In Vivo Monitoring of Rat Spinal Cord Metabolism Using Hyperpolarized Carbon-13 MR Spectroscopic Imaging

Materials and Methods

A total of 6 healthy male Sprague-Dawley rats (median weight, 320 g) were included in this study. Animal studies were approved by the Institutional Animal Care and Use Committee.

Animals were scanned on a 3T clinical MR imaging system (Discovery MR750; GE Healthcare, Milwaukee, Wisconsin) with 40 mT/m, 150 mT/m/ms gradients, a multinuclear spectroscopy hardware package, and a custom-designed, dual-tuned volume radiofrequency coil with a quadrature ¹³C channel and linear ¹H channel with a length of 9 cm. During each imaging session, rats were placed prone on a heated pad positioned inside the radiofrequency coil and scanner. Cervical lordosis was straightened with padding under the ventral neck to minimize partial volume effects with nonspinal tissue when imaging in the axial plane (Fig 1A). The spine at C4–C5 was placed at the center of the radiofrequency coil so that both the spine at C4–C5 position and the supratentorial brain were located within the 7-cm effective region of the coil. Anesthesia was maintained with a constant delivery of isoflurane (approximately 1.5%). Before each ¹³C imaging session, high-resolution T2WI was obtained in the axial plane by using an FSE sequence (TE, 60 ms; TR, 4000 ms; FOV, 8 cm; matrix size, 256 × 256; 2-mm section thickness; and NEX, 8). For each ¹³C experiment, a mixture of 35 μL of [1-¹³C]-pyruvate, 15 mmol/L OX063 trityl radical (GE Healthcare), and 1.5 mmol/L Gd-DOTA was polarized by using a HyperSense polarizer (Oxford Instruments, Oxfordshire, United Kingdom) at 3.35T and 1.4K by irradiation with 94.1 GHz microwaves by using methods described previously.⁴ After 60 minutes of microwave irradiation, the mixture was rapidly dissolved in a saline solution with 5.96 g/L Tris (40 mmol/L), 4.00 g/L NaOH (100 mmol/L), and 0.1 mg/L Na² ethylenediaminetetraacetic acid. The final solution had a concentration of 100 mmol/L pyruvate and pH ∼7.5. A sample from the dissolved pyruvate solution with volume of 2.8 mL was injected into the tail vein of the rat over a 10-second duration.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

An example of hyperpolarized ¹³C data from a naive rat. A, Horizontal dashed lines encompassing the neck delimit a 30-mm axial slab used for acquiring dynamic data in B. B, A stack plot of dynamic ¹³C data from a 30-mm slab. C, Axial T2WI of the neck centered at C4 with ¹³C spectral grid overlay for 2D-localized ¹³C dynamic data. D, Dynamic pyruvate and lactate signal change over time as measured from a 4.8 × 4.8 × 15 mm voxel primarily limited to the cervical spinal cord (white box, C). E, Axial T2WI with ¹³C spectral grid overlay for 3D ¹³C-MRS imaging. F, The corresponding ¹³C spectra with high spatial resolution (2 × 2 mm in-plane voxel size) from a 5.4-mm axial section centered at the C4 vertebral level. G, Zoomed-in ¹³C spectra targeting the spinal cord show high pyruvate and relatively small lactate signal in the spinal cord (highlighted voxels).

To determine an optimal temporal imaging window for acquiring 3D MRSI data, initial dynamic ¹³C spectroscopic data were acquired from 2 rats. In 1 animal, section-localized data from a 30-mm axial slab encompassing the entire neck were acquired (Fig 1A, -B; TE, 35 ms; TR, 3000 ms; flip angle, 10°; 3-second temporal resolution; and 32 total time points). To more specifically gauge temporal evolution of signal within spinal tissue of the neck, 2D-localized multivoxel data from a 15-mm axial section centered at the C4 vertebral level were acquired in a single animal (TE, 6.1 ms; TR, 130 ms; flip angle, 5°; 3-second temporal resolution; 30 total time points; 10 phase encodes in the right-left direction; and a symmetric echo-planar readout in the anteroposterior direction providing 4.8 × 4.8 mm in-plane resolution).⁸ Section-localized and multivoxel 2D-localized data were acquired from each animal simultaneous with the injection of pyruvate solution.

Compressed-sensing ¹³C 3D-MRSI data then were separately acquired from a total of 6 rats (including 2 rats described above) by using a double spin-echo sequence (TE, 140 ms; TR, 215 ms) with centric k-space encoding, a variable flip angle scheme, and flyback echo-planar readout on the z-axis⁵ at 18 seconds from the start of the injection. Eighty-six phase encodes were collected from a 20 × 16 matrix in 18 seconds, resulting in 2 × 2 mm in-plane resolution with 16 5.4-mm sections.

The methods for processing ¹³C data have been described previously.^5⇓⇓–8 The SNR of lactate, pyruvate, and ratio of lactate over pyruvate were calculated from the ¹³C 3D-MRSI data by using the magnitude spectra. To estimate the SNR, the peak height for lactate and pyruvate were scaled by the standard deviation of the noise estimated from the spectrum that contained no metabolite resonances. The SNR of lactate, pyruvate, and ratio of lactate over pyruvate were estimated from voxels in the spine and compared with the corresponding values from voxels in supratentorial brain by using a 2-tailed paired t test.