Combretastatin-A4 disrupts neovascular development in non-neoplastic tissue

Combretastatin-A4 phosphate (cis -CA-4) is a tubulin-binding agent currently undergoing clinical trials as an anti-tumour drug. We have investigated whether CA-4 functions as a tumour-specific anti-vascular agent using the hyperplastic thyroid as a novel in vivo model of neovascularization. CA-4 elicited pathological changes in normal tissue, manifested as the induction of multiple, discrete intravascular thrombi. These vascular-damaging effects indicate that CA-4P does not function as a tumour-specific agent but targets neovasculature irrespective of the primary angiogenic stimulus. © 2001 Cancer Research Campaign http://www.bjcancer.com

The prerequisite of vascularization to permit tumour development beyond a threshold size has focused attention on the therapeutic potential of anti-angiogenic and anti-vascular agents (Hanahan and Folkman, 1996;Hayes et al, 2000). Cis-CA-4, a tubulinbinding agent currently undergoing Phase I clinical trials, causes vascular disruption in some experimental primary and orthotopic tumours, and in vascularized metastatic tumour deposits (Dark et al, 1997;Beauregard et al, 1998;Grosios et al, 1999;reviewed in Griggs et al, 2001). Furthermore, cis-CA-4 inhibits metastasis of Lewis lung carcinoma. The trans-CA-4 isomer is without effect on either ectopic primary or metastatic Lewis lung carcinoma development (Griggs et al, manuscript submitted). Cis-CA-4 differentially affects quiescent and tumour vasculature, as indicated by the greater magnitude of transient changes induced in vascular pressure and resistance in tumours compared with those in normal tissues (Tozer et al, 1999). The mechanism of action of cis-CA-4 remains to be elucidated but it has been hypothesized that it manifests a selective toxicity for tumour endothelial cells (Dark et al, 1997), because their proliferative activity renders them more susceptible to the drug than the quiescent vasculature which predominates in the adult. The anti-vascular activities of cis-  have been reported to be specific for tumour vasculature. Here we examined this hypothesis by investigating whether proliferating endothelial cells in general are susceptible to CA-4P-mediated damage by studying its effects in rapidly proliferating, non-tumour tissue using chemical induction of goitre in mice.
A number of goitrogens can block thyroid hormone production leading to a rise in thyroid stimulating hormone (TSH) which promotes growth of thyroid follicular cells. This induces a coordinated programme of neovascularization in which endothelial cell proliferation is regulated by pro-angiogenic growth factors, for example vascular endothelial growth factor (VEGF), derived from follicular cells (Bidey et al, 1999). Following goitrogen treatment, thyroid growth is induced in rodents within 2 days and continues at a high rate for about 2 weeks. The weight of the gland approaches a plateau after about 4 weeks (Wynford-Thomas et al, 1982a;Peter et al, 1991). During the phase of active growth the mitotic activity in endothelial cells parallels that of epithelial cells (Wynford-Thomas et al, 1982b) and there is a very large increase in blood flow through the gland. We have utilized this system to compare the effects of cis-CA-4 and trans-CA-4 on normal and hyperplastic (goitrogen-treated) mouse thyroid.

Combretastatin synthesis and administration
Both isomers of di-sodium phosphate CA-4 pro-drug were synthesized as previously described (Pettit et al., 1995Bedford et al, 1996;Orsini et al, 1997;, dissolved in saline and filter sterilized. Goitrogen-treated and non-goitrogentreated groups of mice were given either cis-CA-4 phosphate, trans-CA-4 phosphate or no further treatment. CA-4 was administered by intraperitoneal injection at a dose of 12.5 mg kg Ϫ1 every 12 hours from the initiation of goitrogen treatment for 10 days, at which time all mice were killed by CO 2 asphyxiation. The treatment schedule used was selected based on previous studies (Griggs et al, manuscript submitted) in which a chronic treatment regime of 12.5 mg kg Ϫ1 administered every 12 h by intraperitoneal injection is highly efficacious in the inhibition of metastatic development of Lewis lung carcinoma and manifests no detectable side effects or toxicity in C57B16/J mice.

Immunohistochemistry
Thyroid glands attached to the trachea were removed, fixed in neutral buffered formalin (Sigma) for 24 h and embedded in paraffin. 4 µm thick sections were cut at 10 levels through each gland, taking 10 sections at each level. For each thyroid one section from each level was stained with von Willebrand factor (Dako) using an avidin-biotin-horse radish peroxidase technique with diaminobenzidine as the reporter molecule. All sections were screened microscopically at 100 × magnification by two observers without knowledge of the treatment groups to which they belonged.

RESULTS
All goitrogen-treated groups showed diffuse thyroid enlargement with hypertrophic follicular cells, loss of colloid and reduction in size of follicular lumen, together with loss of the normal central and peripheral zonation ( Figure 1B). All non-goitrogen-treated groups showed peripheral follicles with flattened epithelial cells, abundant colloid and more active central follicles ( Figure 1A).
All animals in the cis-CA-4 groups given goitrogen showed multiple microthrombi either in small veins or capillaries within the thyroid ( Figure 1E-H). No microthrombi were seen in extrathyroid vessels or in a range of other normal tissues from the same animals (data not shown). In a few cases the microthrombi in capillaries had expanded to bulge into the lumen of an adjacent follicle, covered by a thin layer of stretched follicular cells. The incidence of microthrombi was quantified by counting the number of microthrombi detected in sections from 10 levels from each gland. The identification of microthrombi in stained sections was confirmed with von Willebrand factor antibody, which identified endothelial cells and also showed that the microthrombi contained the factor ( Figure 1E, G, H). The abundant level of von Willebrand factor in the microthrombi may have originated from activated platelets in addition to damaged endothelial cells. In untreated goitre, endothelial cells were clearly identifiable by von Willebrand staining but were frequently undetectable at the periphery of microthrombi in cis-CA-4-treated goitres. Microthrombi were identified in the thyroids of all 8 mice treated with cis-CA-4, and were detected in 49 of 74 sections (66%). The mean number of microthrombi per goitre was 19.0 in cis-CA-4P-treated animals ( Figure 2). None of the 8 animals treated with trans-CA-4 showed any microthrombi ( Figure 1D) or any other change in any of the 74 sections, apart from those attributable to the goitrogen treatment. One of the 7 animals receiving goitrogen alone showed 3 microthrombi in 2 levels, no microthrombi were detected in the remaining 63 sections from this treatment group. We presume that this low incidence of spontaneously arising vascular disruption in untreated goitre arises from the hyperproliferative nature of the tissue. No other abnormalities were seen.
No microthrombi were seen in any of the mice not given goitrogen (Figure 2) whether treated with cis-CA-4 ( Figure 1C) or trans-CA-4 (data not shown) or given no treatment ( Figure 1A).
All showed the expected normal gland morphology with no anomalous features.

DISCUSSION
The data indicate that cis-CA-4 induces a high frequency of microthrombus formation in normal thyroid tissue that is undergoing a high rate of proliferation with associated neovascularization. Thyroid hyperplasia was induced by administration of a goitrogen: in the absence of this treatment cis-CA-4 had no detectable effects. The trans isomer of CA-4 was without effect on either normal thyroid tissue or in developing goitres, and neither isomer affected other normal tissues. The thyroid responses to CA-4 were consistent with the reported effects of the 2 isomers on primary and metastatic tumours in mice. Cis-CA-4 retards primary tumour growth by a process that involves severe vascular damage and it is also a potent inhibitor of metastatic development. The trans-CA-4 isomer, in contrast, does not inhibit either primary or secondary tumour development. Cis-CA-4 is a tubulinbinding agent but, unlike other agents with a similar action, for example, colchicine and vincristine, its action in vivo appears to be specifically against endothelial cells. In this study, the damage to the endothelial cells manifested itself as microthrombi in venules and capillaries.
Despite the frequency of microthrombus induction by cis-CA-4, there was no necrosis in any of the glands and no detectable change in the degree of hyperplasia in the follicular cells. The thyroid has a network of capillaries around the follicles and several arteries and veins which interconnect through this network. Thrombosis in all the major veins supplying the gland is therefore likely to be necessary before major changes are seen in the gland itself. In a rapidly growing tumour with new vessels growing into the expanding neoplasm, thrombosis in the newly formed vessels penetrating the tumour may well be more effective in causing necrosis or at least in slowing growth. We have no reason to suppose that vascular growth in the thyroid due to TSH stimulation differs in any fundamental aspect from angiogenesis in other tissues. During goitrogenesis vascular growth is coordinated, the thyroid endothelial cells responding to VEGF, as do endothelial cells of many other tissues, with VEGF being produced locally by follicular cells in response to TSH stimulation (Sato et al, 1995;Viglietto et al, 1997;Wang et al, 1998;Bidey et al, 1999). There is no evidence that the endothelial cells respond directly to TSH. Vascular growth in goitres is therefore considered to occur in response to stimulation from epithelial cells in a way that closely corresponds to angiogenesis in tumours induced by cancer cells. Goitrogenesis is a highly reproducible experimental system for inducing vascular growth and the stimulated thyroid gland has one of the highest rates of blood flow of any tissue. We have shown that the thyroid provides a reproducible in vivo model system for testing and quantifying the effects of putative anti-angiogenic or anti-vascular agents in non-tumour tissue.
The demonstration that cis-CA-4 induces widespread microthrombus formation in the rapidly growing vasculature of the stimulated thyroid indicates that its action is specific not for tumour vasculature but for growing endothelium. It follows that the drug may have considerable side effects if any non-neoplastic vascular proliferation is taking place, for example in inflammation. We note that the action of cis-CA-4 on hyperplastic, nonneoplastic vasculature suggests that this drug may be effective against angioproliferative diseases.
Preliminary evidence from Phase I clinical trials (reviewed in Griggs et al, 2001) suggests that CA-4 exerts an anti-vascular effect in some human primary tumours. Consistent with rapidly proliferating endothelium being the primary target of CA-4, we would speculate that there will be variability in the response of different tumour types to CA-4 according to the rates of vascular proliferation, a parameter which shows significant heterogeneity between tumours.