Implementation of Two-pulse Phase-modulated (TPPM) ¹H Decoupling in a Clinical MR Scanner for the Detection of the C1-glycogen Peak in ¹³C MRS

Abstract

Two-pulse phase-modulated (TPPM) ¹H-decoupling pulse sequence repeats a pair of 180^o RF pulses while changing the signs of the RF phase modulation angle and has been widely used for the ¹³C NMR of organic solids. TPPM was introduced into the ¹³C MRS pulse sequence on a clinical 3T MR scanner, and the ¹H-decoupling performance was compared with conventional ¹H-decoupling schemes using aqueous solutions containing glucose and oyster glycogen. The ¹³C C1-glucose peaks were ¹H-decoupled using TPPM with B₂ = 500 Hz, and the optimal RF phase modulation angle was up to 30^o. Cycling sidebands were not observed when TPPM was used but were observed when WALTZ-16 was used. The ¹³C C1-glycogen peak was ¹H-decoupled even with reducing TPPM duration to 8 ms, which reduced simulated specific absorption rate (SAR) to 39%. In conclusion, the TPPM ¹H decoupling is applicable to clinical MR scanners, and the low-SAR sequence may be more valuable at 7T.

Introduction

Glycogen has attracted a great deal of interest with relation to the energy metabolism for physical fatigue in skeletal muscle. The gold standard of measuring skeletal muscle glycogen levels is needle biopsy.¹ However, as this method is invasive, it is not suitable for dynamic studies, such as the tracking of changes of glycogen levels along with skeletal muscle activity. For a non-invasive assessment, the ¹³C MRS method has been developed via detecting the C1-glycogen peak at 100.5 ppm that is clearly distinguishable from other signals in the in vivo spectrum,² instead of the higher cluster peaks from glycogen C2-C5 in the 72–78 ppm range overlapped with the much higher C2-glycerol peak at 69.5 ppm.³ Despite the challenges in technical development for MR acquisition with a low-natural abundance ratio and low-gyromagnetic ratio, the ¹³C MRS has provided valuable insights into glycogen content and kinetics since it can acquire a series of spectra repeatedly from a single subject.⁴ For improving SNR and/or reducing scan time, high magnetic fields are advantageous,^3,5–6 and the ¹³C peaks coupled with adjacent ¹H spins are detected as singlets using ¹H decoupling.^2–6

The ¹H decoupling is an essential technique for high-resolution ¹³C NMR in vitro and is achieved with RF irradiation at ¹H resonance frequency during ¹³C FID acquisition. The technique has been developed to cover a wider range of ¹H chemical shifts at high magnetic fields and is called broadband decoupling. An important advance in its development was the invention of the composite-pulse sequences, WALTZ.⁷ The WALTZ ¹H decoupling is also often used to detect the C1-glycogen peak as a singlet from human skeletal muscle, but this still requires 5–15 mins at 3T.⁴ Although the doublet peak detected without ¹H decoupling provides the heteronuclear J-coupling constant, which is another evidence for identifying the ¹³C C1-glycogen peaks, the peak heights are halved. Therefore, ¹H decoupling is an essential technique given the need for reliable signal detection even when glycogen levels in muscles are low. The challenges of using WALTZ ¹H decoupling for human ¹³C MRS are the limited decoupling field strength (B₂) due to RF system specifications optimized for RF coils covering large sample volume and specific absorption rate (SAR) limitations, as well as the inhomogeneity of B₂ when using surface coils. Although the WALTZ-16 decoupling scheme requires large RF flip angles of over 180^o, de Graaf⁸ demonstrated the feasibility of the WALTZ-16 for human calf muscle ¹³C MRS at 4T with B₂ = 500 Hz generated using a half-volume ¹H-decoupling coil. Since the duration of a rectangle RF pulse for a 180^o flip angle with the B₂ is 1 ms, the duration for a single WALTZ-16 composite pulse is as long as 48 ms, which is too long to detect the ¹³C C1-glycogen peak with short-T₂* in vivo (~4 ms).³ The RF safety of the ¹H-decoupled ¹³C MRS for human subjects is often ensured by reducing the duty cycle of the ¹H-decoupling period during ¹³C FID acquisition. However, if the shortened ¹H-decoupling period is shorter than the duration of a single composite pulse, the decoupling compensation cycle is no longer effective.

Glycogen is a globular macromolecule with a diameter of 20–30 nm and a molecular weight of 10⁶–10⁷, and therefore has some physical properties of a solid. In ¹³C NMR of organic solids, which requires much higher ¹H-decoupling power, the two-pulse phase-modulated (TPPM) ¹H decoupling has been proposed and developed to increase the sensitivity of solid-state NMR with decoupling the heteronuclear scalar coupling and reducing cycling sidebands as well as cancelling the effects from strong ¹H–¹H dipolar interactions in a solid-state spin environment.⁹ Currently, the TPPM has been utilized as a more efficient method which achieves a higher SNR with a lower ¹H-decoupling power than the continuous wave (CW) decoupling.^9–10 The TPPM pulse sequence is one of the simplest composite pulses: a pair of 180° rectangular RF pulses while changing the positive and negative signs of the RF phase modulation angle. Even when used in clinical scanners, the TPPM ¹H-decoupling period can be reduced on a pulse-pair basis, since the pulse pairs are completed within a few milliseconds, comparable to the T₂* of the ¹³C C1-glycogen peak in vivo. Therefore, it would be valuable if the TPPM ¹H decoupling could be applied to detect the ¹³C C1-glycogen peak from human skeletal muscle. In this study, the TPPM ¹H-decoupling scheme was introduced into the ¹³C MRS pulse sequence for a clinical scanner. To begin with, a glucose phantom was used to optimize the RF phase modulation angle of the TPPM. Then, the line shape of ¹H-decoupled ¹³C peaks and ¹H-decoupling bandwidth were investigated to compare the performance of the TPPM with CW and WALTZ-16. Finally, an oyster glycogen phantom was used as a human surrogate to investigate SAR and the height of ¹H-decoupled ¹³C peak to compare the TPPM with WALTZ-16.

Materials and Methods

All MR experiments were carried out using a 3T clinical scanner (Skyra; Siemens Healthineers, Erlangen, Germany) equipped with broadband transmitters and receivers. A multi-channel RF coil was built for human calf muscle ¹³C MRS measurement with ¹H imaging and decoupling capability based on Ref. 6. The linear-¹³C/quadrature-¹HRF coils (¹³C: a 75-mm loop, ¹H: a pair of 95-mm loops) were formed on the surface of semi-cylindrical acrylic resin tubes with radii of 71 mm and 77 mm, respectively, shown in Supplementary Fig. 1. The coils were connected to the MR scanner via an interface box (Takashima Seisakusho, Tokyo, Japan) containing low-pass/high-pass filters, preamplifiers, and transmit/receive switches. A band-pass filter for ¹³C resonance frequency (FSY microwave, Columbia, MD, USA) was inserted between the ¹³C loop coil and the interface box. The materials of the coil housing, such as the acrylic resin, rubber, and wood, have been confirmed to generate no background ¹³C signals around the chemical shift of the C1-glycogen peak. Two aqueous solution phantoms containing [1-¹³C]glucose (ISOTEC, Miamisburg, OH, USA) and oyster glycogen (Nacalai Tesque, Kyoto, Japan) were prepared and were scanned at room temperature. The molar concentration of the C1-carbons in the glycogen phantom was ~150 mM.

Pulse sequences with TPPM and WALTZ-16 decoupling schemes were prepared as research prototypes based on the product pulse sequence for non-localized ¹³C MRS with ¹H-decouping capability. The same ¹H-decoupling amplitude of B₂ = 500 Hz (i.e., a 1-ms pulse duration for each 180° pulse element) was used in those ¹H-decoupling schemes. For the glucose measurement, the RF offset for ¹H decoupling was set at 4.93 ppm, which is at the midpoint between the H1-α- and H1-β-glucose peaks. The decoupling amplitude was kept constant during ¹³C FID acquisition. To optimize the TPPM, glucose ¹³C spectra were acquired using different RF phase modulation angles from 10° to 90° in 10° steps with 4 averages. To observe the ¹H-decoupling bandwidth of those decoupling schemes, glucose ¹³C spectra were acquired using different ¹H-decoupling RF offsets within ±1 kHz in 100-Hz steps. Other identical ¹³C MRS parameters were TR = 15s, the flip angle for ¹³C excitation = 90°, FID acquisition time = 85 ms and data points = 256, i.e., acquisition bandwidth = 3 kHz. The spectral acquisitions were repeated 3 times. For the glycogen measurement using the TPPM or the WALTZ-16, the duration of ¹³C FID acquisition was divided into 2 ¹H-decoupling phases as shown in Fig. 1a. The first phase (Phase I) consists of the cycles of the TPPM or WALTZ-16 scheme, i.e., the duration is an integer multiple of 1 composite cycle, with B₂ = 500 Hz. The aims for having the second phase (Phase II) of a ramping down of the ¹H-decoupling amplitude are to reduce SAR and to avoid detecting noise spikes when the decoupling power is turned off momentarily. Glycogen ¹³C spectra were acquired using different Phase I durations per the duration of ¹³C FID acquisition (85 ms) from 10% to 100% in 10% steps using the TPPM and those of 50% and 100% (= 1 and 2 composite pulse cycles, respectively) using the WALTZ-16. The RF offset for ¹³C transmit and receive was shifted 100 Hz from the peak frequency of C1-glycogen to distinguish the signal from artifacts that may be observed at the center of the acquisition bandwidth in a low sensitivity spectrum. The RF offset for ¹H decoupling was set at 5.4 ppm on the peak of H1-glycogen with α-1,4-glycosidic linkages.¹¹ Other identical ¹³C MRS parameters for the glycogen spectra were TR = 200 ms, the flip angle for ¹³C excitation = 90°, which had been optimized using the glycogen phantom, and 256 averages. RF irradiation at ¹H resonance frequency from the end of the ¹³C FID acquisition to the next ¹³C RF excitation, so-called nuclear Overhauser effect (NOE) pulses, was not applied in this study. SARs were simulated on the scanner for an adult male with a height of 170 cm and a weight of 70 kg and were recorded at the end of the measurements.

Fig. 1.

(a) TPPM ¹H-decoupling scheme for the ¹³C MRS pulse sequence. For the glycogen measurements, the decoupling power was linearly decreased from the middle to the end of ¹³C FID acquisition (Phase II). For the glucose phantom, the decoupling power was kept constant during ¹³C FID acquisition, i.e., the Phase II was not applied. (b) The C1-glucose peaks in ¹³C spectra from the glucose phantom as a function of the RF phase modulation angle for TPPM ¹H decoupling. The spectra are shown as the phase-sensitive display mode and the display width is 1.5 kHz (= 48.4 ppm). (c) The peak height as a function of the RF phase modulation angle in 10° increments. The peak heights were measured in the absolute-value display mode and averaged among 3 measurements. The corresponding standard deviations are smaller than the size of the points on the graph. TPPM, two-pulse phase-modulated.

For spectral analysis of the glucose spectra, the maximum peak height in the spectrum was measured in the absolute-value display mode on the operational software for the scanner, and the average of 3 measurements was calculated. The decoupling bandwidth was measured as the bandwidth where the average height of the ¹H-decoupled signal exceeded 90% of the maximum value. For spectral analysis of the glycogen spectra, the peaks in the Fourier-transformed spectra with 20-Hz line broadening were fitted as Lorentzian curves using the Advanced Method for Accurate, Robust, and Efficient Spectral fitting (AMARES) algorithm¹² in the jMRUI software.¹³ To compare the effect of ¹H decoupling, the height of the fitted ¹³C C1-glycogen peak was evaluated rather than the amplitude (area under the curve) and was calculated by dividing the amplitude by the linewidth, both of which were estimated by the fitting. Relative SAR (relSAR) was calculated by dividing the simulated SAR with a specific Phase-I duration by the SAR with full ¹H irradiation during ¹³C FID acquisition.

Results

The TPPM ¹H decoupling was operated using the clinical 3T scanner as shown below. Figure 1b shows a series of the ¹³C C1-glucose peaks from the glucose phantom as a function of the RF phase modulation angle for the TPPM ¹H decoupling. The ¹³C peaks are ¹H-decoupled and elevated at RF phase modulation angles up to 30° (Fig. 1c). Based on the results, 30° was chosen for the RF phase modulation angle in the subsequent experiments when using the TPPM ¹H decoupling. Figure 2 shows the ¹³C C1-glucose peaks from the glucose phantom using the TPPM, CW and WALTZ-16 ¹H-decoupling schemes. Cycling sidebands, which symmetrically flank the main decoupled peak, are only observed in the spectra using the WALTZ-16. The height of the largest cycling sideband is more than 8% of that of the main ¹H-decoupled ¹³C C1-β-glucose peak. Figure 3a shows a series of the ¹³C C1-glucose peaks from the glucose phantom as a function of the ¹H RF offset for the different ¹H-decoupling schemes. The ¹H-decoupling bandwidth of the TPPM is slightly wider than that of the CW (<200 Hz in both) and narrower than that of the WALTZ-16 (<900 Hz, Fig. 3b). Figure 4 shows the results using the glycogen phantom. The ¹³C C1-glycogen peaks are ¹H-decoupled using the TPPM (Fig. 4a). Using the TPPM, shortening Phase I to 8 ms does not reduce the height of the ¹H-decoupled ¹³C C1-glycogen peak (Figs. 4b and 4c). It reduced the simulated SAR to 39% compared to the SAR with full ¹H irradiation during ¹³C FID acquisition. No cycling sidebands are detected in ¹H-decoupled ¹³C glycogen spectra (Fig. 4c).

Fig. 2.

The C1-glucose peaks in ¹³C spectra from the glucose phantom using TPPM, CW and WALTZ-16 ¹H-decoupling schemes. The spectra are shown as the phase-sensitive display mode. The spectra in (b) are vertically magnified by a factor of 5. CW, continuous wave; TPPM, two-pulse phase-modulated.

Fig. 3.

(a) The C1-glucose peaks in ¹³C spectra from the glucose phantom as a function of the ¹H RF offset using TPPM, CW and WALTZ-16 ¹H-decoupling schemes. The spectra are shown as the phase-sensitive display mode and the display width is 1.5 kHz (= 48.4 ppm). (b) The peak height as a function of the ¹H RF offset in 100-Hz increments. The heights of the maximum C1-glucose peaks were measured in the absolute-value display mode and were averaged among three measurements. The corresponding standard deviations are smaller than the size of the points on the graph. CW, continuous wave; TPPM, two-pulse phase-modulated.

Fig. 4.

(a) ¹³C spectra from the oyster glycogen phantom with (top) and without (bottom) TPPM ¹H decoupling. (b) The average height of the ¹³C C1-glycogen peaks using different ¹H-decoupling schemes and Phase I durations (see Fig. 1a). T_n and W_n indicate that the ¹H-decoupling scheme (T: TPPM, W: WALTZ-16) and Phase I = n%. Error bars indicate the standard deviations (n = 3). (c) Representative ¹³C spectra with different Phase I durations of TPPM (top) and WALTZ-16 (bottom) ¹H decoupling. relSAR is calculated by dividing the simulated SAR by the SAR of Phase I = 100% with the same decoupling scheme. Line broadening = 20 Hz. TPPM, two-pulse phase-modulated

Discussion

This study demonstrated the implementation of the TPPM ¹H decoupling in the clinical 3T MR scanner. Despite the large difference in the B₂ amplitude required for solid-state NMR (>50 kHz)^9–10 and clinical MRI (<1 kHz),^{3,5,8,14–17} the TPPM pulse sequence in the clinical MRI was able to decouple proton spins with the same characteristics as the solid-state NMR, where small RF phase modulation angles are effective for ¹H decoupling.^9–10 This study was conducted with B₂ = 500 Hz, commonly chosen for human studies.^3,8,17 The decoupling bandwidth for the WALTZ-16 relative to B₂ showed similar values in the study by de Graaf (=1.64)⁸ and the present study (<1.8), both of which used half-volume ¹H coils. Thus, our experimental results using the TPPM ¹H decoupling could be reproduced on the MR system by de Graaf.⁸ The disadvantage of the TPPM ¹H decoupling in clinical MR scanners is the narrow decoupling bandwidth due to low B₂. Although the development of ¹H-decoupling schemes has been concentrated in the direction of expanding the decoupling bandwidth, de Graaf⁸ discussed that the direction for in vivo MRS may differ from that for analytical chemistry. The ¹H-decouping of the 2 ¹³C C1-glucose peaks in solution state using TPPM was successful, where the chemical shift difference of the ¹H H1-α- and H1-β-glucose peaks is 0.6 ppm, the ¹H-decoupling bandwidth of the TPPM was estimated to be wider than 70 Hz in this study. Therefore, the TPPM ¹H decoupling may be sufficient when only specific peaks are focused to be analyzed, such as the case of ¹³C C1-glycogen peak from the muscle.

The glycogen phantom study demonstrated that the ¹³C C1-glycogen peaks were able to be ¹H-decoupled using the TPPM without compromising the quality of ¹H decoupling, even when the duration of Phase-I was shortened to 8 ms in the pulse sequence used. In the measurement condition, the shortened duration of the TPPM ¹H decoupling reduced the SAR to less than 40% compared to full ¹H irradiation during ¹³C FID acquisition. The SAR reduction using the TPPM would be much more valuable at magnetic fields higher than 3T. As discussed by Stephenson et al.,¹⁸ detecting the C1-glycogen peak at 7T results in a 1.6-fold higher SNR compared to 3T and can reduce scan times by over 50%. However, including other applications using the WALTZ-16 at 7T,^5,17 TRs were always set much longer (0.8–1.0s) than the estimated T₁ of the ¹³C C1-glycogen peak in vivo (~0.2s)³ for the primary aim of reducing the SAR. If the narrow ¹H-decoupling bandwidth is sufficient, the WALTZ-16 would be replaced with the TPPM. By further shortening the TR and averaging a large number of FIDs within a limited scan time, the noise level can be reduced and the SNR improved. If no noise spikes are detected in the MR system when the decoupling power is turned off momentarily, the SAR can be further reduced without using Phase-II in the pulse sequence. Another advantage of the TPPM at 7T may be the small maximum RF flip angle: 180° for the TPPM versus 360° for the WALTZ-16. That is extremely important at 7T, where uniform RF excitation is difficult to achieve at large RF flip angles. Clinically approved 7T MRI for humans is not widely available, and studies of muscle glycogen using ¹H-decoupled ¹³C MRS have even fewer reports at 7T.^{3,5,17–18 13}C MRS using the low-SAR TPPM ¹H-decoupling pulse sequence would be highly suitable for studying glycogen metabolism in in vivo human muscle at 7T.

Unlike the WALTZ-16, the cycling sidebands were not observed in the ¹H-decoupled ¹³C spectra from the glucose phantom using the TPPM as well as the CW. The reduction of the undesirable cycling sidebands was attempted with the development of ¹H-decoupling schemes for clinical applications.¹⁴ The cycling sidebands are particularly amplified when an ¹H-decoupling cycle based on composite pulses is longer than the inverse of the heteronuclear J-coupling constant,¹⁴ e.g., 6 ms for the C1-peak of glucose and glycogen, which suggests low efficiency of the ¹H-decoupling compensation cycle. The TPPM ¹H-decoupling cycle at 500-Hz B₂ is 2 ms, which may be the reason why no cycling sidebands were observed using the TPPM in this study. The cycling sidebands are also amplified under B₂ inhomogeneity even in the solid-state NMR.⁹ The small RF flip angle requirement of the TPPM may be another reason for no cycling sidebands having been observed. No cycling sidebands were detected in ¹³C glycogen spectra using TPPM or WALTZ-16 due to high noise levels, however the peak intensities tend to be lower using WALTZ-16, likely due to leakage of signal amplitude into the cycling sidebands.

The TPPM pulse sequence has been improved^9–10 and is widely used for ¹H decoupling in ¹³C NMR of organic solids. Since its improvements mainly focus on the dipolar interactions, which is not likely to be strong in liquid state or biological samples, and since the composite pulse duration of such evolved pulse sequences tends to be long, only the TPPM pulse sequence was implemented in this study. The evolved TPPM pulse sequences have been implemented for water suppression in ¹H-detected multidimensional NMR for protein structure determination¹⁹ and for electron spectral saturation in dynamic nuclear polarization.²⁰ The aim of implementing the TPPM in the latter study was to minimize heating of the sample, which is similar to our in vivo ¹³C MRS study. In addition to the ¹H-decoupling application demonstrated in this study, the TPPM pulse sequence could also be used for other purposes in in vivo MR studies. Possible candidates would be water suppression for ¹H MRS and narrow spectral saturation for CEST imaging, magnetization transfer ³¹P MRS and j-editing ¹H MRS.

Conclusion

The TPPM pulse sequence is applicable to clinical MR scanners as a ¹H-decoupling method to detect the C1-glycogen peak in ¹³C MRS. The performance of this sequence has been confirmed using the RF coils designed for human calf muscle, and therefore the sequence should be suitable for future in vivo applications. The low-SAR sequence may be more valuable at 7T.

Supplementary Material

Supplementary figures are available online.

Supplementary Figure 1.

RF coils for ¹³C MR measurements of human calf muscle with ¹H imaging and decoupling capabilities. The linear-¹³C/quadrature-¹H RF coils (¹³C: a 75-mm loop, ¹H: a pair of 95-mm loops) were formed on the surface of semi-cylindrical acrylic resin tubes with radii of 71 mm and 77 mm, respectively.