3054

Dynamic Metabolic Imaging of Co-Polarized [2-13C]Pyruvate and [1,4-13C2]Fumarate Using 3D-Spiral CSI with Alternate Spectral Band Excitation

Maninder Singh¹, Sonal Josan², and Dirk Mayer¹

¹Diagnostic Radiology and Nuclear Medicine, University of Maryland, Baltimore, MD, United States, ²Siemens Healthcare, Erlangen, Germany

Synopsis

Metabolic imaging of biologically-relevant hyperpolarized agents allows measurement of metabolic processes in real time in-vivo. We demonstrate dynamic metabolic imaging of a single bolus of co-polarized [2-¹³C]pyruvate and [1,4-¹³C₂]fumarate in control as well as in rats with liver necrosis. Chemical shift imaging (CSI) of such a mixture is challenging due to the large spectral dispersion of resulting resonances, which could lead to severe chemical shift displacement artifacts if acquired by conventional slice-selective excitation pulses. Here we obtain CSI information by a volumetric method using alternate 3D spectrally-selective excitation of sub-bands containing fewer resonances.

Introduction

Metabolic imaging of hyperpolarized ¹³C-labeled pyruvate (Pyr) and fumarate (Fum) has shown great potential in characterizing multiple diseases and assessing response to treatment, especially in cancer.¹ Performing these measurements with a single injection of co-polarized substrates would be advantageous particularly for the clinical translation of this technology. Recently, a study reported on combined metabolic imaging of [1-¹³C]Pyr and [1,4-¹³C₂]Fum in a rat model of necrosis.² However, the small chemical shift (CS) dispersion of both substrates and their metabolic products (~25 ppm) does not allow using signal-to-noise (SNR)-efficient differential flip angles for substrate and products³ at clinical field strengths, potentially preventing the measurement of their saturation kinetics that requires effective 90° excitation of the products.⁴ Therefore, the goal of this study was investigating the feasibility of simultaneous metabolic imaging of co-polarized [2-¹³C]Pyr and [1,4-¹³C₂]Fum with different excitation flip angles on a clinical MR scanner taking advantage of the much larger CS dispersion of [2-¹³C]Pyr and its products [2-¹³C]lactate (Lac) and [2-¹³C]alanine (Ala) (>160 ppm).

Methods

Our method is based on 3D spiral chemical shift imaging (spCSI) developed for metabolic imaging of [2-¹³C]Pyr.⁵ To avoid CS displacement artifacts due to the large CS dispersion, the sequence applied frequency-selective RF pulses to alternatingly excite two spectral sub-bands each one followed by a fast 3D-spCSI readout. Although implemented on a clinical MR scanner, that study used a high-performance gradient insert to achieve a high spectral width (SW) of ~500 Hz with only 2 spatial interleaves to sufficiently separate the multiple resonances. However, on our GE 3T 750w MR scanner with maximum gradient strength of 33 mT/m and slew rate of 120 mT/m/ms, achieving this SW would require about 20 spatial interleaves for the same spatial parameters, hence, leading to unacceptable scan times of 24 s per band. Therefore, we applied number of signal averages (NSA) analysis used in least-squares estimation imaging approaches^6,7 to investigate if smaller SW would be sufficient. For dynamic imaging, we applied the two RF pulses shown in Fig. 1. The 14 ms pulse for band-1 only excites Fum (2.5°) and both malate (Mal) resonances (10°). The 2.8 ms pulse for band-2 only excites resonances from Lac, Ala, and Pyr hydrate (Pyh), all with a 10° excitation flip angle. Pyh is measured instead of Pyr itself as it is in exchange with Pyr and not metabolically active. For the design of the RF pulses it is in important to sufficiently suppress excitation of Pyr in both bands and of Fum during excitation of band-2. The performance of the pulses was tested on a sample of co-polarized Pyr and Fum using a simple pulse-and-acquire MRS sequence that alternatingly used the two pulses (TR = 3 s) for 90 s. This acquisition was followed by the same sequence but using a 32-µs hard pulse (5.625°) for 4min. Co-polarization of the two substrates to about 35% was performed similarly as described in Ref. 2 using a GE SPINlab polarizer operating at 5T and 0.9K. In vivo experiments using dynamic 3D-spCSI were done in control as well as in rats with liver necrosis induced by CCl₄ administration. A single-loop ¹³C surface coil placed on top of the liver was used for both signal excitation and reception.

Results and Discussion

The NSA plots shown in Fig. 2 indicate an echo shift of 3.87 ms (corresponding to a SW of 258Hz) is sufficient to separate metabolites excited in the two alternating bands. The feasibility of our strategy is demonstrated by the time-averaged CSI data from a rat with CCl₄-induced necrosis shown in Fig. 3. The spectra from an ROI in the liver show how well the peaks are separated in the spectra from different bands. In band-1, the SW is sufficient to detect both Fum and Mal peaks without aliasing. However, for band 2, relative to Lac, the Ala peaks were aliased -2 times and Pyh 3 times. This required separate reconstructions for the aliased signals to properly demodulate the phase.⁸ The corresponding metabolic maps are shown in Fig. 4. From the MRS measurement of the Pyr/Fum solution we calculated the residual excitation flip angle of the substrate resonances as: for Pyr 0.004° in band-1 and 0.015° in band-2, and for Fum 0.006° in band-2. The spectra from a liver ROI in a healthy control animal after injections of co-polarized Pyr/Fum and Pyr only also indicate sufficient suppression of Pyr (Fig. 5a). Metabolite time courses from liver ROIs demonstrate feasibility of dynamic imaging with a time resolution of 4.5 s per band (Fig. 5b).

Acknowledgements

This work was supported by NIH grants R01 DK106395, R21 CA213020, R21 CA202694, and R21 NS096575.

References

1. Kurhanewicz J, Vigneron DB, Brindle K, et al. Analysis of cancer metabolism by imaging hyperpolarized nuclei: prospects for translation to clinical research. Neoplasia. 2011;13:81–97.

2. Eldirdiri A, Clemmensen A, Bowen S, Kjaer A, Ardenkjaer-Larsen JH. Simultaneous imaging of hyperpolarized [1,4-¹³C₂] fumarate, [1-¹³C]pyruvate and ¹⁸F–FDG in a rat model of necrosis in a clinical PET/MR scanner. NMR in Biomedicine. 2017;e3803.

3. Larson PEZ, Kerr AB, Chen AP, Lustig MS, Zierhut ML, Hu S, et al. (2008). Multiband excitation pulses for hyperpolarized ¹³C dynamic chemical-shift imaging. Journal of Magnetic Resonance. 2008;194(1), 121–127.

4. Xu T, Mayer D, Gu M, Yen Y-F, Josan S, Tropp J, et al. Quantification of in vivo metabolic kinetics of hyperpolarized pyruvate in rat kidneys using dynamic ¹³C MRSI. NMR in Biomedicine. 2011;24(8), 997–1005.

5. Josan S, Hurd R, Park JM, et al. Dynamic metabolic imaging of hyperpolarized [2-¹³C]pyruvate using spiral CSI with alternating spectral band excitation. Magnetic Resonance in Medicine. 2014;71(6):2051-2058.

6. Reeder SB, Wen Z, Yu H, et al. Multicoil Dixon chemical species separation with an iterative least-squares estimation method. Magnetic Resonance in Medicine 2004;51:35–45.

7. Wiesinger, F., Weidl, E., Menzel, M. I., Janich, M. A., Khegai, O., Glaser, S. J., Haase, A., Schwaiger, M. and Schulte, R. F. IDEAL spiral CSI for dynamic metabolic MR imaging of hyperpolarized [1-¹³C]pyruvate. Magnetic Resonance in Medicine. 2012;68:8–16.

8. Mayer, D, Levin, YS, Hurd, RE, Glover, GH, & Spielman, DM. Fast metabolic imaging of systems with sparse spectra: application for hyperpolarized ¹³C imaging. Magnetic Resonance in Medicine. 2006;56(4), 932–937.

Figures

Log-scale frequency profile of the spectrally-selective multiband hamming windowed-sinc RF band-1 (A) and band-2 (B) pulses with the passband in band-1 placed between Fum and Mal peaks and in band-2 centered between the two Lac peaks. The band-1 pulse applies a differential flip angle at Fum and Mal peaks in the ratio 1:4 whereas the band-2 pulse applies the same flip angle for Pyh, Lac, and Ala peaks. In both bands, the signal from the [2-¹³C]Pyr is suppressed.

Number of signal averages (NSA) and its mean is plotted as a function of echo shift for two bands. For band-1 (A and B), NSA is plotted for three resonances (Fum and Mal doublet) and for band-2 (C and D), NSA is plotted for five resonances (Pyh and the doublets of Lac and Ala) at 3.0T for 7 equally spaced echoes. Based on these calculations, an optimum echo spacing of 3.87 ms (~258 Hz) is found to separate all species into different metabolic images.

In vivo 3D-spCSI data (FOV=8×8×6cm³, 16×16×12 matrix, 6s per band from a rat with CCl₄-induced liver necrosis. (A) Time-averaged liver spectrum from band-1. Detection of the Mal doublet indicates necrosis of the liver. (B) Corresponding spectrum from band-2. All resonances from Pyh, Lac, and Ala were well separated. Aliased signals required separate reconstruction to properly demodulate the phase caused by off-resonance.

Time-averaged metabolic maps from the data in Fig. 3 from a slice through the liver superimposed onto corresponding ¹H MRI image. Whereas the metabolic products are only detected in the liver, the substrate signals were also high in the vasculature.

In vivo 3D-spCSI data (FOV=6×6×5cm³, 12×12×10 matrix, 4.5s per band from a healthy rat. (A) Time-averaged liver spectra after injection of co-polarized Pyr/Fum (blue) and Pyr only (red, multiplied by 30). No Pyr signal was detected after the Pyr-only injection. (B) Metabolite time courses after injection of co-polarized Pyr/Fum from a liver ROI. The Fum signal is scaled down by a factor 20.

Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)

3054