Introduction

Deuterium (²H) has an integer spin of one which gives rise to a quadrupolar magnetic moment that offers sensitivity to molecular orientation^1,2 and produces faster relaxation compared with ¹H. The low natural abundance (~0.015%) along with the low gyromagnetic ratio (6.54 MHz/T) of ²H, when compared with ¹H, reduces the available MR signal, but the reduced longitudinal relaxation time (T₁) of ²H allows greater signal averaging in a given acquisition time. This has allowed ²H-labelled substrates to be used as metabolic tracers, and has led to an increased interest in deuterium magnetic resonance measurements following injection or ingestion of ²H-labelled compounds.^3,4 Analysis of body fluid samples during heavy water loading is also used for studies of protein synthesis.⁵ Here, images and relaxation time maps are reported at 7T on healthy human subjects loaded with D₂O to ~1.5% enrichment, which was maintained over a 6-week period, for a parallel study of immune cell proteomics. This allowed the characterisation of ²H relaxation times in different brain tissues in vivo for the first time.

Method

Measurements were made on a 7T Philips Achieva scanner using a dual-tuned ²H/¹H head birdcage coil (Rapid Biomedical). Four healthy subjects were scanned regularly during the loading period. 3D sagittal multi-echo, gradient echo (MEGE) ²H images (6 x 6 x 10 mm³ voxels) were acquired with a range of repetition times (TR) to allow calculation of T₁ and T₂^* maps, using dual-fitting to the saturation recovery and echo-time (TE) variation of each voxel. A low resolution MEGE ¹H image (3 x 3 x 5mm³) was also acquired for comparison and later co-registration. In an additional scanning session, a 32-channel receiver coil (Nova Medical) was used to acquire a high-resolution (0.7 mm isotropic) ¹H MPRAGE for tissue segmentation and co-registration, along with a low resolution MEGE ¹H image to allow calculation of ¹H T₂^* maps. All images were transformed into standard MNI space for analysis. An affine matrix was obtained from a linear two-part registration with six degrees of freedom using FSL, which allowed the segmentation masks to be transformed into the low resolution ²H space. Mean relaxation times were calculated for cerebral spinal fluid (CSF), grey matter (GM) and white matter (WM) regions.

Four subjects (A-D) were scanned and the ²H relaxation time measurements were repeated twice on Subjects C and D. Six sets of ²H relaxation times and four sets of ¹H relaxation times are therefore reported in Table 1. Data from Subjects A and B was acquired with five echoes (TE₁=4.3 ms; ΔTE=8.4 ms) and TR=68, 136, 272, 544 ms, with the number of averages adjusted so the acquisition time (T_AQ) for each image was 487 s. Data from subjects C and D was acquired with six echoes (TE₁=4.3 ms; ΔTE=8.4 ms) and one additional TR value (TR=816 ms; T_AQ = 730 s). All ¹H MEGE data was acquired with 15 TEs (TE₁= 2.5 ms; ΔTE =2.34 ms), and TR = 41ms.

Results

Figure 1 shows example 3D sagittal images obtained by summing the ²H MEGE data for a single subject over TE and TR values. The resulting images, which predominantly show T₂*contrast, clearly depict the brain anatomy and have a similar appearance to T₂^*-weighted, ¹H images. The CSF in the ventricles appears hyperintense, while the white matter in the corpus callosum appears hypointense.

Figure 2 shows images of one slice acquired at the individual TE and TR values for one subject. The persistence of CSF signal at longer TEs and the reduction of contrast at short TE and TR is evident. These images also show that good quality data is obtained in individual acquisitions.

Figures 3 and 4 show sagittal and axial relaxation rate maps from two subjects, with the dominant feature in the ²H maps being the reduced R₁ and R₂^* relaxation rates in the ventricles. The elevated R₂* seen in deep grey matter structures in the ¹H maps is not evident in the ²H maps. Table 1 reports the average and standard deviation of the T₁ and T₂^* values measured in GM, WM and CSF in the four subjects by applying the binarized segmentation masks to the relaxation maps.

Discussion

Images with good signal-to-noise-ratio could be obtained in times of around eight minutes at 7T with 6x6x10 mm³ resolution in subjects loading with D₂O to ~1.5% concentration (~100 times natural abundance). Relaxation time measurements were consistent across the six measurements. Measured T₁ values for deuterium in HDO were significantly shorter than corresponding values for ¹H in H₂O,⁶ as expected, due to the quadrupolar ²H relaxation. ²H T₁ and T₂^* values were significantly higher in CSF than in GM/WM. ²H and ¹H, T₂^* values were similar, but the quadrupolar ²H relaxation does not show strong sensitivity to iron content in deep grey matter (Fig. 3 and 4).

Conclusions

²H T₁ and T₂* relaxation times from water in GM, WM and CSF have been measured at 7T in vivo from human subjects loading with D₂O to 100 times natural abundance. These relaxation times can be applied in research protocols using the natural abundance ²H signal from water for calibration.

Acknowledgements

The pilot study into immune cell responses in cancer is supported by Cancer Research UK [A29820]. The ²H imaging research is co-funded by the NIHR Nottingham BRC and CRF. DC’s Ph.D. studies are funded by the Precision Imaging Beacon at the University of Nottingham.