|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to investigate two non-invasive methods for determining the treatment efficacy of the vascular disrupting agent (VDA) CA4P: gadolinium enhanced dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) for perfusion analysis and ELISA of blood samples. Candidate proteins were identified by Multi-Analyte Profile analysis of plasma from KHT sarcoma-bearing C3H/HeJ mice after CA4P administration. Candidate proteins were further analyzed by ELISA of plasma from treated C3H/HeJ, BALBc, and C57BL6 mice. Changes in selected proteins, tumour perfusion and tumour necrotic fraction after CA4P treatment were then compared in individual animals. The cytokines KC and MCP-1 were observed to increase after CA4P treatment in all tested models. No correlation was found between KC or MCP-1 levels and tumour necrosis. However, tumour perfusion correlated (r=0.89, p<0.00001) with CA4P treatment efficacy as measured by necrotic fraction, suggesting DCE-MRI may have utility in a clinical setting.
Due to the limited diffusion of oxygen and nutrients through tissue, tumours are dependent upon their ability to stimulate angiogenesis (the induction of new blood vessels) in order to grow beyond a size of ~1 mm3 1,2. Tumours also must continuously induce angiogenesis as they grow to compensate for the increasing number of cells the vessels must support, resulting in a level of endothelial proliferation not normally found in the adult tissues3. The resulting tumour vessels are abnormal in structure, often leaky, tortuous, and lacking normal pericyte interaction4–7. The presence of these abnormal vessels, which are unique to the tumour, has led to the development of therapeutic strategies aimed at compromising the existing tumour blood vessels3.
One class, the vascular disrupting agents (VDAs), which includes Combretastatin A4 Phosphate (CA4P), OXi4503, and ZD6126, targets proliferating endothelial cells by binding tubulin subunits and preventing polymerization. This results in the depolymerization of microtubules and reorganization of the actin cytoskeleton8 in dividing endothelial cells. Treatment of solid tumours with these VDAs has been shown to lead to tumour vessel occlusion and significantly decreased tumour vascular density9–11. Although the precise mechanism of this vessel disruption is not completely understood, the endothelial cell shape changes caused by the reorganization of the cytoskeleton and disruption of VE-cadherin signalling are believed to play a key role 12,13. As a result of this tumour vessel disruption, blood flow is significantly reduced to the tumour, and the tumour cells die due to loss of oxygen and nutrient supply3,9,14,15.
After VDA treatment, a thin layer of viable tissue survives at the tumour periphery, probably because the cells in this region of the tumour are supported by the normal vasculature which is not disrupted by VDA treatment16. These surviving tumour cells can repopulate the tumour, thus limiting the use of these VDAs as a single agent therapy17–19. However, these agents effectively destroy regions of the tumour normally resistant to more conventional cancer therapies, raising considerable interest in VDAs as adjuvants to current therapeutic regimens10,20,21.
Pre-clinically, VDA efficacy is typically evaluated by analyzing tumour necrotic fraction following treatment. As this method is not practical in the clinical setting, the present study investigated two alternative techniques for monitoring VDA treatment efficacy that could be applied to the clinical setting: blood analysis and gadolinium dynamic contrast enhanced magnetic resonance imaging (DCE-MRI). Proteins involved in the regulation of tissue damage repair released into the plasma as a result of extensive necrosis and tissue damage induced by VDA treatment would be easily examined in a clinical setting by a simple blood test. Potentially, the increase in such proteins might correlate with the extent of tumour damage resulting from VDA treatment. Alternatively, DCE-MRI could be used to visualize the significant reduction in tumour perfusion as a result of VDA treatment. The added advantage of this latter technique is that it has been used on patients receiving VDAs in the clinical setting 9,11,22. However, to date, there is little evidence to connect a specific change in tumour perfusion, as observed by MRI, to a defined treatment outcome.
In the present study, the microtubule targeting agent CA4P was administered to KHT tumour-bearing mice to identify plasma protein candidates. These candidates were then compared in the individual animal with the laboratory standard marker of treatment efficacy, tumour necrotic fraction, as well as with tumour perfusion as measured by gadolinium DCE-MRI to examine the possibility of correlation with treatment efficacy. The design of this study, therefore, allowed for a direct comparison between plasma proteins, change in tumour perfusion, and treatment outcome following CA4P treatment.
All research was governed by the principles of the Guide for the Care and Use of Laboratory Animals and approved by the University of Florida Institutional Animal Care and Use Committee (IACUC). C3H/HeJ, BALBc, and C57BL6 mice were obtained from Jackson Laboratories, Bar Harbor, ME. Mice were six to eight weeks old and were maintained under a specific pathogen free environment (University of Florida Health Science Center) with food and water ad libitum.
CA4P (OXiGENE Inc, US) was prepared in saline and injected intraperitonealy (IP) at 0.01ml/g of body weight at a dose of 100 mg/kg. Control animals were injected with saline.
Plasma samples were collected from the tail vein of tumour bearing C3H/HeJ mice over a 12 hour period after treatment with 100 mg/kg CA4P. The plasma from two mice was pooled for each time point (n=14). These samples were then analyzed at Charles River Laboratories (Austin, TX) by rodent Multi-Analyte Profiles (MAPs) (designed by Rules Based Medicine, Inc Austin, TX). Each sample was screened for 80 proteins including cancer markers, cytokines, hormones, and those involved in infectious disease.
Candidate proteins identified by the broad screen were further evaluated by ELISA (R&D systems, Minneapolis, MN). Plasma samples from tumour-bearing and non-tumour bearing C3H/HeJ, as well as non-tumour bearing BALBc and C57BL6 mice treated with 100 mg/kg CA4P (n=9) was examined. Blood samples were obtained from the tail vein before and four hours after VDA administration. To investigate dose response, plasma samples were obtained from KHT tumour bearing C3H/HeJ mice four hours after administration of various doses of CA4P (12.5 to 200 mg/kg) or saline.
Pre-treatment plasma samples (2.5 above) and tumour perfusion measurements (2.7 below) were obtained from 30 C3H/HeJ mice bearing KHT sarcomas. These data served as a baseline. The following day, the same mice were treated with various doses of CA4P (5 to 100 mg/kg). Four hours after treatment, plasma samples were again obtained, and tumour perfusion measurements were repeated one hour later (5 hours post treatment). Plasma samples were analyzed by ELISA. Twenty-four hours after CA4P administration, the tumours were removed for analysis of tumour necrosis (2.8 below).
The procedures for relative tumour perfusion as well as the evaluation of the dynamic contrast enhanced (DCE) MRI data sets have been described in detail25. A brief summary of the methods are given below.
Relative tumour perfusion measurements were determined in tumour-bearing mice using DCE MRI. Contrast-enhanced MRI measurements were made before and 5 hours after a single dose treatment of 100 mg/kg CA4P. Images were acquired using a 4.7 T magnet (Oxford Instruments, horizontal boar of 25 cm diameter). The mice were anesthetized with 2% Isoflurane via induction chamber and maintained with 1.25 % Isoflurane via face mask (reducing Isoflurane concentration by 0.25% for every 20 min). A flow of warmed air was used to maintain the body temperature of the animals while in the magnet. Spin-echo images for measurement of relative tumour perfusion were T1-weighted (TE = 12 ms, TR = 130 ms; field of view 20 × 20 mm; 128 phase-encode increments and 256 data points, zero-filled to 256 × 256; 8 to 14 slices at a slice thickness 1 mm).
Maps of the initial rate of inflow of gadolinium into the tumours were generated by dividing the signal intensity (SI) of the five post contract images acquired 3, 6, 9, 12 and 15 min later, by the SI of the pre contrast image. Changes in relative tumour perfusion were assessed quantitatively by plotting the mean signal intensity in a region of interest defined within central three tumour slices (whole section, excluding skin) against time after bolus injection of 0.2 mmol/kg contrast agent (Omniscan). Relative tumour perfusion was then determined by measuring the integrating area under the signal intensity/time curve.
Twenty-four hours after VDA treatment, tumours were removed and fixed in 10% formalin for 24 hours, sectioned and haematoxylin and eosin stained (Molecular Pathology Core, University of Florida). Three sections from each tumour were then imaged by tile field mapping using a morphometric microscope (McKnight Brain Institute, University of Florida). The images were analyzed for necrotic fraction by measuring the percent necrotic area over the whole tumour with the analysis software ImageJ. The percent necrosis for each tumour was then determined by averaging the three sections.
Paired two tailed t tests were performed to establish significance of differences between pre-treatment and post-treatment protein levels using t test analysis. The highest P values are reported. For comparisons between multiple groups, ANOVA and post hoc Scheffe analysis was performed with α set to 0.05. t test, ANOVA, and Scheffe analysis were conducted using Microsoft Excel. Correlations were determined by Pearson’s Product Moment Correlation Coefficient using Sigmaplot 2001.
At various times after CA4P treatment, plasma was collected from tumour-bearing C3H/HeJ mice and proteins were identified through MAP analysis. Compared to untreated mice the levels of four proteins (MCP-1, MIP-1β, KC, IL6) were found to be > two times higher in plasma from mice examined four hours after CA4P treatment (Figure 1). These four cytokines are proteins involved in immune response and wound repair. IL6 is a regulatory protein that is involved in several immune and repair processes including fever control, monocyte infiltration, and reepithelialization 26. MCP-1, or macrophage chemoattractant protein, and MIP-1β, macrophage inflammatory protein 1beta, recruit monocytes or macrophages to the site of injury of infection 26,27. KC is also a chemokine, but is involved in the activation and recruitment of neutrophils 28.
Subsequent studies used ELISA to confirm plasma protein levels in tumour-bearing C3H/HeJ mice before and after CA4P treatment. Figure 2 illustrates the observed plasma concentrations in mice treated with a dose of 100 mg/kg CA4P, which indicates a reproducible increase in cytokine concentration after treatment. Mice treated with saline alone showed no significant change (data not shown). To determine the dose dependency of the observed effects, tumour bearing mice were treated with a range of doses of CA4P and plasma cytokine levels were determined by ELISA. The results showed a dose dependent increase in plasma protein concentration for each of the four tested cytokines (Figure 3).
Non-tumour-bearing C3H/HeJ mice also were treated with CA4P and analyzed for plasma cytokine changes (Figure 4). Such mice were found to express plasma cytokine changes comparable to those obtained in tumour-bearing mice (Figure 2 versus 4).
To examine the mouse strain specificity of these effects, non-tumour-bearing BALBc and C57BL6 mice treated with CA4P also were tested. Only two of the candidate proteins, MCP-1 and KC, increased with CA4P treatment in all tested strains (Table 1). As the goal of this study was to identify markers they are likely to be able to be translated into human testing, the proteins IL6 and MIP-1β were excluded from further study as they failed to demonstrate a consistent response across the tested strains.
Necrotic fraction (an established endpoint for VDA treatment efficacy) and change in cytokines and tumour perfusion were determined for each animal following CA4P exposure. These parameters were then compared for possible correlations. Using this approach, neither KC nor MCP-1 levels correlated with necrotic fraction (Figure 6). Cytokine changes also did not relate to changes in tumour perfusion as measured by DCE-MRI (Figure 6). However, when changes in tumour perfusion were compared to the tumour necrotic fraction, a significant correlation emerged (Pearson’s correlation coefficient 0.89, p<0.00001) (Figure 7). Larger changes in tumour perfusion corresponded with larger necrotic fraction values, while smaller perfusion changes resulted in lesser necrotic fraction values following CA4P treatment.
VDAs offer a potentially attractive therapy for application in combination with conventional cancer treatments due to their effective destruction of microenvironmental regions of the tumour typically resistant to conventional anticancer therapies. Several clinical trials combining VDAs with chemotherapy or radiotherapy21 are currently underway. Pre-clinical evaluations of VDA treatment have typically assessed the percentage of induced tumour necrosis as a measure of their efficacy, but such evaluations are not readily applicable to the clinical setting. The aim of this study, therefore, was to evaluate two alternative potential markers for monitoring VDA treatment efficacy; plasma proteins as measured by ELISA and tumour perfusion as determined by gadolinium DCE-MRI. These potential markers were all analyzed in the same animal after CA4P treatment, allowing for direct comparisons between the surrogate markers and treatment outcome as determined by tumour necrotic fraction.
The results showed a significant cytokine response to CA4P treatment. The proteins KC, MIP-1β IL6, and MCP-1, were observed to consistently increase in the plasma of mice four hours after treatment. A dose dependent increase in cytokine concentration was observed with increasing CA4P treatment, indicating a specific response to this VDA. Non-tumour-bearing mice also displayed the cytokine increase in response to CA4P suggesting that the observed cytokine elevation reflected a host response rather than being a result of tumour damage. Interestingly, two of the four cytokines, IL6 and KC, are chemokines involved in neutrophil activation; a finding that may relate to the increase in neutrophils that has been observed in solid tumours following VDA treatment29. Parenthetically, increased neutrophil infiltration also has been reported in tumours of patients undergoing CA4P treatment (Chaplin, D., personal communication, 2005).
Despite the lack of a direct tumour specific effect, it may be argued that if the observed cytokine increases occurred in a manner paralleling the extents of CA4P induced tumour damage, changes in host cytokine production might still be useful as a surrogate marker of treatment efficacy. Such an outcome would suggest that cytokine measurements in plasma of patients undergoing CA4P therapy might offer important information to regarding CA4P dosing. Unfortunately this was not born out in the experimental studies. Indeed increases in plasma concentration of either KC or MCP-1, the cytokines found to be elevated in all three mouse strains following CA4P exposure, failed to correlate with treatment outcome as determined by tumour necrotic fraction measurement (Figure 5).
Somewhat surprising in these studies was the absence of an elevation of plasma VEGF in the initial broad screen analysis of KHT tumour-bearing mice following CA4P administration. This result was in contrast to a recent study demonstrated significant VEGF increases following treatment of MHEC5-T tumours30 with a related VDA (OXi4503). However, immunohistochemical analysis of KHT tumours also failed to demonstrate significant changes in VEGF expression following CA4P treatment (data not shown). Still, the focus on the four chosen proteins in the present study does not preclude the variations in plasma concentrations of other cytokines nor does the broad screen MAP analysis necessarily indicate their non-involvement in either the tumour or host response to VDA treatment.
Post treatment increases in the evaluated cytokines have not previously been reported for the tubulin binding VDAs. Indeed previous in vitro evaluations of cytokine levels in cell lines following VDA administration failed to identify a significant effect of combretastatin VDAs on cytokine expression10. The in vivo results reported here suggest that the cytokine responses reflect a direct host reaction to CA4P exposure. Further studies investigating the source of the cytokine production and the aspects of VDA action that could induce this normal tissue response are clearly warranted.
Results of the MRI experiments reveal a significant correlation between tumour perfusion and necrotic fraction following VDA treatment, supporting the findings of the previously published study relating these parameters 25. However, unlike the previously published study which compared the response of groups of animals, the current study demonstrates a relationship between tumour perfusion and tumour necrosis in individual mice.
In summary, while changes in KC and MCP-1 cytokine concentrations were observed across three mouse strains, no direct correlation could be established between cytokine concentration changes and either tumour necrotic fraction or DCE MRI based tumour perfusion. In contrast, CA4P-induced tumour necrosis assessments and DCE MRI based tumour perfusion measurements did show a significant correlation, supporting the continued use of this non-invasive imaging technique for the monitoring of patient treatment efficacy after CA4P administration. Given the availability of MRI in most hospitals, the application of this approach in clinical evaluations of VDA therapy is highly feasible.
This work was supported by USPHS grant CA-84408. The authors gratefully acknowledge the Optical Microscopy Facility and the Advanced Magnetic Resonance Imaging and Spectroscopy Facility, Evelyn F. & William L. McKnight Brain Institute of the University Of Florida, Gainesville, Florida in the performance of these studies.
Work supported by USPHS grant CA-89655.
CA4P provided by OXiGENE, Waltham, MA 02451, USA.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest Statement
D.W. Siemann serves on the scientific advisory board of OXiGENE, Inc.