Is HIF-1α a good marker for tumor hypoxia?

Abstract

It is now abundantly clear that tumor hypoxia is a strong prognostic factor for outcome in many forms of cancer. High tumor hypoxia is associated with a poor outcome after treatment with any of the three major treatment modalities: surgery, radiotherapy, and chemotherapy. For radiotherapy, an obvious contributing factor will be intrinsic radioresistance of hypoxic cells, a phenomenon known and wellstudied since the first half of last century. For chemotherapy, contributing factors will be reduced drug delivery to hypoxic cells, usually located at a distance from blood vessels, and, for cycle-dependent drugs, the reduced proliferation rate of hypoxic cells. In more recent years, hypoxia has also been shown to influence the invasive and metastatic properties of tumor cells and to lead to the selection of apoptotic-resistant cells, resulting in a more malignant phenotype. This will affect outcome after all treatment forms, including surgery. Eliminating hypoxic cells, or sensitizing them to treatment, is therefore a useful therapeutic goal, at least when drugs or radiation are employed. This concept has been supported by many clinical studies that included a hypoxic modulator and a meta-analysis showing that such additional treatments were more effective. Current ways of attacking the hypoxia problem are the combined use of carbogen and nicotinamide to reduce or eliminate chronic and acutely hypoxic cells, increasing blood oxygen via increasing the hematocrit—for example, with erythropoietin—and administering hypoxic cytotoxins such as Tirapazamine. Unfortunately, many of these treatments have associated toxicities or are expensive. A method to rapidly and quantitatively assess tumor hypoxia before initiating treatment would therefore be a valuable tool. This would potentially lead to the sparing of unnecessary and potentially toxic or expensive treatments to those who are unlikely to benefit from them. Applied to randomized trials, it would also provide a good monitor of whether the proposed hypoxia-directed therapy is actually working through its proposed mechanism (one would expect greater improvements in patients with high hypoxia tumors). The search for a good hypoxia marker has thus occupied many of us for many years. Several nitroimadazole drugs emerged out of the hypoxic cell radiosensitizer program 20 to 30 years ago. Pimonidazole is a good example of this; a failed clinical radiosensitizer but a good hypoxic marker when used at low tracer doses. EF5, a fluorinated nitroimidazole, is another. These compounds are now being extensively tested in the clinic as hypoxic markers and show promise. Their disadvantage is simply the need to administer a drug. It would, of course, be far more convenient if there were gene products that were upregulated under hypoxia and that could be detected by standard immunoperoxidase or immunofluorescence methods with appropriate antibodies. These would be highly useful endogenous markers of hypoxia. The discovery of HIF-1 a decade ago brought with it the hope that here indeed was potentially such an endogenous hypoxic marker. One of the subunits of HIF-1— HIF-1 —is specifically upregulated under hypoxia and rapidly degraded under ambient oxygen conditions. This occurs by largely understood mechanisms involving regulation of protein stability via the ubiquitin pathway, in turn regulated by oxygen dependent hydroxylases (1). There have been some recent papers evaluating HIF-1 as an endogenous hypoxic marker, the latest appearing in this issue by Vordermark and Brown. Although their data broadly support a reasonable correlation between this intrinsic marker and the extrinsic marker pimonidazole, the authors remain somewhat pessimistic in their conclusions about the value of HIF-1 for predictive purposes. Is this pessimism justified? Vordermark and Brown used a human tumor cell line, U87 MG, both in vitro and grown as xenografts, as their test system. A new aspect of this article is that they developed a flow cytometry protocol to quantify the expression of the HIF-1 . They then compared HIF-1 expression with pimonidazole staining, also quantified by flow cytometry. The results can be split into good news and bad news. The good news is that they showed a nice oxygen dependence of HIF-1 expression, with progres-