Diffusion limited hypoxia estimated by vascular image analysis: comparisonwith pimonidazole staining in human tumors

Purpose: To assess diffusion limited hypoxia in human tumors using image an alysis of vasculature and to compare it with the bioreductive marker pimoni dazole as an independent method. Materials and methods: To set up the method, nine rectal adenocarcinomas an d ten squamous cell carcinomas were analyzed. To validate the method, ten s quamous cell carcinomas of the cervix were analyzed from patients who were injected with pimonidazole and biopsied approximately 24 h later. Sections of the rectal and esophageal tumors were stained for vasculature, while cer vix tumor sections were double stained for vasculature and pimonidazole. Tu mor areas were delineated on digitized images, and the proportion of tumor tissue greater than a fixed distance from the nearest blood vessel (called diffusion limited fraction, DLF) was then calculated. The proportion of tum or area stained for pimonidazole was also measured. Results: There was a wide variation between turners in both the vascular-de rived DLF and in the pimonidazole-stained fraction. Average DLFs varied bet ween 1.5 and 92% for different tumors, with significant differences between them. The area stained by pimonidazole was significantly smaller than DLF for all tumors. The correlation between pimonidazole ar ea and DLF was sign ificant in three of seven tumors containing greater than or equal to 3 imag es. When images from all tumors (n = 123) were analyzed together, the corre lation was highly significant (r = 0.47, P < 0.0001). Conclusion: The vascular derived DLF correlates significantly with pimonida zole staining, but then was large scatter. Both methods may underestimate p erfusion limited hypoxia. (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.