Multiparametric MR and PET Imaging of Intratumoral Biological Heterogeneity in Patients with Metastatic Lung Cancer Using Voxel-by-Voxel Analysis.

OBJECTIVES: Diffusion-weighted magnetic resonance imaging (DW-MRI) and imaging of glucose metabolism by positron emission tomography (FDG-PET) provide quantitative information on tissue characteristics. Combining the two methods might provide novel insights into tumor heterogeneity and biology. Here, we present a solution to analyze and visualize the relationship between the apparent diffusion coefficient (ADC) and glucose metabolism on a spatially resolved voxel-by-voxel basis using dedicated quantitative software.

MATERIALS AND METHODS: In 12 patients with non small cell lung cancer (NSCLC), the primary tumor or metastases were examined with DW-MRI and PET using 18F-fluorodeoxyglucose (FDG). The ADC's from DW-MRI were correlated with standardized-uptake-values on a voxel-by-voxel basis using custom made software (Anima M3P). For cluster analysis, we used prospectively defined thresholds for 18F-FDG and ADC to define tumor areas of different biological activity.

RESULTS: Combined analysis and visualization of ADC maps and PET data was feasible in all patients. Spatial analysis showed relatively homogeneous ADC values over the entire tumor area, whereas FDG showed a decreasing uptake towards the tumor center. As expected, restricted water diffusivity was notable in areas with high glucose metabolism but was also found in areas with lower glucose metabolism. In detail, 72% of all voxels showed low ADC values (<1.5x10(-3) mm2/s) and high tracer uptake of 18F-FDG (SUV>3.6). However, 83% of the voxels with low FDG uptake also showed low ADC values, increasingly towards the tumor center.

CONCLUSIONS: Multiparametric analysis and visualization of DW-MRI and FDG-PET is feasible on a spatially resolved voxel-by-voxel respectively cluster basis using dedicated imaging software. Our preliminary data suggest that water diffusivity and glucose metabolism in metastatic NSCLC are not necessarily correlated in all tumor areas.