the dynamics and modification of...

Imaging the tumor microenvironment

THE DYNAMICS ANDMODIFICATION OF HYPOXIA

Anna Ljungkvist

UMEÅ UNIVERSITY MEDICAL DISSERTATIONSNew Series No. 835- ISSN 0346-6612-ISBN 91-7305-432-1

Department of Radiation Sciences, OncologyClinical Bioemdical Research, Pathology

Imaging the tumor microenvironment:

The dynamics and modification of hypoxia

Anna Ljungkvist

Umeå 2003

Cover: Milstolpe, milstone, mijlpaal, at road 1078, Nymölla, Sweden.

Built 1780, when Gustav III was king and Gabriel Sparre was county governor.

At those days 1 Swedish “mil” was 36 000 feet e.g. about 10688 m.

© Anna Ljungkvist

Printed in Sweden by Print & Media, Umeå universitet

Umeå 2003:302018

ISBN 91-7305-432-1

ISSN 0346-6612

To my

Mother, Father and Brother

“In science one tries to tell people, in such a

way as to be understood by everyone,

something that no one ever knew before. But

in poetry, it's the exact opposite.”

Paul Dirac

TABLE OF CONTENTS

TABLE OF CONTENTS

ABSTRACT.......................................................................................................................... 7

ABBREVIATIONS ............................................................................................................. 9

LIST OF PUBLICATIONS.............................................................................................. 10

INTRODUCTION ............................................................................................................. 11

General background.......................................................................................................... 11

Tumor microenvironment................................................................................................. 12Tumor vasculature................................................................................................... 13

Organization ...................................................................................................... 13Prognostic value ................................................................................................ 14Vascular targeting.............................................................................................. 14

Perfusion .................................................................................................................. 15Proliferation ............................................................................................................. 16

Proliferation patterns and prognosis ................................................................. 17Tumor hypoxia ........................................................................................................ 18

Definitions of hypoxia....................................................................................... 18Radiobiological hypoxia ................................................................................... 19Hypoxia and prognosis...................................................................................... 20Assays for detection of hypoxia........................................................................ 21Patho-physiology of hypoxia ............................................................................ 22Dynamics of hypoxic cells ................................................................................ 24Hypoxia and malignant progression ................................................................. 25

Cell loss, apoptosis and necrosis............................................................................. 27Relation between proliferation, hypoxia and cell loss ........................................... 27

AIMS OF THE PRESENT STUDY ................................................................................ 29

MATERIAL AND METHODS ........................................................................................ 31Tumor models.......................................................................................................... 31Exogenous cell markers and administered drugs ................................................... 31Immunohistochemistry............................................................................................ 34Image acquisition..................................................................................................... 36Resolution ................................................................................................................ 36

TABLE OF CONTENTS

Scanning and thresholding, 8-bit system................................................................ 38Scanning and thresholding, 12-bit system.............................................................. 38Post processing of images exported from IP-Lab to TCL-image .......................... 40Quantitative parameters: ......................................................................................... 41

SUMMARY OF RESULTS .............................................................................................. 45Characterization of the tumor microenvironment (paper I) ................................... 45The lifespan of hypoxic cells in three different tumor lines (paper II).................. 46Dynamic measurements of hypoxia after oxygenation modifying treatment

(paper III) ...................................................................................................................... 46Dynamic changes in hypoxia after treatment with the hypoxic cytotoxin

hydralazine (IV)............................................................................................................ 47

DISCUSSION.................................................................................................................... 49Tumor microenvironment and model systems ....................................................... 49Double hypoxic marker assay................................................................................. 50

Binding profiles of the two hypoxic markers ................................................... 50Spatial match/mismatch of the hypoxic markers ............................................. 51

Detection of perfusion limited (acute) hypoxia...................................................... 52Hypoxic patterns...................................................................................................... 53Hypoxic lifespan and its relation to hypoxic patterns............................................ 53Hypoxia and anti-cancer treatments ....................................................................... 54Reproducibility of results: impact of different image analysis systems and

staining protocols.......................................................................................................... 55Future Aims ............................................................................................................. 56

CONCLUSIONS................................................................................................................ 59

ACKNOWLEDGEMENTS............................................................................................... 61

REFERENCES ................................................................................................................. 63

ABSTRACTThe tumor vasculature is poor and heterogeneous which may result in

inadequate oxygenation and changed energy status. In addition the balance betweencell proliferation and the rate of cell death is disturbed, which results in tumor growth.

The aims of this study were 1) to gain more insight into the relation betweentumor vascularity, hypoxia, and proliferation in solid tumors, and 2) to study thechanges and dynamics of tumor oxygenation in relation to the vascular architecturewithin individual tumors. For this purpose a double hypoxic marker method wasdeveloped, which was subsequently used to 3) determine the turnover rate of hypoxiccells in three different tumor models and 4) to study the effect of cytotoxic drugs ontumor hypoxia and cell death.

Solid tumor models grown in mice were used. The tumor microenvironment wasinvestigated with exogenous cell markers for hypoxia (pimonidazole and CCI-103F),cell proliferation (BrdUrd) and blood perfusion (Hoechst 33342). The vasculature andthe exogenous cell markers were visualized with immunohistochemical techniques.The tumor sections were scanned and quantified with an image analysis systemconsisting of a fluorescence microscope, CCD camera and image analysis software.

The spatial organization of hypoxia, proliferation, and tumor vasculature wasanalyzed in several xenograft lines. The study revealed two main hypoxic patterns thatseemed to be the consequence of complex relations between vasculature, oxygendelivery, proliferation, and cell loss. The novel double hypoxic cell marker method,with sequential injection of two hypoxic markers, was developed to study dynamicchanges of the tumor oxygenation. Based on varying injection intervals between themarkers the hypoxic cell half-life was determined in three tumor lines, and rangedfrom 17 to 49 hours. Intra-tumoral changes in oxygenation status upon oxygenmodifying treatments were measured with the double hypoxic marker method. Bothdecreased levels of tumor hypoxia after carbogen breathing (95%O2 and 5% CO2) andincreased levels of tumor hypoxia, as a result of reduced tumor perfusion afterhydralazine treatment was detected. Finally the double hypoxic marker assay was usedto analyze the effects of the hypoxic cytotoxin tirapazamine in relation to the hypoxiccell population, which caused a reversible decrease of the hypoxic fraction.

The results presented in this thesis now form the basis for further studies toidentify subpopulations of cells that represent specific targets for therapy, and toinvestigate the effects of different treatment modalities.

Key words: tumor microenvironment, double hypoxic marker method, hypoxia,hypoxic half-life, and proliferation.

-7-

ABBREVIATIONSARCON accelerated radiotherapy with carbogen and nicotinamideBrdUrd bromodeoxyuridinecarbogen 95% O2 + 5% CO2

CCI-103F 1-(2-hydroxy-3-hexafluoroisopropoxy-propyl)-2-nitroimidazole

DiOC73 3,3-diheptyloxycarbocyanineEF5 2-(2-nitro-1h-imidazol-1-yl)-n-(2,2,3,3,3-pentafluoropropyl)

acetamideEIH elongation index of hypoxiaEPR electron paramagnetic resonanceglut-1 glucose transporter 1glut-3 glucose transporter 3H&E hematoxylin and eosine stainingHbO2 intravascular blood oxygen saturationsHF hypoxic fractionhif-1 hypoxic inducible factor 1hoechst 33342 bisbenzamide fluorochrome hoechst 33342IdUrd iododeoxyuridineip intraperitoneallyiv intravenouslyIVD inter-vascular distancesLI labeling indexNITP 7-(4'-(2-nitroimidazole-1-yl)-butyl)-theofyllineOER oxygen enhancement ratioPCNA proliferating cell nuclear antigenPF perfused fractionpimonidazole 1-[(2-hydroxy-3-piperidinyl)propyl]-2-nitroimidazole

hydrochloridepO2 oxygen tensionT-pot Potential tumor doubling timeTPZ TirapazamineVAMP vascular architecture and microenvironmental parametersvd vascular densityVEGF vascular endothelial growth factor

-9-

LIST OF PUBLICATIONS

This thesis is based on the following papers, which in text will be referred to by theirroman numerals.

I. Ljungkvist, A.S.E., Bussink, J., Rijken, P.F.J.W., Kaanders, J.H.A.M., van derKogel, A.J. & Denekamp, J. Vascular architecture, hypoxia, and proliferation inthe first passage of xenografts of human head and neck squamous cell carcinomas.International Journal of Radiation Oncology Biology Physics, 54, 215-228, 2002.

II. Ljungkvist A.S.E., Bussink J., Rijken P.F.J.W., Begg A.C., Raleigh J.A. KaandersJ.H.A.M & van der Kogel A.J. Imaging of the turnover rate of hypoxic cells insolid tumors. Manuscript

III. Ljungkvist, A.S.E., Bussink, J., Rijken, P.F.J.W., Raleigh, J.A., Denekamp, J. &van der Kogel, A.J. (2000). Changes in tumor hypoxia measured with a doublehypoxic marker technique. International Journal of Radiation Oncology BiologyPhysics, 48, 1529-1538.

IV. Ljungkvist, A.S.E., Bussink, J., Rijken, P.F.J.W., Kaanders, J.H.A.M., Raleigh,J.A., & van der Kogel, A.J. Immunohistochemical imaging of the dynamics oftumor hypoxia after tirapazamine treatment. Manuscript

The original papers have been reproduced with permission from the publishers.

-10-

INTRODUCTION

INTRODUCTION

General backgroundSolid tumors are the end result of a multi step process that transforms normal

cells into a tumor. The process begins with a series of mutations resulting in neoplasticcells that respond abnormally to signals regulating cell growth and death. Furthergrowth of these cells may result in avascular cell aggregates that are provided withoxygen and nutrients by simple diffusion from the surrounding tissue. However,growth beyond 1-2 mm diameter requires the ability to induce formation of tumorstroma and endothelial cells, a process known as angiogenesis (Folkman, 1976). Innormal tissues the vascular bed is regularly organized and able to respond to the energydemand via feedback mechanisms. Therefore, cells in normal tissues rarely experienceinsufficient oxygen and nutrient levels. In contrast, the blood vasculature formed bytumor-induced angiogenesis is poorly organized and has a heterogeneous architecture(Denekamp, 1992; Konerding et al., 1995; Vaupel et al., 1989), which may result in aninadequate tissue oxygenation as well as in an altered pH and bio-energetic status(Höckel & Vaupel, 2001a; Vaupel et al., 1989). The proliferation rate in regions nearthe tumor vasculature is often high. The tumor cell proliferation rate in these regionsmay be stimulated by growth factors secreted by hypoxic tumor cells situated furtherdown the oxygen gradient. These characteristic features of the tumormicroenvironment are believed to cause treatment resistance and favor tumorprogression into a more malignant phenotype, but can also be used as specific targetfor cancer treatment.

It has been known for decades that hypoxic tumor cells are more radioresistantthan well-oxygenated cells, and hypoxic cells are found in most solid tumors. Thenegative effect of hypoxia has been confirmed in clinical studies where pretreatmentoxygen tension (pO2) values were predictive for local tumor control or survival afterradiotherapy in advanced cancer of the uterine cervix and in lymph node metastasis ofhead and neck cancer (Gatenby et al., 1988, Brizel, 1997 #191; Höckel et al., 1993;Nordsmark et al., 1996). Different strategies have been developed to increase theradiosensitivity of hypoxic cells including the use of hypoxic cell sensitizers (such asmisonidazole and nimorazole), or increasing the tumor oxygenation with vasoactivedrugs (such as nicotinamide), or high oxygen gas breathing at normo- or hyper-baricconditions (Henk & Smith, 1977; Kaanders et al., 1998; Overgaard et al., 1998).Accelerated Radiotherapy combined with Carbogen (95%O2 and 5%CO2) and

-11-

INTRODUCTION

Nicotinamide (ARCON) resulted in significant radio-sensitization in the mousemammary carcinoma tumor line CaNT (Rojas et al., 1996). Accelerated radiotherapytargets tumor cell repopulation during the course of treatment, while carbogen andnicotinamide increases the oxygenation status of tumors to reduce hypoxia-mediatedradioresistance. In a phase II study ARCON has resulted in a local control rate of 80%at 3 years for 100 patients with advanced laryngeal carcinoma, which was better thanany previous report for these advanced tumors (Kaanders et al., 2002). In addition,hypoxic cells can be selectively targeted by the use of bio-reductive cytotoxins such astirapazamine (Brown & Siim, 1996).

Flow-cytometric studies have been used to study the lifetime of hypoxic cells.By using consecutive injections of two thymidine analogues it was shown in one cellline that the hypoxic lifetime was dependent on the tumor microenvironment. Thehypoxic lifetime varied from 3-5 days in spheroids in vitro to 4-10 days in vivo inxenografts (Durand & Sham, 1998). Pre-treatment differences in the dynamics ofproliferating or hypoxic tumor cell populations may account for varying efficiencies ofdifferent tumor specific treatment strategies. Optimal combinations of treatmentmodalities with different modes of action may result in complementary cell kill such astreatment strategies to increase pO2 during radiation treatment combined with ahypoxic cytotoxin to kill residual hypoxic tumor cells. These treatment modalities alterthe physiological balance between proliferating and hypoxic cell populations.Therefore it is important to understand how dynamic processes of themicroenvironment in non-treated tumors affect the efficacy of different reatmentmodalities to enable the design and optimization of new treatment combinations.

This thesis is focused on the spatial organization of the tumor microenvironmentand the dynamics of the hypoxic cell population in tumors with different micro-environmental characteristics. The dynamics of hypoxic cells in non-treated tumors aswell as in tumors treated with the high oxygen content gas carbogen or the hypoxic cellcytotoxin tirapazamine were studied with a newly developed double hypoxic markerassay.

Tumor microenvironment

A poorly organized, heterogeneous vascular bed is one of the characteristics ofthe tumor microenvironment. The blood vessels have a reduced functionality resultingin a regional heterogeneity of tumor cell proliferation and hypoxia. The tumormicroenvironment plays a critical role in malignant tumor progression and treatment

-12-

INTRODUCTION

resistance. The following sections cover the most important aspects of the differentcomponents of the tumor microenvironment. Since the focus of this thesis is on tumorhypoxia this will be covered most extensively.

Tumor vasculature

Organization

As mentioned above, tumor growth beyond 1-2 mm requires supply of oxygenand nutrients via blood vessels. The vasculature also drains the tumor of toxiccatabolites and provides a route for tumor cells to escape the primary tumor mass andform distant metastases.

The tumor vasculature is a mixture of pre-existing normal blood vessels, andnewly formed immature blood vessels formed by angiogenesis. Tumor cells that haveoutgrown the oxygen and nutrient supply may induce this process, the neo-vasculatureis often formed from the venous side (Vaupel et al., 1989; Vaupel et al., 2001).

The resulting tumor vasculature is often heterogeneously distributed throughoutthe tumor tissue and the neo-vasculature generally lacks smooth muscle cells. Tumorvessels are often dilated, tortuous, elongated, and saccular. Arterio-venous shunts andbranching with blind endings also occur frequently. The vessels may have anincomplete, or may even lack an endothelial lining, which often results in leaky vessels(Vaupel et al., 1989; Vaupel et al., 2001). These characteristic properties of the tumorvasculature have been shown in three-dimensional studies with corrosion casting(Steinberg et al., 1990). The intervascular distances are generally larger in tumors thanin normal tissue.

It has been shown that solid tumors may contain subpopulations of tumor cellsthat are more or less dependent on the proximity to the vasculature. Tumor cellsisolated from the distant cell population from the vasculature in a xenograft line had abetter take rate in nude mice than cells originating from a cell population that weremore proximal to the tumor vasculature. The tumors of the distant sub line grew fasterand had a lower vascular density compared to the proximal sub line (Yu et al., 2001).

-13-

INTRODUCTION

Figure 1. The left panel shows the regular organization of the vasculature in normal colon andthe right panel shows the heterogeneous and irregular organization of the vasculature in amelanoma xenograft (Konerding et al., 1998).

Prognostic value

Both the quantity and quality of the vasculature affects the tumormicroenvironment. These characteristics may affect the response to different anti-cancer treatments, such as irradiation due to reduced oxygenation status or poor drugdelivery (chemotherapy).

Numerous studies have indicated that vascular density is a prognostic factor forhuman cancers, but conflicting results have been reported both for metastatic potentialand local tumor control (Aebersold et al., 2000; Albo et al., 1994; Dray et al., 1995;Gasparini et al., 1993; Janot et al., 1996; Jenssen et al., 1996; Leedy et al., 1994;Moriyama et al., 1997; Murray et al., 1997; Salven et al., 1997; Sarbia et al., 1996;Shpitzer et al., 1996; Zätterström et al., 1995). A recent study showed a U-shapedrelation between microvascular density and survival in head and neck squamous cellcarcinomas. Both a very low (poor oxygenation and drug availability) and a very high(intensified angiogenesis) vascular density correlated with poor prognosis afterchemoradiotherapy (Koukourakis et al., 2000). These results may (at least partially)explain the conflicting data with respect to the correlation between vascular densityand treatment outcome.

Vascular targeting

Tumor growth is dependent on the formation of new blood vessels throughangiogenesis. This process only occurs in tumors, normal wound healing and themenstrual cycle. Angiogenesis may be induced by tumor cells that express mutantoncogenes or non-mutated proto-oncogenes (Kerbel et al., 1998).

-14-

INTRODUCTION

Around each tumor vessel thousands of cells are situated, that are dependent onthe supply of oxygen and nutrient supply provided by that vessel. Theoretically,inhibition of angiogenesis could stop tumor growth while damage of existingendothelial cells (vascular targeting) may result in massive tumor cell death.

Angiogenesis inhibitors have been pointed out as a treatment target, but there areseveral growth factors that stimulate angiogenesis, and targeting only one of these mayresult in treatment resistance (Kerbel et al., 2001). Drugs directed against angiogenesisonly attack the newly formed immature vasculature, and not the pre-existing maturevessels in the tumors. Therefore anti-angiogenic therapy should be carried out incombination with treatment against the existing mature vessels (vascular targeting) tobe effective. Even if a combination of anti-angiogenic therapy and vascular targetingwould be fully successful, still a viable piece of tumor would remain that is presumedto be fed by the normal tissue vasculature surrounding the tumor. These normal vesselsare generally less responsive to vascular targeting agents (VTAs). Vascular targetinggenerally results in a central necrosis of cells that are often hypoxic and moreradioresistant than peripheral tumor cells. Therefore radiation treatment is likely tobenefit from a combination with vascular targeting.

Perfusion

In experimental and xenografted human tumors two markers, the bisbenzamidefluorochrome Hoechst 33342 (Olive et al., 1985; Smith et al., 1988) and carbocyaninedye 3,3-diheptyloxycarbocyanine (DiOC7(3)) (Olive & Durand, 1987; Trotter et al.,1989b), are available for determining the fraction of perfused vessels at themicroscopic level. For patients such markers cannot be used due to carcinogenicity. Itwas shown that simultaneous injection of the two markers Hoechst 33342 and DiOC73

resulted in a match, while an injection interval of 20 min resulted in a partial mismatchof perfused vessels (Trotter et al., 1989a). This provided evidence for intermittentclosure and opening of tumor vessels. A similar study with lectins supported thepresence of transient tumor perfusion as well (Debbage et al., 1998). In contrast, astudy on melanoma xenografts using Hoechst 33342 and DiOC73 did not show thepresence of transient perfusion (Tufto & Rofstad, 1995).

In a dorsal skin flap window chamber it was shown that fluctuations in tumorblood flow could involve both individual vessels and clusters of vessels (Dewhirst etal., 1996). Regional fluctuations of red blood cell velocity have been detected inhuman tumors with laser doppler electrodes (Pigott et al., 1996). Mismatch

-15-

INTRODUCTION

experiments in combination with cell sorting techniques have shown that only a smallproportion of the tumor vessels experience complete cessation or reopening, but that alarge proportion of the vasculature exhibited fluctuations in the perfusion rate (Durand& LePard, 1995; Durand & LePard, 1997). This was confirmed in recent studiessuggesting that diffusion-limited, chronic, hypoxia may be present in the direct vicinityof blood vessels, and could result from tumor blood vessels that have plasma flow butvery low or absent red blood cell flux (Dewhirst et al., 1996). This was supported bythe finding of steep longitudinal gradients of intra vascular pO2 levels (Dewhirst et al.,1999).

Immunohistochemical studies on human tumors grown as xenografts haveshown that blood perfusion is heterogeneous throughout the tumor sections and that theperfusion fraction ranges from 20 to 85% (Bernsen et al., 1995; Bussink et al., 1999;Rijken et al., 2000). The heterogeneities in vascular architecture and blood perfusionmay cause hypoxia and nutritional starvation in the tumor tissue.

Proliferation

Tumor growth is the net-effect of an altered homeostasis, with a high tumor cellproliferation that is not balanced with a similar rate of cell death. Proliferating cells canbe detected with different exogenous and endogenous cell markers.

Both the PCNA and Ki67 proteins can be detected in all cell cycle phases. ThePCNA gene is constitutively transcribed in all cells, but the PCNA mRNA is onlydetectable in proliferating cells due to increased stability in these cells (Chang et al.,1990). The highest levels of PCNA are detected during S-phase, and the mostintensively stained cells may be used to determine the S-phase fraction. Theproliferation marker Ki67 is also detected in the nucleus during all phases of the cellcycle. In contrast, the synthetic and non-radioactive thymidine analogues,bromodeoxyuridine (BrdUrd) and iododeoxyuridine (IdUrd) compete with theendogenous thymidine for a place in the DNA chain during DNA synthesis (Ellwart &Dörmer, 1985). BrdUrd and IdUrd are frequently used to estimate tumor cellproliferation in vivo (Wilson et al., 1988). Since the in vivo plasma half-life of BrdUrdor IdUrd is short (15 minutes), only cells that are in S-phase immediately afteradministration of these markers incorporate them in the DNA (Grätzner, 1982).Figure 2 shows a schematic drawing of where in the cell cycle the differentproliferation markers can be detected.

Antibodies have been developed against BrdUrd and IdUrd and these substancescan be detected either by flow cytometry or immunohistochemistry (Grätzner, 1982).

-16-

INTRODUCTION

Multiple staining protocols on tissue sections allow analysis of proliferating cells inrelation to other microenvironmental parameters such as vasculature and hypoxia(Bennett et al., 1992). Several studies have shown that proliferating cells can bevisualized and analyzed qualitatively or quantitatively in relation to tumor vasculature,perfusion and hypoxia in tumor sections (Bussink et al., 1998; Kennedy et al., 1997;Rijken et al., 2000; Rijken et al., 2002; Varia et al., 1998; Wijffels et al., 2001; Wijffelset al., 2000).

The immunohistochemical analysis allows analysis of tumor sections at highspatial resolution, but it is more time consuming than flowcytometry. Although flow-cytometry allows a large number of cells to be analyzed in a short period of time, itlacks any spatial information of the tumor microenvironment. In addition tocalculations of the labeling index (LI), flow-cytometry can be used to determine theduration of the S-phase (Ts) and the potential tumor volume doubling time from asingle sample (T-pot) (Begg et al., 1985).

Proliferation patterns and prognosis

A high tumor cell proliferation rate has been shown to occur in the vicinity oftumor vasculature and decreases rapidly with increasing distance from the bloodvessels. This was probably due to reduced oxygen tension and lack of nutrients(Bussink et al., 1998; Hirst & Denekamp, 1979; Tannock, 1968). It is likely thatproliferating cells experiencing transient reductions in tumor oxygenation stopproliferating but can be recruited back into the cell cycle upon re-oxygenation, whilecells at extreme and prolonged hypoxia are doomed to die.

In vitro experiments have shown that proliferation may occur down to 0.2-1mm!Hg (Åmellem et al., 1994; Koch et al., 1973a; Koch et al., 1973b). This impliesthat cells at intermediate oxygen tensions (0.5-20 mm Hg) showing reduced radio

Figure 2. The cell cycle. In each phase thedetectable cell proliferation markers areindicated. The cell cycle phases are notdrawn in proportion to their duration.

-17-

INTRODUCTION

sensitivity still are capable of proliferation. This is supported by studies in vivo, whereadministration of both bio-reductive hypoxic cell markers and cell proliferationmarkers indicated that a small number of hypoxic cells are capable of proliferation(Kennedy et al., 1997; Varia et al., 1998; Zeman et al., 1993), although the oppositehas also been shown (Durand & Raleigh, 1998). The extent of cell proliferation underhypoxic conditions is not clear but may be of therapeutic importance since it couldconfer increased radio-resistance, and lead to rapid repopulation between radiationfractions.

For various endogenous markers of proliferation, such as PCNA, Ki-67 andcyclins, conflicting results in the correlation with treatment outcome have beenreported (Björk-Eriksson et al., 1999; Lera et al., 1998; Nylander et al., 1997; Sittel etal., 2000; Sittel et al., 1999). In a large study of 467 patients treated with radiotherapyalone, a weak correlation between treatment outcome and the proliferation status oftumor cells before the start of treatment was found in head and neck cancer (Begg etal., 1999). Although the correlation with pretreatment proliferation status was oftenweak, the importance of proliferation for treatment outcome is clearly illustrated by thedecrease in local control when radiation treatment periods are longer than 4 to 5 weeks(Withers et al., 1988).

Tumor hypoxia

Definitions of hypoxia

Tumor hypoxia is the net result of an imbalance between the O2 supply and(metabolic) demand in stroma, endothelial cells, and the tumor cells. Severaldefinitions of hypoxia exist, and depend on the experimental endpoint or detectionmethods.

Although no sharp threshold exists between normoxia and hypoxia, the literatureshows that median pO2 levels below 10 mm Hg generally result in intracellularacidosis, ATP depletion, and a drop in energy supply. Oxidative phosphorylation forATP production measured in vitro continues to 0.5-10 mmHg depending on the cellline and the experimental setup, see for review (Höckel & Vaupel, 2001b). Cell cycleprogression is affected (prolonged) at 0.2-1 mm Hg, and hypoxia may inducetranscriptional, posttranscriptional or posttranslational changes resulting either in celldeath or malignant progression in the range 1-15 mm Hg (Höckel & Vaupel, 2001b).

-18-

INTRODUCTION

Table 1. Definitions of hypoxia.pO2 (mm Hg) Definition/ usage Reference

159 At sea level

100 Alveolar blood (Stainsby et al., 1988)

95 Arterial blood (Stainsby et al., 1988)

40 Venous blood (Stainsby et al., 1988)

≈24-66 Tissue, organ dependent (Vaupel, 1990)

% <5 or

% < 2.5Typical Eppendorf resolution

≤ 20 Malignancies (Vaupel, 1990)

≈25The level below which a steep

increase of radioresistancebegins

(Höckel & Vaupel, 2001b)

10-20 Bioreductive hypoxic markers (Gross et al., 1995; Koch et al.,1995)

0.5-20 Intermediate hypoxia (Wouters & Brown, 1997)

0 Anoxia

Together with the evidence presented in the previous paragraph, these resultsindicate that cells at radiobiological levels of hypoxia may show treatment resistance,but still be capable of proliferation. Due to different definitions of hypoxia in theliterature there is a risk for unintentional inconsistency of the definition of hypoxia. Inthe experimental results presented and discussed in this thesis hypoxia is defined byreduction of bioreductive hypoxic cell markers e.g. ≤ 10 mm Hg.

Radiobiological hypoxia

Indirect evidence of oxygen-tension as an important factor for radiation responsewas provided by a study of skin reactions, where pressure on the skin providedradioresistance (Schwarz, 1909). In the 1920:s it was shown that the radiation doserequired to prevent hatching in ascaris eggs was three times higher during anaerobicthan aerobic conditions (Holthausen, 1921). Later it was shown that occlusion of blood

-19-

INTRODUCTION

flow to tissue during radiation treatment also reduced the tissue damage (Mottram,1924). During the following years more suggestive evidence for the effect of oxygenappeared, and a differential cell kill in peripheral and central parts of tumors wasdemonstrated (Mottram, 1936). Preferential damage after irradiation was seen in themarginal cells, and it was hypothesized that these cells were exposed to a bettervascular supply, and consequently to higher oxygen tensions than central hypoxic cellsthat were radioresistant. In the 1950:s Gray et al showed that the sensitivity of mousetumor cells was about three times larger when irradiated in a well-oxygenated mediumthan under anoxic conditions and that the radiosensitivity increased in tumors duringoxygen breathing (Gray et al., 1953). The increased radiosensitivity caused by oxygenwas later called oxygen enhancement ratio (OER), and in 1963 Powers and Tolmachprovided the first quantitative evidence for the OER with the demonstration of abiphasic curve with a steep initial part and a more flat section at higher doses (Powers& Tolmach, 1963). The OER was calculated to be 2.3 (Powers & Tolmach, 1963) andat present it is assumed to range from 2.5 to 3.0.

Hypoxia and prognosis

Recently tumor hypoxia has been shown to be predictive for patients receivingradiotherapy for carcinomas of the head and neck (Brizel et al., 1997; Gatenby et al.,1988; Nordsmark et al., 1996) and the uterine cervix (Höckel et al., 1993). Severalmethods have been developed to overcome radio resistance due to tumor cell hypoxia.These include the use of hypoxic cell sensitizers (such as misonidazole andnimorazole), or vasoactive drugs, and high oxygen gas breathing at normo- orhyperbaric conditions (Henk & Smith, 1977; Kaanders et al., 1998; Overgaard et al.,1998). A meta-analysis of 83 randomized clinical trials comparing radiotherapy aloneand radiotherapy combined with treatments aimed to modify tumor hypoxia showedthat both local control and survival could be improved by reducing or specificallytargeting tumor hypoxia (Overgaard & Horsman, 1996). Although oxygenation-modifying treatment protocols in general have increased both local control rate andoverall survival in several patient categories, not all of the treated patients benefitedfrom these new treatment approaches. Furthermore, these patients experiencedincreased acute side effects compared to conventional radiotherapy (Kaanders et al.,1995). If it was possible to select patients based on tumor oxygenation parameters itcould prevent unnecessary modification or intensification of treatments for a subgroup

-20-

INTRODUCTION

of patients who are unlikely to benefit from oxygenation modifying treatmentapproaches.

Assays for detection of hypoxia

Several methods are available to analyze tumor cell hypoxia. The pO2 in tissuecan be measured directly with polarographic needle electrodes (Vaupel et al., 1991).During these pO2 measurements, small amounts of oxygen are consumed excluding thepossibility to measure continuously at one position in a tumor (Collingridge et al.,1997). Clinical studies with the Eppendorf electrode showed that tumors with a lowpre-treatment pO2 do worse after radiotherapy compared to tumors that were betteroxygenated (Brizel et al., 1997; Nordsmark et al., 1996). Recently a fiber optic basedsystem (OxyLite) was developed. The pO2 measurements with this method are basedon pO2-dependent changes in the half-life of an excited luminophor at the tip of theprobe. This process does not consume oxygen, but relies solely on the presence ofoxygen (Collingridge et al., 1997; Young et al., 1996). The fiber-optic probe can be leftat one position for a longer period of time, which makes it possible to measure bothtemporal pO2-changes as well as the effect of oxygen modifying treatments such ascarbogen over time in tissue (Bussink et al., 2000b; Bussink et al., 2000c).

There are two methods for calculation of radiobiological hypoxia. Withclonogenic assays cell survival after irradiation under normoxic and anoxic conditionsis compared (Steel, 1977; Van Putten & Kallman, 1968). The second more recentlyintroduced comet assay is based on DNA damage induced by radiation (Olive &Durand, 1992).

Tumor cell hypoxia can also be studied with the use of bio-reductive chemicalmarkers such as the 2-nitroimidazoles (Chapman et al., 1981; Koch et al., 1995; Longet al., 1991; Raleigh et al., 1999; Raleigh et al., 1987). These markers are bio-chemically reduced and bound in tissues at oxygen tensions below approximately 10mm Hg (Arteel et al., 1995; Gross et al., 1995), which not necessarily represents thesame subset of cells that is detected as the radiobiological hypoxic fraction (Fenton etal., 1995). Recent studies comparing pimonidazole binding with radiobiologicalhypoxia, suggest that the bio reductive hypoxic cell markers detect hypoxic cellshaving pO2-levels corresponding with radiobiological hypoxia (Olive et al., 2000;Raleigh et al., 1999).

Antibodies have been raised against several bio-reductive markers, such aspimonidazole, EF5, NITP and CCI-103F, of which the first two are being applied in

-21-

INTRODUCTION

clinical studies (Evans et al., 2000; Kennedy et al., 1997; Raleigh et al., 2000; Wijffelset al., 2000). The hypoxic fraction can be determined either by analysis of the fractionlabeled hypoxic cells by flow cytometry (Durand & Raleigh, 1998; Hodgkiss et al.,1991) or by image analysis of microscopic sections (Bussink et al., 1999; Cline et al.,1997; Evans et al., 1995; Kennedy et al., 1997; Webster et al., 1995; Wijffels et al.,2000). By combining several immunohistochemical techniques the bio-reductivebinding of hypoxic markers can be related to other micro-environmental factors oftumors such as vascular architecture, tumor blood perfusion and proliferation (Bussinket al., 1999; Cline et al., 1994; Kennedy et al., 1997).

Besides the OxyLite system, mentioned above, only a limited number of assaysare available by which dynamic information of the oxygenation status of tumors can beobtained. These methods include electron paramagnetic resonance (EPR) oximetry,spin-lattice relaxation of fluorinated Magnetic Resonance agents (Mason et al., 1999)and cryospectrophotometry of intra-vascular oxyhemoglobin saturation profiles(Fenton et al., 1999). With a double hypoxic cell marker technique, based onscintillation counting, a reduction of tumor hypoxia after carbogen breathing andnicotinamide administration was demonstrated (Iyer et al., 1998). A disadvantage of allthese methods is the poor spatial resolution; therefore changes of the tumoroxygenation status cannot be related to other micro-environmental parameters.

In this thesis a double hypoxic marker method will be presented that can be usedto detect changes in hypoxia, with high spatial resolution, relative to the tissuearchitecture.

Patho-physiology of hypoxia

In the 1950:s a corded tumor structure was described, where endothelial stromasurrounded the cords that enclosed a central region of tumor necrosis (Thomlinson &Gray, 1955). It was concluded that the necrosis was caused by diffusion-limitedstarvation, and that the crucial nutrient in this process was oxygen. Based on early dataof oxygen consumption rates, Thomlinson and Gray postulated that the expectedoxygen diffusion distance in tissue would be approximately 150 µm, and the measuredthickness of the viable tumor cell rims never exceeded 180 µm (Thomlinson & Gray,1955). Later an inverted structure of tumor tissue with tumor cords was described thatcontains a central blood vessel, surrounded by a rim of viable tumor cells, followed bytumor necrosis at larger distances from the vessels. The radius of the viable tumorcords generally ranged from 60-120 µm (Tannock, 1968), and classical corded tumors

-22-

INTRODUCTION

described in the literature generally have the histological structure that was describedby Tannock (1968). Figure 3 shows the differences between the two types of cordedtumors.

Both types of corded tumor structures show diffusion limited hypoxia occurringat increasing distance from the tumor vasculature, a phenomenon called chronichypoxia. Later it was shown that hypoxia also could occur in the direct vicinity ofblood vessels that may be caused by transient opening and closure of tumor vessels(Brown, 1979). The above mentioned two forms of hypoxia, chronic, diffusion limitedand acute, perfusion limited hypoxia are shown in figure 4. This is a simplification of acomplex situation, and confers the extreme cases of a more dynamic situation. Recentstudies suggest that complete occlusion of vasculature is rather uncommon, but smallchanges of the perfusion rate may occur frequently (Dewhirst et al., 1996). Thesechanges can result in very low or absent red blood cell flux rate resulting in steeplongitudinal gradients of oxygen and nutrients along vessels and may causing transientperiods of hypoxia (Dewhirst et al., 1999; Walenta et al., 2001).

Nutrients with longer diffusion distances than oxygen such as glucose mayprovide chronically hypoxic cells with enough energy go into a state of anaerobicmetabolism (Vaupel, 1990). Therefore these cells may remain viable for prolongedperiods of times.

Tumor cord according toThomlinson and Gray.

Tumor cord according toTannock.

Viable tumor ≈100-150 µm

Stroma with tumorvasculature

Central bloodvessel

N

Figure 3. Schematical drawing of the two tumor cords described in theliterature (Tannock, 1968; Thomlinson & Gray, 1955). N= necrosis

-23-

INTRODUCTION

Dynamics of hypoxic cells

The tumor microenvironment is constantly changing due to tumor growth as wellas temporal and spatial variations in blood supply. This results in regional changes inhypoxia, apoptosis and proliferation rate within a tumor. Considering the heterogeneitythe tumor vasculature and amount of hypoxia it is likely that there are large differencesin the lifetime of hypoxic cells.

In vitro studies showed that the life time of hypoxic cells could extend to severaldays under severe hypoxic conditions (Koch et al., 1973a; Littbrand & Revesz, 1968).Later it became evident that the inner, nutrient deprived cells in V79 spheroids weremore vulnerable to acute hypoxia than the more viable outer cells (Franko &Sutherland, 1978). The hypoxic cells in the center of a spheroid died within 4 h, but inthe rim of a spheroid cells died after more than 6 h. Cell loss in human tumors occurspredominantly in the nutrient and oxygen deprived cell compartment. It may beassumed that the lifetime of hypoxic cells depends on various factors such as forinstance cell turnover from any labeled cell cohort of known age, such as a populationthat is pulse labeled with IdUrd. Based on this assumption the turnover rate of cellslabeled with IdUrd was used to estimate the hypoxic cell lifetime in the SiHa tumorline. In the spheroids from this line the hypoxic lifetime was 3-5 days, while in thexenografts this was 4-10 days (Durand & Sham, 1998). From these studies it wasconcluded that the lifetime of hypoxic cells depends on the tumor microenvironment(Durand & Sham, 1998).

According to the theory by Thomlinson and Gray, new cells were formed in theproliferating cell compartment that consequently pushed cells away from the vessel and

Figure 4. Schematicdrawing of diffusionlimited, chronic hypoxia andperfusion limited, acutehypoxia within tumor cords.Modified from Horsman etal 1992.

-24-

INTRODUCTION

exposed the cells to a gradual decrease of oxygenation and nutrients. Cells in thehypoxic cell compartment would be pushed further down the oxygen gradient andeventually die of oxygen deficiency and starvation. According to this theory theproliferation rate would determine the hypoxic cell turnover rate.

Normal cells that become hypoxic induce angiogenesis and start a suicidalprogram leading to apoptosis if the hypoxia is severe enough. However if tumor cellshave adapted to the harsh hypoxic environment they may still produce angiogenicfactors, combined with an acquired ability to survive under hypoxic conditions andremain viable for a longer period of time (Höckel & Vaupel, 2001a; Schmaltz et al.,1998; Vaupel & Hockel, 2003; Wouters et al., 2003). An in vitro study has shown thatthe mechanism of apoptosis induced by hypoxia seems to require acidosis and that thecells were rescued by enhanced buffering. When uncoupled from acidosis hypoxia wasshown to enhance tumor cell viability and clonogenicity relative to normoxia due todown regulation of p53 (Schmaltz et al., 1998). This suggests that the viability andlifetime of hypoxic cells may depend on their response to the tumor microenvironment.

Hypoxia and malignant progression

As mentioned above, tumor hypoxia may induce genetic changes, cellquiescence, differentiation, apoptosis, or necrosis. Generally, extreme hypoxia leads toG1/S phase cell cycle arrest, and below a critical energy state hypoxia may causenecrotic cell death. Moderate hypoxia may induce apoptosis via the p53 pathway or thep53-independent hypoxia inducible factor 1 (Hif- ) pathway. In addition to induced celldeath, hypoxia may also promote cell growth and malignant progression by adaptationto nutritional starvation. A large set of molecules and transcription factors is inducedby hypoxia and has been related to tumor progression. Extended periods of tumorhypoxia results in increased genomic instability that promotes tumor cell adaptation tothe hostile microenvironment. Clonal selection and expansion of these cells may leadto tumor propagation and formation of new hypoxic cells, followed by furtheradaptation and clonal selection resulting in a vicious circle for review see (Höckel &Vaupel, 2001b).

Among the genes induced by hypoxia are glycolytic enzymes, glucosetransporters (glut-1 and glut-3) that facilitate anaerobic metabolism and carbonicanhydrase IX (CA9) to maintain a constant intracellular pH level. These factors areregulated via the von Hippel-Lindau tumor suppressor protein and Hif-1a pathway.

-25-

a

INTRODUCTION

Glut-1, CA9 and Hif-1a expression is often found at increasing distance from bloodvessels, suggesting that they could be a marker for diffusion-limited hypoxia.

Hif-1a helps to restore oxygen homeostasis by induction of glycolysis,erythropoiesis and angiogenesis. Hif1-a is mainly found near necrotic areas, suggestinga relation with diffusion limited hypoxia, although expression also is found close toblood vessels (Aebersold et al., 2001; Talks et al., 2000; Vukovic et al., 2001; Zagzaget al., 2000; Zhong et al., 1999). Expression of HIF-1a is found closer to blood vesselsthan bio-reductive hypoxic cell markers which is suggestive for upregulation at higherpO2-levels than at which reduction of agents such as pimonidazole and EF5 take place(Vukovic et al., 2001). In oropharynx tumors the expression of Hif-1a was reported tocorrelate with loco-regional control after radiotherapy (Aebersold et al., 2001).

The carbonic anhydrases are involved in maintaining intracellular pH and over-expression in tumor cells causes a reduction of the extracellular pH thereby facilitatingthe breakdown of the extracellular matrix (Giatromanolaki et al., 2001). Thesemechanisms may play an important role towards a more aggressive tumor behavior andincreased metastatic potential. A steep increase of CA9 expression was demonstratedduring 4 to 24 h after hypoxic induction and the expression of CA9 was shown tooccur at pO2 levels below 20 mmHg (Wykoff et al., 2000). Several studies found acorrelation between pimonidazole binding and CA9 expression, and generally thetissue area showing CA9 expression was larger than the pimonidazole-positive area(Lal et al., 2001; Olive et al., 2001; Wykoff et al., 2001). The areas of mismatch mayreflect differences in oxygenation levels at which the two markers are induced, but itmay also reflect existence of temporal changes in hypoxia (Olive et al., 2001).Expression of CA9 correlates with poor outcome in several tumor types such ascarcinoma of the breast, non-small lung cancer and carcinoma of the cervix(Giatromanolaki et al., 2001; Loncaster et al., 2001; Wykoff et al., 2001). Correlationwith outcome has also been investigated for the glucose transporters and a weak butsignificant correlation between Glut-1 expression and pO2 measurements in advancedcancers of the uterine cervix was found (Airley et al., 2001).

Tumor hypoxia induces many processes via Hif-1a such as reduction ofproliferation rate and increase in apoptosis (Akakura et al., 2001; Carmeliet et al.,1998). Several studies show that there is a relationship between hypoxia and Hif-1a,but because it is an upstream transcription factor that regulates several pathways it maynot be the ideal surrogate marker for diffusion limited hypoxia.

-26-

INTRODUCTION

Cell loss, apoptosis and necrosisCell loss may occur through apoptosis induced by hypoxic cells that experience

intermediate hypoxia. In contrast, necrosis occurs under extreme hypoxia, when theenergy supply in the tumor cells is too low to be able to initiate apoptosis. On frozenH&E stained tissue sections it is not easy to distinguish apoptosis from necrosis and inthis thesis, large confluent areas of cell death are denoted as being necrotic. Apoptoticcells can be detected with antibodies against proteins involved in the apoptoticpathway such as the TUNEL method or staining of activated caspase-3 (Gavrieli et al.,1992; Krajewska et al., 1997).

Alterations of genes involved in apoptosis are important for malignantprogression and for treatment outcome (Reed, 1999), and tumor hypoxia incombination with a low apoptotic index predicts for poor survival in carcinomas of theuterine cervix (Höckel et al., 1999). The relationship between apoptosis and treatmentoutcome depends on the histological tumor type. Adenocarcinomas of the uterinecervix with a high apoptotic index have a good prognosis after treatment (Sheridan etal., 1999; Wheeler et al., 1995) while squamous cell carcinomas of the same site with ahigh apoptotic index have a poor prognosis (Levine et al., 1995; Tsang et al., 1999a;Tsang et al., 1999b). The relationship between hypoxia and apoptosis suggests thattumor reoxygenation during radiotherapy could partially be explained by cell death viaapoptosis in adenocarcinomas of the cervix (Sheridan et al., 2000).

Relation between proliferation, hypoxia and cell loss

From the preceding paragraphs it is obvious that although the differentmicroenvironmental parameters are correlated with treatment outcome it is a reflectionof a dynamic environment. Single parameter analysis shows that there are large intertumor heterogeneities in all studied parameters. By combining these into a multiparameter profile, the interplay between these factors can be analyzed. It is likely thatthe dynamic interactions between different cell populations provide information of whysome tumors respond well to treatments targeting specific subpopulations in the tumorcells, while others don’t.

Cytotoxic treatments alter the complex tumor microenvironment. Afterirradiation, pO2 in tumors increases due to a decreased oxygen consumption andalteration of blood supply. The decrease in consumption is due to a reduction in thenumber of respiring cells as a result of cell cycle arrest, apoptosis and necrosis. Thisdirect effect of irradiation causes shrinkage of the tumor and leads to a decrease of the

-27-

INTRODUCTION

hydrostatic pressure that eventually results in an improved blood supply (Kallman &Dorie, 1986). In experimental tumors an improvement of the oxygenation status wasfound with maximum oxygenation several hours after irradiation (Bussink et al., 2000a;Fenton, 1997; Olive, 1994). Analysis of the temporal changes of hypoxia in a humanxenografted head and neck squamous cell carcinoma tumor line, revealed that 3-6 dayspost-irradiation, hypoxia had returned to the pretreatment level (Bussink et al., 2000a).The initial reduction of tumor hypoxia could not be explained by an increase in tumorblood perfusion but was most likely due to a decrease in oxygen consumption (Bussinket al., 2000a; Olive, 1994). For cancers of the uterine cervix it was shown that changesin hypoxia in the early phase of fractionated radiotherapy depend on the balancebetween changes in cell density and vascular damage such as endothelial swelling andhypertrophy (Lyng et al., 2000).

The above-mentioned studies provide only limited quantitative insight into theactual spatial and temporal changes in hypoxic cell populations induced by cytotoxictreatments, but they clearly show the need to understand how these treatments affecttumors with different microenvironmental characteristics and dynamics of hypoxiccells, in order to design optimal treatment schedules aimed at complementary cell killof different tumor cell populations.

-28-

AIMS OF THE PRESENT STUDY

This thesis is focused on the dynamic relationships of tumor cells within thetumor microenvironment. Tumor vasculature, hypoxia, and proliferation wasvisualized by triple stainings of frozen tissue sections and analyzed with imageanalysis techniques.

The specific aims of this thesis are to:

I. Characterize the physiological states of a panel of human tumors, grown asxenografts, based on the vascular architecture, distribution of hypoxic cells, and thetumor cell proliferation rate.

A novel double hypoxic marker method was developed:

II. To determine the half-life of hypoxic tumor cells in three tumor lines that differ intheir hypoxic cell organization with the double hypoxic marker method.

III. To study dynamic changes of the hypoxic cell population in relation to the functionalvasculature upon oxygen modifying treatments.

IV. To determine the effect of the hypoxic cytotoxin tirapazamine on the oxygenationstatus during the first 48 hours after administration of the drug.

-29-

MATERIAL AND METHODS


Tumor models

The majority of the experiments were performed on human head and necksquamous cell carcinomas grown as xenografts in nude mice (Balb/c nu/nu mouse) (allpapers). In paper III also the tumor lines MEC82 and C38 were used. MEC82 is ahuman head and neck xenograft line derived from mucoepidermoid cell carcinoma ofthe mouth floor grown in nude mice. C38 is a murine colon tumor line transplanted inC75BL/6 black mice (Peters et al., 1987). The tumor lines were maintained by serialtransplantation of a small (1-2 mm3) piece of tumor on the right flank into 5 recipients.

In all experiments tumors with a mean diameter of 6-8 mm were used. Thetumors were measured with a caliper and the volume was calculated from threeorthogonal diameters according to the formula: tumor volume (mm3)=XxYxZx(π/6)(Dethlefsen et al., 1968).

The nude mice were kept in a specific-pathogen-free unit in accordance withinstitutional guidelines. The experiments with cytotoxins were performed at theDepartment of Toxicology of the Central Animal Laboratory. Approval of the localethical committee for animal use was obtained for all experiments.

Exogenous cell markers and administered drugs

As markers of hypoxia, CCI-103F (1-(2-hydroxy-3-hexafluoroisopropoxy-propyl)-2-nitroimidazole) (Raleigh et al., 1986; Raleigh et al., 1987) and pimonidazole(1-[(2-hydroxy-3-piperidinyl)propyl]-2-nitroimidazole hydrochloride) (Raleigh et al.,1995; Zeman et al., 1993) were used. CCI-103F and pimonidazole hydrochloride arebioreductive chemical probes, which only differ in the immuno-recognisable side chain(Figure 5).

-31-


The addition of the first electron in the bioreductive activation is reversiblyinhibited by oxygen futile cycling with a half maximal pO2 of inhibition of about3!mm!Hg, and with complete inhibition occurring below 10 mm Hg (Gross et al.,1995). Although the pharmaco-kinetics and the partition coefficients are different forthe two markers the oxygen dependence of CCI-103F and pimonidazole has beenshown to be consistent with misonidazole binding (Arteel et al., 1995; Raleigh et al.,1987). Pimonidazole accumulates with a tumor to plasma ratio of 3.0 (Saunders et al.,1984) and CCI-103F with a tumor to plasma ratio of 0.8 (Raleigh et al., 1986). Theplasma half-life in mice is about 30 min for pimonidzole and about 41 to 90 min CCI-103F (Jin et al., 1990; Raleigh et al., 1999; Raleigh et al., 1986). This taken togethermakes it likely that the absolute amounts of the reduced drugs in the individual tumorcells may differ, but that the binding profiles follow the same pattern. Figure 6 showsan example of binding the profiles for pimonidazole and CCI-103F the SCCNij3 tumorline.

Pimonidazole was administered intravenously (iv) at a dose of 80 mg/kg in avolume of 0.1 ml in saline. CCI-103F was dissolved in 10% dimethyl sulfoxide(DMSO) and 90% peanut oil and was given at a dose of 80 mg/kg intraperitoneally (ip)in a total volume of 0.5 ml. The antibodies that have been raised against these twohypoxic markers did not show any cross reactivity (Arteel et al., 1995; Azuma et al.,1997; Cline et al., 1990; Raleigh et al., 1987).

The S-phase markers bromodeoxyuridine (BrdUrd, Sigma, St Louis, MO) wasgiven ip at a dose of 25 mg/ml in a volume of 0.5!ml saline. As marker of tumor bloodperfusion the fluorescent dye Hoechst 33342 (Sigma), dissolved in saline, was given iv

Figure 5. Chemical structure ofpimonidazole and CCI-103F.

-32-


at a dose of 15 mg/kg in a volume of 0.1 ml. Hoechst 33342 is cleared from the bloodstream in mice with a half live of 110 s, and is therefore only available for DNAbinding during the first few minutes after administration (Olive et al., 1985). Hoechst33342 binds to the nuclei of cells adjacent to perfused endothelial cells.

Oxygen modifying agents: For carbogen breathing (95% O2, 5% CO2), animalswere placed in a perspex box with a continuous flow of carbogen at 5-7 l/min. Thevasodilator hydralazine (Sigma) was dissolved in saline and injected iv in the tail veinat a dose of 10 mg/kg in a total volume of 0.1 ml. Hydralazine causes an acutevasodilatation of the vascular bed in the normal tissue surrounding the tumor. Therebythe blood perfusion is rapidly decreased in the tumor, resulting in an acute decrease ofthe oxygen delivery to the tumor tissue.

Cytotoxic agents: The hypoxic cytotoxin tirapazamine (TPZ, Sanofi,Netherlands) was dissolved in saline and given iv in the tail vein at a dose of 36 mg/kgin a total volume of 0.1 ml.

The above mentioned exogenous cell markers and drugs were administered indifferent combination and time schedules depending on the assays. Table 2 shows thecombinations that were used in the different papers. For further details of the injectionsequences see the individual papers.

Figure 6. Intensity profiles of thetwo hypoxic markers pimonidazoleand CCI-103F the SCCNij3 line.

-33-


Table 2. Combinations of cell markers and drugs

Marker Paper I Paper II Paper III Paper IV

CCI-103F X X XPimonidazole X X X XBrdUrd XHoechst X X X XCarbogen XHydralazine XTirapazamine X

At the end of the experiment the tumor specimens were removed and directlystored in liquid nitrogen until cut into frozen sections. Consecutive tissue sections werecut at the largest circumference of the tumor, and stored at –80°C until furtherprocessed.

Immunohistochemistry

After thawing the sections were fixed in aceton. The Hoechst signal was scanneddry prior to any immunohistochemical staining. Then the slides were hydrated inphosphate buffered saline (PBS) 0.1 M pH 7.4 (paper I and III) or in polyclonal liquiddilutent immunostain PLD (DPC Breda Diagnostic Products Corporation) (paper II andIV). Between all consecutive steps of the staining procedures the sections were rinsedthree times for two minutes in PBS. The antibodies were diluted in PBS with 0.1%Bovine Serum Albumin c and 0.1% Tween (PBS-BT) in paper I and III, and with PLDin paper II and IV. The endothelium was always stained in combination with one ortwo other cell markers. The combinations stained in the different papers are indicatedin Table 3.

-34-


Table 3. Staining combinations used in the different papers(PerfusionHoechst)*

Endo-thelium

Hypoxiapimonidazole

HypoxiaCCI-103F

ProliferationBrdUrd

Additionalmarkers

Paper

X X X I / IIIX X X Fast Blue IX X X IIIX X X IIIX X X X II / IVX X X Caspase 3 IV

* Auto-fluorescent, scanned dry prior to immunohistochemcal staining.

Depending on the combinations of markers visualized, different combinations ofsecondary and tertiary antibodies were used to prevent cross reactivity. This wasalways checked in control sections. For details see the different studies. In addition thestaining intensities were adjusted to the dynamic ranges of the CCD cameras of the twoimage analysis systems. The following primary antibodies were used in this study:

Endothelium 9F1, rat monoclonal antibody (Mab), Department ofPathology, UMC St Radboud, Nijmegen, The Netherlands,(Bussink et al., 1998; Westphal et al., 1997)

Pimonidazole rabbit anti-pimonidazole antiserum, polyclonal antibody (Pab),(Arteel et al., 1995; Azuma et al., 1997)

CCI-103F rabbit anti-CCI-103F antiserum, Pab, (Cline et al., 1990;Raleigh et al., 1987)

S-phase cells Br-3, mouse Mab to BrdUrd, Caltag laboratories, SanFransisco, CA, (Bussink et al., 1998)

-35-


Image acquisitionIn paper I and III, an 8-bit image analysis system was used, and in paper II and

IV a 12-bit image analysis system was used. The 8-bit system has previously beendescribed in detail (Rijken et al., 1995). It has been used to analyze tumor cell hypoxiaand proliferation (Bussink et al., 1998; Bussink et al., 1999). In Table 4 the maindifferences between the two systems are summarized. The newer 12-bit system has ahigher resolution, and more memory capacity, and it has one extra filter.

The filters used in the two systems are listed in Table 4. The extra filter in the12-bit system is used for staining vasculature with a fluorochrome that is barely visibleto the human eye (far red), but it can easily be recorded with the CCD camera. Thisallows recording of up to four signals per tumor section. Another feature of the CCDcamera of the 12-bit system worth mentioning is the possibility to record bright fieldstaining. Thus it is possible to combine fluorescent and bright field microscopy inindividual tumor sections. This possibility is not used in this thesis but will bedeveloped in future studies. The scans were performed at 100X magnification fortissue sections stained for hypoxia and at 200X magnification for tissue sectionsstained for BrdUrd.

ResolutionThe starting point of each scan (upper left) was determined and x-y stepping the

coordinates were stored in the computer memory. The number of recorded fieldsdepends on the tumor size and may be up to 12 x 12 fields (e.g. up to 176.6 mm2 forthe 8-bit system and up to 343.2 mm2 for the 12-bit system). The tumor sections werescanned field by field in a meander pattern.

Due to the above-mentioned limitations in memory two reduction steps of 50%were performed during image recording and processing in the 8-bit system incomparison with one 50% reduction step in the 12-bit system. This results in a finalresolution of 7.200 µm/pixel for the 8-bit system and 2.668 µm/pixel in the 12-bitsystem.

-36-


Table 4. Selected specifications for the two image analysis systems.Parameter 8 bit system (Paper I/II) 12-bit system (Paper III/IV)Lamp HBO 100 HBO 100 adjustable from 1-100%

Microscope (Zeiss) Axisokop Axioskop II

Uv filterExcitation 365 nm

Emission 420 nm

Excitation 365 nm

Emission 420 nm

GreenExcitation 450-490 nm

Emission 520 nm

Excitation 450-490 nm

Emission 515-565 nm

RedExcitation 510-560 nm

Emission 690 nm

Excitation 546/12 nm

Emission 575-640 nm

Far redExcitation 575-625 nm

Emission 660-710 nm

Detection range CCD camera Fluorescence Fluorescence and bright field

Video signal Analogue Digital

Sensitivity Manual gain control Integration signal

No of images averaged to minimizenoise 10 1

Grey levels 256 4096

Resolution CCD camera 752 x 504 pixels 1300 x 1030 pixels

Resolution of the composite binaryimages

25% (Two 50% reductionsteps) 50% (one reduction step)

µm per pixel in binary compositeimage at 100X 7.2 µm/pixel 2.668 µm/pixel

Maximal scan size (12 x12 fields) 16.2 x 10.9 mm 20.8 x 16.5 mm

Software for image recording andprocessing

TCL-image (TNO, Delft,Netherlands)

IP-Lab (Scanalytic Inc. Fairfax,VA, USA)

Saved image mode Binary Gray scale + binary

Software for outlining tumor areaand masks TCL-image IP-Lab

Quantitative analysis TCL-image TCL-image

-37-


Scanning and thresholding, 8-bit systemThe left flow-chart (page 39) shows the steps in the scanning procedure for the

8-bit system. The segmentation for each recorded signal was set in one representativefield after background reduction, on which a max-min operation was applied. Thethresholds were interactively set to isolate the foreground pixels from the background.The max-min filter was applied on all recorded fields. This corrected for unevenillumination and reduced small differences of the background levels of the stained cellmarkers both within and between tumor sections. This resulted in one relativethreshold for each marker within one experiment.

For all markers studied in this thesis, except the hypoxic cell markers, thethreshold was tumor line independent. Pilot experiments had shown that for tumors ofdifferent passages (first passage versus third passage) of the same xenograft tumor lineone common threshold could be set per hypoxic marker (unpublished data). Differentthresholds were set for the hypoxic markers CCI-103F and pimonidazole. Thus aseparate threshold was set for the two hypoxic markers (pimonidazole and CCI-103F)in each tumor line.

Since the individual fields were recorded and immediately processed into abinary image without storage of the original gray level image it was not possible tochange the threshold afterwards on the composite binary image.

Scanning and thresholding, 12-bit systemThe right flowchart in (page 39) shows the sequence of the different steps in the

scanning procedure for the 12-bit system. The scans were performed with constantlight intensity and exposure time for each cell marker within one experiment. Thissystem has more setting options than the 8-bit system that allows setting of thesegmentation level (threshold) both at the beginning of a scanning session and after thescanning procedure in an already stored composite gray scale image. A backgroundimage, an illuminated clean glass slide, recorded before each scanning session wassubtracted from the individual recorded fields. The thresholds for tumor vasculature

-38-


8 bit system 16 bit system

Initialization and calibration of thescanning stage

One threshold value for segmentingthe structures from the backgroundis recorded in a representative fieldand stored for later use

During scanning individual fieldsare recorded, corrected for unevenillumination (using local min-maxfilter operation), binarized with therecorded threshold, and montagedinto one composite binary image

Tumor area and masks are drawnand added as to the correspondinglayered image to exclude stainingartefacts and necrosis

The composite images of all signalsare layered into one image forfurther quantitative analysis

Binarizing: one threshold perexperiment for vasculature andperfusion; one threshold for eachtumor section for the hypoxicmarkers.

The tumor area and masks aredrawn to exclude staining artefactsand necrosis (binary)

The composite images of all signalsare layered into one image forfurther quantitative analysis

Initialization and calibration of thescanning stage

Illumination is set to obtainmaximal dynamic range of signalintensities

One background image is recordedusing a clean glass slide

During scanning the sectionindividual fields are recorded asgray value images. Afterbackground subtraction the image ismontaged into one composite image

Illumination and camera gain are setfor optimal image recording

Camera exposure time is set toobtain maximal dynamic range ofsignal intensities

-39-


and perfusion were set once for the whole experiment in analogy with the 8-bit system.This was not possible for the hypoxic cell markers with the 12-bit system.

To allow absolute measurements of hypoxic marker binding intensities asfunction of distance to the vasculature, the staining intensity was adjusted to makeoptimal use of the dynamic range of the camera. For the same reason a max-minoperation was not applied on the recorded fields. The combination of a more sensitivesystem and inter-tumor differences in hypoxic cell marker gradients prevents setting ofa tumor-line specific absolute threshold for the hypoxic markers.

The human eye can distinguish about 255 gray levels while the 12-bit CCDcamera detects 4096 gray levels, therefore it was difficult to interactively set a correctthreshold of the hypoxic cell marker signal. Instead a standardized numerical methodfor thresholding was developed. Briefly, a gray-value histogram of the intensity valueswas generated from the total number of pixels in the whole recorded image. Thehistogram was visualized with a linear axis for the number of pixels and a logarithmicscale for the intensity values. A curve fit was calculated on the part of the histogramthat showed a linear decrease in the number of pixels over a logarithmic range ofintensity values. The intersection of the curve fit with the intensity values (x-axis) wascalculated and used as threshold for the tumor-section. Composite binary images werethen created based on the calculated segmentation value.

A consecutive H&E stained tumor section was used to create a binary imagewith a mask of the tumor area, excluding non-tumor tissue and necrosis from theanalysis. A comparable “mask image” was defined in the same way in which thehypoxic cell markers were defined without artifacts. Thus, for each tumor scan sevenbinary images were generated, consisting of four scans (vasculature, perfusion marker,first and second hypoxic marker) and three masks (tumor area, first and secondhypoxia masks).

Post processing of images exported from IP-Lab to TCL-image

As indicated in Table 4, different software packages were used for imageacquisition and image processing. On the 12-bit system IP-Lab was used for scanning

-40-


and creation of binary images. After generating the masks, the binary images from thescans were exported to the image analysis software TCL-image (TNO, Delft, TheNetherlands). This resulted in one image with the seven different scans layered forspatial analysis of hypoxia and other tumor microenvironmental parameters.

Before performing quantitative analysis two image analysis operations wereperformed: 1) “dilation” of Hoechst 33342 signal with two pixels, and 2) “closing” oftwo pixels of the hypoxia signal.

The pixels positive for endothelial cells and Hoechst 33342 positive cellssurrounding these did not overlap due to the high resolution. Therefore the firstoperation was performed to facilitate analysis of overlap of the vasculature andperfusion markers Hoechst 33342 e.g. the perfused vessels.

The hypoxic cell markers pimonidazole and CCI-103F are mainly detected inthe cytoplasm of hypoxic cells. Tumor cells containing cross-sections of nucleicontains visible dark non-stained “holes” appear in the tumor sections. These were alsopresent in the binary images. The second operation prevented underestimation of thehypoxic fraction caused by negative areas of cross sections of nuclei in hypoxicregions. This operation was performed for both hypoxic cell markers. It was chosen notto make a complete fill of these negative areas, to prevent that small necrotic fociaccidentally would be analyzed as hypoxia. The maximal area that was completelyfilled was less than the size of an average mammalian cell.

Quantitative parameters:Although the image acquisition and the post-processing of the images differed

between the two systems, the quantitative parameters were identical for the imageanalysis systems. Table 5 shows a list of all quantitative parameters that were used inthis thesis and the references for these. Considering the fact that different stainingprotocols were used for the two image analysis systems, which have differentresolutions and different post-processing of the images, differences of the absolutevalues from the quantitative analysis were expected to occur.

-41-


Table 5. A list of the parameters calculated from the binary images.

Parameter Abb-reviation

Definition Reference: Used inpaper

Perfusedfraction PF

The area of vascular structures labeledwith the perfusion marker Hoechst33342 divided by the total vasculararea.

(Bernsen et al.,1995; Rijken et al.,1995)

I / II / III

Hypoxicfraction HF

The fraction of the tumor surfacestained for the hypoxic marker relativeto the total tumor surface.

(Bussink et al.,1999; Rijken et al.,2000)

All

Labelingindex LI the BrdUrd-labeled area divided by the

Fast Blue labeled area(Bussink et al.,1998) I

Elongationindex hypoxia EIH

The longest (A) and the shortest (B)axis of each hypoxic area wasdetermined, and the ratio was thencalculated according to A/B.

I

Vasculardensity VD The number of vascular structures per

mm2.

(Bernsen et al.,1998; Rijken et al.,2000; Rijken et al.,1995)

I

Inter vasculardistances IVD The shortest distance between tumor

vessels (µm)(Rijken et al.,2000) I

HF per zone HFZ

HF was determined in six zones at 0-22, 22-50, 50-100, 100-150, 150-200,and 200-250 µm around each vessel,and was presented as the mean of allvessels

(Bussink et al.,1999; Rijken et al.,2000; Wijffels etal., 2000)

I

LI per zone LI perzone

The LI was analyzed in seven zones at0-25, 25-50, 50-75, 75-100, 100-150,150-200, and 200-250 µm around eachvessel and presented as the mean of allvessels.

I

Fraction nonviable tissue I

Dynamics ofhypoxia HF1/HF2 The first hypoxic fraction divided by

the second hypoxic fraction II / IV

-42-


A comparison of the values for the two hypoxic markers CCI-103F andpimonidazole in the SCCNij3 line was done for two experiments (Table 6). Theexperiment in paper II was analyzed with help of the 12-bit system and the experimentin paper III was analyzed with help of the 8-bit system. The hypoxic fractions (HF)obtained for control tumors in the two experiments were similar, but the median HF forboth pimonidazole is CCI-103F is smaller when analyzed with the 12-bit systemcompared with the 8-bit system. However, the differences between the systems arelarger for CCI-103F than for pimonidazole, which results in a larger mismatch betweenthe two markers on the 12-bit system. The resulting ratio is around 1.6 and correspondsto a one to maximally two cell layers difference in distance to the start of hypoxia forthe to markers, in a tumor cord of about 150 µm.

Table 6. Comparison of HF in the SCCNij3 line for pimonidazole and CCI-103F for the 12-bit (n=5) and the 8-bit system (n=7).

Pimonidazole CCI-103F12-bit system 8-bit system 12-bit system 8-bit system

Median 6.9 7.9 4.3 6.5Mean 10.6 7.0 6.7 6.1SEM 4.6 1.5 3.5 1.3

-43-

SUMMARY OF RESULTS

SUMMARY OF RESULTS

Characterization of the tumor microenvironment (paper I)

Tumor cell hypoxia and proliferation was analyzed in relation to the tumorvasculature. Fourteen tumors from 11 first generation xenograft lines of human headand neck squamous cell carcinomas were studied.

Hypoxia, detected with pimonidazole, was found in all tumors. The tumors weredivided into three groups: patchy, mixed, and ribbon-like, based on qualitativemorphological features. The hypoxic patterns were quantified according to anelongation index (EIH, length/width) that ranged from 2.0 to 28.3 and showed a goodcorrelation with the qualitative analysis. The bands of hypoxia in the ribbon-liketumors often seemed to be in close relation to networks of vessels arranged in sheets(lower panels of Figure 1, paper I). These networks were not found in the patchytumors. Analysis of the distribution of hypoxia throughout the tumor section or asfunction of distance to the vasculature did not reveal any differences between tumorswith different hypoxic patterns.

Analysis of proliferation (BrdUrd labeling) did not reveal any distinguishableproliferation patterns, neither in relation to the hypoxic patterns nor in relation to thedistance to the vasculature.

Analysis of EIH showed a significant correlation with the hypoxic fraction (HF)but not with other measured parameters. However, there was a trend towards a higherlabeling index and a larger fraction of necrosis with increased EIH.

Analysis of the hypoxic fraction as function of distance to the vasculatureshowed that in all tumor lines varying amounts of hypoxia was present already withinthe first 50 µm from the vasculature, indicating the presence of perfusion limited(acute) hypoxia. In most tumors the HF increased with the distance to the vasculature.In addition, the labeling index of BrdUrd was generally highest adjacent to thevasculature with a subsequent decrease at increasing distance. Although theseparameters were analyzed on consecutive tissue sections the results indicated co-existence of hypoxic and proliferating cells that was most pronounced at 100-150 µmfrom the vasculature (Figure 5 and 8 paper I).

-45-

SUMMARY OF RESULTS

The lifespan of hypoxic cells in three different tumor lines (paper II)A double hypoxic marker assay was applied to determine the lifespan of the

hypoxic tumor cell population. Three tumor lines were studied, SCCNij3 (patchyhypoxic pattern), and the MEC82 and C38 (ribbon like hypoxic patterns). TheSCCNij3 and MEC82 are derived from human head and neck tumors of squamous andmucoepidermoid origin, respectively. C38 is a murine colon tumor line transplanted inC75BL/6 black mice.

Pimonidazole was always injected first, followed by CCI-103F at varying timeintervals after pimonidazole, but always 2.5 h before tumor harvest. There was a goodspatial match between the two markers at short injection intervals, but pimonidazolealways covered a larger surface fraction than pimonidazole in the binary images(Figure 3 paper II). Therefore, the ratio for the mean hypoxic fractions of the firsthypoxic marker pimonidazole (HF1) and the second hypoxic marker CCI-103F (HF2)was set to 1.0 for the control tumors of each tumor line. The ratios for the subsequentinjection intervals were normalized relative to the control values, and from theresulting curves the half-life of the hypoxic cells was calculated for each line. The half-life (95% confidence interval) of the hypoxic cells was 49h (28-185h) for SCCNij3,23h (16-41h) for MEC82 and 17h (15-19h) for C38 (Figure 4 paper III). The rankingof the potential volume doubling times in these tumor lines was related to the rankingof the hypoxic lifespan.

The hypoxic markers are only reduced and bound in viable hypoxic tumor cells,and are not reduced in necrotic tumor areas. Once reduced and bound to intracellularproteins the, hypoxic marker adducts are stable in the tissue. Therefore label fromdying cells was expected in necrotic areas at longer time intervals. Qualitative analysisof the tumor sections showed that the rapid disappearance of pimonidazole labeledhypoxic cells in C38 and MEC82 indeed corresponded with cell debris stainingpositive for pimonidazole in the necrotic areas. This was already evident at the 16hinjection interval in the C38 line and at the 24h injection interval in the MEC82 line.

Dynamic measurements of hypoxia after oxygenation modifying treatment(paper III)

The experiments were done with the human laryngeal squamous cell carcinomaxenograft line SCCNij3. A double bioreductive hypoxic cell marker assay wasdeveloped to study changes in tumor hypoxia after tumor oxygenation modifyingtreatments. The two hypoxic markers, pimonidazole and CCI-103F, were sequentially

-46-

SUMMARY OF RESULTS

injected in each tumor. The oxygen modifying treatment; carbogen breathing (95% O2

and 5% CO2) or hydralazine was administered between the hypoxic marker injections.A good spatial match was seen between the hypoxic markers in non-treated

animals (Figure 1A and 2A paper III). Carbogen caused a drastic and rapid reductionof the mean hypoxic fraction. It decreased from 7% detected by CCI-103F beforetreatment to 3% detected by pimonidazole administered during carbogen breathingp=0.009), (Figure 4 paper III). The decrease of hypoxia did not occur randomly, andwas most pronounced close to well-perfused regions where hypoxia often completelydisappeared.

Hydralazine caused vasodilatation of the vasculature in the surrounding normaltissue. This resulted in a reduction of the tumor perfusion, and subsequently in asignificant increase of the mean hypoxic fraction from 16% to 40% (p≤0.05).

Thus with consecutive injections of two bio-reductive markers changes in tumorhypoxia can be quantified in relation to the (perfused) tumor vasculature.

Dynamic changes in hypoxia after treatment with the hypoxic cytotoxinhydralazine (IV)

The effect of the hypoxic cytotoxin tirapazamine (TPZ) on the hypoxic tumorcell population was studied with the double hypoxic marker method described in paperIII. TPZ was injected between the subsequent hypoxic marker injections. The dynamicchanges in hypoxia were analyzed with a triple staining of the endothelium plus thetwo hypoxic markers within one tumor section. Therefore analysis could be done on acell-by-cell basis. Qualitative and quantitative analysis showed that TPZ induced adecrease of the HF reaching a nadir about 3.5h after TPZ injection followed byrehypoxification 24h after treatment (Figure 1 and 2 paper IV). Qualitative analysisshows that the effect of TPZ is largest at intermediate distances from the tumorvasculature. In addition the triple staining showed that the same cells that were hypoxicbefore TPZ injection had become hypoxic again 24h after TPZ treatment.

A triple staining of vasculature, caspase 3 and CCI-103F showed a trend towardsincreased apoptosis in the hypoxic cell compartment at 24h and 48 after TPZadministration.

-47-

DISCUSSION

DISCUSSION

Tumor microenvironment and model systemsThe main focus of this thesis was to study the dynamics of hypoxic tumor cells

in relation to the vascular network in an experimental setting. These were mainlyperformed in human head and neck tumors grown as xenografts. Xenografts consist oftumor cells of the donor genotype (for example human), while the vascular and stromacomponents are of host origin (for example mouse). This was shown by comparison ofbinding patterns of anti-human and anti-mouse antibodies in xenografted human headand neck tumors (Warenius, 1980).

The use of xenografts allows in vivo analysis of the interplay between differenttumor cell populations in solid tumors. Clinically relevant treatment strategies can beevaluated with respect to the tumor microenvironment, using endpoints such as growthdelay. Most human xenograft lines only survive in vivo, and must be passaged fromanimal to animal. The advantage of xenografts is that these tumor lines are the bestexperimental model systems for human tumors. A disadvantage is however the highcost and the large number of animal needed for maintenance of the tumor lines. Anumber of well-known tumor lines are further transformed and can be grown in cellculture, as spheroids and as xenografts, such as the SiHa and WiDr senografts (Durand& Sham, 1998). An advantage of these tumor lines is that they can be used both in vivoand in vitro. A disadvantage is that these tumor models are further transformed andcould have lost some of the original properties that may affect the tumormicroenvironment.

Comparisons of several human head and neck tumors grown as xenografts(paper I) showed that the human tumor cells induced specific growth patterns of thetumor vasculature. It was evident that these vascular characteristics were tumorspecific since the vascular architecture differed between xenograft lines, and remainedconstant when passaged. This is in agreement with other studies of human xenografts.Examples of the donor dependence of the vascular architecture have been reported forhuman glioma and melanoma xenograft lines in nude mice (Bernsen et al., 1995;Solesvik et al., 1982). In addition, vascular corrosion casting studies showed that thethree-dimensional architectures of tumor vessels were tumor line dependent(Konerding et al., 1999).

In the present study the hypoxic patterns identified by morphologicalclassification were tumor line dependent and were retained over subsequent passagesin the individual tumor lines (paper I). The hypoxic patterns seemed to be related to the

-49-

DISCUSSION

vascular architecture, These were similar to the hypoxic patterns found in primaryhuman head and neck tumors (Wijffels et al., 2000). It can be concluded that thesexenograft lines can be used as models for the microenvironment of human head andneck tumors. The hypoxic patterns were also similar to hypoxic patterns reported forhuman and canine tumors elsewhere (Cline et al., 1994; Kennedy et al., 1997).

Double hypoxic marker assay

To study the dynamics of hypoxic tumor cells in relation to the tissuearchitecture a double hypoxic marker assay was developed. This was first presented inpaper III. The two hypoxic markers pimonidazole and CCI-103F were consecutivelyinjected with oxygenation modifying treatment in between, revealing both increasedand decreased levels of hypoxia. Moreover this assay was used to quantify hypoxic cellhalf-lives (paper II) and the effect of treatment with the hypoxic cytotoxintirapazamine on tissue oxygenation (paper IV).

Pimonidazole accumulates more rapid in hypoxic areas than CCI-103F. Inaddition reduced and bound pimonidazole adducts have been shown to be more stablethan CCI-103F adducts in canine tumors (Azuma et al., 1997). Therefore the injectionsequence of the two hypoxic markers varied depending on the length of theexperiments.

Binding profiles of the two hypoxic markers

Measurements of intensity profiles in paper II of the two consecutively injectedhypoxic markers, pimonidazole and CCI-103F, resulted in hypoxic profiles with asimilar shape but different absolute intensities. This indicates that metabolism andbinding of the two markers occurred in the same cells, but that the absolute stainingintensities differed. Pimonidazole and CCI-103F differ in the side chain that affects thehydrophilicity and intracellular accumulation of the markers. This affects the stainingintensities of the two markers as well as primary antibody affinities, dilutions andincubation times with the antibodies. This is supported by the fact that the bindingpatterns of pimonidazole and CCI-103F are similar to that for misonidazole (Arteel etal., 1995; Raleigh et al., 1987). Misonidazole shows increased binding at and below10"mm Hg (Gross et al., 1995) which indicates similar oxygen dependence for thepimonidazole and CCI-103F.

-50-

DISCUSSION

Another clinically applicable 2-nitroimidazole is EF5 [2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl) acetamide]. Based on the biochemical propertiesEF5 is expected to have similar oxygen dependence as pimonidazole and CCI-103F,which would result in similar binding profiles of hypoxic cells labeled with any of thethree markers. EF5 has another unique side chain that can be detected with antibodies(Evans et al., 1995) and is easier to dissolve than CCI-103F. Therefore EF5 is expectedto be a good candidate for use in the double hypoxic marker assay.

Spatial match/mismatch of the hypoxic markers

One would expect from the similar intensity profiles of pimonidazole and CCI-103F that the hypoxic fractions estimated by the two hypoxic markers would beidentical when injected concurrently or with short intervals. However this was not thecase. The ratios of the hypoxic fractions for pimonidazole over CCI-103F were alwayslarger than one. One reason is the difference in signal to noise ratio, which is lower forCCI-103F than for pimonidazole. This might be caused by larger variations of thebackground levels of CCI-103F than of pimonidazole. These differences affect thethreshold value and the resulting hypoxic fraction calculated from the binary images. Auniversal correction factor for the hypoxic fraction of CCI-103F could not beestablished due to the larger intratumoral differences observed for CCI-103F than forpimonidazole binding.

Delivery problems of CCI-103F compared to pimonidazole could be ruled outsince intensity measurements showed similar intensity profiles for both markers. Inaddition, differences in spatial match between the markers were seen at intermediatedistances from the vasculature, while at further distances the spatial match for the twomarkers was better (Figure 3 and 5 paper II). Moreover comparison of the untreatedcontrol tumors of the hypoxic life-span study (paper II) and the tirapazamine study(paper IV) showed that the ratio between the two markers was not affected by theinjection sequence.

Therefore it was concluded that differences in signal to noise ratio of the twohypoxic markers and a difference in staining intensity were the main reasons for thesmall differences in hypoxic fractions for pimonidazole and CCI-103F at shortinjection intervals. However, an excellent overall spatial match of these two markersstill was found.

-51-

DISCUSSION

Detection of perfusion limited (acute) hypoxiaTwo forms of hypoxia have been shown to exist in solid tumors; diffusion-

limited chronic hypoxia (Tannock, 1968; Thomlinson & Gray, 1955) and transientperfusion-limited acute hypoxia (Brown, 1979). These forms of hypoxia represent themost extreme forms of hypoxia, and intermediates between these two forms are mostlikely to be present in the tumor tissue. It was demonstrated that longitudinal gradientsin pO2 exist within tumor vessels, which could be related to the tissue pO2 (Dewhirst etal., 1999). In addition, it has been shown that small fluctuations in the blood perfusionare more common than complete vascular closures or openings (Durand, 2001).

The binding of the bioreductive hypoxic markers depends of the absolute levelsof pO2 and the total distribution of hypoxia. The immunohistochemical stainingpatterns of these markers correspond with the expected areas of diffusion-limitedhypoxia. The presence of hypoxia within the first 50 µm from tumor vessels, whichwas found in the characterization study (paper I), provides circumstantial evidence fordetection of perfusion-limited hypoxia with bio-reductive hypoxic markers. However,it could be argued that these short distances to hypoxia would be caused by a very lowpO2 in the tumor vessels or of a complete vascular shutdown of long duration.

In the hypoxic life-span study (paper II) evidence for detection of perfusionlimited hypoxia with hypoxic markers was found at the 2h injection interval in theMEC82 line (Figure 2 paper II). A mismatch, with increased staining of the secondhypoxic marker, CCI-103F, relative to the first hypoxic marker, pimonidazole, wasfound (see arrows). Measurement of hypoxic marker intensity profiles (unpublisheddata) showed large spatial variations of the CCI-103F staining profiles relative to thepimonidazole staining profiles in areas with perfusion-limited hypoxia.

In addition, a gradient of CCI-103F staining intensities along the feeding vessel(Figure 2, paper II, FV) from the tumor periphery to the area with transient perfusionlimited hypoxia (Figure 2, paper II, AH) was indicative for a decreasing oxygengradient along the vessel. This indicates that temporal fluctuations of the perfusion ratein MEC 82 have resulted in regions with transient hypoxia in this specific tumor. Acloser examination of all tumors in the MEC82 line and in the other two tumor lines(C38 and SCCNij3) has indicated that these fluctuations of hypoxia occur at varyingdegrees in all three lines but that this was most pronounced in MEC82.

-52-

DISCUSSION

Hypoxic patterns

The hypoxic patterns describe the global organization of hypoxic cells inrelation to the vascular architecture. A corded structure of the hypoxic cell organizationin relation to hypoxia with either peripheral stroma, followed by hypoxia and centralnecrosis (Thomlinson & Gray, 1955) or a central blood vessel surrounded by hypoxiccells at a distance followed by necrosis (Tannock, 1968) has been described. Thus bothmodels were associated with diffusion-limited hypoxia. Generally only experimentaltumors have been described as being truly corded. In the xenograft lines with a patchystructure of hypoxia, the tumor cells were organized as clusters of grapes around thevessels, while in the xenograft lines with a ribbon-like hypoxic structure the tumorcells were lined up as bands along the tumor vasculature (Figure 1 paper 1).

The patchy hypoxic pattern could be explained as several tumor cords with acentral blood vessel and at a distance, diffusion-limited hypoxia of the type describedby (Tannock, 1968). Between the tumor cords corners of tissue could be present atinter-vascular distances exceeding the oxygen diffusion distance, resulting in “spots”of diffusion- limited hypoxia with necrosis at further distances from the vessels inlarger hypoxic areas.

The ribbon-like hypoxia pattern is characterized of hypoxic cells located alongelongated networks of vessels with necrosis at further distances. This organizationcorresponds with the description of Tomlinson and Gray with peripheral stroma andcentral necrosis (Thomlinson & Gray, 1955). Diffusion-limited hypoxia would arisebeyond the diffusion distance of oxygen, followed by central necrosis. A trend towardshigher hypoxic fractions, and BrdUrd labeling indices together with a higherproportion of non-viable tissue in the ribbon-like tumor lines indicated a higher cellturnover in these tumors than in tumor lines with a patchy hypoxic pattern.

Hypoxic lifespan and its relation to hypoxic patterns

Hypoxia inhibits cell proliferation and eventually leads to cell death, but hypoxiaalso induces signals for angiogenesis and may lead to a more aggressive phenotype.The double hypoxic marker assay (paper II) showed that hypoxic cells in the tworibbon-like tumor lines, MEC82 and C38, had a higher turnover rate than in the patchytumor line, SCCNij3. These results were consistent with the trend towards a higherlabeling index and a larger proportion of non-viable tissue for the ribbon-like xenograftlines that described in paper I.

-53-

DISCUSSION

Generally murine tumor lines tend to have faster growth kinetics then humanxenograft lines, and it cannot be excluded that this plays a role in the rapid kinetics ofthe C38 line. However, differences between the labeling indices and T-pot values forthe ribbon-like murine C38 line and human xenografted MEC82 line were smaller thanthe differences between the two xenograft lines with different hypoxic patterns(MEC82 and SCCNij3). This suggests that the hypoxic patterns could be related to thedynamics of the hypoxic cell population. Tumor lines with a ribbon-like hypoxicpattern might have a shorter half-life of hypoxia, indicating that these cells have alimited ability to adapt to the hypoxic environment. Tumor lines with patchy hypoxicpatterns tend to have a lower amount of necrosis and a lower labeling index. Thesefeatures may be extrapolated to the data for the SCCNij3 line, and may reflect acquiredhypoxic tolerance.

Hypoxic tolerance requires the ability of hypoxic cells to reduce the energydemand and translation resulting in an altered proteome. It has been suggested thattumors with hypoxic tolerance e.g. prolonged survival at low oxygen tensions couldplay a role in tumor growth by providing a continuous angiogenic signal (Wouters etal., 2003). This has been supported by recent observations, showing that the tumormicroenvironment affects the half-life of hypoxic cells. The lifespan of hypoxic cellsof the SiHa line was 3-5 days in spheroids and 4-10 in xenografts (Durand & Sham,1998). Furthermore, the extra-cellular pH has been shown to affect the viability andclonogenicity of hypoxic cells (Schmaltz et al., 1998), and cell-sorting experimentsindicated that selection of tumor cell subpopulations with a reduced vasculardependence might occur during disease progression (Yu et al., 2001). Selection of sub-lines with low vascular dependence resulted in higher take rates and a more rapidgrowth of a melanoma xenograft line in nude mice (Yu et al., 2001).

Hypoxia and anti-cancer treatments

Tumor cell hypoxia reduces the radiosensitivity, and three general treatmentstrategies have been designed to overcome hypoxic treatment resistance duringradiotherapy. These are: radiosensitizers such as misonidazole, oxygen-modifyingtreatments where carbogen breathing is applied and selective hypoxic cytotoxins suchas tirapazamine. A combination of tirapazamine with irradiation could result incomplementary cell kill (Brown & Siim, 1996; Wouters et al., 2002).

By using the newly developed double hypoxic marker method we were able tostudy the dynamics of the hypoxic cell population after oxygenation modifying

-54-

DISCUSSION

treatment (paper III) and after treatment with the hypoxic cytotoxin tirapazamine,(paper IV).

The SCCNij3 line was treated with the high oxygen content gas carbogen (95%O2 and 5% CO2), and the effect of carbogen was compared with pretreatment levels inthe individual tumors (paper III). The effect of carbogen was most pronounced in thevicinity of perfused vessels, indicating that the intermediate hypoxic cell might benefitmost from the increased oxygenation. Also increased levels of hypoxia could bemeasured after treatment with the vasodilator hydralazine, which divert the blood flowfrom the tumor vasculature to the surrounding normal vasculature.

Treatment with tirapazamine in the SCCNij3 line (paper IV) was most efficientat changing the oxygenation at intermediate distances from the vasculature.Tirapazamine is a highly diffusible molecule that is most effective at very low pO2

levels (Brown, 1993), but it was shown that tirapazamine might be fully redused beforeit reaches the most distant hypoxic cells (Kyle & Minchinton, 1999). Repeatedadministrations are feasible in mice (Spiegel et al., 1993) and increased cell kill isexpected upon repeated administrations.

Tirapazamine treatment caused a transient decrease in tumor hypoxia thatreached a nadir approximately 3.5h after administration. Subsequently, tumor hypoxiaslowly increased, and was almost back to pre-treatment levels at 48h after tirapazamineinjection. The kinetics of reoxygenation and rehypoxification in the SCCNij3 aftertirapazamine (paper IV) was similar to the kinetics observed in this line afterirradiation (Bussink et al., 2000a). After tirapazamine treatment a similar pattern ofreoxygenation followed by rehypoxification was reported for the SCCVII line both fortirapazamine and irradiation (Kim & Brown, 1994). The kinetics differed betweenSCCNij3 in this study and the SCCVII line. When the nadir for SCCNij3 was reachedalready pretreatment levels were seen in the SCCVII line. Although methodologicaldifferences cannot be excluded, this suggests a tumor line dependence of thetirapazamine response that might be related to the dynamics of hypoxic cells.

Reproducibility of results: impact of different image analysis systems andstaining protocols

The larger hypoxic fraction for pimonidazole compared to CCI-103F wasconsistent both for experiments analyzed on the 8-bit and 12-bit image analysissystems. A comparison of the results obtained for the SCCNij3 line with the two image

-55-

DISCUSSION

analysis systems was made for control tumors in two different double hypoxic markerstudies (paper II and III).

The staining protocols and detection of the markers were optimized for thesystem used in each experiment. For the 8-bit system the two markers were stained onconsecutive tissue sections, but were detected with the same fluorochrome. For the 12-bit system, a triple staining for both hypoxic markers and vasculature was done, usingdifferent fluorochromes for the hypoxic markers (green for pimonidazole and red forCCI-103F) to achieve an exact spatial analysis of the distribution of the two hypoxicmarkers. In addition the image processing steps and the resolution were different in thetwo studies. The same definition of the hypoxic fraction (the tumor surface stained forthe hypoxic marker relative to the total tumor surface) was used for the two differentimage analysis systems. Although the differences between the hypoxic fractions in thetwo experimental settings were small, the differences were larger for CCI-103F thanfor pimonidazole. Thus, it can be concluded that pimonidazole is less sensitive tovariations in staining protocols and experimental settings than CCI-103F.

The results in this study show that although the differences are not dramatic,different experimental settings affect the hypoxic fractions. In addition there areseveral different definitions of hypoxic fractions in the literature further complicatingcomparisons of absolute values between different studies. Therefore comparisons ofabsolute values must be done with caution, while comparisons of relative differencessuch as the ranking of hypoxic fractions or relative differences between control tumorsand treated tumors between the different systems are less problematic.

Future AimsThe ultimate goal of ongoing research is to achieve individualized treatments

based on the vascular architecture and microenvironmental parameters. The resultspresented in this thesis is one step forward towards understanding the role of thedynamics of the tumor microenvironment in the response to anti-cancer treatments.

In the studies in paper II and paper IV the proliferation marker BrdUrd wasinjected together with pimonidazole, and the proliferation marker IdUrd was injectedtogether with CCI-103F. A triple staining protocol of hypoxia, proliferation andvasculature, as well as a triple staining of vasculature, BrdUrd, and IdUrd are currentlyunder development. The relation between hypoxia and proliferation will be analyzed innon-treated tumors. Furthermore the double hypoxic marker method will be used to

-56-

DISCUSSION

analyze the effects of cytotoxic drugs on the balance between proliferating and hypoxicsubpopulations.

The double hypoxic marker method is a robust tool that can be used to designand test the efficiency and toxicity of clinically relevant complementary treatmentstrategies, an will be used to further explore the relationships between the dynamics ofthe tumor microenvironment and the effect of different treatment modalities.

-57-

CONCLUSIONS

CONCLUSIONS

The tumor microenvironment is characterized by an aberrant and dysfunctionalvascular supply that may lead to hypoxic regions. Generally a gradient of oxygentensions is present in the tumor tissue, with relatively oxic proliferating cells near thevasculature. At increasing distances the level of tissue hypoxia increases, with acorresponding decrease of the proliferation rate. Beyond a critical threshold in theoxygen and nutritional gradients necrotic cells are found.

In this study the tumor microenvironment was analyzed with a multi parametermethod including the vasculature, proliferating BrdUrd-labeled cells and hypoxic cellslabeled with pimonidazole or CCI-103F. A double hypoxic marker assay wasdeveloped to study the dynamics of the hypoxic cell population in experimental tumormodels.

From the papers included in this thesis the following conclusions can be drawn:

I. Characterization of 11 human head and neck cancer xenografts revealed twomain hypoxic patterns: patchy and ribbon-like. These seem to be correlated withthe vascular architecture. However, no difference of hypoxia as a function ofdistance to nearest blood vessel was found. Moreover, the hypoxic patterns werenot related to tumor differentiation grade. There was a trend towards a higherhypoxic fraction, and BrdUrd labeling index in the ribbon-like tumors Inaddition, there was a larger amount of necrosis in these tumors. Thus thehypoxic pattern may reflect the dynamics of the tumor microenvironment.

II. The double hypoxic marker method developed in this study was used to studythe half-life of hypoxic tumor cells. These were labeled with pimonidazole atdifferent time intervals and with CCI-103F 2.5 h before harvest. The ratiosbetween the hypoxic fractions of these two markers were used to determine thehalf-lives of the hypoxic cells in three different tumor lines. The half-life of thehypoxic tumor lines cells ranged from 17h to 49 h. The ranking of the hypoxichalf-lives corresponded with the ranking of the potential tumor doubling time,but not with the actual tumor volume doubling times.

III. The double hypoxic marker assay was used to study changes caused byoxygenation modifying treatments in the SCCNij3 line. Carbogen (95% O2 and

-59-

CONCLUSIONS

5%CO2) caused a decrease of hypoxia, which was most pronounced atintermediate distances from tumor vessels. Hydralazine, a vasodilator thatdiverts the blood flow from the tumor vasculature to the surrounding normaltissue was used to show that both increased and decreased levels of hypoxiacould be measured with this assay.

IV. The double hypoxic marker method was used to study the effect of the hypoxiccytotoxin tirapazamine on the hypoxic tumor cell population. A decrease in thehypoxic fraction was seen shortly after tirapazamine injection. The nadir wasreached between 1-3.5 h after tirapazamine administration. The reoxygenationwas followed by a rehypoxification, and at 48h after tirapazamine injection thehypoxic fractions were almost back to control levels.

The three studies with the double hypoxic marker assay have shown this to be arobust method. It could be used to measure both dynamic changes of the hypoxiccell population in non-treated tumors, as well as the effects of oxygen modifyingtreatments and the effect of cytotoxic drugs on hypoxic cell populations withhigh spatial resolution relative to the tissue architecture.

-60-

ACKNOWLEDGEMENTS

ACKNOWLEDGEMENTS

This work would not have been accomplished without the help and support fromcolleagues, friends, and family. I would like to use this opportunity to thank you all forthe help and I would like to emphasize a few of you here.

I especially want to thank my supervisor professor Bert van der Kogel for the advice,constructive criticism during this project and for letting me work in Nijmegen andprofessor Anders Bergh for the support. I want to thank professor Julie Denekamp (inmemorandum) who first encouraged me to study the field of radiobiology. Hercontribution through discussions and advice has been invaluable for me.

This work was performed both at the Departments of Radiation Sciences, Umeå,Clinical Bioemdical Research the unit for Pathology Umeå and at the Department ofRadiation Oncology, University Medical Center Nijmegen, The Netherlands. I wouldlike to thank all of you that have been my colleges during this work. I especially wantto mention some of you.

Jan Bussink, for the feedback on experimental designs and critical reading of themanuscripts.

Paul Rijken for stimulating discussions and the continuous development of the imageanalysis systems, and for being a supportive roommate during the years in Nijmegen.

Hans Kaanders for providing the primary material for establishment of the xenograftlines and stimulating co-autorship.

Hans Peters, for development of immunohistochemical staining protocols. Jasper Lokand Wenny Peeters, for all help with stainings and scanning.

Marielle Phillipens for all the support.

Geert Poelen, Bianca Lemmers and Debbie Smits at the Central Animal Laboratory,Nijmegen for excellent animal care.

Professor Jim Raleigh and professor Adrian Begg, for co-authorship and stimulatingcollaborations.

-61-

ACKNOWLEDGEMENTS

Professor emeritus Bo Littbrand for giving me the opportunity to do this work and forthe support.

Among the colleagues in Umeå, I especially want to mention Christina Bergström,Iuliana and Alexandru Dasu, and Irina Goritskaya.

Katrin Sundh, Cecilia Elofsson, and Anna Wernblom provided excellent administrativeservice.

Last but not the least I want to thank my family. Especially my mother Inger for allsupport and long phone calls, my father Stig, and my brother Peter, you are alwaysthere for me when I need you. Without the stable basis and all support you have givenme I would not have made it.

Funding was kindly provided from the Cancer Research Foundation in NorthernSweden, the Sir Samuel Scott of Yews Trust, the United Kingdom, and the DutchCancer Society.

-62-

All my friends in Sweden, the Netherlands, and the rest of the world for the supportduring my scientific and personal ups and downs. Ylva Hedberg, Christer Mehle,Maria Nikolova and Charlotte Thysell though far away always close to me.Mariska Drenth for the support and introduction to the Netherlands. Chris Gerristfor encouragement and consideration.

REFERENCES

REFERENCES

Aebersold, D.M., Beer, K.T., Laissue, J., Hug, S., Kollar, A., Greiner, R.H., Djonov,V., Murray, J.D., Carlson, G.W., McLaughlin, K., Pennington, M., Lynn, M.,DeRose, P.B., Williams, J.K. & Cohen, C. (2000). Intratumoral microvessel densitypredicts local treatment failure of radically irradiated squamous cell cancer of theoropharynx. Int J Radiat Oncol Biol Phys, 48, 17-25.

Aebersold, D.M., Burri, P., Beer, K.T., Laissue, J., Djonov, V., Greiner, R.H. &Semenza, G.L. (2001). Expression of hypoxia-inducible factor-1alpha: a novelpredictive and prognostic parameter in the radiotherapy of oropharyngeal cancer.Cancer Res, 61, 2911-6.

Airley, R., Loncaster, J., Davidson, S., Bromley, M., Roberts, S., Patterson, A., Hunter,R., Stratford, I. & West, C. (2001). Glucose transporter glut-1 expression correlateswith tumor hypoxia and predicts metastasis-free survival in advanced carcinoma ofthe cervix. Clin Cancer Res, 7, 928-34.

Akakura, N., Kobayashi, M., Horiuchi, I., Suzuki, A., Wang, J., Chen, J., Niizeki, H.,Kawamura, K.-i., Hosokawa, M. & Asaka, M. (2001). Constitutive Expression ofHypoxia-inducible Factor-1{alpha} Renders Pancreatic Cancer Cells Resistant toApoptosis Induced by Hypoxia and Nutrient Deprivation. Cancer Res, 61, 6548-6399.

Albo, D., Granick, M.S., Jhala, N., Atkinson, B. & Solomon, M.P. (1994). Therelationship of angiogenesis to biological activity in human squamous cellcarcinomas of the head and neck. Annals of plastic surgery, 32(6), 588-94.

Åmellem, O., Löffler, M. & Pettersen, E.O. (1994). Regulation of cell proliferationunder extreme and moderate hypoxia: the role of pyrimidine (deoxy)nucleotides. BrJ Cancer, 70, 857-66.

Arteel, G.E., Thurman, R.G., Yates, J.M. & Raleigh, J.A. (1995). Evidence thathypoxia markers detect oxygen gradients in liver: pimonidazole and retrogradeperfusion of rat liver. Br J Cancer, 72, 889-95.

-63-

REFERENCES

Azuma, C., Raleigh, J.A. & Thrall, D.E. (1997). Longevity of pimonidazole adducts inspontaneous canine tumors as an estimate of hypoxic cell lifetime. Radiat Res, 148,35-42.

Begg, A.C., Haustermans, K., Hart, A.A., Dische, S., Saunders, M., Zackrisson, B.,Gustaffson, H., Coucke, P., Paschoud, N., Hoyer, M., Overgaard, J., Antognoni, P.,Richetti, A., Bourhis, J., Bartelink, H., Horiot, J.C., Corvo, R., Giaretti, W.,Awwad, H., Shouman, T., Jouffroy, T., Maciorowski, Z., Dobrowsky, W.,Struikmans, H., Wilson, G.D. & et al. (1999). The value of pretreatment cell kineticparameters as predictors for radiotherapy outcome in head and neck cancer: amulticenter analysis. Radiother Oncol, 50, 13-23.

Begg, A.C., McNally, N.J., Shrieve, D.C. & Karcher, H. (1985). A method to measurethe duration of DNA synthesis and the potential doubling time from a singlesample. Cytometry, 6, 620-6.

Bennett, M.H., Wilson, G.D., Dische, S., Saunders, M.I., Martindale, C.A., Robinson,B.M., O'Halloran, A.E., Leslie, M.D. & Laing, J.H. (1992). Tumour proliferationassessed by combined histological and flow cytometric analysis: implications fortherapy in squamous cell carcinoma in the head and neck. Br J Cancer, 65, 870-8.

Bernsen, H.J.J.A., Rijken, P.F.J.W., Oostendorp, T. & van der Kogel, A.J. (1995).Vascularity and perfusion of human gliomas xenografted in the athymic nudemouse. Br J Cancer, 71, 721-6.

Bernsen, H.J.J.A., Rijken, P.F.J.W., Peters, J.P.W., Bakker, H. & van der Kogel, A.J.(1998). Delayed vascular changes after antiangiogenic therapy with antivascularendothelial growth factor antibodies in human glioma xenografts in nude mice.Neurosurgery, 43, 570-5.

Brizel, D.M., Sibley, G.S., Prosnitz, L.R., Scher, R.L. & Dewhirst, M.W. (1997).Tumor hypoxia adversely affects the prognosis of carcinoma of the head and neck.Int J Radiat Oncol Biol Phys, 38, 285-9.

Brown, J.M. (1979). Evidence for acutely hypoxic cells in mouse tumours, and apossible mechanism of reoxygenation. Br J Radiol, 52, 650-6.

-64-

REFERENCES

Brown, J.M. (1993). SR 4233 (tirapazamine): a new anticancer drug exploitinghypoxia in solid tumours. Br J Cancer, 67, 1163-70.

Brown, J.M. & Siim, B.G. (1996). Hypoxia-Specific Cytotoxins in Cancer Therapy.Semin Radiat Oncol, 6, 22-36.

Bussink, J., Kaanders, J.H.A.M., Rijken, P.F.J.W., Martindale, C.A. & van der Kogel,A.J. (1998). Multiparameter analysis of vasculature, perfusion and proliferation inhuman tumour xenografts. Br J Cancer, 77, 57-64.

Bussink, J., Kaanders, J.H.A.M., Rijken, P.F.J.W., Peters, J.P.W., Hodgkiss, R.J.,Marres, H.A. & van der Kogel, A.J. (1999). Vascular architecture andmicroenvironmental parameters in human squamous cell carcinoma xenografts:effects of carbogen and nicotinamide. Radiother Oncol, 50, 173-84.

Bussink, J., Kaanders, J.H.A.M., Rijken, P.F.J.W., Raleigh, J.A. & Van der Kogel, A.J.(2000a). Changes in blood perfusion and hypoxia after irradiation of a humansquamous cell carcinoma xenograft tumor line. Radiat Res, 153, 398-404.

Bussink, J., Kaanders, J.H.A.M., Strik, A.M. & van der Kogel, A.J. (2000b). Effects ofnicotinamide and carbogen on oxygenation in human tumor xenografts measuredwith luminescense-based fiber optic probes. Radiother Oncol, 57, 21-30.

Bussink, J., Kaanders, J.H.A.M., Strik, A.M., Vojnovic, B. & van der Kogel, A.J.(2000c). Optical sensor-based oxygen tension measurements correspond withhypoxia marker binding in three human tumor xenograft lines. Radiat Res, 154,547-555.

Carmeliet, P., Dor, Y., Herbert, J.M., Fukumura, D., Brusselmans, K., Dewerchin, M.,Neeman, M., Bono, F., Abramovitch, R., Maxwell, P., Koch, C.J., Ratcliffe, P.,Moons, L., Jain, R.K., Collen, D., Keshert, E. & Keshet, E. (1998). Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis.Nature, 394, 485-90.

Chang, C.D., Ottavio, L., Travalli, S., Lipson, K.E. & Baserga, E. (1990).Transcriptional and posttranscriptional regulation of the proliferating cell nuclearantigen gene. Mol Cell Biol, 10, 3289-3296.

-65-

REFERENCES

Chapman, J.D., Franko, A.J. & Sharplin, J. (1981). A marker for hypoxic cells intumours with potential clinical applicability. Br J Cancer, 43, 546-50.

Cline, J.M., Rosner, G.L., Raleigh, J.A. & Thrall, D.E. (1997). Quantification of CCI-103F labeling heterogeneity in canine solid tumors. Int J Radiat Oncol Biol Phys,37, 655-62.

Cline, J.M., Thrall, D.E., Page, R.L., Franko, A.J. & Raleigh, J.A. (1990).Immunohistochemical detection of a hypoxia marker in spontaneous caninetumours. Br J Cancer, 62, 925-31.

Cline, J.M., Thrall, D.E., Rosner, G.L. & Raleigh, J.A. (1994). Distribution of thehypoxia marker CCI-103F in canine tumors. Int J Radiat Oncol Biol Phys, 28, 921-33.

Collingridge, D.R., Young, W.K., Vojnovic, B., Wardman, P., Lynch, E.M., Hill, S.A.& Chaplin, D.J. (1997). Measurement of tumor oxygenation: a comparison betweenpolarographic needle electrodes and a time-resolved luminescence-based opticalsensor. Radiat Res, 147, 329-34.

Debbage, P.L., Griebel, J., Ried, M., Gneiting, T., DeVries, A. & Hutzler, P. (1998).Lectin intravital perfusion studies in tumor-bearing mice: micrometer-resolution,wide-area mapping of microvascular labeling, distinguishing efficiently andinefficiently perfused microregions in the tumor. J Histochem Cytochem, 46, 627-39.

Denekamp, J. (1992). Inadequate vasculature in solid tumours: consequences forcancer research strategies. BJR Suppl, 24, 111-7.

Dethlefsen, L.A., Prewitt, J.M. & Mendelsohn, M.L. (1968). Analysis of tumor growthcurves. J Natl Cancer Inst, 40, 389-405.

Dewhirst, M.W., Kimura, H., Rehmus, S.W., Braun, R.D., Papahadjopoulos, D., Hong,K. & Secomb, T.W. (1996). Microvascular studies on the origins of perfusion-limited hypoxia. Br J Cancer Suppl, 27, S247-51.

-66-

REFERENCES

Dewhirst, M.W., Ong, E.T., Braun, R.D., Smith, B., Klitzman, B., Evans, S.M. &Wilson, D. (1999). Quantification of longitudinal tissue pO2 gradients in windowchamber tumours: impact on tumour hypoxia. British journal of cancer, 79, 1717-22.

Dray, T.G., Hardin, N.J. & Sofferman, R.A. (1995). Angiogenesis as a prognosticmarker in early head and neck cancer. Ann Otol Rhinol Laryngol, 104, 724-9.

Durand, R.E. (2001). Intermittent blood flow in solid tumours--an under-appreciatedsource of 'drug resistance'. Cancer Metastasis Rev, 20, 57-61.

Durand, R.E. & LePard, N.E. (1995). Contribution of transient blood flow to tumourhypoxia in mice. Acta oncologica, 34, 317-23.

Durand, R.E. & LePard, N.E. (1997). Tumour blood flow influences combinedradiation and doxorubicin treatments. Radiother Oncol, 42, 171-9.

Durand, R.E. & Raleigh, J.A. (1998). Identification of nonproliferating but viablehypoxic tumor cells in vivo. Cancer Res, 58, 3547-50.

Durand, R.E. & Sham, E. (1998). The lifetime of hypoxic human tumor cells. Int JRadiat Oncol Biol Phys, 42, 711-5.

Ellwart, J. & Dörmer, P. (1985). Effect of 5-fluoro-2'-deoxyuridine (FdUrd) on 5-bromo-2'-deoxyuridine (BrdUrd) incorporation into DNA measured with amonoclonal BrdUrd antibody and by the BrdUrd/Hoechst quenching effect.Cytometry, 6, 513-520.

Evans, S.M., Hahn, S., Pook, D.R., Jenkins, W.T., Chalian, A.A., Zhang, P., Stevens,C., Weber, R., Weinstein, G., Benjamin, I., Mirza, N., Morgan, M., Rubin, S.,McKenna, W.G., Lord, E.M. & Koch, C.J. (2000). Detection of hypoxia in humansquamous cell carcinoma by EF5 binding. Cancer Res, 60, 2018-24.

Evans, S.M., Joiner, B., Jenkins, W.T., Laughlin, K.M., Lord, E.M. & Koch, C.J.(1995). Identification of hypoxia in cells and tissues of epigastric 9L rat gliomausing EF5 [2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl)acetamide]. Br J Cancer, 72, 875-82.

-67-

REFERENCES

Fenton, B.M. (1997). Effects of carbogen plus fractionated irradiation on KHT tumoroxygenation. Radiother Oncol, 44, 183-90.

Fenton, B.M., Kiani, M.F. & Siemann, D.W. (1995). Should direct measurements oftumor oxygenation relate to the radiobiological hypoxic fraction of a tumor? Int JRadiat Oncol Biol Phys, 33, 365-73.

Fenton, B.M., Paoni, S.F., Lee, J., Koch, C.J. & Lord, E.M. (1999). Quantification oftumour vasculature and hypoxia by immunohistochemical staining and HbO2saturation measurements. Br J Cancer, 79, 464-71.

Folkman, J. (1976). The vascularization of tumors. Sci Am, 234, 58-64, 70-3.

Franko, A.J. & Sutherland, R.M. (1978). Rate of death of hypoxic cells in multicellspheroids. Radiat Res, 76, 561-72.

Gasparini, G., Weidner, N., Maluta, S., Pozza, F., Boracchi, P., Mezzetti, M., Testolin,A. & Bevilacqua, P. (1993). Intratumoral microvessel density and p53 protein:correlation with metastasis in head-and-neck squamous-cell carcinoma. Int JCancer, 55, 739-44.

Gatenby, R.A., Kessler, H.B., Rosenblum, J.S., Coia, L.R., Moldofsky, P.J., Hartz,W.H. & Broder, G.J. (1988). Oxygen distribution in squamous cell carcinomametastases and its relationship to outcome of radiation therapy. Int J Radiat OncolBiol Phys, 14, 831-8.

Gavrieli, Y., Sherman, Y. & Ben Sasson, S.A. (1992). Identification of programmedcell death in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol,119, 493-501.

Giatromanolaki, A., Koukourakis, M.I., Sivridis, E., Pastorek, J., Wykoff, C.C., Gatter,K.C. & Harris, A.L. (2001). Expression of hypoxia-inducible carbonic anhydrase-9relates to angiogenic pathways and independently to poor outcome in non-small celllung cancer. Cancer Res, 61, 7992-8.

Grätzner, H.G. (1982). Monoclonal antibody to 5-bromo- and 5-iododeoxyuridine: Anew reagent for detection of DNA replication. Science, 218, 474-5.

-68-

REFERENCES

Gray, L., Conger, A.D., Ebert, M., Hornsey, S. & O.C.A., S. (1953). The concentrationof oxygen dissolved in tissues at the time of irradiation as a factor in radiotherapy.Br. J. Radiol., 28, 638-648.

Gross, M.W., Karbach, U., Groebe, K., Franko, A.J. & Mueller Klieser, W. (1995).Calibration of misonidazole labeling by simultaneous measurement of oxygentension and labeling density in multicellular spheroids. Int J Cancer, 61, 567-73.

Henk, J.M. & Smith, C.W. (1977). Radiotherapy and hyperbaric oxygen in head andneck cancer. Interim report of second clinical trial. Lancet, 2, 104-5.

Hirst, D.G. & Denekamp, J. (1979). Tumour cell proliferation in relation to thevasculature. Cell Tissue Kinet, 12, 31-42.

Höckel, M., Knoop, C., Schlenger, K., Vorndran, B., Baussmann, E., Mitze, M.,Knapstein, P.G. & Vaupel, P. (1993). Intratumoral pO2 predicts survival inadvanced cancer of the uterine cervix. Radiother Oncol, 26, 45-50.

Höckel, M. & Vaupel, P. (2001a). Biological consequences of tumor hypoxia.Seminars in Oncology, 28 (2) Suppl. 8, 36-41.

Höckel, M. & Vaupel, P. (2001b). Tumor hypoxia: Definitions and current clinical,biologic, and molecular aspects. J. Natl Cancer Inst., 93, 266-276.

Hodgkiss, R.J., Jones, G., Long, A., Parrick, J., Smith, K.A., Stratford, M.R. & Wilson,G.D. (1991). Flow cytometric evaluation of hypoxic cells in solid experimentaltumours using fluorescence immunodetection. Br J Cancer, 63, 119-25.

Holthausen, H. (1921). Beitrage zur biologie der strahlenwirkun. Untersuchungen anaskorideneirem. Pflügers Arch, 187, 1-24.

Iyer, R.V., Kim, E., Schneider, R.F. & Chapman, J.D. (1998). A dual hypoxic markertechnique for measuring oxygenation change within individual tumors. Br JCancer, 78, 163-9.

Janot, F., Klijanienko, J., Russo, A., Mamet, J.P., de Braud, F., El Naggar, A.K.,Pignon, J.P., Luboinski, B. & Cvitkovic, E. (1996). Prognostic value of

-69-

REFERENCES

clinicopathological parameters in head and neck squamous cell carcinoma: aprospective analysis. Br J Cancer, 73(4), 531-8.

Jenssen, N., Boysen, M., Kjaerheim, A. & Bryne, M. (1996). Low vascular densityindicates poor response to radiotherapy in small glottic carcinomas. Pathol ResPract, 192, 1090-4.

Jin, G.Y., Li, S.J., Moulder, J.E. & Raleigh, J.A. (1990). Dynamic measurements ofhexafluoromisonidazole (CCI-103F) retention in mouse tumours by 1H/19Fmagnetic resonance spectroscopy. Int J Radiat Biol, 58, 1025-34.

Kaanders, J.H., Pop, L.A., Marres, H.A., Liefers, J., van den Hoogen, F.J., van Daal,W.A. & van der Kogel, A.J. (1998). Accelerated radiotherapy with carbogen andnicotinamide (ARCON) for laryngeal cancer. Radiother Oncol, 48, 115-22.

Kaanders, J.H.A.M., Pop, L.A.M., Marres, H.A.M., Bruaset, I., van den Hoogen,F.J.A., Merkx, M.A.W. & van der Kogel, A.J. (2002). ARCON: experience in 215patients with advanced head-and-neck cancer. Int J Radiat Oncol Biol Phys, 52,769-78.

Kaanders, J.H.A.M., Pop, L.A.M., Marres, H.A.M., van der Mazen, R.W.M., van derKogel, A.J. & van Daal, W.A.J. (1995). Radiotherapy with carbogen breathing andnicotinamide in head and neck cancer:feasability and toxicity. Radiother. Oncol.,37, 190-198.

Kallman, R.F. & Dorie, M.J. (1986). Tumor oxygenation and reoxygenation duringradiation therapy: their importance in predicting tumor response. Int J Radiat OncolBiol Phys, 12, 681-5.

Kennedy, A.S., Raleigh, J.A., Perez, G.M., Calkins, D.P., Thrall, D.E., Novotny, D.B.& Varia, M.A. (1997). Proliferation and hypoxia in human squamous cellcarcinoma of the cervix: first report of combined immunohistochemical assays. IntJ Radiat Oncol Biol Phys, 37, 897-905.

Kerbel, R.S., Viloria Petit, A., Okada, F. & Rak, J. (1998). Establishing a link betweenoncogenes and tumor angiogenesis. Mol Med, 4, 286-95.

-70-

REFERENCES

Kerbel, R.S., Yu, J., Tran, J., Man, S., Viloria Petit, A., Klement, G., Coomber, B.L. &Rak, J. (2001). Possible mechanisms of acquired resistance to anti-angiogenicdrugs: implications for the use of combination therapy approaches. CancerMetastasis Rev, 20, 79-86.

Kim, I.H. & Brown, J.M. (1994). Reoxygenation and rehypoxiation in the SCCVIImouse tumor [see comments]. Int J Radiat Oncol Biol Phys, 29, 493-7.

Koch, C.J., Evans, S.M. & Lord, E.M. (1995). Oxygen dependence of cellular uptakeof EF5 [2-(2-nitro-1H-imidazol-1-yl)-N-(2,2,3,3,3-pentafluoropropyl)a cet amide] :analysis of drug adducts by fluorescent antibodies vs bound radioactivity. Br JCancer, 72, 869-74.

Koch, C.J., Kruuv, J. & Frey, H.E. (1973a). The effect of hypoxia on the generationtime of mammalian cells. Radiat Res, 53, 43-8.

Koch, C.J., Kruuv, J., Frey, H.E. & Snyder, R.A. (1973b). Plateau phase in growthinduced by hypoxia. Int J Radiat Biol Relat Stud Phys Chem Med, 23, 67-74.

Konerding, M.A., Malkusch, W., Klapthor, B., van Ackern, C., Fait, E., Hill, S.A.,Parkins, C., Chaplin, D.J., Presta, M. & Denekamp, J. (1999). Evidence forcharacteristic vascular patterns in solid tumours: quantitative studies usingcorrosion casts. Br J Cancer, 80, 724-32.

Konerding, M.A., Miodonski, A.J. & Lametschwandtner, A. (1995). Microvascularcorrosion casting in the study of tumor vascularity: a review. Scanning Microsc, 9,1233-43.

Konerding, M.A., van Ackern, C., Fait, E., Steinberg, F. & Streffer, C. (1998).Morphological aspects of tumor angiogenseis and microcirculation. In Bloodperfusion and microenvironmnet of human tumors, Molls, M. & Vaupel, P. (eds)pp. 6-17. Springer Verlag: Berlin Heidelberg.

Koukourakis, M.I., Giatromanolaki, A., Sivridis, E. & Fezoulidis, I. (2000). Cancervascularization: implications in radiotherapy? Int J Radiat Oncol Biol Phys, 48,545-53.

-71-

REFERENCES

Krajewska, M., Wang, H.G., Krajewski, S., Zapata, J.M., Shabaik, A., Gascoyne, R. &Reed, J.C. (1997). Immunohistochemical analysis of in vivo patterns of expressionof CPP32 (Caspase-3), a cell death protease. Cancer Res, 57, 1605-13.

Kyle, A.H. & Minchinton, A.I. (1999). Measurement of delivery and metabolism oftirapazamine to tumour tissue using the multilayered cell culture model. CancerChemother Pharmacol, 43, 213-20.

Lal, A., Peters, H., St. Croix, B., Haroon, Z.A., Dewhirst, M.W., Strausberg, R.L.,Kaanders, J.H.A.M., van der Kogel, A.J. & Riggins, G.J. (2001). TranscriptionalResponse to Hypoxia in Human Tumors. J Natl Cancer Inst, 93, 1337-1343.

Leedy, D.A., Trune, D.R., Kronz, J.D., Weidner, N. & Cohen, J.I. (1994). Tumorangiogenesis, the p53 antigen, and cervical metastasis in squamous carcinoma ofthe tongue. Otolaryngol Head Neck Surg, 111, 417-22.

Levine, E.L., Renehan, A., Gossiel, R., Davidson, S.E., Roberts, S.A., Chadwick, C.,Wilks, D.P., Potten, C.S., Hendry, J.H., Hunter, R.D. & et al. (1995). Apoptosis,intrinsic radiosensitivity and prediction of radiotherapy response in cervicalcarcinoma. Radiother Oncol, 37, 1-9.

Littbrand, B. & Revesz, L. (1968). Survival of cells in anoxia. Br J Radiol, 41, 479-80.

Loncaster, J.A., Harris, A.L., Davidson, S.E., Logue, J.P., Hunter, R.D., Wycoff, C.C.,Pastorek, J., Ratcliffe, P.J., Stratford, I.J. & West, C.M.L. (2001). CarbonicAnhydrase (CA IX) Expression, a Potential New Intrinsic Marker of Hypoxia:Correlations with Tumor Oxygen Measurements and Prognosis in LocallyAdvanced Carcinoma of the Cervix. Cancer Res, 61, 6394-6399.

Long, A., Parrik, J. & Hodgkiss, R. (1991). An efficient procedure for the 1-alkylationof 2-nitroimidazoles and the synthesis of a probe for hypoxia in solid tumours.Synthesis, 9, 709-713.

Lyng, H., Sundfor, K. & Rofstad, E.K. (2000). Changes in tumor oxygen tensionduring radiotherapy of uterine cervical cancer: relationships to changes in vasculardensity, cell density, and frequency of mitosis and apoptosis. Int J Radiat OncolBiol Phys, 46, 935-46.

-72-

REFERENCES

Mason, R.P., Constantinescu, A., Hunjan, S., Le, D., Hahn, E.W., Antich, P.P., Blum,C. & Peschke, P. (1999). Regional tumor oxygenation and measurement of dynamicchanges. Radiat Res, 152, 239-49.

Moriyama, M., Kumagai, S., Kawashiri, S., Kojima, K., Kakihara, K. & Yamamoto, E.(1997). Immunohistochemical study of tumour angiogenesis in oral squamous cellcarcinoma. Oral Oncol, 33, 369-74.

Mottram, J.C. (1924). On the skin reactions to radium exposure and their avoidance intherapy. Br J Radiol, 29, 174-180.

Mottram, J.C. (1936). A factor of importance in the radiosensitivity of tumours. Br. J.Radiol., 9, 606-614.

Murray, J.D., Carlson, G.W., McLaughlin, K., Pennington, M., Lynn, M., DeRose,P.B., Williams, J.K. & Cohen, C. (1997). Tumor angiogenesis as a prognostic factorin laryngeal cancer. Am J Surg, 174, 523-6.

Nordsmark, M., Overgaard, M. & Overgaard, J. (1996). Pretreatment oxygenationpredicts radiation response in advanced squamous cell carcinoma of the head andneck. Radiother Oncol, 41, 31-9.

Olive, P.L. (1994). Radiation-induced reoxygenation in the SCCVII murine tumour:evidence for a decrease in oxygen consumption and an increase in tumourperfusion. Radiother Oncol, 32, 37-46.

Olive, P.L., Aquino-Parsons, C., MacPhail, S.H., Liao, S.Y., Raleigh, J.A., Lerman,M.I. & Stanbridge, E.J. (2001). Carbonic anhydrase 9 as an endogenous marker forhypoxic cells in cervical cancer. Cancer Res, 61, 8924-9.

Olive, P.L., Chaplin, D.J. & Durand, R.E. (1985). Pharmacokinetics, binding anddistribution of Hoechst 33342 in spheroids and murine tumours. Br J Cancer, 52,739-46.

Olive, P.L. & Durand, R.E. (1987). Characterization of a carbocyanine derivative as afluorescent penetration probe. Cytometry, 8, 571-5.

-73-

REFERENCES

Olive, P.L. & Durand, R.E. (1992). Detection of hypoxic cells in a murine tumor withthe use of the comet assay. J Natl Cancer Inst, 84, 707-11.

Olive, P.L., Durand, R.E., Raleigh, J.A., Luo, C. & Aquino Parsons, C. (2000).Comparison between the comet assay and pimonidazole binding for measuringtumour hypoxia. Br J Cancer, 83, 1525-31.

Overgaard, J., Hansen, H.S., Overgaard, M., Bastholt, L., Berthelsen, A., Specht, L.,Lindelov, B. & Jorgensen, K. (1998). A randomized double-blind phase III study ofnimorazole as a hypoxic radiosensitizer of primary radiotherapy in supraglotticlarynx and pharynx carcinoma. Results of the Danish Head and Neck Cancer Study(DAHANCA) Protocol 5-85 [see comments]. Radiother Oncol, 46, 135-46.

Overgaard, J. & Horsman, M.R. (1996). Modification of Hypoxia-InducedRadioresistance in Tumors by the Use of Oxygen and Sensitizers. Semin RadiatOncol, 6, 10-21.

Peters, G.J., Van Dijk, J., Nadal, J.C., Van Groeningen, C.J., Lankelma, J. & Pinedo,H.M. (1987). Diurnal variation in the therapeutic efficacy of 5-fluorouracil againstmurine colon cancer. In Vivo, 1, 113-7.

Pigott, K.H., Hill, S.A., Chaplin, D.J. & Saunders, M.I. (1996). Microregionalfluctuations in perfusion within human tumours detected using laser Dopplerflowmetry. Radiother Oncol, 40, 45-50.

Powers, W.E. & Tolmach, L.J. (1963). A multicomponent x-ray survival curve formouse lymphosarcoma cells irradiated in vivo. Nature, 197, 710-711.

Raleigh, J.A., Chou, S.C., Arteel, G.E. & Horsman, M.R. (1999). Comparisons amongpimonidazole binding, oxygen electrode measurements, and radiation response inC3H mouse tumors. Radiat Res, 151, 580-9.

Raleigh, J.A., Chou, S.C., Calkins Adams, D.P., Ballenger, C.A., Novotny, D.B. &Varia, M.A. (2000). A clinical study of hypoxia and metallothionein proteinexpression in squamous cell carcinomas. Clin Cancer Res, 6, 855-62.

-74-

REFERENCES

Raleigh, J.A., Franko, A.J., Treiber, E.O., Lunt, J.A. & Allen, P.S. (1986). Covalentbinding of a fluorinated 2-nitroimidazole to EMT-6 tumors in Balb/C mice:detection by F-19 nuclear magnetic resonance at 2.35 T. Int J Radiat Oncol BiolPhys, 12, 1243-5.

Raleigh, J.A., Miller, G.G., Franko, A.J., Koch, C.J., Fuciarelli, A.F. & Kelly, D.A.(1987). Fluorescence immunohistochemical detection of hypoxic cells in spheroidsand tumours. Br J Cancer, 56, 395-400.

Raleigh, J.A., Zeman, E.M., Calkins, D.P., McEntee, M.C. & Thrall, D.E. (1995).Distribution of hypoxia and proliferation associated markers in spontaneous caninetumors. Acta Oncol, 34, 345-9.

Reed, J.C. (1999). Dysregulation of apoptosis in cancer. J Clin Oncol, 17, 2941-53.

Rijken, P.F.J.W., Bernsen, H.J.J.A., Peters, J.P.W., Hodgkiss, R.J., Raleigh, J.A. & vander Kogel, A.J. (2000). Spatial relationship between hypoxia and the (perfused)vascular network in a human glioma xenograft: a quantitative multi-parameteranalysis. Int J Radiat Oncol Biol Phys, 48, 571-82.

Rijken, P.F.J.W., Bernsen, H.J.J.A. & van der Kogel, A.J. (1995). Application of animage analysis system to the quantitation of tumor perfusion and vascularity inhuman glioma xenografts. Microvasc Res, 50, 141-53.

Rijken, P.F.J.W., Peters, J.P.W. & van der Kogel, A.J. (2002). Quantitative analysis ofvarying hypoxic profiles in relation to functional vessels in different human gliomaxenograft lines. Rad Res, 157, 626-632.

Rojas, A., Vojnovic, B., Johns, H., Joiner, M.C., Martindale, C., Fowler, J.F. &Denekamp, J. (1996). Radiosensitisation in normal tissues with oxygen, carbogen ornicotinamide: therapeutic gain comparisons for fractionated x-ray schedules.Radiother Oncol, 39, 53-64.

Salven, P., Heikkila, P., Anttonen, A., Kajanti, M. & Joensuu, H. (1997). Vascularendothelial growth factor in squamous cell head and neck carcinoma: expressionand prognostic significance. Mod Pathol, 10(11), 1128-33.

-75-

REFERENCES

Sarbia, M., Bittinger, F., Porschen, R., Dutkowski, P., Willers, R. & Gabbert, H.E.(1996). Tumor vascularization and prognosis in squamous cell carcinomas of theesophagus. Anticancer Res, 16, 2117-21.

Saunders, M.I., Anderson, P.J., Bennett, M.H., Dische, S., Minchinton, A., Stratford,M.R. & Tothill, M. (1984). The clinical testing of Ro 03-8799--pharmacokinetics,toxicology, tissue and tumor concentrations. Int J Radiat Oncol Biol Phys, 10,1759-63.

Schmaltz, C., Hardenbergh, P.H., Wells, A. & Fisher, D.E. (1998). Regulation ofproliferation-survival decisions during tumor cell hypoxia. Mol Cell Biol, 18, 2845-54.

Schwarz, G. (1909). Ueber desensibilisierung gegen Röntgen- und radium-strahlen.München Med Wchnschr, 56, 1217-1218.

Sheridan, M.T., Cooper, R.A. & West, C.M. (1999). A high ratio of apoptosis toproliferation correlates with improved survival after radiotherapy for cervicaladenocarcinoma. Int J Radiat Oncol Biol Phys, 44, 507-12.

Sheridan, M.T., West, C.M., Cooper, R.A., Stratford, I.J., Logue, J.P., Davidson, S.E.& Hunter, R.D. (2000). Pretreatment apoptosis in carcinoma of the cervix correlateswith changes in tumour oxygenation during radiotherapy. Br J Cancer, 82, 1177-82.

Shpitzer, T., Chaimoff, M., Gal, R., Stern, Y., Feinmesser, R. & Segal, K. (1996).Tumor angiogenesis as a prognostic factor in early oral tongue cancer. ArchOtolaryngol Head Neck Surg, 122(8), 865-8.

Smith, K.A., Hill, S.A., Begg, A.C. & Denekamp, J. (1988). Validation of thefluorescent dye Hoechst 33342 as a vascular space marker in tumours. Br J Cancer,57, 247-53.

Solesvik, O.V., Rofstad, E.K. & Brustad, T. (1982). Vascular structure of five humanmalignant melanomas grown in athymic nude mice. Br J Cancer, 46, 557-67.

-76-

REFERENCES

Spiegel, J.F., Spear, M.A. & Brown, J.M. (1993). Toxicology of daily administrationto mice of the radiation potentiator SR 4233 (WIN 59075). Radiother Oncol, 26,79-81.

Stainsby, W.N., Snyder, B. & Welch, H.G. (1988). A pictographic essay on blood andtissue oxygen transport. Med Sci Sports Exerc, 20, 213-21.

Steel, G.G. (1977). Growth kinetics of tumours. Oxford University Press: Oxford.

Steinberg, F., Konerding, M.A. & Streffer, C. (1990). The vascular architecture ofhuman xenotransplanted tumors: histological, morphometrical, and ultrastructuralstudies. J Cancer Res Clin Oncol, 116, 517-24.

Talks, K.L., Turley, H., Gatter, K.C., Maxwell, P.H., Pugh, C.W., Ratcliffe, P.J. &Harris, A.L. (2000). The expression and distribution of the hypoxia-induciblefactors HIF-1alpha and HIF-2alpha in normal human tissues, cancers, and tumor-associated macrophages. Am J Pathol, 157, 411-21.

Tannock, I.F. (1968). The relation between cell proliferation and the vascular system ina transplanted mouse mammary tumour. Br J Cancer, 22, 258-73.

Thomlinson, R.H. & Gray, L.H. (1955). The histological structure of some human lungcancers and the possible implications for radiotherapy. Br J of Cancer, IX, 539-549.

Trotter, M.J., Chaplin, D.J., Durand, R.E. & Olive, P.L. (1989a). The use offluorescent probes to identify regions of transient perfusion in murine tumors. Int JRadiat Oncol Biol Phys, 16, 931-4.

Trotter, M.J., Chaplin, D.J. & Olive, P.L. (1989b). Use of a carbocyanine dye as amarker of functional vasculature in murine tumours. Br J Cancer, 59, 706-9.

Tsang, R.W., Fyles, A.W., Li, Y., Rajaraman, M.M., Chapman, W., Pintilie, M. &Wong, C.S. (1999a). Tumor proliferation and apoptosis in human uterine cervixcarcinoma I: correlations between variables. Radiother Oncol, 50, 85-92.

Tsang, R.W., Wong, C.S., Fyles, A.W., Levin, W., Manchul, L.A., Milosevic, M.,Chapman, W., Li, Y.Q. & Pintilie, M. (1999b). Tumour proliferation and apoptosis

-77-

REFERENCES

in human uterine cervix carcinoma II: correlations with clinical outcome. RadiotherOncol, 50, 93-101.

Tufto, I. & Rofstad, E.K. (1995). Transient perfusion in human melanoma xenografts.Br J Cancer, 71, 789-93.

Van Putten, L.M. & Kallman, R.F. (1968). Oxygenation status of a transplantabletumor during fractionated radiation therapy. J Natl Cancer Inst, 40, 441-51.

Varia, M.A., Calkins Adams, D.P., Rinker, L.H., Kennedy, A.S., Novotny, D.B.,Fowler, W.C., Jr. & Raleigh, J.A. (1998). Pimonidazole: a novel hypoxia marker forcomplementary study of tumor hypoxia and cell proliferation in cervical carcinoma.Gynecol Oncol, 71, 270-7.

Vaupel, P. (1990). Oxygenation of human tumors. Strahlenther Onkol, 166, 377-86.

Vaupel, P. & Hockel, M. (2003). Tumor oxygenation and its relevance to tumorphysiology and treatment. Adv Exp Med Biol, 510, 45-9.

Vaupel, P., Kallinowski, F. & Okunieff, P. (1989). Blood flow, oxygen and nutrientsupply, and metabolic microenvironment of human tumors: a review. Cancer Res,49, 6449-65.

Vaupel, P., Schlenger, K., Knoop, C. & Höckel, M. (1991). Oxygenation of humantumors: evaluation of tissue oxygen distribution in breast cancers by computerizedO2 tension measurements. Cancer Res, 51, 3316-22.

Vaupel, P., Thews, O. & Hoeckel, M. (2001). Treatment resistance of solid tumors:role of hypoxia and anemia. Med Oncol, 18, 243-59.

Vukovic, V., Haugland, H.K., Nicklee, T., Morrison, A.J. & Hedley, D.W. (2001).Hypoxia-inducible Factor-1{alpha} Is an Intrinsic Marker for Hypoxia in CervicalCancer Xenografts. Cancer Res, 61, 7394-7398.

Walenta, S., Snyder, S., Haroon, Z.A., Braun, R.D., Amin, K., Brizel, D., Mueller-Klieser, W., Chance, B. & Dewhirst, M.W. (2001). Tissue gradients of energy

-78-

REFERENCES

metabolites mirror oxygen tension gradients in a rat mammary carcinoma model.Int J Radiat Oncol Biol Phys, 51, 840-8.

Warenius, H.M. (1980). Identification and separation of mouse and human componentsof heterotransplanted human tumors. In Immunodeficient animals for cancerreserach, S., S. (ed) pp. 207-220. Macmillian press ltd: London and Basingstoke.

Webster, L., Hodgkiss, R.J. & Wilson, G.D. (1995). Simultaneous triple staining forhypoxia, proliferation, and DNA content in murine tumours. Cytometry, 21, 344-51.

Westphal, J.R., van't Hullenaar, R.G., van der Laak, J.A., Cornelissen, I.M.,Schalkwijk, L.J., van Muijen, G.N., Wesseling, P., de Wilde, P.C., Ruiter, D.J. & deWaal, R.M. (1997). Vascular density in melanoma xenografts correlates withvascular permeability factor expression but not with metastatic potential. Br JCancer, 76, 561-70.

Wheeler, J.A., Stephens, L.C., Tornos, C., Eifel, P.J., Ang, K.K., Milas, L., Allen, P.K.& Meyn, R.E., Jr. (1995). ASTRO Research Fellowship: apoptosis as a predictor oftumor response to radiation in stage IB cervical carcinoma. American Society forTherapeutic Radiology and Oncology. Int J Radiat Oncol Biol Phys, 32, 1487-93.

Wijffels, K.I.E.M., Kaanders, J.H.A.M., Marres, H.A.M., Bussink, J., Peters, H.P.W.,Rijken, P.F.J.W., Wilde, P.C.M. & van der Kogel, A.J. (2001). Patterns ofproliferation related to vasculature in human head-and-neck carcinomas before andafter transplanatation in nude mice. Int J Radiat Oncol Biol Phys, 51, 1346-1353.

Wijffels, K.I.E.M., Kaanders, J.H.A.M., Rijken, P.F.J.W., Bussink, J., van denHoogen, F.J., Marres, H.A., de Wilde, P.C., Raleigh, J.A. & van der Kogel, A.J.(2000). Vascular architecture and hypoxic profiles in human head and necksquamous cell carcinomas. Br J Cancer, 83, 674-83.

Wilson, G.D., McNally, N.J., Dische, S., Saunders, M.I., Des Rochers, C., Lewis, A.A.& Bennett, M.H. (1988). Measurement of cell kinetics in human tumours in vivousing bromodeoxyuridine incorporation and flow cytometry. Br J Cancer, 58, 423-31.

-79-

REFERENCES

Withers, H.R., Taylor, J.M. & Maciejewski, B. (1988). The hazard of acceleratedtumor clonogen repopulation during radiotherapy. Acta Oncol, 27, 131-46.

Wouters, B.G. & Brown, J.M. (1997). Cells at intermediate oxygen levels can be moreimportant than the "hypoxic fraction" in determining tumor response to fractionatedradiotherapy. Radiat Res, 147, 541-50.

Wouters, B.G., Koritzinsky, M., Chiu, R.K., Theys, J., Buijsen, J. & Lambin, P.(2003). Modulation of cell death in the tumor microenvironment. Semin RadiatOncol, 13, 31-41.

Wouters, B.G., Weppler, S.A., Koritzinsky, M., Landuyt, W., Nuyts, S., Theys, J.,Chiu, R.K. & Lambin, P. (2002). Hypoxia as a target for combined modalitytreatments. Eur J Cancer, 38, 240-57.

Wykoff, C.C., Beasley, N., Watson, P.H., Campo, L., Chia, S.K., English, R., Pastorek,J., Sly, W.S., Ratcliffe, P. & Harris, A.L. (2001). Expression of the hypoxia-inducible and tumor-associated carbonic anhydrases in ductal carcinoma in situ ofthe breast. Am J Pathol, 158, 1011-9.

Wykoff, C.C., Beasley, N.J.P., Watson, P.H., Turner, K.J., Pastorek, J., Sibtain, A.,Wilson, G.D., Turley, H., Talks, K.L., Maxwell, P.H., Pugh, C.W., Ratcliffe, P.J. &Harris, A.L. (2000). Hypoxia-inducible Expression of Tumor-associated CarbonicAnhydrases. Cancer Res, 60, 7075-7083.

Young, W.K., Vojnovic, B. & Wardman, P. (1996). Measurement of oxygen tension intumours by time-resolved fluorescence. Br J Cancer Suppl, 27, S256-9.

Yu, J.L., Rak, J.W., Carmeliet, P., Nagy, A., Kerbel, R.S. & Coomber, B.L. (2001).Heterogeneous vascular dependence of tumor cell populations. Am J Pathol, 158,1325-34.

Zagzag, D., Zhong, H., Scalzitti, J.M., Laughner, E., Simons, J.W. & Semenza, G.L.(2000). Expression of hypoxia-inducible factor 1alpha in brain tumors: associationwith angiogenesis, invasion, and progression. Cancer, 88, 2606-18.

-80-

REFERENCES

Zätterström, U.K., Brun, E., Willen, R., Kjellen, E. & Wennerberg, J. (1995). Tumorangiogenesis and prognosis in squamous cell carcinoma of the head and neck. HeadNeck, 17, 312-8.

Zeman, E.M., Calkins, D.P., Cline, J.M., Thrall, D.E. & Raleigh, J.A. (1993). Therelationship between proliferative and oxygenation status in spontaneous caninetumors. Int J Radiat Oncol Biol Phys, 27, 891-8.

Zhong, H., De Marzo, A.M., Laughner, E., Lim, M., Hilton, D.A., Zagzag, D.,Buechler, P., Isaacs, W.B., Semenza, G.L. & Simons, J.W. (1999). Overexpressionof Hypoxia-inducible Factor 1{{alpha}} in Common Human Cancers and TheirMetastases. Cancer Res, 59, 5830-5835.

-81-

UMEÅ UNIVERSITY MEDICAL DISSERTATIONS

New Series No. 835

ISSN 0346-6612 • ISBN 91-7305-432-1

Edited by the Dean of the Faculty of Medicine and Oncology

Department of Radiation Sciences, OncologyClinical Biomedical Research, Pathology

Umeå Universitywww.umu.se

the dynamics and modification of...

Documents