Methylation profiling of rectal cancer identifies novel markers of early-stage disease.

BACKGROUND: Radical surgery is the de facto treatment for early rectal cancer. Conservative surgery with transanal endoscopic microsurgery can achieve high rates of cure but the histopathological measures of outcome used to select local treatment lack precision. Biomarkers associated with disease progression, particularly mesorectal nodal metastasis, are urgently required. The aim was to compare patterns of gene-specific hypermethylation in radically excised rectal cancers with histopathological stage. METHODS: Locus-specific hypermethylation of 24 tumour suppressor genes was measured in 105 rectal specimens (51 radically excised adenocarcinomas, 35 tissues adjacent to tumour and 19 normal controls) using the methylation-specific multiplex ligation-dependent probe assay (MS-MLPA). Methylation values were correlated with histopathological indices of disease progression and validated using bisulphite pyrosequencing. RESULTS: Five sites (ESR1, CDH13, CHFR, APC and RARB) were significantly hypermethylated in cancer compared with adjacent tissue and normal controls (P < 0·050). Methylation at these sites was higher in Dukes' A than Dukes' 'D' cancers (P = 0·013). Methylation at two sites (GSTP1 and RARB) was individually associated with localized disease (N0 and M0 respectively; P = 0·006 and P = 0·008). Hypermethylation of at least two of APC, RARB, TIMP3, CASP8 and GSTP1 was associated with early (N0 M0) disease (N0, P = 0·002; M0, P = 0·044). Methylation levels detected by MS-MLPA and pyrosequencing were concordant. CONCLUSION: Locus-specific hypermethylation was more prevalent in early- than late-stage disease. Hypermethylation of two or more of a panel of five tumour suppressor genes was associated with localized disease.

PTTG and PBF repress the human sodium iodide symporter.

The ability of the thyroid to accumulate iodide provides the basis for radioiodine ablation of differentiated thyroid cancers and their metastases. Most thyroid tumours exhibit reduced iodide uptake, although the mechanisms accounting for this remain poorly understood. Pituitary tumour transforming gene (PTTG) is a proto-oncogene implicated in the pathogenesis of thyroid tumours. We now show that PTTG and its binding factor PBF repress expression of sodium iodide symporter (NIS) messenger RNA (mRNA), and inhibit iodide uptake. This process is mediated at least in part through fibroblast growth factor-2. In detailed studies of the NIS promoter in rat FRTL-5 cells, PTTG and PBF demonstrated specific inhibition of promoter activity via the human upstream enhancer element (hNUE). Within this approximately 1 kb element, a complex PAX8-upstream stimulating factor 1 (USF1) response element proved critical both to basal promoter activity and to PTTG and PBF repression of NIS. In particular, repression by PTTG was contingent upon the USF1, but not the PAX8, site. Finally, in human primary thyroid cells, PTTG and PBF similarly repressed the NIS promoter via hNUE. Taken together, our data suggest that the reported overexpression of PTTG and PBF in differentiated thyroid cancer has profound implications for activity of the NIS gene, and hence significantly impacts upon the efficacy of radioiodine treatment.

A potential role for PTTG/securin in the developing human fetal brain.

Human securin, known also as PTTG, has established oncogenic and cell cycle regulatory functions. PTTG/securin transforms cells in vitro, inhibits sister chromatid separation, and regulates secretion of fibroblast growth factor-2. FGF-2 is a key regulator of CNS development and PTTG/securin expression has been reported in murine fetal brain. We examined the expression and function of securin and FGF-2 in the developing human fetal brain and in a fetal neuronal cell line (NT 2). Securin expression was significantly reduced in first and second trimester fetal cerebral cortex compared with adult cerebral cortex, where immunocytochemistry revealed intense securin staining in neuronal cell bodies. FGF-2 protein was concordantly lower in fetal cortex, whereas pretranslational expression of PTTG binding factor (PBF) was not significantly altered in fetal brain compared with adult. PCNA expression demonstrated that high securin levels in adult cortex were associated with absent cell proliferation. In NT-2 cells, securin stimulated FGF-2 expression, which could be abrogated by a carboxyl-terminal mutation. Low transient expression of securin resulted in a significant proliferative effect, whereas high levels of securin expression inhibited cell turnover. We propose a potential role for human PTTG/securin in modulating cell proliferation and FGF-2 expression during human neurogenesis.

Vascular endothelial growth factor, its receptor KDR/Flk-1, and pituitary tumor transforming gene in pituitary tumors.

Pituitary tumorigenesis is a poorly understood process involving dysregulation of the cell cycle, proliferation, and angiogenesis. The novel securin pituitary tumor transforming gene (PTTG) disrupts cell division and stimulates fibroblast growth factor (FGF)-2-mediated angiogenesis. We investigated expression of the angiogenic vascular endothelial growth factor (VEGF) and its receptor KDR/Flk-1 in 103 human pituitary tumors, and we assessed functional relationships between these genes in vitro. Nonfunctioning tumors (n = 81) demonstrated markedly raised VEGF mRNA (3.2-fold, P < 0.05) and protein concentrations, compared with normal pituitaries (n = 10). KDR was also highly induced in nonfunctioning tumors (14-fold, P < 0.001, n = 78) as well as in the whole cohort of pituitary tumors, compared with normal pituitary samples (14-fold, P < 0.0001, n = 100). In vitro, PTTG induced VEGF, but not KDR, expression in fetal neuronal NT2 cells (2.7-fold, P < 0.001, n = 8), MCF-7 breast carcinoma cells (1.9-fold, P = 0.03, n = 10), and choriocarcinoma JEG-3 cells (P = 0.0002, n = 8). A mutated PTTG construct that cannot be phosphorylated showed identical VEGF up-regulation (2.9-fold, P < 0.001, n = 8) in NT2 cells, compared with wild-type PTTG, but a further mutated construct with abrogation of the key protein:protein interaction domain of PTTG resulted in a significant reduction in VEGF stimulation, compared with wild-type (0.37-fold reduction, P < 0.001, n = 8). FGF-2 findings mirrored those of VEGF, although antibody depletion of secreted FGF-2 in the cell medium failed to influence VEGF up-regulation by PTTG. Overall, our findings implicate altered VEGF and KDR signaling in pituitary tumorigenesis, and we propose that PTTG stimulation of FGF-2 and VEGF expression in the presence of up-regulated growth factor receptors may account for angiogenic growth and progression of human pituitary tumors.

Pituitary tumor transforming gene and fibroblast growth factor-2 expression: potential prognostic indicators in differentiated thyroid cancer.

Differentiated thyroid cancers are the most common endocrine cancers, but there are no reliable molecular markers of prognosis. Pituitary tumor transforming gene (PTTG) plays several potential roles in tumor initiation and progression, including regulating mitosis and stimulating expression of fibroblast growth factor (FGF)-2. Increased expression of PTTG has been demonstrated in follicular thyroid lesions, and expression of this oncogene has been identified as a potential prognostic marker in pituitary adenomas and colon carcinomas. We assessed the expression of PTTG and FGF-2 and its receptor FGF-R-1 in 27 differentiated thyroid cancers, and we compared this with expression in 11 normal thyroids, 25 multinodular goiters, and 13 Graves' disease specimens. We also examined the relationship between gene expression and clinical markers of tumor behavior. PTTG and FGF-2 were overexpressed in thyroid carcinomas (9.5-fold increase, P = 0.003, and 5.0-fold increase, P < 0.001, respectively) compared with normal thyroid. Increased FGF-2 mRNA expression was independently associated with the findings of lymph node invasion (R(2) = 0.71; P < 0.001) and distant metastasis (R(2) = 0.55; P = 0.009) at tumor presentation, after taking into account known prognostic factors such as age and gender of the patient and size and type of the tumor. High PTTG expression was independently associated with tumor recurrence (R(2) = 0.64; P = 0.003). We conclude that PTTG and FGF-2 expression are potential prognostic markers (and perhaps therapeutic targets) for differentiated thyroid cancer.

Regulation of arginine vasopressin mRNA in rat fetal hypothalamic cell culture. Role of protein kinases and glucocorticoids.

Studies have been performed to investigate the regulation of arginine vasopressin (AVP) mRNA expression in fetal hypothalamic cultures. AVP mRNA-positive neurones were identified by in-situ hybridization histochemistry, and changes in mRNA expression were quantitated by nuclease protection assay. Both protein kinase C and protein kinase A activators increased the expression of AVP mRNA, in contrast to dexamethasone, which inhibited the responses to both protein kinase C and protein kinase A activation.

PTTG's C-terminal PXXP motifs modulate critical cellular processes in vitro.

Human pituitary tumor-transforming gene (PTTG), known also as securin, is a multifunctional protein implicated in the control of mitosis and the pathogenesis of thyroid, colon, oesophageal and other tumour types. Critical to PTTG function is a C-terminal double PXXP motif, forming a putative SH3-interacting domain and housing the gene's sole reported phosphorylation site. The exact role of phosphorylation and PXXP structure in the modulation of PTTG action in vitro remains poorly understood. We therefore examined the mitotic, transformation, proliferation and transactivation function of the C-terminal PXXP motifs of human PTTG. Live-cell imaging studies using an EGFP-PTTG construct indicated that PTTG's regulation of mitosis is retained regardless of phosphorylation status. Colony-formation assays demonstrated that phosphorylation of PTTG may act as a potent inhibitor of cell transformation. In proliferation assays, NIH-3T3 cells stable transfected and overexpressing mutations preventing PTTG phosphorylation (Phos-) showed significantly increased [3H]thymidine incorporation compared with WT, whereas mutants mimicking constitutive phosphorylation of PTTG (Phos+) exhibited reduced cell proliferation. We demonstrated that PTTG transactivation of FGF-2 in primary thyroid and PTTG-null cell lines was not affected by PTTG phosphorylation but was prevented by a mutant disrupting the PXXP motifs (SH3-). Taken together, our data suggest that PXXP structure and phosphorylation are likely to exert independent and critical influences upon PTTG's diverse actions in vitro.

Interleukin-1 beta induces corticotropin-releasing factor-41 release from cultured hypothalamic cells through protein kinase C and cAMP-dependent protein kinase pathways.

Interleukin-1 beta (IL-1 beta) induces a dose-dependent increase in the release of corticotropin-releasing factor-41 (CRF) from dispersed rat fetal hypothalamic cells in culture. This release of CRF could be inhibited by the protein kinase C inhibitor H-7, and by the protein kinase A inhibitor IP-20. This suggests that both protein kinase C and protein kinase A-dependent pathways are involved in the response of CRF to IL-1 beta. Dexamethasone also blocked the CRF response to IL-1 beta, indicating that activated glucocorticoid receptors can inhibit the response of CRF to IL-1 beta.

Release of corticotrophin-releasing factor-41, arginine vasopressin and oxytocin from rat fetal hypothalamic cells in culture: response to activation of intracellular second messengers and to corticosteroids.

The effects of the activation of protein kinase A (PKA), protein kinase C (PKC) and corticosteroids were investigated on the release of corticotrophin-releasing factor-41 (CRF), arginine vasopressin (AVP) and oxytocin from rat fetal hypothalamic cells in culture. Both forskolin and PMA (phorbol 12-myristate 13-acetate) increased CRF, AVP and oxytocin release, while dexamethasone and aldosterone only reduced basal secretion of CRF. Both steroids also inhibited forskolin-induced CRF, AVP and oxytocin responses to PMA. These data provide direct evidence for a role for both PKC- and PKA-mediated mechanisms in the regulation of CRF, AVP and oxytocin release and for differential effects of both glucocorticoids and mineralocorticoids on PKA- and PKC-stimulated responses.

Corticotrophin-releasing factor-41, vasopressin and oxytocin release into hypophysial portal blood in the rat: effects of electrical stimulation of the hypothalamus, amygdala and hippocampus.

The role of the paraventricular nuclei (PVN), amygdala and hippocampus in the control of the hypothalamic-pituitary-adrenal axis has been studied by determining the effect of electrical stimulation of the PVN, amygdala and hippocampus on the release of corticotrophin-releasing hormone (CRF-41) and arginine vasopressin (AVP) into hypophysial portal blood and ACTH and corticosterone into peripheral blood. Adult female Wistar rats were anaesthetized with sodium pentobarbitone and stimulation was carried out through previously implanted bipolar, glass-insulated platinum electrodes. Hypophysial portal blood was collected 30 min before and 30 min during the application of the stimulus which consisted of trains (30 s on and 30 s off) of biphasic rectangular pulses with a frequency of 50 Hz, pulse width 1 ms and amplitude 1 mA. Bilateral stimulation of the PVN increased while unilateral stimulation of the amygdala decreased the release of CRF-41 into hypophysial portal blood. The threefold increase in release of CRF-41 induced by PVN stimulation correlated with a marked increase in peripheral plasma concentrations of ACTH and corticosterone. Stimulation of the hippocampus had no significant effect on CRF-41 release, and stimulation of each of the three brain regions had no effect on AVP release into portal blood. These findings were extended in a second study to compare the effects of unilateral bipolar electrical stimulation of the PVN and of the supraoptic nucleus (SON) on the release of CRF-41, AVP and oxytocin. This study was carried out on adult male rats, anaesthetized with sodium pentobarbitone, in which the stimulus was applied through previously implanted concentric stainless-steel electrodes.(ABSTRACT TRUNCATED AT 250 WORDS)

5-Hydroxytryptamine stimulates corticosteroid-sensitive CRF release from cultured foetal hypothalamic cells. Role of protein kinases.

5-Hydroxytryptamine (5-HT) has been shown to activate the hypothalamo-pituitary-adrenal axis, possibly by a direct action on hypothalamic CRF synthesis and release. In order to study the mechanisms involved in this effect, foetal hypothalamic cells were cultured and corticotropin-releasing factor-41 (CRF) release was measured by radioimmunoassay. 5-HT induced CRF release in a dose-dependent manner. Further studies were performed with a specific protein kinase C inhibitor, H-7 (1-(5-isoquinolinesulfonyl)-2-methyl-piperazine) and a specific cyclic adenosine monophosphate-dependent protein kinase inhibitor, IP-20. Basal release of CRF-41 from the cultured hypothalamic cells was unaffected by IP-20 and was only diminished at a high (50 microM) concentration of H-7. 5-HT stimulated-CRF release, however, was blocked by both H-7 and IP-20. Dexamethasone and aldosterone both caused a dose-dependent inhibition of 5-HT induced CRF release. These results demonstrate that CRF can be released from hypothalamic neurons in response to 5-HT through a protein kinase C and protein kinase A dependent mechanism and that 5-HT stimulated CRF release can be inhibited by dexamethasone and aldosterone.

Different patterns of molecular forms of somatostatin are released by the rat median eminence and hypothalamus.

The molecular forms of somatostatin (SOM) released from hypothalamic slices and from the isolated median eminence (ME) in vitro were compared by high-performance liquid chromatography and radioimmunoassay. SOM-14 was the predominant form of the peptide released from hypothalamic slices, although small amounts of SOM-28 were detected. Perifusates of ME tissue contained a larger proportion of SOM-28 and higher molecular weight peptides were present; depolarization increased the rates of release of all molecular forms of SOM. These results suggest that the capacity to release SOM-28 and high-molecular-weight forms of SOM may be a specialized function of nerve terminals in the ME.

Effects of adrenalectomy and glucocorticoids on the peptides CRF-41, AVP and oxytocin in rat hypophysial portal blood.

1. The effects of adrenalectomy (3 weeks) and dexamethasone (3 h) treatment on the release of corticotrophin-releasing factor-41 (CRF-41), arginine vasopressin (AVP), oxytocin (OT), adrenocorticotrophin (ACTH) and corticosterone were studied in adult female Wistar rats. 2. The animals were anaesthetized with sodium pentobarbitone which, as assessed by the effects on the circadian rhythm of plasma ACTH and corticosterone, appeared to be a better anaesthetic than either urethane or alphaxalone for studies on the hypothalamic-pituitary-adrenal system. 3. Adrenalectomy increased the concentrations of ACTH in peripheral plasma and the output of CRF-41 and AVP into hypophysial portal plasma. 4. Dexamethasone administered to adrenalectomized rats significantly reduced the concentration of ACTH in peripheral plasma and the amount of AVP released into portal plasma. However, dexamethasone did not affect the output of CRF-41 into portal blood. 5. The output of OT into portal plasma was unaffected by either adrenalectomy or dexamethasone treatment. 6. Dexamethasone administered to adrenalectomized rats reduced significantly the ACTH response to CRF-41. 7. These results show that the feed-back action of glucocorticoids is mediated by two mechanisms. The increased release of ACTH which follows adrenolectomy [corrected] is produced predominantly by an increased release of both CRF-41 and AVP into hypophysial portal blood. The intermediate negative feed-back of glucocorticoids is produced by a reduction in the output of AVP but not CRF-41 into portal blood and, as well, by a significant reduction in the responsiveness of the anterior pituitary gland to CRF-41.

Dexamethasone and aldosterone modulate corticotropin-releasing factor-41 release from cultured rat fetal hypothalamic cells through type II and type I corticosteroid receptors respectively.

Dexamethasone and aldosterone inhibit hypothalamic corticotropin releasing factor-41 (CRF) release. The possible receptors through which these adrenal steroids affect CRF release were studied using rat fetal hypothalamic cell cultures. Neither the antimineralocorticoid RU 28318 nor the antiglucocorticoid RU 38486 alone had any effect on forskolin-stimulated CRF release. RU 38486 and RU 28318 however suppressed dexamethasone (1 microM)- and aldosterone (1 microM)-induced inhibition of forskolin (20 microM)-stimulated CRF release, respectively, suggesting that dexamethasone and aldosterone reduce CRF release through type II and type I corticosteroid receptors, respectively. RU 38486 had no effect on aldosterone-induced inhibition of forskolin-stimulated CRF release, nor did RU 28318 have any effect on dexamethasone-induced inhibition of forskolin-stimulated CRF release, indicating specificity of the binding of aldosterone with type I receptors, and dexamethasone with type II receptors in the hypothalamic cell cultures.

Mechanisms of norepinephrine mediated corticotropin-releasing factor-41 release from cultured fetal hypothalamic cells.

Catecholamines have been shown to activate hypothalamic corticotropin-releasing factor-41 (CRF) synthesis and release. In order to study the mechanisms involved, fetal hypothalamic cells were cultured and CRF release was measured by radioimmunoassay. Norepinephrine (NE) induced CRF release in a dose-dependent manner. Further studies were performed with a protein kinase C inhibitor, H-7(1-(5-isoquinolinesulfonyl)-2-methylpiperazine) and a protein kinase A inhibitor, IP-20. NE-stimulated CRF release was reduced by H-7 (5 and 50 microM) in a dose-dependent fashion, while 5 microM IP-20 resulted in a small but significant inhibition. Pretreatment of the cells for 15 h with 20 and 200 nM 12-O-tetradecanoylphorbol-13-acetate, which down-regulates protein kinase C activity, blocked the release of CRF in response to NE (1 microM), further supporting protein kinase C as a mediator for NE-activated CRF release. Pretreatment with 50 and 500 ng/ml pertussis toxin (15 h) resulted in a dose-dependent inhibition of NE-activated CRF release. Both dexamethasone and aldosterone at the concentrations of 1 microM reduced NE-induced CRF release. These results suggest that CRF can be released from hypothalamic neurons in response to NE through both protein kinase C- and protein kinase A-dependent mechanisms, and that pertussis toxin-sensitive G-proteins are also involved in this response. Furthermore, glucocorticoids and mineralocorticoids can reduce NE-activated CRF release from cultured hypothalamic cells.

Comparison of adrenocorticotropin control in Brattleboro, Long-Evans, and Wistar rats. Measurement of corticotropin-releasing factor, arginine vasopressin, and oxytocin in hypophysial portal blood.

The purpose of this study was to compare the control of adrenocorticotropin (ACTH) and corticosterone secretion in homozygous Brattleboro rats with their syngeneic controls, Long-Evans rats, and with rats of the Wistar strain. Plasma concentrations of ACTH and corticosterone were measured by radioimmunoassay in trunk blood, and corticotropin-releasing factor 41 (CRF-41), arginine vasopressin (AVP), and oxytocin were assayed in hypophysial portal vessel blood. Portal plasma was extracted with methanol for CRF-41 determination, and four different antisera and several different high-performance liquid chromatography (HPLC) systems were used to investigate AVP release. The peripheral plasma concentrations of ACTH and corticosterone were significantly higher in Long-Evans and homozygous Brattleboro than in Wistar rats. This difference was due, at least in part, to an approximately twofold greater release of CRF-41 into hypophysial portal blood of the Long-Evans and Brattleboro compared with Wistar rats. There was no significant difference between the strains in the output of oxytocin into portal blood. While no AVP could be detected in the neural lobe of homozygous Brattleboro rats, a small amount of AVP-like immunoreactivity was detected in unextracted hypophysial portal blood from homozygous Brattleboro rats. However, this AVP-like immunoreactivity was clearly distinct from authentic AVP in several HPLC systems, had no antidiuretic activity, and on gel filtration had a relative molecular mass greater than 5 kD. In contrast, the AVP-like immunoreactivity in hypophysial portal blood from Long-Evans rats co-eluted with authentic AVP in all HPLC systems tested.(ABSTRACT TRUNCATED AT 250 WORDS)

Dysregulation of iodothyronine deiodinase enzyme expression and function in human pituitary tumours.

OBJECTIVE: Thyroid hormones (THs) perform essential roles in pituitary function. They regulate anterior pituitary hormone secretion and are also key determinants of pituitary cell proliferation and differentiation. The critical role of deiodinase enzymes, which serve as prereceptor regulators of TH action, remains largely unexplored. Three deiodinase enzymes metabolize active and inactive THs and thereby determine tissue concentrations of the biologically active ligand, tri-iodothyronine (T3). We hypothesized that aberrant expression of deiodinase enzymes and/or altered enzyme activity in pituitary tumours may change tissue concentrations of THs and influence their growth and secretory characteristics. STUDY DESIGN AND PATIENTS: We studied 105 pituitary tumours and 10 normal pituitaries for expression of deiodinase enzyme mRNAs encoding types 1 (D1), 2 (D2) and 3 (D3) using real-time RT-PCR. Enzyme activity data from 20 pituitary samples were also obtained. RESULTS: Pituitary tumours expressed significantly increased D3 mRNA (6.5-fold, P < 0.0005) compared with normal pituitaries. D2 mRNA was also increased 2.6-fold (P = 0.005) in pituitary tumours compared with normals. The rare TSH-secreting pituitary tumour subtype expressed a 13.1-fold excess of D3 mRNA and reduced D2 mRNA (0.1-fold of normal pituitaries). D2 mRNA expression in ACTH-secreting tumours was similarly reduced to 0.1-fold that in normal pituitaries. CONCLUSIONS: Pituitary adenomas express abnormal levels of deiodinase enzymes compared to normal pituitaries. These abnormalities may have functional consequences on pituitary tumour growth. In the case of TSH-secreting pituitary adenomas, the observed pattern of deiodinase mRNA expression may explain the 'resistance' of this tumour type to TH feedback.

Expression of pituitary tumour transforming gene (PTTG) and fibroblast growth factor-2 (FGF-2) in human pituitary adenomas: relationships to clinical tumour behaviour.

OBJECTIVE: Pituitary tumour transforming gene (PTTG) encodes a multifunctional protein that is implicated in initiating and perpetuating pituitary adenoma growth. PTTG appears to have key regulatory functions in determining control of many fundamental cellular events including mitosis, cell transformation, DNA repair and gene regulation. Several of these events are mediated through interactions with PTTG binding factor (PBF) and fibroblast growth factor-2 (FGF-2). Given this background, we have determined the expression of PTTG, PBF, FGF-2 and its receptor FGF-R-1 in a large cohort of pituitary adenomas and have sought associations between levels of gene expression and clinical markers of tumour behaviour. PATIENTS AND METHODS: We used real-time reverse transcriptase-polymerase chain reaction (RT-PCR) and Western blot analyses to measure PTTG, PBF, FGF-2 and FGF-R-1 expression in ex vivo pituitary tumours (N = 121). Clinical data, including accurate radiological assessment of tumour characteristics, were used to determine any associations between gene expression and tumour behaviour. RESULTS: PTTG was increased significantly (fivefold, P = 0.005) in adenomas compared with normal pituitaries. We also demonstrated that PBF was similarly raised in adenomas (sixfold, P = 0.0001), and was significantly correlated with PTTG expression. FGF-2 and its receptor FGF-R-1 were also raised in adenomas compared with normal pituitary tissue. Moreover, significantly enhanced expression of FGF-R-1 was observed in invasive adenomas compared with other pituitary tumours. CONCLUSIONS: Our data support a fundamental role for PTTG-mediated upregulation of FGF-2 signalling in pituitary tumorigenesis and growth, and suggest that receptor-mediated mechanisms of growth factor action may be critically important. Further prospective studies are required to determine whether measurement of FGF-R-1 mRNA will be of clinical use as a prognostic marker in patients with pituitary adenomas.