Deregulation of genetic pathways in neuroendocrine tumors

Introduction

Neuroendocrine tumors (NET) are uncommon diseases occuring sporadically or in a familial context of autosomal dominant inherited syndromes such as Multiple Endocrine Neoplasia (MEN). During the last decade, at least six major genes involved in initial steps of NET have been characterized and one would expect that clinical screening syndromic forms of NET will include now genetic studies as a common tool for presymptomatic diagnosis (Calender, 2000). Four major MEN syndromes are MEN1 MEN2, von Hippel Lindau disease (VHL), and Carney Complex (CC) which represent the most common forms of inherited predisposition to NET with variable but high penetrance of proliferations in various neuroendocrine tissues. Less commonly, endocrine tumors of the pancreas , parathyroids and adrenal glands have been observed in phacomatosis such as Recklinghausen disease (NF1) , Tuberous sclerosis (TSC). Lastly, familial occurrence of a single endocrine lesion, such as primary hyperparathyroidism or pituitary adenoma have been identified as putative new genetic diseases fo which genetic pathways remain to be identified. Most NET predisposing diseases were related to inactivation of growth suppressor genes, according to the Knudson model, except MEN2, an inherited form of medullary thyroid carcinoma, which occur through dominant activation of a proto-oncogene, the RET tyrosine kinase receptor. Nevertheless, recent knowledges on the biological roles of most genes involved in cancer predisposition teaches us that a single gene might have pleiotropic effects at various stages of the cell physiology as for instance both a negative and a positive regulator function depending on the cellular context and sometimes the type of mutation (Zheng et al., 2000).

Cloning of genes involved in genetic diseases predisposing to NET has led to a better insight in the molecular nature of tumor initiation of (neuro)endocrine tissues in most endocrine glands. Even if the genes and related syndromes are presented separately, we would expect that genetic studies in patients with NET will now help to the differential diagnosis of syndromic diseases and that a single type of endocrine tumor may be related to the deregulation of distinct genes and related pathways. Major syndromes predisposing to NET and overlapping symptoms leading in some patients and/or families to adress the differential diagnosis by molecular genetic tools are presented in Figure 1.

Figure 1 : Overlapping lesions in genetic syndromes predisposing to neuroendocrine tumors. Major endocrine lesions described are MTC (medullary thyroid carcinoma), PHEO (pheochromocytoma), pHPT (primary hyperparathyroidism), pNET (pancreatic endocrine tumors), PiT (anterior pituiray tumors), cADR (adrenal cortical tumors), NET (bronchic or thymic 'carcinoids'), THYR (thyroid papillary or follicular carcinoma), NecT (neuroectodermic-derived tumors such as meningioma, paraganglioma, ependymoma), Carc midgut and hindgut carcinoids). Major genetic syndromes showed are MEN1 (Multiple Endocrine Neoplasia type 1), MEN2 (Multiple Endocrine Neoplasia type 2), VHL (von Hippel Lindau disease), NF-1 (Recklinghausen disease), TSC (Tuberous Sclerosis) and Fam. Isolated which represent genetic diseases with occurrence of a single and homogeneous lesion in probands and first or second degree relatives, such as familial isolated MTC or hyperparathyroidism. The most probable diagnosis suggested for each lesion are indicated inside the boxes. When designed by ? , the genes involved in these familial occurrences are unknown to date.

MEN1, a major genetic context predisposing to NET

From the disease to the gene

Multiple Endocrine Neoplasia type 1 (MEN1, OMIM 131100) is an autosomal dominant syndrome characterized by hyperplasia and/or multiple tumors of the parathyroid, endocrine pancreas, anterior pituitary, foregut-derived neuroendocrine tissues and adrenal cortical glands (Sheperd, 1991). Less common lesions have been associatedi wth MEN1 including cutaneous proliferations such as angiofibroma, collagenoma, lipoma and melanoma (Darling et al., 1997; Nord et al., 2000) and peripheral or central nervous system tumors such as ependymoma (Kato et al., 1996; Giraud et al.). The MEN1 locus has been localized to chromosome 11q13 in 1988 by linkage studies and the MEN1 gene itself was identified in 1997 after a 9 years-long period of extensive positionnal cloning (Larsson et al., 1988; Chandrasekharappa et al., 1997; The European Consortium on MEN1, 1997). The MEN1 gene consists of 10 exons, the first of which is untranslated, spanning around 7-10 Kb of genomic sequence and encoding menin, a protein of 610 amino acids. A major menin 2,8 Kb transcript and the protein were expressed in most human tissues analyzed, the menin RNA presenting a striking 4,2 Kb large form suggesting alternative promotor triggering and/or alternative splicing in 5', a question remaining to date unresolved (The European Consortium on MEN1, 1997).

Menin does not reveal homologies to any other known proteins nor includes consensus motifs from which the putative function of the protein could be deduced. Nevertheless, menin was characterized as a nuclear protein containing two nuclear localization signals (NLS) in exon 10 (Guru et al., 1998) which have been functionnally defined by in vitro expression of deletion constructs labelled by epitope-tagging with enhanced green fluorescent protein (EGFP). Despite the nuclear localization suggesting a major role of menin in the regulation of genome expression, the MEN1 protein was recently shown to move from nucleus to cytoplasm during the cell cycle and mainly the mitotic process (Huang et al., 1999). These data contrast with the absence of any variation of size and amount of menin throughout the cell cycle as studied in synchronized cells (Wautot et al., 2000).

Major function of menin in normal and MEN1-related conditions

Major function of the MEN1 gene product might be related to the ability of menin to bind JunD, a transcription factor belonging to the multimeric AP1 transcription regulation complex (Agarwal et al., 1999). Deletion constructs and natural mutants of menin have shown that interaction with JunD needs three major domains of the menin protein, the first 40 aminoacids of the N-terminal region and two central sequences at positions 139-242 and 323-428 respectively. The menin-JunD interaction seemed to be specific and established through the N- terminal sequences (amino acids 1-70) of JunD and a co-activator, JAB1, involved in the transcriptionnal regulation activity of JunD in the AP1 complex (Agarwal et al., 1999). Specific JunD missense mutants in the N- terminal domain failed to bind menin with a subsequent activation of their transcriptionnal activity suggesting escape of menin control (Knapp et al., 2000). Wild-type menin repressed transcriptionnal activation mediated by JunD upon in vitro co-transfection assays. Independantly, some authors have confirmed these data and showed that menin mediated repression of JunD activity was related to histone deacetylation mechanisms which are involved in chromatin-mediated regulation of gene expression (Gobl et al., 1999).

Despite extensive characterization of menin-JunD interaction, the physiopathogeny of MEN1-related endocrine and non- endocrine lesions remains poorly understood, mainly considering the fact that AP1 is an ubiquitous factor in all tissues which regulates most of cellular process such as, mitosis, DNA replication, transcription, apoptosis and response to physical or chemical stress (Calender, 2000). JunD was considered as a transcriptionnal activator but paradoxical observations showed that overexpression of JunD in NIH3T3 cells suppresses cell growth. This suggests that JunD might be both a co-repressor and a co-activator in AP1 depending the cellular context. Recent studies showed that overexpression of menin in ras- transformed NIH3T3 cells inhibits cell growth supporting the hypothesis that menin is a growth-suppressor through complex interactions within the JunD-AP1 complex (Kim et al., 1999).

More than 300 unique germline mutations have now been identified in MEN1 family probands analyzed throughout the world (Agarwal et al., 1997; Shimizu et al., 1997; Bassett et al., 1998; Giraud et al., 1998; Teh et al., 1998; Poncin et al., 1999; Mutch et al., 1999). Somatic mutations have also been reported in sporadic forms of endocrine tumors with variable incidence of 20-30% in parathyroid, endocrine pancreas ( gastrinomas, insulinomas), lung carcinoids and less than 1% in pituitary and adrenal cortical tumors (Heppner et al., 1999; Toliat et al., 1997; Zhuang et al., 1997; Debelenko et al., 1997; Zhuang et al., 1997; Prezant et al., 1998). All these mutations were recently included in an unpublished mutation database developped in the frame of UMD (Universal Mutation database) software allowing an easy evaluation of genotype-phenotype corralations (Beroud et al., 2000). Over 70% of germline mutations related to the MEN1 disease are nonsense and frameshifts predicting truncation and/or absence of the abnormal protein. Missenses, in-frame deletions or insertions and splice-site alterations account for the remaining ♠ 30% mutations described in the clinical context of MEN1. A putative effect of truncating mutations might be a premature degradation of the truncated menin through the protein catabolism pathway as suggested by the failure of Western-Blot detection of mutant protein in most cases analyzed (Wautot et al., 2000)]. This observation do not exclude a transient and abnormal effect of the truncated menin within the nucleus and cytoplasm, a question which will be further adressed by immunohistochemical competent monoclonal antibodies against menin. Intronic and splice-site mutations were shown to alter RNA splicing with abnormal exon skipping and/or intronic retention (Mutch et al., 1999; Engelbach et al., 1999; Roijers et al., 2000).

Homologs of MEN1 sequences in other species

The murine Men1 gene was mapped and cloned in the pericentromeric region of murine chromosome 19 which is syntenic to human 11q13 [37]. The genomic organization of Men1 is similar to that of the human gene including 10 exons and a noncoding region covering ♠ 6,7 Kb of genomic DNA. Suggestive of what might happen in humans, two major transcripts of 2,8 and 3,2 Kb were detected in most embryonic and adult tissues, resulting from alternative splicing of intron 1 (Bassett et al., 1999). The predicted protein is 611 amino acids in length and characterized by 97% homology with the human sequence. Two independant teams subsequently cloned the murine and rat Men1 and showed similar data on structure and expression (Bassett et al., 1999; Karges et al., 1999). The expression of Men1 is detected as earlier as gestationnal day 7 in the whole embryo, and a strong expression limited to thymus, liver, nervous system and gonads at day 17. Hence, the expression of Men1 was not confined to organs affected in MEN1. In testis, Men1 expression was found to be higher in Sertoli cells than in germ cells (Stewart et al., 1998). These data assess independant observations suggesting that menin-JunD related pathways within the AP1 complex might directly control the transition of granulosa cells to terminally differentiated, non-dividing luteal cells in ovarian gland (Sharma et al., 2000). Lastly, the zebrafish and drosophila homologs of MEN1 have been recently identified and show respectively ♠ 75% and ♠ 47% homologies with the human sequence (Khodaei et al., 1999; Maruyama et al., 2000). Among the amino acid residue substitutions reported as disease-associated missense mutations, most of them (70%) were completely conserved either in zebrafish and drosophila menin, indicating an evolutionnary conserved protein with a fundamental role in biological processes.

Animal model of MEN1 and pathogenic effect of MEN1 gene inactivation

According to the Knudson model, MEN1 appears to be a growth suppressor gene with tumors in MEN1 affected patients showing somatic loss of the wild-type allele, so-called loss of heterozygosity (LOH) at 11q13 (Bystrom et al., 1990; Knudson, 1971). Functionnally, all truncating mutations affect one or both NLS in the C-terminal domain of menin. Conversely, missense mutations have never been observed inside the menin NLSs sequences, suggesting that the functionnal role of these sequences might be critical for cell survival and mainly at the embryonic level. Recently, the first mouse model of MEN1 was produced through homologous recombination of the mouse Men1 gene (Crabtree et al., 2001). A major mice strain lacking exons 3-8 of the Men1 gene was produced and heterozygous Men1+/- were bred to generate Men1-/- homozygotes. Homozygous inactivation of the Men1 gene was lethal at around days 10-14 of embryogenesis with developmental delay, defects in cranial and facial developments. Interestingly, Men1+/- heterozygotes develop hyperplastic pancreatic islets and first small tumors after 9 months of age and significantly larger tumors with capsular invasion after 16 months of age. Other tumors observed in heterozygous Men1+/- mices were parathyroid hyperplasia/adenoma, pituitary adenoma and adrenal cortical carcinomas in 24, 26 and 20% of cases analyzed throughout the course of the time study. Most of the tumors observed in this animal model showed loss of heterozygosity on the wild-type allele (Crabtree et al., 2001).

Clinical implications of MEN1 gene analysis

Taken together, updated data on MEN1 suggest that inactivation or absence of both MEN1 alleles is a critical factor in MEN1-related endocrine tumors initiation. Identification of nonsense or any other truncating mutations in a MEN1 suspected patient clearly assess the diagnosis. With missense or intronic mutations, we might consider that pathogenic effects may be related to disturbances in menin-JunD (or forthcoming new functionnal interactions) related pathway or abnormal splicing respectively. In such cases, in vitro functionnal tests will be required to assess for instance that an amino acid substitution must be considered as a mutation and not a rare polymorphism. Clinical criteria used for diagnosis of MEN1 are crucial and in our experience, MEN1 germline mutations were found in 95% of probands/families with as a patient sharing at least three major lesions of the syndrome and a first-degree relatives affected by one (or more) MEN1-related lesions (Giraud et al., 1998). Most families without demonstrable MEN1 mutations display an atypical clinical pattern, which might suggest genetic heterogeneity of the disease or the occurrence of phenocopies with lesions which are commonly observed in the non-MEN1 individuals, such as primary hyperparathyroidism and prolactinoma (Teh et al., 1987; Stock et al., 1997). Lastly, when MEN1 is strongly suspected, mutations might occur in unknown part of the MEN1 sequence, as the 5' regulatory region for which functionnal characterization is in progress (Khodaei-O'Brien et al., 2000). Large intragenic rearrangements and/or deletions, either within or encompassing the MEN1 gene might have been missed by routine PCR and sequencing procedures. A MEN1 deletion has been suggested in a japanese MEN1 pedigree by quantitative Southern-Blot analysis and shown recently in a large french MEN1 family using molecular cytogenetics procedures (Kishi et al., 1998; Lespinasse et al, in preparation). Finally, we might conclude that clinical screening of patients remains a prerequisite of genetic analysis and that functionnal knowledges on menin-related pathways must be kept in mind of clinicians following MEN1 affected patients.

Oncogenic activation of RET, a tyrosine-kinase membrane receptor, induces MEN2

Clinical expression of RET germline mutations

The two major variants of Multiple Endocrine Neoplasia type 2 (MEN2A or Sipple's syndrome, OMIM 171400 ; MEN2B or Gorlin syndrome, OMIM 162300) result from missense activating (or oncogenic) mutations of RET, a gene localized on chromosome 10q11-2 and encoding a transmembrane tyrosine-kinase (TK) receptor (Donis-Keller et al., 1993; Mulligan et al., 1993). MEN2 might be considered as the inherited form of medullary thyroid carcinoma (MTC) , a constant lesion in MEN2, associated or not with pheochromocytoma and/or primary hyperparathyroidism. MEN2B is a rare variant characterized by an early-occurrence and aggressive MTC, pheochromocytoma, mucosal neuromas on the gastrointestinal tract and a marfanoid habitus. FMTC (Familial isolated Medullary Thyroid Carcinoma) might be considered as the classical MENA syndrome with very low penetrance of pheochromocytoma and hyperparathyroidism (Farndon et al., 1986; Moers et al., 1996).

Functionnal insights on the c-ret protein

RET genomic size is 60 Kb and the gene contains 21 exons (Takahashi et al., 1988). The c-ret protein is characterized by an extracellular cysteine-rich homodimerization domain and an intracellular TK catalytic site. The N-terminal part of the extracellular region contains a cadherin-like domain which mediates calcium-dependant cell-cell adhesion. This receptor bounds at least four ligands, GDNF (Glial cell line Derived Neurotrophic Factor), neurturin, artemin and persephrin, all of them inducing homodimerization of the c-ret protein through the cysteine-rich region, thereby triggering the TK catalytic site (Airaksinen et al., 1999; Taraviras et al., 1999). Intracellular events after ligand binding and c-ret dimerization involve cross-phosphorylation of TK domains and the downstream activation of Ras-MAP-kinase and PI3/AKT transduction pathways (Ohiwa et al., 1997; Segouffin-Cariou and Billaud, 2000). Biological properties of c-ret and ligands is related to the genesis of the peripheral and central nervous sytem and renal excretory tract. Strikingly, both GDNF -/- and c-ret -/- knock-out mices share the same phenotype with an early death after birth, lack of neurons in the whole digestive tract and kidney agenesis (Schuchardt et al., 1994; Sanchez et al., 1996; Enomoto et al., 1998). The c-ret protein is highly expressed in many normal endocrine tissues, developping kidney and in human tumors of the neural crest derivatives, such as MTC, neuroblastoma and pheochromocytoma (Nakamura et al., 1994; Avantaggiato et al., 1994). Recently, it has been proven that a constitutively activated RET-MEN2A allele promotes cell survival in vitro in the absence of any growth factors and that this effect might be controlled by a specific domain around c-ret tyrosine 1062 through the PI3/AKT / MAP kinase pathways (De Vita et al., 2000).

Germline RET mutations in MEN2 and downstream deregulated pathways

All germline mutations of RET observed in MEN2 were missense mutations affecting either the cystein-rich extracellular dimerization domain (exons 8 to 13) or intracellular TK catalytic sites (exons 15 and 16) (Pasini et al., 1996; Eng et al., 1996). Genotype- phenotype correlations have been clearly established, mutations occuring in exons 8 to 14 being mostly related to MEN2A and FMTC variants, mutations observed in exons 15 and 16 being always related to a MEN2B phenotype. Conversely, transgenic mice expressing an activating MEN2A-related mutation in exon 11 develop bilateral MTC as in humans but never pheochromocytoma (Michiels et al., 1997; Kawai et al., 2000). Mutations occuring in exons 8 to 11 induce spontaneous homodimerization of the c-ret protein in absence of the ligand(s) (Santoro et al., 1995). Mutations described in exon 13 and 14 affect the catalytic site by inducing an inappropriate binding to substrates of the intracellular signalling pathway (Santoro et al., 1995; Rossel et al., 1997). Lastly, MEN2B-related mutations in exons 15 and 16 switch the substrate specificity of the c-ret TK from a membrane receptor towards an intracellular tyrosine-kinase, thus inducing abnormal signalling pathways (Santoro et al., 1995; Iwashita et al., 1999).

Clinical implications of RET sequencing in MEN2

To date, germline RET mutations were found in ♠ 100% of MEN2A, ♠ 90% of FMTC and ♠ 95-100% of MEN2B families. In terms of clinical use, we might conclude that routine screening of exons 8, 10, 11, 12, 13, 14, 15 and 16 of RET is now an useful tool for an accurate presymptomatic diagnosis in this disease. Germline RET mutations were also found in 5-7% of sporadic cases of MTC (Eng et al., 1995; Wohllk et al., 1996), mostly in patients for which clinical and genetic informations have been less informative at initial diagnosis, suggesting de novo mutations or missed familial cases. MEN2 syndrome represents a model in genetic predisposition to cancer by the fact that molecular genetic analysis of RET might be considered as an early preventive action, leading to prophylactic thyroidectomy in (young) asymptomatic gene-carriers (Lallier et al., 1998).

Mutations in the VHL gene predisposes to pancreatic NET

Clinical features

In a recent series of 158 patients affected by VHL (von Hippel Lindau or OMIM 193300) disease, Hammel et al showed pancreatic involvement in 77% of cases, including true cysts, serous cystadenomas and neuroendocrine tumors in 12% of VHL patients (Hammel et al., 2000). None of the patients with pancreatic NET had symptoms of hormonal hypersecretion suggesting that VHL-related pancreatic NET are mostly non-functionnal. VHL-related pancreatic NET might be distinguished from MEN1-related tumors based on 1) the absence of primitive duodenal tumors 2) frequent non-functionnal lesions with focal positivity for pancreatic polypeptide, somatostatin, glucagon and insulin 3) a clear-cell morphology related to intracytoplasmic lipid and myelin accumulation and 4) the frequent occurrence of microcystic adenomas around the clear-cell tumors. VHL disease is an autosomal dominant disease which main lesions are retinal hemangioblastomas, retinal hemangioblastomas and/or cerebellar hemangioblastomas, renal cancers, pheochromocytoma, pancreatic lesions and auricular endolymphatic epithelial proliferations (Richard et al., 1998; Couch et al., 2000).

VHL gene and function

The VHL gene was cloned in 1993 on chromosome 3p35-36 and contains only three exons encoding a protein with pleiotropic functions (Latif et al., 1993). The VHL protein interacts with the elongin family involved as regulators of transcriptionnal elongation (Aso et al., 1995; Kibel et al., 1995). Other functions of the VHL protein have been related to the hypoxia-induced cell regulation by enhanced expression of VEGF (Vascular Endothelial Growth Factor), regulation of extracellular matrix fibronectin expression and stabilization and lastly, VHL protein interacts with Cullin-2 and Rbx1, two major components of the cellular ubiquitination or protein catabolism machinery (Mukhopadhyay et al., 1997; Ohh et al., 1998; Iwai et al., 1999). Many naturally occuring mutations in VHL have either been shown or are predicted to abrogate assembly with elongins and Cullin-2, suggesting a functionnal role for these interactions in VHL-related tumors and digestive NET (Bonicalzi et al., 2001; Libutti et al., 2000).

Clinical implications of VHL gene analysis

VHL gene sequencing has been useful in VHL disease presymptomatic diagnosis and in some clinical states suggesting differential diagnosis with MEN1 and MEN2. In fact, germline mutations of the VHL gene have been shown in patients with pancreatic endocrine tumors without any lesions suggesting MEN1 (Aubert-Petit et al., 1999) and in rare cases of familial occurrence of pheochromocytomas (Garcia et al;, 1997).

Forthcoming genes involved in neuroendocrine tumors through the Carney Complex

Carney complex (CC or OMIM 160980) is an autosomal dominant disease predisposing to various types of tumors including cardiac and cutaneous myxomas, spotty pigmentation of the skin and nonneoplastic hyperfunctioning endocrine states, as nodular adrenocortical hyperplasia associated with Cushing syndrome, pituitary and thyroid adenomas (Carney et al., 1985). Clinical evaluation and genetic linkage analysis of families affected by Carney complex suggested at least two distinct loci for disease genes, the first on chromosome 2p16, the second on chromosome 17q24 (Stratakis et al., 1996; Casey et al., 1998). Recent positional cloning studies demonstrate that 17q-linked Carney disease was caused by mutations in the R1α regulatory subunit of cAMP-dependant protein kinase A (PKA), so- called the PPKAR1αgene (Kirschner et al., 2000; Casey et al., 2000). Most mutations were nonsense or frameshifts inducing haploinsufficiency of the PPKAR1α subunit. This protein acts as a tumor suppressor gene in most tissues by down-regulating the PKA activity. Loss of the PPKAR1α regulatory subunit promotes cell proliferation and growth of benign tumors in multiple tissues. Hyperendocrine states are likely to be a result of increased PKA activity and cAMP levels by disruption of the tetrameric inactive form of the enzyme (McKnight et al., 1998). The second gene involved in Carney complex and localized on 2p16 remains to be identified.

Discussion

Apart from well-defined genes involved in genetic predisposition to NET, we were not able to show a common pathway leading to the proliferation of neuroendocrine cells. We might expect that various mechanisms occur depending the embryological origins of the organs concerned. Figure 2 summarizes the mechanisms related to MEN1, MEN2, VHL, CC and NF1/TSC mutations in the genetic syndromes described previously. The genetic mechanisms underlying malignant progression of benign neuroendocrine tumors remain unknown. Some genes, as NF1 (Neurofibromatosis type 1) and TSC1/2 (Tuberous Sclerosis) have been suggested as putative loci involved in tumoral progression from the fact that both genetic syndromes related to mutations in these genes, either Recklinghausen Neurofibromatosis (NF1) and Tuberous Sclerosis (TSC) might rarely predispose to endocrine tumors such as pheochromocytomas, primary hyperparathyrodism and pancreatic somatostatinomas or insulinomas (Van Basten et al., 1994; Kim et al., 1995). The NF1 (chromosome 17) and both TSC1 (chromosome 9) and TSC2 (chromosome 16) genes are involved in the membrane signal transduction pathway by acting as negative regulators of ras (NF1) or rab-5 (TSC1/2)-related small G-proteins (van Slegtenhorst et al., 1997; Xiao et al., 1997). NF1, TSC1/2 disrupt GTP-ras or rab-5 complex and are considered as GTPase-activating proteins (GAP), an independant class of growth-suppressor genes.

Figure 2 : Schematic summary of the complexity of pathways involved in genetic predisposition to neuroendocrine tumors. All proteins, membrane receptors, pathways and interactions are detailed in the text. When designed by +, the gene involved in the syndrome (such as MEN2) acts as an oncogene with a dominant mechanism. When designed by -, the genes related to syndromes (such as MEN1, NF1, TSC, Carney complex and VHL) are considered as growth suppressors, acting by negative regulation of various pathways shown in this figure.

Most cancers result from multistep scenarios and we might consider that many other genes are involved in the progression of benign NET towards a fully malignant phenotype. Mutations of well-known growth-suppressor or onco-genes, such as p53, Ki-RAS , HER2/NEU, C-MYC, N-MYC , N-RAS , C-JUN, PRAD-1, have been excluded as major events in NET progression even if some studies showed upregulation of gene expression in tumor tissues (Calender, 2000). Up-regulation of genes such as bcl-2, an apoptosis-regulating function and p53 or down or de-regulation of adhesion molecules such as CD44 have been suggested to be of importance as prognostic markers in pancreatic and bronchial carcinoid tumors (Granberg et al., 2000). Nuclear nm23 and Ki-67 proliferation-associated markers have also been considered as useful tools for valuable prognostic information and identification of patients at risk of disease-related death (Farley et al., 1993; Solcia et al., 1998). Nevertheless, more hopes on this topic remain related to the identification of loci-specific loss of heterozygosity (LOH) mainly on chromosomes 1p and 1q (Chung et al., 1997), 3p25-26 (Ebrahimi et al., 1999; Chung et al., 1997), 11q13 and 18q (Chakrabarti et al., 1998; Hessmann et al., 1999). Loss of the MEN1 loci might induce latent genomic and/or chromosomal instability as suggested by a few reports but the basic mechanism leading to this instability has not been proven to date (Calender, 2000). Most of the genes involved in such LOH's remain to be identified and their implications in neuroendocrine cell proliferation must be demonstrated by in vitro functional tests. Malignant progression of neuroendocrine tumors might also be triggered by overexpression of growth factors involved in endocrine and endothelial cell proliferation such as TGFα, EGF, NGF and VEGF/VEGF-related factors ( Nilsson et al., 1995; Bold et al., 1995; Grimmond et al., 1996; Liu et al. 1995).

Finally, most of the genes involved in the initiation of endocrine tumors have been now discovered and are involved in genetic predisposition to NET. In sporadic and malignant forms of neuroendocrine tumors, other genes remain to be found and their respective roles during the multistep progression of tumors to be identified. Genetic studies have contributed to the development of animal models of neuroendocrine tumors in the context of genetic syndromes closed to what have been observed in humans. This represents a powerful tool for therapeutical assays in the future and mainly by understanding the mechanisms leading a normal endocrine cell towards a fully malignant and metastatic clone.