p16 expression in primary malignant melanoma is associated with prognosis and lymph node status

Authors


Abstract

Lymph node (LN) status is an important prognostic factor in melanoma patients. p16 expression and proliferation rate (MIB-1) of primary melanomas have been suggested as a marker of metastatic potential. In this study, the correlation of p16 expression and the proliferation rate (MIB-1) with LN status and tumor-specific survival was investigated in primary melanomas. MIB-1 and p16 expression were analyzed by immunohistochemistry in 64 patients with primary cutaneous melanoma. Thirty four nevi were used as control. All patients underwent sentinel lymph node staging. Three different p16 staining patterns were observed: a combination of nuclear and cytoplasmic staining, only cytoplasmic staining and absence of p16 expression. All 34 nevi displayed a nuclear and cytoplasmic p16 staining, whereas p16 was negative in 14 of 64 (22%) melanomas. The level of p16 expression gradually decreased from benign nevi to melanoma without metastasis to melanoma with metastasis. There was a significant correlation between cytoplasmic p16 expression and absence of metastasis (p < 0.05). Death of disease correlated with absence of p16 immunostaining (p = 0.01). MIB-1 expression was not associated with survival. These results confirm the relevance of p16 expression as a prognostic marker in melanoma patients. In addition, it was shown that cytoplasmic immunostaining for p16 in primary melanoma might serve as a predictor of the LN status. Therefore, immunohistochemical evaluation for p16 expression is of potential value for treatment planning in melanoma surgery. © 2005 Wiley-Liss, Inc.

Cutaneous malignant melanoma remains the leading cause of skin cancer death in industrialized countries. Up to date, the only possible therapy to cure melanoma patients is surgical excision of localized, nonmetastatic primary cutaneous melanoma (stage I and stage II patients).1 Unfortunately, 20% of clinically stage I and II patients already have micrometastatic disease at diagnosis. Currently, there is no cure for patients who present with visceral melanoma metastases. Therefore, identifying patients at increased risk for metastases is one of the most critical issues in the management of melanoma. These patients have to be considered for adjuvant therapy.

Several prognostic factors have been identified. These include age, gender, anatomical site,2 Breslow tumor thickness,3, 4, 5 ulceration, mitotic rate,6 tumor infiltrating lymphocytes,7 vascular invasion and metastasis.8, 9, 10, 11 Recent studies have identified varies molecular markers as additional alterations for the metastatic potential of malignant melanoma.12

The product of the p16/INK4a/CDKN2/MTS tumor suppressor gene acts as a negative cell cycle regulator by inhibiting G1 cyclin-dependent kinases, which are phosphorylating the retinoblastoma protein.13, 14p16 is inactivated by mutation, promotor methylation or loss of heterozygosity in a wide range of human malignancies,15, 16 including familial and sporadic melanoma.17, 18 Furthermore, loss of p16 expression in primary melanomas has been suggested as a predictive marker for tumor progression.19, 20

The Ki 67 antigen is expressed in proliferating cells21 and can be detected in formalin-fixed tissues using the MIB-1 antibody.22 The proliferation marker MIB-1 has been used previously as an adjunct to histological diagnosis of malignant melanoma because the proliferation rate is of value in the distinction of malignant melanoma from benign nevi.23 Furthermore, MIB-1 has been shown to be an independent prognostic marker for melanoma patients with respect to disease-free and overall survival.24

The concept of sentinel lymph node (SLN) biopsy for melanoma patient has been introduced by Morton et al. in 1992.25 Nowadays, the SLN biopsy has become the recommended method for determining the histopathologic status of patients with intermediate-thickness (>1.0 mm, ≤4.0 mm) melanoma.1 Preliminary studies have shown 3 years disease-free survival for patients with negative and positive SLN of 88.5 and 55.8%, respectively.26 Considering the low incidence (4–7%) of SLN metastases in thin melanomas,27, 28 there is no common consent whether SLN biopsy should be done or not for thin melanomas.

For a better preoperative planning of the SLN procedure, predictive markers of the lymph node (LN) status are desired. In this study, the predictive value of p16 and MIB-1 expression for the LN status was determined retrospectively in a patient collective, which has been included in a recent SLN study.29

Material and methods

Patients

The patient collective derives from a previous SLN biopsy study.29 Sixty-four patients of this study with primary cutaneous melanomas and consecutive SLN biopsy were included in the present study. In 54 of 64 patients, a clinical follow-up was available (mean clinical follow-up 34 ± 12 month). All patients had a newly diagnosed primary cutaneous malignant melanoma of Breslow's tumor thickness equal or over 1.0 mm. To exclude metastatic disease before the SLN biopsy, a staging procedure within 4 weeks after primary excision with physical examination, chest radiography, ultrasound of abdomen and regional LNs, and whole body positron emission tomography was performed.30 Wide excision (tumor thickness 1.01–3.99 mm: 2-cm margin; tumor thickness ≥4.0 mm: 3-cm margin) and SLN were performed in the absence of detectable distant metastasis.

SLN was investigated on different levels by hematoxylin and eosin (H&E) and immunohistochemistry (HMB-45, S-100 and Pan Melanoma Plus) according to a SLN-protocol.31, 32 A complete lymphadenectomy was performed if the SLN showed metastases or isolated tumor cells (pN0(i+)).

The patient age ranged from 20 to 75 years. There were 35 nodular, 13 superficial spreading, 1 lentiginous, 7 acral lentiginous, 7 not otherwise specified and 1 desmoplastic malignant melanoma. Thirty patients had a Breslow tumor thickness equal or below 2 mm, 34 patients over 2 mm. Forty-three of 64 patients (67%) had a pN0(sn;i−) SLN stage category. Twenty-one of 64 patients (33%) had a pN0(i+), pN1 or pN2 SLN stage category (5 with isolated tumor cells). The 5 patients (8%) with isolated tumor cells in the SLN (pN0(sn;i+)) underwent lymphadenectomy. Metastatic disease in other LN was detected in 3 of these 5 patients. A total of 19 patients had a pN1 or pN2 after the completion of the lymphadenectomy. Five patients with negative SLN developed clinically obvious LN (n = 4) or organ (n = 1) metastasis during the follow-up. In total, 8 of the 54 (15%) patients died of metastatic disease. All except one of those had positive SLNs at the time of primary diagnosis. For clinical data, see Table I. Thirty-four skin nevi (8 dermal, 22 compound and 4 junctional) were also tested for p16 expression.

Table I. Clinical Data and Immunophenotype (p16, MIB-1) of Malignant Melanoma Patients
Age/SexBreslow (mm)Melanoma typep16 nuc and cyto pos (%)p16 cyto pos (%)MIB-1 (%)pN(sn) statuspN status1
  • SSM, superficial spreading melanoma; NMM, nodular malignant melanoma; LMM, lentigo maligna melanoma; ALM, acral lentiginous melanoma, NOS, malignant melanoma not otherwise specified, Desmo, desmoplastic malignant melanoma; p16 nuc and cyto pos, p16 nuclear and cytoplasmic positive; p16 cyto pos, p16 exclusively in the cytoplasma positive.

  • 1

    Lymphadenectomy status after sentinel biopsy (– not performed).

  • 2

    Patient died of metastatic disease.

  • 3

    Developed lymph node metastasis during the follow-up.

  • 4

    Developed organ metastasis during the follow-up.

70/m2.25NMM0021N0
74/m2.25NMM010014N1(mic)N1(mic)
67/f1.6SMM40601N0
52/f1.25NOS0040N0(i+)N2
61/m1.1SSM07033N0
29/f1.6NMM0605N0
38/m1.6NMM302013N0
39/f2.5NMM0517N0
39/f217NMM0049N1(mic)N1(mic)
45/m2.7NMM04010N0
63/f2.25NMM70303N1(mic)N1(mic)
54/f1.1SSM009N0
58/m22.4NMM01024N1N2
38/f2.9ALM202013N0
26/f7NMM0959N03
51/m2NMM401015N0
71/m3NMM0035N0
20/f210NMM05015N1(mic)N2
69/m4.75NMM01017N0
37/f1.15NMM01009N0
67/m22.75NMM0019N1(mic)N2
44/f1.7NMM007N0
74/f24.4NMM0535N1N2
73/f2NMM0051N2N2
43/m6NMM0030N0
48/f1Desmo0011N1N1
43/2.2ALM0514N0
28/f1.08NOS0530N0(i+)N1
71/f2NOS0605N0
53/f5NMM109048N03
37/m2.36NOS10109N0
44/f3NOS503033N0
75/m2SSM01017N0
49/f2.22NMM35354N0(i+)N1
50/m1.8SSM0709N1N1
46/m2.3SSM15708N0
33/m3.5NMM108025N0
34/m3.5NMM15033N0
54/m2.1NMM04011N0
69/m1.7SSM0015N0
59/m3.8NMM0012N0
49/m3.8NMM100036N2N2
50/m1.2NOS10106N2N2
61/f1.6SSM010013N03
60/m1.35SSM010011N0
59/m1.7NMM703023N0
29/f2NMM005N0
33/m2NOS01021N0
60/m1.3SSM09051N0
66/m29ALM1009N04
64/f2.25SSM07015N0
66/m22ALM0015N2N2
47/f1.05ALM30409N0
73/f2LMM307016N0
21/f21.1SSM03015N1N2
39/f1.7ALM45504N0
52/m4.5NMM901026N1(mic)N1(mic)
36/m1.1NMM40604N0
62/m3.4NMM208019N0
34/f1.6NMM307010N0
56/f2.5ALM40609N0(i+)N0(i+)
57/m2.4NMM50451N0
66/f5.2NMM301030N1(mic)N1(mic)
50/f2.4NMM653027N0(i+)N0(i+)

Approval was obtained from the local Institutional Ethical Committee and written informed consent signed by all study participants.

Immunohistochemistry

All tissue samples were formaldehyde-fixed and paraffin embedded. Hematoxylin and eosin-stained sections of all samples were reviewed by 1 pathologist (DM).

p16 and Ki-67 immunohistochemistry were performed on paraffin sections of formalin-fixed tissues, using a Ventana Benchmark automated staining system (Ventana Medical Systems, Tucson, AZ). For antigen retrieval, slides were heated with cell conditioner. Primary antibodies (p16 from Neomarkers, Freemont, CA and Ki-67 from Dako Cytomation, Glostrup, Denmark) against p16 (Clone 16P04, dilution 1:600) and Ki-67 proliferation antigen (clone MIB-1, dilution 1:20) were applied. Monoclonal mouse p16 and MIB-1 antibodies were detected with the Ventana iVIEW DAB detection kit, yielding a brown reaction product. Slides were counterstained with hematoxylin.

In p16 labeling, we differentiated nuclear and cytoplasmic staining from exclusive cytoplasmic staining (Fig. 1) because imaging of immune labeled p16 by scanning laser confocal microscopy indicated that the p16 protein is distributed to both the cytoplasm and the nuclei of normal and melanoma cells. Therefore, we did not interpret cytoplasmic staining as nonspecific as others did.33, 34 Only nuclear staining was interpreted for MIB-1.

Figure 1.

Compound nevus (a, b and c). (a) H&E (×40); (b) and (c) extensive nuclear and cytoplasmic p16 expression (×80 (b) and ×500 (c)). Malignant melanoma (d, e, f and g). (d) H&E (×40); (e) p16 expression in noninvasive melanoma cells (arrow), but not in invasive melanoma cells (×80); (f) almost all melanoma cells are negative for p16 (×500); (g) Exclusive cytoplasmic p16 expression in melanoma cells (×500) at the dermoepidermal junction.

The extent of each staining pattern of p16 and MIB-1 was recorded as the number of positive nuclei or cells per 100 melanoma respectively 100 nevus cells.

Immunohistochemical labeling in less than 5% of melanoma or nevus cells was recorded as negative and labeling of more than 95% was registered as 100% positive (range 5–95). The p16 expression was separately analyzed in the different cell compartments and evaluated in 3 categories: combined nuclear and cytoplasmic staining; only cytoplasmic staining and absence of p16 staining.

Controls

All positive controls displayed an extensive and intensive positive combined nuclear and cytoplasmic staining in more than 80% of cells. All negative controls lacked cells displaying an immune reaction.

Statistical analysis

Two categories of p16 expression patterns were defined for statistical analysis: p16 nuclear and/or cytoplasmic positive staining in equal or more than 50% when compared to less than 50% positive cells. The 50% cut-off was selected for practical purposes, because this cut-off can easily be used in routine practice. The differences between the percentages of positively labeled nuclei or cells were analyzed using the Mann-Whitney test. p-Values below 0.05 were considered as significant. Survival differences between groups were calculated by a log rank test. The Cox-regression analysis was applied for analysis of the association between tumor thickness (Breslow), LN status, p16 expression pattern, proliferation rate (MIB-1) and tumor-specific survival.

Results

p16 expression

All benign nevi displayed nuclear and cytoplasmic p16 staining. p16 immune staining was totally absent in 14/64 (22%) melanomas (Fig. 1). The percentages of p16 labeled cells were significantly higher in benign nevi when compared to malignant melanoma, even if different cell compartments were separately analyzed (Table II).

Table II. p16 and MIB-1 Expression in Benign Nevi and Malignant Melanomas
 nPercentage of positive cells (±SD)
p16MIB-1
Nucleus and cytoplasmCytoplasm onlyNo expression
Benign nevi3467 ± 229 ± 1624 ± 170.2 ± 0.2
Malignant melanoma6416 ± 2434 ± 3450 ± 4118 ± 13
p-Value <0.001<0.001<0.05<0.001

We found a statistically significant correlation between positive cytoplasmic p16 expression in primary melanomas and LN status (p = 0.022; Table III). The percentage of cells with cytoplasmic p16 staining was higher ((40 ± 34)%) in malignant melanoma with negative LN (pN0) when compared to malignant melanoma with LN metastasis ((19 ± 28)%). There was no correlation between negative or combined positive (nuclear and cytoplasmic) p16 expression and LN status. Tumor thickness (Breslow index >1 mm, ≤4 mm) did not correlate with LN status in our patient collective.

Table III. p16 and MIB-1 Expression Compared to Lymph Node Status in Melanoma Patients
pN-categorynPercentage of positive cells (±SD)
p16MIB-1
Nucleus and cytoplasmCytoplasm onlyNo expression
  1. NS, not significant.

pN04515 ± 2040 ± 3445 ± 4116 ± 11
pN1/N21918 ± 3319 ± 2863 ± 4023 ± 15
p-Value NS<0.05NSNS

Combination of proliferation rate (MIB-1) and Breslow tumor thickness with different staining patterns of p16 did not show any significant associations with the LN status in a multivariate analysis.

Proliferation rate (MIB-1 labeling index)

Breslow tumor thickness was associated with high proliferation rate (p = 0.02). Malignant melanoma with a Breslow tumor thickness >2 mm had a MIB-1 index of (20 ± 12)%, whereas the proliferation rate in malignant melanoma ≤2 mm was (16 ± 13)% (not significant).

The percentage of positive cytoplasmic p16 immune staining in melanoma was related to low proliferation rate (p = 0.03). Also, there was a positive trend between high proliferation rate and absence of nuclear and cytoplasmic p16 staining in primary melanomas, but this trend did not reach significance (p = 0.09). When we compared p16 nuclear and/or cytoplasmic positive staining in equal or more than 50% to less than 50% positive tumor cells, there was a significant difference in proliferation activity between the 2 groups (p = 0.033, Table IV). Proliferation activity was significantly higher in melanomas with low p16 expression. There was a trend between existence of LN metastases and high proliferation rate (p = 0.08). Survival was not associated with the MIB-1 labeling index (p = 0.14).

Table IV. p16 Expression and Proliferation Rate (MIB-1 Labeling Index) in Malignant Melanoma
p16 expressionnMIB-1 LI (±SD)p-Value
≥50% Tumor cells2314.09 ± 12.30.033
<50% Tumor cells3118.68 ± 11.7

Tumor-specific survival

Tumor thickness (p = 0.002) and LN status (p < 0.0001) were associated with tumor-specific survival. There was also an association between tumor-specific survival and p16 expression (Fig. 2). Death of disease correlated significantly with reduced p16 expression (p = 0.01), when tumors were grouped into cases with nuclear and/or cytoplasmic expression in equal or more than 50% of cells and into cases with p16 expression in less than 50% of cells. Exclusive cytoplasmic p16 expression in primary tumor was not associated with survival.

Figure 2.

Breslow tumor thickness, sentinel LN status, p16 expression and tumor-specific survival.

When survival rates were compared with MIB-1 labeling index and different p16 expression patterns in a multivariate analysis, absence of p16 expression remained the only independent risk factor. However, when tumor thickness and LN status were included into this analysis, only tumor thickness was an independent risk factor.

Discussion

Our results confirm that there is an association of patient survival with Breslow thickness4 and LN status.26 In addition, it was shown that nevi and primary cutaneous melanoma differed significantly in the p16 staining pattern. Most nevi displayed a nuclear and cytoplasmic staining in more than 70% of cells. Only 2 melanomas showed a nuclear and cytoplasmic staining of more than 70%. Absence of p16 immune staining was never observed in nevi. In contrast, p16 immune staining was completely negative in 22% of melanomas. Therefore, absence of p16 immune staining in association with an increased proliferation rate are strong arguments for the diagnosis of malignant melanoma and might facilitate the differential diagnosis to benign nevi. Further, the results of this study are consistent with our previous findings showing that p16 is a useful marker to distinguish nevi cells in LNs from those of melanoma metastasis.35 This is also in line with findings of Piepkorn and co-workers34 showing the same p16 immune staining pattern in atypical nevi and benign normal nevi.

About 60% of the patients with cutaneous melanoma have a thin melanoma below 1 mm Breslow tumor thickness.36 Such thin melanoma have a 7–15% risk of recurrence, metastasis or death at 10 years.37, 38 Currently, the SLN biopsy is not recommended for patients with a Breslow tumor thickness below 1 mm. It is evident that there is a need to identify the group of melanoma patients with increased risk for SLN metastases.

p16 is a negative cell-cycle regulator and normally expressed in the nucleus.13 Three major mechanisms of p16 inactivation have been described in melanoma and head and neck carcinomas: promotor methylation, homozygous deletion and inactivating mutations.39 In cervical squamous carcinomas, the functional inactivation of pRb by the HPV oncoprotein E7 leads to p16 overexpression in the nucleus and cytoplasm.40 The biological relevance of cytoplasmic immune reactivity of p16 is controversially discussed. Cytoplasmic p16 expression is common in dysplastic cervical epithelium, but genetic alterations of p16 in melanomas are in contrast to cervical cancers where the Rb tumor suppressor function is thought to be functionally inactivated by HPV oncoproteins, resulting in p16 overexpression.40, 41 Alonso et al.19 and Talve et al.33 regarded cytoplasmic p16 staining as an artifact and interpreted such staining pattern as negative. A cytoplasmic/perinuclear p16 expression was already demonstrated in 1999 by Walker et al.42 These authors transfected melanoma cell lines with different CDKN2A cDNAs. They could show cytoplasmic p16 protein localization by immunofluorescence. In addition, distinct p16 variants were associated with predominantly nuclear or cytoplasmic localization. Furthermore, cytoplasmic localization of immune labeled p16 was recently assured by means of scanning laser confocal microscopy34 and electron microscopy.43 A previous study with familial melanoma has also been shown by sequence analysis that different p16 mutants lead to preferential expression of the protein in the nucleus or cytoplasm, whereas wild-type cells have p16 protein in both cellular compartments.44 It seems that different p16 mutations lead to nuclear or cytoplasmic mislocalization as a consequence of different functional impairment of the protein. These results are consistent with our finding that all benign nevi have p16 positivity in both, the nuclear and cytoplasmic compartment, whereas exclusive cytoplasmic p16 expression was observed in more than 10% of melanomas. It would be interesting to determine if melanomas with cytoplasmic immunostaining are associated with distinct p16 mutations. Interestingly, positive cytoplasmic immunoreactivity was associated with a low proliferation rate assessed by MIB-1 in our study. Straume et al. observed in 200 patients with nodular melanomas a p16 cytoplasmic staining in cases lacking nuclear staining.20 These authors also separately presented the 2 staining pattern results. They reported that weak or absent cytoplasmic p16 was significantly correlated with the presence of vascular invasion. These data are consistent with those from our study and suggest that positive cytoplasmic p16 immune staining may be related to a lower metastatic potential of primary malignant melanoma. Straume et al.20 also showed that loss of nuclear p16 protein expression in nodular melanomas independently predicts decreased patient survival. Alonso et al.19 analyzed 165 melanomas with different histological progression phases (radial, vertical growth phase and metastasis) using tissue microarrays and confirmed that p16 is an independent predictor of survival in patients with vertical growth phase melanoma. The probability of death was around 8 times higher when p16 was lost.

In view of the fact that p16 acts as a negative cell cycle regulator, one would expect high proliferation rate in melanomas with absence of p16 expression. Complete absence of p16 expression was found in about 30% of primary malignant melanomas. Importantly, positive cytoplasmic immunoreactivity was associated with a low proliferation rate. However, some tumors with strong p16 expression showed a high proliferation rate, suggesting the existence of different p16 mutations with different consequences for the proliferation activity of malignant melanoma. This is consistent with Walker's observation that not all “p16 melanoma cell variants” have a diminished ability to inhibit cell growth.42

Importantly, our study has shown that the p16 expression pattern in primary malignant melanoma predicts LN status. Pure cytoplasmic expression with absence of nuclear p16 expression was associated with negative LNs. Given the recent findings that specific p16 mutations lead to preferential p16 expression in the nucleus or cytoplasma,44 one might speculate that specific p16 mutations are associated with an increased metastatic risk of malignant melanoma to the LN. Therefore, the p16 expression pattern might be useful as a predictive factor for LN metastasis. However, the exact evaluation of percentage of cells with either cytoplasmic or combined p16 expression is very labor intensive and time consuming. In routine work, exclusive cytoplasmic staining in less than 50% of melanoma cells can easily be recognized and is highly suggestive of LN metastasis and poor prognosis.

In summary, our results confirmed the relevance of p16 as an additional diagnostic marker to differentiate nevi from melanoma. It was shown that pure cytoplasmic p16 immune staining might serve as a predictor of the LN status in primary malignant melanoma. Therefore, immune staining for p16 is of potential value for treatment planning in melanoma surgery.

Acknowledgements

We are grateful to N. Wey for photographic and computer-assisted reproductions.

Ancillary