Cytostatic Agents in the Management of Malignant Gliomas

Tom Mikkelsen, MD

The basic scientific studies of the angiogenic and migratory capacity of malignant
brain tumors provide new areas for potential therapeutic strategies.

Background:  Cytotoxic therapy for malignant gliomas is limited by poor delivery and drug resistance, and local therapy is ineffective in managing migratory cells. However, recent developments in malignant glioma therapy involve trials of cytostatic rather than conventional cytotoxic agents.
Methods:  The biology of the brain extracellular matrix, tumor invasion, and angiogenesis are reviewed, and the cytostatic agents that inhibit matrix metalloproteinases, angiogenesis, cell proliferation, and signal transduction are discussed, as well as studies of the angiogenic and migratory capacity of malignant brain tumors.
Results:  Two specific and interrelated areas, anti-invasion (migration) and anti-angiogenesis, are potential areas to develop new treatment strategies.  Tumor invasion and angiogenesis are important components of the spread and biologic effects of malignant gliomas.  Several proteinase inhibitors are in clinical trial, as well as anti-angiogenic agents and signal transduction cascade inhibitors.
Conclusions:  Biologic control of brain tumor cell populations may offer a new management approach to add to currently available management options for malignant brain tumors.

Outcomes in Malignant Gliomas

    The prognosis for patients with malignant gliomas has not significantly changed in recent years. Despite debulking surgery, radiation, and cytotoxic chemotherapy, median survival has changed little and is still measured in weeks. In the United States in 1995, these tumors affected 17,200 patients and caused 13,300 deaths, for a case-mortality ratio of 77%. Brain tumors constitute the No. 2 cause of cancer deaths in patients under 15 years of age, the No. 3 cause for adult men, and the No. 4 cause for women aged 15 to 34 years. In the 35-to-54-year age-group, brain tumors remain the No. 4 cause of cancer deaths in men.¹ These statistics bear out the presumption that brain tumors are highly fatal and often strike patients in their most productive years. The incidence of glioblastoma is also rising, especially in older adults,the poorest prognostic group.² Due to the lack of significant progress with conventional cytoreductiv approaches, novel therapies and approaches to therapy are well warranted.

Cytotoxic vs Cytostatic Therapy

    The concept of cytostatic agents being used to restrain tumor progression (rather than induce cytotoxic cytoreduction) has recently emerged.³ This concept questions the current therapeutic model in cancer management derived from microbiology, in which cancer cells are considered to be different from the host and these differences are exploited therapeutically. Continuing the analogy to infection, conventional wisdom has purported that unless cells are killed and totally eliminated, they will overwhelm the host.

    A regulatory model has recently been proposed in which cancer can be viewed as a dynamic maladaptive process that originates within the host, is constantly in evolution, and is potentially reversible.⁴ This model is consistent with the molecular genetic understanding of cancer processes such as clonal evolution that has been demonstrated in gliomas.⁵ One implication of such a model is that by reimposing biological control on a cell population or a malignant phenotype, functional control of a tumor may be gained without requiring complete tumor elimination. Management of these malignant phenotypes, then, constitutes a novel avenue for therapeutic research. Anti-invasion/anti-angiogenic therapy represents one such strategy in malignant gliomas and relies on a molecular understanding of these phenotypes.

Invasion in Human Glial Tumors at Onset and at Clinical Recurrence

    Gliomas in general and more highly anaplastic gliomas in particular infiltrate and spread great distances in the brain. The regional infiltration during tumor progression has been shown most strikingly in the whole-mount studies of Scherer⁶ and Burger et al⁷ in which glioblastoma cells appear to arise within a bed of better-differentiated tumor. In histological sections, most glioblastomas contain a central area of necrosis surrounded by a highly cellular rim of tumor and a peripheral zone of infiltrating cells. Infiltration of tumors cells along white matter tracts, around nerve cells, along blood vessels, and beneath the pia (secondary structures of Scherer^8,9) is responsible for local and widespread recurrence and clinical tumor progression. Angiogenesis, the proliferation of neovasculature, is also a pathologic hallmark of malignant gliomas. The recruitment and proliferation of new vessels, which typically do not form an intact blood-brain barrier, result in the pattern of contrast enhancement seen in magnetic resonance imaging (Figure).

Biology of Tumor Invasion and Angiogenesis

    The invasive nature of glioma cells and the accompanying neovasculature is perhaps the key feature in their persistence beyond therapeutic margins and is the primary reason for tumor recurrence and malignant progression. Both invading glioma cells and neovascular endothelial cells must pass through the brain extracellular matrix (ECM), a process that involves three major interrelated steps: (1) adhesion/disadhesion, (2) enzymatic degradation of the components of the parenchymal matrix, and (3) locomotion through the parenchymal barrier.^10-12

Adhesion and Disadhesion

    Coordinated adhesion and proteolysis of adhesive contacts occurs in many normal and pathologic processes, including trophoblast implantation, wound healing, tumor cell invasion, and angiogenesis. Proteinase/matrix interactions are presumed to regulate process extension by invadopodia and endothelial cells as modification of local matrix interactions permits process extension and cell locomotion. The relation of ECM adhesion and signaling with regard to tumor cell invasion is being pursued in our laboratory as well as many others institutes.

Proteolysis

    Proteolysis of brain ECM has been suggested by the observation of overexpression of all major proteinase classes, including matrix metalloproteinase (MMP),¹³ cysteine proteinase (CP),^14-17 serine/threonine proteinase (SP),^18,19 and aspartic proteinase (CD). Few functional studies have been done to substantiate these observational studies.

Locomotion

    Tumor cell locomotion involves a coordinated set of cellular responses; morphologic polarization (receptor asymmetry for integrin/cytoskeletal contacts), membrane extension (invadopodia), cell-substratum attachments, contractile force/traction, and release of focal attachments.²⁰ Cathepsin B has been localized to focal regions in breast cancer and glioma cells in contact with the ECM,²¹ and the proteolysis of matrix components can be seen beneath such focal contacts, which can be inhibited by cathepsin B inhibitors. Inhibitors of cathepsin B inhibit melanoma cell motility induced by autocrine motility factor (AMF)²² in melanoma cells and in response to glioma-conditioned media (T.M., unpublished observations, 1997).

Molecular Mechanisms of Angiogenesis

    Angiogenesis, the formation of new blood vessels, occurs in a variety of normal and pathologic conditions.²³ In physiologic states, such as embryonic development and wound healing, neovascularization is a strictly regulated balance of expression of stimulatory and inhibitory angiogenesis factors.²⁴ The disruption of this finely tuned regulatory pathway and the formation of a pathologic capillary network occur in a variety of disease states, including cancer, diabetic retinopathy, hemangiomata, and vasculitides.²⁵ Tumor neovascularization begins with the sprouting of new capillary buds from an existing vessel in response to direct or indirect angiogenic stimuli. The angiogenesis response occurs as a result of proteinase secretion and basement membrane remodeling, endothelial cell proliferation, and endothelial cell migration to form capillary sprouts and neovascular lumina.²⁶ The parallels between tumor cell invasion and endothelial cells in angiogenesis are striking. For example, the role of the lysosomal proteinase cathepsin B (CB) in the process of angiogenesis has been shown in invasive prostate cancer by immunoelectron microscopy and in situ hybridization.²⁷ CB was demonstrated in proliferative neoendothelial cells in the invading zone. Our own work in immunohistochemistry has demonstrated CB expression, not only in tumor cells, including the infiltrating margin, but also in neovascular endothelial cells.²⁸

Invasion and Angiogenesis: Proteinases, Inhibitors, and Malignancy

    The proteinases that participate in malignant progression are numerous. Among the proteinases implicated in the progression of animal and human tumors are members of the four classes of endopeptidases: (1) matrix metalloproteinases such as stromelysin and gelatinases A and B, (2) serine proteinases such as urokinase, (3) aspartic proteinases such as cathepsin D, and (4) cysteine proteinases such as cathepsin B. There is an increasing awareness of the role played by cell surface proteinases in the malignant phenotype, due in part to the activation of other matrix metalloproteinases at the cell surface by the recently discovered membrane-associated matrix metalloproteinases.

    Proteinases may affect infiltrative capacity of tumor cells in several ways. First, proteinases are capable of degrading ECM and basement membrane (BM) components, which act as barriers to tumor infiltration and metastasis. Limited degradation of the ECM, upon which cells also migrate, divide, and differentiate, allows movement of tumor cells through perivascular channels and white matter (myelin) tracts of the brain. Expression of ECM components is largely limited to the perivasculature of the brain. Production of several ECM components is altered in intracranial tumors. As the complement of proteinases in both intracranial and extracranial tumors is similar, it is possible that the unique BMs and ECMs of the tumor perivasculature prevent formation of brain tumor metastases to tissues outside the central nervous system (CNS). The mechanisms by which MMPs and uPA degrade ECM and BM surrounding arteries and veins of injured brain have been described.²⁹

    In addition to opening migratory pathways, proteinases can alter cell adhesion properties regulated through several classes of cell surface receptors. These receptors, including cadherins, CD-44, integrins, and receptors for fibronectin, laminin, and vitronectin, negatively regulate cell motility and growth through cell-cell and cell-matrix interactions.³⁰ Thus, proteolytic degradation of receptors and/or ECM components could release tumor cells from these constraints. Proteolysis of cell-matrix interactions is tightly controlled by tumor cells that must maintain a substratum upon which to move at their leading edge while detaching from that same support at their trailing edge. This regulation may be accomplished through increased production of proteinases at the leading edge of the tumor where they are in an ideal location to down-regulate proteolytic activity. As described below, the increased expression of several inhibitors has been positively correlated with increased infiltrative capacity of several tumors. Although contradictive at first glance, up-regulation of inhibitors maintains the balance between proteolysis and inhibition. This balance is required for the cyclic attachment of tumor cells to the ECM, followed by focal dissolution of ECM components and substrate-binding cell surface receptors and release from the ECM. Inhibitors not only protect tumor cells from degradation during this process, but also ensure focal degradation of the ECM. Proteinases and inhibitors are known to be secreted from both tumor and host cells and to be stored in the ECM. Growth factors are also trapped in the ECM and may also be released upon its degradation.

    Immunohistochemical studies of proteinases in both gliomas and extracranial tumors have indicated that they may also play a role in angiogenesis. Further, because the BM of new arterioles and veins is incomplete, tumor cells may be able to migrate through this partial barrier and metastasize to distant regions of the CNS and occasionally to extracranial sites. Several proteinases and proteinase inhibitors have been implicated in these processes leading to tumor progression and infiltration, as already noted. The putative role(s) of individual proteinases and inhibitors in intracranial tumor cell infiltration are discussed in more detail below.

Matrix Metalloproteinases and Inhibitors

    MMPs are metal-dependent endopeptidases that may be divided into two classes: those that are secreted (as inactive zymogens) and the newly described membrane-type MMP (MT-MMP), which is associated with the cell surface via a transmembrane domain near the carboxy-terminus.³¹ Only one study has addressed the expression of MT-MMP in brain tumors.³² Results of Northern blot, reverse transcriptase-polymerase chain reaction, and immunohistochemical analyses in this study indicated that MT-MMP mRNA and protein are expressed in astrocytoma cells but not in normal brain tissue. Furthermore, expression of MT-MMP was shown to positively correlated with gelatinase A expression during malignant progression of gliomas. Interestingly, in both of these studies, using immunohistochemistry MT-MMP protein was localized to tumor cell surfaces. MT-MMP has been shown to activate pro-gelatinase A in the absence of tissue inhibitor of metalloproteinase (TIMP)-2.³³ As discussed below, gelatinase A is expressed in several human brain tumors and tumor cell lines. Amberger et al³⁴ also described a membrane-bound MMP purified from rat C6 glioblastoma cells. Homology of this proteinase and MT-MMP has not been determined. When O-phenanthroline (an inhibitor for MMPs) or a synthetic substrate selective for MMPs was added to C6 cultures, spreading on myelin plates was inhibited. This may indicate a role for MT-MMPs in rat glioblastoma cell migration or invasion.

    Abe and colleagues³⁵ demonstrated a correlation between increasing gelatinase A expression at the mRNA level and glioma cell-line invasion as measured with Matrigel barrier invasion assays. In this study, nine cell lines demonstrating variable abilities to invade Matrigel were examined for gelatinase A expression by Northern blotting. Those cell lines most active in the invasion assay also contained the highest amount of gelatinase A mRNA. Gelatinase A mRNA production has also been detected in glioma cell lines by Costello et al.³⁶ Expression of matrilysin and stromelysin message in glioma cell lines has been shown to be highly variable and does not seem to correlate with invasive capacity.³⁷ Similarly, the expression of interstitial collagenase mRNA seems to vary according to the glioma line examined.³² Expression of gelatinase B and stromelysin-2 has not been examined in intracranial tumor cell lines at the mRNA level.

    At the level of protein activity, several investigators have detected gelatinase A in conditioned media of intracranial human tumor cell lines^35,36 and rat BT5C glioblastoma cells.³⁸ Levels of gelatinase A mRNA production correlate with protein activity and expression.³⁵ Gelatinase A activity as measured by zymography was highest in those cell lines that were most invasive as measured by the Matrigel invasion assay. Both the zymogen and active forms of gelatinase A are secreted by CNS tumor cell lines.^35,36 Although such results may argue for a role for gelatinase A in intracranial infiltration, the ECM of the brain does not resemble the makeup of Matrigel. Thus, these studies, like those in extracranial tumors, may imply that gelatinase A may be involved in intracranial infiltration but do not provide direct evidence of such a phenomenon. As with mRNA expression, the levels of protein and activity have not been determined for gelatinase B, interstitial collagenase, or stromelysin-2 in vitro. Likewise, the examination of matrilysin and stromelysin protein expression and activity has yet to be undertaken in intracranial tumor cell lines.

    The major inhibitor of gelatinase A is TIMP-2, which prevents degradation of solubilized collagen by gelatinase A purified from the rat glioma cell line BT5C.³⁹ Lund-Johansen and coworkers⁴⁰ have shown that gelatinase A purified from BT5C glioma cell-conditioned media is capable of destruction of fetal rat brain aggregates in a manner similar to that observed for normal rat brain spheroids confronted with BT5C spheroids. Such results s