Effects of Angiogenesis Inhibitors on Multistage Carcinogenesis in Mice

See allHide authors and affiliations

Science  30 Apr 1999:
Vol. 284, Issue 5415, pp. 808-812
DOI: 10.1126/science.284.5415.808


Solid tumors depend on angiogenesis for their growth. In a transgenic mouse model of pancreatic islet cell carcinogenesis (RIP1-Tag2), an angiogenic switch occurs in premalignant lesions, and angiogenesis persists during progression to expansive solid tumors and invasive carcinomas. RIP1-Tag2 mice were treated so as to compare the effects of four angiogenesis inhibitors at three distinct stages of disease progression. AGM-1470, angiostatin, BB-94, and endostatin each produced distinct efficacy profiles in trials aimed at preventing the angiogenic switch in premalignant lesions, intervening in the rapid expansion of small tumors, or inducing the regression of large end-stage cancers. Thus, anti-angiogenic drugs may prove most efficacious when they are targeted to specific stages of cancer.

Over the past decade, genetically engineered mouse models of cancer have increasingly been used in studies on mechanisms of carcinogenesis. One strength of these models is that cancers arise from normal cells in their natural tissue microenvironments and progress through multiple stages, as does human cancer (1). Such models of organ-specific cancer also present opportunities for development not only of cancer therapies but also of preventative strategies that block the progression of premalignant lesions into tumors.

The RIP1-Tag2 transgenic mouse model of pancreatic islet carcinogenesis serves as a general prototype of the pathways, parameters, and molecular mechanisms of multistage tumorigenesis. The pathway through which normal insulin-producing β cells of the pancreatic islets are converted into islet cell carcinomas under the influence of the SV40 T antigen (Tag) is increasingly well understood (2,3). Briefly, in the RIP1-Tag2 line, the following stages arise sequentially over the 13.5-week average lifetime of these mice. Normal islets (initially 100% of the ∼400 islets per pancreas) express the Tag oncogene and yet are morphologically asymptomatic until 3 to 4 weeks of age. Hyperplastic islets then begin to appear stochastically (increasing to 50% of the islets by 10 weeks), displaying β-cell hyperproliferation and features of dysplasia and carcinoma in situ. Angiogenic islets (8 to 12%) arise from hyperplastic/dysplastic islets by switching on angiogenesis in the normally quiescent islet capillaries (3). This switch is characterized by endothelial proliferation, vascular dilation, and microhemorrhaging. Solid tumors (∼3% of the islets) emerge at 10 weeks as small encapsulated tumors (adenomas) that progress by 12 to 13 weeks into large adenomas and (less frequently) invasive carcinomas, both of which are intensely vascularized by dilated hemorrhagic vessels.

In earlier work, we showed that a mixture of three angiogenesis inhibitors (AGM-1470, minocycline, and alpha interferon) could inhibit tumor growth in the RIP-Tag model (4). These data supported other preclinical data from subcutaneous transplant models (5–7). Encouraged by the result, we designed the present study to assess the effects of four diverse angiogenesis inhibitors when applied to RIP1-Tag2 mice at three distinct stages of pancreatic islet carcinogenesis. Figure 1illustrates the experimental design, which had three branches: (i) early treatment at the hyperplastic stage to block the angiogenic switch before the initial formation of solid tumors (prevention trial), (ii) treatment of mice bearing small (asymptomatic) solid tumors to determine whether their expansive growth and progression to deleterious stages could be stopped (intervention trial), and (iii) treatment of mice with substantial tumor burden and near death to ascertain whether these agents could induce tumor regression (regression trial). We evaluated four angiogenesis inhibitors (8) (Tables 1 and 2): AGM-1470 (TNP470), BB-94 (batimastat), angiostatin, and endostatin, as well as the combination of angiostatin and endostatin. AGM-1470 (TNP470) is a small molecule inhibitor of endothelial cell proliferation (9) that is thought to act by inhibiting an intracellular enzyme, methionylaminopeptidase-2 (9). BB-94 (batimastat) is a broad-spectrum inhibitor of matrix metalloproteinases (10), which are involved in the remodeling of extracellular matrix and capillary basement membranes during angiogenesis (11). Angiostatin and endostatin are cleavage products of larger cellular proteins. Angiostatin is an internal fragment of plasminogen, whereas endostatin is a COOH-terminal fragment of collagen XVIII, itself a component of the blood vessel wall (6, 7). In this study, we used murine versions of angiostatin and endostatin that were produced as fusions to the Fc fragment of the murine immunoglobulin heavy chain (Tables 1 and 2) (12).

Figure 1

Anti-angiogenic trials designed to target discrete stages of carcinogenesis. The RIP1-Tag2 mice develop pancreatic insulinomas and islet cell carcinomas in a multistep pathway that includes an angiogenic switch before solid tumor formation. In the prevention trials, inhibitors were tested for their ability to block a hallmark of tumor development: the onset of angiogenesis. Thus, 5-week-old transgenic mice, whose islets had not yet activated the angiogenic switch, were treated until the first tumors appeared in control mice at 10.5 weeks of age. For intervention studies, designed to address whether angiogenesis inhibitors can slow or stop tumor growth, 10-week-old transgenics bearing small tumors were treated until end-stage disease (13.5 weeks of age). In the regression trials, inhibitors were tested for their ability to extend life-span by inducing regression of large tumors. Thus, treatment was initiated in 12-week-old transgenics harboring substantial tumor burden and with a life expectancy of less than 2 weeks. Cross at right indicates the death of the animal.

Table 1

Prevention trial. The average number ±SD of angiogenic islets of control (PBS) and treated mice and the reduction in the number of angiogenic islets (%) are shown. The reduction in angiogenic islet incidence in treated (T) versus control (C) mice was calculated by the formula (1 – T/C) × 100. Statistical analysis was done with a two-tailed, unpaired Mann-Whitney test comparing experimental groups to (phosphate-buffered saline (PBS)–injected control mice. P values less than 0.1 are considered statistically significant. “N start” indicates the number of animals at the beginning of the trial; “N stop” indicates the number of animals at the end of the trial; and “N trials” indicates the number of independent experiments. The endostatin fragment of collagen XVIII and the angiostatin fragment of plasminogen were each fused at their respective NH2-termini to the constant region of immunoglobulin G (Fc), the hybrid genes were expressed in murine myeloma cells, and the fusion proteins were isolated as described (12). BB-94 was homogenized in 0.02% Triton and PBS (pH 7.0) and given as an emulsion. AGM-1470 was dissolved in 5% dextrose water. Mice were treated with 25 mg per kilogram of body weight (25 mg/kg) of AGM-1470 s.c. every other day; or with 30 mg/kg of BB-94 i.p. every day; or with 15 mg/kg of Fc-endostatin s.c. every day, and/or with 20 mg/kg Fc-angiostatin s.c. every day. The dose of 15 mg/kg of Fc-endostatin was the maximal tolerable dose because of an inflammatory side effect produced by the fusion protein, which resulted in swollen faces and hair loss (or white hair on the head) but did not provoke an evident specific immune response in the form of T and B cell infiltration in the pancreas (15). Neither the Fc fragment alone nor endostatin without the Fc fragment elicited such allergic reactions.

View this table:
Table 2

The average tumor burden ±SD of control (PBS) and treated animals and the reduction in tumor burden (%) in the treated animals are reported in the intervention and regression trial. The percent reduction in tumor burden was calculated from the average tumor burden of the treated groups (T) and the average tumor burden of the 13.5-week-old PBS-treated RIP1-Tag2 group by applying the formula (1 – T/C) × 100. Statistical analysis was performed with a two-tailed, unpaired MannWhitney test comparing experimental groups to PBS-injected control mice.P values less than 0.1 are considered statistically significant. Tumor burdens of experimental groups in the regression trial were compared to the tumor burden of 12-week-old Rip1-Tag2 animals. “N start” indicates the number of animals at the beginning of the trial; “N stop” indicates the number of animals at the end of the trial; and “Ntrials” indicates the number of independent experiments.

View this table:

In the prevention trial, 5-week-old transgenic mice harboring only hyperplastic nonangiogenic islets were treated for 5.5 weeks, to the point at which tumors first began to appear in sham-treated controls (Table 1) (13). At this age (10.5 weeks), 20 to 25% of the hyperplastic islet nodules (∼45 islets) had become angiogenic (3), which allowed the frequency of the angiogenic switch to be quantitated before tumor burden grossly disrupted the pancreatic architecture. AGM-1470 had no statistically significant effect, and angiostatin had only a modest impact on the number of angiogenic islets, as compared to saline treatment. In contrast, BB94 reduced the incidence of angiogenic switching by ∼50%, whereas endostatin alone and the combination of endostatin plus angiostatin resulted in 61 and 63% reductions, respectively.

In the intervention trial, treatment of the RIP1-Tag2 mice began at 10 weeks of age, when every mouse had a few small but highly vascularized solid tumors, and ended at 13.5 weeks, when sham-treated mice had end-stage disease (13). We assessed tumor burden rather than life-span extension, in order to fully evaluate the tumor size and histological characteristics such as apoptotic index and vessel density. As shown in Table 2, all four compounds were efficacious. Fc-angiostatin was least effective, reducing tumor burden by ∼60%. AGM-1470, BB-94, and Fc-endostatin reduced tumor growth by 82, 83, and 88%, respectively. The Fc-angiostatin plus Fc-endostatin combination was not tested here because of a limited supply of Fc-angiostatin.

In the regression trial, we treated cohorts of 12-week-old tumor-bearing mice that had a life expectancy of <2 weeks. We assessed life-span extension as well as tumor burden. The control mice were moribund by 13.5 weeks and were killed. All of the angiogenesis inhibitors extended life-span by at least 2 weeks. Because a subset of mice in some of the treatment groups were moribund at 16 weeks, the trial was terminated at that time to allow comparison of tumor burden and histological characteristics (Table 2 and Fig. 2). Both AGM1470 and the combination of Fc-angiostatin plus Fc-endostatin caused significant regression of tumor burden. In contrast, Fc-endostatin alone, Fc-angiostatin alone, and BB-94 alone only slowed tumor growth. At 16 weeks, the tumor burden of mice treated with these compounds was similar to that of sham-treated mice at 13.5 weeks and was increased from the 12-week starting point. Thus, in the regression trial, significant differences in efficacy were observed among the tested compounds.

Figure 2

Illustrative histopathology of the regression trial. Morphology [hematoxylin and eosin (H/E) staining] and combined endothelial and apoptotic cell labeling (anti-CD31+TUNEL) of tumors are shown. Tumor-bearing pancreases were collected from 13.5-week-old control RIP1-Tag2 mice and from 16-week-old transgenic mice treated with AGM-1470 or a combination of Fc-angiostatin and Fc-endostatin. Pancreases were then fixed, paraffin-embedded, and sectioned at 5 μm. Yellow arrows indicate the invasive fronts of carcinomas, and yellow arrowheads indicate apoptotic cells proximal to blood vessels. Scale bar indicates 44.6 μm in the H/E and CD31/TUNEL panels and 22.3 μm in the CD31/TUNEL (2×) panels.

In all three trials, we performed histological analyses (14) to compare either premalignant lesions or tumors from treated and control mice. In the prevention trial, the angiogenic islets in treated mice did not show obvious histological differences from those in control mice, in terms of vessel density, hemorrhage formation, apoptotic index, or morphological appearance (15). The treated tumors in the intervention and regression trials had several common characteristics, as noted previously (4, 16). Angiogenesis inhibitor treatment did not induce a significant change in the number of proliferating tumor cells, it caused only a slight reduction in vessel density, and it increased the number of apoptotic cells by a factor of 2 to 3. The histopathology of the regression trial is shown for AGM-1470 and the Fc-endostatin plus Fc-angiostatin combination (Fig. 2). The observation that blood vessel density was not decreased (Fig. 2) supports the notion that some tumor vessels can grow, or at least remain intact, in the face of treatment with angiogenesis inhibitors; these vessels apparently dictate the number of tumor cells that can survive, thus governing tumor size while maintaining a similar density of tumor vessels. Although it was not surprising that apoptosis was induced in endothelial cells treated with the anti-angiogenic compounds, it was intriguing that tumor cells in close apposition to capillaries were also frequently apoptotic. One might have expected the opposite: that oxygen and nutrient deprivation resulting from an impaired vasculature would preferentially affect hypoxic tumor cells most distal to capillaries; the periendothelial apoptotic pattern suggests that the apoptotic induction mechanism involves signals distinct from hypoxia.

There was an appreciable incidence (35 to 52%) of invasive islet cell carcinomas among the small tumors present at 16 weeks in all treatment groups (Fig. 2); they were most pronounced (52%) in mice treated with Fc-endostatin or Fc-angiostatin plus Fc-endostatin. In contrast, multiple large encapsulated adenomas predominated at the 12-week starting point of the regression trial, where only ∼20% of tumors were invasive carcinomas. Thus, these angiogenesis inhibitors did not prevent histological progression to the invasive carcinoma stage. Metastases were not observed in the 16-week-old treated mice with prevalent carcinomas; however, metastasis is rare in control RIP1-Tag2 mice (2, 16).

Our data reveal that the four angiogenesis inhibitors tested had different efficacies depending on the stage of carcinogenesis being targeted (Fig. 3). BB-94, Fc-endostatin, the Fc-angiostatin plus Fc-endostatin combination, and to a lesser extent Fc-angiostatin, but not AGM-1470, prevented activation of the angiogenic switch in up to 63% in the hyperplastic islets that otherwise would have switched. Treatment of RIP1-Tag2 mice bearing substantial tumors with AGM-1470 or Fc-endostatin plus Fc-angiostatin produced significant tumor regression, in contrast to Fc-endostatin, Fc-angiostatin, or BB-94 alone. The former agents reduced tumor volume by ∼60% (as compared to tumor volume at the 12-week starting point) and by ∼72% (as compared to tumor burden in untreated mice near end stage at 13.5 weeks). Given that the time course of each trial was limited, none of the agents tested completely prevented the angiogenic switch, blocked the growth of small tumors, or completely resolved lethal tumor burden. However, it is important to recognize that the RIP1-Tag transgene targets oncogene expression to every one of the ∼500,000 pancreatic β cells in these transgenic mice, resulting in a temporal spectrum of developing neoplastic foci and thus rendering this a very stringent system for assaying pharmacological interventions. Although multifocal disease may present a tough standard in comparison to standard tumor transplant models, such stringency is likely to prove of considerable value, given the biological and clinical resilience of human cancers.

Figure 3

Summary of stage-specific efficacy differences among angiogenic inhibitors. The relative effects of particular angiogenesis inhibitor treatments in the three distinct stages of carcinogenesis are compared here. IT stands for initial tumor burden and reflects the mean tumor burden at the beginning of the intervention trial (10 weeks; IT10w) or regression trial (12 weeks; IT12w). The prevention trial was started at 5 weeks of age, when only hyperplastic/dysplastic islet nodules are present (c in the top panel stands for the number of angiogenic islets in the control animals).

This study strengthens the proposition that angiogenesis inhibitors will become important components of anticancer treatment strategies, by using compounds and dosing regimens fine-tuned to target specific stages of disease progression. A number of future applications for this model system can be envisioned: identifying the most potent formulations, dosing regimens, and combinations of the current compounds as well as of new angiogenesis inhibitors, both for prevention and for therapeutics; exploring broader combinatorial possibilities involving other classes of anticancer drugs; and elucidating the pharmacogenetic mechanisms of the stage-specific efficacy differences. The challenge will be to validate the results from this and other mouse models of organ-specific carcinogenesis as instructive in the design of clinical trials to identify drugs and delivery strategies that are effective for the treatment and prevention of human cancers.

  • * To whom correspondence should be addressed. E-mail: dh{at}


View Abstract

Stay Connected to Science

Navigate This Article