Intrinsic Susceptibility MR Imaging of Chemically Induced Rat Mammary Tumors: Relationship to Histologic Assessment of Hypoxia and Fibrosis

Purpose: To investigate relationships between magnetic resonance (MR) imaging measurements of R2* and carbogen-induced Delta R2* in vivo with subsequent histologic assessment of grade, hypoxia, fibrosis, and necrosis in a chemically induced rat mammary tumor model. Materials and Methods: All experiments were performed in accordance with the local ethics review panel, the UK Home Office Animals Scientific Procedures Act of 1986, and the UK Co-ordinating Committee on Cancer Research guidelines. Of 30 rats injected with N-methyl-N-nitrosourea, 17 developed mammary tumors. Prior to MR imaging, rats were administered pimonidazole. Tumor R2* was then quantified while the host first breathed air and then carbogen (95% O-2, 5% CO2; n = 16). Tumor sections were subsequently stained for pimonidazole, sirius red, cytokeratin 14, and hematoxylin-eosin for quantitative assessment of hypoxia, fibrosis, malignancy, and necrosis, respectively, and graded according to the Scarff-Bloom-Richardson scale. Linear regression analysis was used to identify any correlates of the MR imaging data with histologic data. Results: Tumors exhibited wide heterogeneity in the magnitude of carbogen-induced reduction in R2*, an emerging imaging biomarker of fractional blood volume. Significant correlations were found between pimonidazole adduct formation and both baseline tumor R2* (r = -0.54, P=.03) and carbogen-induced Delta R2* (r = 0.56, P=.02), demonstrating that tumors with a larger fractional blood volume were less hypoxic. There was also a significant correlation between pimonidazole and sirius red staining (r = 0.76, P<.01), indicating that more fibrotic tumors were also more hypoxic. There were no correlations of R2* with grade. Conclusion: In this model of breast cancer, baseline tumor R2* and carbogen-induced Delta R2* are predictive imaging biomarkers for hypoxia and primarily determined by blood volume. (C) RSNA, 2010