GLUT-1 AND CAIX AS INTRINSIC MARKERS OF HYPOXIA IN CARCINOMA OF THE CERVIX: RELATIONSHIP TO PIMONIDAZOLE BINDING Rachel E. AIRLEY 1,2,6 * , Juliette LONCASTER 2 , James A. RALEIGH 3 , Adrian L. HARRIS 4 , Susan E. DAVIDSON 5 , Robert D. HUNTER 5 , Catharine M.L. WEST 2,5 and Ian J. STRATFORD 6 1 School of Pharmacy and Chemistry, Liverpool John Moores University, Liverpool, United Kingdom 2 Paterson Institute, Christie Hospital, Withington, Manchester, United Kingdom 3 Department of Radiation Oncology, University of North Carolina School of Medicine, Chapel Hill, NC, USA 4 Wetherall Institute of Molecular Medicine, John Ratcliffe Hospital, Oxford, United Kingdom 5 Department of Academic Radiation Oncology, Christie Hospital, Manchester, United Kingdom 6 School of Pharmacy and Pharmaceutical Sciences, University of Manchester, Manchester, United Kingdom The presence of hypoxia in tumours results in the overex- pression of certain genes, which are controlled via the tran- scription factor HIF-1. Hypoxic cells are known to be radio- resistant and chemoresistant, thus, a reliable surrogate marker of hypoxia is desirable to ensure that treatment may be rationally applied. Recently, the HIF-1-regulated proteins Glut-1 and CAIX were validated as intrinsic markers of hyp- oxia by comparison with pO 2 measured using oxygen elec- trodes. We compare the expression of Glut-1 and CAIX with the binding of the bioreductive drug hypoxia marker pi- monidazole. Pimonidazole was administered to 42 patients with advanced carcinoma of the cervix, 16 hr before biopsy. Sections of single or multiple biopsies were then immuno- stained for Glut-1 and CAIX, and the area of staining scored by eye, using a “field-by-field” semi-quantitative averaging system. Using 1 biopsy only, Glut-1 (r  0.54, p  <0.001) correlated with the level of pimonidazole binding, and Glut-1 and CAIX expression also correlated significantly (r  0.40, p  <0.009). Thus, our study has shown that HIF-1 regulated genes have potential for future use as predictors of the ma- lignant changes mediated by hypoxia, and warrant further investigation as indicators of response to cancer therapy. © 2002 Wiley-Liss, Inc. Key words: GLUT-1; CAIX; hypoxia; pimonidazole Tumour hypoxia occurs as a consequence of an inadequate supply of blood borne oxygen due to the disorganized and chaotic vascular network that develops in tumours. 1 Traditionally, tumour hypoxia has been considered to be a potential therapeutic problem because hypoxic cells are radiation resistant, 2,3 and recent mea- surements of tumour oxygenation in the clinic have shown clear correlations with outcome of radiotherapy. 4–8 The role of tumour hypoxia for the outcome of chemotherapy is less clear, but there are certain reasons why hypoxia may be important. For example, drugs like bleomycin have an obligate requirement for oxygen to be toxic. Chronically hypoxic cells tend to be out of cycle and resistant to cell cycle phase specific drugs. Finally, Because chron- ically-hypoxic cells lie some 100 –150 m from a functional vascular capillary, many chemotherapeutic drugs find it extremely difficult to diffuse that distance in sufficient concentrations to be toxic. Clinical observations have suggested hypoxia can cause cellular changes that result in a more aggressive phenotype. 9 Such hypoxia-related malignant progression may result in increased potential for local invasiveness and metastatic spread. 10 –12 Results from experimental systems support the notion that these changes are indeed hypoxia-mediated. 13,14 There have been many attempts to overcome “the hypoxic cell problem”, but with limited success. Overgaard and Horsman 15 have reviewed the clinical literature where “modification of hyp- oxia” has been the goal of the trials in radiotherapy. Meta-analysis of 37 trials carried out with 5,380 patients in a variety of tumour types treated with hypoxic cell radiosensitizers showed a signifi- cant overall local control benefit of 4% for the “modification” group, but in few of the individual trials was this significant benefit apparent. A potential reason for this is that not all the patients in the trials would have benefited from hypoxia modification, a conclusion that is consistent with the heterogeneity of pO 2 mea- surements seen between and within tumour types, and the depen- dency of treatment outcome on initial pO 2 16 . Therefore, there is a need to identify those patients most likely to respond to adjuvant therapy directed toward hypoxic cells. Such therapeutic ap- proaches could include methods to improve tumour oxygenation such as carbogen, 17 the use of bioreductive drugs 18 or hypoxia- mediated gene therapy approaches. 19 Currently, the method that has received most widespread use to measure tumour oxygenation is the Eppendorf polarographic ox- ygen electrode. 20 It has limited use, however, in that it is invasive and can only be used on relatively superficial tumours. Neverthe- less, this method of pO 2 measurement has generated many corre- lations with cancer treatment outcome and any subsequent meth- ods of hypoxia measurement must be compared to it. Bioreductive drug markers provide an alternative approach for assessing the level and extent of tumour hypoxia. This was first articulated by Chapman et al. 21 who used radio-labeled analogues of the 2-nitroimidazole, misonidazole. The development of immu- nochemical detection 22 permitted the use of related compounds such as pimonidazole, 23 EF-5 24 and NITP. 25 Fluorinated analogues such as SR4554 26 and CCI-103F 27,28 have also been evaluated using non-invasive imaging procedures. They are all structurally similar and will identify intracellular levels of oxygenation known to cause cellular resistance to radiation. 29 Pimonidazole is at the most advanced stage of clinical evaluation 30,31 and a clinical comparative study has been made recently between Eppendorf measurements of pO 2 and the extent of pimonidazole adduct formation in carcinoma of the cervix. 32 In this latter study, there was a trend for tumours with relatively high pO 2 readings to show Grant sponsor: Medical Research Council; Grant sponsor: Royal Phar- maceutical Society of GB; Grant sponsor: Christie Hospital Endowment Fund; Grant sponsor: Cancer Research UK; Grant sponsor: EORTC Trans- lational Research Fund. The first two authors contributed equally to this paper. *Correspondence to: School of Pharmacy and Chemistry, Liverpool John Moores University, Liverpool L3 3AF, UK. Fax: +44-0-151-231-2170. E-mail: R.Airley@livjm.ac.uk Received 8 July 2002; Revised 18 September 2002; Accepted 1 October 2002 DOI 10.1002/ijc.10904 Int. J. Cancer: 104, 85–91 (2003) © 2002 Wiley-Liss, Inc. Publication of the International Union Against Cancer