MOLECULAR PHARMACODYNAMICS OF CHEMOTHERAPY: FIBROBLAST GROWTH FACTOR (FGF) INHIBITORS AS CHEMOSENSITIZERS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the

Graduate School of The Ohio State University

By

Colin T. Walsh, B.S.

*****

The Ohio State University 2005

Dissertation Committee: Approved by

Professor Jessie L.-S. Au, Adviser

Professor John J. Lannutti ______

Professor Thomas D. Schmittgen Adviser College of Pharmacy Professor M. Guillaume Wientjes ABSTRACT

Au and Wientjes et al. recently reported that acidic (aFGF) and basic (bFGF) fibroblast growth factors confer a broad spectrum chemoresistance in solid tumors, and that suramin, a non-specific FGF inhibitor, enhanced the in vitro anti-tumor activity of several anticancer drugs. Based on this initial finding, the studies proposed in this dissertation are focused on improving the understanding of the mechanisms of the FGF- induced resistance and the molecular pharmacodynamics of suramin. Studies in Chapter 2 show suramin can enhance the therapeutic efficacy of chemotherapy in lung cancer, thus establishing the in vivo efficacy of low dose suramin. Studies in Chapter 3 show bFGF is a clinically significant predictor of chemotherapy and suramin effect. Many literature reports show suramin as having anti-angiogenic properties, therefore the possibly of an anti-angiogenic mechanism for suramin was investigated in both in vitro (Chapter 4) and in vivo (Chapter 5) models. Results from Chapter 4 show low doses of suramin alone had no effect compared to control and suramin in combination with chemotherapy had no additional effects as compared to chemotherapy alone in a monolayer endothelial cell model. This finding was extended to an in vitro tumor histoculture model and the results show the suramin effects were vasculature independent. Results from Chapter 5 show that suramin alone had no effect on the in vivo tumor vessel morphology, and suramin did not show any additional effects when combined with chemotherapy. The functionality of

ii the tumor vessels was also tested and the results show that chemotherapy greatly

increased the functionality of the tumor vasculature; however there were no additional

effects from suramin. Collectively, these studies show suramin is capable of sensitizing

tumors to chemotherapy, bFGF is a valid biomarker for chemotherapy and suramin

sensitization effect, and that the in vivo mechanism of suramin chemosensitization is not due to anti-angiogenesis.

iii

Dedicated to my wife, for her tireless encouragement and support throughout the graduate process.

iv ACKNOWLEDGMENTS

My sincerest thanks and admiration goes to Dr. Jessie L.-S. Au for her inspiration,

guidance, kindness, financial support, and most importantly her patience during my

graduate studies. Dr. Au inspires her students to reach for excellence and I think I

benefited greatly from my time spent under her tutelage. I would like to thank Dr. M.

Guillaume Wientjes for his encouragement, scientific input, and help with statistical data

analysis, and all the other members of my committee, including Dr. Thomas D.

Schmittgen and Dr. John J. Lannutti, for their comments and suggestions, and to all of

my teachers without whose help this would not have been possible. I would also like to

thank my other mentors and collaborators from the Au/Wientjes lab. Drs. SaeHeum Song

and Yuebo Gan for their encouragement and advice on the projects outlined in Chapters 2

and 3 respectively, Dr. Yong Wei, for his encouragement, advice, and stimulating

discussions of my experimental findings, and finally to the remaining members of

Au/Wientjes laboratory especially Dr. Ze Lu, Dr. Adam Odgen, Dr. Ron Ortiz, Dr. Fan

Yang, Dr. Leijun Hu, Dr. Greg Lyness, Yan Xin, and Bei Yu for all their scientific and

technical support. I would also like to thank my friend and collague, Dr. Liang Zhao

whose encouragement, keen insights, and friendship helped me along the road to

graduation. I appreciate all the help I received from Brian Kemmenoe and the people at

the Campus Microscopy and Image Facility at The Ohio State University. Last but not

v least, I would like to thank my wife for her unwavering support and confidence and my mother, father, mother-in-law, father-in-law, and the rest of my family who never once doubted my ability. This work was supported in part by a fellowship from the American

Foundation for Pharmaceutical Education (AFPE). I am honored and grateful to have received the AFPE fellowship.

vi VITA

June 13, 1974 Born - Ann Arbor, MI

1993 - 1999 B.S. Chemical Engineering, University of Kentucky

1996 - 1997 Process Engineer, Cooperative Education Program International Specialty Products, Calvert City, KY

1999 - Present Graduate Teaching and Research Associate, The Ohio State University

PUBLICATIONS

1. SaeHeum Song, M. Guillaume Wientjes, Colin Walsh, and Jessie L.-S. Au. Nontoxic Doses of Suramin Enhance Activity of Paclitaxel against Lung Metastases. Cancer Research. 2001 Aug 15;61(16):6145-50.

PAPERS PRESENTED AT SCIENTIFIC MEETINGS

1. Colin Walsh, SaeHeum Song, M. Guillaume Wientjes, and Jessie L.-S. Au. Suramin enhances the antitumor activity of paclitaxel in mice bearing lung metastasis of human prostate tumors. Proc. Am. Assoc. Cancer Res., 42: 2001.

2. Colin Walsh, Liang Zhao, Bei Yu, Yong Wei, SaeHeum Song, M. Guillaume Wientjes, and Jessie L.-S. Au. Chemosensitization effect of suramin is not due to anti- angiogenesis. Proc. Am. Assoc. Cancer Res., 44: 2003.

FIELDS OF STUDY

Major Field: Pharmacy

vii TABLE OF CONTENTS

Page

Abstract ...... ii

Dedication ...... iv

Acknowledgements ...... v

Vita ...... vii

List of Tables ...... xii

List of Figures ...... xiii

Chapters:

1. Introduction and Backgorund Information ...... 1

1.1 Introduction ...... 1 1.2 The FGF family of growth factors ...... 3 1.2.1 Biological properties and activities ...... 3 1.2.2 bFGF expression and clinical prognosis ...... 4 1.2.3 FGFs and chemoresistance ...... 6 1.2.4 The role of FGFs in angiogenesis ...... 6 1.3 Suramin ...... 9 1.3.1 History ...... 9 1.3.2 Pre-clinical and clinical studies using suramin ...... 9 1.3.3 Potential pharmacologic targets of suramin ...... 10 1.3.4 Reported anti-angiogenic effects of suramin ...... 11 1.4 Angiogenesis ...... 13 1.4.1 Normal angiogenesis ...... 13 1.4.2 Tumor angiogenesis ...... 16 1.4.3 Anti-angiogenic effects of chemotherapy ...... 17

viii 1.5 Image analysis ...... 18 1.6 Overview of dissertation ...... 21

2. Low doses of suramin enhance the activity of paclitaxel in vivo ...... 24

2.1 Introduction ...... 24 2.2 Materials and methods ...... 26 2.2.1 Chemicals and reagents ...... 26 2.2.2 Cell and tumor cultures ...... 27 2.2.3 Drug solutions ...... 27 2.2.4 Animal and drug treatment protocols ...... 27 2.2.5 Histological evaluation of tumors ...... 28 2.3 Results ...... 29 2.3.1 Synergy between paclitaxel and suramin in vivo ...... 29 2.4 Discussion ...... 29 2.5 Acknowledgments ...... 31

3. Expression of basic fibroblast growth factor correlates with resistance to paclitaxel and suramin chemosensitization in human tumors ...... 34

3.1 Introduction ...... 34 3.2 Materials and methods ...... 37 3.2.1 Chemicals and supplies ...... 37 3.2.2 Procurement of tumor specimens ...... 38 3.2.3 Pharmacologic effects of chemotherapy ...... 38 3.2.4 Immunohistochemistry ...... 39 3.2.5 Pathologic scoring of the immunohistochemical result in the paclitaxel treated samples ...... 40 3.2.6 Development of the bFGF standard curve ...... 41 3.2.6.1 Cell culture ...... 41 3.2.6.2 Animal protocol for subcutaneous tumor model . 42 3.2.6.3 Evaluation of bFGF levels in tumor lysates . . . . 43 3.2.6.4 Evaluation of the inter-day and intra-day variability of the microscope light source ...... 43 3.2.6.5 Image capture and processing ...... 44 3.2.7 Quantification of bFGF level in the immuno- histochemically stained samples ...... 44 3.2.8 Statistical analysis ...... 45 3.3 Results ...... 45 3.3.1 Quantitative image analysis methodology ...... 45 3.3.2 Correlation between paclitaxel effects and bFGF expression levels in the 96 patient tumor set ...... 47

ix 3.3.3 Relationship between bFGF expression level, 5-fluorouracil effect, and suramin chemosensitization in the RCC clinical samples ...... 51 3.4 Discussion ...... 53 3.5 Acknowledgments ...... 57

4. In vitro investigation of the anti-angiogenic mechanism of suramin ...... 71

4.1 Introduction ...... 71 4.2 Materials and methods ...... 73 4.2.1 Chemicals and reagents ...... 73 4.2.2 Cell culture ...... 74 4.2.3 Drug and reagent stock solutions ...... 74 4.2.4 Drug treatment and activity evaluation in the HUVEC monolayer model ...... 75 4.2.5 Evaluation of the interaction between paclitaxel and suramin in the HUVEC model ...... 76 4.2.6 Histoculture drug response assay ...... 77 4.2.7 Immunohistochemistry ...... 79 4.2.8 Development of an image analysis method to quantify the morphologic changes in the vessel morphology . . . . 80 4.2.9 Statistical analysis ...... 82 4.3 Results ...... 82 4.3.1 Effects of suramin and paclitaxel in the HUVEC monolayer model ...... 82 4.3.2 Chemosensitization in PC3 histocultures ...... 83 4.3.3 No anti-angiogenic effects by paclitaxel or suramin in the PC3 tumor histocultures ...... 84 4.4 Discussion ...... 84 4.5 Acknowledgments ...... 88

5. In Vivo investigation of the anti-angiogenic mechanism of suramin ...... 96

5.1 Introduction ...... 96 5.2 Materials and methods ...... 99 5.2.1 Chemicals and reagents ...... 99 5.2.2 Drug dosing solution preparation ...... 100 5.2.3 Cell culture ...... 101 5.2.4 Animal protocol for xenograft tumor model ...... 101 5.2.5 Immunohistochemical staining of the tumor vasculature . . 103 5.2.6 Development of a quantitative image analysis method to quantify the morphologic and functional changes in the tumor vasculature ...... 104

x 5.2.7 Statistical analysis ...... 106 5.3 Results ...... 106 5.3.1 No Anti-angiogenic effects by suramin alone or in combination with chemotherapy in tumor-bearing animals as measured by vessel morphology ...... 106 5.3.2 Chemotherapy increases the functionality of the tumor vessels and this effect is independent of the suramin effect...... 107 5.4 Discussion ...... 109 5.5 Acknowledgments ...... 111

6. General discussion and perspectives ...... 118

List of references ...... 121

xi LIST OF TABLES

Table 2.1. Enhancement of in vivo antitumor effect of paclitaxel by suramin ...... 32

Table 3.1. aFGF and bFGF staining results in the paclitaxel treated patient samples as analyzed by the pathologic scoring method ...... 58

Table 3.2. Correlation between pathobiological parameters and paclitaxel activity as measured by IC30 ...... 59

Table 3.3. Correlation between pathobiological parameters and paclitaxel activity as measured by EMAX ...... 60

Table 3.4. Correlation between pathobiological parameters and paclitaxel activity as measured by apoptotic fraction ...... 61

Table 3.5. Correlation between bFGF level as measured by quantitative image analysis versus paclitaxel effect within the tumor subgroups . . . 62

Table 4.1. Pharmacologic effects of paclitaxel and suramin in the HUVEC model ...... 89

Table 4.2. Combination Index values (CI) at the IC50 effect level for paclitaxel and suramin in HUVEC model ...... 90

Table 4.3. Summary of morphologic vessel data obtained using image analysis for the PC3 histoculture model ...... 91

Table 5.1. Summary of functional vessel data obtained using image analysis for the HT29 subcutaneous tumor model...... 112

xii LIST OF FIGURES

Figure 2.1. Enhancement of in vivo antitumor activity of paclitaxel by suramin ...... 33

Figure 3.1. Hue Saturation and Intensity histograms ...... 63

Figure 3.2. Representative images of bFGF staining from the standard tumor set ...... 64

Figure 3.3. Non-linear fit of the bFGF standard curve ...... 65

Figure 3.4. Linear fit of the bFGF standard curve ...... 66

Figure 3.5. Representative images of the aFGF staining from the paclitaxel treated patient samples ...... 67

Figure 3.6. Representative images of the bFGF staining from the paclitaxel treated patient samples (A) and results from the quantatitive image analysis (B) ...... 68

Figure 3.7. Plots of paclitaxel effect versus bFGF level for the three tumor subgroups ...... 69

Figure 3.8. aFGF and bFGF in the RCC patient samples (A) and the RCC Histoculture Samples (B) and the correlation between chemosensitivity and the bFGF level in the RCC histoculture control samples (C) ...... 70

Figure 4.1. Dose response for suramin in the HUVEC model ...... 92

Figure 4.2. Cytotoxicity of the combination of paclitaxel and suramin . . . . . 93

Figure 4.3. Chemosensitization in the PC3 histocultures by low dose suramin ...... 94

Figure 4.4. Immunohistochemical staining of CD-31 in the PC3 histoculture control samples ...... 95

xiii

Figure 5.1. Chemosensitization by low dose suramin in PC3 xenograft tumors ...... 113

Figure 5.2. Chemosensitization by low dose suramin in HT29 xenograft tumors ...... 114

Figure 5.3. Sample Images of the immunohistochemical staining (A) and summary of morphologic vessel data obtained using image analysis for the PC3 subcutaneous tumor model (B) ...... 115

Figure 5.4. Sample Images of the immunohistochemical staining (A) and summary of morphologic vessel data obtained using image analysis for the HT29 subcutaneous tumor model (B) ...... 116

Figure 5.5. Examples of the functional vessel staining (A) and the total vessel staining (B) ...... 117

xiv CHAPTER 1

INTRODUCTION AND BACKGROUND INFORMATION

1.1 INTRODUCTION

The clinical treatment of cancer requires knowledge of the pharmacodynamic

relationship between the drug concentration and the desired therapeutic effect. Preclinical

pharmacokinetic and in vitro data can guide drug development as to the appropriate dose

and schedule but to accurately validate the effect the pharmacodynamic relationship

between the drug and the target must be studied. From 1996 to 2000, 209 new drugs,

aiming at about 20 new targets (including growth factors, signal transduction pathways, and angiogenesis) entered clinical evaluation. Among these 209 drugs, 28 showed sufficient efficacy to enter and complete randomized phase III trials [1].

This trend in drug discovery illustrates the need for more quantitative methods for the validation new and existing drug targets at their respective pharmacodynamic endpoints. This will increase the number of successful compounds entering the clinic.

With this in mind, this dissertation is focused on improving the understanding of the mechanisms of the FGF-induced resistance and the molecular pharmacodynamics of the non-specific fibroblast growth actor inhibitor suramin.

The importance of fibroblast growth factors, and their role in chemoresistance, was recently established by Au and Wientjes et al. when they showed that aFGF and

1 bFGF could induce resistance to anticancer drugs with diverse structures and action mechanisms [2]. They showed that bFGF was required to induce resistance, whereas aFGF amplified the bFGF effect. Looking for a way to inhibit this resistance factor, Au and Wientjes et al. investigated suramin, a non-specific bFGF inhibitor, as a possible chemosensitizer. They found that the chemosensitization effect of suramin is biphasic in nature. In vitro the chemosensitizing effect was only observed at nontoxic/subtherapeutic concentrations of 50 µM, while higher concentrations > 100 µM had no effect

(monolayer) or had an antagonistic effect (tumor histoculture). Further in vitro studies showed that the addition of specific bFGF inhibitors (i.e., anti-bFGF monoclonal antibodies) to chemotherapy/suramin combination did not enhance the effect of chemotherapy, indicating that suramin and the antibody shared the same action mechanism [2].

The ability of bFGF to induce resistance, and for suramin to reverse this resistance has been proven in vitro, however there is still a need to test this chemosensitization hypothesis in vivo. Since bFGF is the target for suramin, characterization of the relationship between bFGF and resistance in the patient setting will help to establish the use of low dose suramin in the clinic. Furthermore the role of bFGF in chemoresistance is even more important since it is also our pharmacodynamic target for suramin therapy, thus necessitating an even greater need for quantification of bFGF’s role in resistance. Finally, in order to give suramin in the most effective manner in the clinic, the mechanism of action must be investigated. Understanding the

2 mechanism of suramin chemosensitization will lead to discovery of new targets and will

allow optimization of the therapeutic scheduling thus resulting in better patient response and survival with fewer side effects.

The goal of this dissertation is focused on improving the understanding of the

mechanisms of the FGF-induced resistance and the molecular pharmacodynamics of the non-specific fibroblast growth actor inhibitor suramin for the purpose of clinical translation. This goal was accomplished by three separate studies. The first showed that the in vivo mechanism of low dose suramin parallels the in vitro mechanism of bFGF inhibition. The second validated bFGF as a clinical predictor for paclitaxel resistance.

This study required the development of a novel image analysis methodology to quantify the immunohistochemical result from the patient samples in real protein amounts. Finally there is a great deal of literature showing that suramin is anti-angiogenic, therefore the last study investigated the possibility of an anti-angiogenic mechanism of suramin. The results show anti-angiogenesis not the main factor in the chemosensitization effect of suramin, thus adding more weight to the original bFGF-induced resistance hypothesis.

1.2 THE FGF FAMILY OF GROWTH FACTORS

1.2.1 Biological properties and activities.

FGFs constitute a large family of 24 growth factors that are present in the intracellular and extracellular environment of endothelial cells and blood vessels, and in parenchymal cells in neural, skeletal and reproductive tissues, and major organs. All of

FGF family members share a highly homologous central core of 28 highly conserved and

6 identical amino acids sequences [3;4]. Ten of these highly conserved sequences are responsible for the interaction between FGF and its receptor, FGFR [4;5]. Another

3 commonality among FGF family members is that they all have a high binding affinity to heparin and heparin sulfate proteoglycans [4;6;7]. It is thought that complexing with

heparin protects FGF from enzymatic or thermal degradation and aids in receptor

binding. Also heparin associated proteoglycans can bind and store FGF within the extracellular matrix [8].

The FGF-2 (also called bFGF) gene encodes multiple proteins of different molecular weight (18, 22, 23, 24 kD) isoforms through the use of a different translation start codon. The 22-24 kD proteins (CUG-mediated initiation) contain a nuclear localization signal and are actively transported into the nucleus of the cell [9]. The function of the higher molecular weight FGFs are unknown [8;10]. The 18 kD isoform does not contain a seceratory signal and remains in the cytosol, however bFGF is routinely found extracellularly. It is unknown exactly how the 18kDa bFGF leaves the cell, either through active secretion or via cell injury. However, it is clear that bFGF functions extracellularly as a soluble factor where it can enhance tumor cell proliferation, survival, motility, angiogenesis, and enhanced carcinogenesis, tumor progression, malignancy, and metastasis [6-8;10-15].

1.2.2 bFGF expression and clinical prognosis.

Compared to healthy individuals, cancer patients show significantly higher serum bFGF levels, e.g., prostate cancer (6.6 vs. 1.3 pg/ml) [16], cervical cancer (31 vs. 4.8 pg/ml) [17], leukemia (48 vs. 5.4 pg/ml) [18], renal cell carcinoma (10.9 vs. 8.3 pg/ml)

[19]. Increased FGF levels in urine have also been observed in patients with a wide variety of solid tumors, lymphoma, or leukemia; the range of median bFGF levels for the

patients with different cancers is 164 to 1311 pg/g creatine, and 37 % of the cancer

4 patients show high bFGF levels compared with normal controls [20]. Further reports have shown that elevated serum levels or increased systematic and/or local tissue bFGF level are associated with worse prognosis and shorter survival in many types of human solid tumors including, leukemia and lymphoma [20-22], NSCLC [23;24], SCLC [25;26], colorectal cancer (reviewed in [14]), renal cell carcinoma [27], advanced carcinoma of head and neck [28], gastric cancer [29;30], non-Hodgkin's lymphoma [31;32], esophageal carcinomas [33], thyroid carcinomas [34], malignant solitary fibrous tumor [35], small adenocarcinomas [36], mesothelioma [37], and Wilms' tumor [38].

The role of bFGF expression in clinical prognosis is still unclear. In pancreatic cancer, although bFGF level is not correlated with postoperative recurrence and survival, increased FGF receptor expression is associated with shorter survival [39]. A similar observation was made in NSCLC patients [40]. There are also reports indicating an opposite relationship in patients for some cancer types. One study showed increased local bFGF expression was associated with shorter survival in nodal-negative breast cancer

[41]. While another study in 1307 primary breast cancer patients showed increased bFGF is related with better prognostic and longer survival [42]. This later result is supported by several other studies in breast cancer showing similar trends [43-46] and the same contradictory bFGF relationship has also been reported in ovarian cancer [47] and in pediatric high-grade gliomas [48]. This contradictory data clearly suggests the need for further study.

5 1.2.3 FGFs and chemoresistance.

Several studies, focusing on bFGF, have shown support for bFGFs role in

chemoresistance. However, the role of aFGF in chemoresistance was only demonstrated

recently by Au and Wientjes et al. [2]. Studies from other laboratories showing bFGFs

role in chemoresistance show that transfection into epithelial cells causes amplification of

several genes in purine and pyrimidine biosynthesis pathways and results in resistance to

antimetabolites (i.e., methotrexate, hydroxyurea, and PALA) [49;50], and resistance to

cisplatin and radiation [51-53]. A high intracellular bFGF level is correlated with

resistance to fludarabine in patients with chronic lymphocytic leukemia [54].

Previous studies have shown that extracellular bFGF induces chemoresistance and radioresistance in solid tumor and leukemia cells [51-53;55], but the bFGF concentration

required to induce resistance far exceeds the concentration in patient plasma and urine

samples (20-100 ng/ml vs. <1 ng/ml, see reference [2]), thus raising questions on the

clinical relevance of this mechanism. However, the Au/Wientjes laboratory has since

demonstrated, in prostate tumor cells, that aFGF amplifies the bFGF effect by about 50-

fold such that combinations of aFGF and bFGF, at clinically relevant concentrations,

induce up to 10-fold resistance to several anticancer drugs [2].

1.2.4 The role of FGFs in angiogenesis.

bFGF was one of the first tumor-derived factors known to stimulate endothelial

cell proliferation and induce vascularization in vivo. It was first isolated by Folkman et al.

in 1971 but the protein remained uncharacterized until 1984 at the advent of heparin

affinity chromatography [56;57]. Acidic and basic FGF are both potent inducers of

angiogenesis and have show the ability to stimulate endothelial cell proliferation,

6 migration, enzyme production, and to induce angiogenesis in vitro and in vivo [58-61].

FGFs are also known to synergize with VEGF in the induction of angiogenesis, probably

by up-regulating VEGF and VEGF receptors in endothelial cells [62-65].

Unlike VEGF whose involvement in tumor angiogenesis has been demonstrated

repeatedly [66], there is little experimental evidence for a direct involvement of FGFs in

the regulation of tumor angiogenesis. VEGF is mainly mitogenic for endothelium while

FGFs are highly pleiotropic and can stimulate cell division in a number of different cell

types including tumor cells, endothelial cells, smooth muscle, and fibroblasts. Another

impediment for the acceptance of the role of bFGF as a primary factor in tumor

angiogenesis is the fact that bFGF lacks a secretion signal while many of the other

angiogenic growth factors possess an export signal [8;67]. Furthermore FGFs are stored

in the extracellular matrix which suggests a secondary role since one of the first steps in

the angiogenic cascade is to degrade the matrix thus releasing FGF.

Evaluation of the role FGFs play in angiogenesis is seen in the literature from studies showing the use of antibodies against FGF and the resulting effects in vivo. For example, Matsuzaki et al. found that specific monoclonal antibodies against bFGF could inhibit endothelial cell proliferation in vitro but had no effect on angiogenesis in vivo

[68]. Dennis et al. also found that anti-bFGF antibodies inhibited the endothelial cell migration in vitro, but failed to produce anti-angiogenesis or reduce the size and growth of three tumor models in animals [69]. Compagni et al. performed a similar study using a soluble form of FGFR2 IIIb and Flt (VEGFR-1). It should be noted that FGFR2 IIIb is a high affinity receptor for aFGF but not bFGF (see Table 1 in [70]). The soluble FGFR was able to suppress tumor growth and decreased tumor vessel density [71], however the

7 soluble FGFR was not effective as a soluble VEGFR construct. Furthermore the soluble

FGFR impeded tumor growth at a later time point after tumor induction than did the

soluble VEGF receptor, suggesting that VEGF acts to initiate tumor angiogenesis while

FGF may be more important for maintenance of the vasculature [71]. This concept of

vessel maintenance is supported by the fact that bFGF knockout mice are phenotypically

indistinguishable from FGF2 +/+ littermates, except for neuronal defects and delayed

wound healing [72].

Clinically the suspected angiogenic role of bFGF persists. Some of the literature

reports discussed in section 1.2.2 include correlative studies between angiogenic potential

(typically measured by microvessel density) and intratumoral bFGF levels. For example,

Smith et al. found that intratumoral bFGF levels were not an independent predictor of

survival in breast cancer while Pazgal et al. found serum levels of bFGF correlated with

worse progression-free and overall survival in non-Hodgkin’s lymphoma and finally

Strizzi et al. found that high serum bFGF levels correlated with decreased survival in

malignant pleural mesothelioma. These different studies however were unified in that all

three did not find a correlation between bFGF level and tumor angiogenesis as measured

by microvessel density [31;37;45].

It is not obvious that bFGF inhibitors will produce an anti-angiogenic effect and

thereby an antitumor effect in vivo and there is intense research interest on using anti-

angiogenic agents to enhance the efficacy of chemotherapy. Many anti-angiogenic

candidate compounds are now in clinical testing, however only one, the anti-VEGFR

antibody AvastinTM, has demonstrated effect in patients [73]. Many other anti-angiogenic compounds have failed to demonstrate clinical benefit. The clinical program for the anti-

8 angiogenic compound SU5416 was closed because the compound failed to produce a

survival advantage in colon cancer patients [74]. Hence, the anti-angiogenic approach

remains an open question, thus making it of paramount importance to distinguish the

FGF/FGFR-targeting approach from anti-angiogenesis approach for the purpose of directing future therapy development efforts.

1.3 SURAMIN

1.3.1 History.

Suramin, a polysulfonated naphthylurea, was first synthesized in the early 1900’s and used as an antiparasitic agent. However, the antitumor properties of suramin were not discovered until the early 1980’s when suramin was under investigation as a reverse transcriptase inhibitor for the treatment of HIV/AIDS. Studies showed that suramin could inhibit multiple growth factors, which sparked considerable interest and effort in developing suramin as an antitumor agent [75]. To be effective as an antitumor agent, suramin must be present at very high plasma concentrations (100-200 µM), at which suramin is highly toxic.

1.3.2 Pre-clinical and clinical studies using suramin.

Preclinical studies demonstrated the antitumor activity of suramin in mice bearing adrenal and prostate cancer, and osteosarcoma [76-78]. The suramin doses in these studies were between 100-260 mg/kg per week (plasma concentrations >100 µM) and mainly resulted in tumor growth delay and not tumor regression. Furthermore, suramin had no discernable antitumor activity at 50 mg/kg per injection and was highly toxic at

100 mg/kg, in mice [78].

9 Clinically, suramin has shown some activity in prostate cancer and has undergone evaluations in a wide variety of solid tumors, either as single agent or in combination with other chemotherapeutics. At least 33 trials have been reported [79-92]. In all of these trials, suramin was used as a cytotoxic agent at therapeutic plasma concentrations of 100 to 200 µM. At these concentrations, suramin showed significant toxicities, modest activity in patients, and combination therapy with high dose suramin did not show a benefit over monotherapy. Based on these findings, multiple investigators recommended against the future use of high dose suramin [81-87;89;91].

1.3.3 Potential pharmacologic targets of suramin.

Suramin has many other pharmacological activities that could be responsible for the established chemosensitization effect [61;75;88;93-110]. Suramin can inhibit DNA polymerase α, reverse transcriptase, binding of IL-2 and transferrin to their receptors, phosphorylation of PKC, glycosaminoglycan metabolism, Na/K-ATPase, tumor necrosis factor (TNF-α), and topoisomerase II. It is well documented that suramin can bind and inhibit many different growth factors (i.e., platelet-derived growth factor, FGFs, transforming growth factor-β, epidermal growth factor, vascular endothelial growth factor, and insulin-like growth factor-1) thus preventing these growth factors from binding to their respective receptors. As a specific example, Manetti et al. describes how suramin prevents the interaction of bFGF with its receptor by attaching its naphthylsulfonic moiety to both the heparin and receptor binding sites of bFGF [104]. It is this effect that is of most interest to the current hypothesis and the reason why many studies have investigated suramin both as an anti-tumor and anti-angiogenic agent.

10 1.3.4 Reported anti-angiogenic effects of suramin.

The anti-angiogenic effects of suramin have been extensively studied, both in

vitro and in vivo using a wide variety of models derived from both normal and tumor

tissues. In vitro studies concerning suramin effect on several processes of the

angiogenesis cascade (i.e., inhibition of bFGF binding to endothelial cell receptor,

inhibition of VEGF signaling, VEGFR expression, endothelial cell migration,

proliferation, and protease activity) are best reviewed by Takano and Waltenberger et al.

Studies by Takano et al. on bovine capillary endothelial cells showed that the suramin

effect is concentration-dependent. Inhibition of bFGF binding to endothelial cells was

achieved at suramin concentrations below 60 µM, whereas inhibition of endothelial cell

migration occurred at >150 µM and inhibition of cell proliferation and protease activity

required suramin concentrations of >170 µM [61].

Waltenberger et al. studied suramin effects on VEGF stimulated angiogenesis,

using a porcine endothelial cell model that over express the VEGF receptor KDR. They

show that 2 µM suramin was able to inhibit KDR phosphorylation by 50% and > 200 µM suramin completely abolished KDR phosphorylation and decreased its expression level.

Suramin appeared to have a biphasic effect on endothelial cell proliferation. Cells stimulated by 3 ng/ml VEGF showed that 0.2 µM suramin could further enhance proliferation while > 100 µM suramin was able to reduce endothelial cell proliferation to a level below baseline. Endothelial cell migration using 10 ng/ml VEGF as a stimulus showed that VEGF could increase cell migration by 7 fold and 200 µM suramin could

11 inhibit this migration by 85%. However, 200 µM suramin could not completely reverse

the VEGF stimulated migration and the 200 µM suramin treated cells showed migratory

rates 2 fold greater then the 0 ng VEGF control [110].

The anti-angiogenic effects of suramin are well documented in many ex vivo

normal tissue models. Gagliardi et al shows that 50 µg and 200 µg suramin were able to

inhibit angiogenesis by 46% and 79% respectively in the chorioallantoic membrane

(CAM) assay [59]. Suramin (50 µg/ml; ~35 µM) can significantly inhibit endothelial

sprouting in tissue explants from rat aorta [58]. This finding was confirmed by Stiffey-

Wilusz et al. who showed 50 µg/ml suramin had the same effect in tissue explants from porcine carotid arteries [111].

The literature details several studies showing in vivo activity of suramin in several normal tissue derived models. Suramin at 30 mg/kg per day was able to significantly inhibit neo-angiogenesis in the rat mesentery as induced by compound 40/80.

Furthermore suramin (1.6 mg/eye per day) was able to suppress neo-angiogenesis in the rat cornea induced by chemical injury [58]. The cornea results were confirmed in a study by Takano et al. showing suramin (200 mg/kg suramin given i.v. 48 hours prior) could significantly inhibit neo-angiogenesis cause by the implantation of 100 ng of bFGF into the cornea [61].

The anti-angiogenic effects of suramin in several in vivo tumor derived models show that higher doses of suramin (200 mg/kg, single treatment) were required to produce an anti-angiogenic effect in mice bearing M5076 murine reticulosarcoma

12 xenografts [60]. This effect was only observed when (200 mg/kg) dose was administered prior to tumor/vasculature establishment. This later finding has relevance in the clinical application of suramin since the vasculature of human solid tumors is well established.

Results from Au and Wientjes et al. show that the chemosensitization by 10-20

µM suramin was observed in monolayer cultures of tumor cells, a system that does not involve angiogenesis [2]. Hence, anti-angiogenesis is not the mechanism of in vitro chemosensitization. Furthermore, the addition of bFGF monoclonal antibody did not enhance the suramin effect [2]. This indicates suramin shares the same mechanism as the antibody by inhibiting bFGF. The current hypothesis is that bFGF inhibition is the main mechanism responsible for the chemosensitization effect of low-dose suramin. However, suramin has multiple targets (presented above and reviewed in the next section), therefore the possibility of an anti-angiogenic mechanism of suramin is worthy of study.

1.4 ANGIOGENESIS

1.4.1 Normal angiogenesis.

The adult vasculature is derived from a rudimentary network of vessels that is first created in the embryo by a process called vasculogenesis. In this process the first vessels are formed from angioblasts which are endothelial cell precursors [112]. These angioblasts proliferate and coalesce to what is known as the primary capillary plexus, which is a primitive network of vessels that supplies the growing embryonic tissues.

After the primary capillary plexus is formed, it is remodeled by the sprouting and branching of new vessels from preexisting ones in the process of angiogenesis [113].

13 Unlike in the embryo, normal adult vessels are mostly quiescent and very little endothelial cell turnover occurs in normal adult vessels. Normal angiogenesis only occurs in adult tissues during the ovarian cycle and in normal physiological repair processes

(wound healing). However, when angiogenesis does occur, it is accomplished by the coordination of many diverse processes and is best described as a highly ordered step wise process that is under tight regulation. The first step in the angiogenic cascade is the removal of any accompanying mural cells (pericytes) followed by the destabilization of the cell-cell contacts between the endothelial cells. Destabilization is mainly accomplished by angiopoietin-2 (Ang2) which displaces the Ang1 from its Tie2 receptor.

Angiopoietin-1 (Ang1) is a powerful vessel stabilizer that keeps the vessel walls intact and mitogenically resistant. This is a critical step because endothelial cells are very resistant to mitogenic stimulus when in stable endothelial sheets, but by breaking the cell- cell contacts the endothelium adopts a more plastic and proliferative phenotype [113].

Next, the vasculature is made hyperpermeable by VEGF so local extravasation of proteases and matrix components from the bloodstream can occur. These proteases and blood components aid in the degradation and remodeling of the basement membrane and extracellular matrix surrounding the vessel. A family of enzymes known as the matrix metalloproteinases (specifically MMP-2, MMP-3, and MMP-9) play a central role in the matrix degradation and the soluble factor TGF-β has been shown to play a key role in matrix remodeling. A byproduct of this step is the liberation of many growth factors, including bFGF, VEGF, and insulin-like growth factor-1 (IGF-1), which are normally sequestered in the extravascular matrix [113]. Many of these soluble growth factors

(especially, bFGF, VEGF, and EGF) help the existing endothelial cells proliferate and

14 migrate into the newly remodeled matrix space. Cell migration is directed in part by signaling through αvβ3 and α5β1 integrins, PECAM-1 (or CD-31), and the eph/ephrin receptor-ligand pairs [114].

After sufficient cell division has occurred and the cells have migrated into position, the endothelial cells begin to reinitiate cell-cell contacts and form rudimentary tubules and begin to acquire a lumen. The list of molecules aiding in this step include both membrane bound (VE cadherin, eph/ephrin, αvβ3 and α5β1 integrins, PECAM-1) and soluble (bFGF, PDGF, TNF-α) mediators [114]. As the vessels form, angiopoietin-1

(Ang1) signaling thought the Tie2 receptor shifts the endothelial cells back into their normal motigenicly resistant phenotype. Next the basement membrane is reestablished and mural cells (pericytes in the microvasculature, smooth muscle cells in larger vessels) are recruited to the abluminal surface of the endothelium and blood flow is reestablished in the vessel [113;114].

Interestingly, the biomechanical forces (pressure and shear stress) generated by the blood flow have a profound effect on the growth of the vasculature. Lack of flow (i.e., non-perfused vessels) regress apparently through endothelial cell apoptosis [113;114], whereas vessels with established blood flow persist. Laminar blood flow stabilizes and protects vessel walls by increasing the expression of stress fibers while turbulent blood flow promotes endothelial cell division and stimulates the transcription of genes for

PDGF and TGF-β which in turn promotes the angiogenic cascade discussed above

(reviewed in [113]).

15 In summary, angiogenesis is a complex process that coordinates many different

activities from several different cell types including endothelial cells, pericytes, smooth

muscle, and fibroblasts. These different cell types interact thru a myriad of mediators

including growth factors (bFGF, VEGF, EGF, PDGF, TGF-β, TNF-α), soluble factors

(Ang1, Ang2, ), enzymes (MMPs, u-Pa), cell surface molecules and receptors (VE-

cadherin, integrins, PECAM-1, Tie2, Eph/Ephrin) to coordinate, such processes as,

proliferation, migration, matrix remodeling, tube formation, and finally reestablishment

of the circulation.

1.4.2 Tumor angiogenesis.

It is traditionally thought that the tumor starts as a small avascular mass and must

subsequently induce angiogenesis in order to grow beyond a size of 1-3 mm3 [115;116].

This process, termed the angiogenic switch by Dr. Flokman, is true for many primary

tumors particularly primary epithelial tumors that are initially separated from the

vasculature by a basement membrane. This is also true for many forced avascular models

were the tumor is placed in an avascular space like the subcutaneous tumor model

discussed in Chapter 5. However many tumors, metastases in particular, do not initiate in

an avascular manner rather tumor cells hone in on and grow near existing host vessels thus starting off as a small well vascularized tumor.

Tumor angiogenesis proceeds in a similar manner as normal angiogenesis, except the pathologic state of the tumor is reflected in the resulting vasculature. For example,

tumor tissues contain greater amounts of growth factors (aFGF, bFGF, VEGF), have

higher proliferative rates, abnormal expression of integrins and cadherins, and have

16 higher incidence of hypoxia as compared to normal tissues. These differences cause a loss of regulation of the angiogenic cascade such that the tumor induced vessels are unable to stabilize and differentiate into mature vasculature.

For example, tumor vessels are typically convoluted, torturous, dilated, and are exceptionally leaky due to the presence of fenestrae, transcellular holes, and non-uniform cell junctions [117]. Tumor vessels do not have a complete basement membrane [118] and often lack functional pericytes [119]. Recent findings also suggest that tumor vessels are lined with tumor cells that have attained vasclogenic properties or that the vessel itself is comprised of a mosaic of both tumor an endothelial cells [120;121]. Even though these structural abnormalities of tumor vessels reflect the pathological nature of their induction, their ability to support tumor cell growth underlies the tumor’s ability to commandeer a normal physiological process for it’s own survival [113;114].

1.4.3 Anti-angiogenic effects of chemotherapy.

First proposed in 1986 [122], it is now widely accepted that cancer chemotherapy works, in part, through toxicity to endothelial cells. The rapidly proliferating endothelial cells show greater chemosensitivity to chemotherapeutic agents compared to epithelial tumor cells [123]. Furthermore, cell death in the vasculature of in vivo tumors preceded that of epithelial tumor cells (~12 hr vs. ~ 3 day) [124]. There is also evidence that treatment schedule, due to the differential drug effects on the epithelial tumor cells and the vascular endothelial cells, affects the therapeutic outcome. For example, Lewis lung carcinoma, that was resistant to maximal tolerated dose schedules of cyclophosphamide, responded to frequent low-dose administration [124]. Patient chemotherapy treatment

17 regimens that are more anti-angiogenic in a mouse cornea model were also more

effective in patients [125;126]. Metronomic therapy, where chemotherapy is given in

continuous low dose, is an emerging treatment paradigm under evaluation.

1.5 IMAGE ANALYSIS

The goal of quantitative histopathology is to bring a measure of objectivity to the diagnostic assessment while at the same time improveing the diagnostic and prognostic capability of the measurement [127]. Increasing both the objectivity and the quantitative nature allows for better evaluation and validation of the target of interest in a clinically relevant model (i.e., actual patient samples). The standard image analysis system necessary for evaluation of histological sections are a microscope, color video camera, digital frame grabber, personal computer, and image analysis software. The camera views the light passing through the histological section and transmits an electrical signal to the

frame grabber which interprets the signal and saves the information digitally.

Commercially there are a great number and variety of cameras, frame grabbers,

microscopes, and image analysis software packages from which to choose. Selection of each component should be based on the needs of the investigator.

As an example of the image capture process, an analog signal from a black and white camera would be translated, by the frame grabber, into numerical data that would

record the location and the grey level of each pixel within the viewed image. The pixel

size and number is based on the resolution of the image. Resolution is defined as the

number of pixels within the image. Typical frame grabbers resolve images at resolutions

of 512 x 512 (262,144 total pixels) or 640 x 480 (307,200 total pixels), however higher resolutions of 1280 x 1024 (1,310,720 total pixels) are used. The more pixels present the

18 better chance an image has of representing the true visual information. The grey level of each pixel is defined by the pixel depth. For example an 8-bit pixel depth divides the monochromatic image into 256 distinct grey values where 0 is defined as pure black and

255 is pure white. Higher pixel depths are also used, for example a 12-bit depth yields

4095 grey values and a 16-bit depth yields 65535 grey values.

For color images, the camera breaks the image down into three separate RGB channels (red, green, and blue). The frame grabber saves each of the three sub-images

separately so that analysis can be performed on one or all of the sub-images. RGB is a

traditional image format however newer modes for viewing images have been developed.

For example, the Hue, Saturation, and Intensity (HSI) format is based on heuristics

relating to human perception, where Hue is defined as the shade of color (red, orange,

yellow, etc.), intensity is the brightness of the color, and saturation is a measure of how

much “white” is present in the color. For example, red and pink are two different

“saturations” of the same hue, red. The HSI format is a transformation of the RGB

format,

where H, S, I, R, G, and B are the values in the hue, saturation, intensity, red, green, and blue channels, respectively [128]. This transformation greatly simplifies the analysis of

19 histological stained tissue because the HSI format separates the intensity of a color from its chromaticity. Meaning the identification of a stain’s color is now defined by one

channel (Hue) as apposed to a combination of the three channels as it is in the RGB

format.

Once the image is digitally stored analysis occurs in two main phases: processing

(or selection of the regions of interest, or ROI, while discarding the irrelevant regions)

and measurement (or extracting and calculating the data based on the selected ROI). ROI

selection is typically accomplished using a threshold that selects pixels based on

automatic or user-defined ranges. For the studies presented in this dissertation ranges

were set based on the Hue, Saturation, and Intensity channels such that the pixels

associated with the brown DAB stain were selected out from the hematoxylin blue

nuclear counterstain. Calculations were performed based on the ROI and included such

parameters such as surface area, length, optical density, and number.

The types of image analysis systems available today are quite diverse and use

many different methods for the detection of colors and shapes. Many of these systems

offer interactive programs, customizable software, or even have motorized stages and

optics. Therefore, each system is unique and can range from interactive, where the

investigator is involved in each step of the image analysis process, to a fully automated

system that runs unsupervised. Each setup offers advantages and disadvantages. The

greater degree of supervision will lead to greater consistency and accuracy, since the

operator must check each phase of the process. However this method is not an effective

use of time.

20 The second setup has the disadvantage of requiring a greater expenditure of time

and money up front since motorized stages and optics are quite expensive, however after

setup is complete the system requires far less time and money then the first. One goal of

this dissertation is to develop more quantitative ways of measuring the clinically relevant

target, be that the amount of bFGF present in an immunostained patient tumor or

quantifing a blood vessel’s cross-sectional area in real engineering units. Based on the

varied needs of these studies presented herein a compromise of the two previously

mentioned setups was reached that allowed for a great degree of automation but still

providing the amount of user interaction needed for the analysis of clinical samples.

1.6 OVERVIEW OF DISSERTATION

This dissertation is focused on improving the understanding of the mechanisms of

FGF-induced resistance and the molecular pharmacodynamics of suramin and is divided

into three main parts. The first establishes in vivo efficacy of suramin in a clinically

relevant model (Chapter 2). The second details the investigation of the mechanism of

bFGF-induced resistance and the role of bFGF in suramin chemosensitization in clinical

samples (Chapter 3). Since the target of suramin is bFGF, establishment of bFGF as a

relevant marker for resistance is important. The last part details the investigation of the

possibility of an anti-angiogenic mechanism and its role in the chemosensitization of

suramin (Chapters 4-5). Elucidation of the in vivo mechanism is important in the clinical

development of suramin so that suramin can be most effectively utilized in the clinic.

The remainder of this dissertation is divided into 5 chapters, including 4 chapters detailing experimental studies and one final discussion chapter. Each chapter is followed by tables and figures relevant to that chapter. Chapter 2 deals with the establishment of

21 the efficacy of low dose suramin in vivo. The hypothesis is that suramin inhibits FGFs

thus preventing them from binding to their cell surface receptors and conferring a

survival advantage to the tumor cells. Addition of monoclonal antibodies against bFGF to

the same experimental system resulted in the same effect this suggesting that the

antibodies and suramin work through the same mechanism [2]. Establishment of the in

vivo efficacy allows for further investigation into the in vivo mechanism with the intent of developing suramin for use in the clinic.

With the in vivo efficacy established, the next step is to test the correlation of intertumoral bFGF levels in with drug resistance and suramin chemosensitization in clinical samples. bFGF is the hypothesized therapeutic target and drug resistance is the relevant pharmacodynamic effect therefore evidence of a positive correlation between the two will strengthen the clinical utility of suramin. Furthermore a positive correlation between the bFGF level and suramin chemosensitization will strengthen the in vivo

hypothesis that suramin is acting through bFGF. To accomplish these goals, a

quantitative image analysis methodology was developed. The method included

development of a bFGF standard of immunohistochemically stained samples and a

custom image analysis program to convert the immunohistochemical data into actual

protein amounts. The new method was then applied archival bank of

immunohistochemically stained patient tumors. The pharmacodynamic data of the

antiproliferative and apoptotic effects of paclitaxel and the chemosensitization effects of suramin on the patient samples were correlated to the intratumoral bFGF level. Results indicate bFGF was the single best indicator for drug resistance and bFGF level correlated

well to suramin effect.

22 Understanding the in vivo mechanism of suramin will lead to new treatment

paradigms and targets for suramin and the optimization of the dosing schedule. This will

in turn increase quality of life, response rate, progression free and overall survival while

decreasing toxicity for the patients enrolled in suramin clinical trials. These studies were

based on reports of the literature showing the bFGF to be highly pro-angiogenic and

conversely suramin to be anti-angiogenic in many experimental systems. Therefore the

possibility of an anti-angiogenic was tested in both in vitro (Chapter 4) and in vivo

(Chapter 5) models. Results from the in vitro studies showed that suramin alone had no effects at low doses nor did suramin enhance the cytotoxicity of chemotherapy in a

monolayer HUVEC system. This was further confirmed in an in vitro histoculture system

and the results showed that the chemosensitization effect was independent of the tumor

vasculature. The in vitro findings were further conferred using a well established in vivo

solid tumor model (Chapter 5). Results from Chapter 5 show that chemotherapy had a

profound effect on the tumor vessel morphology and functionality, however suramin did not affect the vessels when used as a single agent, nor did it enhance the chemotherapy effect. Overall, the results from the last two Chapters show that suramin does not act

through an anti-angiogenic mechanism, thus this result separates the Au/Wientjes bFGF hypothesis from the angiogenesis paradigm as an independent and novel area of study.

23 CHAPTER 2

LOW DOSES OF SURAMIN ENHANCE THE ACTIVITY

OF PACLITAXEL IN VIVO

2.1 INTRODUCTION

Au and Wientjes et al. recently reported that acidic (aFGF) and basic (bFGF) fibroblast growth factors confer a broad spectrum chemoresistance in solid tumors, and that suramin, an inhibitor of multiple growth factors including aFGF and bFGF, enhanced the in vitro antitumor activity of several anticancer drugs including paclitaxel, doxorubicin, and 5-fluorouracil [2]. The suramin concentration used in these studies was

15 µM and did not cause cytotoxicity in cultured human tumor cells. Specific inhibitors of aFGF and bFGF, monoclonal antibodies, were also able to reverse the FGF-induced resistance suggesting the same mechanism [2;129].

The in vitro mechanism of bFGF-induced resistance is clear; however the in vivo mechanism is unknown. Therefore the goal of the present study was to evaluate whether suramin can be used to enhance the therapeutic efficacy of chemotherapy in lung cancer, thus establishing the in vivo efficacy of low dose suramin. This is first step in the pre- clinical investigation for the clinical translation of suramin and in the investigation of the in vivo mechanism of suramin.

24 The present study investigated the in vivo interaction between paclitaxel and suramin in a human PC3-LN cells which metastasized to the lung. It is a clinically relevant model because tumor metastasis is the major cause of treatment failure for cancer patients [130]. About 60% of patients have microscopic or clinically evident metastases at the time of diagnosis of primary tumors. Surgery and localized radiation are of limited value, therefore systemic chemotherapy constitutes an important treatment modality for metastatic diseases. The differential response of primary and metastatic tumors to chemotherapy has been documented in preclinical experimental models and in clinical practice (reviewed in [131]).

In general, metastatic tumors are more resistant to chemotherapy compared with

their corresponding primary tumors [131-133]. The microenvironment of the tumor-

bearing organ may play an important role in lowering the chemosensitivity of metastatic

tumors (reviewed in [131;133]). For example, earlier work by Au and Wientjes et al.

showed the antitumor activity of paclitaxel in lymph node metastases was 20-fold lower

than in subcutaneously implanted primary tumors of the parent cell line (a metastatic sub-

clone of rat prostate MAT-LyLu cells). When the metastatic tumor was subcutaneously

reimplanted as a primary tumor in a second animal, the resistance was lost but regained in

the second generation metastases. Further studies confirmed that the observed

chemoresistance was not due to reduced intracellular drug accumulation or retention

ruling out differences in drug delivery between the two sites [129].

Because paclitaxel is among the most effective agents against lung tumors

[134;135], the present study was performed using suramin and paclitaxel. It has been shown that the FGF proteins expressed in lung metastases are the cause of the drug

25 resistance, therefore it more important to use a model that resides in the lungs as apposed

to a subcutaneously implanted model derived from lung tumor cells. Accordingly, in vivo

studies were performed using human prostate PC3-LN tumor cells which, when injected

i.v. in immunodeficient mice, result in well-established lung tumors within 4–6 weeks

with a 100% success rate [136].

The single-drug therapy with paclitaxel or suramin did not reduce body weight,

and single agent suramin had no antitumor activity. Paclitaxel alone reduced the tumor

size by ~75%, reduced the density of nonapoptotic cells by ~70% in residual tumors, and

enhanced the fraction of apoptotic cells by ~3-fold. The addition of suramin to paclitaxel

therapy enhanced the antitumor effect, resulting in an additional 5-fold reduction of

tumor size, an additional 9-fold reduction of the density of nonapoptotic cells, and an additional 30% increase in the apoptotic cell fraction. These data indicate significant enhancement of the efficacy of paclitaxel by suramin and support the use of nontoxic doses of suramin with paclitaxel in the treatment of lung cancer.

2.2 MATERIALS AND METHODS

2.2.1 Chemicals and reagents.

Paclitaxel was obtained from Bristol-Myers Squibb (Princeton, NJ), the National

Cancer Institute (Bethesda, MD), or Hande Tech (Houston, TX). Suramin was purchased from Sigma Chemical Co. (St. Louis, MO). Cefotaxime sodium from Hoechst-Roussel

(Somerville, NJ) and cell culture supplies from Life Technologies, Inc. (Grand Island,

NJ).

26 2.2.2 Cell and tumor cultures.

Human prostate PC3-LN tumor cells were a gift from Dr. Joy Ware (Virginia

Commonwealth University, Richmond, VA). Tumor cells were maintained as monolayer

cultures at 37°C in a humidified atmosphere containing 5% CO2 in RPMI 1640 supplemented with 9% fetal bovine serum, 2 mM L-glutamine, 90 µg/ml gentamicin, and

90 µg/ml cefotaxime.

2.2.3 Drug solutions.

Suramin stock solution (1.1 mg/ml) was prepared in physiological saline.

Paclitaxel stock solution (15 mg/ml) was prepared in Cremophor EL and ethanol (50:50) and diluted with 9 volumes of physiologic saline or suramin stock before administration to animals. Drug solutions were filtered with a 0.2 µm diameter filter before use.

2.2.4 Animal and drug treatment protocols.

Male BALB/c nu/nu mice (6–8 weeks of age) were used. Animal care was in accordance with institutional guidelines. Human PC3-LN cells (106 in 0.1 ml

physiological saline) were injected i.v. via tail vein. After 5 weeks, tumor establishment was determined by visual examination of the lungs of two randomly selected animals, and drug treatment in the remaining animals was initiated when these two animals showed at least five tumor nodules of ~1 mm in diameter.

Mice received i.v. injection, over 1 min, via a tail vein, of 200 µl of a Cremophor

EL/ethanol/saline solution delivering 15 mg/kg paclitaxel, 10 mg/kg suramin, or a combination of both drugs, twice weekly for 3 weeks. The control group received only vehicle. Pharmacokinetic studies in normal mice (i.e., without tumors) indicated that the

27 selected doses yielded a peak plasma concentration of 50 µM and a concentration of 1

µM at 72 h for suramin and a peak plasma concentration of 1 µM and a concentration of

4 nM at 72 h for paclitaxel [132].

2.2.5 Histological evaluation of tumors.

Three days after completion of the drug treatment the animals were euthanized

and their lungs were removed, fixed in Bouin’s solution to visualize tumor nodules, and

then processed for histological evaluation. Histological sections (5 µm) at a depth of

between 200–300 µm from the ventral surface and containing all five lobes of the lungs

were obtained. The lung surface area (counted as the number of pixels) occupied by the

tumor was calculated as a fraction of the total lung area, using Optimas® image analysis

software. The number of tumor cells in residual tumors and the fraction of apoptotic cells in each tumor was also determined.

Cells that showed condensed nuclei and membrane blebbing were considered apoptotic. Others have shown that apoptotic cells identified by these morphological

changes are identical to the apoptotic cells identified by TUNEL staining [137;138].

Because apoptotic cells disappear over time, a second measure of the extent of apoptosis

was the density of nonapoptotic cells in the residual tumors. This was determined by

counting the number of nonapoptotic tumor cells in randomly selected microscopic fields

at x400. On average, the number of cells counted were, 1230 ± 760 cells/animal in the

control and suramin groups and 530 ± 530 cells/animal in the paclitaxel group. In the case of combination therapy, where few tumor cells remained after treatment, therefore all the residual cells were counted; 150 ± 160 cells, between 16–525 cells/animal.

28 2.3 RESULTS

2.3.1 Synergy between paclitaxel and suramin in vivo.

Results of the in vivo antitumor activity evaluation are summarized in Figure 2.1 and Table 2.1. Suramin alone had no antitumor effect nor toxicity, consistent with the results in other mouse tumor models [2;139;140]. Single drug therapy with paclitaxel or suramin did not reduce body weight. Paclitaxel alone significantly reduced tumor size by

~75%, reduced the density of nonapoptotic cells by ~70%, and increased the fraction of apoptotic cells in the residual tumors by ~3-fold. Addition of suramin to paclitaxel therapy significantly enhanced the antitumor effect, resulting in an additional 5-fold reduction of tumor size, an additional 9-fold reduction in the density of nonapoptotic cells, and an additional 30% increase in the apoptotic cell fraction. It is of interest that a small fraction of animals (< 30%) in the control and single drug therapy groups showed extrapulmonary tumors in the neck and diaphragm, whereas no extrapulmonary tumors were found in the combination therapy group.

2.4 DISCUSSION

Several growth factors, including bFGF, IGF, and epidermal growth factor, have been shown to induce tumor resistance to anticancer drugs [49-51;53;55;141-143].

Earlier studies by Song et al. also shows that aFGF, although not required to induce drug resistance, enhanced the FGF effect such that the two proteins, at clinically relevant concentrations, can induce a 10-fold resistance [2]. Results of the present study indicate that suramin, a nonspecific inhibitor of aFGF and bFGF, enhanced the activity of paclitaxel in animals bearing lung metastases. This interaction between suramin and paclitaxel was achieved at low and nontoxic concentrations and doses of suramin.

29 The clinical relevance of the FGF resistance mechanism is supported further by the observation presented in Chapter 3 where FGF expression was found to be a better predictor of paclitaxel resistance in human tumors, as compared with other known prognostic indicators, such as mutated p53, overexpression of Bcl2 and Pgp, and tumor pathology (grade, stage, and labeling index; see Chapter 3).

In addition to aFGF and bFGF, suramin also inhibits the action of several other polypeptide growth factors, including platelet-derived growth factor, vascular endothelial growth factor, transforming growth factor-β, and IGF-1 (reviewed in Chapter 1)

[93;94;99;103;108]. The decision to use suramin in this study was in part because clinical pharmacological data are readily available for this compound. The literature shows that suramin has moderate activity in prostate cancer (see Chapter 1) [79;80;92;144] with a therapeutic window of 100-200 µM (140–280 µg/ml) [90] in the plasma. The two important limitations of suramin are: (a) its broad spectrum of toxicity including neurotoxicity, renal toxicity, adrenal insufficiency, and immune- and glycosaminoglycans anticoagulant-mediated blood dyscrasias [145-151]; and (b) difficulty in dose administration because of its exceedingly long terminal plasma half-life of >21 days

[152;153].

The relatively modest activity of suramin led to the development of combination therapies of suramin with other agents, where suramin was again given at doses that resulted in a >200 µM concentration; these combinations have either shown limited benefit or have resulted in toxicity that discouraged additional evaluation of these regimens [82;83;89;154-156]. The major difference between the previous preclinical and clinical studies with suramin and this study is the intended use of suramin and thus the

30 selection of the dose. In previous studies, suramin was used as a therapeutic agent and, therefore, required the maintenance of a target concentration of >200 µM (the maximum tolerated dose). In the current study, suramin is used to reverse the FGF-induced resistance, an effect requiring <20 µM, which has no cytotoxicity in cultured tumor cells nor toxicity in animals or patients.

In summary, results of the present study indicate that low and nontoxic doses of suramin significantly enhance the in vitro and in vivo antitumor activity of paclitaxel, and support a new treatment paradigm using combinations of chemotherapy with aFGF/bFGF inhibitors.

2.5 ACKNOWLEDGMENTS

This work was supported in part by research grants R37CA49816 and

R01CA78577 from the National Cancer Institute, NIH, and by a research fellowship from the American Foundation for Pharmaceutical Education. I would like to thank my advisor

Dr. Jessie Au for her expertise, support, and patience. Special thanks go to Dr. M. Guill

Wientjes, Dr. Saeheum Song, Dr. Yong Wei, Dr. Liang Zhao, and Dr. Yilong Zhang for their technical support and scientific input in the study.

31

No. of mice with % lung surface area occupied Density of Apoptotic cell End-of-experiment body extrapulmonary by tumor nonapoptotic fraction weight (% of Treatment (n) tumors Median Mean cells/x400 field (%) pretreatment value) Control (9) 1 9.5 12.6 ± 10 180 ± 60 18 ± 6 94.5 ± 7.9 Suramin (9) 3 10.3 18.7 ± 23.1 163 ± 32 19 ± 8 96.6 ± 10.5 Paclitaxel (15) 2 2.4 3.2 ± 3.8a 61 ± 34a 63 ± 15a 94.3 ± 7.1 Combination (15) 0 0.2 0.7 ± 1.2a,b 7±9a,b 94±5a,b 92.1 ± 5.8 a P < 0.05 compared with control and suramin groups. b P < 0.05 compared with all other groups.

32

Table 2.1. Enhancement of in vivo antitumor effect of paclitaxel by suramin. Immunodeficient mice bearing human PC3-LN

lung metastases were treated with physiologic saline (i.e., control), 15 mg/kg paclitaxel, 10 mg/kg suramin, or a combination of

both drugs. The average pretreatment weights for the four groups ranged from 21 to 22 g. Density of viable nonapoptotic cells

and apoptotic cell fraction were determined using randomly selected microscopic fields, at x400 magnification. Mean ± SD.

Statistical analysis was performed using ANOVA with post hoc Tukey analysis of group differences.

32

Figure 2.1. Enhancement of in vivo antitumor activity of paclitaxel by suramin.

Animals with well-established PC3-LN lung metastases were treated with physiological saline (i.e., control), suramin (10 mg/kg), paclitaxel (15 mg/kg), or a combination of both drugs. Top, visible, large tumors on the ventral and dorsal surfaces of the lungs in all animals in the control and suramin groups, small tumors in the paclitaxel group, and yet- smaller tumors in the combination group. Bottom, histological sections (x100), with tumors outlined in red.

33 CHAPTER 3

EXPRESSION OF BASIC FIBROBLAST GROWTH FACTOR CORRELATES

WITH RESISTANCE TO PACLITAXEL AND SURAMIN

CHEMOSENSITIZATION IN HUMAN TUMORS

3.1 INTRODUCTION

Recent reports by Au and Wientjes et al. established that extracellular acidic and basic fibroblast growth factors (FGFs) expressed in solid tumors induce broad spectrum

chemoresistance [2]. The results from Chapter 2 suggest that low doses of suramin

(plasma concentration <50 µM) can reverse the FGF-induced chemoresistance and

enhance the activity of chemotherapy. Therefore the first goal of the studies presented in

this chapter was to investigate the relationship between the intratumoral bFGF level and

drug effect. Establishing the intratumoral level of bFGF is a predictor for drug resistance

in clinical tumor samples will add weight to the original findings of Au and Wientjes and

show that bFGF is a valid target for therapeutic intervention.

Many literature reports detail studies linking elevated serum and intratumoral

levels of bFGF to worse patient prognosis (reviewed in Chapter 1). These studies

typically measure intraturmoral protein levels by homogenizing the tissue and assaying

using an immunosorbent assay (ELISA) or Western blotting. The advantage is that the

results are quantitative, however the disadvantages are that the 3-D structure of the tissue

34 is lost. To maintain the tissue structure, immunohistochemistry is often employed. The

resulting staining is then viewed by a pathologist and a discontinuous score (typically 0 to

3 in arbitrary units) is assigned to represent the staining intensity. This method is semi-

quantitative in nature, subjective, time consuming, and requires someone trained in

pathology.

This outlines a clear need for assays that are able to quantify the amounts of clinically relevant biomarker proteins (bFGF) within the tumor while still maintaining the

important three dimensional tumor structure. To address this need, the second goal of this

study was to develop an assay that would allow for rapid quantification of bFGF levels in

immunohistochemically stained patient samples without the need for additional

pathological evaluation. Immunohistochemistry was chosen because the intensity of the

staining is proportional to the protein concentration and the spatial distribution of the

stain is accurate to the protein location. Current image analysis technology can be readily

applied to interpret and quantify the protein levels in the sample based on the parameters.

To address the first and second goal, the correlation between intratumoral bFGF

level and drug effect was evaluated in a set of 96 patient samples collected from bladder,

breast, ovarian, prostate, and head and neck. The tissues were obtained from a previous

study [157]. The purpose behind using the large variety of tumor types was included to

establish the broad spectrum applicability of bFGF as a clinically relevant predictor of

resistance. All the tumor samples were immunostained for bFGF and analyzed by the

new method and by the traditional pathologic scoring method for comparison and

validation of the novel image analysis method. The bFGF level was correlated with the

paclitaxel effects in each of the 96 tumors.

35 The dose response relationship was established in each of the 96 tumors for the drug paclitaxel via histoculture system and resistance was expressed as IC30, maximum drug effect (EMAX), and % apoptosis values. The data was adopted from a previous study

[157]. The major advantages of the histoculture system are the maintenance of tissue architecture, cell-cell interaction, and inter- and intratumoral heterogeneity. The clinical relevance of the human tumor histoculture system has been demonstrated in retrospective and semi-retrospective preclinical and clinical studies, which show that drug response in histocultures correlates with sensitivity and resistance of cancer patients to chemotherapy and patient survival [158-160].

In addition to the paclitaxel dose response data, several traditional pathobiological parameters were also adopted from previous work to weigh the overall importance of bFGF as a predictor for resistance [157]. The parameters evaluated were tumor stage, grade, labeling index (LI), bcl-2, pgp, and p53 status. Tumor stage and grade were included because they are readily available and represent a measure of the malignant potential of the tumor. Labeling index refers to the % of cells labeled positively by BrdU and relates directly to the proliferation rate of the tumor sample. Paclitaxel acts during the

M phase of the cell cycle and its ability to induced apoptosis has been correlated to cell proliferation [161-163]. P53 and pgp represent classical mechanisms of drug resistance and are therefore valuable in comparison when considering the bFGF-induced paradigm.

Pgp has also been reported as playing a role in paclitaxel resistance [164] and p53 has shown involvement in drug induced apoptosis [165]. Bcl-2 was added because high levels of bcl-2 have been shown to delay the onset of paclitaxel induced apoptosis and bcl-2 expression is changed as a result of bFGF signaling [165].

36 Results show that the bFGF levels were the best single parameter predictor of drug effect and that the quantitative image analysis method was the best method for quantifying the correlation. This validated bFGF as a clinically relevant predictor of resistance and therefore a target for therapeutic intervention. Therefore, the third goal of the study was to evaluate the effects of suramin on the intratumoral bFGF level and test the correlation between the bFGF level and IC50 for 5-Fu and the fold chemosensitization caused by low dose suramin. Renal cell carcinoma (RCC) samples were exposed to doses of 5-Fu alone or in combination with suramin in the same previously mentioned histoculture model. Results show that the post-treatment bFGF level was strongly correlated with the chemosensitization effect of suramin.

3.2 MATERIALS AND METHODS

3.2.1 Chemicals and supplies.

The chemicals used to study the pharmacodynamics of paclitaxel and to detect

Pgp, p53, and bcl-2 were as previously described [137;157]. Monoclonal antibodies against aFGF and bFGF, bicinchoninic acid solution, copper (II) sulfate solution, 5- fluorouracil, suramin, and bromodeoxyuridine (BrdU) were purchased from Sigma

Chemical Co. (St. Louis, MO). Labeled Streptavidin-Biotin detection kit and anti-BrdU antibodies were purchased from Dako (Carpiteria, CA). RIPA lysis buffer was purchased from Upstate (Lake Placid, NY) and Protease Inhibitor Cocktail Set III was purchased for

Calbiochem (San Diego, CA). Human basic FGF ELISA kit was purchased from

Oncogene (Cambridge, MA). Collagen gel was purchased from Ethicon (Somerville, NJ) and cefotaxime sodium from Hoechst-Roussel (Somerville, NJ). DAB (3,3

37 diaminobenzidine tetrahydrochloride) substrate kit was purchased from BioGenex (San

Ramon, CA). All medium and cell culture supplies were purchased from GIBCO (Grand

Island, NY). All chemicals and reagents were used as received.

3.2.2 Procurement of tumor specimens.

Specimens of human bladder, breast, head and neck, ovarian and prostate tumors

were obtained via the Tumor Procurement Service at The Ohio State University

Comprehensive Cancer Center. Tumor pathology was determined by the University

pathologists. Of the 96 tumors studied, 95 were from chemotherapy naive patients. The

remaining tumor was from a head and neck cancer patient that had received paclitaxel

treatment. Human specimens of RCC were obtained from the Cleveland Clinic

Foundation. RCC tumor samples were placed in specimen jars with MEM in 5% FBS and

shipped on ice via overnight express.

3.2.3 Pharmacologic effects of chemotherapy.

The pharmacodynamic data of the antiproliferative and apoptotic effects of

paclitaxel and 5-Fu were obtained from previous studies [157;166]. Briefly, the patient

tumors were cut into 1 mm3 pieces. Four to six tumor pieces were placed on a 1 cm2 collegen gel presoaked in medium in six-well plates. Paclitaxel was chosen as the chemotherapeutic agent based on its clinical utility in bladder, breast, ovarian, prostate, and head and heck [167]. Tumor histocultures were treated with paclitaxel for 2 hrs

(bladder tumors) or 24 hrs (all other tumors). The 2 hrs treatment is the duration of intravesical therapy of superficial bladder cancer, whereas 24 hrs exposure is one of the

38 commonly used treatment schedules in patients. 5-Fu is currently the standard treatment

for renal cell carcinoma [168]. RCC histocultures were treated with 5-FU in the presence

or absence of either 20 or 50 µM suramin, for 48 hrs.

After treatment all histoculture specimens were fixed in 10% formalin and

embedded in paraffin for subsequent sectioning and histological analysis. The

antiproliferative effect of chemotherapy was measured as the reduction in the fraction of

tumor cells labeled by the DNA precursor, BrDU. In the paclitaxel treated tumors,

apoptosis was identified by morphological changes, TdT-mediated dUTP nick end

labeling, and/or DNA fragmentation.

Paclitaxel effects were reported as IC30, EMAX, and the maximum apoptotic index.

The apoptotic index was counted for each histoculture at each of the drug concentrations, and the maximum value was recorded for analysis. 5-Fu effects were reported as IC50 of

5-FU alone, IC50 of the 5-FU and suramin combination, and the ratio of IC50 of the combination divided by IC50 of 5-FU alone. The ratio value was used as a measure of the

degree of chemosensitization effect of suramin. All the IC30/50 and EMAX values were

calculated based on the dose response curves generated from the BrdU staining result.

3.2.4 Immunohistochemistry.

The data of pgp, p53, and bcl-2 expression before and after paclitaxel treatment

were obtained from previous studies [157]. Original patient samples (collected and fixed

prior to histoculture) for bladder, breast, ovarian, prostate, RCC and head and neck; along

with the RCC untreated histoculture controls were stained for aFGF and bFGF using the

same protocol. The protocol was adopted from methods as previously described [137].

Briefly, after de-waxing and rehydration sequentially in xylene, ethanol and water, tissue

39 sections were boiled for 5-7 min in a 0.1 M citrate buffer, pH 6.0, in a microwave oven, then cooled and washed in phosphate buffered saline (PBS). After wiping off the excess

PBS, lines were drawn on the glass slide between the tissue sections, using a hydrophobic marking pen. This created a barrier that confined the different antibody solutions applied to different tissue sections. The tissue sections were incubated with Dako blocking solution for 10 minutes and then with mouse anti-human aFGF antibody (1:50, 1:100,

1:200 dilution) or bFGF antibody (1:20, 1:50, and 1:100 dilution) for 2 hours in a humidified chamber at room temperature. The antibodies were diluted in PBS containing

1 mg/ml bovine serum albumin. The negative controls used mouse IgG of the same isotype as the primary antibody. After washing with PBS, the tissue sections were incubated with the linker solution (DAKO kit) for 40 minutes, and then with peroxidase- conjugated streptavidin solution for 40 minutes (DAKO kit). After washing twice with

PBS, tissue sections were incubated for 6 min with diaminobenzidine and counterstained with hematoxylin.

3.2.5 Pathologic scoring of the immunohistochemical result in the paclitaxel treated samples.

Development of a novel assay requires comparison of the new assay to the excepted method. Immunohistochemical staining is typically scored by a trained pathologist who assigns a score based on the visual assessment of the staining intensity.

The pathologic scoring results from the pgp, p-53, and bcl-2 were adopted from previous studies [157].The aFGF and bFGF staining results were scored using the same method as

40 above. Briefly, this method employs a grading system where a score of 0-2 is given to

represent the intensity of the observed staining. This is illustrated in the following two

tables.

Staining intensity of bFGF at each Staining intensity of aFGF at each Final Final antibody dilution antibody dilution Score Score 1:20 1:50 1:100 1:50 1:100 1:200 0 - ~ +/- - - 0 - ~ +/- - - 1+ +/-- 1+ +/-- 2 + + +/- ~ + 2 + + +/- ~ + -: negative, +/-: weak positive, +: strong positive -: negative, +/-: weak positive, +: strong positive

As an example, if the tumor in question had an observed bFGF staining of strong positive

(+) at the 1:20 dilution, weak positive (+/-) at the 1:50 dilution, and was negative at the

1:100 dilution level then that tumor would be given a score of 1.

3.2.6 Development of the bFGF standard curve.

The purpose was to develop a reproducible immunohistochemical standard that could be stained in parallel to the clinical samples for the purpose of quantifying the immunohistochemical stain in real protein amounts. This required the development of a bFGF staining standard and an image analysis protocol to analyze and quantify images of the staining result.

3.2.6.1 Cell culture.

The tumor cell lines PC3, MiaPACA, SKOV3, HT29, and FADU were purchased from American Type Culture collection (Rockville, MD). Cell lines were chosen based on their intercellular bFGF content ([169] and unpublished data). PC3 and MiaPACA cells were maintained in DMEM medium, HT29 and SKOV3 cells were maintained in

McCoy’s 5A medium, and FADU cells were maintained in MEM medium. All medium was supplemented with 10% fetal bovine serum, 90 mg/ml gentamicin, 2 mM L-

41 glutamine, and 90 mg/ml cefotaxime. McCoy’s 5A and MEM medium was additionally

supplemented with 0.1% non-essential amino acids. All cells were incubated at 37 °C in a

humidified atmosphere containing 5% CO2.

3.2.6.2 Animal protocol for subcutaneous tumor model.

Male/female balbc/nu.nu and athymic nude mice (all mice 5 weeks old) were

purchased from the National Cancer Institute (Bethesda, MD), housed in air-filtered

laminar flow cabinets and cared for in accordance with institutional guidelines. All

tumors cells were harvested from sub-confluent cultures using trypsin, and suspended in

serum free medium for tumor cell implantation. Tumor cells were injected

subcutaneously into the flank on both sides of a mouse (2x106 cells/200 µl per injection

site). PC3 and SKOV3 cells were injected in male and female (respectively) balbc/nu.nu

mice; HT29 and MCF-7 were injected in female athymic nude mice; and MiaPACA,

FADU, and HS766T were injected in male anthymic nude mice. Animals were

euthanized, and the tumors excised, when the longest diameter of the tumor was > 5 mm.

At time of excision, all visible non-tumor tissue was removed and half the tumor was

fixed in 10% formalin in a manner identical to the patient samples and the other half was

weighed and lyzed by adding ice cold lysis buffer (1 ml of buffer per 0.1 gram of tumor)

followed by sonication on ice. RIPA buffer contained 0.2% protease inhibitor cocktail,

0.1% SDS, and 1 mM PMSF to help protect the bFGF from degradation. Tumor lysates

were then centrifuged at 14,000 rpm for 15 min and the supernatant was then removed, assayed for protein content, and stored at -70 oC.

42 3.2.6.3 Evaluation of bFGF levels in tumor lysates.

bFGF levels of the tumor lysates were determined using a bFGF ELISA kit.

Assay was performed per manufacture instruction. Briefly, 50 µl of assay diluents RD1-

43 was added to a well of a 96 well plate followed by 200 µl of each tumor lysate or

bFGF standard. After incubation for 2 hours (at room temperature) the wells were washed

with washing buffer and 200 µl of a solution containing murine monoclonal anti-bFGF

antibody conjugated to horseradish peroxidase was added. Following another 2 hour

incubation, the wells were again washed with buffer and 200 µl of the substrate solution was added. After 30 min, 50 µl of the stop solution was added and the color intensity was determined at 450 nm using a microplate reader. Detection limit was 2.5 pg/ml [169].

3.2.6.4 Evaluation of the inter-day and intra-day variability of the microscope light source.

The inter and intra-day variability of the light source was tested because the intensity of the light source directly effects the resulting image. One of the stained sections from one of the tumor standards was selected and the same location of that section was photographed on the three consecutive days over the course of 9 hours (9am-

6pm) with one image taken every 15 minutes (37 images per day). Since the mean intensity of each image is directly related the intensity of the light source, the intra-day variability was measured by calculating the coefficient of variation of the mean intensity of each image taken throughout each day. The average coefficient of variation of each of the 3 days was 3.16%. Since the same area was photographed each of the three days, the inter-day variability was measured by calculating the coefficient of variation of the three mean intensities, one for each day. This value was <1%.

43 3.2.6.5 Image capture and processing.

All the images used in this study were captured at a magnification of 400x using a

Hamamatsu (Hamamatsu-City, Japan) color chilled 3CCD camera attached to a ZEISS

(Thornwood, NY) Axiovert 35 microscope and saved in 8-bit TIF format. All images were analyzed in Hue, Saturation, and Intensity (HSI) format. A custom macro was

written in Optimas® (version 6.51, Media Cybernetics, Silver Spring, MD) to allow for

unsupervised, high throughput processing of the images. The macro opened each image

in turn and applied a user defined threshold to select the brown bFGF from blue

hematoxylin counterstain then extracted the sum of the optical densities (see Results

section 3.3.1) for each image and exported the data to an Excel spreadsheet.

3.2.7 Quantification of bFGF level in the immunohistochemically stained samples.

Inspection of the pathological scoring results in the 96 paclitaxel treated tumor set

showed that the 1:50 dilution had the greatest range in staining intensity. Therefore the

1:50 antibody dilution condition was adopted for the paclitaxel treated samples and then

extended to the RCC samples. Tumor standards were photographed (5-10 images per

tumor) first and the sum of the optical densities (summed OD values) were calculated

using the macro and plotted against the ELISA result to generate the standard curve

(Figure 3.2 and 3.3). The macro was then modified to convert the summed OD values

directly to bFGF level (units of pg bFGF / mg total protein). The patient samples were

then photographed (average 5.6 of images per tumor sample) and the macro calculated

the bFGF concentrations automatically. Any summed OD values for images from the

clinical samples above the top of the linear curve were set to the maximum value of the

standard curve.

44 3.2.8 Statistical analysis.

Non-linear fitting of the standard curve was accomplished using Sigma Plot

(Point Richmond, CA) and the linear fitting was accomplished using linear regression in

SAS (Cary, NC). Predictive relationships between tumor pathologic parameters and

tumor chemosensitivity were evaluated by linear regression analysis using the maximal r2 selection method and the REG software routine of SAS (Cary, NC). This model determines which model has the highest coefficient of determination for combinations of predictors. An accepted principle of development of a model is to select the simplest model that gives a good description of the data [170]. Generally, an increase in model complexity or number of predictors increases the goodness of fit or r2. The Akaike

Information Criterion (AIC) was used to balance model simplicity and goodness of fit

[170].

3.3 RESULTS

3.3.1 Quantitative image analysis methodology.

All images were analyzed in HSI format over the standard RGB because it separates the intensity of a color from its chromaticity (reviewed in Chapter 1). Meaning the brown bFGF stain and the blue hemotoxlian counterstain appear in different regions

of the Hue channel and are therefore easily separated from one another (Figure 3.1).

Choice of the threshold parameters to select the bFGF stain was accomplished by

selecting images from the patient image set that exhibited different staining

characteristics. For example strong and weak cytoplasmic, strong and weak stromal, and

strong nuclear staining were analyzed to see in what HSI ranges contained these different

types of staining. The ranges of the Hue and Saturation channels were very similar across

45 all staining types, however the Intensity values exhibited a much greater range. For the final threshold the middle range of the Intensity values was selected. Once the threshold was selected, the same threshold was throughout the entire study.

The most difficult obstacle to overcome was selection of a parameter to numerically represent the heterogeneous nature present in the immunostaining and in the tumor structure. Initial choices for quantitative parameters included the mean gray value, mean log inverse grey value, and the total positive area stained per image. The mean gray value and log inverse grey value could not accurately reflect the complexity of the stain and the total area omitted the information contained in the staining intensity.

The logic for the choice of the quantitative image analysis variable is best illustrated by the staining result of the HT29 and FAFU standard tumor samples (Figure

3.2). These two samples show similar bFGF content, as measured by ELISA, but very different staining profiles. The bFGF staining in the HT29 sample is compartmentalized with smaller areas of higher staining intensity. In contrast, the FADU exhibits a less intense, more diffuse staining profile. The final variable choice was to calculate the summation of the optical densities of the positively stained areas with in the image. This variable acknowledges this inherent system heterogeneity. For example, the compartmentalized staining of the HT29 sample consist of a small number of positive pixels with a high OD value. The FADU sample consists of the same amount of bFGF

(per ELISA result) but it is spread out over a larger area thus the number of positively stained pixels increases but each has a decreased OD value. By summing the OD values the stains of each sample can be accurately quantified.

46 Figure 3.2 shows representative images from the standard tumors. The zero bFGF concentration used in the standard curve was measured by applying the image analysis program to the negative control samples. The zero condition is logical since the

summation of the optical densities is zero for a bFGF concentration of zero picograms.

The standard curve was well explained (r2=0.98, p<0.001) using the following equation,

y = a*(1-bx) (1)

where y represents the sum of the optical densities divided by the tissue area, x is the

bFGF concentration, and a and b are constants equal to 0.3808 and 0.9941 respectively

(Figure 3.3). This saturation is expected given the nature of the enzymatic reaction that

generates the brown DAB stain. The quantitative power of the standard is lost beyond the

MiaPACA sample due to the saturation of the stain and the power fit does not accurately fit the tumor samples with lower bFGF concentrations. Therefore the curve was truncated at the MiaPACA sample and linearly fitted (r2 = 0.9979, p-value = 0.0007). The linear fitting was used in all subsequent correlative studies (Figure 3.4).

3.3.2 Correlation between paclitaxel effects and bFGF expression levels in the 96 patient tumor set.

Paclitaxel produced partial antiproliferation (mean ± SD, 46 ± 19%) in 84%

(81/96) of the tumors and induced apoptosis (mean ± SD, 13 ± 7%) in 96% (92/96) of the

tumors. Results from the pathological scoring showed that bFGF was detected in 61/96

(64%) of the tumors (Table 3.1). Representative images of the patient samples stained for

aFGF and bFGF are displayed in Figures 3.5 and 3.6 A, respectively. The location of the bFGF staining was mainly located in the cytoplasm of 12 tumors (20%), nucleus of 12 tumors (20%), and staining was located in both for 37 tumors (60%). Results from image

47 analysis show that bFGF was highly expressed in the ovarian cancers followed by the

bladder, prostate, breast, and head neck (Figure 3.6 B).

Of the 96 available tumor samples, 91 were quantified by image analysis and

included in the correlative study (n = 15, 14, 22, 17, and 23 for Bladder, Breast, Head and

Neck, Ovarian, and Prostate; respectively). The paclitaxel effects (expressed as IC30,

EMAX, and apoptotic fraction) were correlated to the tumor pathobiological parameters using multivariate regression analysis. IC30 values were chosen over IC50 because some

of the samples were very resistant to paclitaxel and the IC50 calculation was not

accurately defined by the dose response curve in these tumors. The tumor pathobiological

parameters included the intratumoral bFGF expression level as measured by a novel

quantitative image analysis method, aFGF and bFGF as measured by traditional

pathologic scoring, and data obtained from previous studies (tumor stage, grade, labeling

index, p53 expression, pgp expression, and bcl-2 expression) [157]. Intratumoral bFGF expression level as determined by image analysis was the most important single parameter predictor of paclitaxel effect as measured by IC30 and EMAX (Table 3.2, A and

B, respectively). Interestingly bFGF did not correlate with paclitaxel-induced apoptosis

(Table 3.2C). Pgp expression was the most important single parameter predictor of

paclitaxel induced apoptosis.

The r2 values for the correlation between the bFGF level and the paclitaxel effect

(IC30 and EMAX) were significant for both the pathological scoring and the image analysis

methods. (Table 3.2, A and B). The strength of the correlation using the image analysis

method was much improved over the pathological scoring method. This suggests that the

two methods are comparable and the image analysis method is more robust. The bFGF

48 level as measured by the pathological scoring method and image analysis method were

highly correlated (r2=0.39, p<0.0001). Direct comparison of the two methods was not

possible because the scoring method requires evaluation of three different antibody

dilutions while the image analysis is performed on only one common antibody dilution.

These results suggest that bFGF expression is broad spectrum resistance factor for

paclitaxel in a wide variety of tumor types. In comparison to classical resistance

mechanisms (pgp, p53) bFGF shows a stronger correlation suggesting intratumoral

growth factors (like bFGF) may play a more important role in chemoresistance and that

the development of molecular targeted therapies to these factors is warranted.

The quantitative nature of the new method and the wealth of supporting pathobiological parameters allowed for the investigation of possible resistant sub-groups within the tumor set. Sub-group analysis can identify co-factors important to bFGF resistance and can help to identify patient populations during the clinical trial process of molecularly targeted bFGF inhibitors (like suramin). The entire data set was divided into subgroups based on each of the pathobiological parameters (stage, grade, labeling index, and expression of aFGF, bcl-2, p53, and pgp) and the correlation between bFGF level as determined by image analysis and paclitaxel effect was reevaluated.

Three tumor subgroups showed an increased correlation between the paclitaxel effects and bFGF level as compared to the correlation found using the entire data set

(Table 3.3, Figure 3.7). Tumors that were either Stage 3-4, p53 positively stained, or aFGF negatively stained exhibited an increased correlation between the intratumoral bFGF level and the paclitaxel effects. Conversely any correlation between bFGF and effect was lost in the stage 1-2, p53 negatively stained, or aFGF positively stained tumor

49 samples. Dividing the tumor subset based on grade, pgp staining, or bcl-2 staining decreased the correlation in all cases. This suggests that p53 expression status, higher stage, and the absence of aFGF plays a role in the effect bFGF has on the paclitaxel effect.

Wild type p53 has a short half life and is present in very small quantities within the cell making it vary hard to detect with immunohistochemistry. Missense p53 mutations often increase the half-life of the protein and cellular accumulation of the protein to levels detectable by immunohistochemistry. Nonsense or frame-shift mutations would affect p53 functionality but would not result in the accumulation of the protein in the cell above wild type levels. Other mutations upstream of p53 could interfere with the feedback signaling that regulates p53 expression resulting in an increase in the cellular level. As a result, immunohistochemistry is not an absolute method for detection of p53 mutations, however it is clear that positive immunohistochemical results can be considered as an abnormality in the p53 system or a mutation in p53 itself [171].

Given this finding, the correlation between bFGF and p53 positive staining means that tumors with mutant p53 show a stronger correlation between bFGF level and paclitaxel effects. P53 is a very well known tumor suppressor and it is possible that wild type p53 masks the bFGF effect such that when it is mutated the bFGF resistance effects become apparent. Higher stage tumors are known to be more resistant, perhaps bFGF level plays a role in tumor progression. The finding that aFGF was involved in bFGF resistance was not surprising. However, the finding that the negative aFGF staining was correlated to effect suggests that presence of aFGF masks the bFGF effect and this result was unexpected.

50 Further evaluation of the sub-group result suggests that there is a threshold level

of bFGF that relates to the paclitaxel effect (Figure 3.7). The three of the tumor

subgroups are clearly separated into a sensitive group and a resistant group with the threshold value of ~100 ng of bFGF / mg total protein. The correlation between bFGF level and paclitaxel effect was lost if the entire data set was divided at the threshold value

(Table 3.3; threshold value set at 95 pg/mg). This suggests that the correlation of bFGF level and paclitaxel effect is not truly linear but appears as such because of the placement of the two subgroups within the data set. This is a valuable finding given the intended use of suramin because it means suramin need only reduce the level bellow the threshold level in order to show effect.

Overall the results from the paclitaxel group show that correlation between bFGF level and drug effect is maintained over many different tumor types. Higher tumor stage, mutant p53, and the absence of aFGF play a role in the effect. Finally there is a non- linear relationship between the bFGF level and the paclitaxel effect suggesting that the bFGF effect operates based on a certain threshold such that response depends more on which side of the threshold the level is on more so then the absolute bFGF level in the tumor.

3.3.3 Relationship between bFGF expression level, 5-fluorouracil effect, and suramin chemosensitization in the RCC clinical samples.

The aFGF and bFGF staining in the RCC tumor samples was very strong (Figure

3.8 A). The bFGF staining for 70% of the RCC clinical samples was above the linear portion of the standard curve making an accurate evaluation of the bFGF levels in these samples impossible. Therfore it was not possible to evaluate the correlation between the

51 intratumoral bFGF level and the 5-Fu IC50 data for the RCC clinical samples. Since

analysis of the original patient samples was not possible, the untreated control samples

from the histoculture experiment were stained for aFGF and bFGF and the staining result

for bFGF was within the linear portion of the standard curve. Example images of the

staining result for aFGF and bFGF in the RCC histoculture controls are shown in Figure

3.8 B.

The correlation between the bFGF level (as calculated by image analysis) in the

RCC histoculture controls and the 5-Fu IC50 values was analysis and no correlation was

found between the IC50 data and the bFGF level (data not shown). This lack of correlation

agrees with the previous findings from the paclitaxel treated samples. The aFGF staining

in the RCC histoculture control samples clearly showed that all the RCC histoculture

control samples exhibited positive aFGF staining. Results from the tumor subgroup

analysis in the 96 paclitaxel treated samples (see Table 3.3 and Figure 3.7) showed that

aFGF positivitly stained tumors exhibited a decreased correlation between bFGF level

and drug effect. Therfore the positive aFGF staining in the RCC histoculture samples

could account for the lack of correlation between 5-Fu IC50 and the bFGF level.

Next the correlation between the suramin chemosensitization effect and the bFGF level in the RCC histoculture samples was evaluated. Suramin effect was quantified as the fold chemosensitization; defined as the ratio of the IC50 of the 5-Fu and low dose

suramin (20 or 50 µM) combination divided by the IC50 of 5-Fu alone. The correlation

between the suramin effect and the bFGF level was evaluated in the same manner as the

above correlations. There was a very strong correlation between the bFGF level and the

degree of chemosensitization caused by the 20 µM suramin (Figure 3.8 C, r2 = 0.68, p-

52 value < 0.05). Interestingly the trend in 20 µM suramin samples showed that the correlation in the higher bFGF containing tumors was much tighter then in the lower bFGF containing tumors suggesting a stronger effect in the tumors with the higher bFGF levels. There was no significant correlation between the fold chemosensitization in the 50

µM suramin and the bFGF effects (Figure 3.8 C, r2 = 0.03, p-value = 0.59).

The results from the RCC histoculture control samples show that the correlation

between bFGF and drug response extends to drugs other then paclitaxel and that the

bFGF level strongly correlates to the amount of chemosensitization produced by low dose

suramin. This is the first evidence showing that suramin in working through bFGF in

RCC.

3.4 DISCUSSION

In the field of cancer research there is always a need for biomarkers that measure

drug effect or that can act as markers for target validation. The present study presents a

new method for quantifying protein biomarkers in immunohistochemically stained

clinical samples. The use of immunohistochemically stained sections has two main

advantages. First formalin fixed, paraffin embedded sections are the easiest to acquire

since most tissue archival protocols adopt this method and secondly, histochemical

staining preserves the 3-D structure of a sample and can therefore provide additional

spatial information often lost in other methods like ELISA or Western Blot. The

challenge that was overcome was the quantification of the often heterogeneous staining

for correlation to real protein concentrations. This was done through the application of

image analysis software and the creation of a novel bFGF standard.

53 Results from this new image analysis method were highly correlated with the

traditional method (pathological scoring) and had higher predictive value than the

traditional method in predicting the relationship between bFGF level and paclitaxel resistance. The results show that the intratumoral bFGF level was best single predictor of paclitaxel resistance over other pathobiological factors, including tumor grade, stage, and expression levels of pgp, p53, bcl-2, and aFGF in patient tumors. Interestingly, bFGF had highest predictive value in the tumors with late stages, positive p53 staining, and negative

aFGF staining. p53 is a known tumor suppressor and it is possible that wild type p53

could effectively mask the bFGF effects. Higher stage tumors are known to be more

resistant and interestingly chemotherapy plays a major role in the treatment of late stage

tumors and therefore bFGF could be a valuable predictor for the chemoresistance in these

tumors. The most surprising finding of the three subgroups is the stronger correlation

with the aFGF negative tumor samples. Results by Song et al. clearly show in vitro that

aFGF is capable of amplifying the effect of bFGF [2]. It is possible that in vivo the

presence of aFGF changes the pathobiological effects such that the resistance is not a

factor of bFGF level alone, but a combination of growth factors and conditions thus the

correlation with bFGF alone is decreased.

Another discovery that was only possible with the increased quantitative power of

the new assay was the suggestion that the response was not linearly related to the bFGF

level. The effect appears to have a threshold value such that the paclitaxel effect was

dictated more by which side of the threshold the bFGF level was on more so then the

absolute amount of bFGF present. However, the study in the 96 tumor set does clearly

54 show that the bFGF and paclitaxel effect is maintained over a wide variety of tumor types

thus showing it is a broad spectrum mechanism resistance as shown in the original in

vitro finding by Au, Wientjes et al. [2].

Addition of the RCC samples allow for two important additions to the study. The

involvement of suramin, and thus the first possible proof that suramin as acting through

bFGF in vivo and a new drug (5-Fu) to help widen the therapeutic implications of the

original paclitaxel findings. Results in the RCC samples did not show a correlation

between 5-Fu effect (IC50) and the bFGF level. This however is in line with the original paclitaxel findings. The aFGF levels in the RCC histoculture were within the criteria of the sub group analysis that showed a masking of the bFGF and drug effect correlation.

The results from RCC samples also show the chemosensitization effect of 20 µM and not

50 µM suramin was well correlated with bFGF level. This is in agreement with previous data showing that low and not high doses of suramin are capable of sensitizing tumors to chemotherapy in vivo [172;173]. The plot of the correlation (Figure 3.8 C) also supports the threshold bFGF level hypothesis proposed in the earlier study. The tumors with higher bFGF levels show a stronger correlation then the tumors with lower bFGF levels.

The most important implication of the suramin finding in RCC was that it is the first proof that suramin is working through bFGF in RCC.

This study is the beginning of the full investigation needed into the resistance mechanism of bFGF and the chemosensitization mechanism of suramin. bFGF is known to activate many downstream signaling cascades and the pluripotent nature of suramin leave many of avenues of exploration open. Fortunately the image analysis methodology can be extended to investigate many of the possible mechanisms of both suramin and

55 bFGF. For example, it has been suggested that induction of bFGF could represent a required drug resistance mechanism [169]. Histocultures could easily be treated with chemotherapy then analyzed via the current method to ask the question of which and how much of a certain type of drug could induce bFGF.

Another advantage of the image analysis software is the ability to analyze morphologic details within the stained images. In fact previous work from this lab developed a method to quantify and count stained nuclei [174-176]. Interestingly 20% of the tumors in the 96 patient set exhibited exclusively nuclear or cytoplasmic bFGF staining. Therefore the next possible iteration of the image analysis protocol could ask if the bFGF location plays a role in drug effect. This could be easily accomplished using the methods established by Weaver et al. and the bFGF level could be separated into nuclear and non-nuclear locations for further quantification or correlation to the pathobiological parameters.

The work outlined in this Chapter details a new quantitative method for measuring an intercellular biomarker, bFGF. This method was then applied to quantify the amounts of bFGF for the purpose of correlating that to resistance. The study was successful in that it shows clearly that bFGF in a valid clinical marker for predicting resistance. The study was extended to evaluate the relationship between bFGF level and the chemosensitizing effects of low dose suramin and found a very strong relationship between the bFGF level in the suramin 20 µM treated groups and the fold chemosensitization.

56 3.5 ACKNOWLEDGMENTS

This work was supported in part by the research grant R01 CA97067 from the

National Cancer Institute, NIH and by a research fellowship from the American

Foundation for Pharmaceutical Education. I would like to thank my advisor Dr. Jessie Au for her expertise, support, and patience. I am grateful for the collaboration with Dr.

Yuebo Gan and for his advisement and permission to use the previously reported data.

Special thanks go to Greg Lyness for his help and scientific input for the RCC portion of this study; and to Dr. M. Guill Wientjes, Dr. Yong Wei, and Dr. Liang Zhao for there technical advisement; and Dr. Ronnie Ortiz and Bei Yu for their help with the Optimas macro.

57

Number of tumors (% frequency) Tumor Type aFGF Staining Score bFGF Staining Score 01 2 0 1 2 Bladder 3 (19) 7 (44) 6 (37) 7 (44) 5 (31) 4 (25) Breast 11 (73) 4 (27) 0 (0) 6 (40) 7 (47) 2 (13) Head/Neck 14 (64) 7 (32) 1 (5) 14 (64) 7 (32) 1 (4) Ovarian 8 (47) 8 (47) 1 (6) 1 (6) 2 (12) 14 (82) Prostate 11 (42) 9 (35) 6 (23) 7 (27) 12 (46) 7 (27) Total 47 (49) 35 (36) 14 (15) 35 (37) 33 (34) 28 (29)

Table 3.1. aFGF and bFGF staining results in the paclitaxel treated patient samples as analyzed by the pathologic scoring method. Patient samples were immunostained for aFGF and bFGF using three different dilutions of primary antibody (1:50, 1:100,

1:200, for aFGF; and 1:20, 1:50, and 1:100, for bFGF) and scored based on the result of the three different dilutions.

58 2 Drug Effect (IC30) r AIC p bFGF (Image Analysis) 0.3051 198 <0.0001 bFGF (Pathological Scoring) 0.1159 220 <0.01 LI 0.0270 229 NS Stage 0.0266 229 NS Grade 0.0033 231 NS p53 0.0026 231 NS aFGF 0.0006 231 NS bcl-2 0.0005 231 NS Pgp 0.0005 231 NS bFGF (IA) + Stage 0.3197 198 <0.0001 bFGF (IA) + Stage + Grade 0.3343 198 <0.0001 bFGF (IA) + Stage + Grade + p53 0.3391 199 <0.0001

59 bFGF (IA) + Stage + Grade + p53 + aFGF 0.3425 201 <0.0001 bFGF (IA) + Stage + Grade + p53 + aFGF + bcl-2 0.3448 203 <0.0001 bFGF (IA) + Stage + Grade + p53 + aFGF + bcl-2 + pgp 0.3456 205 <0.0001 bFGF (IA) + Stage + Grade + p53 + aFGF + bcl-2 + pgp + LI 0.3462 206 <0.0001

Table 3.2. Correlation between pathobiological parameters and paclitaxel activity as measured by IC30. The correlation

between bFGF expression and the antiproliferative and apoptotic effects of paclitaxel were analyzed and compared with other

pathobiological parameters from a previous study [157]. Higher r2 values indicate a better correlation, analyzed by linear

regression using REG procedure in SAS. For the combination of two and more parameters, only the highest r2 was listed. IA,

image analysis; NS, not significant.

59 2 Drug Effect (EMAX) r AIC p bFGF (Image Analysis) 0.2601 540 <0.0001 bFGF (Pathological Scoring) 0.1644 551 <0.0001 Stage 0.1071 557 <0.001 LI 0.0965 558 <0.001 p53 0.0669 561 <0.01 Pgp 0.0410 564 NS Grade 0.0272 565 NS bcl-2 0.0241 565 NS aFGF 0.0202 566 NS bFGF (IA) + Stage 0.3435 531 <0.0001 bFGF (IA) + Stage + bcl-2 0.3624 531 <0.0001 bFGF (IA) + Stage + bcl-2 + LI 0.3775 530 <0.0001

60 bFGF (IA) + Stage + bcl-2 + LI + aFGF 0.3882 531 <0.0001 bFGF (IA) + Stage + bcl-2 + LI + aFGF + p53 0.3969 532 <0.0001 bFGF (IA) + Stage + bcl-2 + LI + aFGF + p53 +Pgp 0.3987 533 <0.0001 bFGF (IA) + Stage + bcl-2 + LI + aFGF + p53 +Pgp + Grade 0.3987 535 <0.0001

Table 3.3. Correlation between pathobiological parameters and paclitaxel activity as measured by EMAX. The correlation

between bFGF expression and the antiproliferative and apoptotic effects of paclitaxel were analyzed and compared with other

pathobiological parameters from a previous study [157]. Higher r2 values indicate a better correlation, analyzed by linear

regression using REG procedure in SAS. For the combination of two and more parameters, only the highest r2 was listed. IA,

image analysis; NS, not significant.

60 Apoptotic Effect r2 AIC p Pgp 0.2852 323 <0.0001 LI 0.2424 328 <0.0001 Grade 0.1002 344 <0.001 bcl-2 0.0453 349 <0.05 Stage 0.0424 349 <0.05 bFGF (Pathological Scoring) 0.0170 352 NS bFGF (Image Analysis) 0.0036 353 NS aFGF 0.0008 353 NS p53 0.0000 353 NS Pgp + LI 0.3941 310 <0.0001 Pgp + LI +Grade 0.4065 310 <0.0001 Pgp + LI +Grade + p53 0.4171 310 <0.0001

61 Pgp + LI +Grade + p53 +bcl-2 0.4219 311 <0.0001 Pgp + LI +Grade + p53 +bcl-2 + bFGF (IA) 0.4244 313 <0.0001 Pgp + LI +Grade + p53 +bcl-2 + bFGF (IA) + aFGF 0.4261 315 <0.0001 Pgp + LI +Grade + p53 +bcl-2 + bFGF (IA) + aFGF + Stage 0.4261 317 <0.0001

Table 3.4. Correlation between pathobiological parameters and paclitaxel activity as measured by apoptotic fraction. The

correlation between bFGF expression and the antiproliferative and apoptotic effects of paclitaxel were analyzed and compared

with other pathobiological parameters from a previous study [157]. Higher r2 values indicate a better correlation, analyzed by

linear regression using REG procedure in SAS. For the combination of two and more parameters, only the highest r2 was listed.

IA, image analysis; NS, not significant.

61

r2 Value Drug Effect Parameter aFGF Subgroup p53 Subgroup Stage Subgroup bFGF Threshold Entire Data Set Possitive Negative Negative Possitive Low Stage (1-2) High Stage(3-4) < 95 pg / mg > 95 pg / mg † † † † IC30 0.305 0.010 * 0.613 0.183 * 0.470 0.0002 0.507 0.0001 0.0958 † † † † EMAX 0.260 0.119 * 0.495 0.041 * 0.546 0.042 0.409 0.0034 0.121 Apoptosis 0.004 0.0007 0.024 0.106 * 0.067 0.002 0.004 0.0026 0.004 †, p-value < 0.0001; *, p-value < 0.05 62

Table 3.5. Correlation between bFGF level as measured by quantitative image analysis versus paclitaxel effect within the

tumor subgroups. The data set was subdivided based on each of the pathobiological variables or by the bFGF expression level

and the correlation between bFGF expression and the antiproliferative and apoptotic effects were analyzed by linear regression

using REG procedure in SAS.

62

Figure 3.1. Hue Saturation and Intensity histograms. A. DAB staining with hematoxylin counterstain, B. hematoxylin staining only, C. DAB staining only.

63 PC3 (39.0 pg bFGF / mg) FADU (60.0 pg bFGF / mg) HT29 (57.3 pg bFGF / mg) 64

MiaPACA (207.9 pg bFGF / mg) SKOV3 (466.2 pg bFGF / mg) Negative Control

Figure 3.2. Representative images of bFGF staining from the standard tumor set. Images shown with their respective

ELISA result (units of pg bFGF per mg of total protein), magnification x400 for all images.

64

0.4 ) 2 0.3

0.2

0.1 Negative Control PC3 FaDu HT29

Image Analysis Result Analysis Image 0.0 MiaPACA SKOV3 (Sum OD / Tissue Area, µm Area, / Tissue OD (Sum Fitted Curve

0 100 200 300 400 500

ELISA Result (pg bFGF)

Figure 3.3. Non-linear fit of the bFGF standard curve. The curve was fit with the equation, y = a*(1-bx) where y is the image analysis result, x is the ELISA Result, and a and b are constants equal to 0.3808 and 0.9941 respectively (r2=0.97, p<0.001). All data

points are mean ± SD. Some deviations are smaller then the data point.

65

0.4 ) 2 0.3

0.2

0.1 Negative Control PC3 FADU

Image Analysis Result Analysis Image 0.0 HT29 MiaPACA

(Sum OD / Tissue Area, µm Linear Fit

0 50 100 150 200 250

ELISA Result (pg bFGF)

Figure 3.4. Linear fit of the bFGF standard curve. Image analysis result = 0.001495 x

ELISA Result (r2=0.9979, p-value = 0.0007). All data points are mean ± SD. Some

deviations are smaller then the data point. This is the fitting used for all subsequent

patient sample analysis.

66 Head and Neck Breast Prostate 67

Bladder Ovarian

Figure 3.5. Representative images of the aFGF staining from the paclitaxel treated patient samples. Specimens of Bladder,

Breast, Head and Neck, Ovarian, and Prostate (n= 15, 14, 22, 17, and 23, respectively) were obtained via the Tumor

Procurement Service at The Ohio State University Comprehensive Cancer Center. Specimens were stained for aFGF using the

protocol detailed in the materials and methods, magnification x400 for all images.

67 Chapter 3 250

200

150

Head and Neck Breast Prostate 100 bFGF Level, pgpg Level, Level, bFGF bFGF 50

0 A. B. Head and Breast Prostate Bladder Ovarian

68 Neck Tumor Type Bladder Ovarian

Figure 3.6. Representative images of the bFGF staining from the paclitaxel treated patient samples (A) and results from

the quantatitive image analysis (B). Specimens of Bladder, Breast, Head and Neck, Ovarian, and Prostate (n= 15, 14, 22, 17,

and 23, respectively) were obtained via the Tumor Procurement Service at The Ohio State University Comprehensive Cancer

Center. Specimens were stained for bFGF using the protocol detailed in the materials and methods, magnification x400 for all

images (A). Images were captured and quantified for bFGF per image analysis (see materials and methods) and the average

bFGF per tumor type was calculated, mean and one SD is shown (B). 68 120 120 2 2 100 r = 0.260 100 r = 0.495 (%) (%) 80 80 MAX MAX 60 60

40 40

20 20

Drug Effect, E Drug Effect, 0 A. Drug Effect, E 0 B.

0 50 100 150 200 250 0 50 100 150 200 250 bFGF Level, pg bFGF Level, pg 120 120 2 2 100 r = 0.409 100 r = 0.546 (%) (%) 80 80 69 MAX MAX 60 60

40 40

20 20

Drug Effect, E Drug Effect, 0 C. E Drug Effect, 0 D.

0 50 100 150 200 250 0 50 100 150 200 250 bFGF Level, pg bFGF Level, pg

Figure 3.7. Plots of paclitaxel effect versus bFGF level for the three tumor subgroups. A. the complete data set (n=91,

r2=0.260, p-value < 0.0001), B. tumors with negative aFGF staining (n=47, r2=0.495, p-value < 0.0001), C. stage 3-4 tumors

(n=54, r2=0.409, p-value < 0.0001), D. tumors with positive p53 staining (n=36, r2=0.546, p-value < 0.0001).

69 RCC Patient Sample C. 10.0 A. 9.0 5-Fu + Suramin 20 µM 8.0 5-Fu + Suramin 50 µM 7.0 6.0 aFGF Staining bFGF Staining 5-Fu+Suramin5-Fu+Suramin 5.0 5050 4.0 RCC Histoculture Control 3.0 Chemosensitivity, Chemosensitivity, 5-Fu/IC5-Fu/IC 2.0 5050 1.0 ICIC 0.0 B. 0 50 100 150 200 250 bFGF level in the 5-Fu Histoculture Control Samples, pg bFGF 70 aFGF Staining bFGF Staining

Figure 3.8. aFGF and bFGF in the RCC patient samples (A) and the RCC Histoculture Samples (B) and the correlation

between chemosensitivity and the bFGF level in the RCC histoculture control samples (C). aFGF and bFGF staining was

carried out as described in the materials and methods. Magnification of all images is 400x. The correlation between the bFGF

level in the untreated histoculture controls and the fold chemosensitization (defined as the ratio of the IC50 of suramin 20/50 µM

/ IC50 of 5-Fu alone) were calculated using SAS in the same manner as previously described for the paclitaxel data. The

correlation for the suramin 20 µM group was significant (r2 = 0.68, p-value = 0.009) and the suramin 50 µM group was not

significant (r2 = 0.03, p-value = 0.59). 70 CHAPTER 4

IN VITRO INVESTIGATION OF THE ANTI-ANGIOGENIC MECHANISM OF

SURAMIN

4.1 INTRODUCTION

Au and Wientjes et al. have shown that the chemosensitization of suramin in vitro is through inhibition of bFGF-induced resistance [2]. Chapter 2 shows that this effect is paralleled in vivo. Chapter 3 showed that bFGF is a clinically relevant predictor for paclitaxel resistance and validated bFGF as a useful therapeutic target for suramin. The final part of this dissertation investigates the in vivo mechanism of suramin. As detailed

in Chapter 1, bFGF has been tightly linked to angiogenesis and conversely suramin has

been linked to anti-angiogenesis, therefore this mechanism presents the greatest interest

and should be investigated first. The literature strongly supports the possibility of an anti-

angiogenic mechanism of suramin therefore a positive result would align with the

literature. However confirmation of the hypothesis that the chemosensitization effect of

suramin is not due to its anti-angiogenic effect supports the bFGF-resistance reversal

approach presented in Chapter 2 as a treatment paradigm that is independent of anti-

angiogenesis and worthy of further development.

71 Suramin has been investigated thoroughly as an anti-angiogenic agent in a large number of studies and in all of the standard in vitro and in vivo assays reviewed previously (Chapter 1.3.4). These literature reports however did not consider the findings of Au and Wientjes et al. that low doses and not high doses of suramin can sensitize tumor cells in vitro [2] and solid tumors in vivo (Chapter 2). Given the extensive amount of literature data, many studies are not necessary, however there still remain some unanswered questions given the new low dose suramin chemosensitization findings of Au and Wientjes et al. The angiogenic mechanism investigation was carried out in both in

vitro (presented here) and in vivo (Chapter 5) models. The in vitro investigation was

carried out in two parts, the first tested if low or high dose suramin is toxic to or enhances

the cytotoxicity of chemotherapy to monolayer endothelial cells. The second was to

establish the chemosensitization of low dose suramin in model systems that do not

involve angiogenesis.

Some studies have shown that cytotoxics can produce antitumor activity by

damaging the tumor endothelial cells. Therefore, the first part of the study tested if the

chemosensitization of suramin enhances the toxicity of chemotherapy resulting in a

greater effect on the endothelial cell compartment. To test this hypothesis, the effects of

low dose suramin (alone and in combination with chemotherapy) were evaluated on

monolayer human umbilical vein endothelial cells (HUVEC). The HUVEC model was

chosen due to its acceptance in the literature, it is human in origin, and primary (non- passaged) cell cultures are commercially available. Ideally tumor-derived endothelial

cells would be the best choice however tumor endothelial cells are difficult to isolate and

their maintenance in culture have limited their use in preclinical studies [177].

72 The purpose of the second study was to confirm the findings of the in vitro and in

vivo studies (presented in Chapter 5). The tumor histoculture model was chosen for this

study, over endothelial tissue explants because of the large amount of literature already

published and this study is focused on the effects on the tumor vasculature and not

normal endothelium. Tumor histocultures retain many features of solid tumors, (i.e., 3-

dimensional multicellular structure, cell-to-cell communication, co-existence of epithelial

tumor cells and normal stromal tissue). However since histocultures are tumor fragments

cultured on a collagen gel matrix they do not contain a functional vasculature [178;179],

thus they are vascular independent.

Results from the monolayer study show that the IC50 for suramin but not paclitaxel were outside the concentration ranges that correlate to the in vivo plasma

concentrations suggesting that low dose suramin in vivo is not functioning as an anti-

angiogenic agent alone. The results also clearly show that low dose suramin does not

further sensitize the HUVEC to paclitaxel insult, suggesting that suramin is not further sensitizing the tumor endothelium in vivo to drug insult. Results in the histoculture system clearly show chemosensitization by low dose suramin is independent of the tumor

vasculature, further suggesting the sensitization mechanism is separate from the tumor

vasculature.

4.2 MATERIALS AND METHODS

4.2.1 Chemicals and reagents.

Paclitaxel was purchased from Yunnan Hande Bio-Tech Co. (Kunming, P. R.

China), suramin, MTT Powder, and phenazine methosulfate (PMS) was purchased from

Sigma Chemical Co. (St. Louis, MO), MTS was purchased from Promega (Madison,

73 WI), labeled streptavidin-biotin (LSAB) horseradish peroxidase kit was purchased from

Dako (Carpinteria, CA), 3,3’-diaminobenzidine tetrahydrochloride (DAB) enzyme substrate kit from BioGenex (San Ramon, CA), gentamicin from Solo Pak Laboratories

(Franklin Park, IL), EGM-2 endothelial growth medium was purchased from Cambrex

(East Rutherford, NJ), and all other tissue culture supplies from GIBCO (Grand Island,

NY). All antibodies used for immunohistochemical staining from BD Pharmingen (San

Diego, CA). All reagents used as received.

4.2.2 Cell culture.

Freshly isolated human umbilical veil endothelial cells (HUVEC) were purchased from Cambrex (East Rutherford, NJ) and were maintained in EGM-2 medium per manufacturer’s instruction. Human prostate tumor cells (PC3) were purchased from

American Type Culture collection (Rockville, MD) and were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 90 mg/ml gentamicin, 2 mM L- glutamine, and 100 units /ml penicillin. All cells were incubated at 37 °C in a humidified atmosphere containing 5% CO2.

4.2.3 Drug and reagent stock solutions.

Stocks for the monolayer HUVEC studies were prepared as follows. Paclitaxel

stock solution (1 mM) was prepared in DMSO (final DMSO content in medium was <

0.1). Suramin stock solution (200 mM) was prepared in basil medium (serum free). MTS reagent powder was dissolved in PBS (pH 7.4) to a stock concentration of 1 mg/ml and the pH was adjusted to between 6.0-6.5. PMS was dissolved in PBS (pH 7.4) to a stock

74 concentration of 2 mg/ml. MTS and PMS stocks were stored at -20 oC until use. At time

of use the MTS and PMS were thawed (in water bath) and mixed per manufacture instruction (50 µl PMS stock per 1 ml of MTS stock).

Drug solution preparation for histoculture study were prepared as follows.

Paclitaxel stock solution (10 mM) was prepared in dimethylsulfoxide (final DMSO content in medium was < 0.1%). Suramin stock solution (20 mM) was prepared in

purified H2O. MTT (2 mg/ml) was dissolved in serum free RPMI medium and stored at

-20 oC until use. All stock solutions were filtered with a 0.2 µm diameter filter before use.

4.2.4 Drug treatment and activity evaluation in the HUVEC monolayer model.

HUVEC are non-immortalized cells, therefore cells were cultured for no more

then 3 passages then discarded. HUVEC cells were harvested from sub-confluent cultures

using trypsin, and seeded at a density of 1,000 cells per well into 96-well plates and

allowed to equilibrate for 18 hours. After which time medium was removed and replaced

with drug containing medium. Control cells were processed similarly but with drug free

medium. Drug exposure was for 72 hours and was based on the doubling time of the cells

(22 hours).

The viable cell number was determined using the MTS reduction assay which is a

colorimetric method for determining the number of viable cells in culture. MTS is

bioreduced by dehydrogenase enzymes in viable cells into a formazan product that is

soluble in culture medium [180]. The quantity of formazan product as measured by the

amount of 490 nm absorbance is directly proportional to the number of living cells in

culture [181;182]. At the end of the 72 hour treatment, 40 µl of the MTS/PMS solution

was added to each well (1 µl MTS solution per 5 µl existing medium in the well). Each

75 96-well plate was then incubated for 3 more hours at 37 oC, after which time the

absorbance of each well was read at 490 nm using an EL340 microplate biokinetics reader (Bio-Tek Instruments, Inc. Winooski, VT). For each experiment blank readings

(wells containing medium and MTS solution only) were subtracted from the sample OD values to give the final result. Growth inhibition was expressed as a percentage of the treated and untreated control (Figure 4.2).

4.2.5 Evaluation of the interaction between paclitaxel and suramin in the HUVEC model.

The nature of the drug interaction was evaluated using fixed concentrations of suramin together with increasing concentrations of paclitaxel, i.e., the fixed-concentration method. This method was previously described by Johnston et al. [183]. The advantage of

this method is that it yields the conventional sigmoidal concentration effect curves

showing increasing effect as a function of increasing paclitaxel concentration. It also

provides a measure of the enhancement of the paclitaxel activity at a fixed suramin

concentration, however the design limits the number of suramin concentrations that can

be studied. Since the pharmacodynamics of the chemosensitization of low dose suramin

has been established and the concentration ranges are already known, this method is well

suited for this study.

All experiments included cells dosed with either paclitaxel (0-100 nM), suramin

(0-10,000 µM), or a combination of the two. For the combination groups, cells were

dosed with varying concentrations of paclitaxel (0, 0.01, 0.03, 0.1, 0.3, 1, 3, 10, 30, and

100 nM) plus a constant concentration of suramin (0, 0.1, 0.3, 1, 3, 10, 30, or 100 µM).

IC50 calculation and dose response curve characterization was accomplished by plotting

76 the % of untreated control versus the log of the respective paclitaxel concentration. The

data was regressed non-linearly in SAS in order to calculate the IC50 values. Each

experiment allocated 3 wells per drug concentration and each experiment was performed

on at least three separate occasions.

The nature of the interaction between paclitaxel and suramin was analyzed using

the combination index method [184]. Concentration effect curves for the paclitaxel and

suramin single agents and their combinations were used to determine the amount of each

agent, either alone or in combination, needed to achieve a given level of effect. The

combination index (CI) was calculated as follows,

CI = (ICPacComb / ICPac) + (ICSurComb / ICSur) (1)

where ICPac and ICSur are the concentrations of paclitaxel and suramin needed to produce

a given level of cytotoxicity when used alone, whereas ICPacComb and ICSurComb are the concentrations needed to produce the same effect when used in combination. A combination index value equal to 1 indicates additive interaction, values <1 indicate synergistic action, and values > 1 indicate antagonistic interaction [185].

4.2.6 Histoculture drug response assay.

Male athymic nude mice (5 weeks old) were purchased from the National Cancer

Institute (Bethesda, MD), housed in air-filtered laminar flow cabinets and cared for in accordance with institutional guidelines. PC3 cells were harvested from sub-confluent cultures using trypsin, and suspended in serum free RPMI medium and injected subcutaneously into the flank on both sides of a mouse (2x106 cells/200 µl per injection

site). Tumors were then allowed to grow for 4-6 weeks until tumor size was ~ 1 cm in

diameter at which time the tumor was excised and cut into 1 mm3 pieces. During the

77 cutting process any obvious non-tumor tissue (i.e., stromal tissue) was removed. Tumor

histoculture pieces were weighted, randomized, and seeded on 1 cm2 pieces of pre-

hydrated collagen gel. Tissue specimens were cultured in 6-well plates, containing RPMI

o medium, in a humidified atmosphere containing 5% CO2 at 37 C for 48 hours to allow

the specimens time to equilibrate.

At the start of treatment, culture medium was exchanged for drug containing

medium and treatment continued for 6 additional days with medium changes every 3

days. At the end of the 10 day schedule all histoculture specimens were either fixed and

embedded for immunohistochemical staining or the relative amount of cellular activity

was determined using the 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyl-2Htetra-zolium

bromide (MTT) assay. MTT methods were adopted from the literature with modification

[57;186;187]. Briefly, 0.5 ml of MTT stock solution (2 mg/ml MTT prepared in serum

free RPMI medium) was added to each well (final MTT concentration 0.1 mg/ml) and the

histocultures were incubated (at 37 oC) overnight (16-20 hours). Next, the gel/tumor

samples were transferred to 1.5 ml epindorf tubes, centrifuged (10,000 rpm for 5 min),

and 0.5-1 ml of DMSO was then added to each tube to extract the MTT-formazan. After

2 hours, 100-250 µl of the extracted solution from each tube was placed in a 96-well plate

and the absorbance was read at 570 nm with an EL340 microplate biokinetics reader

(Bio-Tek Instruments, Inc. Winooski, VT).

Histoculture pieces not evaluated via MTT, were fixed and embedded for immunohistochemical staining. Additional untreated control samples were added for immunohistochemical evaluation to observe any morphologic changes in the vasculature that might occur from the histoculture process. These controls were not part of the MTT

78 assay and included samples collected prior to gel seeding (i.e. day 0, to observe basal vessel morphology) and before the start of treatment (i.e., day 3, to evaluate any changes associated with the culturing). Fixation methods adopted from Beckstead et al. with modification [188]. Briefly, samples were fixed in a solution of zinc acetate, zinc chloride, and Tris-Calcium acetate buffer for 40 hours (0.5g Calcium Acetate, 5.0g Zinc

Acetate, and 5.0g Zinc Chloride dissolved in one liter 0.1M Tris buffer, final pH 6.5-7.0).

Samples were then dehydrated in graded ethanol, cleared in xylene, paraffin embedded, and subsequently sectioned into 5 µm sections for immunohistochemical analysis.

4.2.7 Immunohistochemistry.

Tumor vessels were detected using a monoclonal rat anti-mouse antibody to CD-

31 (clone MEC13.3), which specifically recognizes a 130 kDa trans membrane glycoprotein who expression is almost exclusively limited to endothelial cells and platelets (with the exception of certain subsets of myeloid cells) [189;190].

Immunohistochemical methods were adopted from Lee et al. with minor modifications

[191]. Briefly, after de-waxing and rehydration sequentially in xylene, ethanol and water, slides were washed in PBS (pH 7.4), blocked with protein blocking solution (LSAB kit,

DAKO) for 10 min, and incubated for 1 h with a CD-31 antibody solution (1:25 dilution in 0.5% BSA in PBS, pH 7.4). Slides were then washed in PBS (pH 7.4), incubated for

30 min with a biotinylated goat anti-rat Ig specific polyclonal antibody (1:100 dilution in

0.5% BSA in PBS, pH 7.4), washed in PBS (pH 7.4), incubated for 30 min with streptavidin-conjugated horseradish peroxidase (LSAB kit, DAKO) and then incubated

79 with the DAB chromogen for 2 min. Slides were subsequently counterstained with hematoxylin. Negative controls were performed by substituting rat antibodies of the same isotype as MEC 13.3 for the primary antibody.

4.2.8 Development of an image analysis method to quantify the morphologic changes in the vessel morphology.

Quantification of the morphologic changes in the vasculature of the histocultures was very import because the methodologies established in this study are applied to the in vivo studies presented in Chapter 5. Selection of the vascular areas within the tumor histoculture section was accomplished using the previously reported “hotspot” method

[192;193]. The rationale for employing the “hot spot” method was based on preliminary findings that there existed a degree of tumor heterogeneity, not only within groups but also between tumor groups. In order to make a fair comparison between groups the tumor hot spots were selected [192;193]. Briefly, tumor sections were visually scanned at low magnification (100x) and areas of dense vascularization were identified. A higher magnification (200x) was then used to capture images of these dense regions for image analysis. Histoculture samples were photographed using a Hamamatsu (Hamamatsu-City,

Japan) color chilled 3CCD camera attached to a ZEISS (Thornwood, NY) Axiovert 35 microscope. All images were saved as TIF format images and 1-6 (average of 2) images were photographed and analyzed per histoculture piece per group. The low image number was attributable to inherent small size of the histocultures and the lack of vasculature. At

200x magnification the field size of an image was 0.204 mm2.

80 The Optimas® (version 6.51; Media Cybernetics Silver Spring, MD) image

analysis software discussed in Chapter 3 was used to quantify changes in the vessel

number and morphology. Stained vessels were outlined using Optimas® and the vessel

density, average vessel area, and average vessel length were calculated to evaluate the

effects of the different treatments on the tumor vasculature. Microvessel density was

included as a parameter because it was the first quantitative parameter used to measure

angiogenesis. First developed by Brem et al. in 1972, the measurement of microvessel

density [194], particularly within regions of high vessel concentration, has proven to be a

useful prognostic indicator in prostate and breast carcinomas [193;195;196] as well as a

useful measurement for the assessment of disease stage, metastatic potential,

reoccurrence, and survival [197;198].

The final variables used in the quantification were defined as follows. The

average vessel density was defined as the number of counted vessels within a given field divided by that field’s size (units of, number of counted vessels / field area in mm2).

Vessel area was calculated by averaging the vessel cross sections within the image field

(units of µm2). Due to the large differences in vessel geometry, the vessel length was

calculated by sampling points along the vessel boundary and measuring the distance

between the two boundary points that where the farthest apart, thus this distance is

therefore a representative value for vessel length (units of µm).

81 4.2.9 Statistical analysis.

Statistical differences between vessel parameters were evaluated using Tukey’s

multi-comparison analysis after one-way ANOVA. A p-value of less than 5% was

considered significant. ANOVA calculations were performed using SAS® (The SAS®

Institute Inc. Cary, NC). P-values comparing the suramin effects in the histoculture model were calculated using Students t-test.

4.3 RESULTS

4.3.1 Effects of suramin and paclitaxel in the HUVEC monolayer model.

The dose response curves for suramin alone (Figure 4.1) and in combination with paclitaxel (Figure 4.2 A and B) demonstrate that low doses (<10 µM) of suramin had

little effect on the HUVEC cells and that the combination of these suramin doses did not

significantly alter dose response compared to paclitaxel alone. Higher doses of suramin

(>30 µM) show a large degree of growth inhibition in the HUVEC model, 30% and 70%

in the case of 30 µM and 100 µM suramin, respectively. These results are summarized in

Table 4.1 and clearly show no shift in the IC50 values for the suramin 10 µM combination as compared to the IC50 for paclitaxel alone. The Combination Index (CI) values at the

50% effect level are summarized in Table 4.2. All of combinations yielded a CI value of approximately 1 at the 50% effect level thus indicating additivity between paclitaxel and suramin and not synergism or antagonism.

Overall the monolayer result is against the possibility of an anti-angiogenic mechanism of suramin. The IC50 of suramin alone was 56 µM which is above the target

plasma concentration for the low dose suramin schedule used in vivo (Chapter 2) thus

ruling out a direct suramin effect. The IC50 of paclitaxel alone (3.2 nM) is low enough to

82 suggest that paclitaxel alone may exhibit an effect on the vasculature in vivo, however the results clearly show no synergistic interaction between low dose suramin (10 µM) and paclitaxel. Therefore there is no evidence to support the hypothesis that suramin could amplify the effect of the chemotherapy on the endothelium.

4.3.2 Chemosensitization in PC3 histocultures.

The results of the chemosensitization effect of suramin in the PC3 histocultures is shown in Figure 4.3. Drug concentrations for paclitaxel (10 nM) and suramin (25 µM and

200 µM) where chosen to reflect the plasma concentrations used in the clinic and the in vivo study (presented in Chapter 5) [199] while suramin (50 µM) was chosen to better illustrate the bi-phasic chemosensitization effect of suramin [169;173]. Paclitaxel (10 nM) produced concentration-dependent cytotoxicity in PC3 tumor histocultures, while low concentrations of suramin (25 µM) did not produce any cytotoxicity, whereas higher concentrations (50 µM, 200 µM) produced 20-50% cytotoxicity (Figure 4.3).

The results in Figure 4.3 further showed that low suramin concentrations (< 50

µM) enhanced the antitumor activity of paclitaxel activity in PC3 tumor histocultures while, in contrast, high suramin concentrations (>50 µM) did not produce chemosensitization. In fact, the antitumor activity of combinations of paclitaxel/suramin

50 µM was inferior to paclitaxel alone. These results indicate that chemosensitization occurred only for low and not high suramin concentrations, and were consistent with previous findings (Chapter 2) [2;173;200].

83 4.3.3 No anti-angiogenic effects by paclitaxel or suramin in the PC3 tumor histocultures.

Results from quantitative image analysis show that the vessels present at the start of the study (day 0 control) were not sustained throughout the 10 day histoculture treatment (Table 4.3, Figure 4.4). After the start of the histoculture treatment the vessel density is significantly reduced by 2-3 fold (p<0.05). Vessel area and length remained unchanged throughout the duration of the treatment. These results show that the histoculture system was not capable of supporting the tumor vasculature after culture initiation. This loss of vasculature had no effect on the chemosensitization effect of low dose suramin, thus showing the two are independent of one another.

4.4 DISCUSSION

The purpose of the experiments in this chapter was to investigate the possibility of an anti-angiogenic mechanism of suramin to explain the established chemo-sensitization effect of low dose suramin presented in Chapter 2. The in vitro data presented in this chapter the IC50 of suramin alone is above the low dose range and the combination of low dose suramin with paclitaxel did not shift the IC50 value as compared to paclitaxel alone.

Analysis of the interaction between all combinations of paclitaxel and suramin show additivity and not synergy. This finding is in contrast to the findings by Au and Wientjes et al. that paclitaxel/suramin are synergistic in monolayer tumor cells and this result is paralleled in vivo (Chapter2) [2].

It should be noted that previous data from the Au and Wientjes group show that suramin is heavily protein bound (>99%) and therefore the free suramin concentrations in the monolayer HUVEC model (medium containing 2% FBS) would be higher then they

84 would be in a traditional tumor monolayer system (medium containing 10% FBS) [201].

This suggests that the calculated IC50 values and the suramin effects are slightly lower then they would be in a model that used 10% FBS. The use of 10% FBS medium was not possible however since the medium used with the HUVEC was made specifically for these cells by the company.

The monolayer results were confirmed using an ex-vivo histoculture assay. The results show that the chemosensitization effect by low dose suramin was independent of the effects to the vasculature. Interestingly, the vessels regressed in all groups irrespective of treatment regiment. This could be that survival signals normally present in the functional vasculature of the tumor that was not present in the histoculture system. As reviewed in Chapter 1, sheer stresses, and hemodynamic forces provide a survival signal to the endothelium and since the vasculature in this system is inactive, this survival signal is absent (Chapter 1) [113;114]. Furthermore, the histoculture pieces are approximately 1 mm in size (via experimental design). This size is below the threshold “angiogenic switch” size and therefore the tumor my have decreased the expression of key angiogenesis promoters (i.e. growth factors, chemo attractants etc.) [116]. Due to the dominant regression of all the vasculature, there could still be a less dominant effect that was masked, however this work will be further validated in vivo were such questions are addressed.

The above results do not contradict the previously reported literature findings showing that suramin is anti-angiogenic. For example, Takano et al. show that suramin can inhibit bFGF binding to monolayer bovine capillary endothelial cells (BCE) at concentrations of 60 µM, however it requires much higher concentrations (150-170 µM)

85 to inhibit bFGF stimulated migration, proliferation, and u-PA production in BCE cells

[61]. These concentrations are well above the established chemosensitization effects of

low dose suramin.

Studies involving VEGF and suramin effects on endothelial cells show that 2 µM

suramin can inhibit VEGF mediated activation of KDR by 50% in porcine aortic

endothelial cells (PAE) however down regulation of the KDR receptor only occurred at >

200 µM suramin. Endothelial cells, stimulated by 3 ng/ml VEGF, and showed that only suramin concentrations > 100 µM were able to reduce cell proliferation to a level below baseline. Endothelial cell migration using 10 ng/ml VEGF as a stimulus showed that

VEGF could increase migration 7 fold but suramin concentration as high as 200 µM could not completely reverse this stimulation, i.e., 200 suramin µM was still 2 fold greater then the 0 ng VEGF control [110].

Ex vivo tissue explant studies were excluded from this study because of the large number of studies already present in the literature. For example Bocci et al. demonstrate in the ex vivo rat aortic ring assay that 50 µg/ml suramin (35 µM) is significant to inhibit both endothelial cell proliferation and migration [58]. Stiffey-Wilusz et al. using porcine carotid artery explants show that endothelial cell sprouting is inhibited by 50 µg/ml suramin (the authors calculated an EC50 of 24 µg/ml, or 16.8 µM) [111]. Brown et at.

found that vessel sprouting from explants of human placental blood vessels could be

inhibited by 27% by 10µg/ml (7 µM) suramin and 93% by 100 µg/ml (70 µM) suramin

[202]. Although the suramin ranges reported by these studies are similar to the low dose

suramin ranges that produce sensitization it should be noted that these assays are derived

for normal tissues and are in vitro assays. Therefore these literate results suggest the

86 further need for the in vivo studies in tumor derived models in order to fully ascertain the anti-angiogenic nature of suramin. That is why the experimental design included an in

vivo portion (in vivo studies presented in Chapter 5).

These results suggest that the bFGF inhibitory effects of low dose suramin are

independent of the anti-angiogenesis effects detailed in the literature, thus distinguishing

the Au/Wientjes hypothesis as an independent paradigm for investigation and clinical

development of suramin. This distinction is important because the two approaches of

bFGF inhibition and anti-angiogenesis differ in principle and therefore their practice.

Anti-angiogenic therapy, because the purpose is to inhibit the formation of new vessels,

should be administered continuously in duration that is much longer than is customary for

cytotoxic therapy. In contrast, inhibitors of FGF resistance, because the purpose is to

cause cytotoxicity to tumor cells, should be administered with the cytotoxic therapy.

In addition, while there are intense research interests on using anti-angiogenic

agents to enhance the efficacy of chemotherapy, the clinical usefulness of this approach is

still an open question since only one of many anti-angiogenic compounds (the anti-

VEGFR antibody AvastinTM) shows therapeutic benefits [73]. Verification of the

favorable clinical results of suramin and confirmation of the hypothesis that the

chemosensitization effect of suramin is not due to its anti-angiogenic effect supports the

FGF-resistance reversal approach as a treatment paradigm that is independent of anti-

angiogenesis and worthy of further development.

87 4.5 ACKNOWLEDGMENTS

This work was supported in part by the research grant R21 CA91547-02 from the

National Cancer Institute, NIH and by a research fellowship from the American

Foundation for Pharmaceutical Education. I would like to thank my advisor Dr. Jessie Au

for her expertise, support, and patience. Special thanks go to Dr. M. Guill Wientjes, Dr.

Yong Wei, and Dr. Liang Zhao for their technical support and scientific input in the

study.

88

Treatment Group IC50 values (Mean ± SD) Paclitaxel 3.24 ± 0.59 nM Suramin 56.6 ± 14.5 µM Paclitaxel + 0.3 µM Suramin 3.18 ± 0.39 nM Paclitaxel + 1 µM Suramin 3.47 ± 0.46 nM Paclitaxel + 3 µM Suramin 2.95 ± 0.42 nM Paclitaxel + 10 µM Suramin 3.49 ± 0.07 nM Paclitaxel + 30 µM Suramin 1.31 ± 0.80 nM Paclitaxel + 100 µM Suramin --- ± --- nM

Table 4.1. Pharmacologic effects of paclitaxel and suramin in the HUVEC model.

Human umbilical veil endothelial cells were treated with paclitaxel in the presence or absence of suramin. HUVEC were treated with paclitaxel (0.001 to 100 nM) and the suramin (0.3, 1, 3, 10, 30, and100 µM). Drug activity was measured by the MTS assay.

IC50 calculation of the suramin 100 µM combination was not possible do to the significant amount of growth inhibition that occurs at the 100 µM suramin concentration.

Data represents the mean ± SD for three experimental repeats.

89

Treatment Group CI at the IC50 effect level (Mean ± SD) Paclitaxel + 0.3 µM Suramin 1.034 ± 0.042 Paclitaxel + 1 µM Suramin 1.091 ± 0.064 Paclitaxel + 3 µM Suramin 0.987 ± 0.076 Paclitaxel + 10 µM Suramin 1.162 ± 0.175 Paclitaxel + 30 µM Suramin 0.966 ± 0.194 Paclitaxel + 100 µM Suramin --- ± ---

Table 4.2. Combination Index values (CI) at the IC50 effect level for paclitaxel and

suramin in HUVEC model. Human umbilical veil endothelial cells were treated with

paclitaxel in the absence or presence of suramin. Interaction between paclitaxel and

suramin was determined using the fixed concentration method, where the paclitaxel

concentration was varied (0.001 to 100 nM) and the suramin concentration was kept constant (0.3, 1, 3, 10, 30, and 100 µM). Drug activity was measured by the MTS assay.

CI values were calculated for each of the paclitaxel suramin combinations. Calculation of the CI value for the 100 µM suramin condition was not possible due to the significant amount of growth inhibition at 100 µM suramin. A combination index of <1 indicates synergy and >1 indicated antagonism. Tabulated results are the average of 3 different experiments, mean ± SD.

90

Treatment Group Average Vessel Density Average Vessel Average Vessel (Number of Histocultures) (Vessel Number/mm2) Area (µm2) Length (µm)

Day 0 Control (n=10) 133.3 ± 49.5 181.7 ± 68.4 25.1 ± 4.8 Day 3 Control (n=10) 61.2 † ± 19.3 228.1 ± 129.3 28.5 ± 10.2 Day 10 Control (n=10) 47.4 † ± 34.7 319.9 ± 179.4 33.1 ± 12.0 Suramin 25 µM (n=10) 50.8 † ± 34.8 312.6 ± 138.2 33.8 ± 8.2 Suramin 50 µM (n=10) 46.1 † ± 34.8 293.9 ± 257.5 32.2 ± 13.5 Suramin 200 µM (n=10) 49.8 † ± 31.2 304.4 ± 313.7 33.9 ± 22.4 Paclitaxel 10 nM (n=10) 52.3 † ± 31.3 455.9 ± 241.8 42.5 ± 10.2 Combination 25 µM (n=10) 43.2 † ± 22.2 360.2 ± 132.4 36.4 ± 8.6 91 Combination 50 µM (n=10) 40.5 † ± 30.9 240.4 ± 132.5 29.1 ± 7.0 Combination 200 µM (n=10) 51.7 † ± 36.0 321.8 ± 195.3 30.2 ± 12.9 † p-value <0.05 verses the Day 0 Control.

Table 4.3. Summary of morphologic vessel data obtained using image analysis for the PC3 histoculture model. Average

vessel density, area, and length were calculated using a quantitative image analysis system. Values are mean ± SD. Statistical

analysis was performed using ANOVA with post hoc Tukey analysis of group differences.

91

120

100

80

60

40

Viable Cells, % of Control 20

0 0.01 0.1 1 10 100 1,000 10,000 100,000

Suramin Concentration, µM

Figure 4.1. Dose response for suramin in the HUVEC model. Cytotoxicity of suramin for the human umbilical veil endothelial cells. Results of a represent experiment. Mean ±

SD are shown (n=3). Some standard deviations are smaller then the symbols.

92 120

A. 100

80

60

40

20 Viable Cells, % of Untreated Control of Untreated % Cells, Viable 0 0.0 0.01 0.1 1 10 100 Paclitaxel Concentration, nM

120

B. 100

80

60

40

20

0

Viable Cells, % of Suramin Treated Control Treated of Suramin % Cells, Viable 0.0 0.01 0.1 1 10 100 Paclitaxel Concentration, nM

Figure 4.2. Cytotoxicity of the combination of paclitaxel and suramin. A. Data

plotted as % of untreated control. B. Data plotted as % of suramin treated control. Results of a represent experiment, mean ± SD are shown (n=3), some deviations are smaller then

the symbols. Varying amounts of paclitaxel (0.001 to 100 nM) were combined with

increasing fixed concentrations of suramin: paclitaxel alone (●); paclitaxel plus 0.3 µM

suramin (○); 1 µM suramin (▼); 3 µM suramin (V); 10 µM suramin (■); 30 µM suramin

(□); 100 µM suramin (♦).

93

120 ‡ - 10 nM Paclitaxel 100 + 10 nM Paclitaxel

80 † 60 % Control 40

20

0 0 µM Suramin 25 µM Suramin 50 µM Suramin 200 µM Suramin

Figure 4.3. Chemosensitization in the PC3 histocultures by low dose suramin.

Results of the MTT assay for the PC3 histoculture from a representative experiment.

Mean and one SD shown (n=4). Tumor pieces were cultured for 7 days with suramin (25,

50, 200 µM) alone and in combination with paclitaxel (10 nM). P-values comparing the combination groups (†, p-value 0.065; ‡, p-value 0.0112) versus the paclitaxel alone group were calculated using Students t-test.

94

95 Day 0 Control Day 3 Control Day 10 Control

Figure 4.4. Immunohistochemical staining of CD-31 in the PC3 histoculture control samples. Vessel staining in the day 0,

3, and 10 day histoculture controls. Magnification of all images are 200x.

95 CHAPTER 5

IN VIVO INVESTIGATION OF THE ANTI-ANGIOGENIC MECHANISM OF

SURAMIN

5.1 INTRODUCTION

Au and Wientjes et al. have shown that the chemosensitization of suramin in vitro is through inhibition of bFGF-induced resistance [2]. Chapter 2 shows that this effect is paralleled in vivo. The studies presented in the previous chapter show that suramin was not capable of inducing an anti-angiogenic effect alone or in combination with chemotherapy in an in vitro monolayer HUVEC system. Because monolayer cultures lack the 3-dimensional structure and tumor-stromal interaction that occur in vivo, the suramin effect was further evaluated in an ex vivo histoculture system and the resultant effect on the vasculature supported the monolayer result. Therefore, the final goal of the mechanism study started in Chapter 4 is to translate the in vitro findings to a clinically relevant in vivo model to confirm or deny a possible anti-angiogenic mechanism of suramin as part of the clinical translation of suramin.

The purpose of this chapter is to determine if suramin or chemotherapy alone produce any anti-angiogenic effects in vivo and investigate if low dose suramin can enhance any possible anti-angiogenic effects of chemotherapy. The literature strongly supports the possibility of an in vivo anti-angiogenic mechanism for both suramin and

96 chemotherapy therefore a positive result in both cases would align with the literature.

However a result showing that suramin alone or in combination with chemotherapy showed no effect regardless of the angiogenic chemotherapy effects would support the

FGF-resistance reversal approach presented in Chapter 2 as a treatment paradigm that is independent of anti-angiogenesis and worthy of further development.

As reviewed in Chapter 1, both suramin and chemotherapy have demonstrated anti-angiogenic potential in many in vivo normal and tumor tissue derived assays.

Suramin 30 mg/kg per day was able to inhibit neo-angiogenesis in the rat mesentery [58], suramin 1.6 mg/eye per day or 200 mg/kg given iv was able to suppress neo-angiogenesis in the rat cornea [58;61], and lastly suramin 200 mg/kg and not 100 mg/kg was anti- angiogenic in mice bearing M5076 murine reticulosarcoma xenografts [60]. However in these reports suramin was investigated using doses comparable to the clinical maximum tolerated dose (100-200 µM). The results of Au and Wientjes et al. and those presented in

Chapter 2, clearly show that low and not high doses of suramin are responsible for chemosensitization. Results from Chapter 4 show in vitro evidence against a possible anti-angiogenic mechanism. The finial part of this mechanism investigation is to evaluate the effects both suramin, chemotherapy, and their combination on the tumor vasculature in clinically relevant in vivo model.

For this study two different subcutaneous tumor models (PC3 and HT29) were evaluated. Both showed the low dose suramin chemosensitization effect established in

Chapter 2. To insure the clinical relevance of the models, chemotherapy was initiated only after the tumors were well established (i.e., >3 mm in diameter). The PC3 model included both low and high doses of suramin (10 and 200 µM, respectively) both alone

97 and in combination with paclitaxel. The purpose of the high suramin dose was to parallel the maximum tolerated dose schedule used in previous clinical trials (see Chapter 1) and the purpose of the low doses of suramin were to parallel the established chemosensitization effect (see Chapter 2). PC3 tumors were immunohistochemically stained using a monoclonal rat anti-mouse antibody to CD-31 per the established method outlined in Chapter 4. The quantitative image analysis method developed in Chapter 4 was then used to calculate the average vessel density, area, and length based on the immunohistochemical result (see Chapter 4).

A second tumor model (HT29) of different origin, and using a chemotherapeutic

(CPT-11) with a different mechanism of action, was added to test the broad spectrum applicability of the possible anti-angiogenic mechanism for suramin. Preliminary results from the PC3 model showed that single agent chemotherapy appeared to damage the tumor vasculature. The established immunohistochemical staining protocol did not distinguish the functional and non-functional vessels, therefore a fluorescent dye,

DiOC7(3), was given i.v. to the HT29 baring animals prior to tumor excision [203], to label the functional vessels at the time of animal sacrifice. Tumor sections containing the dye were photographed then immunostained for CD-31 in combination with a fluorescent label (Alexa Fluor 647) and then re-photographed. Quantification of the amount of the percent functionality was accomplished by comparing the fluorescent areas of the two images.

Morphologic results from both models (PC3 and HT29) show that neither low or high dose suramin, as single agent, altered the tumor vessel density, length, or area as compared to untreated control. Chemotherapy (paclitaxel and CPT-11) alone altered the

98 morphology of the tumor vasculature showing a significant increase in vessel density and a significant decrease in vessel area as compared to untreated controls however the addition of low or high dose suramin to chemotherapy did not significantly change the vessel density, area, or length as compared to chemotherapy alone. This is in agreement with the in vitro results in Chapter 4.

Quantification of the vessel functionality in the HT29 model showed that the percent functionality of the tumors treated with CPT-11 alone or in combination with suramin was significantly increased as compared vehicle control. The percent functionality of the tumors that received suramin alone or in combination with CPT-11 were not significantly different from vehicle control or CPT-11, respectively. This result is in line with the morphological data and further illustrates that low dose suramin is not acting through an anti-angiogenic mechanism. Interestingly tumors treated with CPT-11 showed a significant increase in vessel functionality and this effect was independent of suramin effect. Furthermore the functionality differences induced by chemotherapy were maintained for 1.5 mounts after CPT-11 treatment ceased and as the tumor was allowed to re-grow. This suggests that the effects of chemotherapy on the tumor vasculature are not transient in nature and could present a new avenue of study.

5.2 MATERIALS AND METHODS

5.2.1 Chemicals and reagents.

CPT-11 (clinical preparation, CAMPTOSAR®) was obtained from Pharmacia &

UpJohn (Kalamazoo, MI), irinotecan hydrochloride (CPT-11) was purchased from

PolyMed Therapeutics Inc. (Houston, TX), paclitaxel was purchased from Yunnan Hande

Bio-Tech Co. (Kunming, P. R. China), and suramin was purchased from Sigma Chemical

99 Co. (St. Louis, MO). DiOC7(3), Streptavidin-Alexa Fluor 647, and Image-iT™ FX Signal

Enhancer were purchased from Molecular Probes (Eugene, OR). All antibodies used for immunohistochemical staining were purchased from BD Pharmingen (San Diego, CA).

Lactic Acid was purchased from Fisher Chemicals (Fair Lawn, NJ), Sorbitol was purchased from Mallinckrodt Baker (Pillipsburg, NJ), labeled streptavidin-biotin (LSAB) horseradish peroxidase kit from Dako (Carpinteria, CA), 3,3’-diaminobenzidine tetrahydrochloride (DAB) enzyme substrate kit from BioGenex (San Ramon, CA), cefotaxime sodium from Hoechst-Roussel (Somerville, NJ), gentamicin from Solo Pak

Laboratories (Franklin Park, IL), and all other tissue culture supplies from GIBCO

(Grand Island, NY).

5.2.2 Drug dosing solution preparation.

Paclitaxel stock solution (15 mg/ml) was prepared in Cremophor EL and ethanol

(50:50). Suramin stock solutions (1.1 mg/ml, 14.4 mg/ml, and 22.2 mg/ml) were prepared with PBS (pH 7.4) or in normal saline. CPT-11 (clinical preparation, CAMPTOSAR®) was reconstituted to 20 mg/ml in normal saline per manufacturer instruction. When the clinical preparation was not available, irinotecan hydrochloride was prepared to mimic the clinical preparation using the method established by Hardman et al. Drug stock was made such that each milliliter of solution contained 20 mg of irinotecan hydrochloride, 45 mg of sorbitol NF powder, and 0.9 mg of lactic acid, USP in normal saline [204]. The pH of the solution was adjusted to 3.5 (range, 3.0 to 3.8) with sodium hydroxide or hydrochloric acid. DiOC7(3) was dissolved in DMSO at a concentration of 10 mg/ml.

Drug solutions were filtered with a 0.2 µm diameter filter before use.

100 5.2.3 Cell culture.

Human prostate PC3 and human colon HT29 tumors cells were purchased from

American Type Culture collection (Rockville, MD). PC3 cells were maintained in RPMI

1640 medium supplemented with 10% fetal bovine serum, 90 mg/ml gentamicin, 2 mM

L-glutamine, and 90 mg/ml cefotaxime. HT29 cells were maintained in McCoy’s 5A medium supplemented with 10% fetal bovine serum, 90 mg/ml gentamicin, 90 mg/ml cefotaxime, 2 mM L-glutamine, and 0.1% non-essential amino acids. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO2.

5.2.4 Animal protocol for xenograft tumor model.

Male balbc/nu.nu mice and female athymic nude mice, were purchased from the

National Cancer Institute (Bethesda, MD), housed in air-filtered laminar flow cabinets and cared for in accordance with institutional guidelines (all mice 5-6 weeks old). PC3 tumors were implanted in balbc mice and HT29 in the athymic mice. Tumors cells were harvested from sub-confluent cultures using trypsin, and suspended in serum free RPMI medium (PC3) or physiological saline (HT29). Tumor cells were injected subcutaneously into the flank on both sides of a mouse (2x106 cells/200 µl per injection site). Drug treatment was initiated two weeks later or when tumors reached ~5 mm in diameter.

For PC3 tumors, drug treatment was administered intravenously via tail vein.

Animals received either vehicle (i.e., PBS containing 10% of a 50:50 cremophor:ethanol mixture), single agent (paclitaxel, or low or high dose suramin), or a combination of paclitaxel and suramin. Treatments were given twice weekly for three weeks (i.e., on day

1, 4, 8, 11, 15 and 18), animals were euthanized 3 days after the last treatment, and the tumors were excised and flash frozen in liquid nitrogen. The paclitaxel dose was 15

101 mg/kg per treatment. The low dose suramin was 10 mg/kg per treatment. The high dose suramin group received a loading dose of 200 mg/kg during the first treatment, followed by 5 additional doses of 130 mg/kg per treatment. Chemosensitization by low dose suramin in the PC3 model is shown in Figure 5.1 for reference.

For HT29 tumors, animals received intraperitoneal injections of either vehicle

(normal saline), suramin (10 mg/kg), CPT-11 (100 mg/kg), or a combination of the two drugs. Treatments were given twice weekly for a total of seven treatments (i.e., on day 1,

4, 8, 11, 15, 18, and 21) and animals were euthanized 3 days after the last treatment. An additional CPT-11 re-growth group was added for the functional vessel study. Three animals were treated with CPT-11 per above schedule and the tumors were then allowed to grow until the tumor size was equal to that of the vehicle controls (approximately 1.5 mounts after the last CPT-11 treatment). Chemosensitization by low dose suramin in the

HT-29 model is shown in Figure 5.2 for reference.

Intraperitoneal administration of CPT-11 was chosen because of acute death was observed in a pilot study using intravenous CPT-11 injection. Guichard et al. have shown that the intraperitoneal administration of CPT-11 yielded area-under-plasma concentration-time curves of CPT-11 and its active metabolite, SN-38, that are comparable to the intravenous administration [205]. Precipitation was observed when

CPT-11 was mixed with suramin. Hence, for the combination therapy group, suramin was administered 4 hours before CPT-11. Pharmacokinetic evaluation comparing intraperitoneal versus intravenous injection of suramin showed the two administration routes exhibited similar area-under-plasma concentration time curves (unpublished data).

102 The results of the morphologic evaluation of the tumor vasculature showed that chemotherapy increased vessel density and decreased vessel area. Since the established immunohistochemical methodology (see below) does not distinguish between functional and non-functional vessels the HT29 model was extended to include a perfusion marker,

DiOC7(3), to label the functional blood vessels at the time of euthanasia [203]. At time of animal sacrifice, the DiOC7(3) stock was diluted to 0.4 mg/ml in 25% PBS in DMSO. An intravenous dose of 1 mg/kg was administered to the animal one minute prior to sacrifice at which time, tumors were excised and flash frozen in liquid nitrogen and stored at -70 oC for subsequent immunohistological evaluation.

5.2.5 Immunohistochemical staining of the tumor vasculature.

All of the tumor samples from both tumor models (PC3 and HT29) were processed for morphologic evaluation of the vasculature using methods identical to those outlined in Chapter 4, save frozen sections were substituted in place of zinc fixed sections

[191]. Briefly, tumors were cryosectioned (10 µm think), fixed in ice cold acetone (2 min), air dried at room temperature (1 hr). Slides were then washed in PBS (pH 7.4), blocked with protein blocking solution (LSAB kit, DAKO) for 10 min, and incubated for

1 hr with a CD-31 (clone MEC13.3) antibody solution (1:25 dilution in 0.5% BSA in

PBS, pH 7.4). Slides were then washed in PBS (pH 7.4) and incubated for 30 min with a biotinylated goat anti-rat Ig specific polyclonal antibody (1:100 dilution in 0.5% BSA in

PBS, pH 7.4), washed in PBS (pH7.4), incubated for 30 min with streptavidin-conjugated horseradish peroxidase (LSAB kit, DAKO), washed in PBS, and then incubated with

DAB chromogen for 2 min. Slides were subsequently counterstained with hematoxylin and mounted for viewing and image analysis.

103 HT29 tumor samples intended for functional evaluation were cryosectioned (10

µm think) and the DiOC7(3) dye was photographed as described in section 5.2.6. Sections were washed in PBS (pH 7.4), incubated for 1 hour with the same primary antibody

(clone MEC 13.3), washed in PBS (pH 7.4), and incubated for 30 with the same biotinylated secondary antibody mentioned previously. The sections were then incubated for 30 min with Image-iT™ FX Signal Enhancer to block any non-specific binding between the Alexa Fluor 647 and the tumor tissue. Excess blocker was wicked away and the sections were incubated for 30 min with Streptavidin-Alexa Fluor 647 (1:5000 dilution in PBS, pH 7.4). Slides were then mounted using an anti-fade reagent (9:1

Glycerol:PBS containing 0.1% p-phenylenediamine buffered to pH 8.0 with 0.5M

Carbonate/Bicarbonate, [206]) and re-photographed for fluorescent quantification.

Negative controls for all cases consisted of additional tumor sections that were stained by substituting rat antibodies of the same isotype as MEC 13.3 for the primary antibody.

5.2.6 Development of a quantitative image analysis method to quantify the morphologic and functional changes in the tumor vasculature.

Tumor sections containing the fluorescent profusion marker DiOC7(3) were viewed using fluorescent microscopy and the entire section was photographed at 100x using a Leica TCS SP2 AOBS confocal system and a Marzhauser mobile stage attached to a Leica DM IRE2 microscope (Leica Microsystems, Mannheim, Germany). The same section was then washed and stained for CD-31 using the protocol outlined above and the

Alexa Fluor 647 signal was photographed in the same manner. All fluorescent images were saved as 8-bit TIF images and a custom image analysis program was written in

Optimas® to quantify the total area (in mm2) of the two fluorescent stains. Three variables

104 were calculated to quantify the functionality of the tumor vasculature. The total area of

Alexa Fluor 647 stain per section divided by the tissue section area, total area of

DiOC7(3) stain per section divided by the tissue section area, and the total area of the

DiOC7(3) divided by the total area of Alexa Fluor 647 stain.

Tumor sections for morphologic evaluation were stained for CD-31 using the

DAB chromogen and the vessel morphology quantification methods used were the same as those outlined in Chapter 4. Briefly, tumor samples were photographed using a

Hamamatsu (Hamamatsu-City, Japan) color chilled 3CCD camera attached to a ZEISS

(Thornwood, NY) Axiovert 35 microscope. All images were saved as 8-bit TIF format images and analyzed using Optimas® (version 6.51, Media Cybernetics, Silver Spring,

MD) image analysis software. Tumor vascular areas were identified and photographed using the previously reported “hotspot” method [192;193]. An average of 9 (200x) microscopic images were analyzed per tumor (an average of 3 images per section, 3 sections per tumor) per treatment group with the image number dependent on the number of vascular rich areas and the overall surface area of the tumor section. At 200x magnification the field size of an image was 0.204 mm2. Stained vessels were outlined using Optimas® image analysis software and the vessel density (number of counted vessels / field area in mm2), average vessel area (units of µm2), and average vessel length

(units of µm) were calculated to evaluate the effects of the different treatments on the tumor vasculature.

105 5.2.7 Statistical analysis.

Statistical differences between all vessel parameters were evaluated using

Tukey’s multi-comparison analysis after one-way ANOVA. A p-value of less than 5% was considered significant. ANOVA calculations were performed using SAS® (The

SAS® Institute Inc. Cary, NC).

5.3 RESULTS

5.3.1 No Anti-angiogenic effects by suramin alone or in combination with chemotherapy in tumor-bearing animals as measured by vessel morphology.

Morphologically the control and low dose single agent suramin groups for both tumor models were very similar (Figure 5.3 A and Figure 5.4 A). Vessels in these groups looked healthy and tended to be either large cavernous vessels or long multi branching vessels. The high dose suramin tumors (PC3 group only) appeared similar to the control and low dose suramin of the same group, however the larger cavernous vessels were less frequent and the long branching vessels appeared shortened in comparison to the control and low dose groups. Quantitative image analysis showed no statically significant differences between the vehicle controls or any of the single agent suramin treated groups for all three of the morphologic parameters (Figure 5.3 B and Figure 5.4 B). This suggests that suramin alone does not exert any effect on the tumor vasculature in vivo.

Chemotherapy (both paclitaxel and CPT-11) altered the tumor vasculature with similar results, and groups receiving chemotherapy were morphologically different from the control and suramin alone groups. The large cavernous vessels were completely absent and only small punctuated vessels remained. The nuclei of the tumor cells were pinpoint or punctuated in nature and the tumor cells looked compressed. Groups

106 receiving chemotherapy alone showed a significant increase in vessel density and a significant decrease in vessel area and length meaning a higher frequency of smaller vessels. Interestingly, addition of suramin (either low or high dose) to chemotherapy did not appear to morphologically alter the vasculature nor were there any statistical differences between any of the combination suramin groups versus the single agent chemotherapy groups for all three of the vessel parameters (Figure 5.3 and Figure 5.4).

This suggests that suramin was not able to further sensitize the tumor vasculature to chemotherapy in vivo. It should be noted that the effects of the chemotherapy might have masked any possible additional suramin effect. However if this is possible it still speaks against the hypothesis for an in vivo anti-angiogenic mechanism for suramin.

These results indicate that only the chemotherapy treatment was able to alter the morphology of the tumor vasculature. Suramin alone had no morphologic effects nor did suramin exhibit any additional effect when combined with chemotherapy. Meaning suramin alone did not act through an anti-angiogenic mechanism nor did it further amplify the effect caused by chemotherapy (Figures 5.3 and 5.4).

5.3.2 Chemotherapy increases the functionality of the tumor vessels and this effect is independent of the suramin effect.

Results from the quantification of the fluorescent images show that the total amount of endothelium (Alexa Fluor 647 stain / tissue area) is increased in the CPT-11 single agent and the combination low groups as compared to the vehicle control group

(Table 5.1, Figure 5.5), however the CPT-11 re-growth and the suramin low dose groups were not different from the vehicle control groups. This finding parallels the increase in vessel density shown in Figure 5.3 B and Figure 5.4 B and suggests that morphology of

107 the vessels in the CPT-11 group is similar to the combination low group and the morphology of the CPT-11 re-growth group are similar to the vehicle control and the suramin low dose group.

The total amount of functional endothelium (DiOC7(3) stain / tissue area) was significantly increased in the CPT-11, combination low groups, and the CPT-11 re- growth groups versus vehicle control while the suramin low dose groups were not different from control. The percent functionality (total area of DiOC7(3) stain / total area of Alexa Fluor 647 stain) of each of the groups show that the CPT-11 single agent, CPT-

11 re-growth, and combination suramin low groups all have a significantly larger amount of functional vessels as compared to vehicle control (Table 5.1).

Together the data presented in Table 5.1 and Figure 5.5 shows that effects of chemotherapy not only alter the vessel morphology (shown in Figures 5.3 and 5.4) but also increases the functionality of the blood vessels inside the tumor. Suramin alone or in combination with chemotherapy did not significantly alter the functionality; however, suramin alone or in combination with chemotherapy did show a non-significant increase in vessel functionality as compared to vehicle control or CPT-11 single agent, respectively. The effects of chemotherapy on vessel functionality were not transient in nature as shown by the CPT-11 re-growth group. The tumors in the re-growth group were allowed to re-grow until they reached a size equal to that of the vehicle control (~1.5 months). It is possible that the functionality differences are related to tumor size since the tumor volumes of CPT-11 treated tumors and the vehicle control and suramin low groups differ by almost 15 fold. However the tumor sizes of the CPT-11 re-growth tumors were

108 equal to that of the vehicle control and yet the CPT-11 effects on vessel functionality were still present. This speaks against the possibility of tumor size as being the most important factor in vessel functionality.

5.4 DISCUSSION

The chemosensitizing effects of suramin have been proven in both monolayer and histoculture in vitro systems [2;207]. The results from Song et al. show that the in vitro mechanism for this chemosensitization is through the inhibition of bFGF [2]. Further data presented by Song et al. show that the chemosensitization effects of suramin are also repeated in several pre-clinical in vivo models and it is hypothesized that the in vivo mechanism is also through the inhibition of bFGF [2;172;207]. Reports from the literature suggest that suramin has exhibited anti-angiogenic properties in many experimental systems. Therefore the purpose of the in vitro (detailed in the previous

Chapter) and in vivo studies presented herein is to ascertain if the established in vivo chemosensitization mechanism is through anti-angiogenesis.

In vitro drug response from the monolayer HUVEC study showed that suramin did not further sensitize endothelial cells to paclitaxel insult. Results from the histoculture assay show that chemosensitization with suramin occurred regardless of significant regression of the tumor vessels (data presented in Chapter 4). The in vivo results of the this study on well-established xenograft tumors showed no significant differences in the vessel morphology or functionality when comparing the use of suramin (either low or high dosage) as a single agent versus untreated controls or when comparing suramin in combination with paclitaxel/CPT-11 versus single agent chemotherapy. Neither the

109 trends in the in vitro or the in vivo results are in agreement with the established trend that low doses (<50 µM) of suramin show chemosensitization while high doses (>100 µM) of suramin do not [173;207;208].

The literature discussed previously does suggest that suramin has the potential to cause anti-angiogenesis, however most of the experimental systems used showed that higher concentrations of suramin were required to cause anti-angiogenesis. For example

Takano et al. tested the ability of suramin to inhibit of bFGF binding to bovine capillary endothelial cells, endothelial cell migration, proliferation, and protease activity.

Inhibition of bFGF binding was achieved at concentrations 60 µM suramin, whereas inhibition of endothelial cell migration, proliferation, and protease activity required much higher concentrations 150-200 µM[61]. This suggests that low doses of suramin could act to inhibit bFGF thus abolishing the FGF-induced resistance effect without affecting the surrounding vasculature and invoking an anti-angiogenic response.

The study by Pesenti et al. presents an important detail concerning the possibility of an anti-angiogenic mechanism for suramin. The study showed that suramin is able to inhibit in vivo angiogenesis only when a high dose of suramin (200 mg/kg in mice) was administered prior to tumor establishment. The results clearly show suramin doses 100 mg/kg had no anti-angiogenic effect and the 200 mg/kg dose had to be given prior to tumor establishment in order to invoke and anti-angiogenic response [60]. This result is consistent with current findings that the vessel parameters were unchanged after treatment with neither low and high dose suramin because the study was intentionally executed in well established tumor models in order to maintain clinical relevance.

110 The results of this study help to distinguish the FGF-resistance reversal approach apart from the anti-angiogenesis approach which is an important delineation because these two approaches differ in principle and therefore their practice. In anti-angiogenic therapy, the purpose is to inhibit the formation of new vessels therefore, therapy should be administered continuously for a duration much greater than customary cytotoxic therapy. In contrast, for inhibitors of FGF resistance, the purpose is to cause cytotoxicity to tumor cells and therefore should be administered along with the cytotoxic therapy. In addition, while there are intense research interests on using anti-angiogenic agents to enhance the efficacy of chemotherapy, the clinical usefulness of this approach is still an open question.

In conclusion, these results show that the chemosensitization effect of suramin is not due to its anti-angiogenic properties thus supporting the FGF-resistance reversal approach as a treatment paradigm independent of anti-angiogenesis and therefore worthy on its own merit.

5.5 ACKNOWLEDGMENTS

This work was supported in part by the research grant R21 CA91547-02 from the

National Cancer Institute, NIH and by a research fellowship from the American

Foundation for Pharmaceutical Education. I would like to thank my advisor Dr. Jessie Au for her expertise, support, and patience. Special thanks go to Dr. M. Guill Wientjes, Dr.

Yong Wei, Dr. Liang Zhao and, Bei Yu for their technical support and scientific input in the study.

111

Treatment Group Total Amount of Alexa Fluor Total Amount of DiOC (3) 7 % Functionality (Number of Tumors) 647 Stain / Tumor Area Stain / Tumor Area Vehicle Control (n=8) 0.054 ± 0.012 0.012 ± 0.003 22.5 ± 4.9 CPT-11 Re-growth (n=6) 0.052 ± 0.016 0.040 † ±0.020 73.9 † ±21.2 Suramin Low (n=8) 0.055 ± 0.008 0.016 ± 0.006 29.2 ± 13.4 CPT-11 (n=10) 0.100 † ±0.040 0.055 † ±0.013 64.0 † ±30.0 CPT-11 + Suramin Low (n=10) 0.103 † ±0.038 0.069 † ±0.028 77.1 † ±41.0 † p-value <0.05 verses Vehicle Control group.

112

Table 5.1. Summary of functional vessel data obtained using image analysis for the HT29 subcutaneous tumor model.

Values are mean ± SD. Suramin dose was 70 mg/kg/7 doses, CPT-11 dose was 700 mg/kg/7 doses, and the combination group

received both the CPT-11 and suramin doses. The CPT-11 Re-growth group was treated using the same schedule as the CPT-11

group and the tumors were allowed to re-grow until the tumor size was the same as the vehicle control group. Statistical analysis

was performed using ANOVA with post hoc Tukey analysis of group differences.

112

10

1 Relative Tumor Size

0.1 0 7 14 21

Time (days)

Figure 5.1. Chemosensitization by low dose suramin in PC3 xenograft tumors.

Suramin low (○) dose was 60 mg/kg/6 doses (plasma concentration <50 µM), suramin high ( ) dose was 850 mg/kg/6 doses (plasma concentration <100 µM), paclitaxel (▼) dose was 90 mg/kg/6 doses, and the combination low (■) and high (□) groups received both the paclitaxel plus the low/high dosage of suramin, respectively. Control animals received vehicle only (●). Dosing was twice a week for 3 weeks. Mean and one SD are shown, some error bars are smaller then the symbols.

113

10

1 Relative Tumor Size Tumor Relative

0 5 10 15 20 25 30 Time(days)

Figure 5.2. Chemosensitization by low dose suramin in HT29 xenograft tumors.

Suramin ( ) dose was 70 mg/kg/7 doses (plasma concentration <50 µM), CPT-11 (○) dose was 700 mg/kg/7 doses, and the combination (●) group received both the CPT-11 and suramin doses. The control (▼) animals received normal saline. Dosing was twice a week for 3 weeks. Mean and one SD are shown, some error bars are smaller then the symbols.

114

A.) 250 B. 2 *** 200

150

100

50 AverageDensity Vessel

(Vessel Number / Field, µm 0 1,500 ) 2 Control Paclitaxel 1,000

500 * * * Average Vessel Area (µm Area Vessel Average 0 Suramin Low Combination Low 60

40 * * *

20

Suramin High Combination High (µm) Length Vessel Average 0

ol w h el w h tr o ig x o ig n L H ta L H o n n li n n C i i ac io o le am m P at ti ic r ra n na h Su u bi i e S m b V o om C C

Figure 5.3. Sample Images of the immunohistochemical staining (A) and summary of morphologic vessel data obtained using image analysis for the PC3 subcutaneous tumor model (B). Representative images from each treatment group are shown, magnification 200x. Open bars represent control or suramin single agent, filled bars represent paclitaxel alone or in combination with suramin. Bars are mean and one SD.

Statistical analysis of the vessel parameters was performed using ANOVA with post hoc

Tukey analysis of group differences. * represents a p-value of < 0.05 versus vehicle control.

115

A. B. ) 250 2

200 * * 150

100

50

Control AverageDensity Vessel

(Vessel Number / Field, µm / Field, (Vessel Number 0 ) 2 2,000

1,500

1,000 * Suramin Low * 500

AverageArea Vessel (µm 0 100

80

CPT-11 60 * 40 *

20

Average Vessel Length (µm) Length Average Vessel 0 l 1 ro ow -1 ow nt L T L o n P n C i C io le am at ic ur in Combination Low eh S b V om C

Figure 5.4. Sample Images of the immunohistochemical staining (A) and summary of morphologic vessel data obtained using image analysis for the HT29 subcutaneous tumor model (B). Representative images from each treatment group are shown, magnification 200x. Open bars represent control or suramin single agent, filled bars represent CPT-11 alone or in combination with suramin. Bars are mean and one SD.

Statistical analysis of the vessel parameters was performed using ANOVA with post hoc

Tukey analysis of group differences. * represents a p-value of < 0.05 versus vehicle control.

116

1A. 1B. 2A. 2B.

117 4A. 4B. 5A. 5B. 3A. 3B.

Figure 5.5. Examples of the functional vessel staining (A) and the total vessel staining (B). Images from the Vehicle control

(1), CPT-11 re-growth (2), Suramin Low (3), CPT-11 (4), and Suramin Low/CPT-11 Combination (5) groups are shown. All

fluorescent images were captured at 100x using a Leica TCS SP2 AOBS confocal system and a Marzhauser mobile stage

attached to a Leica DM IRE2 microscope (Leica Microsystems, Mannheim, Germany). Total areas of the green stain, red stain,

and the tumor sections were calculated using Optimas®.

117 CHAPTER 6

GENERAL DISCUSSION AND PERSPECTIVES

The studies presented in this dissertation are focused on improving the understanding of the mechanisms of the FGF-induced resistance and the molecular pharmacodynamics of suramin as part of its pre-clinical development. This work is based on a previous finding by Au and Wientjes et al. showing that acidic (aFGF) and basic

(bFGF) fibroblast growth factors confer a broad spectrum chemoresistance in solid tumors, and that low doses of suramin (15 µM), a growth factor inhibitor, enhanced the in vitro anti-tumor activity of several anticancer drugs.

Given the promising in vitro finding, the first study presented here translated that into a clinically relevant in vivo model. The in vivo interaction between paclitaxel and suramin in a tumor cell line which metastasized to the lung. This model was chosen because of its clinical relevance since tumor metastasis is the major cause of treatment failure for cancer patients. The results show that low doses of suramin alone had no antitumor effect nor toxicity, consistent with previous findings. Addition of low dose suramin to paclitaxel therapy significantly enhanced the antitumor effect, as measured by reduction of tumor size, reduction in the density of nonapoptotic cells, and an increase in the apoptotic cell fraction.

118 The next study tested if bFGF level correlates with drug resistance and suramin effect in a clinically relevant tumor model. An archival bank of >100 patient samples from various sites (bladder, breast, head and neck, ovarian, prostate, and RCC) were treated with either paclitaxel or 5-fluorouracil in combination with suramin and immunostained for bFGF. A new image analysis method was developed to quantify the immunohistochemical staining in terms of real bFGF protein amounts. The results show that the intratumoral bFGF level was the best single predictor of paclitaxel effect among a panel of known resistance factors (stage, grade, bcl-2, p53, pgp, stage, and aFGF).

Furthermore, the quantitative nature of the new analysis method allowed for identification of sub-groups within the main data set that showed even higher correlations between bFGF and drug resistance. Since bFGF is the hypothesized target for suramin, the correlation between the fold chemosensitization and the intratumoral bFGF level was evaluated and found to be highly correlated in the clinical samples treated with of low (20

µM) but not high doses (50 µM) of suramin.

The final two Chapters of this dissertation (Chapters 4 and 5) investigated the possibility of an anti-angiogenic mechanism for suramin. The literature shows numerous studies showing that suramin has anti-angiogenic potential therefore the possibility of an anti-angiogenic mechanism was tested in both in vitro and in vivo models. Results from the in vitro studies showed that suramin alone had no effects at low doses nor did suramin enhance the cytotoxicity of chemotherapy in a monolayer HUVEC system. This was further confirmed in an in vitro histoculture system and the results showed that the chemosensitization effect of suramin was independent of the tumor vasculature. The in vitro findings were further confirmed using two well established in vivo solid tumor

119 models. The vessel morphology and functionality was measured after treatment with low and high doses suramin. The in vivo results confirmed the monolayer HUVEC results, that suramin alone had no effects on the vessel morphology nor did suramin enhance the effects of chemotherapy as measured by vessel morphology and functionality.

Interestingly chemotherapy (paclitaxel or CPT-11) had a profound effect on both the morphology and functionality of the tumor vasculature. This data warrants further investigation.

In summary the work presented in this dissertation show that suramin is effective in vivo and the data suggest this is through the proposed mechanism of inhibition of bFGF induced resistance. This work also shows that the intratumoral bFGF level is a clinically relevant predictor of drug resistance and suramin effect. Investigations into the possibility of an anti-angiogenic mechanism suggest that anti-angiogenesis is not the main cause for the observed chemosensitization effects of suramin as the literature suggests.

120 LIST OF REFERENCES

1. Nygren P, Larsson R. Overview of the clinical efficacy of investigational anticancer drugs. J.Intern.Med. 2003; 253:46-75.

2. Song S, Wientjes MG, Gan Y, Au JL. Fibroblast growth factors: an epigenetic mechanism of broad spectrum resistance to anticancer drugs. Proc.Natl.Acad.Sci.U.S.A 2000; 97:8658-8663.

3. Ornitz DM. FGFs, heparan sulfate and FGFRs: complex interactions essential for development. Bioessays 2000; 22:108-112.

4. Ornitz DM, Itoh N. Fibroblast growth factors. Genome Biol. 2001; 2:REVIEWS3005.

5. Plotnikov AN, Hubbard SR, Schlessinger J, Mohammadi M. Crystal structures of two FGF-FGFR complexes reveal the determinants of ligand-receptor specificity. Cell 2000; 101:413-424.

6. Nugent MA, Iozzo RV. Fibroblast growth factor-2. Int.J.Biochem.Cell Biol. 2000; 32:115-120.

7. Okada-Ban M, Thiery JP, Jouanneau J. Fibroblast growth factor-2. Int.J.Biochem.Cell Biol. 2000; 32:263-267.

8. Dow JK, deVere White RW. Fibroblast growth factor 2: its structure and property, paracrine function, tumor angiogenesis, and prostate-related mitogenic and oncogenic functions. Urology 2000; 55:800-806.

9. Slavin J. Fibroblast growth factors: at the heart of angiogenesis. Cell Biol.Int. 1995; 19:431-444.

10. Bikfalvi A, Klein S, Pintucci G, Rifkin DB. Biological roles of fibroblast growth factor-2. Endocr.Rev. 1997; 18:26-45.

11. Cross MJ, Claesson-Welsh L. FGF and VEGF function in angiogenesis: signalling pathways, biological responses and therapeutic inhibition. Trends Pharmacol.Sci. 2001; 22:201-207.

12. Friesel RE, Maciag T. Molecular mechanisms of angiogenesis: fibroblast growth factor signal transduction. FASEB J. 1995; 9:919-925.

121 13. Klint P, Claesson-Welsh L. Signal transduction by fibroblast growth factor receptors. Front Biosci. 1999; 4:D165-D177.

14. Kos M, Dabrowski A. Tumour's angiogenesis--the function of VEGF and bFGF in colorectal cancer. Ann.Univ Mariae.Curie Sklodowska [Med.] 2002; 57:556- 561.

15. McIntosh I, Bellus GA, Jab EW. The pleiotropic effects of fibroblast growth factor receptors in mammalian development. Cell Struct.Funct. 2000; 25:85-96.

16. Cronauer MV, Hittmair A, Eder IE et al. Basic fibroblast growth factor levels in cancer cells and in sera of patients suffering from proliferative disorders of the prostate. Prostate 1997; 31:223-233.

17. Sliutz G, Tempfer C, Obermair A, Reinthaller A, Gitsch G, Kainz C. Serum evaluation of basic fibroblast growth factor in cervical cancer patients. Cancer Lett. 1995; 94:227-231.

18. Aguayo A, Kantarjian H, Manshouri T et al. Angiogenesis in acute and chronic leukemias and myelodysplastic syndromes. Blood 2000; 96:2240-2245.

19. Duensing S, Grosse J, Atzpodien J. Increased serum levels of basic fibroblast growth factor (bFGF) are associated with progressive lung metastases in advanced renal cell carcinoma patients. Anticancer Res 1995; 15:2331-2333.

20. Nguyen M, Watanabe H, Budson AE, Richie JP, Hayes DF, Folkman J. Elevated levels of an angiogenic peptide, basic fibroblast growth factor, in the urine of patients with a wide spectrum of cancers. J.Natl.Cancer Inst. 1994; 86:356-361.

21. Salven P, Teerenhovi L, Joensuu H. A high pretreatment serum basic fibroblast growth factor concentration is an independent predictor of poor prognosis in non- Hodgkin's lymphoma. Blood 1999; 94:3334-3339.

22. Tabone MD, Landman-Parker J, Arcil B et al. Are basic fibroblast growth factor and vascular endothelial growth factor prognostic indicators in pediatric patients with malignant solid tumors? Clin.Cancer Res 2001; 7:538-543.

23. Iwasaki A, Kuwahara M, Yoshinaga Y, Shirakusa T. Basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF) levels, as prognostic indicators in NSCLC. Eur.J.Cardiothorac.Surg. 2004; 25:443-448.

24. Joensuu H, Anttonen A, Eriksson M et al. Soluble syndecan-1 and serum basic fibroblast growth factor are new prognostic factors in lung cancer. Cancer Res 2002; 62:5210-5217.

122 25. Ruotsalainen T, Joensuu H, Mattson K, Salven P. High pretreatment serum concentration of basic fibroblast growth factor is a predictor of poor prognosis in small cell lung cancer. Cancer Epidemiol.Biomarkers Prev. 2002; 11:1492-1495.

26. Ueno K, Inoue Y, Kawaguchi T, Hosoe S, Kawahara M. Increased serum levels of basic fibroblast growth factor in lung cancer patients: relevance to response of therapy and prognosis. Lung Cancer 2001; 31:213-219.

27. Rasmuson T, Grankvist K, Jacobsen J, Ljungberg B. Impact of serum basic fibroblast growth factor on prognosis in human renal cell carcinoma. Eur.J.Cancer 2001; 37:2199-2203.

28. Dietz A, Rudat V, Conradt C, Weidauer H, Ho A, Moehler T. Prognostic relevance of serum levels of the angiogenic peptide bFGF in advanced carcinoma of the head and neck treated by primary radiochemotherapy. Head Neck 2000; 22:666-673.

29. Anzai H, Kitadai Y, Bucana CD, Sanchez R, Omoto R, Fidler IJ. Expression of metastasis-related genes in surgical specimens of human gastric cancer can predict disease recurrence. Eur.J.Cancer 1998; 34:558-565.

30. Noda M, Hattori T, Kimura T et al. Expression of fibroblast growth factor 2 mRNA in early and advanced gastric cancer. Acta Oncol. 1997; 36:695-700.

31. Pazgal I, Zimra Y, Tzabar C et al. Expression of basic fibroblast growth factor is associated with poor outcome in non-Hodgkin's lymphoma. Br.J.Cancer 2002; 86:1770-1775.

32. Salven P, Orpana A, Teerenhovi L, Joensuu H. Simultaneous elevation in the serum concentrations of the angiogenic growth factors VEGF and bFGF is an independent predictor of poor prognosis in non-Hodgkin lymphoma: a single- institution study of 200 patients. Blood 2000; 96:3712-3718.

33. Tanaka S, Ueo H, Mafune K, Mori M, Wands JR, Sugimachi K. A novel isoform of human fibroblast growth factor 8 is induced by androgens and associated with progression of esophageal carcinoma. Dig.Dis.Sci. 2001; 46:1016-1021.

34. Boelaert K, McCabe CJ, Tannahill LA et al. Pituitary tumor transforming gene and fibroblast growth factor-2 expression: potential prognostic indicators in differentiated thyroid cancer. J.Clin.Endocrinol.Metab 2003; 88:2341-2347.

35. Sun Y, Naito Z, Ishiwata T, Maeda S, Sugisaki Y, Asano G. Basic FGF and Ki-67 proteins useful for immunohistological diagnostic evaluations in malignant solitary fibrous tumor. Pathol.Int. 2003; 53:284-290.

123 36. Ito H, Oshita F, Kameda Y et al. Expression of vascular endothelial growth factor and basic fibroblast growth factor in small adenocarcinomas. Oncol.Rep. 2002; 9:119-123.

37. Strizzi L, Vianale G, Catalano A, Muraro R, Mutti L, Procopio A. Basic fibroblast growth factor in mesothelioma pleural effusions: correlation with patient survival and angiogenesis. Int.J.Oncol. 2001; 18:1093-1098.

38. Lin RY, Argenta PA, Sullivan KM, Adzick NS. Diagnostic and prognostic role of basic fibroblast growth factor in Wilms' tumor patients. Clin.Cancer Res 1995; 1:327-331.

39. Ohta T, Yamamoto M, Numata M et al. Expression of basic fibroblast growth factor and its receptor in human pancreatic carcinomas. Br.J.Cancer 1995; 72:824-831.

40. Volm M, Koomagi R, Mattern J, Stammler G. Prognostic value of basic fibroblast growth factor and its receptor (FGFR-1) in patients with non-small cell lung carcinomas. Eur.J.Cancer 1997; 33:691-693.

41. Faridi A, Rudlowski C, Biesterfeld S, Schuh S, Rath W, Schroder W. Long-term follow-up and prognostic significance of angiogenic basic fibroblast growth factor (bFGF) expression in patients with breast cancer. Pathol.Res Pract. 2002; 198:1- 5.

42. Linderholm BK, Lindh B, Beckman L et al. Prognostic correlation of basic fibroblast growth factor and vascular endothelial growth factor in 1307 primary breast cancers. Clin.Breast Cancer 2003; 4:340-347.

43. Blanckaert VD, Hebbar M, Louchez MM, Vilain MO, Schelling ME, Peyrat JP. Basic fibroblast growth factor receptors and their prognostic value in human breast cancer. Clin.Cancer Res 1998; 4:2939-2947.

44. Colomer R, Aparicio J, Montero S, Guzman C, Larrodera L, Cortes-Funes H. Low levels of basic fibroblast growth factor (bFGF) are associated with a poor prognosis in human breast carcinoma. Br.J.Cancer 1997; 76:1215-1220.

45. Smith K, Fox SB, Whitehouse R et al. Upregulation of basic fibroblast growth factor in breast carcinoma and its relationship to vascular density, oestrogen receptor, epidermal growth factor receptor and survival. Ann.Oncol. 1999; 10:707-713.

46. Yiangou C, Gomm JJ, Coope RC et al. Fibroblast growth factor 2 in breast cancer: occurrence and prognostic significance. Br.J.Cancer 1997; 75:28-33.

124 47. Obermair A, Speiser P, Reisenberger K et al. Influence of intratumoral basic fibroblast growth factor concentration on survival in ovarian cancer patients. Cancer Lett. 1998; 130:69-76.

48. Bredel M, Pollack IF, Campbell JW, Hamilton RL. Basic fibroblast growth factor expression as a predictor of prognosis in pediatric high-grade gliomas. Clin.Cancer Res 1997; 3:2157-2164.

49. Huang A, Wright JA. Fibroblast growth factor mediated alterations in drug resistance, and evidence of gene amplification. Oncogene 1994; 9:491-499.

50. Huang A, Jin H, Wright JA. Aberrant expression of basic fibroblast growth factor in NIH-3T3 cells alters drug resistance and gene amplification potential. Exp.Cell Res. 1994; 213:335-339.

51. Cohen-Jonathan E, Toulas C, Monteil S et al. Radioresistance induced by the high molecular forms of the basic fibroblast growth factor is associated with an increased G2 delay and a hyperphosphorylation of p34CDC2 in HeLa cells. Cancer Res. 1997; 57:1364-1370.

52. Cordon-Cardo C, Vlodavsky I, Haimovitz-Friedman A, Hicklin D, Fuks Z. Expression of basic fibroblast growth factor in normal human tissues. Lab Invest 1990; 63:832-840.

53. Konig A, Menzel T, Lynen S et al. Basic fibroblast growth factor (bFGF) upregulates the expression of bcl-2 in B cell chronic lymphocytic leukemia cell lines resulting in delaying apoptosis. Leukemia 1997; 11:258-265.

54. Menzel T, Rahman Z, Calleja E et al. Elevated intracellular level of basic fibroblast growth factor correlates with stage of chronic lymphocytic leukemia and is associated with resistance to fludarabine. Blood 1996; 87:1056-1063.

55. Miyake H, Hara I, Gohji K, Yoshimura K, Arakawa S, Kamidono S. Expression of basic fibroblast growth factor is associated with resistance to cisplatin in a human bladder cancer cell line. Cancer Lett. 1998; 123:121-126.

56. Folkman J, Merler E, Abernathy C, Williams G. Isolation of a tumor factor responsible or angiogenesis. J.Exp.Med. 1971; 133:275-288.

57. Shing Y, Folkman J, Sullivan R, Butterfield C, Murray J, Klagsbrun M. Heparin affinity: purification of a tumor-derived capillary endothelial cell growth factor. Science 1984; 223:1296-1299.

58. Bocci G, Danesi R, Benelli U et al. Inhibitory effect of suramin in rat models of angiogenesis in vitro and in vivo. Cancer Chemother.Pharmacol. 1999; 43:205- 212.

125 59. Gagliardi A, Hadd H, Collins DC. Inhibition of angiogenesis by suramin. Cancer Res 1992; 52:5073-5075.

60. Pesenti E, Sola F, Mongelli N, Grandi M, Spreafico F. Suramin prevents neovascularisation and tumour growth through blocking of basic fibroblast growth factor activity. Br.J.Cancer 1992; 66:367-372.

61. Takano S, Gately S, Neville ME et al. Suramin, an anticancer and angiosuppressive agent, inhibits endothelial cell binding of basic fibroblast growth factor, migration, proliferation, and induction of urokinase-type plasminogen activator. Cancer Res. 1994; 54:2654-2660.

62. Goto F, Goto K, Weindel K, Folkman J. Synergistic effects of vascular endothelial growth factor and basic fibroblast growth factor on the proliferation and cord formation of bovine capillary endothelial cells within collagen gels. Lab Invest 1993; 69:508-517.

63. Pepper MS, Ferrara N, Orci L, Montesano R. Potent synergism between vascular endothelial growth factor and basic fibroblast growth factor in the induction of angiogenesis in vitro. Biochem.Biophys.Res Commun. 1992; 189:824-831.

64. Pepper MS, Mandriota SJ. Regulation of vascular endothelial growth factor receptor-2 (Flk-1) expression in vascular endothelial cells. Exp.Cell Res 1998; 241:414-425.

65. Seghezzi G, Patel S, Ren CJ et al. Fibroblast growth factor-2 (FGF-2) induces vascular endothelial growth factor (VEGF) expression in the endothelial cells of forming capillaries: an autocrine mechanism contributing to angiogenesis. J.Cell Biol. 1998; 141:1659-1673.

66. Ferrara N. Molecular and biological properties of vascular endothelial growth factor. J.Mol.Med. 1999; 77:527-543.

67. Suhardja A, Hoffman H. Role of growth factors and their receptors in proliferation of microvascular endothelial cells. Microsc.Res Tech. 2003; 60:70- 75.

68. Matsuzaki K, Yoshitake Y, Matuo Y, Sasaki H, Nishikawa K. Monoclonal antibodies against heparin-binding growth factor II/basic fibroblast growth factor that block its biological activity: invalidity of the antibodies for tumor angiogenesis. Proc.Natl.Acad.Sci.U.S.A 1989; 86:9911-9915.

69. Dennis PA, Rifkin DB. Studies on the role of basic fibroblast growth factor in vivo: inability of neutralizing antibodies to block tumor growth. J.Cell Physiol 1990; 144:84-98.

126 70. Powers CJ, McLeskey SW, Wellstein A. Fibroblast growth factors, their receptors and signaling. Endocr.Relat Cancer 2000; 7:165-197.

71. Compagni A, Wilgenbus P, Impagnatiello MA, Cotten M, Christofori G. Fibroblast growth factors are required for efficient tumor angiogenesis. Cancer Res 2000; 60:7163-7169.

72. Ortega S, Ittmann M, Tsang SH, Ehrlich M, Basilico C. Neuronal defects and delayed wound healing in mice lacking fibroblast growth factor 2. Proc.Natl.Acad.Sci.U.S.A 1998; 95:5672-5677.

73. Hurwitz H. Bevacizumab (Avastin™) Improves Survival in Metastatic Colorectal Cancer. American Society of Clinical Oncology (ASCO) annual meeting, Chicago, Il. 2003.

74. Golberg, K. B., Goldberg, P., and Whitmore-Wolfe, S. Clinical Trials: Pharmacia closes phase III trial of sugen's SU5416. The Cancer Letter/B&R Report 17(2), 1- 2. 2002.

75. Stein CA. Suramin: a novel antineoplastic agent with multiple potential mechanisms of action. Cancer Res. 1993; 53:2239-2248.

76. Church D, Zhang Y, Rago R, Wilding G. Efficacy of suramin against human prostate carcinoma DU145 xenografts in nude mice. Cancer Chemother.Pharmacol. 1999; 43:198-204.

77. Vincent TS, Hazen-Martin DJ, Garvin AJ. Inhibition of insulin like growth factor II autocrine growth of Wilms' tumor by suramin in vitro and in vivo. Cancer Lett. 1996; 103:49-56.

78. Yoneda T, Williams P, Rhine C, Boyce BF, Dunstan C, Mundy GR. Suramin suppresses hypercalcemia and osteoclastic bone resorption in nude mice bearing a human squamous cancer. Cancer Res. 1995; 55:1989-1993.

79. Ahmann, F. R., Schwartz, J., Dorr, R., and Salmon, S. Suramin in hormone resistant metastatic prostatic cancer: Significant anticancer activity but unanticipated toxicity. Proc.Am.Soc.Clin.Oncol. 10, 178. 1991.

80. Armand, J. P., Bonnay, M., Gandia, D., Cvitkovic, E., De Braud, F., Bertheault, F., Droz, J. P., Carde, P., Schlumberger, M., and Fourcault, D. Phase I-II suramin (SRM) study in advanced cancer patients. Proc.Am.Assoc.Cancer Res. 32, 175. 1991.

81. Dreicer R, Smith DC, Williams RD, See WA. Phase II trial of suramin in patients with metastatic renal cell carcinoma. Invest New Drugs 1999; 17:183-186.

127 82. Falcone A, Pfanner E, Brunetti I et al. Suramin in combination with 5-fluorouracil (5-FU) and leucovorin (LV) in metastatic colorectal cancer patients resistant to 5- FU+LV-based chemotherapy. Tumori 1998; 84:666-668.

83. Falcone A, Antonuzzo A, Danesi R et al. Suramin in combination with weekly epirubicin for patients with advanced hormone-refractory prostate carcinoma. Cancer 1999; 86:470-476.

84. Hussain M, Fisher EI, Petrylak DP et al. Androgen deprivation and four courses of fixed-schedule suramin treatment in patients with newly diagnosed metastatic prostate cancer: A Southwest Oncology Group Study. J.Clin.Oncol. 2000; 18:1043-1049.

85. Miglietta L, Canobbio L, Granetto C, Vannozzi MO, Esposito M, Boccardo F. Suramin/epidoxorubicin association in hormone-refractory prostate cancer: preliminary results of a pilot phase II study. J.Cancer Res.Clin.Oncol. 1997; 123:407-410.

86. Mirza MR, Jakobsen E, Pfeiffer P, Lindebjerg-Clasen B, Bergh J, Rose C. Suramin in non-small cell lung cancer and advanced breast cancer. Two parallel phase II studies. Acta Oncol. 1997; 36:171-174.

87. Motzer RJ, Dmitrovsky E, Miller WH, Jr. et al. Suramin for germ cell tumors. In vitro growth inhibition and results of a phase II trial. Cancer 1993; 72:3313-3317.

88. Myers, C. E., LaRocca, R., Stein, C., Cooper, M., Dawson, N., Choyke, P., Linehan, M., and Urich, M. Treatment of hormonally refractory prostate cancer with suramin. Proc.Am.Soc.Clin.Oncol. 9, 113. 1990.

89. Rapoport BL, Falkson G, Raats JI, de Wet M, Lotz BP, Potgieter HC. Suramin in combination with mitomycin C in hormone-resistant prostate cancer. A phase II clinical study. Ann.Oncol. 1993; 4:567-573.

90. Reyno LM, Egorin MJ, Eisenberger MA, Sinibaldi VJ, Zuhowski EG, Sridhara R. Development and validation of a pharmacokinetically based fixed dosing scheme for suramin. J.Clin.Oncol. 1995; 13:2187-2195.

91. Rosen PJ, Mendoza EF, Landaw EM et al. Suramin in hormone-refractory metastatic prostate cancer: a drug with limited efficacy. J.Clin.Oncol. 1996; 14:1626-1636.

92. Tkaczuk, K., Eisenberger, M. A., Sinibaldi, V., Reyno, L. M., Jodrell, D., Jacobs, S., Melink, T., Pearl, P., Aisner, J., Van Echo, D., and Egorin, M. Activity of suramin (Su) in prostate cancer (PC) observed in a phase I trial. 11, 201. 1992.

128 93. Betsholtz C, Johnsson A, Heldin CH, Westermark B. Efficient reversion of simian sarcoma virus-transformation and inhibition of growth factor-induced mitogenesis by suramin. Proc.Natl.Acad.Sci.U.S.A 1986; 83:6440-6444.

94. Coffey RJ, Jr., Leof EB, Shipley GD, Moses HL. Suramin inhibition of growth factor receptor binding and mitogenicity in AKR-2B cells. J.Cell Physiol 1987; 132:143-148.

95. De Clercq E. Suramin: a potent inhibitor of the reverse transcriptase of RNA tumor viruses. Cancer Lett. 1979; 8:9-22.

96. Eisenberger MA, Reyno LM. Suramin. Cancer Treat.Rev. 1994; 20:259-273.

97. Fortes PA, Ellory JC, Lew VL. Suramin: a potent ATPase inhibitor which acts on the inside surface of the sodium pump. Biochim.Biophys.Acta 1973; 318:262-272.

98. Funayama Y, Nishio K, Takeda Y et al. Suramin inhibits the phosphorylation and catalytic activity of DNA topoisomerase II in human lung cancer cells. Anticancer Res. 1993; 13:1981-1988.

99. Garrett JS, Coughlin SR, Niman HL, Tremble PM, Giels GM, Williams LT. Blockade of autocrine stimulation in simian sarcoma virus-transformed cells reverses down-regulation of platelet-derived growth factor receptors. Proc.Natl.Acad.Sci.U.S.A 1984; 81:7466-7470.

100. Grazioli L, Alzani R, Ciomei M et al. Inhibitory effect of suramin on receptor binding and cytotoxic activity of tumor necrosis factor alpha. Int.J.Immunopharmacol. 1992; 14:637-642.

101. Hawking F. Suramin: with special reference to onchocerciasis. Adv.Pharmacol.Chemother. 1978; 15:289-322.

102. Hensey CE, Boscoboinik D, Azzi A. Suramin, an anti-cancer drug, inhibits protein kinase C and induces differentiation in neuroblastoma cell clone NB2A. FEBS Lett. 1989; 258:156-158.

103. Hosang M. Suramin binds to platelet-derived growth factor and inhibits its biological activity. J.Cell Biochem. 1985; 29:265-273.

104. Manetti F, Cappello V, Botta M et al. Synthesis and binding mode of heterocyclic analogues of suramin inhibiting the human basic fibroblast growth factor. Bioorg.Med.Chem. 1998; 6:947-958.

105. Mills GB, Zhang N, May C, Hill M, Chung A. Suramin prevents binding of interleukin 2 to its cell surface receptor: a possible mechanism for immunosuppression. Cancer Res. 1990; 50:3036-3042.

129 106. Ono K, Nakane H, Fukushima M. Differential inhibition of various deoxyribonucleic and ribonucleic acid polymerases by suramin. Eur.J.Biochem. 1988; 172:349-353.

107. Pollak M, Richard M. Suramin blockade of insulinlike growth factor I-stimulated proliferation of human osteosarcoma cells. J.Natl.Cancer Inst. 1990; 82:1349- 1352.

108. Pollak, M. and Richard, M. Suramin interferes with the binding of insulin-like growth factor I (IGF-I) to human osteosarcoma cells and blocks IGF-I stimulated proliferation. Proc.Am.Soc.Clin.Oncol. 9, 54. 1990.

109. Wade TP, Kasid A, Stein CA et al. Suramin interference with transforming growth factor-beta inhibition of human renal cell carcinoma in culture. J.Surg.Res. 1992; 53:195-198.

110. Waltenberger J, Mayr U, Frank H, Hombach V. Suramin is a potent inhibitor of vascular endothelial growth factor. A contribution to the molecular basis of its antiangiogenic action. J.Mol.Cell Cardiol. 1996; 28:1523-1529.

111. Stiffey-Wilusz J, Boice JA, Ronan J, Fletcher AM, Anderson MS. An ex vivo angiogenesis assay utilizing commercial porcine carotid artery: modification of the rat aortic ring assay. Angiogenesis. 2001; 4:3-9.

112. Risau W. Mechanisms of angiogenesis. Nature 1997; 386:671-674.

113. Papetti M, Herman IM. Mechanisms of normal and tumor-derived angiogenesis. Am.J.Physiol Cell Physiol 2002; 282:C947-C970.

114. Conway EM, Collen D, Carmeliet P. Molecular mechanisms of blood vessel growth. Cardiovasc.Res 2001; 49:507-521.

115. Fidler IJ, Kerbel RS, Ellis LM. Biology of Cancer: Angiogenesis.In: Cancer: Principles and Practice of Oncology eds DaVita VT, Hellman S, Rosenberg SA, Philadelphia: Lippincott Williams and Wilkins, 2001.

116. Folkman J. Anti-angiogenesis: new concept for therapy of solid tumors. Ann.Surg. 1972; 175:409-416.

117. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000; 407:249-257.

118. Carmeliet P, Mackman N, Moons L et al. Role of tissue factor in embryonic blood vessel development. Nature 1996; 383:73-75.

130 119. Benjamin LE, Golijanin D, Itin A, Pode D, Keshet E. Selective ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal. J.Clin.Invest 1999; 103:159-165.

120. Chang YS, di Tomaso E, McDonald DM, Jones R, Jain RK, Munn LL. Mosaic blood vessels in tumors: frequency of cancer cells in contact with flowing blood. Proc.Natl.Acad.Sci.U.S.A 2000; 97:14608-14613.

121. Maniotis AJ, Folberg R, Hess A et al. Vascular channel formation by human melanoma cells in vivo and in vitro: vasculogenic mimicry. Am.J.Pathol. 1999; 155:739-752.

122. Polverini PJ, Novak RF. Inhibition of angiogenesis by the antineoplastic agents mitoxantrone and bisantrene. Biochem.Biophys.Res.Commun. 1986; 140:901-907.

123. Klement G, Huang P, Mayer B et al. Differences in therapeutic indexes of combination metronomic chemotherapy and an anti-VEGFR-2 antibody in multidrug-resistant human breast cancer xenografts. Clin.Cancer Res. 2002; 8:221-232.

124. Browder T, Butterfield CE, Kraling BM et al. Antiangiogenic scheduling of chemotherapy improves efficacy against experimental drug-resistant cancer. Cancer Res. 2000; 60:1878-1886.

125. Hanahan D, Bergers G, Bergsland E. Less is more, regularly: metronomic dosing of cytotoxic drugs can target tumor angiogenesis in mice. J.Clin.Invest 2000; 105:1045-1047.

126. Kerbel RS, Klement G, Pritchard KI, Kamen B. Continuous low-dose anti- angiogenic/ metronomic chemotherapy: from the research laboratory into the oncology clinic. Ann.Oncol. 2002; 13:12-15.

127. Bartels PH. Quantitation in Histopathology: Objectives, Origins, Digital Image Analysis, and Unresolved Issues.In: Image analysis: a primer for pathologists eds Marchevsky AM, Bartels PH, New York: Raven Press, 1994: 1-28.

128. Russ J. Aquiring Images.In: The Image Processing Handbook Boca Raton: CRC Press, 2002: 1-77.

129. Yen WC, Wientjes MG, Au JL. Differential effect of taxol in rat primary and metastatic prostate tumors: site-dependent pharmacodynamics. Pharm.Res 1996; 13:1305-1312.

130. Liotta LA, Stetler-Stevenson WG. Principles of molecular cell biology of cancer: cancer metastasis.In: Principles & Practice of Oncology, 4 Edition, eds DeVita VT, Hellman S, Rosenberg SA, Philadelphia: J. B. Lippincott Co., 1993: 134- 149. 131 131. Fidler IJ, Hart IR. Principles of cancer biology: biology of cancer metastasis.In: Principles & Practice of Oncology eds DeVita VT, Hellman S, Rosenberg SA, Philadelphia: J. B. Lippincott Co., 1990: 80-92.

132. Dong Z, Radinsky R, Fan D et al. Organ-specific modulation of steady-state mdr gene expression and drug resistance in murine colon cancer cells. J.Natl.Cancer Inst. 1994; 86:913-920.

133. Fidler IJ, Wilmanns C, Staroselsky A, Radinsky R, Dong Z, Fan D. Modulation of tumor cell response to chemotherapy by the organ environment. Cancer Metastasis Rev. 1994; 13:209-222.

134. Ferrigno D, Buccheri G. Second-line chemotherapy for recurrent non-small cell lung cancer: do new agents make a difference? Lung Cancer 2000; 29:91-104.

135. Sweeney CJ, Sandler AB. Chemotherapy in non-small cell lung cancer. Invest New Drugs 2000; 18:157-186.

136. Ware JL, Lieberman AP, Webb KS, Vollmer RT. Factors influencing phenotypic diversity of human prostate carcinoma cells metastasizing in athymic nude mice. Exp.Cell Biol. 1985; 53:163-169.

137. Gan Y, Wientjes MG, Schuller DE, Au JL. Pharmacodynamics of taxol in human head and neck tumors. Cancer Res 1996; 56:2086-2093.

138. Gold R, Schmied M, Giegerich G et al. Differentiation between cellular apoptosis and necrosis by the combined use of in situ tailing and nick translation techniques. Lab Invest 1994; 71:219-225.

139. Chahinian AP, Mandeli JP, Gluck H, Naim H, Teirstein AS, Holland JF. Effectiveness of cisplatin, paclitaxel, and suramin against human malignant mesothelioma xenografts in athymic nude mice. J.Surg.Oncol. 1998; 67:104-111.

140. Shin R, Naomoto Y, Kamikawa Y, Tanaka N, Orita K. Effect of suramin on human esophageal cancer cells in vitro and in vivo. Scand.J.Gastroenterol. 1997; 32:824-828.

141. Geier A, Hemi R, Haimsohn M, Beery R, Malik Z, Karasik A. Epidermal growth factor and insulin-like growth factor-1 protect MDA-231 cells from death induced by actinomycin D: the involvement of growth factors in drug resistance. In Vitro Cell Dev.Biol.Anim 1994; 30A:336-343.

142. Guo YS, Jin GF, Houston CW, Thompson JC, Townsend CM, Jr. Insulin-like growth factor-I promotes multidrug resistance in MCLM colon cancer cells. J.Cell Physiol 1998; 175:141-148.

132 143. Shaulian E, Resnitzky D, Shifman O et al. Induction of Mdm2 and enhancement of cell survival by bFGF. Oncogene 1997; 15:2717-2725.

144. Small EJ, Meyer M, Marshall ME et al. Suramin therapy for patients with symptomatic hormone-refractory prostate cancer: results of a randomized phase III trial comparing suramin plus hydrocortisone to placebo plus hydrocortisone. J.Clin.Oncol. 2000; 18:1440-1450.

145. Bitton RJ, Figg WD, Venzon DJ et al. Pharmacologic variables associated with the development of neurologic toxicity in patients treated with suramin. J.Clin.Oncol. 1995; 13:2223-2229.

146. Figg WD, Cooper MR, Thibault A et al. Acute renal toxicity associated with suramin in the treatment of prostate cancer. Cancer 1994; 74:1612-1614.

147. Horne MK, III, Stein CA, LaRocca RV, Myers CE. Circulating glycosaminoglycan anticoagulants associated with suramin treatment. Blood 1988; 71:273-279.

148. Kobayashi K, Weiss RE, Vogelzang NJ, Vokes EE, Janisch L, Ratain MJ. Mineralocorticoid insufficiency due to suramin therapy. Cancer 1996; 78:2411- 2420.

149. La Rocca RV, Meer J, Gilliatt RW et al. Suramin-induced polyneuropathy. Neurology 1990; 40:954-960.

150. Seidman AD, Schwartz M, Reich L, Scher HI. Immune-mediated thrombocytopenia secondary to suramin. Cancer 1993; 71:851-854.

151. Stein CA, Saville W, Yarchoan R, Broder S, Gelmann EP. Suramin and function of the adrenal cortex. Ann.Intern.Med. 1986; 104:286-287.

152. Suramin sodium.In: Cancer Chemotherapy Handbook, 2 Edition, eds Dorr R, Von Hoff D, Norwalk: Appleton & Lange, 1994: 859-866.

153. Lowitt MH, Eisenberger M, Sina B, Kao GF. Cutaneous eruptions from suramin. A clinical and histopathologic study of 60 patients. Arch.Dermatol. 1995; 131:1147-1153.

154. Jodrell DI, Reyno LM, Sridhara R et al. Suramin: development of a population pharmacokinetic model and its use with intermittent short infusions to control plasma drug concentration in patients with prostate cancer. J.Clin.Oncol. 1994; 12:166-175.

155. Qiao L, Pizzolo JG, Melamed MR. Effects of suramin on expression of proliferation associated nuclear antigens in DU-145 prostate carcinoma cells. Biochem.Biophys.Res Commun. 1994; 201:581-588. 133 156. Tu SM, Pagliaro LC, Banks ME et al. Phase I study of suramin combined with doxorubicin in the treatment of androgen-independent prostate cancer. Clin.Cancer Res 1998; 4:1193-1201.

157. Gan Y, Wientjes MG, Au JL. Relationship between paclitaxel activity and pathobiology of human solid tumors. Clin.Cancer Res 1998; 4:2949-2955.

158. Furukawa T, Kubota T, Hoffman RM. Clinical applications of the histoculture drug response assay. Clin.Cancer Res 1995; 1:305-311.

159. Kubota T, Sasano N, Abe O et al. Potential of the histoculture drug-response assay to contribute to cancer patient survival. Clin.Cancer Res 1995; 1:1537- 1543.

160. Robbins KT, Connors KM, Storniolo AM, Hanchett C, Hoffman RM. Sponge- gel-supported histoculture drug-response assay for head and neck cancer. Correlations with clinical response to cisplatin. Arch.Otolaryngol.Head Neck Surg. 1994; 120:288-292.

161. Donaldson KL, Goolsby GL, Wahl AF. Cytotoxicity of the anticancer agents cisplatin and taxol during cell proliferation and the cell cycle. Int.J.Cancer 1994; 57:847-855.

162. Liebmann J, Cook JA, Lipschultz C, Teague D, Fisher J, Mitchell JB. The influence of Cremophor EL on the cell cycle effects of paclitaxel (Taxol) in human tumor cell lines. Cancer Chemother.Pharmacol. 1994; 33:331-339.

163. Lopes NM, Adams EG, Pitts TW, Bhuyan BK. Cell kill kinetics and cell cycle effects of taxol on human and hamster ovarian cell lines. Cancer Chemother.Pharmacol. 1993; 32:235-242.

164. Horwitz SB. Mechanism of action of taxol. Trends Pharmacol.Sci. 1992; 13:134- 136.

165. Tang C, Willingham MC, Reed JC et al. High levels of p26BCL-2 oncoprotein retard taxol-induced apoptosis in human pre-B leukemia cells. Leukemia 1994; 8:1960-1969.

166. Lyness, Greg. Chemosensitization of Urologic Cancers by FGF Inhibitors. 2005. The Ohio State University.

167. Rowinsky EK, Wright M, Monsarrat B, Lesser GJ, Donehower RC. Taxol: pharmacology, metabolism and clinical implications. Cancer Surv. 1993; 17:283- 304.

134 168. Ruiz JA, Raftopoulos H, Petrylak DP. Chemotherapy for Metastatic Reanl Cell Carcinoma (RCC).In: Renal Cell Carcinoma: Molecular Biology, Immunology, and Clinical Management eds Bukowski RM, Novick AC, Totowa, New Jersey: Humana Press, 2000: 283-300.

169. Zhang, Y. Induction of FGF as a Mechanism of Acquired Anticancer Drug Resistance. 2002. The Ohio State University.

170. Ludden TM, Beal SL, Sheiner LB. Comparison of the Akaike Information Criterion, the Schwarz criterion and the F test as guides to model selection. J.Pharmacokinet.Biopharm. 1994; 22:431-445.

171. Greenblatt MS, Bennett WP, Hollstein M, Harris CC. Mutations in the p53 tumor suppressor gene: clues to cancer etiology and molecular pathogenesis. Cancer Res 1994; 54:4855-4878.

172. Yu, B., Song, S., Wientjes, M. G., and Au, J. L-S. Suramin enhances activity of CPT-11 in human colorectal xenograft tumors. Proc.Am.Assoc.Cancer Res. 44. 2003.

173. Zhao, L., Song, S., Wientjes, M. G., and Au, J. L-S. Enhancement of antitumor activity of paclitaxel by suramin: unconventional dose-response relationship. Proc.Am.Assoc.Cancer Res. 43. 2001.

174. Weaver JR, Au JL. Application of automatic thresholding in image analysis scoring of cells in human solid tumors labeled for proliferation markers. Cytometry 1997; 29:128-135.

175. Weaver JR, Au JL. Comparative scoring by visual and image analysis of cells in human solid tumors labeled for proliferation markers. Cytometry 1997; 27:189- 199.

176. Weaver JR, Gan Y, Au JL. Proliferation indices as molecular pharmacodynamic endpoints in evaluation of anticancer drug effect in human solid tumors. Pharm.Res 1998; 15:1546-1551.

177. Taraboletti G, Giavazzi R. Modelling approaches for angiogenesis. Eur.J.Cancer 2004; 40:881-889.

178. Freeman AE, Hoffman RM. In vivo-like growth of human tumors in vitro. Proc.Natl.Acad.Sci.U.S.A 1986; 83:2694-2698.

179. Furukawa T, Kubota T, Watanabe M et al. High in vitro-in vivo correlation of drug response using sponge-gel-supported three-dimensional histoculture and the MTT end point. Int.J.Cancer 1992; 51:489-498.

135 180. Barltrop JAeal. 5-(3-carboxymethoxyphenyl)-2-(4,5-dimethylthiazoly)-3-(4- sulfophenyl)tetrazolium, inner salt (MTS) and related analogs of 3-(4,5- dimethylthiazolyl)-2,5-diphenyltetrazolium bromide (MTT) reducing to purple watersoluble formazans as cell-viability indicators. Bioorg.& Med.Chem.Lett. 1991; 1:611.

181. Cory AH, Owen TC, Barltrop JA, Cory JG. Use of an aqueous soluble tetrazolium/formazan assay for cell growth assays in culture. Cancer Commun. 1991; 3:207-212.

182. Riss TL, Moravec RA. Comparison of MTT, XTT, and a novel tetrazolium compound MTS for in vitro proliferation and chemosensitivity assays. Mol.Biol.Cell 1992; 3 (Suppl.):184a.

183. Johnston JS, Johnson A, Gan Y, Wientjes MG, Au JL. Synergy between 3'-azido- 3'-deoxythymidine and paclitaxel in human pharynx FaDu cells. Pharm.Res 2003; 20:957-961.

184. Berenbaum MC. What is synergy? Pharmacol.Rev. 1989; 41:93-141.

185. Chou TC, Talalay P. Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv.Enzyme Regul. 1984; 22:27-55.

186. Furukawa T, Kubota T, Murata H et al. Antitumor spectra of anthracyclines against gastric cancer tissues obtained from surgical specimens with reference to P-glycoprotein expression. J.Surg.Oncol. 1998; 69:173-177.

187. Yang QF, Sakurai T, Yoshimura G et al. Expression of Bcl-2 but not Bax or p53 correlates with in vitro resistance to a series of anticancer drugs in breast carcinoma. Breast Cancer Res.Treat. 2000; 61:211-216.

188. Beckstead JH. A simple technique for preservation of fixation-sensitive antigens in paraffin-embedded tissues. J.Histochem.Cytochem. 1994; 42:1127-1134.

189. Vanzulli S, Gazzaniga S, Braidot MF, Vecchi A, Mantovani A, Wainstok dC. Detection of endothelial cells by MEC 13.3 monoclonal antibody in mice mammary tumors. Biocell 1997; 21:39-46.

190. Vecchi A, Garlanda C, Lampugnani MG et al. Monoclonal antibodies specific for endothelial cells of mouse blood vessels. Their application in the identification of adult and embryonic endothelium. Eur.J.Cell Biol. 1994; 63:247-254.

191. Lee YM, Jeong CH, Koo SY et al. Determination of hypoxic region by hypoxia marker in developing mouse embryos in vivo: a possible signal for vessel development. Dev.Dyn. 2001; 220:175-186.

136 192. Vermeulen PB, Gasparini G, Fox SB et al. Quantification of angiogenesis in solid human tumours: an international consensus on the methodology and criteria of evaluation. Eur.J.Cancer 1996; 32A:2474-2484.

193. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angiogenesis and metastasis--correlation in invasive breast carcinoma. N.Engl.J.Med. 1991; 324:1- 8.

194. Brem S, Cotran R, Folkman J. Tumor angiogenesis: a quantitative method for histologic grading. J.Natl.Cancer Inst. 1972; 48:347-356.

195. Weidner N, Folkman J, Pozza F et al. Tumor angiogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma. J.Natl.Cancer Inst. 1992; 84:1875-1887.

196. Weidner N, Carroll PR, Flax J, Blumenfeld W, Folkman J. Tumor angiogenesis correlates with metastasis in invasive prostate carcinoma. Am.J.Pathol. 1993; 143:401-409.

197. Borre M, Offersen BV, Nerstrom B, Overgaard J. Microvessel density predicts survival in prostate cancer patients subjected to watchful waiting. Br.J.Cancer 1998; 78:940-944.

198. de Jong JS, van Diest PJ, Baak JP. Hot spot microvessel density and the mitotic activity index are strong additional prognostic indicators in invasive breast cancer. Histopathology 2000; 36:306-312.

199. Villalona-Calero MA, Wientjes MG, Otterson GA et al. Phase I study of low-dose suramin as a chemosensitizer in patients with advanced non-small cell lung cancer. Clin.Cancer Res. 2003; 9:3303-3311.

200. Zhao, L. Evaluation of Combination Cancer Chemotherapy: Theory and Practice. 2003. The Ohio State University.

201. Hu, Leijun. Suramin Pharmacokinetics After Regional or Systemic Administration. 2005. The Ohio State University.

202. Brown KJ, Maynes SF, Bezos A, Maguire DJ, Ford MD, Parish CR. A novel in vitro assay for human angiogenesis. Lab Invest 1996; 75:539-555.

203. Trotter MJ, Chaplin DJ, Olive PL. Use of a carbocyanine dye as a marker of functional vasculature in murine tumours. Br.J.Cancer 1989; 59:706-709.

204. Hardman WE, Moyer MP, Cameron IL. Fish oil supplementation enhanced CPT- 11 (irinotecan) efficacy against MCF7 breast carcinoma xenografts and ameliorated intestinal side-effects. Br.J.Cancer 1999; 81:440-448.

137 205. Guichard S, Chatelut E, Lochon I, Bugat R, Mahjoubi M, Canal P. Comparison of the pharmacokinetics and efficacy of irinotecan after administration by the intravenous versus intraperitoneal route in mice. Cancer Chemother.Pharmacol. 1998; 42:165-170.

206. Krenik KD, Kephart GM, Offord KP, Dunnette SL, Gleich GJ. Comparison of antifading agents used in immunofluorescence. J.Immunol.Methods 1989; 117:91- 97.

207. Song S, Wientjes MG, Walsh C, Au JL. Nontoxic doses of suramin enhance activity of paclitaxel against lung metastases. Cancer Res. 2001; 61:6145-6150.

208. Zhang Y, Song S, Yang F, Au JLS, Wientjes MG. Nontoxic doses of suramin enhance activity of doxorubicin in prostate tumors. J.Pharmacol.Exp.Ther. 2001; 299:426-433.

138