Special Perspectives

Antiangiogenic Therapy: A Universal Chemosensitization Strategy for Cancer?

See allHide authors and affiliations

Science  26 May 2006:
Vol. 312, Issue 5777, pp. 1171-1175
DOI: 10.1126/science.1125950


For more than 50 years, a major goal of research in cancer therapeutics has been to develop universally effective agents that render cancer cells more sensitive to cytotoxic chemotherapy without substantially increasing toxicity to normal cells. The results of recent clinical trials indicate that certain antiangiogenic drugs may produce this long-sought effect. Here, I describe three distinct mechanisms that may help to explain the chemosensitizing activity of these drugs: normalizing tumor vasculature, preventing rapid tumor cell repopulation, and augmenting the antivascular effects of chemotherapy. I then discuss how these potential mechanisms might be exploited to maximize therapeutic efficacy.

In 1971, Judah Folkman first articulated the concept behind what he called “antiangiogenic” drugs: Because progressive tumor growth is dependent on a blood supply, he proposed that treatment with drugs that prevent the formation of tumor blood vessels might be able to constrain cancer for prolonged periods (1). Over the next three decades, roughly 10,000 research papers on angiogenesis were published, culminating in a recent report of the first large-scale clinical success of an antiangiogenic drug for cancer treatment (2). In this phase III trial, patients with metastatic colorectal cancer who had been treated with a combination of conventional cytotoxic chemotherapy plus bevacizumab, a humanized monoclonal antibody directed against vascular endothelial growth factor (VEGF), showed prolonged survival compared with patients treated with chemotherapy alone (2). The combination of bevacizumab (Avastin) and chemotherapy is now approved in the United States and many other countries as a first-line treatment for colorectal cancer. Subsequent phase III trials with bevacizumab and chemotherapy in breast and non–small cell lung cancer have produced similarly promising results (Table 1). In addition, two small-molecule antiangiogenic drugs, SU11248/sunitinib (Sutent) and BAY-43-9006/sorafenib (Nexavar), have been approved as monotherapies for kidney cancer (3). Bevacizumab has also shown activity as a monotherapy in kidney cancer (4).

Table 1.

Results of recent clinical trials of antiangiogenic drugs in cancer. The table presents representative examples; it is not intended to be comprehensive. Bevacizumab is a humanized monoclonal antibody directed against VEGF. SU11248/sunitinib is an orally available small-molecule inhibitor of VEGF receptors 1, 2, and 3 and PDGF receptors alpha and beta c-kit and flt-3. BAY-43-9006/sorafenib is an orally available small-molecule inhibitor of VEGF receptors 2 and 3, PDGF receptor beta, and Raf kinase. The results of the following clinical trials were published (2, 4, 18) or presented at the American Society for Clinical Oncology (5053) and are cited in several reviews and commentaries (3, 7, 8, 15).

DrugTrial typeCancer typeTreatment regimenOutcome
Bevacizumab Phase II Metastatic renal cell cancer Monotherapy Benefit in progression-free survival but not overall survival with higher dose (10 mg/kg) (View inline)
Bevacizumab Phase III Colorectal cancer First-line treatment in combination with chemotherapy (5-FU/leucovorin plus irinotecan) Benefit in both progression-free survival and overall survival (4.7 months) with addition of bevacizumab to chemotherapy (View inline)
Bevacizumab Phase III Colorectal cancer Second-line treatment in combination with “FOLFOX4” chemotherapy (oxaliplatin, 5-FU/leucovorin) Benefit in progression-free survival and overall survival (2.5 months), compared with chemotherapy (View inline) [cited in (View inline, View inline)]
Bevacizumab Phase III Non—small cell lung cancer First-line treatment in combination with paclitaxel plus carboplatin chemotherapy Benefit in progression-free survival and in overall survival (2.3 months) with addition of bevacizumab (View inline) [cited in (View inline, View inline, View inline)]
Bevacizumab Phase III Metastatic breast cancer First-line treatment in combination with paclitaxel chemotherapy Benefit in progression-free survival (about 5 months) based on interim analysis with addition of bevacizumab; overall survival results not yet known (View inline) [cited in (View inline, View inline, View inline)]
SU11248/sunitinib Phase II Cytokine-refractory metastatic renal cell cancer Monotherapy Unusually high rate of objective tumor responses detected (40%); stable disease (27%) also seen; overall survival benefit not yet known (View inline) [cited in (View inline)]
BAY-43-9006/sorafenib Phase III Metastatic renal cell cancer Second-line monotherapy Significant improvement in progression-free survival despite very low (2%) rate of partial tumor responses (View inline) [cited in (View inline, View inline)]

These clinical trial results and some others—e.g., in ovarian and pancreatic cancer (5, 6)—underscore the expanding range of tumor types that respond to this class of drugs (Table 1). Although the survival benefits conferred are modest [generally between 2 and 5 months in the completed phase III bevacizumab trials (Table 1)], the results nonetheless represent one of the most notable advances in cancer research and treatment over the past 3 years (7). They raise the possibility that the addition of antiangiogenic drugs to chemotherapy agents might become standard practice in clinical oncology, thus making such drugs “universal” chemosensitizing agents.

This scenario raises a number of interesting questions. For example, although we know bevacizumab's precise molecular target (VEGF), do we really understand how this drug works as an antitumor agent? Why does it prolong survival of patients with advanced cancers primarily when it is combined with cytotoxic chemotherapy? This is especially puzzling because intuitively one would expect antiangiogenic agents to suppress the intratumoral delivery of coadministered drugs by reducing the number of tumor-associated blood vessels and/or by compromising their ability to perfuse blood. Moreover, these two effects would increase the degree and duration of tumor hypoxia, leading to suppression of tumor cell proliferation—yet, it is proliferating tumor cells that are most sensitive to chemotherapy.

This leads to more questions. Is combination with conventional, maximum tolerated dose (MTD) chemotherapy the optimal (or only) way to use chemotherapy with an antiangiogenic drug? Furthermore, why is it that small-molecule receptor tyrosine kinase inhibitors (RTKIs) with antiangiogenic activity have thus far not shown the same chemosensitizing prowess as antibody-based drugs in randomized clinical trials? For example, based on interim or final phase III clinical trial results, neither SU-5416 nor PTK787 (valatinib), which target the VEGF and/or platelet-derived growth factor (PDGF) receptors, enhanced the efficacy of chemotherapy regimens in metastatic colorectal cancer [cited in (7, 8)]. Perhaps, in some cases, the answer lies in the ability of certain antibody-based drugs directed to cell surface antigens to induce antibody-dependent cell-mediated cytotoxicity, a mechanism that makes target cells more vulnerable to attack by the immune system (9). Alternatively, the activity differences might be related to differences in suboptimal potency, dosing, and/or pharmacokinetics. The results of ongoing trials with drugs such as sunitinib or sorafenib with chemotherapy should answer this question.

Chemotherapy will remain a mainstay of cancer treatment for many years to come, and it will be increasingly used with antiangiogenic agents, as well as other targeted therapies. Thus, research will focus on the mechanisms by which these drugs assist one another. Understanding these mechanisms will almost certainly lead to improvements in the efficacy of the combination therapies and may increase their cost effectiveness, an issue of growing concern. Indeed, the costs of many targeted cancer drugs can be so high as to threaten their widespread usage, thus potentially negating the decades of inspiring research and discovery that contributed to their development (10).

There are other important reasons for focusing on the mechanisms by which these different classes of drugs interact. Conventional chemotherapy can be very toxic, and many antiangiogenic drugs have their own set of toxicities, such as hypertension, blood clotting, and proteinurea, and more rarely, gastrointestinal perforations of the bowel (4). Although most of these complications are not severe (bowel perforations and arterial clots being exceptions), there is limited evidence that some (e.g., blood clot formation) may be exacerbated by combining antiangiogenic drugs with certain regimens of chemotherapy (11). Acquired resistance to antiangiogenic drugs is also emerging as an issue (12, 13), and combinatorial drug strategies, some involving chemotherapy, are one potential approach for prolonging tumor responses to antiangiogenic drugs, just as antiangiogenic drugs may delay the onset of resistance to chemotherapy drugs. Finally, information about the mechanism of action of antiangiogenic drugs will be important for determining each one's optimal biologic dose, when used in combination treatment regimens.

Here, I describe three distinct mechanistic models that may help to explain the chemosensitizing activity of antiangiogenic drugs, and I discuss some possible clinical implications of each.

Model 1: Antiangiogenic Drugs “Normalize” the Tumor Vasculature, Enhancing the Efficacy of Chemotherapeutic Drugs

Much of the vasculature in tumors is disorganized, leaky, and structurally abnormal (14, 15). The structural eccentricities include absent, reduced, or altered basement membranes and periendothelial support cells such as pericytes; excessively dilated or constricted vessels; and corkscrew-like tortuosities. Tumor vessel leakiness can lead to extravasation of plasma proteins and fluid, producing high interstitial fluid pressures (IFPs) within tumors, which in turn may impede the delivery and diffusion of certain cancer drugs, especially large molecules or immune killer cells (14). In addition, the flow and perfusion of blood in many of the tumor vessels may be impaired, which, as noted above, can produce localized and transient areas of hypoxia. As a result, nearby tumor cells may enter a quiescent state that would reduce their inherent sensitivity to chemotherapy and radiation.

It has been postulated that treatment with antiangiogenic agents can transiently reverse some of these abnormalities, a phenomenon called “vessel normalization.” For example, it has been proposed that because VEGF is a potent inducer of vascular permeability (16), agents that block VEGF activity, such as bevacizumab, would suppress vessel leakiness. Such drugs would also prune immature growing blood vessels—generally considered to be the vessels most sensitive to antiangiogenic drugs—leaving behind an increased proportion of mature, functional vessels. Remaining immature vessels might mature rapidly as a result of an influx of pericytes, driven by the action of PDGF acting on PDGF beta receptors expressed by these cells (17). Angiopoietin and the Tie-2 endothelial cell receptor tyrosine kinase may also be involved in this vessel normalization effect (14).

Jain and colleagues have hypothesized that these changes generate transient drops in intratumoral IFPs and decrease the level of tumor hypoxia, after the initial expected increase in hypoxia (14). Improved delivery of cytotoxic drugs to the tumor during this “normalization window” would coincide with a transient burst in tumor cell proliferation, conceivably leading to increased tumor cell killing (15). This hypothesis may explain the failure of certain multitargeted RTKIs to enhance the efficacy of chemotherapy when combined with conventional cytotoxic chemotherapy (8). It has been speculated that because pericyte-associated PDGF receptors contribute to vessel normalization, blocking them with PDGF receptor targeting drugs might suppress vessel normalization and thus reduce the efficacy of cytotoxic chemotherapy, or at least have no effect. Thus, counter-intuitively, stimulating PDGF (or Tie-2) receptor activity may be an effective strategy in combination therapy situations involving chemotherapy or radiation, whereas inhibiting PDGF receptors would be beneficial when such drugs are used as monotherapies (14, 18), as in the case of SU11248 or BAY-43-9006 for kidney cancer (Table 1).

The results of limited preclinical studies on the kinetics of vessel normalization suggest that the administration of an antiangiogenic agent should precede rather than coincide with chemotherapy; otherwise the normalization window could be missed (17). However, many of the successful randomized phase II and III trials of bevacizumab have used concurrent drug administration protocols where one might expect to observe, at least in some cases, a reduction in the efficacy of chemotherapy. In addition, the effects of chemotherapy can be enhanced by an antiangiogenic drug even during sustained increases in hypoxia, with little evidence of vessel normalization (19). Indeed, another class of vascular targeting drugs called vascular disrupting agents, which cause rapid occlusion of existing blood vessels in tumors and subsequent massive hypoxia and tumor necrosis, can nevertheless augment the efficacy of chemotherapy or radiation, even though these treatments are less effective against hypoxic tumor cells (20). Thus, additional or alternative mechanisms of chemosensitization are presumably involved in such cases. Moreover, almost all the randomized trials of bevacizumab have involved patients with advanced metastatic disease, so many of them would have had multiple metastases of varying size in different organs. Vessel normalization induced by an antiangiogenic drug would have to be precisely synchronized in virtually every such lesion in order to be successfully exploited to bring about an overall chemosensitization benefit and prolong patient survival. If so, exploiting vessel normalization to improve chemotherapy may be most practical in situations involving single localized tumors such as glioblastomas, or when using localized radiation therapy.

Model 2: Antiangiogenic Drugs Prevent Rapid Tumor Cell Repopulation After Cytotoxic Chemotherapy

Treatment of most types of cancer with conventional cytotoxic drugs initially induces tumor shrinkage (the so-called “objective tumor response”), but this is almost always followed by tumor cell repopulation. Moreover, the rate of tumor cell growth tends to be faster the smaller the tumor (as a result of “Gompertzian” growth kinetics), so major shrinkage of tumor mass does not necessarily lead to commensurate prolongation of patient survival times (21, 22). In addition, the rate of tumor cell repopulation does not necessarily decline in proportion to the number of successive courses of cytotoxic chemotherapy or radiation; in fact, the observed trend suggests the opposite effect (23). Consequently, it has been argued that implementing therapeutic strategies that dampen the rate of tumor cell repopulation during the break periods between successive courses of MTD chemotherapy may prolong survival times (22, 23). (Such breaks—usually about 3 weeks—permit patient recovery from myelosuppression and other side effects of the chemotherapy.) One way to achieve this goal is to administer the chemotherapy more frequently (e.g., every 2 weeks), a strategy referred to as “dose-dense” chemotherapy (21), along with agents that accelerate recovery from myelosuppression (21, 23). Such dose-dense chemotherapy regimens have shown success in early-stage breast cancer as adjuvant therapy (24).

A second possible approach is to expose the tumor to an antiangiogenic drug during the break periods between courses of chemotherapy, as discussed by Hudis (22). In other words, one should “apply a brake during the break” (22, 23). In a sense, this is the way that bevacizumab is currently used. Unlike most chemotherapy drugs, antibody-based drugs such as bevacizumab can persist in the circulation for 2 to 3 weeks and thus would continue to circulate at relatively high levels during the 3-week break periods. The rationale is that many of the repopulating tumor cells will require oxygen and nutrients delivered by the adjacent tumor vasculature. In this scenario, just when the antibody's inhibitory effects on tumor cell repopulation might begin to wane, the next dose of chemotherapy is administered. As with vessel normalization, this mechanistic model underscores the importance of timing and sequence in achieving the maximal therapeutic benefit from these combination therapies.

Model 3: Antiangiogenic Drugs Augment the Antivascular Effects of Chemotherapy

Many anticancer drugs, including cytotoxic chemotherapy, have been hypothesized to have antiangiogenic activities that contribute to their antitumor efficacy (25, 26). There are at least two plausible explanations for the effect of cytotoxic drugs on the tumor vasculature. First, these drugs could damage or destroy the subset of endothelial cells that proliferate during the formation of new blood vessels (27). Research with preclinical models supports this hypothesis (25, 26). Because VEGF is a potent survival factor for activated endothelial cells (28) neutralization of VEGF function should amplify the ability of chemotherapy to damage such cells (2931) in the local tumor microenvironment.

A second, more systemic way that chemotherapy might exert effects on the vasculature is by impairing the mobilization, function, or viability of circulating bone marrow–derived cell populations that contribute to angiogenesis. These cells include circulating endothelial progenitor cells (EPCs) that can incorporate into the lumens of nascent vessels and differentiate into mature endothelial cells (32, 33), CD45+ Tie-2–expressing monocytes (34), as well as other CD45+ monocyte- and myeloid-like populations (35, 36) that are home to sites of ongoing angiogenesis, adhere to blood vessels and further stimulate angiogenesis (e.g., by secretion of VEGF). Given the well-established myelosuppressive effects of cytotoxic chemotherapy, one might predict that at least some of these proangiogenic bone marrow cell types would be sensitive to chemotherapy. This hypothesis is supported by preclinical evidence showing that chemotherapy can substantially reduce the level of circulating EPCs (37, 38). Moreover, because many of these cell populations can be mobilized into the peripheral blood by growth factors such as VEGF, the combination of a VEGF-targeting agent with chemotherapy would be expected to have an additive, if not synergistic, suppressive effect on these cells. In mice, MTD chemotherapy causes an acute drop in the level of circulating EPCs, but this is followed by a rapid rebound to normal or above normal levels during the subsequent drug-free rest period (37). This rebound has also been observed clinically (39) and is similar to the effects of chemotherapy on hematopoiesis, namely, a decline in hematopoietic (CD34+) progenitor cells, and hence, neutrophils, followed by mobilization and recovery of such cells during the break period. Thus, the presence or administration of certain antiangiogenic drugs during the break period could, in principle, suppress the EPC rebound and thus increase the efficacy of the chemotherapy by maximizing its antivascular properties, without compromising recovery from myelosuppression. The impact of antiangiogenic drugs has not yet been tested in the context of dose-dense chemotherapy regimens that are supported by hematopoietic growth factors such as granulocyte colony-stimulating factor (G-CSF). There is a reason that this should be given consideration, however. Preclinical studies have shown that G-CSF can mobilize EPCs (40, 41), which theoretically could stimulate tumor angiogenesis and tumor growth (40).

In the context of this third mechanistic model, it is interesting to consider a form of dose-dense chemotherapy called “metronomic” or “antiangiogenic” chemotherapy (25, 27, 29, 42). Metronomic chemotherapy refers to the close, regular administration of low, nontoxic doses of chemotherapeutic drugs with no extended drug-free breaks, over prolonged periods (25). The antiangiogenic activity of metronomic chemotherapy appears to involve direct killing of endothelial cells in the tumor neovasculature (27, 29) and suppression of circulating EPCs (37, 38, 43). At least in vitro, dividing endothelial cells are sensitive to very low doses of chemotherapy (25, 44) that do not cause substantial myelosuppression in vivo (45, 46). Because of its low toxicity, metronomic chemotherapy may be well suited for long-term combination with antiangiogenic drugs; such combinations have marked antitumor effects in preclinical models (25, 27, 29, 42). Both antibody-based and small-molecule antiangiogenic drugs enhance the effects of metronomic chemotherapy in preclinical models (29, 42). Phase II trials of metronomic chemotherapy (46, 47), sometimes used in combination with antiangiogenic drugs, have yielded encouraging results in patients with advanced cancer (48), but larger randomized trials are needed to validate the concept. There is also a need for surrogate markers to help determine the optimal biologic dose of this therapy. Circulating EPCs have been used successfully as a marker in preclinical studies (38), but are not yet validated clinically.


Antiangiogenic drugs are on the cusp of fundamentally changing the practice of clinical oncology. While some have shown activity as monotherapies, most clinical trials to date indicate that they are most effective when piggy-backed onto traditional therapies, especially chemotherapy. It may be possible to enhance their chemosensitizing activity by using chemotherapy protocols involving close regular dosing with shortened break periods. I have outlined several distinct mechanisms that might underlie their chemosensitizing activity, but these proposed mechanisms are not mutually exclusive; they may operate concurrently or sequentially. In addition, there may be other mechanisms at work that do not depend on the antiangiogenic or vasculature-modifying properties of these drugs. For example, drugs that target VEGF could act directly on tumor cells that aberrantly express VEGF receptors and depend in part on VEGF for their survival (49).

Considering the enormous, decades-long, but still mostly clinically unsuccessful efforts aimed at the development of effective chemosensitizers and radiation sensitizers, such as hypoxia-activated bioreductive pro-drugs or agents that reverse drug efflux pumps, the chemosensitization prowess of antiangiogenic drugs stands out as a truly unexpected irony. The more we learn about the vascular-modifying properties of antiangiogenic drugs, the more effective they should become, not only as chemosensitizers but possibly as sensitizing agents for other types of cancer drugs, both old and new.

References and Notes

Stay Connected to Science

Navigate This Article