Cytometry Part B (Clinical Cytometry) 70B:4555 (2006)

Original Articles

Comparison Between Pimonidazole Binding,

Oxygen Electrode Measurements, and
Expression of Endogenous Hypoxia Markers
in Cancer of the Uterine Cervix
B. Jankovic,1 C. Aquino-Parsons,1 J. A. Raleigh,2 E. J. Stanbridge,3 R. E. Durand,1 J. P. Banath,1
S. H. MacPhail,1 and P. L. Olive1*
1

Medical Biophysics Department, British Columbia Cancer Agency Research Centre, Vancouver, British Columbia, Canada
2
Department of Radiation Oncology, University of North Carolina, Chapel Hill, North Carolina, USA
3
Department of Microbiology and Molecular Genetics, University of California, Irvine, California, USA

Background: Although tumor hypoxia has been associated with a more aggressive phenotype and lower
cure rate, there is no consensus as to the method best suited for routine measurement. Binding of the
chemical hypoxia marker, pimonidazole, and expression of the endogenous hypoxia markers HIF-1a and
CAIX were compared for their ability to detect hypoxia in tumor biopsies from 67 patients with advanced
carcinoma of the cervix.
Methods: Two biopsies were taken one day after administration of pimonidazole and were analyzed for
pimonidazole binding using ow cytometry or immunohistochemistry. CAIX and HIF-1a expression and
degree of colocalization were measured in sequential antibody-stained sections. Patient subsets were
examined for tumor oxygen tension using an Eppendorf electrode, S phase DNA content, or change in HIF1a expression over the course of treatment.
Results: Approximately 6% of the tumor area stained positive for pimonidazole, HIF-1a, or CAIX. The
CAIX positive fraction correlated with the pimonidazole positive fraction (r = 0.60). Weaker but signicant
correlations were observed between pimonidazole and HIF-1a (r = 0.31) and CAIX and HIF-1a (r = 0.41).
Taking the extent of marker colocalization into consideration increased the condence that all markers
were identifying hypoxic regions. Over 65% of stained areas showed a high degree of colocalization with
the other markers. Oxygen microelectrode measurements and S phase fraction were not correlated with the
hypoxic fraction measured using the three hypoxia markers. HIF-1a levels tended to decrease with time after the start of therapy.
Conclusions: Endogenous hypoxia marker binding shows reasonable agreement, in extent and location,
with binding of pimonidazole. CAIX staining pattern is a better match to the pimonidazole staining pattern
than is HIF-1a, and high CAIX expression in the absence (or low levels) of HIF-1a may indicate a different
biology. q 2006 International Society for Analytical Cytology
Key terms: hypoxia markers; oxygen electrode; CAIX; HIF-1a; pimonidazole; cervical cancer

Hypoxia that develops in many solid tumors is a critical

factor limiting the success of conventional radiation therapy. Hypoxic cells are also less accessible to nutrients and
drugs, more likely to be noncycling, and therefore resistant to many forms of chemotherapy (1). Hypoxia has
been shown to be a driving force in tumor angiogenesis
and has been implicated in promotion of metastasis and
genomic instability (26). The weight of evidence indicating the importance of hypoxia in tumor development,
progression, and response to treatment is now undeniable. Yet in spite of its obvious importance, the presence
of hypoxia in individual human solid tumors is not rou-

q 2006 International Society for Analytical Cytology

tinely measured. This can be attributed in large part to

the limitations associated with the application of prospective, invasive methods such as oxygen microelectrodes

*Correspondence to: Peggy L. Olive, Medical Biophysics Department,

British Columbia Cancer, Research Centre, 675 W. 10th Ave., Vancouver,
British Columbia V5Z 1L3, Canada.
E-mail: polive@bccrc.ca
Received 19 July 2005; Accepted 17 October 2005
Published online 2 February 2006 in Wiley InterScience (www.
interscience.wiley.com).
DOI: 10.1002/cyto.b.20086

46

JANKOVIC ET AL.

and extrinsic hypoxia markers. Endogenous protein markers

have been identied that have the potential to allow for the
implementation of routine measurement of tumor hypoxia
in the clinic (7). However, application of these markers requires careful validation against established methods, especially when many other factors could complicate the use of
hypoxia responsive gene expression as an indication of tumor hypoxia (8,9).
The expression of more than 70 genes is altered under
hypoxic conditions as a result of the change in stability of
a critical transcription factor called hypoxia inducible factor-1 (HIF-1) (10). The HIF-1a subunit of this heterodimer
undergoes stabilization when the oxygen concentration
drops below about 2%. Once stabilized, HIF-1 upregulates
expression of genes involved in oxygen delivery, glycolysis,
and angiogenesis. Retrospective analysis of tumor biopsies
have typically shown that higher expression for HIF-1a or
its downstream targets carbonic anhydrase 9 (CAIX) and
glucose transporters 1 and 3 are correlated with tumor aggressiveness and a worse prognosis (7). However, of the
almost three dozen or so retrospective clinical studies
examining expression of endogenous markers in relation to
prognosis, only a few have compared expression of these
endogenous markers with an established method for measurement of tumor hypoxia, such as oxygen partial pressure
(pO2) or binding of chemical hypoxia markers.
Although a trend was observed between high levels of
pimonidazole binding and low pO2 measurements in 86
cervical carcinoma, this relation was not found to be signicant (11). A study in 28 brain tumors comparing pO2
to EF5 binding also failed to identify a correlation between the two measures of hypoxia (12). The lack of correlation between extrinsic hypoxia markers and oxygen
electrode results is of concern, since it raises the question
whether the proper comparison for endogenous hypoxia
markers is pO2 measurements or measurements of extrinsic hypoxia marker binding. Tumor hypoxia measured
with oxygen microelectrodes is associated with a poorer
outcome (overall survival and/or local control) for many
tumor types including cervical cancer (13), so perhaps
this is the proper comparison. However, chemical hypoxia markers detect hypoxia at the level of the individual cell
and could be viewed as the more appropriate comparison
with hypoxia-regulated gene expression patterns measured in individual cells. It has been suggested that the nature of tumor hypoxia, chronic or transient, may inuence chemical markers and electrode measurements differently (14), and that necrotic regions may inuence
oxygen electrode measurements but not hypoxia marker
binding (11). Of course, sampling variability as a result of
tumor heterogeneity for both oxygen electrodes and hypoxia markers can be substantial (1517). However, in
comparing the extent of binding and colocalization of different hypoxia marker staining patterns on sequential sections, the sampling problem is less important.
A small number of studies have compared endogenous
marker expression with either oxygen microelectrode
measurements or hypoxia marker binding. Airley et al.
(14) compared pimonidazole binding with GLUT-1 and

CAIX expression in 42 patients with cancer of the cervix

using a semi-quantitative scoring method. They concluded
that GLUT-1 correlated with pimonidazole binding, but
there was only a borderline correlation between pimonidazole and CAIX. In other studies, no correlation was found
between oxygen electrode measurements and GLUT-1 expression (18) or HIF-1a expression (15). Recent results indicate that erythropoietin expression was correlated with
pimonidazole binding (r 0.74) (19). Modest correlations
between pO2 and expression of HIF-1a (r 0.4, P < 0.01)
(20) and CAIX (r 0.43, P < 0.001) (21) have been
reported. Unlike HIF-1a expression (20), CAIX expression
was found to have prognostic value in terms of overall survival (21). In the latter study, the measured correlation
between GLUT-1 and pO2 was weak (r 0.28, P 0.04)
(21). In 21 patients with bladder cancer, a strong correlation was observed between pimonidazole and CAIX (r
0.86) as well as between pimonidazole and GLUT-1 (r
0.91) (22). Although the markers did not predict for local
control, both CAIX and GLUT-1 were independent prognostic factors for overall survival. A general conclusion
from these studies has been that endogenous markers offer
promise for the routine measurement of tumor hypoxia,
but they may not provide the same information as pO2
measurements using microelectrodes or binding of chemical hypoxia markers.
Treatment outcome is determined by various patient,
tumor and treatment-related factors. In addition to wellestablished clinical factors including tumor size and nodal
status, tumor proliferation and hypoxia are recognized to
be independent and potentially complementary predictive assays in cervical cancer (23). Since the early work
by Fertil and Malaise, intrinsic radiosensitivity is also known
to be an important and measurable property of both tumors
and normal tissues (24). The challenging approach to apply
functional assays to measure several properties of tumors,
taken by some groups (2527) will undoubtedly increase
the likelihood of developing a useful predictive assay. However, each one of these tumor characteristics will need to
be individually selected and validated. Toward this goal,
we compared several methods for their ability to detect
hypoxic regions within tumors. A quantitative digital image
analysis method was applied to tiled images of sequential
sections from tumor biopsies stained for pimonidazole,
HIF-1a and CAIX. Patients were also examined for tumor
oxygen tension using the Eppendorf oxygen microelectrode and biopsies from a subset of these patients were
examined for S phase DNA content. Therapy outcomes
will be reported separately.

METHODS
Patient Selection and Treatment Protocol
Ethical approval for this study was granted by the British
Columbia Cancer Agency Ethics Board and the University
of British Columbia Ethics committee. Over the course of
the study, more than 100 patients received pimonidazole
as an intravenous infusion before the study was closed at

COMPARISON OF HYPOXIA MARKERS IN CERVICAL CANCERS

the end of 2004. Seventy-eight patients with advanced

cervical carcinoma that participated in this project were
treated at the Vancouver Cancer Centre between September 1999 and June 2003.
All patients underwent initial clinical evaluation in accordance with local protocol, including a complete history
and physical, CT, or MRI of the abdomen and pelvis, chest
X-ray, liver function tests, complete blood count, and serum
creatinine and electrolytes. Staging was according to FIGO
guidelines. Suitable candidates for this study had a clinically
visible and histologically conrmed invasive carcinoma of
the cervix: squamous-cell carcinoma, adenocarcinoma, adenosquamous carcinoma, or carcinoma not otherwise specied, and gave informed consent. Patients were ineligible if
they had liver enzyme tests greater than twice the normal
laboratory values, serum creatinine
150 mmol/l, or a history of a peripheral neuropathy.
Seventy-four patients with cervical carcinoma (FIGO
stages Ia to IVb) were treated with radical radiotherapy,
48 of these patients received adjuvant chemotherapy (cisplatin), and 60 patients inhaled carbogen (95% oxygen, 5%
carbon dioxide) 4 min before and during the daily fractions
of external beam radiation therapy. Three patients underwent palliative treatment for recurrent tumors, and one
patient refused treatment.
All patients received a 20 min i.v. infusion of 0.5 g/m2
of Hypoxyprobe-1 (pimonidazole hydrochloride; Natural
Pharmaceuticals International Inc., Research Triangle Park,
NC) dissolved in 100 ml of 0.9% sterile saline at room temperature. The following day, 24 h later (about 4 plasma
half-lives), the patients underwent tumor oxygen measurements using the Eppendorf electrode, after which two
incisional biopsies (150 mg) were acquired. Oxygen
electrode measurements as well as all biopsies were taken
from clinically representative areas of the tumor. Areas of
obvious necrosis were avoided. Biopsies were transported
to the laboratory immediately after excision and disaggregated into single cells for analysis of pimonidazole binding using ow cytometry, and DNA content. Another biopsy was xed in formalin and embedded in parafn. Sequential sections were prepared by the Pathology Department at
the Vancouver Cancer Centre.
All of the patients were eligible for marker correlation studies, regardless of the therapy they received. Measurement
of the uctuations in HIF-1a over the course of chemoradiotherapy was performed on 15 patients. For each patient,
in addition to one pretreatment biopsy, multiple biopsies
(14 biopsies) obtained during treatment were available for
analysis.
Oxygen Microelectrode Measurements
Measurements of pO2 were performed pretreatment
using an Eppendorf pO2 histograph-6650 with sterile, polarographic probes 250 mm in diameter (Hamburg, Germany). The location of, the number, and length of the
tracks for each site was at the discretion of the clinician,
and were dependent on the clinical size and location of
the lesion, the tolerance of the patient to the procedure,

47

and the clinical suspicion of any measurement artifacts

that may have occurred as previously described (28). Typically 4 tracks and >80 measurements were obtained per
tumor. Median pO2 as well as the percentages of pO2 values 2.5 or 5 mmHg were calculated.
Pimonidazole Analysis by Flow Cytometry
Tumor biopsies were nely minced with scalpels and
disaggregated into single cells using trypsin, collagenase,
and DNase as previously described (29). Ethanol-xed
cells were rinsed in phosphate-buffered saline (PBS) and
resuspended in PST (PBS containing 4% FBS and 0.1% triton
X-100). A uorescein isothiocyanate (FITC)-conjugated
(1:1000 dilution) anti-pimonidazole primary antibody was
incubated with 2  106 alcohol-xed cells for 2 h at 378C.
Samples were rinsed with PST and resuspended for DNA
staining in 1 ml PBS containing 1 mg/ml 4,6-diamidino-2phenylindole dihydrochloride hydrate (DAPI). Single cell
suspensions were analyzed using a Coulter Epics Elite ESP
3-laser, 6-color cyometer (Coulter Corp. Hialeah, FL) for the
intensity of FITC-anti-pimonidazole staining and DNA content. Approximately 100,000 cells were acquired for the
analysis. A sample of antibody-stained cells was also analyzed microscopically in a blinded fashion, and the percentage of brightly stained cells was recorded from a sample of
500 cells.
Univariate histograms from ow cytometry analysis,
plotted as cell number versus logarithm of uorescent
anti-pimonidazole antibody intensity, were analyzed by a
least-squares approach for three Gaussian distributions
representing aerobic, intermediately oxygenated, and hypoxic tumor cell populations. No constraints on the positions of the distribution means were imposed. Cells considered hypoxic were, on average, 10 times more uorescent
than well-oxygenated cells within the population (29)
Measurement of S Phase Fraction
A third incisional biopsy was taken after pimonidazole
administration and disaggregated to produce a single cell
suspension. Cells were xed in 70% ethanol, and analyzed
using cells stained for DNA content using DAPI and for
presence of epithelial cells using SIGMA pan cytokeratin
monoclonal antibody (clone 11; 1:100 dilution). List mode
les were collected and S phase DNA content was determined using MODFIT software (Veristy Software House,
Inc., Topsham, ME). For hyperdiploid tumors, S phase
content was determined using the tumor population only.
For diploid adenocarcinoma, all cytokeratin positive cells
were included in the analysis. For squamous-cell cancers,
all cells were analyzed. More details have been provided
elsewhere (30).
Antibody Staining of Sections
Immunoperoxidase with diaminobenzidine tetrahydrochloride (DAB) substrate was performed to detect hypoxic
regions indicated by the presence of pimonidazole, CAIX,
or HIF-1a. For dewaxing, 5 mm parafn-embedded sec-

48

JANKOVIC ET AL.

tions were incubated at 508C for 1.5 h and immersed in

xylene to complete the dewaxing process. The slides
were hydrated in graded alcohols, and rinsed in distilled
water and PBS. Sections were treated with 3% hydrogen
peroxide in methanol for 10 min to eliminate the endogenous peroxidase activity and thereby prevent nonspecic
reactions with DAB substrate. After rinsing, each section
was treated with 50 ml of protease (1:100 v/v in PBS) for
30 min at 378C to degrade the protein cross-links formed
by formalin xation and expose the antigenic sites. When
staining for HIF-1a, the application of protease was substituted by treatment with high pH target retrieval solution from DAKO (Carpinteria, CA), as this method was
found to have improved efcacy for nuclear antigen exposure compared to protease treatment. The slides were submerged for 30 min in a preheated solution inside a 958C
water bath. Subsequently, they were taken out of the bath
and allowed to cool at room temperature for 15 min. They
were then rinsed several times with PBS. To reduce the
background staining, a blocking agent, PTN, containing
1% BSA (w/v) and 0.2% Tween 20 (v/v) in PBS, was
applied to the sections for 20 min. Fifty to sixty microliters (depending on the size of the tissue section) of antipimonidazole or anti-HIF-a (diluted 1:100 in PTN) was
applied to each drained section for 2 h (pimonidazole) or
1 h (HIF-1a). After washing three times (two times with
PBS and once with PTN) for 5 min, the sections were
incubated in rabbit anti-mouse biotinylated secondary
antibody (1:200 in PTN) for 30 min and subsequently
washed in PBS for 5 min. Eosin (HIF-1a sections) or hematoxlin (pimonidazole sections) were used as counterstains. Staining for CAIX was performed at UC Irvine as
previously described (31) (32).
Quantitative Analysis of Antibody-Stained Sections
A Zeiss Axioplan 2 microscope with an attached monochrome 12 bit CCD camera was used for acquisition of
images of stained tumor sections. Composite images of
the entire tumor tissue were prepared by electronically
tiling up to 200 individual frames. Images of pimonidazole- and CAIX-stained sections were focused and captured in 8-bit grayscale, while the images of HIF-1a
stained sections were captured in 24-bit RGB using an
RGB color lter. Image artifacts, such as bubbles created
upon application of the mounting medium were removed
in NIH image Pieces of tissue present in one section and
absent in a sequential section (stained for a different antigen) were deleted or disregarded. Digitized images were
optimized to obtain maximum contrast between background and stained regions. Thresholding was performed
twice to differentiate tissues from background and
stained tissue from the rest of the image (Fig. 1). The two
highlighted areas were measured and the hypoxic fraction was calculated by dividing the stained tissue area by
the total tissue area. The selected threshold intensities
were adapted to the intensity parameters of each image
analyzed to differentiate between the unstained and
stained tissue as accurately as possible. The threshold in-

FIG. 1. Analysis of hypoxia marker binding in antibody-stained sections and single cells. Panel (a) show a representative tiled tumor section stained for CAIX. In panel (b), this is converted to a gray scale
image. Panel (c) shows the total area of the section and panel (d) shows
the marker positive regions. The ratio of the number of red pixels in (d)
divided by the number of red pixels in (c) gives the percentage of the tumor that is CAIX positive.

tensities depended on the intensity of the staining, the

marker being detected, the extent of the counterstain
absorption, the brightness of the image, and the noise in
the image. Upon completion of the image processing
steps, the images were analyzed for the fractions of tumor
sections stained for pimonidazole, CAIX, and HIF-1a and
for marker colocalization. Images were thresholded and
analyzed independently by three observers for marker
positive fractions and results were averaged.
Although care was taken to maintain consistency in
concentrations and duration of treatment by antibodies,
ABC and DAB reagents between the experiments, the activity of these reagents, and the resulting intensity of the
signal were subject to variation. For this reason, the exact
relationship between the staining intensity and the amount
of marker present could not be precisely identied or
assumed consistent between batches. All of the antibodies
yielded strong signals when present, or no signal relative to
background, and therefore only the percentage of stained
pixels as a fraction of total tumor tissue, and not the staining intensity, was assessed.

Qualitative Analysis of Hypoxia Marker Colocalization

A qualitative colocalization analysis was performed on
all images to calculate the extent of colocalization
between pairs of markers. A discrete scoring system was
used that assigned scores from 0 to 4 to each pair of
stained sections. Pairs of sections with scores greater than
or equal to 2 were considered moderately to highly colocalized. As analysis was applied to marker staining in sequential sections, and there were differences in intracellular localization or oxygen dependency of binding, precise
colocalization was not expected. Instead, pimonidazole-labeled regions that fell within regions of HIF-1a or CAIX
staining (but not necessarily extending to the limits of
staining by those antibodies) were considered completely
co-localized. Similarly, lack of HIF-1a staining in perine-

40.3 (0.089.0)
38.5 (2.087.0)
4.1 (0.630.4)
4.8 (0.418.0)
4.4 (0.022.6)
5.7 (0.223.0)
4.3 (0.627.8)
6.1 (0.420.0)

5.2 (0.328.1)
4.2 (0.227.8)

66.0 (0.0100.0)
54.5 (2.0100.0)
42.0 (3.089.0)
37.0 (0.087.0)
5.6 (0.630.4)
3.7 (0.418.0)
5.0 (0.028.1)
5.3 (0.222.6)
5.4 (0.627.8)
4.7 (0.411.1)

6.7 (0.328.1)
4.3 (0.223.5)

66.0 (0.090.0)
57.0 (4.0100.0)
40.5 (1.080.0)
39.0 (0.089.0)
3.4 (0.430.4)
6.5 (1.318.0)
3.2 (0.020.8)
6.5 (0.328.1)
3.1 (0.411.0)
5.8 (0.627.8)

4.1 (0.223.5)
7.8 (1.128.1)

68.8 (0.0100.0)
57.5 (2.01000)
41.0 (0.028.1)
40.3 (1.089.0)
6.3 (0.430.4)
3.7 (0.5174)
3.8 (0.028.1)
5.4 (0.3230)
4.9 (0.427.8)
4.6 (1.2148)

5.0 (0.228.1)
4.5 (0.3278)

66.0 (9.0100.0)
66.0 (32.096.0)
51.0 (0.092.0)
38.2 (4.092.0)
41.5 (2.087.0)
39.0 (10.081.0)
38.0 (1.083.0)
33.1 (0.089.0)
5.2 (0.630.4)
4.3 (0.421.1)
4.5 (2.217.1)
5.4 (2.96.5)
5.0 (1.023.0)
3.0 (0.020.8)
5.1 (0.728.1)
9.2 (6.322.6)
3.7 (0.627.8)
4.1 (0.411.4)
6.1 (1.511.1)
6.2 (5.611.0)

6.6 (0.328.1)
4.1 (0.223.5)
5.2 (1.114.4)
10.3 (6.412.8)

61.5 (2.0100.0)
57.5 (0.081.0)
70.4 (70.470.4)
39.5 (1.089.0)
35.0 (0.080.0)
64.2 (64.264.2)
4.0 (0.421.1)
7.0 (1.330.4)
6.4 (6.36.5)
4.7 (0.228.1)
7.6 (1.023.5)
12.8 (12.812.8)
5.4 (0.028.1)
2.9 (0.722.6)
7.4 (6.68.2)
5.4 (0.427.8)
1.9 (0.68.9)
9.6 (9.69.6)

HP5
HP2.5
HIF-1a
CAIX
Pimo-FC
Pimo
%

100
75.3
21.9
2.7
100
37.5
37.5
18.1
6.9
100
53.0
47.0
100
50.0
50.0
100
44.9
55.1
100
63.3
36.7

73
55
16
2
72
27
27
13
5
66
35
31
72
36
36
69
31
38
60
38
22

RESULTS
Patient characteristics and percentage of cells considered
hypoxic are given in Table 1. Frequencies of adenocarcinoma and squamous-cell carcinoma in this study reected the
incidence of these two types of cancer in the North American population (34,35). Similar ranges, means, and median
values for hypoxic fraction were found for pimonidazole,
CAIX, and HIF-1a staining. As in many previous studies
(3638), hypoxic fraction measured by three hypoxia
markers did not correlate with the well-established clinical
prognostic factors (including FIGO stage, maximum clinical
diameter, presenting hemoglobin, nodal status, and tumor
grade) (results of statistical analyses not shown).
Figures 1a1d illustrates the method used to determine
the percentage of tumor that was positive for each hypoxia
marker after antibody staining. Colored images were obtained by a x-y stage automatic tiling of entire sections
under 10 magnication (Fig. 1a). Tiled images were converted to gray scale images in Figure 1b. These underwent
thresholding to identify the entire area of the tumor (Fig. 1c)
as well as the marker positive region (Fig. 1d). Areas of
obvious necrosis or tissue folds were not included in the
analysis but no attempt was made to eliminate normal tissue components. Before comparing the staining patterns
for different hypoxia markers, it was important to establish whether immunohistochemical staining of sequential
sections was reproducible and consistent between sections. Fractions of the tumor stained for pimonidazole in
two sequential sections from 16 tumors were compared.
Results indicated a strong correlation (r 0.99, slope
1.1, data not shown).
In addition to immunohistochemical analysis, another
method was employed to measure pimonidazole binding.
A second biopsy was taken at the same time and within
30 min of biopsy, it was disaggregated with enzymes, and
xed in ethanol. Flow cytometry evaluation was performed with anti-pimonidazole antibodies. The percentage of hypoxic cells was calculated by tting ow histograms to three normal distributions representing aerobic,
intermediate, and hypoxic populations (29). For 30% of
the tumor samples, DNA content could be used to discriminate between diploid cells and hyperdiploid tumor
cells (Fig. 2). By gating on DNA content, hypoxic fraction

Histology
Squamous cell carcinoma
Adenocarcinoma
Adenosquamous
FIGO stage
I
II
III
IV
Grade
Well/mod differentiated
Poorly differentiated
Age (years)
<47 (median)

47 (median)
Size (largest diameter in cm)
<5.0 (median)

5.0 (median)
Nodal Status
Negative
Positive

Linear least squares regression analysis was used to

determine the degree of correlation between the fractions
of tissue sections labeled for pimonidazole, CAIX, and
HIF-1a. Bivariate Pearson correlations were performed to
identify pairs of clinical and hypoxia parameters which
correlated signicantly. To examine the differences in hypoxic fraction between populations with different values
of clinical parameters, independent Student t tests were
performed. A P-value less than or equal to 0.05 was considered statistically signicant.

Parameter

Statistical Analysis

Table 1
Clinicopathologic Parameters and Measures of Hypoxic Fraction. The Median Value of the Hypoxia Marker Positive Fraction and the Range of Values are Given

crotic regions (33) was not taken into account in measurement of degree of colocalization.

61.5 (2.096.0)
60.0 (0.0100.0)

49

COMPARISON OF HYPOXIA MARKERS IN CERVICAL CANCERS

50

JANKOVIC ET AL.

FIG. 2. Representative ow cytometric analyses of pimonidazole binding in cells from cervical carcinoma biopsies. Single cells prepared from two
patient samples were xed and stained for pimonidazole adducts and for DNA content. The diploid are hyperdiploid cell populations are evident in panels (a) and (b), and the relation between DNA content and pimonidazole antibody binding is shown in panels (c) and (d). Panels in (e) indicate the distribution of pimonidazole antibody binding for diploid and hyperdiploid populations for each tumor, and the application of a curve tting algorithm to calculate hypoxic fraction. The hypoxic cell population (black) is on average 10 times more uorescent than the aerobic population (gray).

could be determined independently for tumor and normal cells. Invariably, the diploid population showed a
lower hypoxic fraction than the tumor cells, suggesting
that these cells are closer to functional blood vessels. A
reduced ability of diploid cells to metabolize and bind
pimonidazole was also observed, consistent with a
reduced nitroreductase activity. To compare hypoxic fraction measured by ow cytometry with hypoxic fraction
using image analysis of stained sections, DNA content
was not used to discriminate tumor cells from normal
cells even in those cases where this was feasible.
The percentage of hypoxic cells determined by this ow
cytometry approach agreed well with the much simpler
approach of visually scoring the fraction of brightly stained
cells cytospun onto slides after pimonidazole staining (Fig.
3a). Image analysis of pimonidazole-stained tumor sections
from a separate biopsy taken at the same time also showed
a signicant correlation with the ow cytometry method.

However, because two different biopsies were compared

using two different methods, tumor heterogeneity and cell
loss resulted in a poorer correlation (Fig. 3b).
The mean and range of marker positive fractions were
similar for pimonidazole-, HIF-1a-, and CAIX-stained sequential sections (Fig. 4a). The comparison shown in Figure 4b gives condence that the markers are identifying
the same property of the tumors. However, correlations
between the individual pairs, although signicant, are less
convincing with the exception of the correlation between pimonidazole and CAIX (Table 2, Figs. 5a5c). Five
tumors showed low expression of HIF-1a but high
expression of CAIX (>3.5 times higher). Not surprisingly,
the correlations between pimonidazole and CAIX or CAIX
and HIF-1a improved signicantly when these ve tumors
were removed from analysis (data not shown). Interestingly, preliminary outcome analysis indicates that none of
these tumors has progressed.

COMPARISON OF HYPOXIA MARKERS IN CERVICAL CANCERS

FIG. 3. Comparison of pimonidazole binding determined by ow cytometry, visual analysis of cytospun cells, or image analysis of stained
sections. Single cells prepared from biopsies were xed and labeled for
pimonidazole antibody binding. In panel (a), coded samples were analyzed by ow cytometry using the curve tting approach in Fig. 2 or by
visual counting of brightly stained cells after cytospinning. In panel (b),
results from the ow cytometry analysis of pimonidazole binding from
one biopsy were compared to the analysis of pimonidazole positive pixels from a second biopsy taken at the same time. Linear best-t lines
are drawn.

The degree of colocalization between stained regions is

also critical in their evaluation as hypoxia markers. More
than 70% of the stained regions showed extensive colocalization when two markers were compared, and the
degree of colocalization was again greatest for pimonidazole and CAIX (Figs. 5d5f). A comparison between the
semi-quantitative colocalization score and the marker positive fraction was used to estimate the percentage of the
marker positive regions that were occupied by one marker, two markers, or all three markers (Table 3). Few areas
showed pimonidazole binding in the absence of HIF-1a

FIG. 4. Percentage of pimonidazole, HIF-1a,

and CAIX stained pixels in sequential sections of
cervical cancer biopsies. Panel (a) shows hypoxic
fraction distributions for 67 cervical tumors. Panel
(b) compares the percentage of marker positive
pixels for all three hypoxia markers.

51

and CAIX staining, and in this analysis, 6477% of regions

expressed all three markers. Figure 6 shows two examples where pimonidazole and HIF-1a were located in the
same tumor cords. However, pimonidazole-labeled cells
appeared to be associated within the necrotic area, while
HIF-1a labeled cells were located in adjoining regions that
appeared viable.
There was no relationship between Eppendorf oxygen
electrode measurements and hypoxia marker binding. Figures 7a7c show the comparison between the percentage
of pO2 readings 2.5 mmHg and the mean percentage of
hypoxic cells calculated based on pimonidazole, HIF-1a,
and CAIX antibody staining. Similar results were obtained
using the median pO2 or percentage of readings 5 mmHg
(data not shown). For a smaller subset of these patients analyzed for S phase fraction, there was no correlation between S phase content and mean percentage of hypoxic
cells (Figs. 7d7f).
The change in hypoxic fraction over the course of therapy may provide a useful measure of tumor response to
therapy and could extend the usefulness of this approach.
Between two and ve tumor biopsies taken over the
course of treatment from a subset of 15 patients were
available for examination of HIF-1a expression. The average time between the initial pretreatment biopsy to the
last biopsy obtained was 39.3 days. Although no consistent pattern was observed, in 11/15 patients, HIF-1a decreased within two weeks after the rst radiotherapy
treatment (Fig. 8). In 6 out of these 11 patients, HIF-1a decreased by more than 2-fold. Preliminary outcome analysis
indicated that all of the patients who progressed (5/15)
experienced a decrease in HIF-1a during the rst two
weeks of treatment. Although HIF-1a levels have been reported to increase between 48 and 72 h after irradiation
of xenograft tumors (39), this was not evident in the tumors
of these patients while undergoing weekly chemotherapy

52

JANKOVIC ET AL.

Table 2
Pearsons Correlation Coefcients Between Pairs of Hypoxia Measurement Techniques
Marker
statistics
Pimo
Pimo-FC
CAIX
HIF-1a
HP2.5
HP5

Pimo
Pearson
1

P
0

Pimo-FC
Pearson
P
0.45*
1

0
0

CAIX
Pearson
0.60*
0.36*
1

P
0
0
0

HIF-1a
Pearson
P
0.34*
0.13
0.42*
1

0
0.31
0
0

HP2.5
Pearson
0.12
0.07
0.03
0.02
1

P
0.91
0.59
0.83
0.91
0

HP5
Pearson
0.02
0.04
0.04
0.04
0.85*
1

P
0.89
0.76
0.79
0.75
0
0

Asterisk indicate that the correlation is signicant at the 0.01 level (2-tailed).

and daily radiation treatment. This is likely explained by

the fact that biopsies taken during treatment were obtained
on a weekly basis.
DISCUSSION
The search for a practical and robust method for measuring tumor hypoxia in the clinic has been a major catalyst
in the development of chemical and endogenous hypoxia
markers. The main goal of this study was to perform a
quantitative analysis for three hypoxia markers examined
on sequential stained sections from pretreatment biopsies.
The assessment of patient outcome related to markers was
a secondary objective and will be reported later. Previous
studies have shown positive correlations between various
endogenous markers and established methods for measuring hypoxia (9). Nevertheless, some studies have elucidated the shortcomings and the differences between the
methods. Possible causes for marker mismatch patterns in-

clude differences in marker sensitivity with respect to

degree and duration of hypoxia as well as effects of factors
other than hypoxia on expression of endogenous markers
(18). Critical analysis of marker mismatch patterns is essential because colocalization could provide additional information on the nature of hypoxia (Table 3). Although best
accomplished using multiple antibodies in a single section,
analysis using stained sequential sections provides some
information relevant to this question. To our knowledge,
no previous study has quantied spatial correlations between the staining of endogenous markers and chemical
hypoxia marker binding in a patient cohort that has also
been characterized for pO2. Janssen et al (8) analyzed colocalization and hypoxia marker binding in relation to blood
vessels in several patients with head and neck cancers.
Their more quantitative approach indicated a poor correlation between pimonidazole and HIF-1a stained regions and
relatively low colocalization.

FIG. 5. Comparisons between hypoxia marker expression in cervical cancer biopsies. The percentage of
marker positive pixels are compared pair-wise in panels
(a)(c), and linear best-t lines are shown. Panels (d)
(f) present histograms of colocalization frequency between pairs of markers.

COMPARISON OF HYPOXIA MARKERS IN CERVICAL CANCERS

53

Table 3
Estimates of Marker Mismatch and Possible Explanations for Variation in Colocalization of Markers
Pattern observed

Frequency
(% of stained regions)

Pimo only

<5

HIF-1a only

10

CAIX only

10

CAIX & HIF-1a, no Pimo

1216

Pimo & CAIX, no HIF-1a

1417

Pimo & HIF-1a, no CAIX

57

All markers

6477

Possible explanation for mismatch

Almost never observed; possible binding
to keratinizing tissue
Region was normoxic in the past but became
hypoxic very recently
Region was hypoxic in the past, but became
reoxygenated before pimonidazole administration
Region is insufciently hypoxic (time or degree)
to bind pimonidazole
Region was previously hypoxic, but became
reoxygenated recently. Region lacks
nutrients (peri-necrotic)
Region is undergoing transient hypoxia of
insufcient duration to allow CAIX to accumulate
Region is predominantly chronically hypoxic

Note that under the category all markers, areas for comparison are dened based on a single marker
(e.g., do pimonidazole-stained regions express both HIF-1a and CAIX?) so regions of mismatch that do not
include that marker are ignored.

None of the three markers correlated with Eppendorf oxygen electrode measurement (median, HP  2.5 mmHg or
HP  5 mmHg). This is in spite of the fact that tumors with
high oxygen partial pressure (pO2) have been shown to exhibit decreased pimonidazole binding (40) and lower HIF1a expression (20). The relationship between marker expression and oxygen electrode measurements is a complex
one, as these methods do not sample the same tumor
microenvironment or provide directly comparable measures of hypoxia. There are now reports of a lack of correlation between oxygen electrode measurements and GLUT-1
expression (18), CAIX expression (15), and pimonidazole
binding (11).
Signicant correlations were observed between the
three single-cell measures of tumor hypoxia. Pimonidazole binding, HIF-1a, and CAIX expression all indicated
an average hypoxic fraction of about 6% in these tumors.
The mean/median values are similar to those previously
reported by Kaanders et al. for CAIX and pimonidazole in
head and neck cancers (6.4 and 6% respectively) (41),
and for pimonidazole binding in cervical cancers as reported by Azuma et al (4.5 6 4.8%) (42). They are higher
than the median value of 2% measured for HIF-1a in cervical cancers by Haugland et al (20). Other groups have used
a semi-quantitative scoring system making comparisons
more difcult. Underestimates of HIF-1a and CAIX staining
as a result of localization to nucleus or membrane, respectively, might be expected, and the dynamic range of staining
of tissue sections by the endogenous markers was typically
lower than for pimonidazole staining. However, the percentage of stained area is only one aspect of a comparison
between markers. Marker colocalization at a microregional
level was high in most cases, conrming that hypoxia is
likely to be the major factor in the expression of HIF-1a and
CAIX in these tumors. Lack of marker colocalization at a cellular level could be a useful indicator of the nature of tumor
hypoxia and at least some regions of some tumors showed
mismatch between these markers. Although HIF-1a stabilization under hypoxia is largely responsible for expression

of CAIX, once formed, CAIX is typically lost only upon cell

death or division. Conversely, HIF-1a is lost within minutes
upon reoxygenation (33). Therefore, it was interesting to

FIG. 6. Comparison between HIF-1a and pimonidazole antibody staining patterns in sequential sections from three biopsies. Note that pimonidazole, unlike HIF-1a, appears to occur within necrotic regions in all three
of these tumors.

54

JANKOVIC ET AL.

FIG. 7. Comparison between hypoxia marker binding

and Eppendorf oxygen electrode measurements (a)(c) or
S phase fraction (d)(f). Linear best-t lines are shown.

identify ve tumors with high expression of CAIX but low

levels of HIF-1a. This pattern is consistent with transient
hypoxia, or more accurately, recent reoxygenation. It is also
consistent with the reported lack of HIF-1a expression in
perinecrotic regions that maintain CAIX expression because
of its long half-life (33).
Pimonidazole was administered to patients 1820 h before biopsy. This timing was based on the reported pimonidazole plasma half-life of 4.8 h for women and 5.4 h for
men (43), and the requirement that no free pimonidazole be present at the time of biopsy when any unmetabolized drug would be rapidly bound under hypoxia.
However, during this 1820 h, hypoxic cells bordering
necrosis will die, so the cells labeled by pimonidazole will
not include those recently hypoxic but could include
some that are now necrotic. The movement of pimonidazole-stained cells into necrotic regions has been quantied in human tumor xenografts by using comparisons
with HIF-1a expression over time (33) or by expression
of a second hypoxia marker (44). In frozen sections of
SiHa cervical carcinoma xenografts examined 90 min after
pimonidazole injection, 80% of pimonidazole-labeled cells
expressed HIF-1a. However, 48 h later, only 32% of pimonidazole-labeled cells still expressed HIF-1a (33). In our
sequential DAB-stained sections, we also detected evidence of pimonidazole binding extending into necrotic
regions (Fig. 6). Although this degree of mismatch was
not taken into consideration in our colocalization analysis,
it does contribute to observed differences between marker
expression or binding at the cellular level (Table 3). This
potential problem was previously recognized by Denekamp
and Dasu (45).
The possibility that tumors that express high levels of
CAIX but low amounts of HIF-1a are less likely to progress

suggests that combining information from these two

markers could be a more informative approach than relying
on results from a single endogenous marker. Similarly, analysis of response to treatment using a second biopsy would
improve the prognostic value of hypoxia measurement.
Ultimately, information from protein chips that include
these markers for hypoxia, as well as indicators of proliferation and repair capacity, are expected to improve the ability
to predict response to therapy or at least to select patients
likely to respond to hypoxia-directed treatments (e.g. tirapazamine). Factors that continue to limit this approach are
tumor heterogeneity (no single biopsy can be completely
representative) and the value of pretreatment indicators as
opposed to measures of responses to specic therapies.

FIG. 8. Change in expression of HIF-1a as a function of time after the

start of radiochemotherapy. Sequential biopsies from 15 patients were
analyzed for the percentage of HIF-1a antibody-positive area.

COMPARISON OF HYPOXIA MARKERS IN CERVICAL CANCERS

ACKNOWLEDGMENTS
This work was supported by the Canadian Institutes of
Health Research. The assistance of Genevieve Law during
a BCCRC summer studentship is gratefully acknowledged.
LITERATURE CITED
1. Hockel M, Vaupel P. Tumor hypoxia: denitions and current clinical,
biologic, and molecular aspects. J Natl Cancer Inst 2001;93:266276.
2. De Jaeger K, Kavanagh MC, Hill RP. Relationship of hypoxia to metastatic ability in rodent tumours. Br J Cancer 2001;84:12801285.
3. Dunn T. Oxygen and cancer. N C Med J 1997;58:140143.
4. Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, Lowe SW,
Giaccia AJ. Hypoxia-mediated selection of cells with diminished apoptotic potential in solid tumours. Nature 1996;379:8891.
5. Vaupel P, Kelleher DK, Hockel M. Oxygen status of malignant
tumors: pathogenesis of hypoxia and signicance for tumor therapy.
Semin Oncol 2001;28:2935.
6. Young SD, Marshall RS, Hill RP. Hypoxia induces DNA over-replication and enhances metastatic potential of murine tumor cells. Proc
Natl Acad Sci USA 1988;85:95339537.
7. Bussink J, Kaanders JH, van der Kogel AJ. Tumor hypoxia at the
micro-regional level: clinical relevance and predictive value of exogenous and endogenous hypoxic cell markers. Radiother Oncol 2003;67:
315.
8. Janssen HL, Haustermans KM, Sprong D, Blommestijn G, Hoand I,
Hoebers FJ, Blijweert E, Raleigh JA, Semenza GL, Varia MA, Balm AJ, van
Velthuysen ML, Delaere P, Sciot R, Begg AC. HIF-1A, pimonidazole, and
iododeoxyuridine to estimate hypoxia and perfusion in human headand-neck tumors. Int J Radiat Oncol Biol Phys 2002;54:15371549.
9. Vordermark D, Brown JM. Endogenous markers of tumor hypoxia
predictors of clinical radiation resistance? Strahlenther Onkol 2003;
179:801811.
10. Semenza GL. Hydroxylation of HIF-1: oxygen sensing at the molecular level. Physiology (Bethesda) 2004;19:176182.
11. Nordsmark M, Loncaster J, Aquino-Parsons C, Chou SC, Ladekarl M,
Havsteen H, Lindegaard JC, Davidson SE, Varia M, West C, Hunter R,
Overgaard J, Raleigh JA. Measurements of hypoxia using pimonidazole and polarographic oxygen-sensitive electrodes in human cervix
carcinomas. Radiother Oncol 2003;67:3544.
12. Evans SM, Judy KD, Dunphy I, Jenkins WT, Nelson PT, Collins R,
Wileyto EP, Jenkins K, Hahn SM, Stevens CW, Judkins AR, Phillips P,
Geoerger B, Koch CJ. Comparative measurements of hypoxia in human
brain tumors using needle electrodes and EF5 binding. Cancer Res
2004;64:18861892.
13. Milosevic M, Fyles A, Hedley D, Hill R. The human tumor microenvironment: invasive (needle) measurement of oxygen and interstitial
uid pressure. Semin Radiat Oncol 2004;14:249258.
14. Airley RE, Loncaster J, Raleigh JA, Harris AL, Davidson SE, Hunter
RD, West CM, Stratford IJ. GLUT-1 and CAIX as intrinsic markers of
hypoxia in carcinoma of the cervix: relationship to pimonidazole
binding. Int J Cancer 2003;104:8591.
15. Hedley D, Pintilie M, Woo J, Morrison A, Birle D, Fyles A, Milosevic M,
Hill R. Carbonic anhydrase IX expression, hypoxia, and prognosis in
patients with uterine cervical carcinomas. Clin Cancer Res 2003;9:
56665674.
16. Thrall DE, Rosner GL, Azuma C, McEntee MC, Raleigh JA. Hypoxia
marker labeling in tumor biopsies: quantication of labeling variation
and criteria for biopsy sectioning. Radiother Oncol 1997;44:171176.
17. Wong RK, Fyles A, Milosevic M, Pintilie M, Hill RP. Heterogeneity of
polarographic oxygen tension measurements in cervix cancer: an
evaluation of within and between tumor variability, probe position,
and track depth. Int J Radiat Oncol Biol Phys 1997;39:405412.
18. Mayer A, Hockel M, Wree A, Vaupel P. Microregional expression of glucose transporter-1 and oxygenation status: lack of correlation in locally
advanced cervical cancers. Clin Cancer Res 2005;11:27682773.
19. Arcasoy MO, Amin K, Chou SC, Haroon ZA, Varia M, Raleigh JA.
Erythropoietin and erythropoietin receptor expression in head and
neck cancer: relationship to tumor hypoxia. Clin Cancer Res 2005;11:
2027.
20. Haugland HK, Vukovic V, Pintilie M, Fyles AW, Milosevic M, Hill RP,
Hedley DW. Expression of hypoxia-inducible factor-1alpha in cervical
carcinomas: correlation with tumor oxygenation. Int J Radiat Oncol
Biol Phys 2002;53:854861.
21. Loncaster JA, Harris AL, Davidson SE, Logue JP, Hunter RD, Wycoff
CC, Pastorek J, Ratcliffe PJ, Stratford IJ, West CM. Carbonic anhydrase
(CA IX) expression, a potential new intrinsic marker of hypoxia: correlations with tumor oxygen measurements and prognosis in locally
advanced carcinoma of the cervix. Cancer Res 2001;61:63946399.

55

22. Hoskin PJ, Sibtain A, Daley FM, Wilson GD. GLUT1 and CAIX as
intrinsic markers of hypoxia in bladder cancer: relationship with vascularity and proliferation as predictors of outcome of ARCON. Br J Cancer
2003;89:12901297.
23. Tsang RW, Fyles AW, Milosevic M, Syed A, Pintilie M, Levin W, Manchul LA. Interrelationship of proliferation and hypoxia in carcinoma
of the cervix. Int J Radiat Oncol Biol Phys 2000;46:9599.
24. Fertil B, Malaise EP. Inherent cellular radiosensitivity as a basic concept for human tumor radiotherapy. Int J Radiat Oncol Biol Phys
1981;7:621629.
25. Bussink J, Kaanders JH, Rijken PF, Peters JP, Hodgkiss RJ, Marres HA,
van der Kogel AJ. Vascular architecture and microenvironmental parameters in human squamous cell carcinoma xenografts: effects of
carbogen and nicotinamide. Radiother Oncol 1999;50:173184.
26. Eschwege F, Bourhis J, Girinski T, Lartigau E, Guichard M, Deble D,
Kepta L, Wilson GD, Luboinski B. Predictive assays of radiation
response in patients with head and neck squamous cell carcinoma: a
review of the Institute Gustave Roussy experience. Int J Radiat Oncol
Biol Phys 1997;39:849853.
27. Hill RP, Fyles W, Milosevic M, Pintilie M, Tsang RW. Is there a relationship between repopulation and hypoxia/reoxygenation? Results from
human carcinoma of the cervix. Int J Radiat Biol 2003;79:487494.
28. Aquino-Parsons C, Green A, Minchinton AI. Oxygen tension in primary gynaecological tumours: the inuence of carbon dioxide concentration. Radiother Oncol 2000;57:4551.
29. Olive PL, Durand RE, Raleigh JA, Luo C, Aquino-Parsons C. Comparison between the comet assay and pimonidazole binding for measuring tumour hypoxia. Br J Cancer 2000;83:15251531.
30. Durand RE, Aquino-Parsons C. Predicting response to treatment in
human cancers of the uterine cervix: sequential biopsies during
external beam radiotherapy. Int J Radiat Oncol Biol Phys 2004;58:
555560.
31. Liao SY, Brewer C, Zavada J, Pastorek J, Pastorekova S, Manetta A,
Berman ML, DiSaia PJ, Stanbridge EJ. Identication of the MN antigen as
a diagnostic biomarker of cervical intraepithelial squamous and glandular neoplasia and cervical carcinomas. Am J Pathol 1994;145:598609.
32. Olive PL, Aquino-Parsons C, MacPhail SH, Liao SY, Raleigh JA, Lerman
MI, Stanbridge EJ. Carbonic anhydrase 9 as an endogenous marker
for hypoxic cells in cervical cancer. Cancer Res 2001;61:89248929.
33. Sobhanifar S, Aquino-Parsons C, Stanbridge EJ, Olive P. Reduced
expression of hypoxia-inducible factor-1alpha in perinecrotic regions
of solid tumors. Cancer Res 2005;65:72597266.
34. Duarte-Franco E, Franco EL. Cancer of the uterine cervix. BMC Womens Health 2004;4(Suppl 1):S13.
35. Liu S, Semenciw R, Mao Y. Cervical cancer: the increasing incidence
of adenocarcinoma and adenosquamous carcinoma in younger women. CMAJ 2001;164:11511152.
36. Fyles A, Milosevic M, Hedley D, Pintilie M, Levin W, Manchul L, Hill RP.
Tumor hypoxia has independent predictor impact only in patients with
node-negative cervix cancer. J Clin Oncol 2002;20:680687.
37. Hockel M, Schlenger K, Aral B, Mitze M, Schaffer U, Vaupel P. Association between tumor hypoxia and malignant progression in advanced
cancer of the uterine cervix. Cancer Res 1996;56:45094515.
38. Stone JE, Parker R, Gilks CB, Stanbridge EJ, Liao SY, Aquino-Parsons
C. Intratumoral oxygenation of invasive squamous cell carcimoma of
the vulva is not correlated with regional lymph node metastasis. Eur
J Gynaecol Oncol 2005;26:3135.
39. Moeller BJ, Cao Y, Li CY, Dewhirst MW. Radiation activates HIF-1 to
regulate vascular radiosensitivity in tumors: role of reoxygenation,
free radicals, and stress granules. Cancer Cell 2004;5:429441.
40. Nordsmark M, Loncaster J, Chou SC, Havsteen H, Lindegaard JC,
Davidson SE, Varia M, West C, Hunter R, Overgaard J, Raleigh JA.
Invasive oxygen measurements and pimonidazole labeling in human
cervix carcinoma. Int J Radiat Oncol Biol Phys 2001;49:581586.
41. Kaanders JH, Wijffels KI, Marres HA, Ljungkvist AS, Pop LA, van den
Hoogen FJ, de Wilde PC, Bussink J, Raleigh JA, van der Kogel AJ.
Pimonidazole binding and tumor vascularity predict for treatment
outcome in head and neck cancer. Cancer Res 2002;62:70667074.
42. Azuma Y, Chou SC, Lininger RA, Murphy BJ, Varia MA, Raleigh JA. Hypoxia and differentiation in squamous cell carcinomas of the uterine cervix: pimonidazole and involucrin. Clin Cancer Res 2003;9:49444952.
43. Saunders MI, Anderson PJ, Bennett MH, Dische S, Minchinton A,
Stratford MR, Tothill M. The clinical testing of Ro 03-8799pharmacokinetics, toxicology, tissue and tumor concentrations. Int J Radiat Oncol
Biol Phys 1984;10:17591763.
44. Ljungkvist AS, Bussink J, Kaanders JH, Rijken PF, Begg AC, Raleigh JA,
van der Kogel AJ. Hypoxic cell turnover in different solid tumor lines.
Int J Radiat Oncol Biol Phys 2005;62:11571168.
45. Denekamp J, Dasu A. Inducible repair and the two forms of tumour
hypoxiatime for a paradigm shift. Acta Oncol 1999;38:903918.