This item has the following downloads:
1 COMBINATION OF FIRST PASS FLUORESCENCE IMAGING WITH HYPERSPECTRAL IMAGING OF HEMOGLOBIN SATURATION FOR WIDE FIELD IN VIVO ANALYSIS OF MICROVESSEL BLOOD SUPPLY AND OXYGENATION By JENNIFER AMY LEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Jennifer Amy Lee
3 To my family King, Phuc, Kim, Ryan, Phil and Gizmo
4 ACKNOWLEDGMENTS I would like to thank my advisor Dr. Brian S. Sorg for his direction and support in completing my Ph.D. It is because of his mentorship that I have learned how to be a successfully, independent research er. I will forever value the lessons he has taught me and the advice he has given. I am indebted to my committee members, Dr. Brandi K Ormerod, Dr. Benjamin Keselowsky and Dr. David J. Hahn, for their valuable insight in my research. I am especially grate ful to Dr. Ormerod for being my fost er advisor and for always having my back as I finished my work I have the utmost appreciation for Dr. Dietmar W. Siemann for his guidance as a mentor and a collaborator as well as all members of the Siemann Lab for their help in completing my dissertation work. I have special thanks to Dr. Nikolett Molnar for her counsel both as a colleague and as a friend. I am whole heartedly appreciative of my Sorg lab members, Dr. Raymond Kozikowski, Dr. Sewoon Choe and Dr. Mamta Wank hede for their help and contribution to my work and making lab fun I am grateful for my friends and family both near and far away. I thank my friends Amy Dudas, Aditya Asokan and Eric Franca, for the encouragement and fun times throughout this process. I am also grateful for the Jannotti family for being my home away from home. I could not have completed my graduate work without my family and for that I am forever grateful for them To m y Mom and Dad for always being positive about my work and ha ving conf idence in my abilities to accomplish my dreams To m y sister Kim for being a great role model and always giving me someone to look up to. T o m y little
5 bro ther Ryan, for always reminding me to stay a kid at heart. Last but not least, I thank Phillip Jannott i for his love, support, and always believing in me
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 MOTIVATION AND GOAL OF RESEARCH ................................ ........................... 15 2 TUMOR MICROVASCULATURE: TARGETING AND IN VIVO MICROSCOPY .... 19 Tumor Microvasculature ................................ ................................ ......................... 19 Vascular Targeting Therapies ................................ ................................ ................. 20 Vascular Disrupting Agents ................................ ................................ .............. 21 Angiogenic Inhibiting Agents ................................ ................................ ............ 23 Combination of Vascular Targeting Agents ................................ ...................... 24 In Vivo Imaging of Tumor Microvasculature and Microenvironment ....................... 25 3 DEVELOPMENT OF FIRST PASS FLUORESCENCE IMAGING AND ITS COMBINATION WITH HYPERSPECTRAL IMAGING OF HEMOGLOBIN SATURATION ................................ ................................ ................................ ......... 29 Introduction ................................ ................................ ................................ ............. 29 Murine Dorsal Skinfold Window Chamber Mod el ................................ .................... 30 Hyperspectral Imaging of Hemoglobin Saturation ................................ ................... 31 Development of First Pass Fluorescence Imaging ................................ .................. 32 Imaging System ................................ ................................ ................................ 32 Development of Fluorescent Tracer ................................ ................................ 32 FPF Imaging Acquisition ................................ ................................ ................... 37 FPF Image Processing ................................ ................................ ..................... 37 In Vivo Testing ................................ ................................ ................................ ........ 39 Tumor Cel ls ................................ ................................ ................................ ...... 39 Animal Model ................................ ................................ ................................ .... 40 Experimental Design ................................ ................................ ........................ 40 Results ................................ ................................ ................................ .................... 40 Discussion ................................ ................................ ................................ .............. 44
7 4 COMBINATION OF SPECTRAL AND FLUORESENCE IMAGING MICROSCOPY FOR WIDE FIELD IN VIVO ANALYSIS OF MICROVESSEL BLOOD SUPPLY AND OXYGENATION OVER TIME ................................ ............ 60 Introduction ................................ ................................ ................................ ............. 60 Materials and Methods ................................ ................................ ............................ 61 Tumor Cells ................................ ................................ ................................ ...... 61 Animal Model ................................ ................................ ................................ .... 62 Flu orescent Liposome Preparation ................................ ................................ ... 63 Experimental Design ................................ ................................ ........................ 63 Hb Saturation Imaging and Image Processing ................................ ................. 64 FPF Imaging and Image Processing ................................ ................................ 65 Results ................................ ................................ ................................ .................... 66 Discussion ................................ ................................ ................................ .............. 70 5 IN VIVO MICROSCOPY COMPARISON OF MICROVASCULAR FUNCTION AFTER TREATMENT WITH OXI4503, SUNITINIB, AND THEIR COMBINATION IN CAKI 2 TUMORS ................................ ................................ ..... 84 Introduction ................................ ................................ ................................ ............. 84 Materials and Methods ................................ ................................ ............................ 85 Tumor Cells ................................ ................................ ................................ ...... 85 Dorsal Skinfold Window Chamber Preparation ................................ ................ 86 Drug Prepar ation and Experimental Design ................................ ..................... 86 Fluorescent Liposome Preparation ................................ ................................ ... 87 Intravital Microscopy ................................ ................................ ......................... 88 Hb Saturation Imaging and Analysis ................................ .......................... 88 FPF Imaging and Analysis ................................ ................................ ......... 88 Tumor Volume Calculation ................................ ................................ ............... 89 Immunohistochemical Analysis of Window Chamber Tumors .......................... 90 Statistical Analysis ................................ ................................ ............................ 90 Results ................................ ................................ ................................ .................... 91 Morphological Changes of Tumor Blood Vessels ................................ ............. 91 Tumor Growth Inhibition ................................ ................................ ................... 92 Immunohistochemical Analysis ................................ ................................ ........ 93 FPF and Hb Saturation Imaging of Blood Flow and Oxygenation .................... 94 Region of Interest Analysis ................................ ................................ ........ 95 Box and Whisker Plot Analysis ................................ ................................ .. 96 Discussion ................................ ................................ ................................ .............. 97 6 CONCLUSION AND FUTURE DIRECTION ................................ ......................... 114 APPENDIX MATLAB CODE FOR FPF IMAGING ANALYSIS .............................. 118 LIST OF REFERENCES ................................ ................................ ............................. 123 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 133
8 LIST OF TABLES Table page 1 1 Clinical t rials i nvolving OXi4503 ................................ ................................ .......... 22 1 2 Combination of VDAs and AIAs ................................ ................................ .......... 25 3 1 Initial fluorescent tracer attempts ................................ ................................ ........ 34
9 LIST OF FIGURES Figure page 3 1 Murine dorsal window chamber model ................................ ............................... 48 3 2 Overview of hyperspectral imaging of hemoglobin saturation ............................ 48 3 3 Microscope setup for window chamber imaging ................................ ................. 49 3 4 Fluorescence images of SPHERO TM sky blue nanoparticle flow through normal vasculature illustrate the low signal intensity and aggregation of particles ................................ ................................ ................................ .............. 49 3 5 Comparison of TRITC labeled dextrans and DiD labeled liposomes for BST imaging ................................ ................................ ................................ ............... 50 3 6 Extravasated liposomes 24 hours after injection hinder the ability to perform repeated measurements ................................ ................................ ..................... 51 3 7 Improved clearance of liposomes containing DOPS ................................ ........... 52 3 8 Preliminary testing of FPF imaging MATLAB code to determine BST from first pass fluorescence images ................................ ................................ ........... 53 3 9 Overview of BST map creation from first pas s fluorescence imaging ................. 54 3 10 Six frames from first pass imaging of a tail vein injection of ~50 l of DiD liposomes through the vasculature of normal and 4T1 tumor tissue ................... 55 3 11 Single day Hb saturation and BST comparison of tumor vasculature versus normal vasculature ................................ ................................ ............................. 56 3 12 Hb saturation and BST ROI analysis for norm al vasculature reveals a larger number of vessels with lower oxygenation and slower BSTs ............................. 57 3 13 Hb saturation and BST ROI analysis for a 4T1 tumor shows the development of vessels with faster BSTs and higher oxygenations ................................ ......... 58 3 14 ROI analysis of the draining vein of a 4T1 tumor shows the effects of arteriovenous shunts on blood flow and oxygenation ................................ ......... 59 4 1 Six frames from FPF imaging of DiD liposomes through the vasculature of a Caki 2 tumor. ................................ ................................ ................................ ...... 76 4 2 Analysis of a Caki 2 tumor over 4 days illustrates the rapid ch anges of vessel structure and the effects on Hb saturation and BST ................................ ........... 77
10 4 3 ROI analysis for the draining vein of a Caki 2 tumor shows that AV shunts are responsible for faster BSTs and increasin g oxygenation .............................. 78 4 4 FPF and Hb saturation maps show that tumors increase the oxygenation and create faster BSTs in surrounding normal vessels with time .............................. 79 4 5 FPF and Hb saturation imaging analysis of PC3 ML tumor development over four days illustrates the differences in tumor vessel development in varying tumor types ................................ ................................ ................................ ......... 80 4 6 Long term FPF and Hb saturation imaging analysis of normal tissue over 9 days shows the transient nature of normal blood flow ................................ ........ 81 4 7 Combination imaging analysis of a puncture wound charac terizes the tightly regulated process of wound healing ................................ ................................ ... 82 4 8 Combination imaging analysis of Caki 2 tumor initiation il lustrates how tumor cells develop vasculature for further proliferation ................................ ............... 83 5 1 Image acquisition and drug administration schedule ................................ ........ 104 5 2 Daily brightfield images of the tumor area show morphological chan ges to tumor vasculature in response to treatment ................................ ..................... 105 5 3 Daily tumor volume measurements of tumor nodules in window chambers (median interquartile range) reveal enhanced growth inhibition wit h combination treatment ................................ ................................ ...................... 106 5 4 Immunohistologic analysis of tumor vasculature shows the inhibition of vascular development among treatment groups ................................ ............... 107 5 5 Brightfield, Hb saturation, and BST maps for a control tumor reveals with unhindered growth, blood supply times become longer and oxygenation decreases ................................ ................................ ................................ ......... 108 5 6 Brightfield Hb saturation, and BST maps for an OXi4503 treated tumor show Caki 2 vasculature overcoming repeated treatment ................................ ......... 109 5 7 Brightfield, Hb saturation, and BST maps for a Sunitinib treated tumor il lustrates vascular normalization for the duration of this experiment ................ 110 5 8 Brightfield, Hb saturation, and BST maps for a combination treated tumor demonstrates the enhanced antivascular effe cts of the treatment .................... 111 5 9 ROI analysis of Hb saturation and BST maps for each experimental group ..... 112 5 10 Boxplot analysis of H b saturation and BST maps show that treated tumors inhibit the formation of lower oxygenation values and slower BSTs. ................ 113
11 LIST OF ABBREVIATIONS AIA Angiogenic Inhibiting Agent AML Acute Myelogenous Leukemia AV Arteriovenou s BST Blood Supply T ime CA1 Combretastatin A1 CA1 P OXi4503 CA4P Compretastatin A4 Prodrug CCD Charge Coupled Device CT Computed Tomography CEP Circulating Endothelial Progenitors DAPI diamidino 2 phenylindole DID 1,1' dioctadecyl 3,3,3',3' t etramethyl indodicarbocyanine 4 chlorobenzenesulfonate DIR 1,1' dioctadecyl 3,3,3',3' tetramethylindotricarbocyanine iodide DOPC 1,2 dioleoyl sn glycero 3 phosphocholine DOPS 1,2 dioleoyl sn glycero 3 phospho L serine DSPC mPEG2000 1,2 distearoyl sn glycero 3 phosph oethanolamine N [methoxy(polyethyleneglycol) 2000] DV Draining Vein EMCCD Electron Multiplying Charge Coupled Device FLT3 Fetal Liver Tyrosine Kinase Receptor 3 FPF First Pass Fluorescence FRAP Fluorescence Recovery After Photobleaching GIST Gastrointesti nal Stromal Tumor H B Hemoglobin
12 IHC Immunohistochemical IR Infrared KIT Stem Cell Factor Receptor LCTF Liquid Crystal Tunable Filter LLC Lewis Lung Carcinoma MECA Anti Mouse Panendothelial Cell Antigen MRI Magnetic Resonance Imaging NSCLC Non Small Cell Lu ng Cancer OCT Optical Cutting Temperature PBS Phosphate Buffered Saline PDGF Platelet Derived Growth Factor PDGFR Platelet Derived Growth Factor Receptor PEG Polyethylene G lycol PET Positron Emission Tomography RCC Renal Cell Carcinoma ROI Region of Intere st SA Supplying Artery TBS Tris Buffered Saline TKI Tyrosine Kinase Inhibitor TRITC Tetramethylrhodamine 5 (and 6) isothiocyanate VDA Vascular Disrupting Agent VEGF Vascular Endothelial Growth Factor VEGF R Vascular Endothelial Growth Factor Receptor VTA V ascular Targeting Agent
13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COMBINATION OF FIRST PASS FLUORESCENCE IMAG ING WITH HYPERSPECTRAL IMAGING OF HEMOGLOBIN SATURATION FOR WIDE FIELD IN VIVO ANALYSIS OF MICROVESSEL BLOOD SUPPLY AND OXYGENATION By Jennifer Amy Lee August 2013 Chair: Brian S. Sorg Cochair: Brandi K. Ormerod Major: Biomedical Engineering The occurr ence of a bnormal angiogenesis and microvasculature with altered function and morphology is characteristic in a wide variety of pathologies and in many instances lead s to the propagation of these diseases Specifically in the case of cancer, t he identific ation of the microvasculature as a potential target f or tumor treatment has led to drastic increase s in research and development of vascular targeting agents (VTAs) in the past decad e. Two classes of VTAs have emerged from this research, v ascular disruptin g agents which directly target the rapidly dividing endothelial cells of tumor vessels to destroy existing vasculature and angiogenic inhibiting agents which halt the formation of vasculature by disrupting molecular pathways that lead to angiogenesis O ver all as single agents these drugs have failed to produce improved tumor therapy in comparison to current standards of care. However, because of their complementary mechanisms, the investigation into the use of these drugs in combination has become of recent interest. To better characterize the tumor vessel response to these agents and obtain the preclinical data necessary to provide a basis for clinical timing and dosing administration, the use of small animal models and the
14 direct observation of tumor vasc ulature is essential to analyze the relationship between vessel morphology, blood flow and oxygenation. We present the novel combination of two widefield imaging techniques, hyperspectral imaging of hemoglobin (Hb) saturation and first pass fluorescence (F PF) imaging in the murine window chamber model, to simultaneously characterize the oxygenation of and blood flow through microvessel networks over time. This combination of imaging techniques enables a unique analysis of blood flow in microvessels that hig hlights the effects of microvessel network connections on oxygen transport. In this study we focus on the development of the FPF imaging technique its combination with the established method of hyperspectral imaging of Hb saturation and finally the applic ation of this combination imaging to the evaluation of microvasculature in growing tumors and in response to VTAs.
15 CHAPTE R 1 MOTIVATION AND GOAL OF RESEARCH Abnormal vasculature is a characteristic of various pathologies such as diabetes, arthritis and cancer. The irregular growth and developm ent of vessels is thought to propagate the severity of these conditions [1, 2] U nders tanding the process of angiogenesis as well as vascular structure and function in t hese diseases is therefore imperative for further development of vasc ular target ed therapies. Advancements in small animal models of vascularization, angiogenesis and preclinical imaging analysis have made it possible to evaluate pathological vasculature o ver time in a non invasive or minimally invasive manner. The murine dorsal skin window chamber is one such animal model that allows for the serial observation of developing vasculature over time [3, 4] To imag e the vasculature of these animals, various forms of intravital imaging have been developed Transmitted light, fluorescence microscopy, confocal laser scanning microscopy and multi photon microscopy are a few examples of the methods currently employed for the analysis of microvessels in the window chamber [5 9] One specific application for the window chamber model and accompanying imaging methods is the eval uation of tumor microvasculature In 1971, Judah Fol kman proposed the idea that tumor growth is angiogenesis dependent and stifling vascular formation could halt the growth of a tumor  This observation coupled with the fac t that tumor vasculature presents an easily accessible, specific target for therapy resulted in the rapid growth of research into vas cular targeting agents as cancer therapy research that continues to this day There are two main groups of VTAs d isting uished by their method of actio n. Va scular disrupting agents (VDAs) disrupt or destroy existing vasculature directly by targeting the rapidly dividing endothelial cells. Angiogenic
16 inhibiting agents (AIAs) inhibit the formation of tumor vasculature by targeting receptors and molecular pathwa ys that signal for the formation of vasculature  Both of these classes of VTAs have shown to p roduce insufficient antitumor and antivascular effects as singular agents in the clinic, thus current preclinical and clinical studies have begun to use these drugs in combination. Because VDAs destroy existing vasculature and AIAs inhibit new vessels from forming, the combination of the two drugs is hypothesized to produce a greater antitumor and antivascular effect than either drug alone. To further develop and optimize these agents, t he evaluation of tumor microvasculature development and response to VTA s in preclinical models is essential. Window chamber models have been previously used to analyze both VDAs and AIAs as single agents however investigation into the use of these drugs in combination with this model is limited [12 15] Hyperspectral imaging of hemoglobin (Hb) saturation was previously used to analyze the vascular response of 4T1 mouse mammary adeno carcinoma to treatment with the VDA OXi4503. This analysis was then applied to the response of Caki 1 renal cell carcinoma (RCC) tumors to treatment with the VDA OXi4503 and AIA Bevacizumab ( Av astin ) in combination Distinct oxygenation and structural changes were observed after treatment with OXi4503, however treatment with Bevacizumab and the combinati on of the two drugs produced more convoluted results. Bevacizumab showed little change in oxygenation and the combination treatment closely resembled the oxygenation patterns of OXi4503 treated tumors, which led the researchers to question the impact of Be vacizumab in the combination treatment  Moreover, beyond
17 structural and oxygenation data, no information regarding blood flow through these microvessels was collected with this imaging technique. First pa ss fluorescence (FPF) imaging was used previously to characterize the blood flow in tumors of the murine dorsal window chamber  This method of intravital imaging entails an injection of a fl uorescent tracer and simultaneously recording the flow of the tracer as it first enters the field of view in the window chamber. Through post processing of the image stack, a blood supply time (BST) can be assigned to vessels in the field of view. The BST indicates the time it takes for blood to travel to specific vessels in the field of view from a single supplying artery and can be used to evaluate changes in blood flow in a growing tumor. The use of this technique has only been applied as a singular imag ing technique, and thus lacked the ability to attain oxygenation data of these vessels. To further clarify the mechanisms of VTAs and the responses they elicit in tumor microvasculature the goal for this work was to develop our own FPF imaging technique to analyze the blood transit time o f microvessels and combine this technique with hyperspe ctral imaging of Hb saturation to create a novel imaging method for the evaluation of microvasculature in the mouse dorsal skin fold window chamber model. Furthermore we wanted to adapt this combination method to evaluate dynamic functional and oxygenation changes in the microvasculature of different conditions over time. Ultimately, we wanted to utilize this technique to evaluate the microvascular effects of VDAs and AIAs used in combination in comparison to the use of either drug alone.
18 In Chapter 2 w e summarize background data surrounding tumor microvasculature, vascular targeting agents and preclinical methods of ev aluating tumor vessel response The full developm ent of FPF imaging and its combination with hyperspectral imaging of Hb saturation imaging is discussed in Chapter 3. This involve d the discovery and development of fluorescent tracers that would be appropriate for combination imaging, the creation of MATL AB coding to create BST maps and the testing of the method in vivo. Once the method was established we optimized the fluorescent tracers and imaging process to be applied for serial imaging of vessels in the window chamber discussed in Chapter 4. Tumor g rowth, tumor initiation, wound heal ing and normal vessels were all analyzed as proof of principle for the use of this combination method over time. In Chapter 5 we applied our combination imaging modality to analyze the response of tumor vasculature to VDA s AIAs and a combination of the two drugs. Specifically, we utilized Caki 2 RCC tumors, the VDA OXi4503 and the AIA Sunitinib for our analysis. Using our method we were able to show distinct blood flow and oxygenation responses to each drug treatment. In addition we demonstrated t he enhanced antitumor and anti vascular activity of using both drugs in combination in co mparison to either drug alone. Finally i n Chapter 6 we conclude our research and discuss our future directions and applications for our imagin g modality.
19 CHAPTER 2 TUMOR MICROVASCULATURE: TARGETING AND IN VIVO MICROSCOPY Tumor Microvasculature The creation and development of new blood vessels, angiogenesis, is a tightly regulated process in normal tissues Embryogenesis, organ growth, wound he aling an d tissue repair all necessitate angiogen esis for successful completion. Angiogenesis in these processes is regulated by a balance of pro angiogenic and anti angiogenic factors [17, 18] In contrast t um or microv asculature is initiated by an abundance of pro angiogenic factors that results in pathological angiogenesis and vessels of abnormal structure, organization and function. Structur ally, these vessels are tortuous with varying diameters and irregular shape. The vessels are extremely porous and leaky due to endothelial cells that are abnormal in shape and loosely connected which leave gaps. Adding to the porosity, these vessels lack a well developed basement membrane with fewer or abnormal perivascular cell (smooth muscle cells, pericytes) coverage Observations of tumor vascular network as a whole reveal a l ack of vessel hierarchy in which distinct vessels such as arterioles, capillaries and venules are no longer distinguishable [2, 18 20] The abnormalities of tumor vessel structure and organization h ave deleterious effects on blood flow. Spatial and temporal heterogeneity in blood flow is characteristic to tumor vasculature. In individual vessels, blood f low can be slowed, re versed or completely stopped. The formation of arteriovenous (AV) shunts carries blood directly from arteries and into veins, thus blood can bypass systemic circulation as well as areas in the tumor itself. The chaotic spatial distribu tion of vessels within a tumor mass coupled with all of the complications previously mentioned creates numerous issues for
20 drug delivery and treatment of a tumor mass [2, 11, 18, 20] Irregular blood flow throu ghout a tumor and high interstitial fluid pressures can lead to areas of the tumor receiving little to no drug delivery. Moreover, obstacles to blood flow can result in complications for oxygen and nutrient delivery Because of this, small, concentrated re gions of hypoxia are formed which can result in resistances to radiotherapy and chemotherapy not only by decreasing the presence of reactive oxygen species, but also by selecting for more aggressive, malignant cancer cells [21, 22] While the properties of tumor vessels create difficulties for the effective treatment of tumor masses, these characteristics also distinguish tumor vessels enough from normal vasculature to make them viable target s for tumor treatmen t. Vascular Targeting Therapies For tumors to grow beyond a volume of 1 2 mm 3 the development and proliferation of tumor microvasculature is essential  With an unhindered blood supply, primary tumor s c an gr ow and potentially metastasize. The idea to target the tumor vasculature as a cancer therapy was described in the early 1980s [23, 24] Since then, two classes of vascular targeting agents (VTAs) have been developed to either directly or indirectly inhibit the development of tumor vasculature. Vascular disrupting agents (VDAs) aim to destroy existing tumor microvasculature directly cutting off a tumors oxygen and nutrient supply. Distinct from VDAs, a ngiog enic inhibiting agents (AIAs) aim to restrict the fo rmation of microvasculature, limiting the amount of vessels that develop and thus the blood flow to the existing tumor. Along with their mechanisms, these drugs are different in their physiological target the type of disease that can be affected as well as the treatment administration schedule 
21 Vascular Disrupting Agents Endothelial cells of tumor vasculature proliferate up to 20 times faster than normal vessels [23, 24] VDAs take advantage of this characteristic to selectively target tumor vasculatu re by either creating direct apoptotic effects to their endothelial cells or altering the function of the tubulin cytoskeleton of these cells to alter their shape [15, 25 29] Within minutes of administration VDAs cause a significant decrease in tumor blood flow and oxygenation up to 6 hours post injection The use of VDAs results in rapid vessel collapse which can lead to secondary cell death through the restriction of oxygen and nutrients. This vessel damage extends to the edge of the tumor bulk, and can cause necrosis of the tissue in the core of these tumors [26, 30 33] It has been shown that these agents are more active in larger bulk tumors  Because VDAs are fast acting, they are delivered acutely and necessitate a shorter period of drug exposure to be effective. There are two classes of VDAs that are defined by their mechanism of action. Flavanoid VDAs act by inducing apoptosis in endothelial cells of the vasculature while tubulin depolymerizing agen ts bind to the colchicine binding site of tubulin and disrupt t he cytoskeleton of these cells OXi4503 (combretastatin A1 phosphate) is a tubulin binding VDA that is a diphosphate p rodrug of combret a statin A1 (CA1) derived from the South African tree Comb retum Caffrum There have been only three clinical studies using OXi4503 as seen in Table 1 1. Currently this agent is in Phase I of clinical trials for the treatment of patients with acute myelogenous leukemia (AML)  Despite the minimal use thus far in clinical studies, OXi4503 ha s been studied extensively in preclinical studies and has been found to have greater antitumor and antivascular effects than a commonly used tubulin binding agent of the same family CA4P, the
22 prodrug of combretastatin A4 [28, 35] The drug CA4 P has shown to enhance the antitum or effects of current standard of care procedures such as radiation and chemotherapy, showing promise for the use of other VDAs in the clinic such as OXi4503  Table 1 1. Clinical t rials i nvolving OXi4503 Status Phase Conditions Purpose Completed Ib/II Primary or secondary hepatic tumors Safety and clinical activity of intravenous administration Completed  I Advanced solid tumors Safety, establishment of MTD, pharmacokinetic/pharmacodynamics Recruiting I AML, myelodysplastic syndromes Safety, establishment of MTD Information taken from www.clinicaltrials.gov Abbreviations: AML, acute myelogenous leukemia; MTD, maximum tolerated dose. Despite the drastic effects of OXi4503 and other VDAs on microvessels of the tumor core, complete tumor regression has not been achieved  T hese treatments c reate vascular damage u p to the edge of the tumor mass leaving a viable rim of vessels and tumor cells i n the periphery that can overcome treatment and cont inue to proliferate [30, 32, 36, 39] The survival of these vessels and tumor cells is thought to be due to the ir proximity to normal vascula ture surrounding the tumor which can provide the oxygenation and nutrients necessary for survival In addition, it has been reported that there is an increased expression of growth factors in the periphery of the tumor which would aid in the recruitment of endothelial precursor cells and other cells to promote tumor and vessel growth [40 42] The survival and aggressive regrowth of vessels into the tumor mass following VDA treatment has led to the study of these agents in combination with currently used as well as experimental therapies.
23 Angiogenic Inhibiting Agents Rampant angiogenesis seen in tumor growth is a result of the overexpression of pro angiogenic factors. Of these, the most studied is the family of va scular endothelial growth factor s (VEGF) and receptors. VEGF A, the most critical regulator of angiogenesis, is produced by tumor cells and surrounding stromal cells and stimulates the formation of vasculature The VEGF ligands bind specifically to differe nt cell membrane tyrosine kinase receptors (VEGFRs) that are expressed mainly in endothelial cells and in some circumstances tumor cells This binding leads to the activation of signaling cascades that are involved in angiogenesis [43, 44] Because of the large role VEGF and its receptors play in angiogenesis, it was thought that inhibiting this pathway would cause vascular regression and stunt the growth of the tumor thus the development of AIAs  Early preclinical studies using anti VEGF antibodies showed reduction in tumor vascular density and significant tumor growth delay in different tumor cell lines [45, 46] These AIAs that target VEGF directly compete with the ligand for VEGF binding sites on their receptors. O ther AIAs such as small molecule tyrosine kinase inhibitors (TKIs), target VEGFRs TKIs bind to ATP binding sites of the tyrosine kinase domain on VE GFRs thus blocking any intracellular signaling  Sunitinib (sunitinib malate, SU11248, SUTENT) is a TKI with both antiangiogenic and antitumor activities that is currently approved for the treatment of renal cell carcinoma (RCC) and imatinib resistant ga strointestinal stromal tumor (GIST)  This TKI inhibits a wide range of receptors including VEGFR 1, VEGFR 2, fetal liver tyrosine kinase receptor 3 (FLT3), KIT (stem cell factor [SCF] receptor ) platelet derived growth [49, 50] The antitumor and antivascular effects of Suniti nib have been demonstrated in various preclinical studies, indicating
24 decreases in vascular density and tumor growth inhibition with increases in oxygenation and tumor blood flow in the short term [51 54] Whi le preclinical studies continue, Sunitinib is currently being used in over 100 active clinical studies in the United States for a diversity of cancers including glioblastoma, prostate, ovarian and breast (www.clinicaltrials.gov) This agent along with othe r AIAs have shown to be inadequate as singular therapies due to resistances developed in tumors that allow them to continue to grow as well as their induction of tumor invasiveness  Consequently, like VDAs current clinical studies have combined AIAs with conventional therapies in hopes that the AIAs will enhance the effects of chemotherapy and radiotherapy in tumors [56, 57] Combination of Vascular Targeting A gents Both VDAs and AIAs have disadvantages that make them deficient as singular therapies. However, the mechanisms by which they act are complementary, thus the use of these agents in combination are hypothesized to result in enhanced tumor growth delay a nd vascular regression. The notion behind combining VTAs entails the destruction of existing vascul a ture of the bulk tumor by VDAs while the AIAs would inhibit vessel formation and development from the remaining viable vessel rim thereby halting the conti nued growth and development of the tumor. Previous studies combining VTAs has shown dramatic growth delays of tumors treated with both agents in comparison to either treatment alone When OXi4503 was combined with the AIA B evacizumab (Avastin) tumor growt h delay of RCC xenografts was extended from 8 days with OXi4503 and 18 days with Bevacizumab to 27 days with both of the drugs combined [58, 59] Current and completed clinical trials are in the phase I or II s tage s that primarily focus on toxicities and identifying biomarkers for the combination therapy,
25 however these studies have shown promising results with regard to treatment impact [56, 60, 61] A summary of pre clinical and clinical studies is shown in Table 1 2. Table 1 2 Combination of VDAs and AIAs VDA Compound AIA Compound Cancer Type Type of Study Ref. ZD6126 (AstraZeneca) ZD6474 (AstraZeneca) Colorectal, ovarian Preclinical  CA4P or OXi4503 (OXiGENE, Inc.) Bevacizumab (Genentech) Clear cell RCC Preclinical  Zd (AstraZeneca, UK) Tha (Pharmaceutical Factory, China) Liver Preclinical  CA4P Sunitinib (Pfizer) Prostate Preclinical  OXi4503 Bevacizumab AML Preclinical  CA4P Bevacizumab Advanced solid tum ors Clinical (Phase I, completed)  CA4P Bevacizumab Advanced solid tumors Clinical (Phase I, completed)  CA4P Bevacizumab NSCLC Clinical (Phase II, comple ted) CA4P Bevacizumab Ovarian epithelial, fallopian tube, peritoneal cavity Clinical (Phase II recruiting) * Data gathered from www.clinicaltrials.gov Abbreviations: RCC, renal cell carcinoma; AML, acut e myelogenous leukemia; NSCLC, non small cell lung cancer. Timing of therapies, biomarkers, and the combination of agents with conventional therapies are important factors to consider in future investigation of these agents combined. In Vivo Imaging of Tum or Microvasculature and Microenvironment Intravital imaging has been invaluable to the in vivo study of the physiological processes involved in cancer. Wh ile analyses such as histology, vascular casting and molecular methods provide quantitative data at th e tissue, cellular and molecular levels, these methods are usually performed as end stage analyses and are highly invasive  I maging allows observations to be performed in real time, in a non invasive or mi nimally invasive manner. Moreover, functional data can also be determined using
26 imaging that cannot be concluded using histological analysis. The combination of imaging evaluation with histological analysis can provide a comprehensive analysis of pathologi cal processes on both the cellular level and the functional level  There are many imaging modalities currently u sed for the evaluation of tumor microvasculature in the preclinical setting Each individual method has its own advantages and disadvantages that suit them for imaging of specific types of tissues or processes P ositron emission tomography ( PET )  computed tomography ( CT )  and magnetic resonance imaging ( MRI ) [69, 70] techniques are a few of the available methods that can be used in both the preclinical or clinical setting X ray CT conventionally has a limited resolution but the advent of micro CT has provided an extremely useful method for imaging soft tissue, skeletal abnormalities and tumors PET gives information about molecular processes such as glucose metabolism, blood flow and receptor ligand bin ding rates MRI comes in various forms and can provide information surrounding tumor vascular structure, density and blood flow. With the aforementioned methods of analysis the primary advantage is with the evaluation of deep tissue. Unfortunately high co st and low resolution can be an issue. O ptical imaging methods such as transillumination and fluorescence microscopy provides lower cost means of analysis while providing higher spatial and temporal resolution [ 3, 71, 72] One animal model that is commonly used with optical imaging techniques is the murine dorsal skinfold window chamber model. The dorsal skin of the mouse is stretched and the upper layer of skin is removed for the implantation of a titanium win dow chamber to reveal the vasculature of the underlying layer of skin. Tumor cells can be injected into this exposed layer o f skin to initiate tumor growth for analysis 
27 The window chamber model allows for the direct visualization and serial observation of a growing tumor over time. This ability to analyze a tumor over time is essential to evaluate the tumor response to VTAs Regardless of the benefits to this model i t is important to note that there are lim itations to consider T he size of the tumor is limited by the size of the window chamber and tumors are generally s ubcutaneous and not orthotopic placing restric tions on the use of this model. In addition, faster growing tumors are usually necessary in th e window chambers as the viability of the windows are restricted to up to 3 weeks  There is a wide variety of imaging modalities that can be used with window chamber models. Some examples such as fluoresc ence recovery after photobleaching (FRAP), laser scanning  Doppler optical coherence tomography (OCT)  and multiphoton microscopy  are methods which can entail the use of complicated camera and laser systems. More simply, t ransillumin ation microscopy of tumors allows for the direct observation of developing vasculature over time via brightfield images Osus ky et al evaluated the effects of Sunitinib in the Lewis Lung Carcinoma (LLC) carcinoma model in window chambers of C57BL6J mice. This group injected LLC cells and took images of the tumor area daily for 5 days to show the vessel inhibition effects of Sun itinib in comparison to controls  Tozer et al imaged the response of P22 rat carcinosarcomas to the VDA CA4P in dorsal skin chambers of male BD9 rats from 7 to 14 days following surgery. This analysis rev ealed the time course by which the VDA acted and allowed scientists to characterize the vascular damage caused by the drug 
28 T he absorption of native species in the tumor microenvironment can be taken adva ntage of to provide more data about the functionality of the vasculature. Hyperspectral imaging of Hb saturation in microvessels can be determined by utilizing the varying absorption spectra of oxy and de oxyhemoglobin  The oxygenation of vessels can help determine the effectiveness of oxygen transport in specific vessels of the tumor micr ovasculature [8, 75] Wankhede et al utilized hyperspectral imaging of Hb saturation of window chambers to describe the oxygenation changes that occur in 4T1 mammary adenocarcinoma and Caki 1 RCC in response t o treatment with the VDA OXi4503  This study revealed the dynamic changes in oxygenation that can occur within hours after administration of the drug, while also confirming the time course by which these agents act  In addition to transillumination imaging, f luorescence microscopy of the window chamber can be used to easily characteriz e t he tumor microvasculature Specifically, t he injection of fluorescent tracers in fir st pass fluorescence (FPF) imaging has been utilized to characterize the blood flow throughout tumors and assess the permeability of vessels [16, 77 80] Brurberg et al demonstrated the temporal heterogeneity that can occur in blood flow of tumor xenografts by using FPF imaging in tumors at 20 minute intervals. This study showed the blood flow changes that can rapidly occur in single vessels of tumors and postulated that the causes to these changes could be a r esult of arterial/arteriolar vasomotor ac tivity, vessel wall compression and properties of flow rate  While these aforementioned methods do not give 3 dimensional information, and are restricted in the area of vasculature that can be analyzed, the ea se of use, low cost and high spatial resolution make them popular choices for preclinical evaluation.
29 CHAPTER 3 DEVELOPMENT OF FIRST PASS FLUORESCENCE IMAGING AND ITS COMBINATION WITH HYPERSPECTRAL IMAGING OF HEMOGLOBIN SATURATION Introduction The assess ment of tumor microvasculature in small animal models is very important to the development of treatments that target tumor vessels. Neoplastic cells need to be within 100 to 200 m of blood vessels for adequate oxygen diffusion  For tumors to progress and grow, an angiogenic switch occurs in which new vasculature is formed  Thus angiogenesis and existing tumor microvasculature are vital to the maturation and growth of a tumor and fue l the need for methods of analysis that can readily characterize oxygenation and blood flow. Often times, histological samples of tissues are used to examine the vasculature after sacrificing the animals, however this method does not allow fo r flow and oxygenation study  For in vivo study, ultrasound Dopper optic al coherence tomography (OCT) and positron emission tomography (PET) have all been utilized to analyze the structure of vessels and vascular perfusion, but can be time consuming and technically extensive without providing important oxygenation information  Intravital microscopy of the murine dorsal window chamber model has been used e x tensively for the analysis of tumor vasculature  The use of this model has many advantages including the ability to analyze vasculature non invasively, over long periods of time (up to three weeks) and wit h microvessel resolution  First pass fluorescence (FPF) imaging is a simple, repeatable method of analyzing the morphology of blood vessels as well as the directionality of blood flow in th e window chamber tumor model. This analysis is performed by an injection of fluorescent tracer while simultaneously imaging the tracer traveling throughout the vasculature as it first enters the field of view
30 The blood supply time (BST) [16, 77] the time it takes for blood to travel from a supplying artery to a chosen vessel in the tumor, can be determined for all vessels within the tumor. Alone this data provides structural and organizational information for the tumor, however like previous modalities mentioned it does not provide any further information regarding oxygenation To study additional data about microvascular blood flow and oxygenation FPF imaging is combined with hyperspectral imaging of hemoglob in (Hb) saturation to provide corresponding data of oxygen transport within the vessels of the tumor. In this chapter we discuss the development of FPF imaging and its combination with Hb saturation imaging for a novel combination of imaging modalities tha t is a repeatable method of analyzing the morphology directionality of blood flow and oxygenation of tumor microvessels in a window chamber model. Murine Dors al Skinfold Window Chamber Model The murine dorsal skinfold win dow chamber model has been used to analyze different aspects of cancer physiology since 1939  The window chamber has the major benefit of allowing studies to be perfor med serially over a period of days (usually up to 2 3 weeks) With the development and improvement of imaging techniques, this model has been successfully used for study of tumor growth, cell migration, angiogenesis, vascular function and vessel permeabili ty [8, 73, 74, 82, 83] A typical window chamber installation for our experiments is seen in Figure 3 1. The surgical implantation of the window chamber occurs as follows. Mice are first ane sthetized using an injection of xylazine (10 mg/kg) and ketamine (100 mg/kg) intraperitoneally then placed on a temperature controlled heating pad covered by a sterile surgical drape. The dorsal skin of the mice is extended and sutured to a c clamp to hold the skin taut. The front portion of the window chamber is placed on the skinfold
31 and the window area is marked for removal. A 16 gauge needle is used to puncture 3 holes for the window chamber bolts. A 12 mm in diameter circular area of the upper layer of dorsal skin and an y connective tissue beneath the skin is removed. The window chamber is then assembled onto the skin flap and the mouse is removed from the c clamp. At this point, if tumor initiation is intended, a single cell suspension of tumor cells in phosphate buffere d saline ( PBS ) is injected subcutaneously. Following tumor cell injection, a round glass coverslip is placed on the window and held in place with a retaining ring  Once animals have recovered there are hous ing considerations to be made to extend the viability of the window chambers. Humidity can be an issue because the contact between the window chamber and exposed skin is not complete therefore leaving the area susceptible to losing moisture and drying out In addition, the temperature at which animals are kept can also affect the viability of the chamber as well as tumor growth. When the skin is pulled away from the body, heat is lost to the ambient environment which leads to a lower s kin temperature and f urthermore, a potential decrease in tumor growth rate  Animals are thus housed in an environmental chamber at 33C and 50% humidity with free access to food and water with a st andard 12 hour light/dark cycl e to maintain ph ysiological conditions for tumors in the window chamber Hyperspectral Imaging of Hemoglobin Saturation Hyperspectral imaging of Hb saturation is a commonly used method to characterize the transport of oxygen through microvessels in the wi ndow chamber m odel [8, 75, 84] Briefly, a 100 W tungsten halogen lamp was used for transillu mination of the window chamber. For image acquisition, a monochromatic scientific grade charge coupled device ( CCD ) c amera (DVC Company, Austin, TX) therm oelectrically cooled to
32 20C was used. A liquid crystal tunable filter (LCTF) (CRI, Cambridge, MA) with a 400 to 720 nm transmission range was placed in front of the camera to acquire spectral image data sets from 500 nm to 575 nm at 5 nm intervals. With a 2.5x objective the resulting field of view had dimensions of 4.15 mm in width by 3.1 mm in height. Hb saturation pseudocolor maps were created in MATLAB (The Mathworks Inc., Natick MA) using pure oxy and deoxyhemogl obin reference spectra and a previously determined linear least squares regression model  This process is overviewed in Figure 3 2. Development of First Pass Fluorescence Imaging Imaging System A Zeiss Ax ioImager microscope (Carl Zeiss In c. Thornwood, NY) was used as the imaging platform for both hyperspectral imaging and FPF imaging (Figure 3 3) All images for FPF imaging were taken using an ANDO R iXon electron multiplying CCD (EMCCD) camera thermoelectrically cooled to 50C (ANDOR Technology, South Windsor, CT). A 2.5x fluar (Carl Zeiss Inc., Thornwood, NY) long working distance objective was used resulting in a 3.1 mm x 3.1 mm field of view. Ou r imaging system was equipped with Cy5 (ex 64020 nm / em 68030 nm) and Cy7 (ex 74035 nm / em 79550 nm ) filters (Chroma Technology Corp., Rockingham, VT) allowing for analysis using near infrared (IR) fluorescent dyes. Development of Fluorescent Trace r There are many considerations to be made when choosing a fluorescent tracer for FPF imaging. Combining FPF imaging with Hb saturation imaging limits the wavelength of fluorescent dye that can be used. Because Hb saturation imaging utilizes the absorption range of 500 to 575 nm, near IR dyes must be used to avoid affecting
33 the Hb saturation analysis. In addition, tumor microvasculature inherently carries obstacles when using fluorescent tracers to analyze blood flow. Because tumor microvessels can be extre mely porous, particles can be readily leaked into the tumor stroma. Any residual fluorescence signals created by extravasated dye can negatively impact a ny future FPF imaging analysis. Commercially available dye s and particles were initially chosen as fluo rescent tracers for their ease of use. Texas Red labeled dextrans (70 kDa MW; ex 595 nm / em 615 nm) were simple to use but did not overlap with our filter sets enough to produce a strong fluorescent signal. Sky blue SPHERO TM fluorescent particles had an excitation and emission wavelength well outside of the wavelength range of Hb absorption, however the particles readily aggregated and did not have a high enough signal intensity to image smaller vessels as seen in Figure 3 4 P revious experiments done by Oye et al utilized tetramethylrhodamine 5 (and 6) isothiocyanate ( TRITC ) labeled dextrans for FPF imaging  While the absorption spectra of TRITC overlaps with the absorption spectra of oxy and deoxy Hb, we surmised that performing Hb saturation imaging before FPF imaging would negate the spectral overlap as long as the dye was cleared before the next imaging session. The use of TRITC dye required a Cy3 filter set (ex 54525 nm / em 60570 n m (Chroma Technology Corp., Rockingham, VT) This dye had very high signal intensity that illuminated vessels outside of the depth of focus, creating a high background signal for the vessels of interest. The distribution of particles was not uniform and th e dye aggregated, thus there were intermittent bright spots that flowed through the vasculature and became trapped in smaller vessels These areas of high signal intensity
34 would interfere with proper BST analysis and thus this dye was shown to not work for our application. Table 3 summarizes our early attempts at identifying a feasible fluorescent tracer for our experiments. Table 3 1. Initial fluorescent tracer attempts Tracer Particle Size Ex/Em Wavelength Results Texas Red Dextran ~1 m (70 kDa) 595 nm /615 nm Low fluorescent signal SPHERO TM Sky Blue Nanoparticles ~450 nm 600 660 nm/700 nm Particle aggregation, low fluorescent signal TRITC Dextran ~2 um (155 kDa) 544 nm/572 nm Particle aggregation Because of the issues arising with the use of c ommercially available dyes, and the knowledge that our experiments would eventually require FPF imaging over multiple days, we decided to create our own fluorescent liposomes as tracers Fluorescent liposomes are used as carriers and tracers for many diffe rent applications. The composition of liposo mes can be easily tuned to satisfy a variety of specifications making them a suitable candidate as FPF fluorescent tracer s There are many commercially available fluorescent lipid dyes in the near IR range. For our uses, t he fluorescent carbocyanine dyes 1,1' dioctadecyl 3,3,3',3' t etramethylindodicarbocyanine 4 chlorobenzenesulfonate (DiD; ex 644 nm em 665 nm) and 1,1' dioctadecyl 3,3,3',3' tetramethylindotricarbocyanine iodide (DiR; ex 750 nm em 780 nm) (Invitr ogen, Carlsbad, CA) were chosen to work with the Cy5 and Cy7 filter sets in our microscope setup. A comparison of BST fluorescence with T RITC labeled dextrans and DiD labeled liposomes at 5x magnification is shown in Figure 3 5 Figure 3 5 A shows a vascular network filled with TRITC labeled dextrans. The aggregation of dye and high background signal is evident, with bright spots visible in the large main vessel in the
35 field of view. Figure 3 5 B shows the same vascular network filled with DiD labeled liposomes. The fluorescence of liposomes is uniform in all vessels, with no aggregation in smaller vessels and little background signal from ves sels not in the depth of field. We adjusted the dye content in the liposomes to increase the signal intensity and decided on the use of DiD and DiR for our liposomes. Aside from the fluorescent component, t here were several other considerations to be made in the development of our liposomes. Size, composit ion and the use of polyethylene glycol (PEG) were all tested. Unlike many liposomal formulations for use in tumor treatment, we needed our liposomes to avoid extravasation and be quickly taken up by the r eticulo endothelial system (RES) Thus, w e started with a standard liposome composition of 1,2 dioleoyl sn glycero 3 phosphocholine ( DOPC ) as the main component of the liposome, 1,2 distearoyl sn glycero 3 phosphoethanolamine N [methoxy(polyethyleneglycol) 2000] ( DSPC mPEG2000 ) to keep the liposomes from aggregating and cholesterol to provide stability. While PEG is conventionally used to prolong circulation, w e found that without PEG, the liposomes quickly accumulated and created bright spots that became s tuck in small vessels of the tumors. Therefore, a small amount of PEG was added to the formulation to ensure the liposomes did not aggregate. This amount was not enough to prolong the circulation of liposomes for an extended period of time. We initially ch ose 400 nm size liposomes which would be large enough to escape extravasation via tumor microvessel pores. Again, t he se particles suffered a high occurrence of aggregation following injection and often became stuck in smaller vessels of the tumor. Because of this, o ur next iteration of liposomes w as created at 100
36 nm in diameter. There was no occurrence of aggregation or liposomes being entrapp ed in smaller vessels during image acquisition Despite this immediate success, w ith such small liposomes, e xtravas ation of liposomes out of the vessels and into the tumor space was a concern. The accumulation of liposomes is illustrated in images taken of tumor vasculature 24 hours following liposome injection in Figure 3 6 To remedy the issue of liposome extravasa tion, we attempted to reduce the circulation time by adding 1,2 dioleoyl sn glycero 3 phospho L serine ( DOPS ) a negatively charged lipid that would make the particles more readily taken up by the RES  Figure 3 7 shows the improved clearance of liposomes with DOPS 4 hours following injection with the liposomal formula in normal tissue (Figure 3 7 B) and liposome accumulation 24 hours after injection in tumor tissue (Figure 3 7 D) The final liposomal formulation was a c ombination of DOPC, DOPS, PEG and cholesterol in a ratio of 47.4: 19.7: 1: 39.7. The final procedure for the preparation of our fluorescent liposomes using the lipid hydration method  is as follows: All lipids and extruder were purchased from Avanti Polar Lipids, Inc. (Alalbaster, AL). Liposomes were composed of DOPC DOPS DSPC mPEG2000 and cholesterol combined at a molar ratio of 47.4:19.7:1:39.7. Liposomes were fluorescently labeled with either DiD or DiR dye in ethanol (2 mM) added to between 0.5 and 1 mol.% of th e total phospholipid content. This mixture was dried into a thin film with nitrogen gas and held und er vacuum in the dark overnight to remove traces of the solvent. The film was rehydrated in sterile PBS (pH = 7.4) by vortexing at room temperature for 15 minutes to yield a suspension of multilamellar
37 vesicles at a lipid concentration of 10 mg/m L Unilame llar liposomes ~100 nm in diameter were obtained by repeated extrusion of vesicles through a Nuclepore membrane with a 100 nm pore size. FPF Imaging Acquisition For FPF imaging, mice were anesthetized using 1.5% isoflurane in air and placed on a heating pa d on the microscope stage. A catheter was constructed using two 30 gauge needles and polyethylene tubing (PE 10). A syringe of sterile saline was first used to establish that the needle was indeed in the tail vein. The syringe of saline was switched with a syringe of fluorescent liposomes for FPF imaging A single brightfield and fluorescence image was taken at an exposure time of 100 ms before injection to establish location and background signal for the field of view. The following image stack was also ta ke n with an exposure time of 100 ms. With added refresh rates due to camera and computer processing the total exposure time was 117 ms. An injection of 50 L of liposomes w as administered and at least 700 frames were simultaneously recorded as they flowed through the field of view. At a 117 ms exposure time, a 700 frame video acquisition corresponded to 81.9 seconds of recording. Following injection, a single fluorescence image of the vasculature filled with fluorescent liposomes and another brightfield im age were taken to aid in the creation of vascular masks. FPF Image Processing To develop our own image processing method we tested our initial MATLAB code attempts using image stacks of fluorescent dye flowing through a capillary tube. Figure 3 8 illustr ates the various steps in BST analysis. A bolus of dye was injected into a capillary tube and the flow was recorded. A composite of all of the image stacks is shown in Figure 3 8A where the direction of flow is indicated. Figure 3 8B shows the
38 resulting BS T map. Figures 3 8C shows the signal intensity over time for each ROI shown in Figure 3 8A. Lastly Figure 3 8D shows a histogram of the frequency of BST values in pixels within the capillary tube. The timing of our BST could be verified by visual analysis of the dye bolus through the capillary tube and knowing the exposure time of the video MeVisLab (MeVis Medical Solutions AG, Bremen, Germany) and MATLAB software were used to process fluorescent stack images to create BST maps. First, vascular masks of th e vessel areas were created to ensure that only vessels were being analyzed. MeVisLab was used to help isolate vessel structures from brightfield and fluorescent images of the field of view to create vascular masks using built in thresholding, fuzzy c mean s algorithms and Hessian vesselness filters. MATLAB was then used to combine these isolated image s to create a final mask. Once the vessel area was isolated in the vascular masks MATLAB coding was used to create BST maps. Image stacks were truncated to the frame at which fluorescent liposomes first entered the field of view up until the frame at which all vessels were filled. Ten frames from before the liposomes appeared in the field of view were averaged as a background signal to flatfield all images in the truncated stack. Once liposomes have been injected into the tumor space, there are various structures and vessels that are out of the depth of field tha t can scatter the fluorescence and can create increases in the signal intensity of vessels not due to f lowing liposomes in the blood. In order to alleviate the effects of this scatter, each frame in the stack is subjected to a moving filter that travel s pixel by pixel through the vascular mask and subtract s the median value of t he surrounding backgroun d from each individual pixel.
39 By subtracting the background we reduce d the amount of signal attributed to cross talk between adjacent vessel s and other structures and isolate d the fluorescence intensity solely due to the fluorescent liposomes. After being filtered each individual pixel signal was smoothed over time using the smooth function in MATLAB Once smoothed, the time at which the pixel reache d its average intensity wa were placed back into th e vascular mask to then create a BST map. An overview of the entire FPF imaging process is illustrated in Figure 3 9 In Vivo Testing To optimize the combination of Hb saturation imaging and FPF imaging we tested our fluorescent liposomes and image post processing on both normal and tumor vasculature. Tumor Cells 4T1 mouse mammary adenocarcinoma cells (a gift from Mark W. Dewhirst, Duke University Medical Center, Durham, NC) were utilized to demonstrate the utility of Hb saturation imaging and FPF imagin g for the analysis of tumor microvasculature. T umor cells were cultured in vitro in media containing 10% fetal bovine serum (Bio Whittaker, Inc) and 1% penicillin streptomycin (Hyclone) in DMEM (Cellgro,Herndon, VA). Cells were incubated at 37C in a 5% CO 2 atmosphere. A 10 L injection of 7.5x10 3 cells prepared in PBS w as used to initiate tumors for each cell line as determined in previous experiments  Pr o ceedings of SPIE 7902, (2011).
40 Animal Model All in vivo procedures were conducted under a protocol approved by University of Florida Institutional Animal Care and Use Committee. A titanium dorsal skinfold window chamber was surgically implanted in female athymic (nu/nu) nude mice weighing more than 21 g (Harlan Laboratories, Indianapolis, IN) The proc edure for installation of window chambers was described in this chapter previously. Once the upper layer of dorsal skin was removed a single cell suspension of 7.5x10 3 tumor cells in PBS was injected subcutaneously to generate a tumor sized between 2 3 mm in 5 7 days. Following tumor cell injection, a round glass coverslip was placed over the exposed window. For mice not receiving tumor cells, the coverslip was added following removal of the upper layer of dorsal skin. Animals were then housed in an environ mental chamber at 33C and 50% humidity with free access to food and water with a standard 12 hour light/dark cycle. Experimental Design Mice were anesthetized using 1.5% isoflu rane in air and placed on the heated microscope stage. Hb saturation imaging was perform ed first followed by FPF imaging The entire imaging process took up to 30 minutes per mouse. R esults Six frames from first pa ss imaging of liposomes through tumor vasculature in the window chamber are shown in Fig ure 3 10. Figure 3 10 A is ~1.9 3 s after the dye first enters the field of view in the supplying arteries for the tumor. The next frame shows the tumor microvasculature beginning to be filled at a time of ~5.78 s Fig ure 3 10 C at ~7.7 s and 3 10 D at ~11.55 s illustrate the blood from th e tumor beginning to flow to the main
41 veins of the tumor. Lastly, f rames shown in Figure 3 10 E and 3 10 F show the final filling of the main veins with blood traveling from normal tissue outside of the tumor. BST and Hb saturation maps provide a visual rep resentation of the blood flow and oxygenation for microvessels in the entire field of view. Image processing for both Hb saturation a nd BST were performed following image acquisition. Figure 3 11 illustrates the resulting Hb saturation image and BST map fo r normal and tumor vasculature. Figure 3 11 B and 3 11 C show the Hb saturation map and corresponding BST map for normal vasculature. The supplying arter ies of the field of view are straight, narrow vessel s with high oxygenation and fast BSTs. The draining v ein has varying lower levels of oxygen and slower BSTs within the same vessel due to the blood from the systemic flow reaching this vessel at different times from different areas. The BSTs and oxygenations in the intermediate vessels vary, however the oxyg enation in these vessels are generally lower than 50%. T he Hb saturation map of tumor microvasculature (Figure 3 11 D ) shows t he supplying arteries and tumor interior have the highest oxygenation, with the draining veins and tumor periphery showing a lower oxygenation The vessels in the interior of the tumor are very thin and immature compared to normal vessels. The oxygenation is reflected by the BST map (Figure 3 11E) which re veal s trends in blood flow patterns that are influenced by vascular connections and morphology. Vessels in the tumor interior directly form from supplying arteries and drain into the draining veins, acting as arteriovenous (AV) shunt s Region of interest (ROI) analysis was performed on normal vasculature to evaluate the correlation o f BST and Hb saturation s een in Figures 3 12A 3 12D Overall, for a longer BST a lower oxygenation can be expected while for a faster BST a
42 higher oxygenation can be expected S upplying arteries fill first with the fastest BSTs (< 1 s) and highest oxyge nations (> 78%). The draining vein is filled more slowly resulting in longer BSTs (> 12 s) and a lower oxygenation (< 47%) As previously stated, t his vein is filled with blood from different areas of the systemic flow a nd thus the oxygenation is hypothesiz ed to be higher due to mixing of the blood. Within the draining vein there are distinct flow paths of blood with differing BSTs and oxygenations. ROIs 21 and 22 represent two sides of the draining vein that receive blood from different areas. With regard t o BST ROI 21 ( 12.64 .26 s) and ROI 22 ( 12.18 .25 s) are relatively similar however t he disparity between oxygenation is much greater with ROI 21 (23.8 4.1 %) versus ROI 22 (46.8 2.9 %) This is due to the blood being supplied to ROI 21 from an exte rnal vessel (ROI 7; Hb sat. 17.6%) that is carrying deoxygenated blood. The intermediate vessels fill the range between the supplying artery and draining vein mainly consisting of lower oxygenations and slower BSTs in comparison to the supplying arterie s. These vessels do not experience a great degree of mixing of blood and carry vessels of homogenous oxygenation, in many cases lower than that of the draining vein. ROI 13 shows one extreme of low oxygenation and slow BST that can occur in vessels natural ly. The same ROI analysis was repeated for different areas within and around a 4T1 tumor (Figure 3 13 A 3 13 D ). Like normal vasculature, s upplying arteries have a very tight BST to Hb saturation correlation with the fastest BSTs (< 2 s) corresponding to the highest oxygenations (> 84%) In contrast to normal vasculature, the newly formed vessels in the tumor interior are shifted towards faster BSTs and higher oxygenations. Because these vessels overtake the normal vasculature, a larger proportion of vesse ls
43 in the field of view behave more like the supplying arteries in oxygenation and BST. The smaller draining veins in this field of view are dissimilar from the main draining vein seen in normal vasculature and are not receiving as many different blood flo w paths Because of this the oxygenation is not as high and BSTs in these vessels are slow. Figure 3 1 4 illustrates magnified BST and Hb saturations maps of tumor microvasculature for a selected region within the window of Figure 3 1 3 This vasculature in cludes a draining vein into which blood enters from the tumor as well as from two vessels of the normal vasculature. The direction of the blood flow in vessel 1 and vessel 2 cannot be determined using only the Hb saturation map. One would guess that the bl ood would flow from vessel 1 through vessels 2 and 3 and join into vessel 5. However, using BST imaging it is evident that vessel 2 is being supplied by vessel 4 From vessel 4 the blood flow splits and travels both up towards vessel 1 and down towards ves sel 5. Vessel 1 travels into the field of view from outside of the frame, and splits between vessels 2 and 3, encountering obstacles to blood flow in both branches. As previously mentioned, vessel 2 has an opposing flow to slow down blood flow from vessel 1. Merging into vessel 3 are several vessels from the tumor that carry blood to mix with the blood provided from vessel 1. ROI 2 represents an area where the blood from vessel 1 mixes with the incoming blood from the tumor vasculature. This mixing can expl ain the slower BST (39.66 4.39 s) relative to the high er oxygenation value ( 35.2 6.6%) ROI 3 and 5 represent two other blood flows entering the draining vein from the tumor mass. The vessel supplying ROI 5 occurs at a faster time (6.24 .4 s) and hig her oxygenation (46.5 5.7%) with respect to ROI 3 (BST 12.64 .31s; Hb saturation 32.7 4.4%) even though this region is downstream from ROI 3 in the
44 same vessel. The type of mixing seen in ROI 2 can also be seen in ROI 4, where again, even though the region is downstream from ROI 3 and the blood supply time is longer (16.09 .71 s) the oxygenation is similar (34.4 5.9%) due to mixing from the close proximity to another feeder vessel like that feeding into ROI 5. Discussion Tumor microvasculatu re and angiogenesis is necessary for the growth and propagation of a tumor mass  Because of this, research in tumor therapies targeting the microvasculature and angiogenesis of a tumor have become more p opular as scientists search for new meth ods of treating tumor masses [11, 88] With more emphasis being placed on targeting the microvasculature, the importance for effective and repeatable methods of character izing blood flow, oxygenation and vessel structure has consequently increased. We have developed t he combination of BST measurements and Hb saturation data to create a novel method for analyzing the blood flow, structure and oxygenation transport of microv asculature. BST measurements enable observations regarding direction of blood flow, vessel morphology and the timing of blood flow throughout a tumor mass in comparison to normal vasculature. Hb saturation measurements provide quantitative data re garding o xygen transport in blood vessels. The inherent advantage to this multi modal form of imaging is the ability to easily characterize both oxygenation and blo od flow in the same apparatus In addition, both measurements are able to be performed at microvessel spatial resolution which is another advantage over analytical methods such as MRI and PET Hb saturation maps make up for shortcomings in BST measurements by showing the oxygenation within each vessel. The saturation values from our findings show that in
45 normal tissue (Figure 3 12) ther e is a relative correlation in which a longer BST corresponds to a lower saturation value and a higher BST corresponds to a greater saturation value This observation is what one would expect as the blood in arteries is mor e oxygenated than that of in veins. Skala et al. developed a similar combination imaging method using Hb saturation imaging and Doppler OCT to measure blood flow velocity. This study showed a positive correlation between blood flow velocity and oxygenation however found that the correlation was not strong enough to postulate that one parameter could explain the behavior of the other  In our experiments, we demonstrated that the blood transit time can be a nother parameter to help explain the oxygenation of blood in vessels relative to other vessels in the field of view. This correlation is more distinct in individual vascular networks where all observed vessels stem from one common feeder vessel, however th e relationship can va ry in a single field of view where the origin of blood flow in vessels coming from outside of the field of view is unknown The timing and direction of blood flow was elucidated using BST imaging. By analyzing the magnified maps in Fig ure 3 13 we can see disparity in the time that blood from the tumor drains into the main draining vein compared to blood from normal vasculature. F ast BST measurements from the tumor mas s to the draining vein indicate shunting of oxygenated blood into the vein, commonly seen in tumor blood flow  It was previously estimated that up to 30% of blood flow in malignant tumors can be attributed to AV shunts  In accordance with these observations, we have shown that for a tumor in early development, these AV shunts make up the majority of the
46 v asculat ure in the tumor core. Consequently the tumor is dependent on t he oxygen supply from the blood that these shunts provide An other interesting characteristic of blood flow revealed by this imaging technique in both normal and tumor vasculature is distinct parallel flow paths within vessels where blood flowing into th e vessel of interest from different sources continue on the path with little intermixing of the flow. For our study, we observed the shunting and laminar flow behavior in the main draining vein of normal vasculature in Figure 3 11 Here we observed different parallel flows of varying BSTs and oxygenation values across the diameter of the vessel which we attribute to different vessels merging blood from different systemic pathways into this struc ture. This behavior can be attributed to the large diameter of this draining vein and the fact that many vessels probably empty into this vein from different points in the systemic flow. In the main draining vein of a tumor (Figure 3 14) we s ee similar fl ow path behavior, however the behavior in this situation can be explained by the formation of AV shunts. Much more exaggerated than the flow paths in the normal vasculature, blood from the tumor and blood from the systemic flow meet in the same draining ve in with much different BST and oxygenation values. In this instance there are two distinct vessels feeding into the same vein where the blood flow paths divide the vessel in two. Each flow path stay s on the same side of the vessel that they arrived in. Thi s type of flow behavior is also reflected in the oxygenation of the vessel Previous studies utilizing hyperspectral imaging of Hb saturation alone demonstrated the presence of laminar flow in vessels of larger more developed 4T1 mouse mammary carcinoma a nd 4TO7 tumors, a nonmetastatic subclone of the 4T1 tumors. While no direct blood flow information was gathered in this
47 study, the appearanc e of distinct oxygenation values on either side of tumor vessels was observed. Both this study and our results show that the formation of AV shunts create s a more complicated vascular hierarchy and blood flow in all stages of tumor growth. W ithin and around tumor masses there are vessels which show a discrepancy in BST and Hb saturation (i.e. a higher BST value and high oxygenation), which indicate other scenarios such as mixing of blood at junctions where there are opposing flows or where two flow paths join. An instance in which a fast BST results in a low oxygenation value can be explained by the fact that DiD and DiR liposomes represent plasma flow, and blood cells may not be able to travel as readily into certain vessels as the plasma can. Blood flow can be further complicated in tumor s as the flow through a vessel has been observed to stop and even change directions  With this in vivo analysis we have demonstrated the novel combination of first pass fluorescence imaging and hyperspectral imaging of Hb saturation for the wide field analysis of blood flow and oxygen transport in the microvasculature of normal and tumor tissue. The ability to easily and simultaneously characterize the blood flow and oxygenation of t umor microvessels is a benefit to this modality in comparison to other meth ods of vessel analysis. Moreover, our combination imaging modality allows for the observation of tumor microvasculature at a microvessel spatial resolution, an other advantage over c onventional imaging modalities. While only one injection was performed for each mouse, the data obtained from this study will allow for future studies in which BST imaging and Hb saturation imaging can be used to analyze tumor growth over time and tumor re sponse to various vascular targeting treatments.
48 Figure 3 1. Murin e dorsal window chamber model. A) Female, athymic nude mouse (nu/nu) with dorsal skinfold window chamber surgically installed. B) Close up of a dorsal skinfold window chamber with formin g tumor nodule outlined with a black dotted line. The diameter of the window is 12 mm. Figure 3 2. Overview of hyperspectral imaging of hemoglobin saturation. Sixteen brightfield images are acquired at 5 nm intervals from 500 nm to 575 nm. Using a prev iously determined linear least squares regression model, a pseudocolor map of hemoglobin saturation is created in MATLAB
49 Figure 3 3 Microscope setup for window chamber imaging. A) Zeiss AxioImager microscope platform with DVC CCD camer a and ANDOR iXo n EMCCD camera. B) Mouse positioned on the imaging stage on a heating pad under gas anesthesia. Figure 3 4 Fluorescence images of SPHERO TM sky blue nanoparticle flow through normal vasculature illustrate the low signal intensity and aggregation of part icles A) Beginning of nanoparticle flow. B) Flow of nanoparticles ~4 seconds after Fig 3A.
50 Figure 3 5. Comparison of (A) TRITC labeled dextrans and (B) DiD labeled liposomes for BST imaging. Images were taken at 5x magnification. A) TRITC labeled dext ran filled vascular network. Bright spots of dye and aggregation are evident in the main large vessel in the field of view. B) DiD labeled liposome filled vascular network. There is much less background signal while vessels of interest are uniformly illumi nated.
51 Figure 3 6 Extravasated liposomes 24 hours after injection hinder the ability to perform repeated measurements A) Brightfield and B) corresponding fluorescent images of tumor microvasculature 24 hours after liposome injection. Fluorescent lipos omes have aggregated in the tumor space as well as in small vessels of the tumor.
52 Figure 3 7 Improved c learance of liposomes containing DOPS. A) Brightfield image of normal microvasculature. B) Liposome clearance 4 hours post injection. C) Brightfield image of tumor microvasculature. D) Liposome clearance 24 hours post injection.
53 Figure 3 8 Preliminary testing of FPF imaging MATLAB code to determine BST from first pass fluorescence images. A) Fluorescent dye was recorded flowing through a capillary tube. The arrow indicates the direction of flow. Two regions of interest were taken for ROI analysis. B) Test BST map of dye flowing through the capillary tube. C) Region of interest analysis for two areas in the capillary tube. The MATLAB code set the p eak of ROI 1 to be set at 0 seconds. D) Histogram of BST in the field of view.
54 Figure 3 9 Overview of BST map creation from first pass fluorescence imaging. A) Fluorescent liposomes are injected and simultaneously recorded as the liposomes first enter the field of view. B) MATLAB is used to create a vascular mask of the vessel area and assign BSTs to each pixel in the vascular mask. The BST is defined as the time it takes to reach the average pixel intensity for that pixel in the vascular mask. C) Fina l BST map The color r ed corresponds to faster BSTs and blue longer BSTs.
55 Figure 3 10 Six frames from first pass imagin g of a tail vein injection of ~ 50 l of DiD liposomes through the vasculature of normal and 4T1 tumor tissue. A) ~1.93 s after inje ction liposomes first appear in the fi eld of view. Supplying arteries are beginning to be filled. B) ~5.78 s after first appearance, the microvasculature of the tumor has been filled. C) and D) ~7.7 s and ~11.55 s after first appearance, draining from the tumor and into the main draining veins occurs. E) ~19.25 s after first liposome appearance. Veins are being filled with blood that has traveled through the normal tissue hierarchy. F) ~30.8 s later all vessels are fully filled in the field of view.
56 Figu re 3 11 Single day Hb saturation and BST c omparison of tumor vasculature versus normal vasculature. Draining veins (DV) and supplying arteries are indicated (SA). A ) Brightfield image of normal tissue. B ) Hb saturation map of normal tissue. C ) BST map of normal tissue. D ) Brightfield image of a 4T1 mammary carcinoma. E ) Hb saturation map for the field of view including the 4T1 tumor and surrounding normal vasculature. F ) Corresponding BST map. Note that the colorbar scale for each BST map is different.
57 Figure 3 12 Hb saturation and BST ROI analysis for normal vasculature reveals a larger number of vessels with lower oxygenation and slower BSTs. Points are represented as mean standard deviation. A) Brightfield image of ROIs chosen for comparison of B ST and Hb saturation. B) Hb saturation image with ROIs chosen. C) BST map with ROIs chosen. D) Hb saturation versus BST values for different ROIs indicated above. The supplying arteries have closely correlated BSTs and Hb saturations. The draining vein has a varied Hb saturation due to mixing of blood from different vessels merging with the draining vein. Intermediate vessels are generally lower in oxygenation.
58 Figure 3 13 Hb saturation and BST ROI analysis for a 4T1 tumor shows the development of vess els with faster BSTs and higher oxygenations. Points are represented as mean standard deviation. A) Brightfield image of ROIs chosen for comparison of BST and Hb saturation. B) Hb saturation image with ROIs chosen. C) BST map with ROIs chosen. D) Hb satu ration versus BST values for different ROIs indicated above. The supplying artery values and draining vein values are more closely clustered with respect to Hb saturation with the tumor interior and tumor periphery having a wider range of Hb saturation val ues.
59 Figure 3 14 ROI analysis of the draining vein of a 4T1 tumor shows the effects of arteriovenous shunts on blood flow and oxygenation. A) Brightfield magnified region of analysis. B) Hb saturation and C) BST map for the indicated region. Blood f rom the tumor reaches the main draining vein (V3 and V5) more quickly than blood from normal tissue (V1 and V4) resulting in different oxygenation values on either side of the draining vein. D) Hb saturation versus BST for ROIs chosen (mean standard devi ation) ROIs 3 6 represent blood coming from the t umor microvasculature. ROIs 1, 2 and 7 9 represent blood coming from normal tissue vasculature. The blood from the tumor microvasculature empties into the draining veins quicker than th at from the normal va sculature and consequently, this blood is also more oxygenated. ROI 2 represents a region where blood was slower to reach it, however mixed with another incoming vessel from the tumor thus increasing the Hb saturation value for that region.
60 CHAPTER 4 COM BINATION OF SPECTRAL AND FLUORESENCE IMAGING MICROSCOPY FOR WIDE FIELD IN VIVO ANALYSIS OF MICROVESSEL BLOOD SUPPLY AND OXYGENATION OVER TIME Introduction Abnormal microvasculature and angiogenesis are characteristics of a wide variety of pathologies such as diabetes, hypertension and tumors. In many instances, alteration of the function and morphology of microvessels can lead to the propagation of these diseases  As new treatments targeting pathological microvasculatur e and angiogenesis are developed, it is of greater importance to utilize small animal models to analyze the relationship between vessel morphology, blood flow and oxygenation. Currently there are a variety of methods for the analysis of tumor microvasculat ure in small animal models  The murine dorsal skinfold window chamber is on e such model in which it is possible to analyze the same animal over extended periods of time. When dealing with the analysis of changes that occur in physiological processes of tumor vascular development, the ability to analyze the same tumor over time is paramount. Conventional histological analysis provides important information on the cellular and molecular level, however the sacrifice of the animal is required and little is ascertained regarding the function of the vessels in the tumor space. The windo w chamber model allows for high resolution imaging of the same tumors over time, and the development of vasculature can be observed non invasively. We present the novel combination of two widefield imaging techniques, hyperspectral imaging of hemoglobin (H b) saturation and first pass fluorescence (FPF) fluorescence imaging microscopy for wide field in vivo analysis of microvessel blood supply and (3), (2013).
61 imaging of blood transit time in the murine window chamber model to simultaneously characterize the oxygenation of and blood flow through microvessel networks over time This combination of imaging techniqu es enables a unique analysis of oxygen transport in microvessels that highlights the effects of microvessel network connections on oxygen transport. Previously FPF imaging has been used to characterize a tumor at a single time point using one fluorescent tracer. By adding a second fluorescent tracer at a different wavelength, more time is allowed for liposomal clearance and FPF analysis can be performed daily. B oth imaging techniques are easily implemented and allow for serial observations in the same anim al model. Hb saturation imaging provides information regarding the oxygenation of blood within the microvessels and is useful in its own for analysi s of microvascular pathologies FPF imaging involves recording the flow of a fluorescent contrast agent as i t is injected into the circulation. Imaging the appearance of the agent in the microcirculation gives information regarding blood flow, morphology of microvessels and network connections through measurement of the blood supply time (BST) This parameter in dicates the time it takes for blood to travel to individual microvessels in a network from a common reference supplying artery [16, 77, 78] In this study we analyze the vascular networks of growing tumors over different periods of time, characterizing the formation of tumor microvascular networks as well as the vascular networks seen in tumor initiation, wound healing and normal vasculature Materials and Methods Tumor Cells Caki 2 human renal cell carcinoma ( RCC) cells and PC3 ML human prostate cancer cells (gift s from Dr. Dietmar W. Siemann, Department of Radiation Oncology
62 University of Florida College of Medicine, Gainesville, FL ) were utilized to demonstrate the utility of Hb saturation imaging and FPF im aging for the analysis of tumor microvasculature. All tumor cells were cultured in vitro in media containing 10% fetal bovine serum (Bio Whittaker, Inc) and 1% penicillin streptomycin (Hyclone) in DMEM (Cellgro, Herndon, VA). Cells were incubated at 37C in a 5% CO 2 atmosphere. A 10 L injection of 7.5x10 4 cells prepared in phosphate buffered saline (PBS) was used to initiate tumors for each cell line. Animal Model All in vivo procedures were conducted under a protocol approved by University of Florida Ins titutional Animal Care and Use Committee. A titanium dorsal skinfold window chamber was surgically implanted in female athymic (nu/nu) nude mice weighing more than 21 g (Harlan Laboratories, Indianapolis, IN) Mice were first ane sthetized using an injectio n of xylazine (10 mg/kg) and ketamine (100 mg/kg) intraperitoneally. A 12 mm in diameter circular area of the upper layer of dorsal skin was removed and a single cell suspension of tumor cells in PBS was injected subcutaneously for windows of mice used for tumor analysis For all tumor analyses, a bout 7.5x10 4 tumor cells were implanted in each window to generate a tumor sized between 3 4 mm in 5 7 days. This number of cells allowed for tumors to be formed more quickly therefore avoiding window viability iss ues. A single puncture wound in the window area was created with a 16 gauge needle for animals used for wound healing analysis. Following tumor cell injection or wound puncture a round glass coverslip was placed over the exposed window. Animals were then housed in an environmental chamber at 33C and 50% humidity with free access to food and water with a standard 12 hour light/dark cycle.
63 Fluorescent Liposome Preparation Liposomes were prepared using the lipid hydration method  All lipids and extruder were purchased from Avanti Polar Lipids, Inc. (Alalbaster, AL). Liposomes were composed of 1,2 dioleoyl sn glycero 3 phosphocholine (DOPC), 1,2 dioleoyl sn glycero 3 phospho L serine (DOPS), 1,2 distearoyl sn glycero 3 phosphoethanolamine N [methoxy(polyethyleneglycol) 2000] (DSPC mPEG200 0) and cholesterol combined at a molar ratio of 47.4:19.7:1:39.7. Liposomes were fluorescently labeled with dye in ethanol (2 mM) added to between 0.5 and 1 mol.% of th e total phospholipid content. The dyes used were fluorescent carbocyanine dyes 1,1' dioc tadecyl 3,3,3',3' t etramethylindodicarbocyanine 4 chlorobenzenesulfonate (DiD; ex 644 nm em 665 nm) or 1,1' dioctadecyl 3,3,3',3' tetramethylindotricarbocyanine iodide (DiR; ex 750 nm em 780 nm) purchased from Invitrogen (Carlsbad, CA). This mixture w as dried into a thin film with nitrogen gas and held under vacuum in the dark overnight to remove traces of the solvent. The film was rehydrated in sterile PBS (pH = 7.4) by vortexing at room temperature for 15 minutes to yield a suspension of multilamella r vesicles at a lipid concentration of 10 mg/mL Unilamellar liposomes ~100 nm in diameter were obtained by repeated extrusion of vesicles through a Nuclepore membrane with a 100 nm pore size. Experimental Design Hb saturation imaging was performed first and was immediately followed by F PF imaging. The use of DiD and DiR dyes were alternated each day to allow 48 hours of clearance of liposomes from the tumor area. A Zeiss Axioimager microsope (Carl Zeiss, Incorporated, Thornwood, NY) w as used as the imagi ng platform. All experimental images were acquired using long working distance objective of 2.5x magnification
64 Imaging of established tumor vasculature began o nce Caki 2 and PC3 ML tumors reached 2 3 mm in diameter (~4 5 days following tumor cell injecti on), and continued daily for 4 days. Imaging for normal vasculature, tumor initiation and wound healing all started on the day following surgical installation of window chambers. Imaging of normal tissue vasculature and the initiation of tumor vasculature was performed for 9 consecutive days. Imaging of mice for wound healing analysis was performed daily until the wounds healed. Hb Saturation Imaging and Image Processing Hb saturation imaging of tumor microvasculature was described in detail previously [8, 30] Briefly, a 100 W tungsten halogen lamp was used for transillu mination of the window chamber. For image acquisition, a monochrom a tic scientific grade charge coupled device ( CCD ) camera (DVC Company, Austin TX) therm oelectrically cooled to 20C was used. A liquid crystal tunable filter (CRI, Cambridge, MA) with a 400 to 720 nm transmission range was placed in front of the camera to acquire spectral image data sets from 500 nm to 575 nm at 5 nm intervals. W ith a 2.5x objective the resulting field of view had dimensions of 4.15 mm in width by 3.1 mm in height. Hb saturation pseudocolor maps were created in MATLAB (The Mathworks Inc., Natick MA) using pure oxy and deoxyhemoglobin reference spectra and a prev iously determined linear least squares regression model  For comparison to BST maps, Hb saturation maps were cropped to the same width as BST maps to display the same field of view.
65 FPF Imaging and Image Processing Fluorescence images were acquired using an ANDOR iXon electron multiplying CCD (EMCCD) camera thermoelectrically cooled to 50C (ANDOR Technology, South Windsor, CT). The field of view w as 3.1 mm in width by 3.1 mm in height. An injection of ~50 L of liposomes was administered via tail vein injection for FPF imaging. At least 700 stacked TIFF images were acquired during liposome injection at a frame exposure time of 117 m s to capture th e transit of liposomes throughou t the entire vascular network. Cy5 (ex 64020 nm / em 68030 nm) and Cy7 (ex 74035 nm / em 79550 nm (Chroma Technology Corp., Rockingham, VT) filters were used to capture fluorescence from the DiD and DiR labeled liposomes respectively. Image processing was performed using MATLAB and MeVisLab (MeVis Medical Solutions AG, Bremen, Germany) software A vascular mask was created for each image stack in MeVisLab using thresholding, Gaussian smoothing, Hessian vesselness and fu zzy c means algorithms. In MATLAB each image stack was truncated to the frame at which liposomes first entered the field of view and the frame at which the liposomes completely filled the microvasculature. Each data set was flatfielded by dividing them w ith images of the background before liposomes entered the field of view. To create a BST map, every pixel in the vascular mask was designated a BST based on its fluorescence intensity profile over time. The value was determined as the time at which the pix el intensity reached the average of its minimum and maximum value. Vessels that were more rapidly filled with liposomes registered faster BSTs and vessels that filled with liposomes last had slower BSTs.
66 Results Figure 4 1 illustrates frames from an image stack of FPF imaging showing the shunting of blood by newly formed tumor vessels from supplying arteries, through the tumor mass, and into draining veins. The formation of tumor vessels bypasse d the systemic blood flow and allow ed the tumor to receive high ly oxygenated blood. The last vessels to fill were the draining veins which receive d blood from systemic flow as well as flow from the tumor. Figure 4 2 shows (4 2A ) brightfield images, ( 4 2B ) Hb saturation and ( 4 2C ) BST maps of a Caki 2 tumor over 4 days Normal, unaffected vessels surrounding the tumor on d ay 1 exhibit ed expected oxygenation and blood flow, with highly oxygenated vessels (i.e arteries) corresponding to faster BSTs and lower oxygenated vessels (i.e. veins) having slower BSTs. From the BST maps we saw that highly oxygenated vessels of the tumor core ha d fast BSTs, indicating these vessels were quickly filled with blo od from the adjacent supplying artery. These vessels contain ed arteriovenous (AV) shunts that deliver ed blood directly from arteries into veins, resulting in a variability of oxygenation and BSTs in the tumor periphery as blood from the systemic flow ( ) and tumor ( ) converge d into the same vein. As the tumor gr ew and angiogenesis continue d there was evidence of heterogeneity in oxygenation and BST throughout the tumor represented by color changes in the maps. By d ay 4, development of areas of low oxyge nation and longer BSTs was evident on the right side of the tumor. At this time the tumor ha d grown to overtake the supplying arteries on that side. Magnified regions of interest in Figure 4 2B and 4 2C illustrate the utility of using this combination ima ging method to analyze AV formations with great detail. Day 1 maps reve aled that the left venous branch ( ) carrie d blood from the systemic flow
67 while the right branch ( ) wa s supplied with blood via shunts from the tumor. Because of this, blood from the systemic flow arrive d much more slowly compared to the blood flow from the tumor. By d ay 4, the tumor had grown such that the newly formed vessels were shunting enough blood to both branches of the draining vein, decreasing the BST and increasing the over all oxygenation of the network with respect to d ay 1. Figure 4 3 demonstrates region of interest (ROI) analysis for the draining vein vascular network seen in Figure 4 3A. Figure 4 3B shows the mean Hb saturation vs. mean BST for each ROI from Figure 4 3A over time. Both the ranges of Hb saturation and BST shortened over time. Day 4 showed the most dramatic change in which the Hb saturation and BST range were both greatly reduced. The overall values of Hb saturation increased and B ST values decreased over time with the growing tumor as seen in Figure 4 3C with ROI 6, 7 and 8. ROI 6 and 7 represent two sides of the same draining vein while ROI 8 denotes the entire width of the vein downstream. ROI 6 and 7 represent flows that come fr om the systemic flow and tumor respectively, as previously shown in Figure 4 2 ROI 6 show ed the most dramatic changes in BST and Hb saturation as the blood flow originate d from the systemic flow and was slowly altered as more AV shunts are formed. Figure 4 4 is another example of Caki 2 tumor growth over a period of 4 days. On day 1 we s aw that the venous vessels surrounding the tumor mass maintained their lower oxygenation and slower BSTs. By day 2 the tumor vasculature developed such that the surrounding vessels have developed faster BSTs and higher oxygenations. Just as in the previous example we observed draining veins that develop ed faster BSTs and higher oxygenations as the tumor mass overt ook these vessels. B y day 4 an area of
68 hypoxia formed in the tumor core as seen in the pocket of low oxygenation in the Hb sat uration map Figure 4 5 is a demonstration of the combination im aging analysis applied to the slower growing PC3 ML tumor. Imaging began at a later stage once the tumor reached 3 mm in diameter. In comparison to the Caki 2 tumor, the slower growth of the PC3 ML was evident in the lack of development of the tumor vessel s over time Oxygenation in the tumor space stay ed fairly consistent as d id the BST values from day 2 to day 4. Regardless of differences in rate of tumor growth, the same increases in oxygenation and BST of the draining veins surrounding the tumor space w ere observed as a result of AV shunting Other a pplications The combination of FPF imaging and Hb saturation imaging can be used t o analyze other types of vascular processes in addition to established tumor vessel maturation. Normal vasculature, wound healing and tumor initiation all were assessed in the present study. All of the following examples entailed starting i maging analysis on the day following surgical implantation of the window chamber. Figure 4 6 shows normal tissue vasculature over 9 days. The oxygenation and BSTs for the main arterial vessel in the window stay ed consistent over time, however vessels downstream of this ar tery display ed the transient nature of blood flow. In the main draining vein of the field of view, we could see the development of variable flow paths over time as blood merge d into the draining vein from different vessels. The differences in flow patterns were reflected in oxygenation as single vessels displayed varying oxygenations on either side of the vessel. By day 8 and 9 there was cloudiness in the field of view which c ould be an indicator of infection in the window chamber.
69 Concurrently with this ch ange in microenvironment, new vessels were formed and BSTs in the region bec a me faster with an increase in oxygenation. Figure 4 7 is an example of a wound healing study in normal tissue. On day 1 we observe d vessels of the normal tissue space beginning to surround the wound. On day 2 there wa s a defined ring of vessels around the wound, with faster BSTs and higher oxygenation be ginning to develop with the new vascular formation. Blood in the main draining veins of the field of view ha d faster BSTs and higher oxy g enations as well due to the shunting of blood from these newly forming vessels around the wound and into the vein s The network of vessels surrounding the wound bec ame more organized starting on day 3. The vessels surrounding the wound continue d to have fast BSTs and high oxy genation through days 5 and 6 The vessel bed develop ed closer to the wound site until day 7 when t he wound close d and vessels developed through the area where the wound previously existed At this time the BSTs and oxygenation in the field of view beg a n to stabilize and return to normal, with less angiogenesis and slower BSTs developing in vessels of t he surrounding tissue. Figure 4 8 demonstrates the initial stages of angiogenesis following injection of tumor cells into the window chamber. On day 4 it was evident via brightfield images that a small tumor nodule was form ing in the middle of the field of view. During this time the vessels that were forming in this area were extremely thin. On the same day we saw slight increases in oxygenation and faster BSTs in the immediate area of the nodule. In addition to the formatio n of new vessels the existing vessels from the top and bottom of the field of view began to grow towards the middle where the tumor was developing. On day 5 FPF imaging detected the fast blood flow through the tumor nodule area as we
70 observed new vessel f ormation characterized by high oxygenation. On day 6 there was a greater degree of new vessel development and we observed that vessels from the top and bottom of the field of view joined around the periphery of the tumor nodule. This created a tumor core t hat was increasingly oxygenated surrounded by less oxygenated vessels. For the remaining days we saw the same behavior as reviewed previously in this chapter, a rapidly developing tumor with more vasculature being formed and the tumor mass overtaking the n ative vessels in the surrounding tissue space. The formation of tumor microvessels created higher oxygenation values in the vessels surrounding the tumor. This was a result of AV shunting of the blood into these vessels represented by faster BSTs. Finally on day 9 the tumor mass completely overtook these surrounding vessels. Discussion General observations have been made concerning the formation of new pathological connections however t heir effect on tumor microenvironment has not been heavily researched  While H b saturation maps g i ve microvessel oxygenation data, n o information about the behavior of the blood flow is provided The addit ion of BST analysis elucidates the direction of blood flow and blood transit time in the tumor which complements the oxygenation data. The use of two different fluorescent dyes allow ed us to utilize our FPF imaging analysis to observe tumor vessel growth o ver a period of days. With daily analysis we demonstrated how microvessel development of a tumor can rapidly change the behavior of blood flow and oxygenation in a short period of time (Figure 4 2 ). The formation of AV shunts in tumors is a commonly observ ed process  These shunts oxygenate d tumor s during early stages of growth by allowing blood to
71 flow d irectly from supplying arteries through the tumor for fast BSTs. Given enough time to develop, we revealed that tumor microvessels eventually form more complicated networks that create variability in oxygenation and BST. This wa s seen in draining veins sur rounding tumors where Hb saturation was i ncreased and BSTs decreased as the tumor gr e w ( Figure 4 3 ). This shunting behavior was shown in multiple tumor types (Figures 4 4 and 4 5) It is well known that the perfusion of tumors can vary greatly between tumo r types, and even among tumors of the same cell line, size and implantation site [92 95] After tumors bec a me more developed we observed the differences between blood flow and vessel development in different tu mor types. The slower growing PC3 ML tumors show ed much less variation in BSTs and oxygenation over time in comparison to the rapidly proliferating Caki 2 tumors. Unlike the varying vascular behavior of vasculature in larger, more mature tumors, our analys is revealed that the initial stages of tumor vessel formation are similar across tumor lines. Our investigation of the microvasculature is an example of how the combination of Hb saturation imaging and FPF imaging can help identify trends and differences i n blood flow and microvasculature of tumors Further evaluation of the perfusion of different types of tumors at different stages can reveal physiological processes and factors that could aid in the grading and staging of clinical cancers. We have illustr ated the use of FPF imaging and Hb saturation imaging to analyze developed tumor nodules over a short period of time. However, this combination imaging modality can also be used for longer periods of time and for a variety of pathologies. The initial devel opment of a tumor, wound healing and normal vascular
72 function are just a few examples of vasculature that can be observed over a long period of time. The observation of normal vasculature over a period of 9 days demonstrate d how dynamic normal blood flow can be with no outside perturbation aside from the window chamber surgery. The oxygenation and BSTs of the main arterial vessel in the field of view did not change over time. However, any vessel downstream of this vessel d id experience transience in blood flow and oxygenation. Any changes in BST were reflected in similar changes of Hb saturation. This strong correlation between changes in BST and oxygenation were lost in the microvessels of a developed tumor. In a tumor, for a given Hb saturation value, the range of BSTs that correspond to that oxygenation value varies much more in comparison to normal tissue. This can be attributed to the more complicated vascular network that is developed, shunting blood from vessel to vessel without organization. As the o rganization and structure of vessels in a tumor become more chaotic, so does the delivery of oxygen and other nutrients. Infection and inflammation of the window chamber was also observed to create changes in blood vessel function. The changes in the windo w seen on days 8 and 9 are exemplary of these physiological changes aside from tumor development that can create vessel formation and faster BSTs with higher oxygenation  The evaluation of wound healing in small animal models is of great interest for the real time analysis of this physiological process. Overall, the in vivo analysis of changes to blood flow and vasculature during wound healing has been limited. Previously, l aser speckle imaging and murine ear puncture models have bee n used to analyze blood flow in different stages of wound healing  The use of FPF imaging
73 and Hb saturation can be easily implemented to elucidate changes in both the hemo dynamics and oxygenation in vesse ls associated with wound healing (Figure 4 7) With the introduction of a puncture wound to the skin, rapid vessel formation around the wound with fast BSTs and high oxygenation resemble d the vessel formation of an initiating tumor. The n eovascularization in wound healing compared to tumor formation emphasize d how much more regulated angiogenesis is in normal processes [98, 99] On day 1 the wound was in the inflammatory phase during which the vessels clo t t ed an d beg a n to transition into the proliferative phase of wound healing. From day 2 to day 6 the new vessel s formed in a dense network around the wound. These vessels resemble d that of initial tumor vessels with a higher vascular density in comparison to norma l tissue This increase in vessel density was expected in wound healing where it has been shown that the capillary content can reach three or more times of normal tissue for proper wound healing  Blood was shunted to the wound area to bring oxygen and nutrients necessary for closure of the wound. Once the wound closed, the microenvironment in the window chamber immediately beg an to stabilize, with less variance in oxygenation in the field of view and vessels returning to longer BSTs as vessels return ed to their normal hie rarchy. This behavior is seen during the tissue remodeling phase of wound healing with regression of microvessels and differentiation of the newly formed vessels  The density of vessels in the wound area a lso decrease d, which is in stark contrast to a developing tumor wherein the vasculature continues to proliferate for a more complicated network and convoluted delivery of oxygen and other nutrients.
74 Harold F. Dvorak described tumors as wounds that do not heal  Our combination imagi ng analysis of the initial development of tu mors (Figure 4 8) and the healing of a wound (Figure 4 7) supports this comparison. The development of microvasculature in a tumor mass from a bolus of tumor cells is a much slower process in comparison to vessel formation after the introduction of an open wound. Our experiments show ed that the tumor cells utilized existing host vessels to support the initial formation of the tumor 3 to 4 days after implantation  Despite a large number of nearby venous structures, angiogenesis of tumor vessels initiated from existing supplying arteries as fine, highly oxygenated vessels throug h the center of the tumor mass. The venous structures with lower oxygenation above and be low the tumor mass instead gr e w towards the tumor and develop around the tumor periphery, forming into the draining veins seen in previous tumor observations in this chapter. The dramatic changes of native tissue vasculature as a result of the introduction of tumor cells have previously been observed in the window chamber  Once we were able to detect tumor vessels through FPF imaging on day 5, we observed further development of these new tumor vessels whic h beg a n to increase the oxygenation of the vessels immediately surrounding the tumor mass and progress into the tumor vessel proliferation observed earlier in this chapter This same progression of tumor vasculature development following an injection of ce lls was observed by Vajkoczy et al. where they saw the first signs of tumor induced angiogenesis on days 2 to 4 after glioma cell ( 5x10 5 cells) implantation in rat window chambers, and continued development of tumor microvasculature  While our study concentrated on a larger
75 field of view, future considerations for the study of tumor initiation could benefit from evaluation at a higher magnification to better characterize the smaller forming vessels. In summary we have demonstrated the novel combination of hyperspectral imaging of Hb saturat ion and FPF imaging of blood transit time as an effective technique for analyzing the oxygenation of and blood flow through microvasculature in the mouse dorsal skin window chamber model over time This imaging modality allow ed BST and Hb saturation maps t o be created at microvessel spatial resolution. In this experiment the resulting maps provided more insight into how vascular malformations such as AV shunts create an oxygen rich environment for an immature tumor to proliferate. The ROI analysis that can be performed result ed in quantitative data that can help determine blood flow and oxygenation trends that occur with pathological angiogenesis. We demonstrated the analysis of different vascular processes using this technique as proof of principle of the u tility of t he imaging modality in the characterization of microvessel oxygenation and function. Normal vessel function, tumor initiation and wound healing analys e s were performed for duration s of 7 days or longer. For future considerations, t his technique could be be utilized in prospect ive research to analyze the response of tumor microvasculature to different antagonists, including therapeutic vascular targeting agents. Due to the nature of the dorsal window chamber model, it is also possible to take tiss ue samples for histological analysis and allow for alternate data collection.
76 Figure 4 1. Six frames from FPF imaging of DiD liposomes through the vasculature of a Caki 2 tumor. ( A ) ~1.31 s after liposomes first appear in the field of view. The supplying arteries for the tumor start to fill ( B ) ~2.61 s vasculature within the tumor is beginning to be filled. ( C ) ~3.92 s blood is reaching the draining veins of the tumor on the right side ( D ) ~7.83 s blood flow from the tumor has filled the draining veins on either side of the tumor. ( E ) and ( F ) 17.4 s and 34.8 s blood from the normal systemic flow reaches the main draining vein as seen by the veins filling in the top right of the fie ld of view.
77 Figure 4 2 A nalysis of a Caki 2 tumor over 4 days illustrates the rapid changes of vessel structure and the effects on Hb saturation and BST ( A) Bright field images. The supplying artery (SA) and draining vein (DV) of interest are indicated. ( B) Full Hb saturation and ( C) BST maps with magnified region of interest indicate the development of AV malformations. Venous branches carrying blood from systemic flow [arrowh ead symbol in lower left of (B)] and shunted tumor blood flow [arrow symbol in lower left of ( B)] are indicated.
78 Fig ure 4 3 ROI analysis for the draining vein of a Caki 2 tumor shows that AV shunts are responsible for faster BSTs and increasing oxygenation ( A) ROIs chosen for comparison of BST and Hb saturation in tumor draining vein. ( B) Hb saturation as a function o f BST over four days (mean standard deviation) for ROIs seen in A). ( C) Comparison of mean values of BST and Hb saturation for ROIs 6, 7 and 8 over time (mean standard deviation).
79 Figure 4 4 FPF and Hb saturation maps show that tumors increase the oxygenation and create faster BSTs in surrounding normal vessels with time From day 1 to day 2 the tumor microvasculature of a Caki 2 tumor has developed to shunt blood to the normal vessels surrounding the tumor creating faster BSTs and higher oxygenation. This behavior continues over time as the BSTs and oxygenation values become increasingly varied in the tumor core.
80 Figure 4 5 FPF and Hb saturation imaging analysis of PC3 ML tumor development over four days illustrates the differences in tumor vessel development in varying tumor types The vasculature of this tumor type de velops much more slowly with the same vascular structures identifiable from day to day. Consequently, the changes in BST and oxygenation from day to day are minima l
81 Figure 4 6 Long term FPF and Hb saturation imaging analysis of normal tiss ue over 9 days shows the transient nature of normal blood flow Supplying artery (SA) and draining veins are indicated (DV). The blood flow and oxygenation in arteries of the field of view stay consistent over time, however blood flow changes in the draining veins are seen over time. Multiple flow paths of diffe rent BST and oxygenation values are observed in the largest draining vein as blood flow from different sources join in this vein Inflammation is seen in the window on days 8 and 9, creating blood flow and oxygenation changes to the immediate area.
82 Figure 4 7 Combination imaging analysis of a puncture wound characterizes the tightly regulated process of wound healing Rapid angiogenesis around the wound is evident starting on day 2. Vessels form a dense capillary bed aroun d the wound to transport oxygen and nutrients necessary for proper healing. On day 7 the wound is completely closed and vessels in the field of view begin to return to normal function and behavior. The BST values of the vessels slow and oxygenation decreas es as the vessels necessary for wound healing regress.
83 Figure 4 8 Combination imaging analysis of Caki 2 tumor initiation illustrates how tumor cells develop vasculature for further proliferation. A small tumor area with vessel formati on and remodeling is evident on day 4. On day 5 newly formed vasculature through the tumor mass is seen in BST maps with fast BSTs and high oxygenation. Existing venous vessels from the top and bottom of the field of grow towards the tumor mass and eventua lly develop into the main draining veins surrounding the periphery of the tumor. The tumor continues to develop as seen in previous examples.
84 CHAPTER 5 IN VIVO MICROSCOPY COMPARISON OF MICROVASCULAR FUNCTION AFTER TREATMENT WITH OXI4503, SUNITINIB AND THEIR COMBINATION IN CAKI 2 TUMORS Introduction Tumor growth d ependence on the acquisition of adequate blood supply has led to significant effort towards the development of vascular targeting agents (VTAs)  Two major classes of VTAs are the vascular disrupting agents (VDAs) and angiogenic inhibiting agents (AIAs). VDAs act to destroy existing tumor vasculature by targeting their rapidly proliferating endothelial cells [88, 104, 105] The tubulin binding agent OXi4503 is a VDA that interferes with the formation of microtubules in rapidly pr oliferating endothelial cells and results in structural changes that lead to decreases in tumor perfusi on, tumor vascular permeability and ultimately the damage and destruction of tumor vasculature [11, 28, 36] Distinct from VDAs, AIAs aim to inhibit the development of new vasculature by targeting angiogenic stimulators and cytokines that lead to angiogenesis and the formation of tumor vasculature. The multitargeted tyrosine kinase inhibitor Sunitinib is an AIA that is currently approved for treatment of renal cell carcinoma (RCC) and imatinib resistant gastrointestinal stromal tumor (GIST)  The use of VDAs and AIAs on their own has shown to be insufficient for complete treatment of solid tumors, however the complementary mechanisms by which these VTAs act merit investigation into the use of these drugs in combination  Previous studie s combining VTAs showed dramatic growth delays of tumors treated with both VDAs and AIAs in comparison to either treatment alone [58, 62] The goal of the present Manuscript in su b m microscopy comparison of microvascular function after treatment with OXi4503, Sunitinib and their combina tion in Caki
85 stud y was to utilize in vivo analysis of these treatments in real time to assess the effects that the combination of these drugs have on the tumor microvasculature itself. A valuable tool to analyze oxygenation and blood flow through microvessel networks in tumors and their response to VTAs is hypersp ectral imaging microscopy of hemoglobin (Hb) saturation and first pass fluorescence (FPF) imaging of blood transit time in a murine dorsal skin window chamber model. Previously, Wankhede et al. utilized Hb saturation imaging to investigate the oxygenation of tumor microvessels following treatment with OXi4503  While this analysis revealed microvessel functional changes via oxygenation data, no blood flow information was gathered to further characterize the treatment resp onse. FPF imaging of blood transit time has been used previously to characterize the transient changes in blood flow in developing tumors over time by rec ording the flow of fluorescent contrast agents throughout the vasculature [16, 77, 106] This addition of FPF imaging provides insight into changes in blood flow and vessel morphology by measuring the bloody supply time (BST) of vessels in the tumor. The BST is defined as the time it takes for fluorescent trace r s to travel to individual microvessels in a vascular network following injection relative to a reference supplying arteriole In the present study, the combination of hyperspectral imaging of Hb saturation and FPF imaging of blood transit time is used to analyze th e structural, blood flow and oxygenation changes that occur in Caki 2 human RCC during treatment with the VDA OXi4503, AIA Sunitinib and both of these drugs in combination. Materials and Methods Tumor Cells Caki 2 human RCC cells (a gift from Dr. Susan Knox, Stanford University ) were cultured in vitro in DMEM (Cellgro, Herndon, VA) containing 10% fetal bovine serum
86 (Bio Whittaker, Inc) and 1% penicillin streptomycin (Hyclone). Cells were incubated at 37C in a 5% CO 2 atmosphere. Dorsal Skinfold W indow Chamber Preparation All in vivo procedures were conducted under a protocol approved by University of Florida Institutional Animal Care and Use Committee. A titanium dorsal skinfold window chamber was surgically implanted in female athymic (nu/nu) nud e mice weighing more than 21 g (Harlan Laboratories, Indianapolis, IN). Mice were first anesthetized using an injection of xylazine (10 mg/kg) and ketamine (100 mg/kg) intraperitoneally. A 12 mm in diameter circular area of the upper layer of dorsal skin was removed and a single cell suspension of tumor cells in phosphate buffered saline (PBS) was injected subcutaneously. About 7.5x10 4 tumor cells were implanted in each window chamber to generate a tumor ~4 mm 3 in volume in 5 7 days. Following tumor cell i njection, a round glass coverslip was placed over the exposed window. Animals were then housed in an environmental chamber at 33C and 50% humidity with free access to food and water with a standard 12 hour light/dark cycle. Drug Preparation and Expe rimental Design OXi4503 and Sunitinib were received as a gift from Dietmar W. Siemann, Department of Radiation Oncology University of Florida College of Medicine, Gainesville, FL. Both drugs were prepared fresh on the day of administration. OXi4503 (10 mg /kg) was prepared in a solution of sterile saline and 50 L/mL of 5% sodium carbonate solution and given intraperitoneally. Sunitinib (100 mg/kg) was prepared in a vehicle of citric acid monohydrate and sodium citrate dihydrate at a ratio of 1:16.3 and adj usted to a pH of 3.5. Sunitinib was administered via oral gavage. Both drugs were given at a volume of 0.01 mL /g of the mouse body weight.
87 Mice were divided into four groups: untreated (control) (n=6), OXi4503 alone (n=5), Sunitinib alone (n=7) and both d rugs together (n=7). Tumors were allowed to grow until they reached ~4 mm 3 in volume at which time treatment was started. This day was designated as day 1. Figure 5 1 illustrates the drug administration schedule and imaging schedule. Sunitinib was given on days 1 6 and OXi450 3 was administered on days 1, 3 and 5. FPF imaging and Hb saturation imaging were performed on days 1, 3, 5 and 7. All imaging sessions were completed immediately before administration of any drugs for that day. Fluorescent L iposome Preparation Liposomes were prepared using the lipid hydration method  All lipids and extruder were purchased from Avanti Polar Lipids, Inc. (Alalbaster, AL). Liposomes were composed of 1,2 dioleoyl sn glycero 3 phosphocholine (DOPC), 1,2 dioleoyl sn glycero 3 phospho L serine (DOPS), 1,2 distearoyl sn glycero 3 phosphoethanolamine N [methoxy(polyethyleneglycol) 2000] (DSPC mPEG2000) and cholesterol combined at a molar ratio of 47.4:19.7:1:39.7. Liposomes were fluorescently labeled with dye in ethanol (2 mM) added to between 0.5 and 1 mol.% of th e total phospholipid content. The dyes used were fluorescent carbocyanine dyes 1,1' dioctadecyl 3,3,3',3' t etramethylindodicarbocyanine 4 chlorobenzenesulfonate (DiD; ex 644 nm em 665 nm) and 1,1' dioctadecyl 3,3,3',3' tetramethylindotricarbocyanine iodide (DiR; ex 750 nm em 780 nm) purchased from Invitrogen (Carlsbad, CA). This mixture was dried into a thin film with nitrogen gas and held under vacuum in the dark overnight to remove traces of the solvent. The film was rehydrated in sterile PBS (pH = 7.4) by vortexing at room temperature for 15 minutes to yield a suspension of multilamellar vesicles at a lipid concentration of 10 mg/mL Unilamellar liposomes ~100 nm in diamet er were
88 obtained by repeated extrusion of vesicles through a Nuclepore membrane with a 100 nm pore size. Intravital Microscopy For image acquisition, hyperspectral imaging of Hb saturation imaging was completed first followed by FPF imaging. A Zeiss Axio Imager microscope (Carl Zeiss, Inc., Thornwood, NY) was used as the imaging platform for all imaging. A 2.5x fluar (Carl Zeiss, Inc., Thornwood, NY) long working distance objective was used. Mice were placed on a heated platform and anesthetized using 1.5 % isoflu rane in air. Hb Saturation Imaging and Analysis H yperspectral imaging of microvessel Hb saturation was described in detail previously  Briefly, a 100W tungsten halogen lamp was used for transillumination of the window chamber. A monochromatic scientific grade charge coupled device ( CCD ) camera (DVC Company, Austin, TX) thermoelectr ically cooled to 20C was used for image acquisition. A liquid crystal tunable filter (CRI, Cambridge, MA) with a 400 720 nm transmission range was used to acquire spectral image data sets from 500 nm to 575 nm at 5 nm intervals. The final field of view w as 4.15 mm in width by 3.1 mm in height. Pseudocolor maps of Hb saturation were created in MATLAB (The Mathworks Inc., Natick MA) using pure oxy and deoxyhemoglobin reference spectra and a previously determined linear least squares regression model  Hb saturation images were cropped in width to show the same field of view as corresponding BST maps. FPF Imaging and Analysis FPF imaging of blood tra nsit time was discussed in detail previously  Fluorescence images were acquired using an ANDOR iXon electro n multiplying CCD (EMCCD) camera (ANDOR Technology, South Windsor, CT) thermoelectrically cooled
89 to 50C. The final field of view was 3.1 mm in width by 3.1 mm in height. An injection of ~50 L of fluorescent liposomes was administered via tail vein injec tion and 700 stacked TIFF images were simultaneously acquired at a total frame exposure time of 0.1 17 s to capture the transit of liposomes throughout the entire vascular network. The use of DiD and DiR liposomes was alternated to allow for sufficient clea rance of liposomes from the field of view. Cy5 (ex 64020 nm / em 68030 nm) and Cy7 (ex 74035 nm / em 79550 nm (Chroma Technology Corp., Rockingham, VT) filters were used to capture fluorescence from the DiD and DiR labeled liposomes respectively. BST m aps were created using MATLAB and MeVisLab (MeVis Medical Solutions AG, Bremen, Germany) software. A vascular mask was created for each image stack in MeVisLab using thresholding, Gaussian smoothing, Hessian vesselness and fuzzy c means algorithms. In MAT LAB each image stack was truncated to the frame at which liposomes first entered the field of view and the frame at which the liposomes completely filled the microvasculature. To create a BST map, every pixel in the vascular mask was designated a BST bas ed on its fluorescence intensity profile over time. The value was determined as the time at which the pixel intensity reached the average of its minimum and maximum value. Vessels that were more rapidly filled with liposomes registered faster BSTs and vess els that filled with liposomes last had slower BSTs. Tumor Volume Calculation The length, width and height of each tumor nodule were measured daily using calipers. These dimensions were used to calculate the volume of the tumor in mm 3 using the equat ion for half of an ellipsoid.
90 Immunohistochem ical Analysis of Window Chamber Tumors Following image acquisition on day 7, mice were sacrificed and tumors were harvested from the window chambers. Tumors were placed in optimal cutting temperature (OC T ) embedding compound (Sakura Finetek, USA, Inc., Torrance, CA) in plastic molds and frozen in a container containing 2 methylbutane surrounded by dry ice Once sufficiently set, tumors were cut into 5 m slices using a cr yostat. Slices were placed onto microscope slides and placed in a 80C freezer before fixation. Tumor slices were immunolabeled using rat anti mouse endothelial cell antigen (MECA) 32 (1:500, BD Pharmingen, San Diego, CA), then donkey anti rat Alexa Fluor 594 (1:500) to identify endothelial cells. Slides to be labeled were removed from the 8 0C freezer and placed in a 20C freezer for 30 minutes. Slides were removed and immediately suspended in cold acetone for 10 minutes. Slides were then washed with t ris buffered saline ( TBS ) for 5 minutes after which a block of 2% normal horse serum in 1x TBS was placed on the slides for a minimum of 1 hour at room temperature. The primary antibody was then added and slides were incubated overnight at 4C. After incu bation slides were washed with 1x TBS buff e r for 10 minutes. The seco ndar y antibody was added to the slides and incubated at room temperature for 1 hour. Slides were then post fixed in 10% normal buffered formalin for 5 minutes. Tissue sections were cover 6 diamidino 2 phenylindole (DAPI) (Vector Shield, Vector Laboratories Inc., Burlingame CA). Statistical Analysis Statistical analyses were performed using GraphPad Prism software (GraphPad Software Inc., La Jolla, CA) T umor volumes and IHC fluorescence signal
91 intensities were tested using the two tailed Mann Whitney rank sum test. Results with p < 0.05 were considered significant. Results Morphological Changes of Tumor Blood Vessels Daily brightfield image s were acquired to analyze the morphological changes of tumor blood vessels as a function of time (Figure 5 2 ). Control tumors exhibited a visible increase in microvessel density as the vessel network became more complex. Individual vessels became more tor tuous and varied in thickness. From day to day it was difficult to identify the same vessel structures, an indicator of the rapid vessel development in growing tumors. OXi4503 treated tumors displayed characteristic tumor vessel destruction in the tumor co re following the first treatment on day 1. On day 2 we observed a darkened area in the tumor interior indicating blood that leaked into the area as a result of the vessel damage. Vessels of the tumor interior were completely destroyed up to the rim of the tumor mass followed by some vessel recovery in the periphery of the tumor seen 48 hours later on day 3. Following the second treatment on day 3 we observed less vessel damage in the tumor periphery leaving a larger viable rim of vessels on day 4, and more aggressive vessel recovery on day 5. The last OXi4503 treatment on day 5 again resulted in modest vessel damage and aggressive vascular recovery into the tumor core. The vessels that recovered were similar in appearance to those of control tumors. In contr ast to the control tumors, tumors of Sunitinib treated mice displayed a decrease in microvessel density over time. The major vessel structures of the tumor were maintained throughout the experiment with smaller vessels being pruned. This
92 conservation of ve ssel structures allowed for the same vessel s to be identified from day to day. The combination treatment resulted in vascular destruction in the tumor core following treatment on day 1 akin to the destruction displayed in tumors of mice treated only with O Xi4503. However, unlike results seen in the OXi4503 group, there was no vessel regrowth into the tumor core when OXi4503 and Sunitinib treatments were combined. This lack of tumor vessel growth was maintained throughout the course of the combination treatm ents. In contrast to the tumor vessels, normal vessels surrounding the tumor maintained their structures. Only on day 7 was there any visual evidence of the tumor growing outward towards the normal vasculature when vessels surrounding the tumor area becam e slightly less visible as the tumor cells appeared to push these vessels out of the depth of field. Tumor Growth Inhibition Figure 5 3 shows tumor growth as a function of time (median interquartile range) Examination of tumor volumes revealed that all treatment groups showed a statistically significant tumor growth inhibition compared to the control group on day 7 (p<.01) In comparison to the control group, final tumor volumes were attenuated 4.8 fold in the OXi4503 treated group, 5.8 fold in the Sunitinib group and 13.8 fold in the combination group. Among treatment groups, the combinati on treatment significantly arrested tumor volume in comparison to the OXi4503 (p<0.0 1 ) and Sunitinib (p<0. 001 ) single agent treatment groups on day 7. There was no significant difference between Sunitinib and OXi4503 growth by day 7 and n o significant tumo r growth was noted throughout the experiment in mice treated with the combination of OXi4503 plus Sunitinib.
93 Immunohistochemical Analysis Tumor masses from window chambers were harvested and frozen following imaging on day 7 for immunohistochemical (IHC) analysis of vessel staining quantification. Tumor s lices were stained with MECA 32 for endothelial cell s and DAPI for nuclear material. Figure 5 4A shows examples of merged staining images for each experimental group. The vessels of control tumors were in general observed to be much larger and to occur at higher density in comparison to tumors in the three treatment groups. Although OXi4503 treated tumors had a higher presence of MECA 32 staining, the vessels in these tumors were fractured leading us to question their functionality. The cell density overall represented by DAPI staining was also found to be r educed compared to tumors of other groups. Sunitinib treated tumors showed much smaller vessels with a lower density in comparison to control and OXi4503 treated tumors. Although the vessel density and size observed in tumors of the combination treatment g roup were similar to those of the Sunitinib alone treated tumors, the functionality of these vessels was highly suspect when examining these images in conjunction with obs ervations from brightfield, FPF and Hb saturation imaging (Figures 5 1 and 5 5 throug h 5 8 ). Figure 5 4B shows the ratio of MECA 32 to DAPI staining for 10 random regions in tumors from the control (n=5), OXi4503 alone (n=4), Sunitinib alone (n=4) and combination treated tumors (n=4). All treatment groups showed a significantly lower MECA 32 to DAPI ratio in comparison to controls (p<.001) indicating less endothelial cell staining. OXi4503 treated tumors (p<.01) and combination treated tumors (p<.001) had a greater ratio in comparison to Sunitinib There was no significant difference betwe en OXi4503 and OXi4503 plus Sunitinib treated tumors.
94 FPF and Hb Saturation Imaging of Blood Flow and Oxygenation In normal blood flow, faster BSTs correspond to higher oxygenation values and vessels such as arteries. Conc urrently, slower BSTs correspond to lower oxygenation values and vessels such as veins [16, 106] For tumor microvasculature, this relationship becomes less well defined. The use of FPF and Hb saturation imagin g revealed blood flow and oxygenation trends for each experimental group (Figure s 5 5 through 5 8 ). Control tumors (Figure 5 5) developed an increase in overall BST and corresponding decrease in oxygenation with time. There was an increase in heterogeneit y of BSTs over time indicative of the increasingly chaotic microvessel morphology. As more complicated vascular pathways formed, the fluorescent tracer and thus the blood required longer times to travel to specific areas in the tumor. By day 7 the relation ship between BST and Hb saturation was less distinct with a wide range of BSTs corresponding to a particular Hb saturation value. The blood flow and oxygena tion behavior of vessels in OXi4503 treated mice varied wit h repeated treatments (Figure 5 6) After initial treatment on day 1, vessels were destroyed up to the rim of the tumor, thus vessel growth was comprised of newer vessels that originated from norma l vessels surrounding the tumor and surviving peripheral vessels with faster BSTs and high oxygenation. Following treatment on day 3, remaining vessels from the first treatment were likely only partially damaged, leaving a wider rim of vessels that had an established network from which new vessels could develop. Consequently, vessels on day 5 exhibited an increase in tumor BST heterogeneity and variation in oxygenation at the tumor periphery indicator s of a more complicated vascular network. After treatmen t on day 5, tumor vessels were again only
95 partially damaged, leaving an even thicker rim of more developed microvessels. These resulting vessels exhibited BSTs and oxygenation resembling control tumor vessels with a higher variation in both BST and oxygena tion values as tumors overcame the effects of OXi4503 treatment. As also seen in the brightfield images (Figure 5 2 ), BST and Hb saturation maps of tumors treated with Sunitinib (Figure 5 7) displayed a decrease in microvessel density over time. Tumors in this treatment group showed an increase in oxygenation and decrease in BST on day 3. Along with faster BST values, the variability in BSTs decreased as larger vessel structures were retained. The homogeneity in BSTs indicated that Sunitinib trea tment ha d resulted in a less convoluted blood flow pathway. On days 5 and 7, BSTs remained fast while oxygenation values were maintained with little variation in either parameter. Combination treated tumors left no visible tumor vessels to be analyzed (Figure 5 8) The surrounding normal vessels were maintained around the periphery of the tumor until day 7. On day 7 we observed faster blood supply times and higher oxygenations wh ich are indicative of newly forming vessels in the surrounding normal tissue. Region of Interest Analysis Thirty ROIs were chosen from the same areas of each tumor to plot the corresponding Hb saturation values versus BST over time as seen in Figure 5 9. This is a quantitative way to show changes in blood flow and oxygenation. For control tumors, BSTs increased as o verall oxygenation decreased. The range of BSTs increased from 46.83s on day 1 to 56.95s on day 7 with a greater proportion of longer BSTs. The OXi4503 treated tumor demonstrated faster BSTs associated with newly forming
96 vessels and an overall higher oxyge nation on day 3. Days 5 and 7 return to BST and oxygenation patterns seen on day1 as the tumor overcame the treatment. The Sunitinib treated tumor on day 1 had faster overall BSTs in comparison to the other tumor groups, however the BST values on the follo wing days demonstrate the normalizing effect of Sunitnib on vessels. The range of ROI BSTs on day 3, 5 and 7 were 5.64s, 2.71s and 1.72s respectively, showing that Sunitinib inhibits the formation of complicated vascular networks where longer BSTs are form ed. The oxygenation values remained high, with a majority of ROIs having Hb saturation values greater than 50%. The combination treatment group showed similar distributions of BSTs and oxygenation on days 3, 5 and 7 which we can attribute to the fact that these ROIs corresponded to the normal vasculature surrounding the tumor mass once the combination treatment destroyed the vessels in the interior of the tumor. Box and Whisker Plot Analysis Box and whisker plots are used to graphically depict population data. The top and bot tom of the boxes represent the first (Q1) and third quartiles (Q3) of the data. The middle lines represent the median. The top and bottom whiskers represent the highest and lowest data point within a factor of 1.5 of the difference between the third and fi rst quartile ( ). Any red stars outside of these whiskers indicate outliers. Figure 5 10 shows the box and whisker plots for the previous tumors to represent data of the entire field of view. As observed with the images and ROI data, the control tumor shows an obvious decrease in oxygenation over time with an increase in BST. All of the treatments inhibited this behavior and kept BSTs fast and oxygenation high. Although they all change the natural behavior of the tumor blood flow, Sunitinib treatment is shown to be more stable in comparison to OXi4503 treatment as shown by the
97 shortened range of BSTs and more constant Hb saturation. The combination treatment again is representative of the normal vasculature around the tumor area. On day 7 we see a trend of fas ter BSTs which when combined with our imaging observations is an indicator of new vessel formation. Discussion Because the development of new microvas culature is vital to the progression and growth of a tumor mass, tumor microvasculature is currently a popular target for developing therapies [10, 11] OXi4503, a tubulin binding VDA, damages or destroys exist ing vasculature while Sunitinib, a multiple tyrosine kinase inhibitor AIA, inhibits new blood vessel formation. Given their contrasting mechanisms of action, these classes of drugs used in combination might be anticipated to elicit a strong effect on tumor microvasculature and growth. In the current study, we utilize the mouse dorsal window chamber model and the combination of hyperspectral imaging of Hb saturation and FPF imaging of blood transit time to investigate tumor vascular response to the VDA OXi45 03, AIA Sunitinib, and combination of these two agents. An advantage to the window chamber model is the ability to simultaneously assess vascular development and tumor growth as a function of time. Overall, all treatment groups demonstrated significant tu mor growth inhibition in comparison to the growth of untreated control tumors. Among treatment groups the tumors of mice treated with the combination of OXi4503 plus Sunitinib exhibited the greatest tumor growth inhibition, exceeding that of either treatme nt alone. While our analysis dealt with small tumors, Siemann and Shi observed similar enhanced antitumor efficacy when applying VDA plus AIA combination therapy to established larger Caki 1 human RCC xenografts [107, 108] In two separate studies the combinations of OXi4503 with the AIA
98 Bevacizumab (Avastin) and the VDA ZD6126 with AIA Vandetanib (ZD6474) were found to result in significantly greater tumor growth delay than could be achieved with any vascular t argeting therapy administered alone [107, 108] Similar enhanced tumor growth delay following combination treatment in comparison to singular treatments has also been observed in AML treated with OXi4503 and Be vacizumab  and in PC3 prostate cancer xenografts treated with the VDA Combretastatin plus Sunitinib  IHC observations were made on tumors at the end of im aging analysis. Traditionally MECA 32 is used to stain endothelial cells to aid in vessel counting. However, due to the damage done to OXi4503 treated tumors and vessels, counting of vessels would be extremely subjective and thus we utilized the ratio of o verall MECA 32 staining of endothelial cells to DAPI staining of nuclear material. As expected, in comparison to control group tumors, tumors from all treatment groups demonstrated significantly reduced vessel staining (p<.001). There were however signific ant differences between the various treatment groups. OXi4503 treated tumors had a higher staining ratio in comparison to Sunitinib (p<.01) and also had the largest variability in MECA 32 staining. This finding likely reflects some of the temporal variabil ity that can occur in tumors experiencing vascular disruption following VDA treatment. In this study some tumors experienced disruption to the vasculature that was not seen until after the second or third treatment. Furthermore, the degree to which tumors can recover from VDA treatment can also vary  which may account for the results we observed with vessel staining. In contrast to OXi4503, Sunitinib displayed extremely consistent results where smaller vessels were pruned to leave behind mature vessel structures. By inhibiting new vessel formation vessel staining in t he tum or was reduced. Tumors of
99 m ice receiving the OXi4503 plus Sunitinib combination treatment also displayed a greater ratio of MECA 32 to DAPI staining (p <.001) in comparison to Sunitinib however the functionality of these vessels was highly suspect particularly in light of the FPF and Hb saturation imaging results (Figure 5 5 through 5 8 which clearly indicate that no functional vessels could be detected in the tumors of mice receivi ng the combined agent therapy. Our combination imaging approach revealed significant tumor microvasculature structure and function differences in control tumors as well as tumors of mice treated with the VDA, AIA, or their co mbination. Post analysis of BST and Hb saturation maps with ROI analysis and box and whisker plots further helped elucidate trends in the changes in blood flow and oxygenation. ROI analysis allows for specific BST to Hb saturation correlations in specific vessels to be determined while box and whisker plots reveal population data for the entire field of view. It is well known that unhindered tumor growth results in complex vascular networks with variable blood flow and oxygenation. Our observations of incr eased microvessel densities, and trends towards longer BSTs and areas of lower oxygenation in control tumors are consistent with previous studies of tumor growth [20, 52] In contrast to what was observed in co ntrol tumors, measurements made on tumors of mice undergoing vascular targeting therapies showed that all treatments hindered the development of longer BSTs and lower oxygenation. Initial treatment of Caki 2 tumors with OXi4503 resulted in vascular damage up to the edge of the tumor, leaving a viable rim of vessels and tumor cells characteristic to OXi4503 treatment that has been observed previously [27, 36, 42, 109] Forty eight
100 hours following treatment, new vessels formed from the rim, originating from both surrounding normal vessels and surviving peripheral vessels with faster BSTs and a higher overall oxygenation throughout. With a lesser degree of vessel damage and a higher more aggressive degree of vesse l recovery following each subsequent treatment, blood flow and oxygenation within vessels of the tumor began to resemble trends observed in control tumors. Our study suggests that Caki 2 tumor vasculature recovery may overcome OXi4503 treatment efficacy wi th time, but it should be noted that previous studies from our laboratory have shown that the structural recovery of tumors following serial treatments of OXi4503 can vary depending on tumor type. Wankhede et al. used mouse dorsal window chambers and hyper spectral imaging of Hb saturation to evaluate the response of 4T1 mouse mammary carcinoma and human Caki 1 RCC tumors to serial administration of OXi4503 every 48 hours  Like our observations with Caki 2 tumors, all tum ors suffered vascular damage up to the tumor rim following initial treatment. However after 3 treatments, 4T1 tumors eventually developed low oxygenation values and a necrotic tumor core over time with no vessel recovery into this area while Caki 1 tumors displayed the ability to overcome each treatment after 48 hours with almost full vascular recovery and oxygenation  In contrast macroscopic KHT sarcomas previously treated with a VDA (Combretastatin or OXi4503) were fo und to be equally susceptible to repeat treatments with these agents [11, 28, 36] Taken together these findings suggest that the extent of vascular damage and subsequent recovery following VDA treatment may va ry with tumor type. Clearly such factors need to be taken into consideration in the clinical application of these agents when used alone or in combination.
101 Sunitinib has been shown to completely inhibit angiogenesis when administered at the same time as tu mor cell injections in mice bearing window chambers  When dealing with existing tumor vasculature however, the effects of Sunitinib and other AIAs have been debated [110, 111] Rakesh Jain introduced the idea of vascular normalization in which tumor vessels revert towards normal phenotypic vessels following treatment with AIAs  Our results with Sunitinib treatme nt support this theory as we observed an apparent normalization of tumor vasculature with regard to microvessel density, blood flow and oxygenation. Small vessels were cropped as larger, more mature vessel structures were maintained to decrease microvessel density over time. Oxygenation in these vessels increased following initial treatment after which values were maintained. Correspondingly, the blood flow through these vessels was less obstructed as shown by a homogenization of faster BSTs in vessels (Fig ure 5 7 ). These findings would support the notion that AIAs might aid in the delivery of conventional therapy by improving blood flow conditions throughout the tumor. Similar to our results, Czabanka et al. observed faster blood flow through the vessels of SF126 human glioma in mouse dorsal window chambers following treatment with Sunitinib (40 mg/kg)  Using dynamic contrast enhanced magnetic resonance imaging, Hillman et al. also demonstrated an increased tumor perfusion at 20 mg/kg  However, it should be noted that one inherent limitation of the dorsal window chamber model is that assessments of tumor vascularity can be made only over a relatively short p eriod of time. During these times, our data suggest a normalization of vascular network following Sunitinib therapy. Similarly, a window of normalization has also been reported in some macroscopic tumor studies [52, 110, 113, 114] However, this window was transient and
102 reversal of the benefits to anti angiogenic therapy have also been noted [2, 110] Unfortunately it is not possible to utilize the dorsal window cham ber over extended periods of time to determine whether dynamic changes in vascular behavior occur with continued Sunitinib treatment. While both OXi4503 and Sunitinib individually significantly hindered the development of microvessels, the combined treatme nt with both agents resulted in the most effective means of controlling the formation and proliferation of these vessels. The characteristic vascular damage attributed to OXi4503 was complemented by the action of Sunitinib, which inhibited neovascularizati on. Mechanisms by which AIAs interfere with regrowth of the tumor rim following VDA treatment have been previously studied. Madlambayan et al showed that treatment of AML with the VDA OXi4503 resulted in a viable rim of tumor cells with increased VEGF A e xpression leading to angiogenesis back into tumor core. Using the AIA Bevacizumab in combination with OXi4503 however inhibited VEGF A expression and led to an increased antivascular response in accordance with our observations whereby vessel growth back i nto the tumor core was inhibited  Similarly, Taylor et al. showed that the use of Sunitinib after treatment with Combretastatin decreased the number of circulating endothelial progenitor cells that can pro mote tumor vascular development and regrowth, leading to enhanced antitumor and antivascular effects in comparison to either treatment alone  In our experiments, the combined treatment of OXi4503 plus Suni tinib led to necrosis in the tumor core with no functional vessels observed in the tumor interior. Although the IHC analysis showed the presence of vessels and endothelial cells in the tumor interior, perfusion of fluorescent liposomes during FPF imaging w as only
103 observed in normal vessels around the tumor periphery. Despite the lack of observed tumor perfusion, by day 7 there was evidence of tumor regrowth near the surrounding normal tissue. It is likely that this regrowth at the tumor rim is supported by normal tissue vessels that provide oxygen and nutrients necessary for the tumor cells to proliferate  This observation supports our belief that even combination VDAs plus AIAs treatments should still be ad ministered in conjunction with conventional therapies that can effectively eradicate any remaining tumor cells. However, such treatment strategies will require detailed knowledge of optimal timing of agents to maximize antitumor effects while minimizing po tential toxicities [58, 62] As a promising start, studies involving the use of Combretastatin, Bevacizumab and conventional chemotherapies have shown that the addition of Combretastatin to each therapy was wel l tolerated with enhanced tumor effects in comparison to treatments without the use of the VDA [56, 60] These observations clearly are encouraging. In summary, we have used the mouse dorsal skinfold window cha mber model and the combination of hyperspectral imaging of Hb saturation and FPF imaging of blood transit time to measure the longitudinal response of Caki 2 tumors to OXi4503 and Sunitinib used either alone or in combination. We demonstrated the superiori ty of the combined treatment over the use of either agent alone and in particular showed that the OXi4503 plus Sunitinib treatment destroyed existing tumor microvessels, inhibited blood vessel recovery and impaired Caki 2 tumor growth. Furthermore, we show ed that the oxygenation and blood flow behavior among tumor vessels was distinct for each treatment group, demonstrating the utility of our combination imaging methodology as a valuable tool for evaluating VTA treatments in vivo.
104 Figure 5 1 Image ac quisition and drug administration schedule. Imaging was performed first before any drug administration for that day. Hb saturation imaging was performed first followed by FPF imaging. Day 1 is designated as the day that tumors reached ~4 mm 3
105 Figure 5 2 Daily brightfield images of the tumor area show morphological changes to tumor vasculature in response to treatment Sunitinb (red) and OXi 4503 (blue) administration is indicated. Control tumor microvessels became more unorganized and tor tuous in structure with time. OXi4503 treated tumors had major vascular damage following treatment on day 1 with slight recovery by day 3. The treatments wer e less effective as seen on days 5 and 7 as the OXi4503 creates less vessel damage and the tumor vasculature recovers more aggressively. Sunitinib inhibited the formation of new vasculature while larger, more mature vessels were retained throughout treatme nt. There was an observable decrease in vessel density over time. Combination treatment tumors exhibited vessel destruction as a result of OXi4503 while Sunitinib inhibited recovery of new vasculature. This resulted in a complete absence of vessels in the tumor core.
106 Figure 5 3 Daily tumor volume measurements of tumor nodules in window chambers (median interquartile range) reveal enhanced growth inhibitio n with combination treatment. All treatment groups showed a tumor growth delay compared to the controls (p<.01) but the combination treatment significantly arrested the tumor volume in comparison to the OXi4503 (p<.0 1 ) and Sunitinib (p<.001) single agent groups
107 Figure 5 4 Immunohistologic analysis of tumor vasculature shows the inhibition of vascular development among treatment groups ( A ) Slices of tumors (5 m) harvested from window chambers were labele d with MECA 32 (red) for endothelial cells and DAPI (blue) for cell nuclei. Control vessels are much larger and non uniform in shape and size. OXi4503 treated vessel structures are also large but broken up among a lesser amount of cell nuclei. Sunitinib an d combination treated tumors have much smaller, uniform ve ssels. Images were taken at 20x (scalebar=140 m). ( B) MECA 32/DAPI ratio analysis of10 regions per tumor for control (n=5), OXi4503 (n=4), Sunitinib (n=4) and combination treatments (n=4). The medi an interquartile range is indicated. All three treatment groups had significantly less MECA 32 signal in comparison to controls (p < .001). While there was no significant difference between OXi4503 and combination treated tumors, both OXi4503 (p < .01) a nd combination (p < .001) treated tumors had greater ratios in comparison to Sunitinib.
108 Figure 5 5 Brightfield, Hb saturation, and BST maps for a control tumor reveals with unhindered growth blood supply time s become longer and oxygenation decreases. On day 1, the variation of BST is limited but as time goes on the heterogeneity of BST increases as unhindered growth of the vascular network allows complicated vessel pathways to form.
109 Figure 5 6 Brightfield, Hb saturation, and BST maps for an OXi4503 treated tumor show Caki 2 vasculature overcoming repeated treatment The first treatment resulted in vascular damage to the rim of the tumor, resulting in the formation of new, immature vessels from the periphery of the tumor. These vessels have faster BSTs with a higher oxygenation as seen in day 3. Because treatment on day 3 was less effective a larger vascular rim was l eft for vessels to begin recovery. Therefore we start to see variation in BSTs and oxygenation on day 5. Treatment on day 5 was again less effective and left an even larger viable rim of tumor vessels, thus vessels begin to resemble bloo d flow and oxygenat ion behavior of control tumor vessels as the tumor overcomes the treatment.
110 Figure 5 7 Brightfield, Hb saturation, and BST maps for a Sunitinib trea ted tumor illustrates vascular normalization for the duration of this experiment The vessels exhibited faster, more homogenous BSTs with a higher oxygenation on day 3. From this point on oxygenation values and BSTs were maintained throughout the experimen t.
11 1 Figure 5 8 Brightfield, Hb saturation, and BST maps for a combination treated tumor demonstrates the enhanced effects capabilities of the treatment Combination treatment le ft no tumor vessels to be analyzed. Surrounding normal vessels showed maintainence in BST and oxygenation until day 7 where we saw faster BSTs and slightly higher oxygenations which are indicators of new vessel formation.
112 Figure 5 9 ROI analysis of Hb saturation and BST maps for each experimental group. 30 ROIs from random vessels were chosen and corresponding Hb saturation BST values were plotted Control tumors showed a decrease in oxygenation and increas e in BST heterogeneity with longer BSTs OXi4503 treated tumors developed higher oxygenation and faster BST s Sunitinib treated tumors had an overall higher oxygenation and more homogeneous, faster BSTs in retained vessels. Combination treated tumors exhibited destruction of the vasculature in the tumor core with no vessel recovery, thus ROI values are represen tative of vessels located in the periphery of the tumor mass. These vessels resulted in faster BSTs and a higher corresponding oxygenation.
113 Figure 5 10 Boxplot analysis of Hb saturation and BST maps show that treated tumors inhibit the formation of lower oxygenation values and slower BSTs Horizontal red lines indicate the median of the data, the edges of the box are the 25 th and 75 th percentiles, and the whiskers indicate the extreme value that is not an outlier. Control groups saw a decrease in oxygenation with an increase in BST over time Both individual treatment groups halted this behavior by keeping oxygenation high and BST low. Combination treatment evaluation is representative of the remaining vessels following the initial treatment.
114 CHAPTER 6 CONCLUSION AN D FUTURE DIRECTION We have developed a novel combination of FPF imaging of blood transit time with hyperspectral imaging of Hb saturation for the observation of microvessel development blood flow and oxygenation in the murine window chamber model over tim e. Hb saturation imaging revealed oxygenation of the blood within microvessels while FPF imaging complemented this data and provided the blood flow direction and BST of vessels in the window. By determining the BST of vessels we were able to correlate the oxygenation of a vessel to the time it takes to perfuse the vessel from a common artery. In normal tissue with properly functioning vessels, we showed that in general, the longer it took for the blood to reach a vessel, the less oxygenated the vessel was relative to the initial supplying artery We demonstrated the utility of this method to analyze microvasculature of different vascular conditions over time including normal vasculature, wound healing and tum or microvasculature. By observing blood flow and oxygenation in normal vasculature we were able to demonstrate the dynamic changes in hemodynamics that can occur in the absence of any pathology. The demonstration of wound healing as a tightly regulated process was compared with the initial formation of tumor vasculature. The rapid formation of vessels in both wound healing and tumor development result ed in faster BSTs and higher oxygenation as these vessels act ed to provide nutrients and oxy genation for healing and growth. H owever once a wound is healed this vascular growth cease d and BSTs and oxygenation values equilibrate d whereas in a tumor the vessels continue d to grow and proliferate with more complicated vascular networks and varied ranges of BSTs and oxygenation
115 Our continued analysis of tumor microvasculature and the response to VTAs suggested the use of VDAs and AIAs in combination to treat tumor vasculature has enhanced antivascular and antitumor effects in comparison to either treatment alone. We demonstrated that the combination of the VDA OXi4503 and AIA Sunitinib significantly delayed tumor growth in comparison to either treatment alone. With regard to vascular effects, while each treatment individually caused stark changes to the blood flow and oxygenation of tumor vasculature, these effe cts only slowed tumor growth and vessel development. In contrast, t he combination treatment completely eradicated functional vessels within the tumor volume, with no regrowth within the duration of the experiment. The treatment with OXi4503 destroyed exist ing vasculature, while Sunitinib inhibited the formation of any new vasculature that would normally recover from OXi4503 treatment alone. This study advocates for the future exploration into using VDAs and AIAs in combination for better therapeutic outcome s in the clinic. To offer a better understanding of the use of VDAs and AIAs in combination, future testing using these drugs under different experimental parameters should be considered. For example, future experiment s could include a longer experimental duration. In our analysis, we only analyzed the treatment over 7 days. Because there are questions regarding the ability of tumors to overcome both VDAs and AIAs over time, a longer analysis would allow for the question of resistance to be better answered. One issue with extending experimental duration arises with the viability of the window chamber following treatment. VDAs such as OXi4503 create much vascular damage in the tumor mass, generating an immune response and releasing blood and other fluids
116 into the window. This can inhibit the viability of the window chamber thereby shortening the duration of the experiment. Different drugs/doses and tumor types should also be used in future experiments for better characterization of VDAs and AIAs used in combin ation With regard to our experiments, we utilized Sunitinib which is already used in the clinic for the treatment of renal cell carcinoma (RCC). Knowing this, the combination treatment effect on Caki 2 RCC tumor vasculature that we observed may be optimal in our experimentation. In addition, both OXi4503 and Sunitinib treatments alone can produce varying vascular responses among different tumor types. Thus the use of different drug combinations as well as testing on different tumors has the potent ial to pr oduce diverse results that could reveal important implications for future application in the clinic. Lastly, future experimentation should include the use of the combination treatment of VTAs with conventional treatment used in the clinic such as chemother apy and radiotherapy. It is well known that many VTAs on their own have shown better clinical outcome when used in combination with conventional therapies. In our own analysis we observed combination treated tumors beginning to proliferate outward into the normal tissue despite treatment, therefore providing motiv ation to evaluate the use of combination VTA treatment s with conventi onal treatment s to see if tumor regression can be achieved. The combination of FPF and Hb saturation imaging has many advantages and disadvantages to consider for future experiments. The cost and ease of use of this combination imaging is an important advantage, especially considering the high spatial resolution at a microvessel level that can be attained. The use of the window cha mber
117 model and multiple near IR fluorescent liposomes allows for repeated measurements over time, an advantage displayed with our evaluation of tumor vascular response to VTAs. In addition, the ability to harvest window chamber tumors also allows for the h istological analysis of tumors along with real time evaluatio n via imaging. Inherent disadvantages with the window chamber model limit the tumor size and types of tumors that can be used. An o ptical consideration to be made moving forward includes the use of different camera s and objective equipment to widen the field of view while keeping the resolution. Currently we have employed the use of a 2.5x objective resulting in a 3.1mm x 3.1mm field of view, limiting our analysis. A larger field of view would be ben eficial to imaging the whole tumor area as well as surrounding normal tissue during experiments Finally, the use of the same camera to perform both FPF imaging and Hb saturation imaging could allow researchers to easily compare BST and Hb saturation ma ps on a pixel by pixel basis.
118 APPENDIX MATLAB CODE FOR FPF IMAGING ANALYSIS %%%% Complete BST Analysis. %%% Jennifer A. Lee %%% Code to create BST Maps from TIFF image stack %%%% Truncation and flatfielding of original image stack. The original TIFF %% image stack is truncated to the frames at which dye first enters the %% field of view and the frame at which all vessels are completely filled. %% The images are flatfielded with background frames before the dye has %% reached the field of view and saved into the variable 'new_bst_video'. %% 'User Input' denotes parts of the code where the user has to input %% numbers or file names before running the code. % Input file name of image stack file_input = input( 'What is the original filename?' 's' ); % User Input Number of frames for final, truncated image stack number_of_frames = 640; % Find the rows (a) and columns (b) of original image a=501; b=502; siz=[a,b]; % User Change Last frame at which time all vessels have been filled % with fluor escent tracer number_of_frames1 = 700; % Creating new_bst_video to hold flatfielded frames for the desired number % of frames new_bst_video=zeros(a,b,number_of_frames); % User Input Creation of average background frame for flatfielding. 10 % fr ames are chosen from before dye enters the field of view. Each frame % is read and stored into 'background'. For each pixel, the median is % determined and stored into 'new' for use in flatfielding. i=1; for back=20:30 background(:,:,i)=double(imre ad(file_input,back)); i=i+1; end for i=1:a for j = 1:b new(i,j) = median(background(i,j,:)); end end % User Input Creation of truncated image stack w/ beginning cut off. % 'm' determines the frame number for the original tiff stack, 'mm' % determines the frame count for the new, flatfielded video % (new_bst_video).
119 for m=60:number_of_frames1 mm=m 59; image = double(imread(file_input,m)); %read in tif image image2=image(1:a,1:b); image2=image2./(new./mean2(n ew)); new_bst_video(:,:,mm)=image2; end % User Input Save new_bst_video as filename % '(mouse #)_(imaged day)_(number of frames).mat' save( 'notumor_d1_640frames.mat' 'new_bst_video' ); %%%% Moving Filter to subtract effects of background and ne ighboring %% fluorescence. For each pixel, subtract median of surrounding background %% pixels in 25x25 box. % Define 'masked_data' as all data in vessel area. Define 'background_data' % as all data in pixels of the background. for i=1:number_of_frame s image=new_bst_video(:,:,i); masked_data(:,i)=image(mask_ones); background_data(:,i)=image(mask_zeros); end % Creation of variables to analyze surrounding background pixels [mask_size z]=size(mask_ones); [row0 column0]=ind2sub(siz,mask_zer os); [row1 column1]=ind2sub(siz,mask_ones); % Creation of variable to store background subtracted data % (pixel_submedian) pixel_submedian=zeros(mask_size, number_of_frames); for i=1:mask_size clear final_range_index clear final_range clear back_sum_one clear position clear logical_position range_row=(row1(i) 25):1:(row1(i)+25); range_col=(column1(i) 25):1:(column1(i)+25); whole_range=[sort(repmat(range_row(:),length(range_col),1)),repmat(range_col( :),length(range_row), 1)]; final_row=whole_range(:,1)'; final_col=whole_range(:,2)'; row_summary=whole_range(:,1)>=1 & whole_range(:,1)<502; col_summary=whole_range(:,2)>=1 & whole_range(:,2)<503; back_summary = row_summary&col_summary; back_sum_on e=find(back_summary==1);
120 final_range(1,:)=final_row(back_sum_one); final_range(2,:)=final_col(back_sum_one); final_range_index=sub2ind(siz,final_range(1,:), final_range(2,:)); logical_position=ismember(mask_zeros,final_range_index); position=find(logical_position==1); median_values=median(background_data(position(:),:)); pixel_submedian(i,:)=masked_data(i,:) median_values; end % Save background subtracted image data (pixel_submedian) as % 'pixel_submed.mat' save( 'pixel_submed.mat' 'pixel_submedian' ) %%%% Video Smoothing. Smooth data for each pixel over time to determine %% BST. In house command 'smooth' was used from MATLAB The maximum and %% minimum intensity was determined for each pixel. The average intensi ty %% for each pixel was calculated and saved as a threshold for that pixel %% ('th_50_submed'). The final smoothed image data is saved as %% 'sm_vid_submed' sm_vid_submed=zeros(mask_size,number_of_frames); for ii = 1:mask_size sm_vid_submed(ii,:)= smooth(pixel_submedian(ii,1:number_of_frames),.2, 'rloess' ); end max_intensity=max(sm_vid_submed,,2); min_intensity=min(sm_vid_submed,,2); th_50_submed=(max_intensity(:)+min_intensity(:))/2; save( 'sm_vid_submed.mat' 'sm_vid_submed' ) save( 'th_50_submed .mat' 'th_50_submed' ) %%%% BST Map Creation. For each pixel the frame at which it crosses the %% average threshold ('th_50_submed') is determined and marked as the BST %% for that pixel. These values are assigned to the pixel and remapped %% and assigned a predetermined colormap (colormap reverse_jet). % User Input User inputs the total exposure time for the images. time_interval=.117; final_bst_50=zeros(siz); for i=1:mask_size clear lessthan clear pos_slope clear marker clear ma rker2
121 clear marker3 data=sm_vid_submed(i,1:(number_of_frames 1)); data2=sm_vid_submed(i,2:number_of_frames); lessthan=data
122 range_row=(row1(i) 2):1:(row1(i)+ 2); range_col=(column1(i) 2):1:(column1(i)+ 2); whole_range=[sort(repmat(range_row(:),length(range_col),1)),repmat(range_col( :),length(range_row),1)]; final_row=whole_range(:,1)'; final_col=whole_ran ge(:,2)'; row_summary=whole_range(:,1)>=1 & whole_range(:,1)<502; col_summary=whole_range(:,2)>=1 & whole_range(:,2)<503; back_summary = row_summary&col_summary; back_sum_one=find(back_summary==1); final_range(1,:)=final_row(back _sum_one); final_range(2,:)=final_col(back_sum_one); final_range_index=sub2ind(siz,final_range(1,:), final_range(2,:)); logical_position=ismember(mask_ones,final_range_index); position=find(logical_position==1); values=masked_bst(po sition(:),:); value_positions=find(values>=0); median_values(i)=median(values(value_positions)); filtered_bst(i)=median_values(i); end filt_bst_image_50=zeros(siz); filt_bst_image_50(mask_ones)=filtered_bst; % Display final BST map. figure(2) imagesc(filt_bst_image_50) axis image set(gca, 'visible' 'off' ) colormap(reverse_jet)
123 LIST OF REFERENCES  Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;407:249 57.  Jain RK. N ormalization of tumor vasculature: An emerging concept in antiangiogenic therapy. Science 2005;307:58 62.  Palmer GM, Fontanella AN, Shan S, Hanna G, Zhang G, Fraser CL, et al. In vivo optical molecular imaging and analysis in mice using dorsal window chamber models applied to hypoxia, vasculature and fluorescent reporters. Nat Protoc 2011;6:1355 66.  Algire GH. An adaptation of the transparent chamber technique to the mouse. J Natl Cancer Inst 1943;4:1 11.  Lunt SJ, Gray C, Reyes Aldasoro CC Matcher SJ, Tozer GM. Application of intravital microscopy in studies of tumor microcirculation. J Biomed Opt 2010;15:011113.  Baron VT, Welsh J, Abedinpour P, Borgstrom P. Intravital microscopy in the mouse dorsal chamber model for the study of sol id tumors. Am J Cancer Res 2011;1:674 86.  Moy AJ, White SM, Indrawan ES, Lotfi J, Nudelman MJ, Costantini SJ, et al. Wide field functional imaging of blood flow and hemoglobin oxygen saturation in the rodent dorsal window chamber. Microvasc Res 2011; 82:199 209.  Sorg BS, Moeller BJ, Donovan O, Cao YT, Dewhirst MW. Hyperspectral imaging of hemoglobin saturation in tumor microvasculature and tumor hypoxia development. J Biomed Opt 2005;10.  Reitan NK, Thuen M, Goa PE, de Lange DC. Characteriza tion of tumor microvascular structure and permeability: comparison between magnetic resonance imaging and intravital confocal imaging. J Biomed Opt 2010;15:036004.  Folkman J, Bach M, Rowe JW, Davidoff F, Lambert P, Hirsch C, et al. Tumor Angiogenesi s Therapeutic Implications. New Engl J Med 1971;285:1182 6.  Siemann DW. The unique characteristics of tumor vasculature and preclinical evidence for its selective disruption by Tumor Vascular Disrupting Agents. Cancer Treat Rev 2011;37:63 74. [12 ] Wankhede M, Agarwal N, Fraga Silva RA, deDeugd C, Raizada MK, Oh SP, et al. Spectral imaging reveals microvessel physiology and function from anastomoses to thromboses. J Biomed Opt 2010;15.
124  Yuan F, Chen Y, Dellian M, Safabakhsh N, Ferrara N, Jai n RK. Time dependent vascular regression and permeability changes in established human tumor xenografts induced by an anti vascular endothelial growth factor/vascular permeability factor antibody. Proc Natl Acad Sci U S A 1996;93:14765 70.  Tozer GM, Akerman S, Cross NA, Barber PR, Bjorndahl MA, Greco O, et al. Blood vessel maturation and response to vascular disrupting therapy in single vascular endothelial growth factor A isoform producing tumors. Cancer Res 2008;68:2301 11.  Tozer GM, Prise V E, Wilson J, Cemazar M, Shan S, Dewhirst MW, et al. Mechanisms associated with tumor vascular shut down induced by combretastatin A 4 phosphate: intravital microscopy and measurement of vascular permeability. Cancer Res 2001;61:6413 22.  Oye KS, Gula ti G, Graff BA, Gaustad JV, Brurberg KG, Rofstad EK. A novel method for mapping the heterogeneity in blood supply to normal and malignant tissues in the mouse dorsal window chamber. Microvasc Res 2008;75:179 87.  Carmeliet P. Manipulating angiogenesi s in medicine. J Intern Med 2004;255:538 61.  Goel S, Duda DG, Xu L, Munn LL, Boucher Y, Fukumura D, et al. Normalization of the Vasculature for Treatment of Cancer and Other Diseases. Physiol Rev 2011;91:1071 121.  De BK, Cauwenberghs S, Carme liet P. Vessel abnormalization: another hallmark of cancer? Molecular mechanisms and therapeutic implications. Curr Opin Genet Dev 2011;21:73 9.  Nagy JA, Chang SH, Dvorak AM, Dvorak HF. Why are tumour blood vessels abnormal and why is it important t o know? Brit J Cancer 2009;100:865 9.  Mohindra JK, Rauth AM. Increased cell killing by metronidazole and nitrofurazone of hypoxic compared to aerobic mammalian cells. Cancer Res 1976;36:930 6.  Koch S, Mayer F, Honecker F, Schittenhelm M, Boke meyer C. Efficacy of cytotoxic agents used in the treatment of testicular germ cell tumours under normoxic and hypoxic conditions in vitro. Br J Cancer 2003;89:2133 9.  Denekamp J. Endothelial cell proliferation as a novel approach to targeting tumou r therapy. Br J Cancer 1982;45:136 9.  Denekamp J. Vascular endothelium as the vulnerable element in tumours. Acta Radiol Oncol 1984;23:217 25.
125  Chaplin DJ, Pettit GR, Hill SA. Anti vascular approaches to solid tumour therapy: evaluation of com bretastatin A4 phosphate. Anticancer Res 1999;19:189 95.  Kanthou C, Tozer GM. The tumor vascular targeting agent combretastatin A 4 phosphate induces reorganization of the actin cytoskeleton and early membrane blebbing in human endothelial cells. Bl ood 2002;99:2060 9.  Hua JY, Sheng YZ, Pinney KG, Garner CM, Kane RR, Prezioso JA, et al. Oxi4503, a novel vascular targeting agent: Effects on blood flow and antitumor activity in comparison to combretastatin A 4 phosphate. Anticancer Res 2003;23:14 33 40.  Sheng YZ, Hua JY, Pinney KG, Garner CM, Kane RR, Prezioso JA, et al. Combretastatin family member OXI4503 induces tumor vascular collapse through the induction of endothelial apoptosis. Int J Cancer 2004;111:604 10.  Madlambayan GJ, Mea cham AM, Hosaka K, Mir S, Jorgensen M, Scott EW, et al. Leukemia regression by vascular disruption and antiangiogenic therapy. Blood 2010;116:1539 47.  Wankhede M, deDeugd C, Siemann DW, Sorg BS. In vivo functional differences in microvascular respon se of 4T1 and Caki 1 tumors after treatment with OXi4503. Oncol Rep 2010;23:685 92.  Tozer GM, Prise VE, Wilson J, Locke RJ, Vojnovic B, Stratford MR, et al. Combretastatin A 4 phosphate as a tumor vascular targeting agent: early effects in tumors an d normal tissues. Cancer Res 1999;59:1626 34.  Chan LS, Malcontenti Wilson C, Muralidharan V, Christophi C. Alterations in vascular architecture and permeability following OXi4503 treatment. Anti Cancer Drugs 2008;19:17 22.  Siemann DW, Rojiani AM. The vascular disrupting agent ZD6126 shows increased antitumor efficacy and enhanced radiation response in large, advanced tumors. Int J Radiat Oncol 2005;62:846 53.  Siemann DW, Lepler S, Pampo C, Rojiani AM. The novel vascular targeting agent ZD6126 shows enhanced anti tumour efficacy in large, bulky tumours. Eur J Cancer 2002;38:S40.  Hill SA, Tozer GM, Pettit GR, Chaplin DJ. Preclinical evaluation of the antitumour activity of the novel vascular targeting agent oxi 4503. Anticancer Res 2002;22:1453 8.  Salmon HW, Siemann DW. Effect of the second generation vascular disrupting agent OXi4503 on tumor vascularity. Clin Cancer Res 2006;12:4090 4.
126  Patterson DM, Zweifel M, Middleton MR, Price PM, Folkes LK, Stratford MRL, et al. P hase I Clinical and Pharmacokinetic Evaluation of the Vascular Disrupting Agent OXi4503 in Patients with Advanced Solid Tumors. Clin Cancer Res 2012;18:1415 25.  Horsman MR, Siemann DW. Pathophysiologic effects of vascular targeting agents and the im plications for combination with conventional therapies. Cancer Res 2006;66:11520 39.  Rice L, Pampo C, Lepler S, Rojiani AM, Siemann DW. Support of a free radical mechanism for enhanced antitumor efficacy of the microtubule disruptor OXi4503. Microva sc Res 2011;81:44 51.  Ljuslinder I, Melin B, Henriksson ML, Oberg A, Palmqvist R. Increased epidermal growth factor receptor expression at the invasive margin is a negative prognostic factor in colorectal cancer. Int J Cancer 2011;128:2031 7.  Matsuda Y, Ishiwata T, Yamahatsu K, Kawahara K, Hagio M, Peng WX, et al. Overexpressed fibroblast growth factor receptor 2 in the invasive front of colorectal cancer: A potential therapeutic target in colorectal cancer. Cancer Lett 2011;309:209 19.  Nguyen L, Fifis T, Malcontenti Wilson C, Chan LS, Costa PLN, Nikfarjam M, et al. Spatial morphological and molecular differences within solid tumors may contribute to the failure of vascular disruptive agent treatments. Bmc Cancer 2012;12.  Ferrara N. VEGF and the quest for tumour angiogenesis factors. Nat Rev Cancer 2002;2:795 803.  Ferrara N. Vascular endothelial growth factor: Basic science and clinical progress. Endocr Rev 2004;25:581 611.  Kim KJ, Li B, Winer J, Armanini M, Gillett N Phillips HS, et al. Inhibition of Vascular Endothelial Growth Factor Induced Angiogenesis Suppresses Tumor Growth Invivo. Nature 1993;362:841 4.  Warren RS, Yuan H, Matli MR, Gillett NA, Ferrara N. Regulation by Vascular Endothelial Growth Factor o f Human Colon Cancer Tumorigenesis in A Mouse Model of Experimental Liver Metastasis. J Clin Invest 1995;95:1789 97.  Morabito A, De Maio E, Di Maio M, Normanno N, Perrone F. Tyrosine kinase inhibitors of vascular endothelial growth factor receptors in clinical trials: Current status and future directions. Oncologist 2006;11:753 64.  Christensen JG. A preclinical review of sunitinib, a multitargeted receptor tyrosine kinase inhibitor with anti angiogenic and antitumour activities. Ann Oncol 2007 ;18:3 10.
127  Le TC, Raymond E, Faivre S. Sunitinib: a novel tyrosine kinase inhibitor. A brief review of its therapeutic potential in the treatment of renal carcinoma and gastrointestinal stromal tumors (GIST). Ther Clin Risk Manag 2007;3:341 8.  Faivre S, Demetri G, Sargent W, Raymond E. Molecular basis for sunitinib efficacy and future clinical development. Nat Rev Drug Discov 2007;6:734 45.  Hillman GG, Singh Gupta V, Zhang H, Al Bashir AK, Katkuri Y, Li M, et al. Dynamic Contrast Enhance d Magnetic Resonance Imaging of Vascular Changes Induced by Sunitinib in Papillary Renal Cell Carcinoma Xenograft Tumors. Neoplasia 2009;11:910 20.  Ullrich RT, Jikeli JF, Diedenhofen M, Bohm Sturm P, Unruh M, Vollmar S, et al. In Vivo Visualization of Tumor Microvessel Density and Response to Anti Angiogenic Treatment by High Resolution MRI in Mice. Plos One 2011;6.  Czabanka M, Erber R, Vinci M, Ullrich A, Vajkoczy P. Microhemodynamic consequences of tumor angiogenesis targeting using the tyro sine kinase inhibitor SU11248. J Vasc Res 2006;43:564.  Osusky KL, Hallahan DE, Fu A, Ye F, Shyr Y, Geng L. The receptor tyrosine kinase inhibitor SU11248 impedes endothelial cell migration, tubule formation, and blood vessel formation in vivo, but h as little effect on existing tumor vessels. Angiogenesis 2004;7:225 33.  Shojaei F. Anti angiogenesis therapy in cancer: current challenges and future perspectives. Cancer Lett 2012;320:130 7.  Garcia VM, Basu B, Molife LR, Kaye SB. Combining A ntiangiogenics to Overcome Resistance: Rationale and Clinical Experience. Clin Cancer Res 2012;18:3750 61.  Jayson GC, Hicklin DJ, Ellis LM. Antiangiogenic therapy -evolving view based on clinical trial results. Nat Rev Clin Oncol 2012;9:297 303. [5 8] Siemann DW, Shi W. Dual agent targeting of the tumor vasculature: Combining avastin with CA4P or OXi4503. Clin Cancer Res 2005;11:8968S.  Siemann DW, Shi WY. Antivascular combination therapy, using the antiangiogenic agent ZD6474 and the vascular targeting agent ZD6126 in human tumor xenograft models. Clin Cancer Res 2003;9:6140S 1S.  Nathan P, Zweifel M, Padhani AR, Koh DM, Ng M, Collins DJ, et al. Phase I Trial of Combretastatin A4 Phosphate (CA4P) in Combination with Bevacizumab in Patien ts with Advanced Cancer. Clin Cancer Res 2012;18:3428 39.
128  Koh DM, Blackledge M, Collins DJ, Padhani AR, Wallace T, Wilton B, et al. Reproducibility and changes in the apparent diffusion coefficients of solid tumours treated with combretastatin A4 ph osphate and bevacizumab in a two centre phase I clinical trial. Eur Radiol 2009;19:2728 38.  Shi WY, Siemann DW. Targeting the tumor vasculature: Enhancing antitumor efficacy through combination treatment with ZD6126 and ZD6474. In Vivo 2005;19:1045 50.  Chen F, Feng YM, Zheng KE, De Keyzer F, Li JJ, Feng YB, et al. Enhanced Antitumor Efficacy of a Vascular Disrupting Agent Combined with an Antiangiogenic in a Rat Liver Tumor Model Evaluated by Multiparametric MRI. Plos One 2012;7.  Taylor M, Billiot F, Marty V, Rouffiac V, Cohen P, Tournay E, et al. Reversing resistance to vascular disrupting agents by blocking late mobilization of circulating endothelial progenitor cells. Cancer Discov 2012;2:434 49.  Koehl GE, Gaumann A, Geissler E K. Intravital microscopy of tumor angiogenesis and regression in the dorsal skin fold chamber: mechanistic insights and preclinical testing of therapeutic strategies. Clin Exp Metastasis 2009;26:329 44.  Fukumura D, Duda DG, Munn LL, Jain RK. Tumor m icrovasculature and microenvironment: novel insights through intravital imaging in pre clinical models. Microcirculation 2010;17:206 25.  Anderson HL, Yap JT, Miller MP, Robbins A, Jones T, Price PM. Assessment of pharmacodynamic vascular response in a phase I trial of combretastatin A4 phosphate. J Clin Oncol 2003;21:2823 30.  Tai JH, Tessier J, Ryan AJ, Hoffman L, Chen X, Lee TY. Assessment of acute antivascular effects of vandetanib with high resolution dynamic contrast enhanced computed tomo graphic imaging in a human colon tumor xenograft model in the nude rat. Neoplasia 2010;12:697 707.  Nielsen T, Wittenborn T, Horsman MR. Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE MRI) in Preclinical Studies of Antivascular Treatments. 4 ed. Pharmaceutics: 2012. p. 563 89.  Gaustad JV, Brurberg KG, Simonsen TG, Mollatt CS, Rofstad EK. Tumor vascularity assessed by magnetic resonance imaging and intravital microscopy imaging. Neoplasia 2008;10:354 62.  Dufort S, Sancey L, Wen k C, Josserand V, Coll JL. Optical small animal imaging in the drug discovery process. BBA Biomembranes 2010;1798:2266 73.
129  McDonald DM, Choyke PL. Imaging of angiogenesis: from microscope to clinic. Nat Med 2003;9:713 25.  Dreher MR, Liu WG, M ichelich CR, Dewhirst MW, Yuan F, Chilkoti A. Tumor vascular permeability, accumulation, and penetration of macromolecular drug carriers. J Natl Cancer Inst 2006;98:335 44.  Skala MC, Fontanella A, Hendargo H, Dewhirst MW, Izatt JA. Combined hyperspe ctral and spectral domain optical coherence tomography microscope for noninvasive hemodynamic imaging. Opt Lett 2009;34:289 91.  Sorg BS, Hardee ME, Agarwal N, Moeller BJ, Dewhirst MW. Spectral imaging facilitates visualization and measurements of un stable and abnormal microvascular oxygen transport in tumors. J Biomed Opt 2008;13.  Salmon HW, Mladinich C, Siemann DW. Evaluations of vascular disrupting agents CA4P and OXi4503 in renal cell carcinoma (Caki 1) using a silicon based microvascular c asting technique. Eur J Cancer 2006;42:3073 8.  Brurberg KG, Graff BA, Rofstad EK. Temporal heterogeneity in oxygen tension in human melanoma xenografts. Brit J Cancer 2003;89:350 6.  Gaustad JV, Simonsen TG, Brurberg KG, Huuse EM, Rofstad EK. Blood Supply in Melanoma Xenografts Is Governed by the Morphology of the Supplying Arteries. Neoplasia 2009;11:277 85.  Bruns CJ, Koehl GE, Guba M, Yezhelyev M, Steinbauer M, Seeliger H, et al. Rapamycin induced endothelial cell death and tumor vesse l thrombosis potentiate cytotoxic therapy against pancreatic cancer. Clin Cancer Res 2004;10:2109 19.  Guba M, Yezhelyev M, Eichhorn ME, Schmid G, Ischenko I, Papyan A, et al. Rapamycin induces tumor specific thrombosis via tissue factor in the prese nce of VEGF. Blood 2005;105:4463 9.  Berry LR, Barck KH, Go MA, Ross J, Wu XM, Williams SP, et al. Quantification of viable tumor microvascular characteristics by multispectral analysis. Magn Reson Med 2008;60:64 72.  Cao YT, Li CY, Moeller BJ, Yu DH, Zhao YL, Dreher MR, et al. Observation of incipient tumor angiogenesis that is independent of hypoxia and hypoxia inducible factor 1 activation. Cancer Res 2005;65:5498 505.  Vakoc BJ, Lanning RM, Tyrrell JA, Padera TP, Bartlett LA, Stylianop oulos T, et al. Three dimensional microscopy of the tumor microenvironment in vivo using optical frequency domain imaging. Nat Med 2009;15:1219 U151.
130  Dedeugd C, Wankhede M, Sorg BS. Multimodal optical imaging of microvessel network convective oxygen transport dynamics. Appl Opt 2009;48:D187 D197.  Shonat RD, Wachman ES, Niu WH, Koretsky AP, Farkas DL. Near simultaneous hemoglobin saturation and oxygen tension maps in mouse brain using an AOTF microscope. Biophys J 1997;73:1223 31.  Awasth i VD, Garcia D, Goins BA, Phillips WT. Circulation and biodistribution profiles of long circulating PEG liposomes of various sizes in rabbits. Int J Pharm 2003;253:121 32.  Bangham AD, Standish MM, Watkins JC. Diffusion of Univalent Ions Across Lamel lae of Swollen Phospholipids. J Mol Biol 1965;13:238 &.  Thorpe PE. Vascular targeting agents as cancer therapeutics. Clin Cancer Res 2004;10:415 27.  Vaupel P, Kallinowski F, Okunieff P. Blood flow, oxygen and nutrient supply, and metabolic mi croenvironment of human tumors: a review. Cancer Res 1989;49:6449 65.  Endrich B, Intaglietta M, Reinhold HS, Gross JF. Hemodynamic characteristics in microcirculatory blood channels during early tumor growth. Cancer Res 1979;39:17 23.  Park SO Wankhede M, Lee YJ, Choi EJ, Fliess N, Choe SW, et al. Real time imaging of de novo arteriovenous malformation in a mouse model of hereditary hemorrhagic telangiectasia. Journal of Clinical Investigation 2009;119:3487 96.  Jain RK. Determinants of tumor blood flow: a review. Cancer Res 1988;48:2641 58.  Kallinowski F, Schlenger KH, Runkel S, Kloes M, Stohrer M, Okunieff P, et al. Blood flow, metabolism, cellular microenvironment, and growth rate of human tumor xenografts. Cancer Res 1989;49:37 59 64.  Tozer GM, Ameer Beg SM, Baker J, Barber PR, Hill SA, Hodgkiss RJ, et al. Intravital imaging of tumour vascular networks using multi photon fluorescence microscopy. Adv Drug Deliv Rev 2005;57:135 52.  Konerding MA, Malkusch W, Klapthor B van AC, Fait E, Hill SA, et al. Evidence for characteristic vascular patterns in solid tumours: quantitative studies using corrosion casts. Br J Cancer 1999;80:724 32.  Sullivan GW, Sarembock IJ, Linden J. The role of inflammation in vascular disea ses. J Leukoc Biol 2000;67:591 602.
131  Rege A, Thakor NV, Rhie K, Pathak AP. In vivo laser speckle imaging reveals microvascular remodeling and hemodynamic changes during wound healing angiogenesis. Angiogenesis 2012;15:87 98.  Chung AS, Lee J, F errara N. Targeting the tumour vasculature: insights from physiological angiogenesis. Nat Rev Cancer 2010;10:505 14.  Dvorak HF. Tumors: wounds that do not heal. Similarities between tumor stroma generation and wound healing. N Engl J Med 1986;315:16 50 9.  Battegay EJ. Angiogenesis: mechanistic insights, neovascular diseases, and therapeutic prospects. J Mol Med (Berl) 1995;73:333 46.  Holash J, Maisonpierre PC, Compton D, Boland P, Alexander CR, Zagzag D, et al. Vessel cooption, regress ion, and growth in tumors mediated by angiopoietins and VEGF. Science 1999;284:1994 8.  Lin PC. Optical imaging and tumor angiogenesis. J Cell Biochem 2003;90:484 91.  Vajkoczy P, Schilling L, Ullrich A, Schmiedek P, Menger MD. Characterizati on of angiogenesis and microcirculation of high grade glioma: an intravital multifluorescence microscopic approach in the athymic nude mouse. J Cereb Blood Flow Metab 1998;18:510 20.  Siemann DW, Bibby MC, Dark GG, Dicker AP, Eskens FALM, Horsman MR et al. Differentiation and definition of vascular targeted therapies. Clin Cancer Res 2005;11:416 20.  Tozer GM, Kanthou C, Lewis G, Prise VE, Vojnovic B, Hill SA. Tumour vascular disrupting agents: combating treatment resistance. Brit J Radiol 20 08;81:S12 S20.  Lee JA, Kozikowski RT, Sorg BS. Combination of spectral and fluorescence imaging microscopy for wide field in vivo analysis of microvessel blood supply and oxygenation. Opt Lett 2013;38:332 4.  Siemann DW, Shi WY. Efficacy of combined antiangiogenic and vascular disrupting agents in treatment of solid tumors. Int J Radiat Oncol 2004;60:1233 40.  Siemann DW, Shi WY. Dual targeting of tumor vasculature: Combining avastin and vascular disrupting agents (CA4P or OXi4503). An ticancer Res 2008;28:2027 31.  Malcontenti Wilson C, Chan L, Nikfarjam M, Muralidharan V, Christophi C. Vascular targeting agent Oxi4503 inhibits tumor growth in a colorectal liver metastases model. J Gastroen Hepatol 2008;23:E96 E104.
132  Winkl er F, Kozin SV, Tong RT, Chae SS, Booth MF, Garkavtsev I, et al. Kinetics of vascular normalization by VEGFR2 blockade governs brain tumor response to radiation: Role of oxygenation, angiopoietin 1, and matrix metal loproteinases. Cancer Cell 2004;6:553 63  Gaustad JV, Simonsen TG, Leinaas MN, Rofstad EK. Sunitinib treatment does not improve blood supply but induces hypoxia in human melanoma xenografts. Bmc Cancer 2012;12.  Czabanka M, Vinci M, Heppner F, Ullrich A, Vajkoczy P. Effects of sunitinib on tumor hemodynamics and delivery of chemotherapy. Int J Cancer 2009;124:1293 300.  Yang AD, Bauer TW, Camp ER, Somcio R, Liu WB, Fan F, et al. Improving delivery o f antineoplastic agents with anti vascular endothelial growth factor therapy. Cancer 2005;103:1561 70.  Matsumoto S, Batra S, Saito K, Yasui H, Choudhuri R, Gadisetti C, et al. Antiangiogenic Agent Sunitinib Transiently Increases Tumor Oxygenation a nd Suppresses Cycling Hypoxia. Cancer Research 2011;71:6350 9.
133 BIOGRAPHICAL SKETCH Jennifer Amy Lee was born in Monroe, Louisiana in 1986. She grew up in Erie, Pennsylv ania and graduated with honors from Fairview High School in 2004. Jennifer obtained her B.S. in a pplied s ciences with a concentration in b iomedical e ngineering and minors in p hysics and c hemistry from the University of North Carolina at Chapel Hi l l in 2008. She continued her postgraduate work at the University of Florida in the J. Crayto n Pruitt Family Department of Biomedi cal Engineering. Under the tutelage of Dr. Brian S. Sorg in the Biophotonics Imaging, Therapeutics, and Sensing Laboratory (BITS Lab) she developed the novel combination imaging technique of first pass fluorescence imag ing and hyperspectral imaging of hemoglobin saturation for the analysis of tumor microvasculature and the response to vascular targeting agents. After completion of her Ph D program in August 201 3 she will continue research as a p ostdoctoral f ellow in the Department of Radiation Oncology under Dr. Dietmar W. Siemann.