Figure 3.

Comparison of ²³Na peak separation in rats bearing RG2 and U87 tumors. For rats bearing an (a) RG2 and (b) U87 tumor, the tumor boundary is outlined in white, with voxels of interest indicated in yellow squares (with numbers), and spectra acquired before and after TmDOTP⁵⁻ delivery shown in black and green, respectively. Tumor voxels 3 and 4 in (a) RG2 and (b) U87 tumor rats exhibited a fair amount of peak separation due to the leaky BBB. Na⁺_b shift was consistently around 2 ppm, and Na⁺_i shift was at 0 ppm, whereas Na⁺_e shift in the tumor was in the range 0.5–1 ppm. Healthy tissue voxels 5 and 6 in (a) RG2 and (b) U87 tumor rats were slightly shifted in the downfield direction, suggesting the paramagnetic effects of TmDOTP⁵⁻ reach the extracellular space even with limited extravasation. Ventricular voxels 1 and 2 in (a) RG2 and (b) U87 tumor rats displayed a single unshifted Lorentzian peak before and a shifted Lorentzian peak after TmDOTP⁵⁻ injection. This is attributed to the dominant ²³Na signal contribution in the ventricles coming from cerebrospinal fluid (CSF), which contains free (i.e., unbound) aqueous Na⁺. The position of the shifted ventricle peak coincided with the Na⁺_e peak position in other regions of the brain. This agrees with expectation because CSF is in physical contact with the extracellular space with free exchange of aqueous Na⁺ between the two compartments. Similar spectroscopic patterns are observed throughout all voxels in vivo. See Figure S3 for several slices for each rat shown here. All spectra were magnitude-corrected and line-broadened by 10 Hz.