Recently, spin–lock received new interest in MRI: it was demonstrated that on–resonant chemical–exchange–sensitive spin–lock (CESL) and related techniques allow to observe the uptake of administered D–glucose in vivo^1-5 and thus "could be a highly sensitive and quantifiable tool for glucose transport and metabolism studies".¹ The exchange–weighting increases with the magnetic field strength (B₀)^6,7. However, conventional spin–lock produces artifacts owing to B₀– and B₁–field inhomogeneities, which are a common problem especially at high-field whole-body MR scanners. Our aim was the development of an adiabatic water–T_1ρ mapping sequence applicable to studies of glucose metabolism in tumor patients with the following features: insensitivity to field inhomogeneities, SAR compatibility and appropriate temporal resolution.Two on–resonant spin–lock sequences with centric–reordered 2D–GRE readout were implemented on a 7–T whole–body scanner (MAGNETOM; Siemens, Erlangen, Germany): (i) a conventional spin–lock sequence (SL) using a 90° Gaussian–shaped hard pulse (t_p = 1 ms) to tilt the magnetization and (ii) an adiabatic spin–lock sequence (adiaSL), where the Gaussian–shaped pulses are replaced by adiabatic half–passage pulses. Using Bloch–McConnell simulations the sequence parameters were optimized considering technical and SAR limitations and verified in phantom and in vivo experiments. Quantification of T_1ρ was done for each pixel by fitting a monoexponential function to data obtained for different locking times (TSL). A water phantom (∅ ≈ 10 cm, 4g/L NiSO₄ and 5g/L NaCl) and a set of four 30–ml PBS phantoms (pH = 7.00) with glucose concentrations between 5 and 20 mM were used in the in vitro experiments. An in vivo measurement was performed in the brain of a gliosarcoma patient.

Figure 1a shows the measured R_1ρ–decay curves obtained with both sequences for two ROIs (marked in the relative B₁–map, Fig. 1b) of the water phantom. Unlike the two decay curves for the adiaSL sequence (solid lines), the curves for the SL sequence (dashed lines) differ substantially and show oscillating signals for short TSL, which can be explained by unequal orientations of the locking field and the initial magnetization. Figures 1c and 1d show the maps of the fitted magnetization at TSL = 0 ms (S₀) and Fig. 1e and 1f the fitted T_1ρ–maps for the adiaSL and SL sequence, respectively; the maps of both techniques agree in the center. The maps obtained by adiaSL are uniform across the whole phantom, while the values from the SL sequence vary strongly within the homogeneous phantom. Especially the S₀–map shows a strong correlation to the B₁–map. Figure 2a shows the R_1ρ–contrast obtained with the adiabatic spin-lock sequence for the four glucose–containing phantoms. The R_1ρ–values are uniform within the different phantoms, which are well distinguishable among each other. Figure 2b displays R_1ρ as a function of the glucose concentration showing a linear correlation as expected from theory¹. Figure 3 displays the results of the in vivo measurements in the brain of a gliosarcoma patient. The T₂-weighted image is shown in Fig. 3a and the corresponding R_1ρ–maps obtained with the conv. and adiabatic spin-lock sequence are shown in Fig. 3b and 3c, respectively. Several artifacts owing to B₁-imhomogeneities are visible in the R_1ρ-map obtained with the conv. SL sequence (Fig. 3b). These artifacts are clearly reduced in the R_1ρ-map obtained by the adiabatic SL sequence (Fig. 3c).

The implemented adiabatic spin–lock sequence outperforms the conventional spin-lock sequence with respect to its insensitivity to B₁–field inhomogeneities. The optimized adiabatic sequence obeys SAR limits of clinical high–field whole–body scanners; its sensitivity to glucose in the millimolar range as well as its applicability to in vivo studies is proven. So far the approach has been used to examine three brain-tumor patients in a clinical glucose injection trial. However, due to the necessity of motion correction the evaluation is still an ongoing process.