Defining molecular classifications and targets in gastroenteropancreatic neuroendocrine tumors through DNA microarray analysis

in Endocrine-Related Cancer
Authors:
Eva-Maria Duerr Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Eva-Maria Duerr in
Current site
Google Scholar
PubMed
Close
,
Yusuke Mizukami Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Yusuke Mizukami in
Current site
Google Scholar
PubMed
Close
,
Aylwin Ng Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Aylwin Ng in
Current site
Google Scholar
PubMed
Close
,
Ramnik J Xavier Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Ramnik J Xavier in
Current site
Google Scholar
PubMed
Close
,
Hirotoshi Kikuchi Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Hirotoshi Kikuchi in
Current site
Google Scholar
PubMed
Close
,
Vikram Deshpande Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Vikram Deshpande in
Current site
Google Scholar
PubMed
Close
,
Andrew L Warshaw  Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Andrew L Warshaw  in
Current site
Google Scholar
PubMed
Close
,
Jonathan Glickman Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Jonathan Glickman in
Current site
Google Scholar
PubMed
Close
,
Matthew H Kulke Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Matthew H Kulke in
Current site
Google Scholar
PubMed
Close
, and
Daniel C Chung Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology
Gastrointestinal Unit, Department of Medicine, Cancer Center, Center for Computational and Integrative Biology, Department of Surgery, Department of Pathology, Department of Adult Oncology

Search for other papers by Daniel C Chung in
Current site
Google Scholar
PubMed
Close

Free access

Sign up for journal news

Current classifications of human gastroenteropancreatic neuroendocrine tumors (NETs) are inconsistent and based upon histopathologic but not molecular features. We sought to compare a molecular classification with the World Health Organization (WHO) histologic classification, identify genes that may be important for tumor progression, and determine whether gastrointestinal NETs (GI-NETs) differ in their molecular profile from pancreatic NETs (PNETs). DNA microarray analysis was performed to identify differentially expressed genes in PNETs and GI-NETs. Confirmation of expression levels was obtained by quantitative real-time PCR. Immunoblotting and mutational analysis were performed for selected genes. Hierarchical clustering of 19 PNETs revealed a ‘benign’ and ‘malignant’ cluster that corresponded well with the WHO categories of well-differentiated endocrine tumor (WDET) and well-differentiated endocrine carcinoma (WDEC) respectively. FEV, adenylate cyclase 2 (ADCY2), nuclear receptor subfamily 4, group A, member 2 (NR4A2), and growth arrest and DNA-damage-inducible, beta (GADD45b) were the most highly up-regulated genes in the malignant group of PNETs. Platelet-derived growth factor receptor (PDGFR) was expressed in both WDETs and WDECs, and phosphorylation of PDGFR-β was observed in 83% of all PNETs. Malignant ileal GI-NETs exhibited a distinctive gene expression profile, and extracellular matrix protein 1 (ECM), vesicular monoamine member 1 (VMAT1), galectin 4 (LGALS4), and RET Proto-oncogene (RET) were highly up-regulated genes. Gene expression profiles reflect the current WHO classification and can distinguish benign from malignant PNETs and also PNETs from GI-NETs. This suggests that molecular profiling may enhance tumor classification schemes. Potential gene targets have also been identified, and PDGFR and RET are candidates that may represent novel therapeutic targets.

Abstract

Current classifications of human gastroenteropancreatic neuroendocrine tumors (NETs) are inconsistent and based upon histopathologic but not molecular features. We sought to compare a molecular classification with the World Health Organization (WHO) histologic classification, identify genes that may be important for tumor progression, and determine whether gastrointestinal NETs (GI-NETs) differ in their molecular profile from pancreatic NETs (PNETs). DNA microarray analysis was performed to identify differentially expressed genes in PNETs and GI-NETs. Confirmation of expression levels was obtained by quantitative real-time PCR. Immunoblotting and mutational analysis were performed for selected genes. Hierarchical clustering of 19 PNETs revealed a ‘benign’ and ‘malignant’ cluster that corresponded well with the WHO categories of well-differentiated endocrine tumor (WDET) and well-differentiated endocrine carcinoma (WDEC) respectively. FEV, adenylate cyclase 2 (ADCY2), nuclear receptor subfamily 4, group A, member 2 (NR4A2), and growth arrest and DNA-damage-inducible, beta (GADD45b) were the most highly up-regulated genes in the malignant group of PNETs. Platelet-derived growth factor receptor (PDGFR) was expressed in both WDETs and WDECs, and phosphorylation of PDGFR-β was observed in 83% of all PNETs. Malignant ileal GI-NETs exhibited a distinctive gene expression profile, and extracellular matrix protein 1 (ECM), vesicular monoamine member 1 (VMAT1), galectin 4 (LGALS4), and RET Proto-oncogene (RET) were highly up-regulated genes. Gene expression profiles reflect the current WHO classification and can distinguish benign from malignant PNETs and also PNETs from GI-NETs. This suggests that molecular profiling may enhance tumor classification schemes. Potential gene targets have also been identified, and PDGFR and RET are candidates that may represent novel therapeutic targets.

Introduction

Pancreatic neuroendocrine tumors (PNETs) and gastrointestinal neuroendocrine tumors (GI-NETs) are tumors of neuroendocrine origin that share many common biological features. Based on criteria including tumor size, mitotic rate, Ki-67 index, angioinvasion, and distant metastases, the World Health Organization (WHO) classifies these tumors into three groups: well-differentiated NETs (WDET), well-differentiated neuroendocrine carcinomas (WDEC), and poorly-differentiated neuroendocrine carcinomas (PDEC; Kloppel et al. 2004). The WDETs are further subdivided into WDETs of benign behavior and WDETs of low-grade malignant/uncertain behavior (Table 1). Whether these two subgroups are distinct entities or related entities within a spectrum is unknown. It also remains a challenge to distinguish WDETs from WDECs because histologic criteria are imperfect and the only definitive criterion for malignancy is the presence of metastases. Although our understanding of the cellular biology and clinical behavior of NETs has increased in sophistication, insights into their underlying molecular genetics have lagged behind. A number of candidate genes have been implicated in the pathogenesis of PNETs (reviewed in (Duerr & Chung 2007)), including multiple endocrine neoplasia type 1 (MEN1; Shan et al. 1998, Wang et al. 1998), retinoic acid receptor-β (House et al. 2003b), hMLH1 (human mutL homologue 1; House et al. 2003a), RASSF1 (Ras association domain family 1; House et al. 2003b), Her2/neu (herstatin; Evers et al. 1994, Goebel et al. 2002), and the cell cycle regulators cyclin D1 (Chung et al. 2000, Guo et al. 2003), p16INK4a/p14ARF (Muscarella et al. 1998), p18INK4c, and p27Kip1 (Guo et al. 2001), as well as tyrosine kinase receptors (Fjallskog et al. 2003). However, the genetics of tumor progression are poorly defined.

Table 1

The World Health Organization classification of neuroendocrine tumors of the pancreas

CategoryHistologyLocalizationSize (cm)% Ki-67 positive cellsAngioinvasion
Well-differentiated neuroendocrine tumor
 Benign behaviorWell-differentiatedConfined to pancreas<2≤2%No
Well-differentiated neuroendocrine tumor
 Low-grade malignant (or uncertain behavior)Well-differentiatedConfined to pancreas≥2>2%Yes
Well-differentiated neuroendocrine carcinomaWell-differentiatedInvasion of adjacent organs and/or metastases≥2>2%Yes
Poorly-differentiated neuroendocrine carcinoma Poorly-differentiatedInvasion of adjacent organs and/or metastasesAny>30%Yes

It is also uncertain how similar the genetic alterations are that underlie PNETs and GI-NETs, and whether they can be distinguished on a molecular level. Although there are important differences in their clinical behavior, they are still classified in a similar manner by the WHO criteria. DNA microarray technology is a promising tool to better understand gene expression patterns that underlie tumor development. Thus far, only a few studies have investigated gene expression profiles in PNETs. Most of these have focused on differences between tumors and normal tissue (Maitra et al. 2003, Bloomston et al. 2004, Capurso et al. 2006). In the present study, we sought to compare molecular classifications with the WHO histologic classification, identify genes that may be important for tumor progression, and determine whether GI-NETs differ in their molecular profile from PNETs. This was accomplished with the use of DNA microarrays. With such an approach, novel genes that are expressed in a highly differential manner can be identified, and expression patterns of candidate genes that may have biological relevance to neuroendocrine tumorigenesis can be easily defined. This strategy can therefore identify genes of potential interest in both a non-targeted and targeted manner, and confirmation can be obtained through protein analysis.

Materials and methods

Patient samples

Fresh frozen tissue samples of 24 PNETs (5 benign WDETs, 11 low-grade malignant WDETs, and 8 WDECs) and 6 malignant GI-NETs were obtained as surgical discards from Massachusetts General Hospital and Brigham and Women's Hospital/Dana-Farber Cancer Institute respectively. Tumors were classified according to the WHO 2004 criteria. All of the PNET samples were primary tumors. Clinical characteristics are summarized in Tables 2 and 3. This protocol was approved by the institutional review board of each institution.

Table 2

Clinical characteristics of pancreatic neuroendocrine tumor (PNET) samples included in DNA microarray and western blot analyses

ID no.AgeSexTumor typeWHO classificationSpecimen typeTumor size (cm)Ki-67 index (%)MicroarrayPDGFR WB
262MPTHRP-secreting tumorWDECPrimary8.53.9YesNo
5369FNon-functioning PNETWDET LMPrimary13.05.7YesYes
5570MGlucagonomaWDET LMPrimary3.02.0YesNo
5673MInsulinomaWDECPrimary2.51.8YesNo
5746MNon-functioning PNETWDECPrimary10.0<1NoYes
5965MACTHomaWDECPrimary6.52.0YesYes
6158FGastrinomaWDECPrimary12.016.8YesYes
6357FInsulinomaWDET BPrimary1.52.3NoYes
6948FGastrinomaWDECPrimary3.0ndYesNo
7036FInsulinomaWDET BPrimary1.5ndYesYes
7154FInsulinomaWDET LMPrimary2.51.1YesYes
7243FInsulinomaWDET BPrimary1.21.2YesYes
7424FInsulinomaWDET LMPrimary3.04.0YesYes
7615FInsulinomaWDET LMPrimary1.81.8YesNo
7762FGlucagonomaWDECPrimary7.07.5NoYes
7851FInsulinomaWDET BPrimary1.83.9NoYes
7951FInsulinomaWDET LMPrimary5.0<1NoYes
8054MInsulinoma WDET LMPrimary2.2<1YesNo
8174FNon-functioning PNETWDET LMPrimary4.04.2YesYes
8260MInsulinomaWDET BPrimary1.8<1YesYes
8347FNon-functioning PNETWDECPrimary7.0<1YesYes
8459MRecurrent gastrinomaWDECPrimarynkndYesYes
8544MNon-functioning PNETWDET LMPrimary3.06.5YesYes
8683FInsulinomaWDET LMPrimary3.0<1YesYes

WDET B, well-differentiated endocrine tumor – benign; WDET LM, well-differentiated endocrine tumor – low-grade malignant; WDEC, well-differentiated endocrine carcinoma; nd, not done (no tumor sample available); nk, not known.

Table 3

Clinical characteristics of gastrointestinal (GI) neuroendocrine tumor samples included in the DNA microarray analysis

IDAgeSexPrimary siteSpecimen typePrimary tumor size (cm)StageCarcinoid syndrome
2974-172nkIleumPrimary ileal tumor0.4MetastaticYes
11898-152FIleumLiver met lesion2.0MetastaticYes
33762-167MIleumLiver met lesion1.5MetastaticYes
53456-159FColonPrimary colon tumornkMetastaticnk
67494-153FIleumPrimary ileal tumor2.5T3N1M0No
80670-170FIleumPrimary ileal tumor3.5MetastaticNo

nk, not known.

RNA extraction

RNA was extracted from frozen tumors following dissection from normal surrounding tissue using Trizol following the manufacturer's recommendations (Invitrogen, Carlsbad, CA, USA) and purified using the RNeasy MinElute Cleanup kit (Qiagen, Valencia, CA, USA).

DNA microarrays

RNA analyses were performed at the DNA Microarray Core Facility at Massachusetts General Hospital Cancer Center. Amounts, purity, and integrity of RNA were evaluated by u.v. spectrophotometry and an RNA-nano Bioanalyzer (Agilent, Palo Alto, CA, USA). Probe synthesis and hybridization of human U-133A GeneChip DNA microarrays (Affymetrix, Santa Clara, CA, USA) were performed following the manufacturer's instructions.

Microarray data analysis

Data analysis was performed using DChip software (www.dchip.org). CEL files (primary Affymetrix array data files) were loaded and normalized at the probe cell level by the Invariate Set Normalization method (Li & Hung Wong 2001). The model-based method (Li & Hung Wong 2001) was used for probe selection and computing expression values. These expression values were attached with standard errors as measurement accuracy. The lower confidence intervals of fold changes were conservative estimates of real fold changes. The ANOVA test was carried out using a P value <0.05 in order to define a set of significantly up- or down-regulated genes. The resulting genes were filtered for gene presence calls of >20 in >50% of samples. Two-group comparison was employed selecting for increased or decreased gene expression by more than 1.5-fold. Hierarchical clustering analysis (Eisen et al. 1998) was performed on the genes that met the above criteria.

Gene ontology

Enrichments of gene ontology (GO) categories were computed using the hypergeometric probability distribution, which identifies GO molecular function categories overrepresented in the set of differentially induced genes relative to their representation on the Affymetrix U133A array. The analysis was performed using Onto-Tools (Draghici et al. 2003) and GO molecular function categories with P value <0.05 are considered significantly overrepresented.

Protein interaction network

The network was constructed by iteratively connecting interacting proteins, with protein interaction data obtained from the Human Protein Reference Database (Peri et al. 2004). The network uses graph theory, which represents components (gene products) as nodes and interactions between components as edges. Graph layout descriptions were written in the Dot language (Gansner & North 2000) that implements a multidimensional scaling heuristic, which creates a virtual physical model (Spring model; Kamada & Kawai 1989) and is coupled to an iterative solver (Newton–Raphson algorithm) that searches for low-energy configurations to optimize the graph layout.

Quantitative real-time PCR (qRT-PCR)

qRT-PCR of RNA from the 24 tumor samples used in the microarray analysis and 3 normal pancreas samples was performed utilizing the SuperScript III platinum Two-Step qRT-PCR Kit (Invitrogen). The 18S rRNA served as an endogenous control. Primer sequences and PCR conditions for FEV, adenylate cyclase 2 (ADCY2), nuclear receptor subfamily 4, group A, member 2 (NR4A2), growth arrest and DNA-damage-inducible, beta (GADD45β), extracellular matrix protein 1 (ECM1), vesicular monoamine member 1 (VMAT1), LGALS4, RET, and 18S are available upon request. A fluorogenic SYBR Green and MJ research detection system were used for real-time quantification. Relative mRNA expression was calculated using the parameter threshold cycle (CT) values. ΔCT was the difference in the CT values derived from the specific gene being assayed and the 18S rRNA. ΔΔCT represented the difference between the paired samples, as calculated by the formula ΔCT of a sample −ΔCT of a reference (the average ΔCT of three normal pancreas samples). The amount of target, normalized to 18S and the reference, was calculated as .

Protein lysates and western blot analysis

Protein lysates were prepared from 18 snap-frozen PNET samples and 3 normal tissues (two from pancreas and one from duodenum). Thirteen samples were from the same PNETs used for the microarray studies (Table 2). Total cell lysate (150 μg) was separated by SDS-PAGE (NuPAGE, Invitrogen) and transferred to PVDF membranes (Millipore, Billerica, MA, USA). Immunoblotting was performed with anti-platelet-derived growth factor receptor-β (PDGFR-β), anti-PDGFR-α, anti-phospho PDGFR-β Tyr716 (Upstate, Billerica, MA, USA), anti-phospho PDGFR-β Tyr751 (Sugen), and anti-β-Actin (Sigma, St Louis, MO, USA).

RET sequencing

The cDNA of seven WDECs and six GI-NETs was PCR amplified using three different primer sets spanning codons 573–666 (exons 10 and 11), 729–826 (exons 13 and 14), and 858–940 (exons 15 and 16). PCR products were purified (QIAquick gel extraction kit, Qiagen) and sequenced on an ABI 3730XL DNA analyzer (Applied Biosystems, Foster City, CA, USA). Primer sequences are available upon request.

Ret immunohistochemistry

Formalin-fixed paraffin-embedded samples of 21 cases of small intestinal NETs from Brigham and Women's Hospital and 65 cases of PNETs from Massachusetts General Hospital were assembled as part of a tissue microarray. Multiple independent cores from each sample were placed onto the microarray (range 2–6). Ret expression was assessed by immunohistochemistry with a Ret antibody (Santa Cruz, Santa Cruz, CA, USA) at 1:50 dilution after treatment with formic acid, as previously described (Lee et al. 2005). Ret staining was scored from 0 to 3+, and each core sample was scored separately. A papillary thyroid cancer sample was included as a positive control.

Statistical analysis

The P values were calculated utilizing the Wilcoxon rank-sum test with a P value <0.05 considered statistically significant.

Results

Differentially expressed genes in PNETs identified by cDNA microarray

Based upon the WHO criteria, 19 PNET samples were initially classified into 3 histologic groups: WDETs of benign behavior (n=3), WDETs of low-grade malignant behavior (n=9), and WDECs (n=7; Table 2). When comparing benign and low-grade malignant WDETs with WDECs, 112 genes were differentially expressed by at least 1.5-fold with a P value <0.05. Hierarchical clustering revealed two distinct clusters (Fig. 1): the 3 benign WDETs clustered together with 8/9 low-grade malignant WDETs and 1 WDEC (‘benign cluster’), and 6/7 WDECs clustered together with the 1 remaining low-grade malignant WDET (‘malignant cluster’).

Figure 1
Figure 1

Hierarchical clustering of genes in PNET pathogenesis reveals that malignant PNETs cluster apart from benign and low-grade malignant PNETs. Each row represents a cDNA clone on the Affymetrix chip and each column represents an individual tumor mRNA sample. Red represents overexpressed genes, and blue represents underexpressed genes. Benign WDET: samples 70, 72, 82; low-grade malignant WDET: samples 53, 55, 71, 74, 76, 80, 81, 85, 86; and malignant WDEC: samples 2, 56, 59, 61, 69, 83, 84.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

In the ‘malignant cluster’, 71 genes were up-regulated and 41 genes were down-regulated (Fig. 1). Supplementary Tables 1 and 2, which can be viewed online at http://erc.endocrinology-journals.org/supplemental/, contain lists of genes over- and under-expressed in the ‘malignant cluster’. GO analysis revealed that among the up-regulated genes in this cluster, the most frequent and statistically significant molecular function classifiers were ‘transcription regulator’ (11 genes, P=0.006) and ‘binding’ (46 genes, P=0.008; Fig. 2a). Of the genes with binding activity, 28 had protein-binding activity, 19 had ion-binding activity, 16 had nucleic acid-binding activity, 8 had nucleotide-binding activity, 4 had chromatin-binding activity, and 1 had antigen-binding activity (Fig. 2b).

Figure 2
Figure 2

(a) Gene ontology analysis reveals that the molecular function categories ‘transcription regulation’ and ‘binding’ are significantly overrepresented (P<0.05) among the set of up-regulated genes in the ‘malignant’ cluster of PNETs. (b) Detailed GO analysis for the molecular function category ‘binding’ among the set of up-regulated genes in the ‘malignant’ cluster of PNETs reveals several significantly overrepresented sub-categories (P<0.05, dark gray bars). The graphs display the negative log-transformed P values.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

Hierarchical clustering also revealed that genes located on chromosomes 11 and 17 were overrepresented in the 112 differentially regulated genes. Specifically, 8/112 (7.1%) of these genes were located on chromosome 11; 2 were up-regulated in the ‘malignant cluster’; and 6 were up-regulated in the ‘benign cluster’. Of these genes 10/112 (8.9%) were located on chromosome 17, and 8 were up-regulated in the ‘malignant cluster’ but only 2 were up-regulated in the ‘benign cluster’. Most of these genes (80%) were located on chromosome 17q.

Of note, a correlation was observed between mRNA expression and the hormonal profile of these tumors. Insulin mRNA levels were 18.6-fold higher in insulinomas compared with non-insulinomas, gastrin mRNA levels were 31.6-fold higher in gastrinomas compared with non-gastrinomas, and glucagon mRNA levels were 26-fold higher in glucagonomas compared with non-glucagonomas.

Validation of selected genes with quantitative real-time-PCR

The four most highly up-regulated genes in the ‘malignant cluster’ of PNETs (FEV, ADCY2, NR4A2, GADD45β) were selected for further validation by qRT-PCR. In the microarray studies, FEV was up-regulated 11.61-fold, ADCY2 was up-regulated 4.47-fold, NR4A2 was up-regulated 4.45-fold, and GADD45β was up-regulated 3.28-fold in the ‘malignant’ cluster. Statistically significant overexpression of all four genes was confirmed by qRT-PCR (FEV: 37-fold, P=0.007; ADCY2: 55-fold, P=0.026; NR4A2: 15.2-fold, P=0.0006; GADD45β: 5-fold, P=0.002; Fig. 3). In all cases, the microarray studies underestimated the extent of up-regulation.

Figure 3
Figure 3

Analysis of the four most up-regulated genes in WDECs (n=7) relative to WDETs (n=12) by qRT-PCR: (a) FEV, (b) ADCY2, (c) NR4A2, and (d) GADD45β. Values are normalized to three normal pancreas samples.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

Analysis of potential candidate genes

In addition to the identification of FEV, ADCY2, NR4A2, and GADD45β as novel genes that may play a role in tumor progression, we were curious whether specific candidate oncogenes and tumor suppressor genes such as MEN1, retinoic acid receptor-β, hMLH1, RASSF1, Her2/neu, cyclin D1, p16INK4a/p14ARF, p18INK4c, and p27Kip1 were differentially regulated. However, none of these genes was differentially regulated in our microarray study. In addition, angiogenic factors including aFGF, bFGF, or VEGF were not differentially regulated.

Another group of candidate genes are the receptor tyrosine kinases, which are frequently activated in human cancers. These are particularly attractive candidates, as tyrosine kinase inhibitors are promising as molecularly targeted agents. There was no statistically significant difference in expression of PDGFR-α or PDGFR-β, although there was a trend towards higher expression levels (2.3-fold increase) of PDGFR-α in WDECs compared with WDETs. However, given their potential clinical importance and potential biological relevance in neuroendocrine tumorigenesis, we performed immunoblot analysis to evaluate protein expression levels (Fig. 4). PDGFR-α was expressed in 94% of PNETs. It was present in 4/5 (80%) benign WDETs, 8/8 (100%) low-grade malignant WDETs, and 5/5 (100%) WDECs. It was also detected in 1/3 (33%) normal samples, although at much lower levels. PDFGR-β was expressed in 17/18 (94%) PNETs with no difference among tumor stages, and it was also expressed in 2/3 (66%) normal pancreatic samples.

Figure 4
Figure 4

Western blot analysis reveals that expression of PDGFR is common in PNETs and is observed both in early and late tumor stages. Furthermore, the PDGFR-β subunit is frequently activated by phosphorylation in PNETs. Benign WDETs: samples 63, 70, 72, 78, 82; low-grade malignant WDETs: samples 53, 57, 71, 74, 79, 81, 85, 86; malignant WDECs: samples 59, 61, 77, 83, 84; PDGRFα (p.c.), positive control for PDGRF-α; PDGFRβ (p.c.), positive control for PDGFR-β; pPDGFR, phospho-PDGFR-β; N, normal pancreatic tissue; T, tumor tissue; arrow, position of specific band.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

To evaluate PDGFR-β activation, the phosphorylation status at Tyrosine 751 or Tyrosine 716 was determined. Thirteen of eighteen (72.2%) PNETs were phosphorylated at Tyr751 (3/5 benign WDETs, 6/8 low-grade malignant WDETs, and 4/5 WDECs) and no phosphorylation was detected in any of the three normal samples (Fig. 4). Phosphorylation of Tyr716 was observed in 13/18 (72.2%) PNETs, occurring in 20% of benign WDETs, 87.5% of low-grade malignant WDETs, and 80% of WDECs. It was also seen in one of the three normal tissues. Overall, 15/18 (83%) PNETs demonstrated phosphorylation at one or both of these sites. Thus, the PDGFR-β subunit was frequently expressed and activated in PNETs, suggesting that tyrosine kinase inhibition of PDGFR may be a successful therapeutic approach.

Differentially expressed genes in GI-NETs versus PNETs identified by cDNA microarray analysis

The 25 malignant tumor samples were grouped into 6 GI-NETs and 19 PNETs (for sample details see Tables 2 and 3). Between the two groups 385 genes were differentially expressed by at least 1.5-fold with a P value <0.05. Hierarchical clustering revealed that GI-NETs clustered together in one group and PNETs in another (Fig. 5), indicating that gene expression patterns can indeed distinguish these NET subtypes. When compared with PNETs, 157 genes were up-regulated and 228 genes were down-regulated in the GI-NETs (Fig. 5). Supplementary Tables 3 and 4, which can be viewed online at http://erc.endocrinology-journals.org/supplemental/, illustrate the genes over- and under-expressed in GI-NETs. We also performed an analysis excluding samples in which only a liver metastasis was available and confirmed that GI-NETs did differ in their genetic signature from PNETs. The four remaining primary GI endocrine tumors clustered together with one WDEC and the other 18 PNETs represented another cluster (data not shown).

Figure 5
Figure 5

Hierarchical clustering of genes differentiates GI-NETs and PNETs. Each row represents a cDNA clone on the Affymetrix chip and each column represents a tumor mRNA sample. Red represents overexpressed genes, and blue represents underexpressed genes. Sample numbers for PNETs are the same as in Fig. 1, GI-NETs: samples 2974, 11898, 33762, 53456, 67494, 80670.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

GO analysis revealed that among the up-regulated genes in GI-NETs, the most statistically significant molecular function classifiers were ‘transporter’ (19 genes, P=0.00078) and ‘motor activity’ (3 genes, P=0.018; Fig. 6a). Of the up-regulated genes with transporter activity, there were 12 with ion transporter activity, 8 with channel or pore class transporter activity, 3 with ATPase activity, 1 with neurotransmitter transporter activity, and 1 with drug transporter activity (Fig. 6b). This is in sharp contrast to the group of malignant PNETs, where genes involved in ‘binding’ and ‘transcription regulation’ were the most statistically significant.

Figure 6
Figure 6

(a) Gene ontology analysis reveals that the molecular function categories ‘transporter’, ‘motor activity’ and ‘binding’ are significantly overrepresented (P<0.05) among the set of up-regulated genes in GI-NETs. (b) Detailed GO analysis for the molecular function category ‘transporter’ among the set of up-regulated genes in GI-NETs reveals several significantly overrepresented sub-categories (P<0.05, dark gray bars). The graphs display the negative log-transformed P values.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

Hierarchical clustering also revealed that chromosomes 1, 2, 5, 8, 9, 12, 14, 15, 18, and X were overrepresented among the 385 genes. Specifically, more genes on chromosomes 1, 5, 8, and 14 were overexpressed in GI-NETs compared with PNETs, and more genes on chromosomes 2, 9, 12, 15, 18, and X were overexpressed in PNETs compared with GI-NETs.

Validation of selected genes with quantitative real-time-PCR

The three most highly up-regulated genes in GI-NETs identified by the microarray studies (ECM1, VMAT1, and LGALS4) were verified by qRT-PCR. In addition, we analyzed RET because it is critical in the pathogenesis of medullary thyroid cancer, another NET type. In the microarray studies, ECM1 protein was up-regulated 28-fold, VMAT1 by 25-fold, galectin 4 (LGALS4) by 24-fold, and RET by 3.62-fold in GI-NETs compared with PNETs. qRT-PCR confirmed the up-regulation of all these genes in GI-NETs (ECM1: 39-fold, P=0.0011; VMAT1: 523-fold, P=0.0029; LGALS4: 43-fold, P=0.012; RET: 28-fold, P=0.012; Fig. 7). VMAT1 was not detectable in normal pancreatic tissue and most WDECs. Immunohistochemistry was performed for Ret on a larger series of small intestinal NETs and PNETs. There were 21 cases of small intestinal NETs (8 WDETs and 13 WDECs) and 65 cases of PNETs (14 benign WDETs, 43 low-grade malignant WDETs, and 8 WDECs) on the tissue microarrays. Among the intestinal NETs, 11% of the samples displayed weak or no Ret staining (0–1+), whereas 89% exhibited strong staining (2+–3+). By contrast, 65% of PNETs exhibited weak staining (0–1+) and only 35% exhibited strong staining (2+–3+; Fig. 8).

Figure 7
Figure 7

Analysis of the three most up-regulated genes and RET in GI-NETs (n=6) compared with malignant PNETs (n=7) by qRT-PCR. (a) ECM1, (b) VMAT1, (c) LGALS4, and (d) RET. Values are normalized to three normal pancreas samples.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

Figure 8
Figure 8

Representative immunohistochemical staining for Ret in PNETs (a ad b) and small intestinal NETs (c and d). In these PNET cases, Ret staining was scored as 0, and in the small intestinal NETs, Ret staining was scored as 3+.

Citation: Endocrine-Related Cancer 15, 1; 10.1677/ERC-07-0194

RET sequencing

RET is mutated in the MEN2 syndrome and familial medullary thyroid cancer. Because of this critical role in another NET type and the high levels of expression in GI-NETs, we sought to determine whether mutations in RET may also underlie GI-NET pathogenesis. Mutations occur primarily at three ‘hotspot’ regions within the cysteine-rich domain and the tyrosine kinase domains 1 and 2. DNA sequencing of these ‘hotspot’ regions in six GI-NETs and seven WDECs did not reveal any mutations. Incidentally, we identified two single base pair polymorphisms (CTT to CTA at codon 769 and TCG to TCC at codon 904), neither of which resulted in an amino acid change.

Discussion

The present study provides a comprehensive dataset of dysregulated genes in human PNETs and GI-NETs. We sought to identify differences in gene expression patterns between pancreatic WDETs and WDECs and to determine whether GI-NETs differ in their genetic profile from PNETs. Such a strategy may enhance our understanding of tumor pathogenesis and progression, as well as identify novel diagnostic markers and molecular targets for therapy.

Hierarchical clustering revealed that PNETs could be segregated on a molecular level into two groups. The ‘benign cluster’ comprised all benign WDETs, 8/9 low-grade malignant WDETs, and 1/7 WDECs. The ‘malignant cluster’ comprised 1/9 low-grade malignant WDETs and 6/7 WDECs. This is the first demonstration that the histologic subgroup of low-grade malignant WDETs shares more molecular similarities with benign WDETs than with WDECs and provides a molecular correlation of the WHO classification scheme. The clinical behavior of low-grade malignant WDETs is generally good and consistent with this clustering result. We cannot completely exclude an influence of the heterogeneity in the tissue samples on our results. However, due to the rarity of PNETs, it was unfeasible to perform gene expression analyses for each individual hormonal subtype, and it was hypothesized that there may be fundamentally similar mechanisms that underlie all tumor subtypes. Of note, within the group of low-grade malignant WDETs, only 55% were insulinomas, indicating that there was a diversity of tumor types at each stage analyzed.

GO analysis revealed that the molecular functions of ‘binding’ and ‘transcriptional regulation’ were significantly overrepresented in the malignant cluster, possibly reflecting novel pathways that are critical for tumor progression. In addition, genes on chromosomes 11 and 17 were overrepresented in PNETs. This is consistent with published comparative genomic hybridization (CGH) literature, which has shown that genomic gains are common on chromosome 17 (Terris et al. 1998, Speel et al. 1999, Stumpf et al. 2000) and frequently associated with malignant behavior (Speel et al. 2001).

The four most highly up-regulated genes in WDECs (FEV, ADCY2, GADD45β, and NR4A2) have not previously been implicated in the pathogenesis of PNETs. Two of these, GADD45β and NR4A2, regulate apoptosis. GADD45β can block apoptosis induced by IL-1β (interleukin-1β) in cultured islet cells (Larsen et al. 2006). FEV is a member of the ETS family of oncogenic transcription factors (Peter et al. 1997). Further functional studies will be necessary to determine the specific roles these genes may play in PNET pathogenesis and whether they may ultimately serve as novel therapeutic targets that have an impact upon patient management.

We then investigated whether certain target genes that have been previously implicated in PNET pathogenesis were differentially expressed. Immunoblot analysis revealed that PDGFR-α and -β were expressed in PNETs regardless of stage. More importantly, PDGFR-β was activated by phosphorylation in the majority of PNETs. Others have reported high levels of expression of PDGFR-α, PDGFR-β, and c-Kit in PNETs, but no assessment of receptor activation has been previously performed (Fjallskog et al. 2003). The possibility that PDGFR is expressed in mesenchymal components such as fibroblasts or pericytes (Pietras et al. 2003) cannot be excluded. Nevertheless, a specific PDGFR tyrosine kinase inhibitor may be a promising option for the treatment of PNETs. Observations of antitumor activity associated with receptor tyrosine kinase inhibitors further support a potential role for PDGFR in PNETs. In a multi-institutional study, treatment with sorafenib, a small molecule inhibitor with a spectrum of activity that includes VEGFR-2 and PDGFR-β, was associated with objective radiologic partial responses in 11% of PNET patients (Hobday et al. 2007). In a second study, treatment with sunitinib, which targets a similar spectrum of receptor tyrosine kinases, was associated with a 13% partial response rate in PNETs (Kulke et al. 2005). These observations, combined with our findings of PDGFR-β activation in PNETs, support further investigation of specific PDGFR-β inhibition as a clinical strategy in this tumor type. Interestingly, construction of a protein–protein interaction map revealed a novel connection between PDGFR-β and two highly expressed genes in PNETs, GADD45β, and NR4A2 (Supplementary Figure 1, which can be viewed online at http://erc.endocrinology-journals.org/supplemental/). Activation of PDGFR-β may therefore be involved in the regulation of apoptosis in PNETs.

We provide the first description that GI-NETs cluster separately from PNETs by microarray analysis. Although the GI-NETs in this study comprised both primary tumors (n=4) and metastases (n=2), we can exclude an influence of the heterogeneity of samples on our clustering result, as GI-NETs still clustered independently from PNETs when metastases were excluded from the analysis (data not shown). GO analysis revealed that in contrast to PNETs, genes involved in ion transport, channel transport, and neurotransmitter transport were significantly overrepresented in GI-NETs. This may provide new insights into the pathogenesis of GI-NETs. Hierarchical clustering also revealed that genes on chromosomes 9 and 18 were underexpressed in GI-NETs, possibly reflecting chromosomal deletions that have been reported in CGH and LOH (loss of heterozygosity) studies of these tumors (Kytola et al. 2001, Tonnies et al. 2001, Wang et al. 2005).

Of the three most highly up-regulated genes in GI-NETs (VMAT1, ECM1, and LGALS4), VMAT1 and LGALS4 have been previously described in this context (Nilsson et al. 2004, Vikman et al. 2005, Rumilla et al. 2006). Galectin 4 is expressed in the alimentary tract, where it is a component of adherens junctions or lipid rafts in the microvillus membrane (Huflejt & Leffler 2004) and is strongly expressed in ileal carcinoids (Rumilla et al. 2006). ECM1 is expressed in highly vascularized organs (Mongiat et al. 2003) and overexpressed in a number of malignant epithelial tumors (Kebebew et al. 2005). Finally, we demonstrated that RET, an oncogene encoding a transmembrane receptor tyrosine kinase, is up-regulated in GI-NETs. This observation was confirmed by immunohistochemistry of a large panel of intestinal and PNETs. Ret binds glial cell line-derived neurotrophic factor family members and activates MAPK/ERK, PI3K, JNK, p38MAPK, and phospholipase C γ (Arighi et al. 2005). Although no somatic mutations were identified, the high expression of RET in GI-NETs suggests that it may be an attractive therapeutic target. SU11248 is an inhibitor of multiple tyrosine kinases including RET (Kim et al. 2006), and a phase II study of SU11248 as a single agent in 39 patients with advanced GI-NETs revealed a 5% partial response rate (Kulke et al. 2005). RET may therefore play a pathogenic role in GI-NETs, but further investigation of targeted agents is required.

In comparison to published reports of gene expression profiles in NETs, our study has provided several new insights. Previous studies compared WDETs with normal islet controls (Maitra et al. 2003), MEN-1 associated NETs with normal islets (Dilley et al. 2005), PNETs with normal pancreas, pancreatitis, and pancreatic adenocarcinoma (Bloomston et al. 2004), non-functioning PNETs and their metastases with normal islets (Capurso et al. 2006), and metastatic with non-metastatic PNETs, primarily non-functioning (Hansel et al. 2004, Couvelard et al. 2006). Interestingly, there was no significant overlap between the identified genes in these studies and our current analysis. We hypothesize that this poor concordance is most likely a reflection of the different study designs, software platforms, data analysis parameters, and sample subtypes. In contrast to three reports that are most similar to ours (Hansel et al. 2004, Capurso et al. 2006, Couvelard et al. 2006), we studied a broader mix of PNET subtypes, not exclusively non-functioning PNETs, and this may potentially explain the disparity. With respect to technical differences, we utilized an Affymetrix platform, whereas Couvelard et al. obtained microarray chips from the Sanger center. In addition, we utilized the DChip program for data analysis, whereas Couvelard et al. performed their analysis with GeneSpring. Capurso et al. also utilized Affymetrix chips. However, their study differed significantly in that they compared PNETs with normal islets, whereas our comparison was between PNETs of different stages. Although Affymetrix chips were also used by Hansel et al. their analysis comprised only 12 tumors, whereas our analysis included 24 tumors. Nevertheless, it should be noted that there were some similarities, as one study also identified an up-regulation of PDGFR-β in WDECs (Couvelard et al. 2006). In addition, GO analysis in one study also revealed the molecular function classifier ‘binding’ as the most frequent in their up-regulated genes (Capurso et al. 2006). In aggregate, our results enhance the spectrum of genes implicated in NET pathogenesis.

In summary, we have identified a novel set of genes that may play a role in the pathogenesis and progression of PNETs and GI-NETs. Our results reveal a correlation with the WHO histologic classification on a molecular level. Furthermore, there are molecular signatures that distinguish PNETs from GI-NETs, reinforcing the principle that these two groups must be studied separately. By improving the molecular classification of these tumor subtypes, we may ultimately enhance our ability to predict tumor behavior, provide important new insights into the molecular biology and tumor pathogenesis, and design the next generation of targeted therapies. In this context, a potentially important role for PDGFR in the pathogenesis and treatment of PNETs has been revealed.

Acknowledgements

We would like to thank Changzhong Chen for assistance with data analysis. This work was supported in part by funding from the Stephen and Caroline Kaufer Fund for Neuroendocrine Tumor Research, Caring for Carcinoid Foundation, Verto Foundation, and Sugen. There are no conflicts of interest that would prejudice the impartiality of this work.

References

  • Arighi E, Borrello MG & Sariola H 2005 RET tyrosine kinase signaling in development and cancer. Cytokine and Growth Factor Reviews 16 441467.

  • Bloomston M, Durkin A, Yang I, Rojiani M, Rosemurgy AS, Enkmann S, Yeatman TJ & Zervos EE 2004 Identification of molecular markers specific for pancreatic neuroendocrine tumors by genetic profiling of core biopsies. Annals of Surgical Oncology 11 413419.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Capurso G, Lattimore S, Crnogorac-Jurcevic T, Panzuto F, Milione M, Bhakta V, Campanini N, Swift SM, Bordi C & Fave GD et al. 2006 Gene expression profiles of progressive pancreatic endocrine tumours and their liver metastases reveal potential novel markers and therapeutic targets. Endocrine-Related Cancer 13 541558.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Chung DC, Brown SB, Graeme-Cook F, Seto M, Warshaw AL, Jensen RT & Arnold A 2000 Overexpression of cyclin D1 occurs frequently in human pancreatic endocrine tumors. Journal of Clinical Endocrinology and Metabolism 85 43734378.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Couvelard A, Hu J, Steers G, O'Toole D, Sauvanet A, Belghiti J, Bedossa P, Gatter K, Ruszniewski P & Pezzella F 2006 Identification of potential therapeutic targets by gene-expression profiling in pancreatic endocrine tumors. Gastroenterology 131 15971610.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Dilley WG, Kalyanaraman S, Verma S, Cobb JP, Laramie JM & Lairmore TC 2005 Global gene expression in neuroendocrine tumors from patients with the MEN1 syndrome. Molecular Cancer 4 9

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Draghici S, Khatri P, Bhavsar P, Shah A, Krawetz SA & Tainsky MA 2003 Onto-Tools, the toolkit of the modern biologist: Onto-Express, Onto-Compare, Onto-Design and Onto-Translate. Nucleic Acids Research 31 37753781.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Duerr EM & Chung DC 2007 Molecular genetics of neuroendocrine tumors. Nucleic Acids Research 21 114.

  • Eisen MB, Spellman PT, Brown PO & Botstein D 1998 Cluster analysis and display of genome-wide expression patterns. PNAS 95 1486314868.

  • Evers BM, Rady PL, Sandoval K, Arany I, Tyring SK, Sanchez RL, Nealon WH, Townsend CM Jr & Thompson JC 1994 Gastrinomas demonstrate amplification of the HER-2/neu proto-oncogene. Annals of Surgery 219 596601.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Fjallskog ML, Lejonklou MH, Oberg KE, Eriksson BK & Janson ET 2003 Expression of molecular targets for tyrosine kinase receptor antagonists in malignant endocrine pancreatic tumors. Clinical Cancer Research 9 14691473.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Gansner ER & North SC 2000 An open graph visualization system and its applications to software engineering. Software – Practice and Experience 30 12031233.

  • Goebel SU, Iwamoto M, Raffeld M, Gibril F, Hou W, Serrano J & Jensen RT 2002 Her-2/neu expression and gene amplification in gastrinomas: correlations with tumor biology, growth, and aggressiveness. Cancer Research 62 37023710.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Guo SS, Wu X, Shimoide AT, Wong J & Sawicki MP 2001 Anomalous overexpression of p27(Kip1) in sporadic pancreatic endocrine tumors. Journal of Surgical Research 96 284288.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Guo SS, Wu X, Shimoide AT, Wong J, Moatamed F & Sawicki MP 2003 Frequent overexpression of cyclin D1 in sporadic pancreatic endocrine tumours. Journal of Endocrinology 179 7379.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Hansel DE, Rahman A, House M, Ashfaq R, Berg K, Yeo CJ & Maitra A 2004 Met proto-oncogene and insulin-like growth factor binding protein 3 overexpression correlates with metastatic ability in well-differentiated pancreatic endocrine neoplasms. Clinical Cancer Research 10 61526158.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Hobday TJ, Rubin J, Holen K, Picus J, Donehower R, Marschke R, Maples W, Lloyd R, Mahoney M & Erlichman C 2007 MC044h, a phase II trial of sorafenib in patients (pts) with metastatic neuroendocrine tumors (NET): a Phase II Consortium (P2C) study. Journal of Clinical Oncology, 2007 ASCO Annual Meeting Proceedings Part I 25 4504

    • PubMed
    • Search Google Scholar
    • Export Citation
  • House MG, Herman JG, Guo MZ, Hooker CM, Schulick RD, Cameron JL, Hruban RH, Maitra A & Yeo CJ 2003a Prognostic value of hMLH1 methylation and microsatellite instability in pancreatic endocrine neoplasms. Surgery 134 902908.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • House MG, Herman JG, Guo MZ, Hooker CM, Schulick RD, Lillemoe KD, Cameron JL, Hruban RH, Maitra A & Yeo CJ 2003b Aberrant hypermethylation of tumor suppressor genes in pancreatic endocrine neoplasms. Annals of Surgery 238 423431.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Huflejt ME & Leffler H 2004 Galectin-4 in normal tissues and cancer. Glycoconjugate Journal 20 247255.

  • Kamada T & Kawai S 1989 An algorithm for drawing general undirected graphs. Information Processing Letters 31 715.

  • Kebebew E, Peng M, Reiff E, Duh QY, Clark OH & McMillan A 2005 ECM1 and TMPRSS4 are diagnostic markers of malignant thyroid neoplasms and improve the accuracy of fine needle aspiration biopsy. Annals of Surgery 242 353361.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Kim DW, Jo YS, Jung HS, Chung HK, Song JH, Park KC, Park SH, Hwang JH, Rha SY & Kweon GR et al. 2006 An orally administered multitarget tyrosine kinase inhibitor, SU11248, is a novel potent inhibitor of thyroid oncogenic RET/papillary thyroid cancer kinases. Journal of Clinical Endocrinology and Metabolism 91 40704076.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Kloppel G, Perren A & Heitz PU 2004 The gastroenteropancreatic neuroendocrine cell system and its tumors: the WHO classification. Annals of the New York Academy of Sciences 1014 1327.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Kulke M, Lenz HJ, Meropol NJ, Posey J, Ryan DP, Picus J, Bergsland E, Stuart K, Baum CM & Fuchs CS 2005 A phase 2 study to evaluate the efficacy and safety of SU11248 in patients (pts) with unresectable neuroendocrine tumors (NETs). Journal of Clinical Oncology, 2005 ASCO Annual Meeting Proceedings 23 4008

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Kytola S, Hoog A, Nord B, Cedermark B, Frisk T, Larsson C & Kjellman M 2001 Comparative genomic hybridization identifies loss of 18q22-qter as an early and specific event in tumorigenesis of midgut carcinoids. American Journal of Pathology 158 18031808.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Larsen CM, Dossing MG, Papa S, Franzoso G, Billestrup N & Mandrup-Poulsen T 2006 Growth arrest- and DNA-damage-inducible 45beta gene inhibits c-Jun N-terminal kinase and extracellular signal-regulated kinase and decreases IL-1beta-induced apoptosis in insulin-producing INS-1E cells. Diabetologia 49 980989.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Lee S, Hong SW, Moon WC, Oh MR, Lee JK, Ahn CW, Cha BS, Kim KR, Lee HC & Lim SK 2005 High prevalence of c-RET expression in papillary thyroid carcinomas from the Korean population. Thyroid 15 259266.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Li C & Hung Wong W 2001 Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application. Genome Biology 2 RESEARCH0032

  • Maitra A, Hansel DE, Argani P, Ashfaq R, Rahman A, Naji A, Deng S, Geradts J, Hawthorne L & House MG et al. 2003 Global expression analysis of well-differentiated pancreatic endocrine neoplasms using oligonucleotide microarrays. Clinical Cancer Research 9 59885995.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Mongiat M, Fu J, Oldershaw R, Greenhalgh R, Gown AM & Iozzo RV 2003 Perlecan protein core interacts with extracellular matrix protein 1 (ECM1), a glycoprotein involved in bone formation and angiogenesis. Journal of Biological Chemistry 278 1749117499.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Muscarella P, Melvin WS, Fisher WE, Foor J, Ellison EC, Herman JG, Schirmer WJ, Hitchcock CL, DeYoung BR & Weghorst CM 1998 Genetic alterations in gastrinomas and nonfunctioning pancreatic neuroendocrine tumors: an analysis of p16/MTS1 tumor suppressor gene inactivation. Cancer Research 58 237240.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Nilsson O, Jakobsen AM, Kolby L, Bernhardt P, Forssell-Aronsson E & Ahlman H 2004 Importance of vesicle proteins in the diagnosis and treatment of neuroendocrine tumors. Annals of the New York Academy of Sciences 1014 280283.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Peri S, Navarro JD, Kristiansen TZ, Amanchy R, Surendranath V, Muthusamy B, Gandhi TK, Chandrika KN, Deshpande N & Suresh S et al. 2004 Human protein reference database as a discovery resource for proteomics. Nucleic Acids Research 32 D497D501.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Peter M, Couturier J, Pacquement H, Michon J, Thomas G, Magdelenat H & Delattre O 1997 A new member of the ETS family fused to EWS in Ewing tumors. Oncogene 14 11591164.

  • Pietras K, Sjoblom T, Rubin K, Heldin CH & Ostman A 2003 PDGF receptors as cancer drug targets. Cancer Cell 3 439443.

  • Rumilla KM, Erickson LA, Erickson AK & Lloyd RV 2006 Galectin-4 expression in carcinoid tumors. Endocrine Pathology 17 243249.

  • Shan L, Nakamura Y, Nakamura M, Yokoi T, Tsujimoto M, Arima R, Kameya T & Kakudo K 1998 Somatic mutations of multiple endocrine neoplasia type 1 gene in the sporadic endocrine tumors. Laboratory Investigation 78 471475.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Speel EJ, Richter J, Moch H, Egenter C, Saremaslani P, Rutimann K, Zhao J, Barghorn A, Roth J & Heitz PU et al. 1999 Genetic differences in endocrine pancreatic tumor subtypes detected by comparative genomic hybridization. American Journal of Pathology 155 17871794.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Speel EJ, Scheidweiler AF, Zhao J, Matter C, Saremaslani P, Roth J, Heitz PU & Komminoth P 2001 Genetic evidence for early divergence of small functioning and nonfunctioning endocrine pancreatic tumors: gain of 9Q34 is an early event in insulinomas. Cancer Research 61 51865192.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Stumpf E, Aalto Y, Hoog A, Kjellman M, Otonkoski T, Knuutila S & Andersson LC 2000 Chromosomal alterations in human pancreatic endocrine tumors. Genes, Chromosomes and Cancer 29 8387.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Terris B, Meddeb M, Marchio A, Danglot G, Flejou JF, Belghiti J, Ruszniewski P & Bernheim A 1998 Comparative genomic hybridization analysis of sporadic neuroendocrine tumors of the digestive system. Genes, Chromosomes and Cancer 22 5056.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Tonnies H, Toliat MR, Ramel C, Pape UF, Neitzel H, Berger W & Wiedenmann B 2001 Analysis of sporadic neuroendocrine tumours of the enteropancreatic system by comparative genomic hybridisation. Gut 48 536541.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Vikman S, Essand M, Cunningham JL, de la Torre M, Oberg K, Totterman TH & Giandomenico V 2005 Gene expression in midgut carcinoid tumors: potential targets for immunotherapy. Acta Oncologica 44 3240.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Wang EH, Ebrahimi SA, Wu AY, Kashefi C, Passaro E Jr & Sawicki MP 1998 Mutation of the MENIN gene in sporadic pancreatic endocrine tumors. Cancer Research 58 44174420.

  • Wang GG, Yao JC, Worah S, White JA, Luna R, Wu TT, Hamilton SR & Rashid A 2005 Comparison of genetic alterations in neuroendocrine tumors: frequent loss of chromosome 18 in ileal carcinoid tumors. Modern Pathology 18 10791087.

    • PubMed
    • Search Google Scholar
    • Export Citation

Y Mizukami is now at Asahikawa Medical College, Asahikawa, Japan

 

  • Collapse
  • Expand
  • Hierarchical clustering of genes in PNET pathogenesis reveals that malignant PNETs cluster apart from benign and low-grade malignant PNETs. Each row represents a cDNA clone on the Affymetrix chip and each column represents an individual tumor mRNA sample. Red represents overexpressed genes, and blue represents underexpressed genes. Benign WDET: samples 70, 72, 82; low-grade malignant WDET: samples 53, 55, 71, 74, 76, 80, 81, 85, 86; and malignant WDEC: samples 2, 56, 59, 61, 69, 83, 84.

  • (a) Gene ontology analysis reveals that the molecular function categories ‘transcription regulation’ and ‘binding’ are significantly overrepresented (P<0.05) among the set of up-regulated genes in the ‘malignant’ cluster of PNETs. (b) Detailed GO analysis for the molecular function category ‘binding’ among the set of up-regulated genes in the ‘malignant’ cluster of PNETs reveals several significantly overrepresented sub-categories (P<0.05, dark gray bars). The graphs display the negative log-transformed P values.

  • Analysis of the four most up-regulated genes in WDECs (n=7) relative to WDETs (n=12) by qRT-PCR: (a) FEV, (b) ADCY2, (c) NR4A2, and (d) GADD45β. Values are normalized to three normal pancreas samples.

  • Western blot analysis reveals that expression of PDGFR is common in PNETs and is observed both in early and late tumor stages. Furthermore, the PDGFR-β subunit is frequently activated by phosphorylation in PNETs. Benign WDETs: samples 63, 70, 72, 78, 82; low-grade malignant WDETs: samples 53, 57, 71, 74, 79, 81, 85, 86; malignant WDECs: samples 59, 61, 77, 83, 84; PDGRFα (p.c.), positive control for PDGRF-α; PDGFRβ (p.c.), positive control for PDGFR-β; pPDGFR, phospho-PDGFR-β; N, normal pancreatic tissue; T, tumor tissue; arrow, position of specific band.

  • Hierarchical clustering of genes differentiates GI-NETs and PNETs. Each row represents a cDNA clone on the Affymetrix chip and each column represents a tumor mRNA sample. Red represents overexpressed genes, and blue represents underexpressed genes. Sample numbers for PNETs are the same as in Fig. 1, GI-NETs: samples 2974, 11898, 33762, 53456, 67494, 80670.

  • (a) Gene ontology analysis reveals that the molecular function categories ‘transporter’, ‘motor activity’ and ‘binding’ are significantly overrepresented (P<0.05) among the set of up-regulated genes in GI-NETs. (b) Detailed GO analysis for the molecular function category ‘transporter’ among the set of up-regulated genes in GI-NETs reveals several significantly overrepresented sub-categories (P<0.05, dark gray bars). The graphs display the negative log-transformed P values.

  • Analysis of the three most up-regulated genes and RET in GI-NETs (n=6) compared with malignant PNETs (n=7) by qRT-PCR. (a) ECM1, (b) VMAT1, (c) LGALS4, and (d) RET. Values are normalized to three normal pancreas samples.

  • Representative immunohistochemical staining for Ret in PNETs (a ad b) and small intestinal NETs (c and d). In these PNET cases, Ret staining was scored as 0, and in the small intestinal NETs, Ret staining was scored as 3+.