This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Turner, H. E. Articles by Wass, J. A. H. Search for Related Content PubMed PubMed Citation Articles by Turner, H. E. Articles by Wass, J. A. H.

Angiogenesis in Endocrine Tumors

Department of Endocrinology (H.E.T., J.A.H.W.), Churchill Hospital, Oxford OX3 7LJ, United Kingdom; Molecular Angiogenesis Group (A.L.H.), Imperial Cancer Research Fund, Institute of Molecular Medicine, John Radcliffe Hospital, OX3 9DU Oxford, United Kingdom; and Davis Research Institute (S.M.), Cedars Sinai Hospital, Los Angeles, California 90048

Correspondence: Address all correspondence and requests for reprints to: Dr. Helen E. Turner, Department of Endocrinology, Churchill Hospital, Old Road, Oxford OX3 7LJ, United Kingdom. E-mail: jenny.Thomson{at}orh.nhs.uk

	Abstract

Top Abstract I. Introduction II. Mechanism of Angiogenesis III. Pituitary Tumors IV. Adrenal Tumors V. Thyroid Tumors VI. Parathyroid Tumors VII. Carcinoid Tumors VIII. Gastrointestinal... IX. Conclusions References

 
Angiogenesis is the process of new blood vessel development from preexisting vasculature. Although vascular endothelium is usually quiescent in the adult, active angiogenesis has been shown to be an important process for new vessel formation, tumor growth, progression, and spread. The angiogenic phenotype depends on the balance of proangiogenic growth factors such as vascular endothelial growth factor (VEGF) and inhibitors, as well as interactions with the extracellular matrix, allowing for endothelial migration. Endocrine glands are typically vascular organs, and their blood supply is essential for normal function and tight control of hormone feedback loops. In addition to metabolic factors such as hypoxia, the process of angiogenesis is also regulated by hormonal changes such as increased estrogen, IGF-I, and TSH levels.

By measuring microvascular density, differences in angiogenesis have been related to differences in tumor behavior, and similar techniques have been applied to both benign and malignant endocrine tumors with the aim of identification of tumors that subsequently behave in an aggressive fashion.

In contrast to other tumor types, pituitary tumors are less vascular than normal pituitary tissue, although the mechanism for this observation is not known. A relationship between angiogenesis and tumor size, tumor invasiveness, and aggressiveness has been shown in some pituitary tumor types, but not in others. There are few reports on the role of microvascular density or angiogenic factors in adrenal tumors. The mechanism of the vascular tumors, which include adrenomedullary tumors, found in patients with Von Hippel Lindau disease has been well characterized, and clinical trials of antiangiogenic therapy are currently being performed in patients with Von Hippel Lindau disease. Thyroid tumors are more vascular than normal thyroid tissue, and there is a clear correlation between increased VEGF expression and more aggressive thyroid tumor behavior and metastasis. Although parathyroid tissue induces angiogenesis when autotransplanted and PTH regulates both VEGF and MMP expression, there are few studies of angiogenesis and angiogenic factors in parathyroid tumors.

An understanding of the balance of angiogenesis in these vascular tumors and mechanisms of vascular control may assist in therapeutic decisions and allow appropriately targeted treatment.

I. Introduction

II. Mechanism of Angiogenesis

A. Proangiogenic growth factors

B. Measurement of angiogenesis

C. Influence of hormones on angiogenesis and VEGF

D. Inhibitors of angiogenesis

E. Extracellular matrix, invasion, and metalloproteinases

F. Importance of angiogenesis

III. Pituitary Tumors

A. Microvascular density of tumors

B. Pituitary vasculature

C. Angiogenesis in animal models of pituitary tumors

D. Angiogenic factors

E. Genes and regulation of angiogenesis

F. Matrix metalloproteinases

G. Therapeutic application of angiogenesis inhibitors

IV. Adrenal Tumors

A. Microvascular density of tumors

B. Angiogenic factors

C. Matrix metalloproteinases

D. Genes and regulation of angiogenesis

E. Therapeutic application of angiogenesis inhibitors

V. Thyroid Tumors

A. Microvascular density of tumors

B. Angiogenic factors

C. Matrix metalloproteinases

D. Therapeutic application of angiogenesis inhibitors

E. Genes and regulation of angiogenesis

VI. Parathyroid Tumors

A. Microvascular density of tumors

B. Angiogenic factors

C. Matrix metalloproteinases

D. Genes and regulation of angiogenesis

VII. Carcinoid Tumors

VIII. Gastrointestinal Neuroendocrine Tumors

IX. Conclusions

	I. Introduction

Top Abstract I. Introduction II. Mechanism of Angiogenesis III. Pituitary Tumors IV. Adrenal Tumors V. Thyroid Tumors VI. Parathyroid Tumors VII. Carcinoid Tumors VIII. Gastrointestinal... IX. Conclusions References

 
ANGIOGENESIS IS DEFINED as the development of new blood vessels from preexisting vessels, and it is a crucial process in normal physiology (1). Thus, angiogenesis plays an essential role in physiological processes such as embryonic development, wound healing, and the normal menstrual cycle (2, 3). It is however also an important pathogenic process in several diseases such as atherosclerosis (4), arthritis (5), diabetic retinopathy (6), and psoriasis (7).

A major interest has been the relationship of pathological angiogenesis and its role in tumor biology (8). On the basis of experiments that showed that tumors implanted into isolated perfused organs failed to develop, whereas the same tumors implanted within 6 mm of blood vessels not only induced angiogenesis but also both grew and metastasized (9, 10), Folkman (11) proposed that solid tumors are dependent on the process of angiogenesis for growth beyond a few millimeters in size and that increased tumor diameter required a corresponding increase in vascularization.

Angiogenesis (measured as tumor microvessel density) correlates with tumor behavior. In many human tumors, including breast, bladder, and stomach, increased angiogenesis has been shown to be associated with the development of metastases (12, 13), poor prognosis (14, 15), and reduced survival (16, 17).

Endocrine organs are very vascular, with fenestrated epithelium lining the blood vessels, which allows easy transfer of substances across the permeable vessel wall (18, 19). The blood supply is clearly essential for normal metabolic and endocrine functions. Complex feedback loops whereby differences in the serum concentration of various factors lead to alteration in hormonal secretion are essential for endocrine function. The necessity for rapid and tight regulation of these systems, frequently over a very small range of hormone concentrations, clearly requires an efficient blood supply. The purpose of this review is to explore current knowledge regarding angiogenesis in endocrine tumors.

Although the process of angiogenesis has been mainly studied in the context of malignant tumors, precancer has also been associated with increased angiogenic growth factor expression and increased vascularity when compared with nonneoplastic host tissue. Premalignant lesions are more vascular than normal tissue, and precarcinoma of the cervix and breast exhibit increased microvascular density (MVD) (20, 21, 22). Studies using RT-PCR and in situ hybridization (ISH) for the proangiogenic growth factor vascular endothelial growth factor (VEGF) in colonic tissues ranging from normal colon to invasive colorectal tumors show increased VEGF expression in the premalignant phase (23). Histologically, normal lobules derived from breast tissue harboring cancer are more angiogenic than lobules derived from breasts without cancer (24). Transgenic mice with an oncogene in the pancreatic ß-cells demonstrate increased angiogenesis in hyperplastic islets before development of frank neoplastic change (25). Inhibitors of angiogenesis such as angiostatin and endostatin reduce tumor growth if administered at this early stage, suggesting that increasing angiogenesis is important to tumor progression (26).

	II. Mechanism of Angiogenesis

Top Abstract I. Introduction II. Mechanism of Angiogenesis III. Pituitary Tumors IV. Adrenal Tumors V. Thyroid Tumors VI. Parathyroid Tumors VII. Carcinoid Tumors VIII. Gastrointestinal... IX. Conclusions References

 
Angiogenesis is a multistep process, involving both the endothelium and the extracellular matrix (1). After a stimulatory signal (such as a specific growth factor), activated endothelial cells release proteases (such as plasminogen activator), which lead to degradation of the extracellular matrix surrounding the vessel (27), followed by endothelial migration and proliferation. Endothelial cells are reorganized into a tubular structure, followed by fusion with other newly formed vessels, leading eventually to an anastomotic network (28).

The events that occur during development of an angiogenic phenotype are complex, involving stimulation by different proangiogenic growth factors and reduction in inhibitors of angiogenesis. The process involves an interaction between tumor cells, endothelial cells, macrophages, fibroblasts, and the extracellular matrix, all of which are capable of releasing factors influencing the angiogenesis mechanism. It is the net balance of proangiogenic factors and inhibitors of angiogenesis that determines the final angiogenic phenotype of the tumor.

Adult endothelial cells are usually quiescent, but may be stimulated to enter the cell cycle and develop into new capillaries. The switch concept describes the process whereby angiogenesis is activated (29). An understanding of the mechanism of this angiogenic switch in tumors may lead to further understanding of specific determinants of tumor behavior including invasion, malignancy, and metastasis. Studies in transgenic mice, in which tumor development progresses through distinct stages, have demonstrated that the switch to an angiogenic phenotype occurs at an early stage in tumor development and is potentially rate-limiting for subsequent tumor growth (25, 29).

Mechanisms subserving the angiogenic switch relate to a balance of proangiogenic and inhibitory stimuli. Angiogenesis inducers include VEGF and basic fibroblast growth factor (FGF-2), whereas more recently endogenous angiogenesis inhibitors have been described. These latter are often cleaved products of other larger proteins that are not themselves inhibitors. For example, angiostatin is a cleaved product of plasminogen, and endostatin arises from collagen XVIII. Control of proteases and the cleavage process is still not elucidated, but the system is usually effective in inhibiting angiogenesis, as exemplified in pancreatic islets in which VEGF and FGF expression is constitutive but angiogenesis does not normally occur (30). Negative regulators may be particularly important in endocrine organs that are very vascular, to maintain a quiescent endothelium, and rather than increased inducers of angiogenesis, down-regulation of inhibitor production may be required to activate the angiogenic switch (29) (Fig. 1).

View larger version (24K): [in this window] [in a new window]   FIG. 1. Diagram showing proposed balance hypothesis (derived from Ref.29 ).

FGFs were the first angiogenic factors discovered, and they are probably of importance when the tumor becomes large, degrades the extracellular matrix in which FGF is sequestered, and acts through the high-affinity cell-surface tyrosine kinase FGF receptors (FGFR-1 to FGFR-4) (for review, see Ref.33). There are currently 20 described distinct FGFs of which FGF-1, FGF-2, and to a lesser extent FGF-3, FGF-4, FGF-5, and FGF-7 have been shown to possess proangiogenic activity (33, 34). Alternate splicing of the FGFR genes and/or expression of the different FGFR genes themselves leads to different forms of FGFR expression and specificity for different FGFs (33).

VEGF mRNA is up-regulated in virtually every tumor type examined, and in all of these cases VEGF mRNA is invariably expressed in tumor cells (35, 36, 37, 38). Other experiments have shown that the stroma around the tumor may also be an important source of VEGF (39). Immunohistochemistry demonstrates VEGF expression not only in tumor cells but also in the vasculature of these tumors, showing that VEGF is secreted as a paracrine mediator that localizes in target cells (36, 40). High VEGF expression correlates with poor prognosis in breast cancer and non-small cell lung cancer (31, 40, 41).

After the discovery of VEGF, other VEGF-related genes have been identified, encoding placental growth factor and VEGF-B, -C, and -D (38). These demonstrate variable structural homology to VEGF but have different receptor specificities (Table 1). VEGF-B is mainly found in muscle and myocardium but not endothelial cells and signals through VEGF receptor (VEGFR)-1 (42, 43). VEGF-C and VEGF-D signal through the VEGFR-2 and VEGFR-3 to stimulate lymphangiogenesis (44, 45, 46, 47). All three VEGF receptors are characterized by seven extracellular Ig homology domains in which the second and third domains are essential for ligand binding and the fourth to seventh are essential for receptor dimerization (38, 48, 49). Signal transduction involves receptor dimerization and activation of tyrosine kinase (50). Knockout studies in mice demonstrate that Flt-1 (–/–) (VEGFR-1) mice die at embryonic d 8.5 due to disorganization and overgrowth of endothelial cells (51). KDR (–/–) (VEGFR-2) knockout mice are also lethal at embryonic d 8.5 due to the lack of blood vessel development as well as inadequate hemopoiesis (52). VEGFR-3-deficient embryos die due to defective vascular remodeling before lymphatic development (53).

VEGF signals through two tyrosine kinase receptors, VEGFR-1 (Flt-1) and VEGFR-2 (KDR) (Table 1), expressed mainly on endothelial cells; VEGFR-2 is responsible for signaling, whereas VEGFR-1 may play a regulatory role (1). A third VEGF receptor (VEGFR-3) is found in lymphatic endothelium, which binds other members of the VEGF family (VEGF-C and VEGF-D) (58, 59, 60). Recombinant soluble VEGFR-1 or VEGFR-2 inhibits angiogenesis in the retina, corpus luteum, and tumors (61, 62, 63), whereas expression of a dominant negative VEGFR-2 mutant using a retrovirus inhibits signaling through the receptor and suppresses growth of tumors such as glioblastoma multiforme in vivo (64, 65). Expression of a soluble VEGFR-3 has been shown to inhibit lymphangiogenesis in mouse embryos (47).

The human VEGF gene is localized to 6p21.3, and alternative exon splicing generates four different isoforms of VEGF: VEGF 121, 165, 189, and 206 (66). Regulation of VEGF gene expression occurs via several different mechanisms, including hypoxia, growth factors, and cytokines (66). The hypoxic signal is mediated via hypoxia inducible factor (HIF)-1, a heterodimer consisting of a constitutively expressed HIF-1 ß-subunit, (HIF-1ß) and an oxygen and growth factor regulated HIF-1 -subunit (HIF-1). HIF-1 is usually unstable in the presence of oxygen and subject to oxygen-dependent ubiquination and proteasomal degradation, whereas during states of hypoxia, HIF-1 is stable (67). The dimer HIF-1 binds to the hypoxia response element (HRE) within the VEGF promoter and leads to VEGF transcription (68). Another transcription factor, HIF-2, also accumulates during hypoxia and binds to a ß-subunit to form an HIF-1 complex, activating HREs, leading to increased VEGF transcription (69, 70), and undergoing oxygen-dependent degradation. HREs are also found within genes of the glycolytic pathway and the erythropoietin gene leading to oxygen regulation of glycolysis and red cell formation (71). In vitro HIF-defective tumors do not show increased VEGF around areas of necrosis and are less vascularized (72).

Other proangiogenic growth factors and cytokines also up-regulate VEGF mRNA expression, including epidermal growth factor (EGF), TGF-ß, FGF, IL-1, and IGF-I (73, 74, 75, 76, 77). VEGF may therefore also act as a paracrine mediator for other angiogenic factors. Cell density also determines VEGF expression independent of the presence of hypoxia (78).

The angiopoietins are another family of angiogenic growth factors that signal through the Tie family of receptors (79). Ang-1 and Ang-2 are stimulatory and inhibitory ligands, respectively, at the Tie-2 receptor and important in the female reproductive cycle as well as tumor development (80, 81). Recent studies have demonstrated that the angiopoietins and VEGF in combination may fine-tune the process of angiogenesis (82).

A link between hormonal stimulation and increased angiogenesis has recently been elegantly demonstrated (83, 84) (Fig. 2) whereby castration-induced regression of prostate vasculature was reversed by testosterone-dependent angiogenic factors. Although this increase in angiogenesis occurred in the absence of androgen receptors on endothelial cells, the identity of these angiogenic factors responsive to increased testosterone is as yet unknown, but likely includes VEGF.

View larger version (24K): [in this window] [in a new window]   FIG. 2. Diagram showing proposed mechanism for hormonal regulation of angiogenesis in the prostate. [Reproduced with permission from J. Folkman: Endocrinology 139: 441–442, 1998 (84 ). © The Endocrine Society.]

Vascular density of different tumors has been assessed by counting vessels labeled using immunohistochemistry with antibodies to different endothelial markers on both frozen and paraffin-embedded sections. Antibodies that are most commonly used are directed against the endothelial antigens factor eight-related antigen (F8), CD31 (platelet endothelial cell adhesion molecule), CD34, and the lectin ulex europaeus agglutinin 1 (UEA1). These markers differ in their sensitivity for detection of endothelium (88, 89). CD31 and CD34 are the most sensitive and specific markers currently available for use on paraffin sections, although other markers may be easier to use because of practical difficulties with antigen retrieval (87). Different antibodies demonstrate varying sensitivities for detection of endothelium. F8 stains large vessels but does not label smaller microvessels (90). CD31 stains microvessels, but antigen loss can make this reagent unreliable in paraffin-embedded tissue; UEA1 stains all microvessels, but also stains the Golgi in some neoplastic cells (87, 88, 91). Choice of technique may therefore influence the results of MVD measurement.

The use of these markers reflects total vascular density, and new markers being developed are more specific for activated endothelium, but their disadvantage is that, at present, most are only effective when used on frozen tissue. They may, however, be useful to assess specific targets for therapy, e.g., new vessels or the degree of vascular remodeling. These include Tie 2/Tek, which recognizes an endothelial tyrosine kinase (92, 93); E-9, which recognizes endoglin, which is expressed on vascular cells from tumors and inflamed organs (94); and TEC-11, which recognizes a different epitope of endoglin (95). Tumor angiogenesis may also be measured in vivo by magnetic resonance imaging using a contrast agent targeted to the Vß3 integrin, which is more strongly expressed on blood vessels undergoing angiogenesis and may play a role in extracellular matrix adherence during angiogenesis (96). The use of the LH39 antibody recognizing an epitope found only in mature vessels and comparison with vascular counts using CD31, which recognizes both mature and newly formed vessels, allows a measurement of active vascular remodeling rather than a static vessel count. This vascular maturation index correlates well with tumor stage and prognosis in non-small cell lung and breast cancer (97, 98).

There are several methods for quantifying microvessel counts. The original technique described the selection of the most highly vascular areas (hot spots) and counting of individual microvessels (12). The rationale for locating the most vascular area hot spot is based on the fact that tumor behavior is likely to be related to the most vascular area of the tumor (87). Other described techniques are an overall semiquantitative grading technique, Chalkley counting using a microscope eyepiece containing a graticule containing 25 randomly positioned dots (99), and computerized image analysis (87, 100). Clearly the choice of quantification method introduces another important variable when using MVD as a reflection of angiogenesis.

The accuracy of these techniques depends mainly on the correct localization of the hot spot in addition to whether a representative block of tissue is chosen. Despite these potential problems, microvessel counts derived from all techniques have been shown to correlate with tumor prognosis and behavior (12, 15, 87).

C. Influence of hormones on angiogenesis and VEGF
The close relationship between endocrine function and angiogenesis is demonstrated during physiological processes such as the normal menstrual cycle, but also within pathological processes such as the regression in prostate vasculature after castration and the increase in vascularity of the thyroid gland during goiter development.

Table 2A summarizes the in vitro and in vivo effects of hormones on MVD/angiogenesis and/or VEGF production. Most hormones increase angiogenesis, except for glucocorticoids, which reduce VEGF expression.

In Graves’ disease, consistent with the stimulatory effect of TSH receptor antibodies on thyroid function itself, these antibodies also increase angiogenic factors (120). TSH stimulation of cultured thyrocytes was shown in a biological assay to cause endothelial cell proliferation (119).

D. Inhibitors of angiogenesis
The recognition of the therapeutic potential of inhibiting angiogenesis for treating tumors and other angiogenesis-dependent conditions has led to the study of several agents that act at different stages of the angiogenesis pathway (140, 141, 142). The control of physiological angiogenesis and maintenance of homeostasis have also led to the investigation of endogenous angiogenesis inhibitors. It has been recognized for many years that tumor (e.g., breast) metastases may remain dormant and also, curiously, that removal of a primary tumor may be followed by the rapid appearance of metastases.

Studies in a Lewis lung carcinoma mouse model have provided useful information on the mechanism of inhibition of angiogenesis. Removal of the primary tumor leads to rapid growth of tumor metastases (143). Serum and urine derived from mice bearing tumors were shown to inhibit endothelial proliferation (144). This endogenous inhibitor, termed angiostatin, was shown to be a cleaved product of plasminogen and inhibited angiogenesis and metastasis growth in vivo, although intact plasminogen lacked this activity (144). Further investigation of the behavior of dormant metastases showed that although cell proliferation of growing and dormant metastases was not different, metastases remained dormant due to a marked increase in apoptosis (145). Administration of angiostatin leads to inhibition of primary human and murine tumor growth in mice (146). The mechanism of generation of angiostatin from plasminogen has been investigated. Prostate carcinoma cells were shown to possess enzyme activity that cleaved plasminogen to active angiostatin (147, 148). The mechanism whereby angiostatin inhibits angiogenesis likely involves binding to ATP synthase in endothelial cells (149). Angiostatin administration using an adenovirus vector to a glioma tumor model (150), liposomes to a transgenic model of breast cancer (151), and sc to mice with colorectal tumors (152) leads to inhibition of tumor growth or reduction in metastases.

Endostatin, another cleavage product that acts as an endogenous angiogenesis inhibitor, was discovered from a murine hemangioendothelioma (153). This proteolytic fragment of collagen XVIII leads to inhibition of endothelial proliferation as well as increased tumor cell apoptosis leading to metastatic regression and dormancy of tumors.

Thrombospondin (TSP)1 and TSP2 are glycoproteins located within the extracellular matrix and possessing antiangiogenic activity (154). TSP1 suppresses vascular growth in vitro and in vivo, in both normal tissue and tumors (155). Expression of wt p53 has been associated with TSP1 expression providing a link between tumor suppressor gene expression and control of angiogenesis (156).

Of relevance to endocrine glands, somatostatin analogs (157, 158, 159, 160, 161) have been shown to possess inhibitory effects on angiogenesis. An endogenous substance, the 16-kDa fragment of prolactin (PRL), has also been demonstrated to possess angioinhibitory effects (162, 167).

1. Somatostatin analogs.
Using the chick chorioallantoic membrane (CAM), which allows visualization of changes in angiogenesis, the implantation of discs containing the somatostatin analog SMS 201-995 (octreotide) inhibited blood vessel growth in a dose-dependent fashion in vitro (157). Further studies using different somatostatin analogs demonstrated that the inhibition of angiogenesis was dependent on the structure of the molecule (158). Some analogs failed to inhibit angiogenesis. The mechanism for this effect, however, is not known but may involve inhibition of endothelial cell proliferation, although the precise somatostatin receptor involved is unknown. In vivo studies confirm that octreotide inhibits proliferation of human umbilical vein endothelial cells (HUVEC) (159). Studies of the antiangiogenic effect of somatostatin analogs in endocrine tumors have not been reported.

2. 16-kDa PRL and other members of the PRL/GH family (Table 2B).
The PRL molecule has been shown to give rise to a cleaved product, the 16-kDa fragment, which inhibits angiogenesis (167). There is evidence to suggest that the cleaving enzyme may be cathepsin D (168). The 16-kDa fragment has been shown in bioassays to have approximately 10% of the lactogenic effects of total PRL and 65% of the mitogenic activity (169). The full-length PRL molecule in contrast has been shown in the CAM assay to have proangiogenic effects, although it did not have demonstrable effects on angiogenesis in another in vivo assay (corneal angiogenesis) (162). These differences may be due to the maturity of the vessels studied.

Other members of the PRL/GH family and their respective cleaved fragments have also been shown in vitro and in vivo to exert opposing angioinhibitory and proangiogenic effects whereby the full-length hormone is proangiogenic and its N-terminal fragment is antiangiogenic (Table 2B). Further work is required to elucidate which of these effects are of physiological importance in vivo. However, the potential effect of an intrinsic dynamic balance of angiogenesis residing within a single hormone is an attractive concept for regulation of various physiological processes, such as human placental vascularization (162).

E. Extracellular matrix, invasion, and metalloproteinases
The matrix metalloproteinases (MMP) are a family of zinc-containing endopeptidases that are able to degrade the extracellular matrix and allow angiogenesis and tumor invasion, with each MMP acting on a different substrate (177, 178). MMP activity is balanced by tissue inhibitors of metalloproteinases (TIMPs) (179, 180, 181). MMP-2 and MMP-9 are both type IV collagenases that have been shown to be important in tumor invasion in vitro because they are able to break down basement membrane, in particular degrading collagen IV (178, 182). Elevated levels of circulating MMP-9 have been demonstrated in patients with breast cancer (183), and MMP-2 and/or MMP-9 release has been associated with tumor invasion and metastasis (184, 185, 186, 187, 188). Some authors have suggested that altered MMP-9 expression, in addition to MMP-11, characterizes epithelial tumors committed to malignant transformation, possibly relating to underlying genetic events that change the tumor phenotype to invasive (189).

MMP secretion is an important early process facilitating migration of endothelial cells through both the extracellular matrix and angiogenesis (190). In vitro studies have shown that microvascular endothelial cells do not constitutively secrete MMP-9; however, when exposed to an angiogenic stimulus (e.g., TNF-), MMP-9 production is up-regulated (191). Angiogenesis can be inhibited by endogenous TIMPs (192, 193) and also by administration of synthetic MMP inhibitors (e.g., KB-R7785) (194, 195). In addition to permissive effects on angiogenesis, MMP-9 also possesses angiostatin-converting enzyme activity, cleaving plasminogen to angiostatin, thus enhancing inhibition of angiogenesis (196), and potentially leading to a balance of proangiogenic activity and inhibition (29).

There are less data on the interaction between hormones and the MMPs (Table 2C). It is interesting to note that consistent with the down-regulatory effect of glucocorticoids on VEGF and angiogenesis, a reduction in MMP-9 expression with up-regulation of the enzyme TIMP-1 in the submucosa of the airways of patients with asthma treated with glucocorticoids has been demonstrated (197).

F. Importance of angiogenesis
The assessment and understanding of differences in angiogenesis in endocrine tumors may potentially allow accurate distinction of benign from malignant tumors (e.g., adrenal tumors and thyroid follicular neoplasms). In the pituitary gland, where it may be particularly difficult to predict subsequent tumor behavior, differences in angiogenesis may be helpful in assessing tumors more likely to recur or become aggressive and invasive. Because nonfunctioning tumors bear no useful tumor marker, it may be possible to use measurement of an angiogenic factor to determine whether complete tumor removal has been effective or whether tumor regrowth has occurred. Elucidation of angiogenesis mechanisms may help in understanding pathophysiology of hematogenous metastasis of follicular thyroid tumors and lymphatic spread of papillary tumors, or the very vascular phenotype of tumors in Von Hippel Lindau (VHL) syndrome, or why some pituitary tumors do not progress in size. It may be possible to use the measurement of angiogenesis to facilitate therapeutic decision making, for example which tumor is likely to behave aggressively and therefore require close follow-up or subsequent radiotherapy. There are many angiogenesis inhibitors in clinical trials (mainly phase I/II) (Table 3), and an understanding of angiogenesis may enable the use of angiogenesis inhibitors as therapy for endocrine tumors, e.g., synthetic MMP inhibitors for aggressive tumors or tyrosine kinase inhibitors to inhibit VEGF signaling.

	III. Pituitary Tumors

Top Abstract I. Introduction II. Mechanism of Angiogenesis III. Pituitary Tumors IV. Adrenal Tumors V. Thyroid Tumors VI. Parathyroid Tumors VII. Carcinoid Tumors VIII. Gastrointestinal... IX. Conclusions References

A. Microvascular density of tumors
1. Comparison of pituitary tumors with normal gland (Table 4).
Schechter (217) reported in 1972 that the parenchyma of pituitary tumors appeared less vascularized than autopsy specimens of normal tissue. Jugenburg et al. (218) used immunostaining for F8 to assess vascular density in a group of pituitary adenomas and carcinomas and showed that pituitary adenomas had lower vascular densities compared with the nontumorous pituitary.

View larger version (14K): [in this window] [in a new window]   FIG. 3. Comparison of vascular density between normal pituitary and different tumor types. A, Vascular density as measured by CD31 expression. B, Vascular density as measured by UEA1 staining. Normal, Normal anterior pituitary gland; Acro, GH-secreting tumors; MacPRL, macroprolactinomas; MicPRL, microprolactinomas; Cushing’s, ACTH-secreting tumors. Asterisk indicates mean value, and error bars indicate SEM. *, Statistically significant difference between MVD or normal gland and pituitary tumors, P < 0.05.

2. Angiogenesis and pituitary tumor behavior (Table 5).
Jugenburg et al. (218) showed an association between MVD as a measure of angiogenesis and pituitary tumor behavior; higher vascular densities were found in the rare pituitary carcinomas compared with benign adenomas. Another study used F8 to measure vascular density in 22 pituitary adenomas and reported that the highest vascular counts were found in FSH-expressing adenomas and lowest in GH-secreting tumors (224). Vascular density and tumor size or normal pituitary tissue were not compared.

We have shown a relationship between MVD and tumor size in prolactinomas and observed that microprolactinomas are less vascular than macroprolactinomas (Fig. 3). In contrast there was no difference between different-sized GH-secreting tumors (219). This finding is consistent with the notion that microprolactinomas represent a quite different pathological and clinical entity (227, 228) from macroprolactinomas, whereas different-sized GH-secreting tumors are part of the same disease spectrum. Additionally, MVD and serum PRL concentrations correlated positively, although no relationship with ACTH or GH secretion in ACTH-secreting or GH-secreting tumors was found (225). Bromocriptine or metyrapone treatment of patients with prolactinomas or Cushing’s disease, respectively, did not alter MVD of tumors compared with the MVD in tumors removed from patients who did not receive preoperative drug treatment. A recent study, however, has demonstrated that dopamine, signaling through endothelial cell D2 receptors, inhibits VEGF action, possibly by endocytosis of VEGFR-2 (229). Bromocriptine and quinpirole (D2 agonists) inhibit angiogenesis in a mouse ovarian tumor model (229). It might therefore be expected that bromocriptine or cabergoline should lead to angiogenesis inhibition in pituitary tumors. Two explanations for the difference between the animal model and human pituitary tumors are firstly that the typical patient undergoing surgery for a prolactinoma is resistant to, or intolerant of, the drug, and therefore a true comparison may not be valid. Alternatively, VEGF is secondary in importance to other factors (e.g., FGF-2) in stimulation of angiogenesis in pituitary tumors. In vitro studies of D2 agonists and prolactinomas will be helpful.

In terms of tumor behavior, MVD did not predict regrowth of nonfunctioning adenomas, suggesting that factors other than angiogenesis determine nonfunctioning tumor (NFA) recurrence (e.g., surgical technique). In contrast, MVD was related to tumor invasiveness and malignancy (218, 225). Higher vascular densities were found in those tumors showing local invasion of the cavernous sinus and bone and in rare pituitary cancers consistent with a relationship between aggressive pituitary tumor behavior and angiogenesis. Using a malignant pituitary tumor cell line (MGH3), PTHrP has been shown to increase new blood vessel formation in cell diffusion chambers implanted into rat sc tissue (136). Using in vitro techniques with bovine aortic endothelial cells in collagen gel and 8-bromo-cAMP, this study further demonstrated that the proangiogenic effect of PTHrP is unlikely to be directly due to increased endothelial cell proliferation or migration, but rather to capillary tube formation possibly via a cAMP pathway. Although PTHrP expression has only been shown in a metastatic pituitary GH-secreting tumor (230), it is known that PTHrP is produced by other malignant tumor types, and its role as a promoter of angiogenesis may therefore have more universal importance.

3. Relationship with cell proliferation (Table 6).
Assessment of cell proliferation using Ki-67 (a measurement of the proportion of cells within the cell cycle), showed no association between assessment of angiogenesis using MVD and cell proliferation (226), consistent with other tumor types, e.g., breast and colorectal cancer in which intratumoral MVD and Ki-67 labeling index were not related (86, 231, 232). A positive relationship was, however, observed between expression of the antiapoptotic protein bcl-2 and increasing MVD (226), suggesting that there may be an association between angiogenesis and apoptosis; a switch to an angiogenic phenotype and bcl-2 expression (preventing programmed cell death) may be early events in pituitary tumor pathogenesis. VEGF has been shown not only to stimulate endothelial cell proliferation but also to prolong endothelial cell survival by inducing expression of bcl-2 (233). Thus, angiogenesis appears to be related to the ability of pituitary tumor cells to survive, rather than their proliferative capacity.

View larger version (49K): [in this window] [in a new window]   FIG. 4. Diagram of vascular anatomy of pituitary gland. SHA, Superior hypophyseal artery; IHA, inferior hypophyseal artery. [Reproduced with permission from R. M. Bergland and R. B. Page: Science 204:18–24, 1979 (238 ). © American Association for the Advancement of Science.]

A number of controversies remain regarding the normal vascular supply to the anterior gland. Whether the blood supply to the anterior pituitary gland arises exclusively from the portal system is not entirely clear. Some authors also describe the presence of a loral artery, which arises from the anterior hypophyseal artery and enters the superior surface of the anterior lobe; in addition, they suggest that capsular arteries may also provide an arterial supply (237). Others disagree on the presence of a direct systemic source to the anterior gland (238). It is likely that the capsular arteries may supply a small number of cells in the anterior gland, but the predominant vascular supply to the normal adenohypophysis is portal in origin.

There are few studies assessing the source of the vascular supply to human pituitary adenomas. A neuroradiological study performed when angiography and pneumoencephalography were the usual imaging methods performed in the diagnosis of pituitary tumors demonstrated that at least a portion of the blood supply to tumors arose from the systemic vasculature. This was derived from branches of the internal carotid artery, most commonly the capsular and inferior hypophyseal arteries (239). A more recent neuroradiological study has extended these findings by using dynamic magnetic resonance imaging (240) and showed that enhancement of the posterior lobe could be demonstrated before the anterior lobe, as would be expected, but that macroadenoma enhancement also occurred before anterior lobe enhancement, suggesting that pituitary adenomas received a direct blood supply similar to the posterior lobe.

Intrasellar pressure in patients undergoing transsphenoidal surgery for pituitary adenoma is elevated, with a mean pressure of 25 mm Hg, compared with normal pressure of 10 mm Hg (241). This elevated pressure could obstruct the portal venous supply and therefore reduce the delivery of the normal controlling hypothalamic factors. Hyperprolactinemia observed in association with stalk compression syndrome could be related to compression of the portal blood supply. The lack of infarction or ischemia seen in the majority of tumors despite increased intrasellar pressure argues for an additional source of circulating blood. It is suggested that only a systemic arterial supply would allow maintained perfusion to the pituitary gland in this situation. The waveform of intrasellar pressure is pulsatile in two thirds of patients with tumors and in synchrony with the arterial waveform (241). The pressure was higher than venous pressures and may therefore be further evidence for a direct systemic arterial supply to pituitary tumors.

Two morphological studies also provide evidence for a direct arterial blood supply to pituitary adenomas (242, 243). Examination of the pituitary gland at autopsy of subjects with no known endocrine disorders demonstrated the presence of a microadenoma in 27%, and cannulation of the internal carotid artery and injection of dye demonstrated that two thirds of microadenomas had a direct extraportal blood supply (242). An electron microscopic examination of the vasculature within 16 prolactinomas showed that 13 contained arteries, but in contrast morphological evidence for arterial formation was not found in eight normal control pituitary glands (243). Interestingly, tumors with no arterial supply were described as small, although measurements were not provided.

These studies lend support to the notion that pituitary adenomas, in contrast to the normal anterior pituitary gland, receive a direct systemic arterial blood supply. In the liver (another organ with both a portal and direct arterial blood supply), the majority of metastases are supplied directly by the hepatic artery rather than the hepatic portal vein (244). A potential explanation for the relatively low vascular density of pituitary adenomas is that there has indeed been increased angiogenesis, but this has led to in-growth of blood vessels from the systemic circulation and the exclusion of the hypothalamic controlling factors found in the portal circulation.

Studies of the structure of pituitary tumor vasculature show the majority to be thin-walled fenestrated capillaries and give little clue as to the source of the vessel (systemic or portal). Some studies have suggested that ultrastructural changes (endothelial cell swelling and blebbing) seen in capillaries from pituitary tumor specimens may be due to tumor growth leading to compression of the long portal veins and ischemia (217). These studies have also suggested that tumor specimens were less well vascularized than the normal anterior pituitary gland derived from autopsy specimens and that the number of fenestrations may be reduced or even absent in pituitary tumors (217, 245). Altered parenchymatous basement membrane has been demonstrated in pituitary adenomas in which complete absence or fragmentation has been reported (246). This may be due to reduced synthesis or breakdown of the basement membrane by the tumor.

C. Angiogenesis in animal models of pituitary tumors
Studies of estrogen-induced lactotroph hyperplasia and tumor formation in rats with different susceptibility to estradiol have led to useful insights into the blood supply of pituitary tumors. Microspheres injected into the left side of the rat heart were trapped in the first capillary plexus they reached and did not reach the anterior pituitary gland (247). In contrast, rats with estradiol-induced lactotroph hyperplasia and PRL-secreting tumors exhibited numerous microspheres in the anterior pituitary gland, suggesting development of a direct arterial blood supply and confirmed as branches of vessels coursing through the dura (248). A direct arterial supply would be expected to potentiate lactotroph hyperplasia because the dopamine content of this source would be relatively low. Administration of bromocriptine reduced the effect of estradiol on both pituitary weight and serum PRL, as well as decreasing arteriogenesis (248). It was unclear, however, whether bromocriptine alters vasculature once formed. Hypotheses raised in this paper include the possibility that estradiol may induce vascular proliferation directly or via induction of an angiogenic factor such as FGF, or that estradiol suppressed the production of a substance that normally inhibits angiogenesis. Estrogen treatment also led to altered vascular ultrastructure coincident with tumorigenesis in estrogen-sensitive rats (249), in keeping with Schechter’s observation (217) in 1972 that tumors appeared less well-vascularized than the normal pituitary gland and blood flow per unit weight to pituitary tumors induced in rats by prolonged estrogen treatment is reduced (250).

D. Angiogenic factors
The angiogenic factors VEGF and FGF-2 were first isolated in the folliculostellate cells (FSC) of the pituitary gland.

1. FSC.
FSC were first described in the pituitary gland by Rinehart and Farquhar (251) and are characterized by their shape and long slender cytoplasmic processes. They can be identified using immunohistochemistry for S100 (252), although not all FSC immunostain positively for S100 (253). Although they compose follicles early in fetal development, they become stellate and form a reticular network throughout the adenohypophysis, connected to each other and adjacent endocrine cells by junctional complexes. It has been suggested that they play a role in compartmentalization of the anterior pituitary gland and that follicles are interconnected via FSC (254). Apart from a paracrine role, the presence of a Golgi apparatus and endoplasmic reticulum as well as phagocytic vesicles and lysosomal organelles is consistent with an active metabolic as well as a catabolic role (254, 255, 256, 257), and the production of cytokines such as IL-6 production is in keeping with a suggested homology with dendritic cells (253).

Ferrara et al. (258) studied follicular cell function after enzymatic dispersal of bovine anterior pituitary gland and in vitro culture. They demonstrated production of FGF-2 (259), a new growth factor—VEGF (260), and a possible angioinhibitory cytokine-LIF (222) from these cultures. The presence of FSC in pituitary adenomas has been investigated by different groups with conflicting results. We and others have shown that 35% (14 of 40) to 43% (61 of 93) of pituitary tumors contained S100-positive FSC (261, 262). In contrast, others find that FSC are rarely present in pituitary adenomas (263, 264). An unexplained finding is that GH-secreting tumors were most likely to contain FSC (261, 265). A universal finding is that FSC are more common in the normal anterior pituitary gland than in pituitary adenomas.

Animal models of pituitary tumors have demonstrated that FSC become activated during estrogen-induced prolactinoma formation in rats, with the appearance of cytoplasmic phagosomes and evidence of parenchymatous basement membrane degradation (257). This may be an important step in tumor progression, perhaps allowing angiogenesis and therefore tumor growth. The new technique of laser capture microdissection, which allows single cell isolation, will allow a clearer role for the FSC to be elucidated (266).

2. VEGF (Table 7A).
VEGF was first described in pituitary gland extracts (260). In animals, VEGF expression has been shown in rat, ovine, and mouse pituitary. Overexpression of VEGF in Fischer 344 rat pituitary glands and the GH3 cells has been demonstrated during estrogen-induced tumor angiogenesis (267, 273). These studies also demonstrated up-regulation of VEGFR-2 in estrogen-induced rat pituitary tumors and the presence of VEGFR-1 in sheep pituitary gland endothelial cells (267, 269).

View larger version (16K): [in this window] [in a new window]   FIG. 5. VEGF expression in a cohort of human pituitary tumors. Normal, Normal anterior pituitary gland; Acro, GH-secreting tumors; MacPRL, macroprolactinomas; MicPRL, microprolactinomas; Cushing’s, ACTH-secreting tumors. y-axis, Quantitative measure of VEGF labeling in nanoCuries per gram of tissue. Mean values are shown, with error bars indicating SEM. *, P < 0.05, for difference between normal gland and tumors; , P < 0.05, for difference between MacPRL and MicPRL.

Regulation of VEGF expression has been studied in different models (Table 7B). Estradiol, IL-6, and pituitary adenylate cyclase activating peptide (PACAP) have all been shown to increase VEGF expression, whereas dexamethasone (as in other cell types; Table 2A) inhibits VEGF. We have shown that hypoxia is an important stimulus to VEGF expression in cultured human pituitary adenoma cells (Fig. 6). Further studies are required to investigate whether this occurs via the HIF-1 signaling pathway.

View larger version (38K): [in this window] [in a new window]   FIG. 6. Graph showing VEGF concentrations in conditioned medium for primary pituitary tumor cell culture in normoxia (95% air) and hypoxia (0.1% oxygen). x-axis, Case represents each tumor studied; y-axis, VEGF (nanograms per milliliter) concentration in conditioned medium (Turner, H. E., unpublished observations).

4. LIF.
LIF has been detected in conditioned medium from bovine pituitary follicular cells in culture (222). Because LIF inhibited aortic endothelial but not adrenal capillary cell proliferation, it has been suggested that LIF may inhibit arteriogenesis in the pituitary gland, ensuring that the majority of the blood supply is capillary in origin, containing the regulatory hypophyseal factors (222). LIF has been shown to be present in normal adult pituitary tissue as well as human pituitary adenomas (289) and Rathke’s cleft cysts (290).

E. Genes and regulation of angiogenesis
PTTG was cloned from the rat pituitary cell line GH4 (285). Overexpression of rat PTTG in NIH-3T3 fibroblasts induced cellular transformation in vitro and in vivo in nude mice (285). Human PTTG is expressed in most pituitary adenomas at higher levels than in the normal gland, and human PTTG protein is demonstrable in adenoma cell cytoplasm (286). The exact role of PTTG in pituitary tumor development is as yet unclear. Recently, PTTG has been identified as a mammalian securin homolog and shown to inhibit chromatid separation, interfering with mitosis and potentially inducing genetic instability (291). PTTG expression and phosphorylation are related to mitosis, suggesting that PTTG may play a role in cell proliferation (292).

Experiments using the estrogen-induced rat prolactinoma model showed that estradiol also induced PTTG expression, followed within 24 h by increased FGF-2 and VEGF expression, associated with new pituitary blood vessel expression (281). Concordant expression of PTTG and FGF-2 was shown in human pituitary adenomas. Thus, PTTG overexpression may be an important mechanism of estrogen-induced tumorigenesis and angiogenesis. Administration of FGF-2 antibodies to conditioned medium derived from human PTTG-transfected 3T3 cells led to abrogation of angiogenesis in vitro and in vivo. The mechanism whereby PTTG is able to induce FGF-2 is currently unknown. FGF-2 has also been shown to induce cyclin D3 expression in rat pituitary (293). This may therefore provide a mechanism for an increase in cell proliferation associated with angiogenesis because cyclin D3 plays an important role in cell cycle control.

F. Matrix metalloproteinases
Tumor invasion of surrounding tissues is characteristic of more aggressive and often malignant tumor behavior. A previous small study demonstrated that three invasive pituitary adenomas showed high MMP-9 activity in contrast to four noninvasive tumors (294). More recently, an in vitro study of cultured pituitary tumors has shown secretion of another MMP type IV collagenase—MMP-2 from pituitary tumors (295). In addition, this study showed the presence of pro-MMP-9 in two thirds of conditioned medium from cultured tumors.

Assessment of MMP-2 expression using immunohistochemistry in a cohort of human pituitary tumors showed no relationship with pituitary tumor behavior (our unpublished observations). In contrast, however, MMP-9 expression was related to macroprolactinoma invasiveness and nonfunctioning tumor regrowth (296), suggesting that MMP-9 expression is related to aggressive tumor behavior. MMP-9 expression, as expected, was associated with increased angiogenesis, but the mechanism of MMP-9 regulation in pituitary tumors is unknown. Several potential upstream promoters (e.g., cAMP response element binding protein, somatostatin, cyclin D1, FGF-2), may activate MMP-9 expression (297, 298, 299, 300).

G. Therapeutic application of angiogenesis inhibitors
There are few descriptions of the therapeutic use of inhibitors of angiogenesis in endocrine tumors. There have been some preliminary data in the Fischer 344 rat estrogen-induced prolactinoma model, suggesting that a synthetic inhibitor of angiogenesis may have a beneficial therapeutic effect (301, 302). These studies both investigated the effects of the fumagillin analog TNP-470 and demonstrated a reduction in lactotroph proliferation and neovascularization. This is further evidence of the potential importance of angiogenesis in pituitary tumor behavior. Somatostatin analogs have been demonstrated in vitro to possess angioinhibitory activity (157, 158, 159, 160), although it is unknown whether this is clinically relevant in, for example, acromegaly where these analogs are frequently used to control excess GH secretion.

In contrast to other tumor types, pituitary tumors are less vascular than the normal pituitary gland. Whether this is due to endogenous inhibitors of angiogenesis controlling the angiogenic switch, the relatively low growth rate of these tumors, or an ingrowth of a vascular supply from the systemic circulation is not yet known. There is clearly a relationship between tumor behavior and MVD in some tumor types. In the future, it will be important to explore whether these differences can be used to identify tumors with a more aggressive prognosis or to use serum or urinary markers to monitor tumor progression and also assess the role of newly developed antiangiogenesis therapies in treatment.

	IV. Adrenal Tumors

Top Abstract I. Introduction II. Mechanism of Angiogenesis III. Pituitary Tumors IV. Adrenal Tumors V. Thyroid Tumors VI. Parathyroid Tumors VII. Carcinoid Tumors VIII. Gastrointestinal... IX. Conclusions References

 
Both the normal adrenal gland and adrenal tumors are vascular structures. There is a complex arrangement between the different zones of the cortex and medulla whereby different hormones are synthesized in different areas and may play a role in the metabolism of other hormones, for example the effect of glucocorticoids on catecholamine metabolism. Similar to the pituitary gland, incidental adrenal cortical tumors are commonly encountered, but malignant tumors are found more commonly in the adrenal than the pituitary gland, and the differentiation of benign and malignant can be difficult when simply observing a histological specimen. Adrenal medullary tumors are associated with various familial syndromes, e.g., multiple endocrine neoplasia (MEN) 2, neurofibromatosis type 1 (NF-1), and VHL syndrome. Differences in angiogenesis and the balance of angiogenic growth factors and inhibitors may play a role in determining the observed variations in tumor behavior. A better understanding of the underlying mechanisms may help with understanding and predicting tumor biology in the adrenal gland.

A. Microvascular density of tumors
No differences in vascular density between the normal adrenal cortex, adrenal adenomas, and carcinomas were observed using CD34 as the endothelial marker (303) (Table 8). However, MVD was related to tumor behavior in pheochromocytomas, showing that invasive tumors had a significantly higher number of blood vessels than noninvasive tumors (304). Although it is likely that more aggressive tumors do have higher levels of angiogenesis, there was also a difference in tumor size between the two groups, but the relationship between angiogenesis and size was not assessed.

Serum VEGF is elevated in patients with adrenal tumors undergoing surgery (307). One month after surgery, there was a significant reduction in serum VEGF levels in patients with both cortical cancers and benign adrenocortical adenomas. However, whether serum VEGF provides a marker of long-term disease control or recurrence is not known.

Cultured bovine adrenal cortex cells express the FGF-2 gene and release FGF-2,