Background

Hyperpolarized [1-¹³C]pyruvate has been developed as a useful imaging agent for studying in vivo metabolism, and it can be enzymatically converted to [1-¹³C]lactate, [¹³C]bicarbonate and [1-¹³C]alanine. In most applications, lactate and bicarbonate are the products of interest as they are directly related to glycolysis and oxidative metabolism. However, it is still challenging to capture accurate metabolic maps of lactate and bicarbonate primarily due to the transience of hyperpolarized ¹³C signals and dynamic nature of the in vivo metabolism. Various imaging strategies have been developed, ranging from spectroscopic imaging sequences^1-3 to imaging sequences that exploit spectrally-selective (spsp) RF excitations or IDEAL reconstruction.⁴ Spectroscopic imaging maintains spectral information but requires longer acquisition time, more RF utilization and stronger gradient performance. The image quality is extremely susceptible to B₀ field inhomogeneity and often suffers signal contamination from imperfect suppression of large peak (e.g., pyruvate). Moreover, imaging with spsp-RF pulses requires alternating acquisition of lactate and bicarbonate, resulting in asynchronous imaging. In this study, we propose a method to simultaneously excite [¹³C]bicarbonate and [1-¹³C]lactate without perturbing other metabolites. The SNR-advantage of data acquisition with a pyruvate-nulled RF was previously demonstrated.⁵ Lactate and bicarbonate images are spatially separable using spatially-interleaved fly-back echo-planar imaging (EPI) readout with a controlled echo-spacing.

Methods

A spsp-RF pulse that selectively excites lactate and bicarbonate in a hyperpolarized [1-¹³C] pyruvate spectrum was designed using the Spectral-Spatial RF Pulse Design MATLAB Package¹, which is shown in Fig.1A. Four-shot spatially-interleaved fly-back EPI readout was used for the image acquisition (Fig.1B). To achieve the spatial separation of bicarbonate and lactate images, echo-spacing (t_esp) was calculated using the equation: t_esp=2(Φ+N_alias)/Δf, where Φ is the ratio of shifted distance to FOV, N_alias is an integer indicating the periodic shift of the image, and frequency difference Δf=f_lac-f_bic.⁶ In this study, Φ=0.25, N_alias=0 and FOV was set to twice over the size of subject. With the receive frequency as (f_lac+f_bic)/2, the metabolite maps were generated side by side along phase-encoded direction. The RF pulses and the EPI readout were implanted in the MNS Research Pack (GE Healthcare). All the data were acquired using a GE 3T 750W wide-bore scanner. For phantom study, a Gd-doped spherical [¹³C]bicarbonate phantom (0.4M, diameter=18cm) and a Gd-doped cylindrical lactate phantom (6M, diameter=1cm) were used with a ¹H/¹³C 8-channel birdcage human head coil. The designed spsp-RF pulse and the fly-back EPI readout were used for ¹³C imaging (TE/TR=8us/5000ms, thickness=20mm, matrix size=118x128, FOV=36x36cm²). A healthy Wistar rat with a ¹³C/¹H dual-tuned birdcage RF coil was used for in vivo study. A 35-μL sample of 14-M [1-¹³C]pyruvic acid mixed with OX063 trityl (15mM) was prepared and polarized using a SPINlab clinical DNP polarizer (GE Healthcare). The hyperpolarized samples were dissolved, mixed with pH-neutralization media and immediately injected over 10-12s intravenously (~70mM pyruvate, ~7.5 of pH). Dynamic hyperpolarized ¹³C images were acquired from the rat liver 6s after the start of [1-¹³C]pyruvate injection with modified imaging parameters (matrix size=64x46, FOV=16x16cm², 16 scans with time resolution=3s). In order to compensate the signal difference between the interleaves, variable flip angles were applied to each interleave (flip angle=30°, 35°, 45°, 90°).

Results and Discussion

Fig.2A shows the spectral profile of the designed RF pulse. The simulated spectral-spatial profile (Fig.2B) was confirmed by the measured RF profile (Fig.2C). The spatial separation of bicarbonate and lactate images was validated by the phantom study. The two phantoms were positioned as shown in the ¹H image (Fig.3A). The ¹³C image in Fig.3B demonstrates the successful separation of bicarbonate and lactate images. Fig.4A shows the imaged rat liver slice and the time-average metabolite maps of hyperpolarized [1-¹³C]lactate, [1-¹³C]pyruvate (or lipid) and [¹³C]bicarbonate. Each ¹³C image was overlaid on the corresponding ¹H image. It was noted that the pyruvate map shows nonnegligible ¹³C signal, which is likely because of the spatial variation of B₀ field. Although the pyruvate is supposed to be excited by only 0.19 %, small frequency shift in pyruvate peak can result in relatively high pyruvate signal due to the large pyruvate peak. Besides, the [¹³C]bicarbonate signal is relatively low, which may result from the starving state of the animal. Fig.4B shows the dynamic changes of hyperpolarized [1-¹³C]lactate signal from 30 s to 45 s after the start of [1-¹³C]pyruvate injection. Images were partly contaminated by ghosting artifact along phase-encoded direction, which is mainly due to the signal difference between multiple interleaves arising from the incomplete suppression of flowed pyruvate.

Conclusion

We developed an imaging sequence that allows simultaneous assessment of [1-¹³C]lactate and [¹³C]bicarbonate by combining a lactate-bicarbonate-selective RF pulse and a multi-shot fly-back EPI with and appropriate echo-spacing. More animal study will be performed to further optimize the technique for human brain/liver imaging.

Acknowledgements

Funding: The Mobility Foundation; The Texas Institute of Brain Injury and Repair; National Institutes of Health of the United States (P41 EB015908, S10 OD018468).