We present a protocol to measure the magnetic field dependence of the spin-lattice relaxation time of 13C-enriched compounds, hyperpolarized by means of dynamic nuclear polarization, using fast field-cycled relaxometry. Specifically, we have demonstrated this with [1-13C]pyruvate, but the protocol could be extended to other hyperpolarized substrates.
The fundamental limit to in vivo imaging applications of hyperpolarized 13C-enriched compounds is their finite spin-lattice relaxation times. Various factors affect the relaxation rates, such as buffer composition, solution pH, temperature, and magnetic field. In this last regard, the spin-lattice relaxation time can be measured at clinical field strengths, but at lower fields, where these compounds are dispensed from the polarizer and transported to the MRI, the relaxation is even faster and difficult to measure. To have a better understanding of the amount of magnetization lost during transport, we used fast field-cycling relaxometry, with magnetic resonance detection of 13C nuclei at ~0.75 T, to measure the nuclear magnetic resonance dispersion of the spin-lattice relaxation time of hyperpolarized [1-13C]pyruvate. Dissolution dynamic nuclear polarization was used to produce hyperpolarized samples of pyruvate at a concentration of 80 mmol/L and physiological pH (~7.8). These solutions were rapidly transferred to a fast field-cycling relaxometer so that relaxation of the sample magnetization could be measured as a function of time using a calibrated small flip angle (3°-5°). To map the T1 dispersion of the C-1 of pyruvate, we recorded data for different relaxation fields ranging between 0.237 mT and 0.705 T. With this information, we determined an empirical equation to estimate the spin-lattice relaxation of the hyperpolarized substrate within the mentioned range of magnetic fields. These results can be used to predict the amount of magnetization lost during transport and to improve experimental designs to minimize signal loss.
Magnetic resonance spectroscopic imaging (MRSI) can produce spatial maps of metabolites detected by spectroscopic imaging, but its practical use is often limited by its relatively low sensitivity. This low sensitivity of in vivo magnetic resonance imaging and spectroscopy methods stems from the small degree of nuclear magnetization achievable at body temperatures and reasonable magnetic field strengths. However, this limitation can be overcome by the use of dynamic nuclear polarization (DNP) to greatly enhance the in vitro magnetization of liquid substrates, which are subsequently injected to probe in vivo metabolism using MRSI1,2,3,4. DNP is capable of enhancing the magnetization of most nuclei with non-zero nuclear spin and has been used to increase in vivo MRSI sensitivity of 13C-enriched compounds such as pyruvate5,6, bicarbonate7,8, fumarate9, lactate10, glutamine11, and others by more than four orders of magnitude12. Its applications include imaging of vascular disease13,14,15, organ perfusion13,16,17,18, cancer detection1,19,20,21,22, tumor staging23,24, and quantification of therapeutic response2,6,23,24,25,26.
Slow spin-lattice relaxation is essential for in vivo detection with MRSI. Spin-lattice relaxation times (T1s) on the order of tens of seconds are possible for nuclei with low gyromagnetic ratios within small molecules in solution. Several physical factors influence the transfer of energy between a nuclear spin transition and its environment (lattice) leading to relaxation, including the magnetic field strength, temperature, and molecular conformation27. Dipolar relaxation is reduced in molecules for carbon positions with no protons directly attached, and deuteration of dissolution media can further reduce intermolecular dipolar relaxation. Unfortunately, deuterated solvents have limited abilities to extend in vivo relaxation. Increased relaxation of carbonyls or carboxylic acids (such as pyruvate) can occur at high magnetic field strengths due to chemical shift anisotropy. The presence of paramagnetic impurities from the fluid path during dissolution after polarization can cause rapid relaxation and need to be avoided or eliminated using chelators.
Very little data exist for the relaxation of 13C-containing compounds at low fields, where spin-lattice relaxation could be significantly faster. However, it is important to measure T1 at low fields to understand relaxation during preparation of the agent used for in vivo imaging, since the hyperpolarized contrast agents are usually dispensed from the DNP apparatus near or at the earth’s field. Additional physical factors such as 13C-enriched substrate concentration, solution pH, buffers and temperature also influence relaxation, and consequently have an effect on the formulation of the agent. All these factors are essential in the determination of key parameters in optimizing the DNP dissolution process, and the calculation of the magnitude of signal loss that occurs in transportation of the sample from the DNP apparatus to the imaging magnet.
Nuclear magnetic resonance dispersion (NMRD) measurements, i.e., T1 measurements, as a function of magnetic field are typically acquired using an NMR spectrometer. To acquire these measurements, a shuttling method could be used where the sample is first shuttled out of the spectrometer to relax at some field determined by its position in the fringe field of the magnet28,29,30 and then rapidly transferred back into the NMR magnet to measure its remaining magnetization. By repeating this process at the same point in the magnetic field but with increasing periods of relaxation, a relaxation curve can be obtained, which can then be analyzed to estimate T1.
We use an alternative technique known as fast field-cycling relaxometry31,32,33 to acquire our NMRD data. We have modified a commercial field-cycling relaxometer (see Table of Materials), for T1 measurements of solutions containing hyperpolarized 13C nuclei. Compared with the shuttle method, field-cycling enables this relaxometer to systematically acquire NMRD data over a smaller range of magnetic fields (0.25 mT to 1 T). This is accomplished by rapidly changing the magnetic field itself, not the sample location in the magnetic field. Therefore, a sample can be magnetized at a high field strength, "relaxed" at a lower field strength, and then measured by acquisition of a free-induction-decay at a fixed field (and Larmor frequency) to maximize signal. This means that the sample temperature can be controlled during the measurement, and the NMR probe does not need to be tuned at each relaxation field promoting automatic acquisition over the entire magnetic field range.
Focusing our efforts to the effects of dispensing and transporting the hyperpolarized solutions at low magnetic fields, this work presents a detailed methodology to measure the spin-lattice relaxation time of hyperpolarized 13C-pyruvate using fast field-cycling relaxometry for magnetic fields in the range of 0.237 mT to 0.705 T. The main results of using this methodology have been previously presented for [1-13C]pyruvate34 and 13C-enriched sodium and cesium bicarbonate35 where other factors such as radical concentration and dissolution pH have also been studied.
1. Sample Preparation
NOTE: Steps 1.1-1.8 are performed just once
2. Relaxometry
NOTE Please refer to Table 1 for a better understanding of the selection and use of the different parameters described in the following steps. Prior to dissolution, the relaxometer flip angle must be calculate and the relaxometer must be setup and ready for measurement of the hyperpolarized solution (see below).
Figure 2 presents an example of a high-resolution full-range microwave sweep for pyruvic acid. For the presented case, that optimal microwave frequency corresponds to 94.128 GHz, highlighted in the figure insert. Our DNP system can normally work in the range of 93.750 GHz to 94.241 GHz with step size of 1 MHz, polarization time of up to 600 s, and power of up to 100 mW. A full range of frequencies is investigated only for novel substrates. However, based on previous experience with 13C-pyruvic acid, we expect the optimal frequency to be around 94.127 GHz. Therefore, a scan range between 94.117 GHz to 94.137 GHz, with a step size of 1 MHz and a sampling time of 300 s with 50 mW of power, are typically used.
The left column of Figure 3 presents the results for the tip angle calibration for [1-13C]pyruvic acid, which involves acquisition of a series of signal measurements as a function of a linearly varying RF pulse durations to determine the pulse width corresponding to a flip angle of 90° and 180° for 13C nuclei. The pulse width that provides the maximum amplitude corresponds to a flip angle of 90° and the zero crossing corresponds to a flip angle of 180°. The relationship between the two pulse widths should be a factor of two.
The acquisition parameters for the 13C tip angle calibration shown above may require some adjustments depending on the transmit power of the field-cycle relaxometer, the T1 of the sample, and the noise characteristic of the system. Some trial and error may be needed as well to properly find the 90° and 180° without the effects of stimulated echoes, amplifier saturation, and poor SNR.
This procedure, although accurate, is normally time consuming because the poor SNR of thermally polarized 13C compounds requires many averages. An alternative and faster method involves calibrating the flip angle with a gadolinium-doped 1H phantom and scaling the duration of the 90° RF pulse for 13C by multiplying the duration of the 90°-1H RF pulse by the ratio of the gyromagnetic ratios of 1H/13C, which corresponds to a factor of 3.976. For this case, the standard acquisition parameters should be: EXP = ANGLE.FFC, NUC = 1H, TPOL = 0.1 s, BPOL = 30 MHz, SWT = 0.005, BINI = 0 µs, BEND = 15.5 µs, NBLK = 32, MS = 1, RFA = 25, RD = 0.1 s, BS = 652, SW = 1 MHz, FLTR = 100 KHz, SF = 8, RINH = 25, ACQD = 25, EWIP = 10, EWEP = 512, EWIB = 1, and EWEB = 32. The results for this alternative method are shown in the right column of Figure 3. As a comparison, for the presented cases, the total acquisition time for tip angle calibration for 13C was 13.5 minutes whereas for 1H was 7.1 seconds.
Figure 5 illustrates the typical series of decaying FIDs as the hyperpolarized magnetization is sampled. Each T1 measurement at a given BRLX is a separate hyperpolarized dissolution from the DNP apparatus. For this particular case, the relaxation field (BRelax) was 0.2916 mT, with a repetition time of 3.4 s and a flip angle of 5°. All sample temperatures were controlled to 37 °C (±0.5 °C).
Figure 6 presents the relaxation curve for hyperpolarized [1-13C]pyruvate obtained from the data of the previous figure. Each blue point on the curve represents the area under an FID. The T1 value (53.9 ± 0.6s) was obtained by a non-linear least-squares fit of the signal equation to the decay curve data, which included the effects of the flip angle used for excitation. The goodness of fit was assessed by computing the R2 value (0.9995), assuming even weighting of the data points. Fitting residuals (data-fit) are shown as open triangles.
Figure 7 presents the T1 results for all 26 measurements over a range of 0.237 mT and 0.705 T at 37 °C (±0.5 °C). The T1 had an average fitting uncertainty of ±0.33 s for all the results. Analysis of the scatter of measurements repeated at a particular relaxation field yielded an experimental reproducibility several times larger than the statistical uncertainty quoted above, with a T1 of 1.91 s. An uncertainty of 2.24 s was conservatively assigned for all T1 measurements calculated as the sum of the two uncertainties quoted above. The T1-dispersion data are well characterized by the empirical formula T1 = (3.74 ± 0.52) x log10(BRelax) + (63.0 ± 1.2) s; where BRelax is the relaxation field measured in Tesla. The uncertainties for the fitted parameters represent one standard deviation. The solid line on Figure 7 represents the formula along with the dashed lines representing the 95% confidence bands. pHs for these samples ranged from 7.63 to 7.93, with an average pH of 7.75 and a standard deviation of 0.09. Analysis of the results showed that the relaxation time for the C-1 nucleus is ~ 46.9 s at earth’s magnetic field (0.05 mT) compared with ~ 65 s at 3 T, which represents a decrease of 28%.
Figure 1: [1-13C]pyruvic acid molecule. Please click here to view a larger version of this figure.
Figure 2: Full-range microwave sweep and zoom-in section showing the optimal polarization frequency. Please click here to view a larger version of this figure.
Figure 3: Tip angle calibration for 13C (left) and 1H (right) samples. Please click here to view a larger version of this figure.
Figure 4: Field-cycled pulse sequence (HPUB/S) to measure the T1-relaxation time of a hyperpolarized sample at a particular relaxation field (BRLX). Please click here to view a larger version of this figure.
Figure 5: Sequence of FIDs obtained with the HPUB/S pulse sequence. Please click here to view a larger version of this figure.
Figure 6: Relaxation signal (blue dots), curve fitting (red line), and fitting error (open triangles) obtained from the sequence of FIDs presented in Figure 5. This figure has been modified with permission from Chattergoon et al. 201334. Please click here to view a larger version of this figure.
Figure 7: NMRD profile of hyperpolarized [1-13C]pyruvic acid at low magnetic fields. This figure has been modified with permission from Chattergoon et al. 201334. Please click here to view a larger version of this figure.
Parameter | Short Description | コメント | Units |
ACQD | Acquisition delay | Delay required to allow magnetic field to reach steady state after transition and before data acquisition | µs |
BACQ | Acquisition Field | Specified by means of 1H Larmor frequency | MHz |
BEND | End value | Final value of the arrayed parameter | |
BINI | Initial value | First value of the arrayed parameter | |
BPOL | Polarization Field | Specified by means of 1H Larmor frequency | MHz |
BRLX | Relaxation Field | Specified by means of 1H Larmor frequency | MHz |
BS | Block size | Number of data points in a single block | |
EWEB | End block | Any integer number in the range of Number of Blocks (NBLK). 0 means "all" | |
EWEP | End point | Any integer number in the range of Block size (BS). 0 means "all" | |
EWIB | Initial block | From 1 to number of blocks (NBLK) | |
EWEP | Initial point | From 1 to block size (BS) | |
EXP | Experiment | Name of pulse sequence to be used | |
FLTR | Observe filter | Cutoff frequency of the audio signal filters | Hz |
MS | Maximum scans | Desired number of averages | |
NBLK | Number of blocks | Number of sections for the arrayed parameter. The arrayed parameter is "PW90" for "13CANGLE" and "ANGLE" pulse sequences and "T1MX" for "HPUB/S" pulse sequence. PW90 changes after each repetition but T1MX remains constant. | |
NUC | Nucleus | For this protocol 13C or 1H | |
PW | Main RF pulse | Tip angle | Degrees (°) |
PW90 | 90deg pulse | Duration of the 90-degrees pulse | µs |
RD | Recycle delay | Pre-scan magnet-cooling interval | s |
RFA | RF attenuation | RF receiver attenuation | dB |
RINH | Receiver inhibit | Delay required to allow the decay of RF-coil ringing | µs |
SF | System Frequency | Larmor frequency used during acquisition | MHz |
SW | Sweep Width | Spectral window width (Nyquist Frequency) | Hz |
SWT | Switching time | Global magnet-switching time | s |
T1MX | Maximum T1 | Parameter used by the HPUB/S pulse sequence to define the polarization time during each repetition | s |
TPOL | Polarization time | Parameter used by the "ANGLE" and "13CANGLE" pulse sequence to define the polarization time during each repetition | s |
Table 1: Description of parameters used by the field-cycling relaxometer.
The use of DNP to enhance signal acquisition is a technical solution to insufficient magnetic resonance signal available from 13C nuclei at limited concentrations, as those used in animal injections, but presents other experimental challenges. Each relaxation measurement shown in Figure 7 represents a measurement of a uniquely prepared sample because it cannot be repolarized after dissolution for remeasurement. This inevitably leads to experimental variability due to minor differences in sample preparation during weighing of the sample and dissolution media or variations in the dissolution process itself, such as incomplete extraction and thorough mixing of the sample with the dissolution media. This variability may be partially assessed by measuring the pH of each pyruvate solution after relaxometry. Regardless of careful weighing of stock pyruvate/radical mixture and dissolution medium before insertion in the DNP apparatus to better than a milligram, in our experiments the pHs ranged from 5.5 to 8.3. We have chosen to reject any T1 data outside the pH range 7.6 to 8.0.
As mentioned above, the solid-state polarization level for each sample was at least 95%, which was obtained in about one hour. The liquid-state polarization was not estimated for each sample; however, periodic quality assurance of the DNP system, using the same sample preparation, resulted in liquid state polarization levels of about 15%.
During sample preparation, metal ion contamination may occur from contact between the dissolution medium and the DNP dissolution fluid path. This possibility required the addition of disodium ethylenediaminetetraacetic acid (EDTA) to sequester any of the metal ion contamination and preserve spin-lattice relaxation.
Comparing the shuttling method used in reference28 and the fast field cycling presented in this protocol, we can say that the shuttling method is only possible when the shuttle time is small in comparison to the relaxation time; otherwise, the average magnetic fields experienced during the shuttling time may have a significant effect. With the fast field cycling relaxometer we used, the user is in complete control of the switching time, which can go as low as 3 ms. However, for hyperpolarized substrates, a slow switching time is required to keep adiabaticity and not to destroy the polarization of the sample during filed transitions. In our experience, for hyperpolarized 13C-pyruvic acid, a switching time as low as 50 ms does maintain the polarization, but we observed more consistent results using a switching time of 100 or 200 ms. This small transition time from relaxation to acquisition and back to relaxation fields is negligible in comparison to the measured T1 times and has no systematic effect on these measurements. We consider that further research is required to establish the boundaries of adiabaticity of different hyperpolarized substrates at different magnetic fields.
Another important difference between the two methods is the range of magnetic fields, which is 2 mT to 18.8 T for the shuttling method and 0.237 mT to 0.705 T for the field cycling relaxometer. In this regard we can see the two methods as complementary to each other. However, for in vivo studies with hyperpolarized compounds, magnetic fields of up to 3 T are more common.
At field strengths of less than 1 mT, stray magnetic fields from surrounding objects were observed to have a systematic effect on our relaxation measurements. To eliminate these fields, we designed and added a custom magnetic shim around the field-cycling magnet. In comparison, the shuttling method uses µ-metal cylindrical shielding that produces an abrupt change of magnetic field from about 2 mT to 0.2 mT.
Temperature control of the sample was important due to the relatively long acquisition times requiring 300 to 510 s to capture the entire decay curve. We pre-warmed the NMR tubes prior to dispensing the hyperpolarized solution and then maintained the sample temperature by blowing warmed, temperature-regulated (37 °C) air over the tubes during relaxometry. This is an important advantage of the field-cycling relaxometer over the shuttling method because the temperature of the sample can be precisely controlled since the sample is stationary during measurements.
In addition, it was not practical to control the sample exposure to ambient temperature and magnetic field during the brief transfer time between polarizer and relaxometer. The T1 of samples were measured at known magnetic fields and temperature controlled by the relaxometer, so transportation had limited influence. Conditions during transportation can only affect the amount of hyperpolarization that survives for measurement at the relaxometer. A portable holding field magnet (10 mT) was developed for transferring the hyperpolarized solution to the imaging magnet or relaxometer; however, its use was not worthwhile in this experiment given the brief transfer time but may be useful for other hyperpolarized liquids with greater T1-dispersion at lower magnetic fields. A holding field of 0.01 T would increase the T1 of the pyruvate solution by nearly 18% during transportation; however, with our relatively short transfer time of 8 s, these measurements suggest that only a 2.3% increase in signal would be observed.
The authors have nothing to disclose.
The authors would like to thank the Ontario Institute for Cancer Research, Imaging Translation Program and the Natural Sciences and Engineering Research Council of Canada for funding this research. We also like to acknowledge useful discussions with Albert Chen, GE Healthcare, Toronto, Canada, Gianni Ferrante, Stelar s.r.l., Italy, and William Mander, Oxford Instruments, UK.
[1-13C]Pyruvic Acid | Sigma-Aldrich, St. Louis, MO, USA | 677175 | |
10mm NMR Tube | Norell, Inc., Morganton NC, USA | 1001-8 | |
De-ionized water | |||
Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA) | Sigma-Aldrich, St. Louis, MO, USA | E5134 | |
HyperSense Dynamic Nuclear Polarizer | Oxford Instruments, Abingdon, UK | Includes the following: "DNP-NMR Polarizer" software used to control and monitor the whole DNP polarizer; "RINMR" used to monitor the solid state polarization levels; "HyperTerminal" used to communicate the DNP software with the RINMR software that monitors the solid state polarization level. Also includes the MQC bench top spectrometer to monitor the liquid state polarization in conjunction with it own RINMR software | |
MATLAB R2017b | MathWorks, Natick, MA | Include scripts for non-linear fitting of magnetization decay over time and T1 NMRD analysis of hyperpolarized pyruvic acid. | |
OX063 Triarylmethyl radical | Oxford Instruments, Abingdon, UK | ||
pH meter – SympHony | VWR International, Mississauga, ON., Canada | SB70P | |
ProHance | Bracco Diagnostics Inc. | Gadoteridol, Gd-HP-DO3A | |
Pure Ethanol (100% pure) | Commercial Alcohols, Toronto, ON, Canada | P016EAAN | |
Shim Coil | Developed in-house | ||
Sodium Chloride | Sigma-Aldrich, St. Louis, MO, USA | S7653 | |
Sodium Hydroxide | Sigma-Aldrich, St. Louis, MO, USA | S8045 | |
SpinMaster FFC2000 1T C/DC | Stelar s.r.l., Mede (PV) Italy | Includes the software "AcqNMR" that is used to set experimental parameters, monitor the tuning and matching of the RF coil, loading different pulse sequences, calibrate flip angle, data acquisition and curve fitting, among other functions. Also includes a depth gauge, some weights and a depth stopper. | |
Trizma Pre-Set Crystals (pH 7.6) | Sigma-Aldrich, St. Louis, MO, USA | T7943 |