosis of melanocytic neoplasms is not known.
Given the importance of survivin in tumor progression and its potential role in both prognosis and treatment of melanoma, we decided to investigate the diagnostic applications of survivin labeling. In this study, we describe the expression of survivin in a spectrum of melanocytic lesions, ranging from benign nevi and dysplastic nevi to melanoma and metastatic melanoma, and we show that significant nuclear survivin is detectable only in a subset of melanomas and not in nevi.
MATERIALS AND METHODS
All the cases were selected from the files of the Department of Pathology of the UT-MD Anderson Cancer Center and consisted of a total of 90 cases of melanocytic lesions, including benign nevi (BN) (n = 19), dysplastic nevi (DN) (n = 18), primary malignant melanomas (n = 24) (ranging from Clark I to IV, Breslow thickness 0.47–5.5 mm, and with and without ulceration), and metastatic melanomas (n = 31) (subcutaneous, visceral, and nonsentinel Stage III node metastases) (Table 1).
Hematoxylin and eosin (H&E)-stained slides were reviewed from each case to define the representative areas; 0.6 mm (for punch biopsy specimens) and 1.0 mm (for excision specimens) cylindrical tissue cores were punched from the paraffin blocks. At least two tissue cores were taken for each case except for four small nevi and three dysplastic nevi, in which only one core was removed. The selected tissue cores were inserted into three standard 4.5 × 2 × 1 cm paraffin blocks using a manual tissue arrayer (Beecher Instruments, Silver Spring, MD) with an edge-to-edge distance of 0.1–0.15 mm. Two cores from the same case (benign nevus and melanoma) were included in all three blocks as an internal control. Five-micron tissue sections were cut and one section was stained with H&E to verify the presence of the lesions.
Antisurvivin Antibodies and Immunohistochemical Analysis
The three antisurvivin antibodies (two monoclonal and one polyclonal) used were commercially available (rabbit polyclonal antibody and two mouse monoclonal antibodies, clones 32.1 and 58.22; Novus Biologicals, Littleton, CO). The polyclonal antibody, like the others, is affinity-purified and is specific for human survivin, recognizing a single 16.5 kD band on Western blot. Reactivity was confirmed with HeLa cells as a positive control. The tissue microarray slides were reacted with the antisurvivin antibodies using standard procedures. In brief, as an antigen retrieval technique, slides were microwaved for 5 minutes in citrate buffer and incubated overnight at 4 °C with antibody at a 1:1000 dilution (0.0010 mg/mL polyclonal antibody; 0.0185 mg/mL monoclonal clone 32.1 antibody; and 0.0016 mg/mL monoclonal clone 58.22 antibody). The labeling intensity was graded as none to weak (= negative) or moderate to strong (= positive). In addition, the percentage of cells labeling was graded as < 25% (negative), or ≥ 25% (positive).
For both nuclear and cytoplasmic labeling, a chi-square test was performed to determine an association between number of positive cells (negative < 25% and positive ≥ 25%) and intensity (none/weak, negative; moderate/strong, positive). Fisher exact tests were performed to compare all the different groups. An overall significance of 5% (i.e., P < 0.05) was used. However, given the multiple comparisons, a Bonferroni adjustment to the individual significance was employed to prevent spurious statistical significance by chance (i.e., P < 0.005 was used to compare different groups).
Of the three antibodies selected, we found the strongest signal with the polyclonal one. The two monoclonal antibodies tested did not exhibit significant labeling on these paraffin-embedded sections.
Polyclonal Antisurvivin Antibody Labeling of Nuclei (Table 2)
Two parameters, number of cells and intensity of labeling, were compared between the following groups: benign nevi (BN), dysplastic nevi (DN), malignant melanoma primary, subcutaneous/lymph node metastases, and visceral metastases. As shown in Table 2, none of the nevi (benign or dysplastic) exhibited significant nuclear staining (i.e., ≥ 25% of cells labeled, or moderate to strong labeling). With respect to intensity of labeling intensity, moderate to strong labeling was seen only in melanoma (i.e., with a 100% specificity and sensitivity of 71% in primary melanoma, 80% in visceral metastases, and 38% in subcutaneous/lymph node metastases) but not in nevi (BN/DN) (Fig. 1) (P < 0.005). With respect to number of cells labeled, ≥ 25% labeling was seen only in melanoma (i.e., with a 100% specificity and sensitivity of 29% in primary melanoma, 40% in visceral metastases, and 10% of subcutaneous/lymph node metastases; see Table 2) (P < 0.05 for BN/DN as compared with primary melanoma or visceral metastasis; however, given the small numbers of labeled subcutaneous/lymph node metastases [2 of 21], the difference between this subset and BN/DN was not statistically significant). There were no statistically significant differences between intensity of labeling or number of cells labeled between the different melanoma groups (primary vs. subcutaneous/lymph node metastases vs. visceral metastases). Significant differences in nuclear labeling between melanomas of different Clark levels/Breslow thickness and the presence or absence of ulceration were not seen.
Five of 19 BN and 2 of 18 DN exhibited only very weak nuclear labeling, and thus were considered negative.
Polyclonal Antisurvivin Antibody Labeling of Cytoplasm (Table 3)
As for nuclear labeling, the number of cells exhibiting cytoplasmic labeling and intensity of labeling were compared across the different groups (Fig. 2). Although intensity of labeling appeared to be somewhat increased in the different melanoma groups as compared with BN/DN, this was statistically significant (P < 0.05) in only BN versus primary/visceral melanoma. The differences between the other groups (i.e., BN vs. subcutaneous/lymph node metastases, DN vs. melanoma, and between the different melanoma groups) were not significant. Similarly, the apparent differences in number of cells with cytoplasmic labeling were not statistically significant between the BN/DN versus melanoma or between the different melanoma groups.
Polyclonal Antisurvivin Antibody Labeling of Normal Skin Structures
This antibody was also found to decorate keratinocytes, eccrine glands, vascular wall muscle, and arrector pili muscle (Fig. 3). Dendritic cells were also labeled. As these were mostly suprabasal, they are likely Langerhans cells. Serial sections examined with CD1a and polyclonal survivin antibody support this view. Whereas dendritic cells are known to express survivin, to the best of our knowledge survivin expression in cutaneous Langerhans cells has not been previously well documented or characterized. Melanocytes present in the epidermis adjacent to the lesions did not express survivin.
Survivin is a member of the antiapoptotic IAP family. This class of proteins contains 1–3 copies of a zinc finger fold, known as the baculovirus IAP repeat (BIR) and a COOH-terminal RING finger domain. Survivin, a 16.5-kD protein, is the smallest member of this family, with only one BIR. It appears to achieve its antiapoptotic effect by blocking mitochondrial-induced apoptosis and not by caspase inhibition.17–19 In addition, survivin is known to participate in cell division, partly by preserving microtubule stability.20–22 Blocking survivin leads to apoptosis and disruption of the cell cycle.23 These findings may explain why survivin portends more aggressive prognosis in a variety of tumors,2–4, 6, 24–27 including increased resistance to therapy.28–30 An interesting study by Allen et al.31 appears to implicate survivin as a facilitator of tumor progression rather than tumor initiation (via oncogenic transformation).
The most significant finding we report is the expression of nuclear survivin in melanoma but not in nevi. Cytoplasmic survivin was not significantly different between nevi and melanoma (with the exception of nevi exhibiting weaker cytoplasmic immunoreactivity as compared with primary melanoma/visceral metastases).
In this context, it useful to review that survivin exists in two subcellular pools: 80% cytosolic and 20% nuclear.20, 21, 32 The cytosolic pool includes survivin complexed with centromeres, microtubules, and other components of the mitotic apparatus,21, 32, 33 whereas the nuclear survivin localizes with kinetochores of metaphase chromosomes.20, 21 The significance of these pools may be that survivin plays a regulatory role in various phases of mitosis.33 Increased survivin as seen in malignancies may therefore participate in aberrant mitotic activity leading to aneuploidy in tumors.33
Given this background, it is perhaps not surprising that nuclear survivin has been associated with a poor prognosis in several malignancies (e.g., hepatocellular carcinoma,34 esophageal carcinoma,7 mantle cell lymphoma35). Confusingly, nuclear survivin has been found to be a favorable prognostic marker in other malignancies (e.g., lung carcinoma,36 breast,37 osteosarcoma38). Li et al.39 have provided an excellent metaanalysis of nuclear survivin in an attempt to understand and possibly reconcile the contradictory roles of survivin in different tumors. Among other factors, they suggest that some of the confusion in the data may be related to incorrect interpretation of immunohistochemical data, because immunohistochemical processing may lead to darker nuclei that may subsequently be read as positive. To avoid misinterpretation, we chose a cut-off of at least moderate intensity in ≥ 25% of cells as positive labeling.
In contrast to nevi/melanoma, normal melanocytes do not appear to express survivin, as shown previously.9 Because survivin has antiapoptotic activity, it may contribute to the known resistance of melanocytes in either benign or malignant lesions to enter apoptosis.
Our findings support the idea that survivin may prove to be a valuable prognostic factor and a possible therapeutic target. Gradilone et al.10 have shown that survivin expression in melanoma metastatic to sentinel lymph nodes correlates with unfavorable outcome. Melanoma cell lines in which expression of survivin is artificially decreased are more susceptible to treatment with cisplatin.40 Also in cells lines, expression of mutated survivin results in decreased tumor growth and increased apoptosis.15 In addition, spontaneous cytotoxic T-cell responses against survivin-derived epitopes have been detected both in situ and ex vivo in cancer patients.41
In summary, our study confirms the presence of the antiapoptotic protein survivin in a majority of nevi and melanomas. More significantly, our study suggests that a proportion of melanomas, but not nevi, have increased nuclear survivin, a fact that may be helpful in the histologic differential diagnosis of benign versus malignant melanocytic lesions. Also, this finding may have a bearing on the biologic behavior of melanoma and may have prognostic and treatment implications.