Pituitary Tumor Pathogenesis1

Pituitary tumors are mostly benign monoclonal adenomas arising from adenohypophyseal cells in the anterior pituitary. These tumors usually autonomously express and secrete pituitary hormones, leading to endocrinological syndromes, such as hyperprolactinemia (most common), acromegaly, Cushing’s disease, and, rarely, hyperthyroidism. Alternatively, they may be functionally silent and initially diagnosed as an expanding sellar mass resulting in hypopituitarism, usually with central hypogonadism, visual field defects, or headaches. Although our understanding of subcellular mechanisms for the pathogenesis and progression of these common tumors is incomplete, during recent years important new information on the role of intrinsic pituitary genetic alterations, disordered transcription factors and growth factors, and signaling proteins involved in pituitary tumorigenesis has been reported in both sporadic and hereditary adenomas (Table 1). Genetic alterations in pituitary tumors Genetic alterations in pituitary tumors As normal pituitary function is under tight hypothalamic control mediated by the effects of releasing and inhibiting hormones, it has been suggested that hypothalamic hormones may be involved in pituitary tumorigenesis. GHRH, for example, induces somatotroph cell proliferation (1), and ectopic GHRH-secreting tumors (bronchial carcinoids, small cell lung carcinomas, and pancreatic islet cell tumors) and eutopic hypothalamic GHRH-producing tumors (hamartomas and gangliocytomas) result in somatotroph stimulation, hyperplasia, and consequent GH hypersecretion, but rarely in true somatotroph adenoma formation (2). Thus, the majority of these patients do not develop pituitary adenomas. In addition, patients with Cushing’s syndrome caused by ectopic CRH oversecretion from an intrasellar gangliocytoma (3) or prostate carcinoma (4) did not develop corticotroph adenoma despite the presence of corticotroph hyperplasia. Moreover, histological studies of pituitary adenomas clearly reveal distinct tumor borders that are not surrounded by hyperplastic pituitary tissue (5). Furthermore, hormonal secretion from the adenomatous cells is usually independent of physiological hypothalamic control, and surgical resection of small, well defined tumors usually results in definitive short term anatomical and functional cure of pituitary adenomas. The occurrence of mixed plurihormonal tumors originating from primitive pluripotent progenitor cells further supports a de novo pituitary origin for these tumors. Thus, these persistent latter observations strongly suggest that pituitary tumors are derived from an intrinsic pituitary cell defect leading to monoclonal expansion of a single transformed cell, rather than from excessive polyclonal proliferation due to generalized hypothalamic overstimulation. However, hypothalamic hormones and other local growth factors may have an important role in promoting the growth of already transformed pituitary cell clones and the expansion of small adenomas into large or invasive tumors. The clonal origin of pituitary adenomas was determined by X-chromosomal inactivation analysis in female patients heterozygous for variant alleles of the X-linked genes hypoxanthine phosphoribosyltransferase and phosphoglycerate kinase. Using restriction fragment length polymorphisms and differential methylation patterns in these genes and other polymorphic locations that are dependent on whether the gene is active or inactive (6), monoclonality was confirmed in nonfunctioning pituitary tumors (7, 8) as well as in GH-, PRL- (8), and ACTH-secreting adenomas (8–11). Thus, all cells in each tumor contained the same inactivated X-chromosome allele. In contrast, normal pituitary (8) and corticotroph hyperplastic tissue (10) were, not unexpectedly, found to be polyclonal, as these tissues contained an equal number of cells containing either the paternal or maternal activated allele. These data strongly indicate that both secreting and nonsecreting pituitary adenomas result from clonal expansion of a single mutated pituitary cell. However, tumor shrinkage after administration of dopamine agonists to prolactinomas (12) or of somatostatin analogs to GH cell adenomas (13) and, conversely, the rapid tumor enlargement of ACTH-secreting adenomas after bilateral adrenalectomy (14) are clinical demonstrations of the importance of hypothalamic and peripheral hormones in controlling subsequent pituitary tumor progression. Oncogenes are mutated proteins leading to inappropriate or constitutive signaling of downstream effectors without regulated upstream signals. Gain of function mutations are dominant, and thus, heterozygous germ-line or localized somatic mutations will usually be phenotypically expressed in tumors. The G proteins, involved in transmembrane signal transduction, are a group of GTP-binding proteins coupled to a superfamily of polypeptide receptors with seven membrane-spanning domains (15). The G proteins are heterotrimers composed of three distinct subunits: α, β, and γ. The α-subunit (Gsα) binds guanine nucleotides with high affinity and differs from one G protein to another. Up to 40% of screened human GH-secreting adenomas harbor somatic heterozygous missense activating point mutations of the Gsα gene in positions 201 in exon 8 (arginine replaced by cysteine or histidine) or 227 in exon 9 (glutamine replaced by arginine or leucine) (16–19). Surprisingly, gsp mutations are rare (5%) in GH-secreting adenomas encountered in Japan (20). These tumors contain constitutively active Gsα protein, persistently active adenylyl cyclase, and high intracellular cAMP levels (21). Both residues 201 and 227 are critically important in GTP binding and hydrolysis, and their alterations inhibit intrinsic GTPase activity of Gsα protein, thus constitutively activating the protein and converting it into an oncogene (gsp). As cAMP normally mediates GHRH signaling, the mutated somatotroph G protein activation bypasses the requirement for GHRH ligand-mediated receptor activation, and persistent high cAMP activates protein kinase A, leading to phosphorylation of the cAMP response element-binding protein (CREB), which results in constitutive GH hypersecretion and enhanced somatotroph proliferation. Compared with nonmutant tumors, the gsp-expressing adenomas are smaller, have increased intratumoral cAMP, do not respond briskly to GHRH, are frequently responsive to TRH administration, and are extremely sensitive to the inhibitory effect of somatostatin (22–24). However, no difference was found in age, sex, duration of the disease, or cure rate between patients with and without the mutation. Morphologically, tumors expressing the gsp oncogene are usually densely granulated adenomas (25), which are believed to be slowly growing GH cell adenomas. Interestingly, similar, early occurring, postzygotic dominant somatic mutations in codon 201 of the Gsα were identified in several hyperplastic endocrine tissues derived from patients with McCune-Albright’s syndrome, including pituitary hyperplasia and GH-secreting adenomas (26–28). The gsp-activating mutations are rarely detected in nonfunctioning adenomas (<10%) and are absent in prolactinomas (29, 30) and TSH-secreting pituitary tumors (31). Recently, gsp mutations have been identified in 6% of ACTH-secreting adenomas (32). Thus, these mutations are relatively specific to GH cell pituitary tumorigenesis. In addition to mutations in the stimulatory G proteins (gsp), mutations in the α-subunit of the inhibitory GTP-binding protein gene (gip2), at codon 205 of the Gi2α protein, replacing glutamine with arginine, were reported in several nonfunctioning adenomas (30). Interestingly, these mutations result in adenylyl cyclase inhibition and cAMP suppression, in contrast to the gsp mutation effects. The direct mechanism by which cAMP stimulates somatotroph proliferation and GH secretion is unknown, but the phosphorylated cAMP-regulated factor CREB may be involved in mediating the transcriptional effects of cAMP. Transgenic mice overexpressing a phosphorylation-deficient and transcriptionally inactive mutant of CREB in the anterior pituitary exhibit dwarfism and somatotroph hypoplasia (33). In a series of 15 human GH-secreting adenomas studied recently, all tumors contained elevated levels of Ser133-phosphorylated, and hence activated, CREB, compared with low levels of phospho-CREB in nonfunctioning pituitary adenomas (34). Interestingly, CREB phosphorylation was elevated not only in the four pituitary tumors that contained the mutant gsp oncogene, but also in other GH cell tumors expressing wild-type Gsα protein at high levels (34). Thus, Gsα overexpression in some human GH-producing adenomas may promote CREB phosphorylation. Three homologous ras protooncogenes, H-ras, K-ras, and N-ras, encode 21-kDa monomeric proteins (p21) that possess GTPase activity and are structurally related to G proteins (35). Missense mutations at codons 12, 13, and 61, which convert ras protooncogenes into active oncogenes, are commonly identified in a variety of different human cancers (36) and benign and malignant endocrine tumors (37). Four studies (19, 38–40) examining more than 200 secreting and nonsecreting pituitary tumors identified only one H-ras gene mutation in an aggressive prolactinoma (38). H-ras mutations were, however, identified in metastatic pituitary carcinomas in three of five patients studied, but not in the respective primary pituitary tumors or in six other invasive adenomas (41). Thus, ras oncogene point mutation and activation are uncommon events in pituitary tumor initiation, but may be important in aggressive tumors and in the very rare pituitary metastasis formation and growth. In contrast to the dominant protooncogenes, which induce cancer when converted to active oncogenes and which are usually associated with a point mutation, tumor suppressor genes that normally suppress cell proliferation when present as two normal alleles initiate tumor cell growth after both recessive alleles are lost or altered. According to Knudson’s two-hit theory, neoplasia may be generated by two consecutive genetic changes, where the first hit is a dominantly inherited germ-line mutation or a somatic mutation of one allele in the involved tissue, and the second hit is usually a somatic deletion of the second allele of the tumor suppressor gene with its flanking sequences, resulting in a loss of tumor heterozygosity (LOH) that renders the cell hemizygous for the original allelic mutation. This second event, revealed by a comparison of leukocyte and tumor alleles, inactivates the remaining normal suppressor gene allele and leads to unrestrained growth of clonal cancer cell populations. The p53 suppressor gene is the most common gene mutated or deleted in human neoplasia; it is associated with 50% of cancer types. However, p53 mutations were not detected in exons 5–8 (where these mutations usually occur) (42) in nonsecreting and secreting pituitary adenomas (39, 43) or in pituitary carcinomas and their respective metastases (41). Thus, p53 probably has little or no role in pituitary tumorigenesis. MEN-1 is an autosomal dominant hereditary syndrome characterized by the combined occurrence of hyperfunction or tumor formation of the parathyroids, anterior pituitary, pancreatic islets, and, rarely, carcinoid, thyroid, and adrenocortical tumors. The MEN-1 gene has not yet been characterized, but was mapped to chromosome 11q13 (44, 45). LOH involving the 11q13 region has been shown in the majority of parathyroid tumors removed from MEN-1 patients (46–48), in 29% of sporadic parathyroid lesions (48), in pancreatic islet tumors associated with MEN-1 syndrome (44, 49), and in 78% of sporadically occurring carcinoid tumors (50). These observations indicate that inactivation of a tumor suppressor gene in the 11q13 region is pathogenetically involved in the development of both hereditary and sporadic MEN-1-associated tumors. Allelic deletions of chromosome 11q markers were studied in both MEN-1-associated and sporadic pituitary adenomas. Allelic loss involving 11q13 was demonstrated in a prolactinoma (51) and a somatotropinoma (52) obtained from patients with MEN-1 syndrome. Several studies investigated the association of putative MEN-1 chromosomal locus deletion with sporadic pituitary adenomas. Bale et al. (49) and Eubanks et al. (53) failed to demonstrate LOH of chromosome 11 regions in 8 sporadic adenomas. Bystrom et al. (48) studied 26 sporadic nonsecreting and secreting adenomas of all types and observed loss of 11q alleles in only 2 prolactinomas. Herman et al. (39) detected LOH of 11q13 and 11p loci in 1 of 7 sporadic PRL-secreting adenomas, and Thakker et al. (52) revealed allelic loss involving the 11q13 region in 4 of 12 non-MEN-1 GH cell adenomas. Boggild et al. (19) studied 88 sporadic pituitary tumors and identified chromosome 11q13 marker deletions in 20% of nonfunctioning adenomas, 28% of ACTH-secreting tumors, 16% of somatotropinomas, and 12% of prolactinomas. Thus, recessive genetic events in a tumor suppressor gene located on chromosome 11q13 are associated with pituitary tumorigenesis in MEN-1 pituitary adenomas and in 10–15% of all sporadic adenomas. However, this suppressor gene and related oncogenic alterations associated with pituitary transformation have not yet been characterized. The retinoblastoma (Rb) gene product regulates the cell cycle and has an important role in controlling cell differentiation and survival. Inactivation of both Rb alleles on chromosome 13q14 by somatic mutations leads to disappearance of the protein and to sporadic retinoblastomas (54, 55). Breast cancer (56) and nonendocrine carcinomas are associated with LOH on chromosome 13q. Heterozygous dysruptions of the Rb gene in mice result in an approximately 100% incidence of POMC-expressing pituitary tumors (57, 58). These tumors are adenocarcinomas and originate from the pituitary intermediate lobe. In human benign pituitary adenomas, however, no Rb gene deletions or mutations were detected by several groups (59–63). Pei et al. (63) demonstrated LOH in proximity to the Rb locus on chromosome 13q in 13 malignant or highly invasive pituitary tumors and their respective metastases. The Rb protein, however, was identified by immunohistochemistry in all aggressive tumors with chromosome 13 allelic loss (63), suggesting that another suppressor gene located on chromosome 13q adjacent to the Rb locus, and not the Rb gene itself, may be involved in the development of invasive pituitary adenomas and carcinomas. Another tumor suppressor gene that may be associated with pituitary tumorigenesis is the purine-binding factor gene, nm23. In highly metastatic cancer, including breast, hepatic, and colorectal carcinomas, nm23 expression is reduced (64). Recently, nm23 ribonucleic acid (RNA) expression was studied in 22 pituitary tumors, and H2 isoform expression and H2 protein immunoreactivity were significantly reduced in invasive adenomas and correlated inversely with cavernous sinus invasion (65). However, these invasive tumors did not express structural nm23 gene alterations. CDK complexes play a central role in controlling cell cycle progression. Rb phosphorylation by CDK4 to its inactive form neutralizes its ability to regulate the cell cycle (66). p16, the specific inhibitor of CDK4, is inactivated in several human cancer cell lines and was found to be low or undetectable in pituitary tumors (67). However, no p16 mutations or gene loss was detected in these adenomas. p27 is another inhibitor of the kinase activity of the cyclin-CDK complexes and suppresses cell cycle progression. Targeted disruption of the p27 coding sequence in transgenic mice resulted in larger knockout mice with normal serum levels of GH or insulin-like growth factor I (IGF-I). These mice had multiorgan hyperplasia, pituitary intermediate lobe hyperplasia and benign intermediate lobe pituitary tumors, retinal dysplasia, and female sterility (68–70). Interestingly, gene deletions of both Rb (57, 58) and p27 genes uniquely induce pituitary pars intermedia neoplastic growth, suggesting interaction of p27 and Rb proteins in the same regulatory pathway controlling pituitary cell proliferation and differentiation. GH secretion by pituitary somatotrophs is normally regulated by the stimulatory effect of hypothalamic GHRH and the inhibitory effect of somatostatin and IGF-I (71). The GHRH receptor is also involved in somatotroph cell proliferation and differentiation (72). However, no mutations were found in the GHRH receptor in human GH-secreting adenomas (73). Interestingly, some GH cell adenomas expressed an alternatively spliced truncated GHRH receptor (73), whose pathophysiological significance is unknown. Similarly, 19 somatotropinomas screened for IGF-I receptor β-subunit exhibited intact regions of the receptor critical for signal transduction (74). Functional pituitary tumors express heterogeneous somatostatin receptor subtypes (75, 76). The hormonal responses to somatostatin analogs appear to be determined by selective somatostatin receptor subtype signaling in secreting tumors (77). TRH binds to its specific G protein-coupled receptor to normally stimulate TSH and PRL secretion from pituitary thyrotrophs and lactotrophs. In addition, patients with acromegaly or nonfunctioning adenomas may respond to TRH administration by brisk GH and glycoprotein subunit secretion, respectively. Recently, the coding region of the TRH receptor was found to be unaltered in PRL-, GH-, and TSH-secreting and nonfunctioning pituitary adenomas (31, 78). In addition, no dopamine type 2 receptor gene mutations were detected in prolactinomas or TSH-secreting and nonfunctioning adenomas studied recently (79). Therefore, based on mutational analyses of several cell surface receptors for hypothalamic releasing and inhibitory factors, there are no current data to support their role in pituitary tumor pathogenesis. The FGF family comprises at least nine structurally related proteins (80). Two FGF family prototypes (basic FGF, 146 amino acids), FGF-1 (acidic FGF, 140 amino acids) and FGF-2 (basic FGF) lack the characteristic signal peptide sequence of secreted proteins (81). FGF-4, a 206-amino acid protein, is encoded by the heparin-binding secretory transforming gene (hst) and contains the signal sequence for secretion. FGF-2 and FGF-4 both have potent angiogenic activity, which may increase the vascularity of and blood supply to the tissues expressing them. FGF-2 is abundantly found in normal pituitary tissue (82, 83), where it is predominantly expressed within folliculostellate cells. This growth factor stimulates PRL secretion from both normal (84) and adenomatous pituitary cells (85). Human pituitary adenomas express FGF-2 (86, 87). However, FGF-2 is not mitogenic for pituitary cells in vivo (84, 85), and transgenic mice overexpressing FGF-2 in the pituitary do not develop pituitary adenomas (88). Recent observations have identified increased pituitary FGF-2 secretion in MEN-1 patients. Circulating FGF-2 is detectable in 40% of patients with this syndrome and in most MEN-1 patients with untreated pituitary adenomas (89, 90). Interestingly, surgical or medical treatment of tumors resulted in lowering of circulating FGF-2 as by thus suggesting that the pituitary is the of Moreover, several MEN-1 patients with PRL cell pituitary adenomas Thus, the MEN-1 syndrome is associated with loss of a putative tumor suppressor gene on chromosome FGF-2 may have a role in promoting pituitary tumorigenesis in MEN-1 but the interaction between the suppressor gene and the local growth factor is as yet unknown. expression is to tissues do not normally express the However, several human malignant tumors express Interestingly, has been located on chromosome 11q13 in proximity to several other oncogenes, including the putative MEN-1 suppressor gene The oncogene and the protein it FGF-4, PRL gene transcription and secretion in pituitary cells and of pituitary cells is associated with in vivo and of tumor cells after In addition, transforming of the gene were identified in derived from several human prolactinomas for FGF-4 detected the protein product of in a of prolactinomas studied, compared with only in other pituitary adenomas, both secreting and and nonsecreting and no immunoreactivity was present in normal pituitary proliferation activity correlated with prolactinoma However, the mechanism by which the is activated and converted to a transforming gene is as yet unknown, as gene were not detected in prolactinomas of these observations indicate that may have an important role in the pathogenesis of PRL-secreting adenomas, as either a tumor or a for both proliferation and secretion. is a acid mitogenic peptide expressed in normal tissues and in several malignant tumors and its effects the growth factor was from the of anterior pituitary and studies have localized it to PRL-secreting cells where the growth factor receptor has also been detected human pituitary tissue and secreting and nonsecreting adenomas express the growth factor overexpressing transgenic demonstrated hyperplasia and PRL cell adenomas Thus, may play a critical role in prolactinoma pathogenesis. Recently, a transforming gene, that a protein of amino was from GH-secreting pituitary tumors by differential which is not expressed in normal pituitary tissue, transforming effects both in and in As functional human adenomas also appear to express this the presence of a specific transforming gene in human adenomas. is that events may to initiate pituitary adenoma pathogenesis 1). These early chromosomal mutations and expression of factors clonal expansion of the transformed pituitary cell hypothalamic receptor growth factor and disordered cell cycle This of events results in unrestrained transcription and secretion, and cell which are of the pituitary only monoclonality and gsp mutations have been shown to be in sporadic human pituitary adenomas. Thus, despite recent information on the subcellular events associated with the formation and growth of pituitary tumors, the intrinsic defect leading to adenoma formation mechanisms of pituitary tumorigenesis. The gene for has been identified on chromosome 11q13 et al.