Morphological and immunohistochemical profile of pancreatic neuroendocrine neoplasms

Abstract

The study represents a comprehensive retrospective morphological and immunohistochemical profiling of pancreatic neuroendocrine neoplasms (PNENs) in order to reveal the associations between morphological and molecular parameters. The local tumour spread (T), presence of metastases in regional lymph nodes (N) and distant organs (M), tumour grade (G) and resection line status (R) by pathology findings (pTNMGR), mitotic activity, perineural, vascular and lymphatic invasion were assessed in 16 surgically resected PNENs. By immunohistochemistry, expression of Ki-67, p53, p27, p21, cyclin D1, Bcl-2, E-cadherin, CD44, vimentin, cyclooxygenase 2 (COX-2), microvascular density, and cytokeratin (CK) spectrum, along with neuroendocrine, intestinal and squamous markers were detected. Descriptive statistics, Chi-square test, Spearman’s rank correlation, Mann–Whitney and Kruskal–Wallis methods were applied; p < 0.05 was considered significant. Ki-67, CK19, p63, vimentin and COX-2 were significantly up-regulated in PNENs in comparison to benign pancreatic islets. A complex network of morphological and molecular associations was identified. Ki-67 correlated with PNEN size (p = 0.022), the World Health Organization 2004 and 2010 classification grades (p = 0.021 and p = 0.002), stage (p = 0.028) and mitotic count (p = 0.007) but among molecular markers – with CK19 (p = 0.033) and vimentin (p = 0.045). CK19 was significantly up-regulated in PNENs, having higher pT (p = 0.018), pR (p = 0.025), vascular (p = 0.020), perineural (p = 0.026) and lymphatic invasion (p = 0.043). In conclusion, proliferation activity (by Ki-67), E-cadherin, vimentin and CK19 are important molecular characteristics of PNENs due to significant associations with morphological tumour characteristics, pTNMGR and invasive growth.

Introduction

Pancreatic neuroendocrine neoplasms (PNENs) represent a rare entity that nevertheless has a high clinical importance due to the growing incidence and distinct biological properties of neuroendocrine tumours [1, 2]. Although PNENs have significantly better prognosis than does the more common pancreatic ductal adenocarcinoma, novel therapeutic strategies are still needed [3], therefore the cornerstones of PNEN carcinogenesis must be evaluated. Here, we present an integrated molecular and morphological study of PNENs. The proteins that hypothetically can be involved in PNEN development include proliferation markers, notably Ki-67 [4], tumour suppressor p53 [5], cell cycle regulators p21, p27 and cyclin D1 [6-8], anti-apoptotic protein Bcl-2 [5], cell adhesion molecule E-cadherin [9] and cancer stem cell marker CD44 [10, 11]. The cytokeratin (CK) spectrum, in addition to intestinal and squamous markers, is widely used in the assessment of tumour histogenesis; however, cytokeratin-associated cell plasticity or differentiation can confer a pathogenetic role that should be evaluated as well [2, 12, 13]. Epithelial-mesenchymal transition [14], cyclooxygenase 2 (COX-2) induction [3], and angiogenesis [15] also represent reasonable research targets. However, the present data are still controversial. In addition, most of the studies have been devoted to few isolated proteins. We hypothesised that the molecular profile is associated with certain morphological parameters and these correlations can identify target proteins that are significantly involved in the development of PNENs. Thus, the aim of this research is to detect the morphological and immunohistochemical profiles of pancreatic neuroendocrine neoplasms in order to reveal the associations between these parameters.

Material and methods

The study quintessence, ethical principles and case selection

A retrospective morphological and immunohistochemical investigation of PNENs was carried out in accordance with the Declaration of Helsinki and was approved by the Committee of Ethics of Riga Stradins University (E-9(2), issued on October 06, 2011). Consecutive cases were identified using an archive search in a single university hospital (2004-2014). The inclusion criteria comprised a verified unequivocal morphological diagnosis of PNEN in the tissue material submitted after the potentially curative operation. Patients that had a tumour of different histogenesis or of equivocal origin, as well as those that underwent biopsy only or received preoperative chemotherapy or radiotherapy were excluded from the study.

Surgical pathology evaluation

The pathology data were obtained via uniform, protocol-based gross and microscopic examinations of the pancreatic surgery materials. Grossly, tumour localisation and largest diameter were detected, among other findings. Consecutive sections of the whole pancreas were obtained at a distance of 5 mm. The lymph nodes (LN) were entirely submitted for microscopic investigation by anatomic compartments. After specimen inking, the resection margins (RM) were completely submitted for microscopic investigation in accordance with the Leeds Pathology Protocol [16, 17]. In short, proximal and distal transection margins of the stomach and duodenum, respectively, as well as the circumferential (anterior, posterior and superior mesenteric vein groove) RM of the pancreas and transection margins of pancreatic neck and distal bile duct were investigated. The samples were routinely fixed in neutral buffered 10% formalin (Sigma-Aldrich, Saint Louis, USA), processed and stained with haematoxylin and eosin. The slides were examined under light microscopy (Leica DM500, Wetzlar, Germany) to detect histological tumour type and grade [1, 18], the local tumour spread (T), presence of metastases in regional lymph nodes (N) and distant organs (M), summarised into pTNM [19], status of RM [16, 17], mitotic count and vascular, and peri- and intra-neural, as well as lymphatic vessel invasion. To link the study to both the previous and forthcoming publications, the grade was assessed using both the World Health Organization (WHO) clinico-pathological classification of tumour differentiation and behaviour, issued in 2004 (further designated as grade 2004) and the WHO grading, issued in 2010 (further designated as grade 2010), as described in the literature [1, 18]. To detect the mitotic activity, mitoses were counted in 40 consecutive high-power fields (HPFs) within the mitotically most active areas, applying 400× magnification (0.20 mm² per field corresponding to 2 mm² per 10 HPFs) and recalculating to 10 HPFs/2 mm² [1]. The peri- and intra-neural growth, as well as vascular and lymphatic invasion, was evaluated as categorical variables.

Immunohistochemical visualisation and assessment

Immunohistochemical visualisation (IHC) was performed on representative blocks of tumour and non-neoplastic pancreatic tissues. For IHC, 3-micrometre-thick sections were cut using an electronic rotary microtome Microm HM 360 on electrostatic glass slides (Histobond, Marienfeld, Germany). After deparaffinisation, antigen retrieval was performed in a microwave oven (3 × 5 min) using a basic TEG (pH 9.0) buffer, followed by blocking of endogenous peroxidase (Sigma-Aldrich). The sections were incubated with primary antibodies (Table I) at room temperature in the magnetic incubation tray. Bound antibodies were detected by the enzyme-conjugated polymeric visualisation system EnVision, linked with horseradish peroxidase using 3,3’-diaminobenzidine as the chromogen. All IHC reagents were produced by DAKO, Glostrup, Denmark. Positive and negative quality controls were invariably performed and reacted appropriately. The IHC reactivity was classified into three intensity levels: low, moderate and high. Only moderate or high intensity expression was considered positive, and the relative amount of positive cells was detected as the percentage within 400 sequential tumour or islet cells. If less than 400 cells were available in the tissue section, the case was excluded from evaluation of the particular antigen expression. For a case to be considered positive, the relative amount of positive cells had to reach the cut-off value [20-28], as shown in Table I. To detect the microvascular density (MVD), the endothelial layer was highlighted by membranous CD34 expression. Four hot spots of the target tissue were identified by whole slide scanning at low power (100×) and assessed on HPF. The MVD was calculated as the mean amount of microvessels per HPF in a hot spot, counting only vessels with a clearly defined lumen or well-defined linear shape.

Statistical analysis

The statistical analysis was performed using the IBM SPSS Statistics Version 20.0 statistical software package (International Business Machines Corp., Armonk, New York, United States of America). The confidence interval was calculated using the Confidence Interval Analysis software [29]. The assumption check of normality was performed using a Shapiro-Wilk test. The descriptive data were expressed as mean ± standard deviation (SD), median with interquartile range (IQR) or relative frequency with 95% confidence interval (CI). Descriptive statistical methods, such as descriptive and cross tabulation with Pearson’s 2, bivariate correlation as Spearman’s rank correlation coefficient, and non-parametric methods, such as Mann–Whitney U test and Kruskal–Wallis one-way analysis of variance by ranks, were used. The post-hoc analysis with Bonferroni correction was applied to determine differences between three or more groups. The p-values of < 0.05 were considered statistically significant [30].

Results

Clinical and morphological characteristics

During the study period, surgical treatment for PNENs was performed in 16 patients. The mean ± SD age was 59.4 ±9.2 years (range, 47-78) and the median value was 56.5 years (IQR = 16). The study group was characterised by a female predominance (Table II). The clinical data, as well as the morphological tumour profile, including the local tumour spread (T), presence of metastases in regional lymph nodes (N) and distant organs (M), tumour grade (G) and resection line status (R) by pathology findings (pTNMGR), stage and the frequency of such manifestations of invasive growth as vascular, perineural and lymphatic invasion, are represented in Table II. The mean tumour size ± SD was 3.9 ±3.2 cm (range, 1.2-12.5) and the median value was 2.9 cm (IQR = 4.3). The mean LN count ±SD was 6.2 ±10.0 (range, 1-33) and the median was 1.5 (IQR = 6) LN. Mitoses were found in 62.5% (95% CI: 38.6–81.5) of the cases, ranging in number from 1 to 13 per 10 HPFs. The mean mitotic count ± SD reached 2.5 ±3.6 (range, 0-13), and the median value was 1.0 (IQR = 3.8) mitoses per 10 HPFs.

Immunohistochemical profile

The expression of immunohistochemical markers was assessed in the PNENs and non-neoplastic islets of the pancreas (Table III). Significant up-regulation of Ki-67, CK19, p63, chromogranin A, vimentin and COX-2 expression was disclosed in the PNENs. The tumours invariably showed proliferative activity, in contrast to the pancreatic islets. Expression of aberrant p53 protein was uncommon. All p53 positive cases were well-differentiated PNENs with unclear behaviour (by WHO 2004 classification), pT1-2, measuring 1.2-3.0 cm. Using the WHO grade 2010, both G1 and G2 tumours were found in the p53 positive group. The levels of p21 and p27 proteins, as well as cyclin D1, did not differ among normal and neoplastic neuroendocrine tissues. Both target tissues lacked any expression of anti-apoptotic protein Bcl-2, intestinal markers CK20 and CDX2, and two squamous markers, CK5/6 and CK34E12. Epithelial-mesenchymal transition (EMT) by vimentin expression was evident in a significant fraction of PNENs, contrasting with constantly negative islets. COX-2 positive cells also were found only in tumour tissue in low mean number (Table III). In addition, PNENs were characterised by marked IHC heterogeneity (Table IV).

The associations between the studied variables

The statistically significant associations between morphological and molecular variables are shown in Table V. In addition, CK19 was significantly up-regulated in PNENs having higher pT (p = 0.018) with significant differences between pT1 and pT3 (p = 0.022), as well as between pT2 and pT3 (p = 0.004). No associations with clinical, morphological or IHC findings were found regarding p27 expression.

Discussion

Pancreatic neuroendocrine neoplasms are infrequent tumours constituting 1.3-5% of pancreatic malignancies [31, 32]. However, they have recently attracted significant attention due to growing incidence, in parallel with other neuroendocrine neoplasms (NENs). As reported by the Surveillance, Epidemiology and End Results (SEER) registry of the National Cancer Institute, the incidence of gastroenteropancreatic NENs has increased from 1 (1973-1977) to 3.65 (2003-2007) per 100,000. In the same time period, the incidence of PNENs increased from 0.17 to 0.43 per 100,000 [33]. PNENs are associated with significantly better outcomes than is pancreatic ductal adenocarcinoma (PDAC). The median survival of surgically treated non-functional PNENs was 60.4 months and the mean survival of PNEN patients reached 225.3 months [34, 35]. In contrast, the median survival of surgically treated PDAC was 22.8 months [36]. The five-year survival of surgically treated PNEN patients is 84.4%, while of PDAC patients, it is 19-24% [37, 38]. However, the clinical course of PNENs is still difficult to predict [39]. Although knowledge about PNEN pathogenesis and biological potential deepened during the preceding years, the molecular mechanisms responsible for tumour progression and metastasis still remain incompletely understood. Molecular markers are increasingly used to predict patient outcome, and immunohistochemistry has been found to be an appropriate, visually controllable, economically effective and fast surrogate method for evaluating the molecular aberrations on the protein level [40]. Recently, the World Health Organization included the Ki-67 level in the PNEN grading criteria, resulting in improved accuracy of the grading system [4]. Deeper understanding of the molecular pathogenesis of PNENs could help to sharpen the grading system.

Clinical data

The clinical profile was obtained to characterise the general conformity of the study group to the known PNENs characteristics. The observed mean age of PNEN patients, 59.4 years, was slightly higher than in other studies (48-53 years), possibly due to having an adult group, in contrast with the presence of adolescent patients in other studies that showed an age range of 14-78 years [34, 41, 42]. Although in this study, women accounted for 75.0% of patients, PNEN incidence is considered equal in both sexes. The observed difference is probably attributable to the relatively small size of the study group [34, 41]. Regarding the tumour location, in this and other similar studies, PNENs were smoothly distributed within the anatomic compartments of the pancreas [41]. The surgical treatment in the present study generally corresponded to the accepted standards [41, 43, 44], providing a reasonable cure to the patient while also yielding representative tissue material.

Tumour size

Within the study, PNEN size ranged from 1.2-12.5 cm, in accordance with the published results [41, 45-47]. Prognostic value has been attributed to the size of PNEN. The five-year survival rate was statistically significantly better (100%) in patients having tumour ≤ 1.5 cm, in contrast to 86%, 71%, 83% and 48% survival rates regarding PNENs sized 1.6-2.0 cm, 2.1-3.0 cm, 3.1-5.0 cm and larger than 5.0 cm. Consequently, observation has even been suggested as the approach for patients having a tumour smaller than 1.5 cm. However, risk of distant metastases increases sharply with a larger PNEN size [48]. This is in agreement with the present study showing an association between tumour size, pM1 (p = 0.025), and invasion in the blood vessels (p = 0.045), the precursor lesion of distant metastases. In addition, larger PNEN size has been found to correlate with LN metastases [48]. Considering the molecular basis of PNEN growth, the tumour size in the present study showed an association with Ki-67 (p = 0.022) and a trend towards association with vimentin expression (p = 0.051). PNENs larger than 2 cm showed a trend towards association with E-cadherin (p = 0.052) and CK19 (p = 0.054) levels. Thus, proliferation, EMT and loss of cell adhesion are among the pathways of increasing tumour mass, paralleling the metastatic spread.

pTNM

The pT frequency distribution between pT1 (25.0%), pT2 (37.5%) and pT3 (37.5%) was even. A similar occurrence of pT3 tumours (32.4%) has been reported [45]. Significant survival differences were previously reported between pT1-2 vs. pT3-4, as well as between T3 and T4 [49]. It could be attributable to the tumour invasion in extrapancreatic tissues and large arteries in the cases of pT3 and pT4. In our study, there was a significant association between pT, tumour grade 2004 (p = 0.005) and vascular invasion (p = 0.005), explaining the association between pT and prognosis through common mechanisms of anaplasia and invasive tumour properties. pN1 was found in 20.0% of PNEN, in accordance with other observations [41, 45-47, 50] and was strongly predicted by older age. Age exceeding 60 years has been shown to be associated with lower survival, even in patients with low-grade (G1) gastroenteropancreatic PNENs [51]. Similarly, non-functioning PNEN patients older than 55 years had a worse survival [39]. Distant metastases were previously reported in 10.8- 27.0% of surgically treated PNENs and were found in 12.5% of patients in the present study [12, 49, 52]. The pM1 was associated with higher grade 2004 (p = 0.034) and increased mitotic count (p = 0.022), accentuating cell proliferation.

Resection margins

In order to assess the completeness of surgical treatment, as well as the recurrence risk, surgical RM must be examined carefully. In PNENs, R0 can be reached in 68-87.1% of patients [47, 53]. R0 PNENs still recur in 23.1% of patients, leading to death in 11.1% [46]. However, the five-year survival rate in R0 cases is statistically significantly higher than in the R1: 91.2% vs. 55.0%, respectively [47]. In the present study, pR1 status in PNENs was asso­ciated with mitotic count (p = 0.017) and had a trend towards the association with the grade 2004 (p = 0.051). Although shorter survival has been reported in patients with perineural invasion, no association between perineural growth and pR1 has been found in the present study or by other authors [12, 52] that can be explained by limited, mainly intratumoural perineural spread of PNEN, contrasting with pancreatic ductal adenocarcinoma [54].

Ki-67

Proliferation is a characteristic general tumour feature that is also essential in PNEN grading [4]. The observed invariable Ki-67 expression with a median positive cell count of 2% was in agreement with other authors [41, 49]. Regarding elevated Ki-67 expression in PNEN, the most convincing correlation is described with more frequent metastatic spread [55]. In this study, there was strong correlation with tumour stage (rs = 0.688; p = 0.028). Along with a higher ability to cell division, PNENs acquired CK19 expression (p = 0.033) and EMT by vimentin expression (p = 0.045). These markers, similarly to Ki-67, were not observed in islets. Thus, the triad of Ki-67, CK19 and vimentin up-regulation can be used as a tool to identify aggressive behaviour in PNENs. Notably, there was a trend towards negative correlation (p = 0.051) between Ki-67 and stem cell phenotype by CD44 expression, consistent with the stem cell nature of slow, steady renewal.

p53

The “genome guard”, p53 protein is normally found within cells in small quantities due to a short half-life. TP53 mutations can result in the synthesis of aberrant p53 proteins that have longer half-lives and thus accumulate in cells and can be detected by IHC. The TP53 mutation analyses and p53 IHC provide two different levels of molecular examination lacking correlation between gene mutations and aberrant p53 protein expression [56]. The p53 protein expression in PNENs was infrequent in accordance to the previously reported rare occurrence of TP53 mutations and IHC expression of p53 protein in PNENs [57]. Occasionally, a higher frequency (49%) of p53 protein expression has been described in gastroenteropancreatic NENs due to high expression in gastric neuroendocrine neoplasms [58]. Although p53 expression has been attributed to malignant PNENs [5], in the present study, it was observed in well-differentiated PNENs with unclear behaviour. Hypothetically, these tumours have a malignant nature but do not yet demonstrate invasive growth, similarly to PanIN-3 lesion in pancreatic ductal adenocarcinoma. Using the WHO 2010 classification, p53 expression was not limited to any particular grade in accordance with previous publications [58]. In PNENs, no association has been observed between immunohistochemically detected p53 protein expression and survival [59]. It has been hypothesised that p53 is usually not directly altered in gastroenteropancreatic NENs, but the other molecules involved in the whole p53 pathway are inactivated. Accordingly, we did not observe significant differences regarding p53 expression in PNENs and benign endocrine precursors. Analogous considerations are expressed regarding retinoblastoma tumour suppressor protein (Rb) pathway [60] favouring cyclin D1 as one of the key factors for cell cycle dysregulation in NENs, in contrast to epithelial carcinomas [61].

p21

The p21 protein is a cell cycle inhibitor, which controls cell cycle progression, apoptosis and transcription. It has a dual function, including proliferation inhibition and positive modulation, and can play anti- and pro-apoptotic function, depending on the nature of the apoptotic stimulus [62]. Although strong expression of p21 in gastroenteropancreatic neuroendocrine neoplasms has been previously associated with poor outcome [6], in our study, positive p21 expression showed limited correlations, possibly due to this duality. In vitro, induction of p21 has been associated with aspirin-induced inhibition of neuroendocrine tumour cell viability [63], further emphasising the controversies in the interpretation of p21 presence. Regarding other endocrine neoplasms, no difference of p21 levels in thyroid follicular adenomas and carcinomas has been found [64]. In the present study, a significant correlation with E-cadherin was observed, providing evidence for the link between cell cycle regulation and cell adhesion and invasion.

p27

The p27 is a cyclin-cyclin dependent kinase inhibitor and tumour suppressor. It inhibits cell cycle progression, mediating G1 arrest [7]. In cancer, IHC expression of p27 is decreased due to impaired synthesis or accelerated degradation of the relevant protein [65]. In a mouse model, lack of p27 led to neuroendocrine hyperplasia. However, contact inhibition of cell proliferation remained unchanged, signifying necessity for other molecular alterations to complete the tumorigenesis [66]. Consequently, we did not observe significant differences between p27 levels in PNENs and islets. Alternatively, p27 alterations can be late event in NEN development, related to high tumour grade. Loss of p27 in gastroenteropancreatic NENs has been associated with worse prognosis in some [7, 67], but not all [6], studies. Disappearance of p27 was characteristic in poorly differentiated neuroendocrine carcinomas and metastatic well-differentiated endocrine carcinomas (by 2004 WHO classification), as Grabowski et al. [67] reported. Such tumours were infrequent in the present study group. This could explain why no morphological associations were identified in the present study, contrasting with Kim et al. [7].

Cyclin D1

The transcriptional regulator cyclin D1 forms a complex with cyclin dependent kinases 4 and 6 that phosphorylate and thus inactivate the tumour suppressor protein Rb, resulting in the cell cycle progress from the G1 to S phase. It is among the main proliferation control mechanisms [61]. In malignant cells, the level of cyclin D1 can increase due to gene rearrangement, amplification, transcriptional up-regulation [61] or affected degradation [68]. Rb gene is infrequently affected in NENs, favouring involvement of other molecules within the Rb pathway [69].

In the present study, there was no significant difference between the cyclin D1 expression in islets and PNENs. Interestingly, in the mouse model, induced cyclin D1 expression in pancreatic -cells resulted in islet hyperplasia that was not associated with hypoglycaemia, diabetes or tumour [70]. In turn, the relevance of cyclin D1 to the neuroendocrine oncogenesis was emphasised by the interaction with beta-catenin, an important component of the E-cadherin pathway [71]. Extracellular-regulated kinase ERK and p38/mitogen-activated protein kinase MAPK pathways are also involved both in the regulation of cyclin D1 levels [8] and the down-regulation of E-cadherin [72]. Cyclin D accumulation can be the downstream effect of beta-catenin accumulation occurring within E-cadherin/beta-catenin function loss [60]. Alternatively, up-regulation of cyclin D1 is considered an early event in the development of neuroendocrine neoplasms [61].

In PNENs, cyclin D1 possibly should be analysed in conjunction with other proteins to distinguish between hyperplastic and neoplastic pathways. Such complex analysis could include E-cadherin, but should not be limited to it. In the INK4a-ART tumour suppressive network, p21 and p27 loss can have a similar effect as cyclin D1 hyperexpression [60]. Hypothetically, E-cadherin expression could have a higher prognostic value, serving as the limiting measure of molecular deregulation in PNENs, while cyclin D1 expression [73] would be too ubiquitous for prognostic estimates. Other research groups have also failed to identify an association between cyclin D1 expression and survival or clinical and pathological variables [58].

Bcl-2

Bcl-2 is an anti-apoptotic protein. Translocation in the BCL2 gene yields increased expression of the Bcl-2 protein [56]. It is considered that PNEN development necessitates both derangement of cell proliferation and apoptosis [60]. The estimates of Bcl-2 expression are contradictory, ranging from relatively frequent (53.3-53.6%) to no reactivity in PNENs [5, 74]. The hypothetical explanation of this controversy is as follows. In pancreatic carcinoma, over-expression of another anti-apoptotic protein belonging to the Bcl-2 family – Bcl-xL is more common and has a greater relevance [75]. Thus, cross-reactivity can be responsible for most of the positive cases, but was avoided in the present study. It must be emphasised that positive controls (including internal positive control of lymphocytes) were invariably performed and were reactive.

E-cadherin

E-cadherin is an epithelial transmembrane glycoprotein supporting epithelial layer integrity and polarity [56]. Loss of E-cadherin or displacement of it, apart from the cell membrane, has been observed in advanced malignant tumours. The resulting functional derangement of E-cadherin weakens cell-cell adhesion and facilitates cell migration and invasion. Lower E-cadherin levels are associated with a higher incidence of cancer metastasis [76]. In the present study, we confirmed a significant association between E-cadherin loss and advanced tumours, characterised by a higher pT (p = 0.011) in parallel with previously reported association between larger tumour size and the loss of E-cadherin in pancreatic and pulmonary NENs [56, 77]. Both in pulmonary neuroendocrine tumours and in PNENs, reduced E-cadherin expression or displacement of this protein from the cell membrane to the cytoplasm has been associated with the presence of LN metastases [56, 77]. Loss of E-cadherin in PNENs is characterised by more frequent liver metastases [56]. In the present study, decreased E-cadherin levels were associated with more frequent vascular and perineural invasion (both p = 0.003), showing a higher spreading capacity of the E-cadherin-losing tumour. E-cadherin significantly correlated with p53 status (p = 0.011). Wild type p53 proteins can suppress EMT that involves up-regulation of vimentin and loss of E-cadherin; recently, the mechanism has been studied in regard to microRNA regulation. The miR-154-mediated mechanism has been shown in the prostate cancer cell line [78]. Here, we provide in vivo evidence of the p53 and E-cadherin relationship in PNENs.

Molecularly, we also identified a significant association between E-cadherin and the cell cycle regulator p21 (p = 0.046). In hepatocellular carcinoma, miR-148b was recently shown to regulate p21 levels, as well as to influence the Wnt pathway, finally up-regulating E-cadherin expression [79]. Here, we provide in vivo evidence of analogous p21 and E-cadherin association in PNENs. E-cadherin levels in PNENs showed also a strong positive correlation with angiogenesis, as reflected by MVD (p = 0.005). In turn, high vascularity is associated with a lower NEN grade [80]. In the present study, decreased E-cadherin levels were associated with grade 2004 (p = 0.007). Similarly, in pulmonary NENs, functional derangement of E-cadherin complex was associated with a higher proliferation activity, the hallmark of PNEN grading [77].

CD44

CD44 is a transmembrane adhesion molecule that participates in the interaction between different cells or between cells and the matrix, as well as in cell migration [81]. It is also characteristic for stem-like cancer cells, representing self-renewing cells, able to promote clonogenicity, cell growth and migration, metastatic spread and resistance to chemotherapy [11, 27].

The identified trend towards a negative correlation between CD44 and Ki-67 in PNENs (p = 0.051) was typical for the slow, continuous regeneration predicted for stem cells. CD44 expression has been shown in mesenchymal stromal cells that were able to differentiate into islet cells [82]. In accordance with this, a small subpopulation of CD44-positive cells was almost invariably evident in non-neoplastic islets in the present study. However, tissue stem cells exhibit similarities with cancer stem cells, but also show differences [83].

In pulmonary carcinoids, low levels of CD44 mRNA and absence of the CD44 protein are associated with a low 20-year survival [84]. In a combined group of pulmonary and gastrointestinal neuroendocrine neoplasms originally described as carcinoids, the patients with lymph node or distant metastasis lacked CD44 in the tumours significantly more frequently [85]. Thus, the prognostic role of CD44 expression in NENs is different from the negative findings in many carcinomas, including pancreatic ductal adenocarcinoma, gastric adenocarcinoma, clear cell renal carcinoma or non-small cell lung carcinoma, among others [11, 86-88]. In our study, there also was no association between CD44 and such morphological parameters, such as high grade and stage, which are indicative of a poor prognosis.

Thus, in neuroendocrine neoplasms, CD44 exhibits mainly tumour suppressor functions, promoting cell adhesion, limiting metastatic spread and activating apoptosis in accordance with the reported tumorigenicity of CD44-null fibroblasts in nude mice or tumour apoptosis studies [84]. The dual functions of CD44 are pathogenetically important. In practice, pathologists must be aware of the diverse prognostic role of CD44 in tumours of different histogenesis.

Cytokeratins

Although few articles have been devoted to the prognostic value of CKs, the expression of CK19 has been associated with shorter survival in some [12, 13, 52], but not all, studies [32, 39]. In the present study, CK19 was associated with aggressive characteristics, including higher pT (p = 0.011), stage (p = 0.025), grade 2004 (p = 0.002), invasive capacity confirmed by vascular (p = 0.020), lymphatic (p = 0.043) and perineural (p = 0.026) invasion, proliferative activity by mitotic count (p = 0.008) and Ki-67 (p = 0.033), EMT (p = 0.041) and suppressed angiogenesis (p = 0.014). The findings are in accordance with previous studies reporting increased tumour size, higher mitotic activity, lymphatic, vascular and perineural invasion, higher grade and higher TNM parameters in CK19-expressing PNENs [13, 32]. Our findings suggest that CK19 confers increased plasticity to tumour cells, facilitating local invasion and metastatic spread similar to pancreatic ductal adenocarcinoma, which typically expresses CK19 [89] and is known for its fulminant course.

Alternative explanations for CK19 role exist as well. CK19 is present in benign pancreatic ducts [32]. As ductal epithelia can up-regulate the endocrine cells [90], self-stimulation can be provided by CK19-positive PNEN cells. Some authors have hypothesised that PNENs develop from pluripotential ductal stem cells [91], analogous to the unitarian theory regarding colonic carcinoma and endocrine tumours [92]. This hypothesis is supported by the occasional coexistence of PNEN and intraductal papillary mucinous pancreatic neoplasms [93, 94]. If so, CK19 can be the precursor marker of pristine PNENs, retaining some features of the pluripotential primitive ancestor cell. This is in accordance with the embryonic development of the pancreas characterised by CK19 expression in the endocrine component during the early stages, while later it is retained in ducts, but absent from islets [13, 32]. The observed statistically significant (p = 0.006) up-regulation of CK19 in 28.6% of PNENs thus recapitulates early developmental stages.

CK20 is absent from normal pancreatic islets [26, 56]. Although CK20 expression in PNENs has been reported in up to 33% of cases [95], it was not observed in the present study. The published data also vary widely – only 5% of gastroenteropancreatic tumours expressed CK20 in Knosel et al. [96] study.

CDX2

CDX2 is a caudal-related homeobox transcription factor, expressed in the intestinal epithelium. It discloses intestinal differentiation, both in non-neoplastic or malignant cells. The labelling in pancreatic tissues is usually lighter and more variegated than in colorectal carcinoma [56]. Although expression of CDX2 has been reported in a small fraction of PNENs [97, 98], it was absent both from PNENs and islets in the present study.

Squamous differentiation

The endocrine pancreatic tissues and corresponding tumours lacked CK5/6 and CK34E12 proteins that were found in cells exhibiting squamous differentiation [56]. To the best of our knowledge, there are no previously published reports evaluating the trend towards squamous differentiation in PNENs by p63 expression. In the present study, it was observed in 28.6% of cases. Although the expression did not reach the cut-off level of 10% positive cells, the up-regulation was significant in contrast to non-neoplastic endocrine cells in the pancreatic islets (p = 0.034). Thus, association with tumorigenesis could be suspected rather than the squamous differentiation. The pathogenetic role of p63 in a fraction of PNENs is further supported by the trend towards correlation with cell cycle regulator p21 (p = 0.056) and by the invasiveness of p63-expressing PNENs, as p63 expression was significantly higher in tumours that invaded blood vessels (p = 0.024). The association with p53 (p < 0.001) and E-cadherin (p = 0.009) was also indirect evidence of aggressive features. Previously, expression of p63 was reported in pulmonary non-small cell neuroendocrine carcinomas [99] and neuroendocrine carcinomas of the head and neck [100] in studies of diagnostic value.

Vimentin

Vimentin is a major mesenchymal intermediate filament, controlling cellular motility, signalling and directional migration [101]. Malignant pancreatic epithelial cells acquire vimentin and lose E-cadherin during EMT, resulting in tumour invasiveness, metastatic capacity, resistance to chemotherapy and the generation of cancer stem cells. Although EMT has been associated with cancer progression, it has also been attributed to fibrosis due to chronic inflammation. Fibrosis is associated with increased numbers of myofibroblasts arising from epithelial cells through EMT [102]. Since pancreatic islets lack fibrosis even in inflammation, EMT is absent from islets. In PNENs, vimentin was significantly up-regulated in comparison with islets (p = 0.006), suggesting a significant role in neuroendocrine tumorigenesis. Vimentin also showed a positive correlation with CK19 (p = 0.041), which is associated with increased invasiveness in PNENs. Thus, particularly aggressive PNEN could be characterised by an up-regulated triad, i.e., Ki-67, CK19 and vimentin. In pulmonary NENs, vimentin has been significantly associated with higher tumour grade [77]. Appendicular NENs, having favourable clinicopathological profiles, mostly lack vimentin [51].

Alternatively, vimentin could mark a separate molecular subtype of neuroendocrine neoplasms, as evidenced by relatively frequent occurrence and the reported association with low biological potential [51].

The observed frequency of EMT in PNENs in this study is within the reported range (11–58%) in gastroenteropancreatic neuroendocrine neoplasms and/or rectal carcinoids [51, 103]. Coexpression of vimentin and neuroendocrine markers has also been reported in selected peripheral and visceral glomus tumours that show clear cell morphology [104, 105], PNEN associated with tuberous sclerosis [106], and in 80% of renal NENs [107].

Microvascular density

Although rich vascularity is typical in NENs, the differences between these tumours and the precursors are less known. In the present study, PNENs retained the same MVD as islets. In an experimental model of digestive NENs, angiogenesis was associated with well-differentiated cell lines and tumours, and lacked correlations with tumour progression, as well as invasive and metastatic properties [80]. Analogous clinical observations have been published [108]. Low vascularity is characteristic for poorly differentiated neuroendocrine carcinomas [109] and can be quantified by non-invasive xenon-inhalation computed tomography [110]. Pathogenetically, hypoxia-inducible factors are associated with malignant progression of well-differentiated neuroendocrine tumours of the ileum [111].

COX-2

COX-2 is the rate-limiting enzyme in the production of prostaglandins from arachidonic acid. It is expressed in gastric, colorectal and pancreatic carcinomas [56], and has been associated with poor outcome, whereas the risk of cancer mortality is reduced by long-term non-steroidal anti-inflammatory drug therapy. COX-2 selectively promotes the expression and activity of the neuroendocrine marker chromogranin A, and therefore it is a promising research direction in NENs [112]. COX-2 expression has been explored in neuroendocrine tumours, including studies of the biological potential and therapeutic possibilities. It was widely expressed in up to 82% of medullary thyroid carcinoma, showing statistically significant up-regulation in comparison with benign thyroid tissue [113]. In phaeochromocytoma, COX-2 is associated with malignant behaviour and, along with lack of nm-23 and presence of galectin-3, can be applied to predict malignancy [114]. Immunohistochemical expression of the COX-2 protein has been identified in 54-75% of gastroenteropancreatic [115, 116] and 65% of pancreatic NENs [117]. In our study, we showed statistically significant up-regulation of COX-2 expression in PNENs, contrasting with invariably negative islets (p = 0.016) and providing in vivo evidence that COX-2 is involved in the pancreatic neuroendocrine tumorigenesis.

The prognostic role of COX-2 in PNENs has been addressed in a few studies. COX-2 overexpression in gastroenteropancreatic NENs was significantly associated with poor survival. However, only 10/247 NENs in this cohort originated in the pancreas [116]. In a cohort of mid-gut carcinoids comprising NENs of the distal duodenum, jejunum, ileum and proximal part of the colon, a higher COX-2 histoscore was significantly associated with worse survival within the COX-2 positive group [118]. The correlation between COX-2 and proliferation activity by Ki-67 could at least partially explain the negative prognostic meaning of COX-2 [116]. COX-2 expression in gastroenteropancreatic NENs correlated significantly with grade, but not primary location [116]. In contrast, Bergmann et al. concluded that COX-2 expression was independent of the malignant potential of pancreatic endocrine tumours [3]. In the present study, COX-2 expressing PNENs more frequently invaded lymphatic vessels (p = 0.043), suggesting an undesirable course. The association between COX-2 overexpression and lymphatic invasion in gastroenteropancreatic NENs was reported by Ince et al. [115] and Kim et al. [116]. Higher invasive capacity of COX-2-expressing PNEN represents a pathogenetic mechanism for worse survival and is further substantiated by the observed association between COX-2 and the cytokeratin profile, influencing the tumour cell plasticity.

COX-2 inhibitors have been studied in regard of NEN treatment [3]. In vitro NEN studies have been encouraging, revealing the induction of apoptosis [119] and the potentiation of chemotherapy [120] in the medullary thyroid cancer cell line. Enhanced apoptosis has also been observed in PNEN cell lines [3]. The mechanisms of COX-2 inhibitor activity include reduction of the expression of multidrug resistance 1 gene MDR1, resulting in lower levels of the corresponding transmembrane glycoprotein, that prevents intracellular accumulation of cytotoxic medications [120]. Cell proliferation is also decreased due to cell cycle arrest at the G0/G1 transition [119]. However, at present, significant controversies exist regarding the dynamics of COX-2 up-regulation in neuroendocrine neoplasms. In PNENs, COX-2 was more frequently found in the primary tumour than in the metastasis [3]. Cadden et al. [118] reported contrary findings. Considering the focality and heterogeneity of COX-2 expression in this and other studies [121], as well as the beneficial influence of COX-2 inhibition on tumour chemosensitivity, COX-2 inhibitors would be useful only as part of complex therapy.

Neuroendocrine markers

Neuroendocrine markers are the primary diagnostic tools in PNENs. As both chromogranin A and CD56 are invariably expressed in PNENs and islets, both markers can be equally used in PNEN diagnostics. Interestingly, despite the high level of chromogranin A expression, the reactivity is still up-regulated in PNENs at a statistically significant level (p = 0.042). The mechanisms of carcinogenesis can explain these changes, e.g., selective and functionally important up-regulation of chromogranin A by COX-2 [112].

Our report has some limitations, especially with the small size of the study group, despite enrolment of all PNENs within a 10-year long period. We restricted the time period to 10 years in order to avoid loss of immunohistochemical reactivity [122]. Appropriate methods of statistical analysis were selected [30]. The single-hospital approach ensured a homogeneous study group. The strength of our study includes a simultaneous analysis of 20 immunohistochemical markers and 12 gross and microscopic parameters in the same study group using non-neoplastic islets for comparison. The integrated approach allowed us to reveal multiple significant morphologic and molecular associations including the up-regulated triad of Ki-67, vimentin and CK19 as a tool to detect aggressive behaviour of PNEN. Loss of E-cadherin can also be used in the complex with these markers. We also were able to show a link between tumour cell adhesion, invasion and cell cycle regulation as disclosed by the associations between Ki-67, p21, p53, CK19 and E-cadherin. The expression and role of p63 in PNENs has been reported for the first time.

Conclusions

Significant up-regulation of Ki-67, CK19, p63, chromogranin A, vimentin and COX-2 expression was found in PNENs in comparison with non-neoplastic pancreatic islets.

In PNENs, proliferation activity (by Ki-67), E-cadherin, vimentin and CK19 are important molecular traits due to widespread associations with such important oncological parameters, like pTNMGR, tumour size and invasive growth. The triad of Ki-67, CK19 and vimentin up-regulation in association with E-cadherin loss can be used as a tool to identify aggressive behaviour in PNENs.

CD44 expression shows limited associations, contrasting with the negative role in epithelial non-neuroendocrine neoplasms.

The correlations between CK19 and Ki-67, as well as between E-cadherin and p21, provide evidence of the links between cell plasticity, adhesion and proliferation.

To the best of our knowledge, we reported p63 expression in PNENs for the first time. The associations with p53 and E-cadherin suggest that p63 in PNENs has a tumorigenetic role, rather than serving as evidence of squamous differentiation.

PNENs retain the same microvascular density characteristic for islets. The microvascular density shows a negative correlation with mitotic count, thus lower vascularity can be expected in high-grade NENs. MVD is associated with p53, cyclin D1, E-cadherin, p63 and CK19 molecular pathways.

Notably, the correlations between the molecular features of PNENs and tumour grades using the 2004 and 2010 WHO classifications differ remarkably.

The authors declare no conflict of interest.

The present work was carried out within the frames of scientific project Nr. 2013/0004/1DP/1.1.1.2.0/13/APIA/VIAA/020, supported by the European Social Fund (ESF).

References

1. Öberg K, Knigge U, Kwekkeboom D, et al. Neuroendocrine gastro-entero-pancreatic tumors: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol 2012; 23: vii124-vii30.

2. Reid MD, Balci S, Saka B, et al. Neuroendocrine tumors of the pancreas: Current concepts and controversies. Endocr Pathol 2014; 25: 65-79.

3. Bergmann F, Breinig M, Höpfner M, et al. Expression pattern and functional relevance of epidermal growth factor receptor and cyclooxygenase-2: Novel chemotherapeutic targets in pancreatic endocrine tumors? Am J Gastroenterol 2009; 104: 171-181.

4. McCall CM, Shi C, Cornish TC, et al. Grading of well-differentiated pancreatic neuroendocrine tumors is improved by the inclusion of both Ki67 proliferative index and mitotic rate. Am J Surg Pathol 2013; 37: 1671-1677.

5. Yachida S, Vakiani E, White CM, et al. Small cell and large cell neuroendocrine carcinomas of the pancreas are genetically similar and distinct from well-differentiated pancreatic neuroendocrine tumors. Am J Surg Pathol 2012; 36: 173-184.

6. Lee HS, Chen M, Kim JH, et al. Analysis of 320 gastroenteropancreatic neuroendocrine tumors identifies TS expression as independent biomarker for survival. Int J Cancer 2014; 135: 128-137.

7. Kim HS, Lee HS, Nam KH, et al. p27 loss is associated with poor prognosis in gastroenteropancreatic neuroendocrine tumors. Cancer Res Treat 2014; 46: 383-392.

8. Guo SS, Wu X, Shimoide AT, et al. Frequent overexpression of cyclin D1 in sporadic pancreatic endocrine tumours. J Endocrinol 2003; 179: 73-79.

9. Hong SM, Li A, Olino K, et al. Loss of E-cadherin expression and outcome among patients with resectable pancreatic adenocarcinomas. Mod Pathol 2011; 24: 1237-1247.

10. Mohammed A, Janakiram NB, Brewer M, et al. Antidiabetic drug metformin prevents progression of pancreatic cancer by targeting in part cancer stem cells and mTOR signalling. Transl Oncol 2013; 6: 649-659.

11. Hou YC, Chao YJ, Tung HL, et al. Coexpression of CD44-positive/CD133-positive cancer stem cells and CD204-positive tumor-associated macrophages is a predictor of survival in pancreatic ductal adenocarcinoma. Cancer 2014; 120: 2766-2777.

12. Deshpande V, Fernandez-del Castillo C, Muzikansky A, et al. Cytokeratin 19 is a powerful predictor of survival in pancreatic endocrine tumors. Am J Surg Pathol 2004; 28: 1145-1153.

13. Han X, Zhao J, Ji Y, et al. Expression of CK19 and KIT in resectable pancreatic neuroendocrine tumors. Tumour Biol 2013; 34: 2881-2889.

14. Handra-Luca A, Hong SM, Walter K, et al. Tumour epithelial vimentin expression and outcome of pancreatic ductal adenocarcinomas. Br J Cancer 2011; 104: 1296-1302.

15. Chu X, Gao X, Jansson L, et al. Multiple microvascular alterations in pancreatic islets and neuroendocrine tumors of a Men1 mouse model. Am J Pathol 2013; 182: 2355-2367.

16. Verbeke CS, Leitch D, Menon KV, et al. Redefining the R1 resection in pancreatic cancer. Br J Surg 2006; 93: 1232-1237.

17. Menon KV, Gomez D, Smith AM, et al. Impact of margin status on survival following pancreatoduodenectomy for cancer: The Leeds Pathology Protocol (LEEPP). HPB (Oxford) 2009; 11: 18-24.

18. DeLellis RA, Lloyd RV, Heitz PhU, et al. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of Endocrine Organs. DeLellis RA, Lloyd RV, Heitz PhU, et al. (eds). 3rd ed. IARC Press, Lyon 2004; 183-194.

19. Edge SB, Byrd DR, Compton CC, et al. The AJCC Cancer Staging Handbook. Edge SB, Byrd DR, Compton CC, et al. (eds). 7th ed. Springer, New York 2010; 285-296.

20. Chu PG, Weiss LM. Expression of cytokeratin 5/6 in epithelial neoplasms: An immunohistochemical study of 509 cases. Mod Pathol 2002; 15: 6-10.

21. Culhaci N, Sagol O, Karademir S, et al. Expression of transforming growth factor-beta-1 and p27Kip1 in pancreatic adenocarcinomas: relation with cell-cycle-associated proteins and clinicopathologic characteristics. BMC Cancer 2005; 5: 98.

22. Mielko J, Polkowski WP, Skomra DG, et al. Prognostic value of p27 kip1 expression in adenocarcinoma of the pancreatic head region. HPB (Oxford) 2006; 8: 216-222.

23. Burford H, Baloch Z, Liu X, et al. E-cadherin/beta-catenin and CD10: A limited immunohistochemical panel to distinguish pancreatic endocrine neoplasm from solid pseudopapillary neoplasm of the pancreas on endoscopic ultrasound-guided fine-needle aspirates of the pancreas. Am J Clin Pathol 2009; 132: 831-839.

24. Takikita M, Altekruse S, Lynch CF, et al. Associations between selected biomarkers and prognosis in a population-based pancreatic cancer tissue microarray. Cancer Res 2009; 69: 2950-2955.

25. Lozano-Leon A, Perez-Quintela BV, Iglesias-García J, et al. Ductal adenocarcinoma of the pancreas: Expression of growth factor receptors, oncogenes and suppressor genes, and their relationship to pathological features, staging and survival. Oncol Lett 2011; 2: 161-166.

26. Liu H, Shi J, Anandan V, et al. Reevaluation and identification of the best immunohistochemical panel (pVHL, Maspin, S100P, IMP-3) for ductal adenocarcinoma of the pancreas. Arch Pathol Lab Med 2012; 136: 601-609.

27. Tajima H, Ohta T, Kitagawa H, et al. Neoadjuvant chemotherapy with gemcitabine for pancreatic cancer increases in situ expression of the apoptosis marker M30 and stem cell marker CD44. Oncol Lett 2012; 3: 1186-1190.

28. Koyuncuer A. Immunohistochemical expression of p63, p53 in urinary bladder carcinoma. Indian J Pathol Microbiol 2013; 56: 10-15.

29. Altman D, Machin D, Bryant T, et al. Statistics with Confidence. In: Confidence Interval and Statistical Guidelines. Altman D, Machin D, Bryant T, et al. (eds.). 2nd ed. Wiley-Blackwell, Bristol 2000; 1-254.

30. Hill T, Lewicki P. Statistics: Methods and Applications. 1st ed. StatSoft, USA 2006; 1-831.

31. Krampitz GW, Norton JA. Pancreatic neuroendocrine tumors. Curr Probl Surg 2013; 50: 509-545.

32. Son EM, Kim JY, An S, et al. Clinical and prognostic significances of cytokeratin 19 and KIT expression in surgically resectable pancreatic neuroendocrine tumors. J Pathol Transl Med 2015; 49: 30-36.

33. Lawrence B, Gustafsson BI, Chan A, et al. The epidemiology of gastroenteropancreatic neuroendocrine tumors. Endocrinol Metab Clin North Am 2011; 40: 1-18.

34. Morin E, Cheng S, Mete O, et al. Hormone profiling, WHO 2010 grading, and AJCC/UICC staging in pancreatic neuroendocrine tumor behaviour. Cancer Med 2013; 2: 701-711.

35. Yang M, Zeng L, Zhang Y, et al. Surgical treatment and clinical outcome of non-functional pancreatic neuroendocrine tumors: A 14-year experience from one single center. Medicine (Baltimore) 2014; 93: e94.

36. Heestand GM, Kurzrock R. Molecular landscape of pancreatic cancer: Implications for current clinical trials. Oncotarget 2015; 6: 4553-4561.

37. Lin XS, Zhu CL, Liu CH, et al. Analysis of diagnosis and treatment for pancreatic neuroendocrine neoplasms: Results of 47 cases in a single institution. Hepatogastroenterology 2014; 61: 2383-2386.

38. Cameron JL, He J. Two thousand consecutive pancreaticoduodenectomies. J Am Coll Surg 2015; 220: 530-536.

39. Cherenfant J, Talamonti MS, Hall CR, et al. Comparison of tumor markers for predicting outcomes after resection of nonfunctioning pancreatic neuroendocrine tumors. Surgery 2014; 156: 1504-1510.

40. Chan JK, Ip YT, Cheuk W. The utility of immunohistochemistry for providing genetic information on tumors. Int J Surg Pathol 2013; 21: 455-475.

41. Rindi G, Falconi M, Klersy C, et al. TNM staging of neoplasms of the endocrine pancreas: results from a large international cohort study. J Natl Cancer Inst 2012; 104: 764-777.

42. Lam KY, Lo CY. Pancreatic endocrine tumour: A 22-year clinico-pathological experience with morphological, immunohistochemical observation and a review of the literature. Eur J Surg Oncol 1997; 23: 36-42.

43. Bettini R, Boninsegna L, Mantovani W, et al. Prognostic factors at diagnosis and value of WHO classification in a mono-institutional series of 180 non-functioning pancreatic endocrine tumours. Ann Oncol 2008; 19: 903-908.

44. Bartsch DK, Albers M, Knoop R, et al. Enucleation and limited pancreatic resection provide long-term cure for insulinoma in multiple endocrine neoplasia type 1. Neuroendocrinology 2013; 98: 290-298.

45. Takahashi Y, Akishima-Fukasawa Y, Kobayashi N, et al. Prognostic value of tumor architecture, tumor-associated vascular characteristics, and expression of angiogenic molecules in pancreatic endocrine tumors. Clin Cancer Res 2007; 13: 187-196.

46. Kim MJ, Choi DW, Choi SH, et al. Surgical strategies for non-functioning pancreatic neuroendocrine tumours. Br J Surg 2012; 99: 1562-1568.

47. Haynes AB, Deshpande V, Ingkakul T, et al. Implications of incidentally discovered, nonfunctioning pancreatic endocrine tumors: Short-term and long-term patient outcomes. Arch Surg 2011; 146: 534-538.

48. Kishi Y, Shimada K, Nara S, et al. Basing treatment strategy for non-functional pancreatic neuroendocrine tumors on tumor size. Ann Surg Oncol 2014; 21: 2882-2888.

49. Scarpa A, Mantovani W, Capelli P, et al. Pancreatic endocrine tumors: Improved TNM staging and histopathological grading permit a clinically efficient prognostic stratification of patients. Mod Pathol 2010; 23: 824-833.

50. Partelli S, Gaujoux S, Boninsegna L, et al. Pattern and clinical predictors of lymph node involvement in nonfunctioning pancreatic neuroendocrine tumors (NF-PanNETs). JAMA Surg 2013; 148: 932-939.

51. Galvan JA, Astudillo A, Vallina A, et al. Epithelial-mesenchymal transition markers in the differential diagnosis of gastroenteropancreatic neuroendocrine tumors. Am J Clin Pathol 2013; 140: 61-72.

52. Zhang L, Smyrk TC, Oliveira AM, et al. KIT is an independent prognostic marker for pancreatic endocrine tumors: A finding derived from analysis of islet cell differentiation markers. Am J Surg Pathol 2009; 33: 1562-1569.

53. Pomianowska E, Gladhaug IP, Grzyb K, et al. Survival following resection of pancreatic endocrine tumors: Importance of R-status and the WHO and TNM classification systems. Scand J Gastroenterol 2010; 45: 971-979.

54. Bergmann F, Ceyhan GO, Rieker RJ, et al. Fundamental differences in the neural invasion behaviour of pancreatic endocrine tumors: Relevance for local recurrence rates? Hum Pathol 2009; 40: 50-57.

55. Ohike N, Morohoshi T. Pathological assessment of pancreatic endocrine tumors for metastatic potential and clinical prognosis. Endocr Pathol 2005; 16: 33-40.

56. Dabbs DJ. Diagnostic Immunohistochemistry. Theranostic and Genomic Applications. Dabbs DJ (ed.). 4th ed. Elsevier Saunders, Philadelphia 2014; 1-1008.

57. Jonkers YM, Claessen SM, Veltman JA, et al. Molecular parameters associated with insulinoma progression: Chromosomal instability vs. p53 and CK19 status. Cytogenet Genome Res 2006; 115: 289-297.

58. Liu SZ, Zhang F, Chang YX, et al. Prognostic impact of cyclin D1, cyclin E and P53 on gastroenteropancreatic neuroendocrine tumours. Asian Pac J Cancer Prev 2013; 14: 419-422.

59. La Rosa S, Klersy C, Uccella S, et al. Improved histologic and clinicopathologic criteria for prognostic evaluation of pancreatic endocrine tumors. Hum Pathol 2009; 40: 30-40.

60. Simon B, Lubomierski N. Implication of the INK4a/ARF locus in gastroenteropancreatic neuroendocrine tumorigenesis. Ann N Y Acad Sci 2004; 1014: 284-299.

61. Chung DC. Cyclin D1 in human neuroendocrine tumorigenesis. Ann N Y Acad Sci 2004; 1014: 209-217.

62. Cazzalini O, Scovassi AI, Savio M, et al. Multiple roles of the cell cycle inhibitor p21(CDKN1A) in the DNA damage response. Mutat Res 2010; 704: 12-20.

63. Spampatti M, Vlotides G, Spottl G, et al. Aspirin inhibits cell viability and mTOR downstream signaling in gastroenteropancreatic and bronchopulmonary neuroendocrine tumor cells. World J Gastroenterol 2014; 20: 10038-10049.

64. Ferenc T, Lewinski A, Lange D, et al. Analysis of P53 and P21WAF1 proteins expression in follicular thyroid tumors. Pol J Pathol 2004; 55: 133-141.

65. Chu IM, Hengst L, Slingerland JM. The Cdk inhibitor p27 in human cancer: Prognostic potential and relevance to anticancer therapy. Nat Rev Cancer 2008; 8: 253-267.

66. Lloyd RV, Erickson LA, Jin L, et al. p27kip1: A multifunctional cyclin-dependent kinase inhibitor with prognostic significance in human cancers. Am J Pathol 1999; 154: 313-323.

67. Grabowski P, Schrader J, Wagner J, et al. Loss of nuclear p27 expression and its prognostic role in relation to cyclin E and p53 mutation in gastroenteropancreatic neuroendocrine tumors. Clin Cancer Res 2008; 14: 7378-7384.

68. Alao JP. The regulation of cyclin D1 degradation: roles in cancer development and the potential for therapeutic invention. Mol Cancer 2007; 6: 24.

69. Lubomierski N, Kersting M, Bert T, et al. Tumor suppressor genes in the 9p21 gene cluster are selective targets of inactivation in neuroendocrine gastroenteropancreatic tumors. Cancer Res 2001; 61: 5905-5910.

70. Zhang X, Gaspard JP, Mizukami Y, et al. Overexpression of cyclin D1 in pancreatic beta-cells in vivo results in islet hyperplasia without hypoglycaemia. Diabetes 2005; 54: 712-719.

71. Iguchi H, Urashima Y, Inagaki Y, et al. SOX6 suppresses cyclin D1 promoter activity by interacting with beta-catenin and histone deacetylase 1, and its down-regulation induces pancreatic beta-cell proliferation. J Biol Chem 2007; 282: 19052-19061.

72. Gui T, Sun Y, Shimokado A, et al. The roles of mitogen-activated protein kinase pathways in TGF--induced epithelial-mesenchymal transition. J Signal Transduct 2012; 2012: 289243.

73. Capurso G, Festa S, Valente R, et al. Molecular pathology and genetics of pancreatic endocrine tumours. J Mol Endocrinol 2012; 49: R37-50.

74. Proca DM, Frankel WL. Pancreatic endocrine tumors-c-erb B2 (Her-2/neu), bcl-2, and p-53 immunohistochemical testing and their value in assessing prognosis. Appl Immunohistochem Mol Morphol 2008; 16: 44-47.

75. Westphal S, Kalthoff H. Apoptosis: targets in pancreatic cancer. Mol Cancer 2003; 2: 6.

76. Canel M, Serrels A, Frame MC, et al. E-cadherin-integrin crosstalk in cancer invasion and metastasis. J Cell Sci 2013; 126: 393-401.

77. Galvan JA, Astudillo A, Vallina A, et al. Prognostic and diagnostic value of epithelial to mesenchymal transition markers in pulmonary neuroendocrine tumors. BMC Cancer 2014; 14: 855.

78. Diaz-Lopez A, Moreno-Bueno G, Cano A. Role of microRNA in epithelial to mesenchymal transition and metastasis and clinical perspectives. Cancer Manag Res 2014; 6: 205-216.

79. Zhang JG, Shi Y, Hong DF, et al. MiR-148b suppresses cell proliferation and invasion in hepatocellular carcinoma by targeting WNT1/-catenin pathway. Scientific Reports 2015; 5: 8087.

80. Poncet G, Villaume K, Walter T, et al. Angiogenesis and tumor progression in neuroendocrine digestive tumors. J Surg Res 2009; 154: 68-77.

81. Korski K, Malicka-Durczak A, Breborowicz J. Expression of stem cell marker CD44 in prostate cancer biopsies predicts cancer grade in radical prostatectomy specimens. Pol J Pathol 2014; 65: 291-295.

82. Gopurappilly R, Bhat V, Bhonde R. Pancreatic tissue resident mesenchymal stromal cell (MSC)-like cells as a source of in vitro islet neogenesis. J Cell Biochem 2013; 114: 2240-2247.

83. Kaur S, Singh G, Kaur K. Cancer stem cells: An insight and future perspective. J Cancer Res Ther 2014; 10: 846-852.

84. Swarts DR, Henfling ME, Van Neste L, et al. CD44 and OTP are strong prognostic markers for pulmonary carcinoids. Clin Cancer Res 2013; 19: 2197-2207.

85. Sun X, Gong Y, Talamonti MS, et al. Expression of cell adhesion molecules, CD44s and E-cadherin, and microvessel density in carcinoid tumors. Mod Pathol 2002; 15: 1333-1338.

86. Cao X, Cao D, Jin M, et al. CD44 but not CD24 expression is related to poor prognosis in non-cardia adenocarcinoma of the stomach. BMC Gastroenterol 2014; 14: 157.

87. Roudi R, Madjd Z, Korourian A, et al. Clinical significance of putative cancer stem cell marker CD44 in different histological subtypes of lung cancer. Cancer Biomark 2014; 14: 457-467.

88. Mikami S, Mizuno R, Kosaka T, et al. Expression of TNF- and CD44 is implicated in poor prognosis, cancer cell invasion, metastasis and resistance to the sunitinib treatment in clear cell renal cell carcinomas. Int J Cancer 2015; 136: 1504-1514.

89. Cao D, Maitra A, Saavedra JA, et al. Expression of novel markers of pancreatic ductal adenocarcinoma in pancreatic nonductal neoplasms: additional evidence of different genetic pathways. Mod Pathol 2005; 18: 752-761.

90. Webb MA, Chen JJ, Illouz SC, et al. The impact of potential islet precursor cells on islet autotransplantation outcomes. Cell Transplant 2013; 22: 1041-1051.

91. Vortmeyer AO, Huang S, Lubensky I, et al. Non-islet origin of pancreatic islet cell tumors. J Clin Endocrinol Metab 2004; 89: 1934-1938.

92. Swatek J, Chibowski D. Endocrine cells in colorectal carcinomas. Immunohistochemical study. Pol J Pathol 2000; 51: 127-136.

93. Kadota Y, Shinoda M, Tanabe M, et al. Concomitant pancreatic endocrine neoplasm and intraductal papillary mucinous neoplasm: A case report and literature review. World J Surg Oncol 2013; 11: 75.

94. Tewari N, Zaitoun AM, Lindsay D, et al. Three cases of concomitant intraductal papillary mucinous neoplasm and pancreatic neuroendocrine tumour. JOP 2013; 14: 423-427.

95. Cai YC, Banner B, Glickman J, et al. Cytokeratin 7 and 20 and thyroid transcription factor 1 can help distinguish pulmonary from gastrointestinal carcinoid and pancreatic endocrine tumors. Hum Pathol 2001; 32: 1087-1093.

96. Knosel T, Chen Y, Altendorf-Hofmann A, et al. High KIT and PDGFRA are associated with shorter patients survival in gastroenteropancreatic neuroendocrine tumors, but mutations are a rare event. J Cancer Res Clin Oncol 2012; 138: 397-403.

97. Graham RP, Shrestha B, Caron BL, et al. Islet-1 is a sensitive but not entirely specific marker for pancreatic neuroendocrine neoplasms and their metastases. Am J Surg Pathol 2013; 37: 399-405.

98. Denby KS, Briones AJ, Bourne PA, et al. IMP3, NESP55, TTF-1 and CDX2 serve as an immunohistochemical panel in the distinction among small-cell carcinoma, gastrointestinal carcinoid, and pancreatic endocrine tumor metastasized to the liver. Appl Immunohistochem Mol Morphol 2012; 20: 573-579.

99. Masai K, Tsuta K, Kawago M, et al. Expression of squamous cell carcinoma markers and adenocarcinoma markers in primary pulmonary neuroendocrine carcinomas. Appl Immunohistochem Mol Morphol 2013; 21: 292-297.

100. Serrano MF, El-Mofty SK, Gnepp DR, et al. Utility of high molecular weight cytokeratins, but not p63, in the differential diagnosis of neuroendocrine and basaloid carcinomas of the head and neck. Hum Pathol 2008; 39: 591-598.

101. Eriksson JE, Dechat T, Grin B, et al. Introducing intermediate filaments: From discovery to disease. J Clin Invest 2009; 119: 1763-1771.

102. Thiery JP, Acloque H, Huang RY, et al. Epithelial-mesenchymal transitions in development and disease. Cell 2009; 139: 871-890.

103. Kimura N, Sasano N, Namiki T, et al. Coexpression of cytokeratin, neurofilament and vimentin in carcinoid tumors. Virchows Arch A Pathol Anat Histopathol 1989; 415: 69-77.

104. Terada T. Malignant glomus tumor of the foot with obvious neuroendocrine differentiation: A case report with immunohistochemical studies. Pol J Pathol 2014; 65: 241-247.

105. Wang ZB, Yuan J, Shi HY. Features of gastric glomus tumor: A clinicopathologic, immunohistochemical and molecular retrospective study. Int J Clin Exp Pathol 2014; 7: 1438-1448.

106. Bombardieri R, Moavero R, Roberto D, et al. Pancreatic neuroendocrine tumor in a child with a tuberous sclerosis complex 2 (TSC2) mutation. Endocr Pract 2013; 19: e124-e128.

107. Hansel DE, Epstein JI, Berbescu E, et al. Renal carcinoid tumor: A clinicopathologic study of 21 cases. Am J Surg Pathol 2007; 31: 1539-1544.

108. Tan G, Cioc AM, Perez-Montiel D, et al. Microvascular density does not correlate with histopathology and outcome in neuroendocrine tumors of the pancreas. Appl Immunohistochem Mol Morphol 2004; 12: 31-35.

109. Rodallec M, Vilgrain V, Couvelard A, et al. Endocrine pancreatic tumours and helical CT: Contrast enhancement is correlated with microvascular density, histoprognostic factors and survival. Pancreatology 2006; 77-85.

110. Kubota M, Murakami T, Nagano H, et al. Xenon-inhalation computed tomography for noninvasive quantitative measurement of tissue blood flow in pancreatic tumor. Dig Dis Sci 2012; 57: 801-805.

111. Deschamps L, Bacha D, Rebours V, et al. The expression of the hypoxia markers CA9 and CXCR4 is correlated with survival in patients with neuroendocrine tumours of the ileum. Neuroendocrinology 2012; 95: 214-222.

112. Connolly R, Gates D, Loh N, et al. Cox-2 promotes chromogranin A expression and bioactivity: evidence for a prostaglandin E2-dependent mechanism and the involvement of a proximal cyclic adenosine 5’-monophosphate-responsive element. Endocrinology 2007; 148: 4310-4317.

113. Popovtzer A, Morgenstein S, Roizman P, et al. Cyclooxygenase-2 expression in medullary thyroid carcinoma. Head Neck 2007; 29: 559-563.

114. Saffar H, Sanii S, Heshmat R, et al. Expression of galectin-3, nm-23, and cyclooxygenase-2 could potentially discriminate between benign and malignant pheochromocytoma. Am J Clin Pathol 2011; 135: 454-460.

115. Ince AT, Ovunc O, Celikel CA, et al. Cyclooxygenase-2 expression in gastroenteropancreatic neuroendocrine tumors. Hepatogastroenterology 2007; 54: 1125-1129.

116. Kim HS, Lee HS, Kim WH. Clinical significance of protein expression of cyclooxygenase-2 and somatostatin receptors in gastroenteropancreatic neuroendocrine tumors. Cancer Res Treat 2011; 43: 181-188.

117. Ohike N, Morohoshi T. Immunohistochemical analysis of cyclooxygenase (COX)-2 expression in pancreatic endocrine tumors: association with tumor progression and proliferation. Pathol Int 2001; 51: 770-777.

118. Cadden I, Johnston BT, Turner G, et al. An evaluation of cyclooxygenase-2 as a prognostic biomarker in mid-gut carcinoid tumours. Neuroendocrinology 2007; 86: 104-111.

119. Zhang Q, Meng X, Zheng G, et al. Antitumor activity of celecoxib, a selective cyclooxygenase-2 inhibitor, in medullary thyroid carcinoma. Mol Med Rep 2014; 9: 768-772.

120. Vivaldi A, Ciampi R, Tacito A, et al. Celecoxib, a cyclooxygenase-2 inhibitor, potentiates the chemotherapic effect of vinorelbine in the medullary thyroid cancer TT cell line. Mol Cell Endocrinol 2012; 355: 41-48.

121. Lewitowicz P, Koziel D, Gluszek S, et al. Is there a place for practical chemoprevention of colorectal cancer in light of COX-2 heterogeneity? Pol J Pathol 2014; 65: 276-282.

122. Balgley BM, Guo T, Zhao K, et al. Evaluation of archival time on shotgun proteomics of formalin-fixed and paraffin-embedded tissues. J Proteome Res 2009; 8: 917-925.