Immunohistochemical analysis of E-cadherin and zeste homolog 2 expression in endoscopic ultrasound-guided fine-needle aspiration of pancreatic adenocarcinoma

Authors


  • Presented in part at the 101st Annual Meeting of the United States and Canadian Academy of Pathology; March 17-23, 2012; Vancouver, British Columbia, Canada.

Abstract

BACKGROUND

Recent studies have demonstrated that partial or complete loss of E-cadherin (EC) and nuclear accumulation of zeste homolog 2 (EZH2) are hallmarks of poorly differentiated pancreatic adenocarcinoma (PAC). Depletion of EZH2 sensitizes cancer cells to chemotherapy in vitro. The objective of this study was to determine EC and EZH2 expression by immunohistochemistry (IHC) in samples obtained by endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) as potential biomarkers for treatment and disease prognosis.

METHODS

Thirty-eight EUS-FNA samples from patients with a PAC diagnosis were analyzed by IHC for EC and EZH2 expression. Seven corresponding surgical resection specimens were included in the study. The intensity of EZH2 and EC expression in PAC and in normal gastrointestinal pick-ups (internal positive control) was scored by using a 4-tier grading system.

RESULTS

EC demonstrated a focal, weak-to-moderate decrease in 24 PAC samples. Complete loss of EC expression was observed in the poorly differentiated areas represented by single tumor cells. The average staining intensity of EC in samples of poorly differentiated PAC was significantly lower than that of moderately differentiated PAC samples (P = .0005). EZH2 was variably positive in 31 of 38 PAC samples. The average staining intensity of EZH2 in moderately and poorly differentiated PACs did not differ significantly (P = .81).

CONCLUSIONS

EC and EZH2 expression was determined reliably by IHC on paraffin sections of EUS-FNA cell block specimens. The current results were consistent with prior reports indicating a decrease or loss of EC expression in poorly differentiated PAC. However, EZH2 expression did not always correlate inversely with EC expression and was more heterogeneous. Cancer (Cancer Cytopathol) 2013;121:644–52. © 2013 American Cancer Society.

INTRODUCTION

Pancreatic cancer is the fourth deadliest cancer in the United States and affects men and women equally. There were 43,920 estimated new cases of pancreatic cancer with 37,390 estimated deaths in the United States in 2012.[1] Many patients have advanced disease at the time of diagnosis, which makes surgical and medical interventions largely ineffective. According to the American Cancer Society, less than 20% of patients are surgical candidates at the time of diagnosis, and the 5-year relative survival rate is 6%.[1] More than half of patients are diagnosed at a late stage, for which the 5-year survival rate is only 2%. Current treatment options, including surgery, radiation therapy, and chemotherapy, may extend survival but are almost never curative. The lack of early detection markers and limited efficacy of existing treatment regimens are contributing factors to the poor outcome of pancreatic adenocarcinoma (PAC). Despite extensive research on the pathogenesis of PAC and the search for diagnostic and prognostic markers in the past years, the prognosis is still very poor, and the treatment is ineffective for the majority of patients with PAC, especially for those with unresectable disease.

Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) biopsy has become a valuable tool in the diagnosis of PAC[2] by obtaining diagnostic material that may be the only available tissue in inoperable patients. The EUS-FNA samples are not only important in the diagnosis of PAC but also may be useful for further studies on prognostic and predictive biomarkers in unresectable PAC, which could provide important information for the treatment of patients with PAC. However, to our knowledge, no prognostic or predictive biomarkers have been studied or reported on EUS-FNA cell blocks from patients with PAC.

It has been demonstrated that expression levels of multiple molecules correlate with a poor prognosis or resistance to chemotherapy.[3] One of the most recent candidate predictors for PAC outcome is E-cadherin (EC) expression. EC is an important cell adhesion molecule that is encoded by a tumor suppressor gene, CDH1.[4] It has been demonstrated that partial loss of tumor EC expression is an independent predictor of a poor outcome in PAC[5, 6] and is related to resistance to antivascular endothelial growth factor therapy[7, 8] and chemotherapy[9] through epithelial-mesenchymal transformation (EMT). EMT is a pathologic feature in neoplasia and fibrosis, and it is characterized by loss of cell adhesion, repression of EC expression, and increased cell motility. An EMT phenotype reportedly is responsible for the increased aggressiveness of drug-resistant PAC.[7-9] Another prognostic predictor candidate is enhancer of zeste homolog 2 (EZH2), a histone methyltransferase-containing protein in the Polycomb-group (PcG) family. EZH2 acts mainly as a gene silencer by the addition of 3 methyl groups to lysine 27 of histone 3.[10] EZH2 down-regulates tumor suppressors, a modification that leads to chromatin condensation and transcriptional repression. Recent studies have suggested that increased EZH2 protein expression is associated with many cancers,[10] including pancreatic cancer. Nuclear accumulation of EZH2 mediates increased invasiveness and metastasis by silencing downstream targets, including EC, a hallmark of poorly differentiated PAC.[11] It has been demonstrated that the depletion of EZH2 sensitizes cancer cells to chemotherapy in vitro.[12] In addition, it has been reported that an activated point mutation, KRAS mutation at codon 12 (K-Ras[G12D]) in PAC promotes the development of PAC, and knockdown of the K-Ras(G12D) allele leads to a significant increase in the expression of EC both in vitro and in vivo.[13] It has become evident that down-regulation of EC and up-regulation of EZH2 play crucial roles not only in the prognosis but also in the treatment of PAC. The objective of this study was to determine EC and EZH2 expression in PAC using immunohistochemistry (IHC) in EUS-FNA specimens to provide biomarker information about prognostic and potential treatment options for patients with PAC.

MATERIALS AND METHODS

Case Selection

After obtaining the Institutional Review Board approval, 38 cases with a diagnosis of PAC on EUS-FNA samples, which contained diagnostic tumor cells on the cell blocks, were selected from archived material in the Cytopathology Section at the University of Chicago Hospital accessioned from January 2009 to August 2011. All diagnoses were confirmed by at least 2 board-certified cytopathologists. The FNA samples were fixed with CytoLyt solution (Cytyc Corporation, Boxborough, Mass) at least 30 minutes before processing and then embedded in paraffin. The cell block sections were stained with hematoxylin and eosin (H&E). The selected cases were reviewed by a board-certified cytopathologist (W.R.) and a cytopathology fellow (L.G.). The electronic files of the University of Chicago were searched for corresponding histologic diagnoses of PAC as well as patient age, sex, and tumor location. Sixteen cases had histologic correlation, which included 9 core biopsies and 7 resections. The remaining 22 cases did not have corresponding histopathology. Six of the 7 surgical resection cases were available from the archived material in the Surgical Pathology Department. Those cases were reviewed, and 1 representative paraffin-embedded block from each case was selected for IHC study. Table 1 summarizes the clinical data.

Table 1. Clinical Profile and Pancreatic Tumor Location
Age Group, yNo. of PatientsTotal No. of Patients (%)
SexTumor Location: Women + Men
WomenMenHeadBodyTail
40-49111 + 1  2 (5)
50-59342 + 11 + 3 7 (18)
60-69675 + 62 + 00 + 114 (37)
70-79763 + 32 + 11 + 212 (32)
80-89202 + 0  2 (5)
>90101 + 0  1 (3)
Total no. of patients201814 + 115+41 + 338 (100)

Cytologic Grading

The cell block sections (H&E) of PAC were graded using the College of American Pathologists (CAP) Cancer Protocol for the Examination of Specimens from Patients with Carcinoma of Exocrine Pancreas (updated in June 2012). Briefly, the grading is based on the extent of glandular differentiation: grade 1, well differentiated (>95% of tumor composed of glands); grade 2, moderately differentiated (50%-95% of tumor composed of glands); and grade 3, poorly differentiated (<50% of tumor composed of glands).

Immunohistochemistry

IHC for EC and EZH2 was performed on 4-micron-thick paraffin tissue sections using mouse monoclonal antibody anti-EC (clone 4A2C7; Invitrogen, Frederick, Md) at 1:100 dilution and polyclonal rabbit anti-EZH2 (Invitrogen) at 1:250 dilution. IHC was performed on a Leica Bond-Max automated IHC/ISH system (Leica Microsystems Inc., Buffalo Grove, Ill), according to the modified manufacturer's protocol, using Bond Polymer Refine detection (3,3′ diaminobenzidine [DAB]) system (Leica Microsystems). Briefly, after heat antigen retrieval in a high pH buffer (ER2) for 15 minutes and peroxidase block for 5 minutes, sections were incubated with anti-EC antibody for 25 minutes or with EZH2 for 50 minutes, followed by a postprimary step for 15 minutes, and Bond polymer horseradish peroxidase (HRP) for 25 minutes. The peroxidase reaction was developed using the DAB provided in the kit followed by hematoxylin for 5 minutes. Slides were dehydrated in alcohols and mounted in mounting medium (Sakura Finetek, USA, Torrance, CA). The same IHC method was used on formalin-fixed, paraffin-embedded sections from surgical resection tissue specimens, except that anti-EZH2 was used at 1:500 dilution instead of 1:250 dilution.

Double staining on randomly selected FNA cases (n = 12) and resection cases (n = 6) was performed on paraffin-embedded sections. After antigen retrieval, anti-EZH2 antibody (1:500 dilution) was applied on tissue sections for a 1-hour incubation at room temperature in a humidified chamber. Antigen-antibody binding was detected with the Dako EnVision plus HRP system (K4003 anti-rabbit; Dako, Carpinteria, Calif) and the DAB + chromogen (K3468) system (Dako). The stained cell block sections were then incubated with anti-EC antibody (1:100 dilution) for 1 hour at room temperature. Secondary antigen-antibody reaction was observed with the MACH3 antimouse AP-polymer detection kit (M3M532H; Biocare Medical Inc., Concord, Calif) and Vulcan Fast Red Chromogen Kit 2 (FR805S; Biocare Medical Inc.). Hematoxylin was used for counterstaining.

Gastrointestinal (GI) pick-ups on the same cell block section served as an internal control. The intensity of EZH2 (nuclear staining) and EC expression (membranous staining) in PAC and normal GI pick-ups was scored by a 4-tier grading system, from negative (0) to strongly positive (4). Because staining intensity varied from area to area, a weighted average was calculated and reported as negative (0), minimal (<1.0), weakly positive (1.0-1.9), moderately positive (2.0-2.9) and strongly positive (3.0-4.0). For calculation purposes, a middle value was used if the staining intensity was between 2 scores. For example, if an intensity was between 0 and 1.0, then a score of 0.5 was recorded; if it was between 1.0 and 1.9, then a score of 1.5 was recorded. For histology specimens, only overall staining intensity was recorded, and no calculation was used. All IHC slides also were reviewed by a second board-certified pathologist.

Statistical Analysis

The average intensity (score) was calculated based on the sum of percentage of each score. The calculation of a P value was performed in Microsoft Excel using unpaired t test function. A P value < .05 was considered significant.

RESULTS

On the basis of the CAP cancer protocol cancer grading system, 11 PACs were moderately differentiated, and 27 were poorly differentiated. The photomicrographs in Figure 1 reveal areas of moderately differentiated PAC (Fig. 1A,B) and poorly differentiated PAC (Fig. 1C,D) on cell blocks sections (H&E stain). Six cases in our study contained extensive necrotic areas, in which tumor cells were present as single cells in a background of necrosis (Fig. 1D). To produce reliable data, the areas with extensive necrosis in these cases were excluded from calculation.

Figure 1.

These are photomicrographs of fine-needle aspiration cell blocks from diagnostic pancreatic adenocarcinoma (hematoxylin and eosin staining). Staining reveals areas with (A) moderately differentiated adenocarcinoma, (B) moderately differentiated adenocarcinoma, (C) poorly differentiated adenocarcinoma, and (D) poorly differentiated adenocarcinoma in a background of necrosis.

Normal epithelium (internal positive controls), which was present in the majority of cases (33 of 38 specimens), demonstrated strong (4.0), membranous EC expression. In the 38 PAC cases, using single IHC-stained slides, 14 PACs revealed strong, membranous EC expression (3.0-4.0), 10 had moderate EC expression (approximate range, 2.0-2.9), 9 had weak EC expression (1.0-1.9), and 5 had minimal EC expression (<1.0) (Table 2). In all, 28 of 38 PACs revealed focal loss of EC expression. Examples of a membranous EC staining pattern are provided in Figure 2A,B.

Table 2. E-Cadherin and Zeste Homolog 2 Status in 38 Patients With Pancreatic Adenocarcinoma
IHC Staining IntensityNo. of Patients
E-Cadherin Expression (Membranous Staining)aEZH2 Expression (Nuclear Staining)a
  1. Abbreviations: EZH2, zeste homolog 2; IHC, immunohistochemistry.

  2. a

    The average staining intensity on each slide was recorded and then calculated.

Strong (3.0-4.0)148
Moderate (2.0-2.9)1012
Weak (1.0-1.9)98
Minimal (<1.0)58
Negative02
Figure 2.

Single immunohistochemical staining reveals different staining patterns for E-cadherin and zeste homolog 2 (EZH2) on fine-needle aspiration cell block sections. Stains reveal (A) strong membranous staining for E-cadherin; (B) strong (well differentiated pancreatic adenocarcinoma [PAC]) and weak (poorly differentiated PAC), membranous staining for E-cadherin; (C) strong nuclear staining for EZH2; and (D) focal, moderate-to-strong nuclear staining for EZH2.

FNA GI pick-ups (internal positive control) revealed diffuse, moderate nuclear reactivity for EZH2, and normal pancreatic ducts or bile ducts had negative EZH2 nuclear staining. The average nuclear expression of EZH2 on cell block sections was strong in 8 cases, moderate in 12 cases, weak in 8 cases, minimal in 8 cases, and negative in 2 cases (Table 2). Figure 2C,D provides examples of a nuclear EZH2 staining pattern. The individual average staining intensity of EC and EZH2 in 38 PAC cases is listed in Table 3, in which EC expression is scored from negative (0) to strong (4.0), and EZH2 expression is listed for the corresponding EC value.

Table 3. Individual E-Cadherin and Zeste Homolog 2 Immunohistochemistry Staining (Average Intensity) in Patients With Pancreatic Adenocarcinomaa
Patient No.Mean IHC Staining IntensitybPatient No.Mean IHC Staining IntensitybEZH2
E-CadherinEZH2E-Cadherin
  1. Abbreviations: EZH2, zeste homolog 2; IHC, immunohistochemistry.

  2. a

    Patients 1 through 11 had moderately differentiated pancreatic adenocarcinoma, and patients 12 through 38 had poorly differentiated pancreatic adenocarcinoma.

  3. b

    Staining intensity was scored as negative (0), minimal (<1.0), weakly positive (1.0-1.9), moderately positive (2.0-2.9) and strongly positive (3.0-4.0).

11.83.5201.50.0
22.00.5211.73.0
32.20.5221.80.0
42.52.1231.81.0
52.73.8241.82.6
63.42.0252.23.0
73.60.6262.41.5
83.91.2272.42.2
94.00.7282.50.4
104.02.8292.50.7
114.02.3302.51.3
120.22.9313.02.3
130.40.5323.02.5
140.51.0333.04.0
150.81.3343.51.5
160.82.1353.52.5
171.10.5364.02.0
181.23.5374.03.3
191.41.0384.04.0

Table 4 demonstrates that the EC intensity was significantly weaker in poorly differentiated PAC than in moderately differentiated PAC (P = .0005). EZH2 intensity did not differ significantly between moderately and poorly differentiated PACs.

Table 4. Average Immunohistochemistry Staining Intensity of E-Cadherin and Zeste Homolog 2 in 38 Patients With Pancreatic Adenocarcinoma
DifferentiationTotal No. of PatientsIHC Staining Intensitya
ECbSD1EZH2cSD2
  1. Abbreviations: EC, E-cadherin; EZH2, zeste homolog 2; IHC, immunohistochemistry; SD1, 1 standard deviation; SD2, 2 standard deviations.

  2. a

    Staining intensity was scored as negative (0), minimal (<1.0), weakly positive (1.0-1.9), moderately positive (2.0-2.9) and strongly positive (3.0-4.0).

  3. b

    P = .004.

  4. c

    P = .46.

Moderately differentiated113.10.91.81.2
Poorly differentiated272.11.11.91.2

Double IHC staining revealed membranous EC staining (red) and nuclear EZH2 staining (brown) in a moderately differentiated PAC (Fig. 3A) and a poorly differentiated case with a necrotic background (Fig. 3B).

Figure 3.

Double immunohistochemical staining reveals an inverse correlation between E-cadherin and zeste homolog 2 (EZH2) expression on fine-needle aspiration cell blocks for E-cadherin (red) and EZH2 (brown) in pancreatic adenocarcinoma. (A) Moderately differentiated adenocarcinoma is observed with focal, strong, nuclear staining (EZH2) and focal, moderate, membranous staining (E-cadherin). Focal loss of E-cadherin staining is noted. (B) Poorly differentiated adenocarcinoma is observed in a background of necrosis with strong nuclear staining (EZH2) and diminished or lost membranous staining (E-cadherin).

In our study, 7 of 38 cytology cases (18%) had surgical resection specimens with histologic grading of the tumor, and 9 (24%) had core biopsies. Table 5 summarizes patient age, sex, and tumor location as well as the relation among the histologic grade of resection specimens, the grade determined from FNA cell blocks, and the expression of EC and EZH2. We noted focal poor fixation and heterogeneous expression of EC (n = 3) or EZH2 (n = 2) in some resection specimens. Figure 4 reveals morphology and IHC staining for EC and EZH2 in 2 resection specimens (patients 4 and 7) and their corresponding FNA cell blocks.

Table 5. Morphologic Correlation Between Surgical Resection and FNA Cell Blocks With Zeste Homolog 2/E-Cadherin Expression
Patient No.SexAge, yTumor LocationResection Stage/GradeFNA GradeE-Cadherin: Resection/FNAEZH2: Resection/FNA
  1. Abbreviations: FNA, fine needle aspiration; EZH2, zeste homolog; MD, moderately differentiated; PD, poorly differentiated; pT, pathologic tumor classification; ypT, post-treatment pathologic tumor classification; Min, minimal immunohistochemistry (IHC) staining intensity; W, weak IHC staining intensity; Mod, moderate IHC staining intensity; S, strong IHC staining intensity.

1Man46Pancreatic headMD: pT3/grade 2PDMod (variable)/MinMod/W
2Man69Common bile duct in pancreasPD: pT3/grade 3PDW+Mod/Mod, focal SMod+S/Mod
3Man71Ampulla and head of pancreasPD: pT4/grade 3PDMod (variable)/Mod+SMod, focal S/W, focal S
4Man75Pancreatic tailPD: pT3/grade 3PDMod/Mod, focal MinS+Mod/Min+Mod
5Man61Pancreatic headPD: ypT3/grade 3PDMod (variable)/SS+Mod/Min+Mod
6Man52Pancreatic bodyPD: pT3/grade 3MD(NA)/W(NA)/Mod
7Woman63Pancreatic headPD: pT3/grade 3MDMod+S/SS/Mod
Figure 4.

Morphology and immunohistochemical (IHC) staining for E-cadherin and zeste homolog 2 (EZH2) is shown on 2 paraffin-embedded tissue sections from resection specimens with corresponding fine-needle aspiration (FNA) cell blocks. (A-F) These hematoxylin and eosin (H&E)-stained sections are from patient 4, who was diagnosed with poorly differentiated pancreatic adenocarcinoma (PAC) on both (A) histology and (D) cytology. IHC reveals focally moderate, membranous staining for E-cadherin on both (B) histology and (E) cytology; (C) strong nuclear staining for EZH2 on histology; and (F) focally moderate staining on cytology. (G-L) These H&E-stained sections are from patient 7, who was diagnosed with (G) poorly differentiated PAC on histology and with (J) moderately differentiated PAC on cytology. IHC reveals (H) strong, moderate, membranous staining for E-cadherin on histology; (K) strong, membranous staining on cytology; and moderate nuclear staining for EZH2 on both (I) histology and (L) cytology.

DISCUSSION

Although the diagnosis of PAC on FNA biopsy is usually straightforward, it is increasingly important to go beyond simple morphologic diagnosis to provide additional information on prognosis and treatment of patients with PAC. This is particularly important for inoperable patients, for whom FNA samples are the only pathologic material available. In our study, 7 of 38 patients (18%) with PAC underwent surgical resection, 9 (24%) underwent core biopsy, and 22 (58%) had only EUS-FNA specimens available. There were 3 discrepancies between cytologic and histologic grading of the 7 PAC resection specimens (1 cytologic “over call” and 2 “under calls”), as indicated in Table 5. Discrepancies may have been caused by differences in fixation, sampling variations, or the time elapsed between FNA and resection, as well as possible treatment effects on the resection specimen.

Recent studies have suggested that EC and EZH2 are 2 important biomarkers for PAC. Partial loss of tumor EC expression is an independent predictor of poor outcome in patients with PAC,[5, 6] which is related to resistance to treatment with antivascular endothelial growth factor[7, 8] and resistance to chemotherapy with gemcitabine in vitro.[9] Nuclear accumulation of EZH2 mediates increased invasiveness and metastasis by silencing downstream targets, including EC.[11] A recent in vitro study demonstrated that EZH2 inhibitor (DZNeP) synergistically interacts with gemcitabine, which restores EC expression, inhibits cell migration, and promotes apoptosis.[14] That study indicated that reducing EZH2 expression and restoring EC expression could sensitize PAC to chemotherapy. The combination of EZH2 inhibitor and chemotherapy may provide a new treatment regimen for patients with PAC. Therefore, an analysis of EC and EZH2 status on FNA samples may provide valuable information that could be used for predicting the effect of chemotherapy and designing therapy regimens.

Our results indicate that, in most cases, a decrease or focal loss of EC correlates inversely with the degree of PAC differentiation. The average staining intensity score in moderately differentiated cases (mean, 3.1) was higher than in poorly differentiated cases (mean, 2.1). Although there was no statistically significant difference between moderately differentiated PAC (n = 2) and poorly differentiated PAC (n = 5) in our resection specimens, the difference was statistically significant when comparing all (n = 11) moderately differentiated cases with all (n = 27) poorly differentiated cases (P = .004) (Table 4). The decreased EC expression was associated with increased EZH2 expression in most cases. However, in a few cases, this correlation was not observed. A possible explanation for this lack of correlation may be that EC is not the only downstream target gene of EZH2.[15, 16] The limited sampling of FNA also may have contributed to the poor correlation in some of cases. In our study, we noted that tumor necrosis had some impact on negative or minimal staining cases; however, in some necrotic areas, moderate to strong EZH2 nuclear staining in single tumor cell also was observed. However, the complete loss of EC was common both focally in the necrotic areas and in single tumor cells (Fig. 3B). The combination of EZH2 and EC immunostains, especially double stains on the same slide, may provide information that is more accurate than a single IHC stain on those necrotic tumors. In our study, extensively necrotic areas in 6 cases were excluded from calculation.

Unexpectedly, 5 PACs had no or minimal expression of EZH2, whereas EC expression was markedly decreased or lost. The reason for this lack of correlation is unknown, although sampling, tumor necrosis, and technical issues cannot be excluded, as stated above. EZH2 mutations may result in a truncated protein that cannot be recognized by IHC staining; however, EZH2 mutations have not yet been reported in PAC. Furthermore, the results published in the literature were based on resectable PAC, not unresectable PAC. There were 31 patients with unresectable PAC in our study; therefore, it was difficult to correlate the results from our studies with those published based on resectable tumors. Indeed, none of our 7 patients who had resectable PAC had FNA specimens that revealed the loss of both EC and EZH2. Five resection specimens had IHC staining intensity that corresponded closely to the FNA specimen, although a variable degree of discrepancies was noted. Because we only had 6 resection specimens in this study, it is difficult to draw a conclusion. It is worth mentioning that, in addition to tumor heterogeneity, different fixatives and fixation time may have affected the intensity of nuclear staining. All of our FNA specimens were fixed in CytoLyt solution immediately after they were obtained, but our resection specimens were fixed in formalin, and the published data also were based on formalin-fixed specimens.

One recent study demonstrated that loss of EZH2 results in impaired pancreatic regeneration and accelerates K-Ras (G12D)-driven neoplasia, pancreatic intraepithelial neoplasia, in animal models.[17] Those authors suggested that increased EZH2 expression in pancreatitis provided a mechanism of protection against progression to cancerous lesion, and loss of EZH2 in chronic pancreatitis initiates tumorigenesis.[17, 18] In a human study, however, only 33% of patients with chronic pancreatitis had some degree of EZH2 expression.[11] Because the pancreatitis in the animal model was induced by cerulean, it may be different from chronic pancreatitis in humans. Also, the neoplasia in the animal model[17] was pancreatic intraepithelial neoplasia, and not PAC, so it is possible that the level of EZH2 expression would be increased when the animals developed PAC. It is noteworthy that EZH2 expression is detected not only in many cancer cells but also in some normal epithelial cells, such as GI epithelium. However, normal pancreatic or bile duct epithelium is negative for EZH2 expression on IHC.[12] One study indicated that increased EZH2 expression was detected in 40% of patients who had intraductal papillary mucinous neoplasms with moderate dysplasia, in 66% of patients who had high-grade dysplasia, but in no patients who had mild dysplasia.[11] The intensity of EZH2 expression in GI pick-ups in our study was scored as 2.0 to 3.0. Morphologically, it is not difficult to distinguish normal GI pick-ups from pancreatic adenocarcinoma, but it is difficult to differentiate intraductal papillary mucinous neoplasm with severe dysplasia from adenocarcinoma on cell block sections, although the clinical history and imaging studies could provide helpful information.

Because EC expression was not always inversely correlated with the level of EZH2 expression in our study, it seems necessary to determine the expression of both EC and EZH2 for each case when the FNA samples are available. Combined EC and EZH2 status may be important in making treatment decisions, especially for predicting resistance to chemotherapy or therapy with antivascular endothelial growth factor. A combination of EZH2 inhibitors and chemotherapy in the near future could benefit those patients who have chemotherapy resistance because of increased EZH2 expression.[14]

In conclusion, our current results indicate that EC and EZH2 expression can be determined reliably by IHC in EUS-FNA cell block specimens in selected cases with viable tumor cells. We have demonstrated that the degree of differentiation of PAC in EUS-FNA cytology specimens does not always correlate with the histologic grade in the resection specimen. The degree of differentiation in PAC correlates with the EC expression level in most cases, but not with the EZH2 expression level. Therefore, differential analysis of EZH2 and EC expression may be helpful in prognostic stratification and treatment decisions in patients with PAC. Further studies are necessary to better understand the role of EZH2 in PAC progression and it correlation with EC expression.

FUNDING SUPPORT

This study was supported by the Department of Pathology, University of Chicago.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary