Review

Correspondence *Department of Radiation Oncology, Harvard Medical School and Massachusetts General Hospital, 100 Blossom St, Cox-7, Boston, MA 02114, USA

Email jain@steele.mgh.harvard.edu

Introduction

The dependence of tumor growth and metastasis on blood vessels makes tumor angiogenesis a rational target for therapy.^{1, 2, 3, 4, 5, 6, 7} Strategies have been pursued to inhibit neovascularization and/or destroy existing tumor vessels, including direct targeting of endothelial cells, and indirect targeting by inhibiting the release of proangiogenic growth factors by cancer or stromal cells. Unlike preclinical studies in mice, an overall survival benefit with anti-vascular endothelial growth factor (VEGF)-specific agents (e.g. monoclonal antibodies) used as monotherapy has yet to be demonstrated in phase III trials. The addition of a VEGF-specific antibody, bevacizumab (Avastin®; Genentech, Inc., South San Francisco, CA) to current cytotoxic regimens, however, led to improved overall survival (OS) in previously untreated colorectal and lung patients and in previously treated colorectal cancer patients, as well as improved progression-free survival (PFS) in previously untreated breast cancer patients.^{8, 9} By contrast, adding bevacizumab to cytotoxic therapy did not enhance survival in previously treated metastatic breast cancer patients.¹⁰ Moreover, replacing bevacizumab with VEGF-receptor-selective, multitargeted agents such as vatalanib (Novartis, Basel, Switzerland, a receptor tyrosine kinase inhibitor [TKI]) in the combined regimen did not show similar efficacy in chemotherapy-naive or previously treated colorectal cancer patients.¹¹ Nevertheless, monotherapy using multitargeted agents that have a broader spectrum of inhibitory effect on VEGF receptors and a number of growth-factor pathways in cancer cells (sorafenib, Nexavar®, Bayer AG, Leverkusen, Germany and Onyx Pharmaceuticals, Emeryville, CA, and sunitinib, Pfizer, New York, NY) has resulted in significantly prolonged PFS in patients with renal-cell cancer and gastrointestinal stromal tumors (GIST).^{12, 13} Of interest, several agents that target oncogenic signaling pathways (such as the EGFR/HER2-specific antibodies cetuximab, Erbitux®, ImClone, New York, NY or trastuzumab, Herceptin®, Genentech) and may indirectly inhibit angiogenesis, have yielded increased OS with chemotherapy in clinical trials and are approved for human use in Europe, the US and elsewhere.

These contrasting results raise important questions about the use of anti-VEGF agents in clinical practice. Will anti-VEGF monotherapy increase survival in randomized phase III trials? How can tumor-vessel regression by combined anti-VEGF treatment—instead of compromising the delivery and efficacy of cytotoxic treatment—prolong OS in previously treated colorectal cancer, chemotherapy-naive colorectal cancer and lung cancer patients, as well as PFS in breast cancer patients? Why do anti-VEGF agents not prolong survival in certain previously treated cancer patients? Why are some broad-spectrum multitargeted agents, which block signaling from receptors present on both endothelial and cancer cells, effective as monotherapy? What biomarkers of treatment efficacy can be used to guide the optimal use of anti-VEGF agents in cancer patients? In this review, we will summarize the results of recent phase III clinical trials of anti-VEGF agents, address the questions raised above, and suggest new avenues for further investigation (Boxes 1 and 2).

Box 2 Potential mechanisms of tumor escape after anti-VEGF therapy.

 

Beyond achieving proof-of-principle confirmation in clinical trials, there is a need to optimize antiangiogenic therapy to delay tumor relapse beyond a few months. The mechanisms of tumor escape from the anti-VEGF agent might be different—yet not mutually exclusive—during therapy. As tumors progress, they secrete an increasing number of proangiogenic factors and become hypervascular, but the vessels are leaky and the blood flow is spatially and temporally heterogeneous. To achieve an antivascular effect, an antiangiogenic agent should target critical angiogenic pathways in a given tumor; otherwise, tumors might escape this effect soon after the onset of therapy. When efficacious, the effect of antiangiogenic therapy can be threefold: inhibition of new vessel formation and pruning of immature tumor vessels, transient normalization of the remaining vasculature and reduction in the number of blood-circulating endothelial cells and progenitor cells. If the cytotoxic effect of anti-VEGF monotherapy is insufficient for killing of all cancer cells, or the anti-VEGF agent is combined with a cytotoxic agent to which the tumors are refractory, the tumors relapse. Thus, efficacious cytotoxic agents should be added to effective anti-VEGF agent(s). Even if a synergistic effect is achieved by combining the anti-VEGF agent and the cytotoxic agent(s), relapse can occur after combination therapy by at least three mechanisms. First, long-term or high-dose anti-VEGF therapy might lead to a vasculature that is inefficient for drug delivery (in addition to increasing the rate and grade of its side effects). Optimizing the dose and schedule of the anti-VEGF agent might prevent these outcomes. Second, tumors may relapse because they use alternative pathways for neovascularization or as a result of acquired genomic instability by the endothelial cells. These pathways could be targeted with available or novel antiangiogenic agents. Third, cancer cell clones can acquire resistance to the chemotherapeutics used or repopulate between chemotherapy cycles. These clones should be specifically targeted with available or novel therapeutics.

 

Figure adapted with permission from Ref 41 © (2001) Nature Publishing Group

Clinical development of anti-VEGF therapy

Vascular endothelial growth factor A (VEGF-A, also called VEGF) is a potent proangiogenic growth factor expressed by most cancer-cell types and certain tumor stromal cells.^{3, 5, 14} VEGF stimulates endothelial-cell proliferation, migration and survival, expression of adhesion molecules, and potently induces increased vascular permeability.^{2, 3} VEGF can also affect new-vessel formation in tumors by acting as a chemoattractant for bone-marrow-derived progenitor/stem cells.¹⁵ VEGF expression can be triggered during early stages of neoplastic transformation by environmental stimuli (e.g. hypoxia or low pH) or by genetic mutations (e.g. in K-ras, p53, or HER2/ErbB2), and persists during progression.^{2, 16} Proteolytic enzymes produced by cancer cells or stromal cells can also release sequestered VEGF bound to the extracellular matrix.² Treatment itself (e.g. radiation or anti-VEGF therapy^{17, 18, 19}) may increase VEGF production or accumulation, and inhibiting its activity might be important in maintaining antitumor efficacy. Thus, blocking the action of VEGF appears to be a promising antiangiogenic approach to treating multiple types of solid tumors. Such inhibition can be achieved by direct or indirect targeting of the ligand (VEGF) at the mRNA or protein level, direct targeting of its receptors (VEGFR1, VEGFR2, and neuropilin 1 [NP1]; Table 1), or by blocking downstream signaling pathway components. Each of these strategies is currently being investigated in clinical trials (Figure 1).

Figure 1 Schematic representation of direct targeting of cancer, endothelial and perivascular cells by anti-VEGF agents

Combined and direct targeting of cancer cells and endothelial and perivascular cells has yielded increased survival in phase III trials of anti-VEGF agents. This has been achieved by two approaches. The first combines traditional cytotoxic agents (which may kill any proliferating cell) with the VEGF-specific antibody bevacizumab. VEGF blockade will inhibit its signaling pathways in endothelial cells responsible for cell survival, migration, proliferation and vascular permeability; VEGF blockade might also affect cancer cells, when their survival depends on VEGF (e.g. via NP1). The second approach uses low-molecular-weight tyrosine kinase inhibitors with broad inhibitory spectra (i.e. active against VEGFRs, EGFRs, PDGFRs, c-Kit receptors, and/or downstream soluble kinases such as Raf), which may be present in all these cell populations and on other tumor stromal cells (e.g. immune cells and fibroblasts). Alternatively, combinations of antibodies that block the ligands (e.g. VEGF and EGFR/HER2 or PDGF) might be effective in targeting both cancer and endothelial cells. The intracellular tyrosine kinase domains are shown in yellow and the extracellular ligand binding domains are depicted in small pink spheres.

CC, cancer cells; EC, endothelial cells; EGF, epidermal growth factor; EGFR, EGF receptor; HER2, human epidermal growth factor receptor 2; PC, perivascular cells; PDGF, platelet-derived growth factor; PDGFR, PDGF receptor; TKIs, tyrosine kinase inhibitors; TGF-, transforming growth factor-; VEGF, vascular endothelial growth factor; sVEGFR1, soluble VEGF receptor 1.

Full figure and legend (24K)Figures & Tables indexDownload Power Point slide (230K)

Table 1 Anti-VEGF agents currently in clinical development

Full tableFigures & Tables indexDownload Power Point slide (522K)

Direct treatments for antiangiogenesis include antibodies, soluble receptors, low-molecular-weight TKIs, antisense oligonucleotides, APTAMERS and RNA INTERFERENCE (Table 1). Monoclonal antibodies that block VEGFR1 (IMC-18F1, ImClone Systems, New York, NY) or its specific ligand placental growth factor (PlGF) (TB-403, BioInvent International, Lund, Sweden and ThromboGenics Ltd, Dublin, Ireland) are also under clinical development. VEGF-Trap (Regeneron, Tarrytown, NY), a composite decoy receptor based on VEGFR1 and VEGFR2, fused to an antibody Fc segment, potently blocks VEGF and PIGF action. RNA-interference-based approaches (e.g. ICS-283; Intradigm Corporation, Rockville, MD) that target VEGF are currently under preclinical development. Of interest, aptamer (pegaptanib sodium; Macugen®, Eyetech Pharmaceuticals, Inc., New York, NY), as well as antibody fragment (ranibizumab; Lucentis®, Genentech, South San Francisco, CA) that target VEGF, are FDA-approved for patients with age-related macular degeneration. Finally, because the VEGF pathway acts downstream of most oncogenes, treatment with agents that target these oncogenic pathways can indirectly inhibit VEGF (further information is provided in Supplementary Table 1; see website for details).

Anti-VEGF-specific monotherapy has been shown to have antivascular effects on tumor vessels,^{19, 20} but has yet to yield an OS benefit in phase III trials. In lung and colorectal carcinoma trials, bevacizumab monotherapy was discontinued because the primary endpoint (OS) was unlikely to be achieved based on the inferior responses compared with the chemotherapy and combination-regimen groups.^{8, 21}

VEGF blockade by bevacizumab has yielded improved OS or PFS in cancer patients in four phase III trials when combined with standard chemotherapy (Table 2). These results emphasize the potential benefit of targeting and killing both endothelial and neoplastic cells to enhance survival in multiple types of cancer. Impressive results have been achieved by combining anti-VEGF therapy with contemporary cytotoxic agents or by using broad-spectrum multitargeted agents that block VEGF and other growth-factor pathways in both cell types (Figure 1).

Table 2 Completed Phase III trials for the VEGF-specific antibody bevacizumab with standard contemporary chemotherapy

Full tableFigures & Tables indexDownload Power Point slide (452K)

Targeting of cancer and endothelial cells by anti-VEGF therapy and chemotherapy

Bevacizumab, a humanized VEGF-specific antibody with a reported half-life of 17–21 days,⁵ was given to cancer patients in combination with traditional cytotoxic regimens. Some phase I and II trials demonstrated objective responses (including a few complete responses²²) to this combined therapy.²³ The first randomized placebo-controlled phase III trial of bevacizumab, however, failed to show increased PFS or OS when bevacizumab was combined with chemotherapy in previously treated metastatic breast cancer patients.¹⁰ In this trial, despite an increased response rate, no survival benefit was seen in patients receiving bevacizumab with capecitabine versus patients receiving capecitabine alone.

The clinical breakthrough for antiangiogenic therapy came from a randomized phase III trial showing a 4.7-month increase in OS (the primary endpoint) when bevacizumab was used with chemotherapy (irinotecan/5-fluorouracil/leucovorin) in previously untreated, metastatic colorectal cancer patients (Table 2).⁹ Based on these data, bevacizumab became the first anti-VEGF agent to be approved by the FDA for cancer patients. Three other unpublished randomized phase III trials have shown positive results. One trial investigated the efficacy of bevacizumab with standard chemotherapy (paclitaxel) in patients with chemotherapy-naive recurrent or metastatic breast cancer and achieved the primary endpoint of increased PFS.⁸ In another trial, previously treated patients with advanced colorectal cancer who received bevacizumab in combination with proven second-line therapy (an oxaliplatin–5-fluorouracil–leucovorin regimen, FOLFOX 4) had a 2.1-month increase in OS—the primary endpoint—compared with patients who received FOLFOX 4 alone (Table 3). Whether combining bevacizumab with FOLFOX 4 or other chemotherapy regimens will be the best option for first-line therapy for colorectal cancer is under investigation (see reference 24 for review and discussion). Finally, a trial of patients with previously untreated advanced nonsquamous and non-small-cell lung cancer showed a 2.3-month increase in median survival (the primary endpoint) when bevacizumab was added to standard chemotherapy (paclitaxel and carboplatin).²¹ This was the first randomized trial of this agent in combination with chemotherapy that showed a median survival of greater than 1 year in one of the arms, and the first to use a targeted agent and demonstrate a survival advantage in combination with chemotherapy in patients with untreated metastatic lung cancer. The utility and feasibility of using bevacizumab in previously treated lung cancer patients is unknown. Even in renal-cell carcinoma, which is a highly VEGF-dependent malignancy, increase in OS by bevacizumab monotherapy has yet to be demonstrated in a phase III study. A phase III trial of bevacizumab and interferon- versus interferon- alone for renal-cell cancer has been completed, but results are not yet available. Collectively, these phase III trials show that anti-VEGF therapy can increase OS and/or PFS in colorectal, breast and lung cancer patients when combined with cytotoxic agents.

Table 3 Completed Phase III trials for multi-targeted agents with VEGF receptor 2 kinase as one of their targets

Full tableFigures & Tables indexDownload Power Point slide (332K)

Targeting of cancer and endothelial cells by multitargeted anti-VEGF agents with or without chemotherapy

The second approach is to target both cancer cells and endothelial cells with small molecules that inhibit signaling pathways for VEGF and other factors by blocking tyrosine kinase activity with or without chemotherapy. Similar to bevacizumab, VEGFR receptor kinase-selective multitargeted agents have been used in combination with chemotherapeutic agents in phase III clinical trials (Table 3). A randomized phase III study of semaxinib (SU5416; Pharmacia, San Francisco, CA), which primarily targets VEGFR1, VEGFR2 and VEGFR3, and secondarily targets PDGFR- and c-Kit, with 5-fluorouracil/leucovorin and 5-fluorouracil/leucovorin/irinotecan in patients with metastatic colorectal carcinoma, failed to show, at interim analysis, any survival benefit for the SU5416-containing regimens, resulting in the cessation of further development of this compound.²⁵ Vatalanib (PTK787/ZK 222584; Novartis, Basel, Switzerland) also primarily targets VEGFR1, VEGFR2, and VEGFR3, and secondarily targets PDGFR- and c-Kit, and can decrease blood flow at doses greater than 750 mg/ day.²⁶ One trial (CONFIRM 1) compared the efficacy of oral vatalanib in combination with FOLFOX 4 versus FOLFOX 4 alone for first-line treatment of metastatic colorectal cancer. The primary endpoint of this trial, improvement in PFS as judged by an independent central review, was not met. A secondary endpoint of the trial, OS, is perhaps just as clinically relevant, and these data may be available in early 2006.²⁷ Interim results from a phase III trial of vatalanib in combination with FOLFOX 4 chemotherapy as second-line treatment for metastatic colorectal cancer (CONFIRM 2), however, suggested no significant benefit in OS. Thus, multitargeted TKIs are yet to show survival benefit in phase III trials when combined with chemotherapeutics. Questions about the efficacy of these classes of agents and their scheduling can only be answered with additional pharmacokinetic and correlative clinical studies.

Conversely, monotherapy with other multitargeted broad-spectrum TKIs has shown efficacy in two randomized phase III trials for tumors with limited treatment options. Sorafenib (BAY 43-9006; Nexavar®, Bayer Aktiengesellschaft, Leverkusen-Bayerwerk, Germany, and Onyx Pharmaceuticals Inc., Emeryville, CA) targets VEGFR2 and VEGFR3, PDGFR-, Raf, c-Kit and FLT3 (fms-related tyrosine kinase 3), and efficiently inhibits both tumor-cell proliferation and angiogenesis in preclinical models.²⁸ Interim data from a randomized, placebo-controlled phase III trial (with OS as the primary endpoint) showed that renal-cell-carcinoma patients taking 400 mg sorafenib twice daily (sorafenib's half-life is about 20–27 h) had a significant improvement in PFS,^{12, 13} despite a marginal (2%) partial-response rate. Thus, patients on the placebo arm were allowed to cross over to sorafenib after unblinding. A new drug application has been filed with the FDA for sorafenib for use in patients with advanced renal-cell carcinoma. Another recent success was obtained with sunitinib (SU11248; Pfizer, New York, NY) in patients with GIST. In addition to targeting VEGFR2, sunitinib targets c-Kit, PDGFR-, and FLT3. The activated form of the c-Kit receptor is often expressed in GISTs, and is thus a good candidate for treatment with TKIs that inhibit c-Kit activity, such as the FDA-approved agent imatinib mesylate (Gleevec®; Novartis, Basel, Switzerland), which also targets PDGFR- and PDGFR- and BCR/ABL1. A randomized phase III study assessing sunitinib in the treatment of imatinib-resistant GIST successfully met its predetermined efficacy endpoint, time-to-progression (Table 4).²⁹ Sunitinib has a half-life of 40 h and was given daily at a dose of 50 mg in 4-week cycles with 2-week breaks to patients randomized 2:1 to patients receiving placebo. A planned interim analysis of the phase III study data led to the recommendation that the study be 'unblinded' to give all enrolled patients access to sunitinib. Sunitinib showed high response rates in refractory metastatic renal cancers in a large phase II trial.³⁰ An application for FDA approval of sunitinib has also been submitted. Both sorafenib and sunitinib are expected to increase OS in patients in these trials. These contrasting outcomes for multitargeted TKIs when used in monotherapy versus combined with chemotherapeutics call for further investigations on the importance of each target and their mechanisms of action.

Table 4 Surrogate markers under testing for the evaluation of the efficacy of anti-VEGF agents

Full tableFigures & Tables indexDownload Power Point slide (432K)

Safety of anti-VEGF agents

Most agents that target VEGF or its receptors have been well tolerated by patients in clinical trials, either as single agents or in combination with standard chemotherapy.⁶ The toxicity that occurs most consistently for anti-VEGF agents is hypertension, which may be secondary to the inhibition of nitric oxide release. Hypertension is usually manageable with medical treatment and ongoing therapy (Tables 2 and 3).^{26, 34, 35, 36, 37, 38} Fatigue is a more-common side effect across the multitargeted TKIs than for monoclonal antibodies. Other toxicities occur more commonly with specific agents. For example, serious but rare toxicities of bevacizumab include thromboembolic events (primarily arterial), gastrointestinal perforations (seen with greater frequency in colorectal and ovarian cancer patients), impaired wound healing (including surgical), and life-threatening or fatal hemorrhage.³⁵ Sorafenib is associated with fatigue, anorexia, diarrhea, rash and hand–foot syndrome.^{34, 36} Vatalanib has been associated with a higher incidence of fatigue, light-headedness/dizziness, nausea and uncommon thrombotic events.^{26, 38}Fatigue, nausea, diarrhea, mucositis and hand–foot syndrome have been seen with sunitinib.³⁷ Bevacizumab and multitargeted TKIs may also potentiate some of the toxicities associated with the chemotherapeutic agents with which they are given, such as neuropathy, asthenia, and bone marrow suppression. Such toxicities tend to be mild and usually do not entail significant dose modifications of the chemotherapy regimen. One important, albeit relatively uncommon, toxicity potentiated by these agents, however, is the increased risk of congestive heart failure; as noted with the use of bevacizumab (increase from 0.5% to 2.2%, mainly in patients who have received prior or concomitant anthracyclines, or prior left chest wall radiation).¹⁰ As more patients receive these agents for prolonged periods, more unexpected toxicities might occur.⁶

Mechanisms of action of combination therapies that include anti-VEGF targeting

Effective anti-VEGF agents should block the formation of new tumor vessels and prune the existing ones. Despite the successes of some direct anti-VEGF agents (bevacizumab, sorafenib, sunitinib) and indirect inhibitors of angiogenesis (such as trastuzumab), their mechanism of action in patients is far from fully understood. For example, a recent phase I trial in rectal cancer patients has demonstrated that a single infusion of bevacizumab alone can prune tumor vessels and reduce tumor microvascular density by 40–50%.^{19, 20} However, to date, no phase III trial of bevacizumab has shown survival benefit in monotherapy arms over chemotherapy. On the other hand, one would expect that vessel pruning should interfere with the delivery of chemotherapeutic agents given to these patients, and thus fail to provide any survival benefit. Indeed, the very first phase III trial for bevacizumab in breast cancer patients failed to show a benefit. The failure of bevacizumab to increase survival in heavily treated breast cancer patients was initially explained by first, the highly refractory and advanced nature of the tumors in the patients enrolled, and second, the increased expression of other angiogenic factors during breast cancer progression caused by chemotherapy, which rendered VEGF less critical for continued tumor growth.¹⁰ These hypotheses were partially supported by the success of a subsequent trial of bevacizumab combined with a different chemotherapeutic agent in treatment-naive advanced breast cancer patients.⁸ But the hypotheses seem to be contradicted by the efficacy of bevacizumab with chemotherapy in heavily treated colorectal cancer patients. Vatalanib does not seem to confer the same survival advantage as bevacizumab in colorectal cancer patients when combined with chemotherapy.⁸ These contrasting results raise outstanding questions about the mechanisms of action of these agents alone and in combination. Here, we will discuss only three of the many mechanisms of action postulated for anti-VEGF therapy⁶ and the combination with chemotherapy, which can potentially reconcile the differing outcomes in clinical trials.

Normalization of tumor vasculature and microenvironment

More than a decade ago, Teicher³⁹ proposed that combining antiangiogenic therapy with cytotoxic treatments would have synergistic effects because it allows targeting of both the malignant cell compartment and the vascular stroma (Figures 1 and 2). Bevacizumab efficiently reduces the number of tumor vessels;^{19, 20} however, the reduction in the number of tumor vessels achieved by anti-VEGF therapy might antagonize chemotherapy and radiotherapy by impeding the delivery of therapeutics and oxygen (a radio-sensitizer), respectively. Indeed, a number of preclinical studies have demonstrated such antagonism.^{7, 40}

Figure 2 Potential mechanisms of action of bevacizumab on tumor vasculature

Owing to high levels of proangiogenic molecules produced locally, such as VEGF, tumors make the transition from in situ carcinoma to frank carcinoma (1). At this stage, tumors become hypervascular, but the vessels are leaky and the blood flow is spatially and temporally heterogeneous. This leads to increased interstitial fluid pressure (IFP) and focal hypoxia, creating barriers to delivery and efficacy of therapeutics. The proposed mechanism of action of the VEGF-specific antibody bevacizumab is twofold: inhibition of new-vessel formation and killing of immature tumor vessels (2); and transient normalization of the remaining vasculature by decrease in macromolecular permeability (and thus the IFP) and hypoxia, and improvement of blood perfusion (3). Another effect of bevacizumab may be the direct killing of cancer cells in subsets of tumors in which the cells express VEGF receptors. Regardless of the mechanisms involved, monotherapy with bevacizumab is not curative because it cannot kill all cancer cells, and in the longer term leads to a vasculature that is inefficient for drug delivery (4), and to tumor relapse using alternative pathways for neovascularization. Combinations of bevacizumab with chemotherapeutics have therefore been pursued in phase III trials and have led to a survival benefit in patients with chemosensitive tumors, showing the synergistic effect of the two treatment modalities. Synergy may have been achieved as a result of increased cell killing following tumor vascular normalization: the lowered IFP leads to improved delivery of chemotherapeutics and molecularly targeted agents; the improved oxygenation sensitizes cancer cells to cytotoxic therapeutics and reduces the selection of more-malignant phenotype; and, finally, increased cellular proliferation around normalized vessels might increase the cytotoxicity of chemotherapeutics. Normalization of the vasculature might also benefit the direct killing of cancer cells by bevacizumab, in synergy with the chemotherapeutics. Of interest, cytotoxic therapeutics may kill proliferating endothelial cells, and thus may also normalize the tumor vasculature and improve drug delivery to tumors.⁷ Figure adapted with permission fom Ref 41 © (2001) Nature Publishing Group.

Full figure and legend (23K)Figures & Tables indexDownload Power Point slide (228K)

At the same time, such combinations of cytotoxic and antiangiogenic therapies have been successful in a number of preclinical and clinical studies. To resolve this paradox, in 2001 we proposed that anti-VEGF therapy can 'normalize' tumor vasculature.⁴¹ Such normalization can be defined as the structural and functional changes occurring in the context of a significant reduction in the number of tumor vessels that allow, at least transiently,⁴² an increased and/or more-uniform delivery of drugs and oxygen, ultimately leading to improved outcome.^{7, 41} While improved oxygenation can increase the efficacy of radiation therapy or of certain chemotherapeutic agents, it might also accelerate tumor growth. Indeed, we have shown that bevacizumab monotherapy can increase cancer-cell proliferation in some rectal cancer patients.¹⁹ These proliferating cells are likely to be more sensitive to chemotherapy. The increase in proliferation, however, occurs in the context of increased apoptosis of cancer cells. Hence, the effects of vessel normalization in some regions of the tumor seem to be overwhelmed by simultaneous vessel regression in other regions. In addition, the inability of tumors to recruit new vessels during antiangiogenic therapy limits the ability of this transient increase in vascular efficiency to expand the tumor mass. Of interest, we have seen no significant changes in tumor fluorodeoxy glucose (FDG)-uptake in rectal cancer patients receiving bevacizumab, despite a 40–50% reduction in vascular density.^{19, 20} This suggests that the residual 'normalized' vasculature is more efficient. Emerging preclinical results from our laboratory and others support the concept of vascular normalization.^{42, 43, 44, 45, 46, 47, 48, 49} More importantly, the clinical data from a phase I trial of bevacizumab in rectal cancer patients are also consistent with the vascular normalization hypothesis.^{19, 20} Our ongoing phase II study of bevacizumab and radiochemotherapy in rectal cancer patients will provide additional data on these vascular and cellular parameters.

Vascular normalization can potentially explain why bevacizumab is efficacious in combination with chemotherapy, despite the limited efficacy of bevacizumab as monotherapy.^{7, 40} The process suggests that bevacizumab increased survival rates in chemotherapy-naive metastatic breast cancer patients by improving the delivery of chemotherapeutics to chemoresponsive tumors, while such an increase was not seen in chemotherapy-refractory metastatic breast cancer patients, in whom improved delivery of chemotherapeutics might have less of an effect. Vascular stabilization during VEGF blockade potentially also decreased the shedding of metastatic cancer cells from the primary tumors. The alleviation of hypoxia by bevacizumab might make the tumors more chemosensitive and less metastatic. Finally, if during vascular normalization the improved tumor microenvironment led to increased proliferation of the surviving cancer cells, this might have rendered them more sensitive to cytotoxic agents (Figure 2).^{7, 19, 40}

Then why did VEGF blockade by a multitargeted-agent (vatalanib) treatment not show a clear benefit with FOLFOX 4 regimen in metastatic colorectal cancer patients? Besides the simple explanation that it is not as effective an agent at administered doses, vatalanib has a considerably shorter half-life (6 h) than bevacizumab (20 days), and the phase III trials for vatalanib used a single daily dose of the drug. Contradicting these data, however, is the fact that pharmacokinetic data suggest that an active dose of vatalanib is maintained in the blood circulation for 24 h, and that it has a rapid and pronounced anti-vascular effect.⁵⁰ Another explanation could be the off-target effects (i.e. other than on the VEGF receptor kinases). For example, vatalanib might target PDGFR- on perivascular cells. This action was shown in mice to be beneficial for vascular targeting, since the PDGF-B-PDGFR- axis is known to control vascular stabilization/maturation by recruitment of supporting perivascular cells. Blocking PDGFR-, however, may interfere with vascular normalization, by blocking perivascular cell recruitment and excessive vessel pruning, and thus prevent the synergistic effect of combined therapy.⁵¹ Thus, the clinical benefit of targeting perivascular cells in addition to endothelial cells with multitargeted TKIs in the context of chemotherapy remains unclear. How anti-VEGF therapy affects the recruitment or response of cells of the immune system in cancer patients is not known.

Given these facts, why did other multitargeted TKIs, with broader inhibitory spectra and longer half-lives, prove efficacious in other tumor types? Our hypothesis is that broad-spectrum multitargeted TKIs (i.e. those that simultaneously target multiple receptor or soluble kinases such as c-Kit, Raf, FLT3, PGDFR-, etc.) mimic the synergistic effect offered by vascular normalization for combinations of anti-VEGF antibody and chemotherapy more effectively than the combinations of multitargeted VEGF receptor kinases–selective TKIs used with chemotherapeutics. This concept is yet to be proven in the clinic, but is strongly supported by the PFS gain produced by sorafenib in renal-cell carcinoma and by sunitinib in imatinib-resistant GIST patients.²⁹

Collectively, these considerations imply that, if we are to optimally use single-targeted or multitargeted anti-VEGF agents, treatment schemes must be tailored for each agent. For example, it is not yet clear whether the addition to chemotherapy of existing multitargeted TKIs—which selectively target VEGF receptor kinases—to chemotherapy will impact outcome (e.g. vatalanib or semaxinib) to an extent comparable with the responses seen for the combination of bevacizumab—which specifically targets VEGF—with chemotherapy. In choosing a multitargeted agent, the ability to define its spectrum and to match it at the molecular level with the disease will be desirable.

Some agents that target cancer cells directly may indirectly block angiogenesis. In a preclinical model of HER2 overexpressing human breast cancer, trastuzumab (a HER2-specific antibody) decreased the expression of several angiogenic factors (including VEGF by cancer cells), while increasing the expression of the endogenous angiogenesis inhibitor thrombospondin 1, and induced changes (i.e. reduction in tumor vascular permeability, vessel diameter and vascular volume, but not vessel length) consistent with vascular normalization.⁵² In this tumor model, however, trastuzumab actually increased VEGF expression in stromal cells. Thus, efficacious targeting of endothelial and cancer cells might be achieved by regimens that combine tumor-cell targeting by anti-epidermal growth factor receptor (EGFR)-specific agents with direct anti-VEGF agents (such as bevacizumab). Alternatively, monotherapy regimens might consist of multitargeted TKIs that concomitantly target the EGFR/HER2 on cancer cells in addition to VEGF receptor kinases. Preliminary data from the combination of cetuximab and bevacizumab, either alone or in combination with the chemotherapeutic agent irinotecan, for patients with irinotecan refractory colorectal cancer, suggest that these combinations are feasible and have potentially promising response rates.⁵³ Additional trials combining trastuzumab, cetuximab or erlotinib with bevacizumab, as well as trials of Zactima® (ZD6474; AstraZeneca Pharmaceuticals, Cheshire, UK), which is a multitargeted TKI (selective for VEGFR, EFGR and RET), have reached phase II and/or III (in thyroid, breast, colorectal, lung, pancreatic and head-and-neck-cancer patients), and the results will have important implications for the therapy of HER2-positive or EGFR-positive cancers.⁵⁴

Potential biomarkers for anti-VEGF therapy

Currently, there are no proven biomarkers of efficacy of anti-VEGF therapy. These biomarkers are urgently needed to validate the mechanistic hypotheses, to identify responsive patients and optimal doses, to predict efficacy of regimens that include anti-VEGF agents, and to detect and prevent tumor escape (Box 2). These biomarkers would facilitate the optimization of the dose and schedule of bevacizumab for colorectal, breast and lung cancer, and extrapolation of the existing efficacy data to other tumor types. For example, the efficacy and toxicity of bevacizumab with chemotherapy were not always dose-dependent.⁶⁶ Unlike in preclinical studies, the survival benefit for colorectal cancer patients from the addition of bevacizumab to irinotecan/5-fluorouracil/leucovorin was independent of K-ras, BRAF, or p53 mutation status, p53, VEGF or thrombospondin 2 expression, or microvascular density evaluated prior to treatment.^{67, 68}

We have measured significant functional, structural, cellular and molecular changes in tumors in response to VEGF blockade, without a significant reduction of tumor volume.^{19, 20} How these changes during anti-VEGF treatment relate to PFS or OS, however, is not known. Thus, in the absence of an overt cytotoxic effect of the anti-VEGF-specific agents, other surrogate markers for efficacy must be identified. Significant advances have been made in identifying candidate markers; however, no such marker has been shown to be predictive in the clinical setting.⁶⁹ Some of the candidate markers include classic diagnostic or prognostic biomarkers, as well as newly developed, target-based and mechanism-based biomarkers (Table 4).

Biopsy of tumor tissue is a routine but highly invasive diagnostic and prognostic method that has great potential to identify valuable markers for therapeutic efficacy. Where serial biopsies can be obtained, immunostaining for the protein/cell of interest and/or gene profiling using laser capture microdissection of the cellular components of interest might yield invaluable information on the effect of a given anti-VEGF agent. For example, we have evaluated microvascular density, -smooth muscle actin and Ang2 expression, as well as tumor-cell apoptosis and proliferation during bevacizumab treatment in rectal cancer patients.^{19, 20} Interstitial fluid pressure and tissue oxygenation are parameters that reflect vascular function and, by extension, delivery of therapeutics.^{70, 71, 72, 73, 74, 75, 76, 77, 78} Changes in tumor interstitial fluid pressure^{19, 20} or tissue oxygen level during treatment might be valuable surrogate markers of the efficacy of anti-VEGF therapy, and can be measured in several tumor types and sampled from areas of accessible tumors.

Less-invasive methods include measuring changes in protein concentration (e.g. growth factors) in bodily fluids as surrogate markers for therapy. For example, blood plasma levels of VEGF were significantly increased by VEGFR2 blockade in mice and were proposed as a surrogate marker for VEGFR2 blockade.¹⁸ In rectal cancer patients, bevacizumab increased plasma levels of VEGF, but also of PlGF, in all patients analyzed.¹⁹ To date, it is unknown whether commercially available kits detect free VEGF from VEGF bound to bevacizumab. The level of VEGF stored in platelets⁷⁹ might also serve as a surrogate marker, calling for evaluation of VEGF both in the serum and in the plasma. We found that bevacizumab decreased the concentration of viable circulating endothelial cells (CECs) and progenitor cells in rectal cancer patients.¹⁹ We also found that the total number of CECs (which includes non-viable/apoptotic CECs) did not decrease, but rather increased, with bevacizumab treatment, consistent with recent preclinical data.⁸⁰ This result may reflect the shedding of non-viable tumor endothelial cells following antiangiogenic treatment (DG Duda, unpublished observations), and might become an independent surrogate marker for anti-VEGF treatment and vascular normalization if quantitative methods of detecting tumor-derived CECs can be optimized. In preclinical studies, progenitor cells have been extensively studied both as surrogate markers and for their role in cancer growth and treatment,^{15, 81, 82} but their concentration in whole blood in humans is very low (two orders of magnitude lower than viable CECs¹⁹). Elucidation of the biology of different subsets of progenitor cells in humans, and development of improved techniques to reliably quantify cells in this concentration range may allow the future use of circulating cells as surrogate markers in the clinic. In addition, this will help us to understand any putative role that circulating endothelial cell populations might play in tumor response or escape from anti-VEGF therapy.^{19, 20}

Noninvasive techniques have the potential to measure functional parameters and offer surrogate markers for therapy, regardless of tumor type or location.⁸³ Such techniques include dynamic MRI, CT, and positron emission tomography, and are being pursued in clinical trials by our own team and others.^{19, 20, 50, 84, 85, 86} One main limitation of these imaging methods is that they measure composite parameters, which depend on both blood flow and permeability. Another limitation relates to the high heterogeneity of blood flow. The spatial distribution of blood flow determines the distribution of chemotherapeutics and oxygen in tumor tissue, and cannot be evaluated with high spatial resolution by the imaging techniques that are currently available. With improvement of these techniques to measure the spatial and temporal changes in tumor blood flow and vascular permeability with higher resolution, it might be possible to assess more precisely the efficacy of anti-VEGF therapy. Nevertheless, the cost of such investigations might be prohibitive for larger trials. Finally, protein measurements in urine have become increasingly feasible,^{87, 88, 89} and exploring them during therapy might offer independent surrogate markers for the effect of anti-VEGF agents. Combining and comparing multiple parameters obtained using different assays in phase I–II clinical trials hold great promise for defining the effects of individual anti-VEGF agents and identifying simple and meaningful surrogate markers of efficacy.^{19, 20, 84, 90} The best candidates should eventually be validated and used in larger phase III trials and integrated into routine clinical practice.^{83, 91}

Conclusion

The recent successes of the antiangiogenic agents have raised great hope and have taught us important lessons about the significance of the target, timing and dosage of each agent (Box 1). Anti-VEGF agents are now expected to make a difference in cancer patients with a wide variety of tumor types. With the advent of specific and potent new agents—approved or in the process of being approved—oncologists have a variety of direct and indirect antiangiogenic agents to choose from when designing therapy protocols. Determining whether the regimens used in the successful trials are optimal, however, and whether anti-VEGF agents will work in patients outside the rigorous inclusion criteria used for those trials, will be critical for deciding the standard of care for different malignancies. Establishing the most advantageous combinations will require a better understanding of the mechanisms of action of each anti-VEGF agent and the sensitivity of each tumor type, as well as development of robust biomarkers and imaging techniques to guide patient selection and protocol design. A deeper understanding of the mechanisms of antitumor activity of specific and multitargeted anti-VEGF agents in patients, how they can best be combined with other treatment approaches such as chemotherapy and radiation therapy, and how optimization of these effects can be monitored clinically, should contribute to significantly improved cancer treatment and extend survival of cancer patients in the near future, as well as enhance the prospects of developing curative treatment for different cancers in the more-distant future.

Acknowledgments

The authors thank the members of the Steele Lab, especially Y Boucher, E di Tomaso, D Fukumura, K Kozak, S Kozin, L Munn, and T Padera. We also thank T Batchelor, H Chen, V Natarajan, W Novotny, L Rosen, A Ryan, J Samson, C Willett, and J Wood for their helpful comments on this manuscript. The authors' work is supported by grants from the National Cancer Institute.

References

1. Folkman J (1971) Tumor angiogenesis: therapeutic implications. N Engl J Med 285: 1182–1186 | PubMed | ISI | ChemPort |
2. Carmeliet P and Jain RK (2000) Angiogenesis in cancer and other diseases. Nature 407: 249–257 | Article | PubMed | ISI | ChemPort |
3. Dvorak HF (2002) Vascular permeability factor/vascular endothelial growth factor: a critical cytokine in tumor angiogenesis and a potential target for diagnosis and therapy. J Clin Oncol 20: 4368–4380 | Article | PubMed | ISI | ChemPort |
4. Kerbel R and Folkman J (2002) Clinical translation of angiogenesis inhibitors. Nat Rev Cancer 2: 727–739 | Article | PubMed | ISI | ChemPort |
5. Ferrara N et al. (2004) Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer. Nat Rev Drug Discov 3: 391–400 | Article | PubMed | ISI | ChemPort |
6. Carmeliet P (2005) Angiogenesis in life, disease and medicine. Nature 438: 932–936 | Article | PubMed | ISI | ChemPort |
7. Jain RK (2005) Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 307: 58–62 | Article | PubMed | ISI | ChemPort |
8. Bevacizumab (Avastin™) for Treatment of Solid Tumors: NCI Fact Sheet [http://www.cancer.gov/cancertopics/factsheet/AvastinFactSheet] (accessed in 24 November 2005)
9. Hurwitz H et al. (2004) Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 350: 2335–2342 | Article | PubMed | ISI | ChemPort |
10. Miller KD et al. (2005) Randomized phase III trial of capecitabine compared with bevacizumab plus capecitabine in patients with previously treated metastatic breast cancer. J Clin Oncol 23: 792–799 | Article | PubMed | ISI | ChemPort |
11. [No authors listed] News in Brief (2005) Nat Rev Drug Discov 4: 448–449 | Article | ChemPort |
12. Rini BI et al. (2005) Therapy targeted at vascular endothelial growth factor in metastatic renal cell carcinoma: biology, clinical results and future development. BJU Int 96: 286–290 | Article | PubMed | ISI | ChemPort |
13. Branca MA (2005) Multi-kinase inhibitors create buzz at ASCO. Nat Biotechnol 23: 639 | Article | PubMed | ISI | ChemPort |
14. Fukumura D et al. (1998) Tumor induction of VEGF promoter activity in stromal cells. Cell 94: 715–725 | Article | PubMed | ISI | ChemPort |
15. Rafii S et al. (2002) Vascular and haematopoietic stem cells: novel targets for anti-angiogenesis therapy? Nat Rev Cancer 2: 826–835 | Article | PubMed | ISI | ChemPort |
16. Ferrara N et al. (2003) The biology of VEGF and its receptors. Nat Med 9: 669–676 | Article | PubMed | ISI | ChemPort |
17. Gorski DH et al. (1999) Blockage of the vascular endothelial growth factor stress response increases the antitumor effects of ionizing radiation. Cancer Res 59: 3374–3378 | PubMed | ISI | ChemPort |
18. Bocci G et al. (2004) Increased plasma vascular endothelial growth factor (VEGF) as a surrogate marker for optimal therapeutic dosing of VEGF receptor-2 monoclonal antibodies. Cancer Res 64: 6616–6625 | Article | PubMed | ISI | ChemPort |
19. Willett CG et al. (2005) Surrogate markers for antiangiogenic therapy and dose-limiting toxicities for bevacizumab with radiation and chemotherapy: continued experience of a phase I trial in rectal cancer patients. J Clin Oncol 23: 8136–8139 | Article | PubMed | ISI |
20. Willett CG et al. (2004) Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med 10: 145–147 | Article | PubMed | ISI | ChemPort |
21. Sandler AB (2005) Clinical trials comparing carboplatin/paclitaxel with or without bevacizumab in patients with metastatic NSCLC. Lung Cancer Update 2: 6–10
22. Tuma RS (2005) Success of bevacizumab trials raises questions for future studies. J Natl Cancer Inst 97: 950–951 | PubMed |
23. Chen HX (2004) Expanding the clinical development of bevacizumab. Oncologist 9 (Suppl 1): S27–S35 | Article |
24. Rosen LS (2005) VEGF-targeted therapy: therapeutic potential and recent advances. Oncologist 10: 382–391 | Article | PubMed | ISI | ChemPort |
25. Eskens FA (2004) Angiogenesis inhibitors in clinical development; where are we now and where are we going? Br J Cancer 90: 1–7 | Article | PubMed | ISI | ChemPort |
26. Mross K et al. (2005) Phase I clinical and pharmacokinetic study of PTK/ZK, a multiple VEGF receptor inhibitor, in patients with liver metastases from solid tumours. Eur J Cancer 41: 1291–1299 | Article | PubMed | ISI | ChemPort |
27. Hecht JR et al. (2005) A randomized, double-blind, placebo-controlled, phase III study in patients with metastatic adenocarcinoma of the colon or rectum receiving first-line chemotherapy with oxaliplatin/5-fluorouracil/leucovorin and PTK787/ZK222584 or placebo [abstract]. J Clin Oncol 23: LBA3
28. Wilhelm SM et al. (2004) BAY 43-9006 exhibits broad spectrum oral antitumor activity and targets the RAF/MEK/ERK pathway and receptor tyrosine kinases involved in tumor progression and angiogenesis. Cancer Res 64: 7099–7109 | Article | PubMed | ISI | ChemPort |
29. Marx J (2005) Cancer. Encouraging results for second-generation antiangiogenesis drugs. Science 308: 1248–1249 | Article | PubMed | ISI | ChemPort |
30. Beaumont J et al. (2005) Patient reported outcomes in a phase 2 trial of SU11248 for metastatic renal cell carcinoma [abstract #8167]. Proc Am Soc Clin Oncol
31. Slamon DJ et al. (2001) Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 344: 783–792 | Article | PubMed | ISI | ChemPort |
32. Piccart-Gebhart MJ et al. (2005) Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 353: 1659–1672 | Article | PubMed | ISI | ChemPort |
33. Romond EH et al. (2005) Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med 353: 1673–1684 | Article | PubMed | ISI | ChemPort |
34. Clark JW et al. (2005) Safety and pharmacokinetics of the dual action raf kinase and vascular endothelial growth factor receptor inhibitor, BAY 43-9006, in patients with advanced, refractory solid tumors. Clin Cancer Res 11: 5472–5480 | Article | PubMed | ISI | ChemPort |
35. Midgley R and Kerr D (2005) Bevacizumab—current status and future directions. Ann Oncol 16: 999–1004 | Article | PubMed |