Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer

VEGF has a crucial role in physiological and pathological angiogenesis^1–3. Although VEGF blockade, alone or in combination with cytotoxic therapies, is being tested in a number of clinical trials⁴, including the first successful phase 3 clinical trial⁵, the effects of anti-VEGF treatment on the vascular biology of human tumors are not known. To this end, we recently initiated a National Cancer Institute–sponsored phase 1 clinical trial that integrates the VEGF-specific antibody bevacizumab (Avastin; Genentech) into a contemporary treatment program of preoperative chemotherapy and radiation therapy followed by surgery, for patients with primary and nonmetastatic rectal cancer. To gain insight into the mechanisms of action of bevacizumab, we designed the trial to evaluate the effects of bevacizumab alone on (i) tumor physiology (blood perfusion, blood volume, permeability–surface area product, microvascular density (MVD), perivascular coverage, interstitial fluid pressure (IFP) and 18-fluorodeoxyglucose (FDG) uptake); (ii) systemic response (VEGF level in blood, number of circulating endothelial cells (CECs) and progenitor cells); and (iii) tumor response (see Supplementary Note online for methods).

Six patients with primary and locally advanced adenocarcinoma of the rectum have been enrolled in a preoperative treatment protocol of bevacizumab administration alone (5 mg/kg intravenously), followed after 2 weeks—the approximate half-life of bevacizumab in circulation—by concurrent administration of bevacizumab with 5-fluorouracil and external beam radiation therapy to the pelvis and surgery, 7 weeks after treatment completion. Twelve days after bevacizumab infusion, flexible sigmoidoscopy (Fig. 1) revealed that bevacizumab induced tumor regression of >30% in patient 1, and no change in tumor size in the remaining five patients. Functional computed tomography (CT) scans at this time point indicated significant decreases in tumor blood perfusion (40–44%; P < 0.05) and blood volume (16–39% in four of five patients analyzed; P < 0.05; Figs. 1b and 2; see Supplementary Table 1 online for group statistics). This was accompanied by a significant decrease in tumor MVD (29–59% in five patients analyzed; P < 0.05; Fig. 2d). These three sets of data provide direct evidence of the antivascular effects of bevacizumab in human tumors, which is in line with preclinical findings^6,7.

Twelve days after bevacizumab treatment, IFP was reduced in four of four patients (Fig. 2f) and overall mean IFP decreased significantly from 15.0 ± 2.0 to 4.0 ± 2.2 mm Hg (P < 0.01). The decrease in IFP is in concert with our preclinical findings⁶. Elevated IFP, a hallmark of solid tumors, is a result of abnormalities in tumor vessels (such as abnormal structure of the vessel wall). The decrease in IFP after anti-VEGF treatment may be a result of ‘normalization’; that is, the resumption of normal function in the tumor vasculature⁸. The increased fraction of vessels positive for α-smooth muscle actin in four of five patients (Fig. 2e and Supplementary Fig. 1 online) is supportive of vascular normalization⁹. As a result of the decrease in vascular volume and MVD, one would expect a reduction in vascular surface area, and hence a lowering of the permeability–surface area product. Surprisingly, four of five patients had no significant changes in permeability–surface area product (Fig. 2c and Supplementary Table 1), providing indirect evidence for improved extravasation of the CT contrast agent from the normalized vasculature. Finally, on day 12 the follow-up positron emission tomography (PET) scans indicated no change in tumor FDG uptake in five patients, and showed a 40% decrease in patient 3 (Fig. 2g). Collectively, these data suggest that the efficiency of blood vessels after bevacizumab treatment is improved. These clinical findings are consistent with previous preclinical data on tumor oxygenation⁶ and drug uptake¹⁰ after anti-VEGF treatment.

In addition to being a mitogen and survival factor for endothelial cells, VEGF mobilizes progenitor cells from the bone marrow into the circulation¹¹. VEGF blockade decreased the number of progenitor cells circulating on day 3 (Fig. 2h). No decrease in circulating VEGF (plasma, serum and urine) was detected with the current assay (see Methods; data not shown). After bevacizumab treatment, the number of CD31^brightCD45⁻ cells, but not the total number of CD31⁺CD45⁻ cells, decreased in all patients (Fig. 2h). Further analysis indicated that CD31^brightCD45⁻ cells represented viable CECs (Supplementary Fig. 2 online). An increase in the number and viability of CECs was recently found in lymphoma and breast cancer patients¹². In addition, preclinical studies indicate that CEC kinetics might serve as a surrogate marker of response to treatment¹³. Based on these data, we suggest that the kinetics of progenitor cells or viable CECs in peripheral blood should be explored as an early indicator of tumor response to anti-VEGF agents.

Six weeks after completion of the bevacizumab, radiation therapy and chemotherapy regimen, follow-up PET scans (Fig. 1c) showed decreased tumor FDG uptake compared with pretreatment values in five patients (Fig. 2g). Tumor FDG uptake in patient 5 was low, and similar before and at the end of therapy (Fig. 2g). Notably, all six patients completed the combined treatment without dose-limiting toxicity, and underwent surgery without perioperative or postoperative complications. Macroscopic and histologic analysis of the surgical specimens revealed a marked response in all six patients, with only microscopic disease in five of the patients (Fig. 1a and Supplementary Note).

High doses of bevacizumab are more effective when used as monotherapy for highly VEGF-dependent tumors such as renal-cell carcinoma¹⁴. Our data from six consecutive patients show for the first time that even at low doses, bevacizumab alone can decrease perfusion, MVD and IFP in a solid tumor, and decrease the number of CECs. The decrease in IFP and increase in the fraction of vessels with pericyte coverage support the normalization hypothesis⁸. This normalization process may retard the shedding of metastatic cells in the circulation and improve the delivery of therapeutic agents in tumors. Bevacizumab may also sensitize the endothelium to cytotoxic agents. Collectively, these mechanisms may explain the unprecedented efficacy of bevacizumab in recent clinical trials, as well as the possible synergistic or additive interaction between antiangiogenic and cytotoxic therapies that has been observed in preclinical settings for more than a decade^4,15. The identification of valid surrogate markers for antiangiogenic therapy, alone or combined with cytotoxic therapies, has been elusive. The results of this phase 1 study will hopefully facilitate and stimulate future research in this area.