Met Proto-Oncogene and Insulin-Like Growth Factor Binding Protein 3 Overexpression Correlates with Metastatic Ability in Well-Differentiated Pancreatic Endocrine Neoplasms

Pancreatic endocrine neoplasms are neoplastic proliferations of islet cells or islet cell precursors and may be classified as functional or nonfunctional lesions (1). Functional pancreatic endocrine neoplasms frequently present with symptoms related to excessive hormonal secretion such as insulin, gastrin, glucagon, somatostatin, and vasoactive intestinal peptide, whereas nonfunctional pancreatic endocrine neoplasms present secondary to mass effect. Frequently used histopathologic criteria that may herald aggressive behavior in pancreatic endocrine neoplasms include angioinvasion, perineural invasion, and prominent cytologic atypia, among others; however, besides the presence of synchronous metastases, there are few consistently reliable criteria for predicting clinical outcome. A recent study, however, has identified that mitotic index and presence of necrosis may be useful indicators of patient prognosis (2).

Alterations in molecular profiles underlying the development of pancreatic endocrine neoplasms have only recently begun to be elucidated, and little is known regarding the molecular changes that confer less favorable prognostic outcomes. The identification of the gene responsible for the multiple endocrine neoplasia 1 (MEN1) syndrome, MEN1, has led to additional understanding of the subset of pancreatic endocrine neoplasms arising in association with this syndrome, although MEN1 alterations appear to vary in sporadic pancreatic endocrine neoplasms (3, 4). Examination of well-characterized tumor suppressor genes in sporadic lesions has led to the identification of only rare mutations in such genes as p53 (5), p16/MTS1 (6), PTEN (7), among others. Overexpression of KRAS2 has been reported in pancreatic endocrine neoplasms, although this also appears to represent a rare event (8). Finally, loss of a single sex chromosome, either X or Y, in pancreatic endocrine neoplasms appears to correlate with a worsened prognosis, although the genes underlying this phenotype are unknown (9).

Recently, global gene analysis has been used by our group to identify differentially up-regulated genes in well-differentiated nonmetastatic pancreatic endocrine neoplasms versus normal human islet cells (10). This screening technique identified 66 overexpressed transcripts in pancreatic endocrine neoplasms, many of which may function in islet cell carcinogenesis. We have now extended our analysis of pancreatic endocrine neoplasm lesions to examine the differential expression of genes in well-differentiated metastatic versus nonmetastatic pancreatic endocrine neoplasms to determine markers of disease progression. Although the precise pathogenesis of pancreatic endocrine neoplasm development remains unclear, a progression model leading from normal islet cell to primary pancreatic endocrine neoplasm to metastatic pancreatic endocrine neoplasm appears functionally plausible. Furthermore, as only a limited subset of pancreatic endocrine neoplasms metastasize, the identification of altered gene expression profiles in this population may reflect commonly altered pathways relevant to a variety of metastatic cancers.

Permission for this study was obtained from The Johns Hopkins Joint Committee for Clinical Investigation. Specimens were obtained from patients undergoing distal pancreatectomies or Whipple resection for functional and nonfunctional pancreatic endocrine neoplasms. Metastatic (n = 7) and nonmetastatic (n = 5) pancreatic endocrine neoplasms were used for analysis. The male:female ratio was 5:2 and 2:3 for metastatic and nonmetastatic lesions, respectively. Pancreatic endocrine neoplasm tissue was harvested within 10 minutes of surgical resection, snap frozen in liquid nitrogen, and stored at −80°C. H&E-stained sections from adjacent frozen tissue were prepared before sample harvest to confirm the diagnosis (10) and assess neoplastic cellularity. A subset of pancreatic endocrine neoplasms included in this study were additionally classified as functional lesions by clinical hormone production or specific immunolabeling panels.

RNA extraction and processing were performed at GeneLogic, Inc. (Gaithersburg, MD). Sample preparation and processing procedure was performed as described in the Affymetrix GeneChip Expression Analysis Manual (Santa Clara, CA). Briefly, each frozen tissue was crushed to powder by using the Spex CertiPrep 6800 Freezer Mill (Metuchen, NJ). Total RNA was then extracted from crushed metastatic and nonmetastatic pancreatic endocrine neoplasm tissue using TRIzol (Life Technologies, Inc., Rockville, MD) and cleaned using RNeasy columns according to the manufacturer’s protocol (Qiagen, Valencia, CA). Using 5 to 40 μg of total RNA, double-stranded cDNA was synthesized following SuperScript Choice system (Life Technologies, Inc.). T7-(dT24) oligomer was used for priming the first-strand cDNA synthesis. The resultant cDNA was purified using Phase Lock Gel, phenol/chloroform extraction, and precipitated with ethanol. The cDNA pellet was collected and dissolved in the appropriate volume. Using cDNA as a template, cRNA was synthesized using a T7 MegaScript In Vitro Transcription kit (Ambion, Austin, TX). Biotinylate-11-CTP and 16-UTP ribonucleotides (Enzo Diagnostics, Inc., Farmingdale, NY) were added to the reaction as labeling reagents. In Vitro Transcription reactions were performed at 37°C for 6 hours, and the labeled cRNA obtained was purified using RNeasy columns (Qiagen). The cRNA was fragmented in fragmentation buffer (40 mmol/L Tris-acetate (pH 8.1), 100 mmol/L KOAc, and 30 mmol/L MgOAc] for 35 minutes at 94°C. Fragmented cRNA prepared from each sample (10 to 11 μg/probe array) was hybridized to the human GeneChip set (HG_U133A and U133B) noncompetitively at 45°C for 24 hours in a hybridization oven with constant rotation (60 rpm). Fragmented cRNA are hybridized to the GeneChip set by way of multiple 20 to 25 oligonucleotide probes specific for each gene, with each probe corresponding to a different region of the mRNA of interest. The probes specific for each mRNA are scattered across the surface of each GeneChip to control for technical issues that occur in each hybridization. The chips were washed and stained using Affymetrix fluidics stations. Staining was performed using streptavidin-phycoerythrin conjugate (Molecular Probes, Eugene, OR), followed by the addition of biotinylated antibody to streptavidin (Vector Laboratories, Burlingame, CA) and finally with streptavidin-phycoerythrin conjugate. Probe arrays were scanned using fluorometric scanners (Hewlett Packard Gene Array Scanner; Hewlett Packard Corporation, Palo Alto, CA).

The scanned images were inspected and analyzed using established quality control measures, with the hybridization intensities reflecting in a linear manner the mRNA expression in the tissues being assayed. Hybridization was controlled for each probe by the use of a mismatch control that has a single base mismatch. This mismatch control is analyzed using the GeneLogic informatics filter that compares the hybridization intensity of mismatched to perfect matched probes (to eliminate those that are nonspecific over a specified threshold), as well as different probes to the same gene.

The GeneExpress Software System Fold Change Analysis tool was used to identify genes expressed at least 5-fold greater in the metastatic pancreatic endocrine neoplasms compared with nonmetastatic lesions. For each gene fragment, the ratio of the geometric means of the expression intensities in the normal control tissues and the pancreas cancer samples was calculated, and the fold change then calculated on a per fragment basis. Confidence limits were calculated using a two-sided Welch modified t test on the difference of the means of the logs of the intensities.

Tissue microarrays were prepared from 24 nonmetastatic and 15 metastatic pancreatic endocrine neoplasms banked at The Johns Hopkins Hospital from 1993 through 2002. When available, metastatic lesions to the liver (n = 9) or lymph nodes (n = 7) were additionally sampled. Each cancer specimen was represented by four 1.4-mm cores on the tissue microarrays to obtain adequate representation of different regions of neoplastic cells to assess for intratumoral heterogeneity. In addition, nonneoplastic pancreatic islets from the corresponding patients and additional nonneoplastic tissue from the gallbladder, colon, skin, breast, prostate, thymus, and brain were sampled on the tissue microarrays. Slides were deparaffinized in fresh xylenes and rehydrated through sequential-graded ethanol steps. Antigen retrieval was performed by citrate buffer incubation [18 mmol/L sodium citrate (pH 6.0)] using a household vegetable steamer for 60 minutes. Slides were incubated for 5 minutes with 3% hydrogen peroxide, washed in TBS/T [20 mmol/L Tris, 140 mmol/L NaCl, 0.1% Tween 20 (pH 7.6)], and incubated in appropriate antibody dilutions for IGFBP3 (1:50; Santa Cruz Biotechnology, Santa Cruz, CA) and Met (1:1000; Santa Cruz Biotechnology) for 60 minutes at room temperature. Normal saline was substituted for the primary antibody in control sections. The avidin-biotin-peroxidase complex method from DAKO (Glostrup, Denmark) was used, and slides were subsequently counterstained with hematoxylin. Assessment of immunohistochemical labeling in the tissue microarrays was performed by two of the authors (D. Hansel and A. Maitra). IGFBP3 was scored as positive if any level of cytoplasmic staining was detected. Met was scored as positive for membranous staining only. Comparison of immunolabeling between populations was performed using the Fisher’s exact test, and prediction of metastatic spread was performed using positive- and negative-predictive value scores.

Normalization and comparison of the Affymetrix microarray hybridization data were performed using the GeneLogic GeneExpress Software System Fold Change Analysis tool. Sixtyfive overexpressed transcripts and 57 underexpressed transcripts were identified in well-differentiated metastatic versus nonmetastatic pancreatic endocrine neoplasms in our analysis (Table 1). Overexpression of multiple factors involved in growth regulation (MET, IGFBP1, and IGFBP3), cholesterol homeostasis (NPC1L1, LRP5, and SLC27A2), osmotic regulation (AQP3 and solute carrier 6), chemical modification (UGT2B4), cytoskeletal-related molecules (β 1 tubulin and myosin X), coagulation (FGA and coagulation factor V), and hypoxia-inducible factors (IGFBP1 and LDHB) was identified. Underexpressed transcripts were notable for multiple cell-cycle regulatory (CHEK1 and ZNF198), developmental regulatory molecules (MEIS2), second messenger signaling mediators (RAB25 and DYRK), and DNA damage repair (MGMT and GAMT) molecules. Overall pathway modifications were evident for the IGF-signaling cascade (IGFBP1, IGFBP3, EHD1, and ACTN2) in which changes in gene expression would promote pathway activation.

A literature search of PubMed using transcript names paired with “islet cell,” “pancreatic endocrine,” “metastasis,” or “carcinoma” revealed that a subset of overexpressed transcripts had been previously identified in pancreatic endocrine neoplasms, such as IGFBP3 (10), as well as in other cancer types.⁵ Overexpression of MET has been previously described in a wide variety of carcinomas and associated metastases, including pancreatic endocrine neoplasms (11), breast carcinoma (12), and squamous cell carcinoma (13). In addition, three altered transcripts reflected a similar overexpression pattern in our metastatic pancreatic endocrine neoplasm lesions as in human hepatocellular carcinoma specimens analyzed by global gene analysis (IGFBP1, UGT2B4, and VTN; ref. 14). The majority of transcripts identified as over- or underexpressed by our method, however, have not been previously reported in pancreatic endocrine neoplasms or in metastatic lesions and may therefore represent novel cellular targets in this cancer type.

Confirmation of altered gene expression patterns was performed using a tissue microarray comprised of 24 nonmetastatic pancreatic endocrine neoplasms, 15 metastatic pancreatic endocrine neoplasms, 9 liver metastases, and 7 lymph node metastases arising from pancreatic endocrine neoplasm primary lesions (Table 2). Lesions were represented by four cores on the tissue microarray to assess for intratumoral heterogeneity; in all cases, the presence of labeled protein appeared uniform throughout all cores. Expression of two overexpressed transcripts, IGFBP3 (4.88-fold overexpression) and MET (4.90-fold overexpression), was performed using immunolabeling analysis (Table 3).

Nonneoplastic pancreatic tissue demonstrated focal expression of MET in small ductal epithelium cells but not in islet or acinar cells. Examination of metastatic primary pancreatic endocrine neoplasms revealed that 5 of 15 (33%) of these lesions demonstrated MET expression in contrast to only 17% (4 of 24) of nonmetastatic pancreatic endocrine neoplasms (Fig. 1, A and B; Fisher’s exact test, P = 0.4150). Furthermore, analysis of pancreatic endocrine neoplasm metastases identified a still higher proportion of MET-positive lesions metastatic to lymph nodes (4 of 7, 57%; Fisher’s exact test, P = 0.1056) and liver (5 of 9, 56%; Fisher’s exact test, P = 0.0788) as compared with expression in nonmetastatic pancreatic endocrine neoplasms. All MET-expressing lesions demonstrated robustly positive membranous labeling.

IGFBP3 demonstrated a weak granular immunolabeling pattern within the cytoplasm of normal pancreatic islet cells but not in pancreatic ducts or acinar cells. Similar to MET, IGFBP3 also demonstrated an approximate 2-fold increase in the percentage of IGFBP3-positive lesions in metastatic pancreatic endocrine neoplasms (12 of 15, 80%) versus nonmetastatic pancreatic endocrine neoplasms (10 of 24, 42%; Fig. 1, C and D). In this study, any level of cytoplasmic immunolabeling was scored as positive. A Fisher’s exact test comparing these two populations demonstrated a borderline significant P of 0.0408. Prominent expression of IGFBP3 was also confirmed in lymph node (6 of 7, 86%; Fisher’s exact test, P = 0.1001) and liver metastases (9 of 9, 100%; Fisher’s exact test, P = 0.0048) as compared with expression in nonmetastatic pancreatic endocrine neoplasms.

Statistical analysis of MET and IGFBP3 expression was performed on primary pancreatic endocrine neoplasms to determine whether immunolabeling for these molecules could predict metastatic spread. MET immunolabeling demonstrated a specificity of 83% for metastatic spread, and IGFBP3 labeling demonstrated a sensitivity of 80%; neither molecule, however, demonstrated a robust positive predictive value for the development of metastases. The finding of proportionally increased expression within metastases, however, suggests that these molecules may function at a molecular level to promote metastatic spread.

Subclassification of pancreatic endocrine neoplasms by IGFBP3 and MET expression revealed that the majority of glucagonomas (2/2), gastrinomas (1/2), and VIPomas (1/1) were positive for IGFBP3, whereas only the VIPoma demonstrated MET immunoreactivity. Of note, none of the insulinomas (n = 3), which frequently do not metastasize, demonstrated immunoreactivity for MET or IGFBP3. In our study population of 39 patients with primary pancreatic endocrine neoplasms, 3 patients expired 6, 8, and 26 months after initial diagnosis and surgical intervention (the remaining 36 patients were alive without evidence of pancreatic endocrine neoplasm recurrence). Lesions collected from these patients demonstrated either IGFBP3 (2/3) or MET (1/3) expression in all three cases.

Pancreatic endocrine neoplasms encompass a wide spectrum of lesions that often defy standard forms of categorization and therefore lead to challenges in the understanding of the underlying biology of these lesions. Examination of multiple indices of unfavorable patient outcomes and aggressive biological behavior, including proliferation index, necrosis, chromogranin reactivity, and size, have led in certain instances to useful histologic criteria, although molecular alterations that influence patient outcome are lacking. Metastatic spread of pancreatic endocrine neoplasms has been proposed to represent a negative prognostic indicator, as often disseminated disease may not be amenable to surgical cure or adjuvant treatment. To gain additional insight into the molecular alterations that occur in disseminated pancreatic endocrine neoplasm disease, we performed global gene analysis expression of well-differentiated metastatic versus nonmetastatic primary pancreatic endocrine neoplasms and confirmed selected gene expression profiles on tissue microarrays composed of 39 primary pancreatic endocrine neoplasms and 16 pancreatic endocrine neoplasm metastases.

Analysis of gene expression profiles revealed that 65 genes were overexpressed, and 57 genes were underexpressed in metastatic versus nonmetastatic primary pancreatic endocrine neoplasms. In addition to a variety of factors that have been implicated in cell proliferation and metastatic spread in other cancer types, our findings were notable for altered expression of genes involved in cell cycle and DNA repair regulation, cellular growth, cholesterol and lipid homeostasis, intracellular signaling, and coagulation, as well as factors that are induced under hypoxic conditions. To validate the gene expression profiles obtained by our analysis, we examined the expression patterns of IGFBP3 and MET on tissue microarrays containing 24 nonmetastatic and 15 metastatic primary pancreatic endocrine neoplasms, as well as 9 liver and 7 lymph node metastases.

IGFBP3 functions as a carrier molecule for both IGF-I and IGF-II in the circulation (15, 16). IGFBP3 mediates both pro- and antiproliferative effects on various cell types (16), and increased serum levels of IGFBP3 have been associated with progression of breast cancer in several studies (17, 18). We have previously identified the overexpression of IGFBP3 (4.1-fold) in nonmetastatic pancreatic endocrine neoplasms versus normal human islet cells (10). In comparison, IGFBP3 appears to be additionally up-regulated in metastatic versus nonmetastatic pancreatic endocrine neoplasms (4.88-fold), suggesting a continuum of IGFBP3 expression and influence on pancreatic endocrine neoplasm progression and metastases. Analysis of IGFBP3 expression in metastatic versus nonmetastatic pancreatic endocrine neoplasms identified IGFBP3 expression in 42% of nonmetastatic pancreatic endocrine neoplasms versus 80% of metastatic primary pancreatic endocrine neoplasms. In addition, IGFBP3 expression was identified in 86 and 100% of lymph node and liver metastases, respectively.

In addition to IGFBP3, several additional components of the IGF signaling pathway demonstrated altered expression levels, including IGFBP1, ACTN2, and EHD1. IGFBP1 functions as a carrier molecule for IGF-I and IGF-II, undergoes induction by hypoxic conditions (19), and reflects a poorer outcome in cancer patients with elevated circulating levels of this molecule. ACTN2 has been proposed to function as a key transducing molecule in the signaling pathway leading from IGF receptor I activation to cell membrane microspike production, cell-cell separation, and cell migration (20). Finally, EHD1 has been shown to influence the endocytosis of IGF receptor I from the cell surface (21); decreased expression of this molecule, as identified in this study, could potentially lead to an increased cell surface half-life of, and therefore increased signaling through, IGF receptor I. Overall, IGF signaling appears to be enhanced in metastatic versus nonmetastatic pancreatic endocrine neoplasms secondary to altered gene expression.

An additional validation of our data were performed by examining MET expression (overexpressed 4.90-fold) in metastatic versus nonmetastatic pancreatic endocrine neoplasms. MET functions as a transmembrane receptor tyrosine kinase that is activated by hepatocyte growth factor/scatter factor (22). Inappropriate expression of MET has been documented in the majority of solid tumor types (23) and appears to often correlate with worsened prognosis. MET signaling also results in disruption of cell-cell adhesion, branching morphogenesis, and invasive and metastatic behavior of a large array of neoplasms (24). We have identified the expression of MET in 17% of nonmetastatic pancreatic endocrine neoplasms versus 33% of primary pancreatic endocrine neoplasms demonstrating concurrent metastases. MET expression appeared most prevalent in lymph node (57%) and liver (56%) metastases. As with IGFBP3, MET expression may also demonstrate a continuum of expression with neoplastic progression.

Several molecules involved in DNA repair and maintenance of cell-cycle checkpoints appeared to be globally down-regulated in our analysis, including the methyltransferase MGMT, as well as DNA repair and regulatory molecules such as CHEK1 and ZNF198. The expression of MGMT, a DNA repair gene, is regulated by methylation of CpG islands within the promoter region of the gene, and decreased expression of MGMT has been reported in gastric cancer (25), lung (26), and brain (27) cancer. CHEK1 encodes a protein kinase that prevents progression of the cell cycle after double-stranded DNA breaks via a Cdc25A-regulated mechanism (28) and has therefore been suggested to represent a novel tumor suppressor gene (29). Recent analysis of the normal function of ZNF198 has revealed that this molecule may also serve in the DNA repair process through interactions with RAD18 and HHR6 (30).

A final global pathway that demonstrated multiple alterations of gene expression involves lipid metabolism and cholesterol homeostasis. Neoplastic growth requires the formation of new cell membranes, which are dependent upon cholesterol derivatives and subsequent phospholipid formation (31). Renal and brain cancer (32), for example, demonstrate alterations in cholesterol homeostasis that appear to influence cancer growth and metastatic behavior. Our study has identified the overexpression of LRP5 and solute carrier family 27 (SLC27A2), as well as the underexpression of RXRG, all of which influence lipid balance within the cell. LRP5 is a cell surface protein that mediates ligand-internalization, is required for cholesterol balance (33) and is involved in promoting canonical Wnt signaling within cells (34). SLC27A2 is a transmembrane protein that transports long-chain and very long-chain fatty acids into the cell and activates intracellular signaling pathways such as protein kinase C and peroxisome proliferator-activated receptors (35). Additional fatty acid and cholesterol-modifying molecules overexpressed in our study include NPC1-like-1 (NPC1L1, subcellular cholesterol trafficking (36), aquaporin 3 (AQP3, glycerol channel), and peroxisomal matrix protein (catalyzes the oxidation of very long chain fatty acids).

Our study has identified the over- and underexpression of genes involved in multiple facets of cell growth and metastases in primary pancreatic endocrine neoplasm lesions. This study represents the first systematic analysis of altered expression in metastatic lesions of this cancer subtype. Comparison of metastatic versus nonmetastatic primary pancreatic endocrine neoplasms have yielded a limited novel group of altered transcripts that may serve as functional targets for metastases arising from primary pancreatic endocrine neoplasms. Additional analysis using a larger set of pancreatic endocrine neoplasms with paired metastases may be useful in the analysis of various transcripts identified within this screen, especially in the context of predicting metastatic spread. Finally, the comparison of altered gene profiles from islet cell to nonmetastatic to metastatic pancreatic endocrine neoplasms may yield additional information regarding the underlying pathophysiology of pancreatic endocrine neoplasm progression.