Expression of hypoxia-inducible factor-1α in cervical carcinomas: correlation with tumor oxygenation

Int. J. Radiation Oncology Biol. Phys., Vol. 53, No. 4, pp. 854 – 861, 2002 Copyright © 2002 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/02/$–see front matter

PII S0360-3016(02)02815-8

CLINICAL INVESTIGATION

Cervix

EXPRESSION OF HYPOXIA-INDUCIBLE FACTOR-1␣ IN CERVICAL CARCINOMAS: CORRELATION WITH TUMOR OXYGENATION HANS KRISTIAN HAUGLAND, M.D., PH.D.,* VOJISLAV VUKOVIC, M.D., PH.D.,* MELANIA PINTILIE, M.SC.,† ANTHONY W. FYLES, M.D.,‡ MICHAEL MILOSEVIC, M.D.,‡ RICHARD P. HILL, PH.D.,* AND DAVID W. HEDLEY, M.D.*§㛳 Departments of *Medical Biophysics, †Biostatistics, ‡Radiation Oncology, §Medical Oncology and Hematology, and 㛳Oncologic Pathology, Ontario Cancer Institute/Princess Margaret Hospital, Toronto, Ontario, Canada Purpose: To investigate the relations between hypoxia-inducible factor-1 (HIF-1), tumor oxygenation, and clinical correlates in patients with locally advanced carcinoma of the uterine cervix. Methods and Materials: Biopsies from 42 patients with invasive cervical carcinoma and previous polarographic O2 measurements were assessed for the expression of HIF-1␣ using digitized microscopic imaging and analysis. Results: The HIF-1␣ expression levels ranged from <0.1% to 10.7% of the total tumor area; the positive staining was localized exclusively to the nuclei. Three distinct arrangement patterns of HIF-1␣–positive cells in relation to blood vessels were identified using spatial image mapping: (1) most HIF-1␣–positive cells were located within the typical oxygen diffusion distance in tissue (<150 ␮m to the nearest blood vessel); (2) most HIF-1␣–positive cells were located in the vicinity (<60 ␮m) of the blood vessels; and (3) no apparent spatial relationship was found between HIF-1␣–positive cells and blood vessels. A statistically significant association was found between HIF-1␣ expression and tumor oxygenation (Spearman correlation coefficient ⴝ 0.4, p <0.01), as determined with the Eppendorf pO2 histograph. No correlation was found between the level of HIF-1␣ expression and patient outcome, using disease-free survival as the end point. Conclusion: Our results suggest that HIF-1␣ expression may represent a useful biologic marker for hypoxia in uterine cervical cancer. © 2002 Elsevier Science Inc. Image analysis, Hypoxia, HIF-1␣, Eppendorf pO2 histograph, Cervical carcinoma.

[5]). A large number of genes with the potential to affect the growth and survival of tumor cells are upregulated in malignant cells exposed to hypoxia (6 – 8). Several genes involved in erythropoiesis, angiogenesis, and cellular energy metabolism are activated by a common transcription factor termed hypoxia-inducible factor-1 (HIF-1) (9, 10). HIF-1 is a heterodimer consisting of HIF-1␣ and HIF-1␤ subunits, both belonging to a group of transcription factors termed bHLH/PAS proteins. The biologic activity of HIF-1 is determined by the expression and activity of HIF-1␣. HIF-1␣ protein levels are increased in response to decreased cellular O2 concentrations (11), because the HIF-1␣ protein is rapidly ubiquitinated and degraded under aerobic conditions (12, 13). HIF-1 binds to hypoxia response element DNA sequences found in the promoter regions of genes involved in increased delivery of O2 or metabolic adaptation to O2 deprivation (14). These gene products include, among others, erythropoietin (15), vascular endothelial growth fac-

INTRODUCTION Solid tumors often contain regions of hypoxia caused by differences in the O2 carrying capacity of blood, by differences in O2 consumption, and by structural and functional abnormalities of their vasculature. The prognostic significance of hypoxia for human cancer therapy has been demonstrated in several studies using polarographic needle electrodes for measurements of tumor oxygenation. The degree of pretreatment hypoxia has been found to correlate with reduced disease control and metastasis in cervical cancers (1, 2), soft tissue sarcomas (3), and head-and-neck cancers (4). Therefore, tumor hypoxia is associated with resistance to therapy and is predictive of an unfavorable clinical outcome. The increased malignant potential associated with reduced tumor oxygenation is likely mediated by several mechanisms (i.e., genomic instability, inactivation of metastasis suppressor genes, and enhanced gene expression

funds raised by the Terry Fox run. Acknowledgments—We greatly appreciate the technical assistance of Kiyana Javier, Trudey Nicklee, and David Ma. Received Oct 8, 2001, and in revised form Feb 4, 2002. Accepted for publication Feb 27, 2002.

Reprint requests to: David W. Hedley, M.D., Department of Medical Biophysics, Ontario Cancer Institute/Princess Margaret Hospital, 610 University Ave., Toronto, Ontario M5G 2M9 Canada. Tel: 416-946-2262; Fax: 416-946-2984; E-mail: david. [email protected] Supported by the National Cancer Institute of Canada, using 854

HIF-1␣ expression in cervical carcinoma

tor (16) and vascular endothelial growth factor receptor FLT-1 (17), insulin-like growth factor-2 (18), and various glycolytic enzymes (14, 18). Tumor oxygenation is assessed clinically using the Eppendorf pO2 histograph (1, 2). More recently, nitroimidazole derivatives (e.g., EF5 and pimonidazole) have been used to determine tumor hypoxia in patients (19 –21). To the best of our knowledge, no studies have reported the relationship between tumor hypoxia and HIF-1 expression in experimental or patient tumors. We therefore related the expression levels of HIF-1␣ in biopsies obtained from patients with human cervical carcinoma to tumor oxygenation, as determined by the Eppendorf pO2 histograph. Individual microscopic fields of view were tiled to obtain composite images of the biopsy cut sections. HIF-1␣ spatial arrangement profiles were determined by analyzing its expression as a function of distance to the nearest blood vessel. The percentage of HIF-1␣–positive areas was determined in digitally captured images and related to tumor oxygenation and clinical outcome.

METHODS AND MATERIALS Patients Tumor biopsies were available from a total of 101 patients diagnosed with cervical carcinoma between 1994 and 1999. These patients were participating in a prospective study designed to examine the effect of hypoxia on outcome in cervical cancer patients treated with radiation conducted at the Princess Margaret Hospital (2). Patient consent for all research procedures was obtained according to institutional ethical guidelines. A preliminary analysis of HIF-1␣ expression revealed that the mean expression levels were significantly higher (approximately 2 times) in patients enrolled during the first 3 years of the study (1994 –1996) than in patients enrolled during the period of 1997–1999. This observation coincided with a change in tissue-processing protocols at the Princess Margaret Hospital that resulted in extended fixation times. The negative effect of prolonged fixation on HIF-1␣ antigen detection has recently been reported by Talks et al. (22). Therefore, our analysis was limited to the tumor biopsies from a total of 45 patients enrolled in the study between June 1994 and December 1996. Tumor size and stage were assessed prospectively, according to the Federation International de Gynecologie et d’Obstetrique (FIGO) system. Nodal status was determined by staging using CT or MRI. The patients underwent pretreatment oxygen measurements of the tumor using a computerized polarographic oxygen needle electrode (Eppendorf, Hamburg, Germany); punch biopsies were collected from the surface of the visible tumor immediately after the oxygen measurements. All patients were subsequently treated with radiotherapy alone (2).

● H. K. HAUGLAND et al.

855

Immunohistochemical staining Three serial sections, each 3 ␮m thick, were cut from paraffin-embedded tumor tissue; one section was stained with hematoxylin and eosin and was used for histologic evaluation of tissue morphology. The next section was stained with an anti-HIF-1␣ monoclonal antibody (dilution 1:1000, Novus Biologicals, Littleton, CO) using the Catalyzed Signal Amplification System (Dako, Carpinteria, CA). In brief, after deparaffinization and hydration, the sections were subjected to antigen retrieval by pressure cooking for 20 min in citrate buffer, pH 6.0, followed by anti–HIF-1␣ monoclonal antibody incubation (30 min). Hematoxylin was used as a light nuclear counterstain. Negative controls were done using an IgG2b isotype antibody (Dako), ensuring the same concentration of immunoglobins as for the anti–HIF-1␣. The third section was stained for the endothelial marker CD31 (clone JC/70A, Dako). In brief, sections were treated with 0.4% pepsin in phosphate-buffered saline for 10 min in a 42°C waterbath. After blocking for endogenous peroxidase activity, the sections were incubated with a multi-species blocking agent (Signet Laboratories, Dedham, MA), followed by incubation with the anti-CD31 monoclonal antibody (Dako, dilution 1:50) for 1 h. Next, the sections were incubated with a goat antimouse antibody conjugated to horseradish peroxidase (Dako, dilution 1:200); the reaction product was visualized with AEC (Dako). Hematoxylin was used as a light nuclear counterstain. Negative controls were done by omitting the primary antibody. Image acquisition, processing, and analysis HIF-1␣–stained sections were imaged using the MicroComputer Image Device software package (Imaging Research, St. Catharines, ON) linked to a Sony DXC-970 MD, 3CCD color video camera (Sony Corporation, Tokyo, Japan) mounted on a Zeiss Axioskop microscope (Carl Zeiss, Oberkochen, Germany) fitted with a Ludl Biopoint motorized stage (Ludl Electronic Products, Hawthorne, NY). Using a 10⫻ objective lens with a numeric aperture of 0.25 and an automated mini-program, a microscopic field-byfield digitized tiled image of the entire tumor section was generated. Digitally captured images of the tumor sections were stored as 24-bit color TIFF files; primary image processing was done using the Adobe PhotoShop software, version 5.0 (Adobe Systems, San Jose, CA). Obvious areas of necrosis and hemorrhage were excluded from the analysis. The 24-bit images were color segmented to facilitate extraction of HIF-1␣–positive nuclei from the background. Subsequently, the color images were converted to an 8-bit grayscale and then to binary images, using constant threshold values. Finally, the images were “closed” (erosion of image followed by dilatation) with a 2 ⫻ 2 square shape operator to remove noise originating from nonspecific staining. A copy of the original image was used to create a tissue “mask”; the 24-bit image was converted to grayscale, contrast enhanced, inverted, and converted to a binary image using 2 times the average background brightness value in

856

I. J. Radiation Oncology

● Biology ● Physics

corner regions of the image to set the threshold. This tissue “mask” was used to determine the number of pixels corresponding to the total tissue area. HIF-1␣ expression was measured as percentage of the positively labeled tumor area, because this parameter intuitively relates to the number of HIF-1␣–positive cells. To normalize the nuclear area (the directly measured parameter) to the total tumor area, a region of high HIF-1␣ expression was selected in every image, and the local ratio of HIF-1␣–positive to HIF-1␣–negative pixels was determined. The total number of HIF-1␣–positive pixels in every image was multiplied by this ratio (⬃3), to obtain the total HIF-1␣–positive tumor area. Using a software module written in the Interactive Data Language, version 5.3 (IDL, Boulder, CO), the number of pixels corresponding to the HIF-1␣–positive area (see above) was divided by the number of pixels corresponding to the total tumor area to obtain the percentage of HIF-1␣–positive tumor areas. To relate the location of HIF-1␣ staining to blood vessels, images of parallel sections stained for HIF-1␣ and CD31 were aligned with the MicroComputer Image Device software, using features of the tumor sections as fiducial points. Twelve cases were selected for this study using two criteria: (1) high HIF-1␣ expression (⬎2% of total analyzed tumor area), and (2) optimal software alignment of CD31 and HIF-1␣ images. To remove noise originating from nonspecific staining, binary images were “closed” using a 3 ⫻ 3 square shape operator. The spatial arrangement of HIF-1␣ and blood vessel images was analyzed on a pixel-by-pixel basis. The addresses (X and Y coordinates) of all positive pixels in the HIF-1␣ images were calculated. The same procedure was done for the positive pixels in the CD31 images. Subsequently, for every HIF-1␣–positive pixel, the euclidean distance to the nearest positive pixel in the corresponding CD31 image was determined. The resulting set of pixel-to-pixel distances was used to generate histograms of the distances of HIF-1␣ to the nearest blood vessel. Statistical analysis The statistical analysis consisted of two parts. First, analysis of the association between HIF-1␣ staining and the clinical factors, including hypoxia, and second, analysis of the relationship between HIF-1␣ staining and patient outcome. The first part was done using cross tabulations and the chi-square test or determination of the Spearman correlation coefficient. The clinical factors considered were age, FIGO stage, nodal status, tumor size, and histologic features. HIF-1␣ was tested as a continuous and categorical variable, dichotomized at a value of 2% (median). The effect of HIF-1␣ expression on outcome was tested using the Cox proportional hazards model. The outcome end point was disease-free survival (DFS), defined as the time between diagnosis and the first event: relapse or death from any cause. Four patients who had metastases at the time of the primary diagnosis were not included in the outcome analysis. The median follow-up time was 3.6 years (range 1.1–5.4). Only 3 patients had ⬎1 year since their last

Volume 53, Number 4, 2002

follow-up visit, and 1 of them had experienced an event. The proportional hazard assumption was investigated using the scaled Schoenfeld residuals (23). HIF-1␣ was tested by itself in the model and also with adjustment for tumor size and nodal status, known prognostic factors in this group of patients. The DFS graph is based on Kaplan–Meier estimates. The software used for the analysis was Statistical Analysis System and S-Plus. RESULTS HIF-1␣ expression HIF-1␣ protein was detected as a dark brown precipitate in all 45 tumors examined; the staining was confined to the tumor cell nuclei without any consistent staining of the cytoplasm (Fig. 1). In addition, a positive reaction product could be seen in a few stromal cells. The staining intensity varied within the same section and between sections. For image analysis, the threshold was set to include the weakest staining that gave a consistent nuclear signal in each individual section compared with the negative control section. The percentage of tumor area staining positively for HIF-1␣ ranged between ⬍0.1% and 10.7% (median 2%). Staining with the isotype control antibody produced only a weak background staining (data not shown). The spatial arrangement of HIF-1␣ expression indicated a nonuniform, clustered distribution across the tumor areas. In a number of cases, HIF-1␣–positive cells were located in the immediate vicinity of blood vessels, consistent with the presence of intermittent (perfusion-limited) hypoxia (Fig. 1A). In another group of tumors, HIF-1␣–positive cells were arranged around areas of tissue necrosis. This observation is consistent with HIF-1␣ expression in tumor cells located in areas of chronic hypoxia (Fig. 1B). In some tumors, clusters of HIF-1␣–positive cells were located both in close proximity and farther away from blood vessels (Fig. 1C). The HIF-1␣ expression pattern in relation to tumor blood vessels was characterized by determining the distances from HIF-1␣–positive cells to the nearest blood vessel using spatial image mapping. This analysis identified three distinct arrangement patterns of HIF-1␣–positive cells: (1) most HIF-1␣–positive cells were located in the vicinity of the blood vessels (ⱕ60 ␮m to the nearest blood vessel, Fig. 2A), (2) most HIF-1␣–positive cells were located within the typical oxygen diffusion distance in tissue (ⱕ150 ␮m to the nearest blood vessel, Fig. 2B), and (3) no spatial relationship was found between HIF-1␣–positive cells and the blood vessels (Fig. 2C). Relation of HIF-1␣ expression to tumor oxygenation status and parameters of clinical outcome The relationship between HIF-1␣ expression and the clinical/laboratory parameters is shown in Table 1. Data on the tumor oxygenation were available for 42 of the analyzed biopsies. A positive correlation was found between HIF-1␣ expression and tumor oxygenation, as determined with the

HIF-1␣ expression in cervical carcinoma

● H. K. HAUGLAND et al.

857

Fig. 2. Distance profiles for HIF-1␣ to blood vessels. Histograms of distances of HIF-1␣ to blood vessels are shown; lines represent individual tumors. (A) Most HIF-1␣–positive cells are located in the immediate vicinity of blood vessels. (B) Most HIF-1␣–positive cells are located within the typical oxygen diffusion distance in tissue (0 –150 ␮m). (C) HIF-1␣–positive cells are distributed relatively uniformly with respect to the distance to the nearest blood vessel.

Eppendorf pO2 histograph (Spearman correlation coefficient ⫽ 0.4, p ⬍0.01; Fig. 3). No statistically significant correlation was found between HIF-1␣ expression and patient age, disease stage, histologic features, nodal status at diagnosis, or tumor size.

Fig. 1. Overlayed images of serial tumor sections immunohistochemically stained for HIF-1␣ and CD31. (A) HIF-1␣–positive cells (darkly stained nuclei) in close proximity to blood vessels (red staining). Scale bar 100 ␮m. (B) HIF-1␣–positive cells surrounding an area of necrosis; note distance from HIF-1␣–positive cells to blood vessels. Scale bar 200 ␮m. (C) Presence of HIF1␣–positive cells in the vicinity and further away from blood vessels. Scale bar 200 ␮m.

858

I. J. Radiation Oncology

● Biology ● Physics

Table 1. Distribution of HIF-1␣: Clinical and biologic characteristics of cervical carcinomas included in study (n ⫽ 45) HIF-1␣ expression Clinical variable HIF-1␣ (%) Median Range Age (y) ⱕ60 ⬎60 Median Stage IB and IIA IIB III Tumor size (cm) ⱕ5 ⬎5 Median Nodal status Negative Positive Histologic features Squamous Adenocarcinoma Adenosquamous Other Hypoxia (HP5)* ⱕ50% ⬎50% Median (%)

ⱕ2%

⬎2%

29 16 49

13 9 56

16 7 49

18 12 15

9 5 8

9 7 7

24 21 5

15 7 5

9 14 6

34 11

16 6

18 5

33 3 7 2

15 2 4 1

18 1 3 1

24 18 40

14 6 30

10 12 53

Overall 2 0.002–10.7

Abbreviation: HP5 ⫽ proportion of measurements ⱕ5 mm Hg. * Tumor oxygen measurements were available from 42 patients.

The Cox proportional hazards model was used for the analysis of outcome (n ⫽ 41, 4 patients were not included in the analysis because of metastatic disease at diagnosis). HIF-1␣ expression had no influence on DFS when tested by itself (Fig. 4), neither was any effect seen when the expression of HIF-1␣ was tested together with tumor size, nodal status, or tumor oxygenation status. DISCUSSION In this study, we quantified the expression levels of HIF-1␣ in cervical carcinomas and related these levels to the tumor oxygenation status, clinical and laboratory parameters, and DFS. In addition, in a subset of the tumors, we analyzed the spatial expression of HIF-1␣ as a function of the distance to the tumor blood vessels. Our results indicate that increased levels of HIF-1␣ are associated with reduced tumor oxygenation. However, no statistically significant correlation was found with patient outcome, presence of nodal metastases at diagnosis, patient age, disease stage at diagnosis, tumor size, or histologic features. The observed HIF-1␣ spatial expression patterns are consistent with the presence of intermittent and chronic hypoxia in human cervical carcinomas.

Volume 53, Number 4, 2002

Several studies have shown a significant relationship between a high fraction of low pO2 readings in solid tumors and patient outcome after radiation alone or radiation combined with other treatment (2– 4, 24, 25). Hockel et al. (1) also observed an association between tumor hypoxia and poor outcome in patients treated with surgery alone or in combination with chemotherapy, suggesting that hypoxia is associated with biologically aggressive malignant disease. In most clinical studies, tumor oxygenation was assessed invasively using a polarographic needle electrode. One alternative to the use of electrodes is the use of agents that, after systemic administration, bind to areas of low oxygenation. Nitroimidazole derivatives (e.g., EF5 and pimonidazole) have been used extensively to visualize tumor hypoxia (19, 20). The obvious advantage of this approach is the ability to relate tumor hypoxia to tissue architecture and exclude areas of necrosis and tumor stroma from analysis; therefore, tumor hypoxia is assessed specifically in viable tumor cells. This is of particular relevance, because polarographic needle electrode measurements do not distinguish between malignant and nonmalignant or viable and necrotic cells and tissues. The other major advantage of the microscopic assessment of oxygenation is the ability to combine measurements of hypoxia with determinations of other markers of interest, and to characterize their spatial relationships. The potential drawback to the use of chemical tissue markers of hypoxia is that the clinical investigations are restricted to prospective studies. Identification of preexisting biologic markers that could substitute for the chemical markers of hypoxia is of particular clinical significance, because this would allow for the assessment of archival material with adequate follow-up. The transcription factor HIF-1␣ is upregulated at low concentrations of oxygen (11). Increased levels of HIF-1␣ have been found in a number of human tumors (26); HIF-1␣ is predominantly expressed in cells located in the invading edge, adjacent to necrotic areas and surrounding areas of neovascularization. In the highly malignant glioblastoma multiforme, the highest expression of HIF-1␣ was found in pseudopalisading tumor cells located next to areas of necrosis (27). Consistent with these observations, we found increased expression of HIF-1␣ at the perimeter of necrotic areas. These histologic observations are consistent with increased HIF-1␣ expression in regions of chronic hypoxia. Furthermore, using dual fluorescence labeling, we have recently shown that the spatial distributions of HIF-1␣ and the nitroimidazole probe EF5 correlate closely in human cervical carcinoma xenografts (28). Considering its protein stabilization and ubiquitination (rapid DNA association and dissociation) kinetics (12, 29, 30), HIF-1␣ may well also be present in regions of intermittent (perfusion-limited) hypoxia. Our observations of HIF-1␣ expression in cells located in the immediate vicinity of blood vessels is indeed suggestive of the presence of tumor regions with intermittent blood flow and transient hypoxia. Furthermore, Rijken et al. (31) found regions labeled with the hypoxia marker pimonidazole within typ-

HIF-1␣ expression in cervical carcinoma

● H. K. HAUGLAND et al.

859

Fig. 3. Correlation between HIF-1␣ expression and tumor oxygenation. Correlation between HIF-1␣ expression, given as the percentage of total tumor area, and tumor oxygenation, given as the hypoxic fraction ⱕ5 mm Hg (HP5) or proportion of pO2 measurements ⱕ5 mm Hg.

ical oxygen diffusion distances in human glioma xenografts. Wijffels et al. (21) have related pimonidazole staining patterns to the distance to blood vessels and found three distinct patterns of pimonidazole binding: in close proximity to blood vessels (⬍50 ␮m), at intermediate distances (50 –200 ␮m), and beyond the oxygen diffusion distance in tissue

(⬎200 ␮m), as assessed in two-dimensional images. We observed identical patterns of HIF-1␣ expression in relation to blood vessels, suggesting the potential usefulness of HIF-1␣ as a biologic marker for hypoxia. In the present study, we did not find a significant correlation between HIF-1␣ expression and clinical parameters,

Fig. 4. Effect of HIF-1␣ expression on DFS.

860

I. J. Radiation Oncology

● Biology ● Physics

DFS, nodal metastasis, or tumor size. This finding is in contrast to results published recently by Birner et al. (32). Using a different staining procedure and scoring system, these authors found a positive correlation between high levels of HIF-1␣ expression and decreased DFS and overall survival of patients with early-stage invasive cervical carcinoma. The major differences between their report and our study were the absolute levels of HIF-1␣ expression and the tumor stage of the patients included in the study. Birner et al. (32) found a strong HIF-1␣ expression (⬎50% HIF-1␣– positive cells) in most of their analyzed cases (⬃63%). Because the HIF-1␣ protein is effectively stabilized at low pO2 levels (maximal expression at 0.5% O2 [11]), their results imply the presence of significant hypoxia in most of the analyzed tumors; a finding not supported by studies on the oxygenation status of cervical carcinomas using the

Volume 53, Number 4, 2002

polarographic Eppendorf histograph (1, 2). In comparison, the highest expression observed in our study was ⬃11% HIF-1␣–positive cells, even though we used an amplification method to enhance the HIF-1␣ staining intensity. Birner et al. (32) studied cervical carcinoma Stage T1b, and our study included tumor Stages IB–III; the association of HIF-1␣ expression levels with DFS and overall survival in early-stage invasive cervical carcinomas may, therefore, suggest a role for HIF-1␣ in tumor progression. Alternatively, the lack of an association between HIF-1␣ expression and patient outcome seen in the present study might reflect the problems of sampling error due to intratumoral heterogeneity of hypoxia, as has been previously reported (33). Additional studies are warranted to establish whether HIF-1␣ can be used as a marker for hypoxia in human tumor tissue.

REFERENCES 1. Hockel M, Schlenger K, Aral B, et al. Association between tumor hypoxia and malignant progression in advanced cancer of the uterine cervix. Cancer Res 1996;56:4509 – 4515. 2. Fyles AW, Milosevic M, Wong R, et al. Oxygenation predicts radiation response and survival in patients with cervix cancer [published erratum appears in Radiother Oncol 1999;50:371]. Radiother Oncol 1998;48:149 –156. 3. Brizel DM, Scully SP, Harrelson JM, et al. Tumor oxygenation predicts for the likelihood of distant metastases in human soft tissue sarcoma. Cancer Res 1996;56:941–943. 4. Brizel DM, Sibley GS, Prosnitz LR, et al. Tumor hypoxia adversely affects the prognosis of carcinoma of the head and neck. Int J Radiat Oncol Biol Phys 1997;38:285–289. 5. Rofstad EK. Microenvironment-induced cancer metastasis [In Process Citation]. Int J Radiat Biol 2000;76:589 – 605. 6. Dachs GU, Stratford IJ. The molecular response of mammalian cells to hypoxia and the potential for exploitation in cancer therapy. Br J Cancer Suppl 1996;27:S126 –S132. 7. Sutherland RM, Ausserer WA, Murphy BJ, et al. Tumor hypoxia and heterogeneity: Challenges and opportunities for the future. Semin Radiat Oncol 1996;6:59 –70. 8. Blancher C, Harris AL. The molecular basis of the hypoxia response pathway: Tumor hypoxia as a therapy target. Cancer Metastasis Rev 1998;17:187–194. 9. Wang GL, Semenza GL. Purification and characterization of hypoxia-inducible factor 1. J Biol Chem 1995;270:1230 – 1237. 10. Wenger RH. Mammalian oxygen sensing, signaling and gene regulation. J Exp Biol 2000;203:1253–1263. 11. Jiang BH, Semenza GL, Bauer C, et al. Hypoxia-inducible factor 1 levels vary exponentially over a physiologically relevant range of O2 tension. Am J Physiol 1996;271:C1172– C1180. 12. Huang LE, Gu J, Schau M, et al. Regulation of hypoxiainducible factor 1alpha is mediated by an O2-dependent degradation domain via the ubiquitin-proteasome pathway. Proc Natl Acad Sci USA 1998;95:7987–7992. 13. Kallio PJ, Wilson WJ, O’Brien S, et al. Regulation of the hypoxia-inducible transcription factor 1alpha by the ubiquitinproteasome pathway. J Biol Chem 1999;274:6519 – 6525. 14. Semenza GL, Jiang BH, Leung SW, et al. Hypoxia response elements in the aldolase A, enolase 1, and lactate dehydrogenase A gene promoters contain essential binding sites for hypoxia-inducible factor 1. J Biol Chem 1996;271:32529 – 32537.

15. Semenza GL, Wang GL. A nuclear factor induced by hypoxia via de novo protein synthesis binds to the human erythropoietin gene enhancer at a site required for transcriptional activation. Mol Cell Biol 1992;12:5447–5454. 16. Forsythe JA, Jiang BH, Iyer NV, et al. Activation of vascular endothelial growth factor gene transcription by hypoxia-inducible factor 1. Mol Cell Biol 1996;16:4604 – 4613. 17. Gerber HP, Condorelli F, Park J, et al. Differential transcriptional regulation of the two vascular endothelial growth factor receptor genes: Flt-1, but not Flk-1/KDR, is up-regulated by hypoxia. J Biol Chem 1997;272:23659 –23667. 18. Feldser D, Agani F, Iyer NV, et al. Reciprocal positive regulation of hypoxia-inducible factor 1alpha and insulin-like growth factor 2. Cancer Res 1999;59:3915–3918. 19. Evans SM, Hahn S, Pook DR, et al. Detection of hypoxia in human squamous cell carcinoma by EF5 binding. Cancer Res 2000;60:2018 –2024. 20. Varia MA, Calkins-Adams DP, Rinker LH, et al. Pimonidazole: A novel hypoxia marker for complementary study of tumor hypoxia and cell proliferation in cervical carcinoma. Gynecol Oncol 1998;71:270 –277. 21. Wijffels KI, Kaanders JH, Rijken PF, et al. Vascular architecture and hypoxic profiles in human head and neck squamous cell carcinomas [In Process Citation]. Br J Cancer 2000;83:674 – 683. 22. Talks KL, Turley H, Gatter KC, et al. The expression and distribution of the hypoxia-inducible factors HIF-1alpha and HIF-2alpha in normal human tissues, cancers, and tumorassociated macrophages. Am J Pathol 2000;157:411– 421. 23. Grambsch P, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994;81: 515–526. 24. Hockel M, Knoop C, Schlenger K, et al. Intratumoral pO2 predicts survival in advanced cancer of the uterine cervix. Radiother Oncol 1993;26:45–50. 25. Nordsmark M, Overgaard M, Overgaard J. Pretreatment oxygenation predicts radiation response in advanced squamous cell carcinoma of the head and neck. Radiother Oncol 1996; 41:31–39. 26. Zhong H, De Marzo AM, Laughner E, et al. Overexpression of hypoxia-inducible factor 1alpha in common human cancers and their metastases. Cancer Res 1999;59:5830 –5835. 27. Zagzag D, Zhong H, Scalzitti JM, et al. Expression of hypoxia-inducible factor 1alpha in brain tumors: Association

HIF-1␣ expression in cervical carcinoma

with angiogenesis, invasion, and progression. Cancer 2000; 88:2606 –2618. 28. Vukovic V, Haugland HK, Nicklee T, et al. Hypoxia-inducible factor-1␣ is an intrinsic marker for hypoxia in cervical cancer xenografts. Cancer Res 2001;61:7394 –7398. 29. Wang GL, Jiang BH, Rue EA, et al. Hypoxia-inducible factor 1 is a basic-helix-loop-helix-PAS heterodimer regulated by cellular O2 tension. Proc Natl Acad Sci USA 1995;92:5510 –5514. 30. Huang LE, Arany Z, Livingston DM, et al. Activation of hypoxia-inducible transcription factor depends primarily upon redox-sensitive stabilization of its alpha subunit. J Biol Chem 1996;271:32253–32259.

● H. K. HAUGLAND et al.

861

31. Rijken PF, Bernsen HJ, Peters JP, et al. Spatial relationship between hypoxia and the (perfused) vascular network in a human glioma xenograft: A quantitative multi-parameter analysis (1). Int J Radiat Oncol Biol Phys 2000;48:571–582. 32. Birner P, Schindl M, Obermair A, et al. Overexpression of hypoxia-inducible factor 1alpha is a marker for an unfavorable prognosis in early-stage invasive cervical cancer. Cancer Res 2000;60:4693– 4696. 33. Thrall DE, Rosner GL, Azuma C, et al. Hypoxia marker labeling in tumor biopsies: Quantification of labeling variation and criteria for biopsy sectioning. Radiother Oncol 1997;44: 171–176.