Angiogenesis and Apoptosis

Seminars in Cancer Biology 13 (2003) 159–167

Angiogenesis and apoptosis Judah Folkman∗ Department of Surgery, Children’s Hospital and Harvard Medical School, Hunnewell 103 300 Longwood Avenue, Boston, MA 02115, USA

Abstract

This review assembles the laboratory and clinical evidence that cytotoxic chemotherapy and antiangiogenic therapy are each dependent on endothelial cell apoptosis. During cytotoxic chemotherapy, apoptosis of endothelial cells in the vascular bed of tumors precedes apoptosis of tumor cells, even when the tumor has been made drug resistant. Administration of an angiogenesis inhibitor which is not directly cytotoxic to tumor cells can increase tumor cell apoptosis and inhibit tumor growth by inhibiting endothelial proliferation and migration and/or by inducing endothelial apoptosis. Furthermore, oncogene expression and loss of tumor suppressor gene activity can at once protect tumor cells against apoptosis and increase their angiogenic output. Both of these survival advantages conferred on the tumor can be overcome by antiangiogenic therapy. They can also be overcome by cytotoxic chemotherapy administered on a low dose ‘antiangiogenic schedule’ which continuously exposes endothelial cells in the tumor bed to the drug. As a result, endothelial apoptosis can be demonstrated to precede tumor cell apoptosis, and tumors regress or are inhibited, whether or not the tumor cells are resistant to the drug, and with little or no host toxicity. In contrast, cytotoxic chemotherapy administered on a ‘conventional schedule’ of maximal tolerated dose followed by an off-therapy interval, becomes ineffective after drug resistance is acquired. On the basis of these experimental ﬁndings, chemotherapy of cancer may possibly be improved—i.e. decreased drug resistance and decreased toxic side-effects—by changing dose and schedule to maximize apoptosis of endothelial cells in the vascular bed of tumors. Further improvement may be achieved by combining angiogenesis inhibitors with ‘antiangiogenic chemotherapy’. © 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Angiogenesis; Apoptosis; Antiangiogenic chemotherapy

1. Introduction 2. The angiogenic switch in a tumor is followed by decreased apoptosis Angiogenesis, the growth of new capillary blood vessels, is critical for development, reproduction and repair and Tumor growth is angiogenesis-dependent (for experimen- dominates many pathological conditions. The hypothesis tal evidence see reviews [2,3]). In the absence of angiogen- that tumor growth is angiogenesis-dependent was first pro- esis, tumor growth is restricted to a microscopic size; tumor posed in 1971 [1,2] and has subsequently been proven by a cells do not shed into the circulation. In non-angiogenic in variety of experiments including genetic methods [3]. The situ tumors [4] or in dormant micro-metastases [5,6] tumor ability to culture endothelial cells in vitro, the development cell proliferation continues (e.g. up to 30% BrdU labeling in of bioassays for angiogenesis, the discovery of molecules experimental tumors) balanced by high rates of tumor cell which stimulate angiogenesis as well as those which inhibit apoptosis (up to 8–10% apoptotic cells identified by frag- this process, have led to the emergence of an active field mented DNA with the TdT staining technique [5]). High of angiogenesis research now pursued worldwide. A fun- rates of human tumor cell apoptosis can persist for as long damental question of how angiogenesis is linked to apop- as a tumor remains non-angiogenic. Non-angiogenic clones tosis of endothelial cells and of tumor cells is addressed of tumor cells were isolated from human osteosarcomas and here. transplanted from one SCID mouse to another after 8 months residence in each mouse as a <0.5 mm dormant subcutaneous tumor [7]. After 3 years, a small percentage of these dormant tumors spontaneously became angiogenic. In con- ∗ Tel.: +1-617-355-7661; fax: +1-617-355-7662. trast, non-angiogenic dormant human liposarcomas became E-mail address: [email protected] (J. Folkman). angiogenic within 2–4 months of transplantation in SCID

1044-579X/03/$ – see front matter © 2003 Elsevier Science Ltd. All rights reserved. PII: S1044-579X(02)00133-5 160 J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 mice [4]. Other human tumors, such as breast cancer spon- non-angiogenic and angiogenic, are mixed at different ra- taneously switched to the angiogenic phenotype over a pe- tios before implantation into SCID immunodeficient mice riod of 6 months. Therefore, an in situ dormant tumor may reveal that as few as 1% angiogenic cells will lead to a neo- not be harmful to its host until it has recruited microvascular vascularized tumor [7]. However, the latent period before a endothelial cells. detectable tumor appears is longer with a lower percentage The angiogenic switch can be induced at a predictable of angiogenic cells (e.g. months with 1% angiogenic cells time and usually more rapidly, by introduction of oncogenes versus weeks with 50% angiogenic cells). into the tumors. When the human non-angiogenic osteosar- While pre-existing microvessels, are surrounded by nor- coma was transfected with the ras oncogene, neovascular- mal cells, under some conditions such as in brain metastasis, ized tumors appeared within 2 weeks [7]. In transgenic mice tumor cells can displace normal cells around a microvessel, in which the large T-antigen was expressed in all ␤-cells a process of vessel cooption [16]. However, new tissue mass of the pancreatic islets, carcinomas appeared at approxi- requires the recruitment of new microvessels. This principle mately 4 weeks of life and the angiogenic switch occurred applies to tumors as well as to normal tissue (such as fat) at approximately 7 weeks in 4–10% of islets [8,9]. The [17]. Neoplastic tissue usually exceeds the oxygen diffusion angiogenic switch occurred slightly later in fibrosarcomas limit when tumor cell layers accumulate to a thickness of ap- arising in transgenic mice carrying the bovine papilloma proximately 150–200 ␮m from a nearest open microvessel. virus oncogenes [10], or when squamous cell carcinomas Tumor cells beyond this limit undergo apoptosis (Fig. 1). arose in mice carrying the human papillovirus type 16 Therefore, almost any tumor that has reached a diameter of oncogenes targeted to basal cells of the epidermis [11]. The >10–100 mm, is probably already neovascularized. spontaneous angiogenic switch [9] in human tumors can be After the angiogenic switch, new microvessels converge driven by: (1) angiogenic oncogenes [12] which up-regulate on the dormant in situ tumor and tumor cells cluster around expression of pro-angiogenic proteins (i.e. VEGF, bFGF each microvessel in a cylindrical configuration. The radius etc.), and/or down-regulate expression of angiogenesis in- of the microcylinder of tumor cells is limited by the oxygen hibitors, such as thrombospondin); (2) tumor-associated hy- diffusion requirements for that particular tumor, but it would poxic conditions which activate hypoxia-inducible factor-1 be rare to find layers of viable tumor cells beyond 200 ␮m (HIF-1) [13], which itself up-regulates angiogenic proteins; (Fig. 1) [18]. The angiogenic switch is associated with a (3) fibroblasts in the tumor bed which can be induced by marked decrease (3- to 4-fold) in overall tumor cell apoptosis tumor cells to elaborate pro-angiogenic proteins [14]; and [5]. In contrast to apoptosis, the proliferation rate of tumor (4) bone marrow derived progenitor endothelial cells which cells may remain at the same level and in many angiogenic traffic to tumors [15]. Of interest is that not all cells in a tumors appears to be relatively independent of the onset of non-angiogenic dormant human tumor must switch to the angiogenesis or its intensity [19]. angiogenic phenotype for a tumor itself to become neovas- The mechanisms of increased tumor cell survival and de- cularized. Experiments in which human cancer cells both creased tumor cell apoptosis after the onset of angiogenesis,

Fig. 1. Neovascularized transplanted tumors growing in SCID immunodeficient mice to show supported viable tumor cells forming perivascular cuffs (indicated by black dashed ovals). Perivascular cuff size is roughly indicative of the metabolic burden of the tumor cells and their ability to survive under hypoxic conditions. Human melanoma cells (a) within approximately 80 ␮m of the vasculature are viable. Beyond this radius, at the limits oxygen and nutrient diffusion from the microvessel, an abrupt shift to tumor cell necrosis is observed. Prostate carcinoma cells (Dunning rat), (b) cannot exceed an oxygen/nutrient limit of 110 ␮m. Sections in both panels (a) and (b) were stained with hematoxylin for DNA, highlighting areas of necrosis, and with an antibody to CD31 in (b) showing the endothelial lining of blood vessels. (a) From Nava Almog in the Folkman laboratory, unpublished. (b) From Hlatky et al. [18] with permission from the publisher. J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 161 depend not only on delivery of oxygen and nutrients and the hibitors endostatin or TNP-470 (Shay Soker, unpublished removal of catabolites by new microvessels, but also on a data). paracrine release of anti-apoptotic factors from the endothe- Several clinical studies indicate that despite an initial lial cells in these new vessels. Microvascular endothelial increase in tumor blood flow during the early phase of an- cells recruited into a tumor bed supply at least 20 mitogens, tiangiogenic therapy, chronic antiangiogenic therapy causes anti-apoptotic factors, and survival factors (including among total tumor blood flow to reach a steady state or to gradu- others, bFGF, HB-EGF, IL-6, G-CSF, IGF-1 and PDGF) for ally decrease. When endostatin was continued for weeks or those tumor cells apposed to the new microvessels [20–22]. months in cancer patients, PET scans revealed a gradual, It is unclear whether angiogenesis inhibitors also directly dose-dependent reduction in total tumor blood flow [28]. decrease endothelial cell production of paracrine factors, but This could be caused by a dropout of individual microves- this could be tested in vitro. sels, followed by apoptosis of the surrounding tumor cells. Other studies of tumor blood flow during antiangiogenic therapy have been reported by clinical investigators at the 3. Antiangiogenic therapy increases apoptosis of National Cancer Institute [29–31]. In addition, there are sev- endothelial cells and tumor cells eral ongoing trials currently utilizing non-invasive imaging to assess changes in tumor blood flow following treatment Antiangiogenic therapy can inhibit further endothelial with angiogenesis inhibitors. These include two clinical proliferation in a tumor bed or induce endothelial cell trials at the National Cancer Institute evaluating anti-VEGF apoptosis, depending on the potency of the antiangio- antibody, one in renal cancer and the other in breast can- genic therapy versus the total angiogenic output which it cer (Steven Libutti, personal communication). There are must overcome. A tumor’s total angiogenic output can be also clinical trials of “anti-vascular” therapy which shuts thought of as the sum of the positive angiogenic regula- off tumor blood flow early after initiation of therapy, and tors (e.g. VEGF, bFGF, PDGF), and negative angiogenic which may be antiangiogenic later. An example would be regulators (endostatin, angiostatin, tumstatin), which the combretastatin [32]. tumor generates. Angiogenic output can differ from one Despite an initial temporary increased blood flow by some tumor type to another or among tumors of the same type. angiogenesis inhibitors, the resulting decreased leakage of For example, one study showed that 60% of breast cancers plasma proteins from microvessels in the tumor bed [27], express only one pro-angiogenic protein, VEGF, while 4 may contribute to the initiation of tumor cell apoptosis, or 5 additional pro-angiogenic proteins can be expressed which would continue as capillary drop-out continued. This by other breast cancers [23]. A human pancreatic cancer, speculation however, remains to be demonstrated. Never- ASPC-1, generated no detectable negative regulators of theless, these findings suggest that at least one rational de- angiogenesis and grew twice as fast in SCID mice as an- sign of a clinical trial for an angiogenesis inhibitor may be other human pancreatic cancer BxPC3 which generated to provide for long-term therapy with assessment of tumor three angiogenesis inhibitors including high levels of an- blood flow at constant time intervals. If a patient’s tumor giostatin [24]. Tumor cells proliferated at the same rate progresses during therapy, a provision in the trial design to in both tumors, 60% proliferating cell nuclear antigen increase the dose of inhibitor may halt tumor progression or (PCNA). bring about the original stable state. Because antiangiogenic While prolonged antiangiogenic therapy gradually de- therapy does not inhibit DNA synthesis in tumor cells—at creases blood flow to tumors, an apparent paradoxical early least at the beginning of therapy—these cells should remain effect is a temporary increase in blood flow. Teicher et al. susceptible to chemotherapy, unless they have acquired drug demonstrated that antiangiogenic therapy in tumor-bearing resistance mechanisms. animals increased tumor blood flow and oxygen delivery— at least during the first weeks of therapy [25]. This may explain why some tumors increase their size slightly before 4. Cytotoxic chemotherapy causes endothelial decreasing it during uninterrupted antiangiogenic therapy. apoptosis which precedes tumor cell apoptosis While these experiments were short-term, the presumed mechanism was that antiangiogenic agents reduced leak- When conventional cytotoxic chemotherapeutic agents are age of plasma proteins from tumor vessels, resulting in administered systemically, they must cross microvascular decreased intratumoral pressure [26]. Studies by Jain in- endothelium before they reach tumor cells. Because endothe- dicate that antiangiogenic therapy may normalize tumor lial cells in the tumor bed have significantly higher prolif- vasculature to make it less leaky [27]. Soker working with eration rates than quiescent endothelium in the rest of the my laboratory showed that when Evans blue dye was in- body, they should be more susceptible to cytotoxic agents jected intravenously into mice, subcutaneous injection of than resting endothelium (Fig. 2) [33]. VEGF or platelet activating factor (PAF) caused a large Furthermore, endothelial cells are genetically stable and blue stain to appear at the injection site (Miles test), which do not develop p53 mutations, so they should in principle be was prevented by prior treatment with the angiogenesis in- more susceptible to the apoptotic effects of cytotoxic agents 162 J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167

Fig. 2. Mass distributions with respect to turnover time for various tissues. Data-derived 3-dimensional histogram of tissue mass with respect to turnover time of tumor cells (human esophageal cancer) (blue mass) among the treatment-limiting tissues of gut, bone marrow and skin (gray masses). Turnover times are longer with increasing distance to the right. A typical tumor has a small fraction of highly proliferating cells and a substantial fraction of non-growing or slowly growing cells. Conventional cytotoxic chemotherapy targets proliferating cells in the tumor, but its use is limited by coincident injury to normal tissues, in accordance with the individual proliferation rates of those tissues. Repeated application of chemotherapy also resultsin drug resistance because of the genetic instability and high mutation rate of tumor cells. In contrast, antiangiogenic therapy which preferentially or speciﬁcally targets highly proliferating endothelial cells in the tumor bed (red, left) restricts growth of all tumor cell populations. Resistance is less likely to develop because of the genetic stability and low mutation rate of endothelial cells. Moreover, the wide separation in turnover times between the tumor-associated endothelial cells and the non-proliferating endothelial compartment (red mass, right), likely underlies the apparent insensitivity of the normal endothelium to these novel antiangiogenic agents. Therefore, antiangiogenic therapy directed toward the tumor endothelial cell compartment circumvents two major therapeutic issues: toxicity to normal tissues and drug resistance. When low dose chemotherapy is administered at frequent regular intervals (‘antiangiogenic chemotherapy’; also called ‘metronomic’ therapy), it behaves as an angiogenesis inhibitor by its continuous exposure to proliferating endothelial cells in the tumor bed. Tumor growth can be inhibited in experimental animals regardless of whether tumor cells are drug resistant. (From Folkman et al. [33] with permission of the publisher.) than tumor cells. In fact, several chemotherapeutic agents anti-tumor efﬁcacy of chemotherapy in vivo. While it is act as angiogenesis inhibitors in addition to their ability to relatively simple, however, to study chemotherapy-induced induce direct cancer cell death. Paclitaxel, which inhibits mi- apoptosis of endothelial cells or tumor cells separately crotubule polymerization, inhibits vascular endothelial cell in vitro, it is not trivial to analyze apoptosis in these proliferation, motility and invasiveness in a dose-dependent two populations when they are crowded together in a manner in vitro and tumor angiogenesis in vivo [34]. There- tumor. fore, the apoptotic effects of cytotoxic chemotherapy on Browder et al. approached this problem by carrying proliferating vascular endothelium could contribute to the out animal experiments in which the apoptotic effect of

᭤ Fig. 3. ‘Antiangiogenic chemotherapy’ with cyclophosphamide administered systemically at frequent regular intervals induces a peak of endothelial cell apoptosis in tumor vasculature which is followed by a peak of apoptosis in tumor cells 4 days later. Because in this experiment the tumor cells have been made resistant to cyclophosphamide, the inhibition of tumor growth is due primarily to induction of endothelial apoptosis. This experiment reveals that cytotoxic chemotherapy may be endothelial-dependent. (A) In the lower panels endothelial cells of blood vessels in the drug-resistant tumor are stained red by antibody to von Willebrand factor (conjugated to Texas red). Apoptotic cells are stained green (TUNEL). Merged images show that at approximately 12 h after administration of cyclophosphamide on the ‘antiangiogenic schedule’, endothelial cells (yellow nuclei) are undergoing apoptosis. 4 days later (right lower panel), tumor cells are undergoing apoptosis. (B) In the upper panel, (drug-resistant tumor) cyclophosphamide is administered on the conventional schedule, i.e. maximum tolerated dose of 150 mg/kg, every other day for three doses. This schedule suppresses bone marrow, so that a second cycle cannot be administered until 21 days (not shown here) after the last maximum tolerated dose. During that off-therapy interval, endothelial cell apoptosis decreases followed by a reduction in tumor cell apoptosis and the tumor resumes its growth. In contrast, on the antiangiogenic schedule (lower panel) a single dose of cyclophosphamide (170 mg/kg) is administered every 6 days. This schedule does not suppress bone marrow. Each peak of endothelial cell apoptosis is followed by approximately 4 days later by a peak of tumor cell apoptosis. With each wave of tumor cell apoptosis, the tumor is completely inhibited for up to 55 days following which there is very slow growth. When another inhibitor of angiogenesis (TNP-470) is added to the regimen (at a dose which by itself would only inhibit this tumor by approximately 65%), tumors undergo complete regression. (Adapted from Browder et al. [35] with permission of the publisher.) J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 163 164 J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 cytotoxic chemotherapy on endothelial cells in a tumor conventional chemotherapy. Antiangiogenic chemotherapy bed was dissociated from the apoptotic effect of cytotoxic has also been called ‘metronomic’ therapy [45], but the chemotherapy on tumor cells supported by these endothelial two terms do not have precisely the same meaning. ‘An- cells [35]. Browder et al. hypothesized, that the tradition tiangiogenic chemotherapy’ signifies that the target of the of administering chemotherapy at maximum tolerated dose chemotherapy is microvascular endothelium in the tumor (MTD), required a treatment-free interval to allow recovery bed. ‘Metronomic’ therapy indicates that the schedule of ad- of the bone marrow and gastrointestinal tract. During the ministration is at very regular intervals. treatment-free interval, microvascular endothelial cells in In summary, by choosing a murine tumor that is com- the tumor bed could resume their proliferation and support pletely resistant to a standard cytotoxic chemotherapeutic tumor regrowth. When tumor-bearing mice were treated agent, and changing the schedule and dose of this drug to with cyclophosphamide at a conventional maximum tol- provide maximum continuous exposure of microvascular en- erated dose followed by a 21-day interval to rescue bone dothelial cells in the tumor bed, it was possible to show that: marrow, (‘conventional schedule’), tumors recurred after (1) endothelial cell apoptosis precedes tumor cell apoptosis; each cycle. All animals eventually died of large tumors (2) tumors can undergo regression; (3) drug resistance can which had a diminished response to cyclophosphamide be avoided; and (4) toxicity can be dissociated from thera- after each cycle, i.e. the tumors developed acquired drug peutic efficacy. resistance. In contrast, if cyclophosphamide was adminis- Oncogene expression protects tumor cells against apop- tered more frequently and at lower doses (‘antiangiogenic tosis and increases their angiogenic activity. Both survival schedule’), without a prolonged treatment-free interval, all advantages for tumor cells can be overcome by antiangio- tumors regressed and all animals survived without bone genic therapy. marrow suppression or other toxicity. Then the tumors were Kerbel and colleagues reviewed the impact of 15 dif- made drug-resistant to cyclophosphamide. On the ‘conven- ferent oncogenes on tumor angiogenesis [12,46,47]. These tional schedule’, animals died even more rapidly with large pro-angiogenic oncogenes (in addition to others [48,49]) tumors as drug resistant tumor cells continued to proliferate up-regulate expression by tumor cells of angiogenic pro- unaffected by cyclophosphamide. However, the ‘antiangio- teins and/or down-regulate inhibitors of angiogenesis (for genic schedule’ significantly inhibited tumor growth, but recent review see [46]). Other experimental studies support there was still very slow tumor growth. It was not possible this concept. Normal cells immortalized with SV40 large to raise the low dose of cyclophosphamide to a level suffi- T, produce microscopic avascular tumors in mice, which cient to induce apoptosis of the remaining endothelial cells remain dormant [50]. Transfection of the immortalized cells in the tumor bed, without provoking bone marrow suppres- with the ras oncogene resulted in neovascularized tumors. sion. Therefore, a low dose of a pure angiogenesis inhibitor, In a different experimental system, a doxycycline-inducible TNP-470 was added. This inhibitor by itself could not slow H-ras mouse neovascularized melanoma, down-regulation this tumor by more than 65%. However, when combined of the ras oncogene leads to endothelial apoptosis in the tu- with cyclophosphamide on the antiangiogenic schedule, mor bed that precedes tumor cell apoptosis [51]. Activating there was complete tumor regression and long-term sur- mutations in K-ras and H-ras up-regulate VEGF expres- vival in all but one animal. Immunohistochemical analysis sion and down-regulate expression of thrombospondin (an (Fig. 3) revealed extensive endothelial apoptosis that pre- angiogenesis inhibitor) [7,47]. When the Bcl-2 oncogene ceded tumor cell apoptosis by 4–5 days, despite the fact which mediates inhibition of apoptosis [52], was transfected that the tumor cells were resistant to cyclophosphamide into tumor cells, tumor cell apoptosis decreased and VEGF and not directly affected by TNP-470. This experiment re- expression increased significantly [48]. In another study, veals an endothelial-dependence of cytotoxic chemotherapy human osteosarcoma cells implanted in mice formed only that is concealed before tumor cells have acquired drug microscopic avascular, dormant tumors, in which tumor resistance. Klement et al. obtained similar results by admin- cell proliferation was balanced by tumor cell apoptosis. istering continuous low dose vinblastine to tumor-bearing When these cells were transfected with the ras oncogene, mice in combination with a VEGF receptor antibody VEGF expression doubled, expression of thrombospondin [36]. decreased significantly, and large neovascularized tumors Furthermore, these results might also help to explain why grew within approximately 2 weeks [7]. Thus, targeting some patients who receive long-term maintenance or even oncogene products not only affects cancer cell prolifera- palliative chemotherapy have stable disease beyond the time tion and cell death, but also disrupts production of angio- that the tumor would have been expected to develop drug genic factors. These studies predict that certain anti-cancer resistance. Patients with slow-growing cancers who are on drugs developed for their capacity to block an oncogene ‘antiangiogenic scheduling’ of chemotherapy involving con- product (for example, inhibitors of the EGF receptor tyro- tinuous infusion 5-fluorouracil [37–39], weekly paclitaxel sine kinase), may have significant antiangiogenic activity [40,41], or daily oral etoposide [42–44], have shown an [12,53,54]. For example, ras farnesyl transferase inhibitors improved outcome—despite the fact that in some of these block oncogene signaling pathways which up-regulate patients the tumors had already become drug resistant to tumor cell production of VEGF and down-regulate J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 165 production of the angiogenesis inhibitor, thrombospondin-1 6. Antiangiogenic gene therapy [47]. Mutations in the tumor suppressor gene p53 also protect Prolonged use of angiogenesis inhibitors is envisioned tumor cells against apoptosis and increase tumor angiogenic for cancer patients. Therefore, in the future, antiangiogenic activity. Both advantages may be overcome by antiangio- gene therapy [33,66] may be important for protein angiogenic therapy. genesis inhibitors, especially because a constant level of Tumor cells with p53 mutations may become resistant to these inhibitors in the circulation provides more effective apoptosis [55], and thus decrease their response to cytotoxic anti-cancer therapy than intermittent peaks of inhibitor in chemotherapy. In mice, p53 mutations may also decrease mice [62]. There have been many successful experimental the response of cancer cells to antiangiogenic therapies [56]. studies of antiangiogenic gene therapy [68–76], and some Wild-type p53 normally suppresses tumor angiogenesis by negative results [71,77,78]. Systemic delivery of recombi- up-regulating thrombospondin-1 [57], inducing degradation nant adenoviruses encoding the ligand-binding ectodomains of HIF-1-alpha [58], suppressing transcription of VEGF of the VEGF receptors Flk1 and Flt1 resulted in about 80% [59] and down-regulating bFGF-binding protein expression inhibition of tumor growth in animals bearing murine Lewis [60]. In summary, loss of p53 function increases the total lung carcinoma, T241 fibrosarcoma, or human BxPC3 pan- angiogenic output of a tumor. However, it is possible to creatic carcinoma [72]. But, it was puzzling that systemic overcome this increased angiogenic activity of the tumor administration of adenoviruses encoding angiostatin, endo- by raising the dose of an angiogenesis inhibitor, or by com- statin or neuropilin were significantly less effective [71]. bining different angiogenesis inhibitors, in other words, On the other hand, when endostatin was transfected into by titrating against the total angiogenic output of a tumor, tumor cells which were then implanted into mice, there was analogous to the titration of insulin against blood sugar almost complete inhibition of tumor growth [72]. The ap- [61]. We demonstrated a dose-dependent response to a sin- parent difference in anti-tumor efficacy of local endostatin gle angiogenesis inhibitor, varying from 33% inhibition to gene transfer (high efficacy) versus systemic endostatin 97% inhibition of tumor growth with tumor regression [62] gene therapy, when endostatin at very high blood levels in mice bearing a mutant p53-associated human pancreatic (low efficacy), is not clear. One possibility is that systemic tumor [55]. There is currently no quantitative method for gene therapy produces significantly higher plasma levels of determining the total angiogenic output of a patient’s tumor endostatin than systemic protein therapy. If endostatin in the burden. Surrogate markers such as circulating progenitor circulation follows a U-shaped curve of efficacy as is the endothelial cells are being studied [63]. Another possible case for interferon alpha [79], then very high concentrations surrogate marker of angiogenic activity is intravenous in- of the protein in the circulation may be less antiangiogenic jection of technetium-labeled endostatin which localizes to than lower doses. Recent successful studies of systemic the vascular bed of a tumor [64]. endostatin gene therapy in mice [67,74–76] in which levels of circulating endostatin were not as high as in the previous failed reports [77,78], support this speculation. 5. Horizontal transfer of oncogenes by uptake of apoptotic bodies 7. Summary It is clear that tumor cells are genetically unstable. This genetic instability may accelerate tumor progression by Proliferation of microvascular endothelial cells appears promoting mutations that confer a growth advantage [65]. to be essential for growth of tumors or of non-neoplastic A striking example of tumor progression is increased an- tissues [17]. Inhibition of endothelial cell growth can main- giogenic activity of a patient’s tumor over time [23].For tain neoplastic or normal tissue at a stable size. Apoptosis example, in women with newly diagnosed breast cancer, of microvascular endothelial cells can lead to regression 60% of the tumors were producing only one pro-angiogenic of tumor tissue and involution of normal tissue. Therefore, protein, VEGF. But, over subsequent years, recurrent tu- to understand growth control of normal and neoplastic tis- mors or metastases were found to be producing up to six sue, it is essential to continue to develop knowledge of different pro-angiogenic proteins [23]. We have recently the molecular mechanisms which regulate endothelial cell demonstrated that DNA is transferred horizontally by tu- proliferation and apoptosis. mor cell phagocytosis of apoptotic bodies from neighboring tumor cells [65]. The transferred DNA was propagated pro- vided that it conferred a selective advantage to the cell, and References that the phagocytic host cell was p53 negative. Therefore, we have proposed that uptake of DNA via apoptotic bodies [1] Folkman J. Tumor angiogenesis: therapeutic implications. New Engl J Med 1971;285:1182–6. may result in aneuploidy and provide one possible mecha- [2] Folkman J. Tumor angiogenesis. In: Holland JF, Frei III E, Bast nism by which genetic instability and genetic diversity are Jr RC, Kufe DW, Pollock RE, Weichselbaum RR, editors. Cancer generated within a tumor. medicine. 5th ed. Canada: B.C. Decker Inc.; 2000. p. 132–52. 166 J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167

[3] Folkman J. Angiogenesis. In: Braunwald E, Fauci AS, Kasper DL, [23] Relf M, LeJeune S, Scott PAE, Fox S, Smith K, Leek R, et al. Ex- Hauser SL, Longo DL, Jameson JL, editors. Harrison’s textbook pression of the angiogenic factors vascular endothelial growth fac- of internal medicine. 15th ed. New York: McGraw-Hill; 2001. tor, acidic and basic fibroblast growth factor, tumor growth factor-1, p. 517–30. platelet-derived endothelial cell growth factor, placenta growth fac- [4] Achilles E-G, Fernandez A, Allred EN, Kisker O, Udagawa T, tor, and pleiotrophin in human primary breast cancer and its relation Beecken W-D, et al. Heterogeneity of angiogenic activity in a human to angiogenesis. Cancer Res 1997;57:963–9. liposarcoma: a proposed mechanism for “no take” of human tumors [24] Kisker O, Onizuka S, Banyard J, Komiyama T, Becker CM, Achilles in mice. J Natl Cancer Inst 2001;93:1075–81. EG, et al. Generation of multiple angiogenesis inhibitors by human [5] Holmgren L, O’Reilly MS, Folkman J. Dormancy of micrometas- pancreatic cancer. Cancer Res 2001;61:7298–304. tases: balanced proliferation and apoptosis in the presence of angio- [25] Teicher BA, Dupuis N, Kusomoto T, Robinson MF, Liu F, Menon genesis suppression. Nat Med 1995;1:149–53. K, et al. Antiangiogenic agents can increase tumor oxygenation and [6] Gimbrone Jr MA, Leapman SB, Cotran RS, Folkman J. Tumor response to radiation therapy. Radiat Oncol Invest 1995;2:269–76. dormancy in vivo by prevention of neovascularization. J Exp Med [26] Jain RK. Delivery of novel therapeutic agents in tumors: physiolog- 1972;136:261–76. ical barriers and strategies. J Natl Cancer Inst 1989;81:570–6. [7] Udagawa T, Fernandez A, Achilles E-G, Folkman J, D’Amato RJ. [27] Jain RK. Normalizing tumor vasculature with anti-angiogenic ther- Persistence of microscopic human cancers in mice: alterations in apy: a new paradigm for combination therapy. Nat Med 2001;7: the angiogenic balance accompanies loss of dormancy. FASEB J 987–9. 2002;16(11):1361–70. [28] Herbst RS, Mullani NA, Davis DW, Hess KR, McConkey DJ, Charn- [8] Hanahan D. Heritable formation of pancreatic beta-cell tumours sangavej C, O’Reilly MS, Kim HW, Baker C, Roach J, Ellis LM, in transgenic mice expressing recombinant insulin/simian virus 40 Rashid A, Pluda J, Bucana C, Madden TL, Tran HT, Abbruzzese oncogenes. Nature 1985;315:115–22. JL. Development of biologic markers of response and assessment [9] Hanahan D, Folkman J. Patterns and emerging mechanisms of the of antiangiogenic activity in a clinical trial of human recombinant angiogenic switch during tumorigenesis. Cell 1996;86:353–64. endostatin. J Clin Oncol 2002;20(18):3804–14. [10] Kandel J, Bossy-Wetzel E, Radvanyi F, Klagsbrun M, Folkman J, [29] Libutti SK, Choyke P, Carrasquillo JA, Bacharach S, Neumann Hanahan D. Neovascularization is associated with a switch to the RD. Monitoring response to antiangiogenic agents using noninvasive export of bFGF in the multistep development of fibrosarcoma. Cell imaging tests. Cancer J Sci Am 1999;5:252–6. 1991;66:1095–104. [30] Kurdziel K, Bacharach S, Carrasquillo J, Huebsch S, Whatley M, [11] Arbeit JM, Olsen DC, Hanahan D. Upregulation of fibroblast growth Sellers D, et al. Using PET 18F-FDG, 11CO, and 150-water for factors and their receptors during multi-stage epidermal carcinogen- monitoring prostate cancer during a Phase II anti-angiogenic drug esis in K14-HPV16 transgenic mice. Oncogene 1996;13:1847–57. trial with thalidomide. Clin Positron Imaging 2000;3:144. [12] Rak J, Yu JL, Klement G, Kerbel RS. Oncogenes and angiogenesis: [31] Choyke PL, Knopp MV, Libutti SK. Special techniques for imaging signaling three-dimensional tumor growth. J Invest Dermatol Symp blood flow to tumors. Cancer J 2002;8:109–18. Proc 2000;5:24–33. [32] Tozer GM, Prise VE, Wilson J, Cemazar M, Shan S, Dewhirst [13] Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans K, Dew- MW, et al. Mechanisms associated with tumor vasculature shut-down erchin M, et al. Role of HIF-1alpha in hypoxia-mediated apoptosis, induced by combretastatin A-4 phosphate: intravital microscopy and cell proliferation and tumour angiogenesis. Nature 1998;394:485–90. measurement of vascular permeability. Cancer Res 2001;61:6413– Erratum in Nature 1998;395:525. 22. [14] Fukumura D, Xavier R, Sugiura T, Chen Y, Park EC, Lu N, et al. [33] Folkman J, Hahnfeldt P, Hlatky L. The logic of anti-angiogenic gene Tumor induction of VEGF promoter activity in stromal cells. Cell therapy. In: Friedman T, editor. The development of human gene 1998;94:715–25. therapy. New York: Cold Spring Harbor Laboratory Press; 1998. [15] Lyden D, Hattori K, Dias S, Costa C, Blaikie P, Butros L, et p. 527–43. al. Impaired recruitment of bone-marrow-derived endothelial and [34] Belotti D, Vergani V, Drudis T, Borsotti P, Pitelli MR, Viale G, hematopoietic precursor cells blocks tumor angiogenesis and growth. et al. The microtubule-affecting drug paclitaxel has antiangiogenic Nat Med 2001;7:1194–201. activity. Clin Cancer Res 1996;2:1843–9. [16] Holash J, Maisonpierre PC, Compton D, Boland P, Alexander CR, [35] Browder T, Butterfield CE, Kraling BM, Shi B, Marshall B, O’Reilly Zagzag D, et al. Vessel cooption, regression, and growth in tumors MS, et al. Antiangiogenic scheduling of chemotherapy improves mediated by angiopoietins and VEGF. Science 1999;284:1994–8. efficacy against experimental drug resistant cancer. Cancer Res [17] Rupnick MA, Panigrahy D, Zhang C-Y, Dallabrida SM, Lowell BB, 2000;60:1878–86. Langer R, et al. Adipose tissue mass can be regulated through the [36] Klement G, Baruchel S, Rak J, Man S, Clark K, Hicklin DJ, et al. vasculature. Proc Natl Acad Sci USA 2002;99:10730–5. Continuous low-dose therapy with vinblastine and VEGF receptor-2 [18] Hlatky L, Hahnfeldt P, Folkman J. clinical application of antiangio- antibody induces sustained tumour regression without overt toxicity. genic therapy: microvessel density, what it does and doesn’t tell us. J Clin Invest 2000;105:R15–24. J Natl Cancer Inst 2002;94:883–93. [37] Meta-analysis Group in Cancer. Efficacy of intravenous continuous [19] Streit M, Riccardi I, Velasco P, Brown LF, Hawighorst T, Bornstein infusion of fluorouracil compared with bolus administration in ad- P, et al. Thrombospondin-2: a potent endogenous inhibitor of tumor vanced colorectal cancer. J Clin Oncol 1998;16:301–8. growth and angiogenesis. Proc Natl Acad Sci USA 1999;96:14888– [38] Gabra H, Cameron DA, Lee LE, Mackay J, Leonard RC. Weekly 93. doxorubicin and continuous infusional 5-fluorouracil for advanced [20] Rak JW, St. Croix BD, Kerbel RS. Consequences of angiogenesis for breast cancer. Br J Cancer 1996;74:2008–12. tumor progression, metastasis and cancer therapy. Anticancer Drugs [39] Hansen RM, Ryan L, Anderson T, Krzywda B, Quebbeman E, 1995;6:3–18. Benson III A, et al. Phase III study of bolus versus infusion [21] Hamada J, Cavanaugh PG, Lotan O, Nicolson GL. Separable growth fluorouracil with or without cisplatin in advanced colorectal cancer. and migration factors for large-cell lymphoma cells secreted by mi- J Natl Cancer Inst 1996;88:668–74. crovascular endothelial cells derived from target organs for metasta- [40] Abu-Rustum NR. Salvage weekly paclitaxel in recurrent ovarian sis. Br J Cancer 1992;66:349–54. cancer. Semin Oncol 1997;24:62–7. [22] Nicosia RF, Tchao R, Leighton J. Angiogenesis-dependent tumor [41] Loffler TM, Freund W, Lipke J, Hausamen TU. Schedule-and dose spread in reinforced fibrin clot culture. Cancer Res 1983;43:2159– intensified paclitaxel as weekly 1-hour infusion in pretreated solid 66. tumors: results of a phase I/II trial. Semin Oncol 1996;23:32–4. J. Folkman/ Seminars in Cancer Biology 13 (2003) 159–167 167

[42] Kakolyris S, Samonis G, Koukourakis M, Vlachonicolis I, Chalki- implanted osmotic pump improves the efficacy and potency of adakis G, Kalbakis K, et al. Treatment of non-small-cell lung cancer therapy in a mouse xenograft tumor model. Cancer Res 2001;61: with prolonged oral etoposide. Am J Clin Oncol 1998;21:505–8. 7669–74. [43] Chamberlain MC. Recurrent supratentorial malignant gliomas in [63] Monestiroli S, Mancuso P, Burlini A, Pruneri G, Dell’Agnola C, children. Long-term salvage therapy with oral etoposide. Arch Gobb A, et al. Kinetics and viability of circulating endothelial cells 1997;54:554–8. as surrogate angiogenesis marker in an animal model of human [44] Neskovic-Konstantinovic ZB, Bosnjak SM, Radulovic SS, Mitrovic lymphoma. Cancer Res 2001;61:4341–4. LB. Daily oral etoposide in metastatic breast cancer. Anticancer [64] Yang DJ, Kim KD, Schechter NR, Yu DF, Wu P, Azhdarinia A, et Drugs 1996;7:L543–7. al. Assessment of antiangiogenic effect using 99mTc-Ec-endostatin. [45] Hanahan D, Bergers G, Bergsland E. Less is more, regularly: metro- Cancer Biother Radiopharm 2002;17:233–45. nomic dosing of cytotoxic drugs can target tumour angiogenesis in [65] Bergsmedh A, Szeles A, Henriksson M, Bratt A, Folkman J, Spetz mice. J Clin Invest 2000;105:1045–7. A-L, et al. Horizontal transfer of oncogenes by uptake of apoptotic [46] Kerbel R, Folkman J. Clinical translation of angiogenesis inhibitors. bodies. Proc Natl Acad Sci USA 2001;98(11):6407–11. Nat Rev Cancer 2002;2(10):727–39. [66] Pulkkanen KJ, Laukkanen JM, Fuxe J, Kettunen MI, Rehn M, Kan- [47] Rak J, Yu JL, Kerbel RS, Coomber BL. What do oncogenic mutations nasto JM, Parkkinen JJ, Kauppinen RA, Pettersson RF, Yla-Herttuala have to do with angiogenesis/vascular dependence of tumors? Cancer S. The combination of HSV-tk and endostatin gene therapy eradi- Res 2002;62:1931–4. cates orthotopic human renal cell carcinomas in nude mice. Cancer [48] Fernandez A, Udagawa T, Schwesinger C, Beecken W-D, Achilles Gene Ther 2002;9(11):908–16. EG, McDonnell TJ, et al. Angiogenic potential of prostate carcinoma [67] Blezinger P, Wang J, Gondo M, Quezada A, Mehrens D, French M, cells overexpressing bcl-2. J Natl Cancer Inst 2001;93:33–8. et al. Systemic inhibition of tumour growth and tumour metastases [49] Debinski W, Slagle-Webb B, Achen MG, Stacker SA, Tulchinsky by intramuscular administration of the endostatin gene. Nat Biotech E, Gillespie GY, Gibo DM. VEGF-D is an X-linked/AP-1 regulated 1999;17:343–8. putative onco-angiogen in human glioblastoma multiforme. Mol Med [68] Ding I, Sun JZ, Fenton B, Liu WM, Kimsely P, Okunieff P, et al. 2001;7(9):598–608. Intratumoural administration of endostatin plasmid inhibits vascular [50] Arbiser JL, Moses MA, Fernandez CA, Ghiso N, Cao Y, Klauber N, growth and perfusion in MCa-4 murine mammary carcinoma. Cancer et al. Oncogenic H-ras stimulates tumor angiogenesis by two distinct Res 2001;61(2):526–31. pathways. Proc Natl Acad Sci USA 1997;94:861–6. [69] Read TA, Sorensen DR, Mahesparan R, Enger PO, Timpl R, Olsen [51] Chin L, Tam A, Pomerantz J, Wong M, Holash J, Bardeesy N, et BR, et al. Local endostatin treatment of gliomas administered by al. Essential role for oncogenic Ras in tumour maintenance. Nature microencapsulated producer cells. Nat Biotech 2001;19:29–34. 1999;400:468–72. [70] Joki T, Machluf M, Atala A, Zhu J, Seyfried N, Dunn IF, et al. [52] Adams JM, Cory S. The Bcl-2 protein family: arbiters of cell survival. Continuous release of endostatin from microencapsulated engineered Science 1998;281:1322–6. cells for tumour therapy. Nat Biotech 2001;19:35–9. [53] Viloria-Petit AM, Rak J, Hung MC, Rockwell P, Goldstein N, Kerbel [71] Kuo CJ, Farnebo F, Yu EY, Christofferson R, Swearingen RA, Carter RS. Neutralizing antibodies against EGF and ErbB-2/neu receptor R, et al. Comparative evaluation of the antitumor activity of antian- tyrosine kinases down-regulate VEGF production by tumour cells giogenic proteins delivered by gene transfer. Proc Natl Acad Sci in vitro and in vivo: angiogenic implications for signal transduction USA 2001;98(8):4605–10. therapy of solid tumours. Am J Pathol 1997;151:1523–30. [72] Feldman A, et al. Effect of retroviral endostatin gene transfer on [54] Viloria-Petit A, Crombet T, Jothy S, Hicklin D, Schlaeppi JM, Rak subcutaneous and intraperitoneal growth of murine tumors. J Natl J, et al. Acquired resistance to the antitumour effect of epidermal Cancer Inst 2001;93:1014–20. growth factor receptor-blocking antibodies in vivo: a role for altered [73] Scappaticci FA, Smith R, Pathak A, Schloss D, Lum B, Cao Y, tumour angiogenesis. Cancer Res 2001;61:5090–101. et al. Combination angiostatin and endostatin gene transfer induces [55] Furuwatari C, et al. A comprehensive system to explore p53 muta- synergistic activity in vitro and antitumour efficacy in leukemia and tions. Am J Clin Pathol 1998;110:368–73. solid tumours in mice. Mol Ther 2001;3:186–96. [56] Yu JL, Rak JW, Coomber BL, Hicklin DJ, Kerbel RS. Effect of [74] Shi W, Teschendorf C, Muzyczka N, Siemann DW. Adeno-associated p53 status on tumor response to antiangiogenic therapy. Science virus-mediated gene transfer of endostatin inhibits angiogenesis and 2002;295:1526–8. tumor growth in vivo. Cancer Gene Ther 2002;9:513–21. [57] Dameron KM, Volpert OV, Tainsky MA, Bouck N. Control of an- [75] Calvo A, Feldman A, Libutti SK, Green JE. Adenovirus-mediated giogenesis in fibroblasts by p53 regulation of thrombospondin-1. endostatin delivery results in inhibition of mammary gland tumor Science 1994;265:1582–4. growth in C3 (1)/SV40 T-antigen transgenic mice. Cancer Res [58] Ravi R, et al. Regulation of tumor angiogenesis by p53-induced 2002;62:3934–8. degradation of hypoxia-inducible factor 1a. Genes Dev 2000;14: [76] Indraccolo S, Gola E, Rosato A, et al. Differential effects of an- 34–44. giostatin, endostatin and interferon-alpha1 gene transfer on in vivo [59] Zhang L, et al. Wild-type p53 suppresses angiogenesis in human growth of human breast cancer cells. Gene Ther 2002;9:867–87. leiomyosarcoma and synovial sarcoma by transcriptional suppres- [77] Pawliuk R, Bachelot R, Zurikiya O, Eriksson A, Cao Y, Leboulch sion of vascular endothelial growth factor expression. Cancer Res P. Continuous intravascular secretion of endostatin in mice from 2000;60:3655–61. transduced hematopoietic stem cells. Mol Ther 2002;5:345–51. [60] Sherif ZA, Nakai S, Pirollo KF, Rait A, Chang EH. Down-modulation [78] Eisterer W, et al. Unfulfilled promise of endostatin in a gene of bFGF-binding protein expression following restoration of p53 therapy-xenotransplant model of human acute lymphocytic leukemia. function. Cancer Gene Ther 2001;8:771–82. Mol Ther 2002;5:352–9. [61] Browder T, Folkman J, Hahnfeldt P, Heymach J, Hlatky L, Kieran [79] Slaton JW, Perrotte P, Inoue K, Dinney CPN, Fidler IJ. Interferon M, et al. Antiangiogenic therapy and p53. Science 2002;297:471a. alpha mediated down-regulation of angiogenesis-related genes and [62] Kisker O, Becker CM, Prox D, Fannon M, D’Amato R, Flynn E, therapy of bladder cancer are dependent on optimization of biological et al. Continuous administration of endostatin by intraperitoneally dose and schedule. Clin Cancer Res 1999;5:2726–34. 本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP 图书馆。图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具