Main

Cystic tumors of the pancreas are relatively uncommon, accounting for less than 5% of pancreas exocrine tumors (1, 2, 3, 4, 5). Benign and borderline mucinous cystic neoplasms (ICD-O 8470/0 and ICD-O 8470/1) and serous microcystic adenomas (ICD-O 8441/0) account for most of the cystic tumors (2, 3, 4). These two tumors have differences in progression to malignancy and clinicopathological features (6). Mucinous cystic neoplasms are classified as benign, borderline (uncertain malignant potential), carcinoma in situ, and invasive carcinoma based on the severity of dysplasia of the lining epithelium or presence of stromal invasion (2, 3, 4, 7). In contrast, almost all serous microcystic adenomas are benign and rarely progress to malignancy (4). Distinguishing between these tumor types is clinically challenging but important because mucinous cystic neoplasms are more prone to recurrence if incompletely excised and more likely to become malignant.

There are differences in the genetic alterations in these cystic tumors. Activating mutations of K-ras are reported in both benign and malignant mucinous cystic neoplasms (8, 9, 10, 11, 12), but mutations of K-ras proto-oncogene and alterations of p53, p16 and DPC4 tumor suppressor genes are more frequently present in invasive mucinous cystic neoplasms (9, 11, 12, 13), suggesting stepwise increase in genetic alterations in malignant tumors. In contrast, these genetic alterations are not present in serous microcystic adenomas (8, 9, 10, 11, 14), which can occur as a manifestation of von-Hippel-Lindau disease due to germline mutations of the VHL gene (15, 16, 17). In addition, sporadic serous microcystic adenomas frequently have allelic loss of chromosome 3p25 and infrequently, somatic VHL gene mutations (15, 18).

Other molecular abnormalities have been reported in various pancreatic neoplasms. Aberrant methylation of CpG islands is a molecular mechanism that leads to silencing of tumor suppressor genes such as p16 and p14 in a variety of cancers and precursor lesions (19, 20, 21, 22). Recently, methylation of the ppENK gene encoding met-enkephalin was found to be frequent in ductal adenocarcinoma (23) and intraductal papillary-mucinous neoplasm of pancreas (24). Mutations in exon 3 of the β-catenin gene, involving the sites phosphorylated by GSK-3 β for ubiquitination and degradation of β-catenin, stabilize β-catenin protein and in turn up-regulate the Tcf/Lef-1 family of transcription factors. β-catenin mutations have also been reported in diverse types of pancreatic carcinomas including pancreatoblastomas, pancreatic acinar cell carcinomas, and solid pseudopapillary carcinomas (25, 26, 27, 28). Microsatellite instability due to alterations of mismatch repair genes such as human Mut L homologue 1 (hMLH1) and human Mut S homologue 2 (hMSH2) is present in medullary carcinomas of the pancreas (28, 29).

Methylation, β-catenin gene mutations, or microsatellite instability have not been studied in either mucinous cystic neoplasms or serous microcystic adenomas, and their roles in the pathogenesis of the two distinct pancreatic cystic tumors are unclear. In this study, therefore, we compared aberrant methylation of the p16, p14, ppENK, and VHL gene and genetic alterations including K-ras and β-catenin mutations, loss of chromosome 3p25, and microsatellite instability in mucinous cystic neoplasms and sporadic serous microcystic adenomas.

MATERIALS AND METHODS

Patients and Specimens

Fifteen pancreatic mucinous cystic neoplasms (ICD-O 8470/0 or ICD-O 8470/1) and 16 sporadic serous microcystic adenomas (ICD-O 8441/0) were identified from the surgical pathology files of the University of Texas M. D. Anderson Cancer Center (11 mucinous cystic neoplasms and 14 serous microcystic adenomas) and Kyung Pook National University Hospital, South Korea (4 mucinous cystic neoplasms and 2 serous microcystic adenomas) from patients undergoing resection between 1991 and 2001. Formalin-fixed, paraffin-embedded tumor and non-neoplastic tissue specimens were collected. The histopathology of the tumors was reviewed (Fig. 1), and mucinous cystic neoplasms were further classified as benign (ICD-O 8470/0) and borderline (uncertain malignant potential, ICD-O 8470/1), as previously described (2, 3, 4). No mucinous cystic neoplasms with in situ carcinoma or invasive carcinoma were included in this study. Matching non-neoplastic pancreatic tissue was available from 13 patients with mucinous cystic neoplasms and 15 patients with serous microcystic adenomas. The study was approved by the Surveillance Committee (institutional review board) of the University of Texas M. D. Anderson Cancer Center.

FIGURE 1
figure 1

Histology of benign mucinous cystic neoplasm (A–B), borderline mucinous cystic neoplasm (C–D), and serous microcystic adenoma (E–F).

DNA Extraction

Histologic sections were microdissected to obtain >70% neoplastic cellularity. DNA was extracted from microdissected tumors and non-neoplastic tissues as described previously (30).

Bisulfite Treatment of DNA and Methylation-Specific PCR

Bisulfite treatment of DNA for methylation assays was carried out as described previously (31, 32). Methylation of the p16, p14, ppENK, and VHL gene was determined by MSP as previously described (19, 21, 23, 33). Briefly, 2 μL of bisulfite-treated DNA was used as template for PCR reactions using primers specific for methylated and unmethylated alleles (Table 1). PCR products from methylated and unmethylated reactions were electrophoresed on 6% nondenaturing acrylamide gels. The gels were stained with ethidium bromide, and the proportion of methylated allele compared with unmethylated allele was determined by densitometry. The DNA of colon cancer cell line RKO or SssI methylase-treated DNA (New England Biolabs, Beverly, MA) was used as positive control, and water, as negative control.

TABLE 1 Primers and PCR Condition for Methylation-Specific PCR for p16, p14, ppENK, and VHL Genes

Sequencing of Exon 1 of K-ras and Exon 3 of β-Catenin Genes

Exon 1 of the K-ras proto-oncogene and exon 3 of the β-catenin gene were amplified by PCR and sequenced as described previously (34, 35). DNA sequencing was performed on the PCR product and amplification primers for each exon with a commercial DNA sequencing kit, according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA).

Loss of Heterozygosity of chromosome 3p25

Loss of heterozygosity (LOH) of chromosome 3p25 was determined by use of three dinucleotide microsatellite markers near the VHL gene on chromosome 3 (D3S1110, D3S1038, and D3S1435) with PCR amplification using fluorescent dye–labeled primers (Invitrogen, Carlsbad, CA). The 5′oligonucleotide was end-labeled with 6-FAM fluorescent dye. PCR was performed in 15-μL reaction volumes containing 20 ng of DNA, 9 μL of ABI Prism True Allele PCR Premix (Applied Biosystems, Foster City, CA), and 5 pmol of each primer. PCR was performed using a GeneAmp PCR system 9700 thermal cycler (Applied Biosystems, Foster City, CA) with the following conditions: 95° C for 7 minutes; 3 cycles of 94° C for 1 minute, 58° C for 30 seconds, and 72° C for 45 seconds; 42 cycles of 93° C for 45 seconds, 54° C for 30 seconds, 72° C for 40 seconds; and final extension at 72° C for 30 minutes. A 0.25-μL aliquot of each fluorescent-labeled PCR product was combined with 12 μL of formamide and 0.5 μL of GENESCAN 400HD [ROX] size standard (Applied Biosystems). The samples were then subjected to capillary electrophoresis on an ABI 3700 DNA Analyzer using GENESCAN Analysis software (Applied Biosystems). Loss of a marker was considered to be present when the assay showed absence or decrease in intensity by >50% of one of two alleles from a tumor sample as compared with the paired control non-neoplastic tissue. Complete or partial loss of chromosome 3p25 was based on the pattern of loss of the three markers.

Microsatellite Instability and Alteration of the Transforming Growth Factor β Type II Receptor Gene

Microsatellite instability was evaluated by amplification of two mononucleotides repeats (BAT25 and BAT26) and three dinucleotide repeats (D2S123, D5S346, D17S250), as proposed by the NCI workshop (36), and alterations of the polyadenine tract of the transforming growth factor β type II receptor (TGFβRII) gene were evaluated using fluorescent dye-labeled primers as described previously (37). Allelic shift of a microsatellite marker was defined by the presence of at least one additional band in the tumor DNA as compared with in control DNA

Statistical Analysis

The differences in frequencies among clinicopathological factors and epigenetic and genetic alterations were analyzed by χ2 or Fisher’s exact test. All statistical analyses were performed using the SPSS statistical program (SPSS Inc., Chicago, IL).

RESULTS

Clinicopathological Characteristics of Patients and Tumors

The clinicopathological characteristics of the patients and their tumors are summarized in Table 2. Mucinous cystic neoplasms were located mainly in the pancreatic body (80%), whereas serous microcystic adenomas were evenly distributed among the head, body, and tail (44% in head, P = .002). Thirty-three percent of mucinous cystic neoplasms were classified as borderline with uncertain malignant potential, but all serous microcystic adenomas were classified as benign (P = .02).

TABLE 2 Clinicopathological Features of Mucinous Cystic Neoplasms and Serous Microcystic Adenomas

Aberrant Methylation of p16, p14, ppENK, and VHL Genes

Examples of methylation are shown in Figure 2. One control pancreatic tissue adjoining a serous microcystic adenoma was methylated at p14, but no other control pancreatic tissues had methylation of the p14, p16, ppENK, or VHL genes.

FIGURE 2
figure 2

Methylation analysis of p16, p14, ppENK, and VHL genes by methylation-specific PCR. PCR products using primers for unmethylated (U) and methylated (M) alleles of bisulfite-treated DNA from tumor (T) and positive and negative control samples. MCN, mucinous cystic neoplasm; SMA, serous microcystic adenoma; RKO, positive control from colon cancer cell line; M(+), positive control DNA treated with Sss1 methylase; H2O, lanes without DNA used as negative control. MCN 8T and SMA 8T are methylated at p16, MCN 14T and SMA 7T at p14, and SMA 14T at VHL genes. All other samples shown in the gels are unmethylated.

Methylation of p16 and p14 was present in 14% (2/14) and 15% (2/13), respectively, of mucinous cystic neoplasms, and in 12% (2/16) and 37% (6/16) of serous microcystic adenomas (P = not significant, Table 3). Methylation of VHL was present in 7% (1/14) of serous microcystic adenomas but was not found in mucinous cystic neoplasms (0/14). No methylation of ppENK was present in either 14 mucinous cystic neoplasms or 13 serous microcystic adenomas. The frequencies of methylation of p16, p14, and VHL were not different between mucinous cystic neoplasms and serous microcystic adenomas or between benign and borderline mucinous cystic neoplasms. The methylation of p16 and p14 was not concordant in either mucinous cystic neoplasms or serous microcystic adenomas.

TABLE 3 The Frequencies of Epigenetic and Genetic Alterations in Mucinous Cystic Neoplasms and Serous Microcystic Adenomas of Pancreas

Genetic Alterations

Mutation in exon 1 of the K-ras gene was present in 33% (5/15) of mucinous cystic neoplasms, but no mutations were found in serous microcystic adenomas (0/16, P = .004, Fig. 3A, Table 3). Four K-ras mutations were present in codon 12 and one in codon 13. There were no significant differences in mutation frequency between benign and borderline mucinous cystic neoplasms.

FIGURE 3
figure 3

A, nucleotide sequencing of K-ras gene in mucinous cystic neoplasm. The mutation is indicated by an arrow. The wild-type and mutated nucleotide sequences are shown above the arrow. B, allelic loss of chromosome 3p in mucinous cystic neoplasm and serous microcystic adenomas using dinucleotide microsatellite markers (D31110 and D31435). The lanes from normal DNA (N) have two alleles and the lanes from tumor DNA (T) show loss of one allele, indicated by arrows.

LOH of chromosome 3p25 was present in 57% (8/14) of serous microcystic adenomas compared with only 17% (2/12) of mucinous cystic neoplasms (P = .03, Fig. 3B, Table 3). Allelic loss of chromosome 3p25 was more common in mucinous cystic neoplasms with K-ras mutations than tumors without K-ras mutations (40%, 2/5 versus 0/7, P = .04) but was not associated with other clinicopathological characteristics in either type of cystic tumors.

No mutations in exon 3 of β-catenin, microsatellite instability, or alterations of TGFβRII were present in mucinous cystic neoplasms or serous microcystic adenomas (Table 3).

DISCUSSION

Epigenetic alterations have not been studied previously in mucinous cystic neoplasms (benign and borderline) or sporadic serous microcystic adenomas. In our study, we found infrequent methylation of p16 and p14 in mucinous cystic neoplasms and serous microcystic adenomas, rare methylation of VHL gene in serous microcystic adenomas, and no methylation of ppENK gene in either type of cystic tumor. These findings suggest that silencing of these genes by methylation may play a role in the tumorigenesis of some mucinous cystic neoplasms and serous microcystic adenomas. However, there is no distinct methylation profile that can distinguish these two cystic tumor types.

The Rb/p16 pathway is frequently inactivated in pancreatic carcinogenesis by homozygous/heterozygous deletions, mutations, and/or methylation (38). In our study, the frequency of methylation of the p16 gene in mucinous cystic neoplasms and serous microcystic adenomas is similar to the reported frequencies of its methylation in pancreatic ductal adenocarcinomas, intraductal papillary-mucinous neoplasm with carcinoma in situ or invasive carcinoma, and high-grade pancreatic intraepithelial neoplasia (23, 24, 39). In contrast, previous studies of benign and borderline intraductal papillary-mucinous neoplasms and serous microcystic adenomas reported no methylation of p16 gene but more frequent methylation of p16 gene in mucinous cystic neoplasms with invasive carcinoma (11, 23). p14ARF binds to human double minute 2 (HDM2) gene product to prevent HDM2 ubiquitin–mediated degradation of p53 (40). Methylation of p14 has been reported in various cancers and in colitis-associated dysplasia and neoplasia in ulcerative colitis patients (21, 22), but we report for the first time p14 methylation in pancreatic neoplasms.

Recently, frequent tumor-specific methylation of ppENK, which encodes met-enkephalin, has been reported in pancreatic ductal adenocarcinoma, intraductal papillary-mucinous neoplasm, and pancreatic ductal intraepithelial neoplasia (23, 24). Met-enkephalin is a growth inhibitory factor and decreases DNA synthesis, mitosis, and growth of cells (41). However, in the present study, no methylation of ppENK was present in either type of pancreatic cystic tumors. Thus, methylation of ppENK appear to be a selective event in pancreatic adenocarcinomas and intraductal papillary-mucinous neoplasms and may not play any role in the pathogenesis of MCNs and serous microcystic adenomas.

The complex of VHL protein, elongin C, elongin B, and CLU2 enzymatically adds ubiquitin to HIF-1 transcription factor, which is degraded by proteosomal complex. Impaired binding of VHL protein to elongin C and HIF-1 due to mutation, or lack of VHL protein due to deletion or methylation, fails to ubiquitinate HIF-1. This alteration constitutively activates HIF-1 that promotes VHL disease-associated tumors (42). Syndromic serous microcystic adenomas are due to germline mutations of the VHL gene, and sporadic serous microcystic adenomas have somatic alterations of the VHL gene including deletion and mutations (17, 18). In our study, we found more frequent loss of chromosome 3p25 in serous microcystic adenomas compared with mucinous cystic neoplasms. Methylation of the VHL gene was reported in sporadic and VHL disease-associated tumors such as renal cell carcinoma and hemangioblastoma (43). In the present study, methylation of the VHL gene was detected in only one serous microcystic adenoma, indicating that silencing of the VHL gene by methylation has only a minor role in sporadic serous microcystic adenomas.

In our study, K-ras mutations were present in mucinous cystic neoplasms but not serous microcystic adenomas, and the frequency of K-ras mutations was similar between benign and borderline subtypes of mucinous cystic neoplasms. Our finding corroborates results of previous studies that have shown less frequent K-ras mutations in benign and borderline subtypes of mucinous cystic neoplasms (9, 10, 12). In contrast, the frequency of K-ras mutations is more common in mucinous cystic neoplasms with in situ carcinoma or invasive carcinoma (9, 10, 11, 12, 13). Our study and the previous studies (9, 11, 12, 13) support the concept that there is a progressive increase in the frequencies of the genetic alterations, including K-ras mutation, and p53, p16, and Dpc4 gene alterations in mucinous cystic tumors with histological progression: the genetic alterations are more common in mucinous cystic tumors with in situ carcinoma or invasive carcinoma than in benign and borderline subtypes.

We found that mucinous cystic neoplasms and serous microcystic adenomas lacked β-catenin mutation or microsatellite instability, in contrast to pancreatoblastomas, pancreatic acinar cell carcinomas, and solid-pseudopapillary carcinomas that have β-catenin mutation (25, 26, 27) and medullary subtypes of pancreatic carcinomas that have microsatellite instability (28).

It is important to distinguish mucinous cystic neoplasms from serous microcystic adenomas preoperatively because the former are more prone to recurrence if incompletely resected and can transform into invasive carcinoma. There is no distinctive methylation profile based on the limited number of CpG islands analyzed in the current study, but these two cystic tumor types have distinct histology and can be easily distinguished. Evaluation of K-ras gene mutation and chromosome 3p loss in aspiration biopsy specimens may be of help to distinguish mucinous cystic neoplasms from serous microcystic adenomas because of different frequencies in the tumor types: K-ras mutation favors mucinous cystic neoplasms, and 3p25 allelic loss without K-ras mutation favors serous cystic adenoma.

In conclusion, distinct genetic alterations rather than epigenetic alterations may account for the different pathogenesis of these two types of pancreatic cystic tumors. These findings may have diagnostic implications.