Purpose: The purpose of this study was to assess the presence of tumor hypoxia using two independent techniques: binding of the 2-nitroimidazole EF5 and Eppendorf needle electrode measurements. The distribution of tumor hypoxia was assessed with respect to tumor necrosis in corresponding histological studies.
Methods and materials: Each of several rats bearing a subcutaneous 9L glioma or Morris 7777 hepatoma tumor was given EF5 i.v. to a final, whole-body concentration of 100 microM. About 2.5 h later, each rat was anesthetized, and needle electrode measurements were made in the tumor along 1-5 tracks (30-200 individual measurements). At 3 h post-EF5 injection, the tumor was excised and frozen. Frozen sections were analyzed for the presence and distribution of binding of EF5 and necrosis using immunohistochemical techniques followed by staining with hematoxylin and eosin (H&E). The histochemical analysis and electrode readings in similar regions of the tumor were compared.
Results: Electrode measurements were taken at 0.4-mm intervals along one-dimensional tracks, whereas EF5 binding measurements from tissue sections contained two-dimensional information at high spatial resolution ( approximately 2.5 micro). The EF5 measurements showed greater spatial heterogeneity than did the electrode measurements. In tumor regions with minimal necrosis, needle tracks with relatively high pO(2) readings were usually found to contain relatively low EF5 binding, and vice versa. Because EF5 binding is inversely related to tissue pO(2), this result was expected. The expected inverse correlation of the two techniques was most disparate in necrotic tumor regions (confirmed by H&E staining), where needle electrode measurements showed low to zero pO(2) values, but little or no EF5 binding was found.
Conclusion: The two methods compared in this study operate in fundamentally different ways and provide substantially different information. EF5 binding provided detailed spatial information on the distribution of hypoxia in viable tumor tissue. There was no EF5 binding in necrotic tumor tissue because cells in such tissue were unable to metabolize the drug. In contrast, output from the needle electrode method appeared to represent a "track-average" tissue pO(2) and did not distinguish between extreme hypoxia and either macroscopic or microscopic necrosis. At the present time, the importance of tumor necrosis in determining treatment response is unknown. However, our data suggest that the Eppendorf needle electrode technique will overestimate the presence of hypoxia. Both techniques are potentially limited by sampling errors in tumors with heterogeneous distributions of hypoxia.