Effect of carbogen on tumor oxygenation: combined fluorine-19 and proton MRI measurements.

Purpose: Blood oxygen level dependent (BOLD) contrast in magnetic resonance imaging (MRI) has been widely used for noninvasive evaluation of the effects of tumor-oxygenating agents. However, there have been few tests of the validity of this method. The goal of the present work was to use the T(1) of fluorine-19 in perfluorocarbon (PFC) emulsions as a "gold standard" for comparison with BOLD MRI. MATHODS AND MATERIALS: Rats bearing R3230AC tumors implanted in the hind limb were injected with an emulsion of perfluoro-15-crown-5-ether for 2-3 days before experiments, which ensured that the PFC emulsion concentrated in the tumors. We correlated changes in tumor oxygenation caused by carbogen inhalation measured by (1)H BOLD MRI with quantitative (19)F measurements. The (19)F spin-lattice relaxation rate R(1) (= 1/T(1)) was measured to determine initial oxygen tension (pO(2)) in each image pixel containing the PFC, and changes in pO(2) during carbogen (95% O(2), 5% CO(2)) breathing. In a second carbogen breathing period, changes in water signal linewidth were measured using high spectral and spatial resolution imaging. (19)F and (1)H measurements were used to classify pixels as responders to carbogen (pixels where oxygen increased significantly) or nonresponders (no significant change in tumor oxygenation).

Results: The (19)F and (1)H measurements agreed in 65% +/- 11% of pixels (n = 14). Agreement was even stronger among pixels where (1)H showed increased oxygenation; (19)F measurements agreed with (1)H measurements in over 79% +/- 11% of these pixels. Similarly, there was strong agreement between the two modalities in pixels where (19)F reported no change in pO(2); (1)H also showed no changes in 76% +/- 18% of these pixels. Quantitative correlation of changes T(2)* (DeltaT(2)*) in (1)H and changes R(1) (DeltaR(1)) in (19)F was weak during carbogen breathing, and averaged over the whole tumor was approximately 0.40 for 14 experiments. However, the spatial patterns of (1)H and (19)F changes were qualitatively very similar. In hypoxic regions that were identified based on long (19)F T(1) (>2.53 s), (19)F and (1)H MRI agreed that carbogen had relatively weak effects.

Conclusions: These results suggest that (1)H BOLD MRI reliably identifies increases in tumor pO(2). In hypoxic regions where increases in pO(2) are most desirable, carbogen was ineffective. The data suggest that (19)F and (1)H MRI can be used individually or in combination to guide the design of improved tumor-oxygenating agents.