Immunohistochemical Expression of p16 in Melanocytic Lesions: An Updated Review and Meta-analysis

Cyclin-dependent kinase inhibitors (CDKIs) are essential proteins involved in withdrawing from cell cycle progression. Lack of CDKIs allows activation of cyclin-dependent kinases, which may then phosphorylate retinoblastoma protein (Rb). Unregulated hypophosphorylated Rb initially sequesters E2F transcription factors, thereby acting as a tumor suppressor. The phosphorylation of Rb releases the bound E2F. Free E2F then acts to progress cell cycle progression by the transcription of genes required for mitotic S phase entry and DNA synthesis.^16–18 

There are 2 families of CDKIs.¹⁷  The CIP/KIP family of kinase inhibitors includes p21, p27, and p57. The INK4 family of CDKIs include p15^INK4b, p16^INK4a (referred to herein as p16), p18^INK4c, and p19^INK4d. The most studied among these are p21 and p16.¹⁶  It is thought that induction of cellular senescence is established and/or maintained by either or both of the p53/p21 and p16/pRb pathways.^19–22  The upregulation of p16 has been shown to respond to hypermitogenic/oncogenic signals such as overexpression of RAS, MAP kinases, or Myc. Multiple other stimuli for upregulation of p16 have also been proposed, such as response to DNA damage.^16–19 

Among the CDKIs, there has been great interest in studying the relation of p16 with melanoma. This may partially be due to the linkage of chromosome 9p21, which includes the p16 gene, to familial melanoma. Although germline mutations of p16 were strongly correlated with familial melanoma, there was no definitive proof that p16 mutations or deletions were absolute requirements for melanoma, because not all melanomas have p16 mutations. Interestingly, few nondiseased controls were also detected to have p16 mutations. Atypical/dysplastic nevi were also analyzed and had a lesser rate of p16 mutations compared with melanomas.²³ 

The role of p16 in cellular senescence, particularly in melanocytes, is also of great importance, because senescence is considered to be a critical barrier for tumorigenesis or progression of melanoma.²⁴  In vitro cell culture studies further support this view. Analysis of human melanocytes revealed that cells depleted of p16 displayed enhanced proliferation and an extended replicative lifespan in the presence of replication-associated DNA damage.²⁵  Additionally, similar findings were seen with mouse melanocyte culture studies. Mouse Ink4a-Arf^−/− melanocytes were shown to readily become immortal upon culture, whereas wild-type Ink4a-Arf^+/+ cells senesced within 4 to 5 weeks of culture. Restoration of p16 by retroviral vectors containing normal p16 complementary DNAs into Ink4a-Arf^−/− melanocytes stopped growth of the cells.²⁶ 

Within melanocytes, the loss of p16 (by mutation, deletion, or silencing) is thought to be a possible contributing factor for tumorigenesis, consistent with our current understanding of the molecular biology of tumor suppressor genes/products. This is in contrast to malignant human papillomavirus–driven tumors (which include mostly anogenital, gynecologic, and oropharyngeal tumors) that show increased p16 expression by immunohistochemistry. This is thought to be due to the integration and expression of E6 and E7 viral oncoproteins. E7 inactivates pRb, preventing the sequestration of E2F transcription factor. The increase in E2F then leads to upregulation of p16 by a feedback mechanism. E6, on the other hand, acts by degrading p53, preventing apoptosis.^27,28 

Observation of germline mutations within melanoma cases and also through experimental basic science has led to the implication of p16 in the tumorigenesis of melanoma. Many independent groups have therefore studied the possible incorporation of p16 immunohistochemistry for the purpose of distinguishing benign from malignant melanocytic lesions, particularly in sporadic melanoma cases. A review and meta-analysis of studies analyzing nevi, atypical melanocytic lesions, and melanoma is presented.

Multiple studies^29–71  evaluating p16 expression by immunohistochemistry of melanocytic lesions were reviewed, analyzed, and tabulated. Essential information (number of samples, antibody used, collection dates, overall p16 immunohistochemistry results, and general method of interpretation) from each study are listed in the Table. Most studies showed a high proportion of benign cutaneous nevi to stain positively for p16 immunohistochemistry. The reported percentage of benign nevus cases staining positive ranged from 61% to 100%.* Most cases, however, reported p16 positivity close to 100% for benign nevi. For primary cutaneous invasive melanomas,^† the percentage of cases with positive p16 staining reported by multiple studies ranged from 12% to 93%. Most studies (10) reported percentages within the 40s to 60s. There were 4 studies that reported less than 40% and 2 studies that reported greater than 80%. Metastatic melanoma cases,^‡ on the other hand, ranged from 0% to 71% of cases staining for p16 immunohistochemistry. Because each group had differing methodologies and reporting styles, not all studies were included, and the aforementioned range of percentages was derived from studies that defined p16 immunohistochemistry simply as either positive or negative. Still, the definition of what was interpreted as positive versus negative varied from study to study. Excluded were studies^§ that used a scale or grading-type system in reporting their p16 immunohistochemistry results, because they did not clearly distinguish a positive versus negative result but rather a gradient of staining patterns or relative expression. It is evident that the range of reported percentages is wide and appears insignificant. In general, however, there does appear to be a decreasing trend of p16 staining going from benign to advanced melanoma (especially metastatic lesions). This may also be highlighted by one of the studies (Ghiorzo et al⁴⁵ ) that showed 1 subset of primary invasive melanoma with smaller than 0.8-mm Breslow thickness to have 51% relative p16 protein expression, and another subset larger than 0.8 mm to have 21% relative p16 protein expression. Relative expression within this unique study was based on staining intensity derived from using benign nevi as reference for comparison. Unfortunately, however, the wide variation in the numerous studies appears to undermine the possible value of p16 immunohistochemistry in its application for melanocytic lesions. It is uncertain, though, whether this wide variation could be due to subjective interpretation, unstandardized laboratory techniques, source of the antibody, or other factors. There is too much variability to consider; however, it is possible that digital imaging, as used by Ghiorzo et al,⁴⁵  may assist in the reduction of intraobserver variability, although variations could also exist among imaging hardware and software.

Figure 1, a through h, illustrates the variability of p16 immunohistochemistry staining patterns experienced with melanocytic lesions. For baseline comparison, a nevus with corresponding p16 immunohistochemistry characteristically shows strong and diffuse staining pattern of nevocytes (Figure 1, b). A more variable staining pattern is illustrated in primary invasive cutaneous melanoma, in which there is mostly cytoplasmic staining of the invasive cells (Figure 1, d), versus negative staining (Figure 1, f), versus cases in which there is heterogenous positive and negative staining of the invasive cells (Figure 1, h).

Multiple studies have also incorporated specialized types of melanocytic lesions, such as Spitz lesions,^{58–61,64,66,68}  desmoplastic melanomas,^58,67,71  blue nevus–related lesions,^69,71  and mucosal melanocytic lesions^41,49,62,63 ; 2 studies^43,70  focused on nodal metastases, and a recent study (S. S. Koh, unpublished data, 2018) compared nevi of pregnancy and nevoid melanomas. The wide range of variable results is further highlighted when analyzing studies from Spitz nevi. The percentage of Spitz nevi cases positive for p16 immunohistochemistry reported by multiple groups^{30,34,50,61,64}  ranged from 0% to 100%. Thus, the entire range of staining characteristics disfavors the use of p16 immunohistochemistry for spitzoid lesions. Mason et al⁶⁴  showed no difference in p16 staining in Spitz nevi and spitzoid melanomas, concluding that p16 was not a useful marker to distinguish the two. Additionally, George et al⁶⁰  had similar conclusions for atypical Spitz nevi, having shown that their p16 immunoreactivity is intermediate between those of Spitz nevi and melanoma, although they claimed that p16 was useful for discriminating Spitz nevus from melanoma. In contrast, Hilliard et al⁵⁸  (desmoplastic Spitz nevi versus desmoplastic melanoma) and Al Dhaybi et al⁶¹  (Spitz nevi versus spitzoid melanoma) reported profound differences in expression in benign versus malignant cases and proposed p16 immunohistochemistry expression as an aid for differentiation. Although it is inconclusive for diagnostic purposes in spitzoid lesions, Horst et al⁶⁶  and Yazdan et al⁶⁸  proposed p16 immunohistochemistry as a possible screening tool for 9p21 cytogenetic abnormalities. Similar to spitzoid lesions, multiple studies analyzing nevi and primary invasive melanomas have also had mixed conclusions: some studies^52,62  support p16 immunohistochemistry for diagnostic purposes (differentiating nevi from primary invasive melanoma), and some studies^29,65  do not show convincing evidence. Interestingly, in the 2 studies^43,70  analyzing nodal metastases, both supported the use of p16 immunohistochemistry as a diagnostic utility to distinguish nodal nevi from metastatic melanoma.

From the molecular level, some authors^29,54  have suggested that p16 loss is not necessary for tumor initiation, perhaps because it is retained mostly in melanoma in situ and primary invasive melanoma. Others^31,34,63  have, in contrast, supported the possibility that p16 may be contributory as a primary event. Some³⁵  have proposed that the gradual loss of p16 correlates with the progression of melanoma, but is not an initiating event. Most studies did, however, show gradual loss of p16 with melanoma progression (see Table). Furthermore, 2 studies^31,39  correlated loss of p16 with increased Ki-67, supporting its role in the loss of regulating and inhibiting cell cycle entry.

Although some studies incorporated atypical nevi in their studies, most did not analyze p16 immunohistochemistry staining characteristics based on grade of atypia. Chang and Cassarino⁶⁹  evaluated atypical cellular blue nevi, including cases with mild, moderate, and severe atypia, versus melanoma (including cases of malignant cellular blue nevi). They found that mildly and moderately atypical cellular blue nevi could be distinguished from severely atypical cellular blue nevi and melanomas when p16 was used within a panel of markers. Unfortunately, however, severely atypical cases showed loss of p16 similar to melanomas, and were therefore not distinguishable from melanoma based on p16 staining. It may be reasonable to raise the possibility that the severely atypical cases could have been underdiagnosed. However, the authors reported low Ki-67 staining pattern for the severely atypical cases that were similar to mild and moderately atypical cellular blue nevi and distinctly different from the high Ki-67 present in melanoma.

Apart from its use in diagnostics, p16 immunohistochemistry expression has also been explored as a potential prognostic marker. Similar to its use for diagnostics, p16 was found to have mixed results as to its role in prognostication. Some studies proposed lack of p16 to be associated with recurrent disease,³²  predicting decreased patient survival,^39,44  and as a predictor of lymph node metastasis.⁴⁸  Sanki et al,⁵¹  in contrast, proposed that p16 immunohistochemistry expression did not reliably predict recurrence or survival. Additionally, for sinonasal melanomas, loss of p16 did not correlate with prognosis.⁴⁹ 

Analysis of multiple studies shows a wide range of results, as previously discussed. The variation could be due to how the observer is interpreting the results. One of the main differences seen among studies is whether positive staining was considered to be nuclear and/or cytoplasmic. Reanalysis was performed by separating studies depending on whether nuclear alone or nuclear and cytoplasmic staining was considered positive for p16 staining (Table). For benign nevi, 89% to 100% of cases positive for p16 was the range reported^31,34,50,54  if only nuclear staining was considered positive, compared with 61% to 100% for those studies^{29,31,43,52,61}  that used both nuclear and cytoplasmic staining for positivity. For primary invasive melanomas, studies^{31,33,34,50,54}  considering only nuclear staining had a range of 50% to 68% compared with studies^{29,30,37,42,51,52}  considering both nuclear and cytoplasmic staining showing a range of 12% to 91%. For metastatic melanomas, studies^33,34,37,56  considering only nuclear staining had a range of 0% to 64% compared with studies^29,43,51  considering both nuclear and cytoplasmic staining showing a range of 2% to 56%. By analyzing the studies as 2 groups (one group interpreting only nuclear staining, and a second group interpreting both nuclear and cytoplasmic staining), it appears that the group considering only nuclear staining characteristics had a shorter range, or less variability, from study to study. These results may suggest that the use of only nuclear staining for interpreting p16 immunohistochemistry for melanocytic lesions may be more optimal. However, it is unclear at the moment whether only nuclear staining is biologically relevant and whether cytoplasmic staining should be considered.

Figure 2 illustrates the staining patterns encountered with p16 immunohistochemistry in melanocytic lesions. Strong staining is usually encountered with benign nevi. Here we see an example of a nevus showing strong cytoplasmic and nuclear staining in many cells (Figure 2, a). In contrast, an example of primary invasive melanoma with the majority of cells staining is present, but the staining pattern is mostly cytoplasmic, with only a few cells showing nuclear staining (Figure 2, b). In this example, if only nuclear staining is considered within this illustrated field, then positivity may be interpreted as being around 5%. However, if cytoplasmic staining is considered, then overall positivity may be interpreted as being 50% to 60%. Depending on the methodology one uses, results may be very different, as illustrated in this example. As experienced by many pathologists, variability of staining is inevitable, as seen in another example (Figure 2, c) of a primary invasive melanoma showing many cells with weak cytoplasmic staining and no nuclear staining (yellow arrow), cytoplasmic staining without nuclear staining (red arrow), and an adjacent cell showing strong nuclear and cytoplasmic staining (green arrow). The point at which to call a positive-staining cell is quite arbitrary, and unfortunately is left to the discretion of the observer.

The use of p16 for diagnostic purposes in melanocytic lesions appears limited. For the differentiation of metastatic lesions, such as in nodal metastasis versus nodal nevi (although based only on 2 reports), there appears to be some evidence supporting its diagnostic utility. However, for the purposes of distinguishing primary cutaneous melanoma from benign lesions, there is currently a lack of substantial evidence to support its use, especially when it is used alone. When it is used in a panel of other melanocytic markers, however, the possibility for diagnostic utility is likely increased.⁸  Perhaps other potential markers similar to p16 might provide better use for distinguishing benign from malignant melanocytic lesions in the future, such as with the recent report⁷²  of p15 immunohistochemistry in melanocytic lesions. Otherwise, currently, because of the variability of results encountered by analyzing multiple studies, p16 appears to have many limitations, especially for differentiating benign from malignant primary lesions. Nevertheless, if interpretation methods and techniques for its use were better defined, perhaps its role for melanocytic lesions might become more acceptable. Because studies considering only nuclear immunohistochemical p16 staining as positive appeared to show more consistent results, limiting the interpretation to nuclear p16 staining may prove to be more accurate, and may improve interobserver variability, thus potentially making it more useful in the routine distinction of benign versus malignant melanocytic lesions.