You have full access to this article via Jozef Stefan Institute.
Judah Folkman's (Fig. 1) fascination with the vascular circulation can be traced to his boyhood in Bexley, Ohio, USA, during which he could be found most often in his basement laboratory working on his own scientific projects. One such project involved participating in a scientific competition with the goal of keeping a perfused rat heart beating for as long as possible outside of the body. According to the story, Folkman's complex homemade perfusion apparatus and special perfusion formula kept the heart beating well beyond that of all the other entrants. Interestingly, tissue perfusion, vascularization, and tissue growth and survival were crucial elements in Folkman's career in science and medicine (TIMELINE).
After graduating with Phi Beta Kappa honours from the Ohio State University, Folkman enrolled at Harvard Medical School in 1957, where he became involved in surgical research under the tutelage of several prominent surgeons, most notably Robert Gross of the Boston Children's Hospital, whose interest was in repairing cardiac defects in newborn children. Under this influence, Folkman decided to become a paediatric surgeon himself and to combine a career of surgery and research to best serve his patients. His surgical training took place at the Massachusetts General Hospital and he then spent a year at Children's Hospital of Philadelphia training with the noted paediatric surgeon C. Everett Koop.
Tissue perfusion and tumour growth
A critical juncture in Folkman's research career took place immediately after his graduation from medical school but before his actual surgical residency. For two years, from 1960 to 1962, he conducted research at the US National Naval Medical Center in Bethesda, Maryland. The goal of this research was to find blood substitutes that could be stored on naval vessels for long periods of time and then used to replace blood lost after trauma. Remembering his earlier research as a high school student, Folkman and his colleague Frederick Becker decided to perfuse isolated organs with their candidate blood substitutes, which consisted primarily of haemoglobin solutions. After some experimentation, the thyroid was selected as the ideal organ for these experiments. They revealed that the viability of these tissues could be maintained in these solutions for several days.
As a further test of the properties of these perfused organs, Folkman, David M. Long and Frederick Becker implanted small explants of murine melanomas into the perfused thyroids. The result was surprising. The tumours grew to an approximate diameter of 1-2 mm and then stopped growing, even though both the tumour and thyroid tissues remained viable. However, the tumours retained the capacity for growth; when they were transplanted back into mice, they grew and expanded rapidly to produce large tumours. The obvious question to Folkman was why tumour growth was permitted in the intact animal but limited in the perfused tissue. As a surgeon, Folkman had frequently noticed that tumours in both humans and animals were highly vascularized and often bled easily. This had been attributed previously to an "inflammatory" response to the tumour, which was associated with increased vascularity and increased vascular permeability. Confronted with the small avascular tumours in the perfused thyroids and the luxuriant, hypervascular tumours in the mouse, he concluded that the blood vessels must be the crucial element. Tumour growth in the perfused organs would be limited by the inability of oxygen and nutrients to penetrate the tissue beyond the diffusion limits of approximately 1 mm.
Tumour angiogenesis
The perfused organ experiments convinced Folkman that the recruitment of new blood vessels was essential for the continued growth of the tumour. He continued working on this when he returned to Boston to establish his own research laboratory, first at Boston City Hospital and then at Boston Children's Hospital, where he was appointed Chief of Surgery in 1967. It was there, in the late 1960s and early 1970s, that he refined his theory on the dependency of tumours on their blood supply. The hypothesis was synthesized in a landmark paper "Tumour angiogenesis: therapeutic implications", published in the New England Journal of Medicine (NEJM) in 1971 (Ref. 3). Additional experimental evidence underlying the theory was published in the following year in Annals of Surgery.
The 1971 NEJM paper not only elaborated Folkman's full-blown theory of tumour angiogenesis, it provided the underpinnings for nearly four decades of his future research and the blueprint for a field of discovery that barely existed at the time. The hypothesis he outlined had several important premises.
Folkman's principal contribution was not just to notice that tumours could attract blood vessels; this had been noted before. The important insight in this paper was that "solid tumours are far more dependent upon new capillary sprouts than we had previously believed", that is, the notion that tumour growth actually depended on the recruitment of these vessels. This insight was also the source of the therapeutic implications of his theory. Folkman also noted that the tumour cells and the endothelial cells of new blood vessels "constitute a highly integrated ecosystem" in which "the mitotic index of the two cell populations may depend on each other." In contemporary terms, this could be considered an early realization of the role of the tumour microenvironment in regulating tumour cell growth.
An important tenet of the theory of tumour angiogenesis was that tumour cells should be able to release diffusible factors capable of stimulating endothelial cell proliferation. Folkman wrote, "Our own experiments have demonstrated the induction of DNA synthesis in previously resting endothelial cells of capillaries and venules within 2 to 3 mm of microimplant of less than 1 million [tumour] cells." The molecular identity of these factors was not yet known and Folkman chose to call the prototype molecule "TAF" for tumour angiogenesis factor. The actual experimental data that supported this assertion was published the following year using autoradiography and electron microscopy to visualize the nuclei of endothelial cells undergoing DNA replication in response to a nearby tumour. The NEJM paper also notes that tumour-induced blood vessels are not permanent: "...withdrawal of TAF is followed by disappearance of newly formed capillaries". This crucial observation provided the first clue that depletion of angiogenic factors in a tumour environment would result in regression of the new blood vessels.
Perhaps the most powerful statement made in the 1971 paper was that "...antiangiogenesis therapy, perhaps by immunization against TAF, should provide a powerful adjunct to the control of solid neoplasms". The prediction was that control of vascularity within a tumour would facilitate control of the tumour. If, in fact, one substitutes VEGF for TAF, then it is clear that Folkman anticipated the clinical use of anti-angiogenic antibodies such as bevacizumab (Avastin) decades before they were available in the clinic. It is also noteworthy that he anticipated here that angiogenesis inhibitors would ideally be used in combination with other therapeutic modalities. Moreover, Folkman did not expect that the tumour would disappear completely: "If blockade of angiogenesis results in a tiny dormant tumour, what is the mechanism of the dormant state?" Rather, he thought that the tumour would shrink to a size similar to that of the 1-2 mm avascular primary tumour that he had observed in the perfused thyroids. He also expected that these small tumours would be dormant and harmless if and until they were allowed to undergo a new round of angiogenesis and attract more blood vessels.
Initially the Folkman hypothesis was considered to specify principally the relationship between angiogenesis and tumour growth. However, Folkman also noted that "[i]t is possible that metastases will not arise from a nonvascular tumour". This important insight was not immediately appreciated by the scientific community, despite being stated explicitly in this paper.
The fragile nature of the tumour vasculature was also obvious to Folkman: "To my mind, the necrotic center of a large tumour was at an earlier time well vascularized. However, the enormous pressures that build up within a large tumour could diminish blood flow to the center." Over many years, this hypothesis was shown to be true by the elegant work of Rakesh Jain. Increased hydrostatic pressures at the interior of tumours close off the fragile new blood vessels, leading to poor perfusion and central tumour necrosis.
Finally, the therapeutic implications and the need for more information were made abundantly clear: "The mechanism by which tumour implants stimulate neovascularization must be well understood before therapy based on interference with angiogenesis can be devised." Judah Folkman single-mindedly devoted his scientific career to this premise, but always with the long-term goal of developing new treatments for cancer patients. By the time of his death in January 2008, he could not help but see that this goal was well on its way to being realized.
Folkman's hypotheses did not arise in a vacuum. The study of blood vessel growth in development and in wound healing had been an active field of inquiry since the eighteenth century. The term angiogenesis, meaning the formation of new blood vessels, had been coined in 1787 by the British surgeon John Hunter and the vascular anatomy of tumours was studied in detail by anatomists throughout the nineteenth century. A major technical advance was the use of transparent chambers in the ears of experimental animals, which allowed the temporal study of tumour angiogenesis. It had also been shown that tumours released diffusible factors that could stimulate the growth of new blood vessels. Despite these findings, no one before Folkman had suggested so boldly that tumour angiogenesis was a necessary component of tumour growth. What was especially novel was Folkman's assertion that antagonism of the angiogenic process could constitute a new form of cancer therapy (Fig. 2). The 1971 paper was the first to employ the term "antiangiogenesis" to describe a potential therapeutic approach.
Defining the field
Among the several hundred publications that Folkman authored over 45 years, it is possible to discern four major themes that helped to define the field of angiogenesis research. These are the development of reproducible angiogenesis assays, the demonstration that tumours were angiogenesis-dependent, the identification of molecular stimulators of tumour angiogenesis and the identification of naturally occurring inhibitors of tumour angiogenesis.
Development of assays for the study of angiogenesis. At the time that Folkman began his studies on the relationship between angiogenesis and tumorigenesis, there were a few in vivo bioassays that had provided the majority of the earlier data in the field. These included placing tumours in transparent chambers in the skin of an animal in a site such as the rabbit ear or implanting tumours directly into sites such as the rat dorsal air sac or the hamster cheek pouch. Although these assays had yielded considerable descriptive data regarding the vascularization of growing tumours, they were limited by the expense of the animals, the requirement for surgical skill, the presence of background vessels in the surrounding tissue, the presence of an inflammatory response and the lack of quantitation associated with the assays. In the 1960s and 1970s, Folkman and his colleagues developed or perfected nearly all of the in vivo and in vitro assays that are used in the field today. His work started with perfused organs as described above and then moved in vivo, first using the rat air sac model. Using this assay, Folkman and Ramzi Cotran were able to use autoradiography to demonstrate for the first time that the host endothelial cells were undergoing DNA synthesis in response to an implanted tumour.
To eliminate the presence of background vessels, Folkman and Michael Gimbrone sought to find a tissue that was normally avascular but could support the growth of new blood vessels when stimulated by a tumour. After initial experimentation with the iris, they settled on the rabbit cornea. The experimental design was to make a hollow pocket in the cornea and implant the tumour into the pocket. If the tumour was placed proximal to the nearest blood vessels in the limbus of the eye, blood vessel sprouts would move across the previously avascular corneal tissue toward the tumour. This semi-quantitative assay not only beautifully confirmed the capacity of tumours to secrete diffusible angiogenic factors, but quickly became the standard assay in the field. One advantage of the assay was that it clearly showed the movement of new blood vessels through a tissue that had no background vessels to obscure the view.
The next issue was to be able to use this assay to measure the activity of biochemical fractions that were being used to purify the putative tumour angiogenesis factors. Folkman had shown an early interest in the concept of using biomaterials as slow-release depots. He had, in 1964, pioneered the use of silicon as a biomaterial for the slow release of small molecules, a technology that gave rise to the implantable contraceptive Norplant. Capitalizing on the presence in the laboratory of a talented young bioengineer named Robert Langer, Folkman sought to find a polymeric system that could slowly release macromolecular angiogenic factors, a feat that had never been achieved. Together, they developed the use of polymeric systems that could deliver proteins continuously over an extended time period. Now they were able to mix potential angiogenic factors into the polymeric release system and apply that mixture to the corneal pocket. This assay allowed the first analysis of fractionated angiogenic factors that had been isolated from tumour cells. It also opened up an entirely new field of slow-release delivery of macromolecules.
The corneal assay was sensitive and reproducible but was still costly and time-consuming. To deal with these issues, a second assay was developed that used chick chorioallantoic membrane (CAM). Although angiogenesis had been observed when tumours were grown in this site as early as 1913 (Ref. 75), Folkman and his colleagues were the first to use it specifically as a dedicated, semi-quantitative angiogenesis assay. This was easier to perform and less expensive than the corneal assay and it was rapidly and widely adopted. Two limitations of the assay were that there was a background of normal vessels on the CAM and there was often an immune response to dust or other foreign bodies that landed on the membrane. The latter could be prevented by addition of anti-inflammatory steroids to the assay. Inadvertently, this approach answered one of the old concerns that tumour angiogenesis was simply an immune response. Now Folkman could easily show that an angiogenic response to a tumour could take place in the absence of inflammation.
Rapid analysis of angiogenesis factors required an in vitro assay to allow quantitative testing of large numbers of tumour-derived fractions. Starting early in his career, Folkman endeavoured to develop in vitro angiogenesis assays that used cultured vascular endothelial cells derived from the inner walls of blood vessels. The earliest assay he used was to explant aortic fragments and measure DNA synthesis in the aortic endothelium after exposure to tumour factors. The group then turned to the in vitro culture of endothelial cells from human umbilical veins and later were the first to achieve long-term culture of capillary endothelial cells. Because angiogenic vessels were frequently capillary in origin, these cells were thought to more closely duplicate the response of capillary endothelial cells in the host tissue. A further advance was achieved with the finding that the capillary endothelial cells could be induced to form three-dimensional networks in vitro that had some properties of capillary networks in vivo. This assay is frequently used today to assess the in vitro activity of potential angiogenesis inhibitors (Fig. 3).
Demonstration that solid tumours are angiogenesis-dependent. As discussed above, the hypothesis that tumours were angiogenesis-dependent was attractive but unproven and was introduced against a background of thought that held that the tumour blood vessels were innocuous bystanders witnessing tumour growth rather than participating in it. Folkman was obligated to find a way to demonstrate that tumours could not expand without the ability to recruit new blood vessels. In 1990, he looked back to assess all the evidence in support of the theory that "every increase in tumour cell population must be preceded by an increase in new capillaries converging on the tumour". The most significant evidence that he and his colleagues provided were the elegant experiments from the corneal studies. These showed clearly that tumour growth was limited to 1-2 mm spheroids until vascularized, after which the tumours rapidly expanded. This system also showed that the tumour-induced vessels required continued stimulation by tumour-derived factors, as removal of the tumour from the cornea resulted in regression of the new blood vessels. An important implication of this finding was that antagonism of an angiogenic stimulus could be effective even after new blood vessels were established, because the transient new blood vessels would regress in the absence of the stimulus.
In a second model, also in the eye, tumours were implanted into the vitreous fluid in such a way that they remained suspended in the vitreous fluid at a sufficient distance from any blood vessels so that angiogenesis could not take place. Intriguingly, the small tumours remained viable for months and could be induced to grow at any time if placed contiguous to blood vessels in the iris. These experiments provided early evidence for the concept that avascular tumours could survive in a dormant state -- an equilibrium state in which cell division is matched by cell death. Tumour growth could be rapidly induced at the time that the tumours became vascularized. The concept of tumour dormancy became a long-standing interest for Folkman, who sought to understand the role of angiogenesis in this process. He came to believe that delayed onset of angiogenesis could explain why some tumours in humans could be maintained in a seemingly harmless state for years, only to later expand and become life-threatening.
Identification of tumour-derived angiogenesis factors. Folkman initially used the term TAF to denote the material elaborated by the tumour that induced the formation of new blood vessels. The nomenclature suggests an initial thought that there might be just one such factor. As could be easily imagined, there was substantial early interest in the molecular definition of this factor. Early attempts to purify a TAF from tumour samples proved unsuccessful, probably owing to the cumbersome nature of the in vivo assays. Progress remained slow until the switch was made to in vitro assays to guide purification of angiogenic factors. A breakthrough came when Michael Klagsbrun, Yuen Shing and Folkman decided to fractionate chondrosarcoma extracts on the basis of heparin affinity. The first protein so purified turned out to be the fibroblast growth factor (FGF), originally isolated by Gospodarowicz and known to be a potent endothelial cell mitogen. This was the first example of an angiogenesis factor purified directly from tumour lysates, and FGF remains an important example of an angiogenesis stimulator. It soon became apparent, however, that there were several different tumour angiogenesis factors and that different tumours produce different factors. A major concept that has emerged suggests that in certain human cancers there is a progression from production of a single angiogenesis factor to the production of several such factors over a period of time. This finding suggests that antagonizing just one angiogenesis factor would be effective only against tumours that produced just that one factor. More aggressive tumours that produced multiple angiogenesis factors would require more broadly acting antagonists.
Identification of angiogenesis inhibitors. It is fair to say that the principal focus of Judah Folkman's scientific career was the search for inhibitors of angiogenesis that could be useful as adjuncts to tumour therapy. Over the years, he and his colleagues were the first to isolate angiogenesis inhibitors; his own laboratory identified more than a dozen such inhibitors, several of which are in use in patients today. His work also stimulated myriad other laboratories to develop additional angiogenic inhibitors, an activity that continues unabated throughout the academic and pharmaceutical communities. Folkman discerned the possibility of several classes of angiogenesis inhibitor, including endogenous biological molecules derived from avascular tissues, naturally occurring angiogenic inhibitors from normally vascularized tissues, naturally occurring inhibitors produced by tumours themselves and synthetic angiogenesis inhibitors. He also distinguished between what he termed "direct" angiogenesis inhibitors and "indirect" inhibitors. The direct inhibitors were those that worked directly on the endothelial cell processes involved in neovascularization whereas the indirect inhibitors were those that expressly targeted pro-angiogenic factors. An example of the latter would be an antibody to a specific angiogenic factor, a concept that Folkman had proposed in the 1971 NEJM paper.
Certain tissues in the body have small numbers of blood vessels and appear to be relatively avascular. Folkman speculated that sites such as cartilage or the cornea were avascular owing to the presence of naturally occurring angiogenesis inhibitors in those tissues. In 1980, his group showed that partially purified cartilage extracts could inhibit angiogenesis in vivo. They later showed that this activity could be attributed in part to inhibition of metalloprotease activity. This was not totally unexpected as it had been shown that activation of endothelial tissue-degrading proteases was an essential component of the endothelial invasion during the angiogenic process.
Folkman soon came to realize that angiogenesis inhibitors were not restricted to avascular tissues. Over his career, he reported anti-angiogenic activity from a variety of naturally occurring molecules including protamine, platelet factor 4 (Ref. 36), heparin fragments administered in combination with steroids (an early example of the use of combination therapies to suppress angiogenesis), fumagillin, a fungus-derived product that had previously been used as an antibiotic but was found by Ingber and Folkman to have anti-angiogenic activity, 2-methoxyoestradiol, interleukin 12 (Ref. 40), antithrombin III (also known as SERPINC1), the vitamin D binding protein (also known as GC), macrophage-activating factor and a cleaved form of b2-glycoprotein I (also known as APOH). Folkman also pointed out the existence of drugs that, although not anti-angiogenic themselves, could increase levels of endogenous angiogenesis inhibitors.
It is clear that the search for novel angiogenesis inhibitors occupied Folkman over his entire career. Although most of Folkman's own work focused on the search for novel naturally occurring inhibitors, one notable discovery made by Robert D'Amato while working with Folkman showed that synthetic drugs could be predicted to have anti-angiogenic activity based on their side-effect profiles. Using this approach, they suggested and then showed that thalidomide had anti-angiogenic activity, leading to the eventual use of this molecule in the treatment of cancer. Others who were trained by Folkman, or influenced by his work, have continued the search for such molecules, to the point where the number of angiogenesis inhibitors that have been described in the literature is rapidly increasing. Folkman himself described more than 70 angiogenesis inhibitors in a recent review.
Folkman came to see angiogenesis stimulation and inhibition as part of a balanced equation. In most tissues, angiogenesis was held in check by a balance of stimulatory and inhibitory factors. When angiogenesis was necessary, it could be achieved either by increasing the level of angiogenic stimulators or decreasing the level or activity of the angiogenic inhibitors (Fig. 4). Perhaps the most surprising finding was that tumour cells themselves produced angiogenesis inhibitors at the same time that they were expressing angiogenesis stimulators. The novel purification scheme involved fractionation of urine from tumour-bearing mice using an assay for suppression of tumour growth in vivo. This methodology resulted in the identification of two tumour-derived angiogenesis inhibitors that were named angiostatin and endostatin. Intriguingly, both molecules were peptides that had been cleaved from larger naturally occurring, biologically active proteins: angiostatin from plasminogen and endostatin from collagen XVIII. This suggested that proteolytic enzymes produced by the tumour cells were actually crucial for the generation of angiogenesis inhibitors. Consequently, the balance of tumour-derived proteases and protease inhibitors could, in turn, affect the balance of angiogenesis stimulators and inhibitors.
The finding that tumours produced angiogenesis inhibitors caused Folkman to return to the concept of tumour dormancy that he had first addressed in the 1970s. If tumour-derived inhibitors were sufficiently stable to act downstream on small metastatic colonies, then these colonies might be kept small and dormant for extended time periods. This was supported by experiments in which removal of primary tumours resulted in rapid growth of previously dormant metastatic colonies. Administration of exogenous angiogenesis inhibitors could continue to suppress the secondary tumours even after removal of the primary tumours.
Certain aspects of Folkman's work have been controversial. The first of these was the original concept that tumours were angiogenesis-dependent, an idea that is now widely accepted. There has also been criticism regarding the efficacy of angiostatin and endostatin. Because of high levels of publicity surrounding the findings at the time they were made, there were expectations that these drugs might represent a great breakthrough in the treatment of human cancer. Although both are effective at arresting tumours in experimental animals, clinical trials using endostatin in cancer patients have been only sporadically positive. However, the attention surrounding the discovery of these molecules served to galvanize research on a variety of other angiogenesis inhibitors both in Folkman's laboratory and elsewhere. Several of these drugs are now routinely used in the treatment of human cancer.
In the last few years of his life, Folkman worked hard to develop new formulations of a promising drug that had been discovered earlier in his laboratory. TNP-470 is a fumagillin analogue that is an extremely effective angiogenesis inhibitor in animals. Initial clinical trials showed potential anti-tumour activity in humans, but with an associated neurotoxicity. Folkman later reported two formulations that were designed to stabilize TNP-470 and prevent it from crossing the blood-brain barrier. Caplostatin, a conjugate of TNP-470 and hydroxypropyl methacrylate, which was shown to inhibit angiogenesis and tumour growth in animals without associated neurotoxicity. In a posthumous paper in Nature Biotechnology, Folkman, and Ofra Benny, reported a conjugate of TNP-470 and polyethylene glycol called lodamin that could inhibit angiogenesis and tumour growth when administered to mice in their drinking water. Neither of these formulations has yet been tested in humans.
Metronomic therapy
The possibility of treating patients with angiogenesis inhibitors required a change of thinking about the way in which the drugs were administered. Traditionally, chemotherapy had been given using high drug doses over a short time period to try to rapidly eliminate all cancer cells. Chronic dosing was not usually possible owing to the general toxicity of the drugs. Angiogenesis drugs were generally less toxic and could theoretically be given over longer time periods. Furthermore, Folkman realized that the result of anti-angiogenesis treatment would most often be arrested tumour growth or regression to a small, non-vascular tumour. All the tumour cells would not generally be eliminated. Consequently the best treatments would be given at lower doses over long periods of time. Another realization was that some traditional cytotoxic, chemotherapeutic drugs could be anti-angiogenic when given at doses lower than those commonly used in patients. In 2000, Timothy Browder, Folkman and colleagues reported that using cyclophosphamide, a traditional anticancer drug, at lower doses over longer periods of time resulted in angiogenesis inhibition and tumour regression. Similar results were reported at the same time by Robert Kerbel and colleagues, who used a combination of low-dose vinblastine along with an antibody to the cellular receptor for the pro-angiogenic molecule vascular endothelial growth factor (VEGFA). This type of regimen has since come to be known as metronomic therapy and is currently the subject of numerous clinical trials.
Angiogenesis and cancer diagnosis
Early in his career, Folkman realized that the elaboration of angiogenesis activity by tumours might enable their detection. The first indication that this might be possible was the finding that the aqueous fluid of patients with eye tumours possessed angiogenic activity that could be detected using the chick CAM assay. This was followed shortly by reports that patients with transitional carcinoma of the bladder had detectable angiogenic activity in their urine. This activity was later shown to be due to increased urinary levels of basic FGF (FGF2). He then moved from monitoring angiogenesis activity to monitoring the vascular density or number of blood vessels per unit area of tumour. In a landmark paper published with Noel Weidner and colleagues in 1991, they showed that increased vascular density correlated with the development of metastatic disease in breast cancer patients, and later extended this finding to prostate cancer. This correlation between angiogenesis and tumour metastasis is now used as a prognostic indicator in a variety of tumour types.
In the last few years of his career, Folkman turned his attention to the accumulation of angiogenesis factors in platelets. This novel finding had implications both for the biology of metastasis and for tumour diagnosis. Because platelets appear to concentrate the angiogenic factors, detecting these factors in platelets rather than in whole blood provides a more sensitive method for early tumour detection. He argued eloquently that we were in sight of the day when tumours might be detected by the sensitive detection of platelet-sequestered factors at a time when they would be undetectable by any other method. At such a time, he envisioned that broad-spectrum angiogenesis inhibitors might be applied even if the absolute location of the tumour were not known at the time. Over his career, Folkman made many predictions that seemed provocative at the time, only to be proven true, often many years later. Only time will tell whether future cancer regimens include early detection of tumours by their angiogenic factors followed by early anti-angiogenic therapy.
Clinical implications
Judah Folkman entered into the field of angiogenesis research in the 1960s because he thought it provided a chance for a novel approach to cancer treatment. Today, four decades later, it is clear that angiogenesis inhibitors do have a role in cancer therapy. As Folkman suggested in the 1971 NEJM paper, they are best used in combination with other therapeutic modalities. The clinical use of angiogenesis inhibitors in the treatment of cancer is still in its infancy but at the time of his passing it was evident that these inhibitors could act in concert with other therapies to extend the lives of cancer patients. Folkman himself wrote a review that summarized the state of the art of angiogenesis inhibitors just last year. The initial therapeutic advantage is small but is certain to grow as scientists learn to make better inhibitors and physicians learn how best to use the drugs. The initial drugs that are proving to be effective fall into the class of what Folkman would have called "indirect" inhibitors, in that they act as antagonists of specific angiogenesis factors or their receptors. More direct antagonists of endothelial function that work regardless of the particular pro-angiogenic factor produced by the tumour will surely follow. The appropriate dosing schedules and the best combinations to use are still being worked out and much progress will no doubt be made in the next 10 to 20 years. This much is clear: what started as an idea from a young surgeon grew into a hypothesis, then into a field and now into an industry where it is rare to find a pharmaceutical company without an angiogenesis programme. Few ideas generate so much traction. Folkman was very proud of the figure that showed the exponential growth of publications that used the term "angiogenesis" between the three articles published in 1971 and the more than 4,000 published in 2007 (Fig. 5). It may also be that the broadest clinical implications of angiogenesis go well beyond cancer to the treatment of a variety of other diseases that are characterized by the abnormal growth of blood vessels, including inflammatory diseases and vision-threatening eye diseases. His goal of doing research that would have a positive effect on the lives of his patients has most certainly been realized.