Abstract
Background/Aim: Merkel cell carcinoma (MCC) is a rare, aggressive, neuroendocrine skin cancer and most MCCs are related to infection with Merkel cell polyomavirus (MCPyV). Notch signaling modulates cell fate in various tissues including the skin during development and homeostasis, and its aberrant activity relates to onset and progression of various malignancies. Therefore, association of NOTCH1/ NOTCH2/NOTCH3/jagged 1 (JAG1) expression with MCPyV status and prognosis in MCC was investigated. Materials and Methods: A total of 19 MCPyV-positive and 19 MCPyV-negative MCC samples from patients were stained immunohistochemically with antibodies against NOTCH1, NOTCH2, NOTCH3, and JAG1 and analyzed. Results: Expression of NOTCH1 and NOTCH2 was not associated with MCPyV status or prognosis. However, higher JAG1 expression was found in MCPyV-negative than in MCPyV-positive MCC (p<0.001), and NOTCH3 expression was higher in MCPyV-positive MCC (p=0.062). Kaplan–Meier and multivariate analyses showed that patients with MCC with higher NOTCH3 expression had better overall survival than otherwise (p=0.001 and p=0.033, respectively). Conclusion: Expression of NOTCH3, as a tumor suppressor, is an independent predictor of MCC outcome.
Merkel cell carcinoma (MCC) is an aggressive yet rare cutaneous cancer with neuroendocrine features associated with risks factors including advanced age, immunosuppression, and chronic sun exposure (1). The incidence rate of MCC in the United States has continued to rise, by approximately fivefold over the past 30 years: from 0.22 per 100,000 in 1986 to 0.79 per 100,000 in 2011 (2).
In 2008, it was found that about 80% of MCC cases were associated with Merkel cell polyomavirus (MCPyV), in which MCPyV had been integrated into the tumor genome (3). In Northern areas, the majority of cases of MCC are caused by MCPyV, whereas in areas with higher ultraviolet (UV) exposure, carcinogenesis is driven predominantly by UV. On the other hand, UV exposure can also induce local immunosuppressive condition in viral tumorigenesis (4).
The existence of MCPyV infection is related to histological distinction in MCC: tumor cells in MCPyV-positive MCC have homogeneous round nuclei and less cytoplasm, whereas MCPyV-negative MCC tumor cells have pleomorphic nuclei and a plentiful cytoplasm (5). As well as histological differences, MCPyV integration into tumor cells renders different survival prognoses. MCPyV-negative MCC usually leads to shorter survival and worse prognosis compared with MCPyV-positive MCC (6-10).
The Notch signaling pathway regulates both embryonic and adult tissues, and affects the establishment, growth, and regenerative potential of multiple tissues by determining cell fate (11). Its roles in regulation include maintenance and differentiation of stem cells, cell fate determination and cell-cycle regulation, and other functions are being unveiled as investigations continue; it is clear that the role of notch signaling is highly complex and multifaceted (12). Notch signaling is important in regulating cellular behavior, therefore, unsurprisingly, Notch plays an important role in many types of cancers, particularly as it plays such a prominent part in regulation of stem and progenitor cells. Its aberrant activity relates to initiation and progression of various malignancies, and it can play a role either as oncogene or tumor suppressor depending on the tissue and cellular type (13, 14).
There are four receptors (NOTCH1-4) in mammals, and five notch ligands are delta-serrate-lag (DSL) family [jagged 1 (JAG1), JAG2, delta-like 1 (DLL1), DLL3, and DLL4)]. Structurally, Notch receptors have a single-pass transmembrane heterodimers consisting of an extracellular ligand-binding domain and an intracellular domain comprising a transmembrane region and an intracellular portion that mediates signaling upon receptor ligation. The same as Notch receptors, Notch ligands structurally are also transmembrane proteins containing epidermal growth factor (EGF)-like repeats (13, 15). Interaction between the extracellular portion of Notch receptor and the Notch ligand results in a series of events converting the transmembrane form of Notch into a nuclear transcriptional co-activator and activates transcription of target genes, including hairy/enhancer of split (HES) and HES-related with YRPW motif (HEY) family genes (16).
Studies in mice and human tissue indicate that Notch receptors and ligands are distributed in spatially restricted expression patterns throughout the epidermis and its appendages. In the interfollicular epidermis, NOTCH1-4 are expressed at the highest levels in the suprabasal cells of the spinous and granular layers, where cells are undergoing differentiation, whereas Notch ligands JAG1 and JAG2 are expressed in the interfollicular epidermis, predominantly in the suprabasal layers, with limited expression in basal cells (12).
Several studies have been conducted exploring carcinogenesis in MCPyV-positive and MCPyV-negative MCC. Harms et al. performed transcriptome analysis of 30 MCC cases and found that MCPyV-negative tumors displayed a relative up-regulation of mRNA of gene groups associated with Notch signaling (17). Another study performed whole-exome sequencing of 16 MCC cases, nine MCPyV-negative and seven MCPyV-positive, and identified previous known mutations in TP53, retinoblastoma 1 (RB1) and phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) along with novel activating mutations in oncogenes such as HRAS, loss-of-function mutations in prune homolog 2 with BCH domain (PRUNE2) and NOTCH (NOTCH1-4) family genes in MCPyV-negative cases (18). A larger scale exome sequencing study of 49 MCCs confirmed the previous report that MCPyV-negative MCCs have a higher mutation burden, frequent mutations in TP53 and RB1 and additional mutations in genes involved in chromatin modification and DNA damage pathways, and interestingly, both MCPyV-positive and -negative tumors were found to have mutations inactivating the Notch signaling pathway (NOTCH1, NOTCH2) (19).
Panelos et al. performed the first immunohistochemical studies of NOTCH1 expression in MCC regardless of MCPyV status and found 30/31 cases had cytoplasmic and membranous NOTCH1 expression in more than 50% of cells (20). Wong et al. reported that MCC with mutant NOTCH1 showed low or absent nuclear NOTCH1 expression (21). As there are no other studies evaluating Notch signaling expression in MCC using immunohistochemistry, in this study, we evaluated the nuclear expression of activated NOTCH1 and NOTCH2 and membranous/cytoplasmic expression of NOTCH3 and JAG1 in tumor cells of MCPyV-positive and -negative MCC, and examined the association of expression of these markers with clinicopathological factors and the prognosis of MCC.
Materials and Methods
In this study, 38 formalin-fixed paraffin-embedded MCC samples were prepared. These included 19 MCPyV-positive MCCs (15 samples from the United Kingdom and four samples from Japan) and 19 MCPyV-negative MCCs (14 samples from the United Kingdom and five from Japan). The MCPyV-negative MCC samples included 13 of MCC combined with squamous cell carcinoma (SCC) or Bowen disease. A summary of clinicopathological data is listed in Table I. This study was approved by the Institutional Review Board of Medical Faculty, Tottori University, Japan.
Immunohistochemistry. Formalin-fixed paraffin-embedded samples were sectioned into 4-μm-thick pieces, followed by deparaffinization and rehydration. Antigen retrieval was performed by incubating the sections for 40 min at 100°C in Nichirei Heat Pro II (Tokyo, Japan). After blocking endogenous peroxidase activity for 5 minutes, sections were incubated for 60 min with primary antibody, and then incubated with the secondary antibody for 30 min. Sections were then incubated with diaminobenzidine for 10 min; all these processes used a Nichirei Histo Stainer (Tokyo, Japan). After washing the sections using phosphate-buffered saline, they were counterstained with hematoxylin for 5 sec, and then rehydrated and mounted. The primary antibodies (against NOTCH1, NOTCH2, NOTCH3, and JAG1) used in this study and the tissues used as positive controls are listed in Table II.
Immunohistochemical evaluation and scoring system. The stained tissue slides were evaluated by pathologists and researchers who were blinded to the patients' clinical data. The nuclear expression of NOTCH1 or, NOTCH2, and membranous/cytoplasmic expression of NOTCH3 and JAG1 in the stained tumor cells of MCC and combined tumor were evaluated using the modified H-score. The percentage of cells stained was summed and multiplied by values according to the staining intensity level (0=not stained, 1=weakly stained, 2=moderately stained, and 3=strongly stained), and the H-score ranged from 0 to a maximum of 300 (22). The internal negative controls used were non-neoplastic skin and subcutaneous tissues from MCC samples. A summary of H-score data of NOTCH1, NOTCH2, NOTCH3 and JAG1 is listed in Table III.
Statistical analysis. All clinicopathological parameters, such as age, sex, race, and immunohistochemical results, were partitioned based on the MCPyV status and analyzed using the Mann–Whitney U-test. Survival was measured from diagnosis of MCC. The survival analysis was evaluated using the Kaplan–Meier method in accordance with NOTCH1/NOTCH2/NOTCH3/JAG1 expression using the mean H-score to divide patients into groups with low and high expression. The significant differences were examined by the log-rank test. The Cox proportional hazards regression model was applied to perform univariate and multivariate analyses, and those variables that achieved statistical significance in the univariate analysis were included in the multivariable analysis. All data were analyzed statistically using SPSS software (version 21.0J; SPSS Japan Inc., Tokyo, Japan), and a p-value of less than 0.05 was considered statistically significant.
Results
Characteristics of clinical background. The 38 cases of MCC used in this study, comprising 19 MCPyV-positive MCCs and 19 MCPyV-negative MCCs, are described in Tables I and IV. There was a significant difference in the ages of both MCC groups: MCPyV-negative cases (mean=85 years) were older than MCPyV-positive cases (mean=75.06 years), as shown in Table IV (p=0.002). The other clinical parameters, such as sex, race, and staging, were not different between the two groups.
Histological findings in NOTCH1, NOTCH2, NOTCH3 and JAG1 expressions. Immunohistochemical staining was performed to evaluate the expression of NOTCH1, NOTCH2, NOTCH3, and JAG1 in MCC tumor cells and the results are summarized in Tables III, IV and V. Representative immunostaining features of NOTCH1, NOTCH2, NOTCH3 and JAG1 in MCPyV-positive and MCPyV-negative MCC cases are shown in Figure 1. The H-scores for NOTCH1 and NOTCH2 were similar between MCPyV-positive and MCPyV-negative MCCs (Figure 1E and F, Table IV; p=0.51, and Figure 1G and H, Table IV; p=0.173, respectively). NOTCH3 H-score was higher in MCPyV-positive than MCPyV-negative MCC (Figure 1I and J), even though not statistically significant (p=0.062). The H-score for JAG1 was significantly higher in MCPyV-negative (Figure 1L) than in MCPyV-positive MCC (Figure 1K) (Table IV; p<0.001). All MCPyV-positive MCCs, except one focally positive case, showed no JAG1 expression.
In 13 cases of combined MCC and SCC, tumor cells in the MCC component had a significantly higher H-score for NOTCH2 and NOTCH3 expression than did those in the SCC component (Table V; p<0.001, and p=0.021, respectively). There were trends that NOTCH1 and JAG1 expression of tumor cells in the MCC component were higher than those in the SCC component in combined MCC and SCC cases but not statistically significant (Table V; p=0.071 and p=0.877, respectively).
Prognostic analysis of NOTCH1, NOTCH2, NOTCH3 and JAG1 in MCC. The Kaplan–Meier method with log-rank test was used to analyze prognostic survival in this study, and the results are summarized in Table VI. High NOTCH3 expression in MCC tumor cells (H-score ≥199) significantly corresponded to more favorable overall survival (OS) (Figure 2A; p=0.001), but not disease-specific survival (DSS) (Figure 2B; p=0.371).
Clinicopathological parameters and NOTCH1, NOTCH2, NOTCH3, and JAG1 expressions in tumor cells for the prediction of OS and DSS were further investigated by univariate and multivariate analyses with the Cox proportional hazards regression model. The results are shown in Table VII. Results from the univariate analysis indicated that MCPyV negativity [hazard ratio (HR)=3.559; p=0.012] and advanced age (HR=8.918; p=0.034) were unfavorable factors for OS. Female patients had shorter survival than male patients even though not statistically significant [OS (HR=2.3; p=0.118) and DSS (HR=3.735; p=0.226)]. Caucasian patients had better survival than Japanese patients but not significantly [OS (HR=0.413; p=0.105) and DSS (HR=0.325; p=0.198)]. Patients with stage III/IV disease had longer survival than those with stage I/II, although not statistically significant [OS (HR=0.803; p=0.768) and DSS (HR=0.041; p=0.541)]. High NOTCH3 expression (H-score ≥199) significantly lengthened OS (HR=0.213; p=0.003). Only radical excision (HR=0.154; p=0.017) significantly extended DSS. With multivariate analysis, only high expression of NOTCH3 was a significantly favorable prognostic factor for OS (HR=0.313; p=0.033), whereas only radical excision was a significant favorable prognostic factor for DSS (HR=0.107; p=0.019).
Discussion
MCC can be caused either by MCPyV infection, as in MCPyV-positive MCC, or by UV exposure-driven carcinogenesis, as in MCPyV-negative MCC, even though UV exposure can also induce local immunosuppression in viral tumorigenesis. Evaluation of the difference between MCPyV-positive and MCPyV-negative MCC showed that MCPyV-negative MCC has a high frequency of DNA mutations associated with UV damage, disruption of RB1 and TP53, presence of a high degree of aneuploidy, and mutations in genes related to responses to DNA impairment and repair, whereas MCPyV-positive MCC generally has few somatic mutations and little evidence of UV damage, and most MCPyV-positive cases have intact RB1 and wild-type TP53 (4, 19, 21).
Investigation using whole-exome sequencing of 16 MCC cases identified loss-of-function mutations in one or more NOTCH family genes (NOTCH1-4) in six out of eight (75%) MCPyV-negative cases, and NOTCH mutations in MCC were mainly located in EGF or ankyrin repeat regions, which is consistent with a loss-of-function event, and the authors suggested that Notch signaling plays a tumor-suppressive role in MCC similarly to other neuroendocrine malignancies (18).
Wong et al. evaluated targeted capture and massively parallel DNA sequencing of 619 cancer genes to compare the gene mutations and copy number alterations in MCPyV-positive (n=13) and MCPyV-negative (n=21) MCC tumors and cell lines. All MCPyV-negative tumors harbored a high frequency of mutations in NOTCH1, and immunohistochemistry with NOTCH1 antibody revealed that MCC with mutated NOTCH1 showed a marked reduction of NOTCH1 expression (21).
Another larger scale exome sequencing of 49 MCCs by Goh et al. confirmed the previous report and found that both MCPyV-positive and MCPyV-negative MCCs had mutations inactivating the Notch signaling pathway (NOTCH1 and NOTCH2). MCPyV-negative MCC had sporadic somatic single nucleotide variants affecting NOTCH1 and NOTCH2 that were also seen in small-cell lung cancer and they suggested that dysregulation of these genes and pathways were required for the neuroendocrine differentiation of epithelial cells, a common feature of both MCCs and small-cell lung cancer (19).
Whereas the previous studies by Harms et al. (18) and Goh et al. (19) showed that NOTCH genes had loss-of-function events and they suggested that Notch signaling played a tumor-suppressive role in MCC similarly to other neuroendocrine malignancies, we found that there was no difference in nuclear expression of NOTCH1 nor NOTCH2 in both MCPyV-positive and MCPyV-negative MCC (p=0.51 and p=0.173, respectively). Moreover, our finding of high NOTCH1 expression observed in most MCC cases (mean H-score of all cases: 256.13 in Table III) is similar to the result of Panelos et al., as they showed cytoplasmic and membranous NOTCH1 expression in more than 50% of cells of in 30/31 MCC cases regardless of MCPyV status (20). However, we investigated the nuclear expression of NOTCH1 (activated NOTCH1 expression) rather than cytoplasmic/membrane expression reported by Panelos et al. because nuclear NOTCH1 expression reflects the translocation of the activated form of NOTCH1 from the cytoplasm to the nucleus. Our data is different from that for other neuroendocrine tumors such as gut carcinoids, medullary thyroid carcinoma (MTC), and pulmonary typical and atypical carcinoids which show minimal or absence of NOTCH1 signaling (23), and from previous data in MCCs which suggest NOTCH1 is a tumor suppressor in MCCs (18, 19, 21).
What is a reasonable explanation for the discrepancy between our finding of high nuclear NOTCH1 expression in MCC suggesting NOTCH1 to be an oncogene and the suggestion that it is a suppressor by the previous genome-wide studies showing inactivating NOTCH1 and NOTCH2 mutation with somatic single nucleotide variants? It is well known that P53 is often overexpressed or occasionally not expressed immunohistochemically in the nuclei of cancer cells due to mutation of TP53 as a suppressor gene. Overexpression of NOTCH1 and NOTCH2 in our study may reflect the accumulation of mutation-induced abnormal NOTCH1 or NOTCH2 protein in the nuclei of Merkel cell tumor cells, like P53. The antibody used for immunohistochemistry in our study (Abcam, ab8925) was developed for detecting the activated form of NOTCH1 and different from those of previous reports (20, 21). This may be an important reason for our different results of immunohistochemistry for NOTCH1.
In contrast to NOTCH1 and NOTCH2, NOTCH3 expression in MCPyV-negative cases (mean H-score=186.05) was found to be lower than in MCPyV-positive ones (mean H-score=211.32) (p=0.062). In addition, high NOTCH3 expression of tumor cells (H-score ≥199) was associated with more favorable OS. This suggests that NOTCH3 has activity as a tumor suppressor. NOTCH3 dysregulation has been associated with a wide variety of malignancies as it has been shown to affect tumor aggressiveness, maintenance and resistance to chemotherapy, it has roles both as tumor suppressor and oncogene (24). In this study, NOTCH3 appeared to function as a tumor suppressor, and this same function was shown in MTC and ovarian cancer as NOTCH3 induced apoptosis and had an antiproliferative function (24, 25), whereas in breast cancer, NOTCH3 increased chemosensitivity of doxorubicin-resistant breast cancer (24).
Harms et al. used transcriptome analysis of 30 MCC cases and found that MCPyV-negative MCC had relative up-regulation of DLL1, C-terminal binding protein 2 (CTBP2), hairy/enhancer of split 1 (HES1), JAG2 and JAG1 mRNA compared with MCPyV-positive MCC (17). In accordance with their result for up-regulation of JAG1, we showed that JAG1 expression was significantly higher in membranous/cytoplasmic of tumor cells in MCPyV-negative than in MCPyV-positive MCCs (p<0.001), 16 of 19 MCPyV-negative cases expressed JAG1 and no MCPyV-positive case had JAG1 expression, except in one case. In survival analysis, there was no difference in survival according to JAG1 H-score (cut-off ≥17). Notch ligand, JAG1, is overexpressed in many cancer types, and plays an important role in several aspects of tumor biology. JAG1-stimulated Notch activation is directly implicated in tumor growth through maintaining cancer stem cell populations, promoting cell survival, inhibiting apoptosis, and driving cell proliferation and metastasis. In addition, JAG1 can indirectly affect cancer by influencing tumor microenvironment components such as tumor vasculature and immune cell infiltration (26).
In this study, we reconfirmed the findings of our previous studies that patients with MCPyV-positive MCC have a favorable survival (8, 10, 22, 27), elderly patients have worse survival (10, 22), and radical excision significantly extend OS and DSS (10). MCC can occur in both female and male patients (1), although it is almost twice as frequent in males (62.1%) than in females (37.9%) (28). In this study, female patients were the majority (71.1%). This study revealed that female patients had shorter survival (in OS and DSS) compared to male patients, although this was not statistically significant. A previous study supported our result that male patients had significantly improved OS, but not DSS by Kaplan–Meier analysis (27). On the contrary, another study showed that male patients had non-significantly shorter OS and DSS compared to female patients (22), and a large cohort study (n=351) revealed that male sex was an independent prognostic factor of unfavorable outcome in OS and DSS (29).
MCC mainly affects Caucasians (96.4%) and is very rare among individuals of African American (1.2%), and Asian and Pacific Islander (0.8%) descent (28). In this study, Caucasian patients with MCC were the majority (76.3%), whereas only 23.7% were Japanese. Kuromi et al. reported that Japanese ethnicity was linked to significantly longer OS but not linked with DSS by Kaplan–Meier analysis (27). In contrast, this study showed that Caucasians had longer OS and DSS than did Japanese, but not statistically significant. Another study comparable with our result showed that Caucasian patients had better survival than those of Japanese ethnicity, which was also not statistically significant (10).
Harms et al. showed that the extent of MCC was predictive of 5-year OS, with estimated OS of 51, 35, and 14% for those with local, nodal and distant disease, respectively (28). In this study, the majority of patients had stage I/II disease (33/38, 86.8%), and advanced stage was not a predictor of worse OS or DSS. Our previous study showed that patients with stage III/IV disease had shorter OS and DSS compared with those with stage I/II disease, even though not statistically significant (10). However, unexpected result of longer OS or DSS in stage III/IV in this study can be attributed to the small number of total samples (n=38) and small number of stage III/IV samples (n=5, only 13.2%) compared to stage I/II samples (n=33). On the other hand, patient age can influence survival even when diagnosed with early-stage MCC.
This study has a limitation given that we used relatively few samples (38 samples). In order to get reliable results statistically, for example in evaluating the association of clinicopathological variables and Notch signaling markers with patients' survival, in the future, our findings should be confirmed in a larger sample population cohort.
NOTCH3 has roles as an oncogene and a tumor-suppressor gene. Therapy targeting NOTCH3 as an oncogene aims to inhibit the level of NOTCH3 expression, with use of agents such as gamma secretase inhibitors, which are not specific to Notch subtypes; furin-like convertase and a disintegrin and metalloprotease (ADAM) inhibitors for inhibition notch enzymatic processing; monoclonal antibodies to block NOTCH3 to avoid pan-Notch blockage; and microRNA targeting specific Notch subtypes. In contrast, in therapy targeting NOTCH3 as a tumor suppressor, it would be beneficial to induce expression of NOTCH3 rather than to inhibit it (24). Several histone deacetylase inhibitors (HDACi) have been shown to increase Notch expression in a variety of malignancies in which Notch plays a tumor-suppressive role (24). Jaskula-Sztul et al. investigated the role of NOTCH3 signaling as a tumor suppressor in MTC using doxycycline-inducible NOTCH3 intracellular domain (NICD3, the post-γ-secretase cleavage product of NOTCH3) and HDACi AB3, and showed that induction of NOTCH3 signaling can inhibit tumor proliferation and suppress neuroendocrine markers (30). Comparable to that of Jaskula-Sztul et al., another study revealed that NOTCH3 expression declined as thyroid cancers became less differentiated and more malignant, resembling its conserved pattern in development, therefore predicting disease prognosis, and NOTCH3 activation in a gain-of-function follicular thyroid carcinoma cell line inhibited cell proliferation and migration, activated the intrinsic apoptotic cascade, and reduced tumor burden in vivo (25).
Several clinical trials of immune checkpoint inhibition using an antibody to programmed cell death protein 1 (PD1) or PD-ligand 1 in patients with advanced-stage MCC show higher and more durable response rates than conventional chemotherapy. However, not all patients have durable responses to immune checkpoint inhibitors (31-33). Other immunotherapies for MCC that act through mechanisms other than inhibition of PD1 or PD-ligand 1 including therapeutic combinations of anti-cytotoxic T-lymphocyte antigen 4 are still under investigation (34). Combination therapy of immune-checkpoint inhibitors and HDACi for increasing NOTCH3 expression as a tumor suppressor might be a candidate for novel therapy for MCC.
In conclusion, we showed that MCPyV-negative MCC is significantly associated with higher JAG1 expression in tumor cells than MCPyV-positive MCC. NOTCH3 expression is associated with a significantly longer OS by using Kaplan–Meier analysis, and high NOTCH3 expression is an independent predictor of favorable OS in multivariate analysis. This study suggests that NOTCH3 and JAG1 may play a role in MCC tumorigenesis.
Acknowledgements
This study was supported, in part, by grants from the Japan Society for the Promotion of Science [no. 26460433 (K.H.) and no. 26860238 (M.M.)], and the Authors thank the laboratory staff at the Division of Molecular Pathology, Department of Pathology, Tottori University, Faculty of Medicine, particularly Mr. H. Sugihara for the skillful preparation of tissue slides.
Footnotes
Competing Interest
There is no conflict of interests.
Funding/support
This study was supported, in part, by grants from Japan Society for the Promotion of Science [no. 26460433 (K.H.) and no. 26860238 (M.M.)].
- Received November 18, 2018.
- Revision received December 3, 2018.
- Accepted December 12, 2018.
- Copyright© 2019, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved