History of Early Research in Angiogenesis

Angiogenesis, the process by which new blood vessels are formed from preexisting vessels, plays a major role in normal biology such as embryonic and adult development. This process is also essential to pathologic states such as the development and proliferation of tumors. The term angiogenesis is distinct from neovascularization, which is used to describe the formation of new vascular networks from vascular progenitor cells rather than from preexisting blood vessels as occurs with angiogenesis.

Scientists, including Rudolf Virchow, first made the link between tumor growth and their specific blood supply leading to increased vascularity more than a century ago. In the late 1920s, Warren Lewis was the first to describe that spontaneously growing tumors in rats had different types of vasculature depending on the type of tumor. This discovery was the first step toward the understanding that the tumor environment plays a crucial role in the morphologic characteristics as well the growth rate of each tumor’s blood vessels.

Intravital analysis with transparent chambers, a process in which a transparent chamber is implanted into a rabbit’s ear, allowing microscopic visualization of vessels, was an important tool that allowed the observation of angiogenesis, and was developed by J.C. Sandison in 1928. Using this technique a decade later, Ide and colleagues began to examine the relationship between growth rates of carcinoma in relation to its vascular supply in rabbits. They found that, as a tumor began to grow, a widespread and rapid establishment of blood vessels accompanied it, evolving as the tumor increased in size. The investigators made the important leap of establishing that not only did this blood vessel formation help the tumor grow because it provided a growing tumor with necessary nutrients and oxygen but that tumor growth depended on this vessel formation and without this new vessel formation a tumor would not grow. This observation was the foundation of attempts to inhibit angiogenesis decades later.

An important article from Algire and colleagues at the National Cancer Institute further expanded on this hypothesis in 1945. Adapting the transparent chambers to a murine model, they counted the number of blood vessels daily, thereby advancing this technique to a quantitative modality for assessing blood vessel growth in relation to tumor proliferation. Similar to Ide and colleagues, they too observed that it was the tumor and not the normal tissue that permitted increased blood vessel formation and growth. Furthermore, they observed that vascular growth took place before the tumor entering its rapid growth state. They concluded that a tumor’s ability for blood vessel formation is perhaps one of the most essential steps for tumor formation and growth, a notion that eventually gave rise to the term angiogenic switch.

There were minimal advancements in this field until a renewed interest developed in the 1960s. Using a transplantable mammary gland carcinoma model, Tannock and colleagues continued to explore the correlation between tumor-cell and endothelial-cell proliferation by applying newly arising autoradiographic techniques. They were able to show that, as tumor cells moved further away from endothelial cells, their mitotic index was decreased proportionally. This finding was the first direct evidence to show that tumors depend on the nutrients and oxygen diffused from endothelial cells, serving as the rate-liming step for the growth of tumor cells.

Defining Proangiogenic Tumor Secreted Factors

Although many investigators had speculated that a specific factor released by tumor cells led to neovascularization and tumor angiogenesis, direct experiments on this topic were not conducted until the late 1960s. By adapting transparent chambers to a pouch inserted in hamster cheeks, Greenblatt and Shubi at Chicago Medical School and Ehrmann and Knoth from Harvard Medical School showed in parallel that choriocarcinoma or melanoma cells transplanted with a filter interposition between the host and the tumor still promoted the proliferation of blood vessels. Both groups concluded that the most conceivable explanation for the witnessed phenomenon was the diffusion of tumor-produced secreted factors driving tumor vessel formation.

In 1971, Judah Folkman introduced the concept of anti-angiogenic compounds in order to combat human cancers. Folkman and colleagues also published on their isolation attempts of a tumor angiogenesis factor (TAF) from animal and human tumors, which stimulated angiogenesis in the dorsal air-sac model of rats. In addition, they showed the in vitro production of TAF activity from cultured cells, by assessing its ability to stimulate new vessel development in the chick chorioallantoic membrane. In the ensuing 15 years, several pro angiogenic molecules were discovered, including transforming growth factor-alpha, angiogenin, fibroblast growth factor 1 (FGF; also referred to as aFGF), and bFGF (basic fibroblast growth factor), although their specific roles in angiogenesis regulation remained a mystery.

In 1983, Senger and colleagues isolated a protein through partial purification from guinea pig cancer cell conditioned medium. The identified protein was named tumor vascular permeability factor, because of its ability to stimulate vascular leakage in the Miles assay. By 1989, Ferrara and colleagues at Genentech successfully isolated and identified an endothelial-cell mitogen from medium conditioned by folliculostellate cells of bovine pituitary. Because of the high mitogenic levels present, they concluded that the isolated protein was most likely secreted, unlike bFGF, which is produced by the same cell line, but stored internally as previously reported. The protein was named VEGF because of its vascular endothelial cell–specific growth stimulation.

Bevacizumab as Monotherapy for Recurrent Glioblastoma

By 2009, several studies had been published that assessed bevacizumab as a monotherapy for recurrent glioblastoma. The results from these studies were promising, with the 6-month PFS in phase II trials ranging between 29% and 43%, whereas the overall response rate increased to 28%. These results outperformed the findings of studies in which patients received the traditional treatment modality for recurrent glioblastoma, such as various chemotherapeutic regiments and radiation therapy with 6-month PFS ranging from 4% to 9% and 9% to 21%, respectively. When treated with bevacizumab, patients with recurrent glioblastoma had a 35% radiological response rate under the Macdonald criteria, whereas this number increased to 71% when the assessment used the Levin criteria. Taken together, these phase II clinical trial results proved the initial safety and efficacy of bevacizumab for recurrent glioblastoma and led to the FDA approving bevacizumab for recurrent glioblastoma in 2009. This approval made bevacizumab only the second systemic drug to be approved for the treatment of glioblastoma in almost 4 decades. Unlike temozolomide and the locally applied carmustine wafers, bevacizumab’s approval received an accelerated designation because of its FDA approval without a randomized phase III clinical trial.

Bevacizumab in Combination with Other Treatments for Recurrent Glioblastoma

Although the numerous targets of VEGF receptor (VEGFR) inhibitors have allowed them to become effective as monotherapeutic agents, many clinicians consider bevacizumab and other VEGF-targeted therapeutics to perform with an increased efficacy when administered in combination with other agents.

When irinotecan, a topoisomerase I inhibitor, was administered as a monotherapy to patients with glioblastoma, its antitumoral effects were minimal, with response rates ranging from 0% to 17%. These results were comparable with those of other chemotherapeutics targeting glioblastoma that were used at that time. Phase II clinical trials in which irinotecan and bevacizumab were administered concomitantly went on to show a significantly increased PFS ranging from 38% to 50.3% with the 6-month overall survival (OS) ranging between 72% and 77%. Several explanations can account for the increased efficacy that ensued as a result of combining bevacizumab with irinotecan. In the presence of bevacizumab, irinotecan may benefit from an increased uptake into the central nervous system, and/or the ability of bevacizumab to target glioma stem cells, whereas irinotecan targets the differentiated tumor cells. When comparing bevacizumab as monotherapy with bevacizumab and irinotecan combination therapy, the PFS was increased from 29% with monotherapy to 46% with combination therapy. The radiological response rate using the Macdonald criteria also showed an increase from 35% for patients who received bevacizumab as a monotherapy to 57% for the combination trial, once again supporting the notion of an increased benefit when irinotecan is added to bevacizumab. However, combining bevacizumab treatment with temozolomide in a study with 32 patients did not prove to be as promising as irinotecan, and the results were inferior compared with the bevacizumab monotherapy trial. For this study, the 6-month PFS was only 18.8%, with a median PFS of 15.8 weeks. The 6-month OS was 62.5% with a 37-week median OS.

The encouraging results from the irinotecan and bevacizumab combination study led to the search for other potential therapeutic partners for bevacizumab. Sorafenib, an inhibitor of Raf kinase and VEGFR, was of great interest because of the promising preclinical results it produced with antitumoral activity against gliomas. Because sorafenib only proved to have a modest effect as a single agent, it was combined with bevacizumab to inhibit the VEGF/VEGFR axis. However, sorafenib combined with bevacizumab did not improve the outcomes significantly, with 6-month PFS ranging from17% to 26% compared with 18% with bevacizumab as a single agent or historical controls, which may have been attributed to sorafenib’s inadequacy in crossing the blood-brain barrier. At present, sorafenib is not recommended in combination with bevacizumab for glioblastoma. In addition to sorafenib, several other agents were combined with bevacizumab and studied for recurrent glioblastoma, including etoposide, erlotonib, fotemustine, carmustine, and carboplatin. Not only were the benefits minimal but these drugs were sources of potential side effects.

A recently completed randomized trial in Europe made significant advances relative to the studies cited earlier in defining the role of bevacizumab in treating recurrent glioblastoma. This phase III trial, the European Organisation for Research and Treatment of Cancer (EORTC) 26101 trial, explored the combination of bevacizumab and lomustine in patients with a first recurrence of a glioblastoma. Results presented at the 2015 Society for Neuro-Oncology annual meeting revealed that OS was not superior in patients receiving lomustine plus bevacizumab (n = 149 patients) compared with those receiving lomustine alone (n = 288 patients; hazard ratio [HR], 0.95; confidence interval [CI], 0.74–1.21; P = .65), whereas locally assessed PFS was longer with the addition of bevacizumab to lomustine (HR, 0.49; CI, 0.39–0.61).

Bevacizumab in Newly Diagnosed Glioblastoma

Most of the early studies of bevacizumab in glioblastoma were based on treatment of recurrent glioblastoma. In order to fully investigate the therapeutic potential of bevacizumab for patients with glioblastoma beyond its benefits for recurrent disease, the reasonable subsequent step was to determine the drug’s efficacy earlier in the course of the disease. Phase II trials were conducted in which bevacizumab was given in addition to the Stupp protocol of temozolomide and radiation therapy. The results were not promising because there was little to no improvement on OS, the primary end point, which ranged from 19.6 to 23 months, compared with an OS of 14.6 to 21.1 months when the Stupp protocol was followed without the addition of bevacizumab. Although the OS was not affected with bevacizumab at diagnosis, the studies showed a significant improvement in PFS, a secondary end point, ranging from 13 to 13.6 months compared with 6 to 9 months with the standard Stupp protocol without bevacizumab. The 6-month PFS rate was also favorable when bevacizumab was added to the Stupp protocol, increasing to 85% to 88%, a value far exceeding the 54% noted with the standard Stupp protocol. These results are consistent with those seen in the disease setting and suggest a short-term benefit of bevacizumab in delaying recurrence that fails to affect OS, perhaps because of the tumor that emerges on recurrence being more biologically aggressive than the originally treated tumor.

The results were further confirmed in 2 randomized phase III trials, the Radiation Therapy Oncology Group (RTOG) 0825 trial and the Avastin in Glioblastoma (AVAglio) trial, which compared the OS (primary end point) and PFS (secondary end point) of patients with newly diagnosed glioblastoma receiving the Stupp protocol alone versus Stupp protocol plus bevacizumab. In the AVAglio phase III trial, median PFS was improved with the addition of bevacizumab to the Stupp protocol, reaching 10.6 months when bevacizumab was added to the Stupp protocol compared with 6.2 months with the Stupp protocol alone ( P <.001). The Stupp protocol plus bevacizumab group had a higher OS 1 year after initiation of treatment (72.4% vs 66.3%, respectively; P <.049), but by 2 years this difference had faded (33.9% vs 30.1%, respectively; P = .24). Median OS, the primary end point of the study, was 16.8 months in the Stupp protocol plus bevacizumab group compared with 16.7 months for patients treated with the Stupp protocol by itself ( P = .1).

The results from the RTOG 0825 phase III trial revealed a similar trend. The median PFS increased with the addition of bevacizumab to 10.7 months compared with 7.3 months in patients who were solely treated with the Stupp protocol ( P = .007). There was no significant difference in median OS observed with the Stupp protocol alone (16.1 months) compared with the Stupp protocol plus bevacizumab (15.7 months) ( P = .21). With the results from these phase III trials, investigators were able to confirm the trends and findings of phase II trials, which showed that bevacizumab given at the time of glioblastoma diagnosis carries a favorable PFS outcome, but that it fails to be of any benefit with regard to OS.

The combination of bevacizumab with the Stupp protocol at the time of diagnosis played a significant role in improving the baseline quality of life, as well as performance measures. The addition of bevacizumab to the Stupp protocol resulted in Karnofsky Performance Scale scores that remained greater than 70 for 9 months, whereas the group receiving the Stupp protocol alone were only able to maintain a score of more than 70 for 6 months. Similarly, performance status was also preserved in the bevacizumab plus Stupp protocol group for a median of 9 months compared with 5.5 months in those treated with the Stupp protocol alone ( P <.001). Furthermore, the time to clinical deterioration for patients receiving bevacizumab in addition to the Stupp protocol was 14.2 months, which was significantly longer than the 11.8 months it took for patients only treated with the Stupp protocol ( P = .02). Note that there was a decreased requirement for glucocorticoid administration for patients who were treated with bevacizumab. However, in the experimental group there was a greater number of complications and serious adverse events deemed attributable to bevacizumab as well as decreased quality of life outcomes and neurocognitive function. These results show that even though bevacizumab added to the standard Stupp protocol at the time of diagnosis has shown results that are promising for PFS in addition to certain clinical outcome measures, the OS for these patients was unchanged, and may be associated with a higher complication rate. These findings confirm the notion that bevacizumab carries limited additional benefits when administered earlier in the course of glioblastoma.