UC Berkeley

Using monoexponential models, magnetic resonance (MR) parameters transverse relaxation time (T₂) and apparent diffusion coefficient (ADC) have been used to non–invasively assess cell density, which characteristically increases with brain tumor malignancy. Multiexponential models of T₂ and ADC may allow for a more accurate evaluation of cell density than the traditionally used monoexponential model. To better understand how cell density affects multicomponent T₂ and ADC, tumor models with a range of cell densities were created by suspending astrocytoma cells in agarose at different densities. T₂ was measured using a Carr–Purcell–Meiboom–Gill (CPMG) sequence with 64 echo times. ADC was measured using a diffusion–weighted sequence with 32 b–values. Three models were used to fit the data and determine the T₂ values, ADC values, and associated fractions: monoexponential, biexponential, non–negative least squares (NNLS). Spearman Rank was used to test the correlation with cell density.

Both the monoexponential T₂ and ADC were significantly negatively correlated with cell density. The biexponential model identified two T₂ components, short and long. Both T₂ components’ values and fractions were significantly correlated to cell density. The biexponential model identified two ADC components, slow and fast. Both ADC components’ values were significantly correlated with cell density, but not the fractions. The NNLS model identified up to three T₂ and three ADC components. For the NNLS identified T₂s, the components were labeled short, medium, and long. Both the NNLS short and medium T₂ values and fractions were significantly correlated with cell density. For the NNLS identified ADCs, the components were labeled slow, intermediate, and fast. Only the fast fraction was significantly correlated with cell density.

The NNLS components obtained from samples of packed cells were further evaluated by washing the samples with a gadolinium contrast agent (Gd–DTPA) to shorten the T₂ associated with the extracellular space. Gd–DTPA did not affect the short T₂, which was considered to be associated with the intracellular space. The medium T₂ component was no longer identified in samples with Gd–DTPA and was considered to be associated with the extracellular space. Overall, the results suggest that separately measurable components exist that could be utilized for compartment specific information.