Introduction

Efficient pharmacological treatment of neurological and psychiatric disorders is often hindered by poor drug penetration of the blood-brain barrier (BBB) after systemic application. Accordingly, current research explores delivery routes that can bypass the BBB. To comprehend the efficacy and temporal dynamics of drug action between different delivery routes, it is crucial to understand the dynamics of drug distribution within the brain. Magnetic resonance imaging (MRI) and spectroscopy (MRS) are powerful, non-invasive diagnostic techniques. MRS allows non-invasive insights into biochemistry and biodistribution of small molecules. It is severely hampered, however, by some intrinsic obstacles: the signals of small metabolites at ≈ 1-10mM are much smaller and often “buried” under the signals of fat and water ([¹H] = 111M). Here, we are developing an alternative MRS approach: deuterium metabolic imaging (DMI or ²H-MRS).¹ DMI enables speedy signal accumulation due to the fast relaxation of deuterium, also, it is less susceptible to B₀-inhomogeneities compared to ¹H, and is practically free of background signal (high contrast). The natural abundance (or mole fraction) of ²H, however, is no more than 0.011%, such that the concentration of ²H in pure water amounts to ≈ 12mM only. Water is the only detectable ²H-background signal and its concentration (and hence signal) is comparable with the concentration of administrated medication and now dedicated signal suppression is required. To explore the feasibility of ²H-MRS approach for the tracking of drug distributions over time, we here used the deuterium labelled anti-epileptic drug ethosuximide (ETX).

Methods

NMR (Figure 1,2). High-resolution (non-localized) ¹H and ²H NMR spectra of ETX and deuterated ETX (ETX-D4,4’) in aqueous solution ([D₂O]:[H₂O]=1:9) were acquired at 400MHz (9.4T, WB Bruker Avance NEO NMR, 5mm tubes, 10mm BBO probe). T1 relaxation times were measured using the inversion recovery method.
MRI (Figure 4,5). Turbo RARE sequence with FOV 20mm x 20mm x 1mm and resolution 256x256 and localized ²H PRESS-MRS were measured for the phantom sample on the same MR-system using a 25mm dual tune ¹H/²H resonator (Bruker). Rodent in vivo experiments are planned for Q1-2/2021.
Synthesize (Figure 1). Pure ETX-D4,4’ with 95% deuteration fraction was obtained in a rapid microwave-assisted deuterium exchange of ETX in methanol-D₄ with potassium tert-butoxide. We achieved over 90% deuteration in the first round and 95% after repetition with fresh methanol-D₄. After an acidic workup, tert-butanol and methanol was evacuated.

Results and discussion

²H relaxation. Although deuterium is a quadrupole nucleus, its relaxation times were found to be not much shorter than the corresponding relaxation times of protons (Figure 2): T1 of ETX was about 0.25s for D4,4’ and 2.1 s for H4,4’ (in the same conditions). Note that T2 relaxation time was also not much different since the linewidth was < 2Hz, allowing to estimate that their T2 > 0.15s. Thus, the relaxation times of deuterium are well suited for the rapid accumulation of the signal. Furthermore, at the same time, relaxation is slow enough for single voxel spectroscopy such as PRESS or STEAM² or chemical shift imaging (CSI).
Optimal TR for ²H-MRS. The MR signal as a function of TR and TE in spin-echo experiments can be estimated with (1-exp(-TR/T1)) exp(-TE/T2). From here it follows that TE should be as small as possible while TR on the contrary long. However, the experimental time is limited (i) and SNR increases only as a square root of a number of scans, NS^1/2, (ii). Thus, there is an optimal TR that provides the highest SNR for the given experimental time. This optimal TR is ~1.25*T1 (Figure 3c). We measured ETX-D4,4’ with ²H-PRESS as a function of TE and TR (Figure 4). The results revealed the optimal parameters for TR ~ 300ms (Figure 4c), that is close to the theoretically predicted values.
SNR of ¹H- and ²H-PRESS. With the optimal parameters for ¹H-PRESS (TE=16.5ms, TR=2000ms, NS=1200) and ²H-PRESS (TE=14.2ms, TR=300ms, NS=8000) and the same voxel size of 4 mm x 4 mm x 4 mm, and total acquisition time of 40 minutes localized spectra were measured (Figure 5). Obtained SNRs for D4,4’, H7₂, H9₃ and H8₃ of 45mM ETX-D4,4’ were 50, 73, 330 and 105 respectively. Among measured protons, H9₃ gives the sharpest singlet peak. D4,4’ is appearing as two partially overlapping signals that correspond to two individual ²H. From this, one can conclude that SNR of ¹H is 110 and SNR of ²H is ~50 per spin. This gives a hint that alternative deuteration would be more preferable: 3 times higher SNR of ~150 is expected for ETX-D9₃.

Conclusion

We have done all preparation steps for the future in vivo DMI of ETX-D4,4’. Although, the magnetogyric ratio of ²H is not in favour of high MR-signal, the rapid accumulation, sharp lines even in an inhomogeneous magnetic field and low the background signal is allowing a successful DMI measurement. The measurement time (40 minutes) and special resolution (voxels 4mm x 4mm x 4mm) are reasonable for ²H-MRS of samples with a 9mM the concentration of ETX-D4,4’ with SNR of 9. Experimentally we achieved the increase in the number of scans by a factor of 7 (TR for ¹H is 2000ms and for ²H is 300 ms) that partially compensates the low signal of ²H.

Acknowledgements

We acknowledge funding from BMBF (01ZX1915C), DFG (HO-4602/2-2, HO-4602/3, GRK2154-2019, EXC PMI1267, FOR5042. TRR287), Kiel University and the Faculty of Medicine for supporting the MOIN CC.