Hypoxia is a condition, common to most malignant tumors, where oxygen tension in the tissue is below the physiological level. Among consequences of tumor hypoxia is also altered cancer cell metabolism that contributes to cancer therapy resistance. Therefore, precise assessment of tumor hypoxia is important for monitoring the tumor treatment progression. In this study, we propose a simple model for prediction of hypoxic level in tumors based on multiparametric magnetic resonance imaging. The study was performed on B16F1 murine melanoma tumors ex vivo that were first magnetic resonance scanned and then analyzed for hypoxic level using hypoxia-inducable factor 1-alpha antibody staining. Each tumor was analyzed in identical sections and in identical regions of interest for pairs of hypoxic level and magnetic resonance values (apparent diffusion coefficient and T 2). This was followed by correlation analysis between hypoxic level and respective magnetic resonance values. A moderate correlation was found between hypoxic level and apparent diffusion coefficient (ρ = 0.56, P < .00001) and lower between hypoxic level and T 2 (ρ = 0.38, P < .00001). The data were analyzed further to obtain simple predictive models based on the multiple linear regression analysis of the measured hypoxic level (dependent variable) and apparent diffusion coefficient and T 2 (independent variables). Among the hypoxic level models, the most efficient was the 3-parameter model given by relation (HL = k ADC ADC + k T2 T 2 + b), where k ADC = 26%/µm2/ms, k T2 = 0.8%/ms, and b = −32%. The model can be used for calculation of the predicted hypoxic level map based on magnetic resonance–measured apparent diffusion coefficient and T 2 maps. Similar prediction models, based on tumor apparent diffusion coefficient and T 2 maps, can be done also for other tumor types in vivo and can therefore help in assessment of tumor treatment as well as to better understand the role of hypoxia in cancer progression.
Read full abstract